You are on page 1of 97

PORTA LINGUARUM ORIENTALIUM

HERAUSGEGEBEN VON BERTOLD SPULER UND HANS WEHR
NEUE SERlE
VI
AN INTRODUCTION
TO
THE COMPARATIVE GRAMMAR
OF
THE SEMITIC LANGUAGES
PHONOLOGY AND MORPHOLOGY
THIRD PRINTING
1980
OTTO HARRASSOWITZ ·WIESBADEN
AN INTRODUCTION
TO
THE COMPARATIVE GRAMMAR
OF
THE SEMITIC LANGUAGES
PHONOLOGY AND MORPHOLOGY
BY
SABATINO MOSCATI
ANTON SPITALER EDWARD ULLENDORFF
WOLFRAM VON SODEN
EDITED BY
SABATINO MOSCATI
THIRD' PRINTING
1980
811.411
MOSCATIS introduC
512527000001
OTTO HARRASSOWITZ· WIESBADEN
AIle Rechte vorbehalten
© Otto Harrassowitz, Wiesbaden 1964, 1969, 1980
Photographische und photomechanische Wiedergaben jeder Art
nur mit ausdriicklicher Genehmigung des Verlages
Gesamtherstellung: Hessische Druckerei GmbH, Darmstadt
Printed in Germany
Table of Contents
page
Preface............ .................................. ...... .. 1
I. The Semitic Languages ............................... . . . . 3
A. Scope of the Survey. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1. Definition ... .................................... 3
2. Classification.......................... . . . . . . . . . . . . 3
3. Nature and Extent of the Survey. . . . . . . . . . . . . . . . . . . . 5
B. North-East Semitic ................................... 6
C. North-West Semitic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1. General Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2. The Languages of the Second Millennium B.C. ........ 8
3. Canaanite......................................... 9
4. Aramaic....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
a. Old Aramaic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
b. West Aramaic .............................. " . . 11
c. East Aramaic.... . .... .. .. .......... ........ .... 12
D. South-West Semitic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1. General Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2. Arabic (incI. South Arabian). . . . . . . . . . . . . . . . . . . . . . . . 13
3. Ethiopic...................................... . . . . 15
E. Proto-Semitic, Hamito-Semitic, Indo-European. . . . . . . . . . . 15
1. Proto-Semitic........ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2. Hamito-Semitic ..... ................ ..... ..... .. .. 16
3. Hamito-Semitic and Indo-European. . . . . . . . . . . . . . . . . . 17
F. Language and Script .................................. 17
II. Phonology............................................... 22
A. Preliminaries ........................................ .
B. The Phonological System .............................. .
1. Classification ..................................... .
2. The Proto-Semitic Consonantal System ............. .
3. Bilabials ........................................ .
22
23
23
24
24
VI
Table of Contents
4. Interdentals ...................................... .
5. Dental Plosives .................................. .
6. Nasal, Lateral, and Rolled Dentals ................. .
7. Dental and Palato-alveolar Fricatives ............... .
8. Velar Plosives .................................... .
9. Velar Fricatives .................................. .
10. Pharyngal Fricatives; Laryngals .................... .
11. Synopsis of the Consonantal System ................ .
12. Semivowels ...................................... .
13. Vowels .......................................... .
a. Akkadian ..................................... .
b. North-West Semitic of the Second Millennium B.C ..
c. Canaanite ..................................... .
d. Aramaic ...................................... .
e. Arabic ........................................ .
f. Ethiopic ...................................... .
14. Diphthongs ...................................... .
C. Conditioned Phonetic Changes ......................... .
1. Assimilation ..................................... .
2. Dissimilation ..................................... .
3. Prosthesis ....................................... .
4. Anaptyxis ....................................... .
5. Syncope and Contraction .......................... .
6. Haplology ....................................... .
7. Metathesis ....................................... .
8. Sandhi .......................................... .
D. Syllable and Stress ................................... .
1. Syllabic Constitution .............................. .
2. Stress and Associated Changes ..................... .
3. Sentence Stress ................................... .
III. Morphology ............................................ .
A. Preliminaries ........................................ .
1. Morphemes ...................................... .
2. The Proto-Semitic Root ........................... .
3. Morphological Development ........................ .
B. The Noun ........................................... .
1. Themes or Patterns ............................... .
2. Nominal Patterns ................................ .
a. Simple Patterns ............................... .
b. Patterns Extended by Gemination or Reduplication
of Radicals .................................... .
27
31
31
33
37
38
41
43
45
46
47
48
49
51
52
53
54
56
56
58
59
60
61
62
63
63
63
63
65
69
71
71
71
72
75
75
75
76
77
78
Table of Contents
c. Patterns Extended by Prefixes .................. .
d. Patterns Extended by Infixes ................... .
e. Patterns Extended by Suffixes ................... .
f. Patterns from Roots with One, Two, Four, and Five
Radicals ...................................... .
3. Gender .......................................... .
4. Number ......................................... .
a. External Masculine Plural ...................... .
b. Internal Masculine Plural ....................... .
c. Feminine Plural ............................... .
d. Plurals of Biconsonantal Nouns .................. .
e. Dual ......................................... .
5. Declension ........................................
6. Definiteness and Indefiniteness ..................... .
C. The Pronoun .........................................
1. Independent Personal Pronouns .................... .
2. Personal Pronoun Suffixes ......................... .
3. Demonstrative Pronouns .......................... .
4. Relative Pronouns ................................ .
50. Interrogative and Indefinite Pronouns .............. .
D. The Numeral ........................................ .
1. Cardinals ........................................ .
2. Ordinals and Fractions ............................ .
E. The Particles ........................................ .
1. Adverbs ......................................... .
2. Prepositions ...................................... .
3. Conjunctions ..................................... .
4. Interjections ..................................... .
F. The Verb ............................................ .
1. Verbal Themes or Stems ...........................
a. Simple Stem .................................. .
b. Stem with Doubled Second Radical .............. .
c. Stem with Lengthened First Vowel .............. .
d. Stems with Prefixes SO, h-, '- ................... .
e. Stem with Prefix n- ............................ .
f. Stems with Prefix (or Infix) t- ................... .
g. Other Stems .................................. .
h. Verbs with Four and Five Radicals .............. .
2. The "Tenses" •••••••••••• 0 ••••••••••••••••••••••••
3. The ~ : [ o o d s ...................................... .
4. Inflexion .........................................
VII
80
82
82
83
84
86
87
88
91
92
93
94
96
102
102
106
III
113
114
115
115
lI8
120
120
121
121
122
122
122
122
124
124
125
126
127
129
130
131
134
137
VIII Table of Content,;
a. Simple Stem: Suffix-Conjugation ................ .
b. Simple Stem: Prefix-Conjugation ....... . ....... .
c. Simple Stem: Imperative ....................... .
d. Simple Stem: Nominal Forms ................... .
e. Derived Stems: Suffix-Conjugation .............. .
f. Derived Stems: Prefix-Conjugation .............. .
g. Derived Stems: Imperative ...................... .
h. Derived Stems: Participle ...................... .
i. Derived Stems: Infinitive ....................... .
5. The so-called "Weak" Verbs ....................... .
6. Verbs with Pharyngals and Laryngals ............... .
7. Verbs with First Radical n ........................ .
8. Verbs with w, y .................................. .
9. Verbs with Identical Second and Third Radicals ...... .
10. Doubly Irregular or Defective Verbs ................ .
11. Semantic Categories in "Weak" Verbs .............. .
12. Verbs with Pronominal Suffixes .................... .
Bibliography ............................................... .
List of Abbreviations used in the Text ......................... .
137
141
145
146
147
154
156
157
158
159
160
163
164
166
168
168
169
171
185
Preface
The nature of this book and the principles which guided its composition
depend in the first place on the series to which it belongs. It is thus an
elementary introduction to the comparative grammar of the Semitic lan-
guages, intended primarily as a textbook and limited in its scope so as to
serve for a beginners' course. Such an introduction must be clear in
expression while respecting scientific terminology; it must concentrate upon
the essential facts while mentioning various particular questions of special
importance; it must avoid doubtful and disputed hypotheses while indicat-
ing certain of the lines of research being pursued and certain more important
divergences of opinion, so as to give an adequate notion of present-day
scientific trends.
To these general principles dependent upon the present series, others
must be added, adopted in view of the particular situation of Semitic
linguistics. The limits in time and space of the material to be taken into
consideration, the classification of the various languages, the manner for
reconstructing the presumed common forms and tracing their development
in the various languages, the reasons for abstaining from a systematic treat-
ment of syntax and of vocabulary, the choice of a system of transliteration,
the solution of the problem of a conventional root for schemes and for para-
digms, the dates up to which the bibliography can be taken into account:
these questions, and others too, called for decisions, and these decisions are
set forth in the course of the exposition (cf. in particular §§ 1.1-9, 6.1-15,
8.66-96, 12.3, introduction to the Bibliography etc.). It must be admit-
ted that many of these decisions are but imperfect ones, but one may well
doubt whether different decisions would have been any more felicitous, or
even equally so.
As is well known, an introduction to the comparative grammar of the
Semitic languages has been a desideratum for several decades, and the need
for such an introduction as a textbook was urgent. Hence, when the directors
of Porta Linguarum Orientalium did me the honour of inviting me to VlTite
this book (an honour for which I offer them my sincere thanks), I was of two
minds. On the one hand, I was but too well aware of the need for someone
at long last to undertake the task; but on the other hand, it was equally
clear to me that the accomplishment of this task, involving as it did the
digestion of an immense linguistic and bibliographical material and its
condensation into an exposition which by reason of its brevity could not
but be partial imd simplified, would inevitably be open to criticism from all
sides. 'When the author is obliged to sift his matter, to make selections, to
confine his exposition within strict limits, then anyone can make a running
criticism of the result and easily cite imperfections and omissions; further-
more, not everyone is ready to recognize the sacrifice represented by the
undertaking of a thankless task whose accomplishment is, after all, indis-
pensable for the common good.
Moscati, Comparative Grammar 1
2
Preface
If, however, I was thus able from the start to foresee the endless difficulties
and the i.nevitable deficiencies connected with such an enterprise, I was
likewise able to endeavour to reduce those difficulties and deficiencies to
a minimum. The procedure which I adopted was the following: I first
composed as best I could a preliminary and summary draft (Lezioni di
linguistica semitica, Rome 1960), and I then requested certain eminently
qualified colleagues to lend me their aid by snggesting, with particular but
not exclusive regard for each one's speciality, an abundance of additions
and alterations. These colleagues, who most generously responded to my
appeal for their collaboration, are Professors A. Spitaler, E. Ullendorff and
W. von Soden; and to them I express my heartfelt gratitude. Having
assembled the list of additions and alterations thus proposed, I proceeded
to a new and fuller redaction of the book, bearing in mind the material
thus furnished. It is, of course, clear that not in every single case was it
possible to bring my views into harmony with those of my three collabo-
rators; this obliged me to make a certain selection, and I have not always
been able to accord full weight to their proposals; it is therefore only just
that I bear the entire responsibility for this work. However, I must add
that I have very largely taken their proposals into account, even in not
a few cases in which I was not in agreement with them, so that the work
is truly the result of a collaboration.
Other eminently qualified colleagues have likewise read my preliminary
draft and favoured me with their suggestions and opinions, namely Pro-
fessors K. Petracek, C. Rabin and S. Segert, to whom I offer my sincere
gratitude. lowe a like debt of gratitude to Professor B. Migliorini, for
his advice in the area of general linguistics, and to Professor T. de Mauro,
who has likewise advised me in this respect. Finally, lowe a special
gratitude to the Publishers, for having so willingly undertaken the not
inconsiderable labour involved in a work composed with the collaboration
of several persons.
Now that the book is written, I can in all frankness declare that I find it
unsatisfactory; and I think that my collaborators would agree with me.
Each one of us could, if he wished, write dozens of pages of criticism of
the book, for the reasons already given. "Ve trust, however, that our
colleagues, in reading this book, instead of dwelling upon the ad:mitted
manifold defects will duly reflect upon the fact that in a work such as this
these defects, were almost inevitable, and that different solutions of the
problems involved would in all probability have been equally unsatisfactory.
We trust, indeed, that our colleagues will take into consideration the fact
that at last, after so many decades, they have at their disposal an elementary
textbook - although necessarily defective - of the comparative grammar
of the Semitic languages.
October 1962 Sabatino Moscati
I. The Semitic Languages
A. Scope of the Survey
1. Definition
1.1. The name "Semitic" is conventionally applied to a group
of languages spoken in western Asia, or generally originating from
that area, and characterized by a large number of common elements
in their phonology, morphology, vocabulary, and syntax; they also
share certain common tendencies in their evolution. These ele-
ments, preserved despite lapse of time and change of place, suggest
the idea of a common origin; at all events, they characterize and
set apart a linguistic group possessed of a remarkable degree of
internal unity.
1.2. The adjective "Semitic" was brought into use by A. L.
Schl6zer (1781) as a designation of the languages spoken by the
Aramaeans, the Hebrews, the Arabs, and other peoples, on the
basis of Gen. 10, 21-31; 11, 10-26. Once introduced, the adject-
ive "Semitic" was applied to all the languages of the group,
including those subsequently discovered. The affinities existing
between the various languages had, of course, been recognized long
before Sch16zer's time; but the group itself had not as such been
identified and marked out: these languages, like others in Asia,
had generally been referred to as "Oriental languages".
2. Classification
1.3. The Semitic languages occupied in ancient times the follow-
ing regions of western Asia (from east to west): Mesopotamia,
Syria-Palestine, Arabia. On the coast that lies opposite south-
western Arabia, waves of migration led to the occupation by
Semitic populations (and thus by Semitic languages) of another
region: Ethiopia. Mesopotamia, Syria-Palestine, Arabia, and
Ethiopia constitute, therefore, the ancient habitat of the Semitic
1*
4 The Semitic Languages
languages. Beyond this area they have spread only as a result of
secondary developments, i.e. migration, colonization, or conquest.
1.4. The grouping of the Semitic languages is usually based on
their geographical distribution: North-East Semitic (Mesopotamia),
North-West Semitic (Syria-Palestine), and South-West Semitic
(Arabia and Ethiopia). It is obvious that this division is closely
connected with that of the peoples who spoke those Semitic lan-
guages; for the purposes of linguistic inquiry, however, a division
which simply projects ethnical entities on to the linguistic plane
is bound to be imprecise. Schemes of classification which identify
and distinguish groupings within the Semitic area on the basis of
specific bundles of isoglosses may derive support from structural,
functional, and genetic criteria. The position of certain languages
in these classification schemes has been, and still is, the object of
controversy; this applies to the position of Ugaritic (cf. § 3.9) and
of Amorite (cf. § 3.8) as between the western and the eastern
languages; to Nabataean and Palmyrene within Aramaic (cf.
§ 3.16); to South Arabian as between Arabic and Ethiopic (cf. §4.1),
etc. Particular connexions have repeatedly been postulated be-
tween individual languages of different areas; but the problem
remains as to how such affinities can be tested on the statistical,
typological, or historical plane. Of some interest in this respect
are certain common elements as well as points of contrast which
have been observed between, for instance, Akkadian and Ethiopic:
such relationships may be explained in accordance with the prin-
ciples of linguistic geography. Similar interest, on the general
Hamito-Semitic level, would attach to the connexions which have
been claimed to exist between Akkadian and Libyco-Berber
(cf. § 5.5). On the whole, it may however be said that the geo-
graphical division indicated above corresponds tolerably well
(though not without certain exceptions) to the distribution of gross
linguistic features. East Semitic exhibits from the outset certain
independent characteristics as compared with West Semitic
(cf. § 3.2); and these features become further accentuated in the
course of time. In the western area, the distinction between the
northern and the southern languages is more evident in their fully
developed form than at the archaic stage (but it must be borne in
mind that our knowledge of the archaic phase is fairly limited).
Scope of the Survey 5
3. Nature and Extent of the Survey
1.5. The present survey has as its object the whole body of the
classical Semitic languages. Data drawn from modern dialects will
only be taken into consideration when they have a contribution
to make to an examination of the earlier phases of the classical
languages, either by reason of archaic elements which these modern
tongues may have preserved or by virtue of any other relevant
characteristics.
1.6. Within these limits the linguistic material of the present
survey extends in time over some four thousand years: from the
third millennium B.C., when we encounter the earliest manifestations
of a Semitic language (Akkadian), until the first millennium A.D.,
when some of the great literary languages (Syriac, Arabic, Ethiopic)
begin to flourish and to exert an influence well beyond those limits.
1.7. Geographically, the bulk of the material originates from the
region indicated above (§ 1.3), while a limited portion of it comes
from outside that area as a result of the spread of Semitic popu-
lations (Assyrians in Anatolia, Phoenicians on the coasts and
islands of the Mediterranean-even as far as the Atlantic, Arabs
in Africa and on the islands of the Indian Ocean, etc.).
1.S. Within the limits of time and space already described, the
present study aims at a reconstruction of the earliest phonological
and morphological units (Proto-Semitic, for which see § 5.1) as
well as their historical development in the principal languages of
the group. From this limitation the following consequences will
inevitably result: in the first place, the exclusion of modern dialects
and especially of secondary developments due to them; and,
secondly, the relegation of a specific treatment of syntax and vo-
cabulary, partly because of the elementary nature of this treatise
and partly on a c c ~ ) U n t of the somewhat backward state of studies
in those fields. Moreover, the present work does not aim at exhaust-
iveness and, while it places its main emphasis on the position in
the great literary languages (Akkadian, Hebrew, Syriac, Arabic,
Ethiopic), it adduces evidence from other Semitic languages and
modern dialects only when of particular relevance-as is indeed
appropriate to a survey of an essentially elementary character.
6
The Semitic Languages
1.9. For a detailed description of the Semitic languages (which
is outside the scope of this study) the reader is referred to the stand-
ard introductory works. The following observations are restricted
to what is necessary for the clarification of the concepts and
nomenclature adopted in this inquiry-especially in view of chang-
ing notions and terminological fluctuations current at the present
time.
B. North-East Semitic
2.1. North-East Semitic is represented by Akkadian, spoken in
Mesopotamia in the pre-Christian era. This language, which in
the civilization of that region prevailed over, and eventually re-
placed, the non-Semitic Sumerian, derives its name from that of
the city of Akkade, the capital of the empire of Sargon the Great
(2350-2294 B.C., according to the "short" chronology). The
principal phases of Akkadian are:
2.2. a) Old Akkadian may be dated between 2500 and 2000 B.C.
approximately; owing to the limited extent of our documentation
(very few texts from Assyria are extant) it is at present impossible
to establish clearly defined differentiations of dialect. About
2000 B.C. the following principal dialects can be distinguished:
2.3. b) Babylonian, the dialect of the southern part of the region,
is divided into Old Babylonian (about 2000-1500 B.C.) with
several dialectal variations, Middle Babylonian (about 1500 to
1000 B.C.), and New Babylonian (about 1000 B.C. till the beginning
of the Christian era). The most recent phase of New Babylonian
(from about 600 B.C.), characterized by the infiltration of Aramaic
words and linguistic peculiarities (cf. § 3.18), is more specifically
called Late Babylonian ("Spatbabylonisch"), while the literary
language used between about 1400 and 500 B.C. in Babylonia as
well as in Assyria (and differing considerably from the spoken
language) may be referred to as Later Babylonian
lonisch").
2.4. c) Assyrian, the dialect of the northern part of the region,
is divided into Old Assyrian (about 2000-1500 B.C.), with texts
principally of Cappadocian origin, Middle Assyrian (about 1500 to
1000 B.C.), and New Assyrian (about 1000-600 B.C.); the last-
named is strongly aramaicized in its final phase.
North-West Semitic 7
2.5. In Old Akkadian and Old Babylonian texts (ca. 2400 to
1700 B.C.) occur hundreds of Semitic names which cannot be
explained either on the basis of Akkadian or of "East Canaanite"
. (Amorite, cf. § 3.8). These names as well as other linguistic pheno-
mena led von Soden (WZKM 56 [1960], pp. 185-91) to the
conclusion that in the second half of the third millennium B.C.
there must have existed a further Semitic language for which he
proposed the name "Old-Amorite". A grammatical sketch of Old
Amorite (from which certain peculiarities of U garitic might possibly
be explained) has not yet been produced.
C. N orth- West Semitic
1. General Characteristics
3.1. North-West Semitic displays notable internal variations
which reflect the rather chequered history of Syria and Palestine.
It is the custom of grammars and works of introduction to the
Semitic languages to divide the languages of this area into two
main groups, i.e. the Canaanite and the Aramaic languages. But
recent studies tend to show that this division is not in accord with
the most ancient phase of Syro-Palestinian linguistic history
(second millennium B.C.), because in that phase some of the iso-
glosses which distinguish the two groups had not yet been clearly
drawn. This does not mean, of course, that in the second millen-
nium B.C. there existed a lesser degree of variation in the local
speech-forms of the North-West Semitic area, but merely that such
distinctions manifested themselves in forms different from those
current in the following millennium (Garbini, SNO). The division
into Canaanite and Aramaic can be assessed only from the time
when Aramaic made its historically and epigraphically attested
appearance, i.e. from the first millennium B.C.; and even then the
homogeneity of Aramaic is not matched by a comparable measure
of uniformity in the languages grouped under the general heading
of "Canaanite" (of. § 3.12).
3.2. In the earliest historically attested phase of North-West
Semitic many of the distinctive features that were later to contrast
it with South-West Semitic had not yet been clearly realized-at
least so far as our limited knowledge goes. Hence it may be sur-
8
The Semitic Languages
mised that in its first historical manifestations West Semitic
displayed a greater degree of unity than can be discerned in its
subsequent development when it was subjected to powerful dis-
integrating tendencies (of. § 1.4). East Semitic was at every stage
notably differentiated from its western neighbour.
2. The Languages of the Second Millennium B.C.
3.3. The most ancient form of North-West Semitic may be placed
in the second millennium B.C. It embraces, in the first place, a group
of texts of doubtful date and interpretation which, precisely for that
reason, will not normally be discussed in this treatise. These texts
include:
3.4. a) The pseudo-hieroglyphic inscriptions of Byblos, originally
assigned by Dunand to the end of the third millennium B.C. but
more recently dated by some scholars several centuries later and
regarded linguistically as the most ancient manifestation of
Phoenician.
3.5. b) The Proto-Sinaitic inscriptions, previously assigned to
about 1800 B.C. but more recently dated by Albright about 1500.
They are considered (but this is a matter of pure hypothesis which
reflects the uncertainty of all schemes of classification) by Albright
as North-West Semitic and by van den Branden as Proto-Arabic
(representing a linguistic stage prior to the differentiation between
North and South Semitic).
3.S. c) A series of short inscriptions originating for the most part
from Lachish and attributable to various dates in the second
millennium (between 1800 and 1300 B.C. approx.).
3.". Along with the group of texts just mentioned there exists
another one (also belonging to the second millennium B.C.) whose
interpretation and dating may be said to be generally established.
The languages of these texts are:
3.S. a) Amorite (also called "East Canaanite", an inappro-
priate term used to designate the North-West Semitic tongues of
the first half of the second millennium), which is reflected in the
proper names and in certain linguistic peculiarities of the Akkadian
North-West Semitic 9
texts of the period of the First Babylonian Dynasty, and in parti-
cular in the Mari texts (speech-forms not lacking in internal varie-
gation which our inadequate knowledge does not as yet allow us
to appreciate fully). The data drawn from the Akkadian texts are
corroborated by transcriptions· of North-West Semitic names of
persons and places in Egyptian execration-texts (cf. also § 2.5).
3.9. b) Ugaritic, the language of the texts discovered at Uaarit _ 0
(Ras Samra) and belonging to the fourteenth and thirteenth cen-
turies B.C. There has been much discussion as to the typological
placing of this language within the framework of the Semitic
hinguages (cf. § 1.4).
3.10. c) The language of the glosses (usually called "Canaanite")
in the Tell Amarna letters (l4th cent. B.C.) which are written in an
Akkadian showing many Canaanite peculiarities. Similar consider-
ations apply to the Akkadian texts from Ugarit (l4th-13th
cent.). The data drawn from this material are supplemented by
a further group of Egyptian transcriptions belonging to the second
half of the second millennium.
3.11. Towards the end of the second millennium B.C. we can
onset of other North-West Semitic languages which,
smce theIr full development belongs almost entirely to the first
millennium, will be dealt with in the following paragraphs. With
these languages, and within the limits set forth above (of. § 3.1),
the distinction between Canaanite and Aramaic may properly be
introduced.
3. Canaanite
3.12. Canaanite represents the non-Aramaic linguistic mani-
festations of the Syro-Palestinian area, from the end of the second
millennium B.C. onwards. The coherence or independence of
Canaanite (except in so far as it is clearly distinguished from
Aramaic) appears somewhat limited; so much so, in fact, that the
very individuality of the group has been questioned by some
scholars (Friedrich). This may, however, be largely attributable
to our deficient means of ready linguistic identification. The
Canaanite languages are:
10
The Semitic Languages
3.13. a) Hebrew-including: the Biblical period whose literature
may be dated approximately between 1200 and 200 B.C. and which
is supplemented by a number of short inscriptions; the post-
Biblical period, beginning with the apocryphal literature and the
documents recently discovered near the Dead Sea (second and first
centuries B.C.) and continuing with the rabbinical writings of the
first centuries of the Christian era (Misna, T6sefta, Midras); the
poetical, philosophical, a.nd exegetical literature of the Middle Ages
and of modern times; and finally Modern Hebrew, nowadays
spoken in Israel.
3.14. b) Phoenician and Punic, represented by the inscriptions
of the ancient Phoenician cities (to be dated between the tenth
and the first centuries B.C.) and by those of their Mediterranean
colonies (between the ninth century B.C. and the second century
A.D.).
3.15. c) Moabite, represented by the inscription of King M ~ s a ' of
Moab of the ninth century B.C.; this inscription, according to the
latest study (Segert), might however be regarded as a Hebrew
text, belonging to the central Palestinian dialect, having possibly
been drawn up by an Israelite in the service of the King of Moab.
4. Aramaic
3.16. Aramaic forms a considerable and wide-spread linguistic
group whose earliest manifestation goes back to the beginning of
the first millennium B.C. and which survives, in a few remnants,
to the present day. We distinguish between an ancient phase, up
to the first century B.C., and a subsequent division into two
branches, West Aramaic (which appears to be a more direct con-
tinuation of Old Aramaic) and East Aramaic. Some scholars are
inclined to ascribe the division into two branches to the second
or the third century A.D. and include, under the name of Old
Aramaic, Nabataean and Palmyrene; in the present survey these
languages will, however, be subsumed under "West Aramaic".
a. Old Aramaic
3.17. a) Old Aramaic is the language (with some dialectal
variants) of the most ancient inscriptions originating from Da-
North-West Semitic 11
mascus, Hama, Arpad, Sam'al, and Assyria, and belonging to the
period between the tenth and the eighth centuries B.C. Of these
inscriptions two from Sam'al (one of Panamuwa I and one of
Bar-Rakib) are of special importance, owing to their independent
characteristics, and represent the type of Aramaic known as
Ya'udic (derived from the name of the state of Sam'al Ya'udi).
3.18. b) Classical or Imperial Aramaic is the language used under
the Assyrian, Babylonian, and Persian empires (seventh to fourth
centuries B.C.) and continued by certain offshoots into the period
which followed. Eviclence of this comes from Mesopotamia (Ara-
maic inscriptions and proper names, words and constructions in
New Assyrian and New Babylonian texts), Persia, western India,
Anatolia, Arabia, and Egypt. The papyri and ostraca from Egypt,
of the fifth and fourth centuries B.C., are of particular importance
and constitute what has been termed Egyptian Aramaic.
3.19. c) A type of Classical or Imperial Aramaic is represented by
Biblical Aramaic, found in certain parts of the Old Testament
(Gen.31,47 [two words]; Jer.10,11; Ezra 4,8-6,18; 7,22-26;
Dan. 2,4-7,28); the age of these documents ranges probably
from the fifth to the second centuries B.C.
b. West Aramaic
3.20. a) Nabataean is the language of an (ethnically) Arab
population which established a state at Petra and flourished be-
tween the first century B.C. and the third century A.D.; Nabataean
papyri have been discovered among the Dead Sea documents, and
Nabataean inscriptions have been identified as far afield as Greece
and Italy.
3.21. b) Palmyrene is the language of an (ethnically) Arab
population which established a state at Palmyra and flourished
between the first century B.C. and the third century A.D. ;Palmyrene
inscriptions have been found as far afield as England.
3.22. c) Jewish Palestinian Aramaic is the language that was
spoken in Palestine at the time of Christ and during the first cent-
uries of the Christian era. In literary sources it is found in the
Genesis A pocryphon (discovered among the Dead Sea documents)
and the Palestinian Targiim (of which a complete manuscript has
12 The Semitic Languages
been identified in the Vatican Library by Diez Macho). Jewish
Palestinian Aramaic survives above all in a sizable body of Jewish
post-Biblical texts of the second to the fifth centuries A.D.; these
may be divided into two groups, one being represented by the
Targiimim of Onkelos and of Jonathan, and the other by the Gali-
lean variety (some Midrasim and the Jerusalem Talmiid).
3.23. d) Samaritan Aramaic is the language of the Samaritan
Targiim to the Pentateuch (probably of the fourth century A.D.)
and of some later writings.
3.24. e) Christian Palestinian Aramaic is the language used by
the Melkites between the fifth and the eighth centuries A.D.; it is
written in Syriac characters and is attested in several Old Testament
passages, Gospel Lectionaries, and liturgical writings.
3.25. Limited and gradually disappearing survivals of West
Aramaic can still be heard in the villages of Ma'liila, Gubb'adin
and BalJ'a, in the neighbourhood of Damascus.
c. East Aramaic
3.26. a) Syriac, originally the language of Edessa, later developed
a rich Christian literature extending from the third to the thirteenth
century A.D., although it was generally replaced, as a spoken
language, by Arabic during the great Islamic conquests of the
8th cent.
3.27. b) Babylonian Aramaic is the language of the Babylonian
Jews, prominently represented in the Babylonian Talmiid (fourth
to sixth centuries A.D.) and in a series of magical texts composed
in the fifth and sixth centuries A.D.
3.28. c) Mandaean is the language of the Gnostic sect of the
Mandaeans who flourished in Mesopotamia; their writings extend
from the third to the eighth century A.D.
3.29. Survivals of East Aramaic can still be found in the neigh-
bourhood of Lake Urmia, at ']:'iir 'Abdin, and near Mosul. It should
be mentioned that the "Assyrians" (as these Aramaic-speaking
populations are referred to) were displaced, as a result of the first
world war, and now live in scattered communities in the United
South-West Semitic 13
States and in Russia. The Aramaic of Georgia, in particular, has
been the subject of several recent studies (Tsereteli).
D. South- West Semitic
1. General Characteristics
4.1. South-West Semitic is, in grammars and introductions to
Semitic languages, usually divided into two groups: (1) North
Arabic and (2) South Arabian together with Ethiopic. It has,
however, recently been pointed out by some scholars that (at least
within the limits of the ancient period) the separation of South
Arabian from North Arabic, if intended to contrast the former
with the latter and to place it alongside Ethiopic, is not entirely
justified. It is true that historically Ethiopic makes its appearance
as a successor tongue of South Arabian, but such a genetic relation-
ship differs to some extent from the requirements of a descriptive
classification which is based on the convergence of isoglosses.
From the descriptive point of view it has been noted that ancient
South Arabian is in several respects in agreement with North
Arabic and at variance with Ethiopic (and vice versa). Ethiopic,
laid as it was upon a non-Semitic substratum, has undergone
certain developments not to be found in South Arabian. The
present exposition, based on geographical principles, will present
South Arabian within the area of the Arabian peninsula-without
thereby implying an undervaluation of its independent character-
istics or of its agreements with Ethiopic.
4.2. As regards the relationship between South-West and North-
West Semitic, cf. § 3.2.
2. Arabic (incl. South Arabian)
4.3. For the reasons explained in § 4.1, we have chosen to take
the term "Arabic" as a linguistic complex embracing all the tongues
of the Arabian peninsula-with the exception of some Aramaic
infiltrations (Nabataean and Palmyrene) in the extreme north.
This complex, containing many dialectal divergences, may be
subdivided as follows:
14 The Semitic Languages
4.4. a) Ancient or Epigraphic South Arabian (ESA-for whose
independent position cf. the remarks in § 4.1) is the language of
the inscriptions of the ancient South-West Arabian city-states.
Their dates range from the eighth century B.C. (this is subject to
considerable current controversy; cf. the studies of J. Pirenne)
to the sixth century A.D.; the following dialects, corresponding
to the regions of the principal states, are being distinguished:
(1) Sabaean, (2) Minaean, (3) Qatabanian, (4) I.Iagrami, (5) Awsan-
ian.
4.5. b) Pre-classical North Arabic is the language embodied in
a series of inscriptions which may be dated between the fifth
century B.C. and the fourth century A.D. (approximately) and
be divided into the following regional and dialectal groups:
(1) Tamudic (a conventional term which lacks precision and covers,
in fact, many of the tongues over a wide area of pre-Islamic central
and northern Arabia), (2) Lil).yanite, (3) ~ a £ a i t i c .
4.6. c) Classical North Arabic, the "Arabic" par excellence, is
attested from the fourth century A.D. in a few inscriptions and in
some dialectal samples preserved by Islamic writers. It attains
its full realization in pre-Islamic Arabic poetry and later in the
Qur' an (seventh century A.D.); it owes its diffusion and survival
to Islam which turned Arabic into a great literary language as
a result of the Arab conquests and the enormous expansion of this
dynamic religion. The generally attested form of classical Arabic is
the result of a process of systematization by Arab grammarians;
this linguistic material is represented by the pre-Islamic standard
speech ("Hochsprache") and was nurtured by the ample flow of
Arabian dialects. This process has concealed original dialectal
divergences as well as other elements of subsequent evolution
(Fuck).
4.7. The modern Arabic dialects are numerous and will only be
dealt with peripherally in this treatise (0£. § 1.5). In the South
Arabian area there exists a separate group of languages which,
according to some scholars, represent the continuation and develop-
ment of the ancient speech-forms: the principal ones among these
are Me1).ri, Sl].awri, and Soqotri. The large number of dialects
developed from Classical Arabic are most appropriately classified
according to regional groupings: Central-Asian, Iraqi, Arabian,
Proto-Semitic, Hamito-Semitic, Indo-European 15
Syro-Lebanese and Palestinian, Egyptian, North African or Maghrebi.
A separately developed form, owing to its long historical severance
and its exposure to non-Semitic influences, is Maltese.
3. Ethiopic
4.8. Ancient Ethiopic (or Gg'gz) is first attested in epigraphic
material of the first few centuries A.D. and, above all, in the great
Aksum inscriptions of the fourth century. It later developed an
extensive, predominantly religious, literature reaching up to modern
times.
4.9. The modern Semitic languages of Ethiopia are represented
by Tigrifia, Tigre, Amharic, Harari, and Gurage; Gafat and Argobba
are now virtually extinct.
E. Proto-Semitic, Hamito-Semitic, Indo-European
1. Proto-Semitic
5.1. By Proto-Semitic (or Common Semitic or simply Semitic)
we refer to the ensemble of elements which an examination of the
historically documented Semitic languages leads us to regard as
common property of the Semitic group in its most ancient phase
(Semitic isoglosses); hence we discover here the starting-points for
developments peculiar to each individual language. Whether such
postulated reconstructions invariably possessed historical reality
it is difficult to determine, but this uncertainty is not necessarily
an obstacle to comparative inquiry. It is reasonable to suppose
that the dialectal fluctuations with which we are confronted by
the existing historical evidence formed part also of the pre-historic
phase (in contrast to the concept of the genealogical tree). It must
not be forgotten that 'Proto-Semitic' is merely a linguistic conven-
tion or postulate, but such a convention is a necessary pre-requisite
for an understanding and reconstruction of linguistic history.
5.2. The concept of Proto-Semitic would seem comparable to
that of Proto-Indo-European. The problems of the former do,
however, appear more manageable owing to the lesser degree of
geographical dispersion of the Semitic languages and the greater
16
The Semitic Languages
measure of affinity between them. It would, therefore, be more
appropriate to compare Hamito-Semitic with Indo-European, on
the one hand, and Semitic with the Romance. Slavonic. or Ger-
manic languages, on the other.
5.3. The value and importance of individual Semitic languages
to a reconstruction of Proto-Semitic has been variously estimated
at different periods in the history of our studies. Account has to
be taken of archaizing tendencies in some languages in contrast to
genuinely old material which may at times appear in strangely
disguised forms. The central position long occupied by Arabic
as either the proto-type or true image of primitive Semitic has come
to be challenged in recent times. The rich phonological structure
of Arabic is now paralleled by that of Ugaritic and South Arabian,
and its highly developed verbal system is regarded as the result
of systematization rather than archaism. A more profound knowl-
edge of Akkadian, of some of the North-West Semitic languages,
of the modern Ethiopian and South Arabian languages, etc., has
to some extent modified our ideas about Proto-Semitic and those
of the classical tongues which were alleged to resemble it most
closely.
2. Hamito-Semitic
5.4. It has long been held that Semitic is not an isolated group
but forms part of a larger complex of languages, conventionally
called Hamito-Semitic. In addition to Semitic, this larger grouping
comprises Egyptian, Libyco-Berber, and Cushitic; thus Hamito-
Semitic is also sometimes referred to in purely geographical terms
as Afro-Asiatic. There is no "Hamitic" unit comparable to the
Semitic one: Semitic possesses a much greater measure of struct-
. ural uniformity than can be detected among the "Hamitic"
languages. The relationship between the various units of the
Hamito-Semitic group cannot be explained as a secondary develop-
ment, and this makes the concept of an original Hamito-Semitic
linguistic body one of great cogency. We have to aim at the
reconstruction of Proto-Hamito-Semitic forms; though naturally
with all the reservations called for by such a conjectural recon-
struction. Semitic is, of course, the group that is more fully attested
and generally also the most replete with ancient forms.
Language and Script 17
5.5. Certain studies (Rossler) have asserted that Libyco-Berber
is possessed of an essentially Semitic character and have claimed
a particular affinity with Akkadian; this is based on correspondences
of a phonological, morphological, and lexical nature. If this theory
were shown to be correct, the independence of Semitic would, to
some extent, be impaired. However, the similarities adduced in
support of the thesis seem to be in part open to question and in
part inconclusive, for most of the parallels can much more readily
be explained within the framework of the long-established general
Hamito-Semitic affinity (Cohen). Consequently, there is at present
no cogent reason to question the independence of Semitic within
the larger Hamito-Semitic complex.
3. Hamito-Semitic and Indo-European
5.S. A few points of contact have long been noticed between
Hamito-Semitic and Indo-European languages. These are generally
concerned with relations of a phonological and especially lexical
character and have given rise to the so-called "Aryo-Semitic"
(Ascoli) or "Nostratic" (Pedersen, Cuny) hypothesis which is
claimed as common ancestor of Hamito-Semitic and Indo-Euro-
pean. Such conjectures are, however, very highly speculative,
especially on account of deep-seated morphological divergences
between those groups, although the inflexional structure appears
to be common to both. A more reliable explanation is to be sought
in the common Mediterranean environment (especially as regards
lexical elements) and consequent historical contacts and influences
(particularly marked in Anatolia and the Eastern Mediterranean).
Such limited links as may exist between Indo-European and Hamito-
Semitic should not, therefore, be regarded as a heritage from
a 'parent' language, but rather as a haphazard collection of isoglosses
not unconnected with the geographical proximity of the two
groups and certain historical contacts between them.
F. Language and Script
S.1. A treatment of Semitic writing lies outside the scope of the
present work. However, since systems of writing may condition
and at times even influence linguistic elements, it is well to recall
certain essential facts.
Moscati, Comparative Grammar 2
18
The Semitic Languages
6.2. North-East Semitic (Akka.dian) is written in cuneiform
characters, inscribed with a pointed instrument on tablets of
clay or more rarely on stone or metal; this form of writing was taken
over from the non-Semitic Sumerians who preceded the Semites in
Mesopotamia. The cuneiform system possesses many hundreds
of signs which have ideographic or syllabic value and are often
multivalent, so that their reading offers considerable difficulties
(quite apart from the fact that the tablets are not always in a good
state of preservation and that there existed notable divergences in
orthography in different areas and times). The script indicates both
consonants and vowels which (in contrast to the majority of Semitic
alphabets denoting consonants only) is of considerable assistance
to our knowledge of the language. On the other hand, the conso-
nantal inventory of Sumerian (for which this form of writing was
originally devised) differed materially from the Semitic sound
system, so that the graphic representation of Akkadian consonants
by means of Sumerian writing exhibits many imperfections and
difficulties. In the transliteration of Akkadian (for the purposes
of this comparative study) syllables which are separately represented
in cuneiform will, as a rule, be joined together in the same word
(e.g., inaddin "he gives", instead of i-na-ad-di-in). It is important
to realize that length of vowels and doubling of consonants are
not consistently expressed in cuneiform. It is, therefore, difficult
to reach satisfactory conclusions; at times data drawn from com-
parative Semitic grammar will prove useful for the reconstruction
of relevant features.
6.3. West Semitic, both northern and southern, is represented
in consonantal alphabetic scripts with a limited number of signs
(generally less than thirty). The origins of this alphabetic form of
writing are to be sought in the Syro-Palest.inian area in the first
half of the second millennium B.C. After the first attempts (their
interpretation is still in doubt: cf. §§ 3.3-6) we witness, in the
second half of the same millennium, the appearance of the Ugaritic
alphabet, the only one in the Semitic west to use characters of
cuneiform type-though alphabetic in structure. From about the
same period dates the formation of the so-called Phoenician alpha-
bet which was carved on stone and was not of the cuneiform type.
This alphabet has a long history connected with the emergence of
Language and Script 19
the ancient Hebrew as well as the Moabite and Samaritan alpha-
bets. The Aramaic script, too, is derived from the Phoenician, but
its independent evolution gave rise, in its turn, not only to the
various alphabets of the Aramaic languages but also to the Hebrew
"square" script and the classical Arabic alphabet.
6.4. Another alphabet whose origin has probably to be sought in
the Syro-Palestinian area is that which makes its appearance in
the ancient South Arabian inscriptions. Closely connected with
the latter are, on the one hand, the pre-classical Arabic scripts
(1'amiidic, Lil).yanite, and i;jafaitic) and, on the other, the Ethiopic
syllabary (see § 6.9).
6.5. The West Semitic alphabets are, as has been said, purely
consonantal in character. From this fact arise some of the principal
difficulties in the study of the Semitic languages as well as many of
the obscure points in our understanding of their comparative
grammar. However, the principle of not denoting vowels was in
practice subjected to certain subsequent modifications (cf. §§ 8.69
to (6). Among these are the following:
6.6. a) In Ugaritic the consonant' has three forms, according to
its vocalization with a; i, u.
6.7. b) In some alphabets the use of matre8 lectioni8 is developed,
i.e. of the consonant-signs w, y, ',h. This device, limited at first
to final vowels, is later extended to internal long vowels and some-
times, though rarely, even to short ones (East Aramaic).
6.S. c) In some writing-systems additional signs were later intro-
duced to indicate voweis: this applies to Syriac, Hebrew, and Arabic,
where these signs are, however, used only for certain texts of parti-
cular importance (the Bible, the Qur'an, etc.).
6.9. d) In Ethiopic the script has been adapted to denote seven
vowels by a variety of changes in the structure of the consonantal
symbol. Vowels have thus become an integral part of Ethiopic
writing which now assumes a quasi-syllabic character-yet without
sacrificing the general Semitic concept of the predominance of
consonants over vowels.
6.10. Vowel notation by means of the methods just described
still leaves certain deficiencies and problems (cf. § § 8.73-(6).
2*
20 The Semitic Languages
We shall here limit ourselves to mentioning the artificial character
of seemingly very precise vocalizations, such as those of Hebrew
and Biblical Aramaic; or the ambiguity in Hebrew of the sign which
indicates either a vowel of the 'd type or the absence of a vowel
(and the similar ambiguity of the sixth order in Ethiopic); and,
finally, the situation in Syriac which possesses no symbol either
for 'd or zero vowel.
6.11. Another notable deficiency in most of the Semitic writing-
systems concerns the marking of gemination or consonant-doubling
-even though this may be a feature of phonemic significance.
Such doubling (inconsistently expressed already in cuneiform-
§ 6.2) lacks specific symbols in the West Semitic alphabets, though
Hebrew, Biblical Aramaic, and Arabic have developed a gemination
mark along with their general vowel system.
6.12. The transliteration of the Semitic languages which is
employed in this book is based on certain principles which it seems
well to explain beforehand. It is obvious that these principles are
open to a great deal of argument and are not exempt from certain
disadvantages, but it appears to be beyond doubt that any other
set of principles would be subject to an equal measure of ambiguity:
6.13. a) Our mode of graphic representation is, in fact, a trans-
literation rather than a transcription, for it aims at reproducing,
as far as possible, each symbol by one sign, in order to permit the
reconstruction of the original orthography. It need hardly be
mentioned that proper transcription has not been abandoned with-
out regret, but in the case of many of the ancient Semitic languages
the conjectural element involved in such a course seemed unjustifi-
ably prominent.,
6.14. b) The system of transliteration has been kept as simple
as possible-in accordance with the requirements of an elementary
grammar; it eschews the notation of sub-phonemic variants
(allophones)-except where this is called f o ~ by special circum-
stances. Non-distinctive variants can generally be determined in
accordance with grammatical rules (for example, the fricative
articulation of consonants in postvocalic position [of. § 8.10] or
the not always consistently 8mployed matres lectionis [cf. §§ 8.81,
8.87]).
Language and Script 21
6.15. c) The choice of transliteration symbols takes into consider-
ation the usual conventions (which it is well not to alter in an ele-
mentary grammar, unless there are compelling reasons) as well as
phonetic and (especially) etymological data which are of consider-
able importance in a comparative study: thus, for example, in
Ethiopic the transliterations (j and 8 will be used, although in
their pronunciation these consonants became early identified with
~ and s, respectively (of. §§ 8.20, 8.37); for Ethiopic vowels the
quantitative indications derived from etymological comparison
will be retained-in preference to the purely qualitative distinctions
on which the Ethiopic vowel system appears to be based (cf. §§ 8.95
to 96).
H. Phonology
A. Preliminarie8
7.1. A phonological study of the Semitic languages must be based
on the clear distinction, now generally accepted in linguistic work,
between phonetics and phonemics: on the one hand, the articulatory,
acoustic, and auditory characteristics of actual speech (phonetics)
and, on the other, phonemes, i.e. minimal distinctive sound units
relevant to meaning (phonemics). For ancient languages recon-
struction is of necessity phonemic, i.e. the analysis of data obtained
by means of a study of distinctive oppositions (cf. especially Can-
tineau's studies).
7.2. The pronunciation-or, more exactly, the complex of phonetic
expressions-of the ancient Semitic languages is reconstructed and
assessed on the basis of indications of various kinds: a) traditional
pronunciation-in the case of languages which have been trans-
mitted (living or otherwise) up to the present time (Hebrew,
Aramaic, Arabic, Ethiopic); b) testimony of grammarians-for
languages in which a grammatical tradition exists (Hebrew,
Syriac, Arabic); c) transcriptions of Semitic words and phrases in
other languages (Greek, Egyptian, Aramaic and Hebrew for
Akkadian; Greek and Latin for Hebrew and Phoenician-Punic,
etc.) and vice ver8a; d) orthographic peculiarities indicative of
phonetic characteristics not otherwise expressed; e) comparative
Semitic linguistics (reconstruction of Akkadian, Ugaritic, or
ancient South Arabian pronunciations based on parallels in Arabic,
Hebrew, etc.).
7.3. This body of evidence is neither complete nor sufficient.
There remain, consequently, doubts and uncertainties, and our
reconstructions are by and large schematic and conventional.
Sometimes, moreover, tradition proves to be an obstacle rather
than a help-as in the case of classical Ethiopic whose pronun-
ciation has been preserved by speakers of languages derived from
The Phonological System
23
it, with the result that that pronunciation embodies developments
peculiar to these modern languages. Similar considerations apply
to Hebrew whose transmission is affected not only by differences
between the Sephardi and Ashkenazi pronunciations but also by
influences due to the substratum-languages of those who have
gone back to the use of Hebrew as a spoken tongue.
B. The Phonological SY8tem
1. Classification
8.1. The Semitic phonological system is made up of consonants,
semivowels, and vowels as well as certain stress patterns. Their
classification may be based either on the musical principles of
acoustic phonetics or on the physiological elements of articulatory
phonetics. In the latter case, classification is related to the place
and the manner of articulation. According to the place of articula-
tion we have: bilabials, interdentals, dentals, palato-alveolars,
velars, pharyngals, and laryngals; according to the manner of arti-
culation: plosives (or stops), fricatives, laterals (and lateralized con-
sonants n, rolled consonants, and nasals.
8.2. Within the groups so classified a further distinction may
be made according to the voiced or "emphatic" character of some
consonants. As regards sonority, doubts have been raised as to its
nature in Proto-Semitic and it has been termed a "correlation of
tenseness or pressure" (Cantineau). It is, however, difficult to
arrive at concrete results beyond the indications furnished by the
historically attested languages. As regards "emphasis", this term
is used, not altogether properly, to denote a quality characteristic
of the Semitic (and Hamito-Semitic) languages: velarization in
Arabic, glottalization in Ethiopic. It is uncertain which of these
types is primary. According to some scholars, the fact that glot-
talization is not found in Semitic outside Ethiopic, yet occurs in
certain Cushitic languages, would point to its being a secondary
feature. In favour of this thesis one might refer to the pheno-
menon of labialization which is probably also due to Cushitic
influence (cf. § 8.43). According to others, there are reasons for
maintaining, on the contrary, that the Ethiopic glottalized ejectives
are primary, because: a) the Ethiopic "emphatics" are voiceless
24 Phonology
and, apart from Arabic, so the Semitic "emphatics" -almost
without exception; b) the Ethiopic "emphatics" do not appear to
influence the timbre of neighbouring vowels and, again apart from
Arabic, this seems to be the norm in the Semitic languages (cf. how-
ever for certain facts in Akkadian von Soden, GAG, p. 12); c) the
phenomenon q >' in some Arabic dialects can only be explained
by way of glottalization.
2. The Proto-Semitic Consonantal System
8.3. The Proto-Semitic consonantal system may be hypothetically
reconstructed as follows (with such reservations as will be expressed
below) :
Plosive Fricative Lateral Lateralized? Rolled Nasal
Bilabial p, b m
Interdental t, (1" t ¢
Dental t, d,
t
s, z, ? 8 r n
Palato-alveolar S
Velar k, q, g lP, g
Pharyngal "If,
,
Laryngal h
8.4. Following the distinctions of voice and of "emphasis", we
may note, in certain positions of the classification-chart, the
juxtaposition of "triadic" groups, i.e. voiceless-voiced-emphatic.
Such "triads" exist: a) for the interdentals; b) for the dental
plosives and the dental fricatives; c) for the velar plosives.
3. Bilabials
8.5. Proto-Semitic has two bilabial plosives, voiceless p and
voiced b; it has further a bilabial nasal m. In the South Semitic
languages p is replaced by the voiceless labiodental fricative t:
e.g. Akk.Heb.Syr. pqd "to seek", Ar.Eth. tqd.
8.6. Ethiopic possesses a p in addition to the t (probably derived
from Proto-Semitic p); and in Egyptian and Cushitic likewise
both p and t occur. The problem therefore arises whether these two
consonants did not co-exist already in Proto-Semitic. As regards
Ethiopic p, it is rare and usually confined to transcriptions of
The Phonological System 25
Greek words (e.g. pisii = n{aaa). Its shape, too, is borrowed from
Greek p, and its late appearance in the Ethiopic alphabet is corro-
borated by its place at the tail of the Ethiopic syllabary. As regards
Egyptian and Cushitic, the correspondences tend to show that the
two phonemes of these languages had as their Semitic counterpart
the single phoneme p (southern f): hence the differentiation in these
languages seems to be of a secondary character; and in any case,
for Proto-Semitic one single phoneme may be posited. This is
likely to have been p rather than t, for it is easier to suppose an
evolution p > t than one in the opposite direction (there are, more-
over, signs pointing to a similar development-not conditioned by
the position of the consonant-in the northern Semitic area, e.g.
in the language manifested in the Egyptian transcriptions of the
second millennium).
8.7. Ethiopic possesses also a voiceless bilabial plosive p which is
emphatic (or ejective) and, like p, of rare occurrence. This conso-
nant may well be of Cushitic origin, and it is worthy of note that,
for the purposes of its graphic expression, the form of another
ejective sound (though quite different in its basis of articulation)
was slightly modified: p was then placed, in the order of the alpha-
bet, next to the character (?) whose shape it had imitated. The new
consonant seems soon to have overstepped its original function,
for, like p, it is usually employed to transcribe Greek words (e.g.
pariiqlitos = nae6.xArJTOC;). However, since Semitic counterparts
are lacking, it should probably be regarded as secondary. Likewise
secondary, and similarly used for transcriptions from Greek, is
the special p which is to be found in Syriac (James of Edessa) and
in Christian Palestinian Aramaic. A secondary development in
the sphere of emphatic labials may be encountered in some modern
Arabic dialects which have emphatic 9, t, and rp" influenced by the
proximity of other emphatic consonants or of back vowels (posi-
tional variants).
8.8. Interchanges between consonants of the bilabial group, as
well as between them and the bilabial semivowel w, take place in
several languages, but we are not always in a position to ascribe
these interchanges to clearly identifiable linguistic reasons. Thus
we find in Ugaritic p:b:m: e.g. npk "well" from Sem. nbk, sps
"sun" from Sem. sms. In Ya'udic the change p > b is attested:
26 Phonology
e.g. nbS "soul" from Sem. nps. In several Aramaic dialects we
find the change b > w, evidently by way of the fricative articulation
of b: e.g. Syr. *rabrabr'ine "great ones" > rawrabr'ine (dissimilation
may conceivably have been an additional factor in this process);
Mand. 'wd "to perish" from Sem. 'bd. The transition b >9 >w
(of. Ullendorff, SLE, p. 106) is well represented in modern Aramaic
(zabna > *zawna > zona "time"; gabra > *gawra > gora "hus-
band"), modern South Arabian (*lbn > lun "white"), modern
Ethiopic (*sb' > Amharic saw "man"; nbr > Amh. nora "he
stayed"). The change to w affects also other labials, though to
a lesser extent: cf. Syr. qwz "to leap" compared with Aram. Ar.
qpz and Heb. The transition m > b would account for the ESA
bn "from" corresponding to Sem. mn. Ethiopic, too, has alter-
nations b:m:w, e.g. and "to be weak", etc. Akkadian
is in a peculiar situation: the use of m and b for w is frequent, but
owing to the absence of w in Sumerian this is often due to purely
graphic reasons rather than phonetic causes. A change of inter-
vocalic m to w is, however, attested soon after the earliest period
of Akkadian (of. von Soden, GAG, pp. 21-22, 31-32).
8.9. As regards the development of the bilabials in the various
languages, the situation in Akkadian is of particular interest, for
the cuneiform system of writing does not adequately distinguish
between p and b (or between voiced and voiceless consonants
generally): never in final position nor in other positions as far as
Old Akkadian and Old Assyrian are concerned; in Babylonian
and in the later phases of Assyrian special symbols are often em-
ployed for syllables with p and with b. New Assyrian appears to
have lost any consistent distinction between p and b in pronun-
ciation; this, in turn, has led to considerable graphic fluctuations.
Some consonantal interchanges (e.g. awilum and abilum "man"
in Old Akkadian) suggest the possibility of spirantization, i.e.
fricative articulation (of. § 8.10).
8.10. In North-West Semitic (or more precisely in Biblical
Hebrew and in the Aramaic of the Christian era) spirantization
of p > f, b > v occurs as a regular positional variant (the traditional
pronunciation represents the resultant consonants as labiodental
fricatives, like [f; v] in I.P.A. symbols, but this does not exclude
their having been originally bilabial fricatives, [ep, f3] in I.P.A.
The Phonological System 27
symbols). This change affects the non-emphatic plosives (p b t d
k g) which in postvocalic position come to be articulated as frica-
tives, i.e. I.P.A. [f v ff 5 x y]. This is, of course, a conditioned
phonetic phenomenon (partial assimilation of consonant to vowel:
of. § 9.5) and of non-phonemic character (a sub-phonemic positional
variant). As regards the period when spirantization became opera-
tive, there is no certain proof that it pre-dates the Christian era:
neither the Egyptian transcriptions of North-West Semitic names
nor Greek and Latin transcriptions of elements in the pre-Maso-
retic text furnish sufficient indications of the existence of this
distinction (cf. Garbini, SNO, passim). At any rate, it would
appear that such signs of non-plosive articulation as we encounter
(cf. § 8.6) are not necessarily connected with post-vocalic position.
4. Interdentals
8.11. Proto-Semitic has two non-emphatic interdental conso-
nants, voiceless t and voiced rj, i.e. I.P.A. [ff] and [5], respectively.
These consonants are, in fact, attested in certain languages only,
but their proto-Semitic status appears to be vouchsafed by the
ensemble of correspondences which can be satisfactorily explained
on this assumption only.
8.12. Proto-Semitic had in addition an emphatic interdental,
probably voiceless (t). This consonant is represented in Arabic
(where it is usually, but rather inappropriately, transliterated ?),
in South Arabian, and in Ugaritic by a graphic symbol of its own.
The phonetic and phonological correspondences in the other
languages can be explained only by accepting the consonant as
Proto-Semitic.
8.13. Finally, Proto-Semitic appears to have possessed a con-
sonant which Brockelmann and those who follow his system
transcribe rj, i.e. regarding it as a voiced emphatic interdental;
others (Cohen, Cantineau etc.) transcribe it ij, i.e. as a lateralized
(or lateraI1-Martinet) voiced emphatic interdental. Apart from
the question of its precise articulation (which it is difficult to deter-
mine with certainty), this consonant retains its independent status
in the South Semitic languages only: phonological correspondences
would nevertheless suggest its existence in Proto-Semitic. It is
probable that it was voiced, for this is the position in the languages
28 Phonology
which have retained this sound. Its lateralization appears to be
borne out by indications furnished by Arab grammarians and by
the evidence of certain modern South Arabian languages. There
are further indications in the more ancient languages: d. Akk.
Rulda'u or Rulta'u for the Arabic name of the god Ruif.,a'.
8.14. The following table (subject, of course, to certain additional
explanations and reservations to be set forth below) displays the
development of the Proto-Semitic interdentals in the principal
Semitic languages:
Proto- Akkadian Ugaritic Hebrew S yriac Arabic E SA Ethiopia
Semitic
t
8
t
8 t
t
[ 8
iJ
z d or iJ z d
iJ iJ
z
I I t ? (=/) ?(I)
r/
rf, rf, (r/) rf,
(=r/ 1)
Examples: [: Akk. 8uru "bull", Ug. [r, Heb. 80r, Syr. tawn]"
Ar. tawr, ESA twr, Eth. 80r; - iJ: Akk. 'lJz "to take", Ug. 'lJd (iJ),
Heb. 'Jy,z, Syr. 'ly,d, Ar. 'lJiJ; ESA 'lJij, Eth. 'lJz; -/: Akk.
"shadow", Ug. Il, Reb. Syr. tellaZa, Ar. ?ill, (cf. ESA ?ll "to
cover"), Eth. - r/: Ug. Heb. "to be hostile", Syr. 'rr,
Ar. ESA Eth. rf,rr; Akk. er§etu "earth", Ug. dr§, Heb. 'ere§, Syr.
'ar'a, Ar. 'arrf" ESA 'rrf,.
8.10. In Old Akkadian t appears to remain independent, for it
is written by means of the series of symbols for Sumerian 8, while
Proto-Semitic 8 is written with the series for 8. Later on the change
t > 8 takes place. Certain orthographical divergences have been
interpreted as traces of an independent iJ in Old Akkadian (Gelb,
OA, p. 38). The problem remains, however, unresolved.
8.16. In North-West Semitic, the most ancient phase probably
reflects the situation in Proto-Semitic. In Ugaritic, [ and I retain
their independence (cf. Garbini, SNO, pp. 193-94); for a different
opinion d. Rossler, ZA 54 [1961], pp. 158----,72); as regards I,
the fact that it is at times represented by the graphic symbol "g"
suggests the possibility of a development analogous to that later
The Phonological System 29
found in Aramaic (cf. § 8.18). iJ generally merges with d, but in
some cases it is represented by another sign which we may hypo-
thetically (cf. Gordon, UM, pp. 22-23) equate with iJ (d., for
example, Ug. iJr' "arm", Ar. iJira', Heb. z(}roa', etc.); it may, there-
fore, be conjectured that either the evolutionary process was as
yet fluid or that we have to envisage different orthographical
phases, perhaps in documents of different dates. Proto-Semitic d
has developed into §, but some spellings with I have encouraged
the assumption that this process passed through a prolonged
indecisive stage. In North-West Semitic of the second millennium
B.C. (as reflected in Egyptian transcriptions) we may note the
survival of Proto-Semitic t; the existence of iJ (as an independent
phoneme) is very doubtful, and I, r/ are seen to have merged
8.17. These developments-as indeed the process t > 8 and
iJ >z-continue in Canaanite; for this we possess the evidence of
Hebrew as well as that of Phoenician-Punic and Moabite. Some
trace of an original differentiation in the Phoenician area has been
claimed on the basis of Greek transcriptions and l:lCJWY)
of two place-names which in Phoenician both begin with but
this phenomenon may conceivably have a different explanation
altogether (d. Garbini, SNO, pp. 32-33).
8.18. In the most ancient Aramaic inscriptions we find the
symbols "8", ":;;" for t, iJ, I, respectively, i.e. the same symbols
as in Canaanite. Later, about the middle of the first millennium
B:C., there takes place the change t >t, iJ>d, I> t (given in the
table above for Syriac) which applies to other Aramaic languages.
This state of affairs is to be explained, according to some scholars,
as "Canaanisms" of a purely graphic type-at a time when the
characteristic evolution of the Aramaic consonants had already
taken place. According to others, however, the symbols employed
in the most ancient phase are to be regarded as an attempt to
represent, at least approximately, the Proto-Semitic interdentals-
following upon the adoption of an alphabet (the Phoenician) which
had no proper signs for them. The development might thus be
exemplified as follows: Proto-Semitic *ytb "to sit" > Old Aram.
ytb (written "Y8b") > common Aram. ytb; Proto-Semitic *iJhb
"gold" > Old Aram. iJhb (written "zhb") > common Aram. dhb;
Proto-Semitic *nlr "to guard" > Old Amm. nlr (written "n/?r") >
30 Phonology
common Aram. ntr. Proto-Semitic ¢ undergoes a peculiar develop-
ment in Aramaic: the most ancient inscriptions employ the symbol
"q", while later" , " takes its place (e.g. "'rq'" "earth", later
" 'r" "; cf. Ar. 'ar¢). Mandaean has preserved "q" in some cases.
The phonetic process reflected in these changes is far from being
clear: Noldeke (Mandaische Grammatik, p. 73) considers the pos-
sibility that "q" might here be used to represent the articulation
of g, i.e. fricative q (as in fact happens extensively in modern
Ethiopian languages: d. Ullendorff, SLE, pp. 64-65). g might
have been an intermediate stage in the transition ¢ > '.
8.19. In the Arabic sphere, classical Arabic maintains the four
interdental consonants as independent phonemes and also furnishes
some traditional indications as to their pronunciation. In this
tradition t (conventionally transcribed ?) appears to have become
a voiced consonant; in some Arabic dialects it is pronounced as
a voiced emphatic interdental in others as a voiced emphatic
dental plosive [<;1]. As for the fourth consonant of the series,
transliterated ¢, indigenous grammarians have described its original
character as a voiced emphatic lateralized interdental. In modern
dialects its pronunciation is in general the same as that of the
previous consonant, i.e. or [<;1]. South Arabian and pre-Islamic
North Arabic agree with classical Arabic as far as the retention of
the four interdentals is concerned. It would obviously be impossible
to say anything definite about their pronunciation, yet South
Arabian offers some interesting indications: the spelling of
for Sl?y might connote, at a late period, a change t > of the type
also found in other Semitic languages (including Ethiopic); "nz'"
for n¢' suggests a fricative articulation of Proto-Semitic ¢' since
the change ¢ > z is phonetically not improbable, while ¢ > z would
certainly be far less likely.
8.20. In Ethiopic the pronunciation of ¢ merges with that of
soon after the early Aksum inscriptions, and the respective graphic
symbols become liable to arbitrary interchange. It may be noted
that the coalescence in pronunciation of ¢ and may suggest the
survival, in the early period, of a fricative articulation of Proto-
Semitic ¢ (as is also made probable by the spelling of "z" for ¢ in
South Arabian: cf. § 8.19).
The Phonological System 31
5. Dental Plosives
8.21. Proto-Semitic has two non-emphatic dental plosives,
voiceless t and voiced d, as well as an emphatic plosive which was
probably voiceless and is for that reason transliterated t. (The
term "dental" is used to the exclusion of the interdentals which
have already been dealt with.)
8.22. The voiceless nature of Proto-Semitic t is corroborated
by the traditional pronunciation of Arabic and Ethiopic. The fact
that Old Babylonian t is predominantly represented by the graphic
element for d is probably due to the inconsistencies of the cuneiform
system; yet it has to be observed that in northern Babylonia t is
generally expressed by t at that period. Also, Egyptian tran-
scriptions of North-West Semitic names (second millennium B.C.)
show d for t (e.g. dbl} = Tebaly,). The weight to be attached to these
considerations is nonetheless limited, and the balance of prob-
abilities is clearly on the side of the voiceless character of t.
8.23. Akkadian does not appear to distinguish between t, d, t in
final position. In other positions, Old Akkadian and Old Assyrian
writing likewise lacks the distinction; on the other hand, it does
exist in Babylonian and in later Assyrian for the majority of the
syllabic symbols concerned. The lack of differentiation is due to
the peculiarity of cuneiform (cf. § 8.9) which evolved some separate
symbols for syllables with voiced and voiceless consonants or with
emphatic and non-emphatic ones only after 2000 B.C. The reason
for the latter deficiency is to be found in the absence in Sumerian
(and hence in its writing system as taken over by the Akkadians)
of emphatic phonemes. Some consonantal interchanges between t
and s suggest the possibility of spirantization (cf. § 8.9): the avail-
able examples (tit'aru "glittering" and sit'aru, tabsutu "midwife"
and sabsutu) are, however, substantially different from the post-
vocalic fricatives found in North-West Semitic in the first millen-
nium B.C. (more relevant, perhaps, are instances of non-positional
spirantization encountered in Egyptian transcriptions of North-
West Semitic names in the second millennium B.C., §§ 8.6, 8.10).
6. Nasal, Lateral, and Rolled Dentals
8.24. Proto-Semitic has one nasal dental consonant n, one
lateral dental l, and one rolled dental r. A non-phonemic variant
32 Phonology
of n (the palatal nasal IJ) seems to exist in Akkadian, while a variant
of l (the emphatic lateral n, probably non-phonemic (for the contrary
view see Ferguson), is found in Arabic.
8.25. The dental basis of articulation of the consonants just
listed is suggested by their traditional as well as their modern
pronunciation. Certain reservations have however been expressed
(Cantineau) for nand r: n is frequently found contiguous to other
dentals, although Semitic languages generally shun homorganic
radicals in neighbouring position. r is pronounced as a uvular
(I.P.A. [R]) in certain spheres of the traditional pronunciation
of Hebrew and shares several of the characteristics peculiar to
pharyngals and laryngals-thus pointing to a uvular articulation.
A similar situation exists, though in a somewhat less systematic
form, in Syriac where it extends to l as well.
8.26. Interchanges between the consonants of this series occur
in various languages. Those involving nand l are especially fre-
quent: Akk. and "fly" (an occasional change
n >l may be observed in Old Assyrian: kulka "seal!" for kunka);
Phoen. bl "son" for bn; N abo "statue" for Eth. sansal
"chain" in contrast to Ar. silsilat. Interchanges between nand r
(a typical case is the Aramaic br "son" for bn) and between land r
(e.g. Akk. raqraqqu and laqlaqqu "stork") are also fairly common.
In a modern Ethiopian language, Gurage, n, l, and r have become
positional variants and are thus members of the same phoneme
(H. J. Polotsky, BSLP 39 [1938], pp. 137-75).
8.27. Consonants of this group are sometimes dropped or reduced
to '. A notable example of this is the surrender of nunation (or
mimation) in the course of the historical development of Arabic
(and Akkadian): cf. for Arabic (where the omission of nunation is
always connected with that of the case-endings) § 12.68, and for
Akkadian § 12.71. The consonant n is frequently dropped in Jewish
Aramaic and in Mandaean when it is the final element in plural
morphemes: hence nouns in the stat. absol. of the plural often have
the ending -y' instead of -yn, with the result that the construct and
emphatic states become formally identical. This may, however,
be a morphological phenomenon, i.e. an extension of the use of
-y' at the expense of -yn. Analogous cases of the shedding of final n
The Phonological System 33
are found in modern Aramaic dialects: thus in the Ma'liila dialect
*lyablin "ropes" >7y,abli. Instances of the fall of final r occur in
Jewish Aramaic: e.g. 'm' for 'mr "to say" (in some forms of the
imperfect and of the imperative).
7. Dental and Palato-alveolar Fricatives
8.28. Proto-Semitic has two non-emphatic dental fricatives,
voiceless s and voiced Z. It also possesses an emphatic dental
fricative which, unlike some other emphatics, is always voiceless.
It has more than once been maintained that this consonant was
originally an affricate (of the type largely on the basis of
the pronunciation of over a wide sector of the Jewish tradition,
but this pronunciation is probably secondary.
8.29. There is another consonant of this series whose attribution
to Proto-Semitic is debatable and whose character moreover has
hitherto defied precise definition: it is usually transliterated a-but
at times also in other ways (in particular §). This consonant appears
in Hebrew and in Biblical Aramaic, but without a graphic sign of its
own (the symbol for 8 is used, and a diacritic mark was introduced
at a late date as part of the Masoretic pointing). Hence it may be
thought that it is merely a secondary differentiation of 8; yet an
examination of the correspondences in the other languages suggests
its original autonomy (cf. the comparative table below). Old
South Arabian has three symbols for the non-emphatic voiceless
dental fricatives for which the following correspondences with
Proto-Semitic have been claimed: n (Sl) = 8; (S2) = a; 3£ (S3) =S.
At any rate, it may be inferred that the three symbols correspond to
three separate consonant phonemes. An independent a with lateral
articulation is attested in modern South Arabian. North-West
Semitic in its most ancient phase shows traces of an autonomous 8:
at all events, the inferences drawn from Egyptian transcriptions
and from a gloss in the Tell Amarna letters (cf. § 8.33) seem to
point in this direction. On the other hand, the 8 symbol often used
in the transliteration of Old Akkadian does not appear to connote
a phonemic distinction from 8, and cannot, therefore, be regarded
as an independent consonant. The considerations adduced in the
foregoing would seem to be sufficient, at least cumulatively, to
claim a as an independent consonant phoneme in Proto-Semitic.
Moscati. Comparative Grammar 3
34 Phonology
As for the character of this consonant, it has been suggested
(Cantineau) that it was lateralized and that its distinction from s
lay in that peculiarity. This hypothesis is mainly based on the
lateralized B of modern South Arabian, where B is held to correspond
to Hebrew B, Arabic s, and Proto-Semitic B (e.g. Bb' "to be sated",
cf. Heb. Bb', Ar. sb').
8.30. Finally, Proto-Semitic has a voiceless palato-alveolar s (i.e.
I.P.A. rI]). Yet another sibilant (s",) has been proposed for Proto-
Semitic by Goetze (RA 52 [1958J, pp. 137-49) on the basis of certain
correspondences adduced from Old Akkadian and Old South Baby-
Ionian; but this hypothesis has not been generally accepted.
8.31. Looking at the manifestation of the consonants of this
series in the various languages, we notice that s, z, ? have regular
correspondences, while Band s have those indicated in the following
table (the identification in this table of Proto-Semitic with Hebrew
is, of course, purely conjectural):
Proto- Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic
Semitic
8 8 S S S S
S3
S
B s s B s S
S2
S
S S S S S S
S1
S
Examples: s: Akk. kusitu "garment", Ug. kst, Heb. k'asut, Syr.
kussaya, Ar. kuswat, ESA ks
3
w - B: Akk. eser "ten", Ug. 'sr, Heb.
'e8er, Syr. '",sar, Ar. 'asr, ESA 's2r, Eth. 'asru; - s: Akk. l}ame8
"five", Ug. l}ms, Heb. Syr. ly,ammes, Ar. l}ams, ESA l}msl,
Eth. l}ariws.
8.32. In Akkadian the system of writing is, as usual, inadequate
for indicating the difference between voiceless, voiced, and emphatic
(s, z, ?). Where these occur at the end of syllables the same symbols
are almost always used whatever the consonant, and the same is
true at the beginning of syllables in Old Akkadian and Old Assyrian,
and to a large extent in Old Babylonian. From the Middle Assyrian
and Middle Babylonian period onwards, syllables beginning with
z or ? are still, for the most part, written with the same symbols
(the special symbols for ?i, ?U, ?ir are an exception), while s has
either symbols of its own or shares them with 8' (e.g. s/sar, s/sab).
The Phonological System 35
In Middle and New Assyrian originals often appears as s, especially
before bilabials (e.g. usbat "she sits" in contrast to Bab. wasbat;
sapal "under" as well as sapal). Hebrew transcriptions of Assyrian
names confirm this fact: for Assyrian Barruken we have Hebrew
Sargon. The change s > s occurs also in Amorite: e.g. skn "to
put" for skn. In Assyrian (and less often in Babylonian) we have
some instances of initial s for z (e.g. siqqurratu for ziqqurratu
"temple tower"). Interchanges of s and may occur in the neigh-
bourhood of n (e.g. psn and "to veil"). From the Middle
Babylonian and Middle Assyrian period onwards the substitution
of 1 for s before a dental has become a characteristic feature (e.g.
New Babylonian iktaldu "they arrived" instead of iktaSdu). The
causes of this phenomenon are still not clear.
8.33. In the North-West Semitic of the second millennium B.C.,
Ugaritic has three symbols for the non-emphatic voiceless fricatives
of this group. An examination of the correspondences shows that
Proto-Semitic s remains, while Band s coalesce in S. In Amorite,
too, B, 8 merge in s which, in its turn, develops into s (as has been
pointed out in the preceding §). There are however indications of
an autonomous B in the Tell Amarna glosses and in the Egyptian
transcriptions of North-West Semitic names. As for the glosses,
letter 286,56 (of Jerusalem) shows sa-te-e (corresponding to Hebrew
Bade): hence B is rendered by S, whereas it would have been rendered
by s if it had merged with s as in U garitic (but this is the only
example, and the possibility of a purely graphic variant cannot be
excluded). In the Egyptian transcriptions B is rendered by s,
whereas s remains unaltered: e.g. s'r = Heb. B?,ir; snm = Heb.
(but transcriptions vary at times even for the same name:
sk and sik = Heb. Boko).
8.34. In the first millennium all the Canaanite languages, except
Hebrew, show the merging of B with s (the spelling 'sr for 'sr "ten"
in Phoenician is an isolated case). In the late Phoenician inscript-
ions from Cyprus the use of the symbol "8" for s is a noteworthy
feature (e.g. ptlmys for "Ptolemaios"). In late Punic, interchange
between dental and palato-alveolar fricatives is frequent (e.g.
sb'm for sb'm "seventy", s'w'r' for "Severus", etc.). As for Hebrew,
it is, of course, well known that the Masoretes indi{lated a graphic
distinction between Band s by placing a point either above the
3*
36
Phonology
left side of the letter (for s) or its right (for 8), the same symbol
having always served for both consonants. The distinction may
be based on ancient tradition, but we have no reliable evidence
for this: the indications furnished by the Tell Amarna letter from
Jerusalem are insufficient (§ 8.33), and the famous passage in
Judges 12,6, according to which the Ephraimites pronounced 8
as s, probably points to a dialectal differentiation rather than to the
existence of an independent phoneme s. In any case, the phenomen-
on which formed the basis of the Masoretic distinction must have
been of fairly limited extent, since by and large sand 8 appear to
have coalesced in one single consonant (just as they possessed one
graphic symbol only). It has, therefore, been conjectured that the
Masoretes may have generalized a purely dialectal differentiation.
Indeed, the Akkadian, Greek, and Latin transcriptions of Hebrew
names do not distinguish between sand 8; and St. Jerome, in
a well-known passage (Onomastica sacra, p. 36), shows that he
knows of s, ~ and 8, but not of s.
8.35. The most ancient Aramaic inscriptions show the symbol
"s" corresponding to Proto- Semitic s; the Egyptian papyri like-
wise have "s"-except for a few doubtful cases. The development
to s takes place gradually during the second half of the first millen-
nium B.C. and may be said to have been completed, save for rare
exceptions, about the beginning of the Christian era. The position
in the ancient inscriptions may be based either on a graphic
"Canaanism", at a time when the phonetic process characteristic
of Aram(:Lic had already taken place, or on an approximate render-
ing of the Proto-Semitic consonant which still survived (cf. §8.18).
8.36. It has already been mentioned that Old South Arabian
has three symbols, and their probable correspondences with
Proto-Semitic consonants have been listed (of. § 8.29). Pre-
classical North Arabic has only two symbols which correspond
to sand 8; the changes characteristic of classical Arabic (s > 8,
8 >s) seem already to have been accomplished (e.g. Lil).yanite sn
"year" compared with Ar. sanat, Heb. 8ana). It has been observed,
however, that these changes may not be very ancient, for in borrow-
ings from Aramaic they are in part accomplished and in part not
(e.g. sakkin "knife" >sikkin).
The Phonological System 37
8.37. In Ethiopic we encounter the same development as in
classical Arabic. However, the distinctive articulation of 8 has been
lost since the earliest time and has merged with that of s; conse-
quently, the symbol for s has gradually extended its scope to cases
where etymology would require the symbol for 8 (though there also
exist many instances of 8 usurping the place of s), and spelling
conventions have become quite arbitrary. It is interesting to note.
that modern Ethiopian languages have developed a new con-
sonant 8 for which they do not use the ancient character for 8 but
an adaptation of the symbol for s (Ullendorff, SLE, p. Ill).
8. Velar Plosives
8.38. Proto-Semitic has two velar plosives, voiceless k and
voiced g. It also possesses an emphatic velar plosive q, generally
regarded as the emphatic consonant corresponding to k and therefore
also transliterated 7;,.
8.39. The characterization of this last consonant as voiceless is
not completely certain. The traditional Arabic articulation is
indeed voiceless, but some indigenous grammarians and a few
modern dialects support a voiced pronunciation. In Akkadian q
is frequently written with the symbol for g (see, however, § 8.40);
in Mandaean there are many cases of g for q (e.g. g'yt' "summer",
Syr. qayta). Nevertheless, the voiceless correspondences in the
other Semitic languages confirm the voiceless character of q;
and from the phonemic point of view any uncertainty may be
accounted for by the absence of a distinctive opposition.
8.40. In Akkadian, the writing system is, as usual, inadequate
to indicate the distinction between voiceless, voiced, and emphatic.
This differentiation is entirely lacking for consonants in final posi-
tion, as well as for other positions in Old Akkadian and Old As-
syrian. In Babylonian and later Assyrian initial k and g are con-
sistently kept distinct in the majority of the symbols used, but not
in all of them: e.g. g/kir, g/kil; as for q, a special symbol for qa
occurs in Old Babylonian at Mari and Esnunna, while for other
syllables containing q separate symbols do not appear until a later
period.
8.41. In the North-West Semitic languages of the second millen-
nium B.C., certain interchanges between the consonants of this
38 Phonology
series are to be found in Amorite, in the Egyptian transcriptions
of Semitic names, in the Tell Amarna glosses, and in Ugaritic.
The series seems to attain stability later, in the first millennium:
in the Canaanite area it is not until Neo-Punic times that inter-
changes between the voiceless and the emphatic members are
attested; in the Aramaic area interchanges between the voiceless
and the voiced members are found in transcriptions of Assyrian
names, but this phenomenon is due to Assyrian (cf. § 8.40) rather
than to Aramaic factors.
8.42. In Classical Arabic g develops into g (affricate and palato-
alveolarized). The pronunciation as g is, however, attested by
Arab grammarians (although regarded as faulty) and also occurs
in sOme modern dialects of Egypt and Arabia. An analogous
tendency lc > c (though again considered "faulty") is noted by
Arab grammarians and appears in ancient and modern dialects
(in the neighbourhood, it is true, of palatal vowels and thus as
an aspect of assimilation: cf. § 9.5). A similar process of assimilation
(under the influence of front vowels) underlies the transition
k > 8 which occurs in modern South Arabian dialects as well as
in the modern languages of southern Ethiopia and in the Aramaic
dialect of Ma'liila.
8.43. In the Ethiopian sphere many cases of spirantization and
palatalization of velar plosives can be observed, but none of them
is certain for the classical period (Ullendorff, SLE, pp. 49-74).
In addition, Ethiopic has evolved, under the impact of its Cushitic
substratum, a series of labio-velars which exist alongside the
ordinary velars. This labialization embraces, in addition to the
three velar plosives, the velar fricative o,-thus producing: kW,
gW, qW, 7;w. At times the labialized consonants take the place of
the simple ones in such correspondences as: Akk. kalU "all",
Ug. kl, Heb. kol, Syr. kol, Ar. kull, Eth. kWol.
9. Velar Fri ca ti ves
8.44. Proto-Semitic has two velar fricatives, voiceless 7; and
voiced g, i.e. I.P.A. [x] and [yJ.
8.45. In a series of studies Ruzicka has maintained that g is not
a Proto-Semitic consonant but an Arabic innovation. Originally,
The Phonological System 39
the argument in support of this thesis lay in the fact that g was to
be found in Arabic only, and even there it was in some cases second-
ary, i.e. derived from the pharyngal' (e.g. musawwag "permitted",
a variant of musawwa'). When an independent g-or at least an
independent graphic symbol-was identified in South Arabian
and in Ugaritic, Ruzicka dismissed the South Arabian evidence as
a mere extension of the Arabic phenomenon and claimed that the
symbol taken for g in Ugaritic corresponded in some instances to '.
From this he inferred that g did not exist in Ugaritic but that the
symbol in question was simply one of a number of attempts at
fashioning a suitable graphic sign for '. Ruzicka's contentions have
been partially supported by Petracek (ArOr 21 [1953], pp. 240 to
262; 23 [1955], pp.475-78) who has endeavoured to show, in
a statistical investigation, that in Arabic g is of a complex phonemic
nature, being partly a variant of ' and partly an independent
phoneme. This condition can be explained in terms of the acquisi-
tion of independent phonemic status of what was originally a mere
variant. Against this set of observations there still remains the
fact that in classical Arabic, South Arabian, and Ugaritic g possesses
a clearly circumscribed independence which is not invalidated by
a number of peripheral developments. Moreover, it has recently
been pointed out (Rossler, ZA 54 [1961], pp. 158-72) that Proto-
Semitic g-as distinct from Proto-Semitic '-does not always
occasion the Old Akkadian change a > e: a fact which would point
to its independent existence in the most ancient phase of East
Semitic. It appears, therefore, that g is to be retained among the
Proto-Semitic consonants.
8.46. The correspondences of the velar fricatives in the principal
Semitic languages are as follows:
Proto- Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic
Semitic
7; 7; 7;
lJ lJ
7; 7; 7;
g g g g
Examples: 7;: Akk. a7;u "brother", Ug. 67;, Heb. 'alJ" Syr. 'alyii,
Ar. 'a7;, ESA '7;, Eth. 'o7;W;-g: Akk. 'rb "to enter", Heb. Syr.
Eth. 'rb, Ar. ESA grb; Ug. film "boy", Heb. 'elem, Syr. 'olayma,
Ar. g ~ t l i i m , ESA glm.
40
Phonology
8.47. In Akkadian interchanges occur between lJ and k (e.g. lJns
instead of the usual kns "to submit"). Akkadian lJ corresponds
in some cases to Semitic 7;, (e.g. lJkm "to understand", cf. AI'. 7;,km,
whereas the normal Akkadian development [see § 8.54] is 7;, >')
or to g (e.g. flJr "to be small", cf. AI'. fgr).
8.48. In North-West Semitic of the second millennium B.C. the
existence of lJ and g is attested both in U garitic and in the Egyptian
transcriptions of Semitic names. In these transcriptions "lJ" is
used for lJ, and "g" (or "q") for g, while "1;1" is employed for 7;,
and "," for ' (e.g. Egyptian nlJr = Reb. na7;,al, Akk. nalJlu;
Egyptian mgrt = Reb. m,/ara, AI'. magarat; Egyptian qrJt = Reb.
'Azza, AI'. Gazzat). Some Ugaritic uses of the symbol "g" for lJ as
well as the correspondence between U garitic "g" and Akkadian
"1;1" in a syllabary from Ugarit may suggest voiceless articulations
of Ugaritic g (Garbini, SNO, p. 52).
8.49. In North-West Semitic of the first millennium the process
lJ > 7;" g >' is complete. The only problem in this connexion is
posed by the Greek and Latin transcriptions of Rebrew which
show for 7;,: X (ch), e, zero; for': y (g), e, zero. It has been suggested
, that the transcriptions X (ch), y (g) stand for the original con-
sonants lJ, g, respectively, while the others stand for original 7;" '.
This hypothesis can, of course, be tested by a comparative ex-
amination, and the result of such a test militates against. the
hypothesis (e.g. X corresponds to 7;, in A{JLXm)", Reb. 'abi7;,ayil;
y corresponds to ' in rorpeea, from the root 'fr). The variations
in the transcriptions seem instead to relate to different periods:
X (ch) and y (g) predominate in the earlier period, while e and
zero belong to a later one. Another feature characteristic of
late North-West Semitic, which appears in Eastern Syriac and
in a sector of the Jewish-Ashkenazi tradition (though European
languages are bound to have affected Ashkenazi pronunciation),
is the rendering of 7;, as lJ. The correspondences exclude the possi-
bility of a survival of an original !J.
8.50. In Ethiopic the pronunciation of !J gradually coalesces
with that of 7;,; this is reflected in'graphic interchanges of increasing
frequency and arbitrariness.
The Phonological System 41
10. Pharyngal Fricatives; Laryngals
8.51. Proto-Semitic has two fricative pharyngals, voiceless 7;, and
voiced' (I.P.A. En] and [<1], respectively).
8.52. Proto-Semitic has two laryngals: one glottal plosive,
, (I.P.A. [?]) and one voiceless laryngal fricative h (of which,
however, there are some voiced manifestations in modern Arabic
dialects).
8.53. The consonants of the pharyngal fricative and laryngal
series have regular correspondences in the various Semitic lan-
guages, with the exception of Akkadian where they are reduced
to' (or to zero). There are, however, extensive phonetic reductions
and losses which it is well to examine individually (for phenomena
of syncope, cf. § 9.20).
8.M. In Akkadian (as has just been mentioned) these consonants
have been reduced to '-under the influence of Sumerian which
did not possess the consonants of this series. The reduction is not
yet complete in Old Akkadian (cf. the use of the symbol E for the
phonetic values 'it and it', probably corresponding to the Proto-
Semitic consonants hand M; in Old Babylonian, too, there are
indications that SOlIle laryngals at least were still pronounced
('adanum "limit" written with initial !J); in New Assyrian it is
probable that h reappears, because anniu "this" is often spelt
!Janniu (pronounced [hanniu]?). It is only from the Middle Baby-
lonian and Middle Assyrian periods onwards that ' has symbols of
its own which even then are not regularly employed. Apart from
the use of specific symbols, ' may be graphically expressed in various
ways: by the symbol for the vowel which follows (e.g. is-a-am
for is'am) or by the symbols for !J (e.g. e-!Ji-il-tum for e-'},-il-tum).
It should be observed that the graphic notation of " partial and
irregular in medial position, is usually absent at the beginning of
words (cf. von Soden, GAG, p. 24 for some rare exceptions). The
Assyriological custom of not transliterating even initial' is followed
in the present work; it should be clear, however, that the absence
of a symbol does not necessarily coincide with phonetic reality.
There is no reason to suppose that the situation in Akkadian ran
counter to the general Semitic rule which requires that every
42 Phonology
syllable should begin with a consonant (cf. § 10.2). An identification
of the consonants which had coalesced in ' is at times possible on
the basis of modifications to which neighbouring vowels have
been subjected: for' derived from g, ly" ' occasions the change a into
e (e.g. *'aprum "dust" > eprum). This transition does not, however,
always take place in the case of g (cf. § 8.45), while on the other
hand it sometimes occurs with h (e.g. ewilm "to become" compared
with Aram. hJwii).
8.55. In Canaanite, a weakening of the pharyngals is suggested
for pre-Masoretic Hebrew by Greek and Latin transcriptions
(of. § 8.49) and by interchanges with' and h which are attested in
the Dead Sea documents. It is, therefore, not altogether impossible
that the Masoretes may have aimed at restoring the ancient pro-
nunciation by means of their peculiar system of vocalizing the
pharyngals. A characteristic feature of Punic, as distinct from
Phoenician, is the gradual weakening and eventual reduction to '
(or zero) of ly" ',h. This phenomenon becomes manifest to only
a limited degree in official documents where the traditional ortho-
graphy prevails; but it is prevalent in popular inscriptions in which
constant interchanges and losses occur in the pharyngal and
laryngal series (e.g. 'd for 'ly,d "one").
8.56. Aramaic, prior to the division into West and East Aramaic,
retains by and large the independent articulation of the pharyngals
and laryngals (some weakening which may be observed in the
Aramaic of Assyria is probably due to Assyrian influence: e.g.
, >' in ' r ~ t ' for ' r ~ t ' as well as many cases in which intervocalic' is
dropped: mry for mr'y, etc.). Later, extensive areas of phonetic
uncertainty occur which are reflected in the orthography. In the
languages of the Western group the consonants in question are
frequently interchanged or dropped altogether; in those of the
Eastern group the reductions '>', ly, > h are very frequent and
may, in fact, extend further to '> zero, h >' > zero. Syriac, in
particular, shows many cases in which ' loses its consonantal value
and is then dropped in the current spelling convention (e.g. ly,ad
"one", cf. Heb. 'ely,ad, 'Ar. 'aly,ad, etc.); h often loses its consonantal
character (e.g. the pronouns hil, hi "he, she" lose h in enclitic
position): for details cf. Brockelmann, SG, pp. 25-26.
The Phonological System 43
8.57. In the Arabian area, Old South Arabian displays the
transition' >' in the dialect of the l;IaQ.ramawt (e.g. 'd "up to"
for 'd). Classical Arabic exhibits a remarkable stability of the
pharyngals and laryngals, though a few traces of the development
'>', ly, >h are attested in some ancient dialects. As for " it is
possessed of an exceptional constancy in the orthography of the
classical language.
8.58. In Ethiopic we observe a gradual phonetic reduction of
ly, to h and of ' to '; this uncertainty (resulting eventually in almost
complete arbitrariness) does not appear in the most ancient in-
scriptions of Aksum and may well be due to the influence of
Amharic. The latter affects the orthography of classical Ethio-
pic, so that with the passage of time inconsistencies become ever
more prevalent. But there are, of course, no grounds for denying
the original phonemic independence of the consonants of this
series.
11. Synopsis of the Consonantal System
8.59. To return to the table of the Proto-Semitic consonant
system (cf. § 8.3), the evolution of this system in the principal
languages of the group may be envisaged as follows:
Proto- Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic
Semitic
p p p p p
f f f
b b b b b b b b
m m m m m m m m
t 8
t
8 t
t t 8
0
z d (0?) z d
0 0
z
t ~ t ~ t
:;:: :;::
~
¢
~ ~ ~
¢ ¢ ¢
t t t t t t t t
d d d d d d d d
t t
n n n n n n n n
Phonology
Proto- Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic
Semitic
r
8
Z
~
8
8
k
g
q
If
g
lJ
h
r
8
Z
~
8
k
g
q
If
r
8
Z
~
8
8
k
g
q
If
g
lJ
h
r
8
Z
~
8
8
k
g
q
lJ
h
r
8
Z
~
8
k
g
q
lJ
h
r
8
Z
~
8
8
k
g
q
If
g
lJ
h h
r
8
Z
~
8
8
k
g
q
If
h
8.60. The extent to which the original consonantal system has
actually survived in the various languages (irrespective of etymo-
logical relationships) is shown in the following table:
Proto- Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic
Semitic
P
b
m
t
rJ
1
rt
t
d
t
n
r
8
Z
~
8
P
b
m
t
d
t
n
r
8
Z
~
P
b
m
t
(rJ)
1
t
d
t
n
r
8
Z
~
P
b
m
t
d
t
n
r
8
Z
~
8
P
b
m
t
d
t
n
r
8
Z
~
f
b
m
t
rJ
?
rJ
t
d
t
n
r
8
Z
~
f
b
?
rJ
t
d
t
n
r
8
Z
~
8
f
b
m
rJ
t
d
t
n
r
8
Z
~
The Phonological System 45
Proto- Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic
Semitic
g
q
If
g
lJ
h
8
g
q
If
g
q
If
g
lJ
h
g
q
h
g
q
h
12. Semi vowels
h
g
q
If
g
lJ
h
g
q
If
h
8.61. Proto-Semitic has a bilabial semivowel wand a palatal
semivowel y, i.e. I.P.A. [j].
8.62. Both semivowels have regular correspondences in the
various Semitic languages. They are, however, subject to changes
and reductions (for some phenomena of syncope cf. § 9.20).
8.63. Since w is rare, and probably secondary, in the Sumerian
writing-system (and language), its graphic notation in Akkadian
is somewhat uncertain and imperfect. Up to the Old Babylonian
and Old Assyrian periods the syllables wa, we, wi, wu are written with
the Sumerian symbol PI; later on, for w the symbols for mare
predominantly used in Babylonia, and in Assyria those for b (e.g.
Old Bab. Ass. awatum "word", Mid.-Bab. amatu, Mid.-Ass. abatu).
At the beginning of words, w is generally preserved until the Old
Babylonian and Old Assyrian periods; afterwards it is either
dropped (or reduced to ') or written with symbols for m (e.g.
wU88urum "to send" > u88uru and mussuru; waradum "to descend"
>aradu). For w in medial position cf. § 9.20. As regards the other
semivowel, y, Sumerian possessed only the phonetic sequence i-a,
and the symbols for this are used not only for the graphic expression
of the Akkadian syllable ya, but also for yi, ye, yu; the symbols
A-A are employed for the sequences ay, aya, ayya, ayyi, ayye,
ayyu (other sequences are not encountered owing to elisions and
contractions). In initial position y almost invariably disappears
(or is reduced to '), at times leaving behind the vowel which
46 Phonology
accompanied it (e.g. yu >tt), while at other times the homorganic
vowel i (e.g. ya>i) remains. Finally, Old Akkadian spellings of
the type i-ik-mi, i-ig-mu-ur (Gelb, OA, p. 158) seem to suggest
the possibility of a prefix yi- as an intermediate phase in the change
ya- >i-. For y in medial position cf. § 9.20.
S.M. In North-West Semitic there is a characteristic development
w > y in initial position (e.g. Akk. Ar. Eth. wld "to bear", Ug.
Reb. Syr. yld). This phenomenon can already be seen in Amorite,
in the Egyptian transcriptions of the second millennium, in Uga-
ritic and in the Tell Amarna glosses. Some exceptions in the Egyp-
tian transcriptions might suggest that this process was then still
in the evolutionary stage. Initial w is kept in the conjunction w
"and" and in a few nouns (e.g. Reb. walad "child"). Some survivals
of w in Nabataean (in cases where the other North-Western lan-
guages have y) may be explained as due to Arabic influence.
8.65. In ancient Arabian dialects an occasional change w > y
in initial position is suggested by such cases as yazi'ahum "their
protector" for wazi'ahum (Rabin, W A, pp. 65, 83). Reductions
of initial w, y to ' are also found (e.g. 'uguhuhum "their faces"
for wu{juhuhum, 'iqa' "protection" for wiqa'). Another phenomenon
attested in these dialects is the pronunciation of y as {j; but the
stock example usually cited, 'iyyal "deer" > 'i{j{jal (Rabin, WA,
p. 199), may conceivably be the result of dissimilation of the
semivowel in relation to the homorganic vowel i which precedes it.
13. Vowels
8.66. Proto-Semitic has three short vowels: open back velar
a, i.e. I.P.A. [a], close front palatal i, and close back velar u with
strongly rounded lips. Proto-Semitic also possesses the three
corresponding long vowels: a, i, u. Traces of vocalic land r have
also been claimed (von Soden, GAG, p. 11), but further study is
required.
8.67. There are no certain grounds for supposing that Proto-
Semitic had once possessed additional vowel phonemes. In partic-
ular, the addition to the vowel-system of e, which has more than
once been postulated (cf. most recently Rabin, WA, pp. 110-11),
meets with difficulties in demonstrating the phonemic status of
The Phonological System
47
this vowel. From the phonetic point of view it may be taken for
granted that not only this vowel but numerous other varieties
have existed in Semitic since its most ancient phase.
8.68. The Proto-Semitic vowel system has an exact reflection
in that of Arabic whose full network of graphic symbols mirrors the
phonemic position. The history of Arabic and its dialects shows
clearly in what manner vowels of other timbres have evolved in
the Semitic languages and have, in the course of time, acquired
phonemic status. These vowels have arisen in two main ways:
by change under the influence of neighbouring consonants and by
. contraction of diphthongs (aw > 0, ay>e). The non-phonemic
variations e for a, 0 for u, e for i are so common that Arabic vowels
are rightly classified according to the place of articulation rather
than on the grounds of timbre (Fleisch, TPA, p. 63).
8.69. The graphic notation of vowels in the various Semitic
languages is bound up with the system of writing adopted by each
one; and these systems vary between certain extremes of phonemic
and phonetic representation. The examination which follows will
be especially concerned with the vowel-systems of those languages
for which we possess the best sources of information. For other
languages, and especially those of the North-West Semitic group,
the consonantal system of writing does not offer sufficiently
solid grounds for adequate reconstruction, even though there
exists a good deal of circumstantial evidence (general Semitic
comparisons, foreign transcriptions, matres lectionis); the principal
data available will, of course, be recorded.
a. Akkadian
8.70. Akkadian presents a vowel-system identical with that of
Proto-Semitic, but with the addition of the vowel e, either short
or long (e, e), which appears to be derived from a or i (a, a; i, i).
In the writing-system the series of symbols with e is very incomplete;
in the southern dialect of Old Babylonian i occurs so frequently for e
that this feature has been regarded as reflecting a dialectal pecu-
liarity.
8.71. The graphic interchanges between u and i and between
u and a in certain forms (e.g. for "pain",
48 Phonology
for i-na-far "he watches") have been claimed as evidence for the
