Skip to main content

Full text of "Comparative grammar of the Semitic languages"

See other formats





Demy 8vo, dark green cloth, gilt. 

ALBERUNI : India. An Account of the Religion, Philosophy, 
Literature, Geography, Chronology, Astronomy, Customs, Laws, 
and Astrology of India, about a.d. 1030. By Dr. Edward C. 

ARNOLD (Sir E.) : Indian Poetry and Indian Idylls. Containing 
" The Indian Song of Songs ", from the Sanskrit of the Gita 
Govinda of Jayadeva : Two books from " The IHad of India " 
(Mahabharata) : " Proverbial Wisdom," from the. Shlokas of 
the Hitopadesa, and other Oriental Poems. 

BARTH (Dr. A.) : The Religions of India. Authorized 

Translation by Rev. J. Wood. 
BIGANDET (B. P.) : Life or Legend of Gaudama, the Buddha 

of the Burmese ; With Annotations, the Ways to Neibban, and 

Notice on the Phongyies or Burmese Monks. 
BEAL (Prof. S.) : Life of Hiuen-Tsiang. By the Shamans 

Hwui Li and Yen-Tsung. With a Preface containing an Account 

of the Works of I-Tsing. 
BEAL (Prof. S.) : Si-Yu-Ki : Buddhist Records of the Western 

World. Translated from the Chinese of Hiuen-Tsiang. 
COWELL (Prof. E. B.) : Sarva-Darsana-Samgraha ; or. 

Review of the Different Systems of Hindu Philosophy. By 

Madhava Acharya. Translated by Prof. E. B. Cowell, M.A., 

and Prof. A. E. Gough, M.A. 
DOWSON (Prof. J.) : Classical Dictionary of Hindu Mythology 

and Religion, Geography, History, and Literature. 

DUTT (R. C.) : Economic History of India under Early British 

Rule ; from the Rise of the British Power in 1757 to Accession of 
Queen Victoria. 

DUTT (R. C.) : Economic History of India in the Victorian 

Age ; from Accession of Queen Victoria to Commencement of 
Twentieth Century. 
EDKINS (Dr. J.) : Chinese Buddhism : A Volume of Sketches, 
Historical and Critical. New and Revised Edition. 

HAUG (Dr. M.) : Essays on the Sacred Language, Writings, 

and Religion of the Parsis. 
ROCKHILL (W. W.) : The Life of the Buddha and the Early 

History of his Order. Derived from Tibetan works in the 
Bkah-hgyur and Bstan-hgyur. Followed by notices on the early 
history of Tibet and Khoten. 
WEBER (Dr. A.) : History of Indian Literature. Translated by 
John Mann, M.A., and Theodore Zachariae, Ph.D. Fourth 

New volumes uniform with above 

BOULTING (Dr. W.) : Four Pilgrims : Hiuen-Tsiang, S»wulf, 
Mohammed Ibn Abd Allah, Ludovico Varthema of Bologna. 

FARBRIDGE (M. H.) : Studies in Biblical and Semitic 

McGOVERN (Dr. W.) : Manual of Buddhist Philosophy. 
Vol. I (Cosmology). 

O'LEARY (Dr. De L.) : Arabic Thought and its place in History. 

O'LEARY (Dr. De L.) : Short History of the Fatimid Khalifate. 

Other Volumes to follow. 






Lecturer in Aramaic, etc., Bristol University ; author of Arabic Thought 
and its Place in History, A History of the Fatimid Khalifate, etc. 





Printed in Great Britain by Steiihen Austin d Sons, Ltd., Heitfcrd. 



Barth, J. Die Nominalbildung in den semitisclien Sprache. 
(2) Leipzig, 1894. 

Pronominalbildung. Leipzig, 1913. 

Brookelmann, C. Grundriss der vergleiclienden Grammatik 
der semitischen Sprachen. Berlin, (i) 1908 ; (ii) 1913. 

Semitisohe Sprachwissenschaft. Leipzig, 1908, 

Cooke, G. A. Textbook of North Semitic Inscriptions. Oxford. 

Corpus Inscriptionum Semiticaruni. Paris, 1881, etc. 

Haupt, P. Oldest Semitic Verb Form. JRAS., N.S., x 

(1878), pp. 244-52. 
Hurwitz. Root Determinatives in Semitic Speech. New York, 

de Lagarde, P. Uebersicbt iiber die in Aram., Arab., und 

Hebr. iibliche BilduQg. Gottingen, 1889. 
Lambert, M. Observations sur la Theorie des Formes 

Nominales {Journal Asiatique, Fev.-Mars, 1890). 
Naville, E. L'evolution de la Langue ^figyptienne et lea 

Langues Semitiques. Paris, 1920. 
Kenan, E. Hist. Gen. des Langues Semitiques. Paris, 1863. 
Wright, W. Comparative Grammar of the Semitic Languages. 

Cambr., 1890. 
VoUers. System of Arabic Sounds. Trans, ix Internal. 

Congress of Orientalists, vol. ii, Lond., 1893 (pp. 130-54). 
Zunmern. Vergleichende Gramm. d. semitischen Sprache. 

Berlin, 1898. 

Arabic (Classical) 

Addai Scher. Les Mots Etrangers Introduits dans I'Arabe. 

Fluegel, G. Die grammatischen Schulen der Araber. 1862. 
Howell, M. S. Grammar of the Classical Arabic Language. 

7 vols. Allahabad, 1880-1911. 
Lane, E. W. Arabic Lexicon, 8 vols. London, 1863-89. 
de Sacy, S. Grammaire Arabe. Paris, 1810. 
Wright, W. Arabic Grammar. (3) Cambridge, 1896-8. 
Zamakshari. Kitab al-Moufassal. Boulaq, a.h. 1291. 


Arabic (Arabian Dialects) 
Bittner, M. Studien . . . Mehri-Sprache, 1909. 
Jalin, A. Grammatik der Mehri-Sprache. 1905. 
Jayakar, A. The Omanee dialect of Arabic. 1889. 

Omanee Proverbs, n.d. 

Landberg, C. de. Etudes sur les Dialectes de I'Arabie 

Meridionale. (i) Hadramout, 1901. (ii) Datinah, 1905-9. 
Mueller, D. H. Die Mehri und Soquotri Spracbe. (Suedarab. 

Expedition.) Vols, iv, 1902 ; vi, 1905 ; vii, 1907. 
Reinhardt. Ein arabiscber Dialekt in Oman und Zanzibar. 

Rhodokanakis, N. Zur Forraenlebre des Mebri. 1910. 
Snouck Huxgronje, C. Mekkaniscbe Spricbworter. 1887. 
Socin. Diwan aus Centralarabien. Leipzig, 1900-1. 
Stace, E. V. An Englisb-Arabic vocabulary. 1893. 

(Syria and Palestine) 

Bauer, L. Das Palaestiniscbe Arabiscb, 1910. 
David. Etude sur le Dialeote Arabe de Damas. 1888. 
Loebr, M. Der vulgararabiscbe Dialekt von Jerusalem. 1905. 
Mattson, E. Etudes Phonologiques sur le Dialecte . . . 
de Beyrouth. 1910. 

— ('Iraq) 

Socin, A. Der arabische Dialekt von Mosul und Maerdin. 

Van Ess, J. Spoken Arabic of Mesopotamia, Oxford, 1917. 

— (Egypt) 

Dulac, H, Contes Arabes en Dialecte de la Haute-Egypte. 

Gairdner, W. H. T. Egyptian Colloquial Arabic. Cambridge, 

Spitta Bey, G. Gramra. des ar. Vulgardialektes von Aegypten. 

Leipzig, 1880. 
Vollers, K. Lehrbuch der aegypto-arab. Umgangssprache. 

Cairo, 1890. 
Willmore, J. S. Spoken Arabic of Egypt. London, 1905. 

— (North Africa) 

Clermont, J. L' Arabe Parle Tunisien. Tunis, 1909. 
Cotelle, H. Le langage Arabe . . . de I'Algerie. 1873. 
Crussard, E., et H. Chergui. Manuel. 19io. 


Delaporte, J. Principes de I'ldiome . . . Alger, 1839. 
Maryais, W. Le Dialecte Arabe Parle a TIcmyen. Paris, 1902. 

Le Dialecte Arabe des Ulad Brahim. Paris, 1908. 

Stumme, H. Grammatik dea tunisisohen Arab. Leipzig, 

Aeabic (Maltese) 

Stiimme, H. Maltesiscbe Studien. Leipzig, 1904. 

Vasalli, M. Grammatica della lingua maltese. Malta, 1827. 

Abyssinian (Ge'ez) 

Dillmann, A, Grammatik der aetbiopiscben Spracbe. Leipzig, 

Cbaine, M. Grammaire Etbiopienne. 1907. 
Praetorius. Aetbiopiscbe Grammatik. Berlin, 1886. 

- — - (Anabaric) 

Afevork, G. Grammatica della lingua amarica. 1906. 
Alone, J. Sbort Manual of tbe Ambaric Language. 1909. 
Armbruster, C. H. Tnitia Ambarica. 1908. 

Gramm. Elementare della Lingua Amarinna. Rome, 1891. 

Mabler, L. Praktiscbe Grammatik der ambariscben Spracbe. 
Wien, 1905. 

(Tigre) .  . , 

Conti-Rossini, C. Docutrenti per lo studio della lingua tigre. 

Perini, R. Manuale teoretico-pratico della lingua tigre. 1893. 


Praetorius, F. Grammatik der Tigrina Spracbe. 1871. 

de Vito. Grammatica elementare della lingua tigrigna. Rome, 


MiN^AN ; Sab^an 

Inscriptions in tbe Himyaritio cbaracter . . . now in tbe 

Britisb Museimi. London, 1863. 
Langer und Mueller. Sabaeisobe Inscbriften. 1883. 
Prideaux, W. Sabaean Grammar. London, 1877. 

Hebrew, etc. 
Bickell, G. Outlines of Heb, Gram, annotated by S. Curtiss. 

Cbwolson, D. Die Quiescentes in der alt-Heb. Ortbogr. 1878. 


Gesenius, W. Hebrew Grammar, enlarged by Kautzsch, 

revised by A. E. Cowley. Oxford, 1910. 
Driver, S. R. Use of the Tenses in Hebrew. 1881. 
Rosenberg, J. Phoenikische Sprachlehre. 1907. 

Aramaic and Syriac (Biblical Aramaic) 

Kautzsch, E. Grammatik des Bibl.-Aram. Leipzig, 1884. 
Marti, K. Knrzgefasste Gramm. der Bibl. Aram. Berlin, 1896. 
Sachau. Aram. Papyrus und Ostraka. 1911. 
Sayce and Cowley. Aramaic Papyri discovered at Assouan. 


Petermann, J. H. Grammatica Samaritana. 1873. ! 

(Syriac) ; 

Brockelmann. Syrische Grammatik. Berlin, 1905. ] 

Duval. Traite de Grammaire Syriaque. 1881. I 

Maclean. Grammar of the Dialects of Vernacular Syriac. ' 

Cambridge, 1895. 
Merx. Hist, artis gramm aticae apud Syros. Leipzig, 1889. 

Noldeke, Th. Neusyrische Grammatik. Leipzig, 1868. | 

Knrzgefasste syrische Grammatik. Leipzig, 1898. i 

Uhlemann. Grammatik der syrischen Spraohe. 1857. i 


(Jewish Aramaic) 

Dalman. Grammatik des jiidisch-palaatin. Aramaeisch. 1894. ; 

Levy, J. Neuhebraisches und Chaldaisches Worterbuch iiber 

die Talmudim. Leipzig, 1875. 
Margolis. Lehrbuch der aramaeischen Sprache des bab. 

Talmuds. 1910. I 

(Mandaean) ' 

Noldeke, Th. MandaLiche Grammatik. Halle, 1875. | 

Babylonian- Assyrian 

Delitzsch, F. Assyxische Grammatik. Berlin, 1906. ' 

Assyrian Phonology {Hehraica, vol. i). 

King, L. W. First Steps in Assyrian. 1898. 

Hammurabi's Letters. 3 vols. 1898-1900. 

Koehler and Ungnad (cited as K.U.). Hammurabi's Gesetz ^ 

1904, etc. j 

Ungnad, A. Babylonische-assyrische Grammatik. Berlin, 

1906. I 






(i) The Semitio Group 


(ii) Babylonia and, Assyria . 


(ui) Canaan 


(iv) Aramaic 


(v) Arabic 


Ilijaz dialect . 


Nejd . 




Syria and Palestine 


Egypt . 


North Africa 


Malta . 


Hadramaut . 


Oman . 

Southern Arabic . 


(vi) Abyssinian . 


The Function of a Consonant . 

7. (a) The Consonant in a syllable 

8. (6) Transcription. 

9. (c) Classification of Consonant Sounds 

10. (d) The Consonant Sound in detail 

(i) Laryngal Hamza. 

11. (ii) Laryngal h 

12. (iii) Laryngals h, h 

13. (iv) Laryngal ' (^) 

14. (v) Velar ^(fr) 

15. (vi) Velar q 






(vii) The Palatals 



(viii) The Dentals and Sibilants . 


(a) Transmission 


(&) Simple Dentals . 


(c) Final -t . . . 


(d) Aspirate Dentals 


(e) Covered Sounds . 


(/) Uncovered Sibilants . 



(ix) Labials . . . . 



(x) Sonants . . . . 



(xi) Semi-vowels 








(i) Assimilation of Consonants . 



(a) Assimilation of reflexive t 



(b) Other assimilations oi t 



(c) Other Dentals and Sibilants 



(d) Laryngals, Velars, Palatals . 



(e) Labials .... 

. 76 


(/) Sonants (i) m, n 

. 77 


(ii) I . . . 



(iii) r . . . 

. 80 


(g) Semi- vowels . . . . 



(ii) Dissimilation of Consonants . 


(a) Dissimilation of doubles in contact 

; 81 


(1) Dissimilation of the first 

. 82 


(2) Of the second 



(6) Dissimilation of repeated initials o 


finals .... 

. 87 


(c) Other repeated consonante . 

. 87 


(d) Special case of sonant I, r 

. 87 


Consonants affected by Vowels 

. 88 

(a) Aspiration .... 

. 88 


(6) Palatalization 

. 89 


(c) Change of Labial to Semi-vowel 

. 90 


{d) Semi -vowels assimilated to Vowels 

. 90 








General .... 

. 91 


Accent ..... 

. 93 


(i) Long Vowel a . 

. 94 


(ii) Long Vowel I . 

. 101 


(iii) Long Vowel u 

. 102 


(iv) Short a . 

. 103 


(v) Short * .... 

. 108 


(vi) Short ii . 

. 110 


(vii) Diphthong ay 

. 113 


(viii) Diphthong aw 

. 114 


(ix) Contraction of Vowels with Semi- vowels — 








(1) Semi-vowel as closure . .116 

(2) Semi-vowel as initial . . 116 



(i) Influence of Consonants on Vowels . 
(a) a and laryngals 
(6) iju become a near laryngal 
(c) Long Iju insert a before laryngal 
{d) a and Emphatic Consonants . 
(e) iju with Laryngals or Emphatics 
(/) Influence of Sibilants and Dentals 
{g) Influence of Labials 
(h) Influence of Semi-vowels. 
{j) Vowel Assimilation assisted by 
Laryngals .... 

ijc) Change of quantity due to Laryngals 

(ii) Influence of Vowels on Vowels — 

(a) Assimilation . . . . . 
(&) Dissimilation .... 







(I) Formation of new Syllables by the use of 

prosthetic and inserted vowels . .129 
(i) Group of two consonants in inception 129 
(ii) Medial group . . . .132 








(iii) Final group .... 

(iv) Inserted half-vowel 
(II) Haplology and Elision .... 
(a) Consecutive open Syllables with 

same consonant and same vowel 
(6) Consecutive open Syllables with 

same consonant and short vowel 

(c) Omission of initial or closure when 

homogeneous .... 

(d) Elision of one of two homogeneous 

consonants separated by a third 






. 136 


(e) Vowel Elision 

. 137 


(III) Metathesis 

. 137 


. 139 


(A) The Absolute Form . 

. 139 


(1) First Person Singular Absolute 

. 139 


(2) First Person Plural Absolute 

. 141 


• (3) Second Person Absolute 

. 142 


(4) Third Person Absolute 

. 145 


(B) The Suffixed Pronoun 

. 149 


(1) First Person Singular — 

(a) Suffixed to Nouns 



. 149 


(b) Suffixed to Verbs. 

. 150 


(2) First Person Plural . 

. 152 


(3) Second Person 

. 153 


(4) Third Person . 

. 156 


. 158 


General ...... 

. 158 


(i) Demonstrative da, di . 

. 158 


(ii) ha 

. 162 


(iii) hay, 'ay ... . 

. 164 


(iv) la 

. 164 


(v) to 

. 166 






(vi) na . . . . 

. 166 


(vii) ma ... . 

. 168 


(viii) ta . . . . 

. 169 


(ix) ya . . . . 

. 170 


(x) aga . 

. 171 









(A) Relative Pronouns — 

(i) dhu, etc. 
(ii) ('a)Z . 
(iii) sa 
(iv) man, ma 

(B) Interrogative Pronouns- 

(i) ma 
(ii) mxin 
(iii) mi 
(iv) min 
(v) 'ay 


(A) Nouns Generally 

(I) Two Consonant Roots 

(II) Vocalization of noun steins 

(III) Noun Stems with Preform atives 

(i) Preformative m- 
(ii) Preformative t- 
(iii) Preformative y- 
(iv) Preformative Hamza 

(IV) Informatives 

(V) Noun AfEorm atives — 

(a) Afiormative -at, -t 

(b) Afforraative -y, -i 

(c) Afformative Hamza 
{d) Afiormative -n . 
(e) Afiormative -m . 
(/) Afiormative -u . 












Gender ..... 




Number ...... 


(a) Plural ..... 



(b) Plural Forroatives 



(c) Broken Plurals .... 



(d) Dual ..... 




The Cases ...... 



(a) The Construct 



(b) Eelative as denoting the Genitive 

. 200 


(c) Other means of avoiding the direct 




(d) Later forms of the Accusative 



(e) Adverbial use of the Accusative 



(E) Determination and Indetermination 

. 204 


(F) Comparison of Adjectives 









(i) The Verb Stem . 


(ii) Verb Themes 


(1) The Prim,ary 

. 211 

(2) The Intensitive . 


(3) The Causative . 


(4) The Passive with n- . 


(5) Eeflexive with t- 


(6) General Note on the Verb Themes 


(7) Vocalization of Stem Forms anc 


Tenses . . . . . 


(8) Passive by Vowel Change . 


(iii) Verb Inflexions .... 


(a) The Tenses 


(6) The Moods . . . . 


(c) Persons of the West Semitic 


Perfect , . . . . 


(d) Persons of the Imperfect . 


(e) Imperative 











(/) Secondary Tenses — 

(i) Assyrian Permansive. 
(ii) Syriao Participial Tenses . 
(iii) Introduction of time relation 
into tenses . 
(g) The Participles .... 
(h) The Infinitive .... 
(iv) Verbs showing Phonetic Changes . 





(A) Verbs with first radical Semi-vowel 250 

(B) Verbs with medial Semi-vowel . 254 

(C) Verbs with final Semi-vowel . 257 

Paradigms of verbs with radical 

Semi-vcwels .... 258 

(D) Verbs with initial Hamza . . 261 

(E) Verbs with assimilating Radical . 262 

Paradigm of Verb with assimi- 
lating Radical . . . 263 

(F) Verbs mediae geminatae . . 263 

Paradigm of verbs med. gem. . 266 


. 268 


(a) Prepositions 


(6) Prepositions govern ii 



(c) Exclamatory Particles 


(d) Negative Particles 


(e) Interrogative forms 


(/) Conditional Particles 

Clauses and 






1 (i) The Semitic Group 

The Semitic languages are a group of closely aUied members 
spread over a clearly defined and limited area. We may 
regard these as consisting of five branches, Arabic, Abyssinian, 
Hebrew, Aramaic, and Assyrian. In the case of Abyssinian 
we include only certain languages spoken in Abyssinia which 
are obviously of Semitic kinship ; but besides these there are 
various other dialects surviving from an older population 
which are different in character and obviously have an indepen- 
dent origin. 

Already in the eleventh century a.d. the Rabbi Jehuda 
Hayyug (Abu Zakaria Yahya) began to apply the methods of 
the Arabic grammarians to Hebrew and thus unconsciously 
laid the foundation of the comparative philology of the 
Semitic languages. It was already known that a close 
relationship existed between Aramaic and Hebrew, but it was 
commonly supposed that Aramaic was a corruption from 
Hebrew. Theological prepossessions incfined the Jews to regard 
Hebrew as the parent, not only of Aramaic and Arabic, but of 
all other languages as well, and this opinion was generally 
adopted by Christian writers also. Even this view, however, 
admitted that a much closer relationship existed between 
Hebrew, Arabic, and Aramaic, than between Hebrew and any 
other language ; and to this closely related group a fourth 
member, Ethiopic, was added in the seventeenth century, the 
name Ethiopic being used by Europeans to designate Ge'ez, the 
ancient classical language of Abyssinia. The decipherment 
of the cuneiform inscriptions in the nineteenth century added 
Baby Ionian- Assyrian as a fifth member. 




The study of comparative grammar in the nineteenth 
century brought about the general recognition that languages 
can be grouped in families whose members bear such definite 
relationship one to another that we are compelled to regard 
them as descended from a common stock. It was frequently 
assumed, quite erroneously, that these language families were 
conterminous with racial groups, and so the Semitic languages 
were treated as the speech of a Semitic race. It is now 
recognized that the transmission of language, racial descent, 
and culture drift are three distinct things. Race depends on 
physical descent ; language and culture are transmitted by 
education, conquest, imitation and other means ; they are 
learned either in childhood or maturity and not passed from 
parent to child as a matter of heredity. 

The name Semitic as applied to a group of languages had 
already been used by Schloezer towards the end of the 
eighteenth century to denote a racial group which included 
the Arabs, Hebrews, Aramaeans, and Abyssinians as descend- 
ants of Shem. This theory was based on the genealogy 
given in Genesis x, according to which both Aram and 
Arphaxad are made the children of Shem, and the further 
genealogy in Genesis xi, which makes Arphaxad the ancestor 
of Abraham from whom were descended the Israelites and the 

^ . ,, Arabs who claimed to be the children of Ishmael. 
rV| 1 --..^ — — _ u 

— Closer scrutiny of these genealogies shows that the members 

' * are grouped simply according to pohtical relations. Thus 
> Cl'^ Elam and Lud are noted in Genesis x, 22, as brothers of 
[JlIliJAsshur and sons of Shem : but the Elamites, Lydians, and 
^\^<. Assyrians are not kindred races, and they are so grouped 
f lllfS simply because they were united under Assyrian rule at the 
l^-ptime when the genealogies were composed. Strictly, we 

' cannot justify the name of Semitic, but it is a term in common 

f-M^i use and convenient. No doubt it would be preferable to 
denote language groups by sjrmbols such as letters or figures, 


and to avoid names such as Semitic or Indo-European which 
imply racial or geographical ideas : but it is practically more 
convenient to accept a term in general use than to invent a 
new terminology. 

Language is learned, not inherited, and it may be learned in 
childhood because it is the language of the family or the 
tribe, or it may be learned later in life from motives of con- 
venience or necessity. Very often the language of a conquering 
race has been imposed on the conquered, sometimes, perhaps, 
as a matter of deliberate policy, more often because the 
conquered find it extremely inconvenient not to be able to 
understand or use the language of the rulers. Commercial 
intercourse also tends to spread particular languages, and 
sometimes rehgious propaganda does the same. Occasionally 
we find communities with two languages in use side by side, 
but most often one of the two attains the pre-eminence and 
the other is gradually squeezed out. Thus Arabic was 
introduced into Egypt by the Muslim conquest in the seventh 
century, it became the language of poUtical and legal 
administration, and the common medium of intercourse with 
the rest of the Muslim world, and gradually the native Coptic 
language began to be discarded until in the sixteenth century 
it ceased to be spoken at all, and survives only as a Hturgical 
language amongst the native Christians. 

But when a language is thus introduced it begins to be 
seriously modified. Those who learn as adults, when the 
larynx has already been adapted to the sounds of another 
language, tend to alter the pronunciation and thus produce a 
dialect differing in its phonology ; they find much that is 
irksome in new grammatical rules and forms, and so they tend 
to discard or confuse some of the forms. Thus a new dialect, 
or even a new language, is produced with points of resemblance 
to, and of difference from, the parent speech. The resem- 
blances centre in the morphology, that is to say in the 


formatives by means of which the stems of nouns or verbs are 
formed from roots, and in the additions or modifications by 
means of which tenses, moods, numbers, cases, etc., are 
denoted. All these form a kind of framework which is 
transmitted unaltered save by the subtraction of forms which 
fall into disuse, but modified on perfectly regular Hues by con- 
sistent phonetic changes. Substantially, therefore, relation- 
ship between two languages always means a common morpho- 
logical basis, and a regular correspondence in phonology. 

So far we have not mentioned vocabulary. To a large 
extent vocabulary is more concerned with culture drift than 
with the transmission of language. Increase of civihzation 
tends to introduce new words, but these do not affect language 
directly, they are fitted into the existing framework and made 
to conform to it. It is even conceivable that nothing of the 
original vocabulary may be left, yet the language retains its 
own character unimpaired by preserving its own morphology 
and adapting the new vocabulary to it. 

Although we are able to associate the five members, Arabic^ 
Abyssinian, Hebrew, Aramaic, and Assyrian in one family 
to which we apply the conventional term Semitic, we are quite 
unable to draw up a genealogical table showing their relation 
one to another. In some respects Arabic is associated with 
Abyssinian against all others, in other respects Abyssinian and 
Ass}Tian are grouped against the other three, and sometimes 
Aramaic is isolated against the other four. Constantly the 
five members form new groups and re-form in other groupings, 
so that such a classification as " south Semitic " to denote 
Arabic and Abyssinian is no more than a geographical 
grouping which holds good for a moment but does not denote 
a closer relationship between these two than between any 
two out of the five. Still less are we able to designate any one 
of them as the parent language, or to construct a " proto- 
Semitic " as representing the mother speech. To some extent 


Arabic seems to represent the purest Semitic because the 
least affected by alien elements, but frequently we find in 
Hebrew and Assjrrian forms in full vigour which have 
disappeared from, but left their traces in, Arabic. Whilst we 
are able to group together these five members as forming a 
Semitic family, we have indications of relationships of varying 
closeness with other languages which are commonly classified 
as Hamitic. It seems, indeed, that the Semitic group is but 
one member of a much larger Hamitic family. 

The Hamitic languages, as the term is commonly applied, 
fall into two main groups, the Berber or Libyan languages of 
North Africa, that is to say, the non-Arabic dialects of the 
area between the Nile valley and the Atlantic, and the East 
African group, comprising the Bishari, now found between the 
Sudan and the Red Sea, originally the language of Nubia, the 
older non-Semitic languages of Abyssinia, Somali, Galla, and 
probably the Hausa in the west, which seems to have passed 
westwards from Abyssinia or from the Horn of Africa by the 
waterways which run across the continent and have served as 
a route for culture drift as well. All these languages have a 
clearly defined common element, and the element common to 
them all appears also in Semitic. 

Besides these we have the ancient Egyptian language which 
has recently been classified as Semitic. It shows the elements 
common to the Hamitic-Semitic group as well as some of the 
characteristics which are distinctive of Semitic, some even of 
those which appear in later dialects of Semitic. Probably we 
ought to regard it as a rapidly, perhaps a prematurely, 
developed dialect of Semitic. Closely akin to the five members 
of the Semitic group, it does not fit in exactly with all their 
distinctive features, so that it is better, perhaps, to class it as 
sub-Semitic. It is imperative to employ ancient Egyptian 
and the various Hamitic languages to ilhistrate and explain 
the forms found in Semitic. 


The outspread of the Semitic languages has been from 
Arabia as their centre. This does not imply that Arabia was 
the home of the Semitic race, nor that the Semitic languages 
may not have been derived from a Hamitic parent in Africa 
or elsewhere, but simply that Arabia was the locus in which 
the Semitic languages specialized in their distinctive character- 
istics so that on passing out from Arabia they already showed 
the features which we have to regard as distinctively Semitic. 
Outside Arabia we are able to state, with more or less 
confidence, the period, and sometimes the exact date, at which 
a Semitic language was introduced ; but no historical 
evidence enables us to form any theory as to when Semitic was 
first spoken in Arabia. History shows us successive migrations 
of Semites to Mesopotamia, Canaan, and S}Tia, Abyssinia, and 
North Africa : the spread of the Semitic languages is connected 
with these migrations, and the formation of difierent languages 
and dialects is due, mainly, to the fact that, outside Arabia, 
each Semitic language is spoken by a mixed population whose 
non-Semitic element has caused phonetic changes, modification 
and disuse of grammatical forms, and has introduced many 
additions to the vocabulary. In some cases Semitic languages 
are, or have been, spoken by an entirely non-Semitic people, 
as for example, Phoenician, a Semitic speech closely akin to 
Hebrew, Moabite, and other dialects which can be classed as 
Canaanite, but the Phoenicians or Phihstines were the 
descendants of refugees from Crete, and by race in no way 
connected with the Semites. 

Race and language are distinct, but transmission of 
language implies contact, and so is necessarily connected 
with race dispersal, by migration, conquest, commercial 
expansion, etc. 

The early Semites seem to have been a branch of the 
Hamitic stock segregated in the highlands of south-western 
Asia, perhaps owing their isolation to the expansion of the 


Mediterranean civilization and the civilization of Mesopotamia 
which, operating from different points, tended to constrict 
them within this area. At one time, as Prince Caetani de 
Teano shows, Arabia was fertile and capable of supporting an 
abundant and varied hfe, but that was at a remote geological 
age long anterior to any probable date of Semitic migration. 
Those movements seem to have been caused by the tempting 
vicinity of cultivated land occupied by a settled population 
both in the Euphrates valley and in that of the Nile, and to a 
less degree by the pastures of Canaan which was, indeed, a 
" land flowing with milk and honey " when contrasted with 
the barren highlands of Arabia, The earher Semites were 
nomad survivors of the neoHthic age, hunters and shepherds, 
who had never learned the arts of husbandry. From the 
earhest dawn of history to the present day these nomads or 
Bedwin have made incursions into, and plundered, any 
neighbouring agricultural territory ; sometimes the invaders 
have settled down and learned agriculture and adopted a 
settled hfe, and sometimes a strong State, Egypt, Persia, 
Rome, or another, was able for a time to place a barrier which 
restrained these incursions. Undoubtedly, desiccation, or 
rather increased sahne deposit, has progressed in Arabia, but 
the Bedwin are not the descendants of cultivators, squeezed 
out by the growing inability of Arabia to support them, so 
much as survivals of an earher social stratum, 

2 (ii) Babylonia and Assyria 

The earhest recorded movements of the Semites were into 
Mesopotamia, where they entered in a series of migrations 
extending from about 4000 to 2000 B.C. Mesopotamia was 
already inhabited by Sumerians who had attained a high 
degree of civihzation of the type which may be described as 
" garden culture ", the intensitive cultivation by hand of 
ground immediately connected with a river or irrigating canal. 


This produced the peculiar conditions noted by Bevan {House 
of Seleucus, i, 21-2) as characteristic of the civihzation of 
western Asia : culture was confined to the river valleys and 
the lower slopes of the hills ; the inhabitants of the deserts, 
which are mostly elevated plateaux, were practically untouched 
by it and remained, indeed, to a large extent they still remain, 
very much as they had been in prehistoric times. A 
Babylonian Empire did not mean a large territory which 
could be marked out on the map, but only the low-lying 
watered valleys within that territory : in the desert highlands 
the tribes might sometimes be visited by punitive expeditions, 
but for the most part they hved their own independent life, 
ignorant of any government other than that of their tribal 
chiefs, and careless of the culture which flourished in the river 
valleys. From time to time they poured down into the 
cultivated land as marauders, and sometimes these who came 
to plunder remained to learn the advantages of culture and 
settled hfe. No doubt the general course of events was very 
similar to that which occurred at the time of the Arab conquest 
of Egypt. It has generally happened that the results are 
exactly opposite in their bearing on culture and on language : 
the ruder conquerors imposed their own language on the 
conquered, this language being in return deeply affected by 
the speech of those who thus had to learn it as a foreign 
tongue, and they learned the culture of their subjects which 
was somewhat modified thereby, and so developed distinctive 
though minor pecufiarities. Some settlements in Babylonia 
were Semitic from their foundation, others show a population 
mixed in varying proportions, but gradually the Semitic 
language prevailed throughout the whole territory. Babylon 
itself seems to have become entirely Semitic-speaking about 
2400 B.C., but Nippur remained Sumerian-speaking much 
later, and Sumerian long continued to be the official language 
of Arad-Sin, Erech, and Rim-Sin. The ancient Sumerian 


literature was alive down to about the year 2000. The bulk 
of the religious formularies, in which Sumerian and Semitic 
appear together, seems to have been produced between 
2900 and 2470 B.C., and we may regard this as the period 
during which Sumerian was passing out of use as a spoken 

Gradually the course of poHtical events brought about the 
amalgamation of the various States in Mesopotamia and 
produced the Semitic Empire of Babylon. At a later date 
this was supplanted by the purely Semitic kingdom of 
AssjTia, But this latter had a briefer history and passed 
into dechne to be replaced by a revived Babylon. In both 
these two States the language was Semitic, but in Babylon it 
was more influenced by the Sumerian element than was the 
case in the north. The study of the Sumerian language in 
Europe is of comparatively recent date, and it seems probable 
that the future will see much more drawn from it to illustrate 
the phonology as well as the vocabulary of Babylonian- 
Assyrian. The earhest material we possess of the Semitic 
speech of Babylonia appears in the letters and legal code of 
Hammurabi (circ. 1500), and in the Semitic versions attached 
to the Sumerian rehgious literature, this latter as yet little 
worked, and the latest material is approximately of the fourth 
century B.C. 

3 (iii) Canaan 

After the earlier Semitic colonies were formed in Mesopo- 
tamia we find that these settled and civihzed Semites were 
constantly vexed by incursions of their nomadic kinsmen who 
preferred brigandage to agriculture. These nomads were 
called Arimi or Ahlame in the Babylonian- Assyrian documents, 
and are generally described as invading from the west. 
Geographically, indeed, the northern desert of Arabia passes 
on without break to Syria, and it is well to remember that this 


desert higliland which lies east of Palestine and Syria is a 
unity with Nejd in northern Arabia. From this high ground, 
as from a watershed, the Semites flowed down into Mesopo- 
tamia on the east, and Canaan on the west. But whilst there 
were at an early date civiHzed, or partly civiUzed, inhabitants 
of Canaan, these were neither so advanced in culture nor so 
numerous, it would appear, as the Sumerians in Mesopotamia : 
indeed, the culture seems to have been chiefly along the sea 
coast and so comparatively remote from the desert highlands. 
No doubt, also, in some cases as in that of Terah and Abraham 
(Genesis xi, 29-31), Semites who had settled in Mesopotamia 
but dishked the political conditions resulting from the 
estabhshment of the Babylonian Empire, preferred to migrate 
to Canaan where they were able to continue the tribal hfe 
and free pastoral occupations to which they had been attached. 

The earhest material illustrating the Semitic languages of 
Canaan appears in the Amarna Letters (fifteenth century B.C.), 
which were reports sent to Egypt at a time when Canaan was 
an Egyptian province, and which contain information about 
the language for the guidance of Egyptian officials. The 
dialect in these letters was the parent of the Phoenician 
(Cooke, NSI, xix). The later form of Phoenician is known to 
us from inscriptions (circ. 600 B.C., etc.) and, in its general 
character, was very hke, though not identical with Hebrew. 
In some respects it seems to have approached Aramaic, but 
this appears to have been the case also with vernacular 
Hebrew. Like Hebrew, also, it tended to approach more 
closely to Aramaic at the time when this latter was a kind of 
lingua franca of all western Asia, and in its latest stages outside 
observers inchned to identify the two (thus Cyril in Isa. iv, 
293, and Theodoret, Quaes. 19, in Jud.). 

Whilst Phoenician proper was the language of Asiatic 
Phoenicia the Phoenician colonies around the Mediterranean 
developed a vernacular which approached more and more to 


Aramaic and suffered a general decay of grammatical forms. 
This is known to us as neo^Jlunic and appears in inscriptions, 
in a transcription in Plautus, Poenulus, v, 1, and in scattered 
words cited by later Greek and Latin writers and by the 
Church Fathers. 

The most important of all the languages of Canaan was 
Hebrew, the language of the Semitic settlers in Canaan proper 
as modified by the speech of the earlier non-Semitic inhabi- 
tants and, as we read it, vocahzed by the Masoretes at a date 
long subsequent to its disuse as a spoken dialect. We are not 
here called upon to discuss the dates of the materials contained 
in the Old Testament, which is a matter of literary criticism 
rather than of philology. The redaction of the Law is 
possibly not earUer than the eighth century B.C., but it contains 
much earlier material, and purely classical Hebrew continued 
to be written down to the time of Jeremiah (circ. 600 B.C.). 
In the pre-Jeremian text there are difierences of hterary style 
but practical uniformity of language, save that evidences 
occur of a northern dialect differing from that of the south. 
/After the time of Jeremiah Hebrew begins to be affected by 
/ Aramaic, but some of the finest material both from the 
literary standpoint and from that of grammatical purity is 
found amongst that written during the exile. The exiles in 
Babylon seem to have developed the classical speech as a 
literary idiom, whilst the vernacular of the people left in 
Judah tended more towards Aramaic until the hterary speech 
was no longer intelHgible to them (cf. Neh. viii, 8) : of course 
we cannot say which deviated most from the ancient speech, 
the Hterary forms elaborated in Babylonia, or the vernacular 
of the fellahin left in Palestine. In later Hebrew we find 
traces of Aramaic influence and loan words introduced from 
Persian, Greek, and Latin. 

After the Old Testament the next important Hebrew 
document is the Mishna, compiled at Tiberias in the course of 


the second century a.d., but incorporating older elements. 
Its language is called D''^3n 7 " the speech of the 
learned ", by R. Johanan, and represents a later stage of the 
literary idiom. It shows marked differences from the Old 
Testament, in language tending towards Arabic dialect, and 
in orthography approaching Aramaic. Time relations are 
marked in the tenses, compound tenses and analytical forms 
are introduced, a double passive (Nif 'al reflexive) appears, and 
the vocabulary shows an increasing importation of loan words 
from Aramaic, Greek, and Latin. In all this, however, it 
must be noted the spoken language as it appears in Amos and 
the earher prophets, in the song of Deborah, and in the 
prophecy of Balaam, shows more inclination towards Aramaic 
characteristics than is the case with the literary form. On 
the other hand the inscription of Siloam (circ. 700 B.C.) 
endorses the classical speech. 

The Mishna represents the speech of the learned, a language 
no longer spoken by the people who, after the destruction of 
the Temple and the disintegration of Jewish life, had adopted 
(Aramaic, the common medium of intercourse throughout 
western Asia, and this Jewish Aramaic as used in the 
Babylonian Talmud was the common speech of the Jewish 
community until it was replaced by Arabic in the tenth 
century a.d. The use of Arabic had been growing since the 
Arab conquest in the eighth century, and almost simultaneously 
with its final triumph in Palestine there was a revival of 
classical Hebrew as a written language amongst the Jews of 
Spain and North Africa. This revival produced a large 
output of homiletical Hterature, but its main importance to 
us lies in the impetus it gave to grammatical studies. 
Nevertheless the language it used was a conscious imitation of 
the classical Hebrew ; there was no continuity of oral tradition 
in the use of Hebrew as a living language. 

Closely akin to Hebrew was the Moabite language which is 


known to us only from the inscription of the Mesa stele (circ. 
900 B.C.). In it we find a reflexive of the primary theme (as 
Arabic conj. viii), which is lost in Hebrew. 

4 (iv) Aramaic 

We give the name of Aramaic to the languages of the 
Semitic invaders of Palestine and Syria who advanced further 
north and settled in the Aram or " highlands " : those who, 
for the most part, were pre-Israelite immigrants. The 
earHest extant material in Aramaic is the ZinjirH and Nerab 
inscriptions found near Damascus and dating from about the 
sixth century B.C. In Old Testament times Aramaic was the 
northern neighbour of Hebrew, but after the decUne of 
Phoenicia, when Carchemish became the centre of the trade 
of western Asia, Aramaic gradually became the common 
language of political and commercial life throughout all 
western Asia and Egypt (cf. Xenophon, Cyropaed., 1, v, 31), 
and thus used as a lingua franca passed through a rapid decay. 

The earliest material illustrating Aramaic after it had 
become a common medium of international intercourse 
appears in the papyri discovered in Egypt, mainly due to 
Jewish settlers there during the period following the Baby- 
lonian conquest of Judea, and a little later in the book of 
Ezra (chaps, iv, 8-6 and vii, 12-26). Considerably later is the 
book of Daniel (ii. 4b-vii. 28). The Aramaic passages of 
Ezra and Daniel present what is commonly known as 
Biblical Aramaic, and this, in the main, agrees with the dialect 
used in the papyri. 

Next in date are the Palmyrene inscriptions (third century 
B.C. to first century a.d.), and then the Nabataean (first 
century B.C., etc.), but these latter, though written in Aramaic, 
are composed by Arabs who spoke their own language and 
used Aramaic only in inscriptions. 

/ In the time of Christ the language of GaHlee and of Palestine, 
'generally, was Aramaic, Hebrew remaining the property of the 



learned. When the actual words of Christ are quoted in the 
Gospels they approximate to Aramaic (cf. St. Mark vii, 34 : 
V, 41), and this appUes also to the citation of terms in common 
use such as Bethesda (St. John v, 2) and Gahhatha (id. xix, 13). 
After the destruction of the Temple and dispersal of the Jews, 
Aramaic entirely replaced Hebrew in Palestine and amongst 
the Jews in neighbouring lands until replaced by Arabic in the 
tenth century a.d. Of this Jewish Aramaic the earliest 
material survives in the Onqelos Targum (fourth century a.d.) 
and the Jerusalem Targum (sixth century a.d.), this latter in 
an artificial language in which eastern and western dialects are 
mingled. Later material exists in the Jerusalem Talmud 
(T.J.), i.e. the Tiberian commentary (Gemara) on the Mishna 
representing the western or Gahlaean dialect, and the 
Babylonian Talmud (T.B.) representing the eastern type. 

It is not so easy to define the exact position of Samaritan. 
It may be that it is a form of Aramaic which passed southwards 
during the time of the exile, or it may represent the vernacular 
speech of Israel corrupted by Aramaic elements, but from its 
general character it is usually classed as one of the Aramaic 
group. It must be remembered that Hebrew and Aramaic 
are not so much different languages as dialects of one parent 
speech, and it is not always easy to decide what dialectal 
peculiarities justify inclusion in one group or the other (cf . 89). 
Samaritan is known to us in a version of the Hexateuch which 
shows the influence of the Onqelos Targimi and has been 
retouched in Muslim times, and in some poetry of later date. 

Eastern social groups depend very much more on religious 
than on political factors, and sectarian divisions tend to develop 
or accentuate differences of dialect by the artificial barrier 
which restricts intercourse to members of the same rehgious 
community. Such is very clearly the case with Syriac or 
Christian Aramaic. The condemnation of Nestorius and his 
followers at the council of Ephesus in a.d. 431 made a definite 


separation between the eastern Christians who followed 
Nestorius and their co-rehgionists of the west. The council 
of Chalcedon a few years later (a.d. 448), on the other hand, 
excommunicated the Jacobites who represented the element 
most opposed to Nestorius, leaving only a minority of Syrians, 
mostly in the south, in communion with the State church of 
the Byzantine Empire. Amongst these latter developed what 
is known as Palestinian Syriac of which some few documents 
still survive, a Gospel version in a Vatican MS., the so-called 
" liturgy of the Nile ", and a small amount of material 
discovered in Sinai, in Damascus, and in Egypt. As a spoken 
language Palestinian Syriac ceased at the Arab conquest but 
it remained in liturgical use to a much later date. 

The Hterary history of Syriac begins about the second 
century a.d., and shows a later form of Aramaic with a 
vocabulary very strongly influenced by Greek. After the 
separation between the Jacobites and the Nestorians in the 
west and east respectively, we find a marked diSerence in 
dialect between the material produced in one community and 
that of the other, a difference not, of course, caused by the 
schism but accentuated by the cessation of intercourse brought 
about by the division, and by the fact that the Nestorians, 
persecuted in the Roman Empire, were welcomed in the 
Parthian and Persian dominions where they had an inde- 
pendent career, and spread eastwards by missionary enterprise 
well into central Asia. In the seventh century came the 
Arab conquest and the introduction of Arabic, until in the 
thirteenth century West Syriac ceased to be any longer a 
spoken language save in a few isolated parts. It survives 
now only in Ma'lula, Bukh'a, and 'Ain et-Tineh in the Anti- 
Lebanon about J.Dinha, N.E. of Damascus.^ East Syriac 

^ Ma'lula and Bukh'a are Christian villages. 'Ain et-Tineh became 
Muslim in 1860, the change being made, it is locally reported, to save 
the famous pistachio grove threatened with destruction by the Muslims. 


had a more prosperous career and is still used in various 
village communities in the Euphrates valley, at Tur Abdin, 
Urmi, and amongst the fellahin around Mosul. These neo- 
Syriac-speaking groups are thus classified by McLean : (1) 
North, the plain of Salamis in Persia, Qadsanis, Gawar, and 
Gilu ; (2) the Ashirei group in central Kurdistan, upper 
Tiari, Tkhuma, etc., and Ashita in lower Tiari, Marbishu, and 
Shamsdin ; (3) South, Alqosh (near Mosul), Bohtan, and Zaklu 
(about 60 miles from Mosul). 

The dialect known as Mandaean was connected with a 
Gnostic sect in lower Mesopotamia isolated by its rehgious 
character from Christianity, Judaism, and Islam, and of great 
value not only because of a fairly abundant Uterary material, 
but also because its isolation protected it from Greek, Hebrew, 
and Arabic influences, and so it displays an independent 
development of Eastern Aramaic. In it we see the further 
evolution of tendencies already noticeable in older Aramaic. 

5 (v) Arabic 

As we have already noted, Arabia was the centre of distri- 
bution of the Semitic languages and, in all probabihty, their 
area of speciaHzation. But Arabia itself was not untouched 
by outside influences ; along the east and south the Minaean 
civihzation showed the influence of Mesopotamia, and at a 
later date Greek culture filtered down through intercourse 
with Syria. Yet the fact remains that the speech of Arabia 
was less affected by alien influences than any other of the 
Semitic languages. 

Arabic is divided into two branches, north Arabic, i.e. what 
is known as classical Arabic and its derived dialects, and 
south Arabic or Himyaritic and its derived dialects. Classical 
Arabic denotes the Arabic employed as a literary medium by 
Arabic writers from the time of the pre-Islamic poets to that 
of Shawki Bey and Hafiz Ibrahim in our own day. Its basal 


literary form was definitely laid down by the composition of 
the Qur'an which is accepted not only as a divine revelation, 
but as a perfect model of grammar and composition. The 
Qur'an was edited soon after the Prophet's death, but its 
vocaUzation, at first dependent on oral tradition, was not 
marked in the written text until some sixty-five years later 
when Islam was already taking a cosmopoHtan character, and 
Arabic had become the language of government and pubUc 
Hfe in Mesopotamia, Syria, and Egypt. So rapidly had 
Arabic been corrupted where it was spoken by those who had 
learned it as a foreign language, that the Qur'an was unintelU- 
gible to the foreign converts who now formed the majority 
of the MusUm community, and they were unable to read aloud 
the written text correctly. The insertion of vowel points and 
other orthographical signs was primarily intended to secure 
the correct reading of the sacred text. Although the Qur'an 
was published in the current speech of the Hijaz, and its 
orthography showed the pecuHarities of that speech (cf. 10), 
the punctators did not have recourse to the Hijaz dialect but 
to the older and purer speech of the nomadic Arabs of the 
Nejd hinterland. And they were right in taking this standard, 
for it was the dialect least affected by aUen influences. Modern 
philology has shown a marked distaste for " literary " 
languages, which are generally regarded as artificial. Truly, 
much valuable material is to be found in the Arabic dialects, 
but the classical forms cannot be hghtly disregarded. The 
early grammarians took a correct attitude and formed their 
standard from the purest, because the most isolated, dialect, 
and have left us some important notes of dialectal differences 
as they then existed. Of necessity classical Arabic must be 
the starting point for Semitic philology. It certainly does not 
represent proto-Semitic, for it evidently has passed through 
changes ; still it is free from the more violent alterations which 
have taken place in Assyrian and Hebrew because of its more 


isolated position whicli preserved it from contamination by 
non-Semitic languages, and because it represents the speech 
of those amongst whom there were no ahens who had learned 
Semitic as a foreign language. 

Classical Arabic is the literary form of the northern dialects 
of Arabia in the seventh century a.d., the dialects which, for 
the most part, were the parents of the different forms of 
vernacular Arabic in the countries where that language has 
been introduced by the spread of Islam. This northern 
Arabic must be regarded as existing in Arabia in two forms, 
not greatly different the one from the other, and neither much 
affected by any non-Semitic language. 

(1) The dialect of the Hijaz (i.e. " barrier "), the western 
side of Arabia along the coast of the Red Sea. Here was the 
cradle of Islam, and here are the two holy cities of Mecca and 
Medina. It is the Hijazi dialect which is written in the 
Qur'an and in the compositions of the pre-Islamic poets. But 
the Arabs of the Hijaz were partly settled as town dwellers in 
the time of the Prophet and had commercial intercourse with 
foreigners, especially Syrians and Abyssinians. The Qur'an 
itself shows the use of foreign words introduced in the course 
of trade and by the influence of Grseco-Syrian civilization. 
The Hijaz was at the time a partly settled and cultured district. 
On the eve of the Prophet's mission there were both 
Christian and Jewish Arab kingdoms. The Arabs were 
spreading northwards as conquerors and beginning to assimi- 
late Grseco-Syrian culture, and these conditions continued 
under the Umayyad Khahfas of Damascus. The influence 
of Syria appears in its extreme form in Nabatsean conditions 
where Arabic was the spoken language, but Aramaic was used 
in writing. Hijazi Arabic was not absolutely pure even in the 
days of the Prophet. 

(2) The dialect of Nejd, the speech of the desert tribes of the 
hinterland, was not appreciably, if at all, affected by foreign 


influence and so it was purer than that of the Hijaz. It was 
this dialect which the grammarians took as the standard of 
good Arabic, and to this they endeavoured to make the Qur'an 
conform, not without occasional violence to the dialectal 
pecuharities of the text. They studied the Nejd dialect, and 
some of the greatest of them quahfied themselves by a sojourn 
amongst the Bedwin of the desert. Various differences of 
dialect are recorded as existing amongst the desert tribes, 
the B, Tamin, B. Oqeyl, Qays, etc., and these pecuharities 
are all recognized as equally of classical standard. 

From this speech of Nejd and Hijaz Arabic was carried in 
all directions by the spread of Islam, its form varying 
according to the proportion of Arabic to non- Arabic population 
in the invaded areas. In Upper Egypt, for example, the 
people did not become entirely Arabic-speaking until the 
sixteenth century a.d., although the land was conquered by 
the Arabs some nine hundred years earher, and the Arabic 
which made this slow progress against another language was 
necessarily much modified in the process. In every case the 
modifications concern phonology, morphology, and syntax. 
In the last two the change was everywhere the same, namely, 
the discarding of case endings and of the more difficult 
syntactical constructions, and a general relaxation of the 
rules of grammar, whilst new methods of forming a tense 
system as well as the development of a time sense in tenses 
rather than the earher aJctionsart is a regular phenomenon. 
It is worth noting that when, some centuries before the spread 
of Islam, Aramaic had become for a time the lingua franca of 
Western Asia, it passed through precisely similar changes. 
Indeed, this grammatical decay is the natural result of the 
general use of a language in communities whose mother 
tongue it is not. Changes in phonology and vocabulary 
are rather different : to some extent phonetic changes 
result in the confusion of Uke sounds, but there is also 


the introduction of alien sounds and of new words in 
The chief dialects with which we have to deal are : — • 

(3) The dialect of 'Iraq or Mesopotamia. This is the form 
of speech most often cited by the grammarians of Kufa as illus- 
trating the incorrect pronunciation of those who had learned 
Arabic as a foreign language. But the importance given to 
this dialect was largely due to the fact that the grammarians 
were living in the locality where it was the common use of the 
lower classes. Most of the peculiarities of this dialect, as noted 
by the early grammarians, are still characteristic of the 
vernacular 'Iraq, but in vocabulary there has been an 
increasing corruption due to the influx of Persian, Turkish, 
and more recently of French and Urdu. The 'Iraq dialect ia 
spoken throughout the whole Tigris-Euphrates valley, but 
certain pecuHarities exist in the town dialects of Bagdad, 
Mosul, and of Mardin. 

(4) The dialect of Syria and Palestine shows a number of 
local pecuUarities, in many respects approaching the 'Iraq 
dialect, which is more particularly true of the Syrian fellahin 
speech. Marked local peculiarities exist in the Hauran, 
Petraea, and amongst the Druses of the Lebanon, as well as in 
the town speech of Damascus, Bethlehem, and Jerusalem. 

(5) The dialect of Egypt. In this we distinguish the following 
general types : (1) the dialect of Lower Egypt, (2) that of 
Upper Egypt south of Minia, (3) that of the Egyptian Bedwin 
between the settled parts and the eastern frontier, about 
Salhiya, etc., and the moving population scattered amongst 
the inhabitants of the Delta, and (4) the dialect of Sinai which 
seems to be a mingling of Syrian and Egyptian with some 
Hijazi influences. Settlements along the Canal from Port 
Said to Suez are either of late date or have been greatly 
affected by recent arrivals and do not seem to present any 
distinctive dialect. 


(6) The dialects of North Africa show, in places, the influence 
of Berber or Libyan speech and diverge very markedly from 
the vernacular of Syria and Egypt. In some respects, however, 
the peculiarities seem to be simply the survival of archaic 
forms obsolete elsewhere. The leading types are : — 

{a) The dialect of Tripoli. 

(6) Tunisian. 

(c) Algerian — (I) Tlemsen. 

(2) Ulad Brahin in the departmentof Oran. 

(3) 'Ain Madi in south Algeria. 
{d) The dialect of Morocco or Maghrebi. 

(e) The Moorish Arabic of Spain, now extinct but 
known to us from the fifteenth century 
grammar of Pedro. 

(7) The Maltese, which is essentially alHed to the North 
African group but appears to contain Syrian influences and 
has a vocabulary which is largely Italian. It is in the pecuUar 
position of being the only Arabic dialect usually written in 
Roman characters. 