existence of vowel qualities of the type [y], [0] (cf. von Soden,
JCS 2 [1948], pp. 291-303).
8.72. The construct state (of. §§ 12.78-79) in -i of some mono-
syllabic substantives (in contrast to the usual absence of endings
in that form) has suggested the existence in Akkadian (known also
in other Semitic languages) of a vowel of the 'J type (von Soden,
GAG, pp. 10, 82).
b. North-West Semitic of the Second Millennium B.C.
8.73. Amorite, which has come down to us in cuneiform script,
exhibits a vowel-system identical with that of Akkadian-with
the sole exception that e does not appear to be an independent
phoneme but rather an allophone of i (Gelb, RANL 13 [1958],
pp. 146-47). Some interchanges of i and u (e.g. binum and bunum
"son") may possibly suggest the existence of vowel qualities of
the [y] type. Similar considerations might apply to other inter-
changes, e.g. u and a (sumum and samum "name") where one might
suppose a vowel of the [0] type.
8.74. The language of the Tell Amarna glosses (likewise in cunei-
form writing) also displays a vowel-system like that of Akkadian.
The vowel e, frequently resulting from an original a or i, now
appears to be established as part of the phonemic system, even
though it started as a mere allophone. The glosses show the,
apparently non-conditioned, change a> 0 (e.g. a-nu-ki, of. Heb.
'anoki, against Akk. anaku: the writing of u for 0 is due to the
absence in cuneiform of a proper notation for the vowel 0); cf.
§ 8.83. An instance of the change a> 6 is now attested in an Old
Babylonian inscription from Mari: ClJamafam » !Jam1l§am ilfmuf
"he plundered thoroughly".
8.75. In Ugaritic the writing-system is consonantal, but the
consonant ' has t h r ~ e symbols according to the vowel which
follows, i.e. a/a, i/i, u/u. From this we may probably infer that
the Ugaritic vowel-system corresponds substantially to that of
Proto-Semitic. The problem of representing unvocalized ' is open
to argument, for the data are far from being consistent. In the
majority of such cases the symbol for' with the vowel 1: is used;
The Phonological System
49
but sometimes the symbol used is that of ' plus the vowel identical
with that which precedes it; and on other occasions it is the symbol
for ' with any vowel indiscriminately (or at least so it appears t.o
us in the absence of a rational explanation).
8.76. It has been .observed that in Ugaritic the symbol for' with
u/u corresponds also to '(*aw » 0 and that for ' with i/i to
'(*ay » e (Gordon, UM, p. 17). It might be averred in this con-
nexion that the diphthongs aw, ay can also evolve into ii, i (of. Akk.
*baytu > bitu "house" ; *mawtu > miitu "death"); but this Akkadian
development is not attested in N.orth-West Semitic. Moreover,
it is probable that there existed vowels .of the e, 0 timbre, not as
independent phonemes but as allophones of a, i, u. Some vowel
interchanges similar to those in Amorite (cf. § 8.73) have called
forth the idea of vowel qualities of the [y], [0] type. It is possible
(cf. § 8.74) that the symbol for' with the vowel i may in some
cases represent a 8'Jwa (cf. Garbini, SNO, pp. 63-64).
c. Canaanite
8.n. In the Canaanite area the Phoenician vowel-system (which
can be partially reconstructed by means of Akkadian, Greek, and
Latin transcriptions of Phoenician words) presents the usual Semitic
phonemic vowels (§ 8.66) in a number of varying pronunciations.
Thus we find a pronounced as e ('sQa for *zar' "seed"), i as e
(O'sQfJaAor; for *'Azir-ba'al), u as 0 (Baliahon for * Ba'al-yaly,iin).
The original long vowels appear to be more stable than the short
ones, but in Phoenician we may observe the non-conditioned
change a > 0 (e.g. macom for *maqam "place"), for which cf. §8.83.
8.78. The vowel notation of Biblical Hebrew is the work of the
Masoretes and originated during the second half of the first millen-
nium A.D.; it therefore postdates the consonantal text by a very
considerable margin. The reconstruction of the original vocalization
has been attempted by utilizing the transcriptions of Hebrew
names in other languages (Sperber's studies) as well as by the
application of modern linguistic techniques (Z. S. Harris in JAOS
61 [1941] pp. 143-67). The tentative results show considerable
divergences from the Masoretic system. Within the system itself
three different traditions can be distinguished: the Babylonian,
the Palestinian, and the Tiberian. The first and the second of these
Moscati, Comparative Gramnlar 4
50 Phonology
indicate the vowels by means of supralinear signs, while the third
uses (with one exception) sublinear symbols. A characteristic
feature of the Masoretic vowel notation is the fairly elaborate
representation of qualitative distinctions.
8.79. According to the Tiberian system, which has prevailed
in Hebrew manuscripts and later on in printed books, the Biblical
vowel system may be represented as follows:
i e a o u
8.80. Vowel quantity is not in general indicated by the symbol
as such, but depends on the position of the vowel within the word;
only of a can it be said that it is normally a short vowel. As for ::I,
it has two different pronunciations: it is either long when it cor-
responds etymologically to a or short when it corresponds to an
original u. On etymological grounds, naturally paramount in
a comparative grammar, as well as for other reasons set out above
(cf. §§ 6.12-15), the present treatise will employ the following
transliterations (taking the seven vowels in the same order as in
the preceding paragraph):
i/i e/e a a/o 0/0 u/ii
8.81. Combinations of vowel symbols with matres lectionis serve
to indicate a series of predominantly long vowels: w is used as
mater lectionis for vowels of the timbre u or 0; y for those of timbre
i or e; h at the end of a word for those of timbre e, a or 0; , in the
middle or at the end of a word for those of any timbre. The indi-
cation of vowel quantity by means of matres lectionis is very
imperfect-in contrast to the position in Arabic (§ 8.91)-for in
Hebrew we have long vowels without matres lectionis and short
ones with matres lectionis. The use of matres lectionis in the Dead
Sea documents is somewhat peculiar: ' is extensively used at the
end of words, and y appears frequently instead of h (at the end of
words) to indicate vowels of the e timbre.
8.82. To the symbols enumerated above we have to add :- (s'ifwa)
which originally marked the absence of a vowel but which has come
to indicate in certain positions (at the beginning of syllables) the
vowel of the type 'if (M urmelvokal). In combination with other
The Phonological System 51
symbols S'ifwa produces the compounds -=:' To' To' employed with the
pharyngal and laryngal consonants and here transliterated ii, e,
0, respectively.
8.83. Compared with the Proto-Semitic system Hebrew vocal-
ization displays a noteworthy development. This is, however,
closely linked to syllabic structure and stress patterns and will,
therefore, be dealt with at the appropriate entries below. Only
one of the vowel changes in Hebrew appears to be non-conditioned,
i.e. a> 0 (e.g. Akk. samane "eight", Heb. s'ifmone). This change,
which can already be observed in the Tell Amarna glosses and in
Phoenician (§§ 8.74, 8.77), has long been considered a characteristic
of "Canaanite" (in the traditional sense of that term); but in fact,
while it does not occur in Ugaritic, it reappears in the early cen-
turies of the Christian era both in West Aramaic (cf. §§ 8.84, 8.88)
and in pre-Islamic Arabic (Rabin, WA, pp. 28, 105-10).
d. Aramaic
8.84. While Biblical Aramaic uses the same system as Hebrew,
some earlier data are furnished by a magical text in cuneiform
from the third century B.C. (A. Dupont-Sommer, RA 39 [1942-44],
pp. 35-62). The vocalization of this text has-as compared with
the general development of North-West Semitic-a rather archaic
appearance: for example, the original i remains in la-bi-is as
against Heb. Some cases of the change a > 0 (not attested
in Old Aramaic) occur in Palmyrene (cf. § 8.83); this process is,
later on, characteristic of Western Syriac (cf. § 8.88).
8.85. Syriac has a vowel notation going back, like that of Hebrew,
to the second half of the first millennium of the Christian era;
again like the Hebrew system it is characterized by the predom-
inance of qualitative over quantitative distinctions.
8.86. There are two different methods of vowel notation, the
Eastern (used by the Nestorians) and the Western (used by the
Monophysites or J acobites); the latter is based on Greek vowel
symbols:
Eastern: -: a, -: a, - ..-e, -;-e, iii, C? u/ii, 00/0
Western: a/a, -'-e/e, "::-'i/i, -'-u/ii
4*
52 Phonology
8.87. This notation makes use, for the vowels i, u, 0, of matres
lectionis which are widely employed in Syriac: w for u/11, o/a;
y for iIi, and in medial position sometimes for e; 'for a (Western a ~ ,
and for e (Western i) in final (occasionally also medial) position.
8.88. In addition to the difference in the actual system of notation
there is a phonetic distinction between the Eastern vocalization
(which is used in the present treatise) and the Western one. East
Syriac preserves a more ancient vocalism, whereas West Syriac
presents the following developments.: a > a (cf. Hebrew: § 8.83);
6 >11; 0 >u; e >i (in certain types of words). Examples: ESyr.
paraqa "saviour", WSyr. par11qa; ESyr. resa "head", WSyr. riS6.
8.89. Syriac possesses no symbol to indicate the absence of a
vowel or to mark a vowel of type ~ , though the existence of such
a Murmelvokal (a central vowel whose precise timbre is determined
by the nature of surrounding consonants as well as by the effects
of vowel harmony) must be assumed in certain positions.
8.90. Compared with the Proto-Semitic system Syriac vocalization
presents important developments which are' connected-as indeed
is the case in Hebrew-with syllabic structure and the incidence
of stress; they will be dealt with at the appropriate place below.
e. Arabic
8.91. Pre-classical Arabic does not furnish sufficient indications
for a reconstruction of its vowel system, and in any event it is
improbable that this system differed appreciably (at least from the
phonemic point of view) from that of classical Arabic (for the vexed
question of e of. § 8.67). The classical language presents a vowel
system which corresponds phonemically to the Proto-Semitic one.
As for its notation, matres lectionis were consistently used for the
indication of long vowels: w for 11, y for i, ' for a. This is natural
in the case of wand y in view of the Aramaic (or rather Nabataean)
origin of the Arabic script; only the use of ' to mark a may be re-
garded as a specifically Arabic development (and is not, in fact,
customary in the more ancient texts). For short vowels (and for
long ones in combination with matres lectionis) a system of symbols
was introduced (in the late ninth century A.D.) which is derived
from somewhat simplified forms of the matres lectionis:
The Phonological System 53
8.92. There is a special symbol ( ~ ) to denote the absence of a
vowel; a vowel of type ~ does not exist in Arabic.
8.93. From the phonetic point of view, traditional grammar
and the history of the dialects provide some idea of the extensive
variations in the timbre of Arabic vowels. The principal tendencies
noted by Arab grammarians are: a) 'imala, i.e. [a:J > [e:J, a
non-conditioned phenomenon of palatalization whose realization
is at times prevented by the operation of conservative forces;
b) taffpim, i.e. [a: J > [0:], a less frequently occurring phenomenon
of velarization, sometimes conditioned by the neighbourhood of
emphatic consonants; c) 'ismam, i.e. [i:] > [u:], another pheno-
menon of velarization whose true nature is, however, somewhat
less certain (cf. Rabin, WA, p. 159).
f. Ethiopic
8.94. Old Ethiopic was at first written without vowel signs, but
in the fourth century A.D. it introduced a very special type of
vowel notation which operates by means of partial alterations in
the form of the consonantal symbol. The vocalism which is mani-
fested by this notation consists of seven elements; and it may
now be regarded as established that these elements reflect essentially
qualitative distinctions.
8.95. The qualitative values of the Ethiopic vowel series are as
follows (for the purposes of exemplification it is convenient to use
the consonant symbol l):
u i a e
A
( ~ ) o
8.96. Etymologically, u i a correspond in general to Proto-
Semitic 11 i a, respectively; e 0 to the diphthongs ay aw, respectively;
and a to a. Two elements of the Proto-Semitic system appear at
first sight to be unrepresented, i.e. short i u, but in Ethiopic they
have coalesced in the vowel ~ (e,g. Ar. 'ug,n "ear", Eth. 'nn;
Ar. sinn "tooth" , Eth. s ~ n ) in conformity with some general
relationship which, in many instances in Semitic, seems to exist
between these two vowels in opposition to a. For reasons of etymo-
logical correspondence, paramount in a comparative grammar,
54 Phonology
the present treatise will use the following transliterations (in the
same order in which the Ethiopic vowels appear in the preceding
paragraph) :
a i
As regards the sixth vowel, the transliteration 'J is not free from
ambiguity, for though this vowel may correspond to the sound
of 8'Jwa mobile of other languages, it is a stable vowel which may
even be long (Ullendorff, SLE, p. 160). The ambiguity of the
Ethiopic sixth order as either 'J or zero causes difficulty not only
to Europeans, but even Ethiopian scholars sometimes disagree
about it in the traditional pronunciation of Ge'ez.
14. Diphthongs
8.97. The combination of semivowels and vowels produces a series
of rising or falling diphthongs; these are subject to a number of
conditioned changes which will be dealt with at the appropriate
place. Some changes, however, are not necessarily conditioned:
they affect the diphthongs aw, ay whose treatment is a differenti-
ating feature between certain Semitic areas.
8.98. In Akkadian the Proto-Semitic diphthongs aw, ay generally
appear as U, i: e.g. *mawtu "death" > mutu, *'aynu "eye" > inu.
In Assyrian, and partly in New Babylonian, e takes the place of i
(enu). The Akkadian phenomenon may possibly be explained as
the result of assiinilation (of. § 9.8). An exception occurs in the
case of ay before y (e.g. ayyabu "enemy") and in some instances of
the vetitive particle ay "not".
8.99. In the North-West Semitic of the second millennium B.C.
Amorite shows the preservation of aw as well as the develop-
ments am and a; for ay we find a and eli. The Egyptian transcrip-
tions of Semitic names attest sometimes to the loss and sometimes
to the retention of the semivowel element: perhaps they reflect
a stage in the course of actual evolution. The reduction is shown
to be complete in the Tell Amarna glosses and in Ugaritic where
we encounter the result of the changes aw > 0, ay > e (cf. however
§ 8.76). Before y (as in Akkadian) ay does not seem to be reduced
in Ugaritic (Gordon, UM, p. 27): it is conceivable that syllabic
The Phonological System 55
extension (through anaptyxis) might have occurred as in Hebrew
(§ 8.100).
8.100. In the Canaanite sphere, Phoenician exhibits the reductions
aw > 0, ay > e: e.g. !wjht).:xov for * Ye!paw-milk, caneth for *qanayti
"I acquired". In Hebrew the same reductions are generally found:
e.g. *yawm > yom "day". In some cases, however, the diphthongs
remain unreduced, especially in final position: e.g. qaw « *qaww)
"cord", !pay « *lpayy) "living". In doubly-closed syllables we
meet instances of syllabic extension through the insertion of a new
vowel: e.g. *mawt > miiwet "death", *bayt > bayit "house".
These are probably cases of anaptyxis (cf. § 9.17).
8.101. "Defective" writing in Aramaic shows that the reduction
of the diphthongs had taken place even in the most ancient in-
scriptions; some exceptions in Egyptian Aramaic are doubtless
to be explained as instances of historical spelling. In Biblical
Aramaic aw is reduced unless it is followed by w; ay sometimes
remains uncontracted and may give rise to syllabic extension as
in Hebrew (§ 8.100). In Syriac the diphthongs are preserved-
except when their preservation would result in a doubly-closed
syllable: e.g. 'ayna "eye", but st. constr. 'en.
8.102. Whereas Old South Arabian shows graphic variations
(e.g. ywm and ym "day") which may suggest that the process of
reduction was at an active stage (of. HOfner, Altsudarabische
Grammatik, pp. 9-11, 22-23), classical Arabic preserves the
original diphthongs in their entirety; but they undergo extensive
contractions in the modern dialects.
8.103. In Ethiopic the diphthongs appear in reduced form (e.g.
*yawm > yom "today", *layl > Wit "night"), but there are a
number of divergent formations (cf. Dillmann, EG, § 39, pp. 78-79).
8.104. A number of secondary diphthongizations are to be found
in the Semitic languages, and those in Ethiopic (for which cf.
Ullendorff, SLE, pp. 170-83) are particularly noteworthy. They
are, however, secondary phenomena in the various languages,
even if they occurred at an ancient period: of. E-Byr. haykal from
Akk. ekallu "palace" (a Sumerian word).
56 Phonology
O. Oonditioned Phonetic Ohanges
9.1. The manifold phenomena of conditioned phonetic evolution
have not yet been sufficiently investigated from the point of view
of comparative Semitic linguistics: a study of these phenomena and
processes in the various languages and groups, as well as the
determination of their frequency, will undoubtedly contribute
to a better comparative appraisal of the Semitic languages. In
the treatment which follows some of these instances will be
identified and illustrated; attention will be drawn to salient aspects
and characteristic features in individual languages and groups,
while for detailed discussion the reader is referred to the grammars
of the various languages concerned.
1. Assimila.tion
9.2. The Semitic languages present assimilatory processes of
various kinds: assimilation may take place between consonants,
or between vowels, or of consonant to vowel, or vowel to consonant,
or of diphthongs; it may be progressive or regressive or reciprocal;
it may be partial or total; and it may be contiguous or at distance.
9.3. a) Between consonants.-Progressive, partial and
contiguous: e.g. Ar. "it was dyed" > Akk.
(New Ass.) *aqtirib "I approached" > aqtirib, "I fought"
> (later Bab.) (voicing). Contiguous assimilation may
be the cause of the vuicing of t, in some roots with second radical b,
which occurs in some West Semitic languages, both Northern and
Southern: e.g. Akk. 'bt, WSem. 'bd "to perish"; Akk. kbt, WSem.
kbd "to be heavy". It was probably contiguous assimilation that
gave rise to the "emphaticization" (in North-West Semitic) of
t in the root q[l "to kill" as compared with Ar. Eth. qtl (though
dissimilation might conceivably have occurred in the South
Semitic languages).-Progressive, partial and at distance: possibly
Syr. purq(Jsii "tower", from Greek partial
and contiguous: Eth. 'agii'(Jzt lords" > 'agii'(Jst (devoicing).-Re-
gressive, partial and at distance: e.g. Ar. buq'at "plain", Heb.
biq'ii, but Syr. p(Jqa'tii (devoicing).-Progressive total: e.g. Ar.
*ittalaba "he sought" > ittalaba; Akk. *attarad "I sent" > attarad
(the assimilation of t in the Akkadian infixes ta and tan is always
Conditioned Phonetic Changes
57
total when following d, t, z, s: e.g. "he imprisoned" >
total: e.g. Heb. *yintr:n "he gives" > yittr:n ;
the assimilation of vowelless n to the following consonant is
characteristic of North Semitic; it does not occur in South Semitic
with the exception of some instances in South Arabian (cf. §§ 16.116
to 117); the assimilation of vowelless 1 to the following consonant
(which is sporadically found in various languages) takes place most
prominently in the case of the Arabic article before all interdental,
dental, and palato-alveolar consonants, yet 1 continues to appear
in the graphic pattern: e.g. 'al-sams "the sun", pronounced
[afJams]. Reciprocal. assimilation: e.g. Ar. *'i(!takara "he re-
membered" > 'iddakara.
9.4. b) Between vowels.-Assimilation of vowels (or vowel
harmony) is always at distance, since the structure of the Semitic
syllable does not admit vowels in positions of direct contact (cf.
§ 10.2). Vowel harmony is particularly extensive in Akkadian
(von Soden, GAG, pp. 12-13).-Progressive partial: e.g. Akk.
lJibliitu "damage" > lJibletu.-Regressive partial: e.g. Akk. ulJappi
"he struck" > ulJeppi.-Progressive total: e.g. Ar. *riglihu "of
his foot" > riglihi (vowel harmony in the suffix-pronoun of the
third-person m. singular is standard in Arabic).-Regressive total:
e.g. Ug. ulp "prince", cf. Heb. 'allilp; Ar. *sanina (c.ob1.) "years" >
sinina. A typical case of regressive total assimilation occurs in
Assyrian vowel harmony whereby a is assimilated to the vowel
of a case-ending which follows it: e.g. nom. qaqqudu "head", gen.
qaqqidi, acc. qaqqada; cf. also Ar. 'imru'un "man", 'imri'in, 'imra'an.
9.5. c) Consonant to vowel.-In Hebrew and in Aramaic,
after the latter's division into Eastern and Western dialects, the
plosives p, b, t, d, k, g are articulated as fricatives in postvocalic
position (0£. § 8.10): e.g. Heb. diibiir "word" is pronounced
This phenomenon, consisting of the transition of plosives to frica-
tives, may be regarded as an instance of partial assimilation: i.e. the
plosive articulation of the consonant passes towards the continuant
pronunciation characteristic of vowel articulation. This spiranti-
zation may continue after the elision of the vowel which occasioned
it: e.g. Syr. *dahabii "gold" > dahbii, pronourfced Another
phenomenon of assimilation of consonant to vowel must be seen
in the palatalization of k in the neighbourhood of palatal vowels:
58 Phonology
for example, in Arabic dialects, die for dik "cock" (of. § 8.42).
This process is wide-spread also in some of the modern Ethiopian
languages. Some phenomena of spirantization in Akkadian require
further investigation.
9.6. d) Vowel to consonant.-The pharyngal and laryngal
consonants frequently occasion a change of other vowels to a
(cf. for details § 16.ll0): e.g. Ar. *yaftul}u "he opens" > yaftal}u;
Heb. *yislol} "he sends" > yislal}. The meticulous phonetic nota-
tion which characterizes Hebrew Masoretic pointing marks the
appearance of a-timbre vowels after the consonants' and h (and
more rarely l}) in conditions when they would otherwise be vowelless :
thus *ya'mod "he stands" > ya'amod; Such vowels are also indi-
cated between long vowels of other timbres and pharyngal or
laryngal consonants (e.g. *rul} "spirit" > rUiily,). This is the
so-called patal} furtivum which does not, incidentally, run counter
to the rules of Semitic syllabic structure (where two contiguous
vowels are impossible-except when a glottal stop intervenes
between them: cf. § 10.2), because it serves merely as a "catalyst"
in the articulatory process (a similar phenomenon may be observed
in Arabic, though its strictly phonemic vowel notation fails to
indicate it). Labial consonants are liable to cause other vowels to
change into u, generally in preceding rather than following position:
e.g. Sem. *libb "heart" > Ar. lubb.
9.7. e) Rising Diphthongs Assimilated or Reduced.-
Assimilated: e.g. Ar. *'aywam "days" > 'ayyam. Reduced by
assimilation: e.g. Akk. *yakSud "he conquered" > *yiksud > ikSud
(total progressive assimilation). In Hebrew a reduction might
possibly be inferred from Greek transcriptions: e.g. / a a a ~ for
Yi§lyiiq (and cf. later on Syr. '/sl}aq).
9.8. f) Faliing Diphthongs Assimilated or Reduced.-
Assimilated: e.g. Ar. *kawy "burning" > kayy. Reduced by
assimilation: e.g. Akk. *iwbil "he carried" > ubil (reciprocal
assimilation), *baytu "house" > bitu (total regressive assimilation).
Of. for this section the treatment of the diphthongs in §§ 8.97-104.
2. Dissimilation
9.9. The Semitic languages present phenomena of dissimilation
between consonants, between semivowels, between vowels, and
Conditioned Phonetic Changes 59
between semivowel and vowel; both progressive and regressive,
contiguous as well as at distance.
9.10. a) Between Oonsonants.-Progressive and contiguous:
e.g. Ar. lJarrub "carob-bean" and lJarnub (but dissimilation is
not the only way in which this variant may be accounted for, and
in any event this type of dissimilation is infrequent).-Progressive
and at distance: e.g. Ar. layl "night", lUn "to spend the night"
(Heb. lUn, lin, Ug. lyn).-Regressive and contiguous: e.g. Akk.
inaddin "he gives" and inandin (or inamdin); dissimilation by
means of n is extensive in Akkadian, and particularly in Babylonian,
in respect of d, b, and despite phonetic difficulties also z: e.g.
inazziq "he grieves" > inanziq.-Regressive and at distance: e.g.
Sem. *sams "sun" > Ar. *sams (n > sams, cf. Akk. samsu,
Ug. sps.
9.11. b) Between Semivowels.-This occurs particularly in
Arabic: e.g. *wawaqi "ounces" > 'awaqi (regressive and at dis-
tance).
9.12. c) Between Vowels.-E.g. Ar. *madiniy "Medinese" >
madaniy > madaniy (qualitative and quantitative). In Hebrew
and Syriac the succession of two vowels of u or 0 timbre occasions
the dissimilation of one of them to i or e: e.g. Heb. *l}u§on "ex-
ternal" > l}i§on (qualitative); Syr. 8'dlemon for Heb. 8'dlomo
"Solomon" (qualitative).
9.13. d) Between Semivowel and Vowel.-E.g. Ar. wuguh
"faces" > 'uguh (regressive and contiguous); Eth. *z'druw "sown" >
Z'dr'dW (regressive and contiguous), though in the Ethiopian example
other factors may be at work as well.
3. Prosthesis
9.14. As will be explained when dealing with syllabic structure
(cf. § 10.2), the Semitic languages do not permit the presence of
more than one consonant at the beginning of a word. To obviate
such initial consonant clusters a supplementary vowel (introduced
by') is generally prefixed to the first consonant to produce a new
syllable. In some cases, though more rarely, the new vowel is
instead placed after the first consonant (cf. §§ 9.16-17).
60
Phonology
9.15. In Hebrew and in Syriac the prosthetic vowel is e: e.g.
Heb. *zroa' "arm" > 'ezroa' (but also zaroa' [in this case the more
usual form], i.e. the alternative procedure, just mentioned); Syr.
*tqattal "he was killed" > 'etqattal. In Hebrew we have hi- in
the verbal theme Hithpael-possibly by analogy with the theme
Hiphil: e.g. Syr. 'etqattal, Heb. hitqatt(}l (so also in the imperative of
the Niphal). In Arabic the prosthetic vowel is i: e.g. *bn "son" >
'ibn; *nqatala "he was killed" > 'inqatala; more rarely it is u:
e.g. *qtul "kill!" > 'uqtul (vowel harmony n. In Ethiopic the
vowel is a: e.g. *mna "from" > 'amna, *gzi' "lord" > 'agzi'. The
process continues in some modern dialects and becomes operative
also in foreign borrowings such as the modern Eastern Aramaic
'us tal "table" from the Russian stol. Further examples in Ullendorff,
SLE, pp. 198-201.
4. Anaptyxis
9.16. A consonant cluster at the end of a word (which would
be contrary to the principles of Semitic syllabic structure, cf.
§ 10.2) is frequently resolved by the insertion of a secondary vowel
and the consequent creation of a new syllable. The same method
of resolving consonant clusters is employed (as we have seen-cf.
§§ 9.14-15) also at the beginning of a word.
9.17. In Akkadian the anaptyctic vowel is generally identical
with that of the principal syllable: e.g. in the construct state,
*uzn "ear" > uzun, *kalb "dog" > kalab (in Assyrian a occurs at
times after i or u: e.g. uzan). Similarly at the beginning of words:
*ksud "reach!" > kuSud, *I}bat "take!" > I}abat (but some verbs
use i instead of a: e.g. *lmad "learn!" > limad). In late Akkadian
the weakening of stress favours the rise of secondary vowels in the
middle of words: e.g. New Bab. sipiretu "letters" alongside sipretu.
In Hebrew the anaptyctic vowel is e which assimilates to itself the
vowels a, i (but not u) of the preceding syllable: this is the origin
of the "segolate" nouns, e.g. *'abd "slave" > 'ebed (but before
laryngals and pharyngals a remains and even harmonizes the
anaptyctic vowel), *sifr "book" > s(}fer, *'uzn "ear" > 'ozen
(the original forms reappear upon attachment of suffixes: e.g.
'aMi "my slave"). For the diphthongs aw, ay (when they are not
contracted) we have the development aw > awu > awe> awe
Conditioned Phonetic Changes 61
(e.g. *mawt "death" > mawet) and ay > ayi (e.g. *bayt "house" >
bayit); cf. also § 8.100. In Syriac, too, the anaptyctic vowel is e;
the preceding vowel tends to be reduced or dropped as its position
becomes pre-tonic: e.g. *'abd "slave" > *'abed > 'abed.
9.18. In Arabic the case-endings prevent the formation of con-
sonant clusters at the end of a word: 'abd
un
and 'abd
u
"slave",
rigl
un
and rigl
U
"foot". A special situation may, however, arise
as a consequence of the effects of sentence stress (cf. § 10.14).
As for Ethiopic, the ambiguity of the sixth vowel (which represents
both a and zero) does not allow us to arrive at safe conclusions; it
does appear, however, that consonantal clusters were avoided
either by the addition of final a (e.g. gabr "slave", pronounced
[gabrg]) or by the insertion of a between the consonants ([gabgr]):
cf. Ullendorff, SLE, pp. 201-207.
5. Syncope and Contraction
9.19. The syncope of vowels or consonants, when occasioned
by the succession of two of them, is basically a phenomenon
of dissimilation: e.g. Heb. qiqalon "shame", cf. Syr. qulqala. In
the example just quoted there occurs compensatory lengthening of
the vowel.
9.20. There is ample evidence, throughout the whole of the
Semitic area, of the syncope of " w, y and, more rarely, h in inter-
vocalic or paravocalic position. The syncope brings about vowel
contraction or compensatory lengthening. The concomitant action
of assimilation, dissimilation, analogy, as well as interchanges
between the "weak" consonants, makes it difficult to establish
generally valid rules of contraction or lengthening. In many
cases, moreover, the explanation of forms as resulting from syncope
or contraction is purely conventional. What may, in fact, have
happened is much rather the secondary constitution of "weak"
consonants or the lengthening of originally short vowels through
the adaptation, by analogy, of biliteral roots to the predominant
triliteral system (of. §§ 11.5-9, 16.108-27). For detailed inform-
ation the grammars of the various languages have to be consulted;
but it may be said here that as a general rule vowel lengthening is
62 Phonology
unaccompanied by changes in the quality of the vowel-unless it
is occasioned by assimilation to a semivowel. Of contraction it is
generally true that (a) the combination of t}Vo like vowels results
in the same vowel; (b) where one vowel is long and the other short
it is the timbre of the long vowel that tends to prevail (but in
some such cases contraction does not, in fact, take place: of. e.g.
the participles Akk. sa'imu "determining", Ar. qa'im "standing",
in verbs with medial w/y); (c) the quality of a stressed vowel tends
to prevail over that of an unstressed one; (d) two vowels markedly
distant in their basis of articulation may produce a vowel with
an intermediate point of articulation; (e) the vowel resulting from
contraction is generally a long one; (f) Old Akkadian and Arabic
agree that contraction does not generally take place when the second
of the two vowels is a, either short or long; in later Akkadian con-
traction usually takes place, with a prevailing. The following may
serve as examples of the principal changes: (a) postvocalic ': e.g.
Sem. *ra's "head" > Akk. resu, Reb. ros, Syr. resa, against Ar. ra's,
Eth. rii's; (b) intervocalic ': e.g. Sem. *baaa'a "he began" > Reb.
baaa, Syr. biida (the Reb. and Syr. verbs differ in meaning from
Arabic), against Ar. baaa'a; Sem. *tami'a "he thirsted" > Reb.
against Ar. ?ami'a; (c) intervocalic w; e.g. Sem. *gawir
"guest" > Reb. (but Ar. gar); Sem. *dalawa "he drew" > Reb.
dala, Syr. aiila, Ar. dalii, against Eth. dalawa; (d) intervocalic y:
e.g. Sem. *bakaya "he wept" > Reb. Mka, Syr. biika, Ar. bakii,
against Eth. bakaya; (e) h: e.g. Sem. *qatalahu "he killed him" >
Reb. qiitalo, Eth. qatalo, against Syr. qatleh, Ar. qatalahu. For
a detailed treatment of syncope and contraction in Arabic cf.
Fleisch, TPA, pp. 98-138.
6. Raplology
9.21. The omISSIOn of one of two contiguous syllables with
identical consonants (and sometimes vowels) is a phenomenon
of dissimilatory origin which occurs in various Semitic languages.
Certain combinations arise from Arabic morphology and may be
eliminated by haplology: e.g. tataqataluna "you fight" > taqata-
lUna; yaqtulunana "they kill us" > yaqtulUna (cf. Fleisch, TP A,
pp. 149-53). A few cases of haplology may also be observed in
other languages: e.g. Syr. *'aryaya "lion" > 'arya.
Syllable and Stress 63
7. Metathesis
9.22. Examples of metathesis are to be found in all the Semitic
languages: e.g. Akk. *dipsu "honey" > dispu; Reb. simla "coat"
and salma; Syr. *ta'ra "gate" > tar'a; Ar. 'atraba and 'artaba
"he was poor"; Eth. nsk and nks "to bite". Some metatheses can
only be detected by comparison with other languages: e.g. Akk.
simmiltu "ladder", Reb. sullam. Very wide-spread in the Semitic
languages is the metathesis of t (as part of the verbal theme, of.
§§ 16.17-23) with the first radical of the verb when this is a dental
or palato-alveolar fricative: e.g. Reb. *hitsammr;r "he was on his
guard" > histammr;r; Syr. *'etsiimek "he leaned" > 'estiimek; Ug.
*ttff:i,wy "she is prostrate" > ts{f:i,wy. Consequently, when the verbal
theme with prefix s is combined with that with prefix t, the two
consonants change places with each other (cf. § 16.21): Akk. 8utaqbur,
Ar. 'istaqbara, Eth. 'astaqbara (for the paradigm qbr cf. § 12.3). In
Akkadian the non-prefixed forms of the verbal themes with t and
tn and the adjectives of the pattern qitbar, in which the element t
would normally be infixed, show a metathesis in the opposite direc-
tion, i.e. the t becoming a prefix, when the first radical is z, s,
and sometimes s and also d: e.g. >
"he desires" > *dituku "combat" > tiduku.
8. Sandhi
9.23. A particular aspect of sentence-phonetics is that constituted
by the extension of certain phenomena beyond the limits of the
word itself, i.e. by their effect on the boundaries of neighbouring
words (syntactic phonetics or sandhi). Thus, in the case of assi-
milation, the fricative pronunciation in post-vocalic position
of the consonants p, b, t, d, k, g (§ 9.5) becomes operative also at the
beginning of a word when the preceding one ends in a vowel.
Phenomena of syncope and assimilation are widely attested in
Arabic-affecting contiguous words-by the tradition of Koranic
reading: thus in sura 24,44 lJalaqa kull
a
dabbat
in
"he created all the
animals" > [lJalakkulla dabbatin].
D. Syllable and Stress
1. Syllabic Constitution
10.1. There are two types of syllables in Semitic: a) consonant
followed by vowel (open syllable); b) consonant followed by vowel
64
Phonology
followed by consonant (closed syllable). Quantitatively, a syllable
may be: a) short, when it ends in a short vowel; b) long, when it
ends in a long vowel or in a consonant. For example: qa, open short
syllable; qa, open long syllable; qab, closed (and therefore long)
syllable. The term "ultra-long" is used of syllables (cf. § 10.3)
which are closed in addition to having a long vowel (e.g. qab).
For syllables in final position, ending in two consonants, see next §.
10.2. It follows from § 10.1 that in Semitic every syllable normally
begins with one consonant and one only. Two vowels cannot be in
contact. Two consonants may generally be contiguous only in the
middle of a word (final consonant of a closed syllable and initial
consonant of the following syllable). A sequence of two consonants
at the end of a word may result from the shedding of final
vowels. There prevails in the Semitic languages a widespread
tendency to eliminate exceptions to these rules, either by means
of prosthetic vowels (graphically supported by') or anaptyctic
ones (cf. §§ 9.14-17) or else through word juncture. As examples
of prostehtic vowels cf. Ar. *nkasara > 'inkasara "it was broken";
and in transcriptions of foreign words Syr. 'espera from (JqJaie
a
,
Eth. 'atrones from {}eovor;. For anaptyctic vowels cf. *'uzn "ear" >
Akk. uzun, Reb. 'ozen. Liaison of words occurs in Ar. *tumma
nkasara [tummankasaraJ "then it was broken". Among modern
Semitic languages, the Ethiopian tongues make fairly extensive
use of prosthetic and anaptyctic vowels (Ullendorff, SLE, pp. 199
to 200). In North-West Semitic, conclusions about syllabic con-
stitution depend to a large extent on one's judgment as to the nature
of the s'dwa (§ 8.82). The pronunciation of s'dwa as 'd in certain
conditions has been considered by some scholars as a secondary
phenomenon of an anaptyctic character (Gesenius-Bergstrasser,
Hebraische Grammatik, pp. 134-35); and in this connexion one
may compare fluctuating pronunciations of the type of Syr.
dely,elta for dely,l'dta "fear". If one maintains the primarily vocalic
character of s'dwa, one has yet to recognize that Masoretic pointing
acknowledges the succession of two consonants in initial position
(it does not register the second consonant as a fricative-as it
should have done in post-vocalic position) in stayim "two" (to
which sta corresponds in Syriac). It is possible, however, that the
infringement of the general rule is due to the workings of analogy.
Syllable and Stress 65
10.3. According to Brockelmann (GVG, I, p.63), Semitic
originally postulated short vowels in closed syllables. This rule is
mainly based on the position in Arabic, and its general application
over the Semitic field may be subject to some doubt. Nevertheless,
it is a fact that long vowels show a tendency to become short
when their syllable closes. This phenomenon is connected with
the incidence of stress and will, therefore, be dealt with in
that connexion (§§ 10.5-11). In Arabic the shortening of long
vowels in closed syllables is a rule (e.g. *qum > qum "rise !");
the only exceptions occur in certain syllables of secondary origin,
e.g. where the final vowel is dropped in pause (naziliina > naziliin
"descendants"). The shortening of long vowels in closed syllables
is characteristic also of Eastern Syriac (e.g. *'almin "eternity" >
'almin).
10.4. Fairly common is the tendency to lengthen short vowels
in open syllables. This trend is likewise connected with the opera-
tion of stress and will, therefore, be dealt with in that connexion
(§§ 10.5-11). But some cases occur also irrespective of stress:
e.g. Akk. *mil'u "fullness" > *milu > milu (compensatory length-
ening to restore the syllable rhythm of the word). Sometimes
consonant-doubling (gemination) takes the place of vowel-length-
ening and so restores the closed syllable with short vowel: e.g. Akk.
1J,itfu for 1J,itu "sin" (this occurs predominantly, though not ex-
clusively, in the later period of the language: cf. Old Bab. k u ~ ~ u m
alongside k u ~ u m "cold"); Reb. gamal "camel", pluT. g'dmallim.
There are also some cases of consonant-doubling without any parallel
vowel-lengthening, Etspecially in Aramaic: e.g. Syr. *qalil "little" >
qallil, *'atana "she-ass" > 'attana.
2. Stress and Associated Changes
W.o. We lack sufficient data to determine the position of stress
in Proto-Semitic or to distinguish clearly between expiratory stress
and pitch accent. To restrict ourselves to the more readily identi-
fiable expiratory stress, we may say, in the first place, that in
Proto-Semitic it is unlikely to have had distinctive or phonemic
status; and, secondly, that the almost complete agreement between
Arabic and Akkadian might facilitate a hypothetical reconstruct-
ion of Proto-Semitic stress modelled on these two languages.
Moscati, Comparative Grammar 5
64 Phonology
followed by consonant (closed syllable). Quantitatively, a syllable
may be: a) short, when it ends in a short vowel; b) long, when it
ends in a long vowel or in a consonant. For example: qa, open short
syllable; qa, open long syllable; qab, closed (and therefore long)
syllable. The term "ultra-long" is used of syllables (cf. § 10.3)
which are closed in addition to having a long vowel (e.g. qab).
For syllables in final position, ending in two consonants, see next §.
10.2. It follows from § 10.1 that in Semitic every syllable normally
begins with one consonant and one only. Two vowels cannot be in
contact. Two consonants may generally be contiguous only in the
middle of a word (final consonant of a closed syllable and initial
consonant of the following syllable). A sequence of two consonants
at the end of a word may result from the shedding of final
vowels. There prevails in the Semitic languages a widespread
tendency to eliminate exceptions to these rules, either by means
of prosthetic vowels (graphically supported by') or anaptyctic
ones (cf. §§ 9.14-17) or else through word juncture. As examples
of prostehtic vowels cf. Ar. *nkasara > 'inkasara "it was broken";
and in transcriptions of foreign words Syr. 'espera from IJ'cpaie
a
,
Eth. 'atrones from {}e6vor;. For anaptyctic vowels cf. *'uzn "ear" >
Akk. uzun, Reb. 'ozen. Liaison of words occurs in Ar. *iumma
nkasara [tummankasaraJ "then it was broken". Among modern
Semitic languages, the Ethiopian tongues make fairly extensive
use of prosthetic and anaptyctic vowels (Ullendorff, SLE, pp. 199
to 200). In North-West Semitic, conclusions about syllabic con-
stitution depend to a large extent on one's judgment as to the nature
of the s(Jwa (§ 8.82). The pronunciation of s(Jwa as (J in certain
conditions has been considered by some scholars as a secondary
phenomenon of an anaptyctic character (Gesenius-Bergstrasser,
Hebraische Grammatik, pp. 134-35); and in this connexion one
may compare fluctuating pronunciations of the type of Syr.
delwlta for del}l(Jta "fear". If one maintains the primarily vocalic
character of s(Jwa, one has yet to recognize that Masoretic pointing
acknowledges the succession of two consonants in initial position
(it does not register the second consonant as a fricative-as it
should have done in post-vocalic position) in stayim "two" (to
which sta corresponds in Syriac). It is possible, however, that the
infringement of the general rule is due to the workings of analogy.
Syllable and Stress 65
10.3. According to Brockelmann (GVG, I, p. 63), Semitic
originally postulated short vowels in closed syllables. This rule is
mainly based on the position in Arabic, and its general application
over the Semitic field may be subject to some doubt. Nevertheless,
it is a fact that long vowels show a tendency to become short
when their syllable closes. This phenomenon is connected with
the incidence of stress and will, therefore, be dealt with in
that connexion (§§ 10.5-11). In Arabic the shortening of long
vowels in closed syllables is a rule (e.g. *qum > qum "rise !");
the only exceptions occur in certain syllables of secondary origin,
~ . g . where the final vowel is dropped in pause (nazilUna > nazilUn
"descendants"). The shortening of long vowels in closed syllables
is characteristic also of Eastern Syriac (e.g. *'almin "eternity" >
'almin).
10.4. Fairly common is the tendency to lengthen short vowels
in open syllables. This trend is likewise connected with the opera-
tion of stress and will, therefore, be dealt with in that connexion
(§§ 10.5-Il). But some cases occur also irrespective of stress:
e.g. Akk. *mil'u "fullness" > *milu > milu (compensatory length-
ening to restore the syllable rhythm of the word). Sometimes
consonant-doubling (gemination) takes the place of vowel-length-
ening and so restores the closed syllable with short vowel: e.g. Akk.
1J,iHu for 1J,itu "sin" (this occurs predominantly, though not ex-
clusively, in the later period of the language: cf. Old Bab. k u ~ ~ u m
alongside k u ~ u m "cold"); Reb. gamal "camel", plur. g(Jmallim.
There are also some cases of consonant-doubling without any parallel
vowel-lengthening, especially in Aramaic: e.g. Syr. *qalil "little" >
qallil, *'atana "she-ass" > 'attana.
2. Stress and Associated Changes
10.5. We lack sufficient data to determine the position of stress
in Proto-Semitic or to distinguish clearly between expiratory stress
and pitch accent. To restrict ourselves to the more readily identi-
fiable expiratory stress, we may say, in the first place, that in
Proto-Semitic it is unlikely to have had distinctive or phonemic
status; and, secondly, that the almost complete agreement between
Arabic and Akkadian might facilitate a hypothetical reconstruct-
ion of Proto-Semitic stress modelled on these two languages.
J\[oscati, Comparative Grammar 5
66 Phonology
The risks inherent in this procedure need hardly be underlined
when we recall that the situation in Arabic, and particularly in
Akkadian, is subject to much uncertainty (cf. §§ 10.6-7).
10.S. In Akkadian, so far as our limited evidence permits.. a re-
construction, the position of the stress may be expressed as follows:
a) if the final syllable is the result of contraction it generally bears
the stress; b) otherwise stress does not fall on the ultima, even if
it is long, but recedes as far as possible until it meets a long syllable
(if there is no long syllable stress comes to rest on the first syllable
of the word). Only in rare instances (cf. von Soden, GAG, p. 38)
does stress fall on a short syllable in the middle of a word. Ex-
amples: a) accent on ultima: san1i ( < * saniyu) "second"; b) accent
on long syllable: bel1itu "lordship" (long vowel), napistu "life"
(closed syllable); on antepenult (short): kUbburu "stout". If
stress falls on a short syllable it may cause it to be lengthened,
either by the lengthening of the vowel, e.g. imqutU > imq1itu
"they fell", or by the doubling of the following consonant, so as to
form a closed syllable, e.g. iskUnu > iskunnu "they put". Strong
expiratory stress may occasion reduction in neighbouring vowels,
e.g. *wdsibat > wdsbat "she dwells": cf. §§ 10.3-4. Secondary
stress patterns arise in compound words of some length, e.g. with
pronominal suffixes: so "they let him enter". In any
case, stress is non-phonemic in Akkadian.
10.7. For classical Arabic the rule given in the preceding para-
graph is of universal application, i.e. stress does not fall on a final
syllable (even if it be the result of contraction) but goes back as
far as possible till it meets a long syllable or, failing that, the
initial syllable. Examples: qatdltum "you killed"; qdtalU "they
killed"; mdmlakat "kingdom". We do not, however, know any
express Arab tradition of acceptable antiquity which might eluci-
date for us the origin of the stress rules now observed in reading
classical Arabic. According to some recent studies (Birkeland) it
would appear that these rules might derive from the stress patterns
of certain Arabic dialects. In these dialects considerable develop-
ments have taken place which-with regard to this particular
feature-have brought about some affinity to the situation which
prevails in other Semitic languages (cf. §§ 10.8-10), in notable
Syllable and Stress
67
contrast to the pattern in classical Arabic. As stress in Arabic is
bound, it can obviously not be phonemic.
10.S. In Hebrew (at least as far as can be judged from the Maso-
retic tradition) stress falls on the last syllable-save for some cases
of penultimate patterns. In contrast to Akkadian and Arabic,
stress in Hebrew may have distinctive or phonemic value: e.g.
sabu "they returned", but siib1i "they took prisoner". Stress patterns
and syllabic constitution are bound up with complex rules of vowel
evolution which (leaving out of account the difficult question of
their origin) may be summarized as follows:
a) final short vowels are dropped (*qdbara > *qdbar);
b) stress shifts to the last syllable which the development under
(a) has left closed and therefore long (*qdbar > *qabdr);
c) short accented vowels undergo lengthening or change of timbre,
or both, either under the influence of the word-accent or by context-
ual stress patterns (pause) (0£. §10.13):a>ii, u>o
(*ddbaru > diibdr; *qabiru > *ydqburu > yiqb6r); before
two successive consonants, however, i > a instead of i >
(*ziiqinta > ziiqdnta);
d) in contrast to the general Semitic tendency, and probably
by a relatively late process of restoration, open pre-tonic syllables
undergo lengthening and sometimes change of vowel quality:
a > ii, i > ? (or else J according to the development referred to
under g); u remains, but the following consonant is doubled
(cf. § 1004): e.g. *ddbaru > diibdr, *'inabu > '?nab (but *lJimaru >
*lJJmBr > lJamBr), *luqdlJ > luqqdlJ;
e) short vowels in closed unstressed syllables may undergo
change of quality: a > i, i > e, u > 0 (*madbdr > midbdr [dis-
similation 'imriitB and 'emriitB; *'u!!ni > 'oznt);
f) in final open stressed syllables i becomes e (Ar. tamiini, Heb.
sJmi5ni) ;
g) short vowels in open unstressed syllables are reduced to J in
accordance with the general Semitic tendency and in contrast to
the instances listed under (d) where pre-tonic syllables frequently
undergo lengthening; it is likely that these two opposed tendencies
were operative at different periods: e.g. *dabartm > dJbiirtm;
*qiibar1i > qiibJr1i.
5*
68 Phonology
As for the pre-Masoretic stress-accent, this must have diverged
notably from its later Masoretic version (as has been shown by
Bmllllo from Greek transcriptions): cases such as cpifh{}a for
pittal}ta, avvw{}Ev for CW'rJp,8eOV for zamm'iwii, r56.f3e'rJ for
etc., testify to stress conditions contrasting with our notions
derived from the Masoretic recension of the Hebrew text.
10.9. Compared with the ample documentation of Masoretic
Hebrew, the evidence concerning North-West Semitic of the
second millennium B.C. and the rest of Canaanite is exceedingly
scanty:
a) there are indications of a reduction of short vowels in unstressed
open syllables (cf. § 10.8 g) in Amorite (e.g. A-ma-na-nu-um
and Am-na-nu-um, ya-ta-ra-tum and ya-at-ra-tum), in the Tell
Amarna glosses (e.g. mi!Wu for ma!Ji§u, §ilJru for §alJiru) , and
possibly in Ugaritic (Garbini, SNO, pp. 75-77);
b) in Phoenician we have some lengthenings of short stressed
vowels, accompanied by changes of vowel quality (a > 0, i > e,
u > 0), which reveal remarkable similarities to the Hebrew
changes (cf. § 10.8 c).
10.10. In the Aramaic area, while Biblical Aramaic reflects the
situation in Masoretic Hebrew, Syriac stress always falls on the
final syllable. As in Hebrew (indeed, the Masoretes worked under
the impact of Aramaic) there are complex rules of vowel develop-
ment, connected with the incidence of stress and with syllabic
constitution, which may be summarized as follows:
a) final vowels, whether long or short, are dropped (*qabara >
*qabar; *qabaru > *qabar [the final u is written but not pronounced]) ;
b) stress passes to the final syllable which is now closed and
hence long (*qabar > *qabdr);
c) short vowels in open unstressed syllables are reduced to 9 or
dropped (*qabdr > q9bdr);
d) in closed syllables short a and i may become e (*qabrat >
qebrat; *si/rd > se/rd);
e) a short stressed u becomes 0, whether by the action of the
word-accent (as for the change u > 0, of. also the opposition
between West and East Syriac, § 8.88) or by analogy with pro-
nominal forms and verbal suffixes (*qabdrtumu > *qabdrtum >
*qabartum > *qabartBn > q'JbartBn) ;
Syllable and Stress 69
f) i becomes e in final open and stressed syllables (Ar. tamani,
Syr.
After the close of the classical period (about A.D. 700) final
open syllables tend to lose their stress: e.g. nehwe "he is" in Maronite
usage; the Nestorians stress the penult even in cases when the final
syllable is closed: e.g. ketbat "she wrote".
10.11. In Ethiopic it is usually assumed that stress falls on the
final syllable of a noun but the penult of a verb. Recent research
by Ullendorff (SLE, pp. 189-97) shows, however, that the whole
question remains complicated and more than a little uncertain.
The very existence of expiratory stress in Ge'ez is doubtful, and in
the traditional pronunciation it is difficult to distinguish between
stress and pitch. In any event, the accent in Ethiopic (whatever
its precise nature may have been) is non-phonemic.
3. Sentence Stress
10.12. In addition to word-accent, the Semitic languages have
a sentence stress determined (especially in pause) by the traditional
recitation of the text. This stress occasions a number of changes
in some languages.
10.13. In Hebrew the principal alterations are as follows:
a) stress is thrown back on to the penult (e.g. 'anoki > 'ani5ki);
b) a short accented vowel is lengthened (e.g. mayim > mdyim)
sometimes causing change of quality (e.g. 'ere§ > 'dre§).
10.14. In Arabic the principal changes are as follows (cf. Fleisch,
TPA, pp. 172-90):
a) final short vowels are dropped (e.g. qatala > qatal); this may
in some cases affect the constitution of a final consonant group
(of. § 9.18) with consequent anaptyxis (e.g. al-bakru "young
[camel]", in pause al-bakur).
b) the indefinite case-endings -un, -in are dropped, and -an
becomes -a (e.g. malik
un
> malik, malik
in
> malik, malik an>
malika) ;
c) the feminine noun ending -at becomes -ah (e.g. malikat
un
>
malikat > malikah); for a possible Hebrew and Syriac parallel
cf. §. 12.33.
70 Phonology
10.15. For the other languages we possess no adequate indications
about sentence stress, or, at any rate, no changes occur of the
type we have witnessed in Hebrew and Arabic. There are, however,
one or two hints: thus in Akkadian the word on which sentence
stress falls, in interrogative sentences, shows a shift of the stress
on to the penult or ultima with consequent secondary vowel length-
ening: e.g. ippuSa or ippasu "will they do?" instead of ippusu.
III. Morphology
A. Preliminaries
1. Morphemes
11.1. The Semitic languages present a system of consonantal
roots (mostly triconsonantal), each of which is associated with
a basic meaning range common to all members of that root:
e.g. ktb "to write", qbr "to bury", qrb "to approach", etc. These
roots (root morphemes) constitute a fundamental category of
lexical morphemes (cf. Petracek, ArOr 28 [1960], pp.564-68).
The linguistic reality of consonantal roots is shown not only by
their lexical implications but also by the laws governing the
compatibility or otherwise of radicals (which do not concern the
vowels: cf. § 11.10) and in the transcription of foreign words.
Only the pronouns and some particles lie outside this system of
roots.
11.2. The task of lexical individualization (lexical morphemes)
and grammatical categorization (grammatioal morphemes) is
assumed by vowels and by affixes (prefixes, infixes, suffixes): e.g.
in Arabic, from the root ktb "to write" : kitiib "book", kiitib "writer",
maktabat "library", kataba "he wrote", yaktubu "he writes", etc.
The linguistic reality of vocalization and affixes, in their morphemic
function, is clearly attested by their specific semantic implications.
11.3. Grammatical morphemes may be external, internal, or
syntactical. External morphemes are elements attached to a root
(affixes, cf. the preceding paragraph). Internal morphemes are
constituted by the nature or disposition of certain phonetic elements
(consonants, vowels, stress), and in the Semitic languages they
appear especially in the "inner" (or "broken") plurals and in the
passive conjugation of the verb: e.g. Ar. kitiib "book", pI. kutub
"books"; qatala "he killed", qutila "he was killed". Petracek's
important studies (ArOr 28 [1960], pp.547-606; 29 [1961],
pp. 513-545 and to be continued) show that inner inflection is
72 Morphology
particularly developed in South Semitic, though it is not without
precedent in Hamito-Semitic generally. The importance of vowel-
alternation (apophony) in Semitic morphology has been stressed
by Kurylowicz. Syntactical morphemes are constituted by the
order of words or by independent elements; the latter are of rela-
tively low frequency in the Semitic languages (for example in the
Arabic formation of the future tense by means of the particle
sawfa).
11.4. The above account is concerned with consonantal radicals
only, and it has long been usual to conceive of Semitic roots as
purely consonantal; such a reconstruction is unreservedly main-
tained by some scholars (so Fleisch, TPA, pp.247-51) but is
disputed by others. Von Soden, in particular, (GAG, pp. 51-52,
96-97) holds that vowel elements should be regarded as forming
part of the root (see also E. Ullendorff in Or 28 [1958], pp. 69-70).
Such elements, which can be identified in the imperative of verbs
and in the prefix conjugation, are short in triconsonantal roots
(e.g. pqid "to guard, to deposit") but predominantly long in bicon-
sonantal ones (e.g. dilk "to kill", bni "to build"). It should be
observed that these latter roots exhibit stable vowels in other parts
of the Hamito-Semitic area as well (§ 11.6 c), so that the incorpora-
tion of original semivowel radicals (dwk, bny) may be ascribed to
artificial reconstruction. A similar stability is seen in the vowel
elements of nominal roots (e.g. Sem. kalb "dog") which have to be
differentiated from the verbal ones which, in their turn, are to be
divided into those indicating states or conditions and those con-
noting actions (cf. § 16.2). The distinction between the three
semantic spheres of noun, adjective and stative verb, and active
verb is reflected in a differentiation in the structure of the root.
2. The Proto-Semitic Root
11.5. In the historically attested Semitic languages triconsonantal
roots form the great majority; roots with two or with four radicals
are much less numerous, while those with one or with five are rare
(in roots with more than three consonants there is a possibility
of secondary formations by metaplasm, dissimilation, etc.). Ex-
amination of· the dictionary reveals the following phenomenon:
there are many groups of roots having two radicals in common
Preliminaries 73
which express identical or similar meanings. Thus for example in
Hebrew: prd "to separate", prm "to tear", prs "to split", p r ~ "to
break down", prq "to pull apart", prr "to dissolve", prs "to dis-
tinguish" etc. All these verbs have in common the radicals pr
and the basic notion "to divide". This phenomenon, which is
widespread in the Semitic lexicon, raises the question whether many
triconsonantal roots are not, in fact, derived from biconsonantal
ones; and whether a system of biconsonantal roots may, perhaps,
have preceded the triconsonantal theme in Semitic.
11.6. For a solution of this problem the following data must be
borne in mind:
a) The Semitic languages have many biconsonantal nouns (in
addition to some mono consonantal ones) which, from the objects
they denote, must be adjudged fairly ancient: e.g. dam "blood",
yad "hand", yam "sea", etc. The assignation of these nouns to
triconsonantal roots must be ruled out as contrived and far-
fetched.
b) The so-called "weak" verbs exhibit many biradical forms:
e.g. Heb. qam "he rose" (root qwm), 'f,-seb "I dwell" (root ysb) ,
Ar. ram-at "she threw" (root rmy), etc. It is our grammatical
systematization which looks upon these forms as having "dropped"
a radical, while one might maintain with as much reason that the
"weak" radical-in those forms which contain it-was, in fact,
added to the root for the sake of adaptation to the triconsonantal
system. This consideration seems particularly cogent where the
roots in question coincide semantically with others on a bicon-
sonantal basis.
c) Comparison with other languages of the Hamito-Semitic
group strengthens the biconsonantal hypothesis: e.g. Sem. qil "to
kill", Cushitic qal; Sem. p'l (I'l) "to make", Cushitic fal; (it should
be noted-as indeed appears from these examples-that Cushitic
possesses biconsonantal roots with stable vowel).
11.7. The data just set forth show that biconsonantal roots in
the Semitic languages are not a hypothesis relating to a pre-
historic period but constitute an historical reality attested by
a group of nouns and by a series of verbal forms; this is further
supported by the semantic concurrence of many roots in two of
their radicals. There is, however, no sufficient reason for main-
74 Morphology
taining, as some have done, that the entire Semitic stock of roots
was originally biconsonantal. It is a more likely supposition that
originally there existed roots with either two or three consonants
(as well as a smaller number with one only or with more than three)
and that at a certain stage in the development of the Semitic
languages the triconsonantal system prevailed-extending by
analogy and thus bringing into line biconsonantal roots through
the adoption of a third radical. Cf. also Fleisch, TPA, pp. 247-6l.
11.S. As regards this third radical, or "determinant", the follow-
ing questions arise: which consonants can be so used and with what
specific semantic value? A lexical probe leads to the following
tentative conclusions:
a) all consonants may be used as "determinants";
b) apart from some grammatical formatives which retain some
trace of their original function (e.g. s- causative), it may be said
that the present state of the dictionary does not appear to permit
the identification of specific semantic values attached to these
"determinants" .
11.9. In the examination of biconsonantal roots it is to be borne
in mind that the radicals may have undergone certain phonetic
changes: thus alongside the series pr "to divide" (§ 11.5) Hebrew
also possesses the groups pl and br, i.e. by interchange between
consonants with the same (or a similar) point of articulation (plly,
"to furrow", brr "to separate", sbr "to break", etc.).
11.10. Leaving aside biconsonantal roots and their development,
the Semitic languages reveal certain structural incompatibilities
which reduce the number of possible combinations in triconsonantal
roots. In no Semitic language can two identical consonants-or
two consonants with a similar point of articulation-appear next
to each other in first and second position; and it is rare for such
consonants to be found as first and third radicals (e.g. Akk. 7;asa7;u
"to desire"). In positions two and three, identical consonants are
frequently found but not different consonants with a similar basis
of articulation. There exist other incompatibilities in individual
languages or groups: thus in Akkadian g and z are never found in
third position, nor can all three radicals be voiced, and of two
emphatic consonants one is reduced to non-emphatic status (e.g.
The Noun 75
Akk. qatnu "thin", Sem. root q(n); in West Semitic, both Northern
and Southern, there are many incompatibilities between dental
and velar plosives: in Hebrew gt, tg, kt, tk, qt, tq are normally in-
compatible-but not tq, qt; in Arabic (taking into account the
transition g > g) tg, gt, fg, gt, tq, tq, tk, tk, kt are generally incom-
patible. The foregoing considerations apply particularly to verbal
morphemes. In nominal morphemes the position varies to some
extent: e.g. Sem. sams "sun", nun "fish", layl "night", surS
"root", etc.
3. Morphological Development
11.11. Morphology manifests the action of two fundamental
forces in the development of forms:
a) phonetic laws (e.g. assimilation, dissimilation, etc.) which
have already been considered in the section on phonology (cf.
§§ 9.1-22);
b) analogy, both morphological and lexical, sometimes in opposi-
tion to phonetic laws; an example of morphological analogy (cf.
the evidence adduced in §§ 16.44-45) is offered by the Proto-
Semitic endings of the first and second persons singular in the verbal
suffix-conjugation (*-ku, *-ta, *-ti) and their development in
Arabic (-tu, -ta, -ti) or Ethiopic (-ku, -ka, -ki), with analogical
extension of the elements t and k, respectively; as an example
of lexical analogy one might mention Heb. ly,amissa "five" instead
of *ly,amsa, by analogy with siMa "six" (cf. § 14.2).
Further aspects of morphological development are shown in such
contrasting forms as (in the case of verbs primae ') Akk. i"abit
< "was destroyed" alongside innabit "he fled"-despite identical
phonetic origin. In later phases of the language the possibility of
the normative influence of grammarians on morphological as well
as phonological development cannot be discounted.
B. The Noun
1. Themes or Patterns
12.1. In the Semitic dictionary the system of roots is combined
with that of themes or patterns. These are morphological types
which are frequently associated with specific meanings or uses.
For example: Ar. 'abyarf, "white", 'aly,mar "red", 'azraq "blue" are
76 Morphology
formed from the roots byg" ly,mr, zrq, respectively, and the pattern
'aqbar (cf. § 12.3) to render the names of colours.
12.2. The system of nominal themes weakens in the course of
morphological development, particularly in those languages in
which stress and syllabic constitution affect the vocalic structure.
Hebrew is a typical example of such a language: e.g. dabar "word",
construct state d"bar, construct before "light" suffixes d"bar,
construct plural dibr( -e, plural ending): four "themes" which are
conditioned allomorphs or combinatory morphological variants.
2. Nominal Patterns
12.3. Nominal patterns may be "simple" (when the root is modi-
fied by vowels only) or "extended", i.e. when affixes are added.
From a semantic point of view (partly also owing to the insufficiently
developed state of these studies) we are only occasionally able to
assign specific values and uses to individual patterns. In the follow-
ing, some of the principal nominal themes will be presented together
with their main spheres of employment-where these can be reason-
ably well established. For these purposes the paradigm used for
the identification of patterns will be qbr ("to bury"): this is, of
course, merely an arbitrary choice and may therefore serve even
when the resultant forms are not in fact attested. The root qbr is
found throughout the Semitic area and is one of very few verbal
roots which combine this advantage with a suitable phonetic
constitution. It is greatly preferable to the usual paradigm qtl
"to kill" which is not so far attested in Ugaritic, while in Hebrew
and Syriac it is subject to the assimilation t > t (cf. § 9.3); but,
above all, qtl contains the obnoxious t which may so easily be
confused with its other functions (infixes etc.) in the morpho-
logical scheme. For these reasons qbr will be employed for para-
digmatic purposes throughout this book. The fact that in Akkadian
the verbal inflexion of qbr shows the vowel i as against u in the
other languages carries little or no weight, as we are merely con-
cerned with conventional patterns. For further details the gram-
mars of the various languages concerned have to be consulted;
for phonological· developments which have a bearing on certain
nominal patterns see the section on phonology.
The Noun 77
12.4. Noun-patterns can be studied to best advantage in Akkadian
and Arabic, while some of the other languages (apart from those
whose vocalization is not attested) exhibit a lesser number of inde-
pendent themes but a great variety of formal developments result-
ing from the incidence of stress and syllabic constitution. From an
examination of Akkadian and Arabic patterns certain preliminary
conclusions may be drawn: a) not all patterns are necessarily
differentiated in origin: some of them can be explained as secondary
developments brought about by phonetic processes and analogy-
not infrequently corroborated by their meaning (on the other hand,
the same forces have also caused a measure of uniformity and
. levelling as well as reductions in the number of patterns); b) nominal
themes with rising rhythm (with final long syllable) appear to be
predominating (cf. the winer diffusion of qabiir as compared with
qabar); c) the distinctions between noun and adjective, concrete
and abstract, are not always apparent from a purely thematic
point of view: while in some cases differentiation and opposition
may be recognized (particularly in Akkadian), by and large these
patterns occur indiscriminately for the various categories just
mentioned (cf. Fleisch, TPA, pp. 349-76).
a. Simple Patterns
12.5. a) Monosyllables with short vowel: qabr, qibr, qubr. Owing
to the requirements of Semitic syllabic constitution these patterns
are liable to anaptyxis: e.g. Ar. 'abd, Akk. abdu, but Heb. 'ebed
(the "segolate" nouns: cf. § 9.17).
12.6. b) Disyllables with short vowels: qabar, qabir, qabur, qibar,
qibir, qubar, qubur. These themes may be variants of the preceding
ones occasioned by the influence of stress or by anaptyxis or by
the extension of pausal forms: an example of the elision of a vowel
under the impact of stress is Akk. *rapasu > rapsu "wide", fem.
rapastu (in Akkadian the first three patterns are generally employed
as adjectives).
12.7. c) Disyllables with long vowel or diphthong in the first
syllable: qabar, qabir, qabur, qaybar, qaybiir, qaybur, qawbar.
qawbiir. Of these patterns qabir usually has the function of an
active participle and is widespread throughout the Semitic lan-
78 Morphology
guages (cf. § 16.68): Akk. maliku "counsellor", Ar. katib "writer",
Reb. kot?b, Syr. kateb, Eth. waras "heir". The other patterns are
more common in Arabic than elsewhere; qabur serves in Syriac
as nomen agentis (e.g. paroqa "saviour"; for the alternation u: 0
see § 8.88).
12.8. d) Disyllables with long vowel or diphthong in the second
syllable; here the long vowel may be replaced by the feminine
ending -at (whose addition places these patterns into the category
of extended themes): qabar and qabarat, qabir and qabirat, qabur,
qibar and qibarat, qubar and qubarat, qubayr, qubur. Of these
patterns qabir and qabUr are predominantly adjectival (e.g. Ar.
kabir "great", Reb. ?a'ir "small", Eth. marir "bitter"; Ar. taruq
"timid", Reb. 'a?um "strong", Akk. batulu "yo,ung"); qabir, in
particular, is used in Syriac as a passive participle (*qabir >
qabir, with reduction of the pretonic vowel-of. § 10.10 c), and
qabur in Rebrew (*qabur > qabur, with lengthening of the pre-
tonic vowel-of. § 10.8 d) as well as similar uses of both forms to
be sporadically encountered in other languages (e.g. Ar. qatil
"killed", rasul "messenger"-cf. § 16.69). The pattern qubayr is
largely used as a diminutive: it occurs principally in Arabic (e.g.
'abd "servant", 'ubayd "little servant" -of. Fleisch, TPA, pp. 380
to 389), though traces of it are found in other Semitic languages
(e.g. Syr. *'ulayma "lad" > 'alayma; in Akkadian in nouns with
diminutive or pejorative connotation: e.g. *kusaypu > kusipu
"morsel of bread"). The pattern qibar is employed in some lan-
guages for tools or instruments: e.g. Ar. nitaq, Reb. '?zor, Eth.
qanat (all three meaning "belt").
b. Patterns Extended by Gemination or Reduplication of Radicals
12.9. a) Patterns with doubled second radical: qabbar, qabbar and
qabbarat, qabbir, qabbir, qabbur, qibbar, qibbar, qibbir, qibbir,
qibbur, qubbar, qubbar, qubbur, qubbur. Of these patterns, qabbar
is widespread throughout the Semitic languages and characterizes
intensives and names of professions (e.g. Ar. 'allam "man of great
learning", gammal "cameleer"; Akk. dayyanu "judge"; Reb.
?ayyad "huntsman"; Eth. gabbar "worker"). Other patterns with
long second vowel are chiefly employed to indicate adjectives
with intensive meaning (e.g. Ar. farruq "very timid", quddus
The Noun 79
"very holy", ?iddiq "very sincere", kurram "very generous"). Of
the patterns with short second vowel, qabbar and qabbir are used
in Akkadian for adjectives with iterative or intensive significance
(e.g. kabbaru "very thick"), and similarly qubbur, the adjective
of the verb-stem with doubled second radical and intensive meaning
(e.g. dummuqu "very good").
12.10. b) Patterns with reduplication of the second radical:
qababar, qubabir and others are attested in Akkadian (e.g. zuqaqipu
"scorpion") and the modern Ethiopian languages (Amh. tallaq
and talallaq "great").
12.11. c) Patterns with doubled third radical: qabarr, qaburr,
qibarr, qibirr, qubarr, quburr are fairly rare and occur chiefly in
Akkadian and Arabic. In Akkadian they produce both nouns
(e.g. arammu "dyke", kirissu "needle") and adjectives with inten-
sive meaning (e.g. namurru "shining", da'ummu "pitch-dark");
in Arabic they are employed for adjectives (e.g. nia,ibb "timid",
qumudd "strong"). A development peculiar to Akkadian is the
pattern quburra' used for regular actions (e.g. mua,urra'u "recep-
tion") and for certain situations or conditions (e.g. uturra'u "super-
fluity"); it belongs, formally, to the class of patterns extended by
affixes.
12.12. d) Patterns with repeated third radical: qabrar, qabrar,
qabrir, qabrur, qabrur, qibrar, qibrar, qibrir, qibrir, qubrar, qubrar,
qubrur, qubrur. These are infrequent types-attested chiefly as
adjectives (e.g. Reb. *ra'nan > ra'anan "green", 'umlal "languish-
ing"; Ar. ri'did "cowardly"), sometimes with diminutive or
pejorative significance (e.g. Ar. qu'dud "ignoble" from the root
q'd "to sit"). A few nouns of this type occur in Akkadian (e.g.
namriru "splendour", kulbabu "ant") and in Syriac (e.g. zahrira
"ray", partUta "piece"), again on occasion with diminutive or
pejorative meaning.
12.13. c) Patterns with reduplicated second and third radicals:
qabarbar, qabarbar, qabarbir, qabarbur, qabirbir. These are fairly
infrequent and mainly attested in West Semitic as adjectives
(e.g. Reb. *yaraqraq > yaraqraq "greenish", Ar. 'arakrak "thick",
Eth. ly,amalmil "greenish"), but there also exist a few nouns (e.g.
Reb. 'asatsut "common people", Syr. paraly,ruly,ta "spark").
80 Morphology
c. Patterns Extended by Prefixes
12.14. a) Patterns with vowel prefixes (introduced by'): 'aqbar,
'iqbar, 'iqbir, 'uqbur. The most frequent of these is 'aqbar (of.
§ 12.68 for Wehr's study, and see also Fleisch, TPA, pp. 408-17)
which serves in Arabic to characterize elatives and colours (e.g.
'al}san "most beautiful" from l}asan «beautiful"; 'al}mar "red",
'abya¢ "white", 'aswad "black", etc.) and in Hebrew for certain
other types of adjectives (e.g. 'akzar "cruel", 'akzab "mendacious").
In the other languages this pattern is rare (some animal names
occur in Ugaritic: e.g. dn7;r "dolphin"). As for the other types,
they are fairly rare in Arabic and are probably variants of 'aqbar;
in Akkadian 'iqbir is found as a variant of qibr.
12.15. b} Patterns with prefix y-: yaqbar, yaqbiir, yaqbir, yaqbur,
etc. These are rare and confined to West Semitic where yaqbur in
particular is used for names of animals (e.g. Ar. Heb. yal}mur,
a kind of antelope) and, infrequently, of plants (e.g. Ar. yabrul},
Syr. yabrul}a "mandrake"} as well as for adjectives (e.g. Ar.
yal}mum "black").
12.16. c} Patterns with prefix m-: maqbar, maqbiir, maqbir,
maqbir, maqbur, maqbur, miqbar, miqbar, miqbir, miqbir, muqbar,
muqbiir (for prefix n- occasioned by dissimilation of m- before
a labial see § 12.19; for prefix m- and suffix -n [maqbaran] cf.
§12.21). Four principal meaning-variants are connected with the
prefix m-: local, temporal, instrumental, abstract. In the ex-
pression of these meanings the various patterns appear to be used
indiscriminately, yet in individual languages a measure of differen-
tiation can be observed: In Akkadian maqbar(t} and maqbar are
employed for nouns of place and time (e.g. maskanu "place"),
maqbartalso for nouns of instrument, and muqbar(t) and muqbiir
for nouns of time (e.g. m u ~ l a l u "midday"). Arabic indicates nouns
of place by maqbar and maqbir (e.g. maw¢i' "place" from the root
w¢' "to put") and nouns of instrument by miqbar and miqbiir
(e.g. miftal} "key" from the root ftl} "to open"). Hebrew uses for
abstracts maqbar and miqbar (e.g. mamliikii "kingdom" from the
root mlk "to reign", mispa( "judgement" from the root spt "to
judge"), while maqbir and miqbir often designate instruments
(e.g. maptrill} "key"). In Ethiopic m'Jqbar « *miqbiir) predomi-
The Noun 81
nates for nouns of place (e.g. m'Jsraq "east"), while maqbar(t) and
1naqb'Jr(t) refer mainly to instruments (e.g. malbas "dress"). The
theme maqbur expresses the passive participle of the simple verb-
stem in Arabic; for the participles of the derived stems-which are
formed with the prefix m- in most of the Semitic languages-cf.
§§ 16.96-101. Outside the Semitic area, patterns with prefix
m- are attested in Egyptian (e.g. m.sdm.t "cosmetics" from the
root sdm). For a detfl..iled study of patterns with m- prefix of.
Fleisch, TPA, pp. 422-34.
12.17. d) Patterns with prefix t-: taqbar, taqbiir and taqbarat,
taqbir, taqbir and taqbirat, taqbur, taqbur and taqburat, tiqbiir.
These themes produce, for the most part, verbal nouns (e.g. Ar.
tardad "repeating", Akk. tallaktu "going"), generally of the verbal
stem with doubled second radical (e.g. Ar. tibyan "explaining",
Eth. t a f ~ a m "completing"). In particular, taqbir is used in various
languages as the verbal noun of the verbal stem with doubled second
radical (e.g. Akk. tamsilu "image", Syr. ta'dira "help", Ar. tafriq
"distribution"); taqbur, taqbur and taqburat occur as verbal nouns
of the simple stem (e.g. Akk. tapsu7;tu "repose", Heb. tagmul
"favour, recompense", Syr. taktusa "battle", Ar. tahlukat "per-
dition"); taqbiir and taqbur(a)t appear in Akkadian as verbal nouns
of the simple stem with infixed t (e.g. tam7;aru "encounter",
taqrubtu "approach").
12.18. e) Patterns with prefix s-: saqbar, saqbur and saqburat,
suqbur and suqburat, etc. These themes are used in Akkadian for
verbal nouns of the stem with prefix s- and causative value (e.g.
sa7;luqtu "ruin" from the root 7;lq "to perish") and also, though less
frequently, as adjectives with intensive meaning (e.g. sanudu
"very famous", surbu "huge"). To this group also belong the nomi-
nal forms of the verbal stem with sand t, for which of. § 16.21.
Outside Akkadian there are only a few traces in North -West
Semitic: e.g. Heb. salhebet, Jewish-Aram. salhObita "flame", and
perhaps Ug. s'tqt "she who causes to pass".
12.19. Patterns with prefix n-: naqbar, naqbur, etc. These themes
are attested in Akkadian, generally as variants (by dissimilation)
of the theme with prefix m- (e.g. *map7;aru "sum" > nap7;aru,
by dissimilation of the labials m and p), and in this sense do not
Moscati, Compa.rative Grammar 6
82 Morphology
constitute an independent category (cf. § 12.16). They also appear,
however, as nouns and adjectives of the verbal stem with prefix n- :
e.g. namungatu "paralysis"; nalbubu "enraged". Outside Akkadian
a possible example is Ug. nbldt "flames".
d. Patterns Extended by Infixes
12.20. Patterns with infixed t: apart from the nominal forms of
the verbal stems with t (for which cf. § 16.17-23) we find in Akka-
dian qitbar for adjectives with intensive value (e.g. gitmalu "very
complete", itbaru "very friendly").
e. Patterns Extended by Suffixes
12.21. a} Patterns with suffix -an: qabaran, qabran, qibran,
qubran, maqbaran. These themes (in which the suffix -an is attached
to other patterns already discussed) occur especially in abstracts
(e.g. Ar. fayaran "flight", Reb. *pitran > pitron "solution",
Syr. puqdana "order", ESA 'l;wn "brotherhood", Eth. r(}s'an
"old age"); also in adjectives (e.g. Ar. sakran "intoxicated", Reb.
*qadman > qadmon "eastern", Syr. 'ar'an "terrestrial"); and, finally,
in diminutives (e.g. Ar. 'aqraban "little scorpion", Reb. *'isan >
'isOn "[little-man =] pupil [of the eye]", Akk. miranu "little
animal"). Rebrew has a number of nouns in which the usual
change a > 0 does not take place; the question arises, therefore,
whether they belong to the category at present under consideration:
e.g. sul"f},an "table", qorban "sacrifice", etc. A particular Akkadian
type of the patterns with suffix -an is that which describes a
special person in a special condition (e.g. nadinu "vendor",
the seHer in a particular case of the particular object referred to).
The pattern maqbaran appears in a few cases in West Semitic
(e.g. Reb. massa'on "deceit", from a root ns', Palm. madd(}'an
"knowledge", Syr. ma'b(}rana "passage"). In some cases (e.g. Ar.
taysal as well as fays "great number", hidmil alongside hidm
"patched garment"; possibly Reb. karme1, cf. kerem) we may have
to identify independent patterns with suffix -1, even though of
rare occurrence ..
12.22. b)· Patterns with suffix -m: they are infrequent and occur
predominantly in Arabic adjectives (e.g. fU8lJnm "wide", sadqam
The Noun 83
"wide-mouthed"). In Rebrew one may mention safam "mous-
tache" (from safa) , and in Ethiopic qastam "bow" (from qast);
it is, however, conceivable that in some cases the -m might be a
relic of mimation (0£. §§ 12.73, 76).
12.23. c) Patterns with suffixes -iy, -ay, -awi. When attached to
other themes these suffixes produce adjectives :with the meaning
"belonging to" (e.g. Ar. 'ari}iy "terrestrial", Akk. mal;ru « *mal;-
riyn) "first", Bibl.-Aram. kasday "Chaldaean", Reb. y(}hUdi
"Jewish", etc.). The suffix -awi is characteristic of Ethiopic (e.g.
n'Jgusawi "royal"). The ending -iy is called nisba in Arabic, and
this name has been extended to cover the same type of formation
even outt;lide the Semitic languages (the ending is attested, for
example, in Egyptian: lJmw.ty "artist" from lJmw.t "art").
12.24. d) Patterns with suffixes -ut, -it: when attached to other
patterns (resulting in qabrut, qabrit, etc.) these suffixes produce
themes connoting abstracts. Patterns with -ut occur in Akkadian
(e.g. sarrutu "kingship"), Rebrew (e.g. malkUt "kingship"), Syriac
(e.g. dakyuta "purity"), Ethiopic (e.g. l;irut "goodness"). In
Ethiopic these forms are, however, infrequent, and the same
applies to Arabic. Themes in -it are found in North-West Semitic
(e.g. Reb. resit "beginning", Pun. swyt "curtain", Syr. '(}rau,witii
"fever", Bibl.-Aram. [as well as Reb.] 'alyiirit "end"), perhaps
originally as feminine morphemes (0£. § 12.35). In Akkadian they
appear only as feminines of the pattern -iy, while in Ethiopic they
produce abstract nouns (e.g. dalftirit "end", qadamit "beginning",
etc.). In Ethiopic we also encounter the ending -at (e.g. na'asat
"youth", q(}asat "holiness"). Outside Semitic, nominal formation
with the suffix -t is attested in Egyptian (e.g. m.sdm. t "cosmetics",
from the root sdm).
f. Patterns from Roots with One, Two, Four, and Five Radicals
12.25. a} Monoconsonantal patterns; these are fairly rare:
e.g. Akk. pu "mouth", Ug. p, Reb. pe, Phoen. p, Ar. /ii (of. Eth.
'af); Ug. s "sheep", Reb. se (of. Akk. su'n, Ar. sa'). Further ex-
amples are found in the various languages, e.g. Ug. g
12.26. b) Biconsonantal patterns with short vowel: qab, qib, qub
(e.g. Akk. al;n "brother", Ug. dl;, Reb. 'alJ, Syr. 'alJa, Ar. 'al;,
6*
84 Morphology
Eth. ''J!Jw; Akk. sumu "name", Ug. sm, Heb. Aram. sum,
Ar. 'i8m, Eth. 8'Jm).
12.27. c) Biconsonantal patterns with long vowel or diphthong:
qab, qayb, qawb, qib, qub (e.g. for qab: Akk. tabu "good", Heb.
tob, Syr. tabii, Ar. tab; for qawb: Ar. iawr "bull", Syr. tawra, Akk.
suru, Heb. sor).
12.28. d) Biconsonantal patterns with doubled second radical:
qabb, qibb, qubb (e.g. for qabb: Akk. kappu "palm [of hand]",
Heb. kap [kappi "my palm"], Syr. kappa, Ar. kaf!).
12.29. e) Biconsonantal patterns with reduplication of both
radicals: qabqab, qabqab, qabqub, qibqib, qubqub, qubqub, etc. (e.g. for
qabqab: Sem. *kabkab "star" > Akk. kakkabu, Heb. kOkiib, Syr.
kawk'Jbii, Ar. kawkab, Eth. kOkab; for examples in individual
languages: Akk. kimkimmu "wrist"; Ug. 'p'p "eye"; Heb. galgal
"wheel"; Syr. garg'Jra "threshing flail"; Ar. dakdak "plain"; Eth.
lJazlJaz "swamp").
12.30. f) Four-consonant patterns. In addition to those already
discussed (constituted by the attachment of affixes or the redu-
plication of radicals), forms are attested over the entire Semitic
area on the pattern C
1
aC
2
C
a
aC
4
u: e.g. Heb. 'aqrab, Syr. ''Jqarbii,
Eth. 'aqrab "scorpion". Examples of other four-radical patterns
are Akk. "mouse", Heb. 'alckabis "spider", Syr. 'uqb'Jra
"mouse", Ar. qunfurJ "hedgehog", Eth. lJanbal "saddle". Names
of animals figure prominently in these patterns; the same is true
of words of foreign origin.
12.31. g) Five-consonant patterns. These are infrequent (e.g.
Akk. "potsherd", Heb. "frog") and often of foreign
origin. Others of this type are formed, by attachment of affixes
or reduplication of radicals, from roots with a smaller number of
radicals.
3. Gender
12.32. The Semitic languages distinguish two genders: masculine
and feminine. The masculine possesses no special endings (zero
morpheme), ·whereas the feminine is associated with a special
morpheme which probably goes back to a more complex and ancient
system of classes (cf. §§ 12.34-35), i.e. the ending -(a)t which
The Noun 85
extends over the whole of the Semitic area (and beyond: cf. Egyp-
tian 8; "son", 8;. t "daughter"). For example: Akk. sarrat-u
"queen" (from sarr-u "king"); Ug. ilt "goddess" (from U "god");
Heb. talJtit "lower" (fem.) (from talJti "lower"); Syr. biSta "bad"
(fem.) (from bisa "bad", in the emphatic state: cf. § 12.74); Ar.
malikat "queen" (from malik "king"); Eth. b'J''J8it "woman"
(from b'J''J8i "man"). The fem. gender is not always marked-in
relation to the corresponding masculine-by the feminine mor-
pheme, but is sometimes expressed by means of lexical opposition
(e.g. Ar. lJimar "he-ass", 'atan "she-ass"). Grammatical gender
d.oes not necessarily and invariably correspond either to sex or to
the formal constitution of the noun (cf. §§ 12.34-35).
12.33. In Hebrew and Syriac the Proto-Semitic feminine ending
(retained consistently in the construct state: e.g. mamliika "king-
dom", constr. mamleket) develops in the majority of cases (in the
status absolutus) into -a: e.g. Heb. tobii (fem. of tob) "good";
Syr. biSa (fem. of bis) "bad". According to Brockelmann (GVG,
I, p.409), this development may be understood from the Arabic
pausal form -ah which has extended beyond its original function
(-at> -ah > -a > -a), but this explanation appears somewhat
doubtful. A process similar to that in Hebrew and Syriac seems to
take place in Neo-Punic-to judge by Latin transcriptions such as
Anna for /:tnt, alma for 'lmt, while neither Phoenician nor Moabite
manifests this phenomenon. In Arabic also exist the feminine
endings -a' (type qabra', fem. of 'aqbar, for colour-nouns: e.g.
fem. of "yellow") and -a (type qubra, fem. of 'aqbar,
elative: e.g. fem. of "smallest"). It should be noted,
however, that these morphemes are attached to nominal patterns
different from those of the corresponding masculine-thus con-
stituting an instance of inner morphemes (cf. § II. 3). Finally,
there are the very rare endings, in Hebrew and Syriac, -ay (for
which may be adduced such words as Heb. saray "lady" and
Syr. tn'yay "error") and (from -ay the feminine numeral
Heb. "ten", Syr. ''J8re, in the compounds from 11 to 19). The
Ethiopic ending -e (e.g. 8arwe "army", 'arwe "beast", gize "time")
is not associated with the feminine.
12.34. The Semitic languages show instances of the fairly general
phenomenon of masculine nouns with feminine gender morphemes
86 Morphology
and feminine nouns without them. Examples of masculine nouns
with (apparently) feminine ending: Ar. {}ali/at "caliph", Reb.
qohelet n.pr.m. Examples of feminine nouns without feminine
ending: Ar. nafs, Reb. nepes, Syr. napsa, Eth. na/s "soul" (also
masc.); Ar. 'arq" Reb. 'eref, Syr. 'ar'a "earth". It is interesting
to note that in these latter cases Akkadian attaches the feminine
ending (napistu, erfetu). Some nominal patterns in Akkadian present
variant forms with or without the feminine ending but without a
corresponding difference in meaning (qibr: qibirt; maqbar: maqbart;
taqbir: taqbirt, etc.); and similarly in Rebrew: naqam and n'Jqama
"vengeance", ma'on and m'J'ona "dwelling". The names of paired
parts of the body are generally feminine, yet without feminine
ending (e.g. Akk. uznu "ear", Ar. 'urJ,n, Reb. 'ozen, Syr. 'edna,
Eth. ''Jz'Jn). Finally, it is noteworthy that the cardinal numerals
from 3 to 10 use the forms without gender-ending as feminine and
those with the usual feminine morpheme as masculine (cf. § 14.2).
12.35. The feminine morpheme is employed not only to indicate
the corresponding natural gender but also nomina unitatis, dimi-
nutives and pejoratives, abstract and collective nouns. This
multiplicity of function points to the probable origin of the feminine
ending in a more complex system of classes within which the cate-
gory of number has to be included as well (by way of the collective).
Examples of nomina unitatis: Syr. z'Jbatta "time" = French "fois"
(cf. zabna "time" = "temps"), Reb. 'oniyya "ship" (cf. 'oni
"fleet"), Ar. wamqat "leaf" (cf. wamq "foliage"). Examples of
diminutives or pejoratives: Reb. m'JZuna "hut" (cf. maZon "inn"),
Syr. yamm'Jta "lake" (cf. yamma "sea"). Examples of abstracts:
Ar. waqalfat "insolence" (cf. waqalf "insolent"), Reb. ra'a "wicked-
ness" (cf. ra' "wicked"), Eth. sannayt "goodness" (but also the
masc. sannay). Examples of collectives: Ar. fufiyyat "the mystics"
(cf. fU/iy "mystic"), Reb. 'or'Jlfa "caravan" (cf. "guest",
"traveller") .
4. Number
12.36. The Semitic languages possess three numbers: singular,
plural, and dual. The dual is formed by the attachment to the
singular of special endings; the plural may be formed by the addition
of endings, and in that case it is referred to as "external" or "sound"
plural; or it may be expressed by a change of pattern (i.e. the use
The Noun
87
of a pattern different from that employed for the singular), and in
that case it is called an "internal" or "broken" plural. An example
of the "external" plural is Ar. qaffab "butcher", pI. qaf{Jabuna;
of the "internal" plural, Ar. malik "king", pI. mulUk. For the
characteristics of inner inflexion, cf. § 11.3. In some languages we
find hybrid plurals, i.e. pluralizations of forms already plural; this
occurs both in the form of internal plurals of internal plurals (e.g.
Ar. balad "locality", pI. bilM, further pI. buldan) or in that of
external plurals of internal ones (e.g. tariq "way", internal pI.
furuq plus external fem. {uruqat}.
a. External Masculine Plural
12.37. A comparative examination of the Semimc languages
suggests the following Proto-Semitic morphemes for the external
masculine plural: nominative -u (cf. also Egyptian -w: e.g. nIl'
"god", pI. ntl'.w), genitive/accusative -i. These endings seem to
be the result of the lengthening of the corresponding singular
morphemes of the nominative (-u) and genitive (-i), while the in-
dependent ending of the accusative singular (-a) merges with that
of the genitive in the plural.
12.38. In Akkadian the endings of the masculine plural are nom.
-u, gen./acc. -i (Assyriap -e, later extended in part to Late Babylo-
nian) from the earliest occurrence until the Neo-Babylonian and
Neo-Assyrian period, where -i (-e) prevails and is extended to the
nominative: e.g. sing. sarl'u "king", pI. nom. sarru, gen./acc.
sal'l'ile; Neo-Bab. and Neo-Ass. pI. nom./gen./acc. sal'rile. The
fairly frequent appearance of plural endings of feminine form
(cf. § 12.56 and von Soden, GAG, pp. 77-78) for masculine nouns
(e.g. ikkaru "peasant", pI. ikkaratu; naru "river", pI. naratu)
recalls the similar phenomenon in Ethiopic (cf. § 12.41); for some
cases in other languages cf. § 12.66. Akkadian adjectives exhibit
the special endings nom. -utu, gen./acc. -uti: e.g. rabu "great",
pI. nom. mbutu, gen./acc. l'abUti.
12.39. In North-West Semitic the original endings -u, -i are
preserved in Ugaritic: e.g. nom. gen./acc. l'plm "demigods,
shades of the dead" (for the final -m or -n, with or without an
accompanying vowel, d. below §§ 12.70-77); and laterin Ya'udic:
88 Morphology
e.g. mlkw "kings", gen./acc. mlky. Elsewhere -i predominates and
is extended to the nominative: e.g. Reb. 8U8 "horse", pI. 8u8im;
Syr. bis "bad" (status absolutus), pI. bisin.
12.40. In classical Arabic the Proto-Semitic endings remain:
nom. -u, gen./acc. -i: e.g. qa§l§lab "butcher", pI. nom. qa§l§labuna,
gen./acc. qa§l§labina. For the pre-classical phase of Arabic, the
indications furnished by South Arabian and Lil).yanite (drawn
from forms in the construct state, since in the absolute state the
purely consonantal spelling does not allow any valid conclusions)
suggest the same state of affairs: e.g. Lil).yanite pI. nom. constr.
bnw "sons", gen./acc. bny.
12.41. In Ethiopic the plural endings are -an for the masculine
and -at for the feminine (cf. below § 12.52): e.g. §lad"q "just",
pI. masc. §lad"qan, fem. §lad"qat. However, the feminine ending has
greatly expanded at the expense of the masc. morpheme: e.g. may
"water", pI. mayat (cf. also §§ 12.38, 12.56). The ending -an
remains- in use for adjectives, participles, and a smaller number
of substantives: e.g. ly,adi8 "new", pI. ly,adi8an; ma8i(/' "Messiah",
pI. ma8ily,an.
12.42. The masculine external plural -an current in Ethiopic
(cf. the preceding paragraph) appears also in other parts of the
Semitic area. In Akkadian we find nom. -anu, gen./acc. -ani
(combination of -an with -u, -i) in the Old and Middle periods,
-ani for all cases in Late Akkadian: e.g. sarru "king", pI. nom.
sarranu, gen./acc. sarrani (Late period sarrani for all cases). In
Syriac we find -anin (combination of -an with -in): e.g. rabba
"master", pI. rabbanin. According to Goetze (Language 22 [1946],
pp. 121-30) the ending -an designates "individual" plurals as
distinct from "general" ones (e.g. Akk. ilU "the gods", ilanu
"some gods" or "the gods taken individually"). Gelb (Morphology
of Akkadian, pp. 14-15) regards it as an ending without specific
significance which is used to reinforce short nouns. The ending
-an appears also with internal plurals (cf. § 12.50).
b. Internal Masculine Plural
12.43. Internal plurals are formed, as has been explained, by the
use of patterns different from those of the singular. The patterns
The Noun 89
so used may be regarded as original collectives; their employment
as plurals cannot be established-except when they are construed
as plurals in terms of grammatical concord. Cf. Fleisch, TP A,
pp. 470-505 for a detailed discussion.
12.44. The internal plurals may be regarded as Proto-Semitic,
in the sense that their patterns are Proto-Semitic. However, their
use as plurals is regularly found only in the South. Semitic
area (Arabic and Ethiopic). In North-West Semitic there are some
late and rather doubtful traces: e.g. Heb. rekeb from the singular
r o k ~ b "horseman" (Ar. ralcb from sing. rakib) , Syr. qurya from the
singular q"rita "village" (Ar. qura from sing. qaryat) , Syr. ly,emra
from the singular ly,,,mara "ass". The existence of internal plurals
in Ugaritic has not been demonstrated. For Akkadian, attention
might be invited to Old Assyrian §lu7}rum, a collective corresponding
to the singular Ass. §la7}rum, Bab. §le7}rum "small". It must there-
fore be supposed that internal plurals are a particular development
of South Semitic, although Petracek's studies (cf. § 11.3) have
shown the existence of precedents in Hamito-Semitic. The follow-
ing paragraphs will indicate the principal patterns used for internal
plurals-together with the singular themes to which they cor-
respond.
12.45. a) Disyllabic patterns with short vowel: qabar, qibar,
qubar, qubur. Of these themes, qibar corresponds in Arabic to
the singular qibrat (e.g. qita', pI. of q'Wat "piece") and qubar to the
singular qubrat (e.g. 'ulab, pI. of 'ulbat "box"). In South Arabian
the situation must be similar, though the vocalization is, of course,
unknown to us (e.g. §1M, pI. of §lMt "trench" [ ?]; tnw, pI. of tnwt
"plain"). qub'ur corresponds in Arabic to various singular patterns,
but chiefly to those with the second vowel long (e.g. kutub, pI. of
kitab "book"). In Ethiopic, the merging in " of the Proto-Semitic
short vowels i and u (cf. § 8.96) produces the pattern q"bar, cor-
responding to singular q"br (e.g. 'nan, pI. of '"zn "ear").
12.46. b) Monosyllabic patterns with short vowel: Arabic qubr,
plural of 'aqbar (nouns of colour, cf. § 12.14): e.g. ly,ttmr, pI. of
'aly,mar "red".
12.47. c) Disyllabic patterns with long vowel in the second
syllable, or with short second vowel and feminine suffix: qabir,
90 Morphology
qibiir, qubur; qabarat, qibarat, qubarat. Of these themes, qibiir
and qubur correspond in Arabic mainly to the monosyllabic singu-
lars qabr, qibr, qubr: e.g. bi7:uir, pI. of ba"!Jr "sea"; gunild, pl. of
gund "army". The situation in South Arabian may have been
similar, though we remain ignorant of the vocalization: e.g. lJrwf
and lJryf, plurals of lJrf "year". The themes qabarat and qubarat
correspond in Arabic to the participial pattern qabir in the singular:
e.g. kafarat, pI. of kafir "unbeliever". In Ethiopic the situation is
similar: the internal plural pattern qabart corresponds to the singu-
lar participial theme qabiiri (e.g. pl. of "writer")
as well as to the pattern qabir (e.g. tababt, pI. of tabib "wise").
12.48. d) Patterns marked by doubling: Arabic presents qubbar
and qubbiir, both corresponding to the singular participial theme
qabir (e.g. rugga', pI. of ragi' "returning"; kuttab, pI. of katib "scribe").
12.49. e) Patterns extended by prefixes: 'aqbiir, 'aqbirat, 'aqbira',
'aqbur, 'aqburat. In Arabic all these themes, except the last two,
are frequent: 'aqbiir corresponds mainly to the singular pattern
qabar (e.g. 'amrarf" pI. of mararf, "illness"); 'aqbirat correlates to
singular themes with long vowel following the second radical (e.g.
'agribat, pI. of gurab "crow"); 'aqbira' corresponds principally to
the singular pattern qabir (e.g. 'aqribii', pI. of qarib "kinsman");
'aqbur corresponds frequently to the singular theme qabr (e.g.
'anfus, pI. of nafs "soul"). Noteworthy is also the use of the themes
'aqbar, 'aqbirat, 'aqbur (as well as qibrat) for the so-called plural
of paucity, i.e. a plural for quantities not exceeding ten. In
South Arabian the series of patterns extended by prefixes is
fairly widespread: the consonantal schemes 'qbr, 'qbrt, 'qbrwadmit
of the existence of all the patterns mentioned above, although
their vocalization cannot be established (cf. for example 'byt,
pI. of byt "house"; 'lJrft, one of the plural patterns of lJrf "year";
'kbrw, pI. of kbr "great"). In Ethiopic we have 'aqbiir, 'aqb3r
(= 'aqbur), 'aqbur, 'aqb3rt (= 'aqburat or 'aqbirat): of these patterns,
'aqbiir often corresponds to singular themes qabr, q3br, qabar (e.g.
'a'miid, pI. of 'amd "pillar"; 'albas, pI. of l3bs "dress"; 'adwal,
pI. of dawal "district"); 'aqb3r corresponds mainly to the singular
qabr (e.g. 'awg3r, pI. of wagr "hill"); 'aqbur frequently corresponds
to singulars with two short vowels or with a single vowel (e.g.
'ahgur, pI. of hagar "city"; Ja"!Jqul, pI. of "!Jaql "field"); 'aqb3rt
The Noun 91
correlates mainly to a singular theme qabr (e.g. 'agb3rt, pI. of gabr
"servant") .
12.50. f) Patterns extended by suffixes: qibran, qubran, qubara'.
These are frequent in Arabic (e.g. giran, pl. of gar "neighbour").
qubara' corresponds most often to a singular theme qabir, but at
times also to qabir (e.g. 'umara', pI. of 'amir "Emir"; su'ara', pI. of
sa'ir "poet").
12.51. g) Four-consonant themes: These are formed on the
patterns C
1
aC
z
aC
a
iC
4
and C
1
aC
2
aC
a
iC4,' They correspond to four-
consonant singular patterns (e.g. Ar. 'aqarib, pI. of 'aqrab "scor-
pion"; Eth. sanas3l, pI. of sansal "chain"; Ar. 'anaqid, pI. of 'unqud
"bunch") including those formed from triconsonantal roots by the
attachment of an affix (e.g. Ar. manazil, pI. of manzil "dwelling",
root nzl; Eth. mala' 3kt, pI. of mal' ak "messenger", root l' k plus
suffix -t) ; they also correlate to triconsonantal singular patterns with
long vowel (the plural theme being characterized by the insertion
of ' or w or y): e.g. Ar. fawaris, pI. of faris "horseman"; Ar. 'ara'is,
pI. of 'arus "bride"; Eth. k3saw3d, pI. of k3Siid "neck".
c. Feminine Plural
12.52. The feminine plural is of the external type. It appears
to be formed on the same principle as the external masculine plural,
i.e. by the lengthening of the vowel-contrasting with the short
vowel of the singular:-(a)t, fern. pI. -at; e.g. Ar. malikat "queen"
(fern. of malik "king"), pI. malikat.
12.53. In the languages which retain case-endings, these are
attached to the feminine suffix; the same applies to the morphemes
-m or -n (of. §§ 12, 70-78). Thus we find in Akkadian sal'rum
"king" (nom. sg.), fern. sg. (nom.) sarratum, fern. pI. nom. sarra-
tum, gen./acc. sarratim; and in Arabic malik
un
"king" (nom.
sg.), fern. sg. (nom.) malikat
un
, fern. pI. nom. malikat
un
, gen./ace.
malikat
in
(of. on declension §§ 12.64-69 below).
12.54. In the North-West Semitic area, Hebrew (owing to the
well-known transition a > 0 [of. § 8.83]) presents the feminine
plural suffix -ot: e.g. b3raka "blessing", pI. b3rakot. Syriac has -at
in the construct and emphatic states, but in the absolute state -an,
92 Morphology
probably by analogy with the masculine plural -in: e.g. Msa "bad"
(fem. sg. abs.), pI. cstr. bisat, emph. biSiita, abs. bisan (this 'in-
novation' in the Aramaic area is attested" as early as the eighth
century B.C., in the Arpad inscriptions where -n occurs alongside
the more common -t; later on it appears as the established form
in Biblical Aramaic).
12.55. At times the feminine ending of the plural is superimposed
upon that of the singular-instead of being substituted for it.
This phenomenon is not uncommon in Ethiopic (e.g. barakat
"blessing", pI. barakatiit); here it may, however, be accounted for
by the fact that the plural ending -at is no longer limited to the
feminine and has, in fact, been extended to masculine nouns as well
(cf. § 12.41); it is rare in other languages (e.g. Akk. isatu "fire",
pI. isatatu; Heb. qeset "bow", pI. qasatot).
12.56. In several of the Semitic languages we encounter instances
of nouns of feminine form in the singular and of masc. form in the
plural: e.g. Heb. sana "year", pI. sanim (but in the construct state
a feminine morpheme reappe.ars: sanot); Syr. gannatii "garden",
pI. ganne; Ar. ly,arrat "stony ground", pI. ly,irruna (also ly,arrat).
These formations, which do not normally occur in Akkadian and
Ethiopic (of. however Akk. kabutu "dung", pI. kabu), are rare in
Arabic, though fairly frequent in North-West Semitic. The con-
verse of this phenomenon, i.e. the appearance of feminine endings
in the plural of nouns of masculine singular, is rather more wide-
spread: it occurs particularly in nouns which are of feminine gender
but lack the feminine morpheme in the singular (e.g. Akk. lJ,arranu
"road", pI. a,arranatu; Heb. 'aton "she-ass", pI. 'atonot; Syr.
ly,aqlii "field", pI. ly,aqliita); similarly in some nouns of trade or
occupation (e.g. Akk. ikkiiru "peasant", pI. ikkiiratu; Syr. 'asya
"doctor", pI. 'asawwata) and a number of other substantives. This
phenomenon, developed to varying extents in individual languages
(esp. in Akkadian), has assumed somewhat larger prop6rtions in
Ethiopic where the original feminine ending -at has been extended
to masculine nouns over a fairly wide range (§ 12.41).
d. Plurals of Biconsonantal Nouns
12.57. In West Semitic many biconsonantal nouns form their
plurals by adding a third consonant to the singular pattern;
The Noun 93
this consonant is mostly h: e.g. Heb. 'ama "maid-servant", pI.
'amahOt; Syr. sama "name", pI. samahe; Ar. sa/at "lip", pI. si/ah
and sa/awat, sanat "year", pI. sanawat. In Akkadian some bicon-
sonantal nouns double the second consonant in the plural: e.g. abu
"father", pI. abbu, i$u "tree", pI. i$$u.
12.58. In West Semitic there are also traces of a plural formation
by reduplicating the singular of biconsonantal nouns (and then
adding the external plural ending): e.g. Syr. rab "great", pI. *rab-
rabe > rawrabe; ESA 'l "god", pI. 'l'U. Similar cases occur in
Hebrew with nouns which were perhaps originally monoconsonan-
tal, such as mayim "water", cstr. mf or mfmf. It is possible (Brockel-
mann, GVG, I, p. 439) that this type of plural arose from a dis-
tributive context: of. Syr. ly,adly,adane and Amharic 'andand "some"
(from ly,ad and 'and, respectively, "one"), and Tigre kalkal' at
"every two".
e. Dual
12.59. The dual is used for the linguistic expression of natural
pairs, but it also serves, in some of the Semitic languages, to indi-
cate duality outside these narrow limits. Its extensive use in Old
Akkadian, Ugaritic, and Arabic suggests that the restricted employ-
ment in other languages is secondary (cf. below). A comparative
examination proposes the following Proto-Semitic endings: nom.
-ii, gen./acc. -ay, followed by nunation or mimation.
12.60. The Akkadian dual has -an for the nominative and
(*-ayn » -en/-in for the genitive-accusative. Nunation is dropped
in the more recent period. The distinction between the cases is
gradually lost, and in Middle Akkadian -en/-in predominate over
-an. An interesting feature in Akkadian is the use of the dual as
a plural of paucity: e.g. ubiiniisu "his fingers", sinniisu "his teeth" .
12.61. In North-West Semitic, the unvocalized Ugaritic texts
reveal no formal distinction between dual and plural: Gordon
(UM, pp. 43, 223) reconstructs the dual endings as nom. -ami,
gen./acc. -emi (the latter by contraction of original -ay followed by
mimation). Hebrew exhibits a restricted use of the dual (at any
rate during the historically attested period) with the ending -ayim
predominating and extending to all the cases. The position is
very similar in Aramaic, but here we have nunation instead of
94 Morphology
mimation: -ayn; and in Syriac the dual appears to occur in only
two words (taren [tarten] "two" and maten "two hundred").
12.62. Arabic presents the Proto-Semitic endings followed by -ni:
nom. -ani, gen./acc. -ayni. It has been suggested that -ni is a second-
ary derivative of -na (the suffix having been added to the plural
ending); and the hypothesis has been advanced (though with
doubtful justification) that the change took place by way of vowel
dissimilation (Brockelmann, GVG, I, p.456). In Arabic dialects
,
the ending of the oblique cases predominates over that of the nomi-
native-just as it does in the other Semitic languages. South
Arabian has the dual morphemes -an and -ayn, but their employ-
ment appears to be quite indiscriminate in relation to the various
cases.
12.63. Ethiopic preserves only a few traces of the dual,represented
by the ending (*ay » e: kal'e "two" (0£. § 14.2), 'ade "hands"
(before suffixes), ly,aqWe "loins".
5. Declension
12.64. The Semitic languages originally possessed three basic
cases: nominative (subject), genitive (complement governed by
a noun), accusative (complement governed by a verb). For the
plural and dual endings of these cases see the preceding paragraphs;
if we append to them the endings of the singular the following
picture emerges:
Singular Plural Dual
Nominative -u -u -a
Genitive -I.
}
Accusative
-i -ay
-a
According to Brockelmann (GVG, I, p. 459) the singular endings
are quantitatively ambivalent (-11, -f, -(1); but since their opposition
to the plural morphemes depends on a quantitative distinction
they must be regarded as short.
12.65. To the three basic cases in Proto-Semitic we might have
to add a locative in -u which is attested in Akkadian and traces of
which may perhaps be detected in other languages as well-esp.
The Noun 95
in adverbs such as Ar. taJ;,tu "below" and qablu "before", Eth.
lii:ZU "above" and kanW_ "gratuitously". Some of these examples
may be open to doubt. They also raise the problem of the quantity
of the locative ending -u (on this cf. the following paragraph).
12.66. Akkadian retains the basic case-endings in their entirety:
nom. tabu "good", gen. tabi, acc. taba. Only in the course of time
do these distinctions become progressively blurred: in Neo-Baby-
Ionian and Neo-Assyrian the three case-morphemes are used in-
discriminately or even omitted altogether. The Akkadian locative
in -u (cf. § 12.65) is often used with prepositions (e.g. ina libbu
f'in the midst of, within") or appears joined to prepositions (e.g.
balu[m] "without", istu[m] "from"; for the ending -m o£. § 12.71).
According to Gelb (OA, pp. 144-45) the vowel of the locative
ending was originally long; according to von Soden (GAG, pp. 87-
88) its length is a late and secondary development. The material
at our disposal seems to favour this latter view. Finally, Akkadian
has a fifth case, the dative-adverbial in -is (cf. the element s which
characterizes the dative of the pronouns: § 13.3). As a dative this
occurs in only the most ancient phase of the language (e.g. muatis
"to die", amaris "to see"); for -is in comparisons note ilis "like
a god". As an adverbial, however, it remains throughout the entire
period of Akkadian (e.g. madis "much", damqis "well", etc.). In
conjunction with the ending -am (more rarely -um), used ad-
verbially, -is assumes either terminative or distributive function
(e.g. annisam "hither", umisam "daily").
12.67. In the North-West Semitic area, Amorite and the Tell
Amarna glosses retain the Proto-Semitic case-endings-and so
does Ugaritic, as may be seen in nouns whose final consonant
is ' (vocalized with a, i, n). In the later languages the endings
disappear and with them the formal distinction between the cases,
leaving only a few faint traces: in Hebrew the ending -a denoting
motion towards a place (e.g. Babel "Babylon", Babela "towards
Babylon") is regarded by some scholars as such a survival; in
Aramaic, Brockelmann (GVG, I, p. 465) considers adverbs such as
taJ;,ta "below", bara "outside", etc., also as survivals, but this
remains highly conjectural.
12.68. Classical Arabic (as indeed the pre-classical language-so
far as we can judge) retains the Proto-Semitic declension system
96 Morphology
in its entirety: nom. malik
u
"king", gen. maliki, ace. malik
a

There exists, however, a series of nouns whose declension is limited
to two endings only: -u for the nominative and -a for the genitive/
accusative (cf. Fleisch, TPA, pp. 269-80). Wehr's studies on the
nominal pattern 'aqbar
U
conjecture that this diptote declension
originated within that theme. It would then have spread to other
adjectival themes (qabran
U
) and internal plurals ('aqbira', qubara',
C
1
aC
2
aC
a
iC
4
u, C
1
aC
2
aC
a
iC
4
u, and others) as also to a series of proper
nouns with the feminine ending (e.g. Makkat
U
"Mecca"), to some
verbal forms (e.g. Yazid
u
, "Yezid") and to some names of foreign
origin (e.g. 'Ibrahim
u
"Abraham"). For a possible occurrence of
diptotes in Ugaritic see Gordon, UM, p. 45. It is noteworthy that
diptotes do not take nunation. In modern Arabic dialects case-
endings have disappeared altogether-just as they have done in
the other Semitic languages.
12.69. Ethiopic retains only one oblique case ending (indicating
also motion towards a place): e.g. nom. gabr "servant", acc. and
constr. gabra. The ending -a appears both in the singular and in
the plural of nouns terminating in a consonant: e.g. nom. 'agbewt
"servants", acc. and constr. 'agb3rta. The ending -a of the construct
state has possibly arisen from an analogical extension of the accu-
sative morpheme. The final -il of certain numerals may be a survival
of the nominative ending: e.g. 'aly,adil "one" (thus Dillmann, EG,
§ 142, p. 318, though other explanations have been proposed for
this element-of. Brockelmann, GVG, II, p.274). Proper names
are either indeclinable or form an accusative by the attachment of
stressed -ha (e.g. Y38ly,aqha "Isaac" [acc.]).
6. Definiteness and Indefiniteness
12.70. It is not possible to reconstruct Proto-Semitic forms for
the expression of definiteness or indefiniteness. It will, therefore,
be well to examine first the various languages separately and then
to consider what common features emerge from this investigation.
12.71. In Akkadian all nouns, whether definite or indefinite,
have the ending -m (mimation) in the masculine singular and in the
feminine singular and plural, the ending -n (nunation) in the dual,
and neither mimation nor nunation in the masculine plural:
The Noun 97
Singular Plural
(Masc.) (Fem.) (Masc.) (Fem.)
Dual
(Masc.) (Fem.)
Nom. sarrum sarratum sarril sarratum sarran sarratan
Gen. sarrim sarratim
}
sarri sarratim
Acc. sarram sarratam
sarren sarraten
It can thus be seen that mimation and nunation co-exist in
Akkadian, but they do not possess the function of distinguishing
definiteness and indefiniteness-as is the case in some other
languages. In any event, mimation and nunation fall into disuse
from the end of the Old Babylonian and Old Assyrian periods.
It should be observed that in Old Babylonian nunation occurs
instead of mimation in some personal pronouns (yattun, yztttun,
kattun, kuttun, suttun: of. § 13.18) and demonstratives (anm7tun,
annatun: of. § 13.29); in the case of the latter there may possibly
be some connexion with the indication of nearness, but the re-
lationship of this phenomenon with the problem under examination
remains somewhat uncertain. Noteworthy is also the absence of
mimation in some proper names in Old Akkadian as well as in
some common nouns used as proper names: e.g. abu "father"
and a!Ju "brother" when denoting gods (von Soden, GAG, p. 80;
Gelb, OA, p. 145).
12.72. In the North-West Semitic area mimation is found in
the majority of Amorite proper names (of the type $aduqum,
Yapla!Jum etc.) as well as in some Egyptian transcriptions. In
the Tell Amarna glosses we encounter only a few traces of this
phenomenon. Ugaritic shows neither mimation nor nunation in
the singular or in the feminine plural, but it presents endings with
-m in the dual and in the masculine plural. Gordon's hypothetical
vocalization produces (UM, pp. 43-44, 223) the following picture:
Nom.
Gen./Acc.
Plural
(Masc.)
tabilma
tabima
Moscati. Comparative Grammar
(Fem.)
tabatu
tabati
(Masc.)
tabami
tabemi
Dual
(Fem.)
tab (a)tami
tab(a)temi
7
98
Morphology
In North-"West Semitic, as indeed in North-East Semitic, no
properly established relationship can be ascertained between
mimation and nunation, on one hand, and aspects of determination,
on the other.
12.73. Hebrew possesses neither mimation nor nunation in the
singular. Traces of mimation have been seen in forms such as
darom "south", ly,artom "soothsayer", etc., and particularly in such
proper names as Milkom and the series of adverbs (ending in -am)
which includes 'omnam "truly", ly,inncim "gratuitously", etc. (cf.
Torczyner's studies). Yet, the interpretation of these data remains
uncertain. Some formations in -n are also found, such as Gid' on,
but they may well be secondary-or, at any rate, can be explained
as nominal patterns with suffixes. Mimation is a prominent feature
in the masculine plural and in the dual (-im, -ayim) which tallies
with the state of affairs in Ugaritic (cf. preceding paragraph).
This mimation is not, however, related to a differentiation between
the definite and the indefinite; instead, definiteness is expressed,
in all three numbers, by a prefixed article ha- followed by doubling
(or reinforcement) of the initial consonant of the noun (e.g. melek
"king", ham-melek "the king"; m3lakim "kings", ham-m3lakim "the
kings"). The position in Phoenician resembles that in Hebrew-at
least as far as can be judged from the purely consonantal script;
the transition, in late Punic, of the article h- to ' - is due to a general
phonetic change in that area (cf. § 8.55). Moabite differs from
the other Canaanite languages and agrees with Aramaic in the
appearance of nunation (-n) in the plural.
12.74. Syriac (and Aramaic in general) agrees with Hebrew in
possessing neither mimation nor nunation in the singular. Nominal
forms such as 'imama "day", pumma "mouth" may, however,
be regarded as preserving traces of mimation. In the plural (there
is no dual) nunation occurs in the ending -in, but-as in Hebrew-
it is not connected with the differentiation between definite and
indefinite. Definiteness is formally expressed, from the most
ancient Aramaic inscriptions onwards, by a suffixed article -a
(emphatic state of the noun) which in the Eastern Aramaic dialects
(and in Syriac in particular) loses this specific function and becomes
the normal ending of all nouns. Remnants of a nominal ending -a,
The Noun 99
though not entirely agreeing in the specifically determinative
function, have been detected in Old Akkadian, Old Assyrian, and
Amorite (cf. Garbini, SNO, pp. 118-21).
12.75. In the Arabic area, Epigraphic South Arabian presents
a fairly complex situation (Beeston). The absolute state occurs
either with or without mimation, and the ending -m does not by
itself necessarily correspond to the indefinite article ('8
1
m "a man"
or "the man"). The factors governing the use of mimated or un-
mimated forms remain obscure, especially as the syntactical function
of mimation appears to be negligible. Mimation occurs in the singular,
the internal plural, and the external feminine plural. In the dual and
the external masculine plural we encounter unmimated forms only.
On the other hand, the ending -n is attached to the singular, internal
plural, and external feminine plural (it is not attested in external
masc. plurals). It has the function either of a demonstrative or of
a definite article, e.g. ?lmn "this statue" or "the statue". In pre-
classical North Arabic -m occurs occasionally in Tamudic to indicate
indefiniteness, while definiteness is expressed by the prefixed
article h-. In the earliest Lil;tyanite inscriptions we still encounter
-n to mark definiteness, whereas more recent epigraphic documents
have brought to light two instances of (')l-. Finally, classical
Arabic has -n for the expression of indefiniteness and the prefixed
element 'al- as a definite article.
12.76. Ethiopic possesses neither mimation nor· nunation, for
the element -an of the masculine external plural can scarcely be
regarded as a form of nunation. Traces of mimation have been
seen by some scholars in nominal forms such as qa8tam "bow"
(alongside qast), but this opinion does not seem to be well substan-
tiated. Ethiopic has no prefix-article, but a suffix-substitute has
been evolved from the pronominal suffixes (cf. Praetorius, Gram-
matica Aethiopica, p. 33, § 38; Dillmann, EG, p. 426, § 172 [b]):
b3'3Sihu "the man", dabru "the mountain", etc. From these elements
of frozen suffixes, no longer dependent on any antecedents, Amharic
has developed a type of suffix-article. Ethiopic possesses very
elaborate syntactical means for the periphrasis of the definite
article (Dillmann, EG, pp. 423-30, §§ 172-73).
12.77. The evidence adduced in the foregoing confirms the view
that it is impossible to identify any Proto-Semitic means of ex-
7*
100 Morphology
pressing definiteness or indefiniteness. In this respect singular,
dual, and plural seem to behave differently, so that observations
valid for the singular cannot always be applied to the other numbers
in which mimation or nunation are sometimes preserved after
having been dropped in the singular. On this basis it may be
suggested, by way of hypothesis, that originally there existed
a mimation of nouns independent of any semantic function as
regards definiteness or indefiniteness; and that this usage is reflect-
ed in the most ancient Semitic linguistic material. It may further be
assumed that mimation developed into nunation in some languages,
such as Arabic and Aramaic, following upon the change m >n
which occurs also in other connexions in these languages. It may
then be shown how, in the course of the historical development of
the Semitic languages, new and special means of indicating definite-
ness make their appearance in a number of different guises: the
prefixes h- in Hebrew (and also in some pre-Islamic Arabic dialects)
and 'al- in Arabic (a connexion between these may be seen both in
the alternation h:' and in the doubling of the ,following consonant
in Hebrew and Arabic-in the latter as a substitute for the total
assimilation of l to certain consonants); the suffixes -a in Aramaic,
-n in South Arabian, and -u in Ethiopic. Where definiteness is
expressed by a prefix it may bring about a modification in the
use of mimation or nunation, i.e. it may become a means of in-
dicating indefiniteness by virtue of the contrast to the definiteness
connoted by the article. This happens in the singular in classical
Arabic (where the article excludes nunation: e.g. qa??ab
un
"a but-
cher", but 'al-qa??ab
u
"the butcher"), while in Hebrew, where
mimation appears in the plural only, it is semantically irrelevant
and is retained together with the article (e.g. mJlakim "kings",
ham-mJlfikim "the kings"). Where definiteness is expressed by a
suffix, this replaces mimation or nunation which now become
an indication of indefiniteness: e.g. in Aramaic where the
endings of the status emphaticus have extended their use at
the expense of the forms with nunation (Syr. bise prevailing
over biSin).
12.78. The Semitic "construct state" is closely connected with
the function of definiteness or indefiniteness; . this is the special
form taken by a noun when it is defined by a following genitive
The Noun 101
(or pronominal suffix). In these conditions the nomen regens
merges with the nomen rectum in a single complex whose principal
stress falls on the rectum, i.e. the "genitival" element. The two
nouns cannot ordinarily be separated, though there are certain
exceptions to this rule, e.g. in South Arabian: nisI wqbr HNTSIR
"HNTSIR's monument and tomb" (classical Arabic would have
changed the construction to "Hntsr's monument and his tomb").
A noun in the construct state has neither mimation (nunation)
nor the article (e.g. Ar. qa??ab
un
"a butcher", 'al-qa??ab
u
"the
butcher", qa??ab
u
... "the butcher of ... ") except in certain in-
stances of what is termed "improper annexation" (Akk. damqam
inim literally "good of eye", Ar. 'ar-ragul
U
l-J:w,san
u
l-wagh
i
literally "the man handsome of face"; cf. von Soden, JNES 19
[1960], pp. 163-71). The case-endings are retained in their
entirety in Arabic. In Akkadian their retention is less complete:
they appear in some nouns ending in a vowel and before suffixes
(e.g. maru-su "his son", gen. mari-su, acc. mara-su), but in general
the autonomous genitive-ending contrasts with a case-element
common to nominative and accusative. In Ethiopic the ending -a,
characteristic of the accusative, serves also for the construct state
(cf. § 12.69: extension by analogy n. The unity of the noun with
the following "genitive" (with the principal stress on the latter) and
the consequent reduction in the case-endings may occasion changes
in the form of the noun in the construct state. For details the
grammars of the various languages have to be consulted, but it
may be said that these changes are often either of an anaptyctic
type (e.g. Akk. uznu "ear", cstr. uzun; Heb. 'ebed "servant", cstr.
*'abd: cf. § 9.17) or involve modifications in vowel quantity owing
to the displacement of the stress (e.g. Akk. mtiru "son", but before
a suffix maru-su "his son"; Heb. daMr "word", but cstr. dJbar
owing to the shift of the principal stress to the following nomen
rectum). In many cases the original Semitic form can still be
recognized in the construct state: so in the example just cited,
Heb. *'abd (nominal pattern qabr: cf. § 1,2.5) and in the Hebrew
and Syriac feminine-ending (e.g. Heb. yona "dove", cstr. yonat:
cf. § 12.33).
12.79. In Akkadian and Aramaic three "states" of the noun may
be distinguished. In addition to the construct state there is that
102 Morphology
called rectus or "emphatic", in which Akkadian exhibits mimation
and Syria? the ending -a (e.g. Akk. al;um "brother", Syr. 'alyii).
These endmgs do not, however, retain any significance in terms of
noun. Thi.rdly, the "absolute" state is devoid of any
endmgs; It IS comparatIvely rare and occurs mainly in distributive
expressions (e.g. Akk. ina kar kar-ma "in every colony", Syr.
rappin rappin "in flocks"), adverbial contexts (e.g. Akk. ana dar
"for ever", Syr. ba-'gal "in haste"), in some numerals (e.g. Akk.
isten "one", Syr. ly,ad) , and as predicates (of. von Soden, GAG
p. 79; Brockelmann, SG, pp. 104-105). '
O. The Pronoun
1. Independent Personal Pronouns
13.1. The independent personal pronouns of the principal Semitic
languages are as follows:
Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic
Sg. I anaku tin(k) 'ani, 'anoki 'ena 'ana 'ana
2 m. alta tit ' atta 'aU 'anta 'anta
f. atti tit 'att 'att 'anti 'anti
3 m. su hw hu hU huwa wa'atu
f. Si hy hi hi hiya YJ'ati
PI. I ninu
2 m. attunu 'attem ' atton 'antumUi) 'antammu
f. attina 'att!3n(a) 'atten 'antunna 'antan
3 m. sunu hm h!3m(ma) hennon hum(u) 'Jm!tntu
f. Sina hrm(na) hennen hunna 'amantu
Du. 2
'antuma
3 [hm]
huma
13.2. A few general remarks on this table:
a) The first and second persons singular and plural belong to the
same system ('an- plus suffixes), while the third person is formed
from elements related to the demonstratives (cf. § 12.32).
13.3. b) The Akkadian series is used for the nominative onlv
while the other cases employ considerably different elements relat;d
to the forms of the suffix pronouns (cf. § 13.14):
The Pronoun 103
Genitive / A ccusati ve Dative
Sg. 1 yati yasi(m)
2 m. kati/a kiisim
f. kati kiisi(m)
3 m. suati/u, sati/u suasim, sasu/i(m)
f. su/iati, sati su/iiiSim,8asi(m)
Pl. 1 niati niiiSim
2 m. kunuti kunusi(m)
f. [kinati] [kinasi(m}]
3 m. sunuti sunusi(m)
f. sinati [sinasi(m)]
13.4. c) Old Assyrian uses the genitive/accusative forms for the
dative as well, and for the second person singular, masculine or
feminine, it possesses a form of its own, ku(w)ati; in later dialects
new forms may be observed: for the genitive/accusative, 2 m. pI.
katun(u), 3 m. suatunu, satun(u), 3 f. satina, suatina; and for the
dative, 2 m. pI. kiisun(u) , 3 m. sasun(u), 3 f. sasina. In the later
dialects there also occur extensive fluctuations between the geni-
tive/accusative and dative forms. For a connexion between the s
element characteristic of the dative forms and the ending -is of
the dative-adverbial case of the noun, cf. § 12.66.
13.5. d) Ugaritic, too, shows variant forms for the genitive/accusa-
tive, but they are confined to the third person: the characteristic
element in these forms is a suffixed -t (hwt, hyt, hmt).
13.6. e) The suffixed element -t in the third person also occurs
in Phoenician (hmt), as a variant form in South Arabian demon-
stratives (cf. § 13.9), and in Ethiopic (wa'atu, ya'ati, 'amuntu,
'amiintU) , but it is not connected in these languages with any
distinction between the cases.
13.7. To pass now to a consideration of individual forms, we may
posit a Proto-Semitic 'ana(ku) for the first person singular; the
additional element -k appears in North-East and North-West
Semitic (cf. also a-nu-ki in the Tell Amarna glosses, ''lik[y] in both
Phoenician and Old Aramaic, 'nk in Moabite; but contrast ana
which is attested as a secondary form in Old Babylonian). In
North-West Semitic there is a distinctiun in the vowel (-ki),
104 Morphology
perhaps by analogy with the suffix-pronoun of the same person (-ni).
In Egyptian, too, the element -k is attested (ink). As regards the
quantity of the final vowel in 'ana as well as in other pronouns,
one should bear in mind the existence of considerable fluctuations
and the difficulty of determining Proto-Semitic vowel-quantity
with any degree of certainty (cf. Brockelmann, GVG, I, pp. 296-
313; Gray, SOL, pp. 61-65).
13.8. For the second person singular we may postulate the
Proto-Semitic forms 'anta, 'anti (Kienast's recent reconstruction
offering as Proto-Semitic forms ka, ti raises several difficulties);
noteworthy are: a) the frequent assimilation of n (the form alta
of the Tell Amarna glosses confirms the extension of this assimila-
tion to a great part of the North Semitic area; 'nt, 'nty in Egyptian
Aramaic and 'ant, 'anti in Biblical Aramaic are probably historical
spellings); b) the dropping of unstressed final vowels in the North-
West Semitic area, in accordance with the phonetic laws of that
region (cf. § 10.8)-with the exception of the second person mascu-
line in Hebrew where the final vowel remains (important survivals
are also the Hebrew spelling '''atti'' and the Syriac form with -y
in the feminine).
13.9. As to the third person singular: a) Akkadian uses for this
pronoun a consonantal element different from that encountered
elsewhere (s instead of h), and this element is to be found (cf. in
Egyptian: masc. sw, fem. sy) in the South Arabian dialects as
well-with the exception of Sabaean. The South Arabian third-
person pronoun has the forms given in the following table (seman-
tically, South Arabian keeps the original demonstrative values):
Sg. 3 m.
f.
PI. 3 m.
Du.
Sabaean
h', hw', hwt
h', hy', hyt
hmw, hmt
Minaean Qatabanian
SlW, slwt
slyt
slmt slm, slmt
slmyt
b) It appears that the series with h and that with s are both of
Proto-Semitic origin. A reconstruction (supported by the Modern
South Arabian dialect of Mel;tri which has a masculine he and
a feminine se) suggests Proto-Semitic huwa for the masculine and
siya for the feminine; this would explain the position in the various
The Pronoun 105
languages as due to analogy working in opposite directions. But
this reconstruction remains highly conjectural, and it is equally
possible to envisage the formation of the third person pronouns
from either series. c) Some (cf. Gray, SOL, p. 62) prefer Proto-
Semitic hu'a (hu'a is now found in the Dead Sea Manuscripts),
hi' a, Sil' a, si' a to huwa, hiya, suwa, siya; but it is easier to explain
the loss of intervocalic wand y tllan their secondary insertion
(though even this is possible by assimilation to the preceding
vowel). d) The Ethiopic forms wa'atu, ya'ati are conceivably due to
the omission of the initial h- followed by the process 'uw > wu > wa
and 'iy > yi > ya-and subsequent attachment of final -til, -ti.
13.10. For the first person plural we may propose the Proto-
Semitic form naJ:tnu (or even nily,nu, on the basis of Akkadian,
taking the a as due to the influence of the following pharyngal).
It is improbable that Proto-Semitic had an initial vowel of a timbre
(,a-naly,nu or 'a-nily,nu); where such a vowel appears, it may owe
its origin to analogy with the singular form ('ana). As for the
variants in the final vowel, this may again be due to analogy with
the vowel of the suffix pronoun of the same person-as may be
seen (except in Arabic) from the tables (cf. §§ 13.1, 13.14; in Akka-
dian, alongside the form ninu, there later appears nini).
litH. For the second person plural we may posit the Proto-
Semitic forms 'antumu, ,antina: a) the distinctive vowel (u, i) and
consonant (m, n) of these forms are subject to the effects of analogy
(the consonant, in some cases, undergoes gemination); in Akkadian
n prevails over m, in Arabic u over i, etc.; b) the dropping of final
unstressed vowels in the North-West Semitic area corresponds to
the phonetic laws of that region (cf. § 10.8), despite some instances
in which the retention of the vowel is attested by Masoretic Hebrew
for the second as well as the third person plural ( ' a t t ~ n a , hemma,
henna). The Dead Sea documents bear witness to a situation when
this phenomenon was even more widespread (spellings 'tmh, 'tnh).
13.12. As to the third person plural it will be observed that a) for
the consonantal element the same holds true as in the case of the
third person singular (cf. § 13.9); b) taking into account the argu-
ments adduced with regard to the singular, we may postulate, by
way of reconstruction of the Proto-Semitic forms, a series humu,
hina and a parallel one sumu, sina; c) in very much the same way
106 Morphology
as has been shown for the second person plural (cf. preceding
paragraph), analogy affects both the first vowel (u, i) and the follow-
ing consonant (m, n): again n prevails over m in Akkadian, u over i
in Arabic, etc. (for instances of gemination of the consonant
of. § 13.11); d) final unstressed vowels are dropped in the North-
West Semitic area, in accordance with the phonetic laws of that
region, save for a few remnants in some Old Aramaic documents
(hmw); e) the Ethiopic forms ''Jmilntil, ''Jmantil may possibly be
explained as due to the disappearance of initial h- (cf. the enclitic
forms 'ennon, 'ennen in Syriac and the Ethiopic singular forms,
§ 13.9 d); the element -til has been added (as in the masculine
singular). Ethiopic makes use also of the variant forms w'J''JtOmil
(masculine) and w'J''Jton (feminine), formed by the attachment of
the suffix pronouns (cf. § 13.14) to the masculine singular base,
thus recalling, in part, the Akkadian developments suatunu,
satina alongside sunu, sina (of. § 13.4).
13.13. For the dual one may assume as Proto-Semitic the Arabic
forms 'antuma, huma. The data available are, of course, extremely
limited.
2. Personal Pronoun Suffixes
13.14. The suffixed personal pronouns in the principal Semitic
languages are as follows:
Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic
Sg. 1 (noun) -ya, -i ( -y) -i -ya, -i -ya
(verb) -nt -n -ni -n -ni -ni
2m. -ka -k -ka -k -ka -ka
f. -ki -k -k -k -ki -lei
3m. -s(u) -h chil, -0 -(h)i, -h -hu -hil, -0
f. -s(a) -h -(h )a, -ah -h -ha -(h)a
PI. 1 -nt -n -nil -n -na -na
2m. -leunu -lem -kem -leon -leum(1L) -bmmil
f. -kina -len -ken -ken -kunna -bn
3m. -sunu -hm -(helm -han -hum(1L) -(h)omil
f. -sina -hn -(he)n -hen -hunna -(h)on
Du.1 -ny
2 -km -kuma
3 -hm -huma
The Pronoun 107
13.15. The attachment of the suffixes may be effected by means
of connecting (or glide-) vowels or by way of contraction (for
details see the grammars of the various languages). In some lan-
guages the consonant n is inserted before the suffixes (cf. Garbini,
SNO, p. 98); for the reappearance of Proto-Semitic endings before
the suffixes of. §§ 16.139-42.
13.16. Some general remarks on the above table:
a) The Akkadian series is used for the genitive, while for the
accusative and dative the following forms appear (they are, for the
most part, very different and correspond closely to those of the
independent personal pronoun for the accusative/genitive and
dative; of. § 13.3):
Accusative Dative
Sg. 1 -nt -a(m), -ni(m)
2 m. -lea -ku(m)
f. -ki -ki(m)
3 m. -s( n) -su(m)
f. -s(i) -si(m)
PI. 1 -niati -niasi(m)
2 m. -knnilti -kunusi(m)
f. -kinati -kinasi(m)
3 m. -sunuti -snnusi(m)
f. -sinati -sinasi(m)
13.17. b) Old Assyrian uses for the accusative the forms of the
genitive (§ 13.14) and for the dative those of the accusative; in
later dialects, the forms of the accusative and dative are no longer
kept apart and are used fairly indiscrIminately.
13.18. c) Furthermore, Akkadian employs an independent posses-
sive pronoun which is based on the endings of the suffixes; it is
infrequent in Old Akkadian and, for want of Semitic parallels,
must be considered a secondary development; its principal forms
are as follows (Assyrian in brackets):
108 Morphology
]Y[asc. sing. Hem. sing.
Sg. 1
2
3
yii'um, yiim yat(t)um/n
kiim (ku'a'um) kattum/n
sii(m) sattum, suitun
PI. 1 ni'a'um,niim
2 7cuniim
3 sumiim
ni' atum, n'uttum
(7cuniitum)
(suniitum)
2I!asc. plur.
ya'iitum, yiitum
kuttun, (kuwa'iitum)
Hem.plur.
yat(t)u,m/n
kattun
(ni'ii'iitum), nuttum (ni'atum)
(7cunii' atum)
13.19. d) At a later period of Akkadian a new type of possessive
occurs which is formed by adding the genitive suffixes to attu (attu-ka,
attu-kunu, etc.); for the variant forms with final -n instead of -m
(yattun, yuttun, kattun, kuttun, suttun) see § 12.71.
13.20. e) An independent possessive pronoun occurs in some
languages which is the result of a combination of the suffix-
pronouns with elements of the relative pronoun (cf. §§ 13.34-39),
at times connected by the particle l "to": thus in post-Biblical
Hebrew se-l (e.g. selli "mine", literally "which [belongs] to me")
and in Syriac di-l (e.g. dilan "ours"). In Ethiopic the independent
possessive is formed by adding the suffix-pronoun to zi'a- for the
masculine singular, ',mti'a- for the feminine singular, ''Jlli'a- for
the plural (e.g. zi'aya "mine" [masc.], ''Jnti'aya "mine" [fern.],
''Jlli'aya "ours").
13.21. £) Ethiopic has also produced independent pronoun forms
by adding suffixes to lali-, for the subject, and to kiya-, for the
object (laliya, lalika, laliki, etc.; kiyaya, kiyaka, kiyaki, etc.).
Hebrew has created an independent pronoun for the direct object
by using the element 'ot- (,oti, 'Qt3ka, etc.); Arabic employs 'iyya-
for the same purpose (,iyyaya, 'iyyaka, etc.).
13.22. Turning to a consideration of individual forms, we may
propose, for the first person singular, a Proto-Semitic form -ya
or -i. If the original form was -ya, then the -i which appears in
certain languages is due to the loss of the final short vowel and the
subsequent transformation of the semivowel into the homorganic
vowel; or, if the original form was -i, the change -i > ya in certain
languages is due to analogy with the suffix of the second person
singular masculine -ka. The predominance of -i in Old Akkadian
and in Amorite favours the latter hypothesis. The suffix attached
to verbs is preceded by the consonantal element n, possibly by
The Pronoun
109
analogy with the suffix of the first person plural. In Syriac the
final -i of the suffix is written but is not pronounced, and the same
is true of the final vowel of the second person singular feminine and
the third person singular masculine (cf. the optional omission of
the final vowel in the suffixes of the Akkadian third person singu-
lar).
13.23. For the second person singular we may suggest the Proto-
Semitic forms -ka, -ki. Reference has already been made (§ 13.7)
to the uncertainty about the quantity of the vowel element of these
suffixes (as well as of the others-except, probably, those of the
third person singular where the vowel appears long). As to the
feminine form, the loss of the final vowel is already encountered in
the form -k of Amorite and the Tell Amarna glosses, whereas
Hebrew, on the other hand, has in some cases -ki (yet Greek
transcriptions point to instances of -k for the masculine: e.g. w(Jr::X
for' Od3kii, 17vaX for '?nekii).
13.24. For the third person singular one should note: a) Akkadian
forms the suffixes of this person with the consonantal element
s instead of h-just as it does with the independent pronoun. The
same occurs (again in consonance with the independent pronoun
forms) in the South Arabian dialects-with the exception of
Sabaean:
Sabaean Minaean Qatabiinian
Sg. 3 m. -h(w) -SI(W) -SI(W), -SIWW
f. -h -S1 -s1, -Sl yw
PI. 3 m. -hm(w) -slm -slm
f. -hn [-sIn]
Du. 3 -hmy -slmn -Slmy
b) The same considerations apply to the Proto-Semitic suffix-
pronoun as were formulated with regard to the independent
pronoun (cf. § 13.9): two series can be established, -hu, -ha (cor-
roborated also by the Amorite documentation) and -su, -sa (the
vowel is not invariably long)-unless it be conjectured that -hu
was originally the masculine and -sa the feminine form and that
110 Morphology
the workings of analogy subsequently acted in opposite directions.
c) In Old Aramaic a variant form -hi is attested for the masculine.
13.25. For the first person plural we may posit the Proto-Semitic
form -na (corroborated by both the Amorite and the Old Aramaic
documentation). The variations in the vowel in North-East and
North-West Semitic (to the Hebrew -nu corresponds -nu in the
Tell Amarna glosses) must be considered the result of analogy with
the final vowels of the independent pronouns (cf. § 13.1).
13.26. For the second person plural we may suggest Proto-
Semitic -kumu, -kina (the observations on the independent pronoun
[§ 13.11] are relevant also in this context): a) the vowels (u, i) and
the consonantal elements (m, n) are subject to analogy working in
various ways: in Akkadian n prevails over m, in Arabic u over i,
etc.; b) final unstressed vowels are dropped in the North-West
Semitic area (cf. § 10.8), yet some cases of retention are attested in
the Dead Sea Manuscripts, for both the second and the third person
plural (kmh,hmh, hnh).
13.27. For the third person plural note the following: a) as regards
the consonantal element, the same observations hold good which
have already been made for the singular suffix and for the inde-
pendent pronoun (cf. §§ 13.9,13.24); b) on this basis we may propose
for the Proto-Semitic pronoun the forms -humu, -hina as well as
-sumu, -sina, while again referring to the possibility of a Proto-
Semitic masculine form -humu and a feminine -stna; c) in conso-
nance with the second person plural (cf. § 13.26) and the independent
pronoun (cf. § 13.12), both the first vowel (u, i) and the following
consonant (m, n) are subject to the workings of analogy: once more
n prevails over m in Akkadian, and u over i in Arabic, etc. ; d) final
unstressed vowels are dropped in the North-West Semitic area
(cf. § 10.8), but at times they are retained, e.g. in the Hebrew of the
Dead Sea documents (cf. § 13.26); e) the Ethiopic forms -(h)omu,
-(h)on (like the -0 of the third person singular masculine) might
be explained by way of contraction of the initial u of the suffix
with the final a of the noun or verb.
13.28. The agreement of Arabic and Ugaritic suggests that the
forms of the dual are Proto-Semitic. The first person dual-found
Sg.
PI.
The Pronoun III
in Ugaritic alone among the Semitic languages-occurs also in
Egyptian and may thus be regarded as Proto-Semitic.
3. Demonstrative Pronouns
13.29. The principal forms of the "near" demonstrative pronoun
("this") are as follows:
Akka- Hebrew Phoen. Bibl. Syriac Arabic ESA Ethiopic
dian Aram.
m. annu
ze }
r
ana han(a) (ha)ila iln zJ(ntu)
z(')
f. annitu z6t da had(e) (ha)ilihi, ilt za(tti)
ili, ti
m.
annutu I rUu(ntu),
'f!lle 'l ' ~ l ( l e ) , 'Wen hallen (ha)'ula('i) 'lnft 'all6nUJ,
f. annatu 'gUa(ntu)
Du.m. (ha)ilani
f. (ha)tani
13.30. Observations on the above table: a) the Akkadian forms
are the result of contraction of more ancient ones: annium >
annu(m), anniutum > annutu(m), anniatum > annatu(m); b) Old
Babylonian presents, apart from annum, the expanded and declin-
able form anni/ummum, while in Neo-Babylonian a new demon-
strative appears: aga, fern. agatu/i, m. pI. agannutu, f. pI. agannetu
and agati; c) for some Akkadian variant forms with final -n instead
of -m (annutun, anniatun, annimmutun) cf. § 12.71; d) Ugaritic is
not listed because the forms of its demonstrative pronoun are rare
and doubtful (hn and hnd are probably a type of demonstrative);
the additional languages adduced in the comparative table, Phoeni-
cian, Biblical Aramaic, and South Arabian, are mentioned on
account of their special interest; e) Hebrew presents the variant
forms zo for the feminine and 'f,l for the plural which correspond to
the Phoenician forms; the indeclinable zu is used as a relative (cf.
§ 13.36), and the expanded forms masc. sing. hallaze, fern. sing.
hallf,zu, sing. masc. and fern. hallaz are related to the Arabic "far"
demonstratives (cf. § 13.37); f) the plural form indicated for South
Arabian is the Sabaean one, while Minaean has 'hlt and Qatabanian
g,tn/w.
112 Morphology
13.in. The principal forms of the "far" demonstrative pronoun
("that") are as follows:
Akkadian Bibl. Aramaic Syriac Arabic Ethiopic
Sg. m. ullu #k, d i k k ~ n haw rJa(li)ka zaku
f. ullitu dak, d i k k ~ n hay ti/aka, tilka 'antaku,
,
antalcti
PI. m. ullutu
' i l l ~ k
han on
}'ula'ilca 'allahL
f. ullatu hanen
13.32. Observations on the above table: a) instead of 1lllu, etc.,
Assyrian has ammiu, fem. ammitu, masc. plur. amm(i)utu, fem.
plur. amm(i)atu; b) Ugaritic does not offer any special forms for
the "far" demonstrative; c) it is very common in the Semitic
languages to use the personal pronoun as a demonstrative-or
rather to employ the same pronominal element ("anaphoric"
pronoun) for both functions: this occurs in Akkadian, Hebrew,
Syriac, and South Arabian.
13.33. It is clear from a comparison of the forms set out in the
preceding §§ that they cannot easily be reduced to Proto-Semitic
forms. Instead, it will be possible to identify certain formative
elements which make their appearance in various languages, either
in isolation or combined with each other. The most frequent of
those elements is the consonant rJ for the singular (taking into
account the phonetic changes which this consonant undergoes in the
various languages: § 8.14) to which 'l corresponds in the plural.
The antiquity of the element rJ is demonstrated by its occurrence
in Amorite (written "zii") and in Old Aramaic (z'). A component
of the "near" demonstrative is ha which appears in conjunction
with rJ in the Arabic series harJa, etc., as well as in the Syriac had(e),
hallen. In other languages the "near" demonstrative is formed
by adding to rJ the consonantal element n: thus in the Phoenician
dialect of Byblos zn, Old Aramaic znh (as well as z'), Biblical Aramaic
dana, South Arabian rJn, Ethiopic zantU, zatti (with regressive
assimilation), 'allontU and 'allantu (with the additional element -tu).
n combines also with ha in the Akkadian anniu, etc., and in the
Syriac han(a). According to Greenberg (JAOS 80 [1960], pp.317
to 321) the element n is characteristic of the masculine singular
The Pronoun
113
and of the plural, but not of the feminine whose characteristic
component is t: the series n/t/n appears to be typical of the deictic
element in the Hamito-Semitic area. The "far" demonstrative
includes, in the majority of the Semitic languages, the suffixed
consonantal element -k (often preceded in Arabic by l); the forms
which result from these combinations are generally rJk in the singular
and 'lk in the plural.
4. Relative Pronouns
13.34. The relative pronouns are connected in the majority of the
Semitic languages with the demonstrative ones, and more partic-
ularly with the consonantal element rJ. However, in the North-East
and North-West Semitic areas there are certain different forms of
the relative which are made up of the element S.
13.35. In Akkadian we have the following series for the most
ancient period:
Singular m.:
f. :
nom. su, gen. si, acc. sa
nom. acc. sat, gen. sati
Plural m. : sut
f.: sat
Dual sa
This series, which can be recognized as being formally connected
with the personal/demonstrative pronoun of the third person,
is reduced to the single form sa from the Old Babylonian period
onwards (only in rare cases have sut and sat survived).
13.36. In North-West Semitic, Amorite has SU (fem. si), while
Ugaritic has two forms: d personal and dt impersonal. In Hebrew
the forms connected with the demonstrative element rJ (zu, ze)
are rare; the usual forms are se-, sa- and 'aser (cf. asar in the
Tell Amarna glosses and ' sr in Moabite, while the occurrence of
dtr in Ugaritic is disputed); later on 'aser makes room for the shorter
form. In Phoenician the use of z (Byblos dialect) is also rare,
the usual form being s or, more frequently, 's. In the Aramaic
area (in accordance with the phonetic development of the conso-
I1foscati, Comparative Grammar 8
114 Morphology
nant g) the most ancient inscriptions show a relative pronoun zi
which later becomes di and, in Syriac, da-.
13.37. Old South Arabian has a masculine sing. g (used also for
the plural), a feminine gt, and a plur. 'lw (Qatabanian variants:
masc. gw, gm, fem. gtm). Classical Arabic has two series (cf. below,
§ 13.39):
Singular m. gu 'allagi
f. gatu 'allati
Plural m. rjawu,
,
ulu 'allagina, 'allii' i
f. rjawatu,
,
ulatu 'allati, 'alla'i
Dual m. rjawa 'allagani
£. gawata 'allatani
In the second series the element g IS preceded by the article
'al- and the infix -la-.
13.38. The Ethiopic relative is in the masculine singular za,
feminine singular 'anta, common plural 'alla.
13.39. The forms of the relative pronouns are used in some
Semitic languages for the expression of the so-called "determinative"
pronouns, conveying the meaning "that of ... ", "he of ... ", or
simply" of ... ": e.g. Akk. sa 1Jutari "he of the stick", i.e. the man
with the stick; sarru sa mati "the king of the region"; Ar. gu l-mali
"he of the money", i.e. the rich man, etc. The Arabic "determina-
tive" gu is fully declinable (nom. gu, gen. gi, acc. ga) and in this
respect differs from the indeclinable demonstrative gao
5. Interrogative and Indefinite Pronouns
13.40. The interrogative pronouns appear to go back to the
Proto-Semitic elements man in relation to persons (the form my,
peculiar to a large part of North-West Semitic, seems to be second-
ary from a comparative point of view) and rna for things. The
interrogative adjectives have the element 'ay in common. The forms
assumed in the various languages are as follows:
The Numeral 115
"Who?" "What?" "Which 1"
Akkadian mannu min1L ayyu
Ugaritic my mh mn(m)
fIebrew mi rna
,- -
f- ze
Phoenician my m
Biblical Aramaic man rna
Syriac man ma(n), mana 'ayna
Arabic rna 'ayy
un
man
South Arabian mn 'y
Ethiopic mannu mant 'ay
13.41. Observations on the above table: a) Amorite has manna
"who?" and rna "what?"; b) Ugaritic also possesses the element
'ay but it is generally employed as an indefinite pronoun ("any")-
in addition to the element mn( m) (cf. the following paragraph);
c) Hebrew preserves the element 'ay in its original form in the inter-
rogative adverb "where?", while it has numerous inter-
rogative adverbs composed of "which 1" and "where?",
"how 1" etc.); d) the Syriac interrogative adjective 'ayna has
a feminine 'ayda and a plural 'aylen in which the interrogative
element is reinforced by a demonstrative one; e) in Arabic man is
not declinable; 'ayy has a feminine 'ayyat, though the masculine
form frequently takes its place.
13.42. The interrogative forms are also used in the Semitic
languages as indefinite pronouns. The element ma may be placed
in apposition to nouns (so in Arabic, e.g. yaum
an
rna "on whatever
day") or be used as a reinforcing suffix to pronominal forms (so
in Akkadian, where the indefinite pronoun is *man-ma > mamma
for persons, *min-ma > mimma for things, and ayyumma as an
adjective; the component ayyum is declinable: ayyum-ma, ayyam-
ma etc.). Finally, Akkadian man-man > mamman produces an
indefinite form by reduplication of the interrogative.
D. The Numeral
1. Cardinals
14.1. The cardinal numerals in the principal Semitic languages
are as follows:
8*
116 Morphology
Akkadian U garitic Hebrew Syriac Arabic Ethiopic
1 m. iSten Mfd 'el;u"id 7;ad 'a7;ad 'a7;adu
f. istiat, istet a{i,t 'a7;at ?udii 'i{i,dii 'a{i,atti
2 m. Sina tnm s(J)nayim trnen 'itniini kgl'e(tu)
f. Sitta s(g)tayim tarten 'itnatiini, tintiini k<J1' eti
3 m. saliisat SJlosii tifliitii !alii!at salastu
f. salas tl1 siilos tJZiit !alii! saliis
4 m. erbet 'arbii'a 'arbr/ii 'arba'at 'arbii'tu
f. erbe, arba' u arb' (t) 'arba' 'arba' 'arba' 'arba'
5 m. lJ,amsat lJamiMa 7;amsa lJ,amsat lJ,amfJstu
f. lJ,amiS lJ,ms 7;iim?s 7;ammes lJ,ams lJ,amJs
6 m. seMet SiMii sta', 'e§tii sittat SfJMstU
f. [sess;} ] H S?S set sitt SfJSSU
7 m. sebet sib 'a sab'ii sab'at sab'atu
f. sebe sb'(t) seba' s9ba' sab' sab'u
8 m. [samanit] sfJmonii tfJmiinyii !amiiniyat samiinitu
f. samiine !mn sfJmone tfJmiine !amanin samani
9 m. tiSit tis'a tes'a tis 'at tfJs'atu
f. tise ts' tf)sa' tfJsa' tis' t9S'U
10 m. eseret 'asara 'esrii 'asarat 'asartu
f. eser 'sr
Oeser 'fJsar 'asr 'asfu
14.2. Observations on the above table: a) the forms given for
Akkadian are those of the absolute state (cf. § 12.79); b) "one"
fem. in Ethiopic might possibly be formed by analogy with the
feminine personal pronoun y3',Jti; c) for the vocalizations of "two"
in Hebrew and "six" in Syriac cf. § 10.2; d) regressive dissimilation
occurs in Syriac "two" *t3nen > t3ren; e) Ethiopic "two" is derived
from a different root which also appears in Akkadian kilalla!un,
Ugaritic kldt, Arabic kilani "both" and Hebrew kil'ayim; f) Hebrew
"five" is formed on the analogy of "six" (*lJamsa > lJi'imissa on
the analogy of sissa); g) Syriac "five" is formed on the analogy
of "four" (*lJ3me8 > lJammes on the analogy of 'arba'); h) Akkadian
"eight" has initials (samanu) instead of s which might be expected in
consonance with the other languages: analogy with the initial s of
"seven" (seMi) might be the cause; i) the numerals "one" and "two"
are adjectives, while the others are substantives; and by a singular
peculiarity, which must be regarded as Proto-Semitic, they are
used in the gender opposite to that of the noun which follows in
the genitive plural (e.g. Ar. 'arba'at
U
rigal
in
"four men"); this
inversion of gender also' operates when the numeral appears without
an accompanying noun. In Ethiopic the 'numerals with the ending t
The Numeral
117
are more frequently used, while in the Ugaritic cardinals from
"t "t "t "th f . h b
wo 0 en e orms WIt out -t may e employed with either
gender.
14.3. The numerals from "11" to "19" are normally formed
by the juxtaposition of the unit-numbers (with inversion of gender
from "13" onwards) and the numeral "10" which appears in some
variant forms: Heb. 'asar, fem. ' e 8 r ~ ; Syr. fem. 'esre; Ar. 'asara,
fem. 'asrata. In Arabic, moreover, all these numerals have the
fixed ending -a also in the unit-numbers. Examples: Hebrew "13"
masc. s3losa 'asar, fem.83los 'esre; Syriac "14" masc. 'arb3ta'8ar,
fem. 'arba's3re; Arabic "15" masc. l:yamsat
a
'asara, fem. hamsa
'a8rata. For the other languages the following observations ~ may
be made: a) in Ugaritic (see the statement in the preceding para-
graph) inversion of gender is not applicable; b) in Akkadian the
numerals from "II" to "19" are not attested except the forms
seMser "17" and 8amaneser "18" (masc.) and sinseret "12", l:yamisseret
"15", and samaneseret "18" (fem.); c) in Ethiopic the component
"ten" precedes that of the units and is joined to it by wa- "and",
while the gender of both components is inverted (e.g. "13" masc.
'asartu wa-sala8tu, fem. 'asru wa-salas).
14.4. It is generally held (Brockelmann, GVG, I, p. 490) that
in Proto-Semitic "20" is expressed by the dual of "10" (*'asra >
'isra by vowel dissimilation); that the numbers from "30" to "90"
are plurals of those from "3" to "9"; that in Akkadian, South
Arabian, and Ethiopic the ending of "20" is analogically extended
to the others, whereas in the remaining languages this is not the
case:
Akkadian Ugaritic Hebrew S yriac Arabic ESA Ethiopic
20 e8ra 'srm 'esrim 'esrin 'isruna 'S2
ry '3sra
30 salasa
tltm s3losim t3latin talatuna (tlty) salasa
etc.
Although this reconstruction enjoys a considerable degree of
probability, it has nevertheless recently been disputed by von
Soden (WZKM 57 [1961J, pp. 24-28) who holds that the Akkadian,
South Arabian, and Ethiopic forms are feminine plurals in the
absolute state. As regards Akkadian, it has to be realized that,
118
Morphology
owing to the adoption of the Sumerian sexagesimal system, its
numerals from "60" to "90" are not of the same type as those of
the other languages; "60" has the form susu or which may
be compared with that for "1/6" (of. § 14.10).
14.5. The numerals "100" and "1000" are clearly derived from
a common origin (with the exception of "1000" in Akkadian
which is connected with a word meaning "people"):
Akkadian U garitic Hebrew Syriac Arabic ESA Ethiopic
100 me'at mit
-, -
mea ma mi'at m't(m) m;/'CIt
1000 lim dlp 'elep 'alpa 'all 'll(m) ''CIll
(of. § 14.6)
14.6. Observations on the above table: a) the usual form for
"1000" is related to the noun meaning "ox"; b) Ethiopic ''CIll is
used for "10,000" ("1000" being expressed as 'asartu m'CI''CIt = "ten
hundred"); c) for higher figures of. Ugaritic rbt, Heb. ribbo, Syr.
rebbo "10,000".
2. Ordinals and Fractions
14.7. The ordinals from "first" to "tenth" are as follows:
Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic
1st ma!J,ru
rison qadmaya 'awwal qadami
2nd sanu #ni
tenyana fani kal'CI'
3rd sa18u tlt
s'CIlisi t'CIlitaya talit sal'CIs
4th rebu rb' r'CIbi'i r'CIbi'iiya rabi' rab'CI'
5th !J,amsu !J,ms 7:dimisi ly,'CImisaya lJiimis !J,am'CIs
6th sessu tdt
siMi s'CItitayii siidis sad'CIs
7th sebu sb' s'CIbi'i s'CIbi'aya siibi' siib'CI'
8th samnu tmn
sJmini tJminaya
t
amin sam'CIn
9th tisu
t'CIsi'i t'CIsi'aya tasi' tasJ'
lOth esru
'ii8iri ''CIsiraya 'asir
,_ v
aSJr
14.8. Observations on the above table: a) "1st" is formed on
a number of varying themes (in Akkadian another form is attested,
istiyu, though thIs is rare) ; the Ethiopic and Syriac forms are connect-
ed with a Semitic root denoting "precedence".
The other ordinals
The Numeral 119
are adjectives derived from the cardinals on the pattern qabir or
qabur in Akkadian, qabir in Hebrew and Syriac, qiibir in Arabic
and Ethiopic; b) in the U garitic documents hitherto discovered
ordinals "7th" and "8th" are attested in the feminine only;
c) In Hebrew, "6th" is formed from "6" (siMi, instead of *hdisi) on
the analogy of the cardinal SI}S, sissa; d) in Arabic, "6th" is siidis
(instead of *siidit) owing to progressive non-contiguous assimilation,
or-possibly-by analogy with !J,amis "5th"; e) in Ethiopic "2nd"
is derived from different themes: in addition to kal'CI' we have
ka'Jb and dagJm, but the normal pattern re-appears in the feminine
sanit with the meaning "the following day (or night)"; f) in Ethiopic
there also exists a parallel series with the ending -awi (qadamawi,
dagmiiwi, etc.); cf. § 12.23.
14.9. Above "tenth" there occur in Akkadian "11th" istenseru ,
"12th" sinseru, "13th" salasseru, "14th" erbeseru, "20th" esru,
"30th" selasu. Otherwise the forms of the cardinal numbers are
used. In Ugaritic no ordinals beyond "8th" have hitherto been
found. In Hebrew and Syriac there are no special forms beyond
"lOth"; thereafter the cardinal forms are used. In Arabic we have
"11th" ly,iidi 'asar
a
; and for the ordinals from "12th" to "19th"
the ordinal is always followed by from "20th" onwards
the cardinal forms are used. In Ethiopic 20th to 90th is expressed
either by the cardinal or by the addition of the suffix -iiwi (''CIsrawi,
salasawi, 'arb'CI'awi, etc.).
14.10. Fractions are generally formed on the pattern qubr: thus
in Arabic tult "a third", rub' "a quarter", etc., in Syriac tultii,
rub'ii, etc., and occasionally in other languages also (Akk. *sudsu >
SUSSll, "a sixth", Heb. ly,omes "a fifth"). Hebrew uses as fractions the
feminine forms ofthe ordinals (ly,amisit "a fifth" , etc.) ; a similarform-
ation occurs in Akkadian (rabitu "a quarter", sebitu "a seventh"),
while Ugaritic (whose vocalization is, of course, unknown) exhibits
feminine endings and a prefix m- (mtltt, mrb't, etc.). In other cases use
is made of forms which cannot readily be reduced to common pat-
terns. Ethiopic forms the fractions with the masculine and feminine
ordinals followed by ''CId "hand" (e.g. rab'CI''CIt ''CId "a quarter").
14.11 .. The distributive numerals are usually expressed by
a repetition of the cardinals, e.g. Heb. sJnayim s'CInayim "two by
two", Syr. s'CIba' sJba' "seven by seven" (for the use of the absolute
120 Morphology
state cf. § 12.79), Eth. 'aly,adu 'aly,adu "one by one", etc. Akkadian
has special forms: for "1" istena, for "2" sinnu, for "3"-"10"
the nominal pattern qubura' (sulusa, rubu'a, etc.).
E. The Particles
15.1. Under the term "particles" are subsumed (for the sake of
convenience rather than as a linguistically accurate classification)
adverbs, prepositions, conjunctions, and interjections. An analysis
of the particles often reveals a nominal or pronominal origin, but
there remain many cases in which such relationships cannot be
established.
1. Adverbs
15.2. In adverbs of nominal origin it is characteristic of Arabic
(and to a lesser degree of other languages as well) to use the accu-
. d' A' b dan" I "v'dd
an
" "yawm
an
satIve en mg: e.g. r. a a a ways, gt very,
"by day"; Akk. umam "by day"; Eth. nagha "in the morning";
etc. With this group are probably to be connected the Hebrew
adverbs in -am (yamam "by day", ly,innam "gratis", etc.). The
accusative ending also appears without the nunation (or mimation):
e.g. Ar. Ijabaly,a masa'a "morning and evening"; Akk. ma!Jra
"before", warka "behind" (already used in Old Babylonian and
Old Assyrian). Another characteristic formation, which is encoun-
tered in North-West Semitic, involves the feminine ending: Heb.
y(}hudit "in Hebrew", risana "at the beginning"; Syr. taba'it "well",
etc. Akkadian uses the ending -is for the formation of adverbs
(cf. § 12.66) and more rarely -atta and -um to which correspond
adverbial formations in -u in other languages (Ar. qablu "first",
Eth. tahtu "underneath", etc.: cf. §§ 12.65-66). Finally, a simple
or substantive may be used as an adverb (e.g. Heb. rab
"much", ya7}ad "together"); and this adjective or substantive
appears in the absolute state in those languages which have a
distinctive form for it (cf. § 12.79) (e.g. Akk. umakkal "one day",
Syr. sappir "beautifully").
15.3. Apart from the adverbs of demonstrably nominal origin,
there are others of some importance which have common roots
in Semitic. These include the demonstrative adverb of place
Heb. sam(ma), Syr. tamman, Ar. tamma "there"; the interrogative
adverb Akk. ayyanu, Heb. 'ayin, Syr. 'ayka, Ar. 'ayna,
The Particles
121
Eth. 'ayte the temporal interrogative Akk. mati, Heb.
matay, Syr. 'emmat, Ar. mata "when?"; the adverb of existence
(affirmative) Ug. Heb. Syr. 'it "there is", and the negative
form Ass. lassu, Ar. laysa, Syr. layt "there is not"; the negative
adverb Akk. la, Ug. l, Heb. la, Syr. la "not" as well as Akk. ul,
Ug.ul, Heb. 'al.
2. Prepositions
15.4. In some prepositions a nominal origin may be detected:
e.g. Heb. 'aly,ar "after" (and as a noun "back"), 'f}l}el "beside" (and
as a noun "side"), ben "between" (and as a noun "interval"), etc.
. 15.5. The more important prepositions are shown in the following
comparative table:
Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic
"in, by"
b b(}
!)(} bi ba
"to"
1 l(}
,,,
li la
"to"
1 'el 'ila
"like" ki(ma) k k(}(ma) 'ak ka kama
"over" eli 'l 'al 'al 'ala la'la
"from"
min men min '(}mna
"with" 'm
,.
tm 'am ma'a
"up to" adi 'd 'ad '"damma
15.6. Observations on the above table: blanks indicate either
missing forms or forms of different origin (Akk. ina "in", ana "to",
istu "from", itti "with" [for this latter cf. also Heb. 'et]; Ar. ly,atta
"up to" (but South Arabian 'd[y]); Eth. m(}sla "with", '"ska
"up to"; etc.).
3. Conjunctions
15.7. The principal conjunctions with common roots are as follows
(among the independent forms note especially Akk. summa "if"):
Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic
"and" u w w(} w(}
wa wa
"and, also"
(&)p 'ap 'ap fa
"or" u u 'a
,
aw 'aw 'aw
"if"
hm
,.
tm 'en
,.
tn '(}mma
"in order kima k
that"
ki kay kama
122
Morphology
4. Interjections
15.8. For the interjections various vocalic elements are used:
Akk. i, e; Reb. 'i, '8; Syr. '8; AI'. 'i, 'ay; Eth. '8. Another common
interjection is Akk. ennujam, Reb. hinne, Ug. hn "behold". Some-
times the imperative is employed as an interjection, without
verbal meaning: e.g. Reb. l ~ k "away!".
F. The Verb
1. Verbal Themes or Stems
16.1. The Semiticverbhasasetof themes or stems (0£. the following
paragraphs) in which formal changes correspond to certain semantic
variations and express different aspects of the action connoted by
the root. The semantic connexions may be somewhat fluctuating
and are not always readily identifiable, nor are all the stems attested
over the entire range of the Semitic languages. The linguistic
evidence brought to light during recent years (especially for
North-West Semitic of the second millennium B.C.) reveals a re-
markable wealth of forms in the most ancient phase of the Semitic
languages. In the course of time (particularly in the North-Western
region) numerous reductions have occurred-accompanied, at the
same time, by innovations and analogical restorations. For the
paradigm qbr of. § 12.3.
a) Simple Stem
16.2. This stem shows the three radicals in their simple form.
Variations in the vowel pattern relate to a distinction between
action and state. These variations are more clearly marked, as
regards their semantic relevance, in the South-West Semitic area
where, in the suffix-conjugation, the pattern a-a-a stands for an
action a-i-a for a transient condition, and a-u-a for a lasting con-
dition 'or state: e.g. AI'. na?ara "he looked at", salima "he was well",
hasuna "he was beautiful". The antiquity of this threefold vocalic
~ c h e m e in Arabic is confirmed by some of the oldest manifestations
in North-West Semitic, i.e. Amorite, Ugaritic, and the Tell Amarna
glosses. In the prefix-conjugation the variation in the second vowel
is at least partly paralleled: u or i corresponding to a, and a to i,
while u generally remains: e.g. AI'. yan?urn, yaslamu, yal,/'snnn.
The Verb 123
In Akkadian the distinction between active and stative verbs is
less pronounced as far as vowel variation is concerned, yet it remains
generally identifiable. In stative verbs the distinctive vowel i
predominates in the prefix-conjugations (pres. ikabbit "he becomes
heavy", pret. ikbit "he became heavy"), while a is rarer-though
ancient and attested in Old Akkadian and Old Babylonian (e.g. iqrab
"he approached", later iqrib). In active verbs a is more common in
the present tense and u in the preterite (e.g. pres. isakkan "he
puts", pret. iskun "he put"); at times, however, a is found in the
preterite as well (e.g. pres. ilammad "he learns", pret. ilmad "he
learnt"); i appears in cases where the action of the verb is regarded
as momentary (e.g. idallip "he disturbs") as well as in some verbs
of motion (e.g. ittiq "he passes"); u occurs in all cases where the
action is intransitive (e.g. irappud "he runs"). In the stative,
which formally corresponds in Akkadian to the West Semitic
suffix-conjugation (cf. § 16.38), active verbs have i as their second
vowel (e.g. sakin "he is placed"), while stative verbs may have
a or u as well, in accordance with the vowel of the corresponding
adjective (e.g. l}alaq "he is lost", alongside l}aliq).
16.3. The passive is formed on the vowel pattern u-i-a in the
suffix-conjugation; it is in full use in Arabic (e.g. kataba "he wrote",
kutiba "it was written") where the prefix-conjugation in the passive
has its own vowel scheme as well (e.g. yalctubu "he writes", ynktabu
"it is written"). There is a purely formal coincidence here with the
vowel distribution in the derived stem with prefix' -. Apart from
Arabic, a passive of the simple stem exists in Ugaritic (where, in
the prefix-conjugation, it cannot however be readily distinguished
from the passive stem with prefix n-; cf. Gordon, UM, p. (5), in
the Tell Amarna glosses, and in Rebrew, where in the suffix-
conjugation it formally coincides with the stem with doubled
second radical (*qubar > qnbbar; of. § 10.8 d) and in the prefix-
conjugation with the stem with prefix h- (yoqbar). It is also possible
that there are some traces of this passive in the Aramaic of the Arpad
inscriptions and in Biblical Aramaic where the second vowel is
lengthened and the form thus coincides with that of the participle
(e.g. kJtib "it was written" as well as "written"); but these forms
may, in fact, be original participles whose functions have been
extended by analogy. According to Petracek's recent studies on
124 Morphology
inner flexion (cf. § 11.3), the internal passive (which is wanting in
Akkadian) is to be regarded as a secondary development of West
Semitic.
b} Stem with Doubled Second Radical
16.4. This stem, which is attested over the whole Semitic area,
seems to have a primarily "factitive" significance, i.e. as a causative
in relation to a state or condition: e.g. Akk. iblut "he lived",
uballit "he made to live"; Syr. l;()san "he was strong", l;assen
"he strengthened". To this meaning-aspect must be added the
denominative one (e.g. Syr. k()Ula "crown", kallel "he crowned")
and the intensive aspect (e.g. Ar. kasara "he broke", kassara "he
shattered"; Akk. ibtuq "he cut", ubattiq "he cut to pieces").
16.5. Arabic possesses in this stem, too, a distinction between
active and passive, brought about by a change of vowel pattern-
in the same manner as shown in the simple stem: e.g. qattala "he
massacred", quttila "he was massacred"; yuqattilu "he massacres",
yuqattalu "he is massacred". The same passive exists in Hebrew-
on the assumption that the Pu'al form (qnbbar) owes the a of its
second syllable to analogy with the prefix-conjugation (y()qnbbar).
c) Stem with Lengthened First Vowel
16.6. This stem seems to have primarily reciprocal significance,
i.e. to indicate an action accomplished together with another
person: e.g. Ar. kataba "he wrote", kataba "he corresponded".
At times this stem also indicates an action directed towards an
object as well as an attempt to accomplish something (conative):
e.g. Ar. qatala "he killed", qatala "he fought" (= "tried to kill").
16.7. This stem is attested in Arabic and, though less frequently,
in Ethiopic,' in the latter the correlation between form and semantic
value has been largely lost (e.g. saqaya "he tormented", wal}aya
"he visited"). Traces of this stem in North-West Semitic are
rather dubious (cf. the discussion in Garbini, SNO, pp. 126-34).
It may therefore be concluded that this stem is typical of South
Semitic.
16.8. Again, Arabic makes use of a variation in vowel pattern
to express the distinction between active and passive-on the
The Verb 125
lines of the vocalization scheme applied to the previous stems,
but with the first vowel long as the distinctive mark: e.g. qabara,
pass. qubira; ynqabirn, pass. yuqabarn.
16.9. A variant of the stem with long first vowel is that with
a diphthong: this is a development of which there exist very few
traces in North Semitic (e.g. Syr. gawzel "he set fire to") but more
ample ones in Ethiopic (qobara, qebara: cf. Dillmann, EG, pp.146 to
147, § 78) and in Arabic, especially in modern Arabic (e.g. gawraba
"he put on socks") in mainly denominal roots (cf. Brockelmann,
GVG, I, pp. 514-15).
d) Stems with Prefixes s-, h-, '-
16.10. The Semitic languages present a series of stems with
prefix s- or h- or '-, all sharing a causative connotation: e.g. Ug.
ll;m "to eat", sll;m "to cause to eat" = "to feed". The causative
may refer to a state or condition, and in such cases it may coincide
with the "factitive" of the stem with doubled second radical: the
two themes are then used alongside each other without appreciable
distinction (e.g. Akk. kunnusn and snknnSn "to subdue"). Another
aspect of the causative is its declarative value (e.g. Heb. hi?diq
"he pronounced just", from the root ?dq, Eth. 'amsala "he pro-
nounced similar", from the root msl). Finally, the causative may
have intransitive significance, i.e. in cases where the action remains
attached to the subject (e.g. Akk. snlbnrn "to grow old", Heb.
hismin "he grew fat", Ar. 'aqama "he remained"). South Semitic
uses this stem widely for denominative verbs (Ar. 'al;sana "he did
well" from l;asan "beautiful", 'af?al;a "he was eloquent" from
fa?il; "eloquent"; Eth. 'asgala "he divined" from sagal "divination",
'ab'ala "he feasted" from ba'al "feast").
16.11. Of the three stems, that with prefix s- (also found outside
Semitic in Egyptian: e.g. 8 . n ~ n "to bring up" from a root n ~ n
"to be a child") occurs in Akkadian, in Ugaritic, and in the South
Arabian dialects (here S2 > Sl) with the exception of Sabaean:
e.g. Akk. nsamqit "he caused to fall", from mqt; Ug. dshlk "I cause
to flow" from hlk; ESA sl'db "he caused to place", from 'db. The
same prefix is attested in Aramaic (e.g. Bibl. Aram. saklilU "they
completed" from kll; Syr. sa'bed "he enslaved" from 'bd); it appears
also in Arabic and Ethiopic (as well as in Amorite and, in a few
126 Morphology
surviving traces, in other North-West Semitic languages) in
combination with the infix -t- (cf. below, § 16.21).
16.12. The stem with prefix h- occurs in Amorite, in the Tell
Amarna glosses (alongside the less frequent one with s-), in Hebrew,
Moabite, Old Aramaic, Sabaean, and in the most ancient phases
of ,!,amudic and LiJ:lyanite: e.g. Heb. hiqdis "he consecrated"
from qds; Bibl. Aram. hanpN "he caused to go out" from npq;
Sabaean hrtr' "he subdued" from rtr'; LiJ:l. hawdaq "he offered"
from wdq. A few remnants of this prefix survive in classical Arabic
(e.g. haraqa alongside 'araqa "he poured").
16.13. The stem with prefix '- appears in the most recent phase
of Aramaic, in pre-Islamic North Arabic, in classical Arabic, and
in Ethiopic: e.g. Syr. 'albes "he clad" from lbS; LiJ:l. 'awdaq, variant
of hawdaq (of. preceding paragraph); Ar. 'aqtala "he caused to kill"
from qtl; Eth. 'astaya "he gave to drink" from sty. It is possible
that this causative pattern might be detected in some disputed
cases in Ugaritic (of. Gordon, UM, p. 6S). Developments in Aramaic
and Arabic, together with the fact that the prefixes h- and' - are
not found simultaneously in the various languages, suggest the
possibility that both go back to one original theme whose prefix
h- later became' -. There is evidence, moreover, of the existence of
a fourth variant of the prefix, apparently of secondary origin, in
the Phoenician form yqbr (suffix-conjugation): e.g. yqds "he
dedicated" .
16.14. For the causative stem, too, Arabic possesses specific
vowel patterns which relate to the distinction between active and
passive (the vocalization is the same as in the preceding stems):
'aqbara, pass. 'uqbira; yuqbiru, pass. yuqbaru. The same pattern
and meaning are represented by the Hebrew and Biblical Aramaic
stem Hophal (hoqbar)-again on the assumption that the a of the
second syllable is due, as it appears to be, to analogy with the
prefix-conjugation (yoqba1'); the 0 of the first syllable has to be
explained in terms of § 10.S e above.
e) Stem with Prefix n-
16.15. This stem has passive and reflexive meaning. It is attested
over the entire Semitic area (with some traces in Egyptian) with
The Verb 127
the exception of Aramaic. In Ethiopic it is rare but occurs in some
quadriradical verbs. Examples: Akk. naprusu "to be separated",
root prs; Heb. ni8'al "he was asked", root s'l; Ar. 'inqafa'a "he
was cut to pieces", root qt'. In Akkadian this theme adopts in part
the vowel distribution of the simple stem (of. § 16.2 and von Soden,
GAG, p. US); with stative verbs its meaning is predominantly
ingressive: e.g. ibassi "he is", ibbassi "he becomes"; nasa'um "to
carry", nansum "to shoulder". In Ugaritic this stem is attested
hut the n is almost invariably assimilated to the following consonant
(of. however nkbd "honoured", root kbd). In Ethiopic-as has been
. mentioned-this stem appears with some quadriconsonantal verbs,
e.g. 'anfar'a?a "he jumped"; from the semantic point of view,
however, Ethiopic shows a development towards a causative
connotation which is, perhaps, connected with the formal identity
of the prefixes (Brockelmann, GVG, I, p. 536).
16.16. Arabic again has variations in the vowel scheme, but in
the present case the expression of the passive becomes effective
only in those instances where this aspect is absent in the normal
theme: 'inqabara and 'unqubira, yanqabiru and yunqabaru.
f) Stems with Prefix (or Infix) t-
16.17. Of all the stems dealt with above under a to d, additional
themes may be formed with the prefix (or infix) t-, producing
reflexive, passive, and sometimes also reciprocal connotations
(for further meaning variants in Akkadian of. von Soden, GAG,
p. 121).
16.18. t- is prefixed to the simple stem in the Aramaic languages
(e.g. Syr. 'etqfel "he was killed", root qfl) and in Ethiopic (e.g.
ta'asra "he was bound", root 'sr); it changes place with the first
radical, and is thus infixed, in the other languages in which it is
attested: in Akkadian (e.g. mitl;uru "to meet", root ml;r), in Amorite
(cf. the proper name Yabtaly,arna), in Ugaritic (eg. yrtly,? "he washes
himself", root rly,?) , in Phoenician (e.g. thtpk "she is being over-
thrown", root hpk), in Moabite (e.g. 'ltly,m "I am fighting", root
lly,m) , and in Arabic (e.g. 'iqtatala "he fought", root qtl).
16.19. In the stem with doubled second radical the metathesis of
t- with the first radical takes place in Akkadian only: e.g. ustallamu
128 Morphology
"they are being kept safe", root slm. In Hebrew and in Biblical
Aramaic the initial t takes a further prefix h- in the suffix-conju-
gation, possibly by analogy with the causative with prefix h-:
Heb. hitqaddf}s "he sanctified himself", root qds; Bibl. Aram.
hitnaddabil "they made a voluntary offering", root ndb. Examples
in other languages: Syr. 'etly,assan "he was fortified", Ar. takassara
"he was shattered", Eth. taqaddasa "he was sanctified". A notable
phenomenon of Hebrew and Aramaic is the metathesis of the prefix
t- before a dental or a palato-alveolar fricative (cf. § 9.22). Note-
worthy in Akkadian is the Neo-Assyrian formation with reduplica-
tion of infixed -ta-: e.g. uktata§§ar "he will be equipped".
16.20. The prefix t- produces a new stem in Arabic and Ethiopic
when joined to the theme with first vowel lengthened: e.g. Ar.
taqatalil "they fought together", root qil; Eth. tamasalil "they
resembled each other", root msl.
16.21. The stem with prefix s- brings about the metathesis (cf.
§ 9.22) of this prefix with t-. The theme is common in Akkadian
where it presents two types which differ in the forms of the present
(ustaqbar and ustaqabbar-for ust- > ult- cf. § 8.32); the type ustaqbar
has the function of a passive of the s- stem (e.g. ustalpat "it will
be destroyed", root lpt), while the type ustaqabbar has various and
not yet fully explored connotations, including that of a causative
of the simple theme with t- (e.g. sutam7}uru "to cause [numbers]
to correspond with one another", root m7}r) and that of an inner
passive of stative verbs (e.g. sutamru§u "to endeavour", root mr§).
This stem also exists in Arabic (e.g. 'istaqtala "he exposed him-
self to death", root qil) and in Ethiopic (e.g. 'astamly,ara "he showed
himself merciful", root mly,r). It is also attested, though rarely
and with certain doubts and reservations, in North-West Semitic
(e.g. Ug. tstly,wy "she prostrates herself", Heb. histaly,awa "he
prostrated himself", Syr. 'estawdi "he promised ,confessed"; and al-
ready in Amorite proper names of the type Batasni-' Il, Yistasni-' Il).
16.22. In the stem with prefix' - the combination with t- produces
a special theme in Aramaic: e.g. Syr. 'ettrim "he was raised",
root rwm (with assimilation t' > ttl.
16.23. For all these stems (except the last which is attested in
Aramaic only) Arabic has the usual variations in vowel pattern,
The Verb 129
but the effective expression of the passive is contingent on a formal
and semantic contrast, i.e. it is realized only when the passive
meaning is wanting in the ordinary form of the stem: 'iqtabara and
'uqtubira, yaqtabiru and yuqtabaru; taqabbara and tuqubbira,
yataqabbaru and yutaqabbaru; taqabara and htqilbira, yataqabaru
and yutaqabaru; 'istaqbara and' ustuqbira, yastaqbiru and yustaqbaru.
g) Other Stems
16.24. In addition to the basic stems set forth above, certain
secondary and rarer types occur in the various languages. One such
case is the theme classified as No. IX in Arabic, with a perfect of
the type 'iqbarra which is used for verbs indicating colours, physical
defects, etc.: e.g. 'i§farra "he was yellow", root §fr. A variant of
theme IX is stem XI: 'iqbarra. In the other Semitic languages,
some forms corresponding to the stem with repeated third radical
occur in Ethiopic (e.g. bardada "he covered with stones", galbaba
"he wrapped"), in Syriac (e.g. 'abded "he enslaved") and perhaps
in Akkadian (e.g. utnennu "to pray"-cf. Kienast's recent studies).
There also exist a few cases where the first radical re-appears after
the second (e.g. Ar. farfaba "he called [camels]", Syr. qarqes "he
shook"). Repetition of the second radical is a common feature in
the modern Ethiopian languages, generally with iterative or
intensive meaning (e.g. Amh. sababara "he smashed", Tigrifia
qatatala "he slaughtered").
16.25. The extension of originally biradical roots gives rise in
West Semitic, both Northern and Southern, to quadriliteral themes
of the type qabqab, etc.: e.g. Heb. gilgf}l "he rolled", Syr. balbel
"he confused", Ar. zalzala "he shook", Eth. badbada "he deva-
stated".
16.26. It appears to be a characteristic tendency of Akkadian
and Ethiopic to form further stems by a combination of those
listed above. In Akkadian we have a series of additional forms with
the infix -tan-, possessing iterative meaning, inserted in the simple
theme (pres. iqtanabbar), in that with doubled second radical (pres.
uqtanabbar) , in that with prefix s- (pres. ustanaqbar) , and in that
with prefix n- (pres. ittanaqbar). An Old Aramaic form evidently
due to Assyrian influence is htn'bw, a Hittanaphal of y'b( ?), which
appears in isolation in the inscription of Bar Rakib. Another
l\ioscati, Comparative Grammar 9
130 Morphology
stem typical of Akkadian is that resulting from the combination
of the prefix s- and the doubling of the second radical (pres.
usqabbar): this is a poetical form used for either of the two themes
which coalesce in it. As for Ethiopic, it forms a stem with doubled
second radical and one with lengthened first vowel from the theme
with prefix' - and from that with the (already compound) prefix st-;
the entire system in Ethiopic looks as follows:
1.1. qabara I. 2. ·qabbara I. 3. qabara
II. l. 'aqbara II. 2. 'aqabbara II. 3. 'aqabara
III. 1. taqab(a)ra III. 2. taqabbara III. 3. taqabara
IV. 1. 'astaqbara IV. 2. ' astaqabbara IV. 3. ' astaqabara
Finally, the combination of more than one stem is widely attested
in the modern Arabic and Ethiopian languages (cf. Brockelmann,
GVG, I, pp. 540-43).
h) Verbs with Four and Five Radicals
16.27. In all Semitic languages we encounter, to a greater or
lesser extent, a number of verbs with four radicals; in Ethiopic
there are also a few with five radicals. As there are virtually no
common Semitic roots among these verbs, we must consider them
innovations in the various languages. All verbs in this category
possess only a fraction of the stems and forms of the triradical
verb. The following principal types are attested:
a) In Akkadian alone we find verbs of the type with prefix s-:
suqammumu "to be dead-silent", suparruru "to expand", sukenu
"to prostrate oneself". Their morphological structure is irregular;
there are no derived stems-apart from the t- stem.
b) In Akkadian and Ethiopic we find a group of verbs belonging
to the n- stem. The Akkadian verbs in this category all exhibit
either l or r as their second radical (e.g. naballcutu "to pass over",
naparqudu "to lie on one's back"). The n- stem takes the place of
the simple stem, while the s- stem serves as causative; both of them
form iterative tan- stems (e.g. ibbalak7cat "he passes", iterative
ittanablakkat; causative usbala7ckat, iterative ustanablakkat). The
Ethiopic verbs of this type do not form a causative, and a passive
with t- prefix occurs only rarely. Examples:' anfabraqa "it shone",
The Verb 131
'anzahlala "he languished" (with "weak" second radical), 'anfaffafa
"it dripped" (with reduplication).
c) Related to b are Arabic roots with n infix after the 2nd
radical (e.g. 'ibransaqa "he flourished") or with doubled 4th radical
(e.g. 'ismalJarra "he was very high").
d) Reduplicated roots (e.g. Ugaritic mgmg "to mix", Hebrew
gilg(}l "he rolled"); geminated roots (e.g. Syriac 'abded "he
enslaved", Ugaritic fly,rr "to burn"); denominative verbs (e.g.
Arabic basmala "he said bismillahi"); original triradicals with
added 2nd (mostly I, r, n) or 4th radical (e.g. Arabic 1;albasa "he
enticed"-cf. 1;alaba); verbs developed from causative stems (e.g.
Syriac saklel "he completed"-cf. Akkadian suklulu). These verbs
form reflexive-passive t- stems as well as inner passives in Arabic,
Hebrew, and Old Aramaic, while 'a- causatives and causative-
reflexive 'asta- stems are limited to Ethiopic (e.g. 'adanga?a "he
confused", 'astasana'awa "he pacified").
e) Ethiopic verbs of five radicals are formed from triradicals by
the repetition of the last two radicals (e.g. 'aly,malmala "it became
green", 'arsaly,s'J/y,a "he sullied").
2. The "Tenses"
16.28. The "tense" system presents one of the most complicated
and disputed problems of Semitic linguistics. In the West Semitic
area, Arabic and most of the other languages exhibit, according
to the traditional approach, two conjugations which are usually
called "tenses". But this nomenclature must be considered im-
proper, as different temporal concepts converge in each of these two
conjugations; it would be more appropriate to speak of "aspects".
One of these conjugations uses prefixes (type yaqburu: the third
person masculine singular is cited in accordance with accepted
practice) and generally indicates an incomplete action which
corresponds, according to circumstances, to our future, present, or
imperfect. (The prefixes, which are the distinctive element of
this conjugation, are in some instances supplemented by suffixes
having the function of contrast and identification of the various per-
sons.) The other conjugation employs suffixes (type qabara: qabarat,
qabarta, etc.) and generally indicates a completed action which
9*
132 Morphology
corresponds, according to circumstances, to our past tenses. The
two conjugations are usually called "imperfect" and "perfect",
respectively, in the etymological sense of these terms.
16.29. East Semitic (Akkadian) presents a system of several
conjugations: one with prefixes for incomplete action (type iqabbar),
called "present"; another, also with prefixes, has a different vowel
and syllable distribution (inner morphemes) and connotes completed
action (type called "preterite"; and a thircl one with suffixes
(type qabir), called "stative". This last type represents in essence
the conjugation of a noun and may constitute a verbal adjective
(e.g. damiq "he is good", baltalcu "I am alive") as well as a sub-
stantive (e.g. zikaraku "I am a man", from zikaru "man"). Finally,
there is the recent tendency (von Soden, GAG, pp. 104-105, and
before him Landsberger and other scholars) to detect in Akkadian
the existence of a fourth conjugation with infix -ta- (type iqtabar),
called "perfect", which expresses an action complete in itself but
still persisting in its effects (or subsequent to another completed
action). This conjugation must be regarded as an Akkadian
innovation.
16.30. Traditional Semitic grammar was inclined to consider the
Arabic situation as original and that in Akkadian as secondary.
More detailed study of the Semitic languages as well as Hamito-
Semitic comparisons have shown, however, that the position is
more complex: in several instances in West Semitic, both Northern
and Southern, we have discovered elements of a distinction between
two prefix-conjugations, and these findings have been corroborated
by evidence furnished by Hamitic languages, in particular by
Libyco-Berber. In U garitic the existence of two differentiated
conjugations on the consonantal pattern yqbr might be suggested
by the fact that the same pattern appears to indicate both completed
and incomplete action, while the qbr conjugation has notable points
of contact with the Akkadian stative (Goetze). In the Tell Amarna
glosses three conjugations have been identified: one of the yiqbur
type for completed action, one of the yiqab(b)ar type for incomplete
action, and a third form qabajijur for completed action (correspond-
ing to the stative n. In Hebrew the conjugation of the yiqbor
type is employed· not only for incomplete but also for completed
action-as is shown particularly by the use of the conversive w.
The Verb 133
Moreover, one cannot exclude the existence of forms of the type
yiqabbar (Meyer, Rossler), and, in fact, forms like Y3qabber, usually
held to belong to the stem with doubled second radical, may
instead reflect a yiqabbar conjugation of the simple theme (Lands-
berger). In Ethiopic the conjugations Y3qabb3r and Y3qb3r (intrans.
Y3qbar) , which now relate to the distinction between indicative
and subjunctive, may originally have possessed a "tense" con-
notation not very different from the type of semantic contrast
observed in Akkadian. In addition Ethiopic has a suffix conju-
gation, the so-called "gerund", which is formed on the pattern
qabir (cf. § 16.70) with pronominal suffixes-rather like the Akkad-
ian stative. In Arabic, on the other hand, we are unable to pene-
trate to a stage preceding the considerable measure of systematiza-
tion to which the language has been exposed. We have also in-
sufficient data for Amorite where, formally at least, we encounter
the opposition qabar: yaqbur, yet we are not in a position to form
any reliable judgement as to the semantic significance of this
opposition. Finally, Hamitic languages present two distinct
prefix-conjugations, a "preterite" and a "present" (or "habitual"
form or "continuative") which reveal definite points of similarity
with the Akkadian verb: compare Akkadian ipru8 and iparra8
with Libyan ifre8 and ifarre8 (Rossler).
16.31. These considerations have brought about a crisis in the
conventional conception of the primary character of the Arabic
system and have stimulated vigorous scholarly discussion (Brockel-
mann, Cohen, Driver, Fleisch, Klingenheben, Kurylowicz, Meyer,
Rossler, Rundgren, von Soden, Thacker, and others). Without
entering into the details of this debate, which is still unresolved
(0£. the bibliography), it now seems safe to say that the Arabic
"tense" system represents the result of a long process of evolution.
Proto-Semitic possessed almost certainly a nominal suffix-con-
jugation (surviving in the Akkadian stative and Ethiopic "gerund")
which in West Semitic has evolved into a verbal conjugation-yet
without differentiation of mood, which might well be a pointer
to its origin outside the verbal system; it is likely to have been the
function of this suffix-conjugation to record a state or condition
and to describe it as having been accomplished. As for the prefix-
conjugations, some scholars maintain that only one of them is
134 Morphology
to be attributed to Proto-Semitic (Cohen). It might have had
the function of indicating action in contrast to state or condition-
without distinguishing between completed and incomplete action.
This latter distinction would have been realized only subsequently
in the historical development of individual Semitic languages-and
in a number of different ways. In West Semitic the prefix-con-
jugation appears to have been set apart for the designation of an
incomplete action in opposition to the suffix-conjugation which
developed into the expression of completed action. In East Semitic
(Akkadian) the prefix-conjugation might have continued in use for
both completed and incomplete action but subsequently evolving
into two types by means of vocalic and syllabic reconstitution
(iqabbar arises secondarily alongside iqbur by a functional re-
assignation of the intensive ?-thus Rundgren). The two different
aspects of action reside in the contrast between these two forms
which remain opposed, as a group, to the suffix-conjugation which
is retained for the designation of a state or condition. To postu-
late the existence of only one prefix-conjugation in Proto-
Semitic is considered by some scholars an inadequate solu-
tion-nor does its indeterminate character as regards tense
commend itself to them. This problem cannot be separated from
that of the moods (cf. §§ 16.32-36), because some have detected
in the West Semitic "jussive" yaqbur a development of the Akkad-
ian preterite iqbur; conversely, the Akkadian subjunctive or
"relative" mood (iqburu) has recently been regarded as the ancestor
of the West Semitic prefix-conjugation yaqburu (Kienast). Mention
has already been made of the hypothesis claiming a secondary
origin of the Akkadian form iqabbar by means of a redesignation
of the stem with geminated second radical, but the view has also
been advanced that iqabbar was dropped or restricted in use in
West Semitic on account of its formal identity with the imperfect
of the geminated stem. A somewhat singular position in the recon-
struction of the Semitic "tense" system is at present held by von
Soden who posits three prefix-conjugations: a preterite yaqbur,
a "momentary aspect" yaqburu and a durative present yaqabbar.
3. The Moods
16.32. A full range of moods in the "imperfect tense" (as regards
the "perfect" cf. the observations in the preceding paragraph)
The Verb 135
is attested in Arabic, where the moods are being expressed by
means of differences in the endings: indicative yaqbur-u, sub-
junctive yaqbur-a, jussive yaqbur, energic yaqbur-an(na} .. Whether
this range of moods could be attributed to Proto-SemItIc, was ra-
ther difficult to determine when Brockelmann wrote on this
question (GVG, I, p. 554). Nowadays remarkable corroboration
of this modal variety has been furnished by Ugaritic which places
itself alongside Arabic with the same set of endings-recognizable
by the vocalisation of ' (though the severely defective
leaves the distinction between the two forms of the energlC 111
doubt; some reservations have also been expressed as regards the
Ugaritic subjunctive in -a: of. Garbini, SNO, p. 144). In other West
Semitic languages (as will be shown in the following paragraphs)
some remnants of moods have been discerned which agree wholly
or in part with the evidence furnished by Arabic and Ugaritic. As
will be seen in § 16.34, the North-West Semitic documentation
suggests a semantic development of the subjunctive into a cohorta-
tive; the differences vis-a.-vis East-Semitic remain considerable.
16.33. In Akkadian the modal system shows a remarkable
divergence from West-Semitic. In the first place, the moods are
expressed not only in the imperfect but in all the "tenses". Secondly,
the endings differ from those in the other languages: the indicative
has none, and the subjunctive or "relative" mood (whose functions
are different from those of the subjunctive elsewhere: cf. von
Soden, GAG, p. 108) has -u (Assyrian -uni); the other mood attested
is a so-called ventive in -am. It should be noted, however, that in
a group of Old Akkadian texts the suffix of the subjunctive appears
as -a (Gelb, OA, pp. 170-71) which would tally with the Arabic
and Ugaritic ending (but B. Kienast, in Or 29 [1960], pp. 152-53,
thinks that the supposed subjunctive in -a is, in fact, a ventive
without mimation).
16.34. In West Semitic, modal differentiation is limited to the
imperfect. In North-West Semitic (leaving aside Ugaritic-see
§ 16.32) Amorite presents yaqbur and yaqburu, but a modal distinc-
tion cannot be determined; in the Tell Amarna glosses we encounter
a volitive in -a and an energic in -na. The Hebrew documentation
is less relevant because the shedding of final vowels includes the
modal morphemes. In addition to the indicative there remains
136 Morphology
a jussive which is characterized (but not in all cases) by vowel
reduction: e.g. imperfect yiiqum "he rises", jussive (way- )yiiqom.
There also exists an energic or cohortative in -ii, used chiefly in the
first person (e.g. 'eqtJlii "let me kill"). In the Dead Sea texts the
forms with ending -ii are used for the simple indicative as well.
The element n appears before suffixes, although it does not seem
to carry energic value (e.g. yiqqiily,ennu "he takes him"); it cannot,
therefore, be identified with certainty as the same morpheme.
In the consonantal spelling of Phoenician the distinction between
indicative and subjunctive is expressed in the third person plural
by the presence of -n in the indicative and its absence in the sub-
junctive. Modal differentiation may be said to have been entirely
discarded in Syriac; but in more ancient Aramaic dialects, i.e. in
Egyptian and Biblical Aramaic, we find-as in Phoenician-a dis-
tinction between indicative and jussive based on the presence or
absence of -n in the third person plural; later on, the forms with
-n predominate. The element n, which appears before suffixes and
is attested in the late phases of Aramaic (except for a Eolitary
instance in the Zkr inscription), seems unlikely to possess energic
connotation (cf. Hebrew above).
16.35. In South-West Semitic the position of classical Arabic
has already been dealt with. That of South Arabian and of pre-
classical Arabic is probably similar, but the absence of vocalization
allows only the identification of the energic morpheme -no Ethiopic
distinguishes two moods, the indicative and the subjunctive, by
means of thematic variants: indicative YJqabbJr, subjunctive YJqb'Jr,
subj. of intransitive verbs YJqbar (0£. § 16.30 for a comparison of
the Ethiopic forms with the Akkadian prefix-conjugations).
16.36. Finally, all the Semitic languages have an imperative
which, in the simple theme, has the vowel pattern characteristic
of that stem (0£. § 16.2). The form of the imperative generally
corresponds to that of the prefix-conjugation (short form) without
its prefixes; any departure from this rule is due to the appearance
of prosthesis or anaptyxis as a consequence of consonantal clusters
in initial position (cf. §§ 9.14-17): e.g. Akk. prefix-conj. iqbur,
imperative q1lbur, Heb. prefix-conj. yiqbor, imperative qJbor (but
prefix-conj. yikbad, imperative kJbad), Eth. prefix-conj. (sub-
junctive) YJqbJr,· imperative qJbJr, etc. In some North-West
The Verb 137
Semitic languages (Ugaritic, Tell Amarna glosses, Hebrew) we also
find an imperative (masculine singular) with the further ending -ii
which is probably the cohortative element previously referred to
(§16.34).
4. Inflexion
a) Simple Stem: Suffix-Conjugation
16.37. Semitic verbal inflexion is effected by means of personal
prefixes and suffixes, probably of pronominal origin (as shown
by their external form). The suffix-conjugation of the simple
stem is inflected in the principal Semitic languages as shown
in table 1. The following paragraphs offer some general observations
on this table:
1. Simple Stem: Suffix - C 0 nj ug a tion
Akkadian U ga- Hebrew Syriac Arabic
ritic (act.) (pass.)
Sing. 3 m. qabir qbr qabar qJbar qabara qubira
f. qabrat qbrt qabfJra qebrat qabarat etc.
2 m. qabrata qbrt qabarta qfJbart qabarta
f. qabrati qbrt qabart qabart qabarti
1 qabraku qbrt qabarti qebret qabartu
PI. 3 m. qabrii qbr qabarii qabar(iin) qabarii
f. qabra qbr qabarii qfJbar(en) qabarna
2 m. qabratunu qbrtm qfJbartem qfJbart6n qabartum(u)
f. qabratina qbrtn qfJbarten qfJbarten qabartunna
1 qabranu qabarnii qabarn(an) qabarna
Du. 3 m. (qabra)
f. (qabirta)
2
qbr
qbrt
qbrtm
qbrny
qabara
qabarata
qabartuma
Ethiopic
qabara
qabarat
qabarka
qabarki
qabarkii
qabarii
qabara
qabarkammu
qabarkfJn
qabarna
16.38. a) The Akkadian stative (if account is taken of the con-
necting vowel -ii-, which characterizes its second and first persons,
and of the consistent suppression of the second vowel) corresponds
in its endings to the West Semitic perfect (see however below
apropos of each individual person). There are some noteworthy
Assyrian variants: qabriiti for qabriita (which occurs also in Old
Babylonian) and qabriini for qabriinu.
138 Morphology
16.39. b) North-West Semitic of the first millennium has carried
out consistently the characteristic changes connected with the
incidence of stress (cf. §§ 10.8, 10.10). This has entailed, for
Hebrew, the shedding of final short vowels (qabar), the lengthening
of short vowels in pretonic open syllables (same example), the
reduction to 3 of short vowels in open unstressed syllables (qab3ra,
q3bartem); for Syriac, the apocope of both short and long final
vowels, though the latter remain part of the graphic pattern (third
plural masculine: q3bar, written qbrw; in verbs with "weak" third
radical the pronunciation conforms to the spelling: e.g. r3maw
"they threw", (wdiw "they rejoiced"), the reduction to 3 or shedding
of short vowels in open unstressed syllables (q3bar, qebrat) , the
change a> e in closed syllables (qebrat).
16.40. c) The personal endings do not vary in accordance with the
internal vowel patterns (qabara, qabira, qabura, qubira are inflected
with the same endings and are not, therefore, listed separately
in the table).
16.41. d) The length of the final vowels, except for the third
person where it is well established (and perhaps for the second
and first persons dual), cannot be fixed with certainty for Proto-
Semitic (this was also the case for the pronouns-cf. § 13.7); it
will not, therefore, be indicated (traces of an originally long vocal-
ization can be detected in some forms before pronominal suffixes:
cf. §§ 16.137-42).
16.42. In the following, Proto-Semitic forms will be posited for
each individual person (and later, in the same manner, for the other
conjugations and stems). It is, however, well to insist from the
outset on the hypothetical character of these reconstructions which
are intended (and ought to be used) as working aids only. In
many cases the reconstructions are subject to ambiguities and
doubts; in others the developments which have been postulated
may have been quite different. Finally, the proposed explanations
often resolve the problems of only some of the Semitic languages.
16.43. While recalling the normal variations of vowel pattern
(§§ 16.2-3) and. bearing in mind the claims of the Akkadian
stative (§ 16.31), we may propose, for the third person singular,
the Proto-Semitic forms qabar(a) (masculine) and qabarat (feminine).
The Verb 139
They appear as such in Arabic and in Ethiopic (as well as in Ugaritic
and in the Tell Amarna glosses: e.g. abadat "she perished"). In
Hebrew the final -a of the masculine reappears as a connecting
vowel before pronominal suffixes (e.g. q3bt'irani, with first person
singular suffix); the -a of the feminine seems to be formed on the
analogy of the feminine morpheme of the noun, but the original
ending -at reappears before pronominal suffixes (e.g. q3baratni,
with first person singular suffix). Similarly, the ending -t of the
feminine, which is otherwise dropped, is attested before pronominal
suffixes in the consonantal spelling of Phoenician (e.g. p'ltn "she
made me", as against p'l "she made").
16.44. For the second person we may propose the Proto-Semitic
forms qabarta (masculine) and qabarti (feminine) which appear
as such in Arabic (probably also in Ugaritic). For the other lan-
guages (cf. §§ 16.38-41) note: a) in Hebrew, the Greek and Latin
transcriptions almost invariably testify to a masculine suffix
-t, so that the Masoretes appear to have adopted an archaic
form, while for the feminine the Biblical k3tib reflects the older
form with suffix -ti; b} in Ethiopic, the consonant of the suffix
has become k, almost certainly by analogy with the first person
singular; a similar process has taken place in the N eo-Assyrian
variant forms qabraka and qabraki.
16.45. For the first person singular we may postulate a Proto-
Semitic form qabarku which appears as such in Ethiopic ifor the
length of the final vowel cf. § 16 AI) and, so far as the flexional
suffix is concerned, in Akkadian (it should be recalled that the
independent personal pronoun of the first person shows the con-
sonantal element k as against t of the second person, cf. § 13.1).
For the other languages (cf. §§ 16.38-41) note: a) the operation
of analogy with the second person singular whereby, in North-
West Semitic and in Arabic, the consonant of the ending becomes t;
b) the phenomenon peculiar to Hebrew (though already attested
in Amorite and the Tell Amarna glosses) where the vowel of the
ending becomes i, probably by analogy with the possessive suffix -i.
16.46. For the third person plural we may propose the Proto-
Semitic forms qabaru (masculine) and qabara (feminine) which
appear as such in Ethiopic (and probably also in Ugaritic). The
140 Morphology
feminine ending -a occurs also in Biblical, Targiimic, and Talmiidic
Aramaic. For the other languages (cf. §§ 16.38-41) note: a) in
Hebrew, as indeed already in the Tell Amarna glosses and later in
Nabataean, the feminine undergoes analogical adaptation to the
masculine; b) in Syriac, the supplementary suffixes -un, -en are
perhaps due to analogy with the personal pronouns 'attan, 'atten
and the suffixes -han, -hen (the analogy is not perfect; in the prefix-
conjugation the feminine -an departs from the analogy); c) in
Arabic, the feminine ending -na is probably due to analogy with
the corresponding ending of the prefix-conjugation.
16.47. For the second person plural we may posit the Proto-
Semitic forms qabartumu (masculine) and qabartin(n)a (feminine)
which seem to occur as such in U garitic. For the other languages
(cf. §§ 16.38-41) note: a) the Hebrew change i > e in the femi-
nine; the analogical formation (so far as the vowel is concerned)
of the masculine; and the stabilization of the original final vowel
attested in the Dead Sea documents by the mater lectionis h (cf.
similarly for the personal pronouns § 13.26); b) the Syriac vowel
changes u > a and i > e (cf. § 10.10); and the formation of the
masculine on the analogy of the final consonant of the feminine
(a process begun in Egyptian Aramaic, which presents both qbrtm
and qbrtn, and completed in Biblical Aramaic, where we find
q3bartun in the masculine and q3barten in the feminine); c) in
Ethiopic the change of the consonantal element of the suffix
into k by analogy with the corresponding person in the sing. (in
Akkadian we encounter the Neo-Assyrian variant form qabrakunu);
the merging of the short vowels u, i in 3 (cf. § 8.96); and the shedding
of the final vowel in the feminine which reappears however before
pronominal suffixes (e.g. qatalbnnahu "you killed him", with
the third person singular suffix).
16.48. For the first person plural we may propose the Proto-
Semitic form qabarna which appears as such in Arabic and in
Ethiopic. For the other languages (cf. §§ 16.38-41) note: a) in
Akkadian and Hebrew the final vowel is changed to 1L, probably
by analogy with the independent and suffixed personal pronouns
(Akk. ninu, Heb. na7y,nu and suffix-nu); b) in Syriac the subsidiary
ending -an is probably due to analogy with the ending -an of the
independent personal
The Verb
141
16.49. For the third person dual we suggest the Proto-Semitic
forms qabara (masculine) and qabarata (feminine) which appear
as such in Arabic (and probably also in Ugaritic). Of the other
languages, Akkadian shows third person dual forms which are
however used only in the Old Akkadian, Old Assyrian and Old
Babylonian periods. In Akkadian the dual may occasionally be
used also for three subjects (von Soden, GAG, p. 186).
16.50. For the second person dual we may propose the Proto-
Semitic form qabartuma which appears as such in Arabic (and
probably also in Ugaritic).
16.51. The first person dual is attested in Ugaritic only, and it
has been doubted (Wagner) that such a form existed in Proto-
Semitic. Hamito-Semitic comparisons, i.e. the presence in Old
Egyptian of the first person dual ending -ny (coinciding with the
Ugaritic morpheme) may possibly favour the assumption of such
a form in Proto-Semitic. As regards its vocalization, we may
perhaps propose a Proto-Semitic form qabarnaya.
b) Simple Stem: Prefix-Conjugation
16.52. The prefix-conjugation of the simple stem is inflected in
the principal Semitic languages as shown in table II. The following
paragraphs offer some general observations on this table:
16.53. a) The two prefix-conjugations of Akkadian ("present"
and "preterite") and of Ethiopic (indicative and subjunctive)
differ from each other formally and semantically, and their genetic
connexion is not accepted by all scholars (cf. §§ 16.30, 16.35).
The prefixes and suffixes attached to the two conjugations are,
however, formally identical; they also agree with those of West
Semitic generally-subject only to the observations in the follow-
ing paragraphs. The Ethiopic subjunctive has two distinct patterns:
transitive Y3qb3r, intransitive y3qbar.
16.04. b) North-West Semitic has put into effect, from the first
millennium B.C., all the changes consequent upon the incidence
of the stress-accent (cf. §§ 10.8, 10.10); this has entailed, for
Hebrew, the shedding of final short vowels (*yaqburu > yiqbor) ,
the transition u > 0 of stressed short vowels (same example),
the change a > i in closed unstressed syllables (same example;
142 Morphology
• .-<
if2
The Verb 143
some scholars, however, regard the vowel i of the prefix as primary,
alongside a, and as peculiar in origin to stative verbs), the reduction
to a of short vowels in open unstressed syllables (yiqbaru). In
Syriac the same changes are operative-save for the process a > e
which takes place in closed unstressed syllables (neqbor).
16.55. c) The variation of vowel pattern reflecting transitive
or intransitive status (cf. § 16.2) has no effect on the form of the
prefixes or suffixes, while the alteration of the vowel pattern in
the passive (cf. § 16.3) brings about a change in the vowel of the
prefix (yu- instead of ya-) which remains constant throughout the
inflexion.
16.56. d) The differentiation of moods (cf. §§ 16.32-36) causes
the addition of -u in the Akkadian subjunctive (Assyrian -uni)
for forms ending in a consonant, and the addition of -am in the
ventive for forms ending in a consonant, of -m for those ending
in -i, and of -nim for the others. In Arabic (and probably in Ugaritic)
the subjunctive substitutes -a for the final vowel and drops the
afformative -na or -ni when preceded by a vowel; the jussive drops
the final vowel and -na, -ni when preceded by a vowel; the energic
substitutes -an or -anna for the final vowel (-anni in the dual and
in the feminine plural) and drops -na, -ni when preceded by a vowel.
These are the principles of modal distinctions; for details the reader is
referred to the paradigms in the grammars of the various languages.
16.57. e) For the Akkadian "perfect" with infix -ta-, a type of
conjugation which has to be regarded as an innovation confined
to Akkadian (§ 16.29), the reader is referred to the paradigms in
von Soden's GAG, both for the simple and the derived stems.
16.58. In the following survey we have to leave out of account
the special thematic patterns of Akkadian and of Ethiopic, but
we shall allow for the variations of vowel distribution dealt with
in §§ 16.2-3. Turning to the individual forms we may propose,
for the third person singular, Proto- Semitic yaqburu (masculine)
and taqburu (feminine) which appear as such in Arabic (and prob-
ably also in Ugaritic). For the other languages (0£. §§ 16.53-56)
note: a) in Akkadian the prefix has evolved: *ya- > *yi > i-
(0£. § 8.63); the same applies, of course, to the third person plural;
b) in Syriac (third person sing and plural) the prefix n-, instead
144 Morphology
of y-, is characteristic; it is an innovation of East Aramaic (Old
Aramaic and West Aramaic retain y-); c) l-, which occurs in Tal-
mudic Aramaic and occasionally in Mandaean as well as in Biblical
Aramaic leh'ewe "he is", may be considered a remnant of precative
l- (cf. Brockelmann, GVG, I, p. 565).
16.59. For the second person singular we may propose Proto-
Semitic forms taqburu (masculine) and taqburi(na) (feminine) which
appear as such in Arabic (and probably also in Ugaritic). For the
other languages (cf. §§ 16.53-56) note the omission of -u even in
Akkadian and Ethiopic. This explanation rests on the assumption
of a single conjugation yaqburu in Semitic; if, however, a distinction
were to be made between the conjugation-patterns yaqburu and
yaqbur (thus von Soden: cf. § 16.31) a different picture would
emerge.
16.60. For the first person singular we may posit a Proto-
Semitic form 'aqburu which appears as such in Arabic (and probably
also in Ugaritic). For the other languages (cf. §§ 16.53-56) note
the shedding of -u even in Akkadian and Ethiopic. The concluding
observations in the preceding paragraph are relevant also in the
present context.
16.61. For the third person plural we may postulate Proto-
Semitic forms yaqburu(na) (masculine) and yaqbura/na (feminine).
Taking into account the general considerations set forth in
§§ 16.53,-56, these postulates agree broadly with the forms in
Akkadian, Syriac (for the prefix n- cf. § 16.58), Arabic, and
Ethiopic. For the other languages note: a) in Ugaritic, in Hebrew,
and in the only Palmyrene occurrence the consonantal prefix is t-
instead of y- in the feminine (in Ugaritic and the Tell Amarna
glosses it may be t- in the masculine as well), by analogy with the
second person plural and the third singular feminine; b) in the
Hebrew of the Dead Sea texts the ending -un instead of -u has been
attested, probably owing to Aramaic influence; c) in Syriac the
feminine ending -an has been adapted, in part, to conform
with the masc. -1tn.
16.62. For the second person plural we may propose the Proto-
Semitic forms taqburu(na) (masculine) and taqbura/nct (feminine)
which are broadly reflected in Arabic and Ethiopic (and probably
The Verb 145
also in Ugaritic). For the other languages (cf. §§ 16.53-56)
note: a) in Akkadian the feminine ending -a takes the place of the
masculine ending -u; b) in Syriac the feminine ending -an has
again been adapted in part to tally with the masculine -un.
16.63. For the first person plural we would posit a Proto-Semitic
form naqburu which appears as such in Arabic (and probably also
in Ugaritic). For the other languages (cf. §§ 16.53-56) note
the Akkadian prefix ni- which may be the result of analogy with
the prefix i- of the third person.
16.64. For the third person dual we may propose a Proto-Semitic
form yaqbura(ni)-in which case the Arabic feminine with prefix
t- and the fluctuating use of y-/t- in Ugaritic are to be attributed
to analogy with the second person dual; or else a Proto-Semitic
feminine taqbura(ni) might be suggested-in which case the Akkad-
ian feminine with prefix i- would be due to analogy with the mascu-
line.
16.65. For the second person dual Arabic and Ugaritic postulate
a Proto-Semitic form taqbura(ni).
c) Simple Stem: Imperative
16.66. The imperative of the simple stem is inflected In the
principal Semitic languages as shown in table III.
Sing. 2 IP.:
f.
PI. 2 m.
f.
Du. 2
III. SimpIe Stem: Imperati ve
Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic
qubur qbr
qubri qbr
qubrii. qbr
qubra qbr
q>Jbor
qibri
'uqbur
'uqburi
q>Jbar
qab(a)ri
qibru q>Jbor(un) 'uqburtt qab(fi)ru
q>Jborna q>Jbor(en) 'uqburna q>Jb(fi)ra
'uqbura
Taking into consideration the merging of short u, i into" in
Ethiopic (cf. § 8.96), we may propose the Proto-Semitic endings
(cf. § 16.36):
2 m.
f.
Singular
-i
l\1oscati, Comparative Grammar
Plural Dual
} -a
-u
-a/na
10
146
Morphology
16.67. Note: a) in Akkadian the feminine ending -a takes the place
of the masculine ending -u; b) the Syriac auxiliary endings -un,
-en might possibly be the result of analogy with the suffix-con-
jugation (the element -n occurs already in the masculine plural
of the imperative in the Arpad inscriptions).
d) Simple Stem: Nominal Forms
16.68. The active participle of the simple stem (table IV) goes
back to a Proto-Semitic form qabir which appears as such in Akkad-
ian, Amorite, Ethiopic, and Arabic (and probably also in Ugaritic).
In Hebrew we have q a b ~ r , owing to the changes a > a (cf. § 8.83)
and i > ~ (cf. § 10.8c). Syriac has qaber as a result of the change
i > e (cf. § 10.10d). The Ethiopic pattern qab'Jr (for i > 'J cf.
§ 8.96) can no longer be formed at will, but it is still used for some
substantives and adjectives (e.g. war'Js "heir", fad'Jq "just"), while
another participial form (or nomen agentis?) assumes the theme
qabari.
IV. Simple Stem: Participle
Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic
Active qiibiru qbr qob?r qiiber qiibir (qabiiri, qiibrJr)
Passive
qiibur qfJbir raaqbur qfJbur
16.69. The passive participle of the simple stem exhibits the
widely used nominal patterns qabir and qabur. Both coexist in
Amorite and occur, though rarely, in Akkadian (von Soden, GAG,
p. 60) where the function of the passive participle is normally
assumed by the verbal adjective. Hebrew qabur seems to pre-
suppose an original qabur, while Syriac q'Jbir is to be referred
back to qabir (already the most ancient Aramaic inscriptions
present the consonantal spelling qbyr); and Ethiopic q'Jbur may
possibly derive from an original qabUr. Arabic adds the prefix m-
(maqbur) , probably by analogy with the participial forms of the
derived stems, and the Nabataean mqbwr is almost certainly the
result of Arabic influence. However, qabir and qabur are also used
in Arabic: e.g. nasig "fabric" (= "woven"), nal),ir "slaughtered",
rasul "envoy" (= "sent"), etc.
16.70. The verbal noun or infinitive has a variety of forms in
the simple stem; they generally merge with the wide range of
The Verb 147
nominal patterns. Among these is the common theme qabar
which occurs in Akkadian (qabaru) , in Hebrew (qabOr, with the
change a > a-cf. § 8.83), and sporadically elsewhere. The "con-
struct" infinitive of Hebrew may be the phonetic result of a different
theme (*qubur > q'Jbor). While the most ancient Aramaic inscrip-
tions exhibit the radical consonants only, Egyptian Aramaic has
the prefix m- which recurs in Biblical Aramaic (miqbar) and later
in Syriac (meqbar). Syriac also makes extensive use of qabar
(e.g. ''Jbada "action", q'Jraba "battle") which in the Modern Aramaic
dialects appears as the regular form of the infinitive. In Ethiopic
the forms qabir and qabirat predominate, but q'Jbrat, m'Jqbar, etc.,
also exist (and may belong to an older phase of the language).
For the Ethiopic "gerund", C£. §§ 16.30-31.
e) Derived Stems: Suffix-Conjugation
16.71. In the other stems the inflexional suffixes and prefixes
remain unaffected and, for that reason, will not be dealt with in
the following; nor will the variations of vowel pattern connected
with the passive. The ensuing analysis will, therefore, be directed
towards a comparison of the stems in the various languages and
a conjectural reconstruction of common forms (the basis as usual , ,
will be the third person singular masculine; cf. table V). It will be
seen that the Akkadian stative differs in vowel pattern from the
West Semitic suffix-conjugation and will therefore be left aside for
our present purposes. The following comparative treatment refers
therefore to the West Semitic area where certain common themes
can be detected. In East Semitic some important divergent features
persist, but this does not imply that they are secondary (cf. §§ 16.31
and 16.43).
16.72. For the stem with doubled second radical we would propose
the common form qabbara which appears as such in Arabic and in
Ethiopic (and probably also in Ugaritic). In Hebrew we have
qibbar or qibb(}r: the change a > i in the first vowel conforms
to the requirements of § 10.8e, while the change a > (} in the
second vowel may be due to analogy with the vowel of the imperfect
(y'Jqabb(}r). In Syriac qabber the change a > e in the second
vowel might be accounted for in the same way as in Hebrew
(imperfect n'Jqabber).
10*
Akkadian
Suff.-Conj. qubbur
. {Uqabbar
Pref.-ConJ. bb'
uqa tr
Imperative
Pa.rticiple
Infinitive
Suff.-Conj.
Pref.-Conj.
Imperative
Participle
Infinitive
Suff.-Conj.
Pref.-Conj.
Imperative
Participle
Infinitive
qubbir
muqabbiru
qubburu
Ugaritic
qbr
yqbr
qbr
mqbr
V. Derived Stems
a) Stem with Doubled Second Radical
Hebrew Syriac
(active) (passive) (active) (passive)
qubbar qabber
ylJqabbf,r ylJqubbar nlJqabber
qabber
mfJqabb?r mflqubbiir mlJqabber mlJqabbar
qabbOr qubbOr mlJqabbiirii
(cstr. (cstr. qubbar)
b) Stem with Lengthened First Vowel
Arabic
(active) (passive)
qiibara qiibira
yuqiibiru yuqiibaru
qiibir
muqiibir muqiibar
qibiir
c) Stem with Prefix 8-
Akkadian
suqbur
{
usaqbar
usaqbir
suqbir
mU8aqbiru
suqburu
d) Stems with Prefix h-, '-
Ethiopic
qiibara
YflqiiblJr
qiiblJr
(maqiibiJr)
qabflr6(t)
Ugaritic
sqbr
ysqbr
sqbr
msqbr
Arabic
(active) (passive)
qabbara qubbira
yuqabbiru yuqabbaru
qabbir
muqabbir muqabbar
taqbir
Hebrew Syriac
Arabic
Suff.-Conj.
Pref.-Conj.
Imperative
Participle
Infinitive
(active) (passive) (active) (passive)
hiqbir
yaqbir