All these dialects of Arabic mentioned so far have their 
origin in the northern branch of the parent language, and 
with them must be associated certain dialects which have now 
spread southwards but are also of northern origin ; these are : 

(8) The dialect of Hadramaut in south-western Arabia east 
of Aden. 

(9) The dialect of Oman, a district colonized by Qatani 
tribes from Yemen. But these, though dwelhng in the south, 
were not of southern origin, and brought down with them a 
northern dialect. 

Southern Arabic centres in the language of the Himyari or 
'OfiTjplrai who were anciently the ruUng race in Arabia. 
In the south of Arabia there was a Minsean kingdom which is 
supposed to have commenced about 1250 B.C., and this was 


suoceetitxi by a Sjibaviu kingilotii in tho eighth contury B.C., 
though it seems hkoly tliiit tho two were conteiupomry for 
some KMig period. Inscriptions exist whieh date fri>m the 
third-fonrth century i!.i\ Minavm and Sjibsean are known 
to us only from tho inscriptions, but those southern diahict« 
are representixi to-day by tJie diah>ct of Mcliri, in the tiistnct 
between lladramaut and Oman, ah hough possibly somewhat 
corrupted by African iniluences, especially by that of the 
Somali. El-Feiyumi in the Misbah (article Maiiar) says that 
'* the language of the people of March, which is a district oi 
Oman, is quick and scarcely or not at all intelligible, and is of 
the ancient llimyari ". Closely akin to tho dialect of Mehri is 
tJiat of Soqotra. 

6 (vi) Abyssinian 

Tho entry of Semitic colonists into Abyssinia seems to have 
commenced about the fourth century ii.c. In 115 B.C. a 
Saba^m settlement was formed there, and by this means, it 
would appear, a culture drift from IMesopotumia through 
south Arabia passcil into Abyssinia. The lino of intercourse 
does not seem to have been finally broken until the rise of 
Islam in the seventh century a.d. With it, in some way, must 
be connected the philological affinities between Assyrian iuul 
Abyssinian, although it does not suffice to account for them 
altogether, and they remain an unsolved problem. 

The Semitic colonists in Abyssinia settled inland in tho best 
and most fertile country, leaving the conquered earlier 
iuhabitAuts in the desert between them and the sea and in the 
more barren highlands. About the fourth century a.d. the 
Abyssinians became Christians under the influence of Coptic 
missionaries, and thus many Coptic and Creek words were 
introduced into their language. The classical Semitic of 
Abyssinia became a literary language by tho versions of the 
Scriptures made in it, and this was followed by a considerable 


literary output, partly tnirmltttionH fn^rri tlj«? (in;«;k and 
partly origirmi rriattfT IxiwmI on Groolf ru<><{i',\H. Thm iurif^uago 
in known aw (Ja'az or f'Jlhiojnc. It (;«!aH«;d to l>n a Hf>okfm 
tonguo in tho fourt«j<inth contury A.o., hut \h Htill roUinivi an tho 
liturgical jangiwg'j of tlio AbyHwinian Churcli, 

Altlniijgli (UiU'/A irt a rload languag<;, it haH living floHccndantrt 
in 'ri/jriiui which i.H Hpokcu in north»;rn AhyHwinia, but in much 
iriflu<inco(i hy the dominant Amiiaric, and Tif/r4 which iH UHiA 
in the Italian colony of Krythrca and in tho inland of Dalilak, 
and which han [)rcHcrv<^l itH original fcaturoH more faithfully 
by reason of itn mf>rc iHolat<rd poHition. 

Arnhmn in the languagr; moHt commonly uhjmI in AbyHrtinia. 
It wan tho larigiwgc of the ruling dynanty of the period a,d, 
1270-1805, and ha!^ afTecixjd (jvery other Hpoken dialect. It iH 
not derived frrim (Je'ez but from a HiHter langiwge. It iH more 
afTocte<l by Arabic and Oalla and Iohh by Onjck ami (Jo[)tic, 
Its literature commenccH about the fifte<!rith century a.i>. 



I The Function of a Consonant 
7 (a) The Consonant in a syllable 

According to tte Arabic grammarians a sentence (-^AJ ) 

is an intelligible group (X^ c-*) ^a) of words after wbich 

silence seems good, i.e. which gives a complete sense that 

naturally terminates with a stop or pause. From this point 

of view of grammar, properly so called, the constituent elements 

in the sentence are words disposed to express subject, object, 

predicate, etc. : from the point of view of morphology each 

word contains a root, and in most cases also a formative 

element, as well as inflections, prefixes, sufl&xes, etc. 

Phonology regards the same sentence as its subject matter, 

but treats it as a series of syllables which is continuous from 

the point at which the sentence begins after a greater or 

lesser interval of silence, until the point at which speech again 

ceases at a stop. Phonetically, we begin with the idea of a 

sentence because it is necessarily preceded and followed by 

silence : but the phonetic unity may be less than a sentence, 

for a pause may be made not only where " silence seems 

good ", i.e. when the sense is complete, but also for the purpose 

of emphasizing a word, or by a natural tendency after each 

item enumerated in a series. In any case the speech which 

lies between two intervals of silence forms a phonetic unity, 

and this most commonly is a sentence. Only very seldom is 

the syllable equivalent to a word : most words contain two or 

more syllables, and very often a syllable hes partly in one 


word and partly in another, and it is only with a monosyllabic 
word that a syllable alone can form a sentence. Yet 
phonology cannot ignore the existence of words because each 
word may furnish an accented syllable, and these accented 
syllables appear as a series of nuclei which in many cases 
afiect the syllabic constitution. Thus, the phonetic unit is the 
syllable, but the syllable considered as a part of continuous 
speech lying between two intervals of silence. 

It is possible for a single syllable to form a complete 
phonetic group which may or may not be a sentence, as the 

imperative A>- " take ! ", a command which gives a 
complete sense and forms a true sentence, or the name of the 

letter ui which is a complete phonetic group but not a 

sentence, though it is natural to make a short pause after the 
name of each letter as always in the enumeration of items in a 
catalogue. More often the phonetic group will be a sentence 
containing a series of syllables. 

Phonology, therefore, has to consider the formation of the 
syllable which is its unit, and then the group collected about 
one accented syllable, usually a word, but sometimes two 
words as in the Hebrew "^^2^ H^?^ rnal-kaih-she-ha 
" queen of Sheba ", the word in the construct (H^?^) 
being without accent, and thirdly the longer group which 
forms consecutive speech between intervals of silence. 

In all human speech the syllable necessarily contains a 
vowel. In the Semitic languages where the syllabic order is 
very clearly defined it cannot contain more than one vowel, 
although the diphthongs ai, au may be allowed as single vowel 
sounds, and also it can happen that a long vowel sound, 
followed by a consonant whose organ of utterance is far 
divergent from that of the vowel, shows a modification of 
timbre in its closing tone so that ruh becomes rudh, 


the d being a deflection of the u sound as the vocal organs 
form themselves to the utterance of h. Also, the syllable must 
begin with a consonant, which may be in the same word as 
the vowel, or may be the final consonant of a preceding word : 
if the vowel, the essential element of the syllable, follows after 
silence, or after a vowel, a laryngal effort known as Hamza has 
to be made in commencing the vowel, and this effort is classed 
as a consonant. It is, indeed, impossible in any language to 
commence a vowel sound after silence without some such 
effort as is represented by the Greek spiritus lenis, but it must 
be admitted that the vocal organs are used differently in 
different communities, which is, indeed, the principal cause of 
variation in dialect, and Semitic speech employs the larynx 
and soft palate mucli more than is the case in the Indo- 
European languages ; consequently the spiritus lenis is 
rendered much more emphatic by being thrown further back 
into the throat. But in the speech of those who have learned 
Arabic as a foreign language, or who have received it from 
those who have so learned it, the Hamza is much weaker, and 

thus c_-)i is sounded 6h in Morocco ; the initial effort has 


not disappeared, but it is much less distinct than in purer 
Arabic. So in Hebrew *wa- " and " before such words as 
"Hyp is sounded as n-, that is to say it has become a vowel 
sound with a laryngal effort so weak as to be practically 

The general theory, however, remains that a syllable must 
commence with a consonant, and then follows a vowel. From 
this it is evident that a syllable cannot commence with two 
consonants. A word mav commence with two consonants, 
for if the preceding word ends in a vowel it will allow the first 
of these two consonants to enter as closure into the syllable 
to which it belongs, but if the word stands after silence or after 
a consonant it will be necessary for the first consonant to be 


vocalized bv an inserted or prosthetic vowel, and if by a 
prosthetic another consonant will have to be prefixed to 
commence the syllable. Every syllable must contain two 
elements, (1) an initial consonant, and (2) a following vowel, 
which may be long, or short, or a half -vowel (ultra short), the 
latter including the vocalized sonant, but only in speech 
which has been strongly influenced by non-Semitic languages : 
possibly even the half-vowel as it appears in Hebrew and 
Aramaic shows ahen influence. 

8 (b) Transcription 

The following table shows the method followed in the 
following pages in transcription with (in brackets) some other 
transcriptions in common use : — 

Arabic Hebrew Abyssinian 

and Aramaic 

» . t^ . • i\ 

h . d . n . • u 

h . 

• c • 


. rh 

h (kh) . 


n . 



• L  

• 5^ , • 

9 (gh) 



• V  

q . 




. t 

9> 9> J 


y ' 

. 1 

k . 


T • 

. n 




e (th) 



d . 


d (dh) . 





. \. 


. j; 


• u^ 

d . 

• & 


' w 

S . 


ah . 





fiph) . 



V ' 






i > 

Hebrew and 


n . 



1 . 




to . 

. m 

to . 

. m 

r • • 


r • 

D . 


b^ . 


^ . 




^ • 


^ • 














w (u) 


Hebrew and 






h . 



It is not necessary to give different scripts, and thus it will be 
needless to add the various forms of the letters used in Syriac, 
etc. It must be noted that b, g, d, k, p, t, are aspirated as hh, 
gh, dh, Ml, jph, th, in Hebrew and Aramaic by the influence of a 
preceding vowel (cf. 37) ; these latter are not separate 
consonants but merely modifications of the consonant sound 
due to the influence of the vowel. The Abyssinian script 
shows a syllabary ; from this the forms above represent the 
consonant with the short vowel -a following, thus U = ha, etc. 

9 (c) Classification of the letters 

The most important classification of these consonants is 
based on the " organ of utterance ", thus : — 

(i) Laryngals, uttered in the throat, ' h, h, h, '. 
(ii) Velars, in the soft palate, g, q. 
(iii) Palatals, with the hard palate, k, g. 
(iv) Dentals, with outlet against the teeth, t, th, d, dk, t, d. 
(v) Sibilants, the same but with a curving of the tongue in 
the rear of the outlet, s, s, S, s, z, z. 


(vi) Labials, with the lips, ^, /, 6. 

(vii) Sonants, m, n, I, r. 
(viii) Semi-vowels, w, y. 

Another important division is that which classifies the first 
six groups as (i) voiced or soft, being those whose utterance is 
accompanied with a musical vibration of the vocal chords as 
in d, h, z, etc., and (ii) the voiceless or hard, from which this 
musical vibration is absent, as in t, p, s, which are the voiceless 
correspondents of d, b, z. 

Certain letters are known as " emphatic ", being distin- 
guished from others by a more emphatic utterance or laryngal 
efiort, thus li is emphatic of ^, g oik, t of t. These emphatic 
consonants are h, ', g, q, t, z, d, s, and in Abyssinian j) as well 
as a 2? similarly emphasized in Aramaic but not distinguished 
by any pecuhar letter : both these latter are non-Semitic in 

There are other sounds, z, c, is, etc., but these are either 
derivative, as i for s by assimilation to a voiced consonant 
following in contact, or pecuharities of certain dialects in which 
the consonant is influenced by a neighbouring vowel (cf. 38), 
or else due to non-Semitic influence as Amharic jh for the 
combination of d and t. 

10 (d) The Consonant Sounds in detail 

(i) Laryngal Hamza 

Hamza (ej-v* " compression " of the throat) is the glottal 

catch or voiceless (stimmlos) " fester Einsatz " of Sievers 
{Phonetik, 385), the larjmgal efiort necessarily made in com- 
mencing a vowel sound after pause, a sound which may be 
made also at the beginning of a medial syllable, or as closure, 
but which does not necessarily occur in continuous speech. 
The Arab grammarians, indeed, state that in the language of 
foreigners it is sounded only after pause. It resembles the 


spiritus lenis of Greek, but is more distinct, because Arabic is 
what we would call a more guttural language, and because in 
Semitic each syllable is more distinctly defined than in the 
case in the Indo-European languages. In general character 
it corresponds with the consonant h, but is not aspirated. 
This sound, quite easy and natural after pause, presents some 
difficulty when following a consonant, and considerable 
difficulty when acting as the closure of a syllable, and it is in 
this last position that we usually find it faUing into disuse in 
dialect and later speech. 

The Arabic grammarians incorrectly classed Hamza as 

" voiced " ijj'^) or " soft ", because they did not 

separate its sound from that of the following vowel " and 
attributed the voice element of the vowel to the consonant 
itself. It was not before the invention of the laryngoscope 
and the splendid experiments of Czermak, that these sounds 
were physiologically elucidated " (VoUers, in ix Congr. of 
Orientalists, Lond., 1893, ii, 137-8). 


In Hebrew and Aramaic this consonant is represented by 
^^ , but in many cases this letter has lost its consonant value 
in the text of the O.T. as it appears with the Masoretic 
pointing, and in later Hebrew as well as in Aramaic it is 
sometimes used as a vowel sign, e.g. in DKp (Hos. x, 14) 
for Dp. In carefully written Hebrew MSS. a consonant 
'Alef is marked by a point placed above, but this is very 
rarely inserted in printed copies of the Hebrew Bible, though 
we usually find it marked in Genesis xhii, 26 ; Lev. xxiii, 
17 ; Ezra viii, 18. In neo-Punic and Mandsean t< has 
become a vowel sign. 

The alphabet passed to the southern Semites early enough 
for 5< to retain its consonantal value, and it so appears in the 


Minsean and Sabaean inscriptions and in the derived alphabet 
of Abyssinia {^). But the Arabic alphabet was of later 
origin, independently derived from the Aramaic, and so 
'AHf came to be employed by the early Arabic writers as a 
symbol for long d, and was thus used in the first editions of the 
Qur'an. These earher Qur'ans were written according to the 
dialect of the Hijaz which was somewhat affected by foreign 
influences due to the trade route which passed down western 
Arabia from Syria and Egypt, and showed such influences by 
a tendency to discard Hamza as the closure of a short syllable, 
giving a compensatory lengthening to the preceding short 
vowel so that -a' became -a, and i\ w' became i, u. As 'Ahf 
was employed to mark long a, and w, y were used for long u 
and *, these three letters often appeared where there ought to 
have been Hamza with a short vowel before it, according to 
the earher and uncontaminated pronunciation. When the 
Khalifa 'AU ibn 'Abi Taleb, noting the growing corruption in 
speech amongst the people of 'Iraq and the consequent errors 
made in reading the Qur'an, commanded Abu-l-Aswad 
ad-DuwaU to prepare an edition in which the true sounds 
would be clearly represented, this grammarian did not refer 
to the speech of Hijaz which, as we have noted, was not 
absolutely pure, but to that used amongst the tribes of the 
highlands of Nejd, and so made the text conform to a dialect 
purer than that in which it had been composed originally, an 
attitude continued by the grammarians Sibawayhi (d. a.h. 
161) and al-Kissai (d. a.h. 102) and their followers in the 
academies of Basra and Kufa. The text thus edited, however, 
admits different readings ; but these do not at all correspond 
to what we know as " variant readings " in the text of the 
Scriptures or classical authors ; they are merely differences of 
pronunciation due to pecuUarities prevaihng in ancient 
dialects, or to traditional practices of certain famous Qur'an 
readers. At the time of ad-Duwali's revision the Qur'an 


text was abeady treated with too much reverence to permit 
the removal or alteration of any of its written letters, and so 
the short vowel sounds, etc., were denoted by diacritical marks 
over or beneath the sacred text. As the Qur'an has been the 
model of written Arabic we find that Hamza was written over 
'Alif, u\ or y, after short d, u, or i, respectively, and this 
Hamza is the consonant, the letter below merely acting as a 
" support " to the Hamza. Thus the words rub's, gi'iu, bus 
were pronounced rds, gitu, bus, in the Hijaz and so at first 

written (j-'lj, <Z^^*>-, cj^Ji> but afterwards pointed 


C -^ *" ^^ 

(T'lj, C^.'>-, {j'y accordmg to the earher pronunciation, 

the Hijazi long vowel being recognized as a variant reading. 
In course of time, however, the latter was generally discarded 
in Qur'an reading, although it has never been regarded as 
erroneous otherwise. 


In Arabic as elsewhere Hamza may stand as the beginning 
or ending of a syllable. Confining ourselves first to Hamza as 
the commencement of a syllable, it may follow after an interval 
of silence, or may be preceded by another syllable. It can 
only follow after silence when it is the first letter in a word, 
that word either being the first in a sentence, or being 
preceded by a pause made intentionally for emphasis, or 
compulsorily in the enumeration of several items in a catalogue. 
Thus we have :— 

(a) Hamza in inception, i.e. when beginning a syllable after 
pause. As the Arabic grammarians hold, and rightly, that 
every syllable must begin with a consonant, an opening vowel 
necessitates a prefixed laryngal effort which is what we term 

Hamza. Thus the word inqatala (AliJl) can stand all 


right in a sentence wliere it is preceded by another word 
whose final consonant becomes the initial of the syllable 
which contains -in-, or whose final vowel replacing the -i- 
helps to form a syllable which begins with the preceding 
consonant and closes with the -n, but if this word stands first 
in the sentence it must become 'inqatala with initial Hamza. 
The only way of dispensing with the Hamza in such a position 
is by changing it to another consonant, either aspirating it so 
that it becomes h, or uttering it so that it becomes emphatic 
as 'Ayin, or altering it to a semi-vowel w or y. 

(i) Change to h occurs even in classical Arabic. Sometimes 
we find hana for 'ana in the first personal pronoun singular, 
and hanta for 'anta in the second, the latter accounting for the 
pronoun hit or het in Mehri, both substitutes for hant (common 
gender). So usually hanna alternates with 'anna in the sense 
" to moan ". In Qur'an reading kiyydka is admitted as a 
variant for 'iyydka twice in Sura i, 4 : and in the ancient 
dialect of the Tayyi 'in " if " could be pronounced kin. But 
these changes are restricted by tradition and may only be 
made where it is recorded that such a change has been made 
in ancient dialect : k may not be substituted for Hamza at 
discretion. In modern dialects, however, the change is made 
in some other words chiefly in Oman where we find 'ay7i 
" where ? " sounded hayn, and so liah&n for 'aid 'ayn, 'ahen for 
'aid 'ayn, and f hen iorfi 'ayn. The ancient grammarians also 

recorded four verbal forms L?'^;*, -^'^r*' TT',^' J^> 

in which they say ha- is substituted for 'a-, but in this they are 
mistaken, for ha- is here the survival of the older causative 
preformative which has generally become 'a- in Arabic, 
Abyssinian, and later Aramaic, rarely in Hebrew (cf. sect. 11, 
136, below). 

(ii) Change to more emphatic 'Ayin is entirely contrary 
to the tendency of modern dialects, but we find it 


recorded of the speech of the tribe of B. Tamim who 

pronounced j 1 as ^c- ; such a change is known as Al»lci . 
(iii) Change to semi- vowel was rare in classical Arabic with 

initial Hamza, but occurs in a few instances as cIjjJ for 

*— >j I " inheritance ". It becomes more common in 

* ".^ ^ 

modern dialects, and thus we find wahidh or yahidh for A>.| , 


and so in Oman ways for ^"1 ( = i^*pL) "thing", 

and yila for ^1 "to"; in Mehri wahhar for "^^1 

"postpone", yems ioT ^^1 "yesterday"; in Tlemsen 

?/ms for ^*| "chosen friend"; Maltese ?/e7iar for J^ I 
" another ", etc. 
(iv) The loss of initial Hamza with an unaccented vowel is 

fairly frequent in modern dialect, thus hadioTX>-\ "one", 

kbar for ^ I , etc. 

(6) Hamza commencing a syllable after closure 

After a consonant Hamza is normally retained, but even in 
classical Arabic it is sometimes ehded, its vowel being 
transferred to the preceding consonant, which thus becomes 
the commencement of a syllable instead of the closure of the 
preceding one, and before a- this change is not only usual but 

obhgatory : thus Ai L^a may be sounded ma-sa-la as 

^^ 'f. 

though ALu*^ , but (^ I j^ must be sounded yarix as 

the Hamza is followed by long (Z : so Jll«<l, where a 


prosthetic vowel is added before the s- "which would otherwise 
be voweUess (cf. sect. 66), as it is immediately followed by 
Hamza, may become sal, the loss of the Hamza consonant 
causing the s- to be vocahzed by the following vowel so that 
the prosthetic i- is no longer necessary. In dialect where it 
happens that Hamza is one of a group of three consonants 
produced by the fall of final short vowels as case endings or 
personal terminations, it becomes necessary either to insert 
a vowel or to omit a consonant, and where one of the conso- 
nants is Hamza the invariable course is for that consonant to 

be omitted ; thus in Egyptian dialect u I CZ) " I was " 
becomes kuntana. 

(c) Hamza commencing a syllable after an open syllable 

In classical Arabic Hamza in this position is generally 
retained, but after long a and before i or w it is given an 
intermediate sound, inchning towards iv before u, and towards 
y before i ; after u ov i and before a it frequently becomes a 

semi-vowel, so that we may get *^*l2>- for original 

*U_Ji2.>- " fault " and o3y^ for »3y^ " read ", etc. 
Less frequently Hamza becomes w between a-a as in j y 

for /vj*, but regularly aJa becomes a as in A-«<U for Xjm I U 

" how hard it is ". Only occasionally does Hamza in this 
position become h, as hJiinnaha for laHnnaJca. Here also 
ehsion may occur, as when -i'ay becomes -ay and so -e in 
Egyptian fen for fi 'ayn, and similarly -a'i becomes -i in 
willa for wa'illd " if not ", and -a in the Omani ivalla for the 
same expression. 


{d) Hamza as the closure of a syllable 

As we have already noted, at the time when the Qur'an was 
pubhshed the Hijazi dialect had abeady inchned to omit 
Hamza in this position and lengthened the preceding vowel in 
compensation, but the Arabs of Nejd, and notably the tribes 
of B. Tamim and Qays, sounded it true. Thus in Arabic 
generally a' becomes a, as rds for ra's " head ", so m' becomes 
u as mumm for mu'mm " behevers ", and i' becomes * as in 
bir for bi'r " a well " ; and so even when the Hamza is in one 
word and the vowel in the preceding one (cf . Qur. 9, 49, etc.). 
In all modern dialect this is the general rule when Hamza and 
the preceding vowel are in the same word, but sometimes in 
dialect Hamza after a may become a semi-vowel, usually y, 
thus in 'Iraq qara'ta is pronounced qaret " thou didst read ", 
the -ay- resulting in -e-, and ma' " water " becomes may. In 
this last instance, as with final Hamza in some other words, 
the change to y must have taken place at an early date as we 
find t}iay for " water " in Abyssinian also, as well as in Hebrew 
plural m^ylm ; and so samu' " heaven ", Abyssinian samay, 
Hebrew sdmayim, Aramaic s^mayd, Assyrian samdw. It is 
possible that we have another instance in sia' " sheep ", 
Hebrew *'*b^ , though we find the root also treated as rT'K^ 
and s'w. 

(e) Hamza in pause 

After a closed syllable Hamza in this position is really the 
commencement of a syllable, but has become final and has 
lost its vowel by changes due to the pause. If the last word 
is an indetermined noun and therefore normally ending in 
tanwin, the final -n falls because in pause and the word thus 
appears with a final vowel, as will often be the case also if it be 
a verb. If the consonant before this vowel ending be 
Hamza and the preceding consonant be not h or y the vowel 
then suffers metathesis, or else the Hamza becomes a semi- 


vowel, or else both Hamza and its vowel fall away, and thus, 

e e 

in pause "i^j necessarily becomes ^^j, and may then be 
read as ridu', riduw (ridu), or rid ; and so i^la.) becomes 


iJa> , butu', butu, or but, and so in the accusative rida', 

ridaw, rid, etc. In the dialect of the B. Tamim the first of 
these forms was preferred, but the vowel was assimilated, 
thus ridi' for ridu\ 

When the loss of tanwin and the fall of the final vowel 
leaves Hamza as final after an open syllable the Hamza becomes 
the closure of that syllable and may either be sounded true as 

^ I for "^^^J I , or following the precedent of the 
Hijazi dialect it may fall away with compensatory lengthening 

of the preceding vowel so that J) I is sounded 'akmu. 


In Abyssinian initial Hamza is generally retained in the 
older form.s of the language, but the tendency we have noticed 
in Arabic dialect to change it in this position to a semi-vowel 
appears in Amharic where, for example, Ge'ez A'^H 
i^ahaza) " took " becomes yaza. As initial of a medial syllable 
following a closed syllable it is retained, e.g. in ^asdn 
" sandals " ; but after an open syllable, though retained in 
Ge'ez, it is in the later dialects treated as we shall find that the 
semi- vowels are treated when intervocalic, that is to say, the 
following vowel is lost and Hamza then becomes the closure 
of the preceding syllable which results in its fall with the 
compensatory lengthening of the preceding vowel, thus in 
Amharic 'ahazu " took " becomes *'a'azu (cf. § 12), then 
*'a'zu and so ydzu. 


As a final Hamza appears in Abyssinian script but usually 
as a traditional survival only. In this position it is silent, 
and the preceding vowel is lengthened, showing the same 
change which we have already noted in the Arabic of the 
Hijaz and in the modern dialects. Thus we find written 
ma'kal " food ", mala'ket " angels ", but these are pronounced 
mdkal, maldket, and stand for original mtVkal, mala'ket. We 
have already noted that in early Abyssinian final Hamza had 
become -y, as in may " water ", samay " heaven ", etc. 


In Hebrew initial Hamza is usually retained. Change to h 
appears in later forms such as '^'^T} for *^''^s " how ? " in 
1 Chron. xiii, 12, and Dan. x, 17, and occasionally in such 
forms as ■Qiih = UTib " tire "• Change to w does not 
occur with initial Hamza, as Hebrew and Aramaic show ^reat 
reluctance to employ this semi- vowel as an initial ; but in 
rare instances change to y appears as in nn"* for 'IPIX in 
^rt'*'''! Kethib in 2 Sam. xx, 5. Occasionally ehsion of 
initial Hamza takes place when it is followed by a half-vowel, 
thus ^ini for -liniX " we ", which appears in six places 
(Gen. xlii, 11; Exod. xvi, 7, 8; Num. xxxii, 32; 2 Sam. 
xvii, 12; Lam. iii, 42), and *Tir "who" for y^^ 

which appears with assimilation of the r in 'TlpDSJ^ 

(Judges V, 7), ^^)>m (2 Kings vi, 11), and in late Hebrew ^W- 

Once only does this elision take place with a short vowel, in 
nn for nnX " one " in Ezek. xxxiii, 30. Such elision of 

Hamza has its parallel in Aramaic and should probably be 
regarded either as evidence of northern dialect or of collo- 
quial Hebrew. Occasionally change to semi-vowel occurs in 
the case of i«{ as the beginning of a medial syllable, as in 
D^^ (for *mwam-) for Di^l " be high ", which latter form 



is found in Zech. xiv, 10. Only very rarely does medial ^^ 
become aspirated as H, as in T\T]^ for TM^2 " be humble " 
and in urh cited above. Rare also is the assimilation of 
initial X to the consonant closing a preceding syllable, as in 
rhp for T\ihj2 " fullness " in Cant, v, 12, which was the 

regular usage in Assyrian. 

As closure of a syllable t< is preserved in the consonantal 
text but is generally lost in the Masoretic pointing, which 
Bhows exactly the same tendency as appears in the Arabic of 
the Hijaz and in later dialect as well as in Abyssinian. 

Occasionally it is retained as in N'13 " wild ass ", i^^'l 
" fresh grass ", and a few other words. According to the 
more general tendency^ tJ^X^I "head", originally no 

doubt rd'S (Arabic ^j'l'j), has become mi and so ros 

(cf. 43c below) by fall of i^ and compensatory lengthening 
of the preceding vowel. In such forms as XpH " sin ", 
i^^-1 " vallev ", ^^1^ " evil " of the types qatl, qitl, qutl, 
in which the K has failed because it appears as a final 
unvocaUzed by the decay of the case endings, it is merely 
a survival in the written text and has no consonant value. 


In Aramaic the general tendencies are much the same as in 
Hebrew. Occasionally initial J< becomes li as in Samaritan 
|1Dn for |1DX " multitude ". Like Hebrew it does not 
change initial K to w, but change to y is rather more 
common, thus Bib. Aram. T\\ for Hebrew Hi^. In 
Syriac as generally in Arabic dialect X has lost its value as 
. a consonant, thus "1tt^^ > '^l (emar), etc. The ehsion 
of initial X with a haH-vowel has become the regular rule, 
and thus we find "IPl " one " for in{< (Dan. ii, 31), 


and similarly in Syriac and Samaritan (e.g. in Gen. xxii, 13) ; 
(AI» for YlL] "sister", etc., and in Tg. Jer. Xi for 
Bib. Aramaic Ni^{ as the 1st person singular of the 
absolute pronoun. As initial of a medial syllable after an 
open syllable Aramaic rarely retains i<, it is either elided or 

changed to a semi-vowel; thus 6^'es (Arabic ^j-i>) "be 

bad " becomes uil^ (6es), and verbs with medial or 
final ' show a general tendency to assimilate to those with 
medial or final wly. Always X as closure is lost or else becomes 
a semi-vowel. 


The Hamza is not represented in Bab.-Assyrian script, but 
this is merely a matter of orthography : its presence is clearly 
proved by assimilations and other phonetic changes. As 
initial it sometimes becomes a semi-vowel, e.g. in yaii for 'aii 
" I " in Tigl. Pil. viii, 60. In contact with another consonant, 
either commencing a syllable after a closed syllable, or itself 
as closure, older Bab.-Assyrian regularly assimilated it to the 
consonant with which it was in contact, but later Assyrian 
dropped it with compensatory lengthening of the preceding 
vowel, thus zar'n " seed " becomes zarru in the older form of 
the language, zaru in later forms, and so kallatum for 
kaVatum in Ham. K.U. 9, r. 74, etc., whilst adcCnum becomes 
older adannutn, later addnum : ^tDH gives older hittu, 

later hitu, and so i'lik "go" (Heb. ^Sn, Arabic tilU) 

shows illik in Tigl. Pil. ii, 65 ; whilst in later forms we get 
ana sa-al alani where sa-al stands for l'^'^ , so zihu for 

Heb. ^XT (Arabic i^>), and ^Vl after becoming /Xl 


(cf. 13 below) finally produces Bel. As medial intervocal, 


i.e. initial of a syllable following an open syllable, Hamza 
is treated as in Abyssinian, first the vowel following 
is dropped, and then Hamza is dealt with as closure of 
the preceding syllable. Thus i-a-haz becomes i'-haz and 
then ihJjaz, and ma-'a-du after becoming md'-du results in 
mddu, thus illustrating the earher and later methods. 

The ' of ancient Egyptian sometimes corresponds with 
Semitic y, as Jim't = Heb. V''^n " salted ", and it also 
appears as equivalent to r, as k'm = UlD " garden ", 
q'b = yip " middle ". 

11 (ii) Laryngal h 

The laryngal h is the voiced correspondent of Hamza. In 
Hebrew the letter H is also employed at the end of a word to 

denote the vowels a, e, as in nj^S?, Tw^' , etc., and 

final consonant h is properly marked by the point Mappiq 

(H). In Arabic o is not used as a vowel letter, but in the 

form of 4.1?^^^ I tUl [o-j it stands for the final -t of 

fem. -at which becomes -a in pause. In the Semitic languages 
generally the tendency is for h to become non-aspirate Hamza. 
In Maltese dialect h regularly becomes either Hamza or 
emphatic h. 

The most striking change of h to Hamza occurs in the 
preformative of the Causative stem of the verb, where ha- 
itseK is derived from an earlier sa- (cf. 17 f.). 

Change of h to semi- vowel occurs occasionally with medial h 

in Hebrew and Aramaic, as TTlD = nJ2 " circumcise ", 
^r\n = nn " age ", Heb. y^^ = Syr. ^ai> " run ". This 
change is more frequent in T.B., but there also h sometimes 
becomes emphatic h and so in East Syriac. In neo-Punic 
n and n often become {< or ^ as in yij^ for XP (Cooke, NSI., 


58, 2), and K^J^yi for tyX:ni (id. 4). In Assryian all the 
laryngals have become Hamza or h. 

12 (iii) The Laryngals h, h 
(a) Arabic 

Of the two laryngals 7i, h, the former is an emphatic aspirate 
h, and consequently a voiced correspondent of P- , so that 

^ : d : : P- : t. Hence, by loss of emphasis, rr- becomes 

A, as in Arabic ^^ for j>~ "make remote", <l4 for ^?cl« 
" draw out ". This change is very frequent in South Arabia, 
where ^ seldom retains its emphatic value, and so in Maltese. 

By loss of aspiration j?- becomes P as in /e-^' for /^l 

The consonant h (Arabic '?-) has a sound like the ch in 
German ach, " but, I suppose, in spite of the contrary state- 
ments of Wallin {Zeitschrifl, ix, 35) that the modern t- had, 

or has here and there, an alternative palatal sound (the 
German and modern Greek icJi sound) " (Vollers, ix 

Congr. of Orient., London, 1893, ii, 141) ; thus 'r- is used to 

transliterate Greek %, as in U^^tlL* for yii\avyo\la, etc., 

and so t-- sometimes appears as k, as in Sabaean DD22D for 

^j^'^ "five", and Omani 'yamse, fern, j^cims (id.). In 
Maltese ^ and r- are confused, both being commonly 


rendered h, less often h ; but in the eighteenth century the 

two sounds were still clearly distinguished. In Mehri ^ 

becomes h or s, as serlr for j>- " murmur", zaylh for 'rJJ 

" rancid oil", and very often this resultant h weakens to h, 
and so we get zaylh, etc. 

Occasionally in classical Arabic both r- and r- tend to 

become palatal k, as in AAj for AJL>- " mole ", cf. AX — 

corresponding to Heb. "in^ "give a present", but such changes 
are rare though they tend to endorse the ich sound and to 

suggest instances of confusion of ^ with r-. 

(b) Abyssinian 

In Ge'ez both h and h are represented, but in Amharic 
they are confused and both tend to be reduced to h, or 
as initial consonants to Hamza. As the original value of 
h was lost in Amharic, a new consonant letter ""tl (^ ) was 
introduced for use in loan words from the Arabic, but in 
Tigre Arabic /' in loan words commonly became Ic, as in tank 

for Arabic ^^ "epoch, history". 

(c) Hehrevj and Aramaic 

In Hebrew and Aramaic the two sounds h and h are not 
distinguished in the written script, both being represented 

by n which is taken as equivalent to h ; thus Arabic yi>- = 

Hebrew ^IfiPl "dig", Arabic ^^ = Hebrew ISH "be 

ashamed ", without difference in the consonant text. In the 
LXX. n is reproduced as x or \ thus 11711 = X"^"X or 14\ae, 


^"7*111 = x°^^ ^^ 'ASSi , etc., with H sometimes weakened to ^s• 
In later forms the change of H to y or X , i.e. loss of aspiration, 
is frequent as in Syiiac jn^iJ= Hebrew plPI "embrace". 
In later Aramaic, especially in Galilaean, Mandsean, and T.B. 
n frequently loses its emphasis and becomes T\ , thus Hebrew 
|nj (in pnil "belly (of serpent)" Gen. iii, 14) =Syriac .01.. 
" bend ", etc. In the neo-S}Tiac of Ma'lula h has generally 
become //, but h occurs in Jjarufa " sheep ", huttuma " servant ", 
and in loan words from the Arabic (cf . Palest. Explor. Fund, 
April, 1890, p. 86). 

In Hebrew and Aramaic the laryngals H, H, ^, ^^, and the 
laryngal sonant r do not admit of doubling. When the 
morphology tends to cause doubling either (i) this is discarded 
and the preceding vowel receives compensatory lengthening, 
as "Tj'lS for birrekh, the e being due to lengthening i (cf. 
sect. 47, below); or (ii) with H, H, occasionally with "f^, and 
very rarely with X, there may be a virtual doubling which does 
not appear in the written script, that is to say, the short 
vowel is retained in an open syllable contrary to ordinary 
usage, thus C^Hinn, b^Hp, yilT}, etc. 

(d) Assyrian 

In Assyrian all the laryngal sounds have become Jj or '. 

Thus, Arabic 'ali = Assyr. ak-u " brother ", Arabic /y2.>- = 

Assyr. liuss "fence", Arabic XoA = Assyr, aUazu "take", 

Arabic ^j-»-a^ = Assyr. lim "cover", etc. Very often ]>\^i 

becomes ', as in *raliamu>*ra^amu>rdmu, *yahsid> 
*ya'sid>esid, etc. Occasionally h becomes k, as in Assyrian 

iptak " open " = Heb. HHS, Arabic ^^. 


13 (iv) Laryngal ' (P) 

Laryngal ' (Arabic P-, Hebrew y) is related to Hamza, so 

that ' : ' :: A : A (of. Sievers, Phonetik, 353-4). The tendency 
in Arabic dialect is for this laryngal to lose its emphasis and 
become Hamza. In Maltese it frequently ends by disappearing 

altogether. In Arabic ^ may lose its emphasis and become 
Hamza, or it may gain aspiration and become h. Thus, in 

Mehri very generally, as mdlem for/jL«^ " teacher ", which 

suggests intermediate mu'allim, followed by vowel 
assimilation, and so in Malta and in North Africa. In classical 

Arabic such weakening is rare, but instances occur as in c-jl I 

for t_jLc^. In the dialects of Egypt and North Africa ^ 
sometimes becomes r- before a consonant, as samiht for 
O*.*-"' " thou hearest " ; and thus in Mehri also before a 


vowel, as liayr for j^ " ass ". Such a change as this is rare in 
classical Arabic, but we find ^ j for *^ j " camel born in the 
spring ", and a few other hke instances, some, at least, of them 
capable of being explained otherwise. Change of P to l> is 
peculiar to Mehri, where P- always becomes Hamza, or h, 
or q, as badaioq for «Jaj " spUt". We also find P changed to 
P in modern Nejd, as ^j^^^ for (3'-*P' "deep" (Doughty, 


^^ ^ 

Travels, ii, 292), ^y2 for ^Ic- "thirst", '^^y^ for ^"J (a 

dialectal variant of -Uj) according to Ibn 'umm Qasim, 

^«^jl , *^_^*»^, etc., in each case with P- for ^. 

In Abyssinian no distinction is made between the ' and 
(/ sounds, and in Amharic both become ', " ut in manuscriptis 
hie quoque multae inveniuntur confusiones " (Praetorius, 
Granim. Aeili., 4). 

In Hebrew and Aramaic the two sounds ' and g are merged 
in one consonant y. In later Hebrew the tendency was to 
change J^ to ^^, which is then often elided. Thus, Hebrew 
7^3 becomes 73 by intermediate *7K1; so ''3 for '•^3 (Gen. xliv, 
18, Judges vi, 13, etc.), yiDJ^^X in Joshua xxi, 14, corresponding 

to ittriSJ^J<» in Joshua XV, 50; so Arabic j^2^,Hebrew p7ptJ^^^ 
(pr. n.), etc. This change of ^ to t^ is carried much further 
in Syriac, neo-Punic, Samaritan, and T.B. Thus Syriac 

Vii] " meet ", corresponding to Arabic ij^^ (Hebrew V"!^ 

" terrify ", a secondary meaning in Arabic), Syriac |j>oio] = 
)j>oiQJ "memory". Cf. DV3 "grief" (Eccles. i, 18) = DD 

In Galilaean ^^ and ^ were confused, for which reason the 
inhabitants of Bethshean, Haifa, and Tabaon were not allowed 
by the Jews to read public prayers (T.B. Erub., 536). In 
T.J. ^s, ^, and H are often confused. In Manda^an all the 
laryngals and spirant velars tend to become J^, whilst y is 
used as a vowel sign. 

In Assyrian ' becomes ' or A,. 


14 (V) Velar g (P) 

The sounds ' and {) are distinguished in Arabic script only. 

According to VoUers (loc. cit)., {j (P) is the voiced corre- 

spondent of A, and thus we find Jhi' for J)i2.>- "vibrate", and 

/j>-\ for ^c-l " speak through the nose " (cf. Howell, Arabic 

Grammar, ii, 39). More commonly P tends to become P-, 

^ y' y ^ . • 

especially in South Arabia, and thus we get A^J for /;^J 

"agree to", and ^J^ for b^^ "young man", etc. So in 
Maltese, although in the eighteenth century (j was still 


distinguished from '. Amongst the Algerian Bedwin ^ 
becomes g, and in all dialects it very frequently tends to 

become g. Greek 7 followed by a, 0, w, becomes P as 

A.0709 = A*J ; so in modern speech the French gaz appears 

as JU. 


In Ethiopic 9 is confused with f- , and both are represented 
by one letter (Q), but sometimes Arabic 9- appears as g, 
thus A«j = hac^, " mule." In Amharic both become '. 

In Hebrew and Aramaic ' and g are represented by y, but 
it seems clear that this letter had two sounds, and so we find 

V = f" frequently transhterated by 7 in the LXX, as HTy 
(Arabic dj^) = Td%a, TV^ = Tofivppa, whilst the pure 


y = c is rendered by h or ', as vj^ = 'H\i. On the basis of 

the value ^ = f- the ^ can permute with g, h, q, or r, as 
pi = "[^3 " flow ", toya = D^/!: " poHsh ", etc. The modern 

Ashkenaz make ^ = 9' sound as ng, thus !lpy'* = Yangqoh. 
In later Hebrew and in Aramaic the tendency is for ^ to become 

K. The neo-Syriac of Ma'lula sometimes revives the 9- sound, 

as in guhura, " dust," but this is due to Arabic influence. 

In Assyrian the g sound is merged in y, and both become 
' or h. 

15 (vi) Velar q 

The consonant l5 is described as having two sounds in 

Arabic (Ibn Khaldun, ed. Quatremere, Paris, 1858, iii, 171-2), 
and these are identified by Vollers (op. cit., 138-9) as soft 
emphatic g (7^) and hard emphatic h {k), for both of which 
5- is a convenient symbol. These two values appear to be of 
equal antiquity, and neither is described as vulgar or dis- 
approved. The soft sound is preserved in Upper Egypt and 
amongst the Egyptian Bedwin, e.g. qdla ^ Bedw. gal, " say," 
and also in Jedda, Mecca, Nejd, and 'Iraq. In Nejd we find 
q = g OT g, thus faqlr =feglr, "poor," but plural /w^^ara ; 
qadum = gedum, " hatchet " ; qatt = gett, " vetches," etc, 
(Doughty, Travels, ii, 605). In 'Iraq q is sometimes hard, 
and sometimes it becomes g ot g or even c, thus qahwa or 
gahioa, " coffee," qarib = garib, " near " ; qamr = gamr, 
" moon," etc. In Mehri q becomes g or k, as letog, " kill," 
for qatal (with metathesis), qaraza >keraz, etc. In the dialect 
of Nazareth q>k. The vernacular of Cairo, of the province 
of Qalyub, Neosta, most of the Fayyimi, of the towns of 



Syria, and Malta reduce g to ' or Hamza, thus biqadde ay = 
badde ay, " how much ? " (Cairo), yaqtul becomes yigtil 
(Upper Egypt), yi'til (Lower Egyjit) ; qadini = gadlm 
(Upper Egypt), 'adlm (Lower Egypt), etc. 

In Tigre q at the end of a syllable tends to become ', as 
te'talo for teqtalo. 

In Hebrew q shows a double value, as in Arabic, the one 
hard, which changes sometimes to k, the other soft, tending 
to change to g, but in some cases both values appear showing 
that the soft and hard are obscured by the laryngal emphasis. 
Thus qadda (Arabic) = Heb. kddad or gddad ; in Hebrew we 
also find the tendency to change q to ', as zdraq = 2am', 
" scatter." 

The same holds good for Aramaic. Thus Heb. sdqed = 
Syriac segdHhd, " almond-tree," Hebrew qassdth = Syriac 
kassdtd, '' archer," Mandaean StOtJ'ID ; from qtl is formed 
ykflnk (-|iSt03\ Cooke, NSL, 64, 11), " he will kill thee." In 
the dialect of Ma'lula q becomes k {Pal. Exp. Fund, Jan., 
1890, p. 87). 

In early Babylonian q generally appears as soft, and thus 
frequently changes to g, e.g. qaqqadu = gagadu, " head " 
{gagidi-su in Ham. KU. 25r. 86), qaat = gaat, " hand," etc. 
A similar change appears in Sumerian, where qal = gal, 
" great," qal = gil, " demolish," etc., so that this tendency 
to confuse q and g is possibly due to Sumerian influence. 

16 (vii) The Palatals 

The original palatals were two in number, hard k and soft 
g, but from these are derived (a) aspirate palatals kh and gh 
in North Semitic, and (6) palatalized c and <j (j). These 
derivatives are, in most cases, due to the action of a neigh- 
bouring vowel (cf. 37, below), but in so far as g appears also 
as a dialectal variant of g, it must be included here. 


In Arabic the sounds Tc and g are represented by the 
consonant iJ (Mufass. 189, 10). But the native grammarians 
censure the sound g as unclassical when associated with this 
letter. The Semitic g is usually represented in Arabic by t", 

but this consonant represents three sounds : (i) palatal g, 
a sound which Ibn Yaish (fourteenth century a.d.) describes 
as confined to Yemen and to the lower classes of Bagdad, 
but its extension to Egypt is proved as having taken place 

as early as the eighth century a.d., when we find (jlk-*«:>- side 

by side with ijus^ and jUs-*^ as transliteration of the Latin 
quaestor in a papyrus of a.h. 90 (a.d. 708-9), and is now 
extended also to Oman ; it is rarely heard in Nejd. In the 
transcription of non-Semitic words g is represented in older 

forms by Arabic r", in later ones by P- or j^ (cf. p. 48 above). 

Only in colloquial Egyptian do we now find g transcribed 77, as 

French ^'az, properly Jlp, but in Egypt commonly as jU-. In 
Morocco r- occasionally sounds as g before a sibilant. 

(ii) Generally 7^ = g (cf. English change of ^ to c in speak, 
speech). In Mehri r" sounds as q or g. In Nejd it becomes 
c, and so in Bethlehem. In 'Iraq, Palestine, the Syrian 
littoral, and amongst the Christians of Jerusalem r;- = s. 
So Persian and Turkish soft g usually become Arabic s. 


(iii) Sometimes also rf lias the value of dy. Thus, 

occasionally, in Sinai, as is reported, although it does not 
seem generally known, in some inland districts of Palestine 

and Syria, and (rarely) in Lower Egypt, e.g. \^ = dyamal, 
" camel." As a result of this, perhaps archaic, sound, we find 

the chansre of 7- to li as in classical , ^JLs>- = , ^i-^i 

" coarsely ground wheat ", and in the dialect of Damascus 

J' " 

A^U- = ddssa. 

As another result of this dy or iotacized d value, we find ^r 

sometimes replaces y, a change to which we shall refer when 
discussing the semi- vowels (cf. section 20). 

(iv) As already noted, emphatic ^r is represented by q (cf. 15), 
and in Mehri g often has this value of q. 

In Mehri h usually becomes Mi, and hence h, as ha->ha- 
" hke," and in the dialects of Egypt, Damascus, and North 
Africa k sometimes becomes h or Hamza. 

(v) Abyssinian shows k and g (not g) corresponding to 
Arabic k and g. In the dialects of Tigrifia and Amharic k 
often becomes h, as Arabic kdna> hona, "become" (cf. 
section 37). 

(vi) Hebrew and Aramaic possess k and g and also corre- 
sponding aspirates kh and gh, the aspiration being due to 
vowel influence (cf. 37). In Aramaic g sometimes becomes 
', as Hebrew galgal = Syriac 'al'ald, " whirlwind." In the 
dialect of Ma'lula g becomes g or g, thus sugrd>segmthd, 
whilst k>kh, the explosive k being retained only in loan 
words from the Arabic, as O.Syr. kHhdbd>Md.\. khathdbd, 
" book." In early Aramaic sometimes k >g, as Assyr. saknu = 
pD (Cowley and Sayce, Ara7n. Papyri, 13). 


(vii) In Assyrian both palatals k and g occur. As already 
noted, g often stands for q in early Babylonian (of. sect. 15). 

17 (viii) The Dentals and Sibilants 

(a) The transmission of the dental and sibilant sounds in 
Semitic may be represented by the following table : — 

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) 

Arabic . . .tdddtzdss s s z 

Hebrew . 
Aramaic . 
Assyrian . 

tdszt.'jidssss z 

tdsztssss s s z 

t d t d t t ',q s s,s s s z 

tdsztssss s s z 

(h) Taking first the simple dentals t (1) and d (2), we note 
that they are transmitted without change throughout the 
Semitic languages. To this, however, there are exceptions, 
(i) in local dialect and (ii) in modifications due to the influence 
of neighbouring vowels or consonants. In North Morocco, 
some parts of Algeria, and especially in Tlemsen, t has become 
ts, a change probably due to the influence of the non-Semitic 
Kabyle population. The dental t appears as ts or th in the 
Kabyle dialect of Zouaoua and Bougie in North Algeria 
and the kindred Berber of Ghdames, and also, it would 
appear, in the now extinct dialect known as Guanche. The 
two first named prevail in very much the same area as the 
Tlemsen and neighbouring Algerian dialects of Arabic. 
Ghdames Hes to the south of Tripoli, and does not seem to 
have produced a parallel influence in the Arabic of TripoH ; 
whilst the Eif dialects of the Berbers of Morocco do not show 
this change of t to ts. But the Berbers of Ghdames and the 
Rif country have not had the same contact with their Arab 
neighbours as the Kabyles of Algeria, and so we may suppose 
that ts for t is a North Algerian change which has passed over 
into Morocco. 


In Ainharic t-\-t>c and d-{-t =J, a change due to the 
influence of non-Semitic languages in close contact. 

In Hebrew and Aramaic t and d regularly become th and 
dh under the influence of a preceding vowel (cf. 37). 

(c) A striking peculiarity of t is its tendency to fall away 
when used as a final. This generally takes place in Arabic, 
Hebrew, and Aramaic when in the feminine noun termination 
-at, which thus becomes -a. In Arabic such endings are 

generally written a - with what is known as aI?^)^^ 1 iLji 

and sounded -a in pause. As the case endings are obsolete in 
modern dialect, it there invariably becomes -a. Occasionally 

this weak -t is found also after a long vowel, as in 6Uj>. 

" mother-in-law ". In Hebrew and Aramaic H- or H- is written 
as circumstances require. Owing to the loss of the case 
endings the feminine -t has almost always disappeared, only 
the addition of a suffix, the close annexation of a following 
genitive leaving the preceding noun in the construct, or, in 
Aramaic, the addition of demonstrative -d, the so-called 
emphatic form, secure the preservation of feminine -at, 
although the alternative -t {-et) is common. In Hebrew the 
failure of final -t from -at is extended to the 3rd fem. sing, 
of the perfect tense of the verb, so that we get Hebrew Til'Dp 

where Arabic has cJci and Aramaic H/JOp. The fall of 

final -t from feminine -at appears in the earliest written 
Hebrew in the Siloam inscription, where we find such forms 
as n^p^n, etc. It is found also in ancient Egyptian in the 
course of dynasties nineteen to twenty-one (Erman, Aeg. 
Gram., 3rd ed., sect. 174) at a time when there was a very 
close relation between Egypt and Canaan. In the case of 
verbs, as in that of nouns, a sufiix preserves the -t, as in 


■•inipp, etc., and a few instances occur in which it is not 
dropped, as Tt'lii (Deut. xxxii, 36) and T\i<1i^ (Deut. xxxi, 29). 
In Aramaic, though not extended to the verb, the fall of -t 
occurs quite early, and so in the papyri we find nPli^ for T\ni^ 
and n*)l for r\1'2. The loss of final -t has become the general 
rule in Aramaic, and appears, not only in feminine -at, but 
in other cases as well, even after a long vowel or diphthong, 
and thus we find ^:3 for H^l (Cooke, NSL, 63, 16, etc.), and 
so in T.B. Normally the addition of a suffix suffices to 
retain the -t, but this is not always so in later Aramaic, e.g. 
"JoqiALo for IZooILd "kingdom " in Syriac, and ^V^ for iny3 
in Bib. Aram. (Dan. vi, 14). 

In Mandsean t frequently becomes d, and so very generally 
in Aramaic transcription from the Greek, but here it must 
be remembered that in mediaeval and modern Greek t is 
often sounded as d, thus dpTci^Tj = N^Tli^, Syriac ]^yy] 
"a measure", ^dro^ = r\^, Syriac ]^ "beam"; or else 
T > ^ as tl/jlt] = Syriac ]^» ^ " reward ". 

In the modern dialect of Ma'lula t becomes th or c, and 
explosive t occurs only in loan words from the Arabic, whilst 
similarly d becomes dh, as in Dhemsel', " Damascus." 

{d) Taking now the two aspirate forms {th) and {dli), we 
find that these are preserved in classical Arabic and in the 
dialects of Oman, Hadramaut, 'Iraq, the Druses of Lebanon, 
the Bed win of Syria, and of the Nejd, Tunis, Spanish Arabic, 
and sometimes in Oran ; elsewhere these sounds have become 
sibilant or non-aspirate. In Arabic dialect they are generally 
t and d, and thus even amongst the Bedwin of Tunis. Where 
the educated try to render the aspirate sounds they usually 
result in s, z, as is also the case in the colloquial of Mecca and 
Jerusalem. In North Algeria, etc., where the t has become 
is, the aspirate after becoming t passes to ts. In Aramaic these 
consonants have become non-aspirate, but new aspirate 


sounds are produced by the action of a preceding vowel on 
t, d (cf. 37). In Abyssinian, Hebrew, and Assyrian the aspirate 
dentals become sibilants, in each case dh becoming z, but 
th which gives hard s in Abyssinian shows palatalized s in 
Hebrew and Ass}Tian. 

Besides the change of th to ts, there is a fairly frequent 
interchange with /, less commonly 6, in Arabic, as fumma for 
thumma, " moreover,"/uw = thum, " garhc," etc. So/for t in 
mahfid for mahthid, " source," and b for th in the possible 
connexion between the roots bhj, "be white," and thlg, "snow." 
Comparison of roots frequently shows kinship of meaning 
between those which have correspondences between f/b on 
the one side and t/th on the other. 