maqbir

(cstr. haqbir)
hoqbar
yoqbar
moqbiir

'aqber
naqber
'aqber
maqber maqbar
maqbiirii
(active) (passive)
'aqbara
yuqbiru
'aqbir
muqbir
'iqbiir
'uqbira
yuqbaru
muqbar
Ethiopic
qabbara
{Yflqebbflr
Yflqabbflr
qabbflr
(maqabblJr)
qabblJr6(t)
Ethiopic
'aqbara
{
yiiqabflr
yiiqblJr
'aqbJr
(maqbflr)
'aqbor6(t)
.....
..,..
00

0
...,
'd
I:l"
0
0'
il'l
'<I
"':l
I:l"
ct>
;$
...,
0"

Suff.-Conj.
Pref.-Conj.
Imperative
Participle
Infinitive
Suif.-Conj.
Pref.-Conj.
Imperative
Participle
Infinitive
Suif.-Conj.
Pref.-Conj.
Imperative
Participle
Infinitive
Suif.-Conj.
Pref.-Conj.
Imperative
Participle
Infinitive
Akkadian
naqbur
{ iqgabbar
iqqabir
naqbir
muqqabru
naqburu
Akkadian
qitbur
{ iqtabbar
iqtabar
qitbar
muqtabru
qitburu
Akkadian
{
uqtabbar
uqtabbir
qutabbir
muqtabbiru
qutabburu
e) Stem with Prefix n-
Ugaritic Hebrew Arabic
(active) (passive)
niqbar 'inqabara 'unqubira
yqbr yiqqab(3T yanqabiru yunqabaru
hiqqab!Jr 'inqabir
niqbar munqabir munqabar
hiqqabOr, niqbOr 'inqibar
(cstr. hiqqab(3T)
f) Simple Stem with t-
Ugaritic Syriac Arabic Ethiopic
(active) (passive)
'etq>Jber' 'iqtabara 'uqtubira taqabra
yqtbr netq>Jber yaqtabiru yuqtabaru yrJtqabar
lqtbr 'etqabr 'iqtabir taqabar
metq>Jber muqtabir muqtabar
metq>Jbaru 'iqtibar taqab(fJ)r6(t)
g) Stem with Doubled Second Radical with t-
Hebrew Syriac Arabic Ethiopic
(active) (passive)
hitqabba/!Jr 'etqabbar taqabbara tuqubbira taqabbara
yitqabb!Jr netqabbar yataqabbaru yutaqabbaru
{ yfJtqebbar
yfJtqabbar
hitqabb!Jr 'etqabbar taqabbar
mitqabb!Jr metqabbar mutaqabbir
hitqabb!Jr metqabbaru taqabbur
h) Stem with Lengthened First Vowel with t-
Arabic
(active) (passive)
taqabara
yataqabar'U
taqabar
mutaqabir
taqabur
tuqubira
yutaqabaru
mutaqabar
mutaqabbar
Ethiopic
taqabara
y>Jtqabar
taqabar
taqabfJr6(t)
taqabbar
taqabbfJr6(t)
I-'
Ol
0
~
0
.."
'"t;I
t::r"
0
-
0
Otl
'<I
1-3
t:r
CD
~
.."
0"
I-'
Ol
I-'
152 Morphology
,
...,
The Verb 153
16.73. For the stem with first vowel lengthened we may propose
the common form qabara which appears as such in the languages
in which this stem is attested (Arabic and Ethiopic).
16.74. For the stems with prefix s-, h-, '- we may posit the
common forms saqbara, haqbara, 'aqbara with which the Arabic
and Ethiopic documentation conforms (and thus probably also in
Ugaritic). In Hebrew we have hiqbir: the change a > i in the
first vowel is a result of the rule stated in § IO.Se, while the tran-
sition a > i in the second vowel may be due to analogy with
the vowel of the imperfect yaqbir (cf. § 16.S3). In Syriac 'aqber
the change a > e in the second vowel might be accounted for in
the same way as in Hebrew (imperfect naqber).
16.75. For the stem with prefix n- we may tentatively suggest, by
analogy with the forms in the preceding paragraph, the common
element naqbara, but no language actually exhibits this form.
In Hebrew we have niqbar, with the customary change a > i
(cf. § IO.Se); Arabic has 'inqabara, with prosthetic 'i- and a syllabic
distribution which may have been influenced by the imperfect
yanqabiru.
16.76. For the simple stem with t- we may perhaps propose, by
analogy with the forms in the preceding paragraphs, a common
element taqbara of which, however, the only possible relic might
be seen in the Ethiopic tansa'a "he rose" alongside tanas'a "he
was raised up". Apart from such a remnant the form taqbara has
been driven out by analogical formations: in Ethiopic we have
taqabra, probably by analogy with the simple stative qabra, as
well as taqabara. Syriac has 'etqaber, perhaps on the pattern of the
imperfect netqaber. Elsewhere we encounter metathesis between
the t and the first radical of the verb (of. § 16.1S): thus, in addition
to Akkadian, the Arabic 'iqtabara, with prosthetic 'i; the thematic
pattern might again be the result of analogy with the imperfect
yaqtabiru.
16.77. For the stem with t- and doubled second radical we postu-
late the common form taqabbara which appears as such in Arabic
and in Ethiopic. In Hebrew we have hitqabbar or hitqabbffr: the
prefix h- may result from analogy with the Hiphil, and the form
as a whole seems to follow the pattern of the imperfect (yitqabbffr).
l54 Morphology
Syriac 'etqabbar is probably again influenced by the imperfect
netqabbar.
16.78. For the stem with t- and first vowel lengthened we may
take as a basis the form taqabara which appears as such in Arabic
and in Ethiopic. It was probably a general South Semitic feature,
but the purely consonantal South Arabian script does not allow
us to arrive at firm conclusions.
16.79. For the stem with prefix s- and t- we may take as a basis
the form 'astaqbara-attested as such in Ethiopic. Arabic'istaqbara
is formed with prosthetic 'i- on the analogy of the stem with prefix
n- and the simple stem with t- (0£. §§ 16.75-76).
16.80. The stem with prefix' - and t- is found in Aramaic only:
Syriac 'ettaqbar is formed by the assimilation t' > tt (cf. § 16.22)
and on the analogy of the imperfect (nettaqbar).
f) Derived Stems: Prefix-Conjugation
16.81. For the stem with doubled second radical we may propose
a Proto-Semitic form yuqabbir(u) which appears as such in Arabic;
Amorite and Ugaritic, however, have a instead of u in the prefix
(yaqabbir). In the Akkadian preterite uqabbir the initial y has been
dropped (cf. § 8.63). In Hebrew y'Jqabbr;r we see reduction to 'J of
the prefix-vowel in the open unstressed syllable as well as the
change i > r; of the short stressed vowel (cf. § 10.8g, c). In Syriac
we find n'Jqabber, with reduction to 'J of the prefix-vowel in the
unstressed open syllable and the change i > e in the closed syllable
(cf. § 10.10c, d). The Ethiopic subjunctive Y'Jqabb'Jr presents the
merging of Proto-Semitic u, i in 'J (cf. § 8.96).
16.82. For the stem with first vowel lengthened we may take as
a basis the form yuqabir(u), attested in South Semitic only; it
appears as such in Arabic, while the Ethiopic Y'Jqab'Jr again shows
the common transition of u, i into (J (cf. § 8.96).
16.83. For the stems with prefix s-, h-, '- we may posit the Proto-
Semitic forms yusaqbir(u), yuhaqbir(u) , yu'aqbir(u). The first
seems to be represented by Ugaritic ysqbr, though the vowel of the
prefix is different (yasaqbir) , and by the Akkadian preterite
usaqbir where the initial y- has been dropped (cf. § 8.63). In
The Verb 155
Hebrew we have *y'J'aqbir > *yaqbir > yaqbir (the change i > i
in the stressed syllable, instead of i >? [cf. § 10.8c], might con-
ceivably be the result of analogy with the imperfect of verbs with
medial w/y [yaqim]). Syriac has *n'J'aqbir > *naqbir > naqber,
with the change i > e in the closed syllable (0£. § 10.10d). The
Arabic development is *yu'aqbiru > yuqbiru. The Ethiopic
SUbjunctive shows *y(J'aqb'Jr (for u, i > 'J cf. § 8.96) > yaqb'Jr.
16.84. For the stem with prefix n- we may postulate a Proto-
Semitic form yanqabir(u) which appears as such in Arabic (but
cf. Amorite yinqabir). The Akkadian preterite iqqabir embodies
the change ya- > i- (cf. § 8.63) and assimilation of n to the first
radical. Ugaritic yqbr, too, appears to show assimilation of n to the
first radical. In Hebrew yiqqabr;r we observe the change a > i
in the initial closed unstressed syllable (0£. § 1O.8e), assimilation
of n to the first radical, and the development i > r; in the stressed
vowel of the final syllable (cf. § 10.8c).
16.85. For the simple stem with t- we may propose a Proto-
Semitic form yatqabi/ar(u}. Akkadian has iqtabar in the preterite,
with the change ya- > i- (0£. § 8.63) and metathesis between t
and the first radical. Ugaritic yqtbr shows metathesis and a different
vowel in the prefix (yiqtabir). Syriac netq'Jber exhibits the usual
reduction to 'J of the short vowel in the unstressed open syllable
(cf. § 10.10c) and the change a > e (i > e) of the two short vowels
in closed syllables (0£. § 10.10d). In Arabic yaqtabiru metathesis
takes place. Ethiopic has Y(Jtqabar: the prefix Y'J- is probably due
to analogy with the other preformatives of derived stems.
16.86. For the stem with t- and doubled second radical we may
postulate the Proto-Semitic form yat(a)qabbi/ar(u) with which
the Arabic yataqabbaru agrees. In the Akkadian preterite we
have uqtabbir, with initial1l- by analogy with other derived stems,
and metathesis between t and the first radical (cf. § 16.19). Hebrew
yitqabbr;r shows the change a > i in the initial closed unstressed
syllable (cf. § 10.8e) and the change i > r; of the short stressed
vowel in the final syllable (d. § 10.8c). Syriac netqabbar has the
customary transition a > e in the initial closed syllable (cf. § IO.lOd).
In the Ethiopic subjunctive Y'Jtqabbar the prefix Y'J- is probably
the result of analogy with the other derived stems.
156
Morphology
16.87. For the stem with t- and lengthened first vowel we may
take as a basis the form yat(a)qabar(u) which is attested in South
Semitic only: Arabic yataqabaru corresponds to this form, while
in Ethiopic Y3tqabar the prefix Y3- is formed as in the other derived
stems.
16.88. For the stem with prefix s- and t- we may posit a Proto-
Semitic yastaqbiru to which the Arabic form corresponds-apart
from the normal change s > 8 (of. however Amorite yistaqbir).
In the Akkadian preterite ustaqbir the initial u- is the result of
analogy with the other derived stems. The Ethiopic prefix-conj.
yiistaqb3r reveals the transition a > a in the initial syllable by
analogy with the stem with prefix '- as well as the change i > 3 in
the final syllable (of. § 8.96).
16.89. The stem with prefix '- and t- occurs in Aramaic only:
derived from a conjectural ya'taqbar(u) is the Syriac nettaqbar,
with the change a > e in the initial closed syllable (cf. § 10.l0d)
and the assimilation of '.
g) Derived Stems: Imperative
16.90. In West Semitic, both Northern and Southern, the forms
of the imperative in the derived stems generally correspond to \
those of the imperfect without its prefixes: any departure from this
rule (which will be examined presently) is almost invariably due
to the appearance of prosthetic vowels (cf. §§ 9.14-15). In East
Semitic some Assyrian variant forms (qabbir for qubbir in the stem
with doubled second radical, saqbir for suqbir in the stem with
prefix s-) suggest a situation that was originally similar, i.e. cor-
respondence with the forms of the preterite without its prefixes.
The Akkadian development is, however, of some complexity,
and the reader is referred to the paradigms; the following treat-
ment is confined to variant formations in West Semitic.
16.91. In the stems with prefix h-, '- Hebrew haqbr;r shows the
evolution *yuhaqbiru > *Y3haqbr;r (§ 10.8) to haqbr;r. Arabic
'aqbir represents *yu'aqbir (§ 16.83) to 'aqbir.
16.92. In the stem with prefix n- Hebrew prefixes prosthetic hi-,
and Arabic prosthetic 'i- (cf. § 9.15).
The Verb 157
16.93. In the simple stem with t- Ugaritic and Arabic add a
prosthetic 'i- (cf. § 9.15). In Syriac retraction of the vowel and
stress produces the form 'etqabr instead of *'etq3ber. In Ethiopic
taqabar shows the insertion of a vowel after the first radical (cf.
§ 9.14), i.e. either as an anaptyctic vowel 3 subjected to vowel
harmony or by analogy with the stems with t- and second radical
doubled and with first vowel lengthened (taqabbar, taqabar).
16.94. In the stem with t and second radical doubled Hebrew
adds prosthetic hi- (cf. § 9.15).
16.95. In the stem with prefix s- and t- Arabic adds prosthetic
'i- (of. § 9.15).
h) Derived Stems: Participle
16.96. For the participles of the derived stems we may propose
Proto-Semitic forms characterized by the prefix mu- and the
vowel i after the second radical in the active, and by the same
prefix mu- and the vowel a after the second radical in the passive.
These forms have been reconstructed on the basis of Akkadian,
Amorite, and Arabic. However, Hebrew and Syriac (causative
stems) and Ethiopic have a as the vowel of the prefix. Apart from
these characteristics, the participles correspond structurally to
the imperfect (in Akkadian to the preterite). For the Ethiopic
forms see § 16.101.
16.97. Akkadian, which presents active forms only (but the
function of the passive participle is often assumed by the verbal
adjective: cf. § 16.69), follows the general principles set forth in the
preceding paragraph. The only exceptions are the stem with
prefix n- (participle [*munqabru > ] muqqabru) and the simple stem
with t- (participle muqtabru) in which the vowel i is dropped owing
to the succession of short syllables (von Soden, GAG, p. 14).
16.98. In Ugaritic the only participles attested are those of the
stems with second radical doubled and with prefix s- in the active:
the consonantal structure (mqbr and msqbr, respectively) does not
indicate the vowel quality of the prefix.
16.99. In Hebrew the prefix ma- is reduced, in accordance with
the conditions affecting short vowels in unstressed open syllables
158 Morphology
(§ 10.8 g), to m9-; m9-, in its turn, is contracted with the prefixes
of the various stems-thus producing, in the stem with h-, the
forms (*m9haqbir » maqbir and (*m9hoqbar » moqbar, in the
stem with t- and geminated second radical the form (*m9hitqabbf}r »
mitqabbf}r. An unusual form of participle occurs in Hebrew (also
in Phoenician, so far as the consonantal spelling indicates) in the
stem with n-: niqbar, formed by analogy with the suffix-conj.; cf.
also U garitic nkbd.
16.100. In Syriac the prefix ma- is reduced, in accordance with
the principles governing short vowels in open unstressed syllables
(§ 10.10c), to m9-; and m9-, in its turn, is contracted with the
prefix ' - producing the form (*m9'aqber » maqber, while in the
remaining stems the change a > e (cf. § 10.lOd) takes place
(metq9ber, metqabbar, mettaqbar).
16.101. In Ethiopic the prefix is stabilized in the form ma-.
The active use of the participle is, however, greatly restricted,
and the -form qabari of the simple stem gives rise to analogical
formations in the derived stems (qabbari, 'aqbari, etc.). In fact,
the participle in Ethiopic has become a lexical item rather than
a regular morphological feature.
i) Derived Stems: Infinitive
16.102. The infinitives of the derived stems have a number of
forms which can more suitably be examined in each of the languages;
we shall confine ourselves to the observation of certain common
features.
16.103. In Akkadian the infinitive coincides with the stative
followed by the nominal morpheme (stative qubbur, infinitive
qubburu; stative 8uqbur, infinitive 8uqburu; etc.).
16.104. In Hebrew the infinitive is formed on the pattern of the
imperfect without its prefixes and thus coincides with the impera-
tive. By analogy with the absolute infinitive qabOr of the simple
stem Hebrew also forms absolute infinitives of various derived
stems with final vowel 0 (qabbor, qubbor, hiqqabOr, niqbor). The
absolute infinitive· is, however, rarely used in Hebrew, while the
slightly different construct forms are widespread.
The Verb 159
lS.105. In Biblical Aramaic the ending -a of the st. abs. fem. is
characteristic of the infinitive of the derived stems (0£. § 16.70).
In Syriac the infinitive retains the prefix m- of the simple stem,
the second radical receiving the vowel a and the third the (abstract)
ending -ii(t) (the final -t, which does not occur in the absolute state,
appears before suffixes): m9qabbarii, maqbarii, metq9barii, etc.
lS.106. In Arabic the infinitive possesses several patterns, some
of which reveal similar schemes and may be grouped together
('iqbar, 'inqibar, 'iqtibar, 'istiqbar; taqabbur, taqabur); apart from
these, we have the infinitives of the stem with doubled second
radical (taqbir), with lengthened first vowel ([qibar >] qibar) , and
others.
16.107. In Ethiopic the infinitive is formed on the same pattern
as the imperative; to this is added the ending -o(t): qabb9ro(t),
qab9ro(t), 'aqb9ro(t) , etc.
5. The so-called "Weak" Verbs
16.108. The following chapters will be concerned with an ex-
amination of some types of verbs which differ from the regular
pattern. These verbs contain either pharyngals and laryngals (in
particular' ), or the alveolar nasal n, or semi vowels (w, y), or their
second and third radicals are identical. In traditional Semitic
grammar (cf. Brockelmann, GVG, I, pp.584-638; also Gray,
SOL, pp. 110-18) these groups of verbs are usually lumped
together under the term "weak verbs". Their forms are regarded
as explicable on a basis of triradical roots, either by means of
phonetic changes characteristic of the consonants concerned or
by the operation of analogy. More advanced linguistic study has
shown, however, that those principles suffice for the explanation
of only a limited number of these verbal forms, i.e. those with
pharyngals and laryngals and those with (original) initial y- (though
not without some reservations). In the remaining groups it may
be shown that we are dealing, for the most part, with biconsonantal
roots, the third radical having arisen secondarily in a process of
integration with the predominant triradical system. This is con-
firmed by the fact (already referred to in §§ 11.5-9) that many
"weak" verbs appear in several forms which have in common two
160 Morphology
radicals and a basic range of meaning-but differ in the third
radical. If this explanation reflects the linguistic reality accurately,
then we must consider as mere working aids those theories which
have been applied, especially to Arabic grammar (cf. Fleisch, TPA,
pp. 118-38), in order to account for all the forms of these verbs
in terms of the triconsonantal system. The ensuing treatment has
purely descriptive aims and restricts the term "weak" to verbs
of probable biradical origin; it will call attention to certain facts
of fundamental importance to the study of comparative Semitic
grammar. For detailed information about independent develop-
ments in the various languages the reader is referred to the relevant
grammars. The fact that we are dealing with trends pulling in
opposite directions (the reduction of triradical roots and the ex-
pansion of biradical ones) as well as the complex nature of certain
phenomena will leave a wide margin of uncertainty which can be
reduced only by specific advance in the study of this branch of
comparative Semitic grammar.
6. Verbs with Pharyngals and Laryngals
16.109. The verbs with pharyngals and laryngals exhibit certain
specific peculiarities. These are connected, for the most part, with
the characteristics inherent in these consonants (cf. particularly
§ 9.6); in some cases, however, they are occasioned by analogy
with the "weak" verbs (of. below). The glottal stop' occupies a
special position which gives rise to phenomena not shared by other
consonants: the verbs with' will, therefore, be dealt with separately.
It should also be recalled in this context that in A k k a d ~ a n all the
consonants of the pharyngal and laryngal series are reduced to '
(§§ 8.53-54); hence Akkadian will only be considered in connexion
with the verbs containing the glottal stop: the change a > e
caused by , derived from ly, and " and sometimes also from g and h
(cf. § 8.54), has produced two verbal classes (with a and with e),
even though there are considerable fluctuations between them.
16.110. Verbs with pharyngals and laryngals (in Hebrew and
Syriac also those with r: cf. § 8.25) are characterized by the tend-
ency to change into a vowels contiguous to those consonants,
no doubt as an aspect of assimilation. In Hebrew this tendency
is the rule (e.g. *yisloly, "he sends" > yislaly,); the question arises,
The Verb 161
however, whether in some cases this phenomenon does not, in
fact, represent the preservation of an original a (e.g. in yaMob
"he thinks", where the a of the prefix reflects the vowel of the
Proto-Semitic yaqburn). In Hebrew a subsidiary vowel of the a type
establishes itself after' and ly, (rarely h) when the consonant would
othenyise be without vowel at the close of a syllable: e.g. *ya'mod
"he stands" > ya'i'imod; *ya'm9du "they stand" > ya'amdu (but
yaly,mol "he shows compassion"). Moreover, a "liaison" a is inserted
into the articulation between the long vowels of other timbres and
a following pharyngal or laryngal: e.g. *hismi' "he caused to
hear" > hismia' (cf. § 9.6). In Syriac the tendency to change
into a vowels contiguous to pharyngals or laryngals is somewhat
sporadic: e.g. neb'at alongside neb'ot "he pushes", *nedkor "he
remembers" > nedkar; the change is regular, however, in cases in
which e would occur before a pharyngal or laryngal in final position:
e.g. *'etd9ker "he remembered" > 'ett9kar (cf. Brockelmann, SG,
pp.86-87, § 186). The phenomenon is fairly rare in Arabic (e.g.
*yaftuly,u "he opens" > yaftaly,u). In Ethiopic the change 9> a
occurs in the prefixes of verbal forms whose first radical has a
laryngal followed by a (by way of vowel harmony): *Y9ly,aww9r
"he goes" > yaly,aww9r. The opposite process a > 9 takes place
when a laryngal is followed by a vowel other than a: *nasa'u "they
raised" > nas9'u. Finally, Ethiopic lengthens a before a vowelless
laryngal: *sama'ku "I have heard" > sama:ku (cf. Praetorius, Gram-
matica Aethiopica, pp. 16-18, § 16; Ullendorff, SLE, pp. 212-14).
16.111. A characteristic feature of Hebrew, as presented to us by
the Masoretic tradition, is its inability to double pharyngals and
laryngals-with consequent compensatory vowel lengthening:
e.g. *ba"!}r "he consumed" > ba'!}r. Biblical Aramaic shares the
same inability in the tradition of the Masoretes (including also r):
but while in the case of " r and sometimes' there is compensatory
vowel lengthening (e.g. *barrik "he blessed" > barik), it is absent
in connexion with h, ly,. Moreover, there are indications that at
certain times and in certain areas the consonant was doubled
(e.g. the dissimilation *ha"!}l "cause to enter!" > han'!}l). The
accurate phonetic notation of the Masoretes rejects simple snaii
with pharyngals and laryngals and uses instead a compound s9wii,
chiefly with a (of. § 16.110) but also with other vowels: e.g. yely,ezaq
Moscati, Comparative Grammar 11
162 Morphology
"he is strong". In this example, as well as in the previously cited
ya'i'imod, we have a typical instance of vowel harmony in Hebrew.
16.112. The first characteristic of verbs with ' is the elision
of postvocalic " with consequent lengthening of the preceding
vowel in the North Semitic area: a.g. from the root'lJ4 "to seize",
Proto-Semitic *ya'lJuiju, Akk. ilJuz, Heb. yi5lJf-z, Syr. nelJod; but
AI'. ya'lJurJu, Eth. ya'alJn (as well as ya'alJaz). In Akkadian, i' pro-
duces i in Babylonian, e in Assyrian: e.g. *i'kul "he ate" > Bab.
ikul, Ass. ekul. In Hebrew, the form yi5lJf-z is the result of an evo-
lution which may be represented as follows: *ya'lJuiju > *yalJuz >
*yi5lJoz > yi5lJf-Z (by dissimilation). This development is, however,
confined to a few verbs with ' as first radical ('bd "to perish",
'by "to want", 'kl "to eat", 'mr "to say", 'py "to bake"); their
inflection is regarded as "weak" by contrast to the others. There
is, however, a good deal of contamination and interference be-
tween "weak" verbs and others of this type (cf. Beer-Meyer,
Hebraische Grammatik, II, pp. 46-47). In South Semitic, where
this phenomenon does not as a rule occur, we encounter in Arabic
a type of dissimilation of two' in the same syllable: e.g. *, a'lJurJu
"I take" > 'alJuiju, *'a'mana "he believed" > 'amana.
16.113. In Akkadian some verbal forms with first radical' show
syncope of intervocalic " followed by contraction in which the vowel
of the prefix prevails: e.g. *i'akkal "he eats" > ikkal, *u'arrak
"he lengthens" > urrak. Another characteristic of Akkadian is
the assimilatory complex nn, commoner than ", resulting from
a meeting of nand' in the verbal stem with prefix n-: e.g. innabbit
"he flees" alongside i"abbat "he is destroyed" (cf. von Soden, GAG,
pp. 126-29). Syriac has some verbs with first radical' which are
formed by analogy with those with first radical w/y (e.g. 'awkel
"he caused to eat", from the root 'kl); and similarly in Arabic (e.g.
*'i'talJaija "he took" > 'ittalJaija by analogy with e.g. *'iwta'ada
"he promised" > 'itta'ada). Some imperative forms which drop
initial ' (e.g. Syr. zel "go!" from 'zl, Ar. lJuij "take!" from '1Jij)
are analogous with those of verbs with first radical w/y.
16.114. In verbs with third radical' the characteristic elision
of postvocalic ' in North Semitic (of. § 16.112) brings about coales-
cence with the verbs with third radical y as far as the resulting
The Verb
163
vowel is concerned (cf. § 16.121): e.g. Proto-Semitic *mali'a "it
was full", Akk. (stative) mali, Heb. mal?, Syr. mali (beside mala);
but AI'. mali'a, Eth. maZ,/a (mal'a). In Hebrew, ' is preserved
at the beginning of a syllable, probably by analogy (or perhaps
restored by the Masoretes?): e.g. mala'1( "they were full". Finally,
a small group of Akkadian verbs (e.g. pr' "to cut") treats final' as
a normal radical (von Soden, GAG, p. 133).
16.115. The peculiarities of the verbs with " outlined in the
preceding paragraphs, do not apply to verbs with second radical '
for they follow the general pattern of pharyngals and l a r y n g a l ~
(cf. §§ 16.109-111). In Akkadian, however, these verbs are
sometimes inflected by analogy with those of second radical w/y:
thus alongside ida"im "it becomes dark" isal "he asks" (but in
Assyrian sometimes isa"al; cf. von Soden, GAG, pp. 130-33).
In Syriac, nes'al "he asks" > nesal (cf. Noldeke, KurzgefaBte
Syrische Grammatik, p. 108, § 171) shows syncope of '. (Note the
imperative sal of the corresponding Arabic verb s'l.)
7. Verbs with First Radical n
16.116. In North Semitic vowelless n is generally assimilated to
the following consonant: e.g. Akk. *indin "he gave" > iddin;
Ug. *ynpl "he falls" > ypl; Heb. *yinfor "he guards" > yiffor;
Syr. *nentor "he guards" > nettor. This assimilation does not
take place in Hebrew before consonants of the pharyngal and
laryngal group, as these consonants cannot be geminated (of.
§ 16.111): e.g. Heb. yin1y,al "he inherits". In Biblical Aramaic
the n is frequently maintained in these circumstances (e.g. yin-
tanun "they give"), but this is probably due to secondary dissi-
milation of the doubled consonant-as may be shown by several
cases in which n cannot be held to be original (e.g. tinda' "thou
wilt know" < *tidda', from the root yd': cf. Rosenthal, Grammar
of Biblical Aramaic, pp. 16-17, 47). In Syriac, the assimilation
of n becomes inoperative in several verbs with second radical h:
e.g. nenhar "it shines", root nhr. The North Semitic imperative
is formed without n: e.g. Heb. gas "approach!" from ngs, Syr. tor
"guard!" from ntr; Hebrew verbs with second vowel 0, on the
other hand, retain their initial n: e.g. nafor "guard!". In Akkadian
the imperative of these verbs generally presents a prosthetic vowel
11*
164 Morphology
(idin "give!" from ndn, uqur "destroy!" from nqr) , but in Old
Akkadian and in Assyrian forms without prosthesis do occur (e.g.
din "give !").
16.117. In South Semitic, n is not subject to such special treat-
ment-except for some cases in South Arabian: e.g. * s t n ~ r "he
asked for help" > s t ~ r .
8. Verbs with w, y
16.118. For the verbs with w, y it is well to recall the phonetic
laws about semivowels (of. §§ 8.61-65, 10.3, 9.7-8, 9.11, 9.13,
9.20) as well as the working of analogy affecting "regular" and
"weak" verbs; they are particularly exposed to the operation of
Systemzwang by which originally biradical verbs are integrated
within the triconsonantal system.
16.119. Verbs with first radical wand y constitute, in origin,
distinct categories; only those with first radical w (and they are
the more numerous group) seem to be genuinely "weak". Reciprocal
influences between the two categories and the passage of verbs
from one group to the other are, however, so common that it is
well to deal with them together. In the first place, we note the
characteristic change in North-West Semitic of w > y when in
initial position (cf. § 8.64): e.g. Akk. Ar. Eth. wld "to bear", Ug.
Heb. Syr. yld. The original distinction between these two cate-
gories of verbs emerges once more in various forms of the derived
stems: e.g. in the Hebrew verb ysb "to sit" the first radical is
originally w-as is demonstrated by the stem with prefix h-:
hasib, whereas in ytb "to be good" the y is primary-as is proved
by the form hMib. Over the entire Semitic area initial w is absent in
the imperative: e.g. from the root wld Akk. lidi, Heb. lCidi, Ar.
lidi, Eth. ladi "give birth!" (but wCijCir "go out !"). Syriac ilad is
exceptional (by analogy with the verbs primae y), but forms like
teb "sit!" and hab "give!", etc., agree with their counterparts in
the other languages. Initial w does not appear in the imperfect:
from the same root wld, Heb. y?l?d, Syr. nelad, Ar. yalidu, Eth.
(subjunctive) YCilad (Akkadian forms a partial exception: ulid;
note in Hebrew and Syriac the vowel lengthening in the prefix).
Finally, certain West Semitic languages have infinitives without w
The Verb 165
and with "feminine" ending: Heb. ledet, Ar. lidat, Eth. l"dat
(Akkadian has waladu or aladu by reduction of initial w according
to § 8.63; but cf. the noun subtu from the root wsb). For the complex
forms of the verbal stem with .$ in Akkadian, of. von Soden, GAG,
pp.141-42.
16.120. In the verbs with medial radical w, y the imperative
exhibits the appropriate long vowel between the first and third
radicals: e.g. Akk. kun "be steady!", Heb. Syr. Eth. qum "rise!"
(Ar. qum owing to the reduction of the long vowel in the closed
syllable); Akk. sim, Heb. sim, SyT. sim, Eth. sim "put!" (Ar.
sim for the same reason as above). A medial vowel a is rare: e.g.
Akk. bas "be ashamed!" (cf. Heb. bos). The characteristic vowel
remains and is long (also in Arabic) in the prefix-conjugation: thus
Akk. preterite ikun, Heb. yaqum, Syr. nCiqum, Ar. yaqumu, Eth.
(subjunctive) YCiqum; Akk. preterite isim, Heb. yasim, Syr. nCisim,
Ar. yasimu, Eth. (subjunctive) Y'Jsim. In the Ethiopic forms, im-
perative as well as prefix-conjugation, the vowels are marked long
on etymological grounds; this does not necessarily reflect actual
pronunciation. In the suffix-conjugation West Semitic presents
Heb. qam, Syr. qam, Ar. qama, Eth. qoma (in the Ethiopic form
the vowel 0 is derived from the diphthong aw; it is noteworthy that
the usual Hebrew change a > 0 does not take place in this case,
though there are traces of it in the form nu!Jti "I am quiet" of
the Tell Amarna glosses and in the Latin transcription chon "he
was" in Phoenician-Punic) ; Heb. sam (but also bin' 'he understood"),
Syr. sam (but mit "he died"), Ar. sama, Eth. sema. Interference
between the two classes is not unusual and is confirmed by the
Akkadian stative kin and sim (Assyrian ken and sem); in the prefix-
conjugation this interference was only sporadic (e.g. Heb. yasum
alongside the usual yasim). The inflexion of the Arabic perfect
shows the reappearance of the characteristic vowel: qama, q1Lmta;
sama, simta. In the derived stem with doubled second radical,
some languages present forms which correspond to those of the
"regular" verbs (Ar. qawwama, Eth. qawwama, Syr. qayyem), while
Hebrew has formations on the pattern of the verbs with doubled
second radical (qom?m). Akkadian (which has gemination of the
second radical also in the present of the simple stem) exhibits
instead the doubling of the third radical, provided it is followed
166 Morphology
by a vowel (e.g. idukku "they kill", isimmu "they place", against
the singulars idiiak > idak and iSiam). The stem with s is formed
similarly: e.g. usmiit "he kills", usmattu "they kill"; usmit "he
killed", uSmittu "they killed". Formations by analogy with other
"weak" verbs occur in Hebrew (e.g. yumat "he is being killed",
on the pattern of the verbs with first radical w).
16.121. It is characteristic of the verbs with third radical w, y
(with which coalesce those with third radical ') that y predominates
over w in North Semitic. Example of a verb with original y (the
prefix- and suffix-conjugations are indicated): root bky, Akk. baki
(stative), ibki (preterite), Heb. bakii, yibke, Syr. b'Jkii, nebke, Ar.
baka, yabki, Eth. bakaya, y'Jbki (subjunctive). Example of a verb
with original w: root dlw, Ar. dalil, yadlu, Eth. dalawa, y'JdlU (sub-
junctive), but Akk. idlu, Heb. dala, yidle, Syr. d'Jla, nedle. These
examples show: a) there are in Akkadian various exceptions to the
North Semitic predominance of the type with y over that with w
(e.g. imnu "he counted", ilylu "he rejoiced"); b) the Ethiopic
forms bakaya and dalawa agree with the regular pattern and appear
to favour the conception of these verbs as original triradicals; this
assumption is scarcely set aside by certain changes awa > 0
(e.g. halawa and halo "he was"); c) the triradical origin is also sup-
ported by certain Ugaritic forms which keep wand y (e.g. dtwt
"she came"); d) some interesting fluctuations between wand y
are exhibited by South Arabian: from the root rif,w "to be content"
the prefix-conjugation of the stem with h- has yhrif,wn and yhrif,yn.
The greater part of these verbal forms can be explained by syncope
of w, y between vowels and subsequent contraction of those vowels.
This process is of some consequence in the historical development
of the Semitic languages generally (e.g. Akk. ibanniu "they
build", later ibannu). Noteworthy is also the tendency to shorten
or even to drop final vowels resulting from contraction: thus Akk.
ibni "he built" for ibni.
9. Verbs with Identical Second and Third Radicals
16.122. In the verbs with identical second and third radicals,
commonly called verba mediae geminatae, the probable bicon-
sonantal origin is particularly evident; integration within the tri-
consonantal system demonstrates the force of analogy.
The Verb
167
16.123. In Akkadian the verbs of this group are completely
adapted to the regular pattern. A biradical form is presented
by the stative of the verbs which indicate a condition (e.g. dan "he
is strong", sar "he is false"); this form (which is standard in Old
Babylonian) is paralleled in Neo-Babylonian by those correspond-
ing to regular verbs (e.g. elil "he is pure"). A small group of verbs
with second radical 1 or r forms a special durative type with n:
e.g. na'arruru "to come to the rescue", naparruru "to disband", etc.
In the inflexion of these verbs gemination of the third radical fre-
quently occurs before vocalic suffixes: e.g.lin'arirru "let them help".
·16.124. In Hebrew the perfect of the simple stem is integrated
with the regular pattern (type sabab), but stative verbs have
biradical forms (type ly,am); the imperfect and imperative also
have biradical elements (type yasob, sob). Some forms with doubled
first radical are attributed to Aramaic influence: e .. g. yissob along-
side yasob (cf. § 16.125). Before vocalic suffixes the second radical
is geminated: e.g. yiisobbu; before consonantal suffixes a connecting
vowel is introduced: 0 in the perfect and e in the imperfect, e.g.
sabbOti, t'Jsubbena (cf. the similar formation of verbs with third
radical w/y: e.g. tiglena). In the derived stems metaplastic forma-
tions are common: Polel, Polal, Hithpolel; others are inflected
by analogy with the "weak" verbs: e.g. the Hophal yosab modelled
on verbs with first radical w (yosab).
16.125. In Syriac biradical forms are widely attested: the perfect
of the simple stem is baz, bezzat, while in the imperfect the first
radical is doubled, by analogy with the verbs primae n (type
nebboz). The masculine singular participle is formed on the pattern
of the verbs mediae w (type ba'ez): this analogy does not extend
to the feminine and the plural (in contrast to Jewish Aramaic and
Mandaean). In the derived stems forms on the model of the
"regular" verbs are widespread, e.g. 'etbnez, etc.
16.126. In Arabic, the verbs of this type appear in the forms
farra, perfect, yafirru, imperfect; and in the derived stems farra,
yufarru; 'afarra, yufirru, etc. When the last radical has no vowel,
analogy with the "regular" verbs operates: e.g. 'afrarta, etc.
16.127. In Ethiopic, integration with the "regular" verb is
prevalent (perfect lJasasa, imperfect y'J7;ass'Js). In the perfect
168
Morphology
of stative verbs (type ly,arnma) and in the simple stem with t-
(type tanabba) biradical forms are attested. In the imperative
and the imperfect the shortened forms exist alongside those
fashioned on the analogy of the "regular" verbs.
10. Doubly Irregular or Defective Verbs
16.128. All the Semitic languages have doubly irregular verbs,
i.e. verbs which combine two of the categories discussed in the
foregoing: e.g. Akk. 'wr "to be awake" (first radical' and second w),
Reb. ns' "to carry" (first radical n and third '), Syr. lwy "to
accompany" (second radical wand third y), Ar. wqy "to take care"
(first radical wand third y), Eth. whb "to give" (first radical wand
second h), etc. In these verbs the inflexion takes account of the
characteristics of both categories concerned. Much rarer are verbs
in which all three radicals are "weak": e.g. Akk. *awu "to speak",
ewu "to become", etc.
16.129. There also exist a number of defective verbs whose forms
diverge from the general patterns hitherto discussed. For such
verbs the reader is referred to the grammars of the individual
languages. In general these anomalies can be resolved artificially by
subsuming such verbal forms under categories to which in essence
they do not belong: thus Rebrew lqly, "to take" behaves like a verb
of first radical n (imperfect yiqqaly,) and hlk "to go" like a verb of
first radical w (imperfect Syriac 'zl "to go" assimilates its 1
in certain circumstances (first person singular perfect: 'ezzet) and slq
"to go up" assimilates the 1 and has forms like a verb with first
radical n (imperfect nessaq); Arabic r'y "to see" presents shortened
forms (imperfect yara); Akkadian verbs such as izuzzu "to stand"
and itulu "to lie" exhibit forms that may be referred to several
different categories (first radical n, medial radical w/y, etc.); of. von
Soden, GAG, pp. 154-56.
11. Semantic Categories in "Weak" Verbs
16.130. B. Landsberger (1 slamica 2 [1926], pp. 362 ff.) was the
first to recognize that there existed a measure of correlation
between several types of "weak" verbs (of. § 6.116ff.) and certain
semantic categories. While details have not yet been worked out,
the most important categories-according to von Soden-are:
The Verb 169
16.131. Verbs primae n: a) verbs whose biradical basis, without
the element n, connotes a noise, e.g. Sem. nbMl} "to bark" (i.e. "to
say buly,"), npl} "to blow" (i.e. "to sound pul}"); b) verbs in which
the element n has locative meaning, e.g. Sem. ns' "to lift up",
Akk. ndy "to throw down", Reb. npl "to fall down", Arab. nzl
"to descend", Eth. nbr "to sit down".
16.132. Verbs primae w, y: a) verbs which describe certain
involuntary actions, e.g. Sem. wld "to give birth", Arab. wgd "to
find", Eth. wdq "to fall"; b) verbs which connote the aim or
target of a motion, e.g. Sem. wrd "to go down", Eth. wsd "to lead
to" .
16.133. Verbs mediae w: a) verbs which describe a change of
condition or transition from one situation to the opposite one, e.g.
Sem. mwt "to die", West Sem. qwm «to get up"; b) verbs which
refer to types of motion, e.g. Akk. dwl "to go to and fro", Reb. and
Eth. "to run".
16.134. Verbs mediae y: a) verbs which describe a physiological
function, e.g. Sem. syn "to urinate"; b) verbs connoting a definite
outcome or result, e.g. Sem. sym "to place, fix", Akk. AI'. Eth.
l}yr "to elect".
16.135. Verbs ultimae w, y: a) (only ultimae y) verbs of ter-
minative meaning, e.g. Sem. (except Eth.) bny "to build", Reb.
Aram. gly "to reveal"; b) verbs which describe durative actions,
e.g. West Sem. r'y "to see", Akk. Reb. Aram. mnw "to count".
16.136. Verbs mediae gerninatae: especially verbs which connote
a number of individual actions ("Kettendurative"), e.g. Akk. sll,
Syr. bzz "to plunder", Arab. 'dd "to count", etc. For the change
of categories of. e.g. Reb. sgg and sgy "to err", Ar. frr and nfr "to
£lee" .
12. Verbs with Pronominal Suffixes
16.137. Before pronominal suffixes we often witness the re-
appearance of Proto-Semitic elements which have underaone
!:l
considerable development in the forms without suffixes. Certain
alterations of morphemes and endings also occur, and some con-
necting vowels are inserted. Some scattered information on these
points has already been given in various places in this book.
170 Morphology
16.138. In Akkadian the pronominal suffixes are appended to
verbal forms without alteration; some endings take the ventive
morpheme -an before the suffixes-and this fact has a certain
comparative interest.
16.139. In Hebrew the Proto-Semitic endings of the perfect
reappear before suffixes: third singular masculine -a, third singular
feminine -at, second singular feminine -ti. In the second person
plural masculine we have tii instead of tem, probably as a result
of shortening of the Proto-Semitic -tumii. The suffix-pronouns can be
appended directly to the forms of the imperfect in the case of the
second person only (type yiqborka); for the suffixes of the other
persons a connecting vowel is inserted on the model of the verbs
with third radical y (type the same occurs in the im-
perative (type For forms with -an before the suffixes
(like yiqqal}ennii) cf. § 16.34.
16.140. In Syriac the Proto-Semitic endings of the perfect
reappear before suffixes: third singular masculine -a, second singular
masculine -ta, second singular feminine -ti, third plural feminine -a,
first plural -na; likewise in the imperative: second plural masculine
-ii, second plural feminine -a. In those persons of the imperfect
which have no afformatives a connecting vowel i appears (type
neqb9riw) which may occur also in the imperative. Finally, Old
Aramaic may insert -an (cf. §§ 16.138-39) before the suffixes:
e.g. Eg. Aram. Y9siminnak "he puts thee".
16.141. In Arabic long final vowels appear in the perfect endings
in the second singular feminine (type qabartini beside qabartini) and
in the second plural masculine (type qabartumiini).
16.142. In Ethiopic long final vowels are substituted in the
perfect endings of the second singular masculine (-ka) and in the
first person plural (-na). In the second plural feminine the full
ending -k9nna appears which is more often shortened to -ka.
Dissimilation is the cause of the change -i > 9 in the second
singular feminine before the suffix pronoun -ni. For further
details cf. §§ 13.14, 13.27.
Bibliography
The following bibliography includes works published up to the end of
1962; in the text, however, only works which appeared up to the end of 1961
have been considered. The bibliography is selective and, except in cases
of special importance, does not include works published before 1908 (the
date of Vol. I of Brockelmann's GrundrifJ ... and of his KurzgefafJte ver-
gleiohende Grammatik ... ). Inclusion of a work in this bibliography does
not necessarily connote agreement with its theses.
The abbreviations used in the bibliography are those of the Bibliographie
semitique published periodically in Orientalia, with the addition of: .AIOK
XXIV = Akten des XXIV. Internationalen Orientalisten-Kongr6ss6s (Wies-
baden 1959); SOLDV = Studi orientalistioi in onore di G. Levi Della Vida,
2 vol. (Roma 1956); ZS = Zeitsohrift fur Semitistik.
I. The Semitic Languages
.A. Scope of the Survey
Barth, J., SpraohwissensohajtlioM Untersuohungen zum Semitischen (Leipzig
1907-11).
Bergstriisser, G., Einfuhrung in die semitischen Spraohen (Miinchen 1928).
Bibliographie semitique, in Or 16 (1947),103-29; 17 (1948), 91-103; 19 (1950),
445-78; 22 (1953), 1 *-38*; 26 (1957), 50*-115*; 28 (1959), 59*-90*;
30 (1961), 42*-61*.
Blake, F. R., Studies in Semitio Grammar, in JAOS 35 (1917), 375-85;
62 (1942), 109-18; 65 (1945), 111;-16; 66 (1946), 212-18; 73 (1953),
7-16.
Brockelmann, C., GrundrifJ der vergleiohenden Grammatik der semitisohen
Sprachen, 2 vol. (Berlin 1908-13).
-, KurzgejafJte vergleichende Grammatik der semitischen SpracMn. Elemente
der Laut- und Formenlehre (Berlin 1908).
-, Preois de linguistique semitique (Paris 1910).
-, Semitisohe Spraohwissensohaft, 2. Aufl. (Leipzig 1916).
de Lacy O'Leary, Comparative Grammar of the Semitic Languages (London
1923).
Dhorme, P., Langues et ecritures semitiques (Paris 1930).
Fleisch, H., Introduotion Ii l'etude des langues °semitiques (Paris 1947).
Fronzaroli, P., Prospettive di metoda statistioo nella classificazione delle lingue
semitiOM, in ANLR Ser. 8, 16 (1961), 348-80.
Gray, L. H., Introduotion to Semitio Comparative Linguistics (New York 1934).
172
Bibliography
Handbuch der Orientalistik, hrsg. von B. Spuler, III. Semitistik (Leiden
1953-54).
Kramers, J. H., De semietische talen (Lei den 1949).
Langues et ecritures semitiques, in Dictionnaire de la Bible, Supplement, V
(Paris 1957), 257-334.
Levi Della Vida, G. (and others), Linguistica semitica: presente e fUtUTO
(Roma 1961).
Noldeke, Th., Beitriige zur sernitischen Sprachwissenschajt (Strasbourg 1904).
_, Die semitischen Sprachen, 2. Ann. (Leipzig 1899).
_, Neue Beitriige zur semitischen Sprachwissenschaft (Strasbourg 1910).
Pedersen, J., Semiten (Sprache), in Reallexikon der Vorgeschichte, XII (Berlin
1928), 14-50.
Renan, E., Histoire generale et systeme compare des langues semitiques, 1,
3eme ed. (Paris 1863).
Rinaldi, G., Le lingue semitiche (Torino 1954).
Segert, S., Considerations on Semitic Comparative Lexicography, in ArOr 28
(1960), 470-87.
_, Semitistische Marginalien, in ArOr 29 (1961), 80-U8.
Ullendorff, E., What is a Semitic Language?, in Or 28 (1958), 66-75.
von Soden, W., Zur Einteilung der semitischen Sprachen, in WZKM 56 (1960),
177-91.
Wright, W., Lectures on the Comparative Grammar of the Semitic Languages
(Cambridge 1890).
Zimmem, H., Vergleichende Grammatik der semitischen Sprachen (Berlin
1898).
B. North-East Semitic
Aro, J., Studien zur mittelbabylonischen Grammatile (Helsinki 1955).
de Meyer, L., L'accadien des contrats de Suse (Leiden 1962).
Finet, A., L'accadien des lettres de Mari (Bruxelles 1956).
Gelb, 1. J., Morphology of Akkadian (Chicago 1952).
-, Old Akkadian Writing and Grammar (Chicago 1961).
Salonen, E., Untersuchungen zur Schrift und Sprache des Altbabylonischen
von Susa (Helsinki 1962).
Ungnad, A., Grammatik des Akkadischen, 3. Auf I. (Miinchen 1949).
von Soden, W., Der hymnisch-epische Dialekt des Akkadischen, in ZA 40
(1932), 163-227; 41 (1933), 90-183, 236.
-, GrundrifJ der akkadischen Grammatile (Roma 1952).
C. North-West Semitic
Aistieitner, J., Studien zur 1!'rage der Sprachverwandtschaft des Ugaritischen,
in Acta Orientalia 7 (1957), 251-307; 8 (1958), 51-98.
-, Untersuchungen zur Grammatik des Ugaritischen (Berlin 1954).
Albright, W. F., The Early Alphabetic Inscriptions from Sinai and their
Decipherment, in BASOR 110 (1948), 6--22.
Altheim, F. und Stiehl, R., Die aramiiisehe Sprache unler den Achaimeniden
(Frankfurt 1960ff.).
Bibliography 173
Bauer, H. und Leander, P., Grammatik des Biblisch-Aramiiischen (Halle
1927).
Bauer, H. und Leander, P., Historische GrammaUk der hebriiischen Sprache
(Halle 1922).
Beer, G. und Meyer, R., Hebriiische Grammatik, 2 vol. (Berlin 1952-55).
Birkeland, H., The Language of Jesus (Oslo 1954).
BohI, F. M. T., Die Sprache der Amarnabriefe mit besonderer Beriicksichtigung
der Kanaanismen (Leipzig 1909).
Brockelmann, C., Syrische Grammatik, 7. Aufl. (Leipzig 1955).
Burchardt, M., Die altkanaaniiischen 1!'remdworte und Eigennamen im
Agyptischen (Leipzig 1909-10).
Cantineau, J., Grammaire d 1 ~ palmyrenien epigraphique (Le Caire 1935).
.-, Le nabateen, 2 vol. (Paris 1930-32).
Dahood, M., The Linguistic Position of Ugaritic in the Light of Recent
Discoveries, in Sacra Pagina, I (Paris-Gembloux 1959), 269-79.
Dalman, G., Grammatik des judisch-paliistinischen Aramiiisch, 2. Aufl. (Leipzig
1905).
Dammron, A., Grammaire de l'arameen biblique (Strasbourg 1961).
Dhorme, E., Dechiflrement des inscriptions pseudo-hieroglyphiques de Byblos,
in Syria 25 (1946-48), 1-35.
-, La langue de Canaan, in RB 22 (1913), 369-93; 23 (1914), 37-59,
344-72.
Ebeling, E., Das Verbum der El-Amarna-Briefe (Leipzig 1910).
Epstein, J. N., A Grammar of Babylonian Aramaic (Jerusalem 1960).
Friedrich, J., Kanaaniiisch und Westsemitisch, in Scientia 84 (1949), 220-23.
--, Phonizisch-punische Grammatik (Roma 1951).
Garbini, G., Ilsemitico di nord-ovest (Napoli 1960).
-, L'aramaico antieo (ANLM, ser. 8, vol. 7, fasc. 5) (Roma 1956).
_ .., Nuo1)o materiale per la grammatica dell'aramaico antico, in RSO 34
(1959), 41-54.
Gardiner, A. H., Once again the Proto-Sinaitic Inscriptions, in JEA 48 (1962),
45-48.
Gelb,1. J., La lingua degli Amoriti, in ANLR ser. 8, 13 (1958), 143-64.
Gesenius, W. und Bergstrasser, G., Hebriiische Grammatik, 29. Auf I. (Leipzig
1918-29).
Ginsberg, H. L., The Classi/,ication of the North-West Semitic Languages, in
AIOK XXIV, 256-67.
Goetze, A., Is Uga-rit'ic a Canaanite Dialect?, in Language 17 (1941), 127-38.
Gordon, C. H., Ugaritic Manual (Roma 1955).
Goshen-Gottstein, :1\'1.. H., Linguistic Structure and Tradition in the Qumran
Docmnents (Jerusalem 1958).
Harris, Z. S., A Grammar of the Phoenician Language (New Haven 1936).
-, Development of the Canaanite Dialects (New Haven 1939).
Janssens, G., Contribution au dechiffrement des inscriptions pseudo-hUro-
glyphiques de Byb/08, in La n01welle CUo 7-9 (1955-57), 361-77
Joiion, P., Grammaire de l'hebreu biblique, 2eme ed. (Rome 1947).
174
Bibliography
Kutscher, E. Y., Studies in Galilean Aramaic (Jerusalem 1952).
_, The Study of the Aramaic Grammar of the Babylonian Talmud, in
Leshonenu 26 (1962), 149-83.
Leander, P., Laut- und Formenlehre des Agyptisch-AramiUschen (Goteborg
1928).
Margolis, M. L., Lehrbuch der aramiiischen Sprache des babylonischen Talmuds
(Miinchen 1910).
Martin, M., A Preliminary Report after Re-Examination of the Byblian
Inscriptions, in Or 30 (1961),46-78.
_, Revision and Reclassification of the Proto-Byblian Signs, in Or 31 (1962),
250-71, 339-83.
Moran, W. L., The Hebrew Language in its Northwest Semitic Background,
in The Bible and the Ancient Near East (London 1961), 54-72.
Moscati, S., Il semitico di nord-ovest, in SOLDV II, 202-21.
_, Sulla posizione linguistica del semitico nord-occidentale, in RSO 31
(1956), 229-34.
NOldeke, Th., KurzgefafJte Syrische Grammatik, 2. Auf I. (Leipzig 1898).
-, Mandaische Grammatik (Halle 1875).
Petermann, J. H., Brevis Linguae Samaritanae grammatica, litteratura,
chrestomathia cum gwssario (Berlin 1873).
Polotsky, J. R., Studies in Modern Syriac, in JSS 6 (1961), 1-32.
Rosenthal, F., A Grammar of Biblical Aramaic (Wiesbaden 1961).
_, Die aramaistische Forschung scit Th. Noldeke's Veroffentlichungen
(Leiden 1939).
_, Die Sprache der palmyrenischen Inschriften und ihre Stellung innerhalb
des Aramaischen (Leipzig 1936).
Rossell, W. H., A Handbook of Aramaic Magical Texts (Ringwood Borough
1953).
Rowley, R. H., The Aramaic of the Old Testament (London 1929).
Schulthess, F., Grammatik des christlich-palastinischen Aramaisch (Tubingen
1924).
Segal, M. R., A ~ t a m m a r of Mishnaic Hebrew, 2nd ed. (Oxford 1958).
Segert, S., Die Sprache der moabitischen Konigsinschrift, in ArOr 29 (1961),
197-268.
Sobelmann, H., The Proto-Byblian Inscriptions: a Fresh Approach, in JSS
6 (1961), 226-45.
Stevenson, W. B., Grammar of Palestinian Jewish Aramaic (Oxford 1924).
Ullendorff, E., The Position of Ugaritic within thc Framework of the Semitic
Languages, in Tarbiz 24 (1954-55), 121--25.
_, Ugaritic Marginalia, in Or 20 (1951), 270-74.
_, Ugaritic Marginalia II, in JSS 7 (1962), 339-51.
van den Branden, A., Anciennes inscriptions 8 emitiques , in BO 17 (1960),
218-22.
_, Le dechiffrement de8 inscriptions protosinaitiques, in al-1I1achriq 52 (1958),
361-95.
_, Le8 inscriptions protosinaUiques, in OrAn 1 (1962), 197-214.
Bibliography 175
D. South-West Semitic
Beeston, A. F. L., A Descriptive Grammar of Epigraphic South Arabian
(London 1962).
Brockelmann, C., Arabische Grammatik, 14. Auf I. (Leipzig 1960).
Cantineau, Accadien et sudarabique, in BSLP 33 (1932), 175--204.
Caskel, W., Lihyan und Lihyanisch (Kiiln-Opladen 1954).
Cohen, D., Le vocabulaire de base semitique et le classement des dialectes
meridionaux, in Sem II (1961), 55-84.
Dillmann, A., Grammatik der Athiopischen Sprache, 2. Aufl. von C. Bezold
(Leipzig 1899), Eng!. trans!. by J. A. Crichton (London 1907).
Fleisch, H., L'arabe classique. Esquisse d'une structure linguistique (Beyrouth
1956).
-, Trait!! de philologie arabe, I (Beyrouth 1961).
Flick, J., 'Arabiya. Untersuchunge'n zur arabischen Sprach- und Stilge8chichte
(Berlin 1950), French transl. by Cl. Denizeau (Paris 1955).
Gaudefroy-Demombynes, M. et Blachere, R., Grammaire de l'arabe classique,
3eme ed. (Paris 1952).
HOfner, M., Altsudarabische Grammatik (Leipzig 1943).
Leslau, W., South-East Semitic (Ethiopic and South-Arabic), in JAOS 63
(1943), 4-14.
-, The Position of Ethiopic in Semitic: Akkadian and Ethiopic, in AlOK
XXIV, 251-53.
Littmann, E., Syria. IV. Semitic Inscriptions C. Safa'itic Inscriptions
(Leyden 1943).
Moscati, S., Nordarabico, sudarabico, etiopico, in RSO 34 (1959), 33-39.
Praetorius, F., Grammatica Aethiopica (Karlsruhe-Leipzig 1886).
Petracek, K., A Study in the Structure of Arabic, in Orientalia Pragensia
1 (1960), 23-38.
Rabin, C., Ancient West Arabian (London 1951).
Ullendorff, E., The Ethiopian Languages and their Contribution to Semitic
Studies, in Africa 25 (1955), 154-60.
van den Branden, A., Les inscriptions thamoudeennes (Louvain 1950).
Les textes thamoudeens de Philby, 2 vol. (Louvain 1956).
E. Proto-Semitic, Hamito-Semitic, Indo-European
Cohen, M., Essai comparatif sur le vocabulaire et la phonetique du chamito-
semitique (Paris 1947).
-, Langues chamito-semitiques, in Les langues du monde, 2eme ed. (Paris
1952), 81--181.
Cuny, A., Invita,tion Ii l'elude comparative des langues indoeuropeennes et des
lungues chamilo-sernitiques (Bordeaux 1946).
Recherches sur Ie vocalisme, Ie consonantisme et la formation des moines
en « N ostratiq1te ,), ancetre de l' I ndo-Europeen et du Chamito- S emitiql1e
(Paris 1943).
Greenberg, J. R., The Afro-Asiatic (Hamito-Semitic) Present, in JAOS 72
(1952), 1-9.
176 Bibliography
Heilmann, L., Camito-se'lnitico e indoeuropeo (Bologna 1949).
Klingenheben, A., Die Prajix- und die Suffixkonjugationen im Hamitosemi-
tischen, in MIOF 4 (1956), 211-77.
Mayer, M. L., Ricerche sul problema dei rapporU fra lingue indouropee e
lingue semitiohe, in Acme 13 (1960), 77-100.
Moscati, S., Sulla piu, antic a storia delle lingue semitiche, in RANL ser.8,
15 (1960), 79-10l.
-, Sulla ricostruzione del protosemitico, in RSO 35 (1960), 1-10.
Pilszczikowa, N., Le haoussa et Ie chamito-semitique a la lumiere de l'Essai
comparatif de JJf.arcel Cohen, in Rocznik Orientalistyczny 24 (1960),
97-130.
Reinisch, L., Das personliche Furwort und die Verbalflexion in den chamito-
semitischen Spruchen (Wien 1909).
Rossler, 0., Akkadisches und libysches Verbum, in Or 20 (1951), 101--107,
366-73.
-, Der semitische Charakter der libyschen Sprache, in ZA 50 (1952), 121-50.
-, Verbalbau und Verbalflexion in den semitohamitischen Sprachen, in
ZDMG 100 (1950), 461-514.
Thacker, T. W., The Relationship of the Semitic and Egyptian Verbal Systems
(Oxford 1954).
Ungnad, A., Das Wesen des Ursemitischen (Leipzig 1925).
Vycichl, Y'll., Gedanken zur agyptisch-semitischen Sprachverwandtschaft, in
Mus 73 (1960), 173-76.
-, Is Egyptian a Semitic Language?, in Kush 7 (1959),27-44.
-, Nouveaux aspects de la langue egyptienne, in BIFAO 58 (1959), 49-72.
F. Language and Script
Cohen, M., La grande invention de l' ecriture et son evolution, 3 vol. (Paris 1958).
Diringer, D., The Alphabet. A Key to the History of Mankind (London 1948).
-, Writing (London 1962).
Driver, G. R., Semitic Writing from Pictograph to Alphabet, Rev. ed. (London
1954).
Fevrier, J. G., Histoire de l'ecriture, Nouv. ed. (Paris 1959).
Gelb,1. J., A Study of Writing (London 1952).
Ullendorff, E., Studies in the Ethiopic Syllabary, in Africa 21 (1951), 207-17.
van den Branden, A., L'origine des alphabets protosinaftique, arabes pre-
islamiques et phenicien, in BO 19 (1962), 198-206.
von Soden, W., Das akkadische Syllabar (Rom 1948).
n. Phonology
A. Preliminaries
Mittwoch, G., Die traditioneZle Aussprache des Athiopischen (Berlin 1926).
Pfeiffer, R., Clues to the Pronounciation of Ancient Languages, in SOLDV II,
338-49.
Bibliography
177
Sperber, A., Hebrew Based Upon Greek and Latin Transliterations, in HUCA
12-13 (1937-38), 103-274.
Hebrew Grammar: a New Approach, in JBL 62 (1943), 137-262.
B. The Phonological System
Aro, J., Die semitischen Zischlaute (1), 8, 8 und s und ihre Vertretung im
Akkadischen, in Or 28 (1959), 321-35.
Bauer, H., Wechsel von p, m, b mit'!! im Ara1naischen und Arabischen, in
ZS 10 (1935), II-13. .
Beeston, A. F. L., Arabian Sibilants, in JSS 7 (1962), 222-33.
-, Phonology of the Epigraphic South Arabian Unvoiced Sibilants, in Trans-
actions of the Philological Society, 1951, 1-26.
. Birkeland, H., The Syriac Phonematic Vowel System, in Festskrift til Prof.
o. Broch (Oslo 1947), 13-39.
Blake, F. R., The Apparent Interchange between a and i in Hebrew, in JNES
9 (1950), 76-83.
Cantineau, J., Cours de phonetique arabe (Paris 1960).
-, Esquisse d'une phonologie de l'arabe classique, in BSLP 43 (1946), 93-140.
-, Essai d'une phonolog"ie de l'hebreu biblique, in BSLP 46 (1950), 82-122.
-, La «mutation des sifflantes >, en sudarabique, in Melanges Gaudefroy-
Demombynes (Le Caire 1935-45), 313-23.
-, Le consonantisme du semitique, in Sem 4 (1953), 79-94.
C. A., T':e Emphatic 1 in Arabic, in Language 32 (1956), 446-52.
FIscher, W., K > S in den sudlichen semitischen Sprachen, in Miinchener
Studien zur Sprachwissenschaft 8 (1956), 25---38.
Fleisch, H., Etudes de phonetiq1w arabe, in MUSJ 28 (1949-50), 225-85.
Fronzaroli, P., La fonetica ugaritica (Roma Hi55).
Goetze, A., The Sibilants of Old Babylonian, in RA 52 (1958), 137-49.
Knudsen, E. E., Cases of Free Variants in the Akkadian q Phoneme, in JCS
15 (1961), 84-90.
LaSo.r, W. S., The Sibilants in Old South Arabic, in JQR 48 (1957), 161-73.
Martmet, A., La palatalisation «spontanee >, de g en arabe, in BSLP 54 (1959)
90--102. '
--, Remarques sur le consonantisme semitique, in BSLP 49 (1953), 67-78.
Morag, S., The Vocalization Systems of Arabic,Hebrew,andAramaic ('S Graven-
hage 1962). .
Moscati, S., Il sistema consonantico delle lingue semitiche (Roma 1954).
-, Preistoria e storia del consonantismo ebraico antico (ANLM, ser. 8, vol. 5,
fasc. 8) (Roma 1954).
Petracek, K., Der doppelte phonologische Charakter des Ghain im klassischen
Arabisch, in ArOr 21 (1953), 240--62.
-, Struktur der semitischen Wurzelmorpheme und der Ubergang 'ain>
ga2n und 'ain > r im Arabischen, in ArOr 23 (1955), 475---78.
-, Zur Artikulation des sogenannten emphatischen lim Arabischen, in ArOr
20 (1952), 509-23.
Polotsky, H. J., Etudes de grammaire gourague, in BSLP 39 (1938), 137-75.
Moscati, Comparative Grammar
12
178 Bibliography
Rabin, C., The Hebrew Development of Proto-Semitic ii, in Tarbiz 30 (1960),
99-111.
Reif, J. A., The Loss of Consonantal Aleph in Ugaritic, in JSS 4 (1959),16-20.
Rosen, H. B., A Marginal Note on Biblical Hebrew Phonology, in JNES 20
(1961), 124-26.
-, ha-'Ibrit se·liinu (Tel Aviv 1956).
Rossler, 0., Ghain im Ugaritischen, in ZA 54 (1961), 158-72.
-, Zur Frage der Vertretung der gemeinsemitischen Laryngale im Akkadischen
(\ = y), in AIOK XXIV, 129-32.
Ruzicka, R., La question de l'existence du if dans les langues semitiques en
general et dans la langue ugaritienne en particulier, in ArOr 22 (1954),
176-237.
Stehle, D., Sibilants and Emphatics in South Arabic, in JAOS 60 (1940),
507-43.
Ullendorff, E., The Semitic Languages of Ethiopia. A Comparative Phono-
logy (London 1955).
Vilencik, Y., Welchen Lautwert hatte g, im Ursemitischen'l, in OLZ 33 (1930),
89-98.
von Soden, W., Aramaischesl} ersoheint im Spatbabylonischen vor mauch als g,
in AfO 19 (1959-60), 149.
Vokalfarbungen im Akkadisohen, in JCS 2 (1948), 291-303.
C. Conditioned Phonetic Changes
Bravmann, M. M., Some Aspeots of the Development of Semitic Diphtongs, in
Or 8 (1939), 244-53; 9 (1940), 45-60.
Cowan, W., Arabio Evidence for Proto·Semitio ""/awa/ and ""/6/, in Language 36
(1960), 60-62.
Ruzicka, R., Konsonantisohe Dissimilation in den semitischen Sprachen
(Leipzig 1909).
Speiser, E. A., Secondary Developments in Semitic Phonology. An Application
of the Prinoiple of Sonority, in AJSL 42 (1926), 145-69.
Spitaler, A., Zur Frage der Geminatendissimilation im Semitischen, in Zeit-
sohrift fur Indogermanisohe Forsohungen 61 (1954), 257-66.
D. Syllable and Accent
Birkeland, H., Akzent und Vokalismus im Althebraischen (Oslo 1940).
-, Altarabische Pausalformen (Oslo 1940).
-, Stress Patterns in Arabic (Oslo 1954).
Blake, F. R., Pretonic Vowels in Hebrew, in JNES 10 (1951), 243-55.
Brockelmann, C., N euere Theorien zur Gesohichte des Akzents und des Vokalis-
mU8 im Hebraischen und Aramaischen, in ZDMG 94 (1940), 332-71.
Bmnno, E., Studien uber hebraische Morphologie und V okalismus (Leipzig 1943).
Bush, F. W., Evidenoe from Mill}amah and the Masoretic Text for a Penulti-
mate Accent in Hebrew Verbal Forms, in Revue de Qumran 2 (1959-60),
501-14.
Bibliography
179
J:, ,la place de l' acoent de mot en hebreu et en arameen biblique,
III d etudes orientales 1 (1931), 81-98.
-, Elimination des syllabes breves en hebreu et en arameen biblique, in Bulletin
d'etudes orientales 2 (1932), 125-44.
Goetze, A., Accent and Vocalism in Hebrew, in JAOS 59 (1939), 431-59.
Sarauw, C., Uber Akzent und Silbenbildung in den alteren semitischen Sprachen
(Kopenhagen 1939).
Segal, J. B., The Diacritical Point and the Aocents in Syriac (London 1953).
III. Morphology
A. Preliminaries
Botterweck, G. J., Der 'l'riliterismus im Semitisohen (Bonn 1952).
Brockelmann, C., Senbitische Keimwortbildungen, in ZS 5 (1927), 6-38.
J. H., The Patterning of Root Morphemes in Semitio, in Word 6
(1900), 162-81.
Heller, J., Neuere Literatur zur Biliterismus-Frage, in ArOr 27 (1959)
678-82. '
A. H., Analogy in the Semitio Languages (Baltimore 1901).
HurwItz,. S. T. H., Root·Determinatives in Semitic Speech (New York 1913).
KurylowlCz, J., L'apophonie en semitique (Warszawa 1961).
Moscati, S., n biconsonantismo nelle Zingue semitiche, in Bib128 (1947),113-35.
Petracek, K., Die innere Flexion in den semitischen Spraohen, in ArOr 28
(1960), 547-606; 29 (1961), 513-45; 30 (1962), 361-408 (wird fort-
gesetzt).
B. The Noun
Aartun, K., Zur Frage des bestimmten Artikels im Aramaischen, in Aota
Orientalia 24 (1959), 5-14.
Barth, J., Die Nominalbildung in den semitischen Spraohen, 2. Aufl. (Leipzig
1894),
Baumgartner, W., Das hebraische Nominalprafix mi-, in ThZ 9 (1953)
154-57. '
Bravmann, M. M., Genetic Aspects of the Genitive in the Semitic Languages
in JAOS 81 (1961), 386-94. '
-, On a Case of Quantitative Ablaut in Semitic, in Or 22 (1953), 1-24.
-, The Plural Ending ·ut- of Jlasculine Attributive Adjectives in Akkadian
in JCS 1 (1947), 343. '
Brockelmann, C., Diminutiv und Augmentativ im Semitischen, in ZS 6 (1928)
109--34. '
Cantineau, J., La notion de scheme et son alteration dans diverses langues
semitiques, in Sem 3 (1950), 73-83.
Cazelles, H., La mimation nominale en Ouest-Semitique, in GLECS 5 (1951),
79-81.
Christian, V., Die Entstehung der semitisohen Kasusendungen, in ZS 3 (1924),
17-26.
Moscati, Comparative Grammar
12a
180 Bibliography
Cohen, M., Noms d' et de plantes a prefixe n en ethiopien, in GLECS 5
(1951), 85-87.
Entretiens sur la determination et l'indetermination, in GLECS 5 (1948-51),
73-76, 78, 81-82, 88-96, 98.
Feghali, M. et Cuny, A., Du genre grammatical en semitique (Paris 1924).
Fleisch, R., Le nom d'agent fa'al, in MUSJ 32 (1955), 165-72.
Friedrich, J., Der Schwund kurzer Endvokale im Nordwestsemitischen, in
ZS 1 (1922), 3-14.
Gelb, 1., La mimazione e la nunazione neUe lingue semitiche, in RSO 12
(1930),217-65.
Goetze, A., The Akkadian Masculine Plural in -anUfi and its Semitic Back-
ground, in Language 22 (1946), 121-30.
Kurylowicz, J., La mimation et l'article en arabe, in ArOr 18, 1-2 (1950),
323--28.
Lagarde, P. de, Ubersicht iiber die im Aramaischen, Arabischen und He-
braischen iibliche Bildung der Nomina (Gottingen 1889).
Lemoine, E., Theorie de l'emphase (Paris 1951).
Leslau, W., Ethiopic Denominatives with Nominal Morphemes, in Mus 75
(1962), 139-75.
Loretz, 0., Die hebraische Nominalform qattal, in Bibl 41 (1960), 411-16.
Matou8, L., Zum sog. inneren Plural im Arabischen, in ArOr 24 (1956),
626-30.
Moscati, S., n plurale esterno maschile neUe lingue semitiche, in RSO 29
(1954), 28-52.
-, Lo stato assoluto deU'aramaico orientale, in AION 4 (1962), 79-83.
-, On Semitic Case-Endings, in JNES 17 (1958), 142-44.
-, Plurali interni in ugaritico?, in RSO 32 (1957), 339-52.
Nyberg, R. S., Wortbildung mit Priifixen in den semitischen Sprachen, in
Le monde oriental 14 (1920), 177-289.
Rin, S., The Termination -6 in the Plural Absolute, in Leshonenu 25 (1961),
17-19.
Speiser, E. A., The "Elative" in West-Semitic and Akkadian, in JCS 6
(1952), 81-92.
Studies in Semitic Formatives, in JAOS 56 (1936), 22-46.
-, The Terminative-Adverbial in Canaanite- U garitic and Akkadian, in
IEJ 4 (1954), 108-15.
Tagliavini, C., Alcune osservazioni sul primitivo valore della mimazione e nuna-
zione nelle lingue semitiche, in Donum natalicium Schrijnen (Chartres
1929), 240-90.
Torczyner, R., Die Entstehung des semitischen Sprachtypus, I (Wien 1916).
Troupeau, G., Le scheme de pluriel Fu'lan en arabe classique, in GLECS 7
(1955), 65-66.
von Soden, W., Status rectus-l?ormen vor dem Genitiv im Akkadischen und
die sogenannte uneigentliche Annexion im Arabischen, in JNES 19 (1960),
163-71.
Wagner, E., Die erste Person Dualis im Semitischen, in ZDMG 102 (1952),
229-33.
Bibliography
181
Wehr, R., Der arabische Elativ (Wiesbaden 1953).
Young, E. J., Adverbial -u in Semitic, in Westminster Theological Journal
I?, 2 (1951), 151-54.
C. The Pronoun
Barth, J., Die Pronominalbildung in den semitischen Sprachen (Leipzig 1913).
Bravmann, M. M., Hebrew stayim in the Light of Syriac and Turkic, in
Proceedings of the American Society for Jewish Research 21 (1952), 1-2.
Caspari, W., Zum hebraischen Demonstrativ, in ZS 7 (1931), 41-52.
Castellino, G., Observations on the Akkadian Personal Pronouns in the Light
of Semitic and Hamitic, in MIOF 5 (1957), 185-218.
-, The Akkadian Personal Pronouns and Verbal System in the Light of
Semitic and Hamitic (Leiden 1962).
'Greenberg, R., An Afro-Asiatic Pattern of Gender and Number Agreement, in
JAOS 80 (1960), 317-21.
Kienast, B., Das Personalpronomen der 2. Person im Semitischen, in AlOK
XXIV, 253--55.
-, Erwagungen zu einer neueren Studie iiber semitische Demonstrativa, in
Or 26 (1957), 257-68.
Loewenstamm, S. E., On Ugaritic Pronouns in the Light of Canaanitic, in
Leshonenu 23 (1958-59), 72-84.
Rosen, H.B., Zur Vorgeschichte des Relativsatzes im Nordwestsemitischen, in
ArOr 27 (1959), 186-98.
Rundgren, F., Uber Bildungen mit (s)- und n-t-Demonstrativen im Semi-
tischen (Uppsala 1955).
Trager, G. L. and Rice, F. A., The Personal Pronoun System of Classical
Arabic, in Language 30 (1954), 224--29.
D. The Numeral
Barth, J., Zur Flexion der semitischen Zahlworter, in ZDMG 66 (1912),
94--102.
Bauer, R., Noch einmal die semitischen Zahlworter, in ZDMG 66 (1912),
267-70.
Cantineau, J., Le nom de nombre «six» dansleslangues semitiques, in Bulletin
des etudes arabes 13 (1943), 72.
Loewenstamm, S_ E., The Development of the Term "First" in the Semitic
Languages, in Tarbiz 24 (1954-55), 249-51.
Reckendorf, R., Der Bau der semitischen Zahlworter, in ZDMG 65 (1911),
550-59.
von Soden, W., Die Zahlen 20-90 im Semitisohen und der Status absolutus
in WZKM 57 (1961), 24--28. '
E. The Particles
Eitan, 1., Hebrew and Semitic Farticles, inAJSL 44 (1928), 177-205,254--60;
45 (1929), 48--63, 130-45, 1117-211; 46 (1930), 22-50.
Garbini, G., La congiunzione semitica *pa-, in Bibl 38 (1957), 419--27.
182 Bibliography
Guidi, I., Particelle interrogative e negative neUe lingue semitiche, in A Volume
of Studies presented to E. G. Browne (Cambridge 1922), 175-78.
Sarna, N. M., The Interchange of the Prepositions beth and min in Biblical
Hebrew, in JBL 78 (1959), 310-16.
F. The Verb
Ahrens, K., Der Stamm der schwachen Verba in den semitisc'hen Sprachen, in
ZDMG 64 (1910), 161-94.
Bauer, H., Die Tempora im Semitischen (Berlin 1910).
Blake, F. R., A Resurvey of Hebrew Tenses (Roma 1951).
Bravmann, M. :M., Notes on the Forms of the Imperative in Hebrew and Arabic,
in JQR 42 (1961), 51-56.
Brockelmann, C., Die" Tempora" des Semitischen, in Zeitschri/t fur Phonetik
5 (1951), 133-54.
Brzuski, W. K., Note sur les themes Ii seconde radicale graphiquement redoublee
en sudarabique epigraphique, in Rocznik Orientalistyczny 25 (1961),
127-31.
Christian, V., Das Wesen der semitischen Tempora, in ZD:MG 81 (1927),
232-58.
Cohen, :M., Le systerne verbal semitique et l'expression du temps (Paris 1924).
Dahood, M., Some Aphel Causatives in Ugaritic, in Bib138 (1957), 62-73.
Dombrowski, B. W. W., Some Remarks on the Hebrew Hitpa'el and Inver-
sative -t- in the Semitic Languages, in JNES 21 (1962), 220-23.
Driver, G. R., Problems of the Hebrew Verbal System (Edinburgh 1936).
Fleisch, H., Les verbes Ii allongement vocalique interne en semitique (Paris 1944).
-, Sur le systeme verbal du semitique commun et son evolution dans les
langues semitiques anciennes, in :MUSJ 27 (1947-48), 39-60.
Goetze, A., The So-Called Intensive of the Semitic Languages, in ,lAOS 62
(1942), 1-8.
-, The Tenses of Ugaritic, in JAOS 58 (1938), 266-309.
Hammershaimb, E., Das Verbum im Dialekt von Ras Schamra (Kopenhagen
1941).
Heidel, A., The System of the Quadriliteral Verb in Akkadian (Chicago 1940).
Janssens, G., De werkwoordelijke "Tijden" in het semietisch, en in het bizonder
in het hebreeuws, in JEOL 15 (1957-58), 97-103.
Jirku, A., Eine 'Aj'e/.Form im Ugaritischen?, in AfO 18 (1957), 129-30.
Kienast, B., Das Punktualthema *yaprus und seine Modi, in Or 29 (1960),
151-67.
-, Verbalformen mit Reduplikation im Akkadischen, in Or 26 (1957), 44-50.
-, Weiteres zum R·Stamm des Akkadischen, in JCS 15 (1961), 59--61.
Kurylowicz, J., Esquisse d'une tMorie de l'apophonie en semitique, in BSLP
53 (1957-58), 1-38.
-, Le systeme verbal du semitique, in BSLP 45 (1949), 47-56.
Leslau, W., Le theme verbal frequentatif dans les langues ethiopiennes, in
RES 1939, 15-31.
-, Le type verbal qat ala en ethiopien meridional, in MUSJ 31 (1954). 15-98.
Bibliography
183
Martin, 'V. J., Some Notes on the Imperative in the Semitic Languages, in
RSO 32 (1957), 315-19.
M e y ~ r , R., Das hebraische Verbalsystem im Lichte der gegenwartigenForschung,
III Congress Volume Oxford 1959 (Leiden 19(0), 309-17.
-, Spuren eines westsemitischen Prasens-Futur in den Texten von Chirbet
Qumran, in Von Ugarit nach Qumran (Berlin 1958), 118-26.
-, Zur Geschichte des hebraischen Verbums, in VT 3 (1953), 225-35.
Moran, W. L., Early Canaanite yaqtula, in Or 29 (1960), 1-19.
Moscati, S., Il participio passivo in semitico, in RSO 37 (1962), 51-57.
Miiller, W. W., Die Wurzeln Mediae und Tertiae y!w im AZtsudarabischen
(Tiibingen 19(2).
Praetorius, F., Zur Kausativbildung im Semitischen, in ZS 5 (1927), 39-42.
Rossler, 0., Die Priifixkonjugation Qal der Verba 1M Nun im Althebraischen
. und das Problem der sogenannten Tempora, in ZA W 74 (1962), 125-41.
-, Eine bisher unbekannte Tempusform im Althebraischen, in ZDMG III
(1961), 445-51.
Rowton, M. B., The Use of Permansive in Classic Babylonian, in JNES 21
(1962), 233-303.
Rundgren, F., Das althebraische Verbum. AbrifJ der Aspektlehre (Uppsala
19(1).
-, Das aUsyrische Verbalsystem, in Sprdkvetenskapliga Sallskapets i Uppsala
Forhandliger 1958-60, 49-75.
-, Der aspektuelle Charakter des aZtsemitischen Injunktivs, in Orientalia
Suecana 9 (1960), 75-101.
--.:, Intensiv und Aspekt-Korrelation. Studien zur athiopischen und akka-
dischen Verbalstammbildung (Uppsala-Wiesbaden 1959).
Saydon, P. P., The Conative Imperfect in Hebrew, in YT 12 (1962), 124-26.
Sola-Sole, J. M., L'infinitif semitique (Paris 19(1).
Speiser, E. A., The Durative Hithpa'el: a tan-Form, in JAOS 75 (1955),
118-21.
Stinespring, W. F., The Active Infinitive with Passive Meaning in Biblical
Aramaic, in JBL 81 (1962), 391-94.
Vycichl, W., Ein passives Partizip qatil im Agyptischen und Semitischen, in
ZD:MG 109 (1959), 253-57.
von Soden, W., Tempus und Modus im Semitischen, in AIOK XXIV,
263-65.
-, UnregelmafJige Verb en im A7ckadischen, in ZA 16 (1952), 163-81.
'Wernberg-:Moller, P., Observations on the Hebrew Participle, in ZA'iV 71
(1959), 54-67.
184 Lexicography
Lexicography
A. North-East Semitic (Akkadian)
Delitzsch, Fr., Assyrisches Handw6rterbuch (Leipzig 1896).
Gelb,1. J., Glossary of Old Akkadian (Chicago 1957).
Muss-Arnold, W., A Concise Dictionary ot the Assyrian Language. Assyrisch-
Englisch-Deutsches Handworterbuch, 2 vol. (Berlin 1894-1905).
von Soden, W., Akkadisches Handworterbuch, unter Benutzung des lexi-
kalischen N achlasses von Bruno MeijJner bearbeitet (Wiesbaden 1959 ff.).
The Assyrian Dietionary of the University of Chicago. Editorial Board
J. Gelb, Th. Jacobsen, B.Landsberger, A.L. Oppenheim (Chicago 1956ff.).
B. North-West Semitic
Brockelmann, C., Lexicon Syriacum, ed. !lecunda (Halis Saxon. 1928).
Brown, F., Driver, S. R., Briggs, C. A., A Hebrew and English Lexicon of the
Old Testament (Oxford 1906).
Dalman, G., Aramaisch-Neuhebraisches Handworterbuch zu Targum, Talmud
und Midrasch, 2. Aufl. (Frankfurt 1922, Reprint 1938).
Gesenius, W., Hebraisches und AramaischesBandworterbuch uber das Alte
Testament, bearb. von Frants Buhl. Unveranderter Neudruck der 1915
erschienenen 17. Auflage (Berlin-Gottingen-Heidelberg 1950).
Jean, Ch. F. et Hoftijzer, J., Dictionnaire des inscriptions semitiques de
l'Ouest (Leiden 1960 ff.).
Koehler, L. et Baumgartner, VV., Lexicon in Veteris Testamenti libros
(Leiden 1953).
-, Supplementum ad Lexicon in Veteris Testamenti libros (Leiden 1958).
Levy, J., Neuhebraisches und Chaldaisches Worterbuch, 4. vol. (Leipzig
1876-89).
Margoliouth, G.P., Supplement to the Thesaurus Syriacus of R. Payne Smith
(Oxford 1927).
Payne Smith, R., Thesaurus Syriacus, 2 vol. (Oxon. 1868-97).
Schulthess, Fr., Lexicon Syropalaestinum (Berlin 1903).
Zorell, F., Lexicon Hebraicum et Aramaicum Veteris Testamenti (Roma
1940ff.).
C. South-West Semitic
Belot, J. B., Vocabulaire arabe-franl)ais Ii l'usage des etudiants (Beyrouth
1883, many reprints).
Conti Rossini, C., Chrestomathia Arabica ]j[eridionalis Epigraphica (Roma
1931), 99-261.
Dillmann, A., Lexicon Linguae Aethiopicae (Lipsiae 1865, Reprint New York
1954).
Freytag, G. W., Lexicon Arabico-Latinum, 4 vol. (Halis Saxon. 1830-37).
Lexicography. List of Abbreviations used in the Text
185
Grebaut, S., Supplement au Lexicon Linguae Aethiopicae de A. Dillmann
et edition du Lexique de Juste d'Urbin (1850-1855) (Paris 1952).
Lane, E. W., Arabic-English Lexicon, 8 vol. (London 1863-1893, Reprint
1955).
Wehr, H., A Dictionary of Modern Written Arabic, ed. by J. Milton Cowan
(Wiesbaden 1961).
Worterbuch der Klassischen Arabischen Sprache. Auf Grund der Sammlungen
von A. Fischer, Th. Noldeke, H. Reckendorf u. a. Quellen hrsg. durch
die Deutsche Morgenlandische Gesellschaft (Wiesbaden 1957ff.).
List of Abbreviations used in the Text
The abbreviations used for reviews are those of the Bibliographie
semitique of Orientalia. Those used for books are as follows:
Brockelmann, SG = Brockelmann, C., Syrische Grammatik, 7. Aufl. (Leipzig
1955).
Brockelmann, GVG = Brockelmann, C., GrundrijJ der vergleichenden Gram-
matik der semitischen Sprachen, 2 vol. (Berlin 1908-13).
Dillmann, EG = Dillmann, A., Ethiopic Grammar, 2nd ed. (London 1907):
Fleisch, TPA = Fleisch, H., Traite de philologie arabe, I (Beyrouth 1961).
Garbini, SNO = Garbini, G., Il semitico di nord-ovest (Napoli 1960).
Gelb, OA = Gelb,1. J., Old Akkadian Writing and Grammar (Chicago 1952).
Gordon, UM = Gordon, C. H., Ugaritic Manual (Roma 1955).
Gray, SCL = Gray, L. H., Introduction to Semitic Comparative Linguistics
(New York 1934).
Rabin, WA = Rabin, C., Ancient West Arabian (London 1951).
Ullendorff, SLE = Ullendorff, E., The Semitic Languages of Ethiopia.
A Comparative Phonology (London 1955).
von Soden, GAG = von Soden, W., GrundrijJ der akkadischen Grammatik
(Roma 1952).