In the dialect of Mehri th is sometimes retained as in A J Uj 

" eight ", sometimes it becomes s, as in slit for dSj^ 
" three " and hblet, " third," and sometimes it appears as 

f, as in harot for <^^y>- " till the ground ", though this last 

may possibly be an assimilation due to the influence of the 
emphatic letters r and h. 
In the Zinjirli inscription th>s, dh>z, as in Hebrew, thus 

7pD " suspend " = Arabic Ali*. In Phoenician th is 
retained in the word Owp " a bull " (Plutarch, Sulla, 17, 

Arabic ^y), but becomes s (s ?) in salus for Hebrew T\'^Y>'>i^, 

Arabic CSj^) " three ", according to St. Augustine {in 
Rom. vii, 3). 

(e) The four letters Js?, is*, ^J^ , and ,J^ are called in Arabic 

" covered " (^_^U2^), because the hngual outlet of the letter 


is covered by the opposite side of the palate (Mufass.). 
Sibawieh says that but for the covering, i? , Ji' , and ^ would 


be ^, ^, and ^ respectively, whilst ^j^ would not exist 
because it has no uncovered equivalent. 

Of these, Js? is sounded as t and corresponds to the 
same sound in all the other Semitic languages, so the 

error of the inhabitants of 'Iraq, who sounded I? as i 
in t-Ju for v^li? "seeking", was that they sounded 
it weakly and thus produced a value between t and t 
(ar-Radi al-Astarabadi), i.e. the i is a ^ uttered with a 
laryngal effort or emphasis. But Sibawieh treats t as an 

emphatic d, and this soft ]a is admitted as a secondary but 

correct value of i?, which thus like (3 has two sounds, a soft 
and a hard. But the soft emphatic sound is now obsolete 
for i?. 

The emphatic ^j^ or .s also appears throughout the Semitic 
languages. Here, again, both the hard and soft sounds are 

admitted as correct ; but the soft z is now used for ii and not 
for f^ . In such forms as ^X.,jla the softened ^j^ is merely a 
case of assimilation. It is admitted that a weakly sounded 
5 resembles s, so that we may regard the hard sound as more 
fully endorsed. 


The original values of Ji> and ^j^ are less easily stated. 
At present the accepted pronunciation is i^ = 5: and^j^ = d, 
the voiced equivalents of (j<? and i?. In Hebrew and AssjTian 
both these are merged in s, but Abyssinian makes 1? = s 
and ij^ =-- d. Thus Abyssinian agrees with Hebrew and 
Assyrian in voiceless ^, as against modern Arabic voiced 
z and Sibawieh, who describes it as a covered form of dh. 
This latter view gives :i \ > W ±? '. ±?, and so, if i?, which he 
describes as voiced, be really voiceless t, then i? is voiceless 

aspirate tli, i.e. the aspirated equivalent of 1?. This seems 
to be supported by the script, where an extra diacritical mark 

over a letter sometimes denotes aspiration (e.g. ^, ^ and 

Cj , CS), and aspirate th is actually the sound heard in some 
Bedwin tribes. It is admitted that, as a weakly pronounced 
1? in 'Iraq sounded like t, so a weakly pronounced ]a> became 

til. In Aramaic ii> = ^, as though I? : ii» : : O '. cl> also 

supporting the voiceless original ; but we also find ji> = .5 
(or /), as Tupo? = "1^ on the Seleucid coins ; so t^^^D (Cooke, 
NSI. 68, 19). The change of sound in Arabic seems to have been 
mediaeval, and possibly due to Turkish or Persian influence. 

The case of ^ is more difficult, for it had no uncovered 


equivalent (Sibawieh). But in this perhaps Sibawieh was 
mistaken, and the reason why he could not find the uncovered 

equivalent of ^ was that its " covering " differs from that of 

the other three consonants. For " coverinsr is your coverino' 
the lingual outlet of the consonant by the opposite part of 
the palate " (Mufass.) ; but the term " outlet ", says ar-Radi 
al-Astarabadi, is " not universally applicable, because the 

outlet of 1^ is the side of the tongue, whereas the side of the 

tongue is covered by the molars " and not by the palate. 
This suggests a lateral covering, and agrees with Mehri 
lateral d and Hadr. \ \ compare also the approximation 

of ^ to / in *3fti2)l for ^otJa^l " he lay down " (but also 

^3tl?l ). It is mentioned by al-Astarabadi as an error of those 
foreigners who have learned to speak Arabic that they sound 

^ as )a> or between ^ji9 and Ji' ; but Mbn. describes a weakly 

sounded ^j^ as approximating to ill, as in ^^\ for :ij\ . 

The Bedwin generally and the fellahin very often sound ^j^ 

as aspirated emphatic dli, and this occurs also in the dialect 
of Tunis. Elsewhere it is usually given the value of d, in 
Morocco d or t. Amongst the Bedwin of East Tunis and in 
other parts of North Africa, and in Maltese, it appears as 
simple d. The sound z often is heard from those who 
endeavour unsuccessfully to aspirate d, and amongst those 
who have been subject to Turkish or Persian influence. 

The soft sound of ^j^ is supported by the rendering d 
in Abyssinian, but Hebrew and Assyrian show equivalent 


hard s, as in Arabic ^j^jl = Hebrew "^^y^, Ass}T:ian irs- 

itu. Aramaic has earlier equivalent p for ^j^ , later ^ ; 
thus Arabic 'ard = Xp'l^^ in the Ass}Tian weights and 
p'Mi in the ZinjirU inscription (Cooke, NSL, 61, 5), but 
later XV^lS, S^Tian 'ara' : so Arabic darra{t) = Svriac 'arid, 
" concubine " (Hebrew sard, 1 Sam. i, 6). 

(/) The uncovered sibilants 

The original sounds are retained, it would appear, in 
Hebrew alone, for there only do we find the four non-emphatic 
sibilants distinguished as s (D), s (b^), which is pronounced 
with the tongue slightly raised and curved so as to produce 
a concave surface, its tip against the alveolars, s {^) 
palatahzed as Enghsh sh, and z (T). The ordinary s is not 
confused with s, for we have b^t^ " behold " distinct from 
^DD "be foolish", IDST "hire" contrasted with ^IDD 
" shut up ", etc. It has already been noted that Hebrew 
s sometimes corresponds with Arabic th (cf. (d) above), 
but there are other cases in which Hebrew s corresponds 
with § in Aramaic and Ass}Tian and with s in Arabic and 
Abyssinian, and here we presume that S is the original 
Semitic sound. The s became obsolete in ancient Egyptian 
about the time of the New Empire, and was replaced by 
s/s, as is usually the case in Aramaic, although we find s 
retained in yisge' (Dan. iii, 31) and in S^b^l " flesh " (T.B.) ; 
this s corresponds with s in Assyrian, Abyssinian, and Arabic. 

Thus the original s and z remain unchanged throughout, 
whilst the other two uncovered sibilants show the 
correspondence : — 

Hebrew s>s in all other Semitic languages save Aramaic, 
where we sometimes find s or s. 


Hebrew s = s in Aramaic and Assyrian, = s in Arabic 
and Abyssinian. The early similarity of s and s appears 
from the fact that Hebrew used one character ^ for the two 
sounds, the diacritical point being unknown as late as the 
time of St. Jerome (cf. Hieron. in Hab. iii, 4 ; ximos iv, 13 ; 
viii, 12). We read in Judges xii, 16, that the sound s for s, 
as in sibbdleth for sibholeth, was distinctive of the tribe of 
Ephraim, as though that tribe then included, as is possible, 
an Arab element not yet completely assimilated. 

Examples of the correspondence of uncovered sibilants : — 

Arabic. Abyssiniayi. Hebrew. Aramaic. Assyrian. 

— M^ 

" liead " 


re es 




" six " 





V V y ' 


" ten " 






" salute " 





" name " 












It seems, however, that the sibilant sounds were con- 
siderably modified in the course of time. We have already 
noted that s began to disappear in the time of the New Empire 
in ancient Egypt, whilst it survives in earlier Aramaic, but 
disappears from later forms. The change of s to s in Arabic 
was operative at quite a late date, as we see in the loan word 
satdn (Hebrew), which becomes saitdn. Ac a very early 
date it would seem that there had been changes from s/s 
to h and thence in normal course to Hamza. Thus, in the 
personal pronoun Assyrian su, si, Minaean ID, Hebrew 
hu\ hi', Arabic huwa, hiya (cf. below), but both retained in 
Mehri, where he, hi appears as masculine, se, si as feminine. 
So in the causative preformative §- in Assyrian and some- 
times in Aramaic, with very rare survivals in Hebrew ; h- 
in Hebrew and in older Aramaic, with a few survivals in 
Arabic ; Hamza in later Aramaic, in Arabic, and Abvssinian, 


but in these two latter with s- retained in the reflexive st- 
(Arabic ista<itala, etc.). Here, again, Mehri retains both s- 
and h-. Minsean shows causative s-, and this becomes h- 
in Sabaean. So in Mehri we often find h for Arabic s-, as 
Arabic sab', " seven," Mehri hoba' ; Arabic sitt, " six," 
Mehri hitt, etc. 

18 (ix) The Labials 

The labial explosives are voiced b and voiceless p, with 
corresponding spirants bh (v) and/. 

In Arabic only b and.f are retained. Original p has become 
spirant/, or less frequently has been softened to b. Under 
non-Semitic influences modern Arabic sometimss sounds 
p in loan words, this sound being reckoned as a second 

(disapproved) value of c-j, but it is commoner to change the 

foreign p to b oi f, thus ponticus becomes binduq or finduq, 
" the hazel nut," Aramaic sapun (for adirwv), Arabic sdbun, 
" soap " ; TTy^o? (Aram. p^DpS) becomes buqus, " box-wood " ; 
police appears as bulls ; padre as badrl, etc. Foreign v becomes 
explosive b, as burkan, Arabic for vulcano. On rare changes of 
b to m, t cf . Howell, ii, 1333-4. For interchange of fjb and 
th cf. section 17 above. 

In Abyssinian, as in Arabic, only / and b are retained, but 
a new p (T) has been introduced as well as an emphatic 
j9 (^ ). These sounds appear either in late derivatives from 
original b or else in loan words, as papa, " bishop," and 
Qopros for Kv7rpo<i. 

In North Semitic all four labial sounds are retained, but the 
spirants are produced from the explosives by the action of a 
preceding vowel (cf . 37 below). Non-Semitic v before a vowel 
is represented 'by the semi-vowels iv/y, as Valens = Syriac 
Walls, and in later Hebrew veteranus becomes pnD^X. In 


the neo-Syriac of Ma'lula b and / are used as in Arabic, but 
occasionally the former acquires the sound of p, as ob, 
" father " (Arabic 'ab), ippai, " my father." 

Sonant m is closely related to the labials (cf. section 19 
below), and tends to become semi-vowel w (cf. section 39 

In Assyrian the labials 6 and p appear without change. 

19 (X) The Sonants 

There are four sonants, I, r, n, and m. Of these m is allied 
to the labials, n to the dentals. All four tend to interchange, 
as the Semitic languages, though very explicit in the laryngals, 
are liable to confuse the sonant sounds. 

(a) Sonants I, r 

Sonant I is pronounced by means of the whole side of the 
tongue and the opposite teeth ; it is consequently described 

by the grammarians as " lateral " (c-5 1^^, Ibn Yaish, ii, 1466), 

and thus approaches the sibilants (cf. section 36 below). 
Sonant r is described in Arabic as intermediate between I 
and the dental n, pronounced by the tip of the tongue and 
the central incisors ; it is characterized by the " trill " 

{j'^^P^"), a sound found also in Irish-English (Sievers, 

Phonetik, 305). In Arabic this sonant is closely allied with the 
" covered " letters, and has very much the same influence 
on neighbouring vowels, though in a less degree (cf. 56, 57 
below), but in North Semitic it is more definitely allied with 
the laryngals and shares with them the incapacity of being 
doubled. In ancient Egyptian I and r were confused, and the 
same confusion often appears in the Semitic languages, thus 
Hebrew sirsdrd, " chain," appears in Aramaic as silseleth, 
and Arabic salsala{t) ; Arabic barsdm has an alternative form 


halsdm, " balsam " ; Arabic farq, " division," infalaqa, " to 
be divided " ; hayli as variant for hayri in Qur an, xxxviii, 31 ; 
Hebrew 'agar or 'dgal, "roll together," etc. This interchange 
of I and r is particularly common in South Arabia, where we 
find ragga = larjga, " shake," rataba = lataba, " be fixed," 
zarama = zalama, "stop," etc. (Landberg, Etudes, ii,1764, etc.). 

Change between I and n is less common. We find Syriac 
n'^thal for Hebrew ndthan, " give," T.B. l^qat = n^qaf, Syriac 
halfd = hanfd, " impious." In Arabic the change of final -I to -n 
in loan words may be partially due to a tendency to assimilate 
to the formative -an (cf. 120). Interchange of I and n occurs 
in South Arabia, as in zalama = zanama, " cut off," kamana = 
Jcamala, " hide," etc. (Landberg, Etudes, ii, 1758), and so 
Omani berdhul for birdhun, " hackney " (Persian), etc. 

Change of Z or r to a dental appears in Arabic in such rare 
instances as gadd for gald, " sturdy " (Lane, Lexicon, 442), 
and in T.B. and neo-Syriac, where r sometimes becomes d 
(cf. Maclean, Grainmar of Vernacular Syriac, 121 ; cf. also 
section 36 below). 

Sometimes we find sonant I changed to a semi-vowel, as 
in Mehri hawh for kalb, " dog." So in Hebrew, hdlak, " go," 
appears also as hdk by intermediate hivk ; cf. Amharic 
sdst {sawst) for saldstu, " three." 

(6) Sonant n 

The change of w to r occurs in Mehri as ber for ibn, " son," 
so Sinaitic ^i, Syriac bar ; Hebrew bdhan= Aramaic b^har, 
etc. In Amharic n-t>n, due to non-Semitic influences. We 
also find change of n to I, as Arabic sanam = Hebrew selem, 
" image," Aramaic rabbiinl = rabbuli, " lord, my lord," as 
a title of respect (cf. interchange of I and n above). 

(c) Sonant m 

Change to n, or of n to m, appears in the indefinite 
termination nunation or mimation (cf. 132 below). By 


aspiration m tends to become w, a change characteristic of 
later Assyrian, thus Babylonian amelu (Ham. ed. King, iii, 
263)>aM'e^w, Hebrew '^wll, Berossus eueX " man " ; Assyrian 
zimu, transhterated in Hebrew as VT ; Assyrian kislimu, 
Hebrew Idslew, etc. Occasionally we find m changed to a 
kindred labial, as Hebrew rdmd = rdfd " throw", s^demd = 
s^defd in 2 Kings xix, 26. Arabic zaman, Hebrew z^mdn, 
Samaritan z^hdn " time ". 

20 (xi) The Semi-vowels 

(a) Semi-vowel w 

Semi-vowel w is closely related to the vowel u, and also to 
the labials in such a way that aspirated hh approaches vjw. 
In all cases uw becomes u, and usually wu also becomes u, 
except as initial (cf. 51, 52 below). 

(1) Arabic 

W becomes y by assimilation to the vowel i preceding or 
following (cf . 40 below) ; change to Hamza appears in divi >d'i 
in the active participle, etc. (cf. 52 below), but this is not 
observed in dialect, e.g. Egyptian qdyil for qd'il, etc. In 
dialect sometimes huwa, hiya become hu'a, hi'a ('Iraq), or 
ku\ hi' (Datina), and so we find 'uqqita for wuqqita, and 
^dhid for wdhid " one ", where Omani dialect preserves wdhi, 
Moroccan wahad. The loss of initial w in certain verbs is 
not a phonetic change, but due to the analogy of the imperfect 
reproduced in the imperative (cf. 149 below). Sometimes 
Minsean and Sabsean show a change of w to y, as in ^H^, 
corresponding to Arabic wathaba, Assyrian iv-s-b. 

(2) Abyssinian 

In Tigre w frequently becomes y and hence g (cf. b below). 


(3) Hebrew 

Generally initial w becomea y, thus Arabic walada, Hebrew 
*l7*, etc. Initial w survives in the. conjunction wa- (1, 1, ^) 
and in 11 "nail", "111 "carry", '^jSl (Gen. xi, 30), and" in 
some proper names, such as DHI (Num. xxi, 14), |"11 (Ezek. 
xxvii, 19), etc. But w which has become y is restored when 
doubled, as in Iti'l* (Nif'al from lE^** for w-s-b). It usually 
becomes quiescent after a vowel (cf. 51, 52), but to this there 
are occasional exceptions, as in the noun form IHD " winter " 
(Song, ii, 11). 

(4) Aramaic 

As in Hebrew, initial w generally, becomes y ; it is retained 

in o, o "and", ]]o "be necessary or suitable", IrQO 
" fish egg ", and a few other words, and in Greek words such 

as ]oo = ova. In later Aramaic medial w becomes h or 
Hamza occasionally, as Hebrew 2^12 = Aramaic tJ'ni "be 
ashamed ", Hebrew T*11 = Aramaic "Dni " run", etc. In the 
active participle w becomes Hamza only in the singular 
(cf. (6) (5) below). 

(5) Assyrian 

The tendency was for initial and medial w to fall away in 
later Assyrian, but in older Babylonian- Assyrian there does 
not appear the marked distaste for initial w which we have 
noted in Hebrew and Aramaic, thus iv-s-b = Hebrew !3K^*, 

Aramaic ȣ3Aj (Arabic i^^), etc. In later forms amvahb > 
an'asab>a"asab, and so a'a^ab, etc. 

(6) Semi-vowel y 

(1) In Arabic the treatment of y is closely parallel to that 

of w. A curious change of y to g, known as the 4.5t»5>eP» 


appeared in the ancient dialects of the B. Tamim, 
Quda'ah, Teiyi, and Asad, especially when y is doubled in 

pause, thus ^_^ becomes t^c- (proper name), so Jj I 

becomes J^l "mountain goat". 

(2) In the dialect of Tigre the change oiytog already noted 
in certain Arabic dialects also appears. 

(3) In Aramaic the change of y to ^^ occurs in some 
adjectives in Bib. Aram, and in West Syriac, thus ^{^b^^ for 
W^2 (Dan. V, 30). 

(4) In the oldest Babylonian initial y was generally lost, 
and so medial y following a closed syllable, but y between two 
vowels was retained. To the loss of initial y, however, there 
are exceptions, as yaf)iu " ocean ", yaahu " enemy ", yaele 
(Hebrew as loan word /V''), etc. 

(5) With both the semi-vowels we note the change of dwi 
or dyi to d'i in the active participle (Primary) of verbs with 

medial wjy. Thus, in Arabic J^U becomes JJ^U , etc. For 
parallels in the other Semitic languages cf. section 153 below. 



21 (i) Assimilation of Consonants 

The consonants transmitted through the Semitic languages 
in the manner already described frequently suffer temporary 
modifications due to the disturbing influences of neighbouring 
consonants or vowels. 

Assimilation of consonants is the change produced in one 
consonant by another consonant in its neighbourhood, which 
modifies it to partial or complete conformity to itself. Such 
influence takes place most easily when the two consonants 
are in immediate contact, without a vowel intervening, less 
easily when they are separated by a vowel, and only in rare 
cases when another consonant intervenes ; this last is for the 
most part confined to the influence exercised by emphatic 
consonants. Assimilation is commonest in the dentals, to 
a less degree in the sibilants and labials, and still less in the 
laryngals. But assimilation is a tendency rather than a 
binding rule ; it is commoner in some dialects and in some 
forms than in others ; phonetic rules can only indicate the 
lines on which it operates when it does occur. 

The strongest assimilating power proceeds from the 
emphatic consonants (cf. 9), which tend to change 
neighbouring dentals, sibilants, and palatals to their corre- 
sponding emphatic forms, operating even when one or two 
consonants intervene. Usually, however, this influence is 
only efiective when the consonant influenced and the one 
exerting influence are in the same word, or if not in the same 
word are in immediate contact. In such assimilations it 


must be noted that s serves as the emphatic form of s and z, 

and that q serves as emphatic of k and g. Thus Arabic JJli 

becomes vt^p in Hebrew and Aramaic by the influence of 
emphatic q, which changes neighbouring t to emphatic t. 
As we have abeady noted (cf. 68e), the emphasis is a kind 
of laryngal sound given to the dental, etc., and this throwing 
back into the larynx tends to give a laryngal tone to a 
neighbouring consonant as well ; or else both are deprived 
of the laryngal effect, and thus in the ancient dialect of 

Kelb jA,»-i3L4 was sounded as j^J^*- 

22 (a) Assimilation of reflexive t 

The reflexive ta- (cf. section 138 below) is the most striking 
instance of assimilation, and appears in all the Semitic 

(i) Arabic 

(1) Conjugations v and vi show taqaUala and taqdtala as 
reflexives of qattala and qatala, but in old Arabic and still in 
Qur'an reading, as well as in various forms of dialect, these 
take the forms itqaitala and itqdtala when the first radical is 
a dental or sibilant, and when this is the case formative t- 

assimilates. Thus we get >«*j 1 for **aj " investigate ", 
and so ^jlA" becomes ijjbl " follow " (Qur'an, xxvii, 68), 

OX..,aZi becomes i^X^^ " and bestow alms " (Qur'an, 
Ixiii, 10), ^jS'j^^ becomes QjS'^jJ^ " they beseech " 
(Qur'an, vii, 92), ^ JI* becomes /^ j^ " be righteous " 


(Qur'an, Ixxx, 3), etc. This assimilation follows elision of the 
vowel of preformative ta-, as will be noted above, and is 
prevented if there has been previous elision of the personal 

preformative ta- (cf. 70 below), thus J) Aaj becomes y X\ 
(Qur'an, vi, 153) and cannot then form *^ -A-.. This 
assimilation survives in various dialects, e.g. in Omani v_j jUaj 

becomes ddoruh, aJ^sj gives ttamma' (Reinhardt, viii, 2), 
etc. In Maltese and often in North Africa it is extended to 
all reflexives of hollow or med'. gem. verbs with initial dental 

or sibilant, thus from (»-5^ we have iddum " delay " 
(Maltese), issih for tsib " find " (id.), isseJczek for tsekzek 
" hiss ", izzomm for tzomm " hold " (id.), etc. In Maltese, 
North African, and occasionally also in Egyptian, this may 
be extended to palatals also, as Maltese tgih>iggih " bring ", 
tcarrat > iccarrat " send ", and Egyptian ikJcab " pour " 

(root t»J ), and iggad'an or idgad'an for ^c-Ast) " behave 
bravely " (cf. Willmore, Spoken Arabic of Egijpt, xxv, 6). 

(2) In conjugation viii preformative ta- occurs with 
metathesis in the form Alls 1 , and assimilation takes place 
with first radical dental or sibilant : (a) with ^, J^, there is 

compulsory assimilation, so that we have j^l for ^Ol and 

4.«Jl]i?l for 4_JL>1 , etc. 


(/3) With radicals S, ^, <1», J, i, ^, ^j^, the t 
may either assimilate or partly assimilate, becoming 

^ with ^, J, and i? with ^^ , ^, J, or may remain 

unassimilated with O, jj->. The complete assimilation 
with ^ is rare, but occurs m ^y y Xa (Qur'an, liv, 15). 

With ^j^, ^j^, J, we can have ^.^i or ^i2,v£>i for 

-.1,^1, and c_j j^l or t_j Ja^l for c-> ,^1, etc. With 

^> tj^> we may get ^-» i or ^1*1 , and here also we may 

"it yo 

find the (rare) contrary assimilation ^"1 , and similarly 9>XLJ\ 
or >c«..*«l (Qur'an, vi, 25). 

(7) Reciprocal assimilation also occurs with first radical 
^> ij^ > J^,as in^stl?) for ^^tT^l , that is to say, we may 

IP Ml 

have O >^, 1»^ > 1?, and lli'> U. In dialect this 

y e 

. . yy 1 yyy t 

assimilation is extended to palatals, as t^^^^ for a*^' 

" collect ", in which t is softened by the preceding soft palatal 
in contact (cf. Mufass. 693 ; Howell, ii, 1371-2 ; and Lane, 
Lexicon, 456). 

(8) Comparatively rare is the assimilation of t in con- 

jugation viii to medial :>, ^j^ , ^ , as in uiSjl for \S^y , 

involving metathesis of the vowel, and so y^^ y* in Qur'an, 

... . y *> >y 

viu, 9, and J^-^^-a-st)^ in Qur'an, xxxvi, 49. 


(e) In conjugation x ( Jsl2l*-«ii ) assimilation occurs only in 

the verb P 11?, as P \j£.^ for P Ik'^^i in Quran, xviii, 96. 

(ii) Abyssinian 

The vowel of the formative ta- is not elided in the perfect, 
but in the imperfect {yetqatel, etc.) it is brought into contact 
with the initial radical and assimilates with it when a dental 
or sibilant, thus yetsamay becomes yessamay " is named ", 
yetdagam becomes yeddagam " do again ", yetsasal becomes 
yessasal " is shaded ", yettamaq becomes yettamaq " is 
baptized ", etc. Tigriiia extends this assimilation to initial 
palatals and sonants, as yeggadaf " is forgiven ", yeqqebal 
" receive ", yemmelas " return ", etc, 

(iii) Hebrew 

In Hebrew reflexive t- assimilates to a first radical dental, 
as nJDin for nS'inn (but D'^pD'^nn without assimilation in 

T-- T-:- )-:-.. 

Judges xix, 22), Tifpl^^ for nDtDH* in Lev. xi, 43. Occasionally 
this assimilation is extended to palatals and sonants, as tk 
becomes kk in HD^n (Prov. xxvi, 26), but HD^n* in Prov. 
xxiv, 3 ; tn becomes nn in )'2^'n (Jer. xxiii, 13) ; tr becomes rr 
and then one r falls away with lengthening of the preceding 
vowel (cf. xii, 62) in Uiy\1^ for DpnHiS (Isa. xxxiii, 10). With 
a first radical sibilant metathesis takes place and then 
assimilation, so that ts >st, ts >st, as in p"^D^i (Gen. xliv, 16), 
ts> ss only in Dpi^ri (Eccles. vii, 16), and zt> zz (cf. x4.ssyrian 
below) in ^ISTH (Isa. i, 16), which happens to be the only 
instance of reflexive with initial z. 

(iv) Aramaic 

Conditions in Aramaic are very similar to those in Hebrew. 
Assimilation occurs with first radical dental, metathesis and 


assimilation with first radical sibilant, thus st >st in Bib. Aram. 
ySlOlf* (Dan. iv, 12), Syriac ^sr^H, etc. ; zt>zd as in Arabic, 
not zz as in Hebrew and Assjnrian, in j^n^^lTn (Dan. ii, 9), 

Syriac pyOn, etc. In neo-Syriac t also assimilates to the semi- 
vowels so that tw>ww and ty>yy. In T.B. and Mandsean 
reflexive t assimilates to any first radical except ', thus T.B. 
tq>qq in 7t^p5^^, etc. In Palmyrene reflexive t assimilates to 
dentals, sibilants, and laryngals. 

(v) Assyrian 

In Assjnrian metathesis always occurs with reflexive t and 
not as in Hebrew and Aramaic only when the first radical is 
a sibilant. Complete assimilation takes place with s, s, z, 
d, as in issahur for istahur " turn ", utteibhi " become good " 
(Ham. K.U., 20r., 47), but also utteibhi (id.), izzakar for 
iztakar "proclaim" (Ham. King, 110, 60, ii 6, line 12). 
With first radical s assimilation may be partial or complete, 
as istabtu or issahtu for istabtu " seize." Partial assimilation 
occurs in some other instances, t becoming emphatic with 
q, as aqterib for aqterib " advance against ", and soft with 
g, n, m, as igdamru for igtamru " complete." Reflexive t also 
assimilates when in immediate contact with a medial dental 
or sibilant, as pissas for pitsas. 

23 (b) Other assimilations of t 

In instances other than those of reflexive t we find 
that dental made emphatic or softened by the influence of 
a neighbouring consonant. In Arabic the assimilation 
td > dd is particularly associated with the ancient dialect 

of the B. Tamim, thus jiAc- for j 1 Ale. " yearhng goat", 
etc. In Qur'an reading t normally assimilates to a dental or 


sibilant following in immediate contact, and this is permitted 
also in the case of t before <j. Anomalous and rare are the 

assimilations of afformative t in such forms as iZ-^aajy- > 
Jal^a^-, <JL*..Iaji- > j ^^s-, cJa>-1 > i2._>-l (Howell, ii, 

O^ )■ O^ 0} i> 

1369), and OJd > i^Jd, OA>. > A>- (Howell, ii, 1373). 

In Abyssinian the feminine afformative -t assimilates to a 
final radical dental (not sibilant) in contact, as waliedd for 
wahedt " only ", masatt for masatt " robbers ", etc. 

In Syriac t often becomes d (cf. 17 above), and sometimes 
this seems to be assisted by a voiced consonant following as 
I^J'I for dprd/3r}, etc. 

In Assyrian feminine afformative -t becomes d after m or n 
or g, and emphatic t after q, as tamtu> tamdu " the sea ", etc. 

24 (c) The other dentals and sibilants 

(i) Soft dental d, unlike t, is not used as a noun or verb 
formative, and so we have only occasional assimilations 
occurring in radicals. Thus in Arabic, etc., radical d 

assimilates to t in O-^ for sidt " six " (Sabsean HID , Mehri 

sidet), and occasionally to formative t in verbs, as 'abattu for 
'ahadtu (cf. Wright, Arabic Grammar, i, 16). Although in 
Abyssinian dt tends to dd, we occasionally find it producing tt. 
In Hebrew dt becomes tt {t aspirated as th) in lath for ledeth 
(1 Sam. iv, 19), and 'aJiath for 'ahedefji " one " (feminine), 
etc. In Syriac also td>tt in 'atta for 'adta " land ", Mandsean 
'atata. Later dialects of Aramaic assimilate cZ to a following 
sonant or labial, as Syriac hanna for hadna, Mandaean 
qammeh for qad^neh, etc. In Assjrrian dt >tt in qardu, feminine 
qarid-tu> qarittu "strong", and ds> ss in edsu> essu 
" new ", etc. 


(ii) Occasionally emphatic t loses its emphasis by 
assimilation, contrary to the general rule that the emphatic 
gives emphasis to the non-emphatic. Thus Abyssinian 
westa for westa (Arabic wasta) ; Aramaic ts> ts> ts (cf . 
Merx, op. cit., 121, 259). So Assyrian /witew iov futiten. 

(iii) In Arabic and Aramaic it is the general rule that sibilant 
s becomes z before a voiced consonant and s before emphatic. 
In Arabic these assimilations are most marked in the ancient 
dialects of Kelb, Odrah, Kaab, and B. I'Ambar, thus razaha 
for rasaba (Kalb), hazdil for basdil (B. I'Ambar), saqy- for 
saqy- (id.), etc. So Hebrew p^ruzbul for irpoalBoXrj (Mishna 
Shabb, X, 3) ; Syriac nesdur sounded nezdur " we dispose ", 
etc., and pm'' for pHk^^ (Cooke, NSI., 64, note on line 2). 

(iv) Emphatic s shows softening to z before a voiced 
consonant or sonant. In Arabic the change of sd to zd is 
characteristic of the ancient dialect of Kelb, as mazdar for 
masdar "root ". In Aramaic a similar change is usual before b, 
d, and in East Syriac s> s before t. Assyrian generally softens 
s to 2 before a voiced consonant, as in zubatu for subatu 
" garment," zida for sed (Targ. sed, Hebrew sed) " on the right 
side ". 

Palatahzed sibilant s becomes z before a voiced consonant 
in Arabic and Syriac, but this change is not shown in the 
written script. 

25 (d) Laryngals, Velars, and Palatals 

Assimilation of laryngal Hamza occurs in Arabic conjugation 
viii in the verbs 'kl "eat", '7nr "command", and 'hdh 
" seize ", and sometimes in 'gr " pay wages ", 't^r " cover ", 
'mw "be loyal", and 'hi "take a wife", as ittakala for 
iHakala, etc. So in Aramaic, in Ethpe'el and Ethpa'al of 
'bd "destroy", 'M "seize", 'Ar "delay", and in all verbs 
in Ethtaf'al (for Eth'af'al). In older Babylonian-Assyrian 


Hamza assimilates to a preceding or following consonant in 
immediate contact in the interior of a word, as iippaal for 
ii'paal (Ham. K.U., 17, 23, 23r., 71), etc., but in later forms 
it falls away with compensatory lengthening of the preceding 
vowel (cf . section 10). 

Laryngal P (') assimilates to h following in contact in 

reading the Qur'an and in the ancient dialect of B. Tamim, 
as in mahhum for ma'hum. Laryngal h assimilates in Hebrew 
in the suffix -hu, -ha, etc., of the 3rd pers. pron. attached 
to feminine -at, as qHdlattu for q^tdlathu " she killed him ", 
and with -n of the energetic -mww for -enhu, etc. Similarly 
in the dialect of Tunis, -hu, etc., assimilates to feminine -at 

Velar q becomes g by assimilation to a following voiced 
consonant in Qur'an reading and in reading Aramaic ; in 
East Sjrriac it softens to g before s, p, t, and in West SjTiac 
before t (Merx, op. cit., 121, 260). 

The palatals show assimilation in Abyssinian in the 
pronominal suffixes -Tea, -hi, etc., attached to final radical 
-g or -q, as 'ehadegga for 'ehadegka " I will leave thee " 
(Praetorius, Aeth. Gr., 82). In Arabic also soft g assimilates 
to following s in reading (Qur'an, xlviii, 29, etc.). 

26 (e) Assimilation of labials 

Assimilation of labials does not assume great importance as 
labials do not appear as formatives. Arabic and Syriac tend 
to soften f/p to b before a voiced consonant, as Hebrew 
par'as = Arabic hurguth " flea ", Arabic qunfud = S}Tiac 
qubdd " hedgehog ". Similarly, Hebrew barzel = Assyrian 
parzilu " iron ". In Tigrina/ assimilates to a following sibilant 
in ness for nafs " self ". Assyrian shows hardening of 6 to 2? 
before t, s, k, h, as inapatu for inabatu ; Arabic hbs, Hebrew 


Jihs = Assyrian e'pesu (Ham. King, 55, 17) ; Arabic 
dibs, Hebrew d^has = AssjTian disjiu (with metathesis) 
" honey ". In Assyrian also a labial assimilates to enclitic 
-ma, as erum7na for erub7na. 

27 (!) Assimilation of sonants 

(i) Sonants m and n 

Of these sonants m is akin to the labials, n to the dentals ; 
hence by assimilation n>m before a labial, m>n before a 
dental or sibilant. 

In Arabic w > m occasionally before h, as 'ambar for 
'anhar, Arabic n, whether the ordinary consonant or 
the -n of tanwin assimilates to following r, I, m., iv, or 
y in Qur'an reading, unless in cases where an ambiguity 
might occur from the assimilation. Similarly in spoken 

Arabic, in the cases oi ^y^, jl,^c^, jl before U, z^*, V, 

as in ^'1 for ^jl, etc. In the dialect of Oman ^y» 
thus assimilates before the article, as in missems for 

^j,^^] I ^yA ; in Egyptian dialect this assimilation is generally 

found before I, as in kallak for <-A3 jlj " it is thine ", and in 
-in of the plural before li, thus ioiming -illi. In 'Iraq the 

-n of wen i^yj) assimilates to following r-, and so in Mehri 

the -n of y^ before r. 

In Arabic generally the preformative n of conjugation vii 
( jIa)* I ) assimilates to first radical m-, thus ^se^\ for /is«^J I . 


In South Arabia and sometimes in North Africa medial n 
assimilates to a consonant following in immediate contact, 


thus Minscan HID for kindat, Mehri hanta {C^ I ) becomes 

het "thou", Moroccan bitt for C^^ "girl", etc. So in 

Abyssinian (Tigre) 'anta becomes 'atta " thou ". 

In Hebrew, Aramaic, and Assyrian n normally assimilates 
to a consonant following in immediate contact, thus Hebrew 
hW for htlp^', 2^T for 5i^5>\ HnX* (Phoen. n«) for nni^, etc. 

•It "It: - - : T - : t : -: 

As laryngals and r cannot be doubled (cf . sections 12, 62), the 
n after assimilating to one of these letters has to fall away and 
compensatory lengthening takes i3lace in the preceding vowel, 
thus yin'amed becomes *yi"amed and thence l^l^V Thus 
even in med. gemin. verbs as "lli " move ", imperfect 11* 
(Nahum iii, 7), but not in verbs with medial semi- vowel as 
n)^ "rest", imperfect Tli^y, as the vowel contraction, it 
would appear, takes place first and thereby n ceases to be the 
closure of a syllable. However, there are occasional exceptions 
to the assimilation of n, as "ibi* " he keejjs " (Jer. iii, 5), 
pausal forms of *l^i (except ll^i*, Prov. xx, 28) " watches ", 
and cases where the medial is a laryngal as 7)13* " he 
possesses ". In the verb jHi the final -n assimilates to 
consonantal terminations, as nHJ for n^n^. In Syriac n as the 

T - T T : - T *^ 

closure of a syllable not final assimilates to a following 
consonant in immediate contact, even though it be a laryngal, 
as -orx^i/ becomes ^ociaZ (tefiuq) "thou goest out", causative 
Jliai] > *Qa] (affeq); *Gai*j7> .^qI.7 (tehhub) "thou dost 
languish ", and so other verbs with initial n unless they are also 
med. gemin. or med. w/y, in which case the vowel metathesis of 
the med. gemin. (cf. section 157) or the contraction of the 
med. w/y ta]<es place first and prevents the assimilation of 
n because n is now no longer the closure of a syllable. So 


pron. alt {Li\) for anta, ]i« " year " (final -d for -at), 
emphatic ]Al» {sattd for santd), etc. The final -n of |D, 
^io, regularly assimilates in Onq. as J^ni"'J''/!D " from the 
garden " (Gen. iii, 23) for ^7)^'^ ^p, but this is very rare 
in Tg. Jerus., T.J., Mand., and S\Tiac, where, however, we 

find both "J^m _k) and ]r'^ " at once ". Earhehrteus gives 
the traditional rule that final -n assimilates to a following 
initial sonant, sibilant, palatal, or »£), ^, »a (Merx, Artis Gram, 
apud Syr., p. 258). 

Assyrian changes m to n before a dental or sibilant, as 
nakantu for nakamtu " treasure ", unsu for U7nsu " hunger ", 
etc., also mq>nq or qq, as in emqu (Hebrew pf2V)> enqu 
"wise", imqut> iqqut, etc. Assimilation of n follows the 
same course as in Hebrew and Aramaic. In the case of n' 
we find either nn (the earlier form), or n falls away (the later 
treatment), thus in'ud> iud, with compensatory lengthening 
of the following vowel : ns > ss when the s is part of the 
pronominal suffix -sii, -si, etc. In Assyrian no assiinilation 
takes place in bintu " daughter ", but otherwise nt > tt as 
in limuttu for limimtu. 

28 (ii) Sonant I 

In Arabic the article -I- assimilates to a following dental, 
sibilant, or sonant n-, thus al-sams becomes as-sams, etc. 

So I of J^ in Egyptian dialect, and in Omani the I of As- 

and JA (Reinhardt, 8, h-c). In some dialects -I of the article 

assimilates also to palatals and semi- vowels, thus Aa 11 > 

akkull " all " (Egypt), jl ji-l > aggazzdr " the butcher " 
(id.), cJ50l > accelib " the dog " (Nejd), etc. 


Generally lr> rr as in minnirruJii for ^ J^l/c--* " I by 

myself " ; so JA, J^) before r (of, Qiir'an, xxiii, 14, etc.) ; 

in the Qur'an this assimilation of I takes place before O, <.!>, 

and ^J^. 

When I is the second radical and n is the third, In > nn 

and so when I is final before suffixed ^J , U in the dialects 

of Egypt and 'Iraq, thus hanna for u)L>- " our uncle " 

(Egypt), kitanni for /o^-llJ ('Iraq). In Mehri I as second or 

third radical assimilates to a following dental, sibilant, or 
sonant, thus kull say becomes kossi. 

In Abyssinian assimilation of I is rare, but appears in 
^akko for 'alko " is not ? " Tigrina shows Id > dd in waddl for 
waldl " my son" ; Amharic lr> rr in the case of negative 
^al before r-. 

In Hebrew the verb Idqah " take " shows assimilation like 
that of verbs with first radical w-, thus W'C have yiqqah for 
yilqah, etc. 

In Aramaic the verb l^qah " take " shows assimilation as in 
Hebrew. In T.B. and Manda?an there is frequent assimilation 
of the I in 'al " upon " ; and there is assimilation of ? to a 
following dental or sibilant in some of the later dialects, as in 
hassd for halsd (Mandsean), or to preceding s in the verb 
sHaq (Bib. Ai'am. and Sviiac), as hassiqu for hasliqu, and so 
to z in the Sjrciac verb '^zal, and to r in the Ma'lula dialect 
regra for regla " foot " (but cf. 19a). 

29 (iii) Sonant r 

Assimilations of r are much less common than those of the 
other sonants. In the Arabic of Tunis \ve find qadd for qadr ; 


in Abyssinian 'ersu (xAjabic ra's) becomes 'essu (Amharic) ; 
Hebrew shows assimilation of the r in .ver from ''^ser in such 
forms as mlldmd for ser-ldmd (Cant, i, 7). In Aramaic we 
find the dialect of Ma'lula assimilating r to a following sonant 
in amellobu for amer luhu " he said to his father ", namelli 
for namer li, qanna for qarna " horn ", etc. 

30 (g) Assimilations of the semi-vowels 

(i) Semi-vowel w 

In Arabic conj. viii we find assimilation of w to t, thus 
ittasala for iwtasala. In North Africa w often assimilates to 
a preceding labial in contact, as mmagen for mwogen " hours ". 

In Abyssinian (Tigre) after a labial iv> y> g, thus 'abaiv> 
'ahay> 'ahag " father " (cf. section 20). 

In Assyrian in the reflexive wt > tt, as ittalad for itwalad. 

(ii) Semi-vowel y 

In Arabic in. conj. viii yt> tt, as ittasara for iytasara. Also 
y assimilates (sometimes) to following Hamza in contact, as 
in yd'asu for yay'asu wheie -yay' a > -y a" a > -ijd' a (Mufass. 

Aramaic shows yJc > kk in yikkul (Dan. iii, 29) for yiykul ; 
and yt > tt in yittab for yiytab in Dan. ^'ii, 26. 

Assyrian ny > nn in innasir for inyasir. 

31 (ii) Dissimilation of Consonants 

(a) Dissimilation of doubles in contact 

The first kind of dissimilation we have to consider appears 
in the treatment of double consonants without an intervening 
vowel, in which case the tendency is for one or other of the 
two to become a sonant or semi-vowel, the change most 
frequently taking place in the first. The commonest instance 
appears in verb forms such as qattal- reduced to qantal-, 
qartal-, etc. Hurwitz (Root Determinatives in Semitic Speech, 
New York, 1913) argues that these are not instances of 



dissimilation but of informative -n-, -r-, etc. Sometimes this 
is perhaps true, but (i) in Abyssinian and Assyrian it occurs 
as a regular treatment of med. gemin. verbs, e.g. Abyssinian 
habbaha becomes hanhaba, Assyrian imaddad becomes imandad, 
etc. (ii) In Aramaic, especially in later forms, such dis- 
similations are very common with a medial double, e.g. 

M^J — v>s»P " roll ", etc. In certain Semitic languages, 

therefore, we find a marked tendency to dissimilation, which 
at least suggests that parallel forms in other Semitic languages 
may be due to the same cause. Moreover, there seems no 
means of classifying such inserted sonants and semi-vowels as 
to assign to them any clear semantic value, nor can r, w, //, 
be identified with any other forniatives, although a verb 
preformative and informative n exists which may become m 
before a labial (cf. 27), as well as noun afiormatives -n and -I. 
It must be admitted that the verb forms showing inserted 
n, r, etc., often differ in meaning from the qattal-, although 
this might be explained by supposing that some of the dis- 
similations were of early date, and so in course of time 
differentiated by obtaining a specialized meaning. Still, as 
some of the forms undoubtedly show dissimilation, it seems 
convenient to enumerate all the types thus unless the infixed 
sonant or semi- vowel can be identified as a formative. 

32 (1) Dissimilation of the first 

(i) The first becomes sonant n 

In Arabic this is reported as characteristic of the ancient 
dialect of Hims, thus 'utrmog for 'utrugg " lemon ", etc. There 
are a number of words in Hebrew of the med. gemin. type, as 
Mkh (liikk-) " palate ", which show this dissimilation in 
Arabic as hanak, etc., but, of course, it is often possible that 
these are instances of medial >?. roots with assimilation in 
Hebrew, Thus 'es, Arabic 'anis-at " fire ", Hebrew 'af 


i'app-), Arabic 'anf " nose " ; Hebrew hitta, Arabic hinta{t) 
" wheat " ; Hebrew gaf (gapp-) " wing ", Ai-abic ganqf 
" reclining on one side ", etc., but undoubted dissimilation 
occurs in Arabic s'unbula{t), which is a loan word from Hebrew 
sibboleth " ear of wheat ". 

So in verbs we frequently find a form qantal for qattal, in 
which it seems that the n is not informative, but simply due 
to dissimilation ; thus from Jiaggama we get hangam " burst 
in violently '"' (Egyptian d), and similarly with the forms 
hantal ''' work with energy " (Algerian dialect), hangar " kill 
by cutting the throat ", etc. 

Abyssinian shows a similar change in verbs with doubled 
medial, generally in cases where the root is med. gemin., 
thus Arabic sagga'a, Abyssinian zange'a " speak or write in 
rhymed prose ", Arabic hahhaba, Abyssinian hanbaba " run 
to seed ". Ge'ez hattasa becomes hantasa in Tigre. 

In Hebrew and Aramaic the tendency to assimilate n to 
a following consonant is so strong that the contrar}? 
dissimilation is very rare. We find it, however, in Bib. Aram. 
tirida'' (Dan. ii, 30), etc., with nd for dd. Syriac nt for tt in 
'antHhd, Aramaic 'iftd, Hebrew 'issd, Arabic 'anthd, but this 
is more probably an assimilation in Hebrew and Bib. Aram. ; 
in nh for hb in Sj^riac ganburd — Arabic gabbdr, Hebrew 
gibbor ; T.B. nd for dd in nigandar (Aboda Zara, 28a). 

In Assyrian this type of dissimilation occurs frequently in 
verbs med. gemin. Thus, from mdd we have imandad, and 
from 'mr we get (Nt.) ittananmar instead of ittanammar. 

(ii) The first becomes sonant r 

Instances occur in Arabic conj. ii, as gartluimaiov gafMhama 
" take the chie[ part ", qarmata " contract ", qammata 
" tie up " ; so in Egyptian dialect hargam " burst in 
Algerian darbal '' be in tatters ", etc. 

Hebrew kissem " chip off ", and Mrs em " devastate 


(Erub. 1026) 


(Ps. Ixxs, 17), m^kubhal bocomes m^kurbdl " girded " 
(1 Chron. xv, 27), sahhel> sarhel "cover". So Arabic 
qaddum = Hebrew qard&m " axe ", Hebrew 'ahkad appears 
as ^pxa^ in the LXX (Gen. x, 10), etc. 

Aramaic ND"13 " throne " (Cooke, NSI. Ixiii, 7) =-- Hebrew 
kisse, and so Bib. Aram. XD^3 (Dan. v, 20), Syriac ^IcDJQO, 

whence Arabic loan word (_^J^ • Syriac Jiass > haras 

(Dan. V, 20 ; Onq. Dt. xxxiii, 1) ; pK^^"! (proper name) 
appears as p^pyi in 2 Chron. xviii, 5, 6, Syriac JiDJ>, Syriac 
"jj " roU ", T.B. nSntJ^ for IStJ' " cause to drag " 
and D'^'^2^5 for DlilJ^ " was lopped ofE " 
(Men. 386). In neo-Syriac -rnr^-^'^Kn - stir ", ^jioiiD " throw 
down ", .nn;»»V> " clasp ". 
(iii) The first becomes sonant m 

Arabic ll> mi in ^JaJL*^ for ^^^ " adorn ", <— AJL«^ for 
"^^ K " make smooth ". Amharic damhara for dahhara 

" speak ". Syriac ,^rnV>». for ^aiL " fortify ". This kind 
of dissimilation is very common in neo-Syriac, as mq for qq 

in \J»«jQiQCD (Arabic ^^\). Assyrian hb > mb in sumbu 

for subhi " wagon ", unambi for unabbi, etc. The preference 
for m rather than n before a labial is of the nature of an 

(iv) The first becomes sonant I 

yy cy c 

Arabic '^t becomes *A^. i^ f^} (passive 
with informative -an-, cf. section 139) " be removed ", 

"Tii^ becomes hahvas " babble " in Egyptian dialect, and 



f^lsJA " expand " from *kd " make broad ". Hebrew 

P]yT " be angry " and t0\ " glow ", and p^^'^D for |DXp 

(cf. laryngal incapacity of doubling) " tranquil " (Job xxi, 23). 
Assyrian hakhatu becomes balJcatu " tear down ". On the 
change by which a dental or sibilant becomes I before another 
dental or sibilant cf. § 36. 

(v) The first becomes semi-vowel w 

J^ 1 ^ 

3t>- > (\>:j>- " be shackled ". Other instances 

of type qattala> qawtala in ^y^j:>- "fill one's stomach ", 

Q^y>- become senile ", (jyj>- " brand a camel ", c-jj^^so 

" be affected with mange ", and in Egyptian dialect (33^ -^ 
L^jj.>- > liozaq " impale " ; in Syrian dialect hatorab 

" be angry ". In Hebrew and Aramaic these produce qawtal > 
qolal, and so confuse with type qotala. 

(vi) The first becomes semi-vowel y 

Arabic /o^l > (^ I conditional particle, jA) " be full 

(moon) ", jA> > jUl> " pile up ", so JjlJ* " feign to 
be stupid ", ^;JU) " be amazed ", j^^ " turn aside ", 

etc. In type qiVM as in ^_^U) >, J'^^'-"' ^^*^-' where 1 = «?/. 
Noun type ^i^^aZ does not occur without this dissimilation 

unless with suffixed -at, as 6jljL*« " head of a spindle ". 
Abyssinian, resultant aij >e (cf . 49), thus daggana >degana, etc. 


Some Aramaic examples occur in such forms as sabbera > 

83 (2) Dissimilation of the second 

This is not at all so common or regular as dissimilation of 
the first. 
(i) The second becomes sonant n 

Arabic 4_->J^>- > t— >j.» ^^ (name of a tree growmg m Syria, 

the Cemtonia siliqua), Ja.^>- " have pain in the belly ", 

JLzl>- > ial*>- > i2.lA>-l " become inflated in the belly ". 
(ii) Second becomes sonant r 

  * y * y^ 

Arabic k^X>- > c-?j-\.>. " hasten ", ^^i^a^^- " cut off " > 

aJiaj>- '' cut off end of a camel's ear ". 
(iii) Second becomes sonant m 

y y y y 

Arabic i? r*** > ^^^ " ^'^^^ ^ pieces ", root Ja.l>- > 
galmat " shave the head " (Egyptian dialect), 
(iv) Second hecom,es semi-vowel w 

y oy ^ ''^ 

Arabic in j5-^>- " rule a book with lines " from Ja>. 

" twist (a rope) " ; Jy>J " cause to withdraw " from 

JaJ " avoid ". 

(v) Second becomes semi-vowel >j 

yyoy "^ ^. 

Arabic in ..^^ " become evil disposed " from ^o- 

y yy 

" be very angry " : itq^anf (Egyptian dialect) for t^^ 
where i — yi. 


34 (3) Dissimilation of repeated initials or finals 

Another important dissimilation occurs in the change 
which takes place in the corresponding opening or closing 
consonants of doubled syllables, as when siLnl becomes 
sinsil or kahkab becomes Jcnwl-ab. This dissimilation appears 
most commonly in sonants, and more often it is the first 
closure which changes. At the bottom of these doubled 
syllables there is generally a med. gemin. root. 

-v/SLL " to extract gently ", Arabic salsala " to connect ", 
silsilat " chain ", > "^sinsil-, Omani sindd, Ethiopic sensel, 
Spanish Arabic cercele : Syriac slslu-thd " diacritical point ", 
where s-ls-l becomes s-ys-l and iy -- I. 

VGLL " to roll ", Hebrew gulgol-eth " slruU ", gilgal 
" wheel ", cf. N.T. FoXyoda = *gulguiv-td : Syriac gdgul-ta -— 
*gawgul- " skull ", gigla " wheel " (= ^giygal). 

V'RBB "to scatter", Arabic *J:abJcab> kawJcah "star", 
Hebrew koJcdh for kawJcah, Syriac kukhd. 

v/RBB "be great", Aramaic rurhd {=^ *rmvrab-) 
" dignitary ", ]^12^ (Cooke, NSI. Ixiii, 10). 

35 (4) Dissimilation of other repeated consonants 

This dissimilation is also extended to consonants which 
are alike but are not members of corresponding syllables. 

Margar- in ^ap'yapirrj becomes S}rriac margdn-ithd, T.B. 
margdl-iihd, Arabic margdn, loan word from "Syriac. 

Mercur- in Latin pr.n. Mercurius, T.B. Merqulis (Abod. 
Zara. 41). 

Nahna, Arabic dialect for nahna " we ", laljna or ivahna 
(Datina dialect), Hebrew sams- (absolute semes), Assyrian 
samm, samas, Arabic sams-> sams " sun ". 

36 (5) Special case of sonant 1, r 

Change of sibilant to sonant I, r, when preceding and in 
contact with a following sibilant or dental, occurs in AssjTian. 


Babylonian hustdru > Assyrian kultdru. 

Babylonian kasdu> Assyrian Jcaldu " Chaldseau ", 

Assyrian selasti > selalti " three ", w,anzaztu (fern, of 
manzazu) > manzaltu, mazzaltu " resting place " (Rawl. iii, 
59a, 35), Hebrew mazznloth " signs of the Zodiac " (2 Kings 
xxiii, 5), from which is derived a false sing. mazzdl{-at) and so 
Arabic manzil " inn ". 

In most of these cases we can have r for h as manzartu 
{*inanzaltu), etc. Cf . Amharic sessa > selsa. 

With this change of sibilant to sonant cf. Sumerian mis = 
mil " dust ", dis = dil " one ", gas ^ gal " house ". 