Table of Contents
page

Preface............ .................................. ...... ..
I. The Semitic Languages ............................... . . . .

1 3 3 3 3 5 6 7 7 8 9 10 10 11 12 13 13 13 15 15 15 16 17 17 22 22

A. Scope of the Survey. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Definition ... .................................... 2. Classification.......................... . . . . . . . . . . . . 3. Nature and Extent of the Survey. . . . . . . . . . . . . . . . . . . .

B. North-East Semitic ................................... C. North-West Semitic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. General Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2. The Languages of the Second Millennium B.C. ........ 3. Canaanite......................................... 4. Aramaic....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . a. Old Aramaic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . b. West Aramaic .............................. " . . c. East Aramaic.... . .... .. .. .......... ........ .... D. South-West Semitic. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. General Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Arabic (incI. South Arabian). . . . . . . . . . . . . . . . . . . . . . . . 3. Ethiopic...................................... . . . .

E. Proto-Semitic, Hamito-Semitic, Indo-European. . . . . . . . . . .
1. Proto-Semitic........ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Hamito-Semitic ..... ................ ..... ..... .. .. 3. Hamito-Semitic and Indo-European. . . . . . . . . . . . . . . . . .

F. Language and Script .................................. II. Phonology...............................................
AIle Rechte vorbehalten © Otto Harrassowitz, Wiesbaden 1964, 1969, 1980 Photographische und photomechanische Wiedergaben jeder Art nur mit ausdriicklicher Genehmigung des Verlages Gesamtherstellung: Hessische Druckerei GmbH, Darmstadt Printed in Germany

A. Preliminaries ........................................ . B. The Phonological System .............................. .
1. Classification ..................................... . 2. The Proto-Semitic Consonantal System ............. . 3. Bilabials ........................................ .

23 23
24 24

VI 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

Table of Contents Interdentals ...................................... . Dental Plosives .................................. . Nasal, Lateral, and Rolled Dentals ................. . Dental and Palato-alveolar Fricatives ............... . Velar Plosives .................................... . Velar Fricatives .................................. . Pharyngal Fricatives; Laryngals .................... . Synopsis of the Consonantal System ................ . Semivowels ...................................... . Vowels .......................................... . a. Akkadian ..................................... . b. North-West Semitic of the Second Millennium B.C .. c. Canaanite ..................................... . d. Aramaic ...................................... . e. Arabic ........................................ . f. Ethiopic ...................................... . 14. Diphthongs ...................................... .

Table of Contents

VII
80 82 82 83

27 31 31 33 37
38

41
43 45 46

3. 4.

47 48 49 51 52
53

5. 6.

Patterns Extended by Prefixes .................. . Patterns Extended by Infixes ................... . Patterns Extended by Suffixes ................... . Patterns from Roots with One, Two, Four, and Five Radicals ...................................... . Gender .......................................... . Number ......................................... . a. External Masculine Plural ...................... . b. Internal Masculine Plural ....................... . c. Feminine Plural ............................... . d. Plurals of Biconsonantal Nouns .................. . e. Dual ......................................... . Declension ........................................ Definiteness and Indefiniteness ..................... .

c. d. e. f.

84
86

87
88

91 92 93 94 96
102 102 106 III
113 114 115 115 lI8

C. The Pronoun 2. 3. 4. 50.

.........................................

54

1. Independent Personal Pronouns .................... .

C. Conditioned Phonetic Changes ......................... . 1. 2. 3. 4. 5. 6. 7. 8. Assimilation ..................................... . Dissimilation ..................................... . Prosthesis ....................................... . Anaptyxis ....................................... . Syncope and Contraction .......................... . Haplology ....................................... . Metathesis ....................................... . Sandhi .......................................... .

56 56
58 59

Personal Pronoun Suffixes ......................... . Demonstrative Pronouns .......................... . Relative Pronouns ................................ . Interrogative and Indefinite Pronouns .............. .

60 61 62 63 63

D. The Numeral ........................................ .
1. Cardinals ........................................ . 2. Ordinals and Fractions ............................ .

E. The Particles ........................................ .
1. Adverbs ......................................... .

120 120 121 121 122
122

D. Syllable and Stress ................................... .
1. Syllabic Constitution .............................. . 2. Stress and Associated Changes ..................... . 3. Sentence Stress ................................... .

63 63 65 69
71 71 71

2. Prepositions ...................................... . 3. Conjunctions ..................................... . 4. Interjections ..................................... . F. The Verb ............................................ .

III. Morphology ............................................ . A. Preliminaries ........................................ . 1. Morphemes ...................................... . 2. The Proto-Semitic Root ........................... . 3. Morphological Development ........................ . B. The Noun ........................................... .
1. Themes or Patterns ............................... .

1. Verbal Themes or Stems

...........................

72 75 75 75 76
77

2. Nominal Patterns ................................ . a. Simple Patterns ............................... . b. Patterns Extended by Gemination or Reduplication of Radicals .................................... .

78

a. Simple Stem .................................. . b. Stem with Doubled Second Radical .............. . c. Stem with Lengthened First Vowel .............. . d. Stems with Prefixes SO, h-, '- ................... . e. Stem with Prefix n- ............................ . f. Stems with Prefix (or Infix) t- ................... . g. Other Stems .................................. . h. Verbs with Four and Five Radicals .............. . 2. The "Tenses" 3. The ~:[oods ...................................... . 4. Inflexion .........................................
•••••••••••• 0 ••••••••••••••••••••••••

122 122 124
124

125 126 127 129 130 131
134

137

... but on the other hand... or even equally so.. Verbs with Pronominal Suffixes ....... would inevitably be open to criticism from all sides.... when the directors of Porta Linguarum Orientalium did me the honour of inviting me to VlTite this book (an honour for which I offer them my sincere thanks). d........ It must be admitted that many of these decisions are but imperfect ones................... after all.. then anyone can make a running criticism of the result and easily cite imperfections and omissions..... I was but too well aware of the need for someone at long last to undertake the task. Verbs with Identical Second and Third Radicals .... 7... the classification of the various languages.... .... Moscati..... 8.... Simple Stem: Nominal Forms . .. the dates up to which the bibliography can be taken into account: these questions. 8..........1-15... called for decisions. the solution of the problem of a conventional root for schemes and for paradigms... Derived Stems: Infinitive ......... Derived Stems: Prefix-Conjugation ..... Verbs with Pharyngals and Laryngals ..... Verbs with First Radical n ........ On the one hand. .......... Comparative Grammar 5.. b.... 12...... it must avoid doubtful and disputed hypotheses while indicating certain of the lines of research being pursued and certain more important divergences of opinion.... The limits in time and space of the material to be taken into consideration... It is thus an elementary introduction to the comparative grammar of the Semitic languages.. ....... To these general principles dependent upon the present series...... to make selections.... in particular §§ 1... 10. Simple Stem: Imperative ... ...... As is well known............ e. .66-96........ . 156 157 158 159 160 163 164 166 168 168 169 Bibliography ........... Semantic Categories in "Weak" Verbs ...... 6. List of Abbreviations used in the Text . Simple Stem: Prefix-Conjugation .. h....... ....... it must concentrate upon the essential facts while mentioning various particular questions of special importance. 9.... a. The so-called "Weak" Verbs ... Derived Stems: Imperative . involving as it did the digestion of an immense linguistic and bibliographical material and its condensation into an exposition which by reason of its brevity could not but be partial imd simplified. adopted in view of the particular situation of Semitic linguistics.......... Doubly Irregular or Defective Verbs ...... intended primarily as a textbook and limited in its scope so as to serve for a beginners' course. Hence. Derived Stems: Participle . 137 141 145 146 147 154 Preface The nature of this book and the principles which guided its composition depend in the first place on the series to which it belongs...... introduction to the Bibliography etc....1-9....... Derived Stems: Suffix-Conjugation .. Simple Stem: Suffix-Conjugation .................. 'When the author is obliged to sift his matter.. and these decisions are set forth in the course of the exposition (cf. it was equally clear to me that the accomplishment of this task. Verbs with w..... .. y ..... .. not everyone is ready to recognize the sacrifice represented by the undertaking of a thankless task whose accomplishment is.. Such an introduction must be clear in expression while respecting scientific terminology. . the reasons for abstaining from a systematic treatment of syntax and of vocabulary........... ............ .......... 11........ and others too..3. g.. .... .. .. i...... .. ...... others must be added..... indispensable for the common good........... furthermore...VIII Table of Content...... I was of two minds...).... so as to give an adequate notion of present-day scientific trends. c... the manner for reconstructing the presumed common forms and tracing their development in the various languages.... 171 185 1 ....... the choice of a system of transliteration. . an introduction to the comparative grammar of the Semitic languages has been a desideratum for several decades... .... and the need for such an introduction as a textbook was urgent. to confine his exposition within strict limits..... but one may well doubt whether different decisions would have been any more felicitous......... f. 12. ....... 6..

they have at their disposal an elementary textbook . and I then requested certain eminently qualified colleagues to lend me their aid by snggesting. I. Migliorini. and other peoples. that our colleagues. in reading this book. 1. Syria-Palestine.2 Preface If. were almost inevitable. Segert. at all events. of course. and syntax. Having assembled the list of additions and alterations thus proposed. and I have not always been able to accord full weight to their proposals.1. for his advice in the area of general linguistics.nevitable deficiencies connected with such an enterprise. after so many decades. the adjective "Semitic" was applied to all the languages of the group. and characterized by a large number of common elements in their phonology. indeed. that our colleagues will take into consideration the fact that at last.of the comparative grammar of the Semitic languages. who most generously responded to my appeal for their collaboration. Spitaler.3. I proceeded to a new and fuller redaction of the book. Definition 1. and I think that my collaborators would agree with me. or generally originating from that area. however. Syria-Palestine. on the basis of Gen. therefore. with particular but not exclusive regard for each one's speciality. Arabia. 10-26. Rome 1960). The Semitic languages occupied in ancient times the following regions of western Asia (from east to west): Mesopotamia. it is therefore only just that I bear the entire responsibility for this work. We trust. instead of dwelling upon the ad:mitted manifold defects will duly reflect upon the fact that in a work such as this these defects. I was thus able from the start to foresee the endless difficulties and the i.2. morphology. Classification October 1962 Sabatino Moscati 1. and Ethiopia constitute. including those subsequently discovered. L. lowe a special gratitude to the Publishers. an abundance of additions and alterations. had generally been referred to as "Oriental languages". Now that the book is written. clear that not in every single case was it possible to bring my views into harmony with those of my three collaborators. Schl6zer (1781) as a designation of the languages spoken by the Aramaeans. and to them I express my heartfelt gratitude. however. Finally. they characterize and set apart a linguistic group possessed of a remarkable degree of internal unity. suggest the idea of a common origin. even in not a few cases in which I was not in agreement with them. It is. On the coast that lies opposite southwestern Arabia. However. I was likewise able to endeavour to reduce those difficulties and deficiencies to a minimum. for the reasons already given. The Semitic Languages A. the ancient habitat of the Semitic 1* . Arabia. for having so willingly undertaken the not inconsiderable labour involved in a work composed with the collaboration of several persons. The procedure which I adopted was the following: I first composed as best I could a preliminary and summary draft (Lezioni di linguistica semitica. The adjective "Semitic" was brought into use by A. who has likewise advised me in this respect. namely Professors K. the Hebrews. E. to whom I offer my sincere gratitude. if he wished. These elements. de Mauro. and to Professor T. 21-31. "Ve trust. lowe a like debt of gratitude to Professor B. bearing in mind the material thus furnished.although necessarily defective . These colleagues. Rabin and S. I can in all frankness declare that I find it unsatisfactory. the Arabs. Scope of the Survey 1. been recognized long before Sch16zer's time. so that the work is truly the result of a collaboration. waves of migration led to the occupation by Semitic populations (and thus by Semitic languages) of another region: Ethiopia. Other eminently qualified colleagues have likewise read my preliminary draft and favoured me with their suggestions and opinions. they also share certain common tendencies in their evolution. write dozens of pages of criticism of the book. Mesopotamia. Petracek. of course. and that different solutions of the problems involved would in all probability have been equally unsatisfactory. The name "Semitic" is conventionally applied to a group of languages spoken in western Asia. this obliged me to make a certain selection. von Soden. C. Ullendorff and W. Each one of us could. I must add that I have very largely taken their proposals into account. 10. are Professors A. 11. preserved despite lapse of time and change of place. but the group itself had not as such been identified and marked out: these languages. Once introduced. vocabulary. 2. The affinities existing between the various languages had. like others in Asia.

and South-West Semitic (Arabia and Ethiopia).5).7. for the purposes of linguistic inquiry.4. to Nabataean and Palmyrene within Aramaic (cf. Nature and Extent of the Survey 1. The present survey has as its object the whole body of the classical Semitic languages. § 5. The position of certain languages in these classification schemes has been. either by reason of archaic elements which these modern tongues may have preserved or by virtue of any other relevant characteristics. Arabic.9) and of Amorite (cf. Similar interest. § 3. § 3.6. partly because of the elementary nature of this treatise and partly on acc~)Unt of the somewhat backward state of studies in those fields. while it places its main emphasis on the position in the great literary languages (Akkadian. Within the limits of time and space already described. 1. and. the exclusion of modern dialects and especially of secondary developments due to them. Arabic. North-West Semitic (Syria-Palestine). or conquest. Ethiopic).. on the general Hamito-Semitic level.). On the whole. however. Particular connexions have repeatedly been postulated between individual languages of different areas. to South Arabian as between Arabic and Ethiopic (cf. Data drawn from modern dialects will only be taken into consideration when they have a contribution to make to an examination of the earlier phases of the classical languages.1) as well as their historical development in the principal languages of the group. etc. . Arabs in Africa and on the islands of the Indian Ocean. when some of the great literary languages (Syriac.5. it adduces evidence from other Semitic languages and modern dialects only when of particular relevance-as is indeed appropriate to a survey of an essentially elementary character. Ethiopic) begin to flourish and to exert an influence well beyond those limits.8) as between the western and the eastern languages. the object of controversy. or historical plane. 1. functional. Moreover. 1. migration. typological. Within these limits the linguistic material of the present survey extends in time over some four thousand years: from the third millennium B. and these features become further accentuated in the course of time.S.3). would attach to the connexions which have been claimed to exist between Akkadian and Libyco-Berber (cf. secondly. 3. From this limitation the following consequences will inevitably result: in the first place. The grouping of the Semitic languages is usually based on their geographical distribution: North-East Semitic (Mesopotamia).C. In the western area. colonization. this applies to the position of Ugaritic (cf. etc.1). It is obvious that this division is closely connected with that of the peoples who spoke those Semitic languages. 1. §4. § 3. a division which simply projects ethnical entities on to the linguistic plane is bound to be imprecise. but the problem remains as to how such affinities can be tested on the statistical. Syriac. and still is. for instance. Schemes of classification which identify and distinguish groupings within the Semitic area on the basis of specific bundles of isoglosses may derive support from structural. the distinction between the northern and the southern languages is more evident in their fully developed form than at the archaic stage (but it must be borne in mind that our knowledge of the archaic phase is fairly limited). until the first millennium A. it may however be said that the geographical division indicated above corresponds tolerably well (though not without certain exceptions) to the distribution of gross linguistic features. Akkadian and Ethiopic: such relationships may be explained in accordance with the principles of linguistic geography. the present study aims at a reconstruction of the earliest phonological and morphological units (Proto-Semitic. when we encounter the earliest manifestations of a Semitic language (Akkadian).e. § 3. the relegation of a specific treatment of syntax and vocabulary.D.4 The Semitic Languages Scope of the Survey 5 languages. Geographically. the present work does not aim at exhaustiveness and. and genetic criteria. Hebrew. East Semitic exhibits from the outset certain independent characteristics as compared with West Semitic (cf.. Of some interest in this respect are certain common elements as well as points of contrast which have been observed between.2). the bulk of the material originates from the region indicated above (§ 1. while a limited portion of it comes from outside that area as a result of the spread of Semitic populations (Assyrians in Anatolia. Beyond this area they have spread only as a result of secondary developments. i. Phoenicians on the coasts and islands of the Mediterranean-even as far as the Atlantic.16). for which see § 5.

It is the custom of grammars and works of introduction to the Semitic languages to divide the languages of this area into two main groups. according to the "short" chronology). the dialect of the southern part of the region.C. This language. The principal phases of Akkadian are: 2. and New Babylonian (about 1000 B.C. North-West Semitic displays notable internal variations which reflect the rather chequered history of Syria and Palestine.8). there existed a lesser degree of variation in the local speech-forms of the North-West Semitic area.1.West Semitic 1. North-East Semitic 2. and eventually replaced. (Amorite.C. spoken in Mesopotamia in the pre-Christian era. the capital of the empire of Sargon the Great (2350-2294 B.C. is divided into Old Babylonian (about 2000-1500 B. of course.). § 3. The most recent phase of New Babylonian (from about 600 B. This does not mean.C. approximately. derives its name from that of the city of Akkade. till the beginning of the Christian era).). B.). For a detailed description of the Semitic languages (which is outside the scope of this study) the reader is referred to the standard introductory works.e.C.2. 2400 to 1700 B.). cf. because in that phase some of the isoglosses which distinguish the two groups had not yet been clearly drawn. In the earliest historically attested phase of North-West Semitic many of the distinctive features that were later to contrast it with South-West Semitic had not yet been clearly realized-at least so far as our limited knowledge goes. Middle Assyrian (about 1500 to 1000 B. Hence it may be sur- . which in the civilization of that region prevailed over. with texts principally of Cappadocian origin. C.5. there must have existed a further Semitic language for which he proposed the name "Old-Amorite".1.3. owing to the limited extent of our documentation (very few texts from Assyria are extant) it is at present impossible to establish clearly defined differentiations of dialect.9. N orth. from the first millennium B. a) Old Akkadian may be dated between 2500 and 2000 B. the following principal dialects can be distinguished: 2. but merely that such distinctions manifested themselves in forms different from those current in the following millennium (Garbini. The division into Canaanite and Aramaic can be assessed only from the time when Aramaic made its historically and epigraphically attested appearance.C. that in the second millennium B. and even then the homogeneity of Aramaic is not matched by a comparable measure of uniformity in the languages grouped under the general heading of "Canaanite" (of. General Characteristics 3.C.) with several dialectal variations.. About 2000 B. the Canaanite and the Aramaic languages. in Babylonia as well as in Assyria (and differing considerably from the spoken language) may be referred to as Later Babylonian ("Jungbaby~ lonisch"). the dialect of the northern part of the region. 2.6 The Semitic Languages North-West Semitic 7 1.4. while the literary language used between about 1400 and 500 B. i. pp. i. the non-Semitic Sumerian.C.2. the lastnamed is strongly aramaicized in its final phase.. characterized by the infiltration of Aramaic words and linguistic peculiarities (cf. and New Assyrian (about 1000-600 B.C.C. 2. SNO).C. The following observations are restricted to what is necessary for the clarification of the concepts and nomenclature adopted in this inquiry-especially in view of changing notions and terminological fluctuations current at the present time. is more specifically called Late Babylonian ("Spatbabylonisch").) occur hundreds of Semitic names which cannot be explained either on the basis of Akkadian or of "East Canaanite" . b) Babylonian.e. But recent studies tend to show that this division is not in accord with the most ancient phase of Syro-Palestinian linguistic history (second millennium B. § 3. These names as well as other linguistic phenomena led von Soden (WZKM 56 [1960].). In Old Akkadian and Old Babylonian texts (ca. 3.C.C. 185-91) to the conclusion that in the second half of the third millennium B.C. Middle Babylonian (about 1500 to 1000 B. § 3. A grammatical sketch of Old Amorite (from which certain peculiarities of U garitic might possibly be explained) has not yet been produced. is divided into Old Assyrian (about 2000-1500 B.).12).18).C. c) Assyrian. North-East Semitic is represented by Akkadian.

from the end of the second millennium B. Along with the group of texts just mentioned there exists another one (also belonging to the second millennium B. The Languages of the Second Millennium B.4). we can onset of other North-West Semitic languages which. precisely for that reason.C. c) A series of short inscriptions originating for the most part from Lachish and attributable to various dates in the second millennium (between 1800 and 1300 B. originally assigned by Dunand to the end of the third millennium B. 3.). and in particular in the Mari texts (speech-forms not lacking in internal variegation which our inadequate knowledge does not as yet allow us to appreciate fully). so much so. East Semitic was at every stage notably differentiated from its western neighbour.1). . previously assigned to about 1800 B.11. § 3. With these languages. which is reflected in the proper names and in certain linguistic peculiarities of the Akkadian 3.S. onwards. These texts include: 3. The coherence or independence of Canaanite (except in so far as it is clearly distinguished from Aramaic) appears somewhat limited.C. 0 3. c) The language of the glosses (usually called "Canaanite") in the Tell Amarna letters (l4th cent. The most ancient form of North-West Semitic may be placed in the second millennium B.C. approx. the distinction between Canaanite and Aramaic may properly be introduced.C. The data drawn from this material are supplemented by a further group of Egyptian transcriptions belonging to the second half of the second millennium. b) The Proto-Sinaitic inscriptions.3. § 1. a group of texts of doubtful date and interpretation which.C. 3. also § 2.C. It embraces. the language of the texts discovered at Uaarit (Ras Samra) and belonging to the fourteenth and thirteenth centuries B. The Canaanite languages are: 3. be largely attributable to our deficient means of ready linguistic identification. There has been much discussion as to the typological placing of this language within the framework of the Semitic hinguages (cf. Canaanite 3. 2.8 The Semitic Languages North-West Semitic 9 mised that in its first historical manifestations West Semitic displayed a greater degree of unity than can be discerned in its subsequent development when it was subjected to powerful disintegrating tendencies (of. _ b) Ugaritic. Canaanite represents the non-Aramaic linguistic manifestations of the Syro-Palestinian area.9.12. Similar considerations apply to the Akkadian texts from Ugarit (l4th-13th cent. texts of the period of the First Babylonian Dynasty. 3. will be dealt with in the following paragraphs. but more recently dated by Albright about 1500.) whose interpretation and dating may be said to be generally established. but more recently dated by some scholars several centuries later and regarded linguistically as the most ancient manifestation of Phoenician.). in the first place. smce theIr full development belongs almost entirely to the first millennium. The languages of these texts are: 3. This may. that the very individuality of the group has been questioned by some scholars (Friedrich).C.C.) which are written in an Akkadian showing many Canaanite peculiarities.5). Towards the end of the second millennium B.S. B. an inappropriate term used to designate the North-West Semitic tongues of the first half of the second millennium). They are considered (but this is a matter of pure hypothesis which reflects the uncertainty of all schemes of classification) by Albright as North-West Semitic and by van den Branden as Proto-Arabic (representing a linguistic stage prior to the differentiation between North and South Semitic). will not normally be discussed in this treatise.5. however. The data drawn from the Akkadian texts are corroborated by transcriptions· of North-West Semitic names of persons and places in Egyptian execration-texts (cf. in fact. a) Amorite (also called "East Canaanite".C. ~iscern t~e 3.4).10.". a) The pseudo-hieroglyphic inscriptions of Byblos.4.C. 3. and within the limits set forth above (of. § 1.

C. Arpad. and Persian empires (seventh to fourth centuries B.22-26.Palmyrene inscriptions have been found as far afield as England. this inscription.47 [two words]. Of these inscriptions two from Sam'al (one of Panamuwa I and one of Bar-Rakib) are of special importance.nd exegetical literature of the Middle Ages and of modern times.C. 3.C. however.C. belonging to the central Palestinian dialect.. nowadays spoken in Israel. Aramaic mascus.C. Nabataean papyri have been discovered among the Dead Sea documents. b) Palmyrene is the language of an (ethnically) Arab population which established a state at Palmyra and flourished between the first century B. found in certain parts of the Old Testament (Gen.18. 3. beginning with the apocryphal literature and the documents recently discovered near the Dead Sea (second and first centuries B. and the third century A. represented by the inscriptions of the ancient Phoenician cities (to be dated between the tenth and the first centuries B. and the third century A. b) Phoenician and Punic. are of particular importance and constitute what has been termed Egyptian Aramaic. Hama.) and by those of their Mediterranean colonies (between the ninth century B. a) Old Aramaic is the language (with some dialectal variants) of the most ancient inscriptions originating from Da- . West Aramaic 3. represented by the inscription of King M~sa' of Moab of the ninth century B. the poetical. having possibly been drawn up by an Israelite in the service of the King of Moab.17. of the fifth and fourth centuries B. and represent the type of Aramaic known as Ya'udic (derived from the name of the state of Sam'al Ya'udi). in a few remnants.C. western India. Babylonian. 2. .8-6. in the present survey these languages will.. 7. Midras). and Nabataean inscriptions have been identified as far afield as Greece and Italy.10 The Semitic Languages North-West Semitic 11 3. might however be regarded as a Hebrew text. under the name of Old Aramaic. We distinguish between an ancient phase..11.22. and Assyria.D. and Egypt. and the second century A.D.. Aramaic forms a considerable and wide-spread linguistic group whose earliest manifestation goes back to the beginning of the first millennium B.31.20. and include. T6sefta. c) Jewish Palestinian Aramaic is the language that was spoken in Palestine at the time of Christ and during the first centuries of the Christian era. Ezra 4. up to the first century B. and belonging to the period between the tenth and the eighth centuries B. be subsumed under "West Aramaic". Eviclence of this comes from Mesopotamia (Aramaic inscriptions and proper names.15. 3. West Aramaic (which appears to be a more direct continuation of Old Aramaic) and East Aramaic. a) Hebrew-including: the Biblical period whose literature may be dated approximately between 1200 and 200 B. Nabataean and Palmyrene. c) Moabite. Some scholars are inclined to ascribe the division into two branches to the second or the third century A. Anatolia.) and continued by certain offshoots into the period which followed. a.19. the postBiblical period.18.13.4-7. In literary sources it is found in the Genesis A pocryphon (discovered among the Dead Sea documents) and the Palestinian Targiim (of which a complete manuscript has 3. and finally Modern Hebrew. Sam'al. Arabia. 3. 4. Old Aramaic 3.28). Jer.) and continuing with the rabbinical writings of the first centuries of the Christian era (Misna. a.21.16. and a subsequent division into two branches. Dan. philosophical.10. b) Classical or Imperial Aramaic is the language used under the Assyrian.14. 3.D.C. a) Nabataean is the language of an (ethnically) Arab population which established a state at Petra and flourished between the first century B.C. to the present day.C. owing to their independent characteristics. The papyri and ostraca from Egypt.D.C. according to the latest study (Segert). and which is supplemented by a number of short inscriptions. the age of these documents ranges probably from the fifth to the second centuries B.C. c) A type of Classical or Imperial Aramaic is represented by Biblical Aramaic.).C. b. Persia. and which survives. 3. words and constructions in New Assyrian and New Babylonian texts).C.

. 3.D. It is true that historically Ethiopic makes its appearance as a successor tongue of South Arabian. laid as it was upon a non-Semitic substratum.29. b) Babylonian Aramaic is the language of the Babylonian Jews. if intended to contrast the former with the latter and to place it alongside Ethiopic. by Arabic during the great Islamic conquests of the 8th cent. e) Christian Palestinian Aramaic is the language used by the Melkites between the fifth and the eighth centuries A. Gospel Lectionaries.2. a) Syriac. although it was generally replaced. Survivals of East Aramaic can still be found in the neighbourhood of Lake Urmia. 3. in particular. 4. prominently represented in the Babylonian Talmiid (fourth to sixth centuries A.28.26. D. Ethiopic..23.3. Arabic (incl.24. General Characteristics 4. but such a genetic relationship differs to some extent from the requirements of a descriptive classification which is based on the convergence of isoglosses. Gubb'adin and BalJ'a. d) Samaritan Aramaic is the language of the Samaritan Targiim to the Pentateuch (probably of the fourth century A.1. and liturgical writings. It has. and the other by the Galilean variety (some Midrasim and the Jerusalem Talmiid). their writings extend from the third to the eighth century A. recently been pointed out by some scholars that (at least within the limits of the ancient period) the separation of South Arabian from North Arabic. From the descriptive point of view it has been noted that ancient South Arabian is in several respects in agreement with North Arabic and at variance with Ethiopic (and vice versa). Jewish Palestinian Aramaic survives above all in a sizable body of Jewish post-Biblical texts of the second to the fifth centuries A.1. cf.West Semitic 1.2.D. Limited and gradually disappearing survivals of West Aramaic can still be heard in the villages of Ma'liila. later developed a rich Christian literature extending from the third to the thirteenth century A.D. in the neighbourhood of Damascus. § 3.D.. these may be divided into two groups.25. at ']:'iir 'Abdin.D. as a result of the first world war. 2. as a spoken language. will present South Arabian within the area of the Arabian peninsula-without thereby implying an undervaluation of its independent characteristics or of its agreements with Ethiopic. is not entirely justified. East Aramaic 3. 3. South-West Semitic is. has been the subject of several recent studies (Tsereteli).D. one being represented by the Targiimim of Onkelos and of Jonathan. 3.) and of some later writings. it is written in Syriac characters and is attested in several Old Testament passages. The present exposition. For the reasons explained in § 4. As regards the relationship between South-West and NorthWest Semitic. This complex. containing many dialectal divergences. however. 3. has undergone certain developments not to be found in South Arabian. c. It should be mentioned that the "Assyrians" (as these Aramaic-speaking populations are referred to) were displaced. South Arabian) 4. The Aramaic of Georgia.12 The Semitic Languages South-West Semitic 13 been identified in the Vatican Library by Diez Macho). based on geographical principles. and now live in scattered communities in the United States and in Russia. c) Mandaean is the language of the Gnostic sect of the Mandaeans who flourished in Mesopotamia. 3.27. may be subdivided as follows: .D. in grammars and introductions to Semitic languages. South. usually divided into two groups: (1) North Arabic and (2) South Arabian together with Ethiopic. we have chosen to take the term "Arabic" as a linguistic complex embracing all the tongues of the Arabian peninsula-with the exception of some Aramaic infiltrations (Nabataean and Palmyrene) in the extreme north. and near Mosul. originally the language of Edessa.) and in a series of magical texts composed in the fifth and sixth centuries A.

Sl].6.D. By Proto-Semitic (or Common Semitic or simply Semitic) we refer to the ensemble of elements which an examination of the historically documented Semitic languages leads us to regard as common property of the Semitic group in its most ancient phase (Semitic isoglosses). 4.14 The Semitic Languages Proto-Semitic. (3) Qatabanian. It is reasonable to suppose that the dialectal fluctuations with which we are confronted by the existing historical evidence formed part also of the pre-historic phase (in contrast to the concept of the genealogical tree). the studies of J. Pirenne) to the sixth century A. in fact.9. Ancient Ethiopic (or Gg'gz) is first attested in epigraphic material of the first few centuries A. (2) Minaean. and Soqotri. In the South Arabian area there exists a separate group of languages which. (4) I. Harari.D.). The large number of dialects developed from Classical Arabic are most appropriately classified according to regional groupings: Central-Asian. and the fourth century A.1.C. Iraqi.D. are being distinguished: (1) Sabaean. The concept of Proto-Semitic would seem comparable to that of Proto-Indo-European. Egyptian. The generally attested form of classical Arabic is the result of a process of systematization by Arab grammarians. but this uncertainty is not necessarily an obstacle to comparative inquiry. The modern Arabic dialects are numerous and will only be dealt with peripherally in this treatise (0£.5). Syro-Lebanese and Palestinian. 3.8.awri. appear more manageable owing to the lesser degree of geographical dispersion of the Semitic languages and the greater . (this is subject to considerable current controversy.ri. (approximately) and be divided into the following regional and dialectal groups: (1) Tamudic (a conventional term which lacks precision and covers. cf. Amharic. (3) ~a£aitic. hence we discover here the starting-points for developments peculiar to each individual language.7. Indo-European 1. the "Arabic" par excellence. Whether such postulated reconstructions invariably possessed historical reality it is difficult to determine. predominantly religious.yanite. but such a convention is a necessary pre-requisite for an understanding and reconstruction of linguistic history. The modern Semitic languages of Ethiopia are represented by Tigrifia. this linguistic material is represented by the pre-Islamic standard speech ("Hochsprache") and was nurtured by the ample flow of Arabian dialects. it owes its diffusion and survival to Islam which turned Arabic into a great literary language as a result of the Arab conquests and the enormous expansion of this dynamic religion. § 1. E. Hamito-Semitic. It later developed an extensive. in a few inscriptions and in some dialectal samples preserved by Islamic writers..D. It must not be forgotten that 'Proto-Semitic' is merely a linguistic convention or postulate. (2) Lil). Their dates range from the eighth century B. the following dialects. 4. 4. 4. in the great Aksum inscriptions of the fourth century. Tigre. Indo-European 15 4. Proto-Semitic. c) Classical North Arabic. owing to its long historical severance and its exposure to non-Semitic influences. literature reaching up to modern times.C.4. 5. is Maltese.2. many of the tongues over a wide area of pre-Islamic central and northern Arabia). A separately developed form. according to some scholars. The problems of the former do. represent the continuation and development of the ancient speech-forms: the principal ones among these are Me1).5.1) is the language of the inscriptions of the ancient South-West Arabian city-states. a) Ancient or Epigraphic South Arabian (ESA-for whose independent position cf. Arabian. corresponding to the regions of the principal states. Hamito-Semitic. Ethiopic 4. (5) Awsanian. It attains its full realization in pre-Islamic Arabic poetry and later in the Qur' an (seventh century A. Proto-Semitic 5.Iagrami. North African or Maghrebi. b) Pre-classical North Arabic is the language embodied in a series of inscriptions which may be dated between the fifth century B. Gafat and Argobba are now virtually extinct. however. above all. the remarks in § 4. is attested from the fourth century A. This process has concealed original dialectal divergences as well as other elements of subsequent evolution (Fuck). and Gurage.D. and.

but rather as a haphazard collection of isoglosses not unconnected with the geographical proximity of the two groups and certain historical contacts between them. Language and Script S. of course. Account has to be taken of archaizing tendencies in some languages in contrast to genuinely old material which may at times appear in strangely disguised forms. morphological. F. Comparative Grammar 2 . Consequently. conventionally called Hamito-Semitic. Semitic is. Certain studies (Rossler) have asserted that Libyco-Berber is possessed of an essentially Semitic character and have claimed a particular affinity with Akkadian. to some extent. We have to aim at the reconstruction of Proto-Hamito-Semitic forms. on the other. this larger grouping comprises Egyptian. be impaired. In addition to Semitic. Cuny) hypothesis which is claimed as common ancestor of Hamito-Semitic and Indo-European. etc. It has long been held that Semitic is not an isolated group but forms part of a larger complex of languages. The relationship between the various units of the Hamito-Semitic group cannot be explained as a secondary development. A more profound knowledge of Akkadian. of some of the North-West Semitic languages. Slavonic. Hamito-Semitic and Indo-European 5. there is at present no cogent reason to question the independence of Semitic within the larger Hamito-Semitic complex. Such conjectures are. be regarded as a heritage from a 'parent' language. However. though naturally with all the reservations called for by such a conjectural reconstruction. however. it is well to recall certain essential facts. the independence of Semitic would. therefore. since systems of writing may condition and at times even influence linguistic elements. A few points of contact have long been noticed between Hamito-Semitic and Indo-European languages. of the modern Ethiopian and South Arabian languages. has to some extent modified our ideas about Proto-Semitic and those of the classical tongues which were alleged to resemble it most closely.16 The Semitic Languages Language and Script 17 measure of affinity between them. therefore. or Germanic languages. the similarities adduced in support of the thesis seem to be in part open to question and in part inconclusive. ural uniformity than can be detected among the "Hamitic" languages. on the one hand.S. A more reliable explanation is to be sought in the common Mediterranean environment (especially as regards lexical elements) and consequent historical contacts and influences (particularly marked in Anatolia and the Eastern Mediterranean). for most of the parallels can much more readily be explained within the framework of the long-established general Hamito-Semitic affinity (Cohen). However. If this theory were shown to be correct. 2. It would. Such limited links as may exist between Indo-European and HamitoSemitic should not. thus HamitoSemitic is also sometimes referred to in purely geographical terms as Afro-Asiatic.3. A treatment of Semitic writing lies outside the scope of the present work. There is no "Hamitic" unit comparable to the Semitic one: Semitic possesses a much greater measure of struct. The value and importance of individual Semitic languages to a reconstruction of Proto-Semitic has been variously estimated at different periods in the history of our studies. Libyco-Berber. Moscati. although the inflexional structure appears to be common to both. These are generally concerned with relations of a phonological and especially lexical character and have given rise to the so-called "Aryo-Semitic" (Ascoli) or "Nostratic" (Pedersen. and Semitic with the Romance. Hamito-Semitic 5. and lexical nature.4. The central position long occupied by Arabic as either the proto-type or true image of primitive Semitic has come to be challenged in recent times. The rich phonological structure of Arabic is now paralleled by that of Ugaritic and South Arabian. 5. and its highly developed verbal system is regarded as the result of systematization rather than archaism. 5.5. this is based on correspondences of a phonological. 3. and Cushitic. the group that is more fully attested and generally also the most replete with ancient forms. be more appropriate to compare Hamito-Semitic with Indo-European. and this makes the concept of an original Hamito-Semitic linguistic body one of great cogency. very highly speculative.. especially on account of deep-seated morphological divergences between those groups.1.

Lil). 6. u. is derived from the Phoenician. Among these are the following: 6. purely consonantal in character. § § 8. so that their reading offers considerable difficulties (quite apart from the fact that the tablets are not always in a good state of preservation and that there existed notable divergences in orthography in different areas and times).4. and Arabic. the principle of not denoting vowels was in practice subjected to certain subsequent modifications (cf. In the transliteration of Akkadian (for the purposes of this comparative study) syllables which are separately represented in cuneiform will. Another alphabet whose origin has probably to be sought in the Syro-Palestinian area is that which makes its appearance in the ancient South Arabian inscriptions. 2* . the only one in the Semitic west to use characters of cuneiform type-though alphabetic in structure. not only to the various alphabets of the Aramaic languages but also to the Hebrew "square" script and the classical Arabic alphabet. 6. i. d) In Ethiopic the script has been adapted to denote seven vowels by a variety of changes in the structure of the consonantal symbol.9). North-East Semitic (Akka. 6. This device. The origins of this alphabetic form of writing are to be sought in the Syro-Palest.jafaitic) and. c) In some writing-systems additional signs were later introduced to indicate voweis: this applies to Syriac.e. a) In Ugaritic the consonant' has three forms. therefore. both northern and southern.C. though rarely. 6. i. On the other hand.yanite.18 The Semitic Languages Language and Script 19 6. It is important to realize that length of vowels and doubling of consonants are not consistently expressed in cuneiform.dian) is written in cuneiform characters.10. but its independent evolution gave rise. where these signs are.inian area in the first half of the second millennium B.7. instead of i-na-ad-di-in). at times data drawn from comparative Semitic grammar will prove useful for the reconstruction of relevant features.h. on the one hand. etc. the consonantal inventory of Sumerian (for which this form of writing was originally devised) differed materially from the Semitic sound system. 6. '.. This alphabet has a long history connected with the emergence of the ancient Hebrew as well as the Moabite and Samaritan alphabets.6. However. be joined together in the same word (e. in its turn. After the first attempts (their interpretation is still in doubt: cf. y. even to short ones (East Aramaic).2. and i. according to its vocalization with a. the pre-classical Arabic scripts (1'amiidic.3-6) we witness. of the consonant-signs w. Vowels have thus become an integral part of Ethiopic writing which now assumes a quasi-syllabic character-yet without sacrificing the general Semitic concept of the predominance of consonants over vowels. Hebrew. difficult to reach satisfactory conclusions. the Ethiopic syllabary (see § 6. the Qur'an. on the other. inscribed with a pointed instrument on tablets of clay or more rarely on stone or metal. §§ 8. From about the same period dates the formation of the so-called Phoenician alphabet which was carved on stone and was not of the cuneiform type. 6.3. §§ 3. The script indicates both consonants and vowels which (in contrast to the majority of Semitic alphabets denoting consonants only) is of considerable assistance to our knowledge of the language. used only for certain texts of particular importance (the Bible. It is. 6.73-(6). this form of writing was taken over from the non-Semitic Sumerians who preceded the Semites in Mesopotamia. as has been said. as a rule.69 to (6). is later extended to internal long vowels and sometimes. in the second half of the same millennium. limited at first to final vowels.S. however. is represented in consonantal alphabetic scripts with a limited number of signs (generally less than thirty).g.). Vowel notation by means of the methods just described still leaves certain deficiencies and problems (cf. Closely connected with the latter are.9. West Semitic. The cuneiform system possesses many hundreds of signs which have ideographic or syllabic value and are often multivalent. so that the graphic representation of Akkadian consonants by means of Sumerian writing exhibits many imperfections and difficulties. b) In some alphabets the use of matre8 lectioni8 is developed. The West Semitic alphabets are.5. too. The Aramaic script. the appearance of the Ugaritic alphabet. From this fact arise some of the principal difficulties in the study of the Semitic languages as well as many of the obscure points in our understanding of their comparative grammar. inaddin "he gives".

. b) The system of transliteration has been kept as simple as possible-in accordance with the requirements of an elementary grammar. Such doubling (inconsistently expressed already in cuneiform§ 6. each symbol by one sign. 6.87]). It need hardly be mentioned that proper transcription has not been abandoned without regret. The transliteration of the Semitic languages which is employed in this book is based on certain principles which it seems well to explain beforehand.15. for example.2) lacks specific symbols in the West Semitic alphabets. §§ 8. but it appears to be beyond doubt that any other set of principles would be subject to an equal measure of ambiguity: 6. the fricative articulation of consonants in postvocalic position [of. 6.14. 6. and Arabic have developed a gemination mark along with their general vowel system. 8. the situation in Syriac which possesses no symbol either for 'd or zero vowel. though Hebrew. §§ 8. in order to permit the reconstruction of the original orthography. a) Our mode of graphic representation is.13. although in their pronunciation these consonants became early identified with ~ and s. §§ 8. It is obvious that these principles are open to a great deal of argument and are not exempt from certain disadvantages.20 The Semitic Languages Language and Script 21 We shall here limit ourselves to mentioning the artificial character of seemingly very precise vocalizations. as far as possible. it eschews the notation of sub-phonemic variants (allophones)-except where this is called fo~ by special circumstances. or the ambiguity in Hebrew of the sign which indicates either a vowel of the 'd type or the absence of a vowel (and the similar ambiguity of the sixth order in Ethiopic). such as those of Hebrew and Biblical Aramaic. § 8.10] or the not always consistently 8mployed matres lectionis [cf. but in the case of many of the ancient Semitic languages the conjectural element involved in such a course seemed unjustifiably prominent. for Ethiopic vowels the quantitative indications derived from etymological comparison will be retained-in preference to the purely qualitative distinctions on which the Ethiopic vowel system appears to be based (cf.95 to 96). respectively (of. in Ethiopic the transliterations (j and 8 will be used. and. .37). Non-distinctive variants can generally be determined in accordance with grammatical rules (for example. a transliteration rather than a transcription. unless there are compelling reasons) as well as phonetic and (especially) etymological data which are of considerable importance in a comparative study: thus. for it aims at reproducing. Biblical Aramaic.12. 6. Another notable deficiency in most of the Semitic writingsystems concerns the marking of gemination or consonant-doubling -even though this may be a feature of phonemic significance. in fact.81. finally. 8.11.20. c) The choice of transliteration symbols takes into consideration the usual conventions (which it is well not to alter in an elementary grammar.

In favour of this thesis one might refer to the phenomenon of labialization which is probably also due to Cushitic influence (cf. and laryngals. to denote a quality characteristic of the Semitic (and Hamito-Semitic) languages: velarization in Arabic. the articulatory. acoustic. because: a) the Ethiopic "emphatics" are voiceless . Arabic. that the Ethiopic glottalized ejectives are primary. now generally accepted in linguistic work. According to others. especially Cantineau's studies). Preliminarie8 7. and our reconstructions are by and large schematic and conventional. 8. i. pharyngals. i. Sometimes. Hebrew. classification is related to the place and the manner of articulation. velars.e. consequently. Aramaic and Hebrew for Akkadian. moreover. It is. Their classification may be based either on the musical principles of acoustic phonetics or on the physiological elements of articulatory phonetics. Phonology it. interdentals. more exactly.3. b) testimony of grammarians-for languages in which a grammatical tradition exists (Hebrew. and auditory characteristics of actual speech (phonetics) and.) and vice ver8a. Classification 8. on the contrary. e) comparative Semitic linguistics (reconstruction of Akkadian. Within the groups so classified a further distinction may be made according to the voiced or "emphatic" character of some consonants.2. 7. and nasals. between phonetics and phonemics: on the one hand. the fact that glottalization is not found in Semitic outside Ethiopic. tradition proves to be an obstacle rather than a help-as in the case of classical Ethiopic whose pronunciation has been preserved by speakers of languages derived from B. there are reasons for maintaining. this term is used. yet occurs in certain Cushitic languages. the analysis of data obtained by means of a study of distinctive oppositions (cf. the complex of phonetic expressions-of the ancient Semitic languages is reconstructed and assessed on the basis of indications of various kinds: a) traditional pronunciation-in the case of languages which have been transmitted (living or otherwise) up to the present time (Hebrew. The Phonological SY8tem 1. difficult to arrive at concrete results beyond the indications furnished by the historically attested languages. Greek and Latin for Hebrew and Phoenician-Punic. As regards sonority.1. § 8. would point to its being a secondary feature. According to some scholars. The Semitic phonological system is made up of consonants. and vowels as well as certain stress patterns. d) orthographic peculiarities indicative of phonetic characteristics not otherwise expressed. however. It is uncertain which of these types is primary. The pronunciation-or. fricatives. etc. Similar considerations apply to Hebrew whose transmission is affected not only by differences between the Sephardi and Ashkenazi pronunciations but also by influences due to the substratum-languages of those who have gone back to the use of Hebrew as a spoken tongue. etc. Syriac.e.1. according to the manner of articulation: plosives (or stops). doubts and uncertainties. As regards "emphasis". palato-alveolars. on the other. laterals (and lateralized consonants rolled consonants. with the result that that pronunciation embodies developments peculiar to these modern languages.43). In the latter case. Arabic). For ancient languages reconstruction is of necessity phonemic. or ancient South Arabian pronunciations based on parallels in Arabic. phonemes. doubts have been raised as to its nature in Proto-Semitic and it has been termed a "correlation of tenseness or pressure" (Cantineau). glottalization in Ethiopic. not altogether properly. A. According to the place of articulation we have: bilabials. c) transcriptions of Semitic words and phrases in other languages (Greek. This body of evidence is neither complete nor sufficient. 7. semivowels. There remain.2. n. minimal distinctive sound units relevant to meaning (phonemics). Ugaritic. dentals. Ethiopic). Aramaic. A phonological study of the Semitic languages must be based on the clear distinction. Egyptian.).The Phonological System 23 H.

6. Bilabials 8. and in Egyptian and Cushitic likewise both p and t occur. Ethiopic possesses a p in addition to the (probably derived from Proto-Semitic p). next to the character (?) whose shape it had imitated. npk "well" from Sem. since Semitic counterparts are lacking. t s. nbk. for Proto-Semitic one single phoneme may be posited. in the language manifested in the Egyptian transcriptions of the second millennium). sps "sun" from Sem.g. the correspondences tend to show that the two phonemes of these languages had as their Semitic counterpart the single phoneme p (southern f): hence the differentiation in these languages seems to be of a secondary character. h Laryngal m Greek words (e.g. Likewise secondary. t. of rare occurrence. GAG.Heb. p.g. voiceless p and voiced b. e. it is usually employed to transcribe Greek words (e. Following the distinctions of voice and of "emphasis".Syr. 8. (1" t t. Interchanges between consonants of the bilabial group. 3. for the purposes of its graphic expression. As regards Egyptian and Cushitic. c) the phenomenon q >' in some Arabic dialects can only be explained by way of glottalization. Proto-Semitic has two bilabial plosives. the form of another ejective sound (though quite different in its basis of articulation) was slightly modified: p was then placed. so the Semitic "emphatics"-almost without exception. In Ya'udic the change p > b is attested: ¢ 8 r n 8. and in any case. and its late appearance in the Ethiopic alphabet is corroborated by its place at the tail of the Ethiopic syllabary. Ar.e. Akk. This is likely to have been p rather than t. However. g Velar . The Proto-Semitic consonantal system may be hypothetically reconstructed as follows (with such reservations as will be expressed below) : Plosive Fricative Lateral Lateralized? Rolled Nasal p. q. Thus we find in Ugaritic p:b:m: e. in certain positions of the classification-chart. g lP. the juxtaposition of "triadic" groups. it should probably be regarded as secondary. 12). In the South Semitic languages p is replaced by the voiceless labiodental fricative t: e. and similarly used for transcriptions from Greek. voiceless-voiced-emphatic. c) for the velar plosives. in the order of the alphabet. is borrowed from Greek p. Its shape. it is rare and usually confined to transcriptions of t . sms. The problem therefore arises whether these two consonants did not co-exist already in Proto-Semitic. b) for the dental plosives and the dental fricatives.8. pariiqlitos = nae6. pqd "to seek". and rp" influenced by the proximity of other emphatic consonants or of back vowels (positional variants). The Proto-Semitic Consonantal System 8. 8.24 Phonology The Phonological System 25 and. signs pointing to a similar development-not conditioned by the position of the consonant-in the northern Semitic area. Such "triads" exist: a) for the interdentals. again apart from Arabic. as well as between them and the bilabial semivowel w. z. Pharyngal "If.Eth. b) the Ethiopic "emphatics" do not appear to influence the timbre of neighbouring vowels and.).7. this seems to be the norm in the Semitic languages (cf. d. for. tqd.g. however for certain facts in Akkadian von Soden. 8. Ethiopic possesses also a voiceless bilabial plosive p which is emphatic (or ejective) and. pisii = n{aaa). take place in several languages. moreover.3. for it is easier to suppose an evolution p > t than one in the opposite direction (there are.5. This consonant may well be of Cushitic origin. i. and it is worthy of note that. As regards Ethiopic p. is the special p which is to be found in Syriac (James of Edessa) and in Christian Palestinian Aramaic.g.4. it has further a bilabial nasal m. but we are not always in a position to ascribe these interchanges to clearly identifiable linguistic reasons. too. 2. ? Dental S Palato-alveolar k. b Bilabial Interdental t. The new consonant seems soon to have overstepped its original function.xArJTOC. A secondary development in the sphere of emphatic labials may be encountered in some modern Arabic dialects which have emphatic 9. apart from Arabic. we may note. like p. like p.

qpz and Heb. modern Ethiopic (*sb' > Amharic saw "man". attested soon after the earliest period of Akkadian (of.A. New Assyrian appears to have lost any consistent distinction between p and b in pronunciation. Akkadian is in a peculiar situation: the use of m and b for w is frequent.P. This change affects the non-emphatic plosives (p b t d k g) which in postvocalic position come to be articulated as fricatives. Garbini. pp.e. [f v ff 5 x y].10. i. qwz "to leap" compared with Aram. i. the situation in Akkadian is of particular interest. as a lateralized (or lateraI1-Martinet) voiced emphatic interdental. in South Arabian. i. for the cuneiform system of writing does not adequately distinguish between p and b (or between voiced and voiceless consonants generally): never in final position nor in other positions as far as Old Akkadian and Old Assyrian are concerned. Some consonantal interchanges (e. attested in certain languages only. 8. The phonetic and phonological correspondences in the other languages can be explained only by accepting the consonant as Proto-Semitic. f3] in I. but their proto-Semitic status appears to be vouchsafed by the ensemble of correspondences which can be satisfactorily explained on this assumption only.g. von Soden. This consonant is represented in Arabic (where it is usually. I. in turn. Ethiopic. symbols. awilum and abilum "man" in Old Akkadian) suggest the possibility of spirantization. 106) is well represented in modern Aramaic (zabna > *zawna > zona "time". but rather inappropriately. respectively. in Babylonian and in the later phases of Assyrian special symbols are often employed for syllables with p and with b. etc. Mand. As regards the development of the bilabials in the various languages. though to a lesser extent: cf.P. [ep. The transition b >9 >w (of. 21-22. I.A.e. Proto-Semitic had in addition an emphatic interdental. e.P. *rabrabr'ine "great ones" > rawrabr'ine (dissimilation may conceivably have been an additional factor in this process).5) and of non-phonemic character (a sub-phonemic positional variant).e. voiceless t and voiced rj.9. but this does not exclude their having been originally bilabial fricatives. In North-West Semitic (or more precisely in Biblical Hebrew and in the Aramaic of the Christian era) spirantization of p > f. The change to w affects also other labials. Syr. The transition m > b would account for the ESA bn "from" corresponding to Sem. SLE. v] in I.e. b > v occurs as a regular positional variant (the traditional pronunciation represents the resultant consonants as labiodental fricatives. in fact. i. has alternations b:m:w. 'wd "to perish" from Sem. there is no certain proof that it pre-dates the Christian era: neither the Egyptian transcriptions of North-West Semitic names nor Greek and Latin transcriptions of elements in the pre-Masoretic text furnish sufficient indications of the existence of this distinction (cf. i. ~absa and ~awasa.e. GAG. gabra > *gawra > gora "husband"). symbols). Finally.g. Cantineau etc. 8.13. [ff] and [5]. Proto-Semitic has two non-emphatic interdental consonants. 'bd. fricative articulation (of. In several Aramaic dialects we find the change b > w.g. for this is the position in the languages .P. transliterated ?). this consonant retains its independent status in the South Semitic languages only: phonological correspondences would nevertheless suggest its existence in Proto-Semitic. 31-32).A.11. As regards the period when spirantization became operative.12. 8. modern South Arabian (*lbn > lun "white"). Proto-Semitic appears to have possessed a consonant which Brockelmann and those who follow his system transcribe rj. A change of intervocalic m to w is. SNO. nora "he stayed"). it would appear that such signs of non-plosive articulation as we encounter (cf. § 8. p. This is.26 Phonology The Phonological System 27 e. and in Ugaritic by a graphic symbol of its own. 8. regarding it as a voiced emphatic interdental. a conditioned phonetic phenomenon (partial assimilation of consonant to vowel: of. passim). nps. like [f. mn. Interdentals 8. Syr. § 8. qp~.) transcribe it ij. evidently by way of the fricative articulation of b: e. this. These consonants are. It is probable that it was voiced. Ar.10). others (Cohen. too. Apart from the question of its precise articulation (which it is difficult to determine with certainty).6) are not necessarily connected with post-vocalic position. has led to considerable graphic fluctuations. At any rate. of course. probably voiceless (t).g. Ullendorff. nbr > Amh. but owing to the absence of w in Sumerian this is often due to purely graphic reasons rather than phonetic causes. however. nbS "soul" from Sem. § 9.A. 4. "to be weak".

Rossler. Ug. Rulda'u or Rulta'u for the Arabic name of the god Ruif. of course. p. Ug. . ?ill. These developments-as indeed the process t > 8 and iJ >z-continue in Canaanite.rr. OA.28 Phonology The Phonological System 29 which have retained this sound. I> t (given in the table above for Syriac) which applies to other Aramaic languages. ESA 'lJij. Proto-Semitic d has developed into §. however. There are further indications in the more ancient languages: d. 32-33). [ and I retain their independence (cf. i. tellaZa. 'lJiJ. but this phenomenon may conceivably have a different explanation altogether (d. 'ar'a. respectively. (cf. er§etu "earth". Heb.r/: Ug. Examples: [: Akk.14. Syr. (=r/ 1) rf. tawr. Proto-Semitic *iJhb "gold" > Old Aram. ESA twr. 'lJz.z. Reb. 'lJd (iJ). 8.C. 193-94). Ug. 'ly. be conjectured that either the evolutionary process was as yet fluid or that we have to envisage different orthographical phases. According to others. z(}roa'. Eth. for this we possess the evidence of Hebrew as well as that of Phoenician-Punic and Moabite. 8uru "bull". ZA 54 [1961].e.. This state of affairs is to be explained. Garbini. Ar. iJ>d.10. pp. the existence of iJ (as an independent phoneme) is very doubtful. the same symbols as in Canaanite.. Its lateralization appears to be borne out by indications furnished by Arab grammarians and by the evidence of certain modern South Arabian languages. pp. the Proto-Semitic interdentalsfollowing upon the adoption of an alphabet (the Phoenician) which had no proper signs for them. but in some cases it is represented by another sign which we may hypothetically (cf.. Heb. there takes place the change t >t. Il. 158----. In North-West Semitic. (r/) rf. as regards I. Ar. 'lJz "to take". Later. pp. The development might thus be exemplified as follows: Proto-Semitic *ytb "to sit" > Old Aram. Garbini. iJr' "arm". Ug. 8. dr§. The problem remains. Heb. therefore. according to some scholars. 22-23) equate with iJ (d. (as reflected in Egyptian transcriptions) we may note the survival of Proto-Semitic t. iJ. . Heb. while Proto-Semitic 8 is written with the series for 8.iJ: Akk. Ar. it may. In the most ancient Aramaic inscriptions we find the symbols "8". 38). as "Canaanisms" of a purely graphic type-at a time when the characteristic evolution of the Aramaic consonants had already taken place.17. ~~l. ytb. iJ generally merges with d. ~illu "shadow". Syr. Syr. pp.. Akk. UM./ : Akk. at least approximately. ~rr "to be hostile". ESA Eth. about the middle of the first millennium B:C. tawn]" Ar. Ar. Later on the change t > 8 takes place..18). SNO. 80r. 80r. Proto-Semitic *nlr "to guard" > Old Amm. Heb. 8. however. for a different opinion d. iJhb (written "zhb") > common Aram. Ar.72). "z'~. SNO. § 8. r/ are seen to have merged with~. ~(}lalot. In Ugaritic. 'Jy. the fact that it is at times represented by the graphic symbol "g" suggests the possibility of a development analogous to that later . 8. In North-West Semitic of the second millennium B. Certain orthographical divergences have been interpreted as traces of an independent iJ in Old Akkadian (Gelb. [r. nlr (written "n/?r") > 8. The following table (subject. to certain additional explanations and reservations to be set forth below) displays the development of the Proto-Semitic interdentals in the principal Semitic languages: Proto. Some trace of an original differentiation in the Phoenician area has been claimed on the basis of Greek transcriptions (Tveo~ and l:lCJWY) of two place-names which in Phoenician both begin with ~." for t. 'ere§. ESA ?ll "to cover"). the most ancient phase probably reflects the situation in Proto-Semitic. and I. .16. 'rr. iJira'. for it is written by means of the series of symbols for Sumerian 8. Syr.18. Akk. Eth. unresolved.d. Ug. the symbols employed in the most ancient phase are to be regarded as an attempt to represent. Eth. In Old Akkadian t appears to remain independent. t iJ I 8 z ~ ~ t d or iJ I ~ 8 z ~ ~ t d t iJ ? (=/) [ 8 iJ ?(I) z ~ t r/ rf. Syr. 'arrf" ESA 'rrf. perhaps in documents of different dates. rf. Gordon. I. ":. but some spellings with I have encouraged the assumption that this process passed through a prolonged indecisive stage.a'. for example. etc. dhb.). ytb (written "Y8b") > common Aram.Akkadian Ugaritic Hebrew S yriac Arabic E SA Ethiopia Semitic found in Aramaic (cf.

C. The fact that Old Babylonian t is predominantly represented by the graphic element for d is probably due to the inconsistencies of the cuneiform system. Also.C.g. while ¢ > z would certainly be far less likely. a change t > ~ of the type also found in other Semitic languages (including Ethiopic). Egyptian transcriptions of North-West Semitic names (second millennium B. A non-phonemic variant . [~] or [<. fricative q (as in fact happens extensively in modern Ethiopian languages: d. In other positions. i. In the Arabic sphere.1]. while later" . Ullendorff. In this tradition t (conventionally transcribed ?) appears to have become a voiced consonant.C. transliterated ¢. (more relevant. g might have been an intermediate stage in the transition ¢ > '. 64-65). indigenous grammarians have described its original character as a voiced emphatic lateralized interdental. it does exist in Babylonian and in later Assyrian for the majority of the syllabic symbols concerned. SLE. yet South Arabian offers some interesting indications: the spelling of "sl~y" for Sl?y might connote. Proto-Semitic has one nasal dental consonant n. Nasal. voiceless t and voiced d.10). 5.23. in others as a voiced emphatic dental plosive [<. Proto-Semitic has two non-emphatic dental plosives. in the early period. It may be noted that the coalescence in pronunciation of ¢ and ~ may suggest the survival.19. "'rq'" "earth". South Arabian and pre-Islamic North Arabic agree with classical Arabic as far as the retention of the four interdentals is concerned. 8. § 8. It would obviously be impossible to say anything definite about their pronunciation. The lack of differentiation is due to the peculiarity of cuneiform (cf. p. substantially different from the postvocalic fricatives found in North-West Semitic in the first millennium B. 8. are instances of non-positional spirantization encountered in Egyptian transcriptions of NorthWest Semitic names in the second millennium B.20. As for the fourth consonant of the series. 73) considers the possibility that "q" might here be used to represent the articulation of g. soon after the early Aksum inscriptions. cf. t in final position. Lateral. (The term "dental" is used to the exclusion of the interdentals which have already been dealt with. d.30 Phonology The Phonological System 31 common Aram.C. Some consonantal interchanges between t and s suggest the possibility of spirantization (cf. § 8. 'ar¢). In modern dialects its pronunciation is in general the same as that of the previous consonant.21. 8.6..19).9): the available examples (tit'aru "glittering" and sit'aru.9) which evolved some separate symbols for syllables with voiced and voiceless consonants or with emphatic and non-emphatic ones only after 2000 B. Akkadian does not appear to distinguish between t.24.e. later " 'r" ".1]. as well as an emphatic plosive which was probably voiceless and is for that reason transliterated t. and Rolled Dentals 8.g. The reason for the latter deficiency is to be found in the absence in Sumerian (and hence in its writing system as taken over by the Akkadians) of emphatic phonemes. The weight to be attached to these considerations is nonetheless limited. and the respective graphic symbols become liable to arbitrary interchange. and one rolled dental r. however. yet it has to be observed that in northern Babylonia t is generally expressed by t at that period. classical Arabic maintains the four interdental consonants as independent phonemes and also furnishes some traditional indications as to their pronunciation. one lateral dental l. In Ethiopic the pronunciation of ¢ merges with that of ~. Mandaean has preserved "q" in some cases. Proto-Semitic ¢ undergoes a peculiar development in Aramaic: the most ancient inscriptions employ the symbol "q". dbl} = Tebaly.) 8. and the balance of probabilities is clearly on the side of the voiceless character of t. pp. Old Akkadian and Old Assyrian writing likewise lacks the distinction. in some Arabic dialects it is pronounced as a voiced emphatic interdental [~]. tabsutu "midwife" and sabsutu) are. The voiceless nature of Proto-Semitic t is corroborated by the traditional pronunciation of Arabic and Ethiopic. at a late period. " takes its place (e. The phonetic process reflected in these changes is far from being clear: Noldeke (Mandaische Grammatik.). i. on the other hand. Ar. § 8. Dental Plosives 8. 8. §§ 8. perhaps.22. 6. ntr.) show d for t (e. of a fricative articulation of ProtoSemitic ¢ (as is also made probable by the spelling of "z" for ¢ in South Arabian: cf. "nz'" for n¢' suggests a fricative articulation of Proto-Semitic ¢' since the change ¢ > z is phonetically not improbable.e.

largely on the basis of the pronunciation of ~ over a wide sector of the Jewish tradition. at least cumulatively.g. Hence it may be thought that it is merely a secondary differentiation of 8. § 8. Interchanges between the consonants of this series occur in various languages. The consonant n is frequently dropped in Jewish Aramaic and in Mandaean when it is the final element in plural morphemes: hence nouns in the stat. Consonants of this group are sometimes dropped or reduced to '.68. N abo ~nm "statue" for ~lm. Akk. n. An independent awith lateral articulation is attested in modern South Arabian. It has more than once been maintained that this consonant was originally an affricate (of the type [~]). and cannot. 7. bl "son" for bn. This may. In a modern Ethiopian language. A notable example of this is the surrender of nunation (or mimation) in the course of the historical development of Arabic (and Akkadian): cf.33) seem to point in this direction. but without a graphic sign of its own (the symbol for 8 is used. sansal "chain" in contrast to Ar.71. the inferences drawn from Egyptian transcriptions and from a gloss in the Tell Amarna letters (cf. however. On the other hand. i. 8. unlike some other emphatics.25. n. the comparative table below). be regarded as an independent consonant.32 Phonology The Phonological System 33 of n (the palatal nasal IJ) seems to exist in Akkadian. it may be inferred that the three symbols correspond to three separate consonant phonemes. North-West Semitic in its most ancient phase shows traces of an autonomous 8: at all events. therefore. A similar situation exists. while a variant of l (the emphatic lateral probably non-phonemic (for the contrary view see Ferguson). BSLP 39 [1938].P. Analogous cases of the shedding of final n are found in modern Aramaic dialects: thus in the Ma'liila dialect *lyablin "ropes" >7y. although Semitic languages generally shun homorganic radicals in neighbouring position. voiceless s and voiced Z. raqraqqu and laqlaqqu "stork") are also fairly common. for Arabic (where the omission of nunation is always connected with that of the case-endings) § 12. though in a somewhat less systematic form.27. [R]) in certain spheres of the traditional pronunciation of Hebrew and shares several of the characteristics peculiar to pharyngals and laryngals-thus pointing to a uvular articulation. is found in Arabic. Old South Arabian has three symbols for the non-emphatic voiceless dental fricatives for which the following correspondences with Proto-Semitic have been claimed: (Sl) = 8. and r have become positional variants and are thus members of the same phoneme (H. ~ (S2) = a. lam~att~ and nam~atu "fly" (an occasional change n >l may be observed in Old Assyrian: kulka "seal!" for kunka). 3£ (S3) =S. to claim a as an independent consonant phoneme in Proto-Semitic. and for Akkadian § 12. 137-75). of the plural often have the ending -y' instead of -yn.A. The considerations adduced in the foregoing would seem to be sufficient. 8. There is another consonant of this series whose attribution to Proto-Semitic is debatable and whose character moreover has hitherto defied precise definition: it is usually transliterated a-but at times also in other ways (in particular §). Phoen. Certain reservations have however been expressed (Cantineau) for nand r: n is frequently found contiguous to other dentals. l. The dental basis of articulation of the consonants just listed is suggested by their traditional as well as their modern pronunciation.29. Those involving nand l are especially frequent: Akk.e. in Syriac where it extends to l as well. an extension of the use of -y' at the expense of -yn. 8. and a diacritic mark was introduced at a late date as part of the Masoretic pointing). but this pronunciation is probably secondary. is always voiceless. absol. Eth. J. Polotsky. be a morphological phenomenon. which. the 8 symbol often used in the transliteration of Old Akkadian does not appear to connote a phonemic distinction from 8. This consonant appears in Hebrew and in Biblical Aramaic.abli.g. r is pronounced as a uvular (I. Dental and Palato-alveolar Fricatives 8. Comparative Grammar 3 .28. It also possesses an emphatic dental fricative ~. 8. Gurage. yet an examination of the correspondences in the other languages suggests its original autonomy (cf. 'm' for 'mr "to say" (in some forms of the imperfect and of the imperative). pp. n Moscati. At any rate. silsilat. Proto-Semitic has two non-emphatic dental fricatives.26. with the result that the construct and emphatic states become formally identical. Instances of the fall of final r occur in Jewish Aramaic: e. Interchanges between nand r (a typical case is the Aramaic br "son" for bn) and between land r (e.

An examination of the correspondences shows that Proto-Semitic s remains.e. ESA 's2r. ESA ks 3 w . . and emphatic (s. Ug. Proto-Semitic has a voiceless palato-alveolar s (i. siqqurratu for ziqqurratu "temple tower"). in its turn. and Proto-Semitic B (e. The causes of this phenomenon are still not clear. rI]). while Band s coalesce in S. snm = Heb. 8.C. skn "to put" for skn.Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic Semitic 8 8 In Middle and New Assyrian originals often appears as s. As for the glosses. z. 137-49) on the basis of certain correspondences adduced from Old Akkadian and Old South BabyIonian. Interchanges of s and ~ may occur in the neighbourhood of n (e. 'sr. written with the same symbols (the special symbols for ?i.ammes. while s has either symbols of its own or shares them with 8' (e. and the same is true at the beginning of syllables in Old Akkadian and Old Assyrian.sar. Yet another sibilant (s".34.g. In late Punic.g. of course. s/sar. it is. Where these occur at the end of syllables the same symbols are almost always used whatever the consonant. cf. too. This hypothesis is mainly based on the lateralized B of modern South Arabian. 8.30. ?ir are an exception). but this hypothesis has not been generally accepted. ?). 'e8er. Syr.am~s. sb'm for sb'm "seventy". Ar. Ar.g.56 (of Jerusalem) shows sa-te-e (corresponding to Hebrew Bade): hence B is rendered by S. kussaya. it has been suggested (Cantineau) that it was lateralized and that its distinction from s lay in that peculiarity.32. Ug. 8.g. B.g. ? have regular correspondences. except Hebrew. '". . show the merging of B with s (the spelling 'sr for 'sr "ten" in Phoenician is an isolated case). inadequate for indicating the difference between voiceless. Ugaritic has three symbols for the non-emphatic voiceless fricatives of this group. ly. The change s > s occurs also in Amorite: e. In the late Phoenician inscriptions from Cyprus the use of the symbol "8" for s is a noteworthy feature (e. etc. pp. of course. while Band s have those indicated in the following table (the identification in this table of Proto-Semitic with Hebrew is. Heb. ESA l}msl. as usual. usbat "she sits" in contrast to Bab. psn and p~n "to veil"). ptlmys for "Ptolemaios"). ly. Arabic s. In the first millennium all the Canaanite languages. and to a large extent in Old Babylonian. syllables beginning with z or ? are still. Ar.A. In the North-West Semitic of the second millennium B. Hebrew transcriptions of Assyrian names confirm this fact: for Assyrian Barruken we have Hebrew Sargon. ?U. Eth.) has been proposed for ProtoSemitic by Goetze (RA 52 [1958J.31. interchange between dental and palato-alveolar fricatives is frequent (e. Heb. There are however indications of an autonomous B in the Tell Amarna glosses and in the Egyptian transcriptions of North-West Semitic names. Bb'. Bb' "to be sated". 8.B: Akk. In the Egyptian transcriptions B is rendered by s.s: Akk. especially before bilabials (e. In Assyrian (and less often in Babylonian) we have some instances of initial s for z (e. Heb. Syr.33. kst. 'asru. l}ariws. l}ame8 "five".ir. z. whereas it would have been rendered by s if it had merged with s as in U garitic (but this is the only example. 8. B?. voiced. kusitu "garment". sapal "under" as well as sapal). whereas s remains unaltered: e. eser "ten". we notice that s. where Bis held to correspond to Hebrew B.g.g. l}ams. s'w'r' for "Severus". kuswat. Ar. From the Middle Assyrian and Middle Babylonian period onwards. Ug. As for Hebrew. well known that the Masoretes indi{lated a graphic distinction between Band s by placing a point either above the 3* B s S s S B S s S S S S S S S S S3 S2 S1 S S S Examples: s: Akk. s/sab).g. New Babylonian iktaldu "they arrived" instead of iktaSdu)..P. Finally. develops into s (as has been pointed out in the preceding §). 'asr. Looking at the manifestation of the consonants of this series in the various languages. k'asut.). 8 merge in s which. Eth. letter 286. Heb. purely conjectural): Proto. Boko). In Akkadian the system of writing is.g. B{in~m (but transcriptions vary at times even for the same name: sk and sik = Heb. l}ms. for the most part. From the Middle Babylonian and Middle Assyrian period onwards the substitution of 1 for s before a dental has become a characteristic feature (e.34 Phonology The Phonological System 35 As for the character of this consonant. sb'). In Amorite. I. wasbat. and the possibility of a purely graphic variant cannot be excluded). s'r = Heb. Syr.g.

however. for in borrowings from Aramaic they are in part accomplished and in part not (e. It is interesting to note. § 8. shows that he knows of s. 8. Preclassical North Arabic has only two symbols which correspond to sand 8.41. since by and large sand 8 appear to have coalesced in one single consonant (just as they possessed one graphic symbol only).Semitic s. at a time when the phonetic process characteristic of Aram(:Lic had already taken place. therefore. the same symbol having always served for both consonants.37. g/kil. In any case. but not in all of them: e. The distinction may be based on ancient tradition.C. 8ana). The characterization of this last consonant as voiceless is not completely certain. Nevertheless. generally regarded as the emphatic consonant corresponding to k and therefore also transliterated 7. Ill). It has already been mentioned that Old South Arabian has three symbols. the phenomenon which formed the basis of the Masoretic distinction must have been of fairly limited extent. 8. and may be said to have been completed. In Akkadian q is frequently written with the symbol for g (see. that modern Ethiopian languages have developed a new consonant 8 for which they do not use the ancient character for 8 but an adaptation of the symbol for s (Ullendorff. been conjectured that the Masoretes may have generalized a purely dialectal differentiation. as usual. in a well-known passage (Onomastica sacra. probably points to a dialectal differentiation rather than to the existence of an independent phoneme s. the symbol for s has gradually extended its scope to cases where etymology would require the symbol for 8 (though there also exist many instances of 8 usurping the place of s). The traditional Arabic articulation is indeed voiceless.g. certain interchanges between the consonants of this .36 Phonology The Phonological System 37 left side of the letter (for s) or its right (for 8).36. and Latin transcriptions of Hebrew names do not distinguish between sand 8. Lil). It also possesses an emphatic velar plosive q. In Akkadian. and emphatic.. voiceless k and voiced g.40). 8. and their probable correspondences with Proto-Semitic consonants have been listed (of. In Babylonian and later Assyrian initial k and g are consistently kept distinct in the majority of the symbols used. §8.. according to which the Ephraimites pronounced 8 as s.g.39. p. The position in the ancient inscriptions may be based either on a graphic "Canaanism".18). The development to s takes place gradually during the second half of the first millennium B.29).g.. Jerome. about the beginning of the Christian era. 8 >s) seem already to have been accomplished (e. save for rare exceptions. as for q. g'yt' "summer". g/kir. and from the phonemic point of view any uncertainty may be accounted for by the absence of a distinctive opposition. or on an approximate rendering of the Proto-Semitic consonant which still survived (cf. in Mandaean there are many cases of g for q (e. This differentiation is entirely lacking for consonants in final position. ~ and 8. but some indigenous grammarians and a few modern dialects support a voiced pronunciation. the changes characteristic of classical Arabic (s > 8. the distinctive articulation of 8 has been lost since the earliest time and has merged with that of s. In the North-West Semitic languages of the second millennium B.40. the writing system is. 8. as well as for other positions in Old Akkadian and Old Assyrian. sanat. while for other syllables containing q separate symbols do not appear until a later period.yanite sn "year" compared with Ar.33). inadequate to indicate the distinction between voiceless. Proto-Semitic has two velar plosives. but we have no reliable evidence for this: the indications furnished by the Tell Amarna letter from Jerusalem are insufficient (§ 8. It has been observed. § 8. and the famous passage in Judges 12. However. the Egyptian papyri likewise have "s"-except for a few doubtful cases. SLE. Indeed. the voiceless correspondences in the other Semitic languages confirm the voiceless character of q. and spelling conventions have become quite arbitrary.C. that these changes may not be very ancient. Heb. however. In Ethiopic we encounter the same development as in classical Arabic. 8. the Akkadian. Greek.6. 8. The most ancient Aramaic inscriptions show the symbol "s" corresponding to Proto. p. consequently.g. Syr. but not of s. qayta).38. a special symbol for qa occurs in Old Babylonian at Mari and Esnunna. 36). voiced. sakkin "knife" >sikkin). It has.35. Velar Plosives 8. 8. and St.

Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic Semitic 7. but this phenomenon is due to Assyrian (cf.38 Phonology The Phonological System 39 series are to be found in Amorite. and Ugaritic gpossesses a clearly circumscribed independence which is not invalidated by a number of peripheral developments. Heb. SLE. 'o7.e. pp. § 8. Heb.. 8. Against this set of observations there still remains the fact that in classical Arabic.W. 7. 8. Ethiopic has evolved. Proto-Semitic has two velar fricatives. but none of them is certain for the classical period (Ullendorff. in the Tell Amarna glosses. A similar process of assimilation (under the influence of front vowels) underlies the transition k > 8 which occurs in modern South Arabian dialects as well as in the modern languages of southern Ethiopia and in the Aramaic dialect of Ma'liila. under the impact of its Cushitic substratum. g~tliim. . 'olayma. and even there it was in some cases secondary. Eth. It appears. I. in the Egyptian transcriptions of Semitic names. g 7. This labialization embraces.A. 'alyii. it has recently been pointed out (Rossler.-g: Akk. it is true. From this he inferred that g did not exist in Ugaritic but that the symbol in question was simply one of a number of attempts at fashioning a suitable graphic sign for '. in addition to the three velar plosives. 158-72) that ProtoSemitic g-as distinct from Proto-Semitic '-does not always occasion the Old Akkadian change a > e: a fact which would point to its independent existence in the most ancient phase of East Semitic. Syr.w. and voiced g. qW. 7.: Akk. ZA 54 [1961]. Ug. the velar fricative o. The series seems to attain stability later. Eth. 23 [1955].46. kWol. 9. ESA '7. Ar. and in Ugaritic.475-78) who has endeavoured to show. kalU "all". [x] and [yJ. Ug. in the first millennium: in the Canaanite area it is not until Neo-Punic times that interchanges between the voiceless and the emphatic members are attested. 'rb. Moreover.45. 67. 240 to 262. Syr.e. 'elem. being partly a variant of ' and partly an independent phoneme.-thus producing: kW. Ruzicka dismissed the South Arabian evidence as a mere extension of the Arabic phenomenon and claimed that the symbol taken for g in Ugaritic corresponded in some instances to '. voiceless 7. South Arabian. derived from the pharyngal' (e. however. attested by Arab grammarians (although regarded as faulty) and also occurs in sOme modern dialects of Egypt and Arabia.g. in the Aramaic area interchanges between the voiceless and the voiced members are found in transcriptions of Assyrian names. 'rb "to enter". Eth.5). pp. 'alJ" Syr.P.43. g lJ lJ 7. a series of labio-velars which exist alongside the ordinary velars. Ruzicka's contentions have been partially supported by Petracek (ArOr 21 [1953]. 8. 'a7. film "boy". i. At times the labialized consonants take the place of the simple ones in such correspondences as: Akk. Ar. a7.. Syr. Examples: 7.u "brother". i.44. Originally. In Classical Arabic g develops into g (affricate and palatoalveolarized). Heb. Ar. of palatal vowels and thus as an aspect of assimilation: cf. In the Ethiopian sphere many cases of spirantization and palatalization of velar plosives can be observed. The correspondences of the velar fricatives in the principal Semitic languages are as follows: Proto. When an independent g-or at least an independent graphic symbol-was identified in South Arabian and in Ugaritic.. pp.42. kull. pp. This condition can be explained in terms of the acquisition of independent phonemic status of what was originally a mere variant. 8. g 7. kol. in a statistical investigation. that g is to be retained among the Proto-Semitic consonants. In addition. gW. ESA glm. Heb. musawwag "permitted". kol. a variant of musawwa'). g 7. In a series of studies Ruzicka has maintained that g is not a Proto-Semitic consonant but an Arabic innovation. Velar Fri ca ti ves the argument in support of this thesis lay in the fact that g was to be found in Arabic only. § 9. 49-74). An analogous tendency lc > c (though again considered "faulty") is noted by Arab grammarians and appears in ancient and modern dialects (in the neighbourhood. ESA grb. Ar. therefore. kl. that in Arabic g is of a complex phonemic nature.40) rather than to Aramaic factors. The pronunciation as g is. Ug. 8.

. the hypothesis (e. voiceless 7. 10. g. >') or to g (e. from the root 'fr). Akk. fgr).al.50. in New Assyrian it is probable that h reappears. 8. flJr "to be small"... it should be clear.47. in A{JLXm)". the use of the symbol E for the phonetic values 'it and it'. magarat. y (g) stand for the original consonants lJ.A. na7. e. whereas the normal Akkadian development [see § 8. Egyptian mgrt = Reb. is the rendering of 7. 24 for some rare exceptions). The only problem in this connexion is posed by the Greek and Latin transcriptions of Rebrew which show for 7. Egyptian nlJr = Reb. extensive phonetic reductions and losses which it is well to examine individually (for phenomena of syncope. The variations in the transcriptions seem instead to relate to different periods: X (ch) and y (g) predominate in the earlier period. GAG. with the exception of Akkadian where they are reduced to' (or to zero).1" is employed for 7.20).52.. that the transcriptions X (ch). 8. En] and [<1]. This hypothesis can. Another feature characteristic of late North-West Semitic. as lJ. p.P. the existence of lJ and g is attested both in U garitic and in the Egyptian transcriptions of Semitic names." for ' (e.P. Proto-Semitic has two laryngals: one glottal plosive." '.. The correspondences exclude the possibility of a survival of an original !J. AI'. In these transcriptions "lJ" is used for lJ.. Reb. respectively). von Soden.. The consonants of the pharyngal fricative and laryngal series have regular correspondences in the various Semitic languages.. (I. § 9. there are some voiced manifestations in modern Arabic dialects)..g. In North-West Semitic of the first millennium the process lJ > 7./ara.g.C. however.. this is reflected in'graphic interchanges of increasing frequency and arbitrariness. 'Azza. while e and zero belong to a later one. and ". X corresponds to 7.40 Phonology The Phonological System 41 8. is usually absent at the beginning of words (cf. e-!Ji-il-tum for e-'}. while "1. and "g" (or "q") for g. while the others stand for original 7. In Akkadian (as has just been mentioned) these consonants have been reduced to '-under the influence of Sumerian which did not possess the consonants of this series.49. In Akkadian interchanges occur between lJ and k (e. The reduction is not yet complete in Old Akkadian (cf.g. (e. zero. SNO.: X (ch). m.-il-tum). in Old Babylonian.M. AI'. be tested by a comparative examination. It has been suggested . respectively. y corresponds to ' in rorpeea.1" in a syllabary from Ugarit may suggest voiceless articulations of Ugaritic g (Garbini. for': y (g). however.g.. Gazzat). because anniu "this" is often spelt !Janniu (pronounced [hanniu]?). zero. In North-West Semitic of the second millennium B.km.54] is 7. 'abi7. e. In Ethiopic the pronunciation of !J gradually coalesces with that of 7. there are indications that SOlIle laryngals at least were still pronounced ('adanum "limit" written with initial !J). There are.51. ' may be graphically expressed in various ways: by the symbol for the vowel which follows (e. is-a-am for is'am) or by the symbols for !J (e. It is only from the Middle Babylonian and Middle Assyrian periods onwards that ' has symbols of its own which even then are not regularly employed. cf. 8.A. Some Ugaritic uses of the symbol "g" for lJ as well as the correspondence between U garitic "g" and Akkadian "1. Laryngals 8.g. 8. Apart from the use of specific symbols. .. 7. too. of course. 8.48.g.53. probably corresponding to the ProtoSemitic consonants hand M. and the result of such a test militates against. that the absence of a symbol does not necessarily coincide with phonetic reality. Proto-Semitic has two fricative pharyngals. 8. AI'. nalJlu." g >' is complete. 52).ayil. Pharyngal Fricatives.g. [?]) and one voiceless laryngal fricative h (of which. and voiced' (I. cf. p. lJkm "to understand". cf. which appears in Eastern Syriac and in a sector of the Jewish-Ashkenazi tradition (though European languages are bound to have affected Ashkenazi pronunciation). however. It should be observed that the graphic notation of " partial and irregular in medial position. Egyptian qrJt = Reb. Akkadian lJ corresponds in some cases to Semitic 7. There is no reason to suppose that the situation in Akkadian ran counter to the general Semitic rule which requires that every . AI'. The Assyriological custom of not transliterating even initial' is followed in the present work. lJns instead of the usual kns "to submit").

g. ly. Old South Arabian displays the transition' >' in the dialect of the l.59.ad.g. This transition does not. § 10.:: z ~ ~ z ~ t t ~ ¢ t d n t t d t d n t d n t d n ¢ t d n ¢ t d ¢ t d n t n t n . 'd for 'ly. shows many cases in which ' loses its consonantal value and is then dropped in the current spelling convention (e. h >' > zero.g. hJwii). 11. In Canaanite. etc. 8. § 8. 'd "up to" for 'd). is the gradual weakening and eventual reduction to ' (or zero) of ly" '.56. >' in 'r~t' for 'r~t' as well as many cases in which intervocalic' is dropped: mry for mr'y. 'aly. ly.IaQ. A characteristic feature of Punic.55.h. p b m p b t 0 8 z ~ ~ t d m 8 p b m f b m f b m f b m 8 t d (0?) t 0 :. however. h often loses its consonantal character (e. 'ely.58. as distinct from Phoenician. . to h and of ' to '.:: t 0 :.). prior to the division into West and East Aramaic. extensive areas of phonetic uncertainty occur which are reflected in the orthography. To return to the table of the Proto-Semitic consonant system (cf. The latter affects the orthography of classical Ethiopic. in fact. *'aprum "dust" > eprum).g.42 Phonology The Phonological System 43 syllable should begin with a consonant (cf.). in those of the Eastern group the reductions '>'. a weakening of the pharyngals is suggested for pre-Masoretic Hebrew by Greek and Latin transcriptions (of. in particular. she" lose h in enclitic position): for details cf.g. § 8. so that with the passage of time inconsistencies become ever more prevalent. while on the other hand it sometimes occurs with h (e. the evolution of this system in the principal languages of the group may be envisaged as follows: Proto. therefore. Classical Arabic exhibits a remarkable stability of the pharyngals and laryngals. cf. not altogether impossible that the Masoretes may have aimed at restoring the ancient pronunciation by means of their peculiar system of vocalizing the pharyngals.2). >h are attested in some ancient dialects. Aramaic. hi "he.3). but it is prevalent in popular inscriptions in which constant interchanges and losses occur in the pharyngal and laryngal series (e.d "one"). Brockelmann.ad "one".Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic Semitic p b m p b m 8. As for " it is possessed of an exceptional constancy in the orthography of the classical language.g.ramawt (e. the pronouns hil. pp. retains by and large the independent articulation of the pharyngals and laryngals (some weakening which may be observed in the Aramaic of Assyria is probably due to Assyrian influence: e. ewilm "to become" compared with Aram.45). An identification of the consonants which had coalesced in ' is at times possible on the basis of modifications to which neighbouring vowels have been subjected: for' derived from g. Heb. of course. 8. this uncertainty (resulting eventually in almost complete arbitrariness) does not appear in the most ancient inscriptions of Aksum and may well be due to the influence of Amharic. ly.57. But there are. etc. SG. § 8. It is. though a few traces of the development '>'.ad. ly" ' occasions the change a into e (e. 'Ar. 25-26. Syriac. 8. > h are very frequent and may. no grounds for denying the original phonemic independence of the consonants of this series. always take place in the case of g (cf. In the languages of the Western group the consonants in question are frequently interchanged or dropped altogether.49) and by interchanges with' and h which are attested in the Dead Sea documents.g. Later. In Ethiopic we observe a gradual phonetic reduction of ly. extend further to '> zero. In the Arabian area. This phenomenon becomes manifest to only a limited degree in official documents where the traditional orthography prevails. Synopsis of the Consonantal System 8.

wi. and in Assyria those for b (e. In initial position y almost invariably disappears (or is reduced to '). ayya. i.-Ass.Phonology The Phonological System 45 Proto. Old Bab. aya.Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic Semitic P b m P b m P b m P b m P b m f b m f b f b m t t (rJ) t rJ ? ? rJ 1 1 t d rt t d rJ t d rJ t d rJ t d t d t d t d t n r 8 Z t n r 8 t n r 8 Z t n t n t n t n r 8 Z t n r 8 Z r 8 Z r 8 r 8 Z ~ Z ~ Z ~ ~ ~ ~ ~ ~ 8 8 8 . subject to changes and reductions (for some phenomena of syncope cf. its graphic notation in Akkadian is somewhat uncertain and imperfect. Since w is rare. The extent to which the original consonantal system has actually survived in the various languages (irrespective of etymological relationships) is shown in the following table: Proto. I. Mid. awatum "word".20). yu. Ass.63. ayyi. at times leaving behind the vowel which h h 8. wU88urum "to send" > u88uru and mussuru. They are. the symbols A-A are employed for the sequences ay.Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic Semitic 8 8 Z ~ 8 Z ~ 8 8 Z 8 Z 8 Z Z ~ ~ ~ ~ ~ g q g q g q g q g q g q g q 8 8 k g q 8 k g 8 8 k g 8 8 k g q 8 8 8 8 8 If lJ g If If lJ h g h h h q q k g q g q k k g If g lJ h If q h h If g lJ h If If lJ h g lJ lJ g lJ h h If If 12. 8. ayye.g. ye. Up to the Old Babylonian and Old Assyrian periods the syllables wa. Sumerian possessed only the phonetic sequence i-a. abatu).60. afterwards it is either dropped (or reduced to ') or written with symbols for m (e. § 9. and probably secondary. § 9. we.62. Proto-Semitic has a bilabial semivowel wand a palatal semivowel y.Akkadian Ugaritic Hebrew Syriac Arabic ESA Ethiopic Semitic r r r r r r r 8 Z Proto. and the symbols for this are used not only for the graphic expression of the Akkadian syllable ya. Both semivowels have regular correspondences in the various Semitic languages. ayyu (other sequences are not encountered owing to elisions and contractions). For w in medial position cf.A. As regards the other semivowel. y.61.g.20. however. w is generally preserved until the Old Babylonian and Old Assyrian periods. wu are written with the Sumerian symbol PI. Mid. waradum "to descend" >aradu).P. amatu. 8. Semi vowels h 8. later on. for w the symbols for mare predominantly used in Babylonia. but also for yi.e. [j]. At the beginning of words. in the Sumerian writing-system (and language).-Bab.