Consonants affected by Vowels 

87 (a) Aspiration 

The aspiration of consonants is often produced by the 
influence of vowels. In Hebrew and Aramaic, where the 
original aspirate dentals th and dh are lost, new aspirates 
are created by the action of a preceding vowel on the dentals 
t and d, and corresponding aspirates are formed from b, g, 
k, p, as bh, gh, kh, ph. Sometimes the new aspirate is retained 
by analogy when in process of inflection the vowel no longer 
precedes it, thus from stem mdlk-, absolute melek>melehh 
with the k aspirated by preceding inserted e, construct plural 
malkhi. In neo-Svriac the prepositions 6*, l^, d^, and the 
conjunction w®, as well as the formative prefix m*, do not 
suffice to produce the aspiration of a following consonant. 
Again, verbs with final d ov t do not aspirate this dental 
after a vowel in neo-Syriac save in the dialect of Alqosh. 

These new aspirates are quite distinct from the th and dh 
of Arabic which correspond with sibilants (s, z, cf. 17 d) in 
Hebrew and with non-aspirates {t, d) in Aramaic ; they are 
entirely new formations solely due to the action of a preceding 
vowel. In an unpointed text nothing distinguishes these 


aspirates from the corresponding non-aspirates, but in a 
pointed text the aspirate in Hebrew should be marked with 
a horizontal stroke above the letter (Rate), a mark which 
rarely appears in printed texts, and the non-aspirate marked 
by inserted Dagesh. In Syriac the non-aspirate consonants 
should have a supra-linear point (Qusoi) and the aspirates an 
infra-linear point (Rukok). 

In Tigrina and Amharic k becomes kh or h under the 
influence of a preceding vowel, thus Amharic pronominal 
suffix of the 2nd masc. sing, -kh or -h but -ka after a consonant. 

In Assyrian very often a dental becomes a sibilant, which 
is a kind of increased aspiration, before the vowel sounds i 
or u, thus md'assu for ma'attu. Possibly this may be a trace 
of Sumerian influence, for in Sumerian a similar change often 
occurs before the sharp vowel ije, as dim = zem, dug = zib, etc. 

38 (b) Palatalization 

(i) In Arabic dialect palatal k is thus affected by a 
neighbouring i, e (a) : in 'Iraq and Nejd, as well as amongst 
the Bedwin of the Syrian desert and the f ellahin of Palestine, 
k often becomes c after e, i (a), thus in 'Iraq 2nd fern, suffix 
-ic for -ik {— -ki), plur. -cen. Less commonly q becomes c, 
as iu ric for riq " saliva ". With a there are usually alternative 
forms as kathir or catMr, etc. In Hacjramaut, Oman, and 
Mehri, as well as in the ancient dialects of B. Amr. and 
B. Tamim, k before i becomes s, as in the Omani proverb 
kiliid minnasi yd sVinnat wa-'aqqai/nasi wara'-al-'annd " we 
have eaten out of thee, basket, and have cast thee behind 
the house " (Jayaker, Omanee Proverbs, No. 10), where -§i 
twice occurs for ki. So the variant tahtasi for tahtaki in 
Qur'an, xix, 24. Sometimes a similar s for k (instead of c) 
occurs before or after i in 'Iraq, the littoral of S3'ria, amongst 
the fellahin of Palestine and the Christians of Jerusalem 
(cf. Palest. Explor. Fund, January, 1890, p. 98). So in Mehri 


we find sebedat for kihd " liver " and Soqctra bosi for balci 
" weeping ". 

(ii) In Abyssinian we find that Amharic regularly makes 
the changes Jc > c and q' > c.\ thus Ge'ez kehela ^-^ Amharic 
H-e'ela> cola "he drank". Before i we also get z> z, 
s> s, s> c as 'azzazi> 'azzazi, etc., but Amharic is much 
influenced by non-Semitic neighbours and palatalization is 
one of the results (cf. 17). 

39 (c) Change o£ labial to semi-vowel 

In later dialects of Abyssinian and Syriac labials tend to 
become semi-vowel rv after the vowel a, thus Tigre oids {nnws 
for nafs) " self ", Amharic saw {sahe') " man ", neo-Syriac 
gord {gaurd for gabrd) " husband ". 

40 (d) Semi-vowels assimilated to vowels 

Very generally the semi-vowels are assimilated to the 
neighbouring vowels so that w becomes y near i, and y becomes 
w near u ; thus iw >iy>% in Arabic, Hebrew, etc. wi >yi >t 
in Arabic and ije in Assyrian, as ukvmi > ukln ; also uy > 
uw> u and yu> wu> u, but in these latter instances it is 
more often the vowel that is assimilated to the consonant. 




The original Semitic vowels show three timbres, a, i, and u. 

Other timbres exist, a, e, o, il, but all these are derived from 
the three original sounds bv dialectal variation, or else by 
the influence of neighbouring consonants. Thus a must 
remain a in Arabic when in contact with a laryngal ; it may, as 
a matter of dialect, become a, e, i, when not in contact with 
a laryngal or emphatic ; it must become a or o, or in North 
Africa u, in contact with an emphatic. Our course will be 
first to treat of the way in which the vowel sounds have been 
transmitted through the Semitic languages, then to consider 
how they may be influenced by neighbouring consonants, 
and finally to examine the way in which they are affected by 
other vowels resulting either in assimilation or dissimilation. 

In each vowel we have to consider three quantities, the 
long a, I, u, the short a, i, u, and the half-vowel or murmured 
vowel such as follows the p in the French pronunciation of 
'pneu. We need not consider the two other asserted quantities, 
the protracted long and the Ismam or timeless half-vowel 
which involves the compression of the lips as though to form 
the vowel u but without the utterance of any sound, for these 
are the artificial creations of Qur'an readers and concern 
liturgy rather than philology. Nor can we formulate any 
actual time measure for the three quantities, but only a relative 
time as between long, short, and half -vowel. But the quantity 
affects the timbre. Almost always the timbre becomes duller 
and more obscure as the quantity is decreased. In modern 


Arabic the short vowels tend to become vague in timbre and 
to difEer, not only between districts, but between the different 
quarters of a town, and even between individual speakers, 
whilst the half-vowel has an indeterminate timbre partaking 
of i-e and u ; in all cases the consonants are more distinct, 
and the whole speech thrown back deeper into the throat 
than is the case with the Indo-European languages. So in 
Hebrew short a imaccented tends to become e or I, whilst u is 
confused with o, and the half- vowel, if not accompanying a 
laryngal, becomes the vague " of Sh'wa mobile. 

Apart from vowel quantity, there is another relation of 
time. In the main we may contrast the tempo lente of 
deliberate narration with the more hurried speech, the tempo 
allegro of command or exclamation. Thus, generally, the 

imperative 'qtul (Heb. 7bp, Arabic Jjli l) and the Jussive 

show more rapid utterance in verbs, and so the vocative in 
nouns. Thus, in Hebrew we find indicative yaqtil, Jussive 
yaqtel, imperative haqtel, where e is shorter than i, both 
increased from original z. Similarly, in Arabic the final -* 
of the fem. imperative becomes short as 'uqtuli, and suffixed 
-t becomes short -i in the exclamation yd qawmi " my 
people ! " in Qur, v, 21. On the other hand, there is a distinct 
tendency to a more deliberate enunciation in liturgical reading 
than in colloquial speech, and this has undoubtedly afiected 
the Masoretic pointing of the O.T. and the traditional method 
of reading the Qur'an so that neither truly represent the 
spoken language. This produces a more serious effect in the 
O.T. than in the Qur'an, because in the time of the Masoretic 
punctators the Hebrew language was no longer spoken, 
whilst the Qur'an pointing as we know was made with 
reference to the comparatively pure speech then actually 
current in the dialects of the tribes of Nejd. 


42 Accent 

Accent is of two kinds, (i) accent of stress and (ii) accent of 
pitch. Generally speaking, the former influences the quantity 
of the vowels, whilst the latter is itself influenced by vowel 

Unfortunately accent is not represented in Arabic script, 
and we have to rely upon the actual usage of living speech. 
According to this we find an accent of pitch, which is concerned 
with musical rhythm and not with emphasis, as the accent 
generally employed in Syria, and Egypt and recognized as the 
accent to be used in Qur'an reading. But the dialects of North 
Africa show an entirely different accent expressive of stress, 
and Arabia shows the accent of pitch encroaching on and 
gradually displacing the stress accent. 

In the dialects of North Africa the tendency is to accent the 
penultimate syllable, and this, it will be noted, in the perfect 
of verbs means that the emphasis is laid on the characteristic 
vowel, as in qatdla, labisa, hasuna. So we get in Tripoli 
hteb (for kataha), impf . ^Jektib ; in Tunis htib, yiktih ; Tlemsen 
ktseb, yektseb ; South Algeria ktih, yikiih ; Morocco kteh, 
yekteb or yekteh. In all modern dialect there is a strong 
tendency to drop an unaccented vowel or, as in the above 
examples of the perfect tense, to replace it by a half -vowel. 

The same accent of stress appears in Ethiopic, Hebrew, and 
Aramaic, The accentuation of Assyrian is so far obscure that 
we are not justified in formulating any conclusions. 

The accent of pitch seems to be a modern or at least a 
mediaeval introduction. It appears in the Arabic of Sjrria 
and Egypt, and is now regarded as the classical accent ; in 
Arabia it is at the present moment gaining ground to the 
excluding of the older accent of pitch. The modern Ashkenaz 
use this accent, contrary to the written tone, in reading 
Hebrew. This accent of pitch falls on the last long or closed 


syllable, or, failing this, on the syllable most remote from the 
final, thus yuqdttilu, kdtih, qdtala, magid, etc. 

43 (i) Long vowel a 

From the general scheme of the vowel sounds it will appear 
that in pronouncing the vowel a the tongue is further removed 
from the palate than in producing any other of the vowels. 
In sounding u or i the lower jaw is raised and the outlet is 
brought, in the case of u, nearer to the soft palate, and in the 
case of i nearer to the hard palate. The sound of o is inter- 
mediate between a and u, that of e is similarly intermediate 
between a and i. In considering modifications of timbre 
in the vowel a, therefore, we have two directions of change, 
(i) through o towards u, and (ii) through e towards i. 

{a) Arabic 

In classical Arabic, that is to say, the form of the language 
traditionally employed in reading the Qur'an, the change 
of a towards o-u is obligatory before or after one of the 

emphatic letters fj^, ^, I?, Ji>, O, and this modification is 

compulsory and unavoidable. This modification results in the 
soimd a, but in dialect is carried on to full o or u. So far as it 
results from the influence of neighbouring consonants, this 
modification will be considered at a later point (cf . 56 below) ; 
we have now to consider a similar modification which is a 
mark of variation of dialect and not caused by a neighbouring 
emphatic consonant. 

(a) The broad a is mentioned by Sibawaih, az-Zamakhshari, 
and ar-Eadi al-Astarabadi, and is defined by the last of 
these as the Alif which is directed towards Waw. On the 
same authority we are told that it was a common mistake in 
writing as in speech for the people of the Hijaz to substitute 

Waw for Alif, as in d^l^ for 6^)^?. 


In all South Arabia east of Datina accented d becomes 6 
(Landberg, Etudes, ii, 295) ; thus bandt becomes handt, 
sdhir>sdher "wizard" (plur. shark for sahamt), hild>hel6 
"without", kamdh>kamdJi "like this", 'arbadt>rbdt 
" four ", sdb'>hdba' " seven " (Mehri), etc. 

In Syria isolated cases occur of a modified to d or o. Change 
of a to a, 5, u, appears also in the dialect of the rural peasantry 
of Malta. Although the Maltese dialect belongs to the North 
African group, a Syrian element seems to occur in it, and this 
is more apparent in the dialect of the rural districts. 

(/3) The hnale (A)uI) or " deflection " is the name given to 

the inclination of a towards a, e, i. The native grammarians 
attribute this to assimilation to a preceding or following i, 
or to a preceding y, or to a following ya, or secondarily to a 
derived by contraction from a syllable containing y (cf. 51, 
52 below), or to the extension of its influence by analogy to 
cases where, although a is derived by contraction from a 
syllable containing w, this iv sometimes becomes y in inflection, 
or else in Qur'an reading to cases where a verse with a vowel 
ending -a is made to correspond with another verse where 
-a suffers Imale for one of the reasons already described. 
Thus all instances of Imale are ultimately traced to 
assimilation of a to * or ^ ; but this theory is disputed by 
European philologists (e.g. Brockehnann, Grundr., 51, b, jd). 
The matter seems to be that as in any case the Imale is 
regarded as optional, that is to say, it is always allowable to 
read pure a even where the sound is modified by some, the 
Qur an readers have established an artificial rule restricting 
the use of Imale to certain particular instances, a restriction 
unknown in the spoken language. How far this rule may 
be based on the ancient dialect of the B. Tamim, who are 
described as those most given to the use of the Imale, and how 


far it has been developed on purely speculative lines, it is 
impossible to say. We can only state that the traditional 
method of reading confines it to certain cases which ultimately 
depend on assimilation, and that the living speech applies 
it to cases which show no influence of assimilation or any 
kind of analogy by which they can be referred to such 
assimilation. It may be, indeed, that the traditional rules 
observed in Qur'an reading are purely artificial creations of 
grammarians and Qur'an readers ; but, whatever they are 
worth, they represent a definite tradition which cannot be 
ignored, even if we ultimately decide to reject its substance. 
As we have already noted, it is always allowable to sound the 
a true. The Imale is said to have been commonest amongst 
the B. Tamim, less common amongst the people of Qays, 
'Asad, and rare in the Hijaz. 

The traditional Imale as admitted in Qur'an reading is of 
four kinds : — 

(i) Due to assimilation 

Change of d to ale after i or y, or before I or ya. 

(a) After % or y Imale may occur immediately as in bayen 
for haydn, " argument," hayye' for bayyd\ " merchant," 
'imed for 'imdd, " tent-pole," etc., or separated by two 
consonants, i.e. in a syllable following a closed syllable con- 
taining i, as simlel for simldl, " brisk (of a she-camel)," and 
'inne for 'innd, " verily we . . ." (Qur. ii, 151), or even after 
I, y, two syllables before, provided that the syllable containing 
d, or the intervening syllable, begin with h, as in haytehe 
for haytahd, which is really assimilation of a to preceding 
e which is the result of Imale of ta-. 

(b) Before i or ya, not before yu, immediately, i.e. in the 
next syllable, as 'elim for 'dlim, " learned," ketib for Jcdtib, 
" clerk," seyartuhu for sdyartuhu, " I kept pace with him." 
In all these instances it is noted that the Hijazi pronunciation 


admitted Imale caused by % only, and not that caused by y 
either before or after. 

(ii) Due to derivation 

(a) In final a derived by contraction from a syllable con- 
taining y, as in the verb hade for hadd, " direct," from root 
lidy ; and so in nouns like sakrd, fem. of saJcrdn, " drunk," 
because this fem. -d is derived from -ay (cf. sect. 118), as -e 
(for -ay) in Hebrew 'esre, " ten," in fem. numerals 11-19, and 
Samaritan fem. numerals in *• ; so T.B. and Targ. ay>e in 
zotarte, " little (finger)," If date, " new (year) " ; and Ethiopia 
'aJiatti (for 'ahadti), fem. of 'ahadu, " one." And similarly in 
noun forms such asfatet, fem. oifata-, as this a has its being in 
a syllable which originally contained radical y. 

(/S) Final -a from w, which is taken for y because in the 
course of inflection it appears as y, an extension of the 
preceding case by (false) analogy which is tolerated but not 
approved. Thus gazd> gaze though from ^/gzw because the 
last radical appears as y in passive gaziya ; and mallid >7nalhe, 
" place of entertainment," though from ^/lhw because it 
forms dual malhaydni. Such exceptions seem to show the 
Qur'an readers straining their self-imposed rules in the 
endeavour to meet the usage of colloquial speech. 

(7) Medial a in verbs with medial ya, yi, or vd, all making 
* in double closure, as kdla, kilta, bdga, haba, etc., but not 
medial d derived from wa, wu, as qdla, qulta, tola, etc. 

(h) Medial d in nouns derived from y, but not if derived from 
w ; thus ndh- {\/nyb), 'dh- {\/'yb), not bob- {\/bwb), ddr- 

(e) Imale of fem. -at, -d in pause is admitted and is the only 
case of Imale with this termination. Thus, raJimat in pause 
becomes rahmd or rahne. But Imale is not permitted with 
-dt, (-a) as in saldt, but only when the long -d is produced by 
the fall of the fem. -t, unless -at is itself due to final -y, in which 



case it comes under the heading of (ii, a) above : nor is such 
Imale permitted when the -a is the added " ^ of silence ", as 
it is called. 

(iii) Restriction of the Imale 

The rules given above for regulating the permissive use of 
the Imale represent a tradition whose authority has been 
questioned, but the recorded " restrictions " in the main 
agree with what is actually observed in existing dialect. These 
restrictions can be reduced ultimately to three headings : 
(i) Imale is prevented by the influence of neighbouring 
consonants, that is to say, a laryngal in contact compels the 
pure a to be retained ; an emphatic or r in contact causes the 
a to tend towards d>o>u, and consequently prevents its 
deflection towards d>e>i. These instances of consonant 
influence will be treated more fully later (cf. 53 seq.). (ii) 
Imale is not admitted in the uninflected, prepositions, etc. — 
this is a Qur'an readers' rule which is not entirely endorsed 
by the usage of Uving speech, (iii) The cause of the Imale, 
i.e. neighbouring i or y, must be in the same word as the vowel 
affected, and it cannot operate if the cause {i or y) is elided. 
These two latter rules are merely based on the theory that 
Imale is due to assimilation, but rule (i) is important and 
must be carefully observed. 

(iv) Arabic Dialect 

In Arabic dialect Imale is much commoner than in the 
traditional system of Qur'an reading, but it is worth noting the 
statement made in Willmore's Spoken Arabic of Egypt (p. 3, 
n. 2) that " most of the numerous examples given by Spitta 
of imdla or thinning of the a- vowels are illustrations of foreign 
(fellah, bedawi, or berberi) pronunciation. Such forms as 
keldm, lamde, do not occur in the dialect of Cairo as spoken 
by natives ". 


(1) d>d in Syria, North Africa, Egypt: thus tdmdn, 
" eight " (Egypt), nds, " men " (Syria), suffixed 1st plur. 
pron. -nd (Egypt). In the Libyan desert commonly d>dlid. 

(2) d>e in Syria, 'Iraq, Oman, North Africa, Egypt, etc. : 
as Jciteb, " book " (Morocco), suffix -ne (Oman), -ne, -m (Dofar), 
tkellemt for taJcallamtu, " I have spoken " (Morocco). 

(3) d>i chiefly in South Arabia and North Africa, as 
nizel, "descending" (Maltese). Fern, -at {-d)>-U in Mehri, 
as in faidit iov fayadat. 

(b) Abyssinian 

In Tigriiia we find occasionally 6 for a, a, as Ge'ez samdni > 
Tna. somonte, " eight," etc. In Amharic and other later 
dialects there are occasional instances of Imale, as in 
semmen-t, "eight," -nd>-nd>-nd/-ne (Amharic), -yd>-ye 
(Tigrina, Tigre), -ydj-ye (Amharic). 

(c) Hebrew 

In Hebrew the general rule is that a becomes o, a 
new a being produced by lengthening a ; thus kd>kdy 
" thus," but kdkd, " so and so " >*kdkd, and thence kdkd 
(Exod. xxix, 35, etc.), so 'aykd, " how ? " >'ekd (Deut. i, 
12), but 'eko, " where ? " (2 Kings vi, 13). Normally original a 
becomes 5, as in soq (Arabic sdq), " leg," sald?n (Arabic 
saldm), " peace," gibbor (Arabic gabbdr), " hero," unless (i) a 
has been shortened to a and afterwards lengthened to a, or 
(ii) in loan words where abnormal forms occur, as M^'ara 
(Arabic Magdra{t) pr.n. Judges xiii, 4, Vulg. Maarah), P^rdO 
(Assyrian purdt, Sumerian pura), " Euphrates," and in such 
words as k^Odb, " book," an Aramaic loan word in later Hebrew 
(Daniel, Esther) replacing the older sefer. 

The Hebrew 6 from original d and that from original aw 
(cf. 50) is longer than the 6 produced by lengthening original 
u (cf. 4:8c), so that by decrease of accent 6>u ; thus gibbor, 


" strong," g^hurd, " strength," mdnos, " flight," fern, m^nusd, 
although it may be that these latter are separate forms of 
type qatul, a form usually employed for the passive participle, 
as 'asur, " imprisoned," h^Ould, " virgin (secluded)," etc., 
and we also find fem. with d as b^sord, " good news " (Arabic 
bisdra{t)), etc. So in double closure d from wa (cf. 52) as in 
naqom becomes u, n^qumotd. But imperative 2nd pi. qumu 
{<u) becomes with double closure qomnd. Very often this 
longer o is marked by the presence of Waw, which is absent 
from d<u, but this should not be taken as a strict rule, 
for even in the twelfth century a.d. medial Waw as a 
" support " of the vowel point was largely at the discretion 
of the copyist, teste, Ibn Ezra. 

But older forms exist in which original d became u not 6. 
Thus in the Amarna letters we have zuru'a, " arm," where 
Hebrew has z'^rod' (Arabic dhira'-), rusu, " head," Heb. 
ros, both for original ra's (Arabic ra's). Amarna anuki = 
Heb. '^noki, aharunu = Heb. ^'^hardno. So we find Heb. 
taffudh, " apple," for Haffdh (Arabic iujfdh). 

{d) Aramaic 

In East Syriac and Bib. Aramaic a is retained, thus '*was 
(Dan. ii, 10) = Heb. '^wds (Arab, nds), 'eldhin (Dan. ii, 11) = 
Heb. 'Hohim, d^rd' (Dan. ii, 32) = Heb. z^rod', but in West 
Syriac d becomes o or a as in Hebrew. Thus saldm (Arabic 
saldm) = Bib. Ar. sHdm (Ezra iv, 17), West Syr. sHom, 
East Syr. sHdm, the written vowel zeqdfd or zeqofo being 
sounded a in East Syr. and 5 in West Syr. So ^m>dpiov = 
East Syr. zoiidrd, West Syr. zunoro ; Arabic Oaldd, " three " = 
Bib. At. tHddd, East Syr. tHd6d, West Syr. tHoOd, neo-Syr. 
t^ldt (Urmi), tlotd (Ma'lula), but in Ma'lula dialect as in South 
Arabia accented d becomes o, unaccented a is retained. In 
the districts of Salamis, Qudshanis, etc., and to a less degree 
elsewhere in East neo-Syr. there is a tendency for d to become 


e, or I, but this Imale of a is not so common as that of d. 
In compensatory lengthening of a at the fall of a following 
consonant in East ^jv. d>d, thus 'd6, 'atv, and even of or 
o. In Nabataean *wd5, " men " (Aiab. nds) = ^)^^ (Cooke, 
NSI. 79, 7 ; 89, 5). 

(e) Assyrian 

In later Assyrian a frequently becomes e or ^, thus sarmdnu 
appears as sarmenu, etc. 

44 (ii) Long vowel i 

The vowel i is thrown forward so that the outlet advances 
to the hard palate. Whilst I proper is a clear sound, its 
obscuring results in e, which is intermediate between a and i 
(on e resulting from ay cf. 49 below). Aiin to vowel i is 
semi- vowel y. Always iy>i just as uw>u ; generally yi->i 
(but cf. 40) ; iw may become iy>i or uw>u according to 
whether the semi-vowel is assimilated to the vowel or vice 

(a) Arabic 

As a rule long i is maintained, but occasionally we find 
final -1 in an open syllable modified to -e, thus hiya, " she " > 
hi or he (Egypt). 

(6) Abyssinian 

Final -1 in an open syllable is occasionally modified, 
especially in later dialects, thus pronominal suffix -m>-ne in 
Tigre. Ge'ez ze for Arabic dM. 

(c) Hebrew 

Generally i retained as in sddlq, " just " ; but the 
modification of final -i accented in an open syllable is frequent, 
thus Arabic dki = Heb. ze, Arabic damdni, " eight " = Heb. 
s^mone, and other instances in '>)'^ verbs, such as *yagliy> 
yigle ; so pe, " mouth," which returns to pi in the construct 
because no longer final. 


(d) Aratnaic 

In Bib. At. accented long 1 not final is preserved, but final 
-I accented usually becomes e- : unaccented it varies between 
i and e, thus Heb. ze but Aram, dl ; verbs final -y, iy>e 
as yih'e (Dan. vi, 8, 13), yiqre (Dan. v, 7), etc. Syriac negle, 
neo-Svr. gale in Urmi, etc. In Syriac original % retains its 
timbre in the west, but in the east becomes e, the written 
sign Hebotso {-r- or — ) being given these different sounds in the 
two areas. Thus Bib. Ar. n^hlyd, " prophet " (Ezra v, 1) = 
W.S. n^hlyo, E.S. n^beyd. So passive ptc. B.A. k^6tb, but 
also t'qel (Dan. v, 25, 27), Syr. W.S. qHll, E.S. qHel In 
Ma'lula ^ becomes e in closed syllable as plur. -en for -m. 

(e) Assyrian 

Long I and e seem to have been confused in the later 
language and freely interchanged. 

45 (iii) Long vowel u 

As e is the obscuring of i, so 6 is the obscuring of u, the 
u being imperfectly formed and so partaking of the a sound. 
(For o resulting from aw cf. 50.) As the vowel i is akin to 
the semi-vowel y so u is akin to w. 

(a) Arabic 

As a rule long u is retained, but in dialect we sometimes 
find it obscured as o, thus huwa " he " >hu or ho (Egjrpt). 
So especially in North Africa with the emphatic letters 
(cf. 57), as soq for suq " market ", etc. 

(6) Abyssinian 

Usually long u is retained, but sometimes it is obscured 
to o, and this tendency is increased in later dialects, thus 
hum > *humU, Ge'ez homU ; Tigre horn, dm ; Tigriua 'om, dm ; 
Amharic dm. 


(c) Hebrew 

Original u as m, shortening to 5, which is thus contrasted 
with 6 {<aw), which by decrease of quantity changes its 
timbre to ii. Thus *yaqus> ydqus, lahus, etc., but we also 
find qatul forms which have become qatol, as yaqus (Ps. xci, 3) = 
yaqos (Hos. ix, 8). On shortening we find instances in which 
u> d> u as ydqum, jussive ydqom, with Waw yayydqom, but 
these are cases of u produced by contraction (cf. 52) or by 
lengthening (cf. 48), not cases of original u. 

(d) Aramaic 

Original u is preserved in Bib. Aram, and West Syriac, 
but becomes 6 in East Syriac, thus ^wvapiov — West Syriac 
zunoro, East Syriac zondrd. In neo -Syriac, especially amongst 
the Jews of Zakhu and in Alqosh, it has a sound of u inclining 
towards 6. 

(e) Assyrian 

In later forms u tends to become il and so inclines 
towards ije. 

46 (iv) Short a 

(a) Arabic 

In classical Arabic a must be retained when in contact 
with a laryngal, it becomes d/6 in contact with an emphatic, 
and may become dfe in other cases. The traditional rules 
already given in connexion with long d apply equally to a, 
and are based, as already noted, on the questionable theory 
that Imale is due to assimilation to a neighbouring % oi y 
(cf. 43, i, a, 13). 

In Arabic dialect the change of a to a, e, i — the last chiefly 
in South Arabia and North Africa — occurs freely when there 
is not a laryngal or emphatic in contact, but its operation is 


somewhat irregular, and differences occur not only as between 
different countries, but in districts quite near one another. 
Thus S^Tian dialect ge^nel (yamal) " camel ", 'e.nt " thou " ; 
Egypt fidmd " heaven ", kelb " dog ", 'ewto " thou " ; 'Iraq 
keteh (kataba) " he wrote ", sems " sun ", heled " country " ; 
Oman swed {'asuad) " black ", 'ene " I ", huwe " he ", 
?nerkeb " ship " ; Tripoli 'awe " I ", 'enta " thou ", sams 
" sun " ; Mehri isicid " black ", 7nilik " king " ; Maltese 
sems " sun ", etc., a list which might be extended very con- 
siderably. Willmore's remark already quoted (cf. 43) does 
not apply to short a where the Imale is extremely common 
unless prevented by a neighbouring consonant. 

Change of a to 6 is generally due to the influence of a 
neighbouring emphatic (cf. 53), but accented d becomes o in 
South Arabia as described above (cf. 43 a, a), thus Mehri 
'arba'at > rhot " four ", 'asr- > oser " ten ", galaqa " go away ", 
fem. galaqat> galqdt, sab'-> hoba "seven", barq->bdreq 
" lightning ", etc. 

(6) Abyssijiian 

Ge'ez generally preserves short a, thus qatala " he killed ", 
malak " angel ", etc., but occasionally Imale occurs as in 
nehna for '^nahia " we ". This Imale becomes more frequent 
in the later dialects, as Tigre 'anta> 'enta " thou ", so 'enti, 
'entum, 'eaten ; Tigriiia nehna as in Ge'ez ; Amharic 'aiia> 
'ennih " I ", suffix -nn (for -nd) often -ne, qatala> qital, etc. 

(c) Hebrew 

Before examining any of the short vowels in Hebrew or 
Aramaic careful note must be made of the syllabic constitution 
and the influence of the accent. The opening or closing of 
syllables, and the incidence or removal of the tone and of the 
pausal accents, account for changes of quantity in the original 
short vowels, and to a less extent for their change in timbre. 


(i) Preservation of original a. 

Original a is preserved (a) in a doubly closed unaccented 
syllable not initial, as qataltum> q''taltem, and nouns of the 
type qattal as sabhaO gannah "thief", etc. But qaUala> 
qittel where the original vocalization is completely changed, 
a change which can be explained by noting that the perfect 
has been assimilated to the imperfect in the derived verb 
stems. Very often even in nouns the initial syllable 
shows d changed to I in double closure, thus yaqtul > yiqtoL 
*haqtala > hiqttl (stem vowel affected by imperfect), 
*haiqattala > hiOqattel, etc. Sometimes a is preserved by 
doubling the following consonant so as to produce double 
closure, a method more common with i and ic than with a, 
thus M- in hdmmelek, etc. 

(/3) In a closed accented syllable in verbs as qatdl, the 
stronger accent in nouns and in pause increasing this to a, 
as noun hakdm, pausal qdtdl. 

(7) Occasionally in an accented syllable in nouns such as 
sugdr " cage " (Ezek. xix, 9), which appear to be Aramaic 
loan words. 

(ii) Change of tinihre 

In Hebrew d becomes e/i as in the Arabic Imale, but this 
change is confined to closed unaccented syllables, and to the 
first syllable of nouns of the type qdfl, which has been opened 
by the insertion of a vowel producing qetel. Perhaps this e 
in qatl is due to assimilation, as qad> qatel> qetel. In 
either case the change to eji is prevented by a laryngal in 
contact (cf. 53). Thus ydqtul> yiqtSl, etc. 

(a) In a closed unaccented syllable the JMasoretic pointing 
varies between i and e, and we can see no reason why we have 
s*' elteni in 1 Sam. xxv, 5, and s^'iltiv in 1 Sam. i, 20. So Arabic 
yddkum — Hobx&w yedkem, but qaryat> qirya. But this 
change to ell was not made in the time of the LXX or of 


St. Jerome, thus Hebrew (M.T.) Mibsar = MaSaapif, Migdol = 
MaySoXov (in Hdt. ii, 159), 3Iidbdr = MaS/Sep, Miktah =- 
Vulg. Maclitab, BiVdm = LXX BaXaa/uu. 

(h) In the case of qatl the change is undoubtedly one of 
later date. The form qatl with the last consonant unvocalized 
bv the disuse of the case endings necessarily inserted a 
vowel which usually was weak e (cf . 68), unless before or after 
a laryngal (cf. 53), in which case it was a, thus malk> melek, 
zar'> zero,', etc. We would have expected the a in the 
accented and now opened syllable to become a, but this 
increase does not take place, and in due course a becomes e 
presumably by assimilation, and the change is extended by 
analogy to cases where a was inserted. Thus *sams (Arabic 
saws) became *snmes and so appears in the I;XX Bedaafxr}^ 
(Joshua X7/, 10), but semes in the M.T. So ^ITahr (Gen. xlvi, 
17) became Haher, and thus LXX "AjSep and XajBep as well 
as Vulg. Haher, but M.T. Heber. In *J5ar' (Gen. xiv, 2) the 
final laryngal produces Bara\ and so we have LXX Bapa 
(E. text) or BaWa (A.D. texts), Vulg. Bara, but M.T. Berd'. 
An increase of accent, however, raises this e to a as 'eres, 
'ares (Jer. xxii, 29, in pause). Final -ay also after becoming 
-e appears finally as -e, thus 'eqrac (1 Sam. xxviii, 15). 

(iii) Change of quantity 

(a) Short a is raised to a usually in an accented syllable in 
nouns, as *zakar (Arabic dhakar)> zdkdr " male ", *sulldm> 
sulldm " ladder ", *zakah (Arabic dhahah) > zdhdh " gold ", 
etc. ; in an open syllable before an accented vowel originally 
short, as qdtdl, zdkdr, etc., and sometimes before accented 
long as kdhud " honour ", snlom " peace ", etc. ; by the 
increased accent in pause as qdfdl, etc., and in compensation 
by the opening of a closed syllable, as ifhdrek, where the 
syllable should be closed by the doubling of the medial radical 
as in y^qattel, but becomes open by the incapacity of a 


laryngal or r to be doubled. When, however, the syllable is 
opened by the decay of J< closure even though subsequently 
closed by the loss of the case vowel, etc., we sometimes find 
a > a > as in *m's >rds" head " ; in such forms presumably 
the opening of the syllable took place at a very early date, 
and so resultant a was established before the change of a to 5. 
On the other hand, da'g> dag " fish ". 

(/6) Short a is decreased to a half- vowel ** or ^ sometimes in 
an open syllable before an accented vowel originally long, and 
always in an open syllable second before an accented vowel 
originally short, as p^rt " fruit ", but sdldm " peace ", so 
hHdm " dream ", qHalkem, etc. In pause this half- vowel is 
raised to e as 'perl, but the timbre a is preserved by a laryngal 
preceding (cf. 53). 

(d) Aramaic 

In Aramaic accented short a in a closed syllable is retained, 
thus qatdl>qHdl, so h^da " new", emph. Jfdtd. Accented 
a in an open syllable in nouns of the type qatl loses its accent, 
which passes on to the inserted vowel, an inevitable result 
of the syllabic constitution in Aramaic, thus *kdlh > k^leh, 
*mdlk > mHek, etc. In an open unaccented syllable either 
the vowel is lost, or the syllable is closed, or the vowel if 
remaining in an open syllable becomes a half-vowel, thus 
qaidl> qHdl, ^kasp> k'sef, kaspd " silveT " (Dan. ii, 32), 
sams> simsd " sun " (Dan. vi, 15), hasar>¥sar, bisrd " flesh " 
(Dan. ii, 11), etc. In Svriac a may be retained or it becomes 
e in an unaccented closed syllable, thus malkd " king ", 
'egartd " letter ", Bib. Aram, 'igg'rd, yaqtul> neqtul, dakar> 
dekrd. The change to e is commoner in Went Syriac, it 
more often remains a in East Svriac. In Mandsean and Onq. a 
in this position is retained or else becomes i. Thus Galilsean 
milhd " salt " (Hebrew mdlah), 'ibbd " father ", sifrd 

morning ", etc. 


(e) Assyrian 

Original a is generally retained in Ass}T:ian, but in later 
forms it sometunes becomes e/L 

47 (v) Short vowel i 

(a) Arabic 

Although short I is retained in classical Arabic, its sound 
inclines to e with a laryngal (cf. 57) ; thus giss> gess, 'ilm> 
'elm, etc., a tendency which becomes more pronounced in 
dialect, and it inclines to ii with an emphatic, as nisf> nilsf 
" half ". In dialect it generally becomes e in an unaccented 
syllable, as tis'at> tcs'od "nine" (Morocco), yihlf> yesuf 
" he sees " (Egypt), sittat> settd " six " (Morocco), etc. This 
change is regular in Syria except in -iy or final -t, thus -Jem 
for -Mn, etc. 

(6) Abyssinian 

In Abyssinian I becomes c and thus is confused with u, 
which also becomes e. Thus Hisdn (Arabic lisdn) > lesdn 
" tongue ", etc. But in a doubly closed syllable i becomes 
a, as *labisa> labesa or labsa, 2nd sing. masc. labdslca. Short 
i is re -introduced into Amharic but does not correspond with 
common Semitic i. 

(c) Hebrew 

(1) In compensatory lengthening caused by the opening of 
a closed syllable i becomes c, thus *birrek>berek, so *mdli' > 
male. Thus we have either yiyras or yeres. 

(2) In an accented syllable open or simply closed * > e, as 
stfr- absolute sifer, opened by the addition of the inserted 
vowel (cf . 68), Icdhid > kdbid. Sometimes, however, we find 
an alternative l resulting from more strongly accented i, 
which will decrease to e by removal of the accent or by 
double closure. Thus in the causative yaqtil, decreasing to e 
in double closure taqtelnd and in the jussive yaqtel and 


imperative haqtel. Very rarely accented i is preserved as in 
bin (Deut. xxv, 2) for the commoner bcit ; so in the particles 
'im, mm, %m. 

(3) In an accented doubly closed syllable i becomes a, as 
kdhed (for kdhid), kabddtd, so *bint> *bd)it> bad, bittl \ 
Arabic siddiq — Hebrew saddtq. So in pause hedez becomes 
heOdz (Isa. xviii, 5). 

(4) In a closed unaccented syllable i is preserved without 
change of quantity or timbre, as sifn, sifrehem, 'immi, etc. 
But original i becomes e after it has become e or half -vowel, 
and is then by change of accent restored to a short vowel. 
Thus e can result as a shortening of e, or lengthening of % the 
original i from which these vowels were produced being 
forgotten ; there is no confusion between i and e here in the 
way there is between e and i resulting from the Imale of a. 
Thus *yaysib (for yaivsib) becomes yeseb, then as the accent 
recedes with prefixed Waw wayyiseb. By the medium of 
the half-vowel we have some (rarer) forms such as *sikm 
(whence sihnl, etc.), s^kem " shoulder ", in pause sekem 
(Ps. xxi, 3), and in the proper noun se^ma " towards Shechem " 
(Hos. vi, 9), but Vulg. Sichem. 

(5) In an open unaccented syllable i may become long t, or 
it may be preserved by doubling the following consonant and 
so closing the syllable, thus *qim,ds " nettle " as qvmos 
(Hos. ix, 6), qirmnos (Isa. xxxiv, 13). This closure by doubling 
the following consonant is more frequent than with a. but 
not so common as with u. Thus in the verb yasbub > yasuh{b) 
(cf. 74) we may have either ydsob, the d lengthened in an 
open syllable (cf. 46, c, iii), or yd- may become yl- in the 
normal course and then produce yissob to preserve short i ; 
in tishi such a doubling is unnecessary as the syllable is already 
doubly closed. 

(6) But in an open syllable before an original long vowel i 


may become a half-vowel, and this is the regular course in an 
open syllable second before an accented short. But after {.^ 
the half-vowel becomes e by the compensatory lengthening 
already described. Thus sifr- > s'fdrtm, Arabic hisdrat = 
Hebrew h^sord " good news ". 

(d) Aramaic 

In Bib. Aram, i varies between i and e in an accented 
syllable sunply closed, thus *salit > sHet, *sdgir > s^gir. In 
closed accented medial or final, if originally closed, and always 
if doubly closed, i> a, thus namir (Hebrew ndmir) > n^mar, 
hint > bad (batt), kabtda > k^'bid, so b^'es (Dan. vi, 15), 'dkim 
(Dan. iv, 6). In closed unaccented syllables i is retained as 
sippar > siff^r-in, 'ilU, etc. 

In Syriac i becomes e, but i is sometimes retained near a 
sibilant (cf . 58), thus naqif> rifqef, sippar > seffar, etc. 
In the Targums i becomes e in unaccented open syllables. 

(e) Assyrian 

In Assyrian i is unchanged unless afiected by a neighbouring 

48 (vi) Short vowel u 

(a) Arabic 

Near the emphatic letters or the laryngals ii is sounded as 
0. In the dialect of Oman this 6 sound is extended to ii in 
all noun forms, as 'uqdb> 'oqdh, etc. So in dialect -kuma> 
-kon (Syria), -Jium> -horn (Hadr.). In doubly closed syllables 
in dialect w sometimes becomes i, thus dubb> dibb " bear " 
(Egypt), surrbm> siwm "eye of a needle" (id.), Qubtiyy> 
Qibti "Copt" (id.), kid> kil "cat" (Oman), hunaka> 
hinak " there " ('Iraq). 

{h) Abyssinian 

In Abyssinian w becomes e and so confuses with i, thus 
*liihh> lebb " heart ". Arabic 'udhn =- ' ezn " ear ", sukdr 


(pass. mF[n.)> sekdr as verbal noun "drunkenness". In 
Tigrina u sometimes becomes o in the syllable -dm, as 

nesom, etc. 

(c) Hebrew 

(1) In compensatory lengthening at the opening of a closed 
syllable u> 6, thus hurrah > hdraJc (cf. 12). 

(2) In an accented syllable open or closed « > 6, 
thus *'dgul> 'agol, Aiabic (juds = Hebrew qodes "holy", 
Arabic kull — Hebrew Jcdl{l), *qafiln > qatdn. The e which 
appears in the pronoun 'ait em, in the personal termina- 
tion -tern, and in the suffix -kem, must not be 
regarded as a modification of original u but as 
produced by the analogy of the corresponding feminine, just 
as in Arabic the fern, 'antunna, -tunna, -kunna, of the 2nd plur. 
fern, is affected by the influence of the masc. The original 
r of the fem. appears in Arabic dialect, and in Assyrian 
masc. attuna, fem.. attina. Parallel to this change in Hebrew 
is the use of the fem. sing, enti in Tunis for the masc, and so 
ntin in Tlemsen. 

(3) In an unaccented open syllable before the accent u 
normally becomes o, as *sii'ar> so'cir " bad " (Jer. xxix, 17), 
but occasionally in older forms such as proper names we find 
u> nin this position as su'dr (Num. i, 8 ; ii, 5). 

Alternative to this is the closure of the syllable by doubling 
the following consonant, and thus we have yussab as alternative 
for yusah for yusah = yushab (cf. 157). 

(4) Second before an accented vowel originally short il 
becomes half- vowel, like a, t, in the same position, its timbre 
as ° being preserved after a laryngal or velar, thus *quds > 
qodes, plur. q°dmim, so po'al, p^'dlim., and similarly u becomes 
half- vowel before accented long, as *huly->*huli>h°ll, in 
pause Jioll. 


(5) In a closed unaccented syllable the Babylonian pointing 
preserves u, but the Tiberian pointing, which is that found in 
the printed text, varies between ii and o, so that sometimes 
both appear in different passages in the same word, thus in 
Isa. xxvii, 11, y^kunnenu, but the same word as ■ifhonnenu 
in Ps. Ix-^/ii, 2. Thus quttal remains quttal, but huqtal becomes 
Jioqtal. Generally 6 appears where u has been raised to 6 
by the accent and then reduced by the passing of the accent, 
as quds > qodes, qodsl. and so when n has been reduced to 
half- vowel, i.e. after a laryngal, as hHi, holyo, in pause lioll, 
which practically gives the same consonant influence as in 
Arabic (cf. above). On the other hand, ruqq becomes roq{q), 

(d) Aramaic 

(1) In an accented syllable ii is generally preserved, and so 
in a closed accented. This is the case in Bib. Aram, in nouns 
of the qutl type, and so in noun forms as JcvMayya. (Ezra v, 8), 
h^'nul-Jcad (construct, Dan. iii, 2), but o in hohitd " wisdom ", 
'orMOdk (Dan. v, 23), and in verb forms as honhad (or ]i°nha9, 
Dan. v, 20) for Viunhat. 

In nouns u becomes o in an open accented syllable as 
hdsok (type qatal, Dan. ii, 22), and u in closed accented as 
'^siir (Dan. iv, 12). 

(2) In Syriac u is preserved in a closed syllable in West 
Svriac, but becomes o in East Syriac (cf. 45), thus yaqtul> 
West Syiiac neqtul, East Syriac aeqtol, Arabic Icurrasat " little 
book " > West S}Tiac Jcurroso, East S}Tiac korrdsd. In 
open accented u becomes West Syriac o and East Syriac a 
(cf. 45). 

(3) In the Targums u is retained in the verb in an open 
accented, as qHulU, but is rare in closed unaccented. In 
closed accented u> o, rarely u retained, thus yiqtol, -hum > 


(4) Mandsean confuses u and o. In neo-S}Tiac a frequently 
becomes eji. 

(e) Assyria7i 

In later Assyrian u sometimes becomes i, thus older putdqu 
appears later as pitequ. 

49 (vii) Diphthong ay 

(a) Arabic 

The diphthong ay is normally retained in classical Arabic, 
the only exceptions being found in the combinations -aya-, 
-ayi-, etc, (cf. 52), and in such noun forms as Halay-, which 
become tola-, and in double closure where ay > a, as laysa 
" is not ", Idsta " thou wast not ". 

In dialect ay commonly becomes e, and in North .Airica 
and sometimes in Egypt this frequently results in i, except 
in those instances where ay is retained under the influence of 
a neighbouring laryngal or emphatic (cf. 54). Thus hayt 
" house " > bet ('Iraq, Egypt), bit (Morocco) ; 'alaykum 
*' on you " > 'alehum ('Iraq) ; ganayna{t) " garden " > genena 
(Egypt) ; hayr " wealth ". kher (Oman), and so through all 
forms of dialect. 

(b) Abyssiniayi 

The diphthong ay is retained in a doubly closed accented 
syllable, before y, or near a laryngal, elsewhere it becomes e. 
Thus bayt " house " > bet, laylat " night " > Mat, 'ayte 
" where ? " haymdnot " faith ", etc. 

(c) Hebrew 

The diphthong is retained in a closed accented syllable, 
but in an imaccented syllable or in an open syllable whether 
accented or not it becomes c. In nouns of the qatl type with 
medial y, which retains its consonantal value, it is necessary to 



insert a short vowel before the final radical ; norraally the 
inserted vowel is e (cf. 68), but under the influence of the 
preceding y this becomes i and thus we get "^hayt > bayith 
"house", etc., in the construct beth and similarly *layl-> 
layil, lei, and in an open syllable haykal> hehal "temple", 
but this is sometimes restrained by analogy as in baythd 
"homewards", layld "by night". In a final syllable as 
'asray>'esre "ten", but also with resultant -e shortened 
to -e, as qanay> qane, etc. Occasionally we find ay 
retained abnormally as in 'ayslrem " I will chastise them " 
(Ilos.vii, 12). 

{d) Aramaic 

The conditions generally are the same as in Hebrew, thus 
hayit {=:bayt) "house" constr. beth (Dan. iv, 27), hayil 
constr. Ml " craft " (Dan. iii, 20), final accented as in yib'e 
(Dan. vi, 8), In Syriac ay is retained only in an open accented, 
thus beth, baytha " house ", and as in Hebrew ay> e (East 
Syriac = i West Syriac) in a final accented, as negle. In 
modern East Syriac ay sounds as e or ei (i.e. as ei in English 
height). In the Targums ay becomes eln closed accented and 
thus we have, as in Hebrew, layil, constr. lei " night ", lelya 
" by night ". In Mandaean and in West Syriac the resultant 
e becomes i, as btth " house ", etc. 

(e) Assyrian 

In Assyrian the diphthong in the negative prefix ay- is 
retained, e.g. ayka, ayna, etc., but generally ay becomes e 
as ayniq> eniq, etc., and sometimes i as bit "house ", and 
sometimes ue as bvet " house ". 

50 (viii) Diphthong aw 

The diphthong aw shows a history very similar to that of 
ay, its tendency being to reduce to o w^here ay becomes e. 


(a) Arabic 

In classical Arabic aw is normally retained save in the 
combinations awa, aw, etc., described elsewhere (cf.p. 119). In 
dialect aw becomes o, and often in North Africa and some- 
times in Egypt this results in u unless restrained by a 
neighbouring consonant (cf. 53). Thus gaw'dn " hungry " > 
go'dn ('Iraq), 'awtdd " mountains " > 'iltdd (Egypt), yawm 
" day " > yom (Egypt, S}Tia, Morocco, etc.), 'awldd " sons " > 
'uldd (Morocco). 

(6) Abyssinian 

Generally aiv > 5, but aiv is retained in a doubly closed 
accented syllable and before w, thus ^yawm > yom " to-day ", 
jnafawwes " physician ", etc. 

(c) Hebrew 

The diphthong aw is usually retained in a closed accented 
syllable, but becomes 6 in an ppen syllable or in closed 
accented, thus tawk> taioeh, constr. toJc, suffixed tokl. Here 
the consonant value of iv requires the insertion of a vowel 
as in all qatl forms. Accented aio appears as div, thus ivdw 

"nail", sdw (written N"1SJ^, Arabic i.j^) "evil", which 

appears as sdiv in Job xv, 31 (Kethib). But aw often becomes 
6 where its retention would have been expected, as yawm> 
yom plur. ydmln as though from sing, ydm (cf . Targ. ifmdmd) ; 
sometimes we find both forms as 'dwld (2 Sam. iii, 24) and 
'old (Isa. Ixi, 8). Very rarely aiv is retained where o would be 
expected, as in sdlaivtl " I am at ease " (Job iii, 26). 

(d) Aramaic 

Generally aw is retained in an open syllable but becomes 
o/u when the syllable is closed, thus yawm> West Syriac 
yUm, East Syriac yom, yawmd " day ", but yom, yomd in 
Bib. .Aram. (cf. Dan. vi, 11), and so Bib. Aram, sdf, sojd 
" end ". 


(e) Assyrian 

The diphthong aw regularly becomes U, thus yawm > um-u, 
awsih > nsih, etc. 

51 (ix) Contraction o£ vowels with semi-vowels 

In treating of vowels with semi-vowels we have to consider 
(i) cases in which the semi-vowel is the closure of a syllable, 
and (ii) cases in which it is the initial either (a) following 
a closed syllable, or (b) following an open one. 

(i) Semi-vowel as closure of a syllable 

With semi-vowel as closure we may have ay, aw, iy, iw, 
uy, uw. Of these ay and aw are the two diphthongs which 
have been already considered ; -iy and -uw necessarily become 
-t and -u. The two remaining cases are iiv and uy. In these 
either the vowel may assimilate to the semi-vowel so that 
iw > uw > u and uy>iy> i, which are treated elsewhere as 
vowels influenced by consonants (cf. 60) ; or the serai- vowel 
may assimilate to the vowel, iw > iy > %, and uy > uw > u, 
which have been treated as consonants influenced by vowels 
(cf. 40). 

52 (ii) Semi-vowel as initial of a syllable 

(a) Following a closed syllable 

This gives the instances of wa, ya, wi, yi, wu, yu, after a 
consonant. If these groups are not maintained but contract 
to long vowels the syllabic structure of the word is completely 
altered, e.g. where ya becomes a in yahyabu > yahdbu the 
first syllable is opened and its former closure becomes the 
initial of a new syllable. It is with such changes as these 
that we are now concerned. 

(1) wa, ya> a 

Arabic yuqwalu > yuqdlu, yusyalu > yusdlu, the resultant 
a shortening to a in the jussive yuqdl, yusdl. In North Africa 


ya sometimes results in 7 as in initial Imm for yamm " right 
hand " (Moro(3co). Abyssinian yabwa> yehd. Hebrew 
resultant a becomes 5 (cf . 43c) as yahwa > ydbo, naqwam > 
ndqom, nahyan > ndbon, yabivas > yeboL Aramaic, a in Bib. 
Aram, and East Syriac, 6 in West Spiac nfqwam > m^qdm^ 
m^qom ; nfsyam > m^sdm, nfsdm. Assyrian ihwa > ihd, 
asyam > asdm. 

(2) wi, yi > I 

Arabic yuqwilu (imperf. conj. iv) > yaqilu, yasyiru> 
yasiru, shortening to i in the jussive : tvi > yi> yyi in mawit > 
tnayyit. Abyssinian yasivim > yestm. Hebrew yaswim > 
ydsim, shortening in the jussive to e, as ydsem. Aramaic 
yaswim > n^svm in West Syriac and so y^stm in Bib. Aram. 
East Syriac tfsem. Assyrian itwih > itih, uhwin > ukm, 
adyin > adln. 

(3) wu, yu > u 

Arabic yaqwulu > yaqulu, shortening to u in the jussive. 
Abyssinian yaqwum> yeqiim. Hebrew yaqwum> ydqum, 
u shortening to 6 in the jussive ; also indicative in o and so 
identical with the jussive, as in yebos. Aramaic yaqtvum> 
y^qum (Dan. vi, 20) shortening occurs only in yadwuran> 
y^duran (Dan. iv, 9, Q're), Syriac n^qum (East Syriac n'qom). 
(h) Following an open syllable 

In this case there are two different methods of treatment, 
(1) that followed in Assyrian, Abyssinian, Hebrew, and 
Aramaic, and (2) an entirely different one followed in Ai-abic. 

(i) Abyssinian 

After a vowel the semi-vowel loses the vowel following 
and thus reduces to a diphthong, as awa, awi, awu, become 
a'lv and aya, ayi, ayu, become ay, thus *qaivama > *qawma > 
qoma, and *sa,yama> *sayma> sema, etc. The only excep- 
tion to this treatment is that in noun forms iw is sometimes 


restored as ew by analogy with aw (cf. sect. 50), thus zerew or 
zeru " scattered ". 
(ii) Assyrian 

In the case of an intervocal semi-vowel Assyrian discards 
the preceding vowel, so that awa, iwa, uwa become wa, etc., 
and the resultant is treated as described above, thus ahiwan> 
ahvan > akdn, etc. But in the Permansive wi becomes e 
instead of i, as in ken for kawin. In the Intensitive (D) the 
medial, though doubled, quiesces and then either (1) the two 
vowels are left and form a long vowel or diphthong, or (2) the 
first vowel falls away and the doubling fails, the resultant 
following the rules given above, thus from uqayyis we have 
either uqais or uqls. But when a vowel suffix is added the 
final radical is doubled and the vowel shortened, as utawiviru > 
utiru > utirru. The Ass^aian treatment of intervocal semi- 
vowels provides the key to the phenomena observed in Hebrew 
and Ai-amaic. 
(iii) Hebrew 

(1) atva, aya> a, i.e. as wa, ya (cf. Assyrian), thus qaivam> 
qdm. But in noun forms this contraction presumably took 
place at an early date, so that resultant a has become 6, as 
tawab> fob. In double closure a > a, as qdrntd. 