WA. Akk. Proto-Semitic also possesses the three corresponding long vowels: a. which appears to be derived from a or i (a. rnu-ru-i~ for mu-ru-~t~ "pain". [a]. 8. TPA. ay>e). foreign transcriptions.67. Reb.g. 'iqa' "protection" for wiqa'). a. 0 for u. i. Old Akkadian spellings of the type i-ik-mi. but with the addition of the vowel e. 110-11).g. p. For y in medial position cf. Initial w is kept in the conjunction w "and" and in a few nouns (e. OA. For other languages. In North-West Semitic there is a characteristic development w > y in initial position (e. i-ig-mu-ur (Gelb. WA. i).g. Reb.66. i. meets with difficulties in demonstrating the phonemic status of .71. 199). while at other times the homorganic vowel i (e. In ancient Arabian dialects an occasional change w > y in initial position is suggested by such cases as yazi'ahum "their protector" for wazi'ahum (Rabin.g. 13.e. either short or long (e. W A. The non-phonemic variations e for a. Another phenomenon attested in these dialects is the pronunciation of y as {j.P. even though there exists a good deal of circumstantial evidence (general Semitic comparisons. but further study is required.g. contraction of diphthongs (aw > 0. y to ' are also found (e. p. Ug. Syr. 8. most recently Rabin. From the phonetic point of view it may be taken for granted that not only this vowel but numerous other varieties have existed in Semitic since its most ancient phase. I. Akkadian presents a vowel-system identical with that of Proto-Semitic. ya>i) remains. may conceivably be the result of dissimilation of the semivowel in relation to the homorganic vowel i which precedes it. and especially those of the North-West Semitic group. Proto-Semitic has three short vowels: open back velar a. There are no certain grounds for supposing that ProtoSemitic had once possessed additional vowel phonemes. Traces of vocalic land r have also been claimed (von Soden. of course. The graphic interchanges between u and i and between u and a in certain forms (e. 158) seem to suggest the possibility of a prefix yi. These vowels have arisen in two main ways: by change under the influence of neighbouring consonants and by . this vowel. the principal data available will. 'iyyal "deer" > 'i{j{jal (Rabin. The examination which follows will be especially concerned with the vowel-systems of those languages for which we possess the best sources of information. 63).A. The history of Arabic and its dialects shows clearly in what manner vowels of other timbres have evolved in the Semitic languages and have. Eth.>i-. the consonantal system of writing does not offer sufficiently solid grounds for adequate reconstruction. The Proto-Semitic vowel system has an exact reflection in that of Arabic whose full network of graphic symbols mirrors the phonemic position. pp. Akkadian 8. a. § 9. Finally. i-na-~ur 8. p.70. and these systems vary between certain extremes of phonemic and phonetic representation. in the Egyptian transcriptions of the second millennium. e for i are so common that Arabic vowels are rightly classified according to the place of articulation rather than on the grounds of timbre (Fleisch. Some exceptions in the Egyptian transcriptions might suggest that this process was then still in the evolutionary stage. Some survivals of w in Nabataean (in cases where the other North-Western languages have y) may be explained as due to Arabic influence. Ar. 65. walad "child"). and close back velar u with strongly rounded lips.20.as an intermediate phase in the change ya. in the course of time. but the stock example usually cited. acquired phonemic status. in the southern dialect of Old Babylonian i occurs so frequently for e that this feature has been regarded as reflecting a dialectal peculiarity. In particular.65.g. 'uguhuhum "their faces" for wu{juhuhum. close front palatal i. in Ugaritic and in the Tell Amarna glosses. Reductions of initial w. 8. e). which has more than once been postulated (cf. matres lectionis).M. The graphic notation of vowels in the various Semitic languages is bound up with the system of writing adopted by each one. In the writing-system the series of symbols with e is very incomplete. p.68. wld "to bear". the addition to the vowel-system of e. be recorded.69. 11). u.46 Phonology The Phonological System 47 accompanied it (e. 8. pp. Vowels 8. yld). GAG. S. yu >tt). This phenomenon can already be seen in Amorite. i. 83).

anaku: the writing of u for 0 is due to the absence in cuneiform of a proper notation for the vowel 0). 63-64). It is possible (cf. it is probable that there existed vowels . not as independent phonemes but as allophones of a. 8. u and a (sumum and samum "name") where one might suppose a vowel of the [0] type.orth-West Semitic. The glosses show the.76. now appears to be established as part of the phonemic system. The problem of representing unvocalized ' is open to argument.74. In Ugaritic the writing-system is consonantal.g.78-79) in -i of some monosyllabic substantives (in contrast to the usual absence of endings in that form) has suggested the existence in Akkadian (known also in other Semitic languages) of a vowel of the 'J type (von Soden. pp. An instance of the change a> 6 is now attested in an Old Babylonian inscription from Mari: ClJamafam » !Jam1l§am ilfmuf "he plundered thoroughly". u/u. The tentative results show considerable divergences from the Masoretic system. u. § 8. The first and the second of these Moscati.iin).48 Phonology The Phonological System 49 for i-na-far "he watches") have been claimed as evidence for the existence of vowel qualities of the type [y].observed that in Ugaritic the symbol for' with u/u corresponds also to '(*aw » 0 and that for ' with i/i to '(*ay » e (Gordon. The original long vowels appear to be more stable than the short ones. Within the system itself three different traditions can be distinguished: the Babylonian. Amorite. frequently resulting from an original a or i. i. change a> 0 (e.83. and Latin transcriptions of Phoenician words) presents the usual Semitic phonemic vowels (§ 8. Comparative Gramnlar 4 . for *'Azir-ba'al). against Akk. In the majority of such cases the symbol for' with the vowel 1: is used. the Palestinian. Akk. cf. but this Akkadian development is not attested in N. pp. pp. 8. ay can also evolve into ii. e. It might be averred in this connexion that the diphthongs aw. 8.o us in the absence of a rational explanation). Greek. In the Canaanite area the Phoenician vowel-system (which can be partially reconstructed by means of Akkadian.. which has come down to us in cuneiform script. it therefore postdates the consonantal text by a very considerable margin. JCS 2 [1948].of the e. S. 8. for the data are far from being consistent.C. Similar considerations might apply to other interchanges. Harris in JAOS 61 [1941] pp. §8. von Soden.78. UM. 146-47). 8. and on other occasions it is the symbol for ' with any vowel indiscriminately (or at least so it appears t. a/a.73. The reconstruction of the original vocalization has been attempted by utilizing the transcriptions of Hebrew names in other languages (Sperber's studies) as well as by the application of modern linguistic techniques (Z. 8. 17). Heb.72. The construct state (of.74) that the symbol for' with the vowel i may in some cases represent a 8'Jwa (cf.g. i (of. 82). i as e (O'sQfJaAor. RANL 13 [1958]. pp.66) in a number of varying pronunciations. 10. [0] (cf. The language of the Tell Amarna glosses (likewise in cuneiform writing) also displays a vowel-system like that of Akkadian. It has been . Canaanite 8. Some vowel interchanges similar to those in Amorite (cf. i. for which cf. [0] type. Thus we find a pronounced as e ('sQa for *zar' "seed"). 291-303).75.D. b.e. of. 'anoki. §§ 12. 143-67). a-nu-ki. SNO. *baytu > bitu "house" . GAG. From this we may probably infer that the Ugaritic vowel-system corresponds substantially to that of Proto-Semitic. § 8. 0 timbre. u as 0 (Baliahon for *Ba'al-yaly.73) have called forth the idea of vowel qualities of the [y].g. The vowel notation of Biblical Hebrew is the work of the Masoretes and originated during the second half of the first millennium A. macom for *maqam "place"). Moreover. but in Phoenician we may observe the non-conditioned change a > 0 (e. *mawtu > miitu "death").83. § 8. p. but the consonant ' has thr~e symbols according to the vowel which follows. binum and bunum "son") may possibly suggest the existence of vowel qualities of the [y] type. Some interchanges of i and u (e. North-West Semitic of the Second Millennium B. apparently non-conditioned. and the Tiberian. exhibits a vowel-system identical with that of Akkadian-with the sole exception that e does not appear to be an independent phoneme but rather an allophone of i (Gelb. even though it started as a mere allophone. i/i. The vowel e. but sometimes the symbol used is that of ' plus the vowel identical with that which precedes it.n.g. c. Garbini.

a/a. characteristic of Western Syriac (cf. §§ 6.88) and in pre-Islamic Arabic (Rabin. While Biblical Aramaic uses the same system as Hebrew.83. This is. The use of matres lectionis in the Dead Sea documents is somewhat peculiar: ' is extensively used at the end of words. On etymological grounds. only of a can it be said that it is normally a short vowel. Combinations of vowel symbols with matres lectionis serve to indicate a series of predominantly long vowels: w is used as mater lectionis for vowels of the timbre u or 0. According to the Tiberian system. Only one of the vowel changes in Hebrew appears to be non-conditioned. 0. The vocalization of this text has-as compared with the general development of North-West Semitic-a rather archaic appearance: for example.84. the Biblical vowel system may be represented as follows: i e a o u 8. § 8. § 8.84.g. has long been considered a characteristic of "Canaanite" (in the traditional sense of that term). y for those of timbre i or e. 8.12-15). later on. samane "eight".(s'ifwa) which originally marked the absence of a vowel but which has come to indicate in certain positions (at the beginning of syllables) the vowel of the type 'if (M urmelvokal). Dupont-Sommer. h at the end of a word for those of timbre e. this process is. "::-'i/i. There are two different methods of vowel notation. This change.50 Phonology The Phonological System 51 indicate the vowels by means of supralinear signs. a or 0.e. In combination with other 8.74. d. s'ifmone). Vowel quantity is not in general indicated by the symbol as such. ~ iii. -'-u/ii . §§ 8.. 8. respectively.-e. the present treatise will employ the following transliterations (taking the seven vowels in the same order as in the preceding paragraph): i/i e/e a 8. as well as for other reasons set out above (cf. 35-62). i. RA 39 [1942-44]. As for ::I. Aramaic a/o 0/0 u/ii 8.91)-for in Hebrew we have long vowels without matres lectionis and short ones with matres lectionis. but in fact. the Eastern (used by the Nestorians) and the Western (used by the Monophysites or J acobites).88). and y appears frequently instead of h (at the end of words) to indicate vowels of the e timbre. 8.-e. Heb. ~%. 105-10). pp. Compared with the Proto-Semitic system Hebrew vocalization displays a noteworthy development. WA. it has two different pronunciations: it is either long when it corresponds etymologically to a or short when it corresponds to an original u. naturally paramount in a comparative grammar. some earlier data are furnished by a magical text in cuneiform from the third century B.82. To the symbols enumerated above we have to add :. C? u/ii. (A. pp.83). -: a. while it does not occur in Ugaritic. symbols S'ifwa produces the compounds -=:' To' To' employed with the pharyngal and laryngal consonants and here transliterated ii.C. while the third uses (with one exception) sublinear symbols. which can already be observed in the Tell Amarna glosses and in Phoenician (§§ 8. again like the Hebrew system it is characterized by the predominance of qualitative over quantitative distinctions.80. lab~s. but depends on the position of the vowel within the word. -'-e/e. which has prevailed in Hebrew manuscripts and later on in printed books. 28. Some cases of the change a > 0 (not attested in Old Aramaic) occur in Palmyrene (cf. -.81. 8. however. e. the latter is based on Greek vowel symbols: Eastern: -: a. Akk. A characteristic feature of the Masoretic vowel notation is the fairly elaborate representation of qualitative distinctions.79. 8. in the middle or at the end of a word for those of any timbre. The indication of vowel quantity by means of matres lectionis is very imperfect-in contrast to the position in Arabic (§ 8. 8.86. to the second half of the first millennium of the Christian era. it reappears in the early centuries of the Christian era both in West Aramaic (cf. the original i remains in la-bi-is as against Heb. .85.77). be dealt with at the appropriate entries below. therefore. Syriac has a vowel notation going back. closely linked to syllabic structure and stress patterns and will. like that of Hebrew. -. 00/0 4* Western: ~. a> 0 (e.

8. respectively. p. For short vowels (and for long ones in combination with matres lectionis) a system of symbols was introduced (in the late ninth century A.: a > a (cf. 8. paraqa "saviour". Old Ethiopic was at first written without vowel signs. From the phonetic point of view. Syriac possesses no symbol to indicate the absence of a vowel or to mark a vowel of type ~.87. This is natural in the case of wand y in view of the Aramaic (or rather Nabataean) origin of the Arabic script. somewhat less certain (cf. but in the fourth century A. customary in the more ancient texts). Compared with the Proto-Semitic system Syriac vocalization presents important developments which are' connected-as indeed is the case in Hebrew-with syllabic structure and the incidence of stress. though the existence of such a Murmelvokal (a central vowel whose precise timbre is determined by the nature of surrounding consonants as well as by the effects of vowel harmony) must be assumed in certain positions. . a vowel of type ~ does not exist in Arabic.92. 159). This notation makes use. East Syriac preserves a more ancient vocalism. another phenomenon of velarization whose true nature is. [i:] > [u:]. a non-conditioned phenomenon of palatalization whose realization is at times prevented by the operation of conservative forces. it introduced a very special type of vowel notation which operates by means of partial alterations in the form of the consonantal symbol. Arabic 8. Examples: ESyr.90. s~n) in conformity with some general relationship which. seems to exist between these two vowels in opposition to a. and for e (Western i) in final (occasionally also medial) position. respectively. b) taffpim. § 8. i. i. The qualitative values of the Ethiopic vowel series are as follows (for the purposes of exemplification it is convenient to use the consonant symbol l): A u i a e (~ ) o 8. Ar. For reasons of etymological correspondence. Eth. 'ug. 8.89. 6 >11. In addition to the difference in the actual system of notation there is a phonetic distinction between the Eastern vocalization (which is used in the present treatise) and the Western one. and in any event it is improbable that this system differed appreciably (at least from the phonemic point of view) from that of classical Arabic (for the vexed question of e of.n "ear". resa "head". As for its notation. WSyr. paramount in a comparative grammar. y for i. Ethiopic 8. a less frequently occurring phenomenon of velarization. in fact. WA. u. and it may now be regarded as established that these elements reflect essentially qualitative distinctions. matres lectionis were consistently used for the indication of long vowels: w for 11. ' for a. Hebrew: § 8.g.91.94. whereas West Syriac presents the following developments. riS6. they will be dealt with at the appropriate place below. There is a special symbol (~) to denote the absence of a vowel. 8. sinn "tooth" . ESyr. The principal tendencies noted by Arab grammarians are: a) 'imala. WSyr. [a:J > [e:J. and a to a.) which is derived from somewhat simplified forms of the matres lectionis: 8. [a: J > [0:]. c) 'ismam.95. Etymologically. e >i (in certain types of words). Pre-classical Arabic does not furnish sufficient indications for a reconstruction of its vowel system. Ar. short i u. par11qa. of matres lectionis which are widely employed in Syriac: w for u/11.52 Phonology The Phonological System 53 8. o/a. sometimes conditioned by the neighbourhood of emphatic consonants. f. in many instances in Semitic. traditional grammar and the history of the dialects provide some idea of the extensive variations in the timbre of Arabic vowels.83).93. 'for a (Western a~.e. only the use of ' to mark a may be regarded as a specifically Arabic development (and is not.e. Two elements of the Proto-Semitic system appear at first sight to be unrepresented. u i a correspond in general to ProtoSemitic 11 i a.67). e 0 to the diphthongs ay aw. y for iIi. and in medial position sometimes for e. i.96. 0. Rabin.88. e.D. i.e. for the vowels i. however. Eth. 'nn. but in Ethiopic they have coalesced in the vowel ~ (e. 8. The classical language presents a vowel system which corresponds phonemically to the Proto-Semitic one.D. 0 >u. The vocalism which is manifested by this notation consists of seven elements.e.

9-11.100.g. 8. pp. The Akkadian phenomenon may possibly be explained as the result of assiinilation (of. In Akkadian the Proto-Semitic diphthongs aw. A number of secondary diphthongizations are to be found in the Semitic languages. however § 8. The reduction is shown to be complete in the Tell Amarna glosses and in Ugaritic where we encounter the result of the changes aw > 0. caneth for *qanayti "I acquired". Before y (as in Akkadian) ay does not seem to be reduced in Ugaritic (Gordon. p. but they undergo extensive contractions in the modern dialects. SLE. EG. "Defective" writing in Aramaic shows that the reduction of the diphthongs had taken place even in the most ancient inscriptions. 170-83) are particularly noteworthy. In the Canaanite sphere. ayyabu "enemy") and in some instances of the vetitive particle ay "not". *layl > Wit "night"). 8. SLE. Altsudarabische Grammatik. ywm and ym "day") which may suggest that the process of reduction was at an active stage (of. 'en. *yawm > yom "today". Phoenician exhibits the reductions aw > 0. but st. These are probably cases of anaptyxis (cf. however. § 9. *yawm > yom "day".102. for though this vowel may correspond to the sound of 8'Jwa mobile of other languages. and those in Ethiopic (for which cf. !wjht). 8. An exception occurs in the case of ay before y (e. especially in final position: e. Some changes. and partly in New Babylonian.g. even if they occurred at an ancient period: of.76). *bayt > bayit "house". Whereas Old South Arabian shows graphic variations (e. !pay « *lpayy) "living". *mawtu "death" > mutu.17). In Assyrian.g. *'aynu "eye" > inu. some exceptions in Egyptian Aramaic are doubtless to be explained as instances of historical spelling. In the North-West Semitic of the second millennium B. 8. Dillmann. constr. however.g.100). classical Arabic preserves the original diphthongs in their entirety. the transliteration 'J is not free from ambiguity.97.8). ay whose treatment is a differentiating feature between certain Semitic areas. *mawt > miiwet "death". 78-79). In Hebrew the same reductions are generally found: e.C. 14. but even Ethiopian scholars sometimes disagree about it in the traditional pronunciation of Ge'ez. ay generally appear as U. 160). In some cases. .g. The combination of semivowels and vowels produces a series of rising or falling diphthongs.99.101. i: e. pp. secondary phenomena in the various languages. are not necessarily conditioned: they affect the diphthongs aw. The Egyptian transcriptions of Semitic names attest sometimes to the loss and sometimes to the retention of the semivowel element: perhaps they reflect a stage in the course of actual evolution. ekallu "palace" (a Sumerian word).g.54 Phonology The Phonological System 55 the present treatise will use the following transliterations (in the same order in which the Ethiopic vowels appear in the preceding paragraph) : a i extension (through anaptyxis) might have occurred as in Hebrew (§ 8. In Biblical Aramaic aw is reduced unless it is followed by w.103.98. In Syriac the diphthongs are preservedexcept when their preservation would result in a doubly-closed syllable: e. ay sometimes remains uncontracted and may give rise to syllabic extension as in Hebrew (§ 8. for ay we find a and eli. haykal from Akk. e takes the place of i (enu).104. 22-23). ay > e (cf. E-Byr. however. § 39. ay > e: e. They are. pp. Diphthongs 8.g. UM. these are subject to a number of conditioned changes which will be dealt with at the appropriate place. it is a stable vowel which may even be long (Ullendorff.g. 27): it is conceivable that syllabic 8. As regards the sixth vowel. In doubly-closed syllables we meet instances of syllabic extension through the insertion of a new vowel: e. HOfner. Ullendorff. 8. Amorite shows the preservation of aw as well as the developments am and a. § 9. p. qaw « *qaww) "cord".:xov for * Ye!paw-milk. 8. The ambiguity of the Ethiopic sixth order as either 'J or zero causes difficulty not only to Europeans. In Ethiopic the diphthongs appear in reduced form (e. the diphthongs remain unreduced. 'ayna "eye".g. but there are a number of divergent formations (cf.100).

) amdalJi~ (voicing). *u~tabbit "he imprisoned" > total: e.g. assimilation: e. the plosive articulation of the consonant passes towards the continuant pronunciation characteristic of vowel articulation.g.5.-Assimilation of vowels (or vowel harmony) is always at distance. or of consonant to vowel. from Greek :n. singular is standard in Arabic). Akk.g. also Ar. It was probably contiguous assimilation that gave rise to the "emphaticization" (in North-West Semitic) of t in the root q[l "to kill" as compared with Ar. This spirantization may continue after the elision of the vowel which occasioned it: e. Ug.-Regressive 9. or of diphthongs. partial and contiguous: e.g. pp. This phenomenon.tion total when following d.g.-Progressive.-Regressive total: e. k. 'agii'(Jzt lords" > 'agii'(Jst (devoicing). *dahabii "gold" > dahbii. Reciprocal. gen. cf. Ar. *'i~tabaga "it was dyed" > 'i~tabaga. The Semitic languages present assimilatory processes of various kinds: assimilation may take place between consonants. it may be partial or total. or between vowels.-Progressive total: e. 12-13).g. pronourfced [dah~a]. Akk.-Progressive.g. buq'at "plain". partial and contiguous: Eth. Ar.-Regressive. qaqqada. as well as the determination of their frequency. § 8. 'imru'un "man".4. Heb.g.veyo~.ob1. ~. 'allilp. *riglihu "of his foot" > riglihi (vowel harmony in the suffix-pronoun of the third-person m. both Northern and Southern: e. 1. Ar. Another phenomenon of assimilation of consonant to vowel must be seen in the palatalization of k in the neighbourhood of palatal vowels: . consisting of the transition of plosives to fricatives. but Syr.g.e. §§ 16. b) Between vowels. Heb. *attarad "I sent" > attarad (the assimilation of t in the Akkadian infixes ta and tan is always 9. the assimilation of vowelless 1 to the following consonant (which is sporadically found in various languages) takes place most prominently in the case of the Arabic article before all interdental. purq(Jsii "tower". qaqqudu "head". *amtalJi~ "I fought" > (later Bab.) "years" > sinina. c) Consonant to vowel. 9.2. the assimilation of vowelless n to the following consonant is characteristic of North Semitic. ulp "prince". a) Between consonants. acc. 'bd "to perish". and it may be contiguous or at distance. g are articulated as fricatives in postvocalic position (0£. *sanina (c. or vowel to consonant. Heb. *ittalaba "he sought" > ittalaba. 'al-sams "the sun". may be regarded as an instance of partial assimilation: i. and palato-alveolar consonants. *yintr:n "he gives" > yittr:n . In the treatment which follows some of these instances will be identified and illustrated. attention will be drawn to salient aspects and characteristic features in individual languages and groups. Oonditioned Phonetic Ohanges 9. 'imri'in. (New Ass. lJibliitu "damage" > lJibletu.116 to 117). Akk.2).-Progressive partial: e. u~~abbit). it may be progressive or regressive or reciprocal. d. Ar. s: e. partial and at distance: possibly Syr. Heb.-Regressive.56 Phonology Conditioned Phonetic Changes 57 O. Contiguous assimilation may be the cause of the vuicing of t. WSem. GAG.1.g. § 10.g.g.g. 9. Ar. dental.-In Hebrew and in Aramaic. b. the plosives p. ulJappi "he struck" > ulJeppi. after the latter's division into Eastern and Western dialects. The manifold phenomena of conditioned phonetic evolution have not yet been sufficiently investigated from the point of view of comparative Semitic linguistics: a study of these phenomena and processes in the various languages and groups. since the structure of the Semitic syllable does not admit vowels in positions of direct contact (cf. diibiir "word" is pronounced [da~arJ. while for detailed discussion the reader is referred to the grammars of the various languages concerned.) *aqtirib "I approached" > aqtirib. Akk. Ar.3. t.-Regressive partial: e. WSem. cf.-Progressive total: e. pronounced [afJams]. kbd "to be heavy". 'imra'an. qtl (though dissimilation might conceivably have occurred in the South Semitic languages).10): e. Vowel harmony is particularly extensive in Akkadian (von Soden. nom. kbt. Akk. p(Jqa'tii (devoicing). in some roots with second radical b. partial and at distance: e.g. t. Eth. it does not occur in South Semitic with the exception of some instances in South Arabian (cf. Assimila. z. biq'ii. 'bt. *'i(!takara "he remembered" > 'iddakara. Akk. A typical case of regressive total assimilation occurs in Assyrian vowel harmony whereby a is assimilated to the vowel of a case-ending which follows it: e. will undoubtedly contribute to a better comparative appraisal of the Semitic languages. which occurs in some West Semitic languages. yet 1 continues to appear in the graphic pattern: e.g. qaqqidi. Syr.

g. Reduced by assimilation: e.g. Some phenomena of spirantization in Akkadian require further investigation. d) Between Semivowel and Vowel. . lUn "to spend the night" (Heb. lJarrub "carob-bean" and lJarnub (but dissimilation is not the only way in which this variant may be accounted for. /aaa~ for Yi§lyiiq (and cf. 9. b) Between Semivowels. and particularly in Babylonian. Labial consonants are liable to cause other vowels to change into u. *yaftul}u "he opens" > yaftal}u.g.97-104. § 10. a) Between Oonsonants.2). and despite phonetic difficulties also z: e. Ar.Assimilated: e. *yakSud "he conquered" > *yiksud > ikSud (total progressive assimilation).-E.14. and in any event this type of dissimilation is infrequent). both progressive and regressive. though its strictly phonemic vowel notation fails to indicate it). Eth. later on Syr.7.g. 9. *kawy "burning" > kayy.g. wuguh "faces" > 'uguh (regressive and contiguous). *wawaqi "ounces" > 'awaqi (regressive and at distance). inaddin "he gives" and inandin (or inamdin).13.g. This is the so-called patal} furtivum which does not. Heb. c) Between Vowels. and between semivowel and vowel. inazziq "he grieves" > inanziq. Dissimilation 9.2).g. Ar.10.-Progressive and contiguous: e. As will be explained when dealing with syllabic structure (cf. *sams (n > sams. between vowels. lubb. for details § 16. for this section the treatment of the diphthongs in §§ 8. § 10. To obviate such initial consonant clusters a supplementary vowel (introduced by') is generally prefixed to the first consonant to produce a new syllable. *madiniy "Medinese" > madaniy > madaniy (qualitative and quantitative). Reduced by assimilation: e. 9.58 Phonology Conditioned Phonetic Changes 59 for example.g. in Arabic dialects.16-17).-Regressive and at distance: e. die for dik "cock" (of.9. dissimilation by means of n is extensive in Akkadian. § 8. Syr.11. Ar.-E. Ar.6. generally in preceding rather than following position: e.). This process is wide-spread also in some of the modern Ethiopian languages. The Semitic languages present phenomena of dissimilation between consonants. contiguous as well as at distance. 9. Ar. 8'dlomo "Solomon" (qualitative). Ug. between semivowels. cf. layl "night".g. d) Vowel to consonant. The meticulous phonetic notation which characterizes Hebrew Masoretic pointing marks the appearance of a-timbre vowels after the consonants' and h (and more rarely l}) in conditions when they would otherwise be vowelless : thus *ya'mod "he stands" > ya'amod. Ar.g. 9. *l}u§on "external" > l}i§on (qualitative). incidentally.-Progressive and at distance: e.ll0): e. In some cases. Such vowels are also indicated between long vowels of other timbres and pharyngal or laryngal consonants (e. In Hebrew and Syriac the succession of two vowels of u or 0 timbre occasions the dissimilation of one of them to i or e: e. Akk. *'aywam "days" > 'ayyam. In Hebrew a reduction might possibly be inferred from Greek transcriptions: e. sps. 8'dlemon for Heb. Akk.g.-This occurs particularly in Arabic: e. Akk. lin.g. *z'druw "sown" > Z'dr'dW (regressive and contiguous).g.Assimilated: e. lUn. Heb. Akk. *baytu "house" > bitu (total regressive assimilation). 2. *sams "sun" > Ar. *libb "heart" > Ar.g. '/sl}aq). though more rarely. Ug. in respect of d.-The pharyngal and laryngal consonants frequently occasion a change of other vowels to a (cf.42). Prosthesis 9. 9. *rul} "spirit" > rUiily. §§ 9.8. e) Rising Diphthongs Assimilated or Reduced. Ar. f) Faliing Diphthongs Assimilated or Reduced.g.12. the new vowel is instead placed after the first consonant (cf.g. 9. Of. Sem. *iwbil "he carried" > ubil (reciprocal assimilation). run counter to the rules of Semitic syllabic structure (where two contiguous vowels are impossible-except when a glottal stop intervenes between them: cf. lyn). *yislol} "he sends" > yislal}. 3. the Semitic languages do not permit the presence of more than one consonant at the beginning of a word. samsu. because it serves merely as a "catalyst" in the articulatory process (a similar phenomenon may be observed in Arabic. b. though in the Ethiopian example other factors may be at work as well.g. Sem.-Regressive and contiguous: e.

*tqattal "he was killed" > 'etqattal. pronounced [gabrg]) or by the insertion of a between the consonants ([gabgr]): cf. *'uzn "ear" > 'ozen (the original forms reappear upon attachment of suffixes: e.2) is frequently resolved by the insertion of a secondary vowel and the consequent creation of a new syllable. is basically a phenomenon of dissimilation: e. qulqala. have happened is much rather the secondary constitution of "weak" consonants or the lengthening of originally short vowels through the adaptation.g. Syr. In Syriac. Further examples in Ullendorff. Heb.5-9. There is ample evidence. In Akkadian the anaptyctic vowel is generally identical with that of the principal syllable: e. In Hebrew and in Syriac the prosthetic vowel is e: e. *mna "from" > 'amna. The process continues in some modern dialects and becomes operative also in foreign borrowings such as the modern Eastern Aramaic 'us tal "table" from the Russian stol.g. § 10. just mentioned). A special situation may.16.108-27). however. In Arabic the prosthetic vowel is i: e. cf. In Hebrew we have hi. The same method of resolving consonant clusters is employed (as we have seen-cf. the anaptyctic vowel is e. *zroa' "arm" > 'ezroa' (but also zaroa' [in this case the more usual form]. § 10. h in intervocalic or paravocalic position. §§ 9. arise as a consequence of the effects of sentence stress (cf. the ambiguity of the sixth vowel (which represents both a and zero) does not allow us to arrive at safe conclusions.g. 9.g. that consonantal clusters were avoided either by the addition of final a (e. gabr "slave". moreover. The concomitant action of assimilation. in fact.g. in the construct state. Syncope and Contraction 4.17.20.g. more rarely it is u: e. Syr. Similarly at the beginning of words: *ksud "reach!" > kuSud.14). Anaptyxis 9. ay (when they are not contracted) we have the development aw > awu > awe> awe 9. 16. *kalb "dog" > kalab (in Assyrian a occurs at times after i or u: e. sipiretu "letters" alongside sipretu. of biliteral roots to the predominant triliteral system (of. rigl un and rigl U "foot".g. (e. the alternative procedure.e. As for Ethiopic. *mawt "death" > mawet) and ay > ayi (e. dissimilation. i. In Hebrew the anaptyctic vowel is e which assimilates to itself the vowels a. too. uzan). 9. when occasioned by the succession of two of them.g.g. more rarely. the preceding vowel tends to be reduced or dropped as its position becomes pre-tonic: e.60 Phonology Conditioned Phonetic Changes 61 9. i (but not u) of the preceding syllable: this is the origin of the "segolate" nouns. pp. pp.g. 5.g. *gzi' "lord" > 'agzi'. *nqatala "he was killed" > 'inqatala. 9. of the syncope of " w.18. it does appear. *sifr "book" > s(}fer. In Arabic the case-endings prevent the formation of consonant clusters at the end of a word: 'abd un and 'abd u "slave". In late Akkadian the weakening of stress favours the rise of secondary vowels in the middle of words: e.g. 'aMi "my slave"). Ullendorff. Heb. SLE. *uzn "ear" > uzun. cf. In many cases. Heb. *bn "son" > 'ibn. makes it difficult to establish generally valid rules of contraction or lengthening. The syncope brings about vowel contraction or compensatory lengthening. *qtul "kill!" > 'uqtul (vowel harmony In Ethiopic the vowel is a: e. A consonant cluster at the end of a word (which would be contrary to the principles of Semitic syllabic structure.in the verbal theme Hithpael-possibly by analogy with the theme Hiphil: e. as well as interchanges between the "weak" consonants. What may. e.15. 'etqattal. n. *'abd "slave" > *'abed > 'abed. throughout the whole of the Semitic area. 201-207. Syr. SLE. y and.g. but it may be said here that as a general rule vowel lengthening is .g. however. In the example just quoted there occurs compensatory lengthening of the vowel. New Bab.14-15) also at the beginning of a word. the explanation of forms as resulting from syncope or contraction is purely conventional. also § 8. *I}bat "take!" > I}abat (but some verbs use i instead of a: e. *bayt "house" > bayit). *'abd "slave" > 'ebed (but before laryngals and pharyngals a remains and even harmonizes the anaptyctic vowel). analogy. §§ 11. The syncope of vowels or consonants. by analogy. qiqalon "shame". hitqatt(}l (so also in the imperative of the Niphal).100. 198-201. For detailed information the grammars of the various languages have to be consulted. For the diphthongs aw. cf.g. *lmad "learn!" > limad).19.g.

sullam.1. *tami'a "he thirsted" > Reb. Ar.22. show a metathesis in the opposite direction. against Syr. Syr. in fact. Reb. *~itmur "he desires" > ti~mur. A few cases of haplology may also be observed in other languages: e. D. i. biida (the Reb. In Akkadian the non-prefixed forms of the verbal themes with t and tn and the adjectives of the pattern qitbar. rii's. Ar. the participles Akk.3).wy. (d) two vowels markedly distant in their basis of articulation may produce a vowel with an intermediate point of articulation.wy "she is prostrate" > ts{f:i. (e) h: e.e.r. Syllable and Stress 1. against Ar. A particular aspect of sentence-phonetics is that constituted by the extension of certain phenomena beyond the limits of the word itself. with a prevailing. in later Akkadian contraction usually takes place. simmiltu "ladder". against Eth. ~am~. Metathesis 9. § 16.g. by their effect on the boundaries of neighbouring words (syntactic phonetics or sandhi). g~r (but Ar. Sem. Sem. 8. in which the element t would normally be infixed. Syr. ra's. Consequently. Of contraction it is generally true that (a) the combination of t}Vo like vowels results in the same vowel. *'etsiimek "he leaned" > 'estiimek. dala. sa'imu "determining". Eth. *'aryaya "lion" > 'arya. Sem. *hitsammr. the t becoming a prefix. g (§ 9.g. *~itbutu"grasp" > ti~butu. Syr. The omISSIOn of one of two contiguous syllables with identical consonants (and sometimes vowels) is a phenomenon of dissimilatory origin which occurs in various Semitic languages.g. take place: of. ~. Eth. *ttff:i. Ar. nsk and nks "to bite". against Eth. Very wide-spread in the Semitic languages is the metathesis of t (as part of the verbal theme. bakaya.17-23) with the first radical of the verb when this is a dental or palato-alveolar fricative: e.r "he was on his guard" > histammr. i. *gawir "guest" > Reb. TPA.g. The following may serve as examples of the principal changes: (a) postvocalic ': e. bakii. Akk. Phenomena of syncope and assimilation are widely attested in Arabic-affecting contiguous words-by the tradition of Koranic reading: thus in sura 24. in verbs with medial w/y). (b) intervocalic ': e. k. either short or long. 149-53). Fleisch. when the first radical is z. aiila. against Ar. *dalawa "he drew" > Reb. qatalo.g. 8utaqbur. Reb. Ug. Reb. b. Syr. (d) intervocalic y: e. Syr. Syllabic Constitution 10. (f) Old Akkadian and Arabic agree that contraction does not generally take place when the second of the two vowels is a. b) consonant followed by vowel .21): Akk. the two consonants change places with each other (cf. in the case of assimilation. Syr. gar). Ar. baaa. *ta'ra "gate" > tar'a.62 Phonology Syllable and Stress 63 unaccompanied by changes in the quality of the vowel-unless it is occasioned by assimilation to a semivowel. qatleh.g. qa'im "standing".g. (e) the vowel resulting from contraction is generally a long one. ?ami'a. TPA. (c) the quality of a stressed vowel tends to prevail over that of an unstressed one. 'atraba and 'artaba "he was poor". 'istaqbara.g. ros. Examples of metathesis are to be found in all the Semitic languages: e.21. Thus. Sandhi 9. *dipsu "honey" > dispu. 98-138. verbs differ in meaning from Arabic).g. 6.g.23. dalawa. and sometimes s and also d: e. the fricative pronunciation in post-vocalic position of the consonants p. Sem. *ra's "head" > Akk. qatalahu. *dituku "combat" > tiduku. Ar. dalii. qiitalo. against Ar. *qatalahu "he killed him" > Reb. s. Fleisch.5) becomes operative also at the beginning of a word when the preceding one ends in a vowel. e. Raplology 9. Mka. e. Sem. simla "coat" and salma.g. Ar. Sem. baaa'a. Certain combinations arise from Arabic morphology and may be eliminated by haplology: e. Eth. pp. pp. when the verbal theme with prefix s is combined with that with prefix t. 'astaqbara (for the paradigm qbr cf. Reb.g.e. of. yaqtulunana "they kill us" > yaqtulUna (cf. resu. *baaa'a "he began" > Reb. §§ 16. Some metatheses can only be detected by comparison with other languages: e.44 lJalaqa kulla dabbat in "he created all the animals" > [lJalakkulla dabbatin]. biika. There are two types of syllables in Semitic: a) consonant followed by vowel (open syllable). Syr. For a detailed treatment of syncope and contraction in Arabic cf. t. (c) intervocalic w. *bakaya "he wept" > Reb. Akk. § 12. and Syr. Sem. 7. d. (b) where one vowel is long and the other short it is the timbre of the long vowel that tends to prevail (but in some such cases contraction does not. resa. Eth. tataqataluna "you fight" > taqatalUna.

10.g. ku~~um alongside ku~um "cold"). and in this connexion one may compare fluctuating pronunciations of the type of Syr. and. It follows from § 10.elta for dely. *qalil "little" > qallil. 'ozen. Old Bab. I. 10. To restrict ourselves to the more readily identifiable expiratory stress. There prevails in the Semitic languages a widespread tendency to eliminate exceptions to these rules. *'uzn "ear" > Akk. that the almost complete agreement between Arabic and Akkadian might facilitate a hypothetical reconstruction of Proto-Semitic stress modelled on these two languages. This trend is likewise connected with the operation of stress and will. Semitic originally postulated short vowels in closed syllables. 1J. Moscati. b) long. Two vowels cannot be in contact. In North-West Semitic. Comparative Grammar 5 . The pronunciation of s'dwa as 'd in certain conditions has been considered by some scholars as a secondary phenomenon of an anaptyctic character (Gesenius-Bergstrasser. closed (and therefore long) syllable. As examples of prostehtic vowels cf. For example: qa. when it ends in a short vowel. *tumma nkasara [tummankasaraJ "then it was broken". p. uzun. pluT. e. and in transcriptions of foreign words Syr. There are also some cases of consonant-doubling without any parallel vowel-lengthening. § 10. A sequence of two consonants at the end of a word may result from the shedding of final vowels. though not exclusively.g. The shortening of long vowels in closed syllables is characteristic also of Eastern Syriac (e.g. We lack sufficient data to determine the position of stress in Proto-Semitic or to distinguish clearly between expiratory stress and pitch accent.itfu for 1J. Akk.g. that the infringement of the general rule is due to the workings of analogy. qab).o. conclusions about syllabic constitution depend to a large extent on one's judgment as to the nature of the s'dwa (§ 8. dely. But some cases occur also irrespective of stress: e. Etspecially in Aramaic: e.3.2. g'dmallim.63). therefore.g. Hebraische Grammatik. For anaptyctic vowels cf. and its general application over the Semitic field may be subject to some doubt.1 that in Semitic every syllable normally begins with one consonant and one only. Fairly common is the tendency to lengthen short vowels in open syllables.g. Two consonants may generally be contiguous only in the middle of a word (final consonant of a closed syllable and initial consonant of the following syllable). Stress and Associated Changes W. gamal "camel". where the final vowel is dropped in pause (naziliina > naziliin "descendants"). secondly. when it ends in a long vowel or in a consonant. SLE. *mil'u "fullness" > *milu > milu (compensatory lengthening to restore the syllable rhythm of the word). Sometimes consonant-doubling (gemination) takes the place of vowel-lengthening and so restores the closed syllable with short vowel: e. Reb. Eth. be dealt with in that connexion (§§ 10.itu "sin" (this occurs predominantly. 'atrones from {}eovor.l'dta "fear".82). §§ 9. in the first place. therefore. in the later period of the language: cf.14-17) or else through word juncture. *'almin "eternity" > 'almin). Ar. This phenomenon is connected with the incidence of stress and will. In Arabic the shortening of long vowels in closed syllables is a rule (e. Reb.. the only exceptions occur in certain syllables of secondary origin. the Ethiopian tongues make fairly extensive use of prosthetic and anaptyctic vowels (Ullendorff. 2. 'espera from (JqJaie a . 199 to 200). pp.5-11). *qum > qum "rise !"). it is a fact that long vowels show a tendency to become short when their syllable closes. The term "ultra-long" is used of syllables (cf. Akk. *'atana "she-ass" > 'attana. either by means of prosthetic vowels (graphically supported by') or anaptyctic ones (cf.4.g. Liaison of words occurs in Ar. that in Proto-Semitic it is unlikely to have had distinctive or phonemic status. It is possible. If one maintains the primarily vocalic character of s'dwa. one has yet to recognize that Masoretic pointing acknowledges the succession of two consonants in initial position (it does not register the second consonant as a fricative-as it should have done in post-vocalic position) in stayim "two" (to which sta corresponds in Syriac). qa.5-11). a syllable may be: a) short. 134-35). however. According to Brockelmann (GVG. we may say. see next §. open long syllable. be dealt with in that connexion (§§ 10.3) which are closed in addition to having a long vowel (e. ending in two consonants. Syr. pp. *nkasara > 'inkasara "it was broken". Nevertheless. This rule is mainly based on the position in Arabic. open short syllable. qab. For syllables in final position. 10.64 Phonology Syllable and Stress 65 followed by consonant (closed syllable). Among modern Semitic languages. Quantitatively.

ending in two consonants.g. a syllable may be: a) short. Ar. see next §. the Ethiopian tongues make fairly extensive use of prosthetic and anaptyctic vowels (Ullendorff.82).g. Stress and Associated Changes 10. For anaptyctic vowels cf. SLE.g. open long syllable. *mil'u "fullness" > *milu > milu (compensatory lengthening to restore the syllable rhythm of the word). plur. 'ozen. I. For syllables in final position. and its general application over the Semitic field may be subject to some doubt. §§ 9. *qalil "little" > qallil. delwlta for del}l(Jta "fear". *'almin "eternity" > 'almin). Syr. that the almost complete agreement between Arabic and Akkadian might facilitate a hypothetical reconstruction of Proto-Semitic stress modelled on these two languages. The pronunciation of s(Jwa as (J in certain conditions has been considered by some scholars as a secondary phenomenon of an anaptyctic character (Gesenius-Bergstrasser. ~. where the final vowel is dropped in pause (nazilUna > nazilUn "descendants"). A sequence of two consonants at the end of a word may result from the shedding of final vowels. Eth. § 10. especially in Aramaic: e. The shortening of long vowels in closed syllables is characteristic also of Eastern Syriac (e. This rule is mainly based on the position in Arabic. Fairly common is the tendency to lengthen short vowels in open syllables. 1J. Reb. Semitic originally postulated short vowels in closed syllables. and in this connexion one may compare fluctuating pronunciations of the type of Syr. qab. Reb. in the later period of the language: cf. For example: qa. There are also some cases of consonant-doubling without any parallel vowel-lengthening. In Arabic the shortening of long vowels in closed syllables is a rule (e. Nevertheless. that the infringement of the general rule is due to the workings of analogy. gamal "camel". g(Jmallim.4. 'atrones from {}e6vor. the only exceptions occur in certain syllables of secondary origin. But some cases occur also irrespective of stress: e. *qum > qum "rise !"). 2. This phenomenon is connected with the incidence of stress and will. it is a fact that long vowels show a tendency to become short when their syllable closes. This trend is likewise connected with the operation of stress and will. Hebraische Grammatik. Akk. 'espera from IJ'cpaie a .1 that in Semitic every syllable normally begins with one consonant and one only. b) long. closed (and therefore long) syllable.g. secondly.5-Il). Two vowels cannot be in contact. Liaison of words occurs in Ar. As examples of prostehtic vowels cf. we may say. 199 to 200)..2.iHu for 1J. p.3) which are closed in addition to having a long vowel (e. open short syllable. when it ends in a short vowel. The term "ultra-long" is used of syllables (cf. 10.3. ku~~um alongside ku~um "cold"). pp. *nkasara > 'inkasara "it was broken". therefore. 63). and in transcriptions of foreign words Syr. be dealt with in that connexion (§§ 10. Among modern Semitic languages. *iumma nkasara [tummankasaraJ "then it was broken". 134-35). There prevails in the Semitic languages a widespread tendency to eliminate exceptions to these rules. though not exclusively. Quantitatively. In North-West Semitic. pp. It is possible. uzun.itu "sin" (this occurs predominantly. According to Brockelmann (GVG. If one maintains the primarily vocalic character of s(Jwa.64 Phonology Syllable and Stress 65 followed by consonant (closed syllable). J\[oscati. It follows from § 10. Sometimes consonant-doubling (gemination) takes the place of vowel-lengthening and so restores the closed syllable with short vowel: e. conclusions about syllabic constitution depend to a large extent on one's judgment as to the nature of the s(Jwa (§ 8. one has yet to recognize that Masoretic pointing acknowledges the succession of two consonants in initial position (it does not register the second consonant as a fricative-as it should have done in post-vocalic position) in stayim "two" (to which sta corresponds in Syriac). *'atana "she-ass" > 'attana. 10. To restrict ourselves to the more readily identifiable expiratory stress.14-17) or else through word juncture. *'uzn "ear" > Akk. We lack sufficient data to determine the position of stress in Proto-Semitic or to distinguish clearly between expiratory stress and pitch accent. however. Old Bab. Two consonants may generally be contiguous only in the middle of a word (final consonant of a closed syllable and initial consonant of the following syllable). qab). qa. Comparative Grammar 5 .5. either by means of prosthetic vowels (graphically supported by') or anaptyctic ones (cf. Akk.g.g. be dealt with in that connexion (§§ 10. and. therefore.g. 10. that in Proto-Semitic it is unlikely to have had distinctive or phonemic status. when it ends in a long vowel or in a consonant. in the first place.5-11).

*ydqburu > yiqb6r). §§ 10. *'u!!ni > 'oznt). *wdsibat > wdsbat "she dwells": cf. tamiini. is subject to much uncertainty (cf. In any case. Examples: a) accent on ultima: san1i ( < *saniyu) "second". contrast to the pattern in classical Arabic. failing that. 10. p. however. Secondary stress patterns arise in compound words of some length.13):a>ii. According to some recent studies (Birkeland) it would appear that these rules might derive from the stress patterns of certain Arabic dialects. f) in final open stressed syllables i becomes e (Ar. For classical Arabic the rule given in the preceding paragraph is of universal application. In contrast to Akkadian and Arabic. 10.g. sJmi5ni) . either under the influence of the word-accent or by contextual stress patterns (pause) (0£. e. *ddbaru > diibdr. Stress patterns and syllabic constitution are bound up with complex rules of vowel evolution which (leaving out of account the difficult question of their origin) may be summarized as follows: a) final short vowels are dropped (*qdbara > *qdbar). know any express Arab tradition of acceptable antiquity which might elucidate for us the origin of the stress rules now observed in reading classical Arabic. napistu "life" (closed syllable). c) short accented vowels undergo lengthening or change of timbre. *dabartm > dJbiirtm. stress is non-phonemic in Akkadian. i > e. Examples: qatdltum "you killed". g) short vowels in open unstressed syllables are reduced to J in accordance with the general Semitic tendency and in contrast to the instances listed under (d) where pre-tonic syllables frequently undergo lengthening. it can obviously not be phonemic. b) stress shifts to the last syllable which the development under (a) has left closed and therefore long (*qdbar > *qabdr). so as to form a closed syllable. *luqdlJ > luqqdlJ. u>o (*ddbaru > diibdr. e.7. it is likely that these two opposed tendencies were operative at different periods: e. qdtalU "they killed". 38) does stress fall on a short syllable in the middle of a word.e. the initial syllable.g. either by the lengthening of the vowel. imqutU > imq1itu "they fell". before two successive consonants. stress does not fall on a final syllable (even if it be the result of contraction) but goes back as far as possible till it meets a long syllable or.g. *'inabu > '?nab (but *lJimaru > *lJJmBr > lJamBr). u remains.g. or both. b) otherwise stress does not fall on the ultima. however. i. Heb. *qabiru > qi5b~r. d) in contrast to the general Semitic tendency.8-10). stress in Hebrew may have distinctive or phonemic value: e. 'imriitB and 'emriitB. but siib1i "they took prisoner". Only in rare instances (cf. Strong expiratory stress may occasion reduction in neighbouring vowels. open pre-tonic syllables undergo lengthening and sometimes change of vowel quality: a > ii. e) short vowels in closed unstressed syllables may undergo change of quality: a > i. In these dialects considerable developments have taken place which-with regard to this particular feature-have brought about some affinity to the situation which prevails in other Semitic languages (cf. 5* . and probably by a relatively late process of restoration. but recedes as far as possible until it meets a long syllable (if there is no long syllable stress comes to rest on the first syllable of the word). We do not. i > a instead of i > ~N (*ziiqinta > ziiqdnta).g. or by the doubling of the following consonant. b) accent on long syllable: bel1itu "lordship" (long vowel). § 1004): e. on antepenult (short): kUbburu "stout". §10.. u > 0 (*madbdr > midbdr [dissimilation ~].S. the position of the stress may be expressed as follows: a) if the final syllable is the result of contraction it generally bears the stress.3-4. In Hebrew (at least as far as can be judged from the Masoretic tradition) stress falls on the last syllable-save for some cases of penultimate patterns. e.66 Phonology Syllable and Stress 67 The risks inherent in this procedure need hardly be underlined when we recall that the situation in Arabic. von Soden. but the following consonant is doubled (cf.6-7). with pronominal suffixes: so us~rib1i-su "they let him enter". In Akkadian. i>~N.g. in notable 10. iskUnu > iskunnu "they put". If stress falls on a short syllable it may cause it to be lengthened. GAG. e. sabu "they returned". *qiibar1i > qiibJr1i. As stress in Arabic is bound. i > ? (or else J according to the development referred to under g). so far as our limited evidence permits. §§ 10. a reconstruction. mdmlakat "kingdom".g.S. even if it is long. and particularly in Akkadian. §§ 10.

68

Phonology

Syllable and Stress

69

As for the pre-Masoretic stress-accent, this must have diverged notably from its later Masoretic version (as has been shown by Bmllllo from Greek transcriptions): cases such as cpifh{}a for pittal}ta, avvw{}Ev for hannot~n, CW'rJp,8eOV for zamm'iwii, r56.f3e'rJ for dibr~, etc., testify to stress conditions contrasting with our notions derived from the Masoretic recension of the Hebrew text. 10.9. Compared with the ample documentation of Masoretic Hebrew, the evidence concerning North-West Semitic of the second millennium B.C. and the rest of Canaanite is exceedingly scanty: a) there are indications of a reduction of short vowels in unstressed open syllables (cf. § 10.8 g) in Amorite (e.g. A-ma-na-nu-um and Am-na-nu-um, ya-ta-ra-tum and ya-at-ra-tum), in the Tell Amarna glosses (e.g. mi!Wu for ma!Ji§u, §ilJru for §alJiru) , and possibly in Ugaritic (Garbini, SNO, pp. 75-77); b) in Phoenician we have some lengthenings of short stressed vowels, accompanied by changes of vowel quality (a > 0, i > e, u > 0), which reveal remarkable similarities to the Hebrew changes (cf. § 10.8 c). 10.10. In the Aramaic area, while Biblical Aramaic reflects the situation in Masoretic Hebrew, Syriac stress always falls on the final syllable. As in Hebrew (indeed, the Masoretes worked under the impact of Aramaic) there are complex rules of vowel development, connected with the incidence of stress and with syllabic constitution, which may be summarized as follows: a) final vowels, whether long or short, are dropped (*qabara > *qabar; *qabaru > *qabar [the final u is written but not pronounced]) ; b) stress passes to the final syllable which is now closed and hence long (*qabar > *qabdr); c) short vowels in open unstressed syllables are reduced to 9 or dropped (*qabdr > q9bdr); d) in closed syllables short a and i may become e (*qabrat > qebrat; *si/rd > se/rd); e) a short stressed u becomes 0, whether by the action of the word-accent (as for the change u > 0, of. also the opposition between West and East Syriac, § 8.88) or by analogy with pronominal forms and verbal suffixes (*qabdrtumu > *qabdrtum > *qabartum > *qabartBn > q'JbartBn) ;

f) i becomes e in final open and stressed syllables (Ar. tamani, Syr. t'Jman~). After the close of the classical period (about A.D. 700) final open syllables tend to lose their stress: e.g. nehwe "he is" in Maronite usage; the Nestorians stress the penult even in cases when the final syllable is closed: e.g. ketbat "she wrote". 10.11. In Ethiopic it is usually assumed that stress falls on the final syllable of a noun but the penult of a verb. Recent research by Ullendorff (SLE, pp. 189-97) shows, however, that the whole question remains complicated and more than a little uncertain. The very existence of expiratory stress in Ge'ez is doubtful, and in the traditional pronunciation it is difficult to distinguish between stress and pitch. In any event, the accent in Ethiopic (whatever its precise nature may have been) is non-phonemic.
3. Sentence Stress

10.12. In addition to word-accent, the Semitic languages have a sentence stress determined (especially in pause) by the traditional recitation of the text. This stress occasions a number of changes in some languages. 10.13. In Hebrew the principal alterations are as follows: a) stress is thrown back on to the penult (e.g. 'anoki > 'ani5ki); b) a short accented vowel is lengthened (e.g. mayim > mdyim) sometimes causing change of quality (e.g. 'ere§ > 'dre§). 10.14. In Arabic the principal changes are as follows (cf. Fleisch, TPA, pp. 172-90): a) final short vowels are dropped (e.g. qatala > qatal); this may in some cases affect the constitution of a final consonant group (of. § 9.18) with consequent anaptyxis (e.g. al-bakru "young [camel]", in pause al-bakur). b) the indefinite case-endings -un, -in are dropped, and -an becomes -a (e.g. malikun > malik, malikin > malik, malik an> malika) ; c) the feminine noun ending -at becomes -ah (e.g. malikatun > malikat > malikah); for a possible Hebrew and Syriac parallel cf. §. 12.33.

70

Phonology

10.15. For the other languages we possess no adequate indications about sentence stress, or, at any rate, no changes occur of the type we have witnessed in Hebrew and Arabic. There are, however, one or two hints: thus in Akkadian the word on which sentence stress falls, in interrogative sentences, shows a shift of the stress on to the penult or ultima with consequent secondary vowel lengthening: e.g. ippuSa or ippasu "will they do?" instead of ippusu.

III. Morphology
A. Preliminaries
1. Morphemes

11.1. The Semitic languages present a system of consonantal roots (mostly triconsonantal), each of which is associated with a basic meaning range common to all members of that root: e.g. ktb "to write", qbr "to bury", qrb "to approach", etc. These roots (root morphemes) constitute a fundamental category of lexical morphemes (cf. Petracek, ArOr 28 [1960], pp.564-68). The linguistic reality of consonantal roots is shown not only by their lexical implications but also by the laws governing the compatibility or otherwise of radicals (which do not concern the vowels: cf. § 11.10) and in the transcription of foreign words. Only the pronouns and some particles lie outside this system of roots. 11.2. The task of lexical individualization (lexical morphemes) and grammatical categorization (grammatioal morphemes) is assumed by vowels and by affixes (prefixes, infixes, suffixes): e.g. in Arabic, from the root ktb "to write" : kitiib "book", kiitib "writer", maktabat "library", kataba "he wrote", yaktubu "he writes", etc. The linguistic reality of vocalization and affixes, in their morphemic function, is clearly attested by their specific semantic implications. 11.3. Grammatical morphemes may be external, internal, or syntactical. External morphemes are elements attached to a root (affixes, cf. the preceding paragraph). Internal morphemes are constituted by the nature or disposition of certain phonetic elements (consonants, vowels, stress), and in the Semitic languages they appear especially in the "inner" (or "broken") plurals and in the passive conjugation of the verb: e.g. Ar. kitiib "book", pI. kutub "books"; qatala "he killed", qutila "he was killed". Petracek's important studies (ArOr 28 [1960], pp.547-606; 29 [1961], pp. 513-545 and to be continued) show that inner inflection is

72

Morphology

Preliminaries

73

particularly developed in South Semitic, though it is not without precedent in Hamito-Semitic generally. The importance of vowelalternation (apophony) in Semitic morphology has been stressed by Kurylowicz. Syntactical morphemes are constituted by the order of words or by independent elements; the latter are of relatively low frequency in the Semitic languages (for example in the Arabic formation of the future tense by means of the particle
sawfa).

11.4. The above account is concerned with consonantal radicals only, and it has long been usual to conceive of Semitic roots as purely consonantal; such a reconstruction is unreservedly maintained by some scholars (so Fleisch, TPA, pp.247-51) but is disputed by others. Von Soden, in particular, (GAG, pp. 51-52, 96-97) holds that vowel elements should be regarded as forming part of the root (see also E. Ullendorff in Or 28 [1958], pp. 69-70). Such elements, which can be identified in the imperative of verbs and in the prefix conjugation, are short in triconsonantal roots (e.g. pqid "to guard, to deposit") but predominantly long in biconsonantal ones (e.g. dilk "to kill", bni "to build"). It should be observed that these latter roots exhibit stable vowels in other parts of the Hamito-Semitic area as well (§ 11.6 c), so that the incorporation of original semivowel radicals (dwk, bny) may be ascribed to artificial reconstruction. A similar stability is seen in the vowel elements of nominal roots (e.g. Sem. kalb "dog") which have to be differentiated from the verbal ones which, in their turn, are to be divided into those indicating states or conditions and those connoting actions (cf. § 16.2). The distinction between the three semantic spheres of noun, adjective and stative verb, and active verb is reflected in a differentiation in the structure of the root. 2. The Proto-Semitic Root 11.5. In the historically attested Semitic languages triconsonantal roots form the great majority; roots with two or with four radicals are much less numerous, while those with one or with five are rare (in roots with more than three consonants there is a possibility of secondary formations by metaplasm, dissimilation, etc.). Examination of· the dictionary reveals the following phenomenon: there are many groups of roots having two radicals in common

which express identical or similar meanings. Thus for example in Hebrew: prd "to separate", prm "to tear", prs "to split", pr~ "to break down", prq "to pull apart", prr "to dissolve", prs "to distinguish" etc. All these verbs have in common the radicals pr and the basic notion "to divide". This phenomenon, which is widespread in the Semitic lexicon, raises the question whether many triconsonantal roots are not, in fact, derived from biconsonantal ones; and whether a system of biconsonantal roots may, perhaps, have preceded the triconsonantal theme in Semitic. 11.6. For a solution of this problem the following data must be borne in mind: a) The Semitic languages have many biconsonantal nouns (in addition to some mono consonantal ones) which, from the objects they denote, must be adjudged fairly ancient: e.g. dam "blood", yad "hand", yam "sea", etc. The assignation of these nouns to triconsonantal roots must be ruled out as contrived and farfetched. b) The so-called "weak" verbs exhibit many biradical forms: e.g. Heb. qam "he rose" (root qwm), 'f,-seb "I dwell" (root ysb) , Ar. ram-at "she threw" (root rmy), etc. It is our grammatical systematization which looks upon these forms as having "dropped" a radical, while one might maintain with as much reason that the "weak" radical-in those forms which contain it-was, in fact, added to the root for the sake of adaptation to the triconsonantal system. This consideration seems particularly cogent where the roots in question coincide semantically with others on a biconsonantal basis. c) Comparison with other languages of the Hamito-Semitic group strengthens the biconsonantal hypothesis: e.g. Sem. qil "to kill", Cushitic qal; Sem. p'l (I'l) "to make", Cushitic fal; (it should be noted-as indeed appears from these examples-that Cushitic possesses biconsonantal roots with stable vowel). 11.7. The data just set forth show that biconsonantal roots in the Semitic languages are not a hypothesis relating to a prehistoric period but constitute an historical reality attested by a group of nouns and by a series of verbal forms; this is further supported by the semantic concurrence of many roots in two of their radicals. There is, however, no sufficient reason for main-

or "determinant". §§ 9. nor can all three radicals be voiced. -ti) or Ethiopic (-ku. etc. The Noun 1. qatnu "thin". -ta. by interchange between consonants with the same (or a similar) point of articulation (plly.g. etc. surS "root". as an example of lexical analogy one might mention Heb.e. B. root q(n). "to furrow". fg. that the entire Semitic stock of roots was originally biconsonantal. i. brr "to separate". 3. respectively.causative). Morphology manifests the action of two fundamental forces in the development of forms: a) phonetic laws (e. 'abyarf. sbr "to break". Further aspects of morphological development are shown in such contrasting forms as (in the case of verbs primae ') Akk. Akk.g.74 Morphology The Noun 75 taining. gt. *-ta. assimilation. In the examination of biconsonantal roots it is to be borne in mind that the radicals may have undergone certain phonetic changes: thus alongside the series pr "to divide" (§ 11. in Arabic (taking into account the transition g > g) tg.). In no Semitic language can two identical consonants-or two consonants with a similar point of articulation-appear next to each other in first and second position.10.S. it may be said that the present state of the dictionary does not appear to permit the identification of specific semantic values attached to these "determinants" . b) apart from some grammatical formatives which retain some trace of their original function (e.g. in West Semitic. qt.g. identical consonants are frequently found but not different consonants with a similar basis of articulation. Sem.5) Hebrew also possesses the groups pl and br.44-45) is offered by the ProtoSemitic endings of the first and second persons singular in the verbal suffix-conjugation (*-ku. as some have done.mar "red". "white". with analogical extension of the elements t and k. tk. Themes or Patterns 12. there are many incompatibilities between dental and velar plosives: in Hebrew gt. It is a more likely supposition that originally there existed roots with either two or three consonants (as well as a smaller number with one only or with more than three) and that at a certain stage in the development of the Semitic languages the triconsonantal system prevailed-extending by analogy and thus bringing into line biconsonantal roots through the adoption of a third radical. b) analogy. by analogy with siMa "six" (cf. sams "sun". tq.asa7. Leaving aside biconsonantal roots and their development. the following questions arise: which consonants can be so used and with what specific semantic value? A lexical probe leads to the following tentative conclusions: a) all consonants may be used as "determinants". and it is rare for such consonants to be found as first and third radicals (e.) which have already been considered in the section on phonology (cf. tq are normally incompatible-but not tq. There exist other incompatibilities in individual languages or groups: thus in Akkadian g and z are never found in third position. kt are generally incompatible. In positions two and three. -ka. ly. tk. gt. also Fleisch. These are morphological types which are frequently associated with specific meanings or uses. As regards this third radical. tk. -ki). 11.11. 11. dissimilation. tg. both Northern and Southern. the evidence adduced in §§ 16. 11. nun "fish". sometimes in opposition to phonetic laws. the Semitic languages reveal certain structural incompatibilities which reduce the number of possible combinations in triconsonantal roots. *-ti) and their development in Arabic (-tu. For example: Ar. In the Semitic dictionary the system of roots is combined with that of themes or patterns. an example of morphological analogy (cf. 'azraq "blue" are . i"abit "was destroyed" alongside innabit "he fled"-despite identical phonetic origin. TPA. 7. Cf. Morphological Development < 11. The foregoing considerations apply particularly to verbal morphemes.1. 247-6l. In later phases of the language the possibility of the normative influence of grammarians on morphological as well as phonological development cannot be discounted. kt.amsa.9.2). tq. § 14.amissa "five" instead of *ly.g. qt. pp.u "to desire"). Akk.1-22). Sem. both morphological and lexical. s. etc. In nominal morphemes the position varies to some extent: e. 'aly. layl "night". and of two emphatic consonants one is reduced to non-emphatic status (e.

qaybur. Akk. the winer diffusion of qabiir as compared with qabar). qabir. c) the distinctions between noun and adjective. In the following. qtl contains the obnoxious t which may so easily be confused with its other functions (infixes etc. particularly in those languages in which stress and syllabic constitution affect the vocalic structure. 12.) in the morphological scheme. pp. It is greatly preferable to the usual paradigm qtl "to kill" which is not so far attested in Ugaritic.g. by and large these patterns occur indiscriminately for the various categories just mentioned (cf. For these reasons qbr will be employed for paradigmatic purposes throughout this book. the same forces have also caused a measure of uniformity and . Nominal patterns may be "simple" (when the root is modified by vowels only) or "extended". The system of nominal themes weakens in the course of morphological development. b) Disyllables with short vowels: qabar.76 Morphology The Noun 77 formed from the roots byg" ly. plural ending): four "themes" which are conditioned allomorphs or combinatory morphological variants. rapastu (in Akkadian the first three patterns are generally employed as adjectives). Ar. some of the principal nominal themes will be presented together with their main spheres of employment-where these can be reasonably well established.7. 'ebed (the "segolate" nouns: cf. Nominal Patterns 12. qawbar. i. a. Noun-patterns can be studied to best advantage in Akkadian and Arabic. The root qbr is found throughout the Semitic area and is one of very few verbal roots which combine this advantage with a suitable phonetic constitution.3. 2. From a semantic point of view (partly also owing to the insufficiently developed state of these studies) we are only occasionally able to assign specific values and uses to individual patterns. b) nominal themes with rising rhythm (with final long syllable) appear to be predominating (cf. when affixes are added. but. while in Hebrew and Syriac it is subject to the assimilation t > t (cf. construct state d"bar. 12.2.levelling as well as reductions in the number of patterns). qibar. fem. Owing to the requirements of Semitic syllabic constitution these patterns are liable to anaptyxis: e. 12.5. 'abd. and the pattern 'aqbar (cf. Hebrew is a typical example of such a language: e. qubur. TPA. Simple Patterns 12. as we are merely concerned with conventional patterns. qaybar. respectively. dabar "word". concrete and abstract. a) Monosyllables with short vowel: qabr. qabir. while some of the other languages (apart from those whose vocalization is not attested) exhibit a lesser number of independent themes but a great variety of formal developments resulting from the incidence of stress and syllabic constitution.e. abdu. c) Disyllables with long vowel or diphthong in the first syllable: qabar. are not always apparent from a purely thematic point of view: while in some cases differentiation and opposition may be recognized (particularly in Akkadian). but Heb. of course. These themes may be variants of the preceding ones occasioned by the influence of stress or by anaptyxis or by the extension of pausal forms: an example of the elision of a vowel under the impact of stress is Akk. qaybiir. construct before "light" suffixes d"bar. qibr. for phonological· developments which have a bearing on certain nominal patterns see the section on phonology. § 9. Fleisch. 12. qabur.6. zrq.17). qubar. above all.mr.g. Of these patterns qabir usually has the function of an active participle and is widespread throughout the Semitic lan- . *rapasu > rapsu "wide". The fact that in Akkadian the verbal inflexion of qbr shows the vowel i as against u in the other languages carries little or no weight. For these purposes the paradigm used for the identification of patterns will be qbr ("to bury"): this is. 349-76). qibir. qawbiir.3) to render the names of colours.3). From an examination of Akkadian and Arabic patterns certain preliminary conclusions may be drawn: a) not all patterns are necessarily differentiated in origin: some of them can be explained as secondary developments brought about by phonetic processes and analogynot infrequently corroborated by their meaning (on the other hand. qubr.4. qabur. § 12. For further details the grammars of the various languages concerned have to be consulted. merely an arbitrary choice and may therefore serve even when the resultant forms are not in fact attested. § 9. construct plural dibr( -e.

waras "heir".g. Ar. kot?b. kateb. katib "writer". Reb. gammal "cameleer". Ar. nia. qubbar. c) Patterns with doubled third radical: qabarr. mua. pp.g. qubarr. here the long vowel may be replaced by the feminine ending -at (whose addition places these patterns into the category of extended themes): qabar and qabarat. Reb. rasul "messenger"-cf. ly. qubayr. qabirbir. qabrar. qubur.g. Reb. da'ummu "pitch-dark"). qubar and qubarat. Eth. 'a?um "strong". 12. to the class of patterns extended by affixes. *ra'nan > ra'anan "green". taruq "timid". quburr are fairly rare and occur chiefly in Akkadian and Arabic. § 16. qibbir. The pattern qibar is employed in some languages for tools or instruments: e. qibar and qibarat. These are infrequent types-attested chiefly as adjectives (e. Eth. a) Patterns with doubled second radical: qabbar. gabbar "worker").13. nitaq.g. *yaraqraq > yaraqraq "greenish". zuqaqipu "scorpion") and the modern Ethiopian languages (Amh. In Akkadian they produce both nouns (e. the adjective of the verb-stem with doubled second radical and intensive meaning (e.12. formally.g. qabrur. namurru "shining". ri'did "cowardly"). A development peculiar to Akkadian is the pattern quburra' used for regular actions (e. 12. Ar.ta "spark"). Ar. Syr.g.g. 12. § 10.8 d) as well as similar uses of both forms to be sporadically encountered in other languages (e. qibrar. qabrir. qibarr. uturra'u "superfluity"). 'arakrak "thick". Eth. Reb. is used in Syriac as a passive participle (*qabir > qabir. ?iddiq "very sincere". namriru "splendour". and similarly qubbur. Ar.88).g. A few nouns of this type occur in Akkadian (e.g. Akk. Eth. qubrur. again on occasion with diminutive or pejorative meaning. qabarbir. c) Patterns with reduplicated second and third radicals: qabarbar. for the alternation u: 0 see § 8. Of these patterns qabir and qabUr are predominantly adjectival (e.g. These are fairly infrequent and mainly attested in West Semitic as adjectives (e. in Arabic they are employed for adjectives (e. qibbar. 380 to 389).10. Ar. The other patterns are more common in Arabic than elsewhere. 'abd "servant".11.g. § 10. qibirr. zahrira "ray". qabbir. but there also exist a few nouns (e. qabbur. qibbur. and qabur in Rebrew (*qabur > qabur. qubabir and others are attested in Akkadian (e. Reb.ung"). qabir and qabirat. § 16. *'ulayma "lad" > 'alayma. qubrur.10 c). qibrar. b) Patterns with reduplication of the second radical: qababar. b.g. qibrir.ibb "timid". qumudd "strong").g.g. Fleisch. 'umlal "languishing". qabbir. kulbabu "ant") and in Syriac (e. qatil "killed". Ar.g. Syr. with lengthening of the pretonic vowel-of. tallaq and talallaq "great").69). in Akkadian in nouns with diminutive or pejorative connotation: e. qabrur.g. The pattern qubayr is largely used as a diminutive: it occurs principally in Arabic (e. d) Disyllables with long vowel or diphthong in the second syllable. qanat (all three meaning "belt"). kurram "very generous").amalmil "greenish"). qubbar. Ar.8. farruq "very timid". qabbar is widespread throughout the Semitic languages and characterizes intensives and names of professions (e. qabur serves in Syriac as nomen agentis (e.78 Morphology The Noun 79 guages (cf. 12. it belongs. paroqa "saviour". ?ayyad "huntsman". kirissu "needle") and adjectives with intensive meaning (e. though traces of it are found in other Semitic languages (e.g. qabbar and qabbir are used in Akkadian for adjectives with iterative or intensive significance (e. Reb. qabbar and qabbarat. maliku "counsellor". qibrir.g. Reb.g. 12.ruly. sometimes with diminutive or pejorative significance (e. qubrar. arammu "dyke". qabarbur.g. qubbur. with reduction of the pretonic vowel-of. ?a'ir "small". dummuqu "very good"). Patterns Extended by Gemination or Reduplication of Radicals 12. Akk. qubrar. qabir. Syr. 'asatsut "common people". paraly. marir "bitter". partUta "piece"). qibbar. quddus "very holy". qibbir. '?zor. kabbaru "very thick"). qaburr.urra'u "reception") and for certain situations or conditions (e. qabarbar. 'ubayd "little servant" -of. *kusaypu > kusipu "morsel of bread"). Ar. 'allam "man of great learning". Ar.68): Akk. qabur.9. dayyanu "judge". TPA. Reb. d) Patterns with repeated third radical: qabrar. in particular. qubbur. qu'dud "ignoble" from the root q'd "to sit").g. .g. Of the patterns with short second vowel. Eth. kabir "great". Other patterns with long second vowel are chiefly employed to indicate adjectives with intensive meaning (e.g. Of these patterns. batulu "yo.

g. maskanu "place").luqtu "ruin" from the root 7. etc. miqbir. TPA. naqbur. 12. though less frequently.g. Four principal meaning-variants are connected with the prefix m-: local. taktusa "battle". recompense". §§ 16. b} Patterns with prefix y-: yaqbar. Akk.19. while maqbir and miqbir often designate instruments (e. 'iqbir.18. tamsilu "image". In the expression of these meanings the various patterns appear to be used indiscriminately. In particular. 'aswad "black". m. § 12. muqbiir (for prefix n. suqbur and suqburat. taqbiir and taqbarat. These are rare and confined to West Semitic where yaqbur in particular is used for names of animals (e. taqbir is used in various languages as the verbal noun of the verbal stem with doubled second radical (e.) and in Hebrew for certain other types of adjectives (e. Outside the Semitic area. Fleisch. 422-34. tafriq "distribution"). maqbur. Ar. Hebrew uses for abstracts maqbar and miqbar (e. pp.sdm.r "dolphin"). muqbar. Jewish-Aram. verbal nouns (e. 12. In the other languages this pattern is rare (some animal names occur in Ugaritic: e. and see also Fleisch. etc. infrequently. sanudu "very famous".lq "to perish") and also. mamliikii "kingdom" from the root mlk "to reign". Ar. 'akzar "cruel". For a detfl. taqbiir and taqbur(a)t appear in Akkadian as verbal nouns of the simple stem with infixed t (e.tu "repose". tallaktu "going"). 12. taqbir. yal}mum "black"). pp.g. 12. miqbar. maqbiir. and in this sense do not Moscati. These themes are used in Akkadian for verbal nouns of the stem with prefix s. To this group also belong the nominal forms of the verbal stem with sand t. taqbur. mu~lalu "midday"). 408-17) which serves in Arabic to characterize elatives and colours (e. maqbur. tahlukat "perdition"). §12. In Ethiopic m'Jqbar « *miqbiir) predomi- nates for nouns of place (e.g. 'al}mar "red". yabrul}. Compa. tapsu7. tagmul "favour. As for the other types.prefix of.g. d) Patterns with prefix t-: taqbar. a kind of antelope) and.g. 'uqbur.g.in most of the Semitic languages-cf. The most frequent of these is 'aqbar (of.aru "encounter". yaqbiir. *map7. taqbur.rative Grammar 6 . they are fairly rare in Arabic and are probably variants of 'aqbar. as adjectives with intensive meaning (e.. maqbir. etc. Syr. tam7.g. generally of the verbal stem with doubled second radical (e. maptrill} "key"). saqbur and saqburat. salhebet.g. 12. m'Jsraq "east"). Ar. abstract.g.14. TPA. The theme maqbur expresses the passive participle of the simple verbstem in Arabic. These themes are attested in Akkadian.(e. maqbartalso for nouns of instrument. Syr. taqrubtu "approach"). and muqbar(t) and muqbiir for nouns of time (e. Ar. miqbir.g. instrumental. c} Patterns with prefix m-: maqbar. patterns with prefix m. taqbur and taqburat.are attested in Egyptian (e. taf~am "completing").and causative value (e. malbas "dress"). § 16.g. These themes produce. tardad "repeating". ta'dira "help".g.aru "sum" > nap7. for the participles of the derived stems-which are formed with the prefix m. Ar. yet in individual languages a measure of differentiation can be observed: In Akkadian maqbar(t} and maqbar are employed for nouns of place and time (e. 'akzab "mendacious"). of plants (e.g. a) Patterns with vowel prefixes (introduced by'): 'aqbar. maqbir. 'abya¢ "white". mispa( "judgement" from the root spt "to judge"). Ar. yaqbir. while maqbar(t) and 1naqb'Jr(t) refer mainly to instruments (e. e) Patterns with prefix s-: saqbar. tiqbiir.g.g. Syr. etc.g.g.g. for prefix m. 'al}san "most beautiful" from l}asan «beautiful". surbu "huge"). maw¢i' "place" from the root w¢' "to put") and nouns of instrument by miqbar and miqbiir (e.g. salhObita "flame". Patterns Extended by Prefixes 12.19. Akk. Heb. Heb. sa7. yal}mur. for the most part.80 Morphology The Noun 81 c.t "cosmetics" from the root sdm). miftal} "key" from the root ftl} "to open").g.15.g. 'iqbar. Outside Akkadian there are only a few traces in North -West Semitic: e. dn7. and perhaps Ug.occasioned by dissimilation of m. tibyan "explaining". Eth.iled study of patterns with m. miqbar.16. yaqbur. Patterns with prefix n-: naqbar. Akk.aru.17. Arabic indicates nouns of place by maqbar and maqbir (e.g. s'tqt "she who causes to pass".96-101. temporal. by dissimilation of the labials m and p).and suffix -n [maqbaran] cf.g.before a labial see § 12. generally as variants (by dissimilation) of the theme with prefix m. for which of.21). taqbur and taqburat occur as verbal nouns of the simple stem (e. yabrul}a "mandrake"} as well as for adjectives (e. in Akkadian 'iqbir is found as a variant of qibr. Ar.21.68 for Wehr's study. taqbir and taqbirat. Heb.

dakyuta "purity").. *qadman > qadmon "eastern". sadqam "wide-mouthed"). qadamit "beginning". dalftirit "end". it is. The suffix -awi is characteristic of Ethiopic (e. therefore. Syr. conceivable that in some cases the -m might be a relic of mimation (0£. qabran. Reb. '(}rau. however. and Five Radicals 12. and in Ethiopic qastam "bow" (from qast). Patterns with infixed t: apart from the nominal forms of the verbal stems with t (for which cf. -it: when attached to other patterns (resulting in qabrut.16). Akk. g "v~ice". even though of rare occurrence . The ending -iy is called nisba in Arabic. Reb. a} Monoconsonantal patterns. Akk. qabrit.24. Syr. Ethiopic (e. massa'on "deceit". In Rebrew one may mention safam "moustache" (from safa) .g. sul"f}.g.20. Reb. Outside Semitic. t "cosmetics".g. l. b) Biconsonantal patterns with short vowel: qab. also in adjectives (e. 12. gitmalu "very complete". itbaru "very friendly"). qubran. qub (e. fU8lJnm "wide".g. § 12.35).g. Ug. -ay.. 'aqraban "little scorpion".g.g. y(}hUdi "Jewish". Ar. su'n. r(}s'an "old age"). § 16. puqdana "order".-Aram. 'ari}iy "terrestrial".g. When attached to other themes these suffixes produce adjectives :with the meaning "belonging to" (e. In Akkadian they appear only as feminines of the pattern -iy.wn "brotherhood". Patterns with -ut occur in Akkadian (e. They also appear. Palm. In Ethiopic we also encounter the ending -at (e. Phoen. q(}asat "holiness"). a} Patterns with suffix -an: qabaran. Ug. however. Ar. p.g.lide the Semitic languages (the ending is attested.g. Reb. infrequent. 6* . miranu "little animal").22. taysal as well as fays "great number". 'af). and. pe. e. -awi. 'al. Ar.g.] 'alyiirit "end").g.an "table". finally. Reb. whether they belong to the category at present under consideration: e. from the root sdm).) these suffixes produce themes connoting abstracts.g. as nouns and adjectives of the verbal stem with prefix n. Ug. Reb. p. fayaran "flight". Patterns Extended by Suffixes 12. Ar.g. Reb. A particular Akkadian type of the patterns with suffix -an is that which describes a special person in a special condition (e.23. nominal formation with the suffix -t is attested in Egyptian (e.t "art").17-23) we find in Akkadian qitbar for adjectives with intensive value (e. resit "beginning".: e.g. karme1. /ii (of. The pattern maqbaran appears in a few cases in West Semitic (e.-Aram. s "sheep". qibran. Ar. Two. Rebrew (e. 'alJ. ma'b(}rana "passage"). malkUt "kingship"). In some cases (e. dl. Reb. Ar.73.n "brother". maqbaran. Patterns Extended by Infixes 12. sa'). Syr.21. Eth.g. in diminutives (e. In Ethiopic these forms are. Pun.ru « *mal..g. Outside Akkadian a possible example is Ug. Syr. al.g.sdm. for example.witii "fever".26. Patterns from Roots with One. c) Patterns with suffixes -iy. Akk. perhaps originally as feminine morphemes (0£. qorban "sacrifice".25. sarrutu "kingship"). etc. qib. ESA 'l. Eth. and the same applies to Arabic. mal.g. Ug. kerem) we may have to identify independent patterns with suffix -1. Four. in Egyptian: lJmw. etc.). etc. pu "mouth". and this name has been extended to cover the same type of formation even outt. [as well as Reb. na'asat "youth". b)· Patterns with suffix -m: they are infrequent and occur predominantly in Arabic adjectives (e. nbldt "flames".82 Morphology The Noun 83 constitute an independent category (cf. d) Patterns with suffixes -ut. while in Ethiopic they produce abstract nouns (e. 12. 'ar'an "terrestrial").ty "artist" from lJmw. these are fairly rare: e. §§ 12. Syr. 12. nalbubu "enraged". Ar. 12. *pitran > pitron "solution". from a root ns'. se (of. n'Jgusawi "royal"). Bibl. madd(}'an "knowledge". Akk. Themes in -it are found in North-West Semitic (e. cf. however. kasday "Chaldaean".riyn) "first". Reb.). Rebrew has a number of nouns in which the usual change a > 0 does not take place. swyt "curtain".g.g. These themes (in which the suffix -an is attached to other patterns already discussed) occur especially in abstracts (e.g. e. m. Further examples are found in the various languages. 'alJa. the question arises. etc. hidmil alongside hidm "patched garment". Syriac (e. d. § 12. nadinu "vendor". Bibl. *'isan > 'isOn "[little-man =] pupil [of the eye]". Ar. nadinan1~ the seHer in a particular case of the particular object referred to).g. 76). namungatu "paralysis".irut "goodness"). sakran "intoxicated". f. Akk. possibly Reb.

but is sometimes expressed by means of lexical opposition (e. garg'Jra "threshing flail". Ar. iawr "bull". mamliika "kingdom". Heb. p. (e. of tob) "good". 'atan "she-ass"). for qab: Akk.g. 3). Ar. Egyptian 8. kimkimmu "wrist". qayb.g. Gender extends over the whole of the Semitic area (and beyond: cf. !Jum~iru "mouse".g. of bis) "bad".34-35). in Hebrew and Syriac. Ar.33. 'esr~ "ten". alma for 'lmt. Heb. Heb. 12. for qabqab: Sem. 12. Heb. In Arabic also exist the feminine endings -a' (type qabra'. fem. of 'aqbar. Finally. ''J!Jw.) (from bisa "bad".g. Names of animals figure prominently in these patterns. "son".g. f) Four-consonant patterns. According to Brockelmann (GVG. for qabb: Akk. -ay (for which may be adduced such words as Heb. Ar. in the compounds from 11 to 19). suru. Syr. Syr. qubb (e. kappa. Ar. Others of this type are formed. Heb. 'i8m. there are the very rare endings. qawb. tn'yay "error") and -~ (from -ay ~-cf. § 12.32. tabii.. Heb. The Semitic languages show instances of the fairly general phenomenon of masculine nouns with feminine gender morphemes .29.) (from talJti "lower").30. kOkiib. In Hebrew and Syriac the Proto-Semitic feminine ending (retained consistently in the construct state: e. of 'a~gar "smallest"). c) Biconsonantal patterns with long vowel or diphthong: qab. talJtit "lower" (fem. Syr. gender is not always marked-in relation to the corresponding masculine-by the feminine morpheme. e) Biconsonantal patterns with reduplication of both radicals: qabqab. for colour-nouns: e. Ar. 'aqrab "scorpion". Syr. tobii (fem. These are infrequent (e. I. the same is true of words of foreign origin. the feminine numeral Heb. Examples of other four-radical patterns are Akk. mamleket) develops in the majority of cases (in the status absolutus) into -a: e. The fem. Eth. Syr. sum.g. *kabkab "star" > Akk. Eth. Syr. galgal "wheel". Heb. 12. from roots with a smaller number of radicals. qunfurJ "hedgehog". while neither Phoenician nor Moabite manifests this phenomenon. The masculine possesses no special endings (zero morpheme).g. Syr. that these morphemes are attached to nominal patterns different from those of the corresponding masculine-thus constituting an instance of inner morphemes (cf. in the emphatic state: cf. 12. Syr. A process similar to that in Hebrew and Syriac seems to take place in Neo-Punic-to judge by Latin transcriptions such as Anna for /:tnt. biSa (fem. fem. qabqab. t "daughter"). kawk'Jbii. lJimar "he-ass". s~m.g. The Semitic languages distinguish two genders: masculine and feminine. is!Jil~u "potsherd". ~afra'. biSta "bad" (fem. qubqub. gize "time") is not associated with the feminine. §§ 12. Akk. constr.34-35). 12. 'uqb'Jra "mouse". Eth. this development may be understood from the Arabic pausal form -ah which has extended beyond its original function (-at> -ah > -a > -a).27. 12. i. but this explanation appears somewhat doubtful. sumu "name". by attachment of affixes or reduplication of radicals. for qawb: Ar. ''J8re. qabqub. For example: Akk. sm. saray "lady" and Syr. qib.g. tob. fem. ·whereas the feminine is associated with a special morpheme which probably goes back to a more complex and ancient system of classes (cf. Aram.31. for examples in individual languages: Akk. § II. 'aqrab. Ug. malikat "queen" (from malik "king"). lJazlJaz "swamp"). g) Five-consonant patterns. 12. Heb. tawra.34. qibb. kOkab.g. In addition to those already discussed (constituted by the attachment of affixes or the reduplication of radicals). Syr. forms are attested over the entire Semitic area on the pattern C1 aC 2CaaC 4u: e. Akk. Akk. etc. Ug. Eth. qibqib. ~'Jfard~a' "frog") and often of foreign origin. 8.oes not necessarily and invariably correspond either to sex or to the formal constitution of the noun (cf. Eth. qub (e. ''Jqarbii. kakkabu. Eth. Heb. of 'a~far "yellow") and -a (type qubra. tabu "good". Ar. 'alckabis "spider". kawkab. 3. d) Biconsonantal patterns with doubled second radical: qabb. §§ 12. Heb. elative: e. however. 8'Jm). 'arwe "beast". ~ugra. the ending -(a)t which 12. The Ethiopic ending -e (e. Ar. Ug.28. of 'aqbar.409). 8arwe "army". Syr. kaf!).74). qubqub. dakdak "plain". 'p'p "eye". Grammatical gender d. lJanbal "saddle". fem. kap [kappi "my palm"].84 Morphology The Noun 85 Eth. kappu "palm [of hand]". b'J''J8it "woman" (from b'J''J8i "man"). Heb. ilt "goddess" (from U "god").e. sor). sarrat-u "queen" (from sarr-u "king").g. tab. It should be noted.

etc. This multiplicity of function points to the probable origin of the feminine ending in a more complex system of classes within which the category of number has to be included as well (by way of the collective). and laterin Ya'udic: 12. Reb. abstract and collective nouns. pluralizations of forms already plural.g. Reb./acc. Reb./acc. sing. further pI. and in that case it is referred to as "external" or "sound" plural. yamm'Jta "lake" (cf.n.3. nepes. sarl'u "king". buldan) or in that of external plurals of internal ones (e. erfetu). Reb. A comparative examination of the Semimc languages suggests the following Proto-Semitic morphemes for the external masculine plural: nominative -u (cf.70-77). § 12. 12. for some cases in other languages cf. {}ali/at "caliph". plural. -i are preserved in Ugaritic: e. Ar.39. zabna "time" = "temps"). maZon "inn"). a.g. § 11. this occurs both in the form of internal plurals of internal plurals (e. 'oni "fleet"). nafs. qaf{Jabuna. bilM. while the independent ending of the accusative singular (-a) merges with that of the genitive in the plural. 'eref. Finally. the plural may be formed by the addition of endings. gen. 'ozen. gen. Some nominal patterns in Akkadian present variant forms with or without the feminine ending but without a corresponding difference in meaning (qibr: qibirt. Akkadian adjectives exhibit the special endings nom. Syr.e. Reb.g. sarru. shades of the dead" (for the final -m or -n. furuq plus external fem. qohelet n. pI. pI. sannay). rabu "great". gen. 'edna. uznu "ear". In Akkadian the endings of the masculine plural are nom. External Masculine Plural 12. pp. it is noteworthy that the cardinal numerals from 3 to 10 use the forms without gender-ending as feminine and those with the usual feminine morpheme as masculine (cf. wamqat "leaf" (cf. with or without an accompanying vowel.56 and von Soden. nom. nom. 'ar'a "earth".pr. tariq "way". fU/iy "mystic"). pI./acc. gen. nIl' "god". of the "internal" plural.g. the use . naratu) recalls the similar phenomenon in Ethiopic (cf. below §§ 12. and dual. ikkaratu. The fairly frequent appearance of plural endings of feminine form (cf. z'Jbatta "time" = French "fois" (cf. waqalf "insolent"). malik "king". The dual is formed by the attachment to the singular of special endings. Neo-Bab. pI.g. mulUk. Examples of collectives: Ar. waqalfat "insolence" (cf. and in that case it is called an "internal" or "broken" plural./acc. l'abUti. sal'rile. rp~im. 'arq" Reb. The names of paired parts of the body are generally feminine. and similarly in Rebrew: naqam and n'Jqama "vengeance". mbutu. -utu. These endings seem to be the result of the lengthening of the corresponding singular morphemes of the nominative (-u) and genitive (-i)./gen. Ar. 'oniyya "ship" (cf. -u. § 12. An example of the "external" plural is Ar. m'JZuna "hut" (cf. Ar. Reb. "traveller") .g. or it may be expressed by a change of pattern (i. fufiyyat "the mystics" (cf. 'urJ. qaffab "butcher". Examples of feminine nouns without feminine ending: Ar. It is interesting to note that in these latter cases Akkadian attaches the feminine ending (napistu. 4. pI. ra'a "wickedness" (cf. ikkaru "peasant". internal pI. In some languages we find hybrid plurals. yet without feminine ending (e. In North-West Semitic the original endings -u. § 12. pI. 'or'Jlfa "caravan" (cf./acc.66. 12. where -i (-e) prevails and is extended to the nominative: e. cf. naru "river". Number 12.41). Syr. Examples of abstracts: Ar. ''Jz'Jn). GAG. genitive/accusative -i. -uti: e. yamma "sea"). Ar. pI. wamq "foliage").e./acc. Eth. Syr. pI. d. sal'l'ile. balad "locality".86 Morphology The Noun 87 and feminine nouns without them. pI. The feminine morpheme is employed not only to indicate the corresponding natural gender but also nomina unitatis.37. The Semitic languages possess three numbers: singular.2). later extended in part to Late Babylonian) from the earliest occurrence until the Neo-Babylonian and Neo-Assyrian period. and Neo-Ass. Akk.38. 77-78) for masculine nouns (e. Eth. na/s "soul" (also masc. {uruqat}.g.g.w). Examples of nomina unitatis: Syr.). Eth. For the characteristics of inner inflexion. napsa. gen. Examples of diminutives or pejoratives: Reb.35. nom. maqbar: maqbart. 'or~alf "guest". also Egyptian -w: e. Syr. -i (Assyriap -e. Examples of masculine nouns with (apparently) feminine ending: Ar. sannayt "goodness" (but also the masc. taqbir: taqbirt.36. ra' "wicked"). of a pattern different from that employed for the singular). § 14. ntl'. l'plm "demigods. nom. diminutives and pejoratives.). Ar.m. ma'on and m'J'ona "dwelling". i.

mara "ass".42. pI. qub'ur corresponds in Arabic to various singular patterns. pI. In Akkadian we find nom. nom.yanite (drawn from forms in the construct state. The ending -an appears also with internal plurals (cf. 14-15) regards it as an ending without specific significance which is used to reinforce short nouns. In classical Arabic the Proto-Semitic endings remain: nom. also §§ 12. qura from sing. -i: e. § 8.44.g.88 Morphology The Noun 89 e. ly.adi8 "new". participles. mlkw "kings"./acc.g./acc. Syr. qa§l§lab "butcher". § 11. as has been explained.adi8an. -ani (combination of -an with -u. bisin. a) Disyllabic patterns with short vowel: qabar. sarrani (Late period sarrani for all cases).52): e. 470-505 for a detailed discussion. § 12. Akk. sarru "king".g. pI. since in the absolute state the purely consonantal spelling does not allow any valid conclusions) suggest the same state of affairs: e. of '"zn "ear").mar "red". It must therefore be supposed that internal plurals are a particular development of South Semitic. pp. pI. b) Monosyllabic patterns with short vowel: Arabic qubr. Syr. Internal Masculine Plural 12. of §lMt "trench" [ ?]. tnw.41.emra from the singular ly.47. In Syriac we find -anin (combination of -an with -in): e. According to Goetze (Language 22 [1946]. cf. pI. fem. but chiefly to those with the second vowel long (e.g.46. In North-West Semitic there are some late and rather doubtful traces: e. §la7}rum. the feminine ending has greatly expanded at the expense of the masc. the merging in " of the Proto-Semitic short vowels i and u (cf. qubar.43. 12.14): e. Bab. In Ethiopic the plural endings are -an for the masculine and -at for the feminine (cf. mayat (cf. -ani for all cases in Late Akkadian: e. The internal plurals may be regarded as Proto-Semitic. Internal plurals are formed. in the sense that their patterns are Proto-Semitic. ilU "the gods"./acc. corresponding to singular q"br (e. attention might be invited to Old Assyrian §lu7}rum. Cf. The masculine external plural -an current in Ethiopic (cf. qurya from the singular q"rita "village" (Ar. The existence of internal plurals in Ugaritic has not been demonstrated. bny. the preceding paragraph) appears also in other parts of the Semitic area. qa§l§labuna./acc. rekeb from the singular rok~b "horseman" (Ar. For Akkadian. gen. Gelb (Morphology of Akkadian.g.g.38. qibar corresponds in Arabic to the singular qibrat (e. rabba "master". rabbanin.g. of 'ulbat "box").3) have shown the existence of precedents in Hamito-Semitic. qibar.g. In South Arabian the situation must be similar. pI. pI. gen. §lad"q "just". 12. §le7}rum "small". pI. The patterns .96) produces the pattern q"bar. rakib) . or with short second vowel and feminine suffix: qabir. their use as plurals is regularly found only in the South. masc.45. 'nan. kutub. §lad"qan. ilanu "some gods" or "the gods taken individually"). and a smaller number of substantives: e. their employment as plurals cannot be established-except when they are construed as plurals in terms of grammatical concord. qaryat) . of q'Wat "piece") and qubar to the singular qubrat (e. bis "bad" (status absolutus). 8u8im. ly.56).. ma8ily. Elsewhere -i predominates and is extended to the nominative: e.g. Semitic area (Arabic and Ethiopic). § 12. of tnwt "plain"). pI. mlky. b. Reb. Fleisch.. pI. pI. 12. 12. Syr. the indications furnished by South Arabian and Lil). unknown to us (e. qita'.in use for adjectives. The following paragraphs will indicate the principal patterns used for internal plurals-together with the singular themes to which they correspond.g. 12.40. The ending -an remains. However.g. gen. pI. -anu. pI./acc. For the pre-classical phase of Arabic.an. by the use of patterns different from those of the singular. nom. qubur. gen. TPA. Lil). may "water". Of these themes.g. so used may be regarded as original collectives. below § 12.g. of kitab "book"). pI. Heb.g. although Petracek's studies (cf. c) Disyllabic patterns with long vowel in the second syllable. §lad"qat. of course.g. constr. sarranu. ralcb from sing.g. pp. pp. ly. 121-30) the ending -an designates "individual" plurals as distinct from "general" ones (e. pI. plural of 'aqbar (nouns of colour. gen.g. 12. ma8i(/' "Messiah". In Ethiopic. 12. gen. bnw "sons".ttmr. though the vocalization is./acc.yanite pI. qa§l§labina. 8U8 "horse". However.50). of 'aly. pI. -i) in the Old and Middle periods. ly. -u. a collective corresponding to the singular Ass. 'ulab. morpheme: e. 12. nom. §1M.

pI. pI. Hebrew (owing to the well-known transition a > 0 [of.g. of 'amd "pillar". nom.64-69 below). 70-78). pI. malikat "queen" (fern.g. 'aqbur (as well as qibrat) for the so-called plural of paucity. fern. of gabr "servant") . 'anfus. 'qbrt. Noteworthy is also the use of the themes 'aqbar. are frequent: 'aqbiir corresponds mainly to the singular pattern qabar (e. of nafs "soul"). -at. pI. 'aqribii'. The situation in South Arabian may have been similar. of ragi' "returning". of kafir "unbeliever". Ar. pI. pI. both corresponding to the singular participial theme qabir (e.).g. 'aqarib. 'amrarf" pI. malikatun .g. In South Arabian the series of patterns extended by prefixes is fairly widespread: the consonantal schemes 'qbr. pI. c. gen.g. 'lJrft. pI. pI. pI. pI. . It appears to be formed on the same principle as the external masculine plural.g.g.52. of gund "army".54. q3br. 'anaqid. §§ 12. 'awg3r. pI. of wagr "hill")./acc. In the North-West Semitic area.g. fern. pl.g. of l3bs "dress". Eth.48. 'aqb3r corresponds mainly to the singular qabr (e. qubara'. 'aqb3r (= 'aqbur). (nom. Eth. fern.g.e. pI. pI. pI. 12. The themes qabarat and qubarat correspond in Arabic to the participial pattern qabir in the singular: e. 'qbrwadmit of the existence of all the patterns mentioned above.g. (nom. 'aqbira'. 12. Feminine Plural 12. Ar. bi7:uir. of sansal "chain". malikat. pI. malikatin (of. nom./ace. b3rakot. pI. Ar. 'agb3rt. In Arabic all these themes.g. "illness"). of ba"!Jr "sea". 'aqbirat.g. root nzl. of 'arus "bride". 'aqbiir often corresponds to singular themes qabr. 'ara'is. 'aqb3rt 12. sg.51. 'umara'. sarratum. pI. § 8. 'aqbur. qibr. Ja"!Jqul. tababt. g) Four-consonant themes: These are formed on the patterns C1 aC zaC aiC4 and C1 aC 2aC aiC4. 'a'miid. Ar. 'aqbur. pI. except the last two. 'kbrw. of sa'ir "poet"). ~a"!Jaft. Ar. of 'unqud "bunch") including those formed from triconsonantal roots by the attachment of an affix (e. pI. Of these themes. qubr: e. of 'amir "Emir". gunild. qubur. pI. Syriac has -at in the construct and emphatic states. but at times also to qabir (e. mala' 3kt. b3raka "blessing". 'aqbur frequently corresponds to singulars with two short vowels or with a single vowel (e. These are frequent in Arabic (e. of 'aqrab "scorpion". pI.53. pI. the same applies to the morphemes -m or -n (of. correlates mainly to a singular theme qabr (e. qabar (e. a plural for quantities not exceeding ten. sg. of gurab "crow"). rugga'. of dawal "district"). 'ahgur. plurals of lJrf "year".83]) presents the feminine plural suffix -ot: e. of k3Siid "neck". these are attached to the feminine suffix.g. d) Patterns marked by doubling: Arabic presents qubbar and qubbiir. although their vocalization cannot be established (cf. In Ethiopic we have 'aqbiir. sg. of ~a"!Jafi "writer") as well as to the pattern qabir (e.g. though we remain ignorant of the vocalization: e. fern. f) Patterns extended by suffixes: qibran.) malikatun .) sarratum.e. sarratim.49.g. pI. of gar "neighbour"). 12. by the lengthening of the vowel-contrasting with the short vowel of the singular:-(a)t. e. kafarat. qubara' corresponds most often to a singular theme qabir. su'ara'. 'albas. pI. of "!Jaql "field"). pI. 12. lJrwf and lJryf. for example 'byt. root l' k plus suffix -t) . The feminine plural is of the external type. of katib "scribe"). gen. i. 'aqb3rt (= 'aqburat or 'aqbirat): of these patterns. In the languages which retain case-endings. sg. 'adwal. qabarat. 'aqbira' corresponds principally to the singular pattern qabir (e. of malik "king"). pI. one of the plural patterns of lJrf "year". qubarat. pI. 'aqburat. of qarib "kinsman"). giran. but in the absolute state -an. on declension §§ 12. and in Arabic malikun "king" (nom. Ar. Eth. qubran.50. pI. pI. 'agribat. pl. 'aqbirat correlates to singular themes with long vowel following the second radical (e. of kbr "great"). fern. qibarat.g. pl. pI. 'aqbur corresponds frequently to the singular theme qabr (e. of hagar "city". 12. fawaris. sanas3l. of mal'ak "messenger". e) Patterns extended by prefixes: 'aqbiir.' They correspond to fourconsonant singular patterns (e. k3saw3d. of manzil "dwelling".). of byt "house". i.g. In Ethiopic the situation is similar: the internal plural pattern qabart corresponds to the singular participial theme qabiiri (e.g. qibiir and qubur correspond in Arabic mainly to the monosyllabic singulars qabr. pI. of faris "horseman". manazil. 'aqbirat. of mararf. they also correlate to triconsonantal singular patterns with long vowel (the plural theme being characterized by the insertion of ' or w or y): e.g. kuttab. Thus we find in Akkadian sal'rum "king" (nom.90 Morphology The Noun 91 qibiir. of tabib "wise").g.

pI.61.g. kabutu "dung". pI. This phenomenon. is rather more widespread: it occurs particularly in nouns which are of feminine gender but lack the feminine morpheme in the singular (e.g. 12.). An interesting feature in Akkadian is the use of the dual as a plural of paucity: e.C. si/ah and sa/awat. 'aton "she-ass". -emi (the latter by contraction of original -ay followed by mimation). qeset "bow". rab "great". isatu "fire"./acc. 223) reconstructs the dual endings as nom. I.59.adly. Syr. in some of the Semitic languages. pI. pI. It is possible (Brockelmann. Dual 12. bisan (this 'innovation' in the Aramaic area is attested" as early as the eighth century B. mf or mfmf. gen. abbu.adane and Amharic 'andand "some" (from ly.92 Morphology The Noun 93 probably by analogy with the masculine plural -in: e. In North-West Semitic. followed by nunation or mimation. In West Semitic there are also traces of a plural formation by reduplicating the singular of biconsonantal nouns (and then adding the external plural ending): e. 'l'U. sama "name". Syr. . -ay.g. developed to varying extents in individual languages (esp.56. Akk. and Arabic suggests that the restricted employment in other languages is secondary (cf.arranu "road". 'asawwata) and a number of other substantives. the appearance of feminine endings in the plural of nouns of masculine singular.g. the unvocalized Ugaritic texts reveal no formal distinction between dual and plural: Gordon (UM. 439) that this type of plural arose from a distributive context: of. Heb. *rabrabe > rawrabe. Nunation is dropped in the more recent period. Plurals of Biconsonantal Nouns 12.g. ikkiiratu. cstr. a.ad and 'and. similarly in some nouns of trade or occupation (e.g. it is rare in other languages (e. "one").arranatu.g. isatatu. which do not normally occur in Akkadian and Ethiopic (of. GVG. This phenomenon is not uncommon in Ethiopic (e. Ugaritic. § 12. in fact. -ii. though fairly frequent in North-West Semitic. this consonant is mostly h: e. qasatot). been extended to masculine nouns as well (cf. pI. pI. pI. gannatii "garden". These formations.58. 43. Akk. 'ama "maid-servant".55. ly. pI.g. In several of the Semitic languages we encounter instances of nouns of feminine form in the singular and of masc. ly.60. Heb. to indicate duality outside these narrow limits.aqliita). pI. pI. and in Middle Akkadian -en/-in predominate over -an. 'amahOt. sanat "year". Akk. kabu). and Tigre kalkal' at "every two". abu "father". but here we have nunation instead of 12. i$$u. pI. barakat "blessing". Ar. biSiita. Heb. has assumed somewhat larger prop6rtions in Ethiopic where the original feminine ending -at has been extended to masculine nouns over a fairly wide range (§ 12.41). Similar cases occur in Hebrew with nouns which were perhaps originally monoconsonantal. Msa "bad" (fem. pI. The position is very similar in Aramaic. emph. 12.57. pI. 'asya "doctor". The converse of this phenomenon. later on it appears as the established form in Biblical Aramaic). Heb. pp. 12. pI. i$u "tree". sg.irruna (also ly. however Akk. are rare in Arabic.. p. The dual is used for the linguistic expression of natural pairs. Ar.aqlii "field". here it may.g. Its extensive use in Old Akkadian. bisat. pI. form in the plural: e.ars: sanot). ly. -ami. The distinction between the cases is gradually lost. ly./acc. but it also serves. Syr. pI. lJ. gen. pI.g. sanim (but in the construct state a feminine morpheme reappe. abs. ganne. pI. barakatiit).e. sanawat. e.arrat "stony ground". be accounted for by the fact that the plural ending -at is no longer limited to the feminine and has. ikkiiru "peasant". Syr. At times the feminine ending of the plural is superimposed upon that of the singular-instead of being substituted for it. ubiiniisu "his fingers". sinniisu "his teeth" .41). Syr. d. In Akkadian some biconsonantal nouns double the second consonant in the plural: e. in the Arpad inscriptions where -n occurs alongside the more common -t.arrat). i. In West Semitic many biconsonantal nouns form their plurals by adding a third consonant to the singular pattern. ESA 'l "god". however. 12. A comparative examination proposes the following Proto-Semitic endings: nom. The Akkadian dual has -an for the nominative and (*-ayn » -en/-in for the genitive-accusative. abs. samahe. pI. cstr. sa/at "lip". pI. respectively. such as mayim "water". Syr. pI. in Akkadian). sana "year". ly. Hebrew exhibits a restricted use of the dual (at any rate during the historically attested period) with the ending -ayim predominating and extending to all the cases. 'atonot. below).

. it remains throughout the entire period of Akkadian (e. The material at our disposal seems to favour this latter view.2). For the plural and dual endings of these cases see the preceding paragraphs. As a dative this occurs in only the most ancient phase of the language (e. the element s which characterizes the dative of the pronouns: § 13.68. 12.g. however. Only in the course of time do these distinctions become progressively blurred: in Neo-BabyIonian and Neo-Assyrian the three case-morphemes are used indiscriminately or even omitted altogether. istu[m] "from". 5.63. The Semitic languages originally possessed three basic cases: nominative (subject). 12. damqis "well". umisam "daily"). -ayni. -is assumes either terminative or distributive function (e. 465) considers adverbs such as taJ. In the later languages the endings disappear and with them the formal distinction between the cases. and the hypothesis has been advanced (though with doubtful justification) that the change took place by way of vowel dissimilation (Brockelmann. Babela "towards Babylon") is regarded by some scholars as such a survival. etc. tabi.). lii:ZU "above" and kanW_ "gratuitously". Eth. gen. but this remains highly conjectural. ly. and in Syriac the dual appears to occur in only two words (taren [tarten] "two" and maten "two hundred"). gen. if we append to them the endings of the singular the following picture emerges: Singular Nominative Genitive Accusative 12. amaris "to see").tu "below" and qablu "before". South Arabian has the dual morphemes -an and -ayn. Amorite and the Tell Amarna glosses retain the Proto-Semitic case-endings-and so does Ugaritic. accusative (complement governed by a verb). 8788) its length is a late and secondary development. In the North-West Semitic area. genitive (complement governed by a noun). according to von Soden (GAG. Finally. ina libbu f'in the midst of.3).ta "below". etc. To the three basic cases in Proto-Semitic we might have to add a locative in -u which is attested in Akkadian and traces of which may perhaps be detected in other languages as well-esp. muatis "to die".. taba. for -is in comparisons note ilis "like a god".66. in Aramaic. -f. 144-45) the vowel of the locative ending was originally long.65) is often used with prepositions (e.g. -ani. p. also as survivals. balu[m] "without". According to Gelb (OA. As an adverbial.94 Morphology The Noun 95 mimation: -ayn.g.g. Brockelmann (GVG.g. In conjunction with the ending -am (more rarely -um). . p. 12. madis "much". § 12. I.g. I. Declension in adverbs such as Ar. n). The Akkadian locative in -u (cf.62. annisam "hither". 459) the singular endings are quantitatively ambivalent (-11. Arabic presents the Proto-Semitic endings followed by -ni: nom. Some of these examples may be open to doubt. pp.456). GVG. p./acc. the dative-adverbial in -is (cf. 12. i. 12. Classical Arabic (as indeed the pre-classical language-so far as we can judge) retains the Proto-Semitic declension system Plural -u Dual -a -ay -u -I. bara "outside".represented by the ending (*ay » e: kal'e "two" (0£. § 14. for the ending -m o£. the ending of the oblique cases predominates over that of the nominative-just as it does in the other Semitic languages. Ethiopic preserves only a few traces of the dual. They also raise the problem of the quantity of the locative ending -u (on this cf. but since their opposition to the plural morphemes depends on a quantitative distinction they must be regarded as short. -(1).71). as may be seen in nouns whose final consonant is ' (vocalized with a. tabu "good". pp.64.65. Babel "Babylon". used adverbially. but their employment appears to be quite indiscriminate in relation to the various cases. 'ade "hands" (before suffixes).67. taJ.. within") or appears joined to prepositions (e. the following paragraph). acc. Akkadian retains the basic case-endings in their entirety: nom. 12.aqWe "loins". In Arabic dialects . § 12. leaving only a few faint traces: in Hebrew the ending -a denoting motion towards a place (e. It has been suggested that -ni is a secondary derivative of -na (the suffix having been added to the plural ending). Akkadian has a fifth case. I. -a } -i According to Brockelmann (GVG.

12. Fleisch. In any event. 80. 318. In modern Arabic dialects caseendings have disappeared altogether-just as they have done in the other Semitic languages.) as well as in some Egyptian transcriptions. the ending -n (nunation) in the dual. kattun.g. kuttun. 12. to some verbal forms (e. Proper names are either indeclinable or form an accusative by the attachment of stressed -ha (e. and constr. It would then have spread to other adjectival themes (qabran U ) and internal plurals ('aqbira'. (Fem.29).g.) (Fem. qubara'. Brockelmann. UM. acc. and neither mimation nor nunation in the masculine plural: 12. p. but the relationship of this phenomenon with the problem under examination remains somewhat uncertain. ace.g.) (Fem.274).g. Ethiopic retains only one oblique case ending (indicating also motion towards a place): e. OA. EG. Definiteness and Indefiniteness It can thus be seen that mimation and nunation co-exist in Akkadian. and others) as also to a series of proper nouns with the feminine ending (e.) sarril sarratum sarratim Dual (Masc. In the North-West Semitic area mimation is found in the majority of Amorite proper names (of the type $aduqum.18) and demonstratives (anm7tun. and constr. Gordon's hypothetical vocalization produces (UM. Ugaritic shows neither mimation nor nunation in the singular or in the feminine plural. In the Tell Amarna glosses we encounter only a few traces of this phenomenon. Yapla!Jum etc.69. in the case of the latter there may possibly be some connexion with the indication of nearness.) Dual (Fem. The ending -a appears both in the singular and in the plural of nouns terminating in a consonant: e. "Yezid") and to some names of foreign origin (e.) Nom. For a possible occurrence of diptotes in Ugaritic see Gordon. GAG. It should be observed that in Old Babylonian nunation occurs instead of mimation in some personal pronouns (yattun. p. maliku "king". 269-80).70.) tabilma tabima tabatu tabati tabami tabemi tab (a)tami tab(a)temi 7 Moscati.71. 'agb3rta. but they do not possess the function of distinguishing definiteness and indefiniteness-as is the case in some other languages. pp.]). It will. though other explanations have been proposed for this element-of. 'Ibrahimu "Abraham"). The ending -a of the construct state has possibly arisen from an analogical extension of the accusative morpheme. § 142. 'agbewt "servants".aqha "Isaac" [acc. nom.) (Masc. whether definite or indefinite. gabr "servant". It is not possible to reconstruct Proto-Semitic forms for the expression of definiteness or indefiniteness. Wehr's studies on the nominal pattern 'aqbarU conjecture that this diptote declension originated within that theme. Singular (Masc. nom. Makkat U "Mecca").adil "one" (thus Dillmann. therefore. p. C1 aC 2aC aiC 4u. In Akkadian all nouns. malika• There exists. C1 aC 2aC aiC 4u. but it presents endings with -m in the dual and in the masculine plural. pp. TPA. GVG. 223) the following picture: Plural (Masc.g. Plural (Masc. be well to examine first the various languages separately and then to consider what common features emerge from this investigation. 145). Gen. 'aly. abu "father" and a!Ju "brother" when denoting gods (von Soden. yztttun. however. 45. gen.) (Fem.g. p. suttun: of. 6. The final -il of certain numerals may be a survival of the nominative ending: e.72. annatun: of. gabra.) sarran sarren sarratan sarraten sarrum sarrim sarram sarratum sarratim sarratam } sarri 12. Acc. II. p. Gelb./Acc. have the ending -m (mimation) in the masculine singular and in the feminine singular and plural. maliki. mimation and nunation fall into disuse from the end of the Old Babylonian and Old Assyrian periods. § 13. Comparative Grammar . 43-44. a series of nouns whose declension is limited to two endings only: -u for the nominative and -a for the genitive/ accusative (cf. acc.g. Y38ly.) Nom. Yazid u . Gen. Noteworthy is also the absence of mimation in some proper names in Old Akkadian as well as in some common nouns used as proper names: e.96 Morphology The Noun 97 in its entirety: nom.g. § 13. It is noteworthy that diptotes do not take nunation.

from the most ancient Aramaic inscriptions onwards. Garbini. have been detected in Old Akkadian. Old Assyrian. in late Punic.as a definite article.. however. m3lakim "kings". related to a differentiation between the definite and the indefinite. pumma "mouth" may. pp. In the dual and the external masculine plural we encounter unmimated forms only.is due to a general phonetic change in that area (cf. can be explained as nominal patterns with suffixes. especially as the syntactical function of mimation appears to be negligible.75. ly. of the article h. instead.to ' . Traces of mimation have been seen by some scholars in nominal forms such as qa8tam "bow" (alongside qast). Traces of mimation have been seen in forms such as darom "south". From these elements of frozen suffixes. but they may well be secondary-or.55). for the element -an of the masculine external plural can scarcely be regarded as a form of nunation. 33. § 8. 12. etc. In the plural (there is no dual) nunation occurs in the ending -in. In preclassical North Arabic -m occurs occasionally in Tamudic to indicate indefiniteness.followed by doubling (or reinforcement) of the initial consonant of the noun (e. Yet. ham-melek "the king". whereas more recent epigraphic documents have brought to light two instances of (')l-. 12. ?lmn "this statue" or "the statue". The absolute state occurs either with or without mimation. as indeed in North-East Semitic. etc.77. and aspects of determination. the ending -n is attached to the singular. etc. preceding paragraph). 423-30. p. but-as in Hebrewit is not connected with the differentiation between definite and indefinite. Amharic has developed a type of suffix-article. but this opinion does not seem to be well substantiated. dabru "the mountain". and external feminine plural (it is not attested in external masc.76. Torczyner's studies). Ethiopic possesses neither mimation nor· nunation. §§ 172-73). 426. on one hand. Some formations in -n are also found. In the Arabic area. while definiteness is expressed by the prefixed article h-. 118-21). ly. e. Mimation occurs in the singular. The evidence adduced in the foregoing confirms the view that it is impossible to identify any Proto-Semitic means of ex7* . The position in Phoenician resembles that in Hebrew-at least as far as can be judged from the purely consonantal script. classical Arabic has -n for the expression of indefiniteness and the prefixed element 'al. Finally. be regarded as preserving traces of mimation. Praetorius. on the other. p.98 Morphology The Noun 99 In North-"West Semitic. This mimation is not. Definiteness is formally expressed. ham-m3lakim "the kings"). Syriac (and Aramaic in general) agrees with Hebrew in possessing neither mimation nor nunation in the singular. but a suffix-substitute has been evolved from the pronominal suffixes (cf.inncim "gratuitously".g. and particularly in such proper names as Milkom and the series of adverbs (ending in -am) which includes 'omnam "truly". (cf. 12. Remnants of a nominal ending -a. such as Gid' on.g. It has the function either of a demonstrative or of a definite article. melek "king". the interpretation of these data remains uncertain. the internal plural. at any rate. § 38.73. Dillmann. by a prefixed article ha. Ethiopic has no prefix-article. by a suffixed article -a (emphatic state of the noun) which in the Eastern Aramaic dialects (and in Syriac in particular) loses this specific function and becomes the normal ending of all nouns.74. On the other hand. 12. no longer dependent on any antecedents. Mimation is a prominent feature in the masculine plural and in the dual (-im. EG. Moabite differs from the other Canaanite languages and agrees with Aramaic in the appearance of nunation (-n) in the plural. Hebrew possesses neither mimation nor nunation in the singular. In the earliest Lil.artom "soothsayer". and Amorite (cf. Ethiopic possesses very elaborate syntactical means for the periphrasis of the definite article (Dillmann. however. no properly established relationship can be ascertained between mimation and nunation. and the external feminine plural. and the ending -m does not by itself necessarily correspond to the indefinite article ('8 1m "a man" or "the man"). § 172 [b]): b3'3Sihu "the man". in all three numbers. SNO. internal plural. -ayim) which tallies with the state of affairs in Ugaritic (cf. Grammatica Aethiopica. the transition.tyanite inscriptions we still encounter -n to mark definiteness. plurals). Epigraphic South Arabian presents a fairly complex situation (Beeston). definiteness is expressed. pp. The factors governing the use of mimated or unmimated forms remain obscure. 12. Nominal forms such as 'imama "day". though not entirely agreeing in the specifically determinative function. EG.

12. § 1. but cstr. uzun. mtiru "son". ham-mJlfikim "the kings"). mara-su). qa??ab un "a butcher". It may further be assumed that mimation developed into nunation in some languages. von Soden. though there are certain exceptions to this rule.g.17) or involve modifications in vowel quantity owing to the displacement of the stress (e. e. uznu "ear". Heb. For details the grammars of the various languages have to be consulted. and that this usage is reflected in the most ancient Semitic linguistic material.2. gen. such as Arabic and Aramaic.e. yonat: cf. cf.69: extension by analogy The unity of the noun with the following "genitive" (with the principal stress on the latter) and the consequent reduction in the case-endings may occasion changes in the form of the noun in the construct state. in South Arabian: nisI wqbr HNTSIR "HNTSIR's monument and tomb" (classical Arabic would have changed the construction to "Hntsr's monument and his tomb"). Akk. The case-endings are retained in their entirety in Arabic. and plural seem to behave differently. Heb. 'al-qa??ab u "the butcher". n. new and special means of indicating definiteness make their appearance in a number of different guises: the prefixes h. 163-71). and -u in Ethiopic. i. The two nouns cannot ordinarily be separated. daMr "word". 12.e. This happens in the singular in classical Arabic (where the article excludes nunation: e. dual. in the course of the historical development of the Semitic languages. Heb. qa??ab un "a butcher". A noun in the construct state has neither mimation (nunation) nor the article (e.33). Akk. i.in Hebrew (and also in some pre-Islamic Arabic dialects) and 'al. In many cases the original Semitic form can still be recognized in the construct state: so in the example just cited. 'ar-ragul U l-J:w. Ar.g. In this respect singular.. The Semitic "construct state" is closely connected with the function of definiteness or indefiniteness.78.100 Morphology The Noun 101 pressing definiteness or indefiniteness. while in Hebrew. In these conditions the nomen regens merges with the nomen rectum in a single complex whose principal stress falls on the rectum. Ar. In Ethiopic the ending -a.sanu l-waghi literally "the man handsome of face". maru-su "his son". damqam inim literally "good of eye". qa??ab u . this replaces mimation or nunation which now become an indication of indefiniteness: e.g.in Arabic (a connexion between these may be seen both in the alternation h:' and in the doubling of the . § 12. in Aramaic where the endings of the status emphaticus have extended their use at the expense of the forms with nunation (Syr. the "genitival" element. but in general the autonomous genitive-ending contrasts with a case-element common to nominative and accusative. where mimation appears in the plural only. cstr. -n in South Arabian. . 'ebed "servant". mari-su. it is semantically irrelevant and is retained together with the article (e. § 12.g. by way of hypothesis. It may then be shown how. cstr.g. Where definiteness is expressed by a prefix it may bring about a modification in the use of mimation or nunation. but it may be said that these changes are often either of an anaptyctic type (e. On this basis it may be suggested. cstr.g. mJlakim "kings". pp. that originally there existed a mimation of nouns independent of any semantic function as regards definiteness or indefiniteness. acc. so that observations valid for the singular cannot always be applied to the other numbers in which mimation or nunation are sometimes preserved after having been dropped in the singular.79. JNES 19 [1960]. § 9. *'abd (nominal pattern qabr: cf. but before a suffix maru-su "his son".g. yona "dove". it may become a means of indicating indefiniteness by virtue of the contrast to the definiteness connoted by the article.. *'abd: cf.g. dJbar owing to the shift of the principal stress to the following nomen rectum). characteristic of the accusative. Heb. In addition to the construct state there is that . but 'al-qa??ab u "the butcher"). In Akkadian and Aramaic three "states" of the noun may be distinguished..following consonant in Hebrew and Arabic-in the latter as a substitute for the total assimilation of l to certain consonants). serves also for the construct state (cf. ") except in certain instances of what is termed "improper annexation" (Akk. the suffixes -a in Aramaic.. Where definiteness is expressed by a suffix. In Akkadian their retention is less complete: they appear in some nouns ending in a vowel and before suffixes (e.g. bise prevailing over biSin). following upon the change m >n which occurs also in other connexions in these languages.this is the special form taken by a noun when it is defined by a following genitive (or pronominal suffix).5) and in the Hebrew and Syriac feminine-ending (e. "the butcher of .

f. In North-West Semitic there is a distinctiun in the vowel (-ki). ly. isten "one". 'nk in Moabite. katun(u). 3 m. § 12. 1 2 m. § 13. § 13. 3 m. Brockelmann. sati/u su/iati. in some numerals (e. Syr. I 2 m. GAG p. in which Akkadian exhibits mimation and Syria? the ending -a (e. In the later dialects there also occur extensive fluctuations between the genitive/accusative and dative forms. These endmgs do not. and for the dative. the additional element -k appears in North-East and North-West Semitic (cf. too. pp. § 12. hyt. masculine or feminine. Akk. f. 3 m. the "absolute" state is devoid of any endmgs. it possesses a form of its own.rdly.4. Syr. 2 m.1. 3 f. Akk. adverbial contexts (e. pI. but it is not connected in these languages with any distinction between the cases. 13. The Pronoun 1. 'anoki 'ena ' atta 'aU 'att 'att hu hU hi hi ('a)na~nu ('ena)~nan 13. Independent Personal Pronouns 13. It IS comparatIvely rare and occurs mainly in distributive expressions (e. Akk. Syr.g. To pass now to a consideration of individual forms. 3 m. f. sasun(u). 2 3 [hm] 13. SG. kiisun(u) . and as predicates (of. c) Old Assyrian uses the genitive/accusative forms for the dative as well.um "brother". cf.9). Syriac Arabic 'ana 'anta 'anti huwa hiya na~nu Ethiopic 'ana 'anta 'anti wa'atu YJ'ati ni)~na anaku alta atti su Si tin(k) tit tit hw hy 'ani. satun(u).d to the forms of the suffix pronouns (cf. shows variant forms for the genitive/accusative. A few general remarks on this table: a) The first and second persons singular and plural belong to the same system ('an. Thi.66. and for the second person singular. suatina. The independent personal pronouns of the principal Semitic languages are as follows: Akkadian Ugaritic Hebrew Sg. f. in later dialects new forms may be observed: for the genitive/accusative.14): 13. s PI.g. f. suatunu. pI. however. satina. e) The suffixed element -t in the third person also occurs in Phoenician (hmt). 'alyii). b) The Akkadian series is used for the nominative onlv while the other cases employ considerably different elements relat.ad) . ' Genitive / A ccusati ve Sg. ana dar "for ever". sati niati kunuti [kinati] sunuti sinati Pl.102 Morphology The Pronoun 103 Dative yasi(m) kiisim kiisi(m) suasim. f. Akk.7.2.g. 104-105). I 2 m.6. ina kar kar-ma "in every colony". hmt).plus suffixes). 3 m. but contrast ana which is attested as a secondary form in Old Babylonian). 13.5. 3 f.32). O. 1 2 m. Syr. 79. . but they are confined to the third person: the characteristic element in these forms is a suffixed -t (hwt. ninu attunu attina sunu Sina hm 'attem 'att!3n(a) h!3m(ma) hrm(na) ' atton 'atten hennon hennen 'antumUi) 'antunna hum(u) hunna 'antuma huma 'antammu 'antan 'Jm!tntu 'amantu Du. al. rappin rappin "in flocks").3. 3 m. 'amuntu. 2 m. For a connexion between the element characteristic of the dative forms and the ending -is of the dative-adverbial case of the noun. and in Ethiopic (wa'atu. while the third person is formed from elements related to the demonstratives (cf. f. sasu/i(m) su/iiiSim. ya'ati. von Soden.8asi(m) niiiSim kunusi(m) [kinasi(m}] sunusi(m) [sinasi(m)] called rectus or "emphatic". 'amiintU) .g. f. as a variant form in South Arabian demonstratives (cf. also a-nu-ki in the Tell Amarna glosses. d) Ugaritic. retain any significance in terms of defi~ing ~h~ noun. we may posit a Proto-Semitic 'ana(ku) for the first person singular. ba-'gal "in haste"). sasina. ku(w)ati. ''lik[y] in both Phoenician and Old Aramaic. yati kati/a kati suati/u. 13.

South Arabian keeps the original demonstrative values): Sabaean Minaean Qatabanian SlW.tri which has a masculine he and a feminine se) suggests Proto-Semitic huwa for the masculine and siya for the feminine. As to the third person plural it will be observed that a) for the consonantal element the same holds true as in the case of the third person singular (cf. 'nty in Egyptian Aramaic and 'ant. etc. SOL. In Egyptian. 61-65). where such a vowel appears. 3 m. It is improbable that Proto-Semitic had an initial vowel of a timbre (. there later appears nini).8. . this may again be due to analogy with the vowel of the suffix pronoun of the same person-as may be seen (except in Arabic) from the tables (cf.9. The South Arabian thirdperson pronoun has the forms given in the following table (semantically. For the second person plural we may posit the ProtoSemitic forms 'antumu. hwt h'. alongside the form ninu. For the second person singular we may postulate the Proto-Semitic forms 'anta. hy'. despite some instances in which the retention of the vowel is attested by Masoretic Hebrew for the second as well as the third person plural ('att~na. GVG. suwa.a-naly. 13. hw'. noteworthy are: a) the frequent assimilation of n (the form alta of the Tell Amarna glosses confirms the extension of this assimilation to a great part of the North Semitic area.8). hmt slmt b) It appears that the series with h and that with s are both of Proto-Semitic origin. § 10. 13. §§ 13. As to the third person singular: a) Akkadian uses for this pronoun a consonantal element different from that encountered elsewhere (s instead of h). SOL.14. too. sw. As for the variants in the final vowel. si' a to huwa. taking the a as due to the influence of the following pharyngal).antina: a) the distinctive vowel (u. 3 m.1. hi'a. in accordance with the phonetic laws of that region (cf. 'anti (Kienast's recent reconstruction offering as Proto-Semitic forms ka. For the first person plural we may propose the ProtoSemitic form naJ:tnu (or even nily. c) Some (cf. p. on the basis of Akkadian. The Dead Sea documents bear witness to a situation when this phenomenon was even more widespread (spellings 'tmh. hina and a parallel one sumu. Sil' a. A reconstruction (supported by the Modern South Arabian dialect of Mel. 13. b) taking into account the arguments adduced with regard to the singular. 'nt. b) the dropping of final unstressed vowels in the North-West Semitic area corresponds to the phonetic laws of that region (cf. and it is equally possible to envisage the formation of the third person pronouns from either series. in Egyptian: masc. and this element is to be found (cf. But this reconstruction remains highly conjectural. § 13. we may postulate. by way of reconstruction of the Proto-Semitic forms. f.. in Akkadian n prevails over m.followed by the process 'uw > wu > wa and 'iy > yi > ya-and subsequent attachment of final -til. in Akkadian. hemma. litH. henna). slmt slmyt languages as due to analogy working in opposite directions. I. undergoes gemination). a series humu. i) and consonant (m. but it is easier to explain the loss of intervocalic wand y tllan their secondary insertion (though even this is possible by assimilation to the preceding vowel). one should bear in mind the existence of considerable fluctuations and the difficulty of determining Proto-Semitic vowel-quantity with any degree of certainty (cf. hyt hmw. Du. n) of these forms are subject to the effects of analogy (the consonant. in Arabic u over i. h'. c) in very much the same way . 13.104 Morphology The Pronoun 105 perhaps by analogy with the suffix-pronoun of the same person (-ni). PI. Gray. it may owe its origin to analogy with the singular form ('ana). slwt slyt slm.10. this would explain the position in the various 13. 296313. 'tnh). ti raises several difficulties). Gray. ya'ati are conceivably due to the omission of the initial h. hiya.nu). fem. Sg. sina. -ti. pp. Brockelmann. § 10. sy) in the South Arabian dialects as well-with the exception of Sabaean.8)-with the exception of the second person masculine in Hebrew where the final vowel remains (important survivals are also the Hebrew spelling '''atti'' and the Syriac form with -y in the feminine). b) the dropping of unstressed final vowels in the NorthWest Semitic area.nu or 'a-nily. As regards the quantity of the final vowel in 'ana as well as in other pronouns. d) The Ethiopic forms wa'atu. 62) prefer ProtoSemitic hu'a (hu'a is now found in the Dead Sea Manuscripts). in some cases. siya.9).12. 'anti in Biblical Aramaic are probably historical spellings). the element -k is attested (ink).nu. pp.

of. f. PI. §§ 16.3): Accusative Dative -a(m). save for a few remnants in some Old Aramaic documents (hmw). f. very different and correspond closely to those of the independent personal pronoun for the accusative/genitive and dative. analogy affects both the first vowel (u. n): again n prevails over m in Akkadian. f. § 13. -nt -lea -ki -s( n) -s(i) -niati -knnilti -kinati -sunuti -sinati Sg. for the reappearance of Proto-Semitic endings before the suffixes of. 3m. 'ennen in Syriac and the Ethiopic singular forms. c) Furthermore. f.139-42. f.14) to the masculine singular base. Some general remarks on the above table: a) The Akkadian series is used for the genitive. etc. Personal Pronoun Suffixes 13.13. ''Jmantil may possibly be explained as due to the disappearance of initial h. the element -til has been added (as in the masculine singular). 98). p. formed by the attachment of the suffix pronouns (cf. § 13. The data available are. 1 2m.4). 3 m. in later dialects. § 13. the enclitic forms 'ennon. 1 2 m. of course. (for instances of gemination of the consonant of. e) the Ethiopic forms ''Jmilntil.14. 13. 3m.1 2 3 -ya. 1 2 m.106 Morphology The Pronoun 107 as has been shown for the second person plural (cf. extremely limited. b) Old Assyrian uses for the accusative the forms of the genitive (§ 13. § 13. huma. Du. sina (of. SNO. -i -nt -ka -ki -s(u) -s(a) -nt -leunu -kina -sunu -sina (-y) -n -k -k -h -h -n -lem -len -hm -hn -ny -km -hm -i -ni -ka -k chil. -i -n -ni -k -ka -k -ki -(h)i. preceding paragraph).15. must be considered a secondary development. its principal forms are as follows (Assyrian in brackets): . -0 -(h )a. the Akkadian developments suatunu. for the most part. -ni(m) -ku(m) -ki(m) -su(m) -si(m) -niasi(m) -kunusi(m) -kinasi(m) -snnusi(m) -sinasi(m) 13.18. In some languages the consonant n is inserted before the suffixes (cf. 2. The suffixed personal pronouns in the principal Semitic languages are as follows: Akkadian Ugaritic Hebrew Syriac Arabic Ethiopic -ya.16. satina alongside sunu. For the dual one may assume as Proto-Semitic the Arabic forms 'antuma. u over i in Arabic. i) and the following consonant (m.14) and for the dative those of the accusative. f. the forms of the accusative and dative are no longer kept apart and are used fairly indiscrIminately. § 13. 13.17. 1 (noun) (verb) 2m. Garbini.11). PI. thus recalling. -h -hu -h -ha -n -leon -ken -han -hen -na -leum(1L) -kunna -hum(1L) -hunna -kuma -huma -ya -ni -ka -lei -hil. Akkadian employs an independent possessive pronoun which is based on the endings of the suffixes. -ah -nil -kem -ken -(helm -(he)n 13. while for the accusative and dative the following forms appear (they are. Ethiopic makes use also of the variant forms w'J''JtOmil (masculine) and w'J''Jton (feminine). 13. 3 m. f.9 d). The attachment of the suffixes may be effected by means of connecting (or glide-) vowels or by way of contraction (for details see the grammars of the various languages). -0 -(h)a -na -bmmil -bn -(h)omil -(h)on Sg. in accordance with the phonetic laws of that region. f.(cf. it is infrequent in Old Akkadian and. for want of Semitic parallels. d) final unstressed vowels are dropped in the NorthWest Semitic area. in part.

7) to the uncertainty about the quantity of the vowel element of these suffixes (as well as of the others-except. lalika. for the object (laliya.niim 7cuniim sumiim ni' atum. As to the feminine form. 13. § 13. on the other hand. The predominance of -i in Old Akkadian and in Amorite favours the latter hypothesis. 3 b) The same considerations apply to the Proto-Semitic suffixpronoun as were formulated with regard to the independent pronoun (cf. the change -i > ya in certain languages is due to analogy with the suffix of the second person singular masculine -ka. 1 2 3 PI.g. 13. PI. etc. the loss of the final vowel is already encountered in the form -k of Amorite and the Tell Amarna glosses. we may propose. -ha (corroborated also by the Amorite documentation) and -su.]. etc. 2I!asc. 3 m. -sa (the vowel is not invariably long)-unless it be conjectured that -hu was originally the masculine and -sa the feminine form and that .21. etc.24.9): two series can be established. plur. the optional omission of the final vowel in the suffixes of the Akkadian third person singular).).for the feminine singular. zi'aya "mine" [masc. ''Jnti'aya "mine" [fern. for the first person singular. sing. or. ''Jlli'aya "ours"). etc. yat(t)u.71. £) Ethiopic has also produced independent pronoun forms by adding suffixes to lali-. The suffix attached to verbs is preceded by the consonantal element n. a Proto-Semitic form -ya or -i. probably. sing. The same occurs (again in consonance with the independent pronoun forms) in the South Arabian dialects-with the exception of Sabaean: 13.34-39). at times connected by the particle l "to": thus in post-Biblical Hebrew se-l (e. f. 13. 1 2 3 yii'um. -Sl yw -slm [-sIn] -Slmy -h(w) -h -hm(w) -hn -hmy Du. ya'iitum. Turning to a consideration of individual forms. whereas Hebrew. 'iyyaka. In Syriac the final -i of the suffix is written but is not pronounced. 3 m. e) An independent possessive pronoun occurs in some languages which is the result of a combination of the suffixpronouns with elements of the relative pronoun (cf. selli "mine". possibly by Sabaean Sg. ''Jlli'a. n'uttum (7cuniitum) (suniitum) (ni'ii'iitum). (kuwa'iitum) Hem.(. for the subject. 13. 'Qt3ka. 17vaX for '?nekii).m/n kattun The Pronoun 109 Sg. for the variant forms with final -n instead of -m (yattun. §§ 13. those of the third person singular where the vowel appears long).).). Morphology Hem. suitun ni'a'um. -SIWW -s1. has in some cases -ki (yet Greek transcriptions point to instances of -k for the masculine: e.108 ]Y[asc.]. yiitum kuttun.23. suttun) see § 12. and the same is true of the final vowel of the second person singular feminine and the third person singular masculine (cf.g.22. dilan "ours"). Hebrew has created an independent pronoun for the direct object by using the element 'ot. literally "which [belongs] to me") and in Syriac di-l (e.iyyaya. Arabic employs 'iyyafor the same purpose (. kattun. w(Jr::X for' Od3kii.g. For the third person singular one should note: a) Akkadian forms the suffixes of this person with the consonantal element s instead of h-just as it does with the independent pronoun. and to kiya-.20. Reference has already been made (§ 13. If the original form was -ya. nuttum (ni'atum) (7cunii' atum) analogy with the suffix of the first person plural.oti. Minaean -SI(W) -S1 -slm -slmn Qatabiinian -SI(W). etc. 13.mti'a. yiim yat(t)um/n kiim (ku'a'um) kattum/n sii(m) sattum. kiyaki. if the original form was -i. -ki. kiyaka. In Ethiopic the independent possessive is formed by adding the suffix-pronoun to zi'a.for the masculine singular.plur. attu-kunu. kiyaya.19. -hu. kuttun.g. laliki. then the -i which appears in certain languages is due to the loss of the final short vowel and the subsequent transformation of the semivowel into the homorganic vowel. yuttun.. '. d) At a later period of Akkadian a new type of possessive occurs which is formed by adding the genitive suffixes to attu (attu-ka. f. For the second person singular we may suggest the ProtoSemitic forms -ka.).for the plural (e.

29. §§ 13. § 13. § 10. and the expanded forms masc. hnh). and u over i in Arabic. both the first vowel (u. 13.m. . § 13. agannetu and agati. The principal forms of the "near" demonstrative pronoun ("this") are as follows: Akkadian Hebrew Phoen. and South Arabian.25. agannutu.27. annu f. fern. etc. m. -(h)on (like the -0 of the third person singular masculine) might be explained by way of contraction of the initial u of the suffix with the final a of the noun or verb.l for the plural which correspond to the Phoenician forms. 'lnft 'all6nUJ. in Arabic u over i. hallf.zu.13. The variations in the vowel in North-East and North-West Semitic (to the Hebrew -nu corresponds -nu in the Tell Amarna glosses) must be considered the result of analogy with the final vowels of the independent pronouns (cf. f.8).36). are mentioned on account of their special interest. 13. Aram. sing.24). in the Hebrew of the Dead Sea documents (cf. the same observations hold good which have already been made for the singular suffix and for the independent pronoun (cf. d) Ugaritic is not listed because the forms of its demonstrative pronoun are rare and doubtful (hn and hnd are probably a type of demonstrative). anniatum > annatu(m). masc. 'gUa(ntu) Du. I r da ana (ha)ila (ha)ilihi. -hina as well as -sumu. annimmutun) cf.26.8). 13. Phoenician.hmh. sing. while again referring to the possibility of a ProtoSemitic masculine form -humu and a feminine -stna. -kina (the observations on the independent pronoun [§ 13. .12). but at times they are retained. annitu z(') PI. for both the second and the third person plural (kmh. § 10. hallaz are related to the Arabic "far" demonstratives (cf. For the third person plural note the following: a) as regards the consonantal element. § 13. c) in consonance with the second person plural (cf. ili. m. anniatun. Biblical Aramaic.110 Morphology The Pronoun III the workings of analogy subsequently acted in opposite directions. m. d) final unstressed vowels are dropped in the North-West Semitic area (cf.30. b) on this basis we may propose for the Proto-Semitic pronoun the forms -humu. ze } z6t Syriac han(a) had(e) Arabic ESA Ethiopic zJ(ntu) za(tti) Sg. f. fern. apart from annum. 'Wen hallen annatu (ha)ilani (ha)tani rUu(ntu). while in Neo-Babylonian a new demonstrative appears: aga. Observations on the above table: a) the Akkadian forms are the result of contraction of more ancient ones: annium > annu(m). while Minaean has 'hlt and Qatabanian g.37).11] are relevant also in this context): a) the vowels (u. § 12. Bibl.26). and fern. e. n) are subject to analogy working in various ways: in Akkadian n prevails over m. For the first person plural we may posit the Proto-Semitic form -na (corroborated by both the Amorite and the Old Aramaic documentation). pI. agatu/i. b) Old Babylonian presents. anniutum > annutu(m). § 13. the indeclinable zu is used as a relative (cf. hallaze. e) the Ethiopic forms -(h)omu. § 13.. The agreement of Arabic and Ugaritic suggests that the forms of the dual are Proto-Semitic. Demonstrative Pronouns 13. etc. the expanded and declinable form anni/ummum. i) and the consonantal elements (m. The first person dual-found in Ugaritic alone among the Semitic languages-occurs also in Egyptian and may thus be regarded as Proto-Semitic. ti (ha)'ula('i) iln ilt 'f!lle 'l '~l(le). b) final unstressed vowels are dropped in the North-West Semitic area (cf.26) and the independent pronoun (cf. e) Hebrew presents the variant forms zo for the feminine and 'f. sing. n) are subject to the workings of analogy: once more n prevails over m in Akkadian. the additional languages adduced in the comparative table. pI. f) the plural form indicated for South Arabian is the Sabaean one.1). i) and the following consonant (m.28. 3.9. For the second person plural we may suggest ProtoSemitic -kumu.tn/w. annutu f. § 13. 13. c) In Old Aramaic a variant form -hi is attested for the masculine.71. -sina. 13. yet some cases of retention are attested in the Dead Sea Manuscripts. c) for some Akkadian variant forms with final -n instead of -m (annutun.g.

sa. sat. tilka 'antaku. etc. si).32.112 Morphology The Pronoun 113 13. m. .33. c) it is very common in the Semitic languages to use the personal pronoun as a demonstrative-or rather to employ the same pronominal element ("anaphoric" pronoun) for both functions: this occurs in Akkadian. it will be possible to identify certain formative elements which make their appearance in various languages. In other languages the "near" demonstrative is formed by adding to rJ the consonantal element n: thus in the Phoenician dialect of Byblos zn. The "far" demonstrative includes. Hebrew. Observations on the above table: a) instead of 1lllu. The most frequent of those elements is the consonant rJ for the singular (taking into account the phonetic changes which this consonant undergoes in the various languages: § 8. the usual forms are se-. In the Aramaic area (in accordance with the phonetic development of the consoI1foscati. plur. 4. but not of the feminine whose characteristic component is t: the series n/t/n appears to be typical of the deictic element in the Hamito-Semitic area. in the majority of the Semitic languages. as well as in the Syriac had(e). plur. However. hallen. acc. It is clear from a comparison of the forms set out in the preceding §§ that they cannot easily be reduced to Proto-Semitic forms. gen. 'allontU and 'allantu (with the additional element -tu). In North-West Semitic. f. South Arabian rJn. either in isolation or combined with each other. 13. The antiquity of the element rJ is demonstrated by its occurrence in Amorite (written "zii") and in Old Aramaic (z').. Instead. ammitu.36. ullu ullitu ullutu ullatu #k. 's.and 'aser (cf. PI. Comparative Grammar 8 . etc. masc. si. while Ugaritic has two forms: d personal and dt impersonal. acc.34.14) to which 'l corresponds in the plural. and more particularly with the consonantal element rJ. antalcti 'allahL and of the plural. In Hebrew the forms connected with the demonstrative element rJ (zu. more frequently. pp. Relative Pronouns han on }'ula'ilca hanen 13. Babylonian period survived). According to Greenberg (JAOS 80 [1960]. su.. in the North-East and North-West Semitic areas there are certain different forms of the relative which are made up of the element S. : Plural Dual nom. Aramaic Syriac Arabic Ethiopic Sg. amm(i)atu. m. The principal forms of the "far" demonstrative pronoun ("that") are as follows: Akkadian Bibl. 13.: This series. gen. the suffixed consonantal element -k (often preceded in Arabic by l). asar in the Tell Amarna glosses and ' sr in Moabite. zatti (with regressive assimilation). dikk~n dak. while the occurrence of dtr in Ugaritic is disputed). f. sa nom. etc. sati sut sat sa m. dikk~n 'ill~k haw hay zaku rJa(li)ka ti/aka. The relative pronouns are connected in the majority of the Semitic languages with the demonstrative ones. b) Ugaritic does not offer any special forms for the "far" demonstrative. Old Aramaic znh (as well as z'). fem. and in the Syriac han(a). Syriac. Amorite has SU (fem. which can be recognized as being with the personal/demonstrative pronoun of is reduced to the single form sa from the Old onwards (only in rare cases have sut and sat formally connected the third person. In Phoenician the use of z (Byblos dialect) is also rare. A component of the "near" demonstrative is ha which appears in conjunction with rJ in the Arabic series harJa.35. In Akkadian we have the following series for the most ancient period: Singular m. and South Arabian. Ethiopic zantU.. Assyrian has ammiu. fem. : f.317 to 321) the element n is characteristic of the masculine singular 13. n combines also with ha in the Akkadian anniu. ze) are rare. 13. the usual form being s or. amm(i)utu. Biblical Aramaic dana.: f. the forms which result from these combinations are generally rJk in the singular and 'lk in the plural. later on 'aser makes room for the shorter form.in.

seems to be secondary from a comparative point of view) and rna for things. Observations on the above table: a) Amorite has manna "who?" and rna "what?".37. gu l-mali "he of the money". i. ". sa 1Jutari "he of the stick". or simply" of . conveying the meaning "that of . f. Interrogative and Indefinite Pronouns 13. Akkadian man-man > mamman produces an indefinite form by reduplication of the interrogative.. The forms of the relative pronouns are used in some Semitic languages for the expression of the so-called "determinative" pronouns. The forms assumed in the various languages are as follows: 14. and ayyumma as an adjective. ulatu rjawa gawata Akkadian Ugaritic fIebrew Phoenician Biblical Aramaic Syriac Arabic South Arabian Ethiopic mannu my mi my man man man mn mannu min1L mh rna m rna ma(n). da-.). common plural 'alla. the rich man. b) Ugaritic also possesses the element 'ay but it is generally employed as an indefinite pronoun ("any")in addition to the element mn(m) (cf. c) Hebrew preserves the element 'ay in its original form in the interrogative adverb 'ayy~ "where?". The interrogative forms are also used in the Semitic languages as indefinite pronouns.1.39..ze f- 'ayna 'ayyun 'y 'ay In the second series the element 'al. etc.41. acc. f. ayyamma etc. gw.. D. the man with the stick. 'ayy has a feminine 'ayyat.38. Cardinals 13. gu. mana rna mant ayyu mn(m) .. § 13. '~ka "how 1" etc. gtm). Ar. feminine singular 'anta. e.e. g (used also for the plural). where the indefinite pronoun is *man-ma > mamma for persons. "Who?" "What?" "Which 1" 13. ulu . in Syriac. gen. £. while it has numerous interrogative adverbs composed of '~. gi. Plural Dual m. The interrogative pronouns appear to go back to the Proto-Semitic elements man in relation to persons (the form my. Old South Arabian has a masculine sing.e. e) in Arabic man is not declinable. peculiar to a large part of North-West Semitic. "he of . gu gatu 'allagi 'allati 'allagina. The element ma may be placed in apposition to nouns (so in Arabic.. ".114 Morphology The Numeral 115 nant g) the most ancient inscriptions show a relative pronoun zi which later becomes di and. 13. 'lw (Qatabanian variants: masc. and a plur. rjawatu.g. d) the Syriac interrogative adjective 'ayna has a feminine 'ayda and a plural 'aylen in which the interrogative element is reinforced by a demonstrative one.. gm. the component ayyum is declinable: ayyum-ma. the following paragraph). The cardinal numerals in the principal Semitic languages are as follows: 8* . below.39): Singular m.and the infix -la-. m. fem.g. Classical Arabic has two series (cf. rjawu. sarru sa mati "the king of the region". 13. Finally.42. *min-ma > mimma for things. though the masculine form frequently takes its place.('~ze "which 1" and "where?". Akk. The Ethiopic relative is in the masculine singular za. yaum an rna "on whatever day") or be used as a reinforcing suffix to pronominal forms (so in Akkadian. The Numeral 1. ": e.). a feminine gt.. 'allii' i 'allati. 'alla'i 'allagani 'allatani . g IS preceded by the article 13. The Arabic "determinative" gu is fully declinable (nom.40. ga) and in this respect differs from the indeclinable demonstrative gao 5. The interrogative adjectives have the element 'ay in common. i.

whereas in the remaining languages this is not the case: Akkadian Ugaritic Hebrew S yriac Arabic 20 30 'asfu 14. GVG. g) Syriac "five" is formed on the analogy of "four" (*lJ3me8 > lJammes on the analogy of 'arba').t Sina tnm Sitta saliisat salas tl1 erbet erbe.ms seMet [sess."and". f. f. 'asar.amsat lJ. fem. Ugaritic kldt. hams a 'a8rata.79). fem. 8 m. 5 m. 14.amJs SfJMstU SfJSSU sab'atu sab'u samiinitu samani tfJs'atu t9S'U 'asartu The Numeral 117 1 m. that in Akkadian. As regards Akkadian. In Ethiopic the 'numerals with the ending t ESA Ethiopic '3sra salasa e8ra salasa 'srm tltm etc. 'esre. and by a singular peculiarity. and Ethiopic the ending of "20" is analogically extended to the others. it has nevertheless recently been disputed by von Soden (WZKM 57 [1961J. b) in Akkadian the numerals from "II" to "19" are not attested except the forms seMser "17" and 8amaneser "18" (masc. f. It is generally held (Brockelmann. 'e8r~. b) "one" fem. 'arba'at U rigal in "four men"). Arabic kilani "both" and Hebrew kil'ayim. 9 m.ammes lJ. s3losa 'asar.u"id 'a7. d) regressive dissimilation occurs in Syriac "two" *t3nen > t3ren. 'asrata. in Ethiopic might possibly be formed by analogy with the feminine personal pronoun y3'.g.adu 'a{i. 'asru wa-salas).3. I. p. c) in Ethiopic the component "ten" precedes that of the units and is joined to it by wa. f. f) Hebrew "five" is formed on the analogy of "six" (*lJamsa > lJi'imissa on the analogy of sissa).) and sinseret "12". this inversion of gender also' operates when the numeral appears without an accompanying noun.ad 7. § 10. 'arba's3re.iim?s SiMii S?S sib 'a seba' sfJmonii sfJmone tis'a tf)sa' 'asara Oeser 'a7.2. that the numbers from "30" to "90" are plurals of those from "3" to "9". istet a{i. while in the Ugaritic cardinals from "two "t0 "ten "the f orms WIt h out -t may b e employed with either . e) Ethiopic "two" is derived from a different root which also appears in Akkadian kilalla!un. gender. South Arabian. South Arabian. Ar. fem. Syr. 2 m. Arabic "15" masc. In Arabic. arba' u arb' (t) lJ. 7 m. h) Akkadian "eight" has initials (samanu) instead of s which might be expected in consonance with the other languages: analogy with the initial s of "seven" (seMi) might be the cause. 'esrim 'esrin 'isruna 'S2 ry s3losim t3latin talatuna (tlty) Although this reconstruction enjoys a considerable degree of probability. fem. tintiini tifliitii !alii!at tJZiit !alii! 'arba'at 'arbr/ii 'arba' 'arba' 7. The numerals from "11" to "19" are normally formed by the juxtaposition of the unit-numbers (with inversion of gender from "13" onwards) and the numeral "10" which appears in some variant forms: Heb.amfJstu lJ.amsa lJ. Observations on the above table: a) the forms given for Akkadian are those of the absolute state (cf. 10 m. 'e§tii sittat set sitt sab'ii sab'at sab' s9ba' tfJmiinyii !amiiniyat tfJmiine !amanin tes'a tis 'at tfJsa' tis' 'esrii 'asarat 'fJsar 'asr are more frequently used.83los 'esre. 24-28) who holds that the Akkadian.amiS lJ. moreover.g.dii ?udii trnen 'itniini tarten 'itnatiini. "13" masc. § 12. 4 m. 490) that in Proto-Semitic "20" is expressed by the dual of "10" (*'asra > 'isra by vowel dissimilation).ams sta'. l:yamsata 'asara. they are used in the gender opposite to that of the noun which follows in the genitive plural (e. Syriac "14" masc. iSten Mfd istiat. which must be regarded as Proto-Semitic. Examples: Hebrew "13" masc. 'arb3ta'8ar.2. while the gender of both components is inverted (e. fem. f. and samaneseret "18" (fem. it has to be realized that.at s(J)nayim s(g)tayim SJlosii siilos 'arbii'a 'arba' lJamiMa 7. For the other languages the following observations ~ may be made: a) in Ugaritic (see the statement in the preceding paragraph) inversion of gender is not applicable. Ar. while the others are substantives. 'asartu wa-sala8tu. f. fem. 'asara.4.Jti.atti kgl'e(tu) k<J1' eti salastu saliis 'arbii'tu 'arba' lJ. . f. f.).} ] H sebet sebe sb'(t) [samanit] samiine !mn tiSit ts' tise eseret eser 'sr 'el. all these numerals have the fixed ending -a also in the unit-numbers. 3 m. f.116 Morphology Akkadian U garitic Hebrew Syriac Arabic Ethiopic 'a7. 6 m. i) the numerals "one" and "two" are adjectives.ad 'i{i. f. l:yamisseret "15". fem. 14. pp.amsat 7. c) for the vocalizations of "two" in Hebrew and "six" in Syriac cf. and Ethiopic forms are feminine plurals in the absolute state.

f) in Ethiopic there also exists a parallel series with the ending -awi (qadamawi. etc.. Heb.11 .g.amis "5th".23.10. 2. mrb't. "13th" salasseru. "60" has the form susu or 8USSt~ which may be compared with that for "1/6" (of. In Arabic we have "11th" ly. Fractions are generally formed on the pattern qubr: thus in Arabic tult "a third". and for the ordinals from "12th" to "19th" the ordinal is always followed by 'asa~. 14. the Ethiopic and Syriac forms are connected with a Semitic root denoting "precedence". § 14.am'CIs sad'CIs siib'CI' sam'CIn tasJ' . or-possibly-by analogy with !J. Ethiopic forms the fractions with the masculine and feminine ordinals followed by ''CId "hand" (e. Above "tenth" there occur in Akkadian "11th" istenseru .amisit "a fifth" .ms tdt sb' tmn rison #ni s'CIlisi r'CIbi'i 7:dimisi siMi s'CIbi'i sJmini t'CIsi'i 'ii8iri 14. in Syriac tultii.9. its numerals from "60" to "90" are not of the same type as those of the other languages. § 12. In Ugaritic no ordinals beyond "8th" have hitherto been found. etc.). unknown) exhibits feminine endings and a prefix m. "6th" is siidis (instead of *siidit) owing to progressive non-contiguous assimilation. Ugaritic rbt. a similarformation occurs in Akkadian (rabitu "a quarter". "20th" esru. 14. ribbo. 14.8. rebbo "10.omes "a fifth"). sJnayim s'CInayim "two by two". rub' "a quarter". and occasionally in other languages also (Akk. e) in Ethiopic "2nd" is derived from different themes: in addition to kal'CI' we have ka'Jb and dagJm. "30th" selasu. b) in the U garitic documents hitherto discovered th~ ordinals "7th" and "8th" are attested in the feminine only. but the normal pattern re-appears in the feminine sanit with the meaning "the following day (or night)". rub'ii. ly.iidi 'asar a . thereafter the cardinal forms are used. 'arb'CI'awi. 14. *sudsu > SUSSll. sissa. d) in Arabic.7.(mtltt. In Hebrew and Syriac there are no special forms beyond "lOth". § 14. b) Ethiopic ''CIll is used for "10. Hebrew uses as fractions the feminine forms ofthe ordinals (ly. The numerals "100" and "1000" are clearly derived from a common origin (with the exception of "1000" in Akkadian which is connected with a word meaning "people"): Akkadian Ugaritic Hebrew Syriac Arabic ESA 100 me'at 1000 lim mit dlp -. Heb. c) In Hebrew.) .000". though thIs is rare) . while Ugaritic (whose vocalization is. salasawi. The other ordinals .118 Morphology The Numeral 119 owing to the adoption of the Sumerian sexagesimal system.amsu sessu sebu samnu tisu esru ~n tlt rb' !J. Syr. Ordinals and Fractions 14.. qabir in Hebrew and Syriac.g. etc.6) are adjectives derived from the cardinals on the pattern qabir or qabur in Akkadian. Syr. Heb. s'CIba' sJba' "seven by seven" (for the use of the absolute 14.ru sanu sa18u rebu !J.). qiibir in Arabic and Ethiopic. etc.10). of course. Observations on the above table: a) "1st" is formed on a number of varying themes (in Akkadian another form is attested. The ordinals from "first" to "tenth" are as follows: Akkadian Ugaritic Hebrew 1st 2nd 3rd 4th 5th 6th 7th 8th 9th lOth Syriac qadmaya tenyana t'CIlitaya r'CIbi'iiya ly./'CIt 'll(m) ''CIll (of.'CImisaya s'CItitayii s'CIbi'aya tJminaya t'CIsi'aya ''CIsiraya Arabic 'awwal fani talit rabi' lJiimis siidis siibi' t amin tasi' 'asir Ethiopic qadami kal'CI' sal'CIs rab'CI' !J. etc. c) for higher figures of.6. "14th" erbeseru. "12th" sinseru. cf. from "20th" onwards the cardinal forms are used. Otherwise the forms of the cardinal numbers are used. "6th" is formed from "6" (siMi. dagmiiwi. Observations on the above table: a) the usual form for "1000" is related to the noun meaning "ox". instead of *hdisi) on the analogy of the cardinal SI}S.._ v aSJr ma!J. In other cases use is made of forms which cannot readily be reduced to common patterns. sebitu "a seventh"). "a sixth".5. e. etc. rab'CI''CIt ''CId "a quarter"). In Ethiopic 20th to 90th is expressed either by the cardinal or by the addition of the suffix -iiwi (''CIsrawi.000" ("1000" being expressed as 'asartu m'CI''CIt = "ten hundred"). istiyu. mea 'elep Ethiopic ma 'alpa mi'at 'all m't(m) m.). The distributive numerals are usually expressed by a repetition of the cardinals.

15. Akk. Ar. The Particles 15.. 'ayka. ly. 'emmat. summa "if"): Akkadian Ugaritic Hebrew "and" u w(} w "and. Akkadian has special forms: for "1" istena. ly. prepositions. Syr. which is encountered in North-West Semitic. Heb. by" "to" "to" "like" ki(ma) "over" eli "from" "with" "up to" adi Ugaritic Hebrew Syriac Arabic b(} b !)(} bi l(} 1 li 1 'el 'ila k k(}(ma) 'ak ka 'l 'al 'al 'ala min men min . ben "between" (and as a noun "interval"). sam(ma). The more important prepositions are shown in the following comparative table: Akkadian "in. ~t. 'f}l}el "beside" (and as a noun "side").atta "up to" (but South Arabian 'd[y])..adu 'aly.adu "one by one". §§ 12. the adverb of existence (affirmative) Ug.5. ana "to". Syr. taba'it "well". rubu'a. Eth. m(}sla "with".79). la "not" as well as Akk. Ug. 15. These include the demonstrative adverb of place Heb. § 12. a simple adjecti~e or substantive may be used as an adverb (e. 2. Heb. gt an " very. tahtu "underneath". mata "when?". Apart from the adverbs of demonstrably nominal origin. and the negative form Ass.120 Morphology The Particles 121 state cf. Heb. 'al. Another characteristic formation.). and this adjective or substantive appears in the absolute state in those languages which have a distinctive form for it (cf. "by day". qablu "first". Ug. ma!Jra "before". Syr. Heb.ul. umam "by day". Heb. 'ayin. etc.2. warka "behind" (already used in Old Babylonian and Old Assyrian). '~ka..79) (e. etc. With this group are probably to be connected the Hebrew adverbs in -am (yamam "by day".66) and more rarely -atta and -um to which correspond adverbial formations in -u in other languages (Ar. ayyanu. for "3"-"10" the nominal pattern qubura' (sulusa.g. Eth. '"ska "up to". ina "in". y(}hudit "in Hebrew". Akkadian uses the ending -is for the formation of adverbs (cf.g. Finally. Adverbs 15. etc. Syr. ya7}ad "together"). 'aly. Under the term "particles" are subsumed (for the sake of convenience rather than as a linguistically accurate classification) adverbs. Eth. '~. satIve end'mg: e. but there remain many cases in which such relationships cannot be established. tamman. 'it "there is". lassu. rab "much". The accusative ending also appears without the nunation (or mimation): e.7.: cf. nagha "in the morning". la. 'm tm 'am ma'a 'd 'ad '"damma Ethiopic ba la kama la'la '(}mna . Ar. also Heb. istu "from".4. 15.1. .ar "after" (and as a noun "back"). E. the temporal interrogative Akk. etc. umakkal "one day". Heb. the interrogative adverb Akk. Ijabaly. etc. Heb. 3. Ar. laysa. Syr. also" (&)p 'ap "or" u 'a u . for "2" sinnu. Observations on the above table: blanks indicate either missing forms or forms of different origin (Akk. tamma "there". Prepositions 15. aw 'en Arabic wa fa 'aw .6. mati.a masa'a "morning and evening". Syr. Eth. 'ayte "where~"..65-66). 'et]. § 12. Akk. Ar. 'aly. In adverbs of nominal origin it is characteristic of Arabic (and to a lesser degree of other languages as well) to use the accu"v'dd "yawman . and interjections. layt "there is not". Ar. The principal conjunctions with common roots are as follows (among the independent forms note especially Akk. r. involves the feminine ending: Heb. Conjunctions 15. the negative adverb Akk.3. "if" hm tm "in order kima k ki that" Syriac w(} 'ap . there are others of some importance which have common roots in Semitic. etc. Syr.g. conjunctions. l.g.). risana "at the beginning". ul. itti "with" [for this latter cf. An analysis of the particles often reveals a nominal or pronominal origin. matay.). Akk. etc.g. etc. 'ayna. y~s. sappir "beautifully").. § 12.innam "gratis". la. 1. tn kay Ethiopic wa 'aw '(}mma kama . A ' a badan" a Iways. Syr. Eth. In some prepositions a nominal origin may be detected: e. Ar.

however. According to Petracek's recent studies on 16. yan?urn. 16. ittiq "he passes"). These variations are more clearly marked. '8. irappud "he runs").g.g. alongside l}aliq). u occurs in all cases where the action is intransitive (e. Reb.1.g. In the prefix-conjugation the variation in the second vowel is at least partly paralleled: u or i corresponding to a. hasuna "he was beautiful". For the interjections various vocalic elements are used: Akk. The antiquity of this threefold vocalic ~cheme in Arabic is confirmed by some of the oldest manifestations in North-West Semitic. In stative verbs the distinctive vowel i predominates in the prefix-conjugations (pres. it is in full use in Arabic (e.C. kJtib "it was written" as well as "written"). § 16. yet it remains generally identifiable. ikbit "he became heavy"). ilmad "he learnt"). The linguistic evidence brought to light during recent years (especially for North-West Semitic of the second millennium B. The semantic connexions may be somewhat fluctuating and are not always readily identifiable. kutiba "it was written") where the prefix-conjugation in the passive has its own vowel scheme as well (e. a) Simple Stem In Akkadian the distinction between active and stative verbs is less pronounced as far as vowel variation is concerned. while a is rarer-though ancient and attested in Old Akkadian and Old Babylonian (e.g. UM. but these forms may. a passive of the simple stem exists in Ugaritic (where. AI'./'snnn. and a-u-a for a lasting condition 'or state: e. in the South-West Semitic area where. Variations in the vowel pattern relate to a distinction between action and state. 'i. active verbs have i as their second vowel (e. pret.g. at times. ynktabu "it is written"). the pattern a-a-a stands for an action a-i-a for a transient condition. Gordon. 'ay. nor are all the stems attested over the entire range of the Semitic languages. § 12. Apart from Arabic. be original participles whose functions have been extended by analogy. Syr. pret. at the same time. 'i. salima "he was well".g. ennujam. i. F. i. ilammad "he learns". e.122 Morphology The Verb 123 4. (5). pres.g. In active verbs a is more common in the present tense and u in the preterite (e. pret.g. Another common interjection is Akk.e. by innovations and analogical restorations. na?ara "he looked at".3. In the stative.) reveals a remarkable wealth of forms in the most ancient phase of the Semitic languages. iqrab "he approached". l~k "away!". of. Interjections 15. AI'.8. This stem shows the three radicals in their simple form. in the suffix-conjugation. cf. Ugaritic. in the prefix-conjugation. The Semiticverbhasasetof themes or stems (0£. it cannot however be readily distinguished from the passive stem with prefix n-. where in the suffixconjugation it formally coincides with the stem with doubled second radical (*qubar > qnbbar.g. iskun "he put"). There is a purely formal coincidence here with the vowel distribution in the derived stem with prefix' -. ikabbit "he becomes heavy".2. § 10. later iqrib). sakin "he is placed"). '8. kataba "he wrote". The passive is formed on the vowel pattern u-i-a in the suffix-conjugation. The Verb 1. l}alaq "he is lost". In the course of time (particularly in the North-Western region) numerous reductions have occurred-accompanied. pres. Reb. For the paradigm qbr of. Eth. in accordance with the vowel of the corresponding adjective (e. It is also possible that there are some traces of this passive in the Aramaic of the Arpad inscriptions and in Biblical Aramaic where the second vowel is lengthened and the form thus coincides with that of the participle (e. and a to i. in fact.8 d) and in the prefixconjugation with the stem with prefix h. while stative verbs may have a or u as well. Sometimes the imperative is employed as an interjection. Verbal Themes or Stems 16. Amorite. .g.3. and the Tell Amarna glosses. in the Tell Amarna glosses. and in Rebrew. hn "behold". as regards their semantic relevance.38). a is found in the preterite as well (e. idallip "he disturbs") as well as in some verbs of motion (e. yalctubu "he writes". p. without verbal meaning: e. AI'.g. hinne. yaslamu. '8.g. yal. isakkan "he puts". Reb. Ug.g. i appears in cases where the action of the verb is regarded as momentary (e.(yoqbar). which formally corresponds in Akkadian to the West Semitic suffix-conjugation (cf. while u generally remains: e. the following paragraphs) in which formal changes correspond to certain semantic variations and express different aspects of the action connoted by the root.g.

g. A variant of the stem with long first vowel is that with a diphthong: this is a development of which there exist very few traces in North Semitic (e. qatala "he fought" (= "tried to kill").assen "he strengthened".10. Ar. 16. Ug. Syr.n~n "to bring up" from a root n~n "to be a child") occurs in Akkadian. qabara. Traces of this stem in North-West Semitic are rather dubious (cf. The causative may refer to a state or condition. quttila "he was massacred".g.g. that with prefix s. especially in modern Arabic (e. l.6. brought about by a change of vowel patternin the same manner as shown in the simple stem: e. Akk.g. seems to have a primarily "factitive" significance.m "to cause to eat" = "to feed". qubira. Syr. yuqattilu "he massacres". b} Stem with Doubled Second Radical lines of the vocalization scheme applied to the previous stems. i. Again.g. d) Stems with Prefixes s-. 514-15). from mqt. This stem seems to have primarily reciprocal significance. 16. as a causative in relation to a state or condition: e. though less frequently. a distinction between active and passive. sll. It may therefore be concluded that this stem is typical of South Semitic.g. Ar. qatala "he killed". pass.g.sana "he did well" from l.7.asan "beautiful". 8 . Heb. Another aspect of the causative is its declarative value (e. which is attested over the whole Semitic area. pass. kasara "he broke". Bibl. 'ab'ala "he feasted" from ba'al "feast"). To this meaning-aspect must be added the denominative one (e. hismin "he grew fat". but with the first vowel long as the distinctive mark: e. This stem. ubattiq "he cut to pieces"). k()Ula "crown". Arabic possesses in this stem. wal}aya "he visited"). gawraba "he put on socks") in mainly denominal roots (cf.e. in Ugaritic. too. '- 16. Of the three stems.(also found outside Semitic in Egyptian: e.124 Morphology The Verb 125 inner flexion (cf. in Ethiopic. the discussion in Garbini. Akk. Syr.9. The same passive exists in Hebrewon the assumption that the Pu'al form (qnbbar) owes the a of its second syllable to analogy with the prefix-conjugation (y()qnbbar). the internal passive (which is wanting in Akkadian) is to be regarded as a secondary development of West Semitic. i.e.m "to eat".or h. ll. ESA sl'db "he caused to place".8. SNO. it appears also in Arabic and Ethiopic (as well as in Amorite and. pp. dshlk "I cause to flow" from hlk. Ar. sa'bed "he enslaved" from 'bd). Finally. kunnusn and snknnSn "to subdue"). 16. from 'db.g.' in the latter the correlation between form and semantic value has been largely lost (e. in cases where the action remains attached to the subject (e. Akk. the causative may have intransitive significance. ibtuq "he cut". GVG.g.g. saklilU "they completed" from kll. The Semitic languages present a series of stems with prefix s. gawzel "he set fire to") but more ample ones in Ethiopic (qobara. snlbnrn "to grow old". yuqattalu "he is massacred".g. Eth. 126-34). This stem is attested in Arabic and.a "he was eloquent" from fa?il. and in such cases it may coincide with the "factitive" of the stem with doubled second radical: the two themes are then used alongside each other without appreciable distinction (e. Eth. from the root msl). ynqabirn. South Semitic uses this stem widely for denominative verbs (Ar. 'al. in a few . pp.e. 16. I. kataba "he wrote".5. § 78) and in Arabic. h-. all sharing a causative connotation: e. Akk. saqaya "he tormented". kataba "he corresponded". Ug.()san "he was strong". hi?diq "he pronounced just". 'amsala "he pronounced similar". Brockelmann.g.146 to 147.3).g.g. The same prefix is attested in Aramaic (e. yuqabarn. nsamqit "he caused to fall". Arabic makes use of a variation in vowel pattern to express the distinction between active and passive-on the 16. Dillmann.or '-. Akk. from the root ?dq. Ar. qattala "he massacred".4. c) Stem with Lengthened First Vowel 16.11.g. At times this stem also indicates an action directed towards an object as well as an attempt to accomplish something (conative): e. qebara: cf. kallel "he crowned") and the intensive aspect (e. EG. pp. i. § 11. Syr.g. to indicate an action accomplished together with another person: e. and in the South Arabian dialects (here S2 > Sl) with the exception of Sabaean: e. uballit "he made to live". iblut "he lived". l. 16. Heb.g. 'aqama "he remained"). "eloquent". Aram. 'af?al. 'asgala "he divined" from sagal "divination". kassara "he shattered".

15. p. connected with the formal identity of the prefixes (Brockelmann. Arabic possesses specific vowel patterns which relate to the distinction between active and passive (the vocalization is the same as in the preceding stems): 'aqbara.17. passive. Examples: Akk. GAG. too. The stem with prefix h. Ar. § 16. root s'l. 'iqtatala "he fought".uru "to meet". US). Arabic again has variations in the vowel scheme. mentioned-this stem appears with some quadriconsonantal verbs. haraqa alongside 'araqa "he poured"). The stem with prefix '. In Akkadian this theme adopts in part the vowel distribution of the simple stem (of.? "he washes himself". suggest the possibility that both go back to one original theme whose prefix h. Of all the stems dealt with above under a to d. UM. In Ugaritic this stem is attested hut the n is almost invariably assimilated to the following consonant (of. Eth. 'albes "he clad" from lbS. in classical Arabic. 'anfar'a?a "he jumped". ibbassi "he becomes".and' . producing reflexive. root prs. and is thus infixed. in Moabite (e.!. Ar. hanpN "he caused to go out" from npq.126 Morphology The Verb 127 surviving traces. 'awdaq. however. Heb.g.is prefixed to the simple stem in the Aramaic languages (e. root kbd).g. in the Tell Amarna glosses (alongside the less frequent one with s-). it changes place with the first radical.g. root lly. Aram.g. pass. 'etqfel "he was killed". and sometimes also reciprocal connotations (for further meaning variants in Akkadian of. naprusu "to be separated".are not found simultaneously in the various languages.later became' -. 16.arna). to analogy with the prefix-conjugation (yoqba1'). together with the fact that the prefixes h. from the semantic point of view. root rly. Gordon. preceding paragraph). with stative verbs its meaning is predominantly ingressive: e. yqds "he dedicated" . Developments in Aramaic and Arabic. 16. in other North-West Semitic languages) in combination with the infix -t. yanqabiru and yunqabaru. For the causative stem. in the Phoenician form yqbr (suffix-conjugation): e.2 and von Soden. Syr.g. ta'asra "he was bound". in the other languages in which it is attested: in Akkadian (e.12. yuqbaru. GAG.occurs in Amorite. the 0 of the first syllable has to be explained in terms of § 10. In the stem with doubled second radical the metathesis of t. 16. 'inqafa'a "he was cut to pieces".13. In Ethiopic-as has been .?) . 6S). e.m "I am fighting".g. 16. LiJ:l. The same pattern and meaning are represented by the Hebrew and Biblical Aramaic stem Hophal (hoqbar)-again on the assumption that the a of the second syllable is due. p.g. yrtly. pass. Sabaean. Sabaean hrtr' "he subdued" from rtr'. f) Stems with Prefix (or Infix) t- 16. in Amorite (cf. 'ltly. Old Aramaic. in Phoenician (e.21). Bibl.19. and in Arabic (e. in Ugaritic (eg. however nkbd "honoured". 16. e) Stem with Prefix n- the exception of Aramaic. as it appears to be. p.14. Syr.g.18. t. nansum "to shoulder". ni8'al "he was asked".16. It is attested over the entire Semitic area (with some traces in Egyptian) with 16.g. in pre-Islamic North Arabic. nasa'um "to carry".g. hiqdis "he consecrated" from qds.g. perhaps. ibassi "he is".m) .(cf. and in the most ancient phases of . This stem has passive and reflexive meaning. GVG. Ethiopic shows a development towards a causative connotation which is. root 'sr). in Hebrew.appears in the most recent phase of Aramaic. In Ethiopic it is rare but occurs in some quadriradical verbs. of the existence of a fourth variant of the prefix. von Soden.with the first radical takes place in Akkadian only: e. root ml. 'astaya "he gave to drink" from sty. moreover.r). p. mitl. root qtl). Heb.S e above. A few remnants of this prefix survive in classical Arabic (e. root qfl) and in Ethiopic (e. There is evidence.g. below.g. 16. yuqbiru. but in the present case the expression of the passive becomes effective only in those instances where this aspect is absent in the normal theme: 'inqabara and 'unqubira. LiJ:l. root qt'. hawdaq "he offered" from wdq. thtpk "she is being overthrown".amudic and LiJ:lyanite: e. root hpk). and in Ethiopic: e. ustallamu . 536). additional themes may be formed with the prefix (or infix) t-. 'aqtala "he caused to kill" from qtl. Moabite. variant of hawdaq (of. It is possible that this causative pattern might be detected in some disputed cases in Ugaritic (of. 'uqbira. apparently of secondary origin. I. § 16. the proper name Yabtaly. 121).

sababara "he smashed". g) Other Stems 16. There also exist a few cases where the first radical re-appears after the second (e. It is also attested. bardada "he covered with stones". 'ettrim "he was raised".g.22) of this prefix with t-. a Hittanaphal of y'b( ?). inserted in the simple theme (pres. 16. Ar. root msl.g. taqabara and htqilbira. galbaba "he wrapped"). 16. Bibl. 'etly. A variant of theme IX is stem XI: 'iqbarra. etc. sutamru§u "to endeavour". hitqaddf}s "he sanctified himself". to quadriliteral themes of the type qabqab. sutam7}uru "to cause [numbers] to correspond with one another". This stem also exists in Arabic (e.g. uktata§§ar "he will be equipped". Syr. § 9. In Akkadian we have a series of additional forms with the infix -tan-. generally with iterative or intensive meaning (e. in Syriac (e.25. including that of a causative of the simple theme with t.cf.21. root §fr. zalzala "he shook".awa "he prostrated himself". Eth.128 Morphology The Verb 129 "they are being kept safe". i.26. § 8.g. Heb. yastaqbiru and yustaqbaru. but the effective expression of the passive is contingent on a formal and semantic contrast. Ug. physical defects. Comparative Grammar 9 . Ar. Syr. 'istaqtala "he exposed himself to death". in that with doubled second radical (pres. root qds. 16. root rwm (with assimilation t' > ttl. 'i§farra "he was yellow".(pres. though rarely and with certain doubts and reservations. and already in Amorite proper names of the type Batasni-' Il. root mly. yataqabaru and yutaqabaru.> ult. Eth. Syr.32). The extension of originally biradical roots gives rise in West Semitic. in North-West Semitic (e. while the type ustaqabbar has various and not yet fully explored connotations. taqabbara and tuqubbira. For all these stems (except the last which is attested in Aramaic only) Arabic has the usual variations in vowel pattern. 'abded "he enslaved") and perhaps in Akkadian (e. with a perfect of the type 'iqbarra which is used for verbs indicating colours. An Old Aramaic form evidently due to Assyrian influence is htn'bw.stem (e. § 9.g. and in that with prefix n. uqtanabbar) . 'astamly. it is realized only when the passive meaning is wanting in the ordinary form of the stem: 'iqtabara and 'uqtubira.g.: e.20. which appears in isolation in the inscription of Bar Rakib. ustalpat "it will be destroyed". Syr. possibly by analogy with the causative with prefix h-: Heb.assan "he was fortified".(pres.brings about the metathesis (cf. It appears to be a characteristic tendency of Akkadian and Ethiopic to form further stems by a combination of those listed above.e.ara "he showed himself merciful". taqaddasa "he was sanctified". Another l\ioscati. root m7}r) and that of an inner passive of stative verbs (e. root qil) and in Ethiopic (e. Heb. certain secondary and rarer types occur in the various languages. root ndb.23.g.produces a new stem in Arabic and Ethiopic when joined to the theme with first vowel lengthened: e.g. gilgf}l "he rolled".22).wy "she prostrates herself". possessing iterative meaning. Aram.produces a special theme in Aramaic: e. Repetition of the second radical is a common feature in the modern Ethiopian languages. 16.(e. farfaba "he called [camels]".r). utnennu "to pray"-cf.g.in the suffix-conjugation.the combination with t. IX in Arabic. Noteworthy in Akkadian is the Neo-Assyrian formation with reduplication of infixed -ta-: e.24. The stem with prefix s. badbada "he devastated". Kienast's recent studies). Ar. Eth.g. yaqtabiru and yuqtabaru. qarqes "he shook"). iqtanabbar). some forms corresponding to the stem with repeated third radical occur in Ethiopic (e. tstly. root mr§). A notable phenomenon of Hebrew and Aramaic is the metathesis of the prefix t.g. balbel "he confused". Amh. tamasalil "they resembled each other". root slm. Ar. In the stem with prefix' .before a dental or a palato-alveolar fricative (cf.g. root qil. root lpt). in that with prefix s. The prefix t. histaly.: e. 16. In addition to the basic stems set forth above. 16. ittanaqbar). etc.g. In Hebrew and in Biblical Aramaic the initial t takes a further prefix h. 'estawdi "he promised . The theme is common in Akkadian where it presents two types which differ in the forms of the present (ustaqbar and ustaqabbar-for ust.g. taqatalil "they fought together". Yistasni-' Il). One such case is the theme classified as No. hitnaddabil "they made a voluntary offering". yataqabbaru and yutaqabbaru. takassara "he was shattered". ustanaqbar) .22.g. both Northern and Southern.g. Examples in other languages: Syr. Tigrifia qatatala "he slaughtered").confessed". In the other Semitic languages. the type ustaqbar has the function of a passive of the s. 'istaqbara and' ustuqbira.

1. One of these conjugations uses prefixes (type yaqburu: the third person masculine singular is cited in accordance with accepted practice) and generally indicates an incomplete action which corresponds. In the West Semitic area. Ugaritic fly. verbs developed from causative stems (e. and a passive with t.stems as well as inner passives in Arabic. d) Reduplicated roots (e. I. naparqudu "to lie on one's back"). as different temporal concepts converge in each of these two conjugations. IV.g.) and generally indicates a completed action which 9* I. n) or 4th radical (e. taqabbara 2. naballcutu "to pass over". ' astaqabbara I. 'ismalJarra "he was very high").g. Arabic 1. sukenu "to prostrate oneself". 540-43).stem. As for Ethiopic. iterative ustanablakkat). 3. 1.27. two conjugations which are usually called "tenses". Examples:' anfabraqa "it shone". Hebrew gilg(}l "he rolled"). according to the traditional approach. Akkadian suklulu).prefix occurs only rarely. denominative verbs (e. 1.) The other conjugation employs suffixes (type qabara: qabarat. 'aqbara III. 'arsaly.stems are limited to Ethiopic (e. 3.alaba). These verbs form reflexive-passive t. a number of verbs with four radicals. or imperfect. III. to a greater or lesser extent. Arabic and most of the other languages exhibit. which are the distinctive element of this conjugation. The "Tenses" 16. original triradicals with added 2nd (mostly I. it forms a stem with doubled second radical and one with lengthened first vowel from the theme with prefix' . The Akkadian verbs in this category all exhibit either l or r as their second radical (e. while the s.causatives and causativereflexive 'asta.28. 3. Hebrew. both of them form iterative tan. qabarta.1. IV. II. 3.g. usqabbar): this is a poetical form used for either of the two themes which coalesce in it. causative usbala7ckat. III. Syriac 'abded "he enslaved".130 Morphology The Verb 131 stem typical of Akkadian is that resulting from the combination of the prefix s. present. c) Related to b are Arabic roots with n infix after the 2nd radical (e. ibbalak7cat "he passes".g. . 'aqabbara 2. according to circumstances. e) Ethiopic verbs of five radicals are formed from triradicals by the repetition of the last two radicals (e. But this nomenclature must be considered improper. 'aly. All verbs in this category possess only a fraction of the stems and forms of the triradical verb. are in some instances supplemented by suffixes having the function of contrast and identification of the various persons.g.g. iterative ittanablakkat. to our future. the entire system in Ethiopic looks as follows: 1.rr "to burn").stems (e. 2. in Ethiopic there are also a few with five radicals. etc. b) In Akkadian and Ethiopic we find a group of verbs belonging to the n.and the doubling of the second radical (pres. The "tense" system presents one of the most complicated and disputed problems of Semitic linguistics. 'ibransaqa "he flourished") or with doubled 4th radical (e. As there are virtually no common Semitic roots among these verbs. ·qabbara 2. GVG. 'adanga?a "he confused". Ugaritic mgmg "to mix".g.a "he sullied"). l. The Ethiopic verbs of this type do not form a causative. Syriac saklel "he completed"-cf. qabara 'aqabara taqabara ' astaqabara Finally. it would be more appropriate to speak of "aspects". 'astasana'awa "he pacified"). geminated roots (e. 'astaqbara 'anzahlala "he languished" (with "weak" second radical). 'anfaffafa "it dripped" (with reduplication).and from that with the (already compound) prefix st-. (The prefixes. and Old Aramaic. h) Verbs with Four and Five Radicals 16. Their morphological structure is irregular. 2. II. pp. Arabic basmala "he said bismillahi"). suparruru "to expand". while 'a. there are no derived stems-apart from the t.malmala "it became green".stem. qabara II. In all Semitic languages we encounter. The n.g.g.stem serves as causative. we must consider them innovations in the various languages.g.albasa "he enticed"-cf. r. taqab(a)ra IV.s'J/y.stem takes the place of the simple stem. The following principal types are attested: a) In Akkadian alone we find verbs of the type with prefix s-: suqammumu "to be dead-silent". the combination of more than one stem is widely attested in the modern Arabic and Ethiopian languages (cf.g. Brockelmann.

(type iqtabar). forms like Y3qabber. some scholars maintain that only one of them is n. zikaraku "I am a man". von Soden.31. called "preterite". another. called "present". In Arabic. may originally have possessed a "tense" connotation not very different from the type of semantic contrast observed in Akkadian. In Ethiopic the conjugations Y3qabb3r and Y3qb3r (intrans. We have also insufficient data for Amorite where. Rossler). which expresses an action complete in itself but still persisting in its effects (or subsequent to another completed action). As for the prefixconjugations. it now seems safe to say that the Arabic "tense" system represents the result of a long process of evolution. baltalcu "I am alive") as well as a substantive (e. from zikaru "man"). and a thircl one with suffixes (type qabir). Finally. Traditional Semitic grammar was inclined to consider the Arabic situation as original and that in Akkadian as secondary. the so-called "gerund". which is formed on the pattern qabir (cf. These considerations have brought about a crisis in the conventional conception of the primary character of the Arabic system and have stimulated vigorous scholarly discussion (Brockelmann. in particular by Libyco-Berber. Klingenheben. a "preterite" and a "present" (or "habitual" form or "continuative") which reveal definite points of similarity with the Akkadian verb: compare Akkadian ipru8 and iparra8 with Libyan ifre8 and ifarre8 (Rossler).132 Morphology The Verb 133 corresponds. which now relate to the distinction between indicative and subjunctive. Cohen. in fact. Meyer. This conjugation must be regarded as an Akkadian innovation. formally at least. according to circumstances. Rossler. § 16. that the position is more complex: in several instances in West Semitic. respectively. Proto-Semitic possessed almost certainly a nominal suffix-conjugation (surviving in the Akkadian stative and Ethiopic "gerund") which in West Semitic has evolved into a verbal conjugation-yet without differentiation of mood. GAG. More detailed study of the Semitic languages as well as HamitoSemitic comparisons have shown. and before him Landsberger and other scholars) to detect in Akkadian the existence of a fourth conjugation with infix -ta.30. one cannot exclude the existence of forms of the type yiqabbar (Meyer. Moreover. we encounter the opposition qabar: yaqbur. in the etymological sense of these terms. on the other hand. Fleisch. called "perfect". there is the recent tendency (von Soden. usually held to belong to the stem with doubled second radical. however. one of the yiqab(b)ar type for incomplete action.29. Driver. pp. Kurylowicz. . while the qbr conjugation has notable points of contact with the Akkadian stative (Goetze). has a different vowel and syllable distribution (inner morphemes) and connotes completed action (type iqb~tr). The two conjugations are usually called "imperfect" and "perfect". 16. also with prefixes.g. In the Tell Amarna glosses three conjugations have been identified: one of the yiqbur type for completed action. This last type represents in essence the conjugation of a noun and may constitute a verbal adjective (e. 16. East Semitic (Akkadian) presents a system of several conjugations: one with prefixes for incomplete action (type iqabbar). Finally. In U garitic the existence of two differentiated conjugations on the consonantal pattern yqbr might be suggested by the fact that the same pattern appears to indicate both completed and incomplete action. 16. the bibliography). and. to our past tenses. yet we are not in a position to form any reliable judgement as to the semantic significance of this opposition. it is likely to have been the function of this suffix-conjugation to record a state or condition and to describe it as having been accomplished. and these findings have been corroborated by evidence furnished by Hamitic languages. we have discovered elements of a distinction between two prefix-conjugations.70) with pronominal suffixes-rather like the Akkadian stative. 104-105. and a third form qabajijur for completed action (corresponding to the stative In Hebrew the conjugation of the yiqbor type is employed· not only for incomplete but also for completed action-as is shown particularly by the use of the conversive w. Rundgren. In addition Ethiopic has a suffix conjugation. we are unable to penetrate to a stage preceding the considerable measure of systematization to which the language has been exposed. Y3qbar) . which might well be a pointer to its origin outside the verbal system. may instead reflect a yiqabbar conjugation of the simple theme (Landsberger).g. both Northern and Southern. and others). called "stative". which is still unresolved (0£. Without entering into the details of this debate. Hamitic languages present two distinct prefix-conjugations. damiq "he is good". Thacker.

in fact. conversely. p. pp. a "momentary aspect" yaqburu and a durative present yaqabbar. 3. and the subjunctive or "relative" mood (whose functions are different from those of the subjunctive elsewhere: cf. the moods are expressed not only in the imperfect but in all the "tenses".32) Amorite presents yaqbur and yaqburu. To postulate the existence of only one prefix-conjugation in ProtoSemitic is considered by some scholars an inadequate solution-nor does its indeterminate character as regards tense commend itself to them. 108) has -u (Assyrian -uni). the other mood attested is a so-called ventive in -am. however. It should be noted.34. In Akkadian the modal system shows a remarkable divergence from West-Semitic.33. A full range of moods in the "imperfect tense" (as regards the "perfect" cf. This latter distinction would have been realized only subsequently in the historical development of individual Semitic languages-and in a number of different ways. the Akkadian subjunctive or "relative" mood (iqburu) has recently been regarded as the ancestor of the West Semitic prefix-conjugation yaqburu (Kienast). Garbini. Mention has already been made of the hypothesis claiming a secondary origin of the Akkadian form iqabbar by means of a redesignation of the stem with geminated second radical. This problem cannot be separated from that of the moods (cf. because some have detected in the West Semitic "jussive" yaqbur a development of the Akkadian preterite iqbur. in the Tell Amarna glosses we encounter a volitive in -a and an energic in -na. As will be seen in § 16. was rather difficult to determine when Brockelmann wrote on this question (GVG. but the view has also been advanced that iqabbar was dropped or restricted in use in West Semitic on account of its formal identity with the imperfect of the geminated stem.134 Morphology The Verb 135 to be attributed to Proto-Semitic (Cohen). GAG. 554). 170-71) which would tally with the Arabic and Ugaritic ending (but B. energic yaqbur-an(na} . Nowadays remarkable corroboration of this modal variety has been furnished by Ugaritic which places itself alongside Arabic with the same set of endings-recognizable by the vocalisation of ' (though the severely defective sp~llir:g leaves the distinction between the two forms of the energlC 111 doubt.-vis East-Semitic remain considerable. the observations in the preceding paragraph) . In other West Semitic languages (as will be shown in the following paragraphs) some remnants of moods have been discerned which agree wholly or in part with the evidence furnished by Arabic and Ugaritic.32. In West Semitic. where the moods are being expressed by means of differences in the endings: indicative yaqbur-u. The Moods is attested in Arabic. The two different aspects of action reside in the contrast between these two forms which remain opposed. p. 144). Whether this range of moods could be attributed to Proto-SemItIc. as a group. OA. in Or 29 [1960]. von Soden. In West Semitic the prefix-conjugation appears to have been set apart for the designation of an incomplete action in opposition to the suffix-conjugation which developed into the expression of completed action. In East Semitic (Akkadian) the prefix-conjugation might have continued in use for both completed and incomplete action but subsequently evolving into two types by means of vocalic and syllabic reconstitution (iqabbar arises secondarily alongside iqbur by a functional reassignation of the intensive ?-thus Rundgren). some reservations have also been expressed as regards the Ugaritic subjunctive in -a: of. that in a group of Old Akkadian texts the suffix of the subjunctive appears as -a (Gelb. to the suffix-conjugation which is retained for the designation of a state or condition. jussive yaqbur. In North-West Semitic (leaving aside Ugaritic-see § 16.34. the endings differ from those in the other languages: the indicative has none.. Kienast. SNO. In addition to the indicative there remains 16. 152-53. thinks that the supposed subjunctive in -a is. a ventive without mimation). The Hebrew documentation is less relevant because the shedding of final vowels includes the modal morphemes. I. 16. 16. subjunctive yaqbur-a. the North-West Semitic documentation suggests a semantic development of the subjunctive into a cohortative. modal differentiation is limited to the imperfect. §§ 16. p. A somewhat singular position in the reconstruction of the Semitic "tense" system is at present held by von Soden who posits three prefix-conjugations: a preterite yaqbur. In the first place. It might have had the function of indicating action in contrast to state or conditionwithout distinguishing between completed and incomplete action. but a modal distinction cannot be determined. Secondly. pp. the differences vis-a.32-36).

· imperative qJbJr.14-17): e. it cannot.g. § 16. 4. The suffix-conjugation of the simple stem is inflected in the principal Semitic languages as shown in table 1. therefore. yiqbor.136 Morphology The Verb 137 a jussive which is characterized (but not in all cases) by vowel reduction: e.g.35. Hebrew) we also find an imperative (masculine singular) with the further ending -ii which is probably the cohortative element previously referred to (§16.Hebrew ritic Sing. 3 m. seems unlikely to possess energic connotation (cf. yikbad. yiqqiily.g. i. a) The Akkadian stative (if account is taken of the connecting vowel -ii-. in Egyptian and Biblical Aramaic. Finally. Eth. 'eqtJlii "let me kill"). imperative qJbor (but prefix-conj.e. imperfect yiiqum "he rises". qabrii f. we find-as in Phoenician-a distinction between indicative and jussive based on the presence or absence of -n in the third person plural. by means of thematic variants: indicative YJqabbJr. §§ 9. but the absence of vocalization allows only the identification of the energic morpheme -no Ethiopic distinguishes two moods. subjunctive YJqb'Jr. There also exists an energic or cohortative in -ii. qabratunu f. In South-West Semitic the position of classical Arabic has already been dealt with. 16. qabir f. Inflexion a) Simple Stem: Suffix-Conjugation 16. but in more ancient Aramaic dialects. The element n appears before suffixes. the indicative and the subjunctive. (qabra) f. In the Dead Sea texts the forms with ending -ii are used for the simple indicative as well. prefix-conj.C 0 nj ug a tion Akkadian U ga.g.) (pass. (qabirta) 2 16. qabarat qabarta qabarka qabarti qabarki qabartu qabarkii qabarii qabarna qabartum(u) qabartunna qabarna qabara qabarata qabartuma qabarii qabara qabarkammu qabarkfJn qabarna qbr qbrt qbrt qbrt qbrt qbr qbr qbrtm qbrtn qbr qbrt qbrtm qbrny qabar qabfJra qabarta qabart qabarti qabarii qabarii qfJbartem qfJbarten qabarnii Du. prefix-conj. the forms with -n predominate. which characterizes its second and first persons. imperative kJbad). probably of pronominal origin (as shown by their external form). In the consonantal spelling of Phoenician the distinction between indicative and subjunctive is expressed in the third person plural by the presence of -n in the indicative and its absence in the subjunctive. qabratina 1 qabranu Syriac qJbar qebrat qfJbart qabart qebret qabar(iin) qfJbar(en) qfJbart6n qfJbarten qabarn(an) Arabic Ethiopic (act.37. qabrata f.)yiiqom. § 16.38. imperative q1lbur. any departure from this rule is due to the appearance of prosthesis or anaptyxis as a consequence of consonantal clusters in initial position (cf. Akk. (subjunctive) YJqbJr. Heb. qabrati 1 qabraku PI. Modal differentiation may be said to have been entirely discarded in Syriac. be identified with certainty as the same morpheme.2). 16. Semitic verbal inflexion is effected by means of personal prefixes and suffixes. The following paragraphs offer some general observations on this table: 1.30 for a comparison of the Ethiopic forms with the Akkadian prefix-conjugations). subj. Simple Stem: Suffix .) qabara qubira qabara qabarat etc. 3 m.ennu "he takes him"). The form of the imperative generally corresponds to that of the prefix-conjugation (short form) without its prefixes. prefix-conj. The element n. qabrat 2 m. iqbur. although it does not seem to carry energic value (e. . jussive (way. and of the consistent suppression of the second vowel) corresponds in its endings to the West Semitic perfect (see however below apropos of each individual person).36. later on.34). In some North-West Semitic languages (Ugaritic. 3 m. used chiefly in the first person (e. etc. qabra 2 m. in the simple theme. Hebrew above). Tell Amarna glosses. which appears before suffixes and is attested in the late phases of Aramaic (except for a Eolitary instance in the Zkr inscription). of intransitive verbs YJqbar (0£. has the vowel pattern characteristic of that stem (0£. There are some noteworthy Assyrian variants: qabriiti for qabriita (which occurs also in Old Babylonian) and qabriini for qabriinu. That of South Arabian and of preclassical Arabic is probably similar. all the Semitic languages have an imperative which.

§ 16 AI) and.46. q3bartem).g.39. in others the developments which have been postulated may have been quite different. the lengthening of short vowels in pretonic open syllables (same example). § 13.40. for the other conjugations and stems).38-41) note: a) in Hebrew. in verbs with "weak" third radical the pronunciation conforms to the spelling: e. for the third person singular.7). 16. 16. however. though the latter remain part of the graphic pattern (third plural masculine: q3bar. (wdiw "they rejoiced"). §§ 10. the change a> e in closed syllables (qebrat). well to insist from the outset on the hypothetical character of these reconstructions which are intended (and ought to be used) as working aids only. the proposed explanations often resolve the problems of only some of the Semitic languages.41. For the other languages (cf. For the third person plural we may propose the ProtoSemitic forms qabaru (masculine) and qabara (feminine) which appear as such in Ethiopic (and probably also in Ugaritic). qebrat) . almost certainly by analogy with the first person singular. listed separately in the table). 16. b) North-West Semitic of the first millennium has carried out consistently the characteristic changes connected with the incidence of stress (cf. in NorthWest Semitic and in Arabic. In many cases the reconstructions are subject to ambiguities and doubts. For the other languages (cf.138 Morphology The Verb 139 16. p'ltn "she made me". Proto-Semitic forms will be posited for each individual person (and later. For the first person singular we may postulate a ProtoSemitic form qabarku which appears as such in Ethiopic ifor the length of the final vowel cf. the Proto-Semitic forms qabar(a) (masculine) and qabarat (feminine).g. it will not. 16.38-41) note: a) the operation of analogy with the second person singular whereby. therefore.g. Similarly. is attested before pronominal suffixes in the consonantal spelling of Phoenician (e. bearing in mind the claims of the Akkadian stative (§ 16. the consonant of the suffix has become k. b) the phenomenon peculiar to Hebrew (though already attested in Amorite and the Tell Amarna glosses) where the vowel of the ending becomes i. In the following. qubira are inflected with the same endings and are not.42.44. b} in Ethiopic. For the second person we may propose the Proto-Semitic forms qabarta (masculine) and qabarti (feminine) which appear as such in Arabic (probably also in Ugaritic).137-42). which is otherwise dropped. while for the feminine the Biblical k3tib reflects the older form with suffix -ti.45.8. It is. but the original ending -at reappears before pronominal suffixes (e. the Greek and Latin transcriptions almost invariably testify to a masculine suffix -t. §§ 16. the reduction to 3 of short vowels in open unstressed syllables (qab3ra. with first person singular suffix). They appear as such in Arabic and in Ethiopic (as well as in Ugaritic and in the Tell Amarna glosses: e.31).10). in Akkadian (it should be recalled that the independent personal pronoun of the first person shows the consonantal element k as against t of the second person. abadat "she perished"). 16. cannot be fixed with certainty for ProtoSemitic (this was also the case for the pronouns-cf. the ending -t of the feminine. In Hebrew the final -a of the masculine reappears as a connecting vowel before pronominal suffixes (e. Finally. the -a of the feminine seems to be formed on the analogy of the feminine morpheme of the noun. except for the third person where it is well established (and perhaps for the second and first persons dual). q3bt'irani. probably by analogy with the possessive suffix -i. the consonant of the ending becomes t. we may propose. cf. c) The personal endings do not vary in accordance with the internal vowel patterns (qabara. so that the Masoretes appear to have adopted an archaic form.g. § 13. a similar process has taken place in the N eo-Assyrian variant forms qabraka and qabraki. written qbrw. therefore.1). r3maw "they threw". qabura. 10. 16.2-3) and. This has entailed. the reduction to 3 or shedding of short vowels in open unstressed syllables (q3bar. as against p'l "she made"). so far as the flexional suffix is concerned. in the same manner. §§ 16. the shedding of final short vowels (qabar). The . for Hebrew. While recalling the normal variations of vowel pattern (§§ 16. with first person singular suffix). for Syriac. 16. qabira. q3baratni. d) The length of the final vowels.g. §§ 16. be indicated (traces of an originally long vocalization can be detected in some forms before pronominal suffixes: cf. the apocope of both short and long final vowels.43.