(2) awi > e, but ayi > i. Thus maivit > meth, nawir > ner, 
bayin> bin, or else the ayi verb, which is comparatively rare, 
assimilates to the aya form as ban. In double closure t/e> 
i> a (cf. 47). In the active participle Qal the divi becomes 
either (a) d and so o, as lot (Isa. xxv, 7), or (b) an'i> a' (cf. 
Arabic qdwil> qail), <T,nd so d as in Idt (Judges iv, 21). 

(3) aivu, ayu > a, but divu, dyu > u, thus bawus > bos, 
qawum >qum. In double closure resultant o >o, as in bostd. 

(iv) Aramaic 

(1) awa, aya> a (Bib. Aram, or East Syriac), 5 (West 
S}Tiac), as qawam> qdm, qdm. 


(2) awi, mji>i (Bib. Arara. and West Syriac), e (East 
Syriac), as sayim> sim, sem. 

(v) Arabic 

Arabic shows a form of vowel contraction which differs 
from both the Abyssinian and the Assyrian-Hebrew types. 

(1) No contraction takes place in the medial syllable of a 
word which has a semi-vowel medial and also a semi-vowel 
final, as sawaya> sawd " roast ". 

(2) Later derivatives, such as denominal verbs, verbs 
specialized to express wonder, and elatives, show no 
contraction with medial semi-vowel. No contraction takes 
place with medial semi-vowel doubled, as qaivwala, nor in 
conj. ii, as qdwala where dwa has been formed from awiva 
(cf. 135). No contraction takes place in iya, as radiya, but 
iwa becomes iya, or it may become d (dialect of Tayyi), or 
I (Mufass. 120). 

(3) awa > d, as qawala > qdla, nadhaiva > nddha ; so nouns 
such as bawab> hob " door ". But this is not a pure d, it 
inclines towards d> o> u, and so contraction in double 
closure will result in u not d, as qulta. 

(4) aivi, ayi, aya > d, but inclining to d> e, and so 
shortening to * in double closure, as sayira > sdra, sirta ; 
*yardaya {&\ibi.)> yardd ; nawir>ndr "light" contrasted 
with Hebrew ner. 

(5) awu, ayu > a, as yundawu > yundd., yardayu > yardd, 
tawula> tola. 

(6) uwi, uyi, iyu > I, shortening to i in double closure, as 
quwila > qtla, qilta ; suyira > sira, sirta ; yarmiyu > yarml.. 

(7) uwu, uyu > u, as yanduwu > yandu. 

In the active ptc. dwi, dyi, become d'i as qdwil > qa'il. 
In common speech, however, awd often becomes uwd, as 
guwdr for gawdr " female slaves " ; a few instances of this 
occur also in classical speech in proper names, as suwd'iq-u, etc. 



Temporary modifications of the vowel sounds are due to 
disturbing factors • parallel to those which we have already 
noted in consonant changes. Such modifications may be 
produced by neighbouring consonants or by other vowels, and 
in the latter case the resultant changes may be classified as 
assimilative and dissimilative. We will consider first the 
phonetic influence of consonants on neighbouring vowels, 

53 (i) Influence of consonants on vowels 

In broad outline it may be said that the laryngals attract 
the vowel sound a, the dentals and sibilants attract i, the 
labials u, and the emphatic letters tend to modify a to o. In 
each case the outlet of the vowel sound is controlled by the 
disposition of the vocal organs due to the preceding or 
following consonant. 

(a) The a sound and the laryngals 

Imale is restrained or prevented by a preceding or 
following laryngal. In Arabic the Imale is prevented by 
preceding or following ', h, h, h, or velar g, but not by the weak 
laryngal Hamza ; also by r preceded by a or u. But this 
preventative power is subject to certain restrictions : (a) 
the laryngal or r does not prevent Imale when it is the initial 
of a verb with medial or final ivjy ; {13) and with rare exceptions 
it is not efiective unless the a is in contact with the laryngal. 
(7) In the case of r, to prevent Imale of a the r must be in 


actual contact, as rasid, etc., and it must not be vocalized 
with i, so Imale cannot occur in 'ahsdrihim " their eyes " 
(Qur'an, ii, 6), 

Yet against this influence we find a contrary principle in 
the modification of a to a, e, i, near a laryngal appearing 
occasionally in particular areas, as in the dialects of Egypt 
and North Africa, where after a laryngal, and more especially 
after a, suffers this Imale, thus 'dsal for 'asal "honey" (Egypt), 
^elm for 'alima " know " (Tunis). 

In Hebrew where a in an unaccented closed syllable becomes 
eji, and in the imperfect preformative usually i, a laryngal 
produces e, as 'eqtdl, yeh°'bas. 

In Assyrian a' becomes e by the fall of ^< (but cf. 43, e), as 
*ra'l > res-u, ba'al > *ba'al > hel ; and so I frequently becomes 
e before r, h, as in unammera, tatneih (Delitzsch, Ass. Gr., 44). 

54 (b) Vowels i/u become a near a laryngal 

In the imperfect of verbs with a as vowel of the medial 
radical in the perfect we usually find ^ or u, but where the 
medial or final radical is a laryngal the imperfect has a. Thus 
qatala, yaqtul, galasa, yaglis, but fa'ala, yafal, and qata'a, 
yaqta". So in Abyssinian qatala, yeqtel, but fataha, yeflah. 
Similarly in Hebrew sdhat, yishat and sdlah, yislah. In Bib. 
Aram. sHah, yislah (Ezra v, 17). 

In Arabic dialect i/u before or after a laryngal tends to 
become a ; thus mihrab> mahrah " prayer niche " (Egypt), 
'inah " grape " > 'aneh ('Iraq), 'anab (Morocco) ; 'usbu' 
" finger " > seba' (Maltese), saba' (Morocco), 'asaba' (id.). 

In Hebrew and Aramaic the short vowel e{i) normally 
inserted after the medial to vocalize nouns of the type qail, 
qitl, quit, becomes a with a medial or final laryngal, thus 
malk-> melehh, but zar'> zero,', Aramaic z^ra' "seed", 
rumh > Hebrew romah, Syriac rumah. 


55 (c) Long vowels i/u insert a before a laryngal 

In disposing the vocal organs to enunciate a laryngal 
following long i or u the timbre of the vowel sound is modified 
and a glide is made which gives the sound of a. Thus in 
Arabic dialect nh becomes ndh " wind " (Egyptian). 
Occasionally in dialect this occurs also with iju, thus 'usbu' > 
sboa' " finger " (Tunisian dialect). In Hebrew a is inserted 
before h, h, \ following r/e or ujo, as in ruah " wind ", sdluah 
"sent", gihodh "high". In the LXX this frequently 
appears as e, thus ydftd' > 'Ia(f)ie, Vulg. Japhie, JapJiia, 
nddh > Nwe, Vulg. Noe, or it is omitted as in ydnddh >Iavcox, 
Vulg. Janoe. 

56 (d) The a sound with the emphatic consonants 

We have so far noted the influence of laryngals on the timbre 
of vowels. Here we have to notice the influence of the emphatic 
consonants (cf. 9) on vowel a. 

In Arabic, before or after the emphatic consonants s, d, t, z, 
q, the vowel sound a is thiclcened to a, o, and sometimes to u. 
Thus sdhr> sahr "patience", fasal>fsol "chapter" 
(Algerian), daraha> drob "beat" (id.), etc. In Oman the 
imperfect of verbs Avith third radical emphatic (or r) shows 
vowel o, as yohqor, yunhor, tjorqot. In the dialect of 'Iraq 
a becomes o near an emphatic, as yaqifu > yogaf. In North 
Africa an- > ad near a laryngal or emphatic, as saiot > sadt 
" vine ". In Arabic the use of the Imale is optional, but the 
thickening of a near an emphatic is obligatory. 

57 (e) The i/u sounds with laryngals or emphatics 

(i) Inlluence of an emphatic or laryngal on vowel i. 

%> ejo with an emphatic in Egypt, North Africa, and 
'Iraq, as qitfa{t), qotta " cat " (Egypt). 

t> o near an emphatic or laryngal in Syria and North 
Africa, in Tunis only after ' or A ; thus hisn > Mm, hosn 
"stronghold" (S}Tia), kirs>k6rs "origin" (Morocco). 


i> dje or uju before or after an emphatic or laryngal, as 
dirs> durs " molar tooth " (Syria), nisf> nusfinuss) " half ", 
himdr> Jiuniur " ass ". 

In S}'Tia a ov i becomes il with s, f, q, h, or '. In Mehri 
ay, i become ai, ei, near a laryngal. In neo-Syriac a/j becomes 
u in the same syllable as s, t, h, q, or ', but there are many 
exceptions, especially with s ; ' modifies the vowel only in 
Alqosh and Kurdistan. 

(ii) Influence of an emphatic or laryngal on vowel u. 

4 > o near an emphatic or laryngal, as mq > soq " market ", 
zufr> zofr "claw", 'arus> 'aros "bridegroom" (Morocco, 
Tunis, Tripoli) 'umr>'omr "life" (Egypt). In Maltese 
ii followed by ', or ^ forms o with loss of the laryngal, as 
sugl> sol " business ". 

(iii) Diphthong aw with emphatic or laryngal. In North 
Africa and sometimes in Egypt niv becomes ii instead of o, 
but the presence of a neighbouring emphatic or lar}Tigal 
restrains this change and preserves o. In North Africa the 
emphatic consonants include in this relation not only r but 
also I. Thus 'awldd> 'oldd (not 'uldd) "sons" (Morocco). 
In Mehri aw and o become auloii near a laryngal or emphatic. 

58 (f) Influence of Sibilants and Dentals 

Dentals and sibilants tend to attract the voAvel i. Thus in 
the dialect of Oman the second vowel of a verb stem is usaally 
eji with a final radical dental, sibilant, I or n ; thus hased, 

In Aramaic a frequently becomes e before a sibilant, thus 
ras> ras > res head . 

In Assyrian a becomes i with sjs as asikin for asakan, 
and near a dental d tends to become eji. 

59 (g) Influence of the Labials 

The general tendency is for the labials, amongst v/hich we 
must include the sonant labial m, to attract the vowel u. 


Thus in Arabic we frequently find change of a/i to u hefore 
a labial, as in 'umm (Hebrew 'em) for 'imrn " mother ", 
^uhl for (J)bl " troop ",fnm ioT fani " mouth " (Morocco), etc. 
Occasionally a similar change takes place after ?n, as mulk 
for malk " king " (Nejd), mular for matar " rain " ('Iraq). 
In Abyssinian dji becomes e {-= u) before a labial in nafs > 
nefs " soul " ; Amharic gamal > gemal " camel ". In Hebrew 
the conjunction wa- becomes u- before a labial or labial 
sonant, thus umelekh, etc. In Aramaic a[i sometimes become 
u before a labial or labial sonant, thus Hebrew sem, Syriac 
sem> Bib. Aram, sum " name " ; Syiiac gamld> Mandsean 
gumld "camel". Arabic ^ism> Syriac gusmd "body". 
Occasionally this occurs after a labial, as Arabic baraha{t)> 
Syriac hurkHd " blessing ". Assyrian a, e, i, become u before 
a labial, as sem> sum-u " name ". 

60 (h) Influence of semi-vowels 

In the case of semi-vowels w is homogeneous to u, y to i. 

uy> iy> I, Arabic hayt " house ", dimin. huyayt or hiyayt : 
'abyad> " white ", plural buyd> bid (Lane, Lexicon, 283). 

yu>yi, yumna>yimna " right hand " (Socin, Diwan, ix, 6) 
in dialect of Nejd, So mabyu' > mabV where yu> yi> i 
(Lane, Lexicon, 285). 

wi > wu> u generally in Hebrew and Aramaic. 

awa> uwa in common speech. 

aw > uw > u, sometimes in the dialect of Egypt and North 
Africa, as 'aivtad> 'utad (Egypt). In Assyrian yawm> um-u 
" day ", awsib> usih, etc. 

61 (j) Vowel assimilation assisted by a laryngal 

(i) In the dialects of Egypt, Sjnria, and sometimes in that 
of Oman nouns of type qatil become qitil with a medial or 
final laryngal, thus Egyptian wihis " dirty ", ivihis " desert " 
(cf. 46, a). So in Nejd, 'Iraq, Oman, and North Africa we find 
qatil > qitil> qiil imder similar circumstances. 


In Abyssinian e (from i or u) beconies a before a laryngal 
followed by a, thus yehawer becomes yahawer, and nouns of 
type quttal become qatldl with medial laryngal. Similarly 
d-ii becomes e-w (= w-w) when there is an intervening laryngal 
as nasa'u> naseu "they removed", but no assimilation 
takes place when the laryngal is one of two consonants in 
contact as in nehnd " we ". 

In Hebrew when by reason of accentuation a short vowel 
becomes a half -vowel, this half- vowel retains its original 
timbre with a laryngal, as 'ddm, 'ddtn, ^°-(Lind " delicate ", 
'el 'il, 'Hohlm " God ", Viiily, Ull, Ml " disease ". WTien a 
syllable before the accent is closed by a laryngal, it is opened 
by adding a half-vowel, and this half-vowel assimilates in 
timbre to the vowel of the preceding syllable, thus pTH 
imperfect pTIT', and when by further change of accent the 
new syllable is closed the inserted half-vowel becomes a short 
vowel as yo'°bad, yo'ohdu, etc. With an intervening laryngal 
d-e becomes d-d, as in V^i^, V^NH. In Hebrew and Aramaic 
la-, ka-, wa-, before a laryngal with a half-vowel assimilate in 
timbre to it, as nt^i " like a lion ", "ibx'?, HIH^ (Dan. ii, 35), 
tJ'i^5^ (Dan. vii, 4) ; in the Targums this assimilation is 

irregular, and in later Aramaic and Syriac it has become 

63 (k) Change of quantity due to a laryngal 

The Arabic dialect of Oman lengthens the final vowel in 
the units before 'ser {'asara) in the numerals 11-19. Thus 
'ahada 'asara becomes hedd'ser " eleven ", etc. In Abyssinian 
the vowel is lengthened in a syllable closed by a laryngal, as 
tufasdJikii for tujasdhku " I rejoiced ". In Hebrew and 
Aramaic laryngals and r are incapable of doubling and there- 
fore all forms in which a laryngal or r should be doubled 


show a failure of the doubling with compensatory lengthening 
of the preceding vowel, thus Hebrew herekh for '^birrekh, etc. 

63 (ii) Influence of vowels on vowels 

(a) Assimilation of vowels 

(1) a-i> i-i 

Arabic, in the dialect of Egypt and to a less extent in that 
of Oman and 'Iraq, verbs with i after the second radical tend 
to change a after the first to i also, thus sanha " drink " 
becomes sirib (Egypt), sereb ('Iraq), etc., a parallel change 
taking place in verbs with u (cf. (2) below). Sometimes this 
change is extended to conj. vii, viii, as yinbirik for yanbarik 
(Eo-ypt). We have already noticed a similar change in nouns 
with a laryngal (cf. 61). In Abyssinian a-e> e-e with a 
laryngal between, or when the first vowel is in a doubly closed 
syllable, as tefseht for tafseht " gladness ", the e standing for 
^/w. In Hebrew there are some instances in which nouns 
of type qatil become qitil as seber, sibrl " breach ", and 
nouns maqtil become miqtil as mizbeah " altar ". 

The change of a-% to i-i is rare in Arabic, but we find in 
Spanish iVrabic midina for madina " city ", possibly only an 
instance of Imale. Sometimes Abyssinian shows the change 
of qatil to qetil as in lehiq " old ". In Assyrian there is 
frequently a change of a to e before final -1, especially in 
verbs with third radical -y, as ubewm for ubannl ; so aqtarib 
becomes aqterib. 

(2) a-u > u-u 

In Egypt, etc., verbs with second vowel u change a of the 
first radical to u by assimilation parallel to that in i verbs, 
thus swjur for sagura (Egypt), etc. So generally in quadri- 
literals of the type qatlul as furhud " short " for farhud, and 
similarly tuJiluk for tahluk (type taqtul), etc. (cf. Wright, 
Ar. Gr., i, 115, C-116, A). Abyssinian shows a similar change 


in the passive participle qetul for qatul, and Tigre dialect 
regularly changes final -a to -e (=w) before suffixed -Aw 
" his ". The change of a-u to u-u occurs also in Assyrian 
when the first a is in a closed syllable. 

(3) i-u > u-u 

This change occurs in Arabic in the imperative of the 
primary stem. Normally this stem, having no vowel after 
the first radical, prefixes i- (cf. 66), but before u in the stem 
we find i replaced by u, as uqtul. In dialect a vowel is 
inserted before a consonantal suffix as the case endings have 
become obsolete ; this is usually -i-, but before -kujn, -hum, 
it is u, as 'umm-u-hum " your mother " (Egypt). So in the 
Assyrian imperative, a vowel is inserted after the first radical 
and this assimilates to the stem vowel, as kusud for *ksud. 
In Babylonian (not Hammurabi) we find izuzzum > uzuzzum, 
itulum > utulum. 

(4) Assimilation of the second to the first 

This is much less common in vowel assimilation. In 
Arabic the suffixes -hu, -hum, -hunna become -hi, -him, -hinna 
after -i or -ay as 'alayhim " upon them ", but such assimila- 
tion is not always observed in dialect. In Assyrian a often ^ 
becomes e/i after i, e, ii, thus helat becomes helit " mistress ", 
imdru> imeru " ass ". 

64 (b) Dissimilation of vowels 

(1) a-a> i-a 

Thus in Arabic in plurals of the type qatldn which become 
qitlan, as ganna " garden ", pi. gindn. In Syrian dialect 
nahnd (vulg. for nahnu) becomes nihnd " we ", and in Nejd 
'asara becomes 'asira " ten " in compounds. 

(2) i-i > a-i 

In Arabic adjectives in -ly from nouns of type qatil, qatilat, 
change i of the stem to a, as habid " liver ", adj. kabadiy, but 


there is no change if the substantive has more than three 
consonants, as Yathrih (ancient name of Medina), adj. 
Yatknhiy. In Hebrew hen [bin) " son ", plur. binim becomes 

(3) 0-0 > i-o 

Aramaic v6/jlo<;> nimos (T.B. Gittin, 6). 

(4) a-a > a-i 

This appears in Mehri in nouns of type qdtal which become 
qdtil, as 'dsag> 'osig (accented d as o, cf. 43, a, a) "box 
thorn or lycium ". In Hebrew and Aramaic a-a-a becomes 

(5) Dissimilative change in quantity 

In Abyssinian we find a dissimilative change in vowel 
quantity in i-i for i-l, as in plural -nl followed by the sufl&x 
-ki, e.g. kehuraneB " thy kinsmen ", qatalkenl " thou hast 
slain me ", etc. Similarly l-ya becomes i-ya as in keburaneya 
for keburanlya, in each case as always e appearing for short i. 

(6) Dissimilation of vowels from semi-vowels 

With y Arabic changes i to a in plurals of type qatdld where 
the third radical is y, thus mandyd for mand'iyu from singular 
maniyd (cf. Wright, Ar. Gr., i, 222, C-D). 

In Aramaic uww becomes iww or eww in adjectives of type 
quttal with medial w, thus Bib. Aram, hiivwar, Syr. Jiewwar 
" white ", it being assumed that adjectives denoting colour 
are as 'ukdm " black ", yurdq " green ", etc. 



65 (I) Formation of new syllables by the use of prosthetic or 
inserted vowels 

A consonant may stand either at the beginning or at the 
end of a syllable. In consecutive speech, therefore, two 
consonants may be in contact, the one as the closure of a 
syllable, the other as the initial of the following syllable, 
the former vocalized by a preceding vowel, the latter by a 
vowel following. But there cannot be three consonants in 
contact, for the middle one would have no vowel, nor are two 
consonants in contact in inception, i.e. after pause, nor in 
closure before pause, for in inception the first would be 
vowelless, and in closure this would be the case with the 
second. Thus we have three conditions to consider — (i) 
a group of two consonants in inception, (ii) a group of 
three in continuous speech, and (iii) a group of two at pause. 

66 (i) A group of two consonants in inception 

This appears sometimes in the natural form of the stem, 
as *qtul imperative " slay ", in foreign words as *qUm = 
KXi^ia, or as the result of vowel elision as when na- becomes 
n- in nqatala. 

(a) Arabic 

Three methods of treatment are possible — (1) one of the 
consonants may be ehded, which is rare ; or (2) a prosthetic 
vowel may be added ; or (3) a vowel may be inserted. When 
a prosthetic vowel is used it is normally i-, but before u it 


becomes u- by assimilation. A prosthetic vowel is only a 
temporary expedient, and so it falls away when no longer 
needed, as, for example, when it follows after another word 
which ends in a vowel, for this vowel naturally vocalizes the 
first consonant of the word following. After silence the 
prosthetic vowel must be preceded by Hamza, for a syllable 
must commence with a consonant. Thus after pause nqatala 
becomes 'inqatala, so the article -I- (for la-), which may be 
vocalized by the final vowel of a preceding word, or, if that 
word end in a consonant, it becomes -il-, although it sometimes 
happens that the preceding word revives an obsolete vowel 
ending in this position as mudJi, which becomes mudhu 
because originally from mundhu, but after silence the article 
becomes 'aZ-, the a prefixed to vocalize the -1-, and the Hamza 
prefixed to it to provide the necessary consonant beginning 
of a syllable. A few nouns also occur which commence with 
two consonants and so add a prosthetic i- if they follow a 
consonant or silence, and in this latter case prefix also a Hamza ; 
such are bn- " son ", thn- (in ithndni) " two ", st- " anus ", 
sm- " name ", mra' (fem. imra'at) " man ", and thus 'ihn, etc. 

In loan words commencing with two consonants sometimes 
one is omitted, as Greek ^i^o? >*ksif>sif " sword ", but, as 
we have already remarked, this is a rare method. More often 
a prosthetic vowel is added, either i-, or less commonly 
d/u by assimilation, and this is preceded by a permanent 
Hamza which becomes part of the stem irrespective of 
whether the word follows pause or is in consecutive speech ; 
thus Greek TrXdroyv > 'afldtiin, KXi/xa > 'iqlim, crToX,oir > 'istul, 
a7r6yyof;> 'isfing (Syrian dialect), etc. 

In dialects which have been in contact with non-Semitic 
influences, such as those of 'Iraq, North Africa, and of the 
large towns with a cosmopolitan population, an initial group 
can often be pronounced by the insertion of a half -vowel 


(cf. Hebrew infra) or by the vocalization of a sonant, whicb is 
not really a Semitic expedient. Thus a Cairene can sound 
initial groups hr-, gr-, fr-, kr-, gl-, as in krumb " cabbage ", 
and this is more especially the case with persons who have 
been educated under European teachers. In Oman a 
prosthetic vowel is often omitted, more particularly when 
the resultant group contains a sonant, semi-vowel, or labial, 
thus swed for 'aswad " black ", hyad for 'ahyad " white ", 
etc. ; but when the resultant does not contain a consonant 
of these kinds a full vowel is usually inserted, as hit for 'aht- 
" sister ", although we find such forms as skur " thanks ", 
etc. In South Arabia and North Africa, especially in 
Morocco, an initial group is often vocalized by a half-vowel, 
as rbd for 'arha'a " four " (Morocco), rbot (id. Mehri), thnerie 
for 'ithndni " two " (Oman), thnu (id. Mehri), khol " black " 
(Morocco), etc. On the other hand, dialect sometimes prefers 
the use of a prosthetic (metathesis) as 'ahmdr for himdr 
" ass " (Syria), 'igbir for kabir " great " (Malta). 

(6) Abyssinian 

In Ethiopic the prosthetic vowel always appears with 
prefixed Hamza, in continuous speech as well as after pause, 
so in the verb conjugations we find 'astaqiala, etc. But an 
initial group is often vocalized by an inserted -e-, as in the 
imperative qetel for *qtid. In Amharic we find a tendency 
to employ a prosthetic vowel before r, thus ra's " head " 
becomes 'ers. 

(c) Hebrew 

The prosthetic vowel is employed only with prefixed K 
or h-, but, as in Abyssinian, the preference is for the inserted 
half-vowel, thus imperative qHol (Arabic uqtul) : reflexive 
itqattala as hitkqattel, and in Phoenician as ^JOpHi^. An 
initial group of two consonants appears in one word as pointed 


in the Tiberian text, viz. stayim " two ", but probably this 
should be pointed sittayim or sintayim, from masc. s^nayim 
{Aiahicithnani); Phoenician DiK'^5 (Cooke, NSI. 12, 3) suggests 
'estayim. Elsewhere hn>ben, sm>sem, st>seth, etc., the 
prosthetic vowel being avoided. Sometimes, however, 
Hebrew prefers the prosthetic as in '°'hdl " but " (Arabic bal), 
"^hattih " melon ", etc., and sometimes we find alternative 
forms, as in z^roa' or 'ezroa' "arm", bd/chd (Gen. xxxviii, 5) 
or ' '^bakh (Joshua xix, 29). In later Hebrew the Mishna 
shows aroa = i<*ltOD\^ (Nidda 1, 6), scutella = i^^DlpD^^ 
(Moed Katon, 3). 

(d) Aramaic 

For the most part the vocalization of an initial group of 
consonants follows the same course as in Hebrew. Prosthetic 
i' occurs in the verb forms ; in Bib. Aram, as in Hebrew with 
prefixed h-, but later forms prefer Hamza, hith-, 'eth-, etc. 
One word appears in Syriac with an initial group of two 
consonants, std " six ". Loan words with two consonant 
initials are usually vocalized by a prosthetic vowel, sometimes 
by an inserted half-vowel, thus Targ. ainlpa — ^'•£DX, 
Syriac ^e^'ta = 'aksen, ^di'6iov>k^santiyun, (TrdhLov>^estadin, 
a-Tv\o<i>'estund, ^i(f)tov>k^sifun, etc. 

(e) Assyrian 

Vocalizes the imperative by an inserted short vowel which 
assimilates to the stem vowel, thus purus, piqid, sabat, etc. 

67 (ii) Medial group of consonants 

(a) Group of three consonants 

Ordinarily this occurs when an initial group produced by 
vowel elision follows a consonant termination. But we have 
already explained the way in which an initial group is 
vocalized by a prosthetic vowel or inserted half-vowel or short, 
and this applies to an initial group of two when following a 


final vowel as well as after pause, only that, of course, 
there need be no prefixed Hamza or h-, etc., as the syllable 
containing the prosthetic vowel will have as its first consonant 
the final of the preceding word. In modern colloquial Arabic 
the disuse of the case endings and of the short final vowels 
of the verb persons tends to produce a group of three 
consonants when the next word has not a prosthetic vowel, 
and this produces a group of consonants which is vocalized 
by an inserted vowel. The vowel thus inserted is commonly 
* or e, but this becomes u by assimilation to a following u : 
thus shuft-e-ragll " I saw a man ", darabt-u-hum " I struck 
them " (Egyptian), etc. In North Africa the partial 
vocalization of the sonants allows a group of three or even 
more consonants to stand if one or more of them are sonants, 
as kull blad " every country " (Morocco), etc. 
(6) Medial group of two consonants 

Normally the first is the closure of one syllable and the 
second commences the next, but sometimes we find a vowel 
inserted so as to form a new syllable with the closure of the 
first, although there is no necessity that this should be done. 
Thus in Arabic, more especially if one of the two be a sonant, 
laryngal, or palatal, and chiefly in the dialects of 'Iraq and 
North Africa, e.g. meharab for mihrab " altar " (Maltese), 
sa'ab for sa'b " difficult " ('Iraq, Egypt), ba'ad for ba'd " after " 
('Iraq). In Abyssinian this is often done with a sonant or 
laryngal, as sa^na'eku for sama'ku,'alabo for ^albo (Tigre dialect), 
and so Amharic verb forms qatalatala for qataltala. In Aramaic 
we find such forms as dalfbd for dahbd " gold " with inserted 
half-vowel, and regularly an inserted half-vowel after t, d, 
preceded by da-, as in wedatjfwatha, etc. 

68 (ill) Final group of consonants 

The commonest instance of a final group of consonants is 
that produced by the fall of case endings after nouns of the 


types qail, qitl, qutl. In Arabic dialect such a group is vocalized 
by the insertion of a short vowel ^ or e (cf. Hebrew below), 
thus tiben for tibn- " straw " (Egypt), duher for duhr " dawn ", 
hamas for hams "five", Omani qador for qadr "power", 
with the inserted vowel modified by the influence of the 
emphatic consonants q-r, Mehri 'ahat for 'aht "sister", etc., 
but in North Africa a duplicate is retained if one member is 
a sonant, as qalb " heart " (Tunis). 

In Abyssinian a final double is vocalized by inserted vowel 
e, as kaleb for kalb " dog " ; but it must be noted that 
Abyssinian e is practically no more than a half -vowel. 

In Hebrew the forms qatl, qitl, qutl, insert e unless with 
medial or final laryngal, in which case the inserted vowel is 
a (cf. 54), thus malk becomes melekh, quds >qddes, sab' >seba\ 
etc. In roots med. gem. one of the duplicates is dropped 
if a final, thus 'anf>'app >'af " nose ". The original stem is 
of course restored when the group of two consonants is relieved 
by the addition of a vowel ending, as inalko, 'affo, etc. 

In Aramaic the course followed is as in Hebrew, but the 
vowel in the opened syllable becomes a half -vowel, as nafs > 
n^fes, nefsd, etc. 

In Assyrian a final group occurs in the construct of nouns 
of the type qatl, etc., and this is vocalized by inserting a short 
vowel which assimilates to the stem vowel, as kalm>kalam, 
tis' >tisit, etc. 

69 (iv) Inserted half-vowel 

We may refer here to the apparent insertion of a half- vowel 
or short vowel with certain consonants, although this is not 
properly an insertion but a " glide ", i.e. change of vowel 
timbre in the passage from the vowel to a following consonant 
which has an outlet removed from that of the vowel sound 
(cf. 55). Thus with laryngals following long i/e or u/o, 
as in Hebrew rildh for ruh " wind ", a change of timbre 


necessitated by the disposition of the vocal organs preparing 
to enunciate the laryngal following. So Arabic rudh " breath " 
(Egyptian dialect), gue' iorgu' (Algerian), and gila (Morocco). 

In Nejd and North Africa a similar glide frequently occurs 
after a labial and before i or a, as mHnni for minni (Nejd). 
Of course, there is no glide after a labial to introduce the 
vowel u, which has an outlet not alien from that used in 
enunciating the labial, just as no glide foUows vowel a 
preceding a laryngal. 

In Tripoli and Morocco an inserted u often occurs after h 
and before a. In Abyssinian this u or w regularly appears 
with q, h, k, and g, as in quasala " was wounded ", etc. 

70 (II) Haplology and Elision 

We have seen the formation of new syllabic groups by means 
of prefixed or inserted vowels ; we have now to consider change 
of syllabic constitution by the loss of vowels and syllables. 

(a) Of two consecutive open syllables each containing the 
same initial consonant followed by the same vowel the first 
is frequently omitted. Thus in Arabic ta- for tafa- with the 
personal prefix ta- in the imperfect of conjugations with 
preformative ta-, as taJcallamu " speak " for tatakallamu 
(Qur'an, 11, 107) ; so 'i'i becomes 'i in minassama 'ila for 
minassama'i 'ila " from heaven to (earth) " (Qiir'an, 32, 4). 
In conj. X of hollow verbs with initial ^ or ^ we find elision of 
formative ta-, and this is sometimes followed by assimilation 
of t, thus for istata'a we may have ista'a or ista'a ; this change 
is particularly common in Egypt and Damascus. 

Amharic shows beioihebem'aqqabet for 'aqqahe het "master 
of the house " (Praetorius, Amhar. Sprach., 159a, 333a). 

Assyrian has se for sese in salaseru for salas eseru " thirteen " 
(and hamisserit for hamis eserit " fifteen "), and na for nana 
in inage for ina nage, etc. 


71 (6) Of two consecutive open syllables eacli containing the 
same initial consonant followed by a short vowel, the first 
is often omitted. Thus Arabic ya- for yuya- in yabhisun 
(assimilated as yibbisun, of. section 63) for yuyabbisun " they 
dry " in the dialect of Hadramaut. So n- for n-n- in the 
imperfect terminations -ina (2nd fern, sing.), -dni (dual), 
-una (2nd and 3rd plur. masc.) before suffixed -na, -ni, so that 
we get -Tnd for -mand, etc. 

In Abyssinian a similar change takes place in the 2nd fern, 
plur. -kennd before suffixed -n%, -nd, thus -kdni for -kenndni, 
-kdnd for -kenndnd, in which case the first nn falls away with 
the preceding vowel. 

Hebrew omits 6*- " in " before h-, p-, m-, in beth for b^beth 
(Gen. xxiv, 23 ; xxxviii, 11, etc.), pathah for b^pathah " at 
the threshold " (Gen. xviii, 1), and Md'on for b^Md'on " at 
Maon " (1 Sam. ii, 29), but it may be that these are instances 
of the accusative denoting the place where. So the omission 
of m* in ma en for m^maen in Exod. vii, 27. 

Assyrian shows this elision generally in conj. D (Arabic 
conj. ii) of verbs med. gem. as ipassu for ipassisu, rmidtallu 
for mudtallilu. 

72 (c) Closely akin to the above is the omission of the initial 
or closure of a closed syllable when these are homogeneous, 
as in Arabic verbs with initial y- after personal preformative 
y-, thus yabasu for yaybasu, yd'su for yay'asu, etc. So sabin 
" bridegroom", a loan word from theSyriac susbind (Hebrew 

73 {d) Less common is the elision of one of two homogeneous 
consonants separated by a third, as in Arabic h-n for n-Ji-n in 
ehna for nahna ('Iraq), ahna (Tunis, Malta), ihna (Egypt), 
the 1st pers. pron. in the plur., etc. So Abyssinian h-t for 
t-h-t in hatte, hante for tahta " under " ; Hebrew sarsd for 
Sars^rd " chain ". 


74 (e) Vowel elision regularly occurs in verbs med. gem. 
in Arabic, Hebrew, and Aramaic, as Arabic farra for 
farara, Hebrew sab{h) for sahah, etc. This elision does not 
always take place if the vowels are different. 

In Arabic the elision of a final short vowel takes place in 
pause, but in dialect and in Hebrew and Aramaic the final 
short vowel is generally obsolete, or else has become long. 
Other instances occur of occasional vowel elision as Assyrian 
anilani for ana Hani, zikru for zikaru, etc. 

75 (HI) Metathesis 

Metathesis does not directly alter the syllabic constitution 
of a word but only its syllabic form, but indirectly it often 
leads to elision, etc., or is the result of such elision. Thus 
qatala with reflexive preformative ta- becomes in Arabic 
not TAqatala but qaTAtala and hence -qtatala with prosthetic 
i-iqtatala. Metathesis occurs most easily between two 
consonants in contact when one or both are sibilant, dental, 
or sonant. In Arabic, as we have just noted, such metathesis 
takes place in the reflexive of the primary conjugation ; it 
does not take place in the reflexive of the intensitive taqattala, 
but in Hebrew it does so if the first radical is a sibilant, as in 
histabbel for hitsabbel, although we find hithsotatnd in Jer. xlix, 
3, without metathesis so as to avoid the collocation of three 
t sounds ; and this metathesis appears in Arabic as a kind 
of survival in Qur'an reading and in some dialects with this 
conjugation (cf. 22). So in Aramaic, where each conjugation 
or theme has its own reflexive form, metathesis taking place 
with a first radical sibilant. In the Shaf'el theme (older 
causative, cf. 136) the reflexive Eshtaf'al shows similar 
metathesis, and so we have 'estaqtal for 'etsaqtal, etc. In 
Assyrian metathesis takes place between reflexive t- and the 
first radical whatever its character, thus aparas, aptaras; 
uparras, uptarras, etc. 


Sometimes we find metathesis in nouns of type qawtal, 
qaytal, wlien the medial is sonant or laryngal : thus Mehri 
generally as in haybob or habyob for Arabic habbab. 

Other sporadic instances of metathesis occur, chiefly when 
one consonant is sonant or semi-vowel, less commonly when 
one is labial, and comparatively rarely in other cases. Thus : — 


rs >sr Arabic f>,j^>'^ (Samaritan DI^H) = Hebrew j^lH 

" grapes ". 
rt >tr Hebrew tJ^£tD*1 " grow fresh again " (Job xxxiii, 25) = 

Arabic (loan word) i^y^ " recover ". 
sr>rs Hebrew n^p " reap ", pp " tear oS ". 

zr>rz Hebrew ^15, Arabic j^^ = Hebrew Hi, Arabic 

jj>- " cut off ". 
mr>rm Assyrian zumru = Hebrew D*1T " shower ". 

y yy 

ql>lq Arabic Jls = Mehri letog " kill " (cf. sect. 15). 

lm>ml Arabic ^^j<:>- = Hebrew 7^n " be gentle ". 

'1>V Hebrew n^^D = Arabic ZUT (Barth, Nominalbild. 

276, n. 2). ^ ^ .^ 

ks>sk 'AXe^avBpo^, Syriac .mn^irnnVj Ara.birjJl!\ *« Jl |. 

st>ts Hebrew n^^ (Isa. xli, 17) = DTl^ (Jer. xviii, 14) 
" dry up ''. 



(A) The Absolute Form 

76 The absolute personal pronoun is a pronoun in tlie 
nominative case (in Assyrian and in sub-Semitic it is found 
also in the oblique cases) and emphatic. The non-emphatic 
personal subject is expressed by the personal prefixes or 
suffixes attached to the verb stem, the absolute is added only 
when the subject is to be emphasized. In the 1st and 2nd 
persons the demonstrative 'aw- is commonly prefixed to the 
pronoun when thus used in the absolute form. This corre- 
sponds with the ancient Egyptian demonstrative particle 
'in- attached to an emphatic nominative pronoun or noun 
(Erman, Mgypt. Gram. (3), 494), although in Egyptian the 
use of this emphatic form generally leads to the disuse of the 
personal suffix to the verb. The emphatic personal pronoun 
in ancient Egyptian ('mA;, ntk, etc.), which corresponds with the 
Semitic absolute, is still rare in the Pyramid texts, but was 
fully developed in later periods, and appears in Coptic as 
^.ItOK, UTOK, etc. 

77 (1) First Person Singular Absolute 

(a) Form 'ana, etc., demonstrative 'an- with pronominal 'a. 
Cf. pronominal 'a- in the 1st pers. sing, of the verb {West 
Semitic imperfect). 

(i) In Arabic this appears as b I , j 1 , j I , (in pause) <> I , 

c $ 

jj I , but the final vowel is always short except in pause and 


sometimes in dialect, as a7id (Morocco, Tlemsen), and (Algeria, 
*Ayn Madi). In dialect we find phonetic modifications due to 
the Imale (cf. 43), thus awa (Algeria above), dne (TripoH), 
ene (Oman), In the dialect of Hadramaut we find masc. 
ana, fern, ani by analogy with the 2nd person. Occasionally 

U I becomes IJJii , and thus Mehri ho, or hu. In Minsean 

inscriptions we get "JX, the final -i being, no doubt, due to the 
influence of the pronominal suffix (cf. 82), and so ani in the 
dialect of 'Iraq and in Spanish Arabic (cf. Hebrew below). 

(ii) Abyssinian 'ana (Ge'ez), Tigriua 'ane (cf. below), 
Amharic 'ennih or 'ene. 

(iii) Hebrew ''i^^ (in pause ''^^), the final -i due to the 
pronominal suffix (cf. Minaean and Samaritan below). 

(iv) Aramaic N^X (Cooke. NSI. 63, 1 ; 77), Bib. Aram. X5^? ; 
Samaritan HJ^ (Gen. vi, 17, etc.) and rarer NJX, also ^J^{ 
with final -i. Onq. "'3^5 (Gen. vi, 17, etc.), also ^^iS ; 

Mandsean J^ii< ; Syriac ^jf . In Targ. Jer. we find i<i 
without the hamza and half-vowel. 

(v) With these forms we may compare sub-Semitic (East 
African group of Hamitic) ane (Bishari), ani (Galla), ace. 
ana (id.), anu (Saho), an (Bilin, Dambea), ana, am, an (Somali). 

(6) Form anak- 

Barth {Pronominalh. 2c) regards this suffixed -Ic as a demon- 
strative, but it appears elsewhere as a personal element. 
Thus in Berber (North Hamitic) the theme of the 1st person 
is n-k, of which n- is the demonstrative (Semitic 'an-) and -k 
is the pronoun, thus nek (dialects of Ahaggar, Awelimidden, 
Ghat), nee, nes (Zenatia), nis (Siwah), etc. So ancient 
Egyptian 'ink, Coptic ^rtOK. Cf. verb person in Mehri -ik, 
-ek, Soqotra -k, Bilin -ko, koti, Galla -ko. So Abyssinian -ku, 
where the other Semitic languages have -tu {-ti, -t). With 


these may be compared the Sumerian 1st sing. pers. pron. 
KU oiGU (bilingual frag. 5 r., 20, No. 4). 

Thus Moabite "[3J< (Mesa stele, 1, 21, 22, etc.), Phoenician 
*]iX (Cooke, NSI. 3, 1, 2, etc.), anek, anech (Plautus, Poenul. 
5, 1, 8 ; 2, 35), Hebrew anuld in Amarna letters, ""DiX in 
the O.T., and Phoenician ^^1^ (Cooke, NSI. 3, 1, 2), Aramaic 
*]3X (id. 61, 1), ^^3iS (id. 62, 19), Assyrian anahu, enclitic 
-aku. The final -i in Hebrew, etc., is due to the analogy of 
the pronominal suffix. 

(c) Forms compounded with demonstrative ya 

These forms are quite secondary. They appear as and-ya 
(Morocco), ana-ya (Algeria, 'AynMadi), dna-ia (id. U. Brahim), 
'awe (= 'ana-ya, Tigrifia), 'eiie, 'enei (Amharic). Prefixed 
ya- appears in Maltese yan, ydna, yin, yina. On demonstrative 
ya cf. 91 below. 

78 (2) First Person Plural Absolute 

Theme n-h-n in which the first n- represents the demon- 
strative n, 'an, as in Hebrew ^J^:^{, Phoenician Vr\y^, etc. 
No doubt the second -n is the plural termination, and this 
suggests that h may be the personal element as in the Berber 
theme n-k-n plural of n-k. 

(a) In Arabic and Abyssinian the prefixed 'a- is lost, but 
it must be remembered that the demonstrative is n{a)-, so 

that this 'a- is only a phonetic addition. Thus Arabic ^l^^ 

y.^ (Oman), L°^ (Tunis, Syria), ^^^ (Syria), ^^^ (id.), 

and Ge'ez nehna, Tigrina nelind, Tigre nahnd. With 
dissimilation as laJma, wahna (Arabic of Datina). But we 
also find forms in which the demonstrative n- is lost, as 
honu, hene, hne (Oman), hona (Morocco), honnd, hannd, 
kmnd (Nejd), hna, hne (Tripoli), ahna (Tunis, Malta), ehnd 


('Iraq, Hauran), ihna ('Iraq, Syria, Egypt). Only occasionally 
do we find the final -n omitted in Arabic dialect as nihd 
(Hadramaut), nahd, 7ihd, nah (Mehri). 

(b) Hebrew shows ^3^^^{, but also ^liHi (six times in the 

O.T., Exod. xvi, 7, 8, etc.) : so Aramaic ^5i^i (Onq. and 

Targ. Jer.). Bib. Aram. X3^3^{, once n;in;X (Ezra iv, 16) ; 

in the papyri ^i^I^^{ Hebrew and Aramaic show a (later) 
tendency to omit the H, as ^it< (Mishna), ^^^^ (Jer. xlii, 6), 

li^^ (T.B.), p^ (T.J.), i^ix (Mand.), j^iii^ (id.), px, pn:i&< 

(Samaritan) : Syriac -J-k». 

(c) Assyrian aninujl, ninujl. 

{d) Parallels. Somali anuna, Galla nu, Saho nanu, Dambea 
anen, Kafa no, Hamara yinne, yin, Bilin yin, Bishari hene, 
henen, Hausa namu. 

79 (3) Second Person Absolute 

(a) Basic forms 

Sing. masc. Sing. fern. Plur. masc. Plur. fern, 
'anto 'anti 'antum 'antin 

Demonstrative 'an- as in the 1st sing. : personal element 
-t. Fem. in -i as in verb persons, e.g. imperative sing. masc. 

Ax3 1 fem. /JUi I ; perfect sing. masc. olxd, fem. cJLv3, 

etc., and so in some noun forms (cf. 116). Plural in -m, -n, 
the fem. plur. retaining %, the masc. plur. with w. 

(6) Arabic 

Sing. masc. C.<J I , and so in dialect enta, inte ('Iraq), 

ente, ent (Arabia), ent, enteh (Hauran), ent, int, ente (Syria), 
enta, inta, inte (Egypt), nte (Oman), ente (Spanish Arabic), 

ent, ente (Ha^ram.), rarer <2X-^, and thence Mehri (common 


e (■ 

gender) Ml, Ifit (for Jiant). Sing, fern, CJ I and dialect enti, 
inti (Central Arabia), nti (Oman), inti ('Iraq), enti (Syria), 

inti (Sjnria, Egypt). Plur. masc. lx> I ( 1.L* I in poetry and 

^ }°i \ \ 

wasl). Plur. fern. ^\) I , the vowel u is due to tlie influence 

of the masc. u, i is preserved in dialect (cf. below). 

In dialect generally the fern. plur. is obsolete, but we find 
some forms which retain the i vowel, as inten ('Iraq), nten 
(Oman), enten (Datina). Occasionally the fern. sing, is 
obsolete, as is the case in Hadramaut and Spanish Arabic. 
Contrariwise the fem. sing, has sometimes replaced the masc, 
as in enti (common gender, Tunis), ntt-n in Tlemsen, enti-n, 
entt-na in Morocco, and Maltese inti, int. In Tunis and 
Tlemsen these forms have displaced the masc, in Morocco 
there are separate forms for the two genders, but if the pronoun 
is reinforced by the addition of the demonstrative -n, -na, 
the fem. enti is used for both genders. The added demon- 
strative -n, -na, appears in Tlemsen ntin, plur. ntuman, and 
in Morocco sing, entin, enttna, plur. ntuna. The demonstrative 
ya is found in Algerian (U. Brahim and 'Ayn Madi) and in 
Moroccan, thus : — 

U. Brahim. 'Ayn Madi. Morocco. 
Sing, masc. . entdya ntdya ntdya 

fem. . entiya ntiyd — 

In the masc. plur. added -a (u) appears in poetry and wasl, 
and so in dialect in Tunis, Tlemsen, Algeria, and Morocco. 

The masc. plur. appears without final -m (as j\> I ) in Syria, 

'Iraq, Hauran, Oman, Egypt, and Morocco (in ntu-na). The 
suffixed pronoun -k occurs in Mardin sing, antek and Maltese 
plur. intkom. In Mehri the plur. forms are masc tern, fem. ten. 


(c) Abyssinian 

Sing. masc. Sing. fern. Plur. masc. Plur. fern. 
Ge'ez . 'anta 'anti 'antemmu 'anten 

Tigre . 'enta 'enti 'entHm 'enten 

Amharic . 'anta 'and Cantu — 

'ant ^ 'elldnt, 'enndnt 

In Amharic -ti becomes c (cf. -ki in 38). In the plur. -m 
falls away with the lengthening of the preceding vowel 

(cf. Arabic ^L* I ) ; the fern. plur. of this form is obsolete in 

Amharic. There is also a plural (common gender) 'elldnt = 
'ella + 'ante (cf. demonstrative in 91 below), which by 
assimilation sometimes becomes 'enndnt. The 2nd personal 
pronoun is obsolete in Tigrina, and is replaced by the stem 

nese- (i.e. ness- = -y*A> " self ") with the pronominal 

suffixes, thus sing. masc. nesekd, fem. nesekz, plur. masc. 
nesekum, fem. neseken (dialect Aksum, masc. nessdtku, fem. 

(d) Hebrew 

In all cases -nt- becomes -tt-. Masc. sing. ^^^{, Pt^, in pause 
^^^{, T]P\i^. ri{< occurs only in Kethib (1 Sam. xxiv, 19 ; 

T T T - T - *' 

Ps. vi, 4 ; Eccles. vii, 22 ; Job i, 10 ; Neh. ix, 6), and flX in 
Num. xi, 15 ; Deut. v, 24 ; Ezek. xxviii, 14. Fem. sing. 
p\if:, in pause Hi^, TlX in Kethib (Judges xvii, 2 ; 1 Kings 
xiv, 2), possibly a mark of northern dialect. Plur. masc. 
DDX, where -um becomes -em. Plur. fem. |Tn{< only in Ezek. 
xxxiv, 31, more commonly |nx. Also H^iriX in Gen. xxxi, 6 ; 
Ezek. xiii, 11, 20; xxxiv, 17. Phoenician sing. masc. HX . 

(e) Aramaic 

Sing. masc. r\^ (Cooke, NSI.64, 5), so Hi^ in Onq. (Gen. xlix, 

8, etc.), Samaritan, Syriac I], and neo-Syriac of Tur hat, 


Ma'lula 'ate, hate ; also P\^'^ (papyri, Bib. Aram., Targ. Jer., 

and Syriac orthography Ai]) ; HnJX (Bib. Aram., Kethib), 
r\r\ii (Samaritan), ^^{J^< (Mandoean). The sing, fem, is rare ; 
we find TliX in the papyri, '•fli^ in Samaritan (Gen. xxiv, 60), 
and Syriac ^L], wiAj]. It does not occur in Bib. Aram., and 
in Onq. and Mandsean resembles the masc. Plur. masc. C^i^^ 
(papyri), pJ^JK (Bib. Aram., and so Syriac .O^AjI), ]^^\i< (Onq., 
Gen. xlv, 8, etc.), pnx (T.B.), and so Samaritan jIHS, Syriac 
.oAj] with silent n, Mandaean jin^iX. Plur. fem. Vr\is (Onq., 
Samaritan, T.B.), and so Syriac —lAj]. 

(/) Assyrian 

Sing. masc. atta, fem. atti, plur. masc. attunu, fem. attina. 

(g) Sub-Semitic parallels 

Ancient Egyptian sing. masc. nth, fem. ntt, plur. nttn. 
Dambea sing, ent, plur. enten ; Bilin sing, enti, inti, plur. 
entin, intm. 

80 (4) Third Person 

(a) Arabic 

(i) Masc. sing. jS*, jSi , huwa (Egypt, Syria, Tunis, Mecca), 

huwe (Oman, Palestine), huwd (Tripoli), Tiouwa (Algeria, 
U. Brahim), heuwa (Tlemsen) ; hence Mta (Tunis, 'Iraq, 
Morocco, Malta), hu'a ('Iraq), hue (Oman, Syria, Spanish 
Arabic), and finally hii (Egypt, Syria, Central Arabia, 
Hadramaut, Malta, Spanish Arabic, and Sab. 111). 

(ii) Fem. sing. ^A hiya (Egypt, Tripoli, Tunis, 'Ayn Madi 

of Algeria, Tlemsen), or hiye (Oman, Syria, Palestine), or 
hSya (Algeria, U. Brahim). Corresponding to masc. hua, etc., 
we have fem. hia (Morocco, Malta, Spanish Arabic, Tunis) 
and hie (Syria, 'Iraq), and hi'e ('Iraq). Corresponding to 


masc. hu is fern. M (Syria, Malta, Central Arabia, Hadramaut, 
Spanish Arabic). Thus as general types : — 
masc. huwa (e) fern. Mya (e) Oman, Syria, Palestine, 

Egypt, North Africa. 
hiVa hfa 'Iraq. 

hua Ma Oman, 'Iraq, Syria, Malta, 

Morocco, Tunis, Spanish 
^^ M Arabia, Syria, Malta, Egypt, 

Spanish Arabic. 

(iii) Masc. plur. ^, ^. In dialect usually hum; 

in Hadramaut as liom. ^ in Egypt, Tripoli, 'Iraq, and 

Mma or huma in Tunis, Algeria, Morocco, and Maltese huma, 
usually with alternative use of hum. In Syria a form with 
hemme with its vowel modified by the fem., or the fem. henne 
appears in use for both genders in addition to the ordinary 

(iv) Fem. plur. Originally hin or hinna, and thus hm in 
the dialects of Oman and Syria, and hinna, hinne in 'Iraq. 
In classical Arabic the vowel is assimilated to that of the 

masc. as A.*. 

(v) Special forms with suffixed -n. Mosul sing. masc. 
hi-nu ; Tlemsen masc. plur. huma-n ; Syrian fem. plur. 
henne-n used as of common gender. 

(vi) Forms with suffixed -t. 

Sing. masc. Sing. fem. Plur. masc. Plur. fem. 
Spanish Arabic hu-et hi-et hum-et hunnat 

Palestine . hu-tu hi-te — 

Sabsean . — — H/Dn 

Cf . use of demonstrative -t in Abyssinian. 


(vii) Forms in s. Where Sabsean shows in, H the older 
Minsean has *ID, D, with which must be compared the Assyrian 
Iw, si, etc. (cf. below). These s forms survive in Mehri and 
Soqotra, but are specialized to denote the fem., thus sing. masc. 
he, hi, sing. fem. se, si, plur. masc. hem, plur. fem. sen. In 
Soqotra, also, we find masc. sing, yhe, a form which Earth 
{Pronomin. 76) connects with Aramaic irT'N, etc. (below). 

(6) Abyssinian 

(i) Ge'ez, Sing. masc. we'etu, sing. fem. ye'eti, masc. plur. 
we'etomu, plur. fem. ye^eton. At the base of these forms we 
have hu'a, etc., with suffixed demonstrative -tu (cf. emphatic 
-tu in Galla, and use of demonstrative -t in Assyrian, etc., as 
well as in Sabsean plur. ; this becomes hu'etu and thence 
we^etu by assimilation of h to the following vowel, and forms 
its plur. in -m (cf. 2nd person). Similarly htati>hi'eti> 
ye'eti with plur. in -n. There is another plur. form with masc. 
'emuntu, fem. 'eniantu, in which we see the element em- with 
gender endings -u, -a, and demonstrative -tu not inflected. 

(ii) Tigre. Masc. sing, hotil, fem. hetd, plur. masc. hetom, 
fem. hetdn. Pronominal ho (= hu), he {= hi) with added 
demonstrative t which takes the gender terminations (fem. -a, 
not -i as in Ge'ez). To this are added the plur. formatives 
-m, -n. 

(iii) Tigrina uses nese- as in the 2nd person (q.v.), giving 
sing. masc. wesM, fem. nesa, plur. masc. nesom, fem. nesen. 

(iv) Amharic also has lost the 3rd personal pronoun proper, 
and uses the stem 'ers- {ra's " head ") with the pronominal 
suffixes, thus sing, masc. 'ersu, fem. 'ersowa, plur. (common) 
'ersaco or ^ersacaw. 

(c) Hebrew 

Sing. masc. ^'^T\, Moabite XH in Mesa stele lines 6, 27. 
Phoenician XH (C.I.S. i, 9, etc.), Plautus hu. Sing. fem. 


^^*^. Plur. masc. DH, "Dn, H^ri. Phoenician HDH (with 
demonstrative -t, cf. Saba^an, etc.). Plur. fem. \7^ "jn 113(1. 

{d) Aramaic 
Masc. sing. J^IH in Bib. Aram., Targ. Jer., and Samaritan, 

^n in Mandsean, and so Syriac ooi ; liuwe in the neo-Syriac of 
Tur. Fem. sing, ^s"•^ in Bib. Aram., Onq., Samaritan, yn 
in Mandsean, and Syriac ^oi ; hiya in the neo-Syriac of Tur. 
Plur. masc. DH in the papyri, and IDH in the papyri and 
Bib. Aram. 

To these simpler forms we must add a series of plurals 
with final -n, thus masc. jlSn in Bib. Aram, (accusative only), 

"t^TT^ in Mandsean, and Syriac .OJOI ; neo-Syriac of Tur 

honne. Fem. pTl in Mandsean and Syriac _»jai. Mandsean 
also as |yyn. 

A third group shows prefixed -''J^ in T.B. sing. masc. ^7V^, 
fem. 'n\^, plur. masc. ^T^y^, fem. ''^y^<. With these compare 
Bib. Aram. plur. masc. J'l^X, fem. pj;^, and Syriac plur. masc. 

.Qj] , fem. ^j . 

T.B. also shows a series with prefixed -0, which is in use as 
a substantive verb, thus sing. masc. irfi, fem. *n"*X plur. masc. 

inyj, fem. -nyi 

(e) Assyrian 

Sing. masc. su, fem. si, plur. sun{u), fem. Una. To these 
may be added the demonstrative -atu as suatu, etc. 

With this sjs form (cf. Minsean, Mehri, above) must be 
compared the ancient Egyptian older absolute pronoun sing, 
masc. sw, fem. si, plur. sn, and the later form sing. fem. nts, 
plur. ntsn. Cf, also the Berber suffixes sing. masc. -s, -is, 
-es, plur. masc. -sen, fem. -sent. Also Galla sing. masc. isa, 
fem. isi, plur. isan ; and Saho sing. masc. ussuk, fem. issi, 
plur. ussun. 


81 (B) The Suffixed Pronoun 

The suffixed pronoun may be attached to a verb or pre- 
position as denoting the objective governed by the word 
to which it is attached, or to a noun as denoting the genitive 
on which the noun depends. In either case the suffix is 
attached to the full word, i.e. root, formative, and termination, 
and so a final vowel or consonant which in ordinary speech 
falls away because it is final is restored before the pronominal 
suffix. In the case of a suffixed pronoun attached to a noun, 
the noun is thereby defined just as it would be by a following 
noun in the genitive (cf. 132 below). 

82 (1) First Person Singular 

(a) Attached to nouns and prepositions 

(i) Arabic. After a long vowel or diphthong -ya (but also 
-u-ya as -uy, and -aw-ay as -ay), in poetry and wasl after a 
consonant or short vowel (which is then dropped) -iya, 
otherwise after consonant or short vowel (which is then 

dropped) -1 ; but in exclamation -1 becomes -i, as ^j^^ 

"0 my people" (Qur'an,5,21); and in pause <» — or li — . Now, 

-% undoubtedly stands for -iy, and so we may regard -iya as 
in all probability the original form. In dialect we find yi, 
ye for ya after long vowels (Egypt, Syria, Palestine, 'Iraq, 
Oman), and sometimes -y (Hadramaut, Egypt as heladiy 
" my countrymen "). Oman also -yine (with demonstrative 

-ne). With the nouns t^ I, ^11, ^>-, Ji, the classical usage is 

y^ I , etc., before suffixes other than the 1st person sing., 

and that suffix appears simply as -i, the final vowel of the 
substantive being elided, thus 'ahl, etc. But in dialect we 
find 'abuya ('Iraq, Petrsea, Egypt), 'abuy (Hadramaut), 
huyd (Morocco), ahuyi (Syria, Oman). 


(ii) Abyssinian. In Ge'ez -ya, Tigre -ye, Tigrifia -y, -tje, 
-ay (the last only after a consonant), the -y causing the 
lengthening of the preceding vowel. Amharic -ya, -ye. 

(iii) Hebrew. After a consonant -i, after a vowel or 
diphthong -y. So Phoenician \ Plur. or dual -ay, -ey, makes 
-ayy, etc. 

(iv) Aramaic -i, -ay, -y. Mandsean -ya. 

(v) Assyrian -i, -ya, -a. 

(vi) Sub-Semitic parallels. Ancient Egyptian -w^i, the older 
form of the absolute employed as a suffix. In the Berber or 
Libyan languages -i is employed as objective suffix to the verb. 
Somali dative -i, possessive -ai. Dambea -yi, -ye, and so 
Bilin prefixed yi-. 

(vii) Noun suffix with demonstrative t (cf. sect. 95) appears 
sometimes in Arabic Avith 'ah, and 'umm, thus 'abati for 'abi 
in Qur'an, 12, 4. Usually it has the form 'abati, 'ummati, but 
'abatd, 'abatd, etc., are also found. In Mehri we have the 
form -tey. 

83 (b) Attached to verbs 

AVhen the suffix of the 1st person is used to express the 
accusative and is attached to the verb it has the inserted 
consonant -n-. This is a purely phonetic addition called by 

^ "* > > 
the Arabic grammarians "the supporting w" (J^LJI j ai*) 

^"^ '■'>>■ ^ '' 

or "the protecting w" (^^l^^JI Jy), and so Brockebiann 

refers to the n as used to avoid the hiatus (Brockelmann, 
Sem. SpracL, Leipzig, 1906, p. 100) ; but Wright {Co7np. 
Gram., p. 96) seems disposed to regard it as in some way 
denoting the accusative. 

(a) Arabic 

In Arabic the form of this suffix is normally -ni, or -niya, 
the latter form in poetry and in the wasl, or -myah with the 


" h of silence " in pause. Sometimes in exclamation -n% is 
shortened to -ni, as ittaquni " fear me ! " 

This -ni appears generally in dialect, but in Ha(Jramaut 
we find two forms, masc. -nd, fern, -m, on the analogy of the 
2nd person sing. 

Properly restricted to use as an accusative after verbs, it 
is, in some dialects, extended to other words. Thus in Tunis 
it is used with ba'd, 'ad, Id, and hd, as Id-ni " it is not I "; 
in 'Iraq it occurs with ba'ad and with ism " name ", which 
then becomes semnl " my name " ; so sma^ii (Tripoli), semni 

In South Arabia (Oman, Hadramaut) we sometimes find 
-in-ni with participles, the inserted -in- being perhaps a form 
of the demonstrative n{a) (cf. 93 and Hebrew below). 

This -nl form does not appear in Mehri, where -tey is used 
with nouns and verbs. 

(b) Abyssinian 

In Tigrina and Amharic the verb takes the same suffix 
as the noun. In Ge'ez we find -nl with verb persons ending in 
a vowel ; in the 2nd plur. fern, -ken before this suffix becomes 
-kennd (cf. Arabic -tunna), and then by haplology -kenndnl 
becomes -kdnl. By dissimilation -ml appears as -e-ni, thus 
giving qatalkeni with the sufiix following the 2nd fem. sing, 
and similarly with the 2nd fem. sing, imperative. In the 
Tigre dialect the suffix appears as -ne. 

In Hebrew the suffix is -m : in the 2nd masc. plur. original 
-turn normally appears as -tem, but with suffixed -ni, ^-tumuni 
becomes -tuni, an instance of haplology similar to that noted 
in Abyssinian. In the imperfect, imperative, and infinitive 
-ni is used whenever the stem ends in a vowel, but when it 
ends in a consonant we find -e- inserted, thus qotUni, etc. 
This inserted -e- may be, as Barth suggests, due to the analogy 
of final -y verbs (Barth, Pronominalbildung, § 126) ; or 


it may be the helping vowel Ije raised to e by the accent. In 
the imperfect, chiefly in pause, we also find the suffix as 
-an-nl or -en-ni where the inserted -an-, -en-, may be (i) a form 
of demonstrative na- where -na-ni- becomes -anni as lala 
becomes alia in 'allddhi ; or (ii) it may be " the old termination 
en of the energetic mood, but without its original force " 
(Brockelmann, Sem. Sprach., p. 160). 

In Aramaic the same suffix appears as -ni, in Bib. Aram, 
also as -e-ni after a consonant in the imperfect, and as 
-inna-ni or -nna-ni (cf. Dan. vii, 16), showing the same inserted 
-n- as we have seen in Hebrew. So Samaritan ""i- (Gen. iv, 
11, etc.) and Onq. -m, and -nam as in Gen. xl, 14. In 
Mandsean we find |^^ after consonants, | after vowels. 
Syriac has »^ (^aJ) after a vowel, including the restored 
vowel of the perfect 3rd masc. sing., etc. In the imperfect, etc., 
-ay (Hebrew -e) is inserted after a consonant ending. We 
also find the form -{e)nan{i) as neqtuleyndn (cf. Hebrew). 

Ass}Tian shows the suffix -ni or -ani as well as the forms 
-anni, -inni, as in userabanni, etc. 

84 (2) First Person Plural 

The ordinary form is -na, and the suffix thus appears in 
Arabic, Tigre, Tigrina, Bib. Aram., Onq., Targ. Jerus., and 

Arabic dialect sometimes shows the Imale, thus -ne (Oman), 
-ne, -nl (Dofar), -n (Mehri). In Ge'ez -no, becomes -na. In 
Amharic with nouns it is -n, or -en after a consonant, with 
verbs -na, -ne. In S3n:iac we find -n, -an, -nan, and in T.B. 
-in, Mandsean |-, Vi^-, jt^'X- ; Samaritan -n (Gen. xx, 60, 
etc.). Assyrian -na in letters. 

Hebrew assimilates the final vowel to the plural termination 
-u, thus -nu, an assimilation which has already taken place 
in the Amarna letters. We find -nu also in Assyrian ; but 



Assyrian shows -m in the possessive and accusatives -nidti, 
-rilti (with demonstrative t), and dative -nidsi, -nisi. 

Forms also occur with added demonstrative -n. Thus in the 
southern dialects of Arabic -in-na attached to participles, as 
hum mahalifin-inna " they are allied to us " (Landberg, 
Etudes, ii, 723), Tigre -anna, Hebrew -en-nil as yimsaennu 
'Hos. xii, 5, rare form), Syriac -nan, Targums -inn^nd, -nand, 
Mandsean, with the verb, |^^3-, jN^-, Samaritan -nan 
(Exod. xxxii, 1, etc.), Assyrian -anniti, dative -annasu, 

Sub-Semitic parallels occur in Galla -ke-na (possess), 
Dambea ana-, yin- as prefixes, BiUn yina-, verbal sufiix -nu. 

85 (3) Second Person Suffixed Pronoun 

(a) Arabic 

Sing. masc. Sing. fern. PI. masc. PI. fem. 



•'r ''r ''r 

Oman, Hadr. . 

'Cth/y "SfCy "fC 

-SI, -s 




-ak, -ek, -k 

{-ki), -ic 



Mehri . . . 




Syria, Palestine 

-ak, -ek 

-ik, -ki 

-kum, -kon 

-kin, -kon 

Palestine fellah. 

-cim, -cu 


Egypt . . . 

-ak, -k 

-ik, -ki 

-kum, -ku 

Tripoli . . . 

-ak, -ek, -ik 



Tunis, Morocco 

-ak, -ek 


Spanish Arabic 

-ak, -ek 


Maltese . 

-ak, -ik, -k 


(i) Sing. masc. becomes -k in pause and very frequently in 
dialect, thus necessitating the insertion of a vowel after a final 
consonant. This vowel is usually -a- by assimilation to the 
(lost) final of the suffix, but we also find -e and -i. (ii) Sing. 


fern, -hi or -hi is sometimes retained, or else the vowel falls 
with insertion of -i- after a stem ending in a consonant. 
Palatalization of the -k occurs under the influence of the -i- 
(cf. 38) in South Arabia as -s, in 'Iraq and with the fellahin 
of Palestine as -c. (iii) Plur. masc. is properly -hum, but we 
also find -hu (cf. absolute, 79). Evidently in the dialect of the 
fellahin of Palestine this is influenced by the fern, both in 

reproducing the vowel -i- (but cf. /J in the dialect of the 
B. Kalb) and in the palatalization of -/;. (iv) Plur. fern. 

originally -hin{na) and so /J in the dialect B. Kalb and 

-hin in South Arabia, 'Iraq, Syria, Palestine, but in classical 
Arabic the vowel has been assimilated to masc. -u- 
(cf. absolute). In North Africa the fem. plur. is obsolete. 

(6) Abyssinian 

Sing. masc. Sing. fem. Plur. masc. Plur. fem. 






Tigre . 




-hen, -x^n 


-ha, -X 




-hen, -dthen 


-X, -h, -ha 



-hu, {-Jiii) 

In Tigre and Tigrina -h becomes -x after a vowel. Amharic 
sing. masc. -x, -h but -ha restored before an enclitic ; sing. fem. 
-h is palatalized as -s ; plur. masc. -hit, which becomes -hu 
after fem. -ac (as -achu) ; the fem. plur. is obsolete. 

(c) Hebrew 

Sing. masc. "?| , HI), "Tj . Hp is rare (Hip'iyy in 1 Kings xviii, 
44). *n- after monosyllabic prepositions with preceding (original) 
vowel lengthened as in TIS for bd-hd ; after other mono- 
syllabic stems ending in double closure either (1) as in pause 
"T]— , thus "^Qi^S " in thy nose ", 2 Kings xix, 28 ; or (2) as 


monosyllabic prepositions, thus 'TjpJD " all of thee " in Isa. 
xxii, 1. 

Sing, fern. % ''^ only in Kethib, as in Ps. cxvi, 7 ; Cant, 
ii, 3 ; Jer. xi, 5 ; and with plnr. in Ps. ciii, 3, 4, 5 ; 2 Kings iv, 
2, 3. Perhaps a mark of northern dialect. 

Plur. masc. DD-. 

Plur. fern. p-. n:JD in HlJsS, Ezek. xiii, 18, 20 ; xxiii, 48, 49. 

I V T V T V T 


(d) Aramaic 

Sing. masc. Sing. fem. Plur. masc. Plur. fem. 

Papyri . . ^ ^^ C3 | 

„ (verbs only) p:)^ p:^^^ p:Di p:):j^ 

^yriac . . ^, ^^ ^, ^, ^ ^QO, ^QD-^ ^, ^^ 

Bib. Aram. IT with imperfect having vowel ending, i 
"nlJ— same with consonant ending, otherwise T]— after 

consonants, "TT- after vowels. So p33 after imperfect with j 

consonant ending. With these cf. verbal suffixes in i 

Mandeean, and -3- forms in Samaritan. j 

(e) Assyrian . 
Sing. masc. -Jcu, -ka. Sing. fem. -ki. Plur. masc, -kunu, -kuni. i 
Dative -kum. -kim. ; 

(/) Sub-Semitic 'parallels 

Genitive with noun, Agau -ki, Hamara ku-, Bilin -ka, i 

Galla -ke, Dambea -ki, Bishari -k, -ok. 1 

Bib. Aram 
Dnqelos . 



86 (4) Third Person 

(a) Arabic 




S. Arabia . -h, -uh, -oh -lid 


'Iraq . 

Syria, Palest 



N. Africa . 




Plur. masc. 

Plur. fem 


-hum, -horn 

- -lien. 






-h, -0 -hd, -he 

-h, -a, -ah -hd, -d 

-oh, -0, -u -hd -hon — 

-h, -0, -u -hd -hum — 

-aJi -hd, -hd -um, -om, -hum -em, -hem 

-u -hd, -hu 

-h, -uh, -u -ha, -a 

-h, -eh -s, -se -hem -sen 


The masc. is evidently -hu with plm^. -hum. Mosul shows 
also -nu with verbs due to the analogy of the 1st person. 
Fem. -hd ; the fem. plur. originally -hin{na), which survives 
in dialect but in classical speech is affected by the analogy 
of the masc. (cf. 2nd pers.). Generally the fem. plur. is 
obsolete in modern dialect. 

(6) Abyssinian 

Sing. masc. Sing. fem. Plur. masc. Plur. fem. 

-hu, -u, -5 -hd, -d -homu, -omu -hon, -on 

-u, -we -od, -ud -dm -au 

. -w -a -hom, -dm -on 

. -u, -6 -d -om,, -'dm — 

In Ge'ez sing. masc. -u appears only in the nominative, 
-0 in the accusative. Fem. plur. shows vowel assimilation as in 
Arabic. Amharic sing. masc. -ive only with verbs ; plur. 
-au after fem. -ac. 

Ge'ez . 
Tigre . 


(c) Hebrew 

Sing. masc. . . nil 1 after vowels. Phoenician \ 

Sing. fern. . . T\ T\ 

Plur. masc. . . DH D i^ Plur. fern, p |. 

{d) Aramaic 

Sing. masc. Sing. fern. Plur. masc. Plur. fern. 

Bib. Aram. ^H H HIJ H [in 

Onqelos . H* H H jin 

bynac . . *a01, 01, ^CTU 01, oi .ooi r^^ 

Mand. . n h^ X* ri ^^ pn p \t\'^ jrj^ p p p^i^ 
„ (verbs) pn3* pi^ pnr p' p 

Samaritan. M 1 Hi H Hi \T\ ?1 p:j pH p 

Nabatsean ^H H DH 

neo-Punic i^ DH D DHi 

Masc. sing, -ih (y\_,, oi, n\ etc.). In plur. a clear distinction 
is made between masc. -u- and fern. -i-. Nabat. masc. ^H 
(Cooke, NSI. 89, 29), fem. M (id. 90, 91, 84), pirn-. DH (id. 76. 
85, 2. 89, 5. 90) ; neo-Punic fem. N- (id. 59, 60), plur. forms 
(id. 53, 55, 59). 

(e) Assyrian 

Sing. masc. -su, -s ; fem. -si ; plur. masc. -§un{u), -sunuti/u 
and -sunUsilu ; fem. -sin{a), -sindti/u, -sindsiju. 



87 The demonstratives are based on particles whicli are of the 
nature of exclamations denoting attraction or aversion, near- 
ness or remoteness. For the most part these demonstratives 
are used in combination with personal pronouns or with one 
another. In some cases they are specialized in such a way 1 hat 
one is used for the masculine another for the feminine, one for 
the singular another for the plm'al, and sometimes they take 
the terminations employed to denote number or gender with 
nouns, but in themselves they have no idea of gender or 
number, and the manner in which they are specialized is not 
the same in all the Semitic languages. 

88 (i) Demonstrative da, di 

The demonstrative da, di, appears throughout West Semitic, 

in Arabic as O, (^^, and by regular phonetic change as 

Abyssinian zd, ze, Hebrew HT, IT, and Aramaic ^^'^, H. Alone 
it denotes nearness and so " this ", a sense often strengthened 
by the addition of demonstrative ha and other particles, 
but it is found also as a merely emphatic addition to particles 
denoting remoteness. In the Arabic dialect of Oman we find 
it used to reinforce the 2nd pers, pron., and thus forming sing, 
masc. dok, fem. dos, plur. dokum. 

(a) The Genders 

In Arabic D is used for the masc, (_$i for the fem., but in the 
other West Semitic languages these genders are reversed, and 


sometimes we find Arabic (_^^ as masc, as in the combination 
{^X>\ . A feminine in -i corresponds with the gender forms 

op 6 

in the personal pronouns, e.g. 2nd sing. masc. cl*> I , fern, C-o I , 
and with the gender suffix in the verb persons as 2nd sing. 

imperf . /^J^i • But -a is the commoner sign of the feminine 

in noun forms. ^ 

The Arabic forms are probably akin to the nounj^, fem. 

Cjb, denoting " owner ", a noun which was actually used as 
a demonstrative in the ancient dialect of the Teiyi, and thus 
used was indeclinable J ^ for all genders and numbers. A 

similar 3>,(S^ is employed as a relative pronoun (cf. 
98 below). 

Although the fem. (_^.i occurs and is frequent in dialect 
as dki (Oman), di (Egypt, North Africa, except Tunis, Mehri), 
and in compounds with -Jca, etc., it is usually replaced in Arabic 

by a fem. ^^ or AT, which sometimes appears as U; this 
U, td, form is possibly borrowed from the demonstrative, 
which occurs in Abyssinian as masc. tu, fem. td (Tigre, Tigrina), 
and as suffixed in we'e-tu, etc. 

In Abyssinian, Hebrew, and Aramaic the genders are 
reversed — di for the masc, da for the fem. In Ge'ez the 
masculine appears as 2e (= zi) and fem. as zd ; but there is 
an alternative fem. zdfi, which corresponds to the Sabgean nS. 
In Tigrina we find the stem 'ez- to which the ordinary gender 
terminations are added, thus masc. 'ezu, fem. 'ezd, and an 


alternative form occurs with demonstrative -iy- (cf. § 96 
below), thus masc. 'eziyu, fern, 'eziyd. Amharic has zi-y 
forming masc. zikha, zikh, fem. zic " this ", and masc. zld, 
fem. zldc " that ". 

In Hebrew we find masc. HT, fem. liT or il and HNT. 
The masc. shows vowel shortening by which the -i becomes 
-e, and similar shortening no doubt takes place in Phoenician T. 
Fem. riT, with regular change of a to o, appears in Eccles. ii, 
2 ; V, 15, 18 ; vii, 23 ; ix, 13 ; and ')], a mere transcriptional 
variant, in Ps. cxxxii, 12, and Hos. vii, 16. In all these 
passages it may be a colloquialism and so inchning towards 
Aramaic X*l. More commonly the fem. afiormative -t is 
added as in PKT (Jer. xxvi, 6, as nHXT). This added -t 
appears also in Moabite HNT (Mesa stele line 3), in Phoenician 
n? (Cooke, NSI. 60) or syth (Plautus, Poen., 5, 1, 1), Sabsean 

n"T, and Ethiopic zdti. An exceptional form ^T occurs in 
poetry (Ps. xii, 8 ; Hab. i, 11), and is also used as a relative 

(cf. below). 

In Aramaic da appears as the feminine, and so in Samaritan 
and Nabatffian (Cooke, NSI. 78), but the masculine has suffixed 
-n, thus Samaritan p (Phoenician JT), Nabateean HH 
(Cooke, NSI. 92, 86), Zinjirli inscription H^T (Cooke, NSI. 63, 
20), Bib. Aram. r\T\. With this must be compared Sabsean 

]"1, Maltese dialect masc. dan, fem. din ; Ethiopic ze-n-tu, 
etc., and Mehri (with change of n to m) masc. dome, fem. 
d:ime. In all these instances the demonstrative da, etc., is 
compounded with the particle -n, which is itself a demon- 
strative (cf. 93 below). 

(b) The Plural 

A plural directly formed from da, di, appears in Arabic 
dialect in compounds with ha-, thus hddun in the dialect of 
TripoH, hddun (Morocco), kddhunva (Tunis), hddhu (Algeria), 


hddu (Tlemsen), and without ha- in Maltese daun. But these 

plurals, JJ^, 3>, are never recognized as tolerable and are 

evidently formed in North African dialect by analogy ; they 
have no equivalents in other Semitic languages. 

The regular plural is formed from an entirely different stem 
'wZ, which is used in West Semitic in the plural only, but 
appears in Babylonian- Assyrian in the singular as well. Thus 

Arabic plur. (Jjl in the dialect of theB. Tamim, or 5^^ I in 

that of the Hijaz. Abyssinian masc. ^ellu, fem. 'ella (Ge'ez) ; 
Amharic masc. 'elldm, fem. 'elldn, and also masc. 'elli, fem. 
'elld ; Mehri masc. liom, fem. lie. It is only in Abyssinian 
and Mehri that we find the genders distinguished. Hebrew 

7{»JI or ^p^{ ; Phoenician 7X (C.I.S. i, 14, 5, etc.) ; neo- 
Punic X7X and Plautus ihj ; Aramaic HyX (Jer. x, 11), I7X 
(papyri), ^7^? (Nabataean, Cooke, NSI. 87, 3). But in Aramaic, 
as in West Semitic generally, these forms appear usually in 
combination with prefixed ha- (cf. § 89 below). Assyrian, 
plur. masc. ulu-utu, fem. ullu-ate, sing, ullum (common 

(c) The Dual 

Classical Arabic also shows a dual formed from the singular, 

thus masc. nom. jl^, oblique /j.O ; fem. nom, jlT, 


oblique A.*) 

(d) Combination of the da and 'ul stems 
Abnormally in Arabic dialect we find the combination of 

1^ and 2J 1 ; thus, in the dialect of Mecca dol, in that of Egypt 

doll, dold, and in the speech of women dolat. This combination 
occurs very often with prefixed ha- (cf. 89 below). 



89 (ii) Demonstrative ha >^. \ a^ " '^^■r^4 . ^^ 

Demonstrative ha appears in Arabic as a particle implying 
nearness to the speaker, as hd huwa " here he is ! " hd hund 
" here ! hither ! " It occurs with the 2nd personal pronoun 
suffixed as hdkd, hdki, hdkum,, hdkunna, " here you are ! " 
in the sense of " take this ! " but it is distinct from the verb 
ha'a " take ", which gives masc. ha'a, fern. Jia'i, plur. masc. 
hd'um, fern, ha'unna, as in Qur'an, 16, 19, etc. So in dialect 
hdk ('Iraq), hdku (Petrsea). In dialect we sometimes find 
Jidhuwa as an emphatic form of the 3rd personal pronoun, 
thus Egyptian sing. masc. aho, fem. ahi, ahe ; Omani ha-uwe 
" this is ". Perhaps it is to be identified with the first part of 
the stem in huwa, hiya, with vowel assimilated to the following 
semi-vowel. In Hebrew the same particle appears as he ' 
in Gen. xlvii, 23 ; Ezek. xvi, 43 ; and in Aramaic as hd 
" lo ! " in Dan. iii, 25 ; and Syriac hd as in Pesh. Matt, x, 
16. Probably it is akin to 'a in Aramaic pX (for jnrt) " this " 


It occurs as the definite article in Hebrew, Phoenician, 
Moabite, and sometimes in Samaritan. In this use it appears 
as hd- with closure by doubling the following consonant, 
or (in Hebrew) as he-, hd-, before the laryngals ; but of these 
hd- is obviously the normal form, although this does not 
preclude its identification with Arabic hd, the shortening 
being due to its use as a prefix. Possibly both are connected 
with hay (cf. below). There does not appear to be any basis 
for Stern's theory which identifies ha- with Arabic al, supposing 
an original hal with I assimilating with the following consonant, 
for it is obviously not a case of assimilation as may be seen from 
the hd- before certain laryngals, but simply an instance of the 
preservation of a short vowel by the expedient of closing the 
syllable by doubling the following consonant. 

The same particle appears as a sufiix in Abyssinian 'ethd 


" this time ", where its use is adverbial, and so Hebrew -o 
in ^attd " now ". As a sujfi&x it is also used in Aramaic in the 
so-called emphatic form which was originally the stem with 
suffixed article, but has now lost its determining power. In 
Samaritan it is used as an alternative for the prefixed article, 
and is the commoner. Before leaving the use of ha as a suffix 
it may be noted that it is employed in the Omani dialect as 
an enclitic for strengthening other demonstratives, as dak-ha, 
dik-ha " that ", and so in Egyptian masc. duk-ha, fem. dik-ha. 
But it is as a prefix reinforcing other demonstratives that it 
most commonly occurs. 

(a) Ha with da, di, and plur. 'ul 

Arabic shows ha- prefixed to da, etc., as the commonest form 
of the demonstrative " this ". Thus sing. masc. hddJid (hddd 
in Hadramaut, Syria, Morocco, Algeria), fem. hddJn {hddl in 
Syria, Morocco, Algeria), or hddhihi or hdtd. In 'Iraq we find 
hadh for both genders. Omani has also masc. dhdhd, fem. 
dhihd. No plural is properly formed from this stem, but we 
find in North Africa hddun (TripoU), hddum (Morocco), 
hddhuma (Tunis), hddhu (Algeria), hadil (Tlemsen), and 
hydaun (Tripoli, Malta). 

Hebrew shows sing. masc. hazze, fem. Mzzoth ; Phoenician TX. 

Aramaic has sing. fem. hadd, Syriac hade, rarer had, and 
enchtic (common gender) hddd ( .^qiljOl ) ; Mandsean 
XTXn, Samaritan J^TSH, T.J. 'add, T.B. hadd or hu. Neo- 
Syriac of Ma'lula sing. fem. hodh, plur. hattin. The masculine 
singular is formed with added -na, as Bib. Aram, hdden, 
Samaritan hadin, T.J. hadin, T.B. hadm or ha'i, Mandsean 
|"'TNn, Syriac hand, hdn (from hdden), Ma'lula hanna (for 
hadna). The plural is regularly formed by ha-ul. Thus 
Arabic hd'ula'i (Hijaz), hau Id'i (B. Oqeyl), ha'Ula (B. Tamim), 
hold ('Iraq). Hebrew ha'el (Pentateuch only), hd'elle (else- 


where in the O.T.), hallu, 'ellu (Mishna). Aramaic Mien, 
Mandaean j^'^yn. 

(6) Combined ha-da-ul 

As we find dha and 'ul sometimes combined in Arabic 
dialect, so we find this combination also with prefixed Jia- in 
hadhol ('Iraq), Jiadol (Meccan dialect), hddul (Tripoli), hddhula 
(Tunis), hddol (Damascus), hddhal and contracted Jidl (Syria 
and Central Arabia), masc. hddhold, fem. hddhdll (Nejd), 
hddhile (Oman) ; and a derived masc. sing, hddold, fem. 
hddenm ('Iraq). 

90 (iii) Demonstrative hay or 'ay 

This particle occurs in Arabic as exclamatory 'ay and as 
'ayya with the pronominal suffixes, the latter chiefly in 
North Africa. Another, probably the original, form appears 
in the interjection hayd. In Palestine we find hei " here is " 
with pronominal suffixes, as heynl " here am I ". So 
Abyssinian heya. Probably it is the same root which appears 
in Hebrew as ''hi " where ? " (Hos. xiii, 10, 14) and in the 
Mishna as he in he'dkh " how ? " hel'kd " therefore ", etc. 

(a) In Arabic we find Jiay compounded with dd, etc., in 
hmjdd, haydi " this " (Syrian dialect), and heydah " that " 
(id.), and with t demonstrative in haytd, haytu " come here ", 
and hatin, had " give, bring " ; and also in hayhdt- " away ". 

(6) In Aramaic ay occurs in T.B. H'i^l n^J< " this and 
that ", where ay strengthens the demonstrative di ; and hay 
in hayidd " this " (fem.) in T.J. So T.B. 'aijdak " that ", with 
plur. 'aynak. 

91 (iv) Demonstrative la 

In Arabic this demonstrative appears as an exclamation 
compounded with yd in ydlaka, ydlahu, etc., used as an 
interjection expressive of surprise or admiration, " thou ! " 
According to the grammarians of Kufa, this ydla was originally 


yd'ala, followed by a proper name, the whole forming the 

aJLaU-I (S^^ "the war-cry of the time of ignorance", 

which was forbidden under Islam. The word ydla could be 
followed by a pr.n, in the genitive or a suffixed pronoun 
denoting the person invoked, and this could be further 
followed by the accusative or min. 

In Arabic this demonstrative appears as the article, usually 
la- becoming -1-, but in Omani also as lo-, Id-, lu-, e.g. lomse 
" the evening ", etc., or assimilating with dentals, sibilants, 
and sonants, as in ssagg, ththora, nnefes, etc. In Morocco 
it occurs as la-, le-, lu- ; and in Southern Arabia as am-. 

Professor Wright says of the article that " though it has 
become determinative, it was originally demonstrative, as 

still appears in such words as ^j.J 1 " to-day ", ,J i( 1 

" now ", etc. (Wright, Arabic Gr. i, 269, B). Hebrew perhaps 
retains traces of this article in such words as 'almoddd (Gen. x, 
26), 'eltoldd (Joshua xv, 30 — toldd in 1 Chron. iv, 29). We also 
find a reduplicated form in Abyssinian la-la- >lali- with a 
pronominal suffix as lalika, denoting " that ". So in Arabic 
la-la- >'al-la- in compounds with dl (cf. below). 

(a) Compounded la-ha 

Abyssinian (Tigre dialect) masc. lahay, fem. laha, with 
pronominal suffix denoting " that ", the gender terminations 
following ze, za (above). So Hebrew halla " that " (Mishna). 

(6) Compounded la-da e^V -<- <ft — ^ * • -^ '^<- 

Arabic, reduplicated stem in ^alladM, 'allati, etc., i.e. 
'alia- compounded with dJii (not dhd), which here shows masc. 
in -I, as in Hebrew, Abyssinian, etc. Hebrew halldze (Gen. xiv, 
65), haldz (Judges vi, 20), fem. hallezu (Ezek. xxxvi, 35), 
halldz (2 Kings iv, 25), where hall- = 'all- of Arabic. 


92 (V) Demonstrative ka -: ^"•^' '^^^ ^ 

This particle is used to denote a remote object, " that 
yonder," etc. Thus dhd " this ", dhdka " that ". It occurs 
as an adverb in Tigre ka " then ", Hebrew ko " thus ". With 
the interrogative ay ( = e) it produces the Assyrian ekd 
" where ? " As a demonstrative it is chiefly used in com- 
bination with others to give an idea of remoteness. 

Arabic sing. masc. dhdka, fem. tdka " that ", in dialect fem. 
dink (Oman, Central Arabia), dik (Mehri, Maltese) ; plur. 
'uldka, 'ulaika, and in dialect dhuk (Morocco), dauka (Maltese). 
Egyptian dialect shows also the compound kide and Mardin 
kide. Abyssinian (Ge'ez) sing. masc. zeku, plur. 'elleku. 
Amharic sing. masc. zika, zik, zih, fem. zlc, zihc, corresponding 
to Arabic dhdka, etc. Aramaic, Bib. Aram. sing. masc. dekh 
(Ezra V, 17, etc.), fem. ddkh (id. iv, 13), plur. ''37^^ in the 
papyri, 'illekh (Dan. iii, 12). Mandaean haek for hadek, and 
T.B. masc. "]\^n, fem. "[XH. 

Arabic also shows prefixed ha- in sing. masc. hddhdka, 
fem. hdtdka, hdtika, etc. Also in dialect hdddk (Mosul, Mardin, 
North Africa), hddhdk (Oman), hdddk (Datina), hddhika 
(Tunis), and fem. hddhlk (Mosul, Mardin, 'Iraq), hadic 
('Iraq), hddhik (Oman), hdtak (Datina) ; inverted order in 
Egyptian masc. dukhd, fem. dikJid, forms which are found also 
with the pronominal suffix as masc. dukhauwa, fem. dikJmiya, 
and in Oman masc. dhdkhd, fem. dJnkhd. Plural by 
substitution of 'ul for dhd as dolak (Mecca), dylakhin 
(Oman), etc. 

Further compounded with la-, in sing. masc. dhdlika, 
fem. tilka, tdlika " that " with plural 'uldlika, 'uldlika. Syriac 
hdrka for hdlka. 

93 (vi) Demonstrative na 

Like I- and certain other demonstratives this may be 
vocalized by a following or by a prefixed vowel, na-, 'an, 'in. 


etc., and just as la-la- appears in Arabic as 'alia-, so na-na- 
becomes 'anna (of. 91), Assyrian annu. As usual, we have 
occasional changes of initial Hamza to h-, and thus we get 
the Sabsean article jH, Hebrew hen " this " {'ane in 1 Kings ii, 
36, etc.), and -n without a prosthetic vowel when vocahzed 
by a preceding vowel, as Jcen (Hebrew and Syriac) " like this ", 
i.e. " thus ". It is worth noting that Sumerian has a personal 
pronoun NI, NA, NE, which is also used as a demonstrative, 
and with this we must compare the particle 'in-, introducing 
the nominative in ancient Egyptian (cf. 76) and Galla ini, 
Dambea ni (demonstrative). 

As a prefix this demonstrative most commonly appears as 
'an- or han-, less commonly as m-, na-. Thus in the absolute 
personal pronoun, 1st and 2nd pers. 'an-a, 'an-ta, etc., showing 
na- or 'an- in the 1st plur. Arabic naJmu, Hebrew "^nahnu, 
etc., and this prefix is sometimes extended to the 3rd person, as 
masc. \)^H, fem. pJX in Samaritan, masc. 'enun, fem. 'enen 
in Syriac in the plural, and Bib. Aram, himtnon, with sing. 
Jiimmo where himm- — hinm- ; less frequently it assists in 
forming an emphatic singular ni-hu, ni-hi "it is he, she " 

As a suffix it appears as -n, -na, -nu, etc. Thus, with 
prefixed interrogative 'ay-, Arabic 'ayna " where ? " 'ayydna 
" when ? " Hebrew 'ayin, Assyrian anu, ani. As a suffix to 
the personal pronoun, it occurs in the 2nd sing. fem. 'anti-n{a) 
(Morocco), antydna (Tunis), 2nd plur. ' antum-dn{a) (Morocco), 
and 3rd sing. masc. hi-nu (Mosul, Bagdad). Occasionally it is 
used instead of the personal pronoun, as in lanu " to him " 
(Mosul, Bagdad), where -nu = -nhu. 

But the commonest use appears in inserted -n- between the 
verb stem and the pronominal suffix. In Arabic this occurs 
only in the dialects of South Arabia, with the suffixes -m, 
-nd, attached to participles which thus become -in-nl, 


-in-nd. Tigre with plural suffixes -an-nd, -kkum, -kken, 
where -kk- is for nk. In Hebrew we find -en- in -mm, -enhu 
(Exod. XV, 2), etc., and rarer -dnni ; 2nd sing, -en-ka ( Jer. xxii, 
14), but commoner -ekka ; once 3rd sing, -nehu (Deut. xxxii, 
10), commoner -enhu (id.), -ennu, -enno (Num. xxiii, 13). 
This inserted -en- occurs with singular suffixes and 1st plural 
only. Aramaic -n- after a vowel, -in- after a consonant, as 
Bib. Aram, -inna, -nnd ; in 2nd plur. -inkom, etc. In the 
papyri we find 2nd sing. masc. "ly, fem. *D> ; so Samaritan 
2nd masc. |133, fem. p^X y^T ; Mandsean 2nd plur. masc. 
P^X ]1D5<i, 3rd masc. pi"*, fem. y^T. Assyrian -in- or -an- 
with assimilation as -akk{u) for -anku, -issu for -imu, -assu 
for -ansu, etc. 

This demonstrative also appears compounded with da, 

'ul in Sabsean p, Aramaic sing, den (Bib. Aram.), |T in 
Phoenician, p in Samaritan, HiT (Zinjirli, Cooke, NSI. 63, 
20), Nabataean T[T\ (id. 92, 86) ; plur. Sabsean |Sx, Aramaic 
'alien, Mandsean Vt^. Compare sing. masc. ddn, fem. dm 
(Maltese), plur. 'ellon (Abyssinian), and plur. masc. 'ellom, 
fem. 'elldn (Tigre). 

With added -k (cf. 92), Aramaic pT (papyri), dikken 
(Bib. Aram.), hanik (Mandsean), hdnek (T.B.), and Syriac masc. 
^QJOi, fem. ouiJOl. 

94 (vii) Demonstrative ma 

This seems to be akin to nunation, i.e. to the final -n used 
in the indeterminate form (cf. 132), thus Sabsean D^^b = 
Arabic malik-un, Sabsean D12J'n= Arabic Jiasid-un. In Arabic 
-ma appears in 'ayma " so far as, as regards ", and in halma 
" come here " from hala or Jialla, and 'ayma " what ? " with 
interrogative 'ay. In another form it appears as -umma 
in the vocative 'alldJmmma " God ! " Perhaps it is akin 
to the article am which appears in Southern Arabic 


(Hadramaut em-, cf. Landberg, Datina, ii, B, 7 ; 8, 9, 10 ; 
8, 6, 15). Suffixed -ma also appears to be related (cf. Mufassal, 
28, 14 ; 74, 4) ; but it must be noted that in these instances 
m may be due to a phonetic change from n. In Abyssinian 
it appears as -em, used adverbially in temdl-em " yesterday ", 
and so in Hebrew silsom " day before yesterday ". Assyrian 
ammu " that ", and adverbially in kiam " also ", kiam 
" thus ". 

This demonstrative occurs in combination with other 
demonstratives in Mehri ddkim, fem. dlkime " that ", and in 
Aramaic DDT, which appears in uyh, DDD in the papyri. 
It is doubtful whether we ought to identify it in the Arabic 
plm\ hdduma, hddumka (Morocco), hddumma (Tripoli), or 
whether this -hddum is merely a dialectal plural of hddd ; 
but it may well be an affix in Mehri sing. masc. dom, dome, 
fem. dim, dime, plur. liom. 

With this Semitic demonstrative we may compare Afar 
ama " this " and Irob-Saho mnma, ammay " this ", both of 
the East African group of Hamitic. 

95 (viii) Demonstrative ta 

This demonstrative appears in Abyssinian in composition 
with prepositions, as hotu, hdii, lotu, Idtt, etc., and in Assyrian 
jdtu, jdti, etc. With this ta, tu, ti. Earth {Pronominalh. 30) 
connects the adverbial termination in Arabic rahha-ta " very 
much ", Aramaic beth " therefore ", rebhath " very much ". 

It most commonly appears in fem. ti (masc. tu, as in pers. 
pron, hu, hi) in Arabic, replacing the fem. of dhd (cf. 886) 
in tdka, tika, tdlika, tilka ; in Abyssinian it occurs as masc. 
tu, fem. ti, suffixed to the personal pronoun in we'etu, ye'eti, 
etc., and as a personal pronoun in Tigre and Tigrma, sing, 
masc. tu, fem. ta, plur. masc. torn, fem. ten (Tigre), or tan 
(Tigrina) ; also in Tigrina in the form sing. masc. 'etu, fem. 
'eto, plur. masc, 'etom, fem. 'etan ; Amharic masc. 'ut, fem. 


'at ; Sabaean HI, Phoenician H^H. In Assyrian as sufiSxed 
to the 3rd pers. pron. (cf. Abyssinian) su-tu, su-dtu, su-dtt, etc. 
We find it also compounded with other demonstratives 
in Arabic hay-tu, -ti, -ta " come here ", hmj-ha-tu, -ti, -ta 
" away ", although these may be cases of the personal endings 
of the verb applied by analogy to demonstrative stems. 
Abyssinian sing. masc. zen-tu, fern, zd-ti, plur. masc. 'ellon-tu, 
fem. 'elldn-tu, also plur. 'ellotu. With these compare Galla 
emphatic -tu as in ani-tu. 

96 (ix) Demonstrative ya 

This demonstrative appears in Arabic as yd, an exclamation 
drawing attention, and as 'ay. Both are combined in the 
reduplicated forms 'ayya, 'iyya, just as Hala becomes 'alia. 
As y- with a half- vowel it is found attached to the personal 
pronoun in Soqotra y-he, and appears also in North African 
ana-ya, anta-ya, etc. (Morocco), anti-ya (Algeria), jyn, jyna, 
ynae (Malta), and ya-h " he " (S3rria). In Oman it is found 
with prepositions as biya in biyadayla " with these ", etc. In 
Tigre it is used in the exclamation yaha to call attention, and 
in Tigriiia and Amharic it occurs as attached to the personal 
pronoun, thus 'afie for 'ana-ya (Tigriiia), 'ene, 'enei (= 'en-ye, 
'en-yei, Amharic). So Aramaic in\^, ^"I'^{ (T.B. and T.J.) 
and Mandaean plural pH'N*, |'n\S* for pn, JTl ; to which 
must be added Assyrian a-a-u (= ay-yu). 

It may perhaps be connected with the particle employed to 
denote the accusative in the form 'iyya with pronominal 
suffix as 'iyyaya, 'iyyaka, etc. Assyrian k-iy-ya with suffix 
as kiyahu " him ". Akin is the interrogative 'ay (cf. 106), as 
in Hebrew 'ay ye " where ? " Tigre 'ayyi, fem. 'ayya. Com- 
pounded with tu, ti, ta, it appears in Hebrew as the particle 
denoting the accusative, 'eth, 'eth-, Phoenician n"*{^, 
Hebrew 'adth with suffixes, Mishna 1^1^5, Aramaic rT'i^ 
(papyri), later yath (Dan. iii, 12, etc.), wath in Syriac 'akwath 


" like ", with personal suffix attached. Tigrina 'etu, for 

Compounded with other demonstratives we find hadulaya 
(dialect Tripoli), and masc. Iiadahaya, fem. hadikaya (id.) : 
Abyssinian zentu, where -e- is possibly for -ay- (Ge'ez), and 
Tigrina masc. 'ezi-yu, fem. 'ezi-yd. Aramaic (papyri) ''2/^^, 
Syriac dm. 

97 (x) Demonstrative aga 

Assyrian, sing. masc. aga, fem. aga-ta " this " ; in plural 
with annutu as agannutu, etc., cf. Hebrew ge " this " in 
Ezek. xlvii, 13, but it may be, as Gesenius thinks, that H^ 
is a transcriptional error for DT: Z.A. iv, 56, regards aga 
as a variant for a'a, and Jensen (Z.A. vii, 173 sqq.) considers 
that aga-n is allied to the root KWN. 



98 (A) The Relative Pronouns 

(i) dhu, etc. (cf. demonstratives). The same form used for 
all genders and numbers, special forms for feminine and plural 
only in colloquial dialects. Arabic dhii, but also dhi in the 

combination 'allddl (cf. 916); Sabsean 1; Ethiopic zd; 
Hebrew ze, zii ; Aramaic zu in the Ben-Hadad inscription, or 
di, d^ (this the commoner form in Syriac, the Targums, T.B., 
and Mandaean). 

Fem. Arabic dju as di in dialect Hadramaut and Morocco ; 

Sabsean H"! ; Ethiopic zi- in combination with the personal 
suffixes, thus sing. 1st zVaya, 2nd zi'aJca, etc., and 'ento, a form 
which Barth regards as connected with Berber relative enta 
{Pronomin. 676). 

Plur. Arabic 'ul in elli, illi (dialect 'Iraq, Syria), yalli 
(= ya-\-elli in Syrian), halU (Egypt, Mecca, and North 

Africa) ; Sabaean vX. 

99 (ii) ('o)i!. Rare use of the article as relative in Arabic ; 
so hd- in the later Hebrew of Chronicles and Esther (cf. 2 Chron. 
xxix, 36, etc.). Arabic dialect li (Dofar, Tunis, Malta), ^el 
('Iraq, Syria, Palestine) ; Sabsean 7^{ ; Tigre la. 

100 (iii) ^a, etc. Hebrew' '*8er, late se-, s* (2 Kings vi, 11 ; 
Judges V, 7, as mark of northern dialect of Israel), Phoenician 
tJ'X (C.I.S. i, 2, cf. assamar in Plautus, Poenul. Iii, 26), 
^ (C.I.S. i, 112, etc.), this latter more frequent in neo-Punic 
{se in Plautus, Poenul. i, 1, 3). Later Hebrew sel as mark 
of the genitive (=:'"«er li " which is to . . . ", cf. 128), so Punic 
(Cooke, NSI. xxxix, 2 ; xli, 2, etc.). Assyrian sa, su. 


The same root appears in Assyrian assam " there ", Hebrew 
sdm, Aramaic tamman, Arabic thamma, thumma. 

101 (iv) The interrogatives man, 7nd (cf. 102 below) are also 
used as relatives. Thus Arabic ma7i " who ", md " which " ; 
Hebrew ml " who ", md, md (cf. 102) " which ", and also mo 
in the poetical forms 6*mo (Ps. xi, 3), k^mo (Ps. Ixxv, 15), l^mo 
(Job xxvii, 14). Aramaic md. So Ethiopic md- in md'ze 
" when ", Tigre ma'aze. Arabic mdtd (cf. interrogatives), 

102 (B) The Interrogative Pronouns 

(i) wa. Arabic md " what ? " shortened md with the 
prepositions bimd, lima, etc. Also mdhuwa " who ? " 
(Moroccan dialect), mhu, mu " who 'i " (Oman). 

Abyssinian ma with preposition kama, etc. 

Hebrew md, md (cf. above), the short retained by closure of 
the syllable, thus mdllakem " what is it to you ? " (Isa. iii, 15), 
mdbhesas " what profit ? " (Ps. xxx, 10), mdzze " what is this ?" 
(Exod. iv, 2). Before h, h, ' (cf. 53), as me, me hddel 
(Ps. xxxix, 5), etc., also at beginning of sentence as in 
2 Kings i, 7, etc., and in lame (1 Sam. i, 8), kamme, bamme, etc., 
but lengthened as md in l^md, bammd, kammd, lammd. 

Aramaic md (Ezra vi, 8), Syriac ma. With prep, k^md, etc. 

The same root is combined with dh ot t (cf. relatives) in 
Arabic mdtd " when ", Ethiopic md'ze " when ? " Hebrew 
Tnatay (id.), Syriac 'emta " when ? " and in Syriac mdnd 
(=*md-de}id) " who ? " and Targ. mddm " why ? " 

103 (ii) man. Arabic manu, mam " who ? " (cf. relative 
man), Ethiopic mannu, ace. manna, plur. 'ella-mannu ; Amharic 
mdn, ace. mdnan, plur. 'ella-mdn ; Tigre, Tigriiia tndn. 
Aramaic man, mdnd in Sp'iac and Targ., also Syriac miln ; 
Samaritan V2 and so Nabatsean (Cooke, NSI. Ixxx, 87) ; 
Syriac manu, T.B. masc. manu, fem. mam. Assyrian, nom. 
mannu, mannum, obhque cases mannim. 


104 (iii) nil. Abyssinian mi-atow " how much ? " (Ge'ez) ; 
Tigre nil " what ? " Hebrew nil " who ? " and as " what ? " 
(Amos vii, 2, 5). Aramaic mi (T.B.) as simple particle of 

105 (iv) min. Arabic min " who ? " men, min (Hadramaut, 
'Iraq, Morocco). Abyssinian men (Amharic). Assyrian 
minu " what ? " ana meni " why ? " Also in Abyssinian 
ment " what ? " (Ge'ez), ment-av (Tigrifia). 

106 (v) 'ay. Arabic 'ayy-, 'ay!/ (Egypt, Hadramaut). Ethi- 
opic 'ay (indeclined). Aramaic masc. 'ay-na, fern, 'ay -da, plur. 
'ay-len. Assyrian ay. All these denote " who ? what ? " 
used adjectivally. 


107 (A) Nouns generally 

The question as to whether nouns or verbs come first in 
the historical evolution of language may be regarded as 
largely one of those theoretical exercises which are but little 
calculated to advance the practical work of philology. So far 
as the Semitic languages are concerned there are undoubtedly 
older forms surviving amongst the nouns than amongst the 
verbs, and the variety of noun forms as contrasted with the 
comparatively stereotyped verb form seems to support the 
view that the nouns present an earlier type than verbs. But 
it is absurd to argue as though all the nouns came into existence 
first and all the verbs were derived from them or vice versa. 
Early Semitic, like other primitive languages, must have 
possessed a limited range of noun ideas and verb ideas, and 
many new nouns were afterwards formed by derivation from 
verbs, as well as new verbs from nouns. We say noun ideas 
and verb ideas because it is quite possible that there may not 
have been a limited vocabulary, but, as appears in several 
African languages, a large though vague vocabulary full of 
synonyms, the progress of development being mainly an 
advance in the formation of accurate ideas, and consequently 
the specializing of synonyms so that those previously used 
indefinitely of a whole genus were now applied to different 
species, and thus the vocabulary became more accurate. 
Such a progress had taken place to a great extent before the 
separation of the different branches of Semitic, but it had not 
taken place before the separation of Semitic from the earlier 
Hamitic group, for there we find hardly any likeness in 
vocabulary although close similarities in morphology. 


But these must be regarded as mere suggestions. Whilst 
we can find primitive languages which are difiuse and vague 
in vocabulary, there are undoubtedly others which are scanty 
and yet fairly accurate in denoting a limited range of common 
objects, as seems to be the case with the Baltic group in the 
Indo-European family. We seem able to state that when 
Semitic was specialized its parent stock was by no means of 
a primitive type, and so proto-Semitic was itself at a fairly 
advanced stage of development. 

The great majority of noun and verb stems show a base 
form of three consonants, but a small number of bi-consonantal 
roots appear amongst the nouns. Some of these are found 
also in ancient Egyptian, whilst others not so found are 
assimilated to tri-consonantal roots. 

108 (I) Two Consonant Roots 

(a) First Group 

Two consonant noun stems which show no assimilation to 
three consonant stems. 

6n " two " : ancient Egyptian sn ; Arabic i^w {dua\i6ndni) ; 
Hebrew s^nayim (dual) ; Assyrian sind. 

sp " lip " : ancient Egyptian sp-t (fern, affix) ; Arabic 
safat ; Hebrew sdfd ; Assyrian sap-tu. 

s' " sheep " : ancient Egyptian sw ; Arabic sa' ; Hebrew se. 

mH " water " : ancient Egyptian mw ; Arabic ma' ; Assyrian 
mi-i \ Abyssinian may (cf. § 19) ; Hebrew '^maij-, in plur. 

mt " man " : ancient Egyptian mt ; Hebrew 7na6 or meO ; 
Abyssinian met ; Assyrian mut-u. 

bn " son " : Arabic ibn ; Hebrew ben ; Assyrian binu 
(Aramaic bar). 

sm " name " : Arabic ism ; Sabaean DD ; Hebrew Im ; 
Assyrian sumu. 


vrC " hundred " : Arabic mi'at ; Abyssinian m^-et (fern.) ; 
Hebrew me' a ; Assyrian Ttieat. 

jpm (pw) "mouth" : Arabic /am ; Abyssinian 'a/ ; Hebrew 
pe ; Assyrian pu. 

yd " hand " : Arabic yad ; Sabsean IX ; Hebrew yad ; 
Assyrian idu. 

(^m" blood": AiSihic dam ; Hebrew dam ; Assyrian damn. 

(6) Second Group 

Noun stems which show two consonants but are frequently 
treated as having three. This may be an assimilation to 
tri-consonantal forms, or it may be that they were originally 

'6 i'bw) " father " ; 7i {'hw) " brother " ; hm (hmw) " father- 
in-law " : thus Arabic nom. 'abil-, ace. 'aha, gen. 'abl before 
suffixes, and so generally in the Semitic languages. This 
may be final -w assimilating, or it may be a long vowel inserted 
as compensation after a bi-Hteral stem. 