-hen (the analogy is not perfect. 16. The first person dual is attested in Ugaritic only. and Talmiidic Aramaic. and the formation of the masculine on the analogy of the final consonant of the feminine (a process begun in Egyptian Aramaic. c) in Arabic. The prefixes and suffixes attached to the two conjugations are. GAG. Heb. The following paragraphs offer some general observations on this table: 16. and it has been doubted (Wagner) that such a form existed in ProtoSemitic. §§ 16.8. Hamito-Semitic comparisons. §§ 10. 10.49. and completed in Biblical Aramaic.38-41) note: a) in Akkadian and Hebrew the final vowel is changed to 1L. the transition u > 0 of stressed short vowels (same example). . they also agree with those of West Semitic generally-subject only to the observations in the following paragraphs. i in 3 (cf. For the first person plural we may propose the ProtoSemitic form qabarna which appears as such in Arabic and in Ethiopic. Akkadian shows third person dual forms which are however used only in the Old Akkadian.38-41) note: a) the Hebrew change i > e in the feminine.96). and the stabilization of the original final vowel attested in the Dead Sea documents by the mater lectionis h (cf. For the third person dual we suggest the Proto-Semitic forms qabara (masculine) and qabarata (feminine) which appear as such in Arabic (and probably also in Ugaritic). the analogical formation (so far as the vowel is concerned) of the masculine. a) The two prefix-conjugations of Akkadian ("present" and "preterite") and of Ethiopic (indicative and subjunctive) differ from each other formally and semantically. For the second person plural we may posit the ProtoSemitic forms qabartumu (masculine) and qabartin(n)a (feminine) which seem to occur as such in U garitic.48. For the second person dual we may propose the ProtoSemitic form qabartuma which appears as such in Arabic (and probably also in Ugaritic). -en are perhaps due to analogy with the personal pronouns 'attan. and their genetic connexion is not accepted by all scholars (cf.10). for Hebrew. qatalbnnahu "you killed him". §§ 16. 16. the change a > i in closed unstressed syllables (same example. 'atten and the suffixes -han.. in the prefixconjugation the feminine -an departs from the analogy).g.53. The Ethiopic subjunctive has two distinct patterns: transitive Y3qb3r. from the first millennium B. which presents both qbrtm and qbrtn. 16. p. Targiimic. §§ 16. formally identical. Of the other languages.51. c) in Ethiopic the change of the consonantal element of the suffix into k by analogy with the corresponding person in the sing. b) North-West Semitic has put into effect. 16. b) in Syriac. with the third person singular suffix). ninu. For the other languages (cf. the feminine ending -na is probably due to analogy with the corresponding ending of the prefix-conjugation.30. the presence in Old Egyptian of the first person dual ending -ny (coinciding with the Ugaritic morpheme) may possibly favour the assumption of such a form in Proto-Semitic.35). § 10.47.52. § 8. similarly for the personal pronouns § 13. the shedding of final short vowels (*yaqburu > yiqbor) . §§ 16. b) Simple Stem: Prefix-Conjugation 16. 16. The prefix-conjugation of the simple stem is inflected in the principal Semitic languages as shown in table II. the merging of the short vowels u.nu and suffix-nu). na7y. where we find q3bartun in the masculine and q3barten in the feminine). the feminine undergoes analogical adaptation to the masculine. this has entailed. b) in Syriac the subsidiary ending -an is probably due to analogy with the ending -an of the independent personal pronoun(['ena]~nan). For the other languages (cf. 186).04.50. (in Akkadian we encounter the Neo-Assyrian variant form qabrakunu). In Akkadian the dual may occasionally be used also for three subjects (von Soden. i. probably by analogy with the independent and suffixed personal pronouns (Akk. as indeed already in the Tell Amarna glosses and later in Nabataean. however. the supplementary suffixes -un.26). b) the Syriac vowel changes u > a and i > e (cf. and the shedding of the final vowel in the feminine which reappears however before pronominal suffixes (e. 16. all the changes consequent upon the incidence of the stress-accent (cf.140 Morphology The Verb 141 feminine ending -a occurs also in Biblical.38-41) note: a) in Hebrew. intransitive y3qbar.C. As regards its vocalization. 16. For the other languages (cf. Old Assyrian and Old Babylonian periods.10).e. we may perhaps propose a Proto-Semitic form qabarnaya.

16. while the alteration of the vowel pattern in the passive (cf. For the other languages (0£. a type of conjugation which has to be regarded as an innovation confined to Akkadian (§ 16. 16.3) brings about a change in the vowel of the prefix (yu. to the third person plural.instead of ya-) which remains constant throughout the inflexion. b) in Syriac (third person sing and plural) the prefix n-. and as peculiar in origin to stative verbs). of -m for those ending in -i.2-3. §§ 16. alongside a.142 Morphology The Verb 143 some scholars. In the following survey we have to leave out of account the special thematic patterns of Akkadian and of Ethiopic.63). Proto.56. and of -nim for the others. of course.Semitic yaqburu (masculine) and taqburu (feminine) which appear as such in Arabic (and probably also in Ugaritic). • . -ni when preceded by a vowel. Turning to the individual forms we may propose. the same applies.58. These are the principles of modal distinctions. In Arabic (and probably in Ugaritic) the subjunctive substitutes -a for the final vowel and drops the afformative -na or -ni when preceded by a vowel. § 8. for details the reader is referred to the paradigms in the grammars of the various languages. however. the reduction to a of short vowels in open unstressed syllables (yiqbaru).-< if2 16. the reader is referred to the paradigms in von Soden's GAG. but we shall allow for the variations of vowel distribution dealt with in §§ 16.2) has no effect on the form of the prefixes or suffixes.32-36) causes the addition of -u in the Akkadian subjunctive (Assyrian -uni) for forms ending in a consonant. e) For the Akkadian "perfect" with infix -ta-. and the addition of -am in the ventive for forms ending in a consonant. regard the vowel i of the prefix as primary. § 16. both for the simple and the derived stems. § 16.> *yi > i(0£. c) The variation of vowel pattern reflecting transitive or intransitive status (cf. the jussive drops the final vowel and -na. In Syriac the same changes are operative-save for the process a > e which takes place in closed unstressed syllables (neqbor). the energic substitutes -an or -anna for the final vowel (-anni in the dual and in the feminine plural) and drops -na. -ni when preceded by a vowel. §§ 16. 16. d) The differentiation of moods (cf.55. for the third person singular. instead .57.29).53-56) note: a) in Akkadian the prefix has evolved: *ya.

f. § 16. Comparative Grammar . § 8. c) in Syriac the feminine ending -an has been adapted. it is an innovation of East Aramaic (Old Aramaic and West Aramaic retain y-). The imperative of the simple stem is inflected principal Semitic languages as shown in table III. For the other languages note: a) in Ugaritic. qubrii. which occurs in Talmudic Aramaic and occasionally in Mandaean as well as in Biblical Aramaic leh'ewe "he is".53. and Ethiopic. or else a Proto-Semitic feminine taqbura(ni) might be suggested-in which case the Akkadian feminine with prefix i.: f. -a/na } -a 10 l\1oscati.65. § 16. Taking into consideration the merging of short u. these postulates agree broadly with the forms in Akkadian. §§ 16. For the third person dual we may propose a Proto-Semitic form yaqbura(ni)-in which case the Arabic feminine with prefix t. For the other languages (cf.of the third person. to conform with the masc. For the second person dual Arabic and Ugaritic postulate a Proto-Semitic form taqbura(ni).96).in the feminine (in Ugaritic and the Tell Amarna glosses it may be t. 16. For the second person singular we may propose ProtoSemitic forms taqburu (masculine) and taqburi(na) (feminine) which appear as such in Arabic (and probably also in Ugaritic).53-56) note the shedding of -u even in Akkadian and Ethiopic. is characteristic.64. a distinction were to be made between the conjugation-patterns yaqburu and yaqbur (thus von Soden: cf.and the fluctuating use of y-/t. probably owing to Aramaic influence.53-56) note the Akkadian prefix ni. c) Simple Stem: Imperative 16. IP. -1tn. For the other languages (cf. b) in the Hebrew of the Dead Sea texts the ending -un instead of -u has been attested. 2 PI.62. if. III. p. Syriac (for the prefix n. §§ 16. The concluding observations in the preceding paragraph are relevant also in the present context. Arabic. -u -i f.(cf.66.63. For the first person plural we would posit a Proto-Semitic form naqburu which appears as such in Arabic (and probably also in Ugaritic). For the other languages (cf. This explanation rests on the assumption of a single conjugation yaqburu in Semitic. may be considered a remnant of precative l. 16. 16. and in the only Palmyrene occurrence the consonantal prefix is tinstead of y.in Ugaritic are to be attributed to analogy with the second person dual. GVG.31) a different picture would emerge. 2 qibru q>Jbor(un) 'uqburtt qab(fi)ru q>Jborna q>Jbor(en) 'uqburna q>Jb(fi)ra 'uqbura Du.-56. qubur qubri qubra qbr qbr qbr qbr q>Jbor qibri 2 m. For the first person singular we may posit a ProtoSemitic form 'aqburu which appears as such in Arabic (and probably also in Ugaritic). 16.61. i into" in Ethiopic (cf. SimpIe Stem: Imperati ve Akkadian Ugaritic Hebrew Syriac Arabic 'uqbur 'uqburi Ethiopic q>Jbar qab(a)ri In the Sing.36): Singular Plural Dual 2 m.60.59. I. §§ 16.53-56) note: a) in Akkadian the feminine ending -a takes the place of the masculine ending -u. For the other languages (cf.58). in Hebrew. 16.53-56) note the omission of -u even in Akkadian and Ethiopic. For the second person plural we may propose the ProtoSemitic forms taqburu(na) (masculine) and taqbura/nct (feminine) which are broadly reflected in Arabic and Ethiopic (and probably also in Ugaritic). however. in part. b) in Syriac the feminine ending -an has again been adapted in part to tally with the masculine -un.in the masculine as well). §§ 16. Taking into account the general considerations set forth in §§ 16.would be due to analogy with the masculine. c) l-.which may be the result of analogy with the prefix i. Brockelmann. § 16.cf. 565). we may propose the Proto-Semitic endings (cf. by analogy with the second person plural and the third singular feminine. 16.144 Morphology The Verb 145 of y-. For the third person plural we may postulate ProtoSemitic forms yaqburu(na) (masculine) and yaqbura/na (feminine). 16.

. but this does not imply that they are secondary (cf. but it is still used for some substantives and adjectives (e. Note: a) in Akkadian the feminine ending -a takes the place of the masculine ending -u. etc. While the most ancient Aramaic inscriptions exhibit the radical consonants only. and the Nabataean mqbwr is almost certainly the result of Arabic influence.146 Morphology The Verb 147 16. while the change a > (} in the second vowel may be due to analogy with the vowel of the imperfect (y'Jqabb(}r). while another participial form (or nomen agentis?) assumes the theme qabari. fad'Jq "just").70.68.10d). cf.96) can no longer be formed at will. though rarely. and Ethiopic q'Jbur may possibly derive from an original qabUr. § 8. The verbal noun or infinitive has a variety of forms in the simple stem.which recurs in Biblical Aramaic (miqbar) and later in Syriac (meqbar). but q'Jbrat. q'Jraba "battle") which in the Modern Aramaic dialects appears as the regular form of the infinitive. However. nasig "fabric" (= "woven"). The active participle of the simple stem (table IV) goes back to a Proto-Semitic form qabir which appears as such in Akkadian. table V). qabir and qabur are also used in Arabic: e. m'Jqbar. and sporadically elsewhere. Syriac also makes extensive use of qabar (e. for that reason.g. will be the third person singular masculine.67. Ethiopic. d) Simple Stem: Nominal Forms 16. in Hebrew (qabOr. while Syriac q'Jbir is to be referred back to qabir (already the most ancient Aramaic inscriptions present the consonantal spelling qbyr).30-31. as usual . Syriac has qaber as a result of the change i > e (cf. C£.69.ir "slaughtered". Among these is the common theme qabar which occurs in Akkadian (qabaru) . Egyptian Aramaic has the prefix m. b) the Syriac auxiliary endings -un. IV. war'Js "heir". § 10. The "construct" infinitive of Hebrew may be the phonetic result of a different theme (*qubur > q'Jbor). The following comparative treatment refers therefore to the West Semitic area where certain common themes can be detected. qiibrJr) qfJbur nominal patterns. p. etc. In Syriac qabber the change a > e in the second vowel might be accounted for in the same way as in Hebrew (imperfect n'Jqabber). with the change a > a-cf. e) Derived Stems: Suffix-Conjugation Active Passive 16. It will be seen that the Akkadian stative differs in vowel pattern from the West Semitic suffix-conjugation and will therefore be left aside for our present purposes. rasul "envoy" (= "sent"). and Arabic (and probably also in Ugaritic). Hebrew qabur seems to presuppose an original qabur. The ensuing analysis will. GAG. Both coexist in Amorite and occur.g.83). -en might possibly be the result of analogy with the suffix-conjugation (the element -n occurs already in the masculine plural of the imperative in the Arpad inscriptions). § 8. Simple Stem: Participle Akkadian Ugaritic Hebrew Syriac qiiber qob?r qbr qiibiru qfJbir qiibur Arabic qiibir raaqbur Ethiopic (qabiiri. will not be dealt with in the following.g.71. 16. 16. For the stem with doubled second radical we would propose the common form qabbara which appears as such in Arabic and in Ethiopic (and probably also in Ugaritic). also exist (and may belong to an older phase of the language). §§ 16. The passive participle of the simple stem exhibits the widely used nominal patterns qabir and qabur. The Ethiopic pattern qab'Jr (for i > 'J cf. In Ethiopic the forms qabir and qabirat predominate. in Akkadian (von Soden.8e. In the other stems the inflexional suffixes and prefixes remain unaffected and.31 and 16. probably by analogy with the participial forms of the derived stems.72. In Hebrew we have qab~r. Arabic adds the prefix m(maqbur) . nor will the variations of vowel pattern connected with the passive. § 10. Amorite. ''Jbada "action". owing to the changes a > a (cf. therefore. they generally merge with the wide range of 16. be directed towards a comparison of the stems in the various languages and a conjectural reconstruction of common forms (the basis . In East Semitic some important divergent features persist. § 8.8c).83) and i > ~ (cf. In Hebrew we have qibbar or qibb(}r: the change a > i in the first vowel conforms to the requirements of § 10.43). For the Ethiopic "gerund". 10* . §§ 16. 60) where the function of the passive participle is normally assumed by the verbal adjective. nal).

rticiple muqabbiru Infinitive qubburu qbr yqbr qbr mqbr qibba/~r Hebrew (passive) qubbar ylJqubbar mflqubbiir qubbOr (cstr... qabb~r) qabbara {Yflqebbflr yuqabbaru Yflqabbflr qabbflr (maqabblJr) muqabbar qabblJr6(t) 0 ..-Conj.r qabb~r mfJqabb?r qabbOr (cstr. Pref..-ConJ.-Conj.-Conj. qubbur . ....-Conj. Imperative Participle Infinitive suqbur usaqbar { usaqbir suqbir mU8aqbiru suqburu d) Stems with Prefix h-. 'Hebrew ct> (active) Suff. Derived Stems a) Stem with Doubled Second Radical Akkadian Ugaritic (active) Suff. uqa bb' tr Imperative qubbir Pa.-Conj.$ .. Imperative Participle Infinitive hiqbir yaqbir haqb~r (passive) hoqbar yoqbar moqbiir hoqb~r Syriac (active) (passive) 'aqber naqber 'aqber maqber maqbiirii maqbar Arabic (active) (passive) 'aqbara yuqbiru 'aqbir muqbir 'iqbiir 'uqbira yuqbaru muqbar . Pref... {Uqabbar Pref.. haqbir) ~ .-Conj. 0 ~ 'd I:l" il'l '<I 0' b) Stem with Lengthened First Vowel Arabic Ethiopic (active) Suff. Ethiopic 'aqbara yiiqabflr { yiiqblJr 'aqbJr (maqbflr) 'aqbor6(t) 0" maqbir haqb~r (cstr....-Conj. 00 V. Pref.. qubbar) Syriac (active) (passive) qabber nlJqabber qabber mlJqabber mlJqabbar mlJqabbiirii Arabic (active) (passive) qabbara yuqabbiru qabbir muqabbir taqbir qubbira Ethiopic ylJqabbf. Imperative Participle Infinitive qiibara yuqiibiru qiibir muqiibir qibiir (passive) qiibira yuqiibaru muqiibar qiibara YflqiiblJr qiiblJr (maqiibiJr) qabflr6(t) c) Stem with Prefix 8Akkadian Ugaritic sqbr ysqbr sqbr msqbr "':l I:l" Suff.

Pref.-Conj. Pref." 0 Otl '<I t::r" Akkadian Ugaritic Syriac Arabic Ethiopic (active) Suif. Imperative Participle Infinitive naqbur { iqgabbar iqqabir naqbir muqqabru naqburu niqbar yqbr yiqqab(3T hiqqab!Jr niqbar hiqqabOr. Pref.-Conj. Imperative Participle Infinitive 'etqabbar netqabbar 'etqabbar metqabbar metqabbaru taqabbara Arabic (passive) tuqubbira yutaqabbaru mutaqabbar Ethiopic taqabbara { yfJtqebbar yfJtqabbar taqabbar 1-3 yataqabbaru taqabbar mutaqabbir taqabbur taqabbfJr6(t) CD t:r h) Stem with Lengthened First Vowel with t(active) Suif.-Conj. niqbOr (cstr. Pref.-Conj." 0" taqabara yataqabar'U taqabar mutaqabir taqabur I-' I-' Ol .-Conj..I 0 0 .. hiqqab(3T) f) Simple Stem with t- (passive) 'unqubira yunqabaru munqabar 'inqabara yanqabiru 'inqabir munqabir 'inqibar ~ '"t.-Conj. Imperative Participle Infinitive qitbur { iqtabbar iqtabar qitbar muqtabru qitburu 'etq>Jber' yqtbr lqtbr netq>Jber 'etqabr metq>Jber metq>Jbaru 'iqtabara yaqtabiru 'iqtabir muqtabir 'iqtibar (passive) 'uqtubira yuqtabaru muqtabar taqab(fJ)r6(t) taqabra yrJtqabar taqabar g) Stem with Doubled Second Radical with tAkkadian Hebrew hitqabba/!Jr uqtabbar { uqtabbir qutabbir muqtabbiru qutabburu yitqabb!Jr hitqabb!Jr mitqabb!Jr hitqabb!Jr Syriac (active) Suif. Imperative Participle Infinitive Arabic (passive) tuqubira yutaqabaru mutaqabar taqabfJr6(t) Ethiopic taqabara y>Jtqabar taqabar ~ .I-' 0 Ol e) Stem with Prefix nAkkadian Ugaritic Hebrew Arabic (active) Suff.-Conj.-Conj.

Se. as well as taqabara. in addition to Akkadian. with the customary change a > i (cf. 'aqbara with which the Arabic and Ethiopic documentation conforms (and thus probably also in Ugaritic). § IO.Se). perhaps on the pattern of the imperfect netqaber. h-.and doubled second radical we postulate the common form taqabbara which appears as such in Arabic and in Ethiopic. the thematic pattern might again be the result of analogy with the imperfect yaqtabiru.S3). . 16.74. In Hebrew we have niqbar. while the transition a > i in the second vowel may be due to analogy with the vowel of the imperfect yaqbir (cf. For the simple stem with t. 16. with prosthetic 'i.we may perhaps propose. '.. 16. For the stem with t.73. For the stems with prefix s-. haqbara. Arabic has 'inqabara. . the Arabic 'iqtabara. a common element taqbara of which. but no language actually exhibits this form. Syriac has 'etqaber. For the stem with prefix n. the only possible relic might be seen in the Ethiopic tansa'a "he rose" alongside tanas'a "he was raised up".77. the common element naqbara. § 16.we may posit the common forms saqbara. For the stem with first vowel lengthened we may propose the common form qabara which appears as such in the languages in which this stem is attested (Arabic and Ethiopic). § 16.1S): thus. In Syriac 'aqber the change a > e in the second vowel might be accounted for in the same way as in Hebrew (imperfect naqber).we may tentatively suggest. by analogy with the forms in the preceding paragraph. however. . 16.and a syllabic distribution which may have been influenced by the imperfect yanqabiru.. In Hebrew we have hiqbir: the change a > i in the first vowel is a result of the rule stated in § IO. and the form as a whole seems to follow the pattern of the imperfect (yitqabbffr).75. by analogy with the forms in the preceding paragraphs.76. probably by analogy with the simple stative qabra. with prosthetic 'i.. Elsewhere we encounter metathesis between the t and the first radical of the verb (of.may result from analogy with the Hiphil. In Hebrew we have hitqabbar or hitqabbffr: the prefix h. Apart from such a remnant the form taqbara has been driven out by analogical formations: in Ethiopic we have taqabra.152 Morphology The Verb 153 16.

on the analogy of the stem with prefix n. It was probably a general South Semitic feature. § IO.r shows the change a > i in the initial closed unstressed syllable (cf.is found in Aramaic only: Syriac 'ettaqbar is formed by the assimilation t' > tt (cf. § 8.> i. 16. i in 'J (cf. in the stressed vowel of the final syllable (cf. § 10. have a instead of u in the prefix (yaqabbir).80.we may propose a ProtoSemitic form yatqabi/ar(u}.82. instead of i >? [cf. § 1O.(0£. of the short stressed vowel in the final syllable (d. '.is probably due to analogy with the other preformatives of derived stems. with the change ya.96).and doubled second radical we may postulate the Proto-Semitic form yat(a)qabbi/ar(u) with which the Arabic yataqabbaru agrees.63). with the change i > e in the closed syllable (0£.78. In Syriac we find n'Jqabber. 16. For the simple stem with t. The Akkadian preterite iqqabir embodies the change ya. yu'aqbir(u).we may posit the ProtoSemitic forms yusaqbir(u). Syriac netqabbar has the customary transition a > e in the initial closed syllable (cf. . § 8.(0£. and the development i > r.r we see reduction to 'J of the prefix-vowel in the open unstressed syllable as well as the change i > r. In the Akkadian preterite uqabbir the initial y has been dropped (cf. § 8.l54 Morphology The Verb 155 Syriac 'etqabbar is probably again influenced by the imperfect netqabbar. too. appears to show assimilation of n to the first radical. attested in South Semitic only. § 10. 16. For the stem with prefix n. while the Ethiopic Y'Jqab'Jr again shows the common transition of u. In Hebrew y'Jqabbr. In the Ethiopic subjunctive Y'Jtqabbar the prefix Y'J. For the stem with t.> i. and by the Akkadian preterite usaqbir where the initial y.has been dropped (cf.79.63) and assimilation of n to the first radical. it appears as such in Arabic. Ethiopic has Y(Jtqabar: the prefix Y'J. of the short stressed vowel (cf.8g.is probably the result of analogy with the other derived stems. The Ethiopic SUbjunctive shows *y(J'aqb'Jr (for u. The Arabic development is *yu'aqbiru > yuqbiru. §§ 16.10c) and the change a > e (i > e) of the two short vowels in closed syllables (0£. § 16. i > 'J cf. assimilation of n to the first radical. with initial1l. In Hebrew yiqqabr. 16. For the stem with first vowel lengthened we may take as a basis the form yuqabir(u).and t.we may take as a basis the form 'astaqbara-attested as such in Ethiopic. 16. though the vowel of the prefix is different (yasaqbir) . might conceivably be the result of analogy with the imperfect of verbs with medial w/y [yaqim]). § 8.81. § 10. The Ethiopic subjunctive Y'Jqabb'Jr presents the merging of Proto-Semitic u. Hebrew yitqabbr.63) and metathesis between t and the first radical. For the stem with t. Ugaritic yqbr. d).10d).96) > yaqb'Jr.19).22) and on the analogy of the imperfect (nettaqbar).and t. and metathesis between t and the first radical (cf.and the simple stem with t. § 8. Syriac has *n'J'aqbir > *naqbir > naqber. § 8. For the stem with doubled second radical we may propose a Proto-Semitic form yuqabbir(u) which appears as such in Arabic.we may postulate a ProtoSemitic form yanqabir(u) which appears as such in Arabic (but cf. § 8.75-76).96). § 16. however. In Arabic yaqtabiru metathesis takes place. Amorite yinqabir). c). Akkadian has iqtabar in the preterite.83. § 10.(cf.10c. 16. yuhaqbir(u) .86. § 10.8c). i into (J (cf. but the purely consonantal South Arabian script does not allow us to arrive at firm conclusions.8e). In s-. 16. Ugaritic yqtbr shows metathesis and a different vowel in the prefix (yiqtabir). § 10.8e) and the change i > r. For the stem with prefix s. § 10. with reduction to 'J of the prefix-vowel in the unstressed open syllable and the change i > e in the closed syllable (cf.lOd). 16. The stem with prefix' . Arabic'istaqbara is formed with prosthetic 'i. f) Derived Stems: Prefix-Conjugation Hebrew we have *y'J'aqbir > *yaqbir > yaqbir (the change i > i in the stressed syllable. Syriac netq'Jber exhibits the usual reduction to 'J of the short vowel in the unstressed open syllable (cf.r we observe the change a > i in the initial closed unstressed syllable (0£.63).by analogy with other derived stems. 16. In the Akkadian preterite we have uqtabbir.and first vowel lengthened we may take as a basis the form taqabara which appears as such in Arabic and in Ethiopic.10d). § 10.8c). The first seems to be represented by Ugaritic ysqbr. For the stems with prefix h-. § 10. Amorite and Ugaritic.8c].85.84.

saqbir for suqbir in the stem with prefix s-) suggest a situation that was originally similar. the following treatment is confined to variant formations in West Semitic.occurs in Aramaic only: derived from a conjectural ya'taqbar(u) is the Syriac nettaqbar.15).88.99. However.83) to 'aqbir.and second radical doubled and with first vowel lengthened (taqabbar.(cf.e. For the stem with t. and Arabic prosthetic 'i.89. Akkadian. Apart from these characteristics.is the result of analogy with the other derived stems. Hebrew and Syriac (causative stems) and Ethiopic have a as the vowel of the prefix. § 10.we may posit a ProtoSemitic yastaqbiru to which the Arabic form corresponds-apart from the normal change s > 8 (of.8) to haqbr.r (§ 10. §§ 9. For the Ethiopic forms see § 16.91. Arabic 'aqbir represents *yu'aqbir (§ 16.(participle [*munqabru > ] muqqabru) and the simple stem with t.and lengthened first vowel we may take as a basis the form yat(a)qabar(u) which is attested in South Semitic only: Arabic yataqabaru corresponds to this form. 16.156 Morphology The Verb 157 16. For the participles of the derived stems we may propose Proto-Semitic forms characterized by the prefix mu. the participles correspond structurally to the imperfect (in Akkadian to the preterite).in the active: the consonantal structure (mqbr and msqbr.98.is formed as in the other derived stems.92. with the change a > e in the initial closed syllable (cf.r shows the evolution *yuhaqbiru > *Y3haqbr. The stem with prefix '.(cf.and t. Amorite. 16.90. of some complexity. § 9.95.e. § 9. in accordance with the conditions affecting short vowels in unstressed open syllables 16.l0d) and the assimilation of '. g) Derived Stems: Imperative 16.15).94. '. § 16.r. p.is reduced.14-15). 14). 16. In the stem with t and second radical doubled Hebrew adds prosthetic hi. In the Akkadian preterite ustaqbir the initial u.Arabic adds prosthetic h) Derived Stems: Participle 16. the forms of the imperative in the derived stems generally correspond to \ those of the imperfect without its prefixes: any departure from this rule (which will be examined presently) is almost invariably due to the appearance of prosthetic vowels (cf. § 9. 16. 16. i.14). and by the same prefix mu. In the simple stem with t. § 9.96. In the stem with prefix n.Hebrew haqbr.(participle muqtabru) in which the vowel i is dropped owing to the succession of short syllables (von Soden.Hebrew prefixes prosthetic hi-.and the vowel a after the second radical in the passive. 16. respectively) does not indicate the vowel quality of the prefix. In Ethiopic taqabar shows the insertion of a vowel after the first radical (cf. The only exceptions are the stem with prefix n. In the stem with prefix 'i. § 8. § 9. follows the general principles set forth in the preceding paragraph. The Ethiopic prefix-conj. These forms have been reconstructed on the basis of Akkadian. In East Semitic some Assyrian variant forms (qabbir for qubbir in the stem with doubled second radical.101.and the vowel i after the second radical in the active.Ugaritic and Arabic add a prosthetic 'i. s- and t. however Amorite yistaqbir). 16. either as an anaptyctic vowel 3 subjected to vowel harmony or by analogy with the stems with t. while in Ethiopic Y3tqabar the prefix Y3. In Ugaritic the only participles attested are those of the stems with second radical doubled and with prefix s. and Arabic. 16. The Akkadian development is.87. In Hebrew the prefix ma. which presents active forms only (but the function of the passive participle is often assumed by the verbal adjective: cf.15).69). i. both Northern and Southern.and t. taqabar). correspondence with the forms of the preterite without its prefixes. yiistaqb3r reveals the transition a > a in the initial syllable by analogy with the stem with prefix '.96). For the stem with prefix s.15).(cf. . In West Semitic.(of.as well as the change i > 3 in the final syllable (of. In the stems with prefix h-. however. In Syriac retraction of the vowel and stress produces the form 'etqabr instead of *'etq3ber. GAG. and the reader is referred to the paradigms. 16.97.93.

also Gray.106. that those principles suffice for the explanation of only a limited number of these verbal forms. I. either by means of phonetic changes characteristic of the consonants concerned or by the operation of analogy. qubbor. 5. to m9-.108. 16. metqabbar. y).104. so far as the consonantal spelling indicates) in the stem with n-: niqbar. rarely used in Hebrew. formed by analogy with the suffix-conj.. In Hebrew the infinitive is formed on the pattern of the imperfect without its prefixes and thus coincides with the imperative. qab9ro(t). in its turn. In Akkadian the infinitive coincides with the stative followed by the nominal morpheme (stative qubbur. to this is added the ending -o(t): qabb9ro(t). or semi vowels (w. By analogy with the absolute infinitive qabOr of the simple stem Hebrew also forms absolute infinitives of various derived stems with final vowel 0 (qabbor. 'aqb9ro(t) . and the -form qabari of the simple stem gives rise to analogical formations in the derived stems (qabbari. taqabbur. stative 8uqbur. In Ethiopic the infinitive is formed on the same pattern as the imperative. etc. m9-. cf. or the alveolar nasal n. i) Derived Stems: Infinitive 16. niqbor). pp. pp. In fact. The active use of the participle is.70). abs. Their forms are regarded as explicable on a basis of triradical roots. The so-called "Weak" Verbs 16.). In the remaining groups it may be shown that we are dealing. 110-18) these groups of verbs are usually lumped together under the term "weak verbs". with biconsonantal roots. we have the infinitives of the stem with doubled second radical (taqbir). 16. also U garitic nkbd. etc. is contracted with the prefix ' .101. In Arabic the infinitive possesses several patterns. In Syriac the infinitive retains the prefix m.(though not without some reservations).). with lengthened first vowel ([qibar >] qibar) . 'iqtibar.and geminated second radical the form (*m9hitqabbf}r » mitqabbf}r.e.105. fem. in accordance with the principles governing short vowels in open unstressed syllables (§ 10. however. 16. we shall confine ourselves to the observation of certain common features. or their second and third radicals are identical. while the slightly different construct forms are widespread. The absolute infinitive· is. § 10. An unusual form of participle occurs in Hebrew (also in Phoenician. maqbarii. some of which reveal similar schemes and may be grouped together ('iqbar. 'aqbari. to m9-. mettaqbar). and others. is characteristic of the infinitive of the derived stems (0£. greatly restricted. however. in its turn. in the stem with t. i. however. while in the remaining stems the change a > e (cf. 16. infinitive qubburu. 'istiqbar.of the simple stem. the participle in Ethiopic has become a lexical item rather than a regular morphological feature.8 g). The infinitives of the derived stems have a number of forms which can more suitably be examined in each of the languages. This is confirmed by the fact (already referred to in §§ 11. in the stem with h-.584-638. § 16. 'inqibar. metq9barii.102. and m9-.5-9) that many "weak" verbs appear in several forms which have in common two . SOL.100. The following chapters will be concerned with an examination of some types of verbs which differ from the regular pattern. 16. These verbs contain either pharyngals and laryngals (in particular' ). In traditional Semitic grammar (cf.10c). lS.lOd) takes place (metq9ber. is contracted with the prefixes of the various stems-thus producing. lS. the third radical having arisen secondarily in a process of integration with the predominant triradical system. GVG.107. etc. More advanced linguistic study has shown. appears before suffixes): m9qabbarii.103. the forms (*m9haqbir » maqbir and (*m9hoqbar » moqbar. In Ethiopic the prefix is stabilized in the form ma-.producing the form (*m9'aqber » maqber. In Biblical Aramaic the ending -a of the st.158 Morphology The Verb 159 (§ 10. infinitive 8uqburu. the second radical receiving the vowel a and the third the (abstract) ending -ii(t) (the final -t.is reduced. taqabur). for the most part. In Syriac the prefix ma. those with pharyngals and laryngals and those with (original) initial y. apart from these. Brockelmann. etc. which does not occur in the absolute state. hiqqabOr.

A characteristic feature of Hebrew.110) but also with other vowels: e.111. *hismi' "he caused to hear" > hismia' (cf. pp. "he sends" > yislaly. These are connected.160 Morphology The Verb 161 radicals and a basic range of meaning-but differ in the third radical. hence Akkadian will only be considered in connexion with the verbs containing the glottal stop: the change a > e caused by . 212-14). in cases in which e would occur before a pharyngal or laryngal in final position: e.u "he opens" > yaftaly.g. § 8. no doubt as an aspect of assimilation.. In Hebrew a subsidiary vowel of the a type establishes itself after' and ly. below).aww9r "he goes" > yaly.53-54). TPA. § 186). The glottal stop' occupies a special position which gives rise to phenomena not shared by other consonants: the verbs with' will. The phenomenon is fairly rare in Arabic (e. derived from ly.g.6). § 16. represent the preservation of an original a (e. Grammatica Aethiopica.g. ly. Finally. The opposite process a > 9 takes place when a laryngal is followed by a vowel other than a: *nasa'u "they raised" > nas9'u. in some cases. *'etd9ker "he remembered" > 'ett9kar (cf. The accurate phonetic notation of the Masoretes rejects simple snaii with pharyngals and laryngals and uses instead a compound s9wii. § 9. *yisloly. however. Ullendorff. If this explanation reflects the linguistic reality accurately.109.mol "he shows compassion"). Praetorius. in order to account for all the forms of these verbs in terms of the triconsonantal system. 118-38).u). *ya'mod "he stands" > ya'i'imod. *yaftuly. whether in some cases this phenomenon does not.110. pp. where the a of the prefix reflects the vowel of the Proto-Semitic yaqburn). in yaMob "he thinks". as presented to us by the Masoretic tradition. *ya'm9du "they stand" > ya'amdu (but yaly. in fact.g.g. Moreover. and " and sometimes also from g and h (cf. 16-18. it will call attention to certain facts of fundamental importance to the study of comparative Semitic grammar. *ba"!}r "he consumed" > ba'!}r. a "liaison" a is inserted into the articulation between the long vowels of other timbres and a following pharyngal or laryngal: e. (rarely h) when the consonant would othenyise be without vowel at the close of a syllable: e.6). Moreover.g. *barrik "he blessed" > barik). pp. they are occasioned by analogy with the "weak" verbs (of. In Syriac the tendency to change into a vowels contiguous to pharyngals or laryngals is somewhat sporadic: e. has produced two verbal classes (with a and with e). chiefly with a (of. Ethiopic lengthens a before a vowelless laryngal: *sama'ku "I have heard" > sama:ku (cf. The ensuing treatment has purely descriptive aims and restricts the term "weak" to verbs of probable biradical origin.86-87. yely.). neb'at alongside neb'ot "he pushes". 6. the question arises. 16.g. it is absent in connexion with h. § 8. however. pp. SLE. is its inability to double pharyngals and laryngals-with consequent compensatory vowel lengthening: e. Biblical Aramaic shares the same inability in the tradition of the Masoretes (including also r): but while in the case of " r and sometimes' there is compensatory vowel lengthening (e. § 16. the dissimilation *ha"!}l "cause to enter!" > han'!}l). particularly § 9. therefore. Verbs with Pharyngals and Laryngals 16. The verbs with pharyngals and laryngals exhibit certain specific peculiarities.54). Comparative Grammar 11 . Fleisch. SG. The fact that we are dealing with trends pulling in opposite directions (the reduction of triradical roots and the expansion of biradical ones) as well as the complex nature of certain phenomena will leave a wide margin of uncertainty which can be reduced only by specific advance in the study of this branch of comparative Semitic grammar.25) are characterized by the tendency to change into a vowels contiguous to those consonants. be dealt with separately. however. especially to Arabic grammar (cf. even though there are considerable fluctuations between them. for the most part. In Hebrew this tendency is the rule (e.aww9r. Verbs with pharyngals and laryngals (in Hebrew and Syriac also those with r: cf. Brockelmann. For detailed information about independent developments in the various languages the reader is referred to the relevant grammars. It should also be recalled in this context that in Akkad~an all the consonants of the pharyngal and laryngal series are reduced to ' (§§ 8. *nedkor "he remembers" > nedkar.g.g. 16. then we must consider as mere working aids those theories which have been applied. with the characteristics inherent in these consonants (cf.g. the change is regular. there are indications that at certain times and in certain areas the consonant was doubled (e.ezaq Moscati. In Ethiopic the change 9> a occurs in the prefixes of verbal forms whose first radical has a laryngal followed by a (by way of vowel harmony): *Y9ly.g.

This development is. §§ 16. 16. 16.113. Some imperative forms which drop initial ' (e. *u'arrak "he lengthens" > urrak. 'by "to want". In Akkadian some verbal forms with first radical' show syncope of intervocalic " followed by contraction in which the vowel of the prefix prevails: e. The first characteristic of verbs with ' is the elision of postvocalic " with consequent lengthening of the preceding vowel in the North Semitic area: a. nenhar "it shines". on the other hand. Akk. mali'a. § 16. In this example. these verbs are sometimes inflected by analogy with those of second radical w/y: thus alongside ida"im "it becomes dark" isal "he asks" (but in Assyrian sometimes isa"al. GAG.g. a good deal of contamination and interference between "weak" verbs and others of this type (cf. In Akkadian the imperative of these verbs generally presents a prosthetic vowel 11* . 130-33). maZ. Heb. 133). as these consonants cannot be geminated (of.g. In verbs with third radical' the characteristic elision of postvocalic ' in North Semitic (of. Verbs with First Radical n 16. but AI'. innabbit "he flees" alongside i"abbat "he is destroyed" (cf. *nentor "he guards" > nettor. the assimilation of n becomes inoperative in several verbs with second radical h: e. Hebrew verbs with second vowel 0. Akk. Syr.112) brings about coalescence with the verbs with third radical y as far as the resulting vowel is concerned (cf. nes'al "he asks" > nesal (cf. yintanun "they give"). *'a'mana "he believed" > 'amana.116.g. The peculiarities of the verbs with " outlined in the preceding paragraphs. tor "guard!" from ntr. GAG.109-111). nelJod. Ass. the form yi5lJf-z is the result of an evolution which may be represented as follows: *ya'lJuiju > *yalJuz > *yi5lJoz > yi5lJf-Z (by dissimilation). Noldeke. gas "approach!" from ngs.g. Another characteristic of Akkadian is the assimilatory complex nn.121): e. pp. II. probably by analogy (or perhaps restored by the Masoretes?): e. In Hebrew. *. do not apply to verbs with second radical ' for they follow the general pattern of pharyngals and laryngal~ (cf. This assimilation does not take place in Hebrew before consonants of the pharyngal and laryngal group.al "he inherits". cf.114. however. ya'alJn (as well as ya'alJaz). ikul. 16-17. *i'akkal "he eats" > ikkal. from the root yd': cf. their inflection is regarded as "weak" by contrast to the others.g. Proto-Semitic *ya'lJuiju. but this is probably due to secondary dissimilation of the doubled consonant-as may be shown by several cases in which n cannot be held to be original (e.g. ilJuz. pp. pr' "to cut") treats final' as a normal radical (von Soden.g. Syr.g.g. § 16. *indin "he gave" > iddin. 'mr "to say".162 Morphology The Verb 163 "he is strong". 108. mal?.g.g. § 16. p. as well as in the previously cited ya'i'imod./a (mal'a). von Soden. however. GAG. In Akkadian. we have a typical instance of vowel harmony in Hebrew.115. Akk. pp. There is. Eth. Syriac has some verbs with first radical' which are formed by analogy with those with first radical w/y (e. 'kl "to eat". *i'kul "he ate" > Bab. a small group of Akkadian verbs (e. tinda' "thou wilt know" < *tidda'.g. *ynpl "he falls" > ypl. *yinfor "he guards" > yiffor.g. mali (beside mala). Syr. Ar. from the root'lJ4 "to seize". 'py "to bake"). root nhr. i' produces i in Babylonian. p. e in Assyrian: e. 16. ' is preserved at the beginning of a syllable. we encounter in Arabic a type of dissimilation of two' in the same syllable: e. ya'lJurJu. zel "go!" from 'zl. (stative) mali. Heb. In North Semitic vowelless n is generally assimilated to the following consonant: e. pp. where this phenomenon does not as a rule occur. In Syriac.a'lJurJu "I take" > 'alJuiju. Eth.g. retain their initial n: e. resulting from a meeting of nand' in the verbal stem with prefix n-: e. § 171) shows syncope of '. however.g. KurzgefaBte Syrische Grammatik. yi5lJf-z. In South Semitic. 46-47). Syr. ekul.g. Heb. Ug.g. Heb. Rosenthal. and similarly in Arabic (e. yin1y. Grammar of Biblical Aramaic. *'iwta'ada "he promised" > 'itta'ada).112. 'awkel "he caused to eat". but AI'. Beer-Meyer. *'i'talJaija "he took" > 'ittalJaija by analogy with e. confined to a few verbs with ' as first radical ('bd "to perish". In Hebrew.g. mala'1( "they were full". von Soden. 16. 47). Heb. from the root 'kl).) 7. In Syriac. Proto-Semitic *mali'a "it was full". nafor "guard!". commoner than ". (Note the imperative sal of the corresponding Arabic verb s'l. Finally. The North Semitic imperative is formed without n: e. 126-29). Hebraische Grammatik.g. In Akkadian. Syr.111): e. In Biblical Aramaic the n is frequently maintained in these circumstances (e. lJuij "take!" from '1Jij) are analogous with those of verbs with first radical w/y.

Syr.g. y 16. sim for the same reason as above). so common that it is well to deal with them together. sema. In the suffix-conjugation West Semitic presents Heb. Heb. (subjunctive) Y'Jsim. note in Hebrew and Syriac the vowel lengthening in the prefix). y it is well to recall the phonetic laws about semivowels (of. simta. qam.20) as well as the working of analogy affecting "regular" and "weak" verbs. Heb. in the prefixconjugation this interference was only sporadic (e.g. lidi. Verbs with w. sam (but also bin' 'he understood"). Eth. 10. Verbs with first radical wand y constitute.. but forms like teb "sit!" and hab "give!". though there are traces of it in the form nu!Jti "I am quiet" of the Tell Amarna glosses and in the Latin transcription chon "he was" in Phoenician-Punic) . yasimu.g. wld "to bear". 9. Eth. Initial w does not appear in the imperfect: from the same root wld. sim. sim.g. while Hebrew has formations on the pattern of the verbs with doubled second radical (qom?m). 16. Syr. qayyem). Eth. bas "be ashamed!" (cf. A medial vowel a is rare: e. sim "put!" (Ar. Akk. qawwama. 9. For the complex forms of the verbal stem with . yaqum. preterite isim. Over the entire Semitic area initial w is absent in the imperative: e. 9.117. Eth. sama. nelad. Heb. y the imperative exhibits the appropriate long vowel between the first and third radicals: e. *stn~r "he asked for help" > st~r.61-65. 16. but cf. Heb. qum "rise!" (Ar. Heb. in origin. yasim. ledet. §§ 8. Eth. 9. In the Ethiopic forms. q1Lmta. Interference between the two classes is not unusual and is confirmed by the Akkadian stative kin and sim (Assyrian ken and sem). Akk. we note the characteristic change in North-West Semitic of w > y when in initial position (cf. yaqumu. pp. ladi "give birth!" (but wCijCir "go out !").g. of.g.11. Syr. nCiqum. they are particularly exposed to the operation of Systemzwang by which originally biradical verbs are integrated within the triconsonantal system. l"dat (Akkadian has waladu or aladu by reduction of initial w according to § 8. Syr. Heb. sam (but mit "he died"). the noun subtu from the root wsb). qoma (in the Ethiopic form the vowel 0 is derived from the diphthong aw.7-8. In the verbs with medial radical w. (subjunctive) YCilad (Akkadian forms a partial exception: ulid. this does not necessarily reflect actual pronunciation. uqur "destroy!" from nqr) . imperative as well as prefix-conjugation. Ar. however.141-42. agree with their counterparts in the other languages. The characteristic vowel remains and is long (also in Arabic) in the prefix-conjugation: thus Akk. kun "be steady!". din "give !"). Eth. yalidu. qum owing to the reduction of the long vowel in the closed syllable). Ar. qam. but in Old Akkadian and in Assyrian forms without prosthesis do occur (e. Ar. y?l?d. nCisim. in the Hebrew verb ysb "to sit" the first radical is originally w-as is demonstrated by the stem with prefix h-: hasib.3. Eth. Akk. Eth. In South Semitic. Syriac ilad is exceptional (by analogy with the verbs primae y). The original distinction between these two categories of verbs emerges once more in various forms of the derived stems: e. von Soden. sim. lidat. it is noteworthy that the usual Hebrew change a > 0 does not take place in this case. Akk. Akkadian (which has gemination of the second radical also in the present of the simple stem) exhibits instead the doubling of the third radical. Ar. Syr. Ar. whereas in ytb "to be good" the y is primary-as is proved by the form hMib. § 8. Heb. Syr.63. Finally. qawwama. sama. lidi. Eth.118. Eth.120. distinct categories.13. (subjunctive) YCiqum. Ar. SyT. Eth. For the verbs with w. Ar. the vowels are marked long on etymological grounds.g. etc. Ug. In the first place. from the root wld Akk. only those with first radical w (and they are the more numerous group) seem to be genuinely "weak". Akk. provided it is followed . certain West Semitic languages have infinitives without w and with "feminine" ending: Heb.g. 16. Ar. Heb.119. n is not subject to such special treatment-except for some cases in South Arabian: e. In the derived stem with doubled second radical. yld. Syr.64): e. some languages present forms which correspond to those of the "regular" verbs (Ar. Heb. 8. GAG. Syr. Heb. bos). yasum alongside the usual yasim). preterite ikun. The inflexion of the Arabic perfect shows the reappearance of the characteristic vowel: qama.164 Morphology The Verb 165 (idin "give!" from ndn. qama. lCidi.$ in Akkadian. Reciprocal influences between the two categories and the passage of verbs from one group to the other are.

yadlu. the verbs of this type appear in the forms farra. e. Ar. Eth. yabki.and suffix-conjugations are indicated): root bky. b'Jkii. nedle.125. analogy with the "regular" verbs operates: e. sar "he is false"). A small group of verbs with second radical 1 or r forms a special durative type with n: e. Polal.has yhrif. In Hebrew the perfect of the simple stem is integrated with the regular pattern (type sabab). ibki (preterite). imnu "he counted".124. y (with which coalesce those with third radical ') that y predominates over w in North Semitic. When the last radical has no vowel. baka. In the verbs with identical second and third radicals. In Ethiopic. These examples show: a) there are in Akkadian various exceptions to the North Semitic predominance of the type with y over that with w (e. perfect. etc. Syr. by analogy with the verbs primae n (type nebboz). e.g. this form (which is standard in Old Babylonian) is paralleled in Neo-Babylonian by those corresponding to regular verbs (e. but Akk. ·16. 16. but stative verbs have biradical forms (type ly. against the singulars idiiak > idak and iSiam). 'afarra.g. Ar. 'etbnez. elil "he is pure").g. . bezzat. 16. ilylu "he rejoiced"). naparruru "to disband". y'Jbki (subjunctive). nebke. isimmu "they place". etc. Heb. bakaya. In the derived stems forms on the model of the "regular" verbs are widespread. yumat "he is being killed". idukku "they kill". yiisobbu. etc. It is characteristic of the verbs with third radical w.g. integration within the triconsonantal system demonstrates the force of analogy. Example of a verb with original w: root dlw. The greater part of these verbal forms can be explained by syncope of w.123. on the pattern of the verbs with first radical w).g.g. sob). Hithpolel. Syr. yibke.g. dala.166 Morphology The Verb 167 by a vowel (e.g.w "to be content" the prefix-conjugation of the stem with h. The masculine singular participle is formed on the pattern of the verbs mediae w (type ba'ez): this analogy does not extend to the feminine and the plural (in contrast to Jewish Aramaic and Mandaean). imperfect y'J7. 16.ass'Js). this assumption is scarcely set aside by certain changes awa > 0 (e. In Syriac biradical forms are widely attested: the perfect of the simple stem is baz. the Hophal yosab modelled on verbs with first radical w (yosab). and in the derived stems farra.126. Formations by analogy with other "weak" verbs occur in Hebrew (e. y'JdlU (subjunctive). b) the Ethiopic forms bakaya and dalawa agree with the regular pattern and appear to favour the conception of these verbs as original triradicals.g. na'arruru "to come to the rescue".125). usmiit "he kills". idlu. g.g. others are inflected by analogy with the "weak" verbs: e. yafirru. usmattu "they kill". In Akkadian the verbs of this group are completely adapted to the regular pattern. In Arabic. In the derived stems metaplastic formations are common: Polel. the similar formation of verbs with third radical w/y: e. usmit "he killed". Before vocalic suffixes the second radical is geminated: e. sabbOti. yidle.127.g. 16. later ibannu).. § 16. 'afrarta. In the perfect 16.g. yufarru. ibanniu "they build". y between vowels and subsequent contraction of those vowels. before consonantal suffixes a connecting vowel is introduced: 0 in the perfect and e in the imperfect.g. The stem with s is formed similarly: e.122. while in the imperfect the first radical is doubled. uSmittu "they killed". dalil.am). d'Jla.lin'arirru "let them help". ibni "he built" for ibni. Akk. c) the triradical origin is also supported by certain Ugaritic forms which keep wand y (e. imperfect.g. In the inflexion of these verbs gemination of the third radical frequently occurs before vocalic suffixes: e. dtwt "she came").g. t'Jsubbena (cf.g. Verbs with Identical Second and Third Radicals 16.yn. baki (stative). A biradical form is presented by the stative of the verbs which indicate a condition (e. dalawa.wn and yhrif. integration with the "regular" verb is prevalent (perfect lJasasa. etc. Heb. This process is of some consequence in the historical development of the Semitic languages generally (e. Example of a verb with original y (the prefix. d) some interesting fluctuations between wand y are exhibited by South Arabian: from the root rif. commonly called verba mediae geminatae. halawa and halo "he was"). tiglena). yufirru. Akk. the probable biconsonantal origin is particularly evident. bakii. 9. yissob alongside yasob (cf. dan "he is strong". the imperfect and imperative also have biradical elements (type yasob. Eth. Some forms with doubled first radical are attributed to Aramaic influence: e. Noteworthy is also the tendency to shorten or even to drop final vowels resulting from contraction: thus Akk.g.121.

Akkadian verbs such as izuzzu "to stand" and itulu "to lie" exhibit forms that may be referred to several different categories (first radical n. dwl "to go to and fro". In general these anomalies can be resolved artificially by subsuming such verbal forms under categories to which in essence they do not belong: thus Rebrew lqly.133. Some scattered information on these points has already been given in various places in this book. 16.g.g. e.134. Much rarer are verbs in which all three radicals are "weak": e. r'y "to see".g. All the Semitic languages have doubly irregular verbs. Reb.g. e. e. Sem.g. wdq "to fall". sll. connotes a noise. Before pronominal suffixes we often witness the reappearance of Proto-Semitic elements which have underaone !:l considerable development in the forms without suffixes. frr and nfr "to £lee" . Verbs primae n: a) verbs whose biradical basis.128.136. "to sound pul}"). y: a) verbs which describe certain involuntary actions. mnw "to count".e. whb "to give" (first radical wand second h). y: a) (only ultimae y) verbs of terminative meaning. pp. GAG. e. "to say buly.) bny "to build". Verbs mediae y: a) verbs which describe a physiological function.116ff. e. There also exist a number of defective verbs whose forms diverge from the general patterns hitherto discussed. Eth. While details have not yet been worked out. Verbs primae w.132. b) verbs connoting a definite outcome or result.130. 11. gly "to reveal". medial radical w/y. l}yr "to elect". etc. b) verbs in which the element n has locative meaning. e. and some connecting vowels are inserted. ndy "to throw down". Reb. e. wsd "to lead to" . Akk. Syr.g. Akk. Sem.g. wld "to give birth". Sem. . rw~ "to run". Sem. ns' "to lift up". bzz "to plunder". Ar. fix". 154-56. Aram. npl} "to blow" (i.g. Certain alterations of morphemes and endings also occur. Arab. "to take" behaves like a verb of first radical n (imperfect yiqqaly. without the element n. etc. Arab. Eth. (except Eth. § 6. For the change of categories of. etc. 'wr "to be awake" (first radical' and second w). sgg and sgy "to err". Sem. Akk. Reb. Landsberger (1 slamica 2 [1926]. Akk. 16. Ar. e. ns' "to carry" (first radical n and third ').g. Akk. West Sem.g. nzl "to descend". ewu "to become". of. AI'. qwm «to get up". von Soden.131. sym "to place. pp.e. wgd "to find". 'dd "to count".) was the first to recognize that there existed a measure of correlation between several types of "weak" verbs (of. Doubly Irregular or Defective Verbs 16. 10.168 Morphology The Verb 169 of stative verbs (type ly.arnma) and in the simple stem with t(type tanabba) biradical forms are attested. 16. 12.135. syn "to urinate". Verbs ultimae w. e.g. Sem. npl "to fall down".137. b) verbs which connote the aim or target of a motion. Reb. In these verbs the inflexion takes account of the characteristics of both categories concerned. Aram. b) verbs which refer to types of motion.) and certain semantic categories. Verbs mediae w: a) verbs which describe a change of condition or transition from one situation to the opposite one.).g. Arab. B. etc.129.e. lwy "to accompany" (second radical wand third y). Eth.) and hlk "to go" like a verb of first radical w (imperfect y~lek). Verbs with Pronominal Suffixes 16. 16. For such verbs the reader is referred to the grammars of the individual languages. b) verbs which describe durative actions. Reb. i. Semantic Categories in "Weak" Verbs 16. and Eth. Akk. mwt "to die". In the imperative and the imperfect the shortened forms exist alongside those fashioned on the analogy of the "regular" verbs. wqy "to take care" (first radical wand third y). Syriac 'zl "to go" assimilates its 1 in certain circumstances (first person singular perfect: 'ezzet) and slq "to go up" assimilates the 1 and has forms like a verb with first radical n (imperfect nessaq).g. e. the most important categories-according to von Soden-are: 16. 16. Syr. West Sem. Verbs mediae gerninatae: especially verbs which connote a number of individual actions ("Kettendurative"). wrd "to go down".g. Arabic r'y "to see" presents shortened forms (imperfect yara). 16. verbs which combine two of the categories discussed in the foregoing: e. Akk. Sem. nbr "to sit down". 362 ff. Eth. nbMl} "to bark" (i."). Sem. Eth. *awu "to speak". e. Reb. e.

GrundrifJ der vergleiohenden Grammatik der semitisohen Sprachen. §§ 16.. Comparative Grammar of the Semitic Languages (London 1923). 19 (1950).103-29. Aram.138-39) before the suffixes: e..141. (Roma 1956). In Syriac the Proto-Semitic endings of the perfect reappear before suffixes: third singular masculine -a. Finally.A. KurzgejafJte vergleichende Grammatik der semitischen SpracMn. and of his KurzgefafJte vergleiohende Grammatik .170 Morphology 16. ZS = Zeitsohrift fur Semitistik. Old Aramaic may insert -an (cf. F. de Lacy O'Leary. Eg. Dhorme. I of Brockelmann's GrundrifJ . Bergstriisser. 375-85. 7-16.34. second singular feminine -ti. Blake. Prospettive di metoda statistioo nella classificazione delle lingue semitiOM. SOLDV = Studi orientalistioi in onore di G. 50*-115*. Semitisohe Spraohwissensohaft. Bibliographie semitique. with the addition of: . Preois de linguistique semitique (Paris 1910).. Fleisch. P. 445-78.g. 16. Dissimilation is the cause of the change -i > 9 in the second singular feminine before the suffix pronoun -ni. I. 17 (1948)... 59*-90*. The bibliography is selective and. does not include works published before 1908 (the date of Vol. Einfuhrung in die semitischen Spraohen (Miinchen 1928). Scope of the Survey Barth. G. C. In the second plural feminine the full ending -k9nna appears which is more often shortened to -ka. Studies in Semitio Grammar. Elemente der Laut. except in cases of special importance. 66 (1946).AIOK XXIV = Akten des XXIV. third plural feminine -a. 26 (1957). § 16. Introduotion Ii l'etude des langues °semitiques (Paris 1947). 16. 16. Fronzaroli. SpraohwissensohajtlioM Untersuohungen zum Semitischen (Leipzig 1907-11). only works which appeared up to the end of 1961 have been considered. (Leipzig 1916).. The abbreviations used in the bibliography are those of the Bibliographie semitique published periodically in Orientalia. in the text. In Ethiopic long final vowels are substituted in the perfect endings of the second singular masculine (-ka) and in the first person plural (-na). H. Introduotion to Semitio Comparative Linguistics (New York 1934). 2 vol. in JAOS 35 (1917). in Or 16 (1947). In Hebrew the Proto-Semitic endings of the perfect reappear before suffixes: third singular masculine -a. In Akkadian the pronominal suffixes are appended to verbal forms without alteration. .140. first plural -na. second plural feminine -a.14. ... 13. L. 73 (1953). 28 (1959). The suffix-pronouns can be appended directly to the forms of the imperfect in the case of the second person only (type yiqborka). 8. Levi Della Vida.-16. Y9siminnak "he puts thee". 22 (1953). In the second person plural masculine we have tii instead of tem. H. Brockelmann. For further details cf. 16 (1961). In Arabic long final vowels appear in the perfect endings in the second singular feminine (type qabartini beside qabartini) and in the second plural masculine (type qabartumiini). 65 (1945). In those persons of the imperfect which have no afformatives a connecting vowel i appears (type neqb9riw) which may occur also in the imperative.. 348-80. (Berlin 1908-13). Bibliography The following bibliography includes works published up to the end of 1962. likewise in the imperative: second plural masculine -ii. Inclusion of a work in this bibliography does not necessarily connote agreement with its theses.. For forms with -an before the suffixes (like yiqqal}ennii) cf. second singular masculine -ta. R. 2 vol. 212-18. some endings take the ventive morpheme -an before the suffixes-and this fact has a certain comparative interest. 42*-61*. probably as a result of shortening of the Proto-Semitic -tumii.27. 91-103. Gray. P. in ANLR Ser.138. second singular feminine -ti.. .139.und Formenlehre (Berlin 1908). §§ 13. 2. 1 *-38*. 109-18. the same occurs in the imperative (type qobr~ni).. 16..142. Aufl.. 30 (1961). .. Internationalen Orientalisten-Kongr6ss6s (Wiesbaden 1959). The Semitic Languages . 111. however. ). 62 (1942). for the suffixes of the other persons a connecting vowel is inserted on the model of the verbs with third radical y (type yiqb9r~ni). third singular feminine -at. J. Langues et ecritures semitiques (Paris 1930).

AufI. III. P.. A.. 256-67. Dammron. 344-72.. in Language 17 (1941). 1. Beitriige zur sernitischen Sprachwissenschajt (Strasbourg 1904). R. Der hymnisch-epische Dialekt des Akkadischen. Ungnad.. W. Grammaire de l'arameen biblique (Strasbourg 1961). Pedersen. . Ebeling. 1. Wright. W.. Gesenius.. . . 163-227. J. A. Phonizisch-punische Grammatik (Roma 1951). Noldeke. H. Gordon. Bauer. (Berlin 1952-55). W. G. in Or 28 (1958). Rinaldi. BohI. 8 (1958).. (Leipzig 1899). Studien zur 1!'rage der Sprachverwandtschaft des Ugaritischen.. J. L. Cantineau. von B. L. L'aramaico antieo (ANLM. J.. Once again the Proto-Sinaitic Inscriptions. Nuo1)o materiale per la grammatica dell'aramaico antico. in AIOK XXIV. .. Renan. The Linguistic Position of Ugaritic in the Light of Recent Discoveries. . _. 14-50. Ullendorff. Grammatik des Akkadischen.. V (Paris 1957). in RSO 34 (1959). Dechiflrement des inscriptions pseudo-hieroglyphiques de Byblos. und Bergstrasser. (Rome 1947). Ilsemitico di nord-ovest (Napoli 1960). Histoire generale et systeme compare des langues semitiques. 66-75. (Paris 1930-32). Supplement. Finet. M. E. 51-98.. H.ication of the North-West Semitic Languages..172 Bibliography Bibliography 173 Handbuch der Orientalistik. Die semitischen Sprachen. E.. Das Verbum der El-Amarna-Briefe (Leipzig 1910). Gelb. Janssens. in ArOr 28 (1960). E. Grammatik des Biblisch-Aramiiischen (Halle 1927).. Contribution au dechiffrement des inscriptions pseudo-hUroglyphiques de Byb/08.. S.. in ZA 40 (1932). Segert.. P. Birkeland. La lingua degli Amoriti. 8. Beer. (Leipzig 1955). 2 vol... (Miinchen 1949).. 7. W. :1\'1.. G. F. Morphology of Akkadian (Chicago 1952). M. F. I (Paris-Gembloux 1959). 3.. 177-91. Untersuchungen zur Grammatik des Ugaritischen (Berlin 1954). G.-. H. Harris. Epstein. E. Grammatik des judisch-paliistinischen Aramiiisch. R. Salonen. in Acta Orientalia 7 (1957). Zur Einteilung der semitischen Sprachen. in ANLR ser. 236. B. in WZKM 56 (1960). W. Burchardt. _. XII (Berlin 1928). C. Vergleichende Grammatik der semitischen Sprachen (Berlin 1898). in JEA 48 (1962). Syrische Grammatik. Is Uga-rit'ic a Canaanite Dialect?. Dahood... und Leander. in ArOr 29 (1961)... Considerations on Semitic Comparative Lexicography.. Semitistische Marginalien. 1-35. Bauer. 7. 127-38. Gardiner. Langues et ecritures semitiques. Altheim. Dhorme. A Grammar of Babylonian Aramaic (Jerusalem 1960). G. in RB 22 (1913). de Meyer... 29. Old Akkadian Writing and Grammar (Chicago 1961)... 369-93. H.1. 5) (Roma 1956). E. Z. Le lingue semitiche (Torino 1954). fasc. . J.). vol. 23 (1914). L'accadien des contrats de Suse (Leiden 1962). Friedrich. North-West Semitic Aistieitner. in BASOR 110 (1948). J. Ugaritic Manual (Roma 1955). The Classi/.. und Stiehl. Levi Della Vida.. M. Spuler. 143-64. 2. J.. The Language of Jesus (Oslo 1954). F.. 2 vol. P.. Albright. A.. The Early Alphabetic Inscriptions from Sinai and their Decipherment.. in La n01welle CUo 7-9 (1955-57). Neue Beitriige zur semitischen Sprachwissenschaft (Strasbourg 1910). in Syria 25 (1946-48). H.. What is a Semitic Language?. 6--22. Die Sprache der Amarnabriefe mit besonderer Beriicksichtigung der Kanaanismen (Leipzig 1909). 220-23. North-East Semitic Aro. H. T. Brockelmann. hrsg. Le nabateen. Historische GrammaUk der hebriiischen Sprache (Halle 1922). C. (Paris 1863).. in Reallexikon der Vorgeschichte.. 41-54. 13 (1958). _ . Aufl. Hebriiische Grammatik. von Soden. Th. Linguistica semitica: presente e fUtUTO (Roma 1961).. Gelb. Lectures on the Comparative Grammar of the Semitic Languages (Cambridge 1890). G. H. 269-79. Untersuchungen zur Schrift und Sprache des Altbabylonischen von Susa (Helsinki 1962). und Leander. von Soden. L'accadien des lettres de Mari (Bruxelles 1956). Ann. (and others). N. La langue de Canaan. in Sacra Pagina. Development of the Canaanite Dialects (New Haven 1939). A. De semietische talen (Lei den 1949). Die altkanaaniiischen 1!'remdworte und Eigennamen im Agyptischen (Leipzig 1909-10). 2. Goetze. 80-U8. A Grammar of the Phoenician Language (New Haven 1936). (Leipzig 1918-29). Die aramiiisehe Sprache unler den Achaimeniden (Frankfurt 1960ff. S. Studien zur mittelbabylonischen Grammatile (Helsinki 1955).. Garbini. 37-59. A. Goshen-Gottstein. 470-87. H. 90-183. 251-307. 45-48. J.. (Leipzig 1905)... J. Semiten (Sprache). Kanaaniiisch und Westsemitisch. Semitistik (Leiden 1953-54). Hebriiische Grammatik. J. Grammaire d1~ palmyrenien epigraphique (Le Caire 1935). Zimmem. ser. 41 (1933).. GrundrifJ der akkadischen Grammatile (Roma 1952)... 2eme ed. Grammaire de l'hebreu biblique. H. Kramers.. Linguistic Structure and Tradition in the Qumran Docmnents (Jerusalem 1958)... 3eme ed. --. . G. Dalman. . 8.. und Meyer. Ginsberg. in Dictionnaire de la Bible.. in Scientia 84 (1949). AufI. G. 361-77 Joiion. C. Aufl. 257-334. _.

Laut. in Or 30 (1961). 154-60. Invita. 197-214. A Grammar of Biblical Aramaic (Wiesbaden 1961). M. H. W. 23-38. The Ethiopian Languages and their Contribution to Semitic Studies. Praetorius. J. . Indo-European Cohen. The Study of the Aramaic Grammar of the Babylonian Talmud. (Louvain 1956). 270-74. trans!. AufI. H.. in JSS 7 (1962). Le vocabulaire de base semitique et le classement des dialectes meridionaux. Le8 inscriptions protosinaUiques. in Sem II (1961). A ~tammar of Mishnaic Hebrew. Studies in Modern Syriac. P. Margolis. in SOLDV II... in Or 20 (1951). in al-1I1achriq 52 (1958). KurzgefafJte Syrische Grammatik. Altsudarabische Grammatik (Leipzig 1943). in Les langues du monde.. Ie consonantisme et la formation des moines en « N ostratiq1te . Eng!. (Oxford 1958)... A. in JAOS 63 (1943). M. E. Moscati. 250-71. Arabische Grammatik.. _. _. A Study in the Structure of Arabic. Accadien et sudarabique.). A. (Leipzig 1960). Dillmann. 2.. French transl. W. Leslau. in The Bible and the Ancient Near East (London 1961). Untersuchunge'n zur arabischen Sprach. South-East Semitic (Ethiopic and South-Arabic). Moran.. Les inscriptions thamoudeennes (Louvain 1950). W. etiopico. Sulla posizione linguistica del semitico nord-occidentale. Rossell. by J.und Stilge8chichte (Berlin 1950). 339-83. _. Die aramaistische Forschung scit Th. in Orientalia Pragensia 1 (1960). in Or 31 (1962). 339-51. in BSLP 33 (1932). F. D. 3eme ed. Il semitico di nord-ovest. 1-32. L. NOldeke. Le dechiffrement de8 inscriptions protosinaitiques. 197-268. M. Beeston. 55-84.. Die Sprache der palmyrenischen Inschriften und ihre Stellung innerhalb des Aramaischen (Leipzig 1936).. J... (Paris 1952). (Leipzig 1898). in JSS 6 (1961). . Essai comparatif sur le vocabulaire et la phonetique du chamitosemitique (Paris 1947). van den Branden. Recherches sur Ie vocalisme.. in Africa 25 (1955). in AlOK XXIV. Ugaritic Marginalia. Aufl. in BO 17 (1960).. in OrAn 1 (1962). 121--25. .. Studies in Galilean Aramaic (Jerusalem 1952). by Cl. M. 218-22. . 202-21. 226-45. Ullendorff. S. 175--204. Proto-Semitic. Cantineau. Fleisch. L'arabe classique. Y. W. sudarabico. The Position of Ugaritic within thc Framework of the Semitic Languages. Lehrbuch der aramiiischen Sprache des babylonischen Talmuds (Miinchen 1910). C.. Leander.und Formenlehre des Agyptisch-AramiUschen (Goteborg 1928). The Afro-Asiatic (Hamito-Semitic) Present. von C. R. in JAOS 72 (1952). 4-14.. C. 2nd ed. Littmann. Crichton (London 1907). The Proto-Byblian Inscriptions: a Fresh Approach. Segal. Revision and Reclassification of the Proto-Byblian Signs. 2eme ed. R... litteratura.. Polotsky. F. J.. in Tarbiz 24 (1954-55). E. 361-95. Lihyan und Lihyanisch (Kiiln-Opladen 1954)... in JSS 6 (1961). Esquisse d'une structure linguistique (Beyrouth 1956). A Preliminary Report after Re-Examination of the Byblian Inscriptions.. 2 vol.. R. Les textes thamoudeens de Philby. Gaudefroy-Demombynes. Mandaische Grammatik (Halle 1875). Denizeau (Paris 1955).tion Ii l'elude comparative des langues indoeuropeennes et des lungues chamilo-sernitiques (Bordeaux 1946). Ancient West Arabian (London 1951). Trait!! de philologie arabe. J. A Handbook of Aramaic Magical Texts (Ringwood Borough 1953). Bezold (Leipzig 1899). _... Die Sprache der moabitischen Konigsinschrift. chrestomathia cum gwssario (Berlin 1873). Ullendorff. Brevis Linguae Samaritanae grammatica. ancetre de l' I ndo-Europeen et du Chamito. E. E.. 54-72. 14. Stevenson. _. in ArOr 29 (1961). Segert. K. W. Hamito-Semitic.. Th. IV. H. Caskel. A Descriptive Grammar of Epigraphic South Arabian (London 1962). Petermann. M..S emitiql1e (Paris 1943). Cuny. Petracek. The Position of Ethiopic in Semitic: Akkadian and Ethiopic.. H. B. Flick.. Rabin. Rowley. 33-39. 2.46-78. Grammatik des christlich-palastinischen Aramaisch (Tubingen 1924). Cohen.. -. South-West Semitic 175 Kutscher. A. et Blachere. F. L. Brockelmann. The Aramaic of the Old Testament (London 1929).. Greenberg. 229-34. Sobelmann.. S.174 Bibliography Bibliography D. 149-83. (Paris 1952). _. Semitic Inscriptions C. Langues chamito-semitiques. Anciennes inscriptions 8 emitiques . I (Beyrouth 1961). Nordarabico. E. A. Grammatik der Athiopischen Sprache. in Leshonenu 26 (1962). van den Branden. _. Safa'itic Inscriptions (Leyden 1943).. Martin. Moscati. Noldeke's Veroffentlichungen (Leiden 1939). Ugaritic Marginalia II.. S.. _.. Grammar of Palestinian Jewish Aramaic (Oxford 1924). A. Rosenthal. Grammatica Aethiopica (Karlsruhe-Leipzig 1886). The Hebrew Language in its Northwest Semitic Background. 81--181. _. A. R.. L. Schulthess. Grammaire de l'arabe classique. HOfner. F. M. 'Arabiya. in RSO 34 (1959). 1-9. in RSO 31 (1956). 251-53. R. AufI. Syria.. H.

in JSS 7 (1962). in BIFAO 58 (1959). 15 (1960). R. Phonology of the Epigraphic South Arabian Unvoiced Sibilants. Remarques sur le consonantisme semitique.. Essai d'une phonolog"ie de l'hebreu biblique. in ArOr 20 (1952).. Comparative Grammar n. Bauer. 93-140.. Y'll. 446-52.. 84-90. A. 103-274. Broch (Oslo 1947). 1-26. The Apparent Interchange between a and i in Hebrew. Preliminaries Mittwoch. L. Knudsen. Das akkadische Syllabar (Rom 1948). H. 137-49.. S. Camito-se'lnitico e indoeuropeo (Bologna 1949). B. . in RSO 35 (1960). o. Histoire de l'ecriture. Rossler. 67-78. E. 207-17. 76-83. L'origine des alphabets protosinaftique. 82-122. Ullendorff.. 338-49. F~rguson..... E. Klingenheben. L. 12 . ' --. Petracek. ... 8. 366-73. W... T':e Emphatic 1 in Arabic. H.. Is Egyptian a Semitic Language?. C. 49-72. 211-77. Etudes de phonetiq1w arabe. fasc. antic a storia delle lingue semitiche. von Soden.. in JCS 15 (1961). Blake. Diringer. Mayer.. F. 8. 509-23. Le consonantisme du semitique. J. Fronzaroli. M. A. 173-76.arcel Cohen. J. 0.. A Key to the History of Mankind (London 1948). in BO 19 (1962). vol. Cours de phonetique arabe (Paris 1960).. S. m. Birkeland. in ArOr 23 (1955). van den Branden. W. ... Studies in the Ethiopic Syllabary. Gedanken zur agyptisch-semitischen Sprachverwandtschaft. Phonology A. in Festskrift til Prof. in ZDMG 100 (1950). Esquisse d'une phonologie de l'arabe classique. in BSLP 43 (1946). in RANL ser. . 8) (Roma 1954). in Or 20 (1951). 97-130. Pfeiffer. in ArOr 21 (1953). 101--107.. Vycichl. Moscati. La palatalisation «spontanee >. 313-23. Cantineau. en sudarabique.. Moscati. Gelb. arabes preislamiques et phenicien.27-44. H. Etudes de grammaire gourague. J. in JQR 48 (1957).. 461-514. -. in Language 32 (1956).. F.r. A.. in Sem 4 (1953). in ZS 10 (1935). La grande invention de l' ecriture et son evolution. 198-206. A. Il sistema consonantico delle lingue semitiche (Roma 1954).. 79-94.. Beeston. Pilszczikowa.176 Bibliography Bibliography 177 Heilmann. A. Sperber. P. in Mus 73 (1960). . The Syriac Phonematic Vowel System. 475---78. Der semitische Charakter der libyschen Sprache. Clues to the Pronounciation of Ancient Languages. ~i~ Struktur der semitischen Wurzelmorpheme und der Ubergang 'ain> ga2n und 'ain > r im Arabischen. Fevrier.. 121-50.. Verbalbau und Verbalflexion in den semitohamitischen Sprachen.1.. in MUSJ 28 (1949-50). Die semitischen Zischlaute (1). in Africa 21 (1951).Hebrew. Cases of Free Variants in the Akkadian q Phoneme. in MIOF 4 (1956). Ungnad. Sulla piu. Akkadisches und libysches Verbum. 8 und s und ihre Vertretung im Akkadischen. LaSo. in JBL 62 (1943). . II-13. Le haoussa et Ie chamito-semitique a la lumiere de l'Essai comparatif de JJf. Das personliche Furwort und die Verbalflexion in den chamitosemitischen Spruchen (Wien 1909).. in Melanges GaudefroyDemombynes (Le Caire 1935-45). Morag. D.. E. 5. A Study of Writing (London 1952). Die traditioneZle Aussprache des Athiopischen (Berlin 1926). Polotsky. Reinisch. in Or 28 (1959). G. (London 1954). The Relationship of the Semitic and Egyptian Verbal Systems (Oxford 1954). 13-39. 240--62. Nouveaux aspects de la langue egyptienne. The Sibilants in Old South Arabic. Martmet. Fleisch. The Alphabet. G. F. 225-85. 3 vol. Arabian Sibilants. Zur Artikulation des sogenannten emphatischen lim Arabischen. .. The Sibilants of Old Babylonian. A. FIscher. L... in BSLP 39 (1938).. 137-262. in HUCA 12-13 (1937-38). Rev.. in Kush 7 (1959).. W. . The Phonological System Aro.. in BSLP 49 (1953). ser.. in Rocznik Orientalistyczny 24 (1960). Der doppelte phonologische Charakter des Ghain im klassischen Arabisch. N. Semitic Writing from Pictograph to Alphabet. R. Hebrew Grammar: a New Approach. 137-75. Thacker...andAramaic ('S Gravenhage 1962). Sulla ricostruzione del protosemitico.. Goetze.. in Acme 13 (1960). The Vocalization Systems of Arabic. .. A. 1951. (Paris 1959). in JNES 9 (1950). S. La «mutation des sifflantes >.. H. 79-10l. Ricerche sul problema dei rapporU fra lingue indouropee e lingue semitiohe. Wechsel von p. M. A. 25---38. . R. . in RA 52 (1958)..8. (Paris 1958). L.. de g en arabe. in BSLP 46 (1950). S.. La fonetica ugaritica (Roma Hi55).und die Suffixkonjugationen im Hamitosemitischen. ed. ed. in ZA 50 (1952). T. in Transactions of the Philological Society. J. Nouv. Language and Script Cohen. Hebrew Based Upon Greek and Latin Transliterations. . in Miinchener Studien zur Sprachwissenschaft 8 (1956). Das Wesen des Ursemitischen (Leipzig 1925). 222-33. b mit'!! im Ara1naischen und Arabischen. 321-35. . K > S in den sudlichen semitischen Sprachen. Driver. J. . Moscati. K.. . 1-10. W. in SOLDV II. Writing (London 1962).. . Preistoria e storia del consonantismo ebraico antico (ANLM. Die Prajix. 161-73. in BSLP 54 (1959) 90--102. G. 77-100.

178

Bibliography

Bibliography

179

Rabin, C., The Hebrew Development of Proto-Semitic ii, in Tarbiz 30 (1960), 99-111. Reif, J. A., The Loss of Consonantal Aleph in Ugaritic, in JSS 4 (1959),16-20. Rosen, H. B., A Marginal Note on Biblical Hebrew Phonology, in JNES 20 (1961), 124-26. - , ha-'Ibrit se·liinu (Tel Aviv 1956). Rossler, 0., Ghain im Ugaritischen, in ZA 54 (1961), 158-72. - , Zur Frage der Vertretung der gemeinsemitischen Laryngale im Akkadischen ( \ = y), in AIOK XXIV, 129-32. Ruzicka, R., La question de l'existence du if dans les langues semitiques en general et dans la langue ugaritienne en particulier, in ArOr 22 (1954), 176-237. Stehle, D., Sibilants and Emphatics in South Arabic, in JAOS 60 (1940), 507-43. Ullendorff, E., The Semitic Languages of Ethiopia. A Comparative Phonology (London 1955). Vilencik, Y., Welchen Lautwert hatte g, im Ursemitischen'l, in OLZ 33 (1930), 89-98. von Soden, W., Aramaischesl} ersoheint im Spatbabylonischen vor mauch als g, in AfO 19 (1959-60), 149. Vokalfarbungen im Akkadisohen, in JCS 2 (1948), 291-303. C. Conditioned Phonetic Changes Bravmann, M. M., Some Aspeots of the Development of Semitic Diphtongs, in Or 8 (1939), 244-53; 9 (1940), 45-60. Cowan, W., Arabio Evidence for Proto·Semitio ""/awa/ and ""/6/, in Language 36 (1960), 60-62. Ruzicka, R., Konsonantisohe Dissimilation in den semitischen Sprachen (Leipzig 1909). Speiser, E. A., Secondary Developments in Semitic Phonology. An Application of the Prinoiple of Sonority, in AJSL 42 (1926), 145-69. Spitaler, A., Zur Frage der Geminatendissimilation im Semitischen, in Zeitsohrift fur Indogermanisohe Forsohungen 61 (1954), 257-66. D. Syllable and Accent Birkeland, H., Akzent und Vokalismus im Althebraischen (Oslo 1940). - , Altarabische Pausalformen (Oslo 1940). - , Stress Patterns in Arabic (Oslo 1954). Blake, F. R., Pretonic Vowels in Hebrew, in JNES 10 (1951), 243-55. Brockelmann, C., N euere Theorien zur Gesohichte des Akzents und des VokalismU8 im Hebraischen und Aramaischen, in ZDMG 94 (1940), 332-71. Bmnno, E., Studien uber hebraische Morphologie und V okalismus (Leipzig 1943). Bush, F. W., Evidenoe from Mill}amah and the Masoretic Text for a Penultimate Accent in Hebrew Verbal Forms, in Revue de Qumran 2 (1959-60), 501-14.

Can~ineau, J:, D~ ,la place de l' acoent de mot en hebreu et en arameen biblique, III Bullet~n d etudes orientales 1 (1931), 81-98. - , Elimination des syllabes breves en hebreu et en arameen biblique, in Bulletin d'etudes orientales 2 (1932), 125-44. Goetze, A., Accent and Vocalism in Hebrew, in JAOS 59 (1939), 431-59. Sarauw, C., Uber Akzent und Silbenbildung in den alteren semitischen Sprachen (Kopenhagen 1939). Segal, J. B., The Diacritical Point and the Aocents in Syriac (London 1953).

III. Morphology
A. Preliminaries Botterweck, G. J., Der 'l'riliterismus im Semitisohen (Bonn 1952). Brockelmann, C., Senbitische Keimwortbildungen, in ZS 5 (1927), 6-38. Greenb~rg, J. H., The Patterning of Root Morphemes in Semitio, in Word 6 (1900), 162-81. Heller, J., Neuere Literatur zur Biliterismus-Frage, in ArOr 27 (1959) 678-82. ' Huizi~ga, A. H., Analogy in the Semitio Languages (Baltimore 1901). HurwItz,. S. T. H., Root·Determinatives in Semitic Speech (New York 1913). KurylowlCz, J., L'apophonie en semitique (Warszawa 1961). Moscati, S., n biconsonantismo nelle Zingue semitiche, in Bib128 (1947),113-35. Petracek, K., Die innere Flexion in den semitischen Spraohen, in ArOr 28 (1960), 547-606; 29 (1961), 513-45; 30 (1962), 361-408 (wird fortgesetzt). B. The Noun Aartun, K., Zur Frage des bestimmten Artikels im Aramaischen, in Aota Orientalia 24 (1959), 5-14. Barth, J., Die Nominalbildung in den semitischen Spraohen, 2. Aufl. (Leipzig 1894), Baumgartner, W., Das hebraische Nominalprafix mi-, in ThZ 9 (1953) 154-57. ' Bravmann, M. M., Genetic Aspects of the Genitive in the Semitic Languages in JAOS 81 (1961), 386-94. ' - , On a Case of Quantitative Ablaut in Semitic, in Or 22 (1953), 1-24. - , The Plural Ending ·ut- of Jlasculine Attributive Adjectives in Akkadian in JCS 1 (1947), 343. ' Brockelmann, C., Diminutiv und Augmentativ im Semitischen, in ZS 6 (1928) 109--34. ' Cantineau, J., La notion de scheme et son alteration dans diverses langues semitiques, in Sem 3 (1950), 73-83. Cazelles, H., La mimation nominale en Ouest-Semitique, in GLECS 5 (1951), 79-81. Christian, V., Die Entstehung der semitisohen Kasusendungen, in ZS 3 (1924), 17-26.
Moscati, Comparative Grammar

12a

180

Bibliography

Bibliography

181

Cohen, M., Noms d' anima1~X et de plantes aprefixe n en ethiopien, in GLECS 5 (1951), 85-87. Entretiens sur la determination et l'indetermination, in GLECS 5 (1948-51), 73-76, 78, 81-82, 88-96, 98. Feghali, M. et Cuny, A., Du genre grammatical en semitique (Paris 1924). Fleisch, R., Le nom d'agent fa'al, in MUSJ 32 (1955), 165-72. Friedrich, J., Der Schwund kurzer Endvokale im Nordwestsemitischen, in ZS 1 (1922), 3-14. Gelb, 1., La mimazione e la nunazione neUe lingue semitiche, in RSO 12 (1930),217-65. Goetze, A., The Akkadian Masculine Plural in -anUfi and its Semitic Background, in Language 22 (1946), 121-30. Kurylowicz, J., La mimation et l'article en arabe, in ArOr 18, 1-2 (1950), 323--28. Lagarde, P. de, Ubersicht iiber die im Aramaischen, Arabischen und Hebraischen iibliche Bildung der Nomina (Gottingen 1889). Lemoine, E., Theorie de l'emphase hebra~que (Paris 1951). Leslau, W., Ethiopic Denominatives with Nominal Morphemes, in Mus 75 (1962), 139-75. Loretz, 0., Die hebraische Nominalform qattal, in Bibl 41 (1960), 411-16. Matou8, L., Zum sog. inneren Plural im Arabischen, in ArOr 24 (1956), 626-30. Moscati, S., n plurale esterno maschile neUe lingue semitiche, in RSO 29 (1954), 28-52. - , Lo stato assoluto deU'aramaico orientale, in AION 4 (1962), 79-83. - , On Semitic Case-Endings, in JNES 17 (1958), 142-44. - , Plurali interni in ugaritico?, in RSO 32 (1957), 339-52. Nyberg, R. S., Wortbildung mit Priifixen in den semitischen Sprachen, in Le monde oriental 14 (1920), 177-289. Rin, S., The Termination -6 in the Plural Absolute, in Leshonenu 25 (1961), 17-19. Speiser, E. A., The "Elative" in West-Semitic and Akkadian, in JCS 6 (1952), 81-92. Studies in Semitic Formatives, in JAOS 56 (1936), 22-46. - , The Terminative-Adverbial in Canaanite- U garitic and Akkadian, in IEJ 4 (1954), 108-15. Tagliavini, C., Alcune osservazioni sul primitivo valore della mimazione e nunazione nelle lingue semitiche, in Donum natalicium Schrijnen (Chartres 1929), 240-90. Torczyner, R., Die Entstehung des semitischen Sprachtypus, I (Wien 1916). Troupeau, G., Le scheme de pluriel Fu'lan en arabe classique, in GLECS 7 (1955), 65-66. von Soden, W., Status rectus-l?ormen vor dem Genitiv im Akkadischen und die sogenannte uneigentliche Annexion im Arabischen, in JNES 19 (1960), 163-71. Wagner, E., Die erste Person Dualis im Semitischen, in ZDMG 102 (1952), 229-33.

Wehr, R., Der arabische Elativ (Wiesbaden 1953). Young, E. J., Adverbial -u in Semitic, in Westminster Theological Journal I?, 2 (1951), 151-54. C. The Pronoun Barth, J., Die Pronominalbildung in den semitischen Sprachen (Leipzig 1913). Bravmann, M. M., Hebrew stayim in the Light of Syriac and Turkic, in Proceedings of the American Society for Jewish Research 21 (1952), 1-2. Caspari, W., Zum hebraischen Demonstrativ, in ZS 7 (1931), 41-52. Castellino, G., Observations on the Akkadian Personal Pronouns in the Light of Semitic and Hamitic, in MIOF 5 (1957), 185-218. - , The Akkadian Personal Pronouns and Verbal System in the Light of Semitic and Hamitic (Leiden 1962). 'Greenberg, R., An Afro-Asiatic Pattern of Gender and Number Agreement, in JAOS 80 (1960), 317-21. Kienast, B., Das Personalpronomen der 2. Person im Semitischen, in AlOK XXIV, 253--55. - , Erwagungen zu einer neueren Studie iiber semitische Demonstrativa, in Or 26 (1957), 257-68. Loewenstamm, S. E., On Ugaritic Pronouns in the Light of Canaanitic, in Leshonenu 23 (1958-59), 72-84. Rosen, H.B., Zur Vorgeschichte des Relativsatzes im Nordwestsemitischen, in ArOr 27 (1959), 186-98. Rundgren, F., Uber Bildungen mit (s)- und n-t-Demonstrativen im Semitischen (Uppsala 1955). Trager, G. L. and Rice, F. A., The Personal Pronoun System of Classical Arabic, in Language 30 (1954), 224--29. D. The Numeral Barth, J., Zur Flexion der semitischen Zahlworter, in ZDMG 66 (1912), 94--102. Bauer, R., Noch einmal die semitischen Zahlworter, in ZDMG 66 (1912), 267-70. Cantineau, J., Le nom de nombre «six» dansleslangues semitiques, in Bulletin des etudes arabes 13 (1943), 72. Loewenstamm, S_ E., The Development of the Term "First" in the Semitic Languages, in Tarbiz 24 (1954-55), 249-51. Reckendorf, R., Der Bau der semitischen Zahlworter, in ZDMG 65 (1911), 550-59. von Soden, W., Die Zahlen 20-90 im Semitisohen und der Status absolutus in WZKM 57 (1961), 24--28. ' E. The Particles Eitan, 1., Hebrew and Semitic Farticles, inAJSL 44 (1928), 177-205,254--60; 45 (1929), 48--63, 130-45, 1117-211; 46 (1930), 22-50. Garbini, G., La congiunzione semitica *pa-, in Bibl 38 (1957), 419--27.

182

Bibliography

Bibliography

183

Guidi, I., Particelle interrogative e negative neUe lingue semitiche, in A Volume of Studies presented to E. G. Browne (Cambridge 1922), 175-78. Sarna, N. M., The Interchange of the Prepositions beth and min in Biblical Hebrew, in JBL 78 (1959), 310-16. F. The Verb Ahrens, K., Der Stamm der schwachen Verba in den semitisc'hen Sprachen, in ZDMG 64 (1910), 161-94. Bauer, H., Die Tempora im Semitischen (Berlin 1910). Blake, F. R., A Resurvey of Hebrew Tenses (Roma 1951). Bravmann, M. :M., Notes on the Forms of the Imperative in Hebrew and Arabic, in JQR 42 (1961), 51-56. Brockelmann, C., Die" Tempora" des Semitischen, in Zeitschri/t fur Phonetik 5 (1951), 133-54. Brzuski, W. K., Note sur les themes Ii seconde radicale graphiquement redoublee en sudarabique epigraphique, in Rocznik Orientalistyczny 25 (1961), 127-31. Christian, V., Das Wesen der semitischen Tempora, in ZD:MG 81 (1927), 232-58. Cohen, :M., Le systerne verbal semitique et l'expression du temps (Paris 1924). Dahood, M., Some Aphel Causatives in Ugaritic, in Bib138 (1957), 62-73. Dombrowski, B. W. W., Some Remarks on the Hebrew Hitpa'el and Inversative -t- in the Semitic Languages, in JNES 21 (1962), 220-23. Driver, G. R., Problems of the Hebrew Verbal System (Edinburgh 1936). Fleisch, H., Les verbes Ii allongement vocalique interne en semitique (Paris 1944). - , Sur le systeme verbal du semitique commun et son evolution dans les langues semitiques anciennes, in :MUSJ 27 (1947-48), 39-60. Goetze, A., The So-Called Intensive of the Semitic Languages, in ,lAOS 62 (1942), 1-8. - , The Tenses of Ugaritic, in JAOS 58 (1938), 266-309. Hammershaimb, E., Das Verbum im Dialekt von Ras Schamra (Kopenhagen 1941). Heidel, A., The System of the Quadriliteral Verb in Akkadian (Chicago 1940). Janssens, G., De werkwoordelijke "Tijden" in het semietisch, en in het bizonder in het hebreeuws, in JEOL 15 (1957-58), 97-103. Jirku, A., Eine 'Aj'e/.Form im Ugaritischen?, in AfO 18 (1957), 129-30. Kienast, B., Das Punktualthema *yaprus und seine Modi, in Or 29 (1960), 151-67. - , Verbalformen mit Reduplikation im Akkadischen, in Or 26 (1957), 44-50. - , Weiteres zum R·Stamm des Akkadischen, in JCS 15 (1961), 59--61. Kurylowicz, J., Esquisse d'une tMorie de l'apophonie en semitique, in BSLP 53 (1957-58), 1-38. - , Le systeme verbal du semitique, in BSLP 45 (1949), 47-56. Leslau, W., Le theme verbal frequentatif dans les langues ethiopiennes, in RES 1939, 15-31. - , Le type verbal qat ala en ethiopien meridional, in MUSJ 31 (1954). 15-98.

Martin, 'V. J., Some Notes on the Imperative in the Semitic Languages, in RSO 32 (1957), 315-19. Mey~r, R., Das hebraische Verbalsystem im Lichte der gegenwartigenForschung, III Congress Volume Oxford 1959 (Leiden 19(0), 309-17. - , Spuren eines westsemitischen Prasens-Futur in den Texten von Chirbet Qumran, in Von Ugarit nach Qumran (Berlin 1958), 118-26. - , Zur Geschichte des hebraischen Verbums, in VT 3 (1953), 225-35. Moran, W. L., Early Canaanite yaqtula, in Or 29 (1960), 1-19. Moscati, S., Il participio passivo in semitico, in RSO 37 (1962), 51-57. Miiller, W. W., Die Wurzeln Mediae und Tertiae y!w im AZtsudarabischen (Tiibingen 19(2). Praetorius, F., Zur Kausativbildung im Semitischen, in ZS 5 (1927), 39-42. Rossler, 0., Die Priifixkonjugation Qal der Verba 1M Nun im Althebraischen . und das Problem der sogenannten Tempora, in ZA W 74 (1962), 125-41. - , Eine bisher unbekannte Tempusform im Althebraischen, in ZDMG III (1961), 445-51. Rowton, M. B., The Use of Permansive in Classic Babylonian, in JNES 21 (1962), 233-303. Rundgren, F., Das althebraische Verbum. AbrifJ der Aspektlehre (Uppsala 19(1). - , Das aUsyrische Verbalsystem, in Sprdkvetenskapliga Sallskapets i Uppsala Forhandliger 1958-60, 49-75. - , Der aspektuelle Charakter des aZtsemitischen Injunktivs, in Orientalia Suecana 9 (1960), 75-101. --.:, Intensiv und Aspekt-Korrelation. Studien zur athiopischen und akkadischen Verbalstammbildung (Uppsala-Wiesbaden 1959). Saydon, P. P., The Conative Imperfect in Hebrew, in YT 12 (1962), 124-26. Sola-Sole, J. M., L'infinitif semitique (Paris 19(1). Speiser, E. A., The Durative Hithpa'el: a tan-Form, in JAOS 75 (1955), 118-21. Stinespring, W. F., The Active Infinitive with Passive Meaning in Biblical Aramaic, in JBL 81 (1962), 391-94. Vycichl, W., Ein passives Partizip qatil im Agyptischen und Semitischen, in ZD:MG 109 (1959), 253-57. von Soden, W., Tempus und Modus im Semitischen, in AIOK XXIV, 263-65. - , UnregelmafJige Verb en im A7ckadischen, in ZA 16 (1952), 163-81. 'Wernberg-:Moller, P., Observations on the Hebrew Participle, in ZA'iV 71 (1959), 54-67.

A Hebrew and English Lexicon of the Old Testament (Oxford 1906). GrundrijJ der vergleichenden Grammatik der semitischen Sprachen.. Lexicon Linguae Aethiopicae (Lipsiae 1865. Ugaritic Manual (Roma 1955). TPA = Fleisch. R. ed. Ullendorff. unter Benutzung des lexikalischen N achlasses von Bruno MeijJner bearbeitet (Wiesbaden 1959 ff. many reprints). H. The Assyrian Dietionary of the University of Chicago. Assyrisches Handw6rterbuch (Leipzig 1896). A Dictionary of Modern Written Arabic. Arabic-English Lexicon.P. Auf Grund der Sammlungen von A. 2 vol. Garbini.. Unveranderter Neudruck der 1915 erschienenen 17. Quellen hrsg.. Milton Cowan (Wiesbaden 1961). et Baumgartner. Briggs. C. W.. 7. Editorial Board J. von Soden. Jacobsen. Dillmann.1. W. (Halis Saxon. Reprint New York 1954). B. WA = Rabin. Lexicon Syriacum. Traite de philologie arabe. by J. SLE = Ullendorff. Neuhebraisches und Chaldaisches Worterbuch. (Berlin 1894-1905). The abbreviations used for reviews are those of the Bibliographie semitique of Orientalia. North-East Semitic (Akkadian) Delitzsch..). Ancient West Arabian (London 1951). Conti Rossini. Those used for books are as follows: Brockelmann. Levy. Syrische Grammatik. 2 vol.. G.. W. Thesaurus Syriacus. Payne Smith (Oxford 1927).. Rabin.). Fr..184 Lexicography Lexicography. 1830-37). Dalman. Lane. Grebaut. von Frants Buhl. C. C... Freytag. H. GAG = von Soden. L.Landsberger. C. AssyrischEnglisch-Deutsches Handworterbuch. H. E. C.. (Leipzig 1955). Glossary of Old Akkadian (Chicago 1957). A Concise Dictionary ot the Assyrian Language. Akkadisches Handworterbuch. vol. J... bearb. Jean.. Schulthess. E. SNO = Garbini. C. Zorell. (Leipzig 1876-89). Lexicon Hebraicum et Aramaicum Veteris Testamenti (Roma 1940ff.. Aramaisch-Neuhebraisches Handworterbuch zu Targum. Dillmann et edition du Lexique de Juste d'Urbin (1850-1855) (Paris 1952). Gelb.L. Supplement au Lexicon Linguae Aethiopicae de A. 2.). A... Reprint 1938). 2nd ed... Reckendorf u. List of Abbreviations used in the Text B. Il semitico di nord-ovest (Napoli 1960). Lexicon Syropalaestinum (Berlin 1903). North-West Semitic Brockelmann.1. Ethiopic Grammar. GVG = Brockelmann. Fischer. B.. Gordon. W. S. Ch. (Oxon. A. J. durch die Deutsche Morgenlandische Gesellschaft (Wiesbaden 1957ff. Gelb. H. Noldeke. A.. VV... Gesenius. G. F. Old Akkadian Writing and Grammar (Chicago 1952). F. Margoliouth.. 4. South-West Semitic Belot.. H.. G. 1868-97). von Soden.). Worterbuch der Klassischen Arabischen Sprache. 8 vol. F. Reprint 1955). Hebraisches und AramaischesBandworterbuch uber das Alte Testament. A. SCL = Gray. GrundrijJ der akkadischen Grammatik (Roma 1952).. SG = Brockelmann.. Introduction to Semitic Comparative Linguistics (New York 1934). R. Auflage (Berlin-Gottingen-Heidelberg 1950). UM = Gordon.. Fr. et Hoftijzer. Lexicon Arabico-Latinum. OA = Gelb. Brown. Gray. 99-261. Dillmann. 1928).. !lecunda (Halis Saxon. (London 1863-1893.. a. Payne Smith. L. . Wehr. J. Lexicon in Veteris Testamenti libros (Leiden 1953).). G. Th. . J. Brockelmann. S. J. Gelb. Vocabulaire arabe-franl)ais Ii l'usage des etudiants (Beyrouth 1883. A Comparative Phonology (London 1955). Aufl. Driver. ed. Oppenheim (Chicago 1956ff. 2 vol. 4 vol. Koehler. (Berlin 1908-13). Chrestomathia Arabica ]j[eridionalis Epigraphica (Roma 1931). C. Aufl. Th. (Frankfurt 1922. C. W.... List of Abbreviations used in the Text 185 Lexicography A. (London 1907): Fleisch. The Semitic Languages of Ethiopia. Dictionnaire des inscriptions semitiques de l'Ouest (Leiden 1960 ff. Talmud und Midrasch.. Muss-Arnold. Supplementum ad Lexicon in Veteris Testamenti libros (Leiden 1958). I (Beyrouth 1961). EG = Dillmann. W. Supplement to the Thesaurus Syriacus of R.