'm {'mm) " mother " : Arabic 'umm ; Hebrew 'em ; 
Assyrian ummu. 

ym [ymm, ywm) " day " : Arabic yawm ; Hebrew yom ; 
Assyrian immu. 

In any case these two consonant stems are not numerous, 
although, as will be noticed, they include some very common 

109 (11) Vocalization of Noun Stems 

Nouns other than proper names may generally be divided 
into three leading classes : (i) common nouns denoting con- 
crete objects ; (ii) adjectives and participles ; and (iii) abstract 
nouns, including the N. Verbi and infinitive. The older and 
commoner nouns belong to class (i), whilst the abstract 
nouns of class (iii) are presumably of later formation. 



(1) qatl, (2) qitl, (3) qutl 

These are three very common noun forms showing words 
of all three classes enumerated above. In Abyssinian (2) and 

(3) are confused together in one qetl (qetel) type (cf. 47, 48). 
Where the case endings are obsolete a vowel is of necessity 
inserted between the second and third radicals (cf. 68 above), 
thus kalb- " dog ", Hebrew keleb, with suffix kalbi ; Aramaic 
k^leb, emphatic kalbd. So Assyrian construct kalab, absolute 
kalb-u, etc. For roots with semi-vowel as medial or final 
radical cf.'SS 51, 52. With initial y- Hebrew adds afformative 
-t, thus root ysb, infin. sebeth (cf. § 149). 

(4) qatal 

Also a common type of all three classes. Hebrew qdtdl 
(cf. § 46, chap, iii), but Aram. qHal, qatld, as (1) with 2nd or 
3rd laryngal, etc. 

(5) qital, (6) qutal 

These two are frequently interchanged : but normally 
(5) appears in concrete nouns, (6) in adjectives, whilst both 
are found in abstracts. They are confused as qetal in 
Abyssinian. In Hebrew (6) may appear as qutdl in more 
ancient forms as the pr.n. Sii'dr, but generally and 
regularly it becomes qdtdl as sofdr " trumpet ", rarely qdtdl, 
qutdl as in Sofdr (pr.n. in Job ii, 11) and sugar " cage ". 

(7) qatil, and by assimilation (8) qitil 

These are found represented in all three classes. In 
Hebrew (7) qatil becomes qdtel, qHeli, or assimilates as (8) qHel 
in the absolute, e.g. p^'er " ornament ". 

(9) qatul, and by assimilation (10) qutul 

(9) in concrete nouns and adjectives, (10) more commonly 
in abstract nouns. 


We turn next to the forms containing a long vowel, as : — 

(11) qdtil 

This is the commonest form for the N. Agentis and active 
participle (cf. sect. 147 below), and appears also as an abstract 
type. Hebrew qdtel, etc. 

(12) qdtil and (13) qdtul 

These are distinctively Aramaic types, the former as the 
usual Syriac passive participle (cf. 147). Arabic contains 
some loan words from Aramaic of type (13) as sd'ur, etc., and 
perhaps some native examples in such words as 'dthur. 

(14) qatdl, (15) qitdl, (16) qutdl 

The two former have examples of all three classes, the last 
is mainly adjectives (especially diminutives) and abstract 
nouns. In Hebrew q^tol seems to correspond with Arabic (15) 
and (16), although we also find qetol for (15). These forms 
make feminine qatalat, qitalat, qutulat shortening the second 
vowel (cf. Barth, Nominalhildung). 

(17) qatil, and by assimilation (18) qitil 

Chiefly adjectives, abstract nouns also of type (17). 
Feminines as qatilat, qitilat. 

(19) qatul, and assimilated (20) qutul 

Adjectives and abstract nouns. Fem. qatulat, etc. 
Abyssinian nouns qetul for (19) and either qetul or qutul 
for (20) ; both these latter are rare. In (19) Hebrew may have 
qdlol or qdtul, as ydqos (Hos. ix, 8), and the same word as 
ydqus in Ps. xci, 3. 

Passing to forms with doubled medial, we have : — • 

(21) qattal, (22) qittal, (23) quttal, (24) qittil, (25) qattul, 
(26) quttul, and (27) quttul 
The types qutayl, etc., are treated as stems with informatives 
(cf. 114). Deverbals from verb stems with added n- or t- 


must be classed under the verb themes which show these 
preformatives, whilst stems with m-, etc., are treated below 
(§§ 110-13). 

110 (in) Noun Stems with preformatives 

(i) Preformative m- 

(a) Arabic 

(!) The preformative ma- (mu- before stems containing 
more than three consonants) denotes the time or place of 
the action or state designated by the root, or the place 
abounding in or producing the material denoted by the 
parent noun ; or the abstract or N. Verbi in deverbal forms, 
or the participle (verbal adjective). 

(a) Time : ojj " bear chUd ", aVJa " birthday ". 

.. . "^  

(/3) Place : X^a^ " repair to ", >X.,,!2JL4 " place aimed at " ; 

J.^^ " enter ", J-^A-"^ " entry " ; "J>^^ " dwell ", 

/-X_.*w« " dwelling " ; c- > »c- " set (sun, etc.) ", »—>»*-• 

"west"; JU-I "lion", XJ\^a "place full of Hons " ; 

/Ju? " pray " (conj. ii), /JUa^ " oratory " ; -\.5t>-» " bow 

down ", Jbt.**-« " mosque ", etc. 

(7) Abstract,, N. Verbi, Infinitive : pi" eat ", JJ V-.4 

" act of eating " ; XU- " be patient ", infin. ^jXj:>s.a- Also 

with afformative -t, as Xs- "praise", infin. 6-\.<>.r»c^. 
Sometimes maqtal is the N. Verbi where maqtil is the noun 


^ » 

of place or time, thus ^Y*^>>• "sit", infin. ^**JL3tL«, but 

^JLst^ " place where tribunal sits ". 
(S) Participle : passive ptc. of the Primary stem as 

J J.-A-4 ; all participles of the derived stems, as J^-* 

(intens.), J^^a^ (caus.), etc. 
(2) Preformative mi- denotes instrument, sometimes also 

the living agent ; thus pid " open ", 7^^ " key " ; so 

^^^ " file " {mabrad in dialect, Egypt, 'Iraq) ; with -t as 

A5e,*«.>w4 " broom ". 

(6) Abyssinian 

(1) ma- instrument, object, agent. Thus 'adada 
" reap ", ma'dad " sickle " ; sdrara " found ", masarat 
" foundation ", etc. 

(2) ma-, me- (for mi-) denoting place, as ^draqa " rise 
(of the sun, etc.) ", mesraq " east " ; sahaha " lie ", meskab 
" bed " : with initial w-, muldd " birthplace ". 

(3) tna-, me-, abstract, etc. In Amharic the infinitive 
regularly with ma-. Thus 'arafa " rest ", m^'raf " rest 
(noun) ". 

(4) ma- with the participles, maqattal, etc. 

(c) Hebrew 

Generally ma-, mi-, are not strictly specialized in meaning, 
but depend more on phonetic influences. Usually md- is 
preserved before a laryngal or in double closure, md- appears in 
an open syllable before the tone, me-, mi-, m*- are produced by 
the operation of the phonetic rules already given (cf. sect. 46). 

(1) Place, tim£. |*y " fountain ", j^^^tt " place abounding 


in fountains " ; |^D " lay up ", ]1DtOp " storehouse " ; Dp 
" stand up ", Dlp^ "place " ; niT " slaughter ", niTp " altar, 
place of slaughter " ; ]yf " dwell ", pS^*b " dwelling " ; etc. . 

(2) Instrument. HHS " open ", Pin^p " key " ; ^3^^ 
" eat ", nSpJ^p " knife " ; 1)D " capture ", n^DD " net " ; etc. 

(3) Abstract, bb^ " complete ", hh:if2 or h'hpt: " com- 
pletion " ; adjective SSi^ " dark ", h^^^ " darkness " ; DfikT 
" to judge ", D3^p " judgment ". 

(4) Participles. Intensitive as 7t3p^, etc., and causative as 
b'tppp, etc. 

Verbs with radical semi- vowel as NX** form deverbal nouns, 
as N^ID " going out ". Verbs med. gem. as v^Tlf^, or 
"TjDp ('s/'liD) or 2DXD, i.e. first syllable opened and vowel 
increased. Initial n- as ^03 " bend (a bow) ", V^^ " weapon " 

(for *ywp). 

{d) Aramaic 

As in Hebrew the vocalization of m- depends mainly on 
phonetic conditions. 

(1) Place, time, nil "to slaughter", nS"!^ "altar" 
(Ezra vii, 17), Syriac l^ ^ ; p^ " dwell ", [^l^p, Syriac 

n»V) " dwelling " ; etc. 

(2) Instrument. |TX " weigh ", |*iT{<2D " scales " (Dan. v, ' 
27) ; l4ai " extract ", y4n\V> " forceps ". f 

(3) Abstract, ^osu " go out ", ..nak) " a going out " ; so j 
infinitives of the derived stems as intensitive q^clSij, j 
causative oV^^Vn qV^^Q*^) etc. 5 

(4) Parlid'ples. Passive I^V^ " work done " (Dan. iv, 1 


34), j1|!d As2), etc. Active U^^Id, etc. Deverbal adjective 
as ,ni " do ", ] \^K<n " active ". 

• • 

(e) Assyrian 

In Assyrian m- becomes n- before a stem containing a 
labial or labial sonant. 

(1) Place, time. — shn " sit ", maskanu " place " ; 7^ " go ", 
mdlaku " way ". 

(2) Instrument. — rkh " ride ", narkabu " chariot ". 

(3) Abstract. — rdmu " love ", nardmu " love (noun) " ; 
phr " collect ", napharu " totality ". 

(4) Participle, agent, object, etc. — banu " create ", nabnitu 
" creature ". So mu-, nu- in participles of the derived 

(/) General note on the m- preformative 

This m- preformative appears in ancient Egyptian, as in 
mh't " scales " from h' " measure ", m7iht " oil " from nh 
" anoint ", msdhr " sleep " from sdhr " to sleep ", etc. 
Cf. Coptic JUL^.- " place ", prefixed in ju.^. Itccjoft^ 
" prison " from ccoit^ " bind " ; JUL^. rt^OJTTl " west " 
from ^{xyrw. " go down ", etc. Similarly, Libyan, though 
rarely, as in moutfen " entry " from atef " enter " (dialect 
Ouargla), messa " food " from es " eat " (Tlemsen) ; and 
in Hausa to express the agent, as gudu " run ", maigudu 
" fugitive ", saki " weave ", masaki " weaver ", etc. 

Ill (ii) Preformative t- 

This preformative appears in derivatives from verbal stems 
in t- (reflexive passive, cf. sect. 140), and sometimes as a kind 
of variant of m-. 


(a) Arabic 

In common use as infinitive of the intensitive, as A^Za) 

a ^^ ** 

«.V e ^ --- 

<Llir ; hence verbal noun and deverbal abstract, as 



AJU- " dispense ", /i-i^ " dispensation ". Cf. ^j^"^ 

" prosperity ", <X Lr " danger ", abstract, where <>v__JL^ 

denotes the concrete " place of danger ". In concrete nouns, 

denoting agent, etc., as «.D J " lean on ", dl_5v-_j* " walking- 
stick ", (Ju^" "image, likeness", more concrete than Ju^ 

c ^ 

" pattern ". Some female names, as ^j*^, etc., cf. final 

fem. -at and Libyan use of fern, as prefix and suffix t-t (cf. 
sect. 115 below). 

(6) Abyssinian 

Apparently only in loan words, as ta^har " work ", te'ezdz 
" commandment ", etc. 

(c) Hebrew 

nSin "generation", from nS"l "bear chUd " ; HoS 
(intens.) " teach ", hence passive T*p/n " pupil ". 

(d) Aramaic 

^^Kc\ J " pupil " (cf. above), IjoAjsiZ " pride ", etc. 

(e) Assyrian 

mahar " oppose ", tamharu " battle " ; anahu " sigh ", 

tanihu "sighing"; talittu = Aiahic dAJ, etc. Also as 

informative (metathesis) in gitmalu " perfect " from gamdlu 
" to complete " ; ritpdsu " wide " from rps " be wide ". 

112 (iii) Preformative y- 

Such a preformative seems to occur in proper names which 

were originally sentences, as ^X — 1j , a^Xoi , etc., or in Hebrew 
which commenced with the name of God, as DHV, etc. Laying 


tliese aside, we have instances of y- equivalent to m-, as ^j^^a^ 

" be green ", jj.CiiJ^^ " verdant meadow ", or as showing 

^°- . • 1. ^ '- 

a modified sense as in » tj^cLi " timid ", where i,^j.^l^ 

means " feeble ", and a,y>s^ " black smoke ", compared with 
^jsco-* " feverish ". In Hebrew we have riD'^y " stem ", 
P|1J^y_ " owl ", niDH* " kind of goat ", etc. In Syriac ^io^)-^ = 

Arabic f'y^J " jerboa ", the derivation is presumably from 

>o J '■ XLol^ " mandrake " is a compound »sou " give " 

and ]lt09 " desire, spirit ", i.e. aphrodisiac, hence Arabic 

> ox- 
loan word 7"-?^' . 

113 (iv) Preformative Hamza 

Most often preformative Hamza simply results from a 


prosthetic vowel (cf. sect. 66), as r6' " four ", Arabic *)jl , etc., 

but there are cases in which it is a true formative, as (1) in 
broken plurals in Arabic and Abyssinian ; (2) in some 
abstract forms ; and (3) in some concrete nouns. Thus (1) 

in broken plurals we have Arabic ^i I , AZi] , AJCs 1 , etc. 

(cf. 124), and in Abyssinian 'ahqul, plur. of haqel " field ", 

'azne'et, plur. of zane' " seed ", etc. (2) As an abstract 

/o 4 

preformative w^e find it in such forms as Arabic J^L^I 

" mistake ", Abyssinian 'awdal " leading out ", from ivada'a 
" go out ", showing the trace of derivation through the 
causative formation of the verb, which probably accounts for 


most noun preformatives of this type. Hebrew ]r\'^i^ 
" strength ", or more commonly as h- (cf. the causative of 
the verb) in 7p^T} " authority ", Tj^nil " a pouring out ", 

etc. (3) As a concrete formative we find it in Arabic A^ J I 

" confused sound ", <_jLr I " stones mixed with earth ", 

Abyssinian 'agaqas " door ", Hebrew P^HNI " wages ", ^^'2'^ 
" girdle ", the A- form being more generally appropriated 
to abstract nouns, Assyrian ikribu " prayer ", from krh 
"ask", etc. 

114 (IV) Informatives 

In a certain number of stems we find an informative semi- 
vowel resulting in types (1) qawtal, (2) qutyal, (3) qatwal, 
(4) qittaui, (5) qitwall, and (6) qityal. Some of these may be 
instances of dissimilation, e.g. qawtal may be dissimilated from 
qattal (cf. 32, v, vi, above). Only the first two types are found 
with any degree of frequency, and it is only in type (2) which 
represents the diminutive that we can assign any particular 
semantic value. 

(1) qawtal, as Arabic OJj^, Hebrew 7T15 " young dove ". 
But many of this type in Hebrew are deverbals of the Po el 
group of med. gem. stems, e.g. 77)]^ " child " (cf. §§ 31, 32). 

(2) qutyal, diminutives, as Arabic c— *A; " little dog ", 
from c-AV' dog ", and Hebrew T^T " little ". 

(3) qatwal, Arabic ip'j'^i " broken down old man ". 

(4) qittawl, Arabic Jj-^, very rare alternative for J^ 
" utmost ". 


(5) qitwall, Arabic (jjtc- " weak minded " (rare). 

(6) qityal, Arabic Jj 'Xo- " short ". 

115 (V) Noun Afformatives 

These all definitely denote feminines, abstracts, collectives, 
etc., or are deverbal derivatives. 

{a) Ajformative -at, -t 

This is common to all the Semitic languages and appears 
also in Hamitic. Thus ancient Egyptian sn " brother ", snt 
" sister ", etc. So in Libyan, but there both as afformative 
and preformative, as faunas " bull ", tefounest " cow " (Siwah 
dialect) ; Bishari as tekJc " man ", teket " woman " ; Galla 
hame " good ", fem. ham-tu ; Hausa karia " dog ", kariata 
" bitch ". 

(1) As -t. Arabic^ 1 "brother", C^l "sister", ^)l 

"son", cJl. "daughter". Hebrew ^Xl^ " Moabite ", 
fem. ^''1^^^D, etc. Assyrian tah-u "good", fem. tah-tu. 
Abyssinian walad " child ", fem. walat for waladt. 

(2) As -at, the commonest form. In Hebrew, Aramaic, 
and Arabic dialect this -at becomes -a as final, i.e. if not 
protected by a suffix or annexed genitive, and so in classical 

Arabic in pause. Thus Arabic /t^^ " great ", fem. ^^^^c-, 
etc. Hebrew D*lD " horse ", DD^D " mare ", construct 


HD^D. Assyrian sarr-u " king ", fem. sarr-at-u. Abyssinian 
more commonlv -t or -et. 

(3) With 'ah, ham we find -at, as though compensatory with a 

bi-literal root, thus Arabic dU>-, Abyssinian hamdt, Hebrew 



In the examples given above -t, -at-, is a feminine afforma- 
tive, but it is also used to derive abstract nouns from 
verbs or adjectives, and similarly it is employed in the 
formation of nouns denoting office or profession ; thus 

Arabic dSll^J " piety ", \'A.:^ " office of Hahf " ; Hebrew 
T\yi^ " goodness " ; Abyssinian sandyt " beauty " (from 
sandy " beautiful ") ; Assjn-ian tab-t-u " goodness ", etc. 

Closely allied to these abstracts are the Nomina Speciei, 
deverbal nouns describing the manner of doing the action of 

the verb, thus in Arabic in the form qitlat as 4-J j " manner 

of riding ", from < J j " ride ". 


As the formative of an adjective we find Assyrian pu 
" mouth ", plat " belonging to the mouth ". 

The termination -t, -at, is also employed to derive the name 

of the individual from the collective, as Arabic ^» " cattle " 

d^^ " COW " ; Abyssinian lul " pearls ", luldt " a single 
pearl " ; Hebrew nyb^ " hair ", TTsW " single hair ", etc. 

116 (b) Afformative -y, -i 

(1) This afformative occurs most often in the formation of 
Gentile names, but is applied by analogy to other adjectival 

forms. Thus Arabic -iyy, -ly, fem. -iyyat, as ^iJli. " Syria ", 
'^ \ 

^^L^ " Syrian " ; Hebrew -iyy, -i, fem. -it, as '•^T'X'lb'* 
" Israelite ", ''llsy '"' Hebrew " ; Abyssinian -7, fem. -It, 
as harrdsi " ploughman ", more commonly in combination 
with -dw, -d, as kerestiydndwi " Christian ", 'aiydivi or 'aiydy 
" like " ; Aramaic as "^ll^oj " Roman ", ^j"^ " Egyptian ", 


etc.; Assyrian -iy {-iyu>-u), fern, -it or -ay {-aya>-d, 
indeclinable), as assuru " Assyrian ", sidund " Sidonian ". 

(2) Abstract nouns ; rare use of -I alone to denote abstracts, 
thus Aramaic ^^^p (Marti. ''Wi^lp, cf. below) " rooting out " 

(Ezra vii, 26), Assyrian -iy-u>-u, as in purrusu " decision ". 
Frequently, however, this -i is combined with -t, as in Arabic 

-lyat, -iyyat, or -ayaf, e.g. ^*1 JJ or <*Ai^ " abhorrence ", 

and A} La) or <Jl2} "intelligence", the added -iy- 

apparently making no difference in the meaning ; Abyssinian 
nesVlt " smallness " from na'asa " be small " ; Hebrew 

n''2J'J<'1 " beginning " ; Aramaic "JAjjoQISd " baptism ". 

(3) Feminine. In the 2nd person singular of the pronoun 
(cf. 79) and of the verb (cf. 146) and in the plur, -In, etc. 
(id.). Also in some Hebrew female names as HSJ' (Gen. xvi, 1), 

Phoenician '^r\'2lb " mistress " (Cooke, NSI. 13, 3), and as 

-e {<-ay) in ''^l^^ " ten " in fern, numerals 11-19 in Hebrew, 

and ''- in Samaritan fem, numerals. Aramaic -i in '•^iPli^, fem. 

' T ; T 

of r^ini^ " another " (Bib. Aram.), and so in the papyri : -e 
i-ey) in Targ. and T.B. as W^IT " Httle (finger) ", ^mH 
" new (year) ". Abyssinian -i in 'ahatti for 'ahadti, fem. of 
'ahadu " one ". 

117 (c) AfEormative Hamza 

(1) As feminine afiormative, in Arabic adjectives denoting 

colour or bodily defect, thus ^a^\ " yellow", fem. >f.\^^a.^, 
etc. Adjectives in -an make fem. in -a or -a, thus ',jiX>- 
" joyful ", fem. iVl>.; 'J\J^^ " drunken ", fem. \'JC^. 


St**, etc., 

and type qittald' for qutdl, and 'aqtila for 'aqtilat, e.g. iLsc.,^ I = 
■St "^ 

\oo^\ , etc. Abyssinian -a or -o with sufi&xed -t, as qedeadt 

" holiness ", infin. nabibot " to advise ", also with afiormative 
-n as qedsend " holiness ", and without suffix in ^afeqro " to 
love " infin. Hebrew with suffixed -t in nitJDn " wisdom ", etc. 

; T 

118 (d) Afformative -n ^^^ ^^ 

(1) Abstract. Arabic -an as jLdA>-" trembhng ", from ^Jyi>- 

" to tremble " ; jl^^^X-**. " thanks ", etc. Abyssinian -an 

as bed' an " blessedness ", or -nd as in qadsend " hoHness ". 
Hebrew -on as tinj;^ "famine" (Gen. xlii, 19), ]m« 
" destruction " ; Aram, -on, -dn as l^iyi (Ezra vi, 2), 
p3'^ (id. iv, 15) " memory " ; Assyrian as in dulhdn-u. 

Also in combination with -t, -at, as Arabic 41*1) , Abyssinian 
'abednat " madness ", Hebrew n^K^^l " shame " (Hos. x, 6), etc. 

(2) Only occasionally does -dn appear as a feminine 
afEormative in North Semitic, thus Hebrew jIH^pV 
" tortuous " (Isa. xxvii, 1), Aramaic '*.p^/tD^{ " fruitful " 
(Dan. vii, 7). ^^o- 

(3) As an adjectival formative in Arabic (j'*!^ " fooUsh ", 

^^t " tale bearing ", jLlliaff- " thirsty " ; Abyssinian 
tekurdn " black " ; Hebrew \\im " last ", [IDH " inner ". 

119 (e) Afformative -m 


Afiormative -m appears in adjectives ^Vj^k^ " hard ", 
lSi^>. " very black ". 


120 (£) Afformative -u 

Afformative -u (o-, 2o.) appears as an abstract formation 
in Aramaic, and so in loan words in Arabic and later Hebrew, 

thus /nnVV) "kingdom", Arabic O^^XJ.-*, Hebrew H^DvD 
(1 Chron. xii, 23). It seems allied to afformative -i, thus 
^7lJ (Ezra vi, 11), as equivalent to v"lJ (Dan. ii, 5). It 

appears in Syriac in the infinitive of the secondary stems of 
the verb. 

121 (B) Gender 

One of the most marked characteristics of the Semitic and 
Hamitic languages lies in the distinction of genders, a point 
which was taken by Lipsius as itself the test of Semitic- 
Hamitic kinship amongst the languages of Africa. 

Grammatical gender to us suggests sex distinction, but this 
does not appear to have been its original import, for most of 
the oldest words used to denote females are not feminine in 
form ; either the male and female are called by quite different 
names, or the same word is used to denote both male and 
female without the sign of the feminine in the latter case. This 
includes many kinds of animals of which we may suppose 
that in earlier times men had not had occasion to note a 
distinction of sex. 

Grammatically we may group gender forms under three 
heads — (i) the older type in which the feminine denotes the 
female, but is not derived from the root used for the corre- 
sponding masculine ; (ii) the nouns in which the same form 
is used to denote either sex ; and (iii) the derived feminines 
made by the addition of an afformative, not always denoting 
the female sex. Assyrian and ancient Egyptian stand alone 
in adding the feminine afformative -t, -at to all names of 
females, thus ancient Egyptian mw-t " mother " as though 


from masc. niw, although no such root exists to denote 

" father " ; Arabic jjmA» " soul ", Assyrian napis-t- with 

feminine afEormative. But, as we have seen above, the 
afiormatives -t, -at, etc., are used not only to form feminines 
but also for abstract and collective nouns, etc. It may be 
that these afformatives are relics of a " class " system such 
as still exists in the Bantu languages. 

122 (C) Number 

(a) The Plural 

Probably the oldest type of plural is that which shows the 
reduplication of the singular, or part of the singular, root ; 
thus in Somali 'ad " white ", plur. 'ad'ad ; der " tall ", plur. 
derder; and by partial reduplication, Kafa haJco "hand", 
plur. bakiJco, buse " maiden ", plur. husise ; Hausa yasa 
" finger ", plur. yasosi ; and probably in such ancient 
Egyptian forms as ^VV " vine ". Some traces of such forms 
survive in Semitic as may " water " in Hebrew, construct 
plur. 7neme (also me) ; so ^^e " mouth ", plur. pipi-oth 
(Ps. cxlix, 6). So Aramaic rob " great ", plur. p^in 
(Cooke, NSI, 63, 10). 

A second type of plural occurs in the use of the singular 
as plural without change, i.e. a singular used collectively 
as Hebrew 'dddm " man " for mankind. 

A third type of plural is the collective proper derived from 
the singular by the addition of an afformative, for which 
purpose the same afformatives -t, -at, -y, etc., as used as for 
the feminine and abstract. 

A fifth kind of plural is formed by internal modification 
(broken plural), which is derivation by vowel change instead 
of by the addition of a formative. In fact, this implies the 
use of some stem form which is collective or abstract as a 
plural, e.g. qutul (cf. 109) as plural of qatdl, qitdl, qutdl, qatll, 


qatul, etc., as firds "bed", -phii. furus. This particular kind 
of plural, or the use of a collective formation in place of a 
plural, is principally developed in Arabic and Abyssinian. 

123 (b) Plural formatives 

The plural or collective formatives are substantially the 
same as those which we have already seen employed to form 
the feminine, abstract, etc. Thus : — 

(i) -at, -t : collective, as Arabic kam' " mushroom ", plur, 
ham' at ; Hebrew dag " fish ", plur. ddgd, cf. 'dyebeO " company 
of exiles " (Mic. vii, 8). All these in -at are essentially 

(ii) -y, -I : Abyssinian, plurals in -i before suffixes, thus 
plur. 'abaw " fathers ", suffix 'abaw-i-hu " his fathers ", etc. 
This -I is added to plurals of every formation before the suffix. 
Hebrew and Aramaic construct plur. -ay, unaccented -e 
(cf. 49). Assyrian -e appears as a plural formative, but in 
reality this is the ace. -gen. termination which in later 
forms of the language is used more or less carelessly for any 
case (cf. 126). 

(iii) -dn : Arabic as in 'ihiddn " servants ", insdn " men " : 
as qitldn, plural of qatl, qitl, qutl, qatal, qutal, qutdl ; as qutldn, 
plural of qatl, qitl, qatal, qatil, qdtil, qitdl, qutdl. Abyssinian 
-dn as regular plural of masculine adjectives, names of trades, 
occupations, etc., as qasisdn " priests ". In Tigre as -dm. 
Hebrew -on in collective 'izbon-lm " traffic " (Ezek. xxvii, 
19). Aramaic, chiefly as plural of nouns denoting rank or 
condiments, as rawr^hdne " magnates ". Assyrian ildni 
" gods ", etc., but here again -dni is properly the termination 
of the oblique cases (cf. 126). 

124 (c) The Broken Plurals 

The broken plurals are noun formations of the type described 
in sect. 109, originally collective, which have to a great extent 



displaced the ordinary plurals in South Semitic, This is most 
obviously the case in those which have the types qatal, qital^ 
qutal, from which the nouns denoting individual members are 

derived by the addition of afiormative -at, thus (_J) j "knees ", 

sing. A<j J . Some others {qatalat, qitdlat, qitlat, etc.) show the 

afiormative employed to denote the collective group. So 
the very common broken plural of the type qatl is most often 

purely collective, as a j,d " people ". Plurals of this kind are 
most frequent in Arabic and Abyssinian, and appear also in 
SabaBan as D^l^DHJ^ " fruits ", sing. C'lDH (D- is mimation, 
of. sect. 132), on'?") " children ", DhD'n^^ " feudatories ", 
sing. D/^*1i^. Although collective formations occur freely in 
North Semitic, we find no developed use of collectives as 
replacing plurals there. 

125 (d) The Dual 

The dual does not appear to be of very early date, as it is 
obNaously secondary and derivative. On the other hand, it is 
in the process of decay in all the Semitic languages. Its 
characteristic is the suflB.x -arj. In Arabic masculine nouns 
this shows in nominative -ay>-d, and in the oblique cases 
-ay ; thus construct, nom. -a, gen. ace. -ay, otherwise nom. 
-dni, gen. ace. -ayni ; and so in the feminine. With this we 
may compare ancient Egyptian -y added either to the plur. 
-w or to the fern. -t. The added nunation gives -dna>-dni 
by dissimilation, the oblique cases -ay-ni by analogy with the 

In Abyssinian rare traces survive in which -ay>-e (cf. 49), 
thus keVe " two ", cf. Arabic kildni " both " ; 'ede- " both 
hands " before suffixes only. 


Hebrew has -ay, whicli appears in the construct, and so 
yaday " two hands " (Ezek. xiii, 18), rarely -e as in ^ne " two " 
in the combination sne 'dsdr " twelve ". With mimation 
yddayim " two days ", etc. Like the ancient Egyptian and 
Arabic the fern, dual adds this -ay to the fern, -at, thus sdfd 
" lip ", dual s^fdthayhn, construct sif^the. 

Aramaic shows -ay in the two words y^day " hands ", 
raglay " feet ". Syriac has -e, which appears only in t^ren 
" two ", fem. tarten, and mathen (_i2]!iD) " two hundred ". 

Assyrian in the old and middle forms of the language has 
masc. -d (from -ay), oblique cases -e, fem. -t-d, -t-e. These 
occur rarely in the later forms of the language. 

In modern Arabic dialect adjectives have entirely lost the 
dual termination, which is obsolescent throughout, and usually 
the feminine singular is employed in qualifying a dual noun, 
thus raglen talyaniya " two Italian men " (Egyptian dialect), 
or the adjective may be in the plural, as ir-raglen taiyibin 
" the two good men " (id.). 

In Hebrew an adjective qualifying two nouns is in the 
plural, not dual, as in Gen. xl, 5. 

126 (D) The Cases 

Three cases appear in Semitic, and are distinguished by 
vowel endings, nominative -u, genitive -i, accusative -a, 
and there are traces also of a fourth, the adverbial case, whose 
vowel ending was -u like the nominative. The genitive -i 
suggests comparison with the formative -iy-, -i (cf. 121), 
as denoting " belonging to . , .", etc., in adjectives. The 
accusative has been associated with the demonstrative ha 
(cf. 89), and actually appears as -hd in Abyssinian proper 
names (cf. below). With rather less confidence the nominative 
has been connected with the personal pronoun hu. 

Originally these case endings seem to have been long in 
quantity, and are generally so preserved in the plural and as 


attached to the words 'ab " father ", 'ah " brother ", ham 
" father-in-law " (cf. 108) before suffixed pronominal forms 
(cf. 82). 

The general history of the Semitic case endings shows 
(i) the shortening of the vowels in the singular, (ii) a tendency 
to discard short final case endings, and (iii) in certain languages 
a tendency to confuse some or all of the cases. 

(a) Arabic 

In classical Arabic all three cases appear as -ii, -i, -a, save 
with the words 'ab, etc., and a few others which retain the 
long vowel endings before suffixed pronouns, as nom. 'abuk, 
gen. 'ahik, ace. 'abdk, etc., and short case endings also 
appear with the -at of the fem. plur. In the sing, nouns 
ending in -iiv, -iy (not -iyij or -iy) the nom. -iyu becomes -i 
and thus assimilates to the genitive so that we have only 
the two case endings nom. gen. -i, ace. -a, as in qadi " a judge "; 
whilst others ending in -aw, -ay show -a for aU three cases. 
In the plural the accusative has been entirely absorbed in 
the genitive, leaving only the two case endings nom. -u, 
gen. ace. -1, and this has been reproduced in the dual (cf. 125). 
The adverbial case is preserved as -u in the adverbs qablu, 
*' before ", tahtu " under ", etc. 

In vernacular Arabic the short case endings are lost ; even 
before consonant suffixes they are commonly replaced by 
inserted vowels as described in sect. 68. It has been noted 
that 'ab, 'ah, etc., retain their case vowels long before 
consonant terminations but not before 1st sing, -i in classical 
Arabic. In dialect, as a rule, one of the case vowels is retained 
and serves for all cases, and sometimes this is employed even 
before the 1st sing., which then appears as -ya or -y (cf. 82). 
In the plural the gen. -i serves for all cases. Thus in dialect, 
even where some of the case endings are preserved, all dis- 
tinction between cases is lost. 


(&) Abyssinian 

Abyssinian preserves ace. -a, the gen. -e (for -i) having 
fallen away as a final but being retained before suffixes, as 
negUs-e-ka, negus-a-ka " thy king ". Proper names, however, 
retain the full accusative -hd, which may be the original form 
of this case. The nouns 'ah, 'ah, ham, show gen. -u and 
ace. -a. The true nominative -u and ace. -a are preserved 
in the numerals 1-10, thus masc. nom. 'ahadu, ace. 'ahadd 
" one ", fem. ace. 'ahatd (for 'ahadta, cf. 23), and nom. fem. 
-ti. The adverbial case appears as -u in the adverbs tahtu 
" under ", qadlmu " formerly ", etc. 

(c) Hebrew 

lu Hebrew the accusative -a is preserved in the sense of 
direction towards, as l^ma'ld " upwards ", Bdbeld " towards 
Babylon ", baydd " homewards ", and so even with a preceding 
preposition as mihbdbeld " from the direction of Babylon " 
(Jer. xxvii, 16), and in the sense of time in 'attd " now ", 
and miyydmim ydmimd " from year to year ". A more 
direct use of the ace. termination appears in Isa. viii, 23, " he 
lightly afflicted the land {'arsd) of Zebulun and the land (id.) 
of Naphtali." 

The nouns 'ah, 'ah, ham, preserve the genitive ending and 
no other before consonantal suffixes, as 'ablka, etc. The gen. 
-1 or -e is the only case ending retained in the plural where 
we find -1m and construct -ay, -e (cf. 49). Genitive -I occurs 
also in Ps. ex, 4, 'al-dibrd6l Malki-sedek " according to the 
order of Melchisedek ", and sometimes is added to the 
construct as b^ni '^6dn6 "his ass's colt" (Gen. xlix, 11), as 
well as in such proper names as Gabri-el, and Punic Hannibal 
(" favour of Baal "). 

Nominative -u is found in some proper names as P^nu-'el, 
Punic Hasdru-hal, and as exclamatory (vocative) -o in b^no 
Sippor (Num. xxiii, 18). The adverbial case appears as -6 


with added mimation (cf. 132) in silsdtn "the day before 
yesterday ". 

(d) Aramaic 

Nominative -u appears with the nouns 'ab, 'ah, ham before 
suffixes. Gen. -i {-in, -e, -ay) is, as in Hebrew, the one case 
ending retained in the plural. It is possibly preserved in -e- 
before suffixed -k of the 2nd fern. sing, and -h of the 3rd 
masc. sing. Accusative -a- is perhaps the connecting vowel 
before masc. sing, -k and before the plural suffixes, but these 
inserted vowels may have a merely phonetic origin, as 3rd 
sing. fern, -ah for -ha, 2nd -ek for -ki by metathesis. 

{e) Assyrian 

The three case endings are employed in the older 
Babylonian, but at an early date they began to be used 
without careful discrimination, and in neo- Assyrian all are in 
use but are confused without any adherence to their original 
meaning. The adverbial case appears in such forms as ^sejiiCa 
" at my feet ". 

127 (a) The Construct 
The Arabic grammarians describe the relation between a 

noun and its dependent genitive as " annexation " (^dL^I). 

In it are two elements, the ruling noun which may be in any 
case according to the part it plays in the logical meaning of 
the sentence, but has certain peculiarities of form v/hich are 
expressed by stating that it is in the " construct ", and the 
genitive annexed to the regent or ruler. 

The annexation forms a very close tie between these two 
nouns 80 that normally no other word can intervene between 
them, although this rule has exceptions, especially in poetical 
and rhetorical expressions. Thus, if the noun has a qualifying 
adjective, which normally follows the substantive it qualffies, 


this adjective must be deferred until after the genitive so 
that " the glorious book of God " becomes " the book-of- 
God the glorious " ; so if two words depend on the same 
genitive, the second is normally deferred and the genitive 
repeated to prevent the second word intervening between the 
first regent and its genitive, thus " Zayd's sword and spear " 
becomes " the sword of Zayd and his spear ", and so we have 
" the sons of David and his daughters " or " the sons of David 
and the daughters of David", not "the sons and daughters 
of David ", although we do get this latter form in poetical 
passages such as " the knowledge and fear of the Lord " 
(Isa. xi, 2). Similarly, a construct should not depend on two 
genitives because the first genitive intervenes between the 
construct and the second genitive, so " the creation of heaven 
and earth " becomes " the creation of heaven and the creation 
of earth " and " the God of the heavens and the God of the 
earth " (Gen. xxiv, 3), and " the bones of the kings of Judah, 
and the bones of his princes, and the bones of the priests, 
and the bones of the prophets, and the bones of the inhabitants 
of Jerusalem " (Jer. viii, 1) ; but we do find " the Lord God's 
creating of earth and heavens " (Gen. ii, 4). 

In annexation the construct comes first, the genitive 
follows, and the two are so closely allied that they are regarded 
as one word. In feminine nouns, for example, where the 
termination -at has become -a (Hebrew, Aramaic, Arabic 
dialect) as final (cf. 17c), the -at is preserved by a suffix or 
by an annexed genitive. In the construct a noun is defined 
by the following genitive, and therefore (i) it does not take 
the article as it is already determinate (cf. 132), and (ii) it 
cannot possess, the indefinite nunation or mimation (cf. 132), 
and so we have Arabic construct qassdhu from qassdbun, and 
plur. qassdhu not qassdbuna, and so Hebrew construct plur. 
malke for malkim, etc. We are thus unable to use the 


genitive to express an indefinite regent, and " a daughter of 
the king " must be expressed by the help of a preposition, 
" a daughter to the king " or " a daughter from the daughters 
of the king ". 

In classical Arabic, early Babylonian, and Abyssinian, the 
case endings are more or less preserved, but the general 
tendency to lose or confuse these endings caused other 
languages either to lay greater emphasis upon the other 
characteristics of the annexation or to introduce accessory 
elements. The former alternative appears in modern Arabic. 
In the genitive proper, known to the grammarians as the 
logical or true genitive, the construct cannot have the article 
whilst the dependent genitive must be defined either by the 
article, or by a possessive pronoun, or by its own nature as 
a proper name. Thus the external mark of the annexation 
is now the undefined noun immediately followed by the defined, 
e.g. hubz el-walad "the boy's bread", or hubz Muhammad 
" Muhammad's bread ", the genitive having no other ex- 
pression than the fact that it is a strictly defined noun 
immediately following a noun not formally defined, and the 
freedom from any word intervening, etc., must be rigidly 

128 (b) The relative as denoting the genitive 

In Hebrew the genitive relation is generally clear enough 
from the fact that the dependent genitive immediately follows 
the ruler, and because the ruler very often, but not always, 
shows that it is in the construct by a difference of form due 
to the shifting of accent and modification of syllabic form as 
the result of close annexation. But this was not always free 
from ambiguity, and in later Hebrew we find a tendency to 
insert a relative followed by the preposition 1-, a circum- 
locution used in older Hebrew to avoid possible confusion 
when a genitive immediately follows another genitive, as " the 



chief of the herdsmen of Saul ", which appears " the chief of 
the herdsmen which (are) to Saul " {im U^^n n^3i^ 

h)i<^h, 1 Sam. xxi, 8). In later Hebrew this S im 
contracts to yp by assimilation (cf. 29). The relative 
itself, which appears as prefixed ^ in Ecclesiastes and 
Song, did not become a genitive prefix in Hebrew without 
the following preposition 7, but it appears thus alone in 
Punic (cf. Cooke, NSI. xxxix, 2 ; xli, 2). These are later 
developments, but in Assyrian the use of the kindred sa as 
a relative pronoun and also as a genitive preposition appears 
very much earlier, thus siru sa Bel " priest of Bel " in the 
tablet of Ramman-Nirari, circ. 1325 B.C. 

Whilst the relative sa, '^ser, thus developed a use as a 
genitive preposition in Assyrian, Hebrew, and Punic, the other 
relative da, etc., developed a similar use in Aramaic and 
Abyssinian. In Aramaic this appears as prefix d"-, da-, which 
is used both as relative and genitive preposition, as 
VLl Icjil^j oiJq " son of the Hving God " (St. Matt, xvi, 16), 
very often, as here, with the possessive pronoun sufiixed to 
the construct. So the Abyssinian relative sing. masc. za, 
fern, 'enta, plur. 'ella is regularly used as a genitive prefix, as 
'elat 'ahdy 'enta kwenane "the great day of judgment", etc., and 
sometimes a parallel use of de, da, etc., occurs in Morocco, 
South Arabic, and with li in Mehri. 

All the Semitic languages except Arabic thus developed 
a genitive use of the relative by which " king of Egypt " 
became " king who [of] Egypt ", thus making the relative 
the construct in apposition with the ruling noun. 

In all the Semitic languages without exception a prepo- 
sitional construction with " to ", " from ", or " in " can 
generally be substituted for the genitive, although this is 
done more freely in Abyssinian than elsewhere, thus Hebrew 
hassofim l^sd'id " the scouts (belonging) to Saul ". This is 


the usual construction when the genitive stands as the 
predicate, as Arabic hadhd al-kitdbu U " this book (is) mine ". 

129 (c) Other means of avoiding the direct construct 

Sometimes the direct use of the construct is avoided by 
means of a noun in apposition used in the construct, a device 
very common when the genitive qualifies and does not 
determine, as in nick-names, genitive abstract replacing an 
adjective, etc. This is parallel with the use of the relative 
which we have already considered, and in later forms tends 
to produce a genitive preposition. Thus for " the man of 
sin " we may have " the man (son) of sin ", etc. There are 
certain words thus employed very commonly in Semitic, and 
the convenience of these is that in such a form as " the man 
(son) of sin " the peculiar construct form is taken by " son ", 
so that only these standard constructs need be used without 
afiecting the stem of the noun which is the real regent. 

The commonest words of this kind are : (i) " father " or 
" mother ", and other names of relationship, as " son of six 
hundred years " for " six hundred years old " (Gen. vii, 6), 
" sons of the poor " (Ps. Ixxii, 4), etc. (ii) " Master " or 
" owner ", Arabic masc. dku, fem, dM, akin to the relative 
pronoun, Hebrew ha'al, Aramaic h^^el, Abyssinian hd'la, 
Tigrina he'dl, Mehri 6a7, as in Hebrew ha'al se'dr " a hairy 
man " (2 Kings i, 8), etc. ; and Aramaic res, Arabic rd% 
(Oman), and Tigrina rdH. (iii) " Man " or " people ", as 
Hebrew 'anse-rd' " bad men " (Prov. xxviii, 5), etc. 

From these have arisen several quasi-prepositions governing 
the genitive and not requiring the regent noun to be placed in 
the construct ; such are, in modern Arabic, betd' {metd') 
" owner " in the dialects of Egypt and Palestine, taba' 
" belonging to " (Palestine), sit (Damascus), mdl ('Iraq), 
hagg or haqq (Hadramaut). 


130 (d) Later forms of the accusative 

The short vowel case endings are obsolete in modern Arabic 
save for some rare survivals as in Moroccan al-yuma " to-day ", 
and so in Hebrew and Aramaic. For the most part the object 
case is denoted only by ^position, the object following the 
subject ; in modern Arabic this position is very commonly 
emphasized by placing the subject before the verb, the 
object after it. 

The accusative of the personal pronoun is properly 
expressed by the pronominal suffix (cf. 81), but we also find 
it denoted by an accusative particle with the pronoun suffixed, 
thus in Arabic 'iyyd, as 'iyyaka " thee ", etc. In the 
Abyssinian of Tigrina this appears as 'et-iy-, e.g. 'etiyu, etc., 
and in Hebrew as '6th-, e.g. 'o^o " him " (Mishna inii^), 
Phoenician yth. In Hebrew, however, it is not confined to 
the accusative of the personal pronoun, but is extended to 
accusative substantives defined by the definite article, by a 
suffix, or in annexation with a following genitive, or as being 
a proper name, in all such cases appearing as 'eth or 'eth-. 
The same particle appears in older Aramaic as H^J^ (papyri), 
later yath (Dan. iii, 12, etc.), and so Syriac, where we also 
find wath in the combination 'akwath- for ka-wath- followed by 
a pronominal suffix denoting " like ". The corresponding ^ 
form in Assyrian is k-iy-ya- with suffix as kiyahu " him ". ^ 

131 (e) Adverbial use of the accusative 

Although there are traces of an adverbial case in -u, as in 
Arabic qattu " ever ", ba'du " afterwards ", etc., we find also 
an accusative form used adverbially chiefly in Arabic, thus 
'abadan " ever ", dahilan " within ", ma'an " together ", 
'aldna " now ", kayfa " how ", etc., the indefinite nunation 
being added (cf. 132). So Arabic laylan " by night ", 
sabaha masa'a " every morning and evening " ; Abyssinian 
ahatta ellata " one day " ; Hebrew atta " now " (Joshua xiv. 


11, -eth " time ") or " presently " (Job vi, 3) ; Assyrian uma 
" now ", eninna " now ", matima " at any time ", la sakipu 
musa u imma " not resting night or day " (King, First Steps, 
p. 127). 

And so, generally, in all the Semitic languages the accusative 
is used adverbially to express time how long, time when, and 

manner, circumstances, etc. ( Jl>-), as Arabic Jcarra zayd-un 
'asadan " Zayd charged (like a) lion ", etc. 

132 (E) Determination and Indetermination 

Determination means the definition of a noun as denoting 
a particular person or thing as distinct from the common noun, 
which is applicable to any member of a class. Thus the proper 
noun is determined by its own nature ; but common nouns 
may become determined in various ways. 

(a) A noun may be determined by the use of a demonstrative, 
either (1) " this " designating the thing referred to as the 
one near at hand, or (2) " that " designating it as the one 
remote, both of these implying a pointing out of the thing 
referred to, or (3) it may be defined by the article " the ", 
which implies that it is already known. For the two former 
demonstratives cf. §§88 sqq. (above). It is here sufficient 
to note that Semitic generally uses these in combination with 
the definite article, where that article exists. 

For the definite article Arabic uses the demonstrative -l- 
(cf. 91), Nabateean Sn (Cooke, NSI. 105). Hebrew uses the 
prefixed demonstrative ha- (cf. 89). In Aramaic this appears 
as suffixed -a, the so-called " emphatic " form, but here the 
determining force is generally lost. In later Aramaic and 
Syriac we often find a redundant use of the pronominal suffixes 
as equivalent to the article, and this is of regular occurrence 
in the Philoxenian version. In Samaritan both prefixed ha- 
and suffixed -a occur, never, of course, together, but the 


latter is much the more common. No definite article appears 
in Assyrian or Abyssinian save in Tigre dialect, where we 
find la-. Like Aramaic, Abyssinian often conveys the 
determining or demonstrative sense by the use of a redundant 
suffix, as be SI " man ", he'slhu " this man ", etc. 

(6) A noun is also determined by the use of a possessive 
suffix, as " my book ", which restricts the common noun 
" book " to one individual article known and defined. 

(c) The construct is also defined by the following genitive. 
Thus the common noun " house " becomes definite and 
individual in the phrase " the house of the king ". English, 
indeed, uses the definite article " the " before " house " in 
this expression, but Semitic does not attach the article to 
the construct, as the noun is sufficiently defined by the 
following genitive (cf. 127). 

Indetermination means that the noun is left indefined, and 
so is applicable to any member of a class. Anciently the 
indeterminate was marked by the addition of -n (Arabic, 
Aramaic, Abyssinian) or -m (Hebrew, Assyrian) to the 
termination, but where the short final vowels have fallen away 
this suffixed njm has also disappeared. Thus classical Arabic 
halb-un " a dog " appears as kalb in modern dialect and 
keleb in Hebrew, etc. In the plural the nim is retained after 
the long vowel in all indetermined nouns and in nouns 
determined simply by the article, but disappears when they 
are defined by a following genitive or by a suffixed pronoun. 

This added -m {-n) is perhaps from ma " something " used 
as a sign of the indefinite (cf. 101, 102). In Assyrian it 
early ceased to convey an indeterminate sense, and there, as 
well as in Hebrew and Aramaic, the only trace we find of its 
indeterminate character is its omission from the construct 
plurals. Mimation occurs also as -am in Hebrew singulars as 
'dmndm from 'omen, hinndm from hen, reqdm from req, and 



adverbially in yomdm " to-day " (Deut. i, 39). In Syiiac 
^ the indeterminate sense is so completely lost that it can take 
the suffixed -a, originally a defining article, as imdmd 
" to-day ". 

133 (F) Comparison o! Adjectives 

In Hebrew the comparative is marked simply by the use of 
the preposition min " than ", as vK^^ID HnX DDH " thou art 
wiser than Daniel " (Ezek. xxviii, 3) ; and the superlative is 
most commonly denoted by kol{l) " all " after min, as " the 
serpent was the most cunning of all (= cunning than all) the 
beasts of the field " (Gen. iii, 1). Sometimes 6(a)- " in " is 
used for min, as " the least of the nations " D*1X3 |bp 
(Jer. xlix, 15), or the adjective is in the construct 
followed by the substantive in the genitive, as D^pn^H *]3pT 
" the oldest of the priests " (Isa. xxxvii, 2). Akin to this is 
the construct adjective followed by the same adjective in 
the construct plural, as Wp'l'^T^ tJ^"lp " holy of hoHes ", i.e. 
" most holy ", and so with indetermined substantives, as 
Dn^y "liy " servant of servants ", " lowest servant ". The 
superlative may also be expressed by the use of some term 
denoting primary importance in the construct with a plural 
adjective or descriptive substantive in the genitive, as 
^nnpb^ 2r■^^n "chief. of my joy" = "my chief joy" (Ps. 
cxxxvii, 6). 

In all these cases, it will be noted, there is no morphological 
type expressing the comparative or superlative. 

In Aramaic we find closely parallel constructions. The 

comparative is denoted by the ordinary (positive) adjective, 

, followed by the prep, men " than ", often reinforced by tdh 

^ or yathir ; the superlative is expressed in the same way as in 

Hebrew. So in Abyssinian we find the comparative denoted 


by the use of ^emna (Tigre 7nen) after the adjective, the super- 
lative by 'emna kuellu " than all ". 

In Arabic the comparative and superlative take the form 
aqtal-u, as 'ahsanu " more beautiful ", etc. As comparatives 
these are followed by min " than ", the same forms becoming 
superlative by the definition of the following substantive, 
either by the use of the article or by the construct and genitive, 
as huwa 'afdalu ragulin " he is the best of men " or ar-ragulu-l- 
'afdalu " the best man ". Certain other forms, qatl, also appear 
as comparatives. The association of these particular noun 
types with the comparative-superlative is probably a later 
development, as it has no parallel in the other Semitic 
languages ; similar forms appear in Hebrew, as 'akhzdb 
" lying ", 'akhzdr " fierce ", but without any trace of com- 
parative or superlative meaning. 


134 (i) The Verb Stem 

In verb forms we may distinguisli four elements— (1) the root 
whicli is normally a framework of three consonants in itself 
neither verb nor noun ; (2) the formatives which appear as 
consonantal additions to the root modifying its meaning in 
derived formations ; (3) the vocalization which is partly 
employed as a means of forming a noun or verb from the 
root with or without added formatives, partly connected with 
the formative additions, and partly due to phonetic require- 
ments, in order that the resultant word may have a syllabic 
form which can be pronounced ; and (4) the additions, suffixed 
or prefixed, which are used to express persons, moods, etc. 
Although the vowels are partly formative, and partly phonetic 
additions, it is convenient to treat the vocalization under one 
heading, as the resultant is modified by the same phonetic 
conditions. Thus the Arabic (conj. x) istaqtala contains the 
root q-t-l, the formatives s- and t-, the verb stem has its vowel 
-a- as s-t-q-fal-, the formatives are properly vocalized sa-, 
ta-, but the resultant sataqtal- becomes -staqtal-, the first 
consonant being vocalized either by the final vowel of a 
preceding word or by a prefixed i-, and if the word comes 
after pause by i- preceded by Hamza (cf. 66) to this {i)staqtal- 
the personal ending -a, -ta, etc., is added. 

(i) The Root 

The root is a consonantal framework from which the verb 
or noun stem is formed. Normally this root appears as three 


consonants. Sometimes the number is apparently reduced 
by the assimilation of one or more of these three, or by the 
quiescence of one if it happens to be a semi-vowel ; and some- 
times we find four or more consonants, either by the addition 
of a formative which has become permanently attached and 
whose original character has been lost sight of, or by the 
combination of two roots with or without subsequent eUsion ; 
or by additions whose nature can no longer be accurately 
accounted for, though it seems most probable that it originally 
was due to a formative or to composition. There has been 
much discussion as to whether a bi-consonantal root may not 
lie beneath the three consonant skeleton which now appears 
as the framework of most Semitic verbs, and it seems very 
probable that this is the case. It will, however, be best to 
reserve this until we discuss noun stems, which show a very 
wide diversity of structure and include some clearly bi- 
consonantal ones : whatever may have been the case in proto- 
Semitic or in the parent language, which we may call 
pre-Semitic, the common Semitic verb form shows definitely 
a tri-consonantal norm which must be presumed to have 
arisen at a very early stage, and which is now characteristic 
of the Semitic languages. Unlike the noun, the verb does not 
show a great variety of stem vocalizations ; for the most part 
these run through a very limited number of variations, and 
these are confined to the primary theme (Arabic conj. i, 
Hebrew Qal). 

135 (ii) Verb Themes 

The several formations of the verb stem are commonly 
known as conjugations, a word which is not altogether con- 
venient, as they do not at all correspond with the conjugations 
so called, for example, in the Latin grammar, and they are 
represented by parallel formations in the Indo-European 
languages, to which the name of conjugations has never been 



applied. Frencli writers call them themes (e.g. Meillet, Les 
Langues Indo-Europ., 4th ed., Paris, 1915), whilst German 
writers call them Stammformen (e.g. Brockelmann, Grundriss, 
etc., Berlin, 1908), and some such term is distinctly preferable 
to " conjugation ". For the purpose we shall employ the name 
" themes ", because " stem " implies a vocalized and there- 
fore more fully developed formation than that which we have 
in view at the moment ; thus q-t-l is the root, qatal-, qattal- 
are stems, but we shall give the name of theme to the root 
with (or without) the addition of a formative previous to 
vocalization, as q-tt-l-. In the Indo-European languages also 
we have verb themes or stems composed of root and formative, 
but there those themes denoting the inceptive, causative, etc., 
have not developed in use like the tenses, moods, and voices, 
but seem to have suffered a check at a fairly early stage. 
Whilst verbs generally have moods, tenses, etc., and transitive 
verbs have different voices, only a certain limited number can 
form inceptives, causatives, etc. — it is no regular part of a 
verb's inflexion. In Semitic the case is otherwise. The 
Semitic tense development did not expand freely : the two 
original Semitic tenses alone appear until we reach the period 
of decay in Aramaic, modern Arabic, and ancient Egyptian, 
which seems to have had a premature development so early 
and so pronounced as to put it out of line with the rest of the 
Semitic languages. On the other hand, the themes or derived 
stems are very much more regular and in general use than is 
the case with the Indo-European languages. 
The themes follow five leading types : — 

(1) The primary theme, which shows the root vocalized 
without the addition of a formative. 

(2) The intensitive, derived from the primary by doubling 
one or more of the radical consonants. 

(3) The causative, with preformative s/s or hj\ 


(4) The passive, with preformative n-. 

(5) The passive-reflexive, with preformative t-. 

Not all these are in equally vigorous life. The n- passive (4) 
is obsolete in Aramaic, save in a few very early instances, 
and nearly so in Ethiopic ; whilst in Hebrew the passive- 
reflexive is generally confined to the intensitive instead of 
being freely combined with any of the other themes. On the 
other hand, in some cases, and notably in Ethiopic and 
Assyrian, there is much freedom of combination by which 
means a large variety of stems is produced. Very often the 
primary theme itself does not exist, but it is very rarely that 
the intensitive is missing, even though the primary may be 
obsolete. Under the intensitive we include the conative 
(Arabic conj. iii), as it seems to be derived from the intensitive 
by a phonetic change, and it is only in Arabic that it has fully 
developed its specialized meaning. 

136 (1) The Primary Theme shows the plain root from 
which the stem is formed by vocalization. This will be 
treated in due course (cf. below). 

137 (2) The Intensitive Theme is formed by doubling one 
or more of the radical consonants. 

(a) Theme q-tt-l 

This is the commonest type, with doubled medial, and this 
appears in general use in all the Semitic languages. It is used 

to express the intensitive, as Arabic A"9 " kill ", \1a 

" slaughter ", the act or state being extended to many objects, 
or done by many agents, or often repeated. This intension 
or extension seems to be the basal meaning of the form, but 
it is also employed to derive a verb stem from a noun form, 

as Arabic *\^*>- " tent ", ^^>- " pitch a tent ", etc., and so 
has developed a semi-causative sense which encroaches on 


the causative proper. Thus it commonly happens in all the 
Semitic languages that an intransitive verb becomes transitive 

x" »»'  


when it is put in the intensitive, as j?- ^ " be glad ", -?- ^ 

" gladden ", and also that a transitive verb becomes doubly 
so, i.e. governs two direct objects, when it is put in the ^ 

intensitive, as ^c- " to know ", ^ic- " to teach ". In North 

Semitic (Hebrew and Aramaic) a laryngal or r is incapable 
of doubling, and so, when such a consonant stands as medial 
radical, the doubling theoretically takes place, and then one 
member falls away with compensatory lengthening of the 
preceding vowel, thus primary 'll^, intensitive *birrekh> 
berekh (^1^). In neo-Syriac this and all other derived themes 

add preformative m^-, as ..^^aSo " put out " from *4^ " S^ 
out ", although in the dialect of Urmi, etc., this m- is silent 
(cf. Maclean, Vernacular Sijriac, sect. 35 (5)), and in East 
Syriac there is a strong tendency to ignore the doubling of the 
medial (id. 87c). This doubling of the medial is the most 
usual intensitive form in ancient Egyptian, as sdm, sddm 
" hear ", etc., and occurs also in Libyan tri-consonantal roots, 
as Kabyle ekrez " boil ", intensitive kerrez, eknez " scratch ", 
kemmez, etc., and appears in a slightly diSerent form in such 
words as egz " fold ", intensitive aqaz, del " cover ", intensitive 
dal, where gg regularly becomes qq and dd becomes dd. 

In Arabic verbs med. gemin. we sometimes find an 

alternative form with elision of the third radical, as ^j^"^ for 

jya>A9 " to plaster ". 

(6) Theme q-t-l-l- 

Doubling of the final radical. In Arabic this is specialized 
as conj. ix, iqtalla, and conj. xi, iqtdlla, chiefly of verbs denoting 


•i ^ 

colour or bodily defect, as >j^ 1 or ^\j^ I " be black ". An 
uncontracted form {iqtalala) occurs with some final w/y verbs, 

as (Sj^J ' " refrain ", and so Ethiopic sardada and the 

forms of the quadriliteral verb as 'akmosasa " ridicule " for 
*kamasasa, etc. 

In Hebrew we find pVI " be green ", \y^^* " be at rest ", 
and similar forms, and in the Amarna letters ushihin, istahihin. 
Aramaic 211)^ (Targ. Jer. i, Gen. xi, 9), and hence Tabu. Jer. 
pnnynD (Keth. 25a), NHiinyD (R. Peth. 23). Assyrian 
usparir, suqalulu. 

(c) Theme q-tq-l 

Chiefly with mediae gemin. verbs, or those having medial 

semi-vowel as Arabic \^\ from \ , etc. Ethiopic lamolama 

" shoot out buds or leaves ", Amharic lamalama, etc., with 
-a- inserted in the medial group as is usual in Amharic. 
Hebrew hh^ " roll " (Jer. Ixi, 25) from hbx V^W " delight " 
(Ps. xciv, 19) from yj^SJ^, and from hollow verbs 7tD7D " cast 
forth " (Isa. xxii, 17), SidS^ " sustain " (Jer. vi, 11), Ip^D 
" undermine " (Num. xxiv, 7 ; Isa. xxii, 6), etc. In Bib, 
Aram, we find iniH " conceive " (Dan. iv, 2), and so in 
Sjrriac JjqqSqo, etc. In the Targums and T.B. it is a very 
common form from hollow verbs, as yiyi " agitate " (Targ, 
Jer. i, Gen. xxxii, 25), THn " suppose " (T.B. Sabb. 8^^), 
b^p " starve " (T.B. Sanhed. 23c, Abod. Zara, 42a), also 
from J^'y as TT/T (T.B. Keth. 50a), and from initial w-verbs 
such as nnnna " thrown" from -)n:i (T.B. Ter. 466), btht:if2 
" agitated " (T.B. Joma, 406, from 7tDi). Similar forms occur 
very frequently in neo-Syriac (Maclean, Vernaxi. Syr., 83a), 
but always with preformative m*- as ]l«^V«^<-n " bubble ", 


U»v-\vi^ " make round ", J>Q**.LQ>jiD "get warm ", etc. It 
appears also in Assyrian, chiefly from med. gem. verbs, as 
dandanu from dnn " be very strong ", etc. 

This form has an interesting parallel in ancient Egyptian, 
where we find verbs containing a serai-vowel radical 
reduplicated and then reduced by elision of the semi-vowel ; 
thus ghy "be weak", intensitive *gbygby>gbgb : such a 
treatment applied to a root with a medial semi-vowel would 
produce exactly the same result as appears in Semitic. In 
the Hamitic languages corresponding intensitives are produced 
from bi-consonantal roots, thus Bishari hemhem-ya " neigh " 
(with auxiliary -ya), Kafa Jcare " fight ", intensitive Jcarkare 
" fight continually ", Galla ademe ademe " keep on going ", 
Somali go " cut ", gogo " cut in pieces ", Hausa sani " sit ", 
causative san-si, causative intensitive sansan-si. 

It would seem that here we have one of the oldest types of 
reduplication, but although there are many early stems formed 
on these lines, it is a curious fact that it is commonest in the 
modern dialects of Arabic and neo-Syriac. Perhaps it may 
be as Hurwitz {Root Determinatives, p. 49) suggests that this 
formation was retained embedded in Semitic consciousness 
as an available means of intensifying a verb meaning. 

(d) Theme q-t-lt-l- 

Duplication of the last two radicals as Hebrew in^lilD 
"palpitate" (Ps. xxxviii, 11) from IPID "go round"; so 
*lD^!3n " ferment violently " (Lam. i, 20 ; ii, 11 ; Job, xvi, 6), 
"iSiSn " be crooked " (Prov. xxi, 8), and parallels in noun 
forms such as HpHplX " reddish ", nip/p7n " slippery 

(places) " (Ps. XXXV, 6), '7n'7n3 " deceitful " (Deut. xxxii, 5), 
lllin^ " black " (Song of Sol. i, 6), etc. In Bib. Aram, we 
find the derived noun ^^^3^S£i' (Dan. vi, 20) and in Syriac 
*kl^Vn\s " render safe " from »i«Q^, »i0^jj^Z] " be excom- 


mimicated ", etc., with traces of the same formation in nouns, 
as 'J^ojOm " milky way " from jOk. " white ", etc. In 

Arabic we have the noun forms jj) ^ " strong ", (*^j^ 

" heavy ", etc. (Barth, Nominalb. 145-7). As a verb theme 
this appears with dissimilation, by which q-t-lt-l- becomes 

q-t-wt-l- (conj. xii, cf. 34 above), as i.^>^Xo^ I " be hump- 

backed ", j^^^^^sJ- I " become dark brown ", etc., and by 

subsequent assimilation q-t-ww-l- (conj. xiii), as ^j^>- 1 " to 

last long ", etc. In Ethiopic this type occurs in natabtaha 
" distil in drops ", etc., and more commonly in derived noun 
forms such as hamalmal " green ", gahatbat " colic " (Barth, 
Nomin. 147), and so in Amharic, where qataltala becomes 
qatalatala as mfatalatala " rub thin between the fingers ", etc. 
None of these are very common forms, but they are 
extremely interesting as showing a parallel to the ancient 
Egyptian methods of reduplication. There, as we have seen, 
the whole root is reduplicated and semi-vowels elided ; 
but when the root does not contain a semi-vowel it is the second 
occurrence of the first radical which is elided, as nJimhm from 
nhm. Perhaps we may be justified in supposing that qtltl 
stands half-way between qtlqtl and qttl in historical evolution. 

(e) Theme q-t-q-l- 

This is a rare form which appears in a few examples such as 

- <<C^,, . - -<^ . . 

(^Jj y " disturb ", from < >J) " twist, distress ". 

(/) Dissimilated themes 

A number of instances occur in which it seems that after 
the medial has been reduplicated -it-, etc., one of the resultant 
consonants has become a sonant or semi- vowel by dis- 


similation (cf. 32), as j>^>- " kill by cutting tlie throat ", from 

root p&>- " scarify and draw blood ", etc. Sometimes the 

meaning is modified so as to suggest that the sonant or semi- 
vowel is an informative, but sufficient instances remain to 
show that in some cases at least this is simply dissimilation. 
It is particularly common with mediae gemin. verbs in 
Abyssinian and Assyrian, as Ethiopic hanbaba " run to seed " 
for habbaba, etc., and in many instances the meaning is 
essentially the same as that of the qttl form. There is 
considerable freedom in producing these dissimilated forms in 
later Arabic dialects and neo-Syriac, thus Egyptian Arabic 

hatigam or Jiargam " burst in " from /^stit^ " intrude ", and 

ta'arqal for Ji^T " be intelHgent " ; T.B. "^Tp " cause 

to drag " (Erub. 1026), Dn^lJi^ " be lopped ofE " (Men. 38&), 
and neo-Syriac m^barqes " stir ", m^hardef " throw down ", 
rrfharbeq " clasp, button ", etc. Most often it is the first 
consonant which is thus dissimilated, but sometimes we find 

the second so treated, as /y-^ " look intently " from {^■X>- 
" look at ", etc. So V^)n (Dan. ii, 30) for y'^tn in Bib. Aram., 
Mandaean 5<"iy^ (from the same root), 7''tD'^^^ (Arabic 

J^), etc. 

The dissimilation of the first into a semi-vowel produces 
a diphthong which results in some dialects in a new vowel, 
thus Ethiopic daggana >*daygana >degana (cf. 32 (1) vi) and so 

Tigre hebaha ; Egyptian Arabic hozaq " impale " for Ojj^- 

from L?^/>-, etc. 


(g) Conative qdtala 

Parallel to these dissimilated forms is that in which the 
duplication of the medial has taken place but has then failed, 
the loss of the consonant producing compensatory lengthening 
in the preceding vowel. We have already noted a later case 

of this kind where *birrekh has become "Tj^S, later because it 
had taken place after the a in the closed unaccented syllable 
had become i, but there are also cases in Arabic, etc., where 
original d has become a, and this a has produced o in Hebrew 
(cf. 43). In Arabic such forms are regular and appear as 

conj. iii MJ*L9 with the specialized meaning of a conative, 

as Ji*L3 " try to kill ". They occur also in Abyssinian. In 

Hebrew this formation is usual, but not universal, with mediae 
gemin. verbs as 2^10, etc., thus VX*|1 = V-S^ " oppress " 
without change of meaning, but sometimes with meaning 
modified as ^!liD " go about " (Eccles. i, 6), D^D " turn, 
change"; S^IH "make foohsh ", S^Pl "exult" (both 
meanings " be foolish " and " be brilliant " in primary 7 /H), 
etc., and is reproduced by analogy in hollow verbs as DDl p from 
qwm, etc., which results in a superficial resemblance to the 
qtR type (cf. h above). Hare instances occur in other verbs 
as --riynV " I have appointed " (1 Sam. xxi, 3), 'p3Jrb 
(participle. Job ix, 15), ^'HK' " take root ". So Aramaic 
pSiD (Onq. and Jerus. Targ. I in Deut. ii, 7), ^^"IDD (Onq. 
and Jerus. Targ. I in Deut. i, 31), P|SiyD (T.B. Hull. 516), 

but the form q^ql (cf. c above) is more usual in Syriac. In 
neo-Syriac we very often find a (o) after the first radical to 
compensate for the disuse of duplication of the medial (cf. 
Maclean, Yernac. Syriac, sect. 87). 


138 (3) The Causative Theme 

The causative is formed by preformative s/s, h, or ', 
originally, it would appear, with vowel -d- {sd-, etc.). Of these 
forms s and s are related, as already described in sect. Ill, s 
appearing in Hebrew, Aramaic, and Assyrian, s in Arabic and 
Abyssinian : leaving out Hebrew this would suggest original 
s, which produces these correspondences in the other four 
languages, and it might be that Hebrew s, which only appears 
in a few rare survivals (cf. below), has suffered some con- 
tamination, perhaps from Assyrian. The s causative actually 
does occur in ancient Egyptian. If this is the original form, 
we can understand the s of Assyrian and Aramaic, and the 
s of Arabic and Abyssinian ; or it might be that we are here 
in face of philological changes earlier in date than those which 
generally appear. In Arabic we find another inconsistency, 
for in Mehri there is s- where we should expect s-. The apparent 
anomaly of s-, which sometimes appears in Aramaic instead of 
S-, is not a real difficulty, for original s sometimes becomes s 
in Aramaic instead of s, and there is also the possibility of loan 
words from the Arabic (cf. below). 

(a) The s/s Form 

This is the only causative in Assyrian where it is in regular 
use, as u-sajpris, u-saparras, u-sparris, etc., from i-paras 
" depart ". 

In Aramaic the sa- preformative {saqtel, etc.) is more or 
less a survival, but still fairly frequent. In Bib. Aram, it 
occurs only with the root 7^3 " finish ", as Hl^l^ (Ezra vi, 

14), infin. Tw?^^/ (Ezra v, 3, 9), and combined with reflexive 

t- in pS'^DrjP^^ (Ezra iv, 13). In Onqelos we find b^b:2^ in 

Hl^Pi^ii (Gen. ii, 1), ^rhl^^ " burn ", implied in noun 

n^n'inS^ (Exod. iii, 2), etc., and Targ. Jer. rmm (Gen. 

XXX, 9), etc., Mandsean TJ^J'IXtJ^. In Old Syriac ^q» = 


;I*ol " delay ", Pesh. »2)2q* " communicate ", ^Lqs " fill ", 
;*^v« " enslave", etc., and neo-Syriac joinak) " be proud", 
 <y>v>>aVn " change ", ,mmk) " oppress ", ,.^/;aV> " sprawl " 
(Old Syriac os*;.^, Aramaic miD " be tired "), IL^AaLd " be 
fulfilled ", etc. In Aramaic we have also the s- formative 
in Syriac ^a^'^ (T.B. nn")p) " hasten ", VJy^QO) " meet ", 
Mandsean ^yn"lXD, and neo-Syriac ;«'^^.rftVn " visit " (;^Lm 

"associate with"), ^jmSo "shiver" (from Arabic Ac- j, 

J^J1^\), >aoi;mV) " hasten ", etc. It seems quite likely 

also that we have early s- formations, now reduced to 
tri-consonantal stems in plK^ " tear away " (Hebrew pp^ 
or pin " be cleft "), DHS^ " burn brown " from □»!! " be 
hot ", -r^x* " burn black " from I^H " burn ", etc. 

In Hebrew there are only survivals of the s- form in the 
nouns 1t?'20 " snail " (Ps. Iviii, 9), implying causative 77125' 
from bh^ " moisten ", ninS^ " flame " (Job xv, 30, root 
mn'? "burn"), nh^ypSJ' "hollows" (Lev. xiv, 37, ^Vp 
" be deep "), and probably yp^ " loathe " from |*1p, inCT 
" break " from 11'2 " separate ". 

In Arabic and Abyssinian we expect a corresponding s-, 

but this occurs in Arabic only in rare cases, as ^^_Ail«»" throw 
prostrate", causative of iJ^-, c- Aa-*" "throw anyone 
on his back " (^..JU " turn upside down "), cJi»l-«i 
" swallow ", and presumably in C^x^m " interrupt " from 

J^^ " incise " from ya " cut ", *.: 

<L^ "sever", .ia-w "incise" irom Jtf> "cut", /%.3c>- 

r' \ y 

" become black " from ^>- " burn ", jTcJa-^- " spread " from 


jTci? or ^ Ll^ (id.), ^ae-w " fill water pipe " from jL>- 

" trickle ". The causative saqtala is no longer used generally 
in Arabic, although it appears in Minsean causative ^JpD, 
but it is implied in the reflexive *sa-ta-qtal- > (by metathesis) 

istaqtala (J^IaJL*.) , conj. x), which is the regular and only form 

of the causative reflexive (cf. Minaean ^57^^D, etc.). 

Precisely similar conditions appear in Abyssinian, where we 
find rare Ethiopic formative 'as- (for sa-), as in 'asqoqawa 
" lament ", and the same preformative in the noun forms 
saqoqdw " lamentation ", sa'dzdz " stiffness ", saqordr " dis- 
dain " ; whilst Amharic has some sa- survivals in sanakJcala 
" stir up strife ", etc. But causative * saqtala is implied in the 
causative reflexive 'astaqtala, 'astaqatala, 'astaqotela, etc. 

Mehri has causative forms J \z2.J^ and J_^ll^ with s-. 

As we have already noted, s- causative appears in ancient 
Egyptian, as hr " fall ", shr " make to fall " ; nfr " be 
beautiful ", snfr " make beautiful ", etc. So Libyan s- as 
ers " descend ", sers " bring down " ; Bishari s- or is-, as 
in efdig " leave ", isfedig " send away ", etc. In East African 
and Hausa this preformative becomes an afiormative, and so 
we get Kafa qay " be complete ", qayis " finish " ; ari-ye 
" know " {-ye the auxiliary verb), arise " teach " ; uwe 
" drink ", use " give to drink " ; Somali ohon " know ", 
ohonsi " teach ", and denominal forms as in raha " happiness ", 
rahaisi " make happy " ; Galla -s{a) or -z{a) by variant of 
dialect, as diga " stand ", digaisa " make to stand " ; fago 
" distant '\fages " remove " ; hino " small ", hanes " lessen "; 
debia " give back ", dehiza " bring back " ; Hausa tsai 
" stand ", causative formative -se, tsase " place " ; ce " eat ", 
cise " feed " ; zamma " sit ", zamnase " give anyone a seat ". 


It has been suggested that this s/s is the rehc of an early root 
denoting " make ", but of this we have no definite knowledge. 
Turning from the Hamitic languages to the Sumerian, we find 
that a verbal infix -si- occurs in verbs when associated with 
nouns having the causative sufl&x -su, -es (cf. Langdon, 
Sumerimi Gramm., §§ 90, 196, 198). Sumerian is not a 
Semitic language, but there seem to be points of contact which 
are worth noting, even though we are at present quite unable 
to explain them. 

(6) The h- Form 

The change from s/s to h appears in the Sabsean causative 

^ipn for Minaean ''^p'D (cf. 17/ above), and in Mehri JcaA, 

J_^I^ as alternatives for jl!l2-^, (JjIaZ). In Arabic a few 

survivals of h occur in ^ \^ " give rest to ", ^1_^ " wish 

for ", L?!^;* " pour out ", and jllA " attach ornamental 
border to cloth, etc.", which the native grammarians treat 

(incorrectly) as substitution (J A) ) of 6 for «. (cf. 10); 

and probably in ^^A " beheve " {^\^\), CjI* "give", 
and ^*A " be voracious ", a combined causative and in- 
tensitive from *l) " swallow ", as well as in -v>*3t* " feel " 

from ^^3^ or (_;-U- " feel for". 

In Hebrew hd-, which becomes M- in a doubly closed 
unaccented syllable, is the regular form, as b'^^pTi, C^pri, 
etc. In Aramaic it appears in older forms, as Bib. Aram. 
-T3iri (Dan. v, 2) and ZinjirU M^^n') (Cooke, NSI. Ixv, 11, 


for retention of PI in the imperfect cf. Ezra iv, 13, pTin^l, and 
Dan. vii, 24, 73^11^), with a very few later survivals as T.B. 
yp'n " hold true ", etc. 

(c) The '- Form 

The change from h- to '- can be seen in actual progress in 
Hebrew and Aramaic. In Hebrew H is the usual form, but 
i^ appears occasionally, as in 'T)'7XIl&«{ " I have defiled " 

(Isa. Ixii, 3) and infinitive D''3Ei'i^ (Jer. xxv, 3). In Aramaic 
n occurs in the older forms, but in the book of Daniel it is in 
course of being replaced by }<{, e.g. ^p"*p^5 (Dan. iii, 1), and in 

Syriac this latter form has entirely superseded the h- as U^^*), 
etc. So usually in T.B. as nii< (Berakh, 5a), cf. Hp^DSt^ 
(Ber. Rab. 59), m'r\i^ (Aboda Zara, 41a), Vj^X (Bab. B. 13), 
and similarly in the Targums as pj^^ " resolve " in Onq. 
and Jerus. I of Gen. vii, 16. In Phoenician the causative 
appears with y- as in ti'"lp\ yitO"*, etc., doubtless due to a 
weakening of Hamza (cf. 10). 

Preformative Hamza is the regular use in Arabic (j^dl 

conj. iv) and Abyssinian {^aqtala, etc.). In Tlemsen and 
Morocco this theme is rarely used, being generally replaced 
by the intensitive (Mar9ais, p. 76), but we find survivals such 

as ssekber (^5v»\-wl), and in Tlemsen an occasional use of 

preformative y- as ll^) (from ^ U?). 

The apparent use of causative m- in neo-Syriac is no more 
than an instance of m- which is prefixed to all derived verb 
stems followed by a stem in which 'a- has become a- by decay 
of the weak laryngal (cf . above) ; thus tt^, causative reflexive 

Vi^Z] , etc. 


The rare appearance of P- as a causative preformative in 
Arabic, as m y-,^^^ "dye yellow" {_j^ ', ^y^l "be 

yellow "), must be regarded as an instance of 41»lc (cf. 10). 
Sometimes this appears with a modification of meaning from 

the Hamza preformative, as \^s^ \ "let one do as he pleases ", 

Jw^aC " leave camel unguarded ". 

139 (4) The Passive with n- 

The n- (na-) preformative is used to denote the passive or 
reflexive sense of the idea conveyed by the verb root, or to 
transfer it into the neuter meaning of the verb of state. It is 

defined by the Arabic grammarians as quasi-passive ( P- J \As.a) 

of the primary, or of the causative, and sometimes, improperly, 
of the intensitive. 

(a) In Arabic preformative n- (na-) has usually become in- 
(Js*l2>l, conj. vii), which assimilates with first radical m- 
(cf. 27), or informative -an- in quadrihterals, as ^^j>-\ " be 

yy ey 

gathered in a crowd " from ^s-^^. In 'Iraq it is the common 

form of the ordinary passive, and it is also used to denote 
impossibility or prohibition, as hed durh ma, yenmesi " this 
road cannot be walked on " (Van Ess., p. 75), and it appears as 
the usual passive in North African, thus Moroccan hand 
" build ", inhdnd " be built " ; qddd " finish ", anqddd " be 
finished " ; Tlemsen nsroq " steal away " ; Algerian qta' 
" cut ", enqata' " be cut " ; grah " wound ", engrah " be 


wounded ", etc. In Egyptian and Tunisian this formation is 
practically obsolete. 

The ordinary Arabic formation in- agrees with Sabsean An- 
as in DSn^n from DSH. 

(6) In Abyssinian the passive-reflexive n- is normally 
replaced by the form in t- (cf. sect. 140), but traces of n- 
occur in noun forms such as nastat " horror " from root 
satata. They appear also in quadriliterals and combined with 
other formatives, as 'anfar'asa, passive of far'asa, and some- 
times in forms of the type tanqdtala. 

(c) In Hebrew this form is in regular use as a passive, either 

with preformative ni- (na-) or kin-, as 7tDpi and 7tDpn for 
*7tpWri. It denotes (1) the passive of the primary stem, as 

n^^ " bear (child) ", "iS^i " be born " ; (2) the passive of the 
intensitive or causative, as *T!33 " honour ", I^DD " be 
honoured", *in5 "conceal", TM^n "cause to disappear", 
^n^i as passive of both ; (3) reciprocal, as tO^D' " judge ", 
tOS^;i " go to law ", hi^m " ask for oneself " ; and (4) it is 
used to form an intensitive denominal, as ^^T " male ", ^IDH 

T T ~ : * 

" be born a male ". 

(d) In Aramaic the n- passive is found only in early forms of 
the language, as in the Zinjirli inscription, where it appears 
combined with reflexive t- in llXinn (Cooke, NSI. Ixiii, 14)^ 
and in the modern dialect of Ma'lula, where it is introduced 
from the Arabic, as inktil " be killed" {Palest. Explor. Fund, 
Jan. 1890, p. 92). 

(e) Assyrian shows the n- formative in free use, either alone, 
as in apparas for anparas, or combined with t-, as in attapras 
for antapras, or as aptanaras (i.e. either nt- or tn-), or in com- 
bination ntn-, as in attanapras for antanapras. 

This n- preformative appears in ancient Egyptian, but its 
use is in process of decay ; usually it is found in combination 


with other derivatives, thus witVthe intensitive in ndddd from 
dd, and with the causative and intensitive in snfhfh from root 
fh. It is common in Libyan, thus Kabyle ets " eat ", passive 
mets ; zer " see ", pass, mzer " be seen ", or as reciprocal 
" see one another " ; in Libyan m- usually appears as the 
correspondent of n-, but in Ahaggar dialect m- alone stands 
for the passive, nm- for the reflexive. As sm- or ms- it is used 
to form the Libyan causative passive. In Bishari we find 
m- passive, as in misoei from esao " increase ", and in umhokuar 
from jehakur " bind ", etc. In Mehri and in the Hamitic 
languages of Abyssinia this n-, m- formative does not occur, 
but in Somali we note passive afformative -an, as in suban 
from sub " melt ", buhsan from buhi " fill " ; and so in Galla 
-ma as egama^ passive of ega " wait for ", and in the formation 
of neuter verbs by the addition of -m, -ma to adjectives, as 
dula " old ", dulama " become old ", kino " small ", hinom 
" become small ". In Hausa also some verbs occur which 
form a passive by prefixing an- or ana-, as Jcasse " kill ", 
ana-kasse " be killed ". 

140 (5) The Reflexive in t- 

The reflexive in t-^ which has practically replaced the n- 
passive in Aramaic and Abyssinian, is well represented in all 
the Semitic languages. 

(a) In Arabic reflexive t- occurs as a preformative with the 
intensitive and co native, thus Ala.7 (conj. v) as reflexive of 

*' ■SO-' ^ ^ ^^ »^ ^ .^ 

A19 (conj. ii), and JJ*l-Ar (conj. vi) as reflexive of Jj*Li 
(conj. iii) ; with metathesis it is attached to the primary, as 

Axli (conj. viii), reflexive of A'^ (conj. i), and with 

metathesis of the formative in the causative, where the older 



s- form is preserved, thus Alil*-) (conj. x), reflexive of 

Jlil (conj. iv). With the intensitive and conative the 
preformative t- commonly appears with a half-vowel in the 

dialects of North Africa and Malta, as tkellem for /^i>wr 

(Tlemsen), and in the Tlemsen dialect conj. viii of verbs with 
initial radical dental or sibilant and medial semi-vowel is 
treated as though conj. v (for assimilations, etc., of. sect. 22 
above). In Egyptian Arabic conj. viii is used with metathesis 
as the reflexive, but a derived form without metathesis serves 
as passive, thus reflexive i'tamad, passive illmhas. In Tlemsen 
dialect there is also found a reflexive of conj. vii, in which the 
t- preformative is combined with the passive n-, a formation 

not found elsewhere, thus ntliel " be eatable " from p 1 " eat", 

nteqrd " be readable " from l^i " read ", nthobb " make 

oneself loved " (Mar^.ais, p. 86). 
Some triliteral stems with initial t- seem to have been formed 

in Arabic from roots with initial semi-vowel, as ^T 

" fear God " from ^\3 " call upon God ", and so causative 

jrcirl "insert" from rd^. Sabsean shows reflexives in t- 

without metathesis when used in the intensitive, as ^{!3i^, and 
with metathesis in formations from the primary stem, as 

in '7XnD from SiS'D. 

(6) Abyssinian shows a free use of reflexive t-. Thus from 
the primary stem we have taqatala, taqatla, etc., intensitive 
taqattala, conative taqatala, causative 'astaqtala, etc., but 
before n the t sometimes becomes d (cf. Assyrian) as dansawa, 
etc. Tigre dangara, Amharic 'adanaqwara. In Tigrina the 


t- regularly assimilates to a consonant with which it is in 
immediate contact, thus tasabra, imperfect yessabra for 
yetsabra, and so yeqqadef " be forgiven ", etc. (cf. sect. 22 
above). As in Arabic the causative shows the older s- form 
in the reflexive 'astaqtala, etc. 

(c) In Hebrew the reflexive t- is used regularly only with the 
intensitive as vtSpDn, etc., although occasional instances occur 
of primary reflexives, as ^"IpGHn (Judges xx, 15), ^pBH^ 
(Judges xxi, 9), and ^IpSHn (passive. Num. i, 47), all, it will 
be noted, from the verb HpS, and in denominatives such as 
"in^nn " make oneself a Jew ". The prefix shows -T^'H, the h- 

perhaps due to the analogy of the causative. Only once is this 
h- absent, in the late form H^nn^^ (2 Chron. xx, 35). Metathesis 

occurs with a first radical sibilant. Moabite also shows a 
reflexive primary with metathesis in Dnn/ll (Mesa stele, 
line 11). Denominal forms occur in such words as y^inn 

" blow a broken blast ", Tlinn " make a beginning ", 

D^'in " translate " (cf. Arabic /v>-j " conjecture "), and 

Dl^l " offer up a heave offering ". 

(d) Aramaic employs reflexive t- freely in all conjugations. 
Thus Bib. Aram, with the primary stem n'lT5ri5< (Dan. ii, 45), 

with the causative p^Yp^l^l (Ezra iv, 16). In Ezra the pre- 
formative is HH as in Hebrew, but in Daniel it has become ^^C 

and so Syriac I] . In T.B. it is HX or n**^, the -t- assimilating 
to the following consonant (cf. sect. 22 above). 

(e) In Assyrian reflexive t- is used freely in combination with 
any stem, metathesis taking place in all cases ; thus primary 
iptaras, causative reflexive uHapras, intensitive reflexive 
aptarras, etc. 


A t- formative occurs in ancient Egyptian, but does not 
appear to convey any reflexive sense ; it appears in the 
earliest Egyptian with transitives and causatives, later with 
intransitives as well. In old Egyptian its form was -t or -t'i, 
in middle Egyptian -t, -tw, in later Egyptian -tw only, and in 
Coptic it has disappeared. Apparently both hpr and hprt 
mean " become ", and ph and pht " arrive " ; no doubt the 
afformative originally modified the meaning, but in the earliest 
existing text the sense is lost, and the formative seems to be 
merely a survival. In Libyan, on the other hand, it is in full 
vigour, as (i) passive preformative t- (ts-), as in tsegzem^ passive 
of egzem " cut " (Kabyle) ; (ii) passive tou- {tsou-) in tsouaf, 
passive of of " find " (id.) ; (iii) iterative t-, ts-, th-, as in tsaf, 
iterative oi af " find " (Doubdou), thebbi, iterative of ebbi 
" cut " (Shilha), thetsou of tsou " forget " (Kabyle) ; (iv) in 
Twareg only as afiormative -t, e.g. emm " die ", emmout 
(Ahaggar), but some rare survivals of this afformative occur 
in other dialects, e.g. emmeth (Haraktas and Zouaoua). As 
for the modification of meaning to iterative, it must be noted 

that in Semitic t- is not a true passive, e.g. /^JU- " teach ", 

j^JLjo " become learned " as mceptive and quasi-passive, 

but distinct from the true passive " be taught ". In Bishari 
we find reflexive et-, as eJctem " arrive ", reflexive etJcetam, 
esem " call ", reflexive etsosam, etc. In East African, as usual, 
this becomes an afformative -te, thus Kafa uw " drink ", 
stative uw-we, stative reflexive uw-we-te, baje " hinder ", 
reflexive bajite, and similarly in Saho, Bihn, and Hamara 
speech. Galla shows corresponding -t {-it) or -d {-da) according 
to dialect, thus of a " make a bed ", reflexive afada, and so we 
find denominatives such as hinot " become small " from hino 
" small ", etc. Hausa has reflexive -ta or -da and causative 
reflexive -sda, thus tara " collect ", tarada " overtake ", 


i.e. add oneself to a crowd, kwana " sleep ", Jcwanta " pass 
the night ", etc. 

141 (6) General note on the verb themes 

In Abyssinian and Assyrian there is considerable freedom 
in combining two or even more of these themes together. 
Thus in Assyrian we have, more or less regularly, Nt, i.e. 
N passive and reflexive, as attapras (for antapras), the primary 
combined with the N passive and reflexive, as aptanaras, 
the causative and intensitive as usparras, this last with the 
reflexive as ustaparras, and even the reflexive of the N 
passive afterwards combined again with the N passive, as 
attanapras (for antanapras). 

In Arabic dialect we sometimes find the causative reflexive 
combined with the intensitive as istanna from root 'ny, which 
appears in the dialect of Tlemsen as ssenna ; so in Tlemsen 
the reflexive is combined with the n- passive in ntkel " be 
eatable ", nteqrd " be readable ", etc., but this occurs only 
with verbs having one radical semi-vowel or Hamza treated 
as a semi- vowel. In Tripoli and Tunis we get a form istahdyl 
which is a combination of the reflexive causative with the 
conative {qdtala^ conj. iii). 

142 (7) Vocalization of the Stem and Tense Forms 

Laying aside the participles and infinitive, which are 
essentially noun forms, there are two methods of vocalizing 
the stem, which may be represented by the Arabic forms 
qatala and yaqtulu of the primary (conj. i). Both these are 
3rd person singular, and discarding the personal formatives 
we have qatal- and -qtulu, for it must be noted that final -a 
of the perfect is as much a personal termination as -ta, etc. 
From the second we also discard the final -u, which is a modal 
distinction denoting the indicative, and thus arrive at qatal- 
and -qtul. In West Semitic the former of these is known as 


the perfect, the latter as the imperfect and imperative. The 
West Semitic perfect has its personal terminations sufi&xed, 
the imperfect and imperative have certain suffixes denoting 
gender and number, and in the imperfect there are personal 
prefixes, but these are not found in the imperative, which 
consequently commences with a group of two consonants, 
and these are vocaHzed as already described (cf. § 66). In 
Assyrian the first qatal stem is commonly called the 
" present ", the second -qtul is the " preterite " and imperative, 
but the propriety of these names is open to serious question ; 
both present and preterite have their persons formed in the 
same way as the imperfect of West Semitic, but a third 
(participial) tense exists, known as the permansive, which 
employs suffixed personal terminations hke the West Semitic 

The essential difEerence between these two stems may be 
regarded as a question of musical tempo, the deliberate 
statement of the West Semitic perfect and Assyrian 
" present " being spaced in tempo lente, the command and 
subordinate statement (imperfect) is in tempo allegro 
Phonetically, therefore, the slower time of the perfect 
necessitates the insertion of a vowel after the first radical 
unless, as is the case in some of the derived stems, it is already 
vocalized by a preformative. The insertion of this vowel in 
the perfect and its absence from the imperfect is not the result 
of a personal ending being suffixed to the former and a personal 
prefix being attached to the latter, for the same stem 
vocalizations occur in Assyrian, where the personal formative 
is prefixed in both cases, and the vocalization of the imperative 
is substantially that of the imperfect, although the imperative 
has no personal prefix. 

In the two stem forms qatal and -qtal the vowel following the 
second radical is the stem vowel proper, not inserted from the 
necessities of phonetic conditions, but essentially a part of the 


stem. In the primary stem this vowel may be a, i, or u, 
in either stem, but in the derived stems it is normally a in 
the perfect and i in the imperfect. The formation of a passive 
by the modification of these vowels will be considered later. 
The Arabic grammarians treat these vowels in the perfect as 
semantic, the qatal stem denoting an act, qatil a transitory 

state as ^^ " be clothed ", qatul a permanent state as 

/,.*«>- " be beautiful ". This is generally but not umversally 

true, for there are cases, especially with medial or final laryngal, 
where the vocalization has evidently been influenced by the 
neighbouring consonants, and it remains an open question 
whether this semantic value of the vocalization was original, 
sometimes affected by consonant influence, or whether it was 
due to phonetic causes and then affected by analogy, i.e. 
the verbs following the class type of some leading roots of 
kindred meaning (cf. Lambert, Journal Asiatique, Feb., 
March, 1890, pp. 169 sqq.). 

Generally perfect qatal corresponds with imperfect -qtul 
or -qtil and perfect qatil has imperfect -qtal. The original 
imperfect corresponding to perfect qatul is unknown ; it now 
appears as -qtul. Exceptions to the above occur in verbs with 
second or third laryngal, which commonly have perfect qatal, 
imperfect -qtal (cf. sect. 54), unless those denoting superiority, 

which have imperfect -c^ul, as j>:A , js-^Oi " surpass m 

glory ". Rare instances occur in which perfect qatil has 
imperfect -qtil. 

Thus in Arabic we have six possible types : —  

(1) J:9 J::2 "kiU". 

(2) Zfj^ ^j^ "strike". 



(3) 7^ ^^^ 


" open ". 

rxfi- (♦•*.»' " be wise ". 

r - 

(5) c..**»^>- <,_/-«^ suppose 


A^ (»,P^ " ^® precious ". 

In Abyssinian * is confused with w, and so type (2) is 
assimilated to (1), and similarly (6) is assimilated to (5). In 
Hebrew and Aramaic the imperfects with vowel i (e) are in 
decay in the primary stem ; in Hebrew they have practically 
disappeared in strong verbs, but appear in jHX '^\ and in 
verbs with initial or final semi- vowel. In Aramaic they survive 
in pQU " make " and _3p " buy " only. In Assyrian 

all the forms of vocalization appear, as present (i.e. West 
Semitic perfect) i-paras, i-pakid, i-balut, preterite (imperfect) 
i-prus, i-pkid, i-sbat. 

The following is a general outline of the vocalization of the 
tense forms in Semitic : — 

(G Ground-form or Primary stem, D Duplicated or 
Intensitive, C Causative, N Passive in n-, t the combination of 
any of these with the reflexive t- ; I the perfect of West 
Semitic, II the imperfect, in each case with the corresponding 
stem in Assyrian.) 

Arabic Abyssinian Hebrew Aramaic Assyrian 

G, I. qatal- qatal- qatal qHal -paras 

labis- I jldbes Vhes -pakid 

hasun-j [qdton q^fud -halut 

II. -qtul "I - - , (-prus 

•Ibis] -^^ '^^^^ -^"^ [-pkid 

-fiah -Ibas -qtan -dlial -sbat 






































• • • 






• • • 





• • • 





• • • 




















• • • 






• • • 



The Gr (primary) stem shows a variety of vocalizations. 
A Hebrew imperfect -qtil stem exists where the root has 
initial or medial y. In Hebrew and Aramaic the perfect of 
the intensitive and causative have had their vocalization 
assimilated to that of the imperfect. Generally, however, 
as will be noted, the perfect of the derived stems has -a- and 
the imperfect -i- ; exceptionally the Arabic and Abyssinian 
imperfect of the intensitive reflexive (Dt) has -a-, and so the 
imperfect of the primary reflexive (Gt) in Assyrian. Whilst 
the perfect is affected by the imperfect in the D, causative, 
and primary reflexive in Aramaic, the converse instance of the 
imperfect affected by the perfect appears in Aramaic in the 
reflexive of the intensitive and causative. Other cases must 
be admitted in which the vocalization is influenced by the 
neighbouring consonants (cf. sect. 53 above). 

143 (8) Passive by Vowel Change 

Besides the passive and reflexive forms in n- and t- there is a 
passive formed by modification of the vocalization. In Arabic 
this produces perfect -u-i-, imperfect -u-a-, thus active 


qatal- becomes passiv^ qutil-, and imperfect active yaqM 
becomes passive yuqtil. This vowel modification extends 
to personal preformative and also to the formative of the 
derived stem, as intensitive passive quttil-, yuqattal- ; causative 
uqtil; yuqtal, etc. In dialect we find Syrian Arabic as passive 
qutil, imperfect yuskan ; Oman passive primary knoq, 
intensitive hurrug, suffid, etc. In North Africa this passive 
is generally lost, but a few forms survive in Tlemsen, as 
yusdb " he will be found ", yoUdi " it is necessary ". 

This passive by vowel change is obsolete in Abyssinian. 

In Hebrew it survives in the primary theme only in the 
participle 7^t3|^, but it is in full vigour in the intensitive and 
causative, as intensitive perfect ^tSp, imperfect 7t3p\ causa- 
tive perfect l^'pj}, imperfect bv^'p\ and mediae gemin. 
intensitive perfect S^ID, imperfect i^iD\ 

In Aramaic it survives only in the participles, as l^HS, 
7pri (Dan. V, 25) in the primary; intensitive W ^Avn 
causative Vs^Q^J, ^Qmk). So in T.B. participle T\'BX:}l2 
" banned " (Sabb. 67a), P|3iyD " folded " (Hull. 516), etc. 

Assyrian shows some relics in the Amarna letters, as perfect 
dika, plur. dA,ku. 

(iii) Verb Inflexions 
144 (a) The Tenses 

The Semitic tenses are two in number, which are called 

"past" {(JAa, ^ny) and "future" (^jU.^, Tfli?) 

in the older grammars, but have been generally known as 
" perfect " and " imperfect " since the time of Ewald. But 
they are not tenses in the sense in which the word is used in 


European grammar, as they are not concerned with time, 
past, present, or future, but only with what is described as 
aktionsart, time and place being expressed adverbially. This 
is true also of the " tenses " in the Indo-European languages, 
where the present describes an enduring act or state, the 
aorist denotes the action or state simply without reference to 
duration, and the perfect deals with an action as finished. 
The introduction of a time sense is a later development by 
means of added particles or by the use of auxiliaries. Such 
a time development appears in later Semitic, in Arabic 
dialect, in neo-Syriac, and in ancient Egyptian, which seems 
to have had a rapid and premature development. 

The " perfect " of West Semitic, corresponding to the 
" present " in Assyrian, expresses a state or action which is 
definitely asserted and regarded as certain as contrasted with 
the imperfect expressing what may be, what is possible, or 
can be treated as an accessory, causal, conditional, etc. The 
perfect is declaratory, with emphasis on the assurance, 
regardless of whether the time is present, past, or future, 
although, of course, the past is usually regarded as known and 
settled more than could possibly be the case with the present 
or future, but in promises or bargains the perfect is used even 
though it must refer to future time because the emphasis 
is on the assurance and certainty of the promise made, and so 
in prayers, blessings, curses, and in prophecy. Thus in 

Arabic, " serve the Lord who created you " (/^X_fli>-, Qur'an, 

ii, 21), and " when your Lord said (J Li) to the angels, 

I am placing (participle) in the earth a ruler, they said ( lj,3l_9), 

. -^ <^ 

wilt thou place (imperfect \^ I ) in it one who will make 


miscliief (imperfect X-^jJcu ) in it and shed ( Xiu^ ) 

blood ? " (Qur'an, ii, 30). The angels refer in the imperfect 
to events which they regard as probable, but of which they 
have no certain knowledge, but the events described as 
actually having taken place are in the perfect. Even in 
conditional clauses the perfect is used if the condition is 
assumed to have taken place, as "if you are in doubt as to 
that which we have revealed " (Qur'an, ii, 23), it being treated 
as a known fact that those whom the prophet addresses 
were at the time in doubt. So in conditions in which no 
emphasis is laid on the conditional character, so that the 
condition is nearly equivalent to a participle or a clause 
introduced by " when ", as "if they dispute with you, 
say . . . ." (Qur'an, iii, 19). Similarly when the conditional 
is introduced by a particle (cf. 164), which implies that the 
condition is impossible or is known or believed not to have 
taken place. 

The perfect is used also with temporal particles denoting 
" when " or " whilst ", even though the acts are continuous 
and so incomplete ; and so in the statement of consequence, 
provided that emphasis is laid in the certainty of the result 
following, as " the Lord will send His angel and 'prosper thy 
way " (Gen. xxiv, 40), and " lest he put forth his hand and 
take and eat " (Gen. iii, 22). 

As the perfect describes a certain and assured act or state, 
so the imperfect describes the incomplete, or dubious, or 
possible, as well as the subordinate statement which is not 
emphasized as the object of assertion. Thus we have in 
the imperfect to continuous or repeated act, as well as the 
nascent, progressive, or potential, and so what often corre- 
sponds with our present, as " the people come to me"(Exod. 
xviii, 15), and the frequentative or customary. The imperfect 


also is used in hypothetical sentences, in the " if " clause if it be 
regarded as dubious or uncertain, and in the consequence if 
emphasis is not laid on the certainty of its following the 
protasis. Besides this, we have also what may be described 
as the " polite " use of the imperfect, which appears in 
courteous inquiry, in request, command, etc., as less harsh 
than the imperative, and invariably in the negative command 
for which Semitic does not admit the imperative. 

As denoting the incomplete the imperfect is used to express 
the customary, frequentative, etc., in which it is often very 
little to be distinguished in meaning from the perfect, but may 
be described as the tense of the nascent, progressive, inceptive, 
or potential ; thus "it is a guide to those ivlio believe in the 
unseen " (Qur'an, ii, 3), " the humble ones who know that they 
shall meet their Lord " (Qur'an, ii, 46) ; " thus did Job 
continually " (Job i, 5), " there went up a mist " (Gen. ii, 6), 
" what seekest thou ? " (Gen. xliv, 7), and hence obtains a 
quasi-present meaning, as " I redeem ", i.e. " I am in the 
habit of redeeming " (Exod. xiii, 15), etc. 

The imperfect is the tense used to express the subordinate 
assertion, the accessory whether descriptive, or expressing 
the purpose, result, etc., which is not the main statement of 
the sentence, thus " he came to a spring of water to drink ", 

i.e. " that he might drink " ((„j JIj ), " he intends to 

turn you out of your land " (" that he turn you ..." Qur'an, vii, 
110), " now surely they fold up their breasts that they may make 
concealment from him " (Qur'an, xi, 5), " lest you sit down 
despised " (Qur'an, xvii, 22) ; so Hebrew " the sickness he 
was to die of " (2 Kings xiii, 14). Other examples of the 
functions of the imperfect will be found under the heading 
of " moods ", for the subjunctive and jussive in Semitic 
are subdivisions of the imperfect (cf. 145 below). 


145 (b) The Moods 1 

In Semitic the moods are concerned only with the imperfect 
tense. In Arabic there are altogether five modal distinctions, 
(i) The simplest of these is commonly known as the 
" Jussive ", and shows only the imperfect stem with the 
personal preformatives (cf. § 147) and certain terminations 
added to denote gender and number, namely : — 
Sing. fem. 2nd pers. ... -I 
Plur. masc. 2nd and 3rd . -u 
fem. 2nd and 3rd . . -7id 

Dual -a 

These terminations are the same as those employed in the 
imperative (cf. 148). 

(ii) The subjunctive shows the same stem, preformatives, 
and terminations, but wherever there is no termination the 
vowel -a is added ; thus sing. 3rd masc. jussive yaqtul, subj. 
yaqtula. This vowel, it will be noted, is the same as that used 
as the termination of the accusative of a noun, and the 
subjunctive shows a parallel use as the verb of many object 
sentences, so that certain particles which cause a noun to be 
in the accusative also cause the verb to assume the -a of 
the subjunctive. 

(iii) The indicative adds short -u where the subjunctive 
has -a, but it also adds -na to each long vowel ending, save in 
the dual, where by dissimilation -d-na becomes -d-ni. 

(iv) The first energetic mood shows the jussive with added 
-anna (cf. 93), which becomes -una after the long vowel 
endings -i, -u, which are then shortened by closure, but dual 
-a remains long, and fem. -na becomes long -na, in each case 
the following -una dissimilates to -nni. '-. ■>' i - 

(v) The second energetic shows -an where there is no vowel 
termination, and simple -n after a vowel ending which is then 
shortened. It does not occur with the dual or feminine plural. 
These results may be thus tabulated : — 



Sing. 2nd fern. 
PI. 3rd, 2nd masc. 

3rd, 2nd fern. 
Dual . 
Elsewhere . 

Indie. Subj. Jussive Energ. I Energ. II 

-ina -I -I -inna -in 

-una -u -u -iinna -un 

-na -na -na -ndnni —  

-dni -a -a -anni — 

-it -d — -dnni -an 

The clearest distinction between the moods thus lies in 
those persons which have not a distinctive termination to 
indicate gender or number, those which we have classed 
together as " elsewhere ". But here the distinction is one 
due to short vowel endings, and in Arabic dialect, Hebrew, 
etc., final short vowels have become obsolete. Consequently, 
it is only in Arabic of what is called the classical type that we 
find the moods fully distinguished. 

In classical Arabic (i) the indicative is essentially the 
mood of narrative and it is thus employed, subject to the 
general character of the imperfect, which we have noted 

(ii) The subjunctive is used to express the purpose, and so 
after li, hd, likd, Wan, hattd, or negative kayld, likayld, Walla, 
expressing the purpose and intention of the agent ; or the 
result introduced by Hdhan, or hy fa-, wa-, after an imperative, 
expression of wish, hope, etc., or after a negative or 
interrogative ; or the object sentence introduced by 'an, 
negative 'an Id, 'alld, Ian, after an expression of hope, 
fear, duty, effort, etc. ; or the exception after 'ilia 'an 
" unless ". 

(iii) The jussive, closely akin to the imperative, is used to 
express command, with or without prefixed li- ; it may also 
express the milder form of a wish, entreaty, or resolve. In 
prohibition it must be substituted for the imperative. It is 
used after the negative lam " not " and after lammd in the 
sense of " not yet ". It appears also in conditional clauses. 


(iv) The energetic forms may be employed for command, 
wish, etc., like the jussive, especially when the clause is 
reinforced by the emphatic particle la- " verily ". They 
are used also in the conditional when the conditional particle 
is reinforced by -md. 

Abyssinian does not retain the short vowel terminations of 
the imperfect, and the subjunctive and jussive are confused 
in one form, yeqtel, etc. Although the indicative does not 
retain its distinctive vowel termination, it is distinguished by 
its accent and time. Whilst the jussive-subjimctive has the 
tempo allegro of the imperative, the indicative in tempo lente 
inserts an additional vowel after the first radical, producing 
yeqatel, and the accented penultimate causes the accent to 
fall on this inserted vowel where there is no added termination. 

In Hebrew no distinction appears between the indicative 
and subjunctive, owing to the loss of the final short vowels, 
but sometimes a distinctive difierence appears in the jussive, 
which never had a vowel termination, so that the syllable 
commencing with the medial radical was always closed and 
did not become closed. Thus in the causative where the 
indicative-subjunctive has -i-, the jussive has the shorter 
vowel -e-, as indicative yaqtll, jussive yaqtel ; or, before final 
laryngal, indicative yasliah (cf. 53), jussive yasldh. So with 
verbs having medial wfy, as Vbyn, indicative ydbin, jussive 
yaben ; \/qwm, indicative ydqum, jussive ydqom. 

A cohortative form appears also in the 1st person singular 
and plural of the imperfect and in the imperative, which is 
characterized by suffixed -d, corresponding to the Arabic 
energetic -an, -anna, which become -a in pause. This has an 
emphatic force in the 2nd and 3rd persons of the imperative, 
and a modifying precatory sense in the 1st. Another obvious 
relic of the energetic appears in the demonstrative -ann-, 
-inn-, which is often added to the imperfect before a suffixed 
pronoun (cf. 93). 



The modal endings are not entirely extinct in Assyrian, 
but, like the case endings of nouns, they have suffered con- 
fusion. It would appear that -a was properly employed in 
continued narrative, whilst final -u is used in subordinate 
clauses, and in this employ appears in later forms as well. 
It is difficult to suppose that we have here the original uses. 

146 (c) Persons of the West Semitic Perfect 

The personal terminations of the perfect in Arabic, 
Abyssinian, Hebrew, and Aramaic, with which may be 
compared the terminations of the permansive in Assyrian, 
may be summarized as follows : — 

Arabic Abyssin. Hebrew Aramaic Assyrian 


Sing. 3 masc. 









-d {-a 




2 masc. 


















Plur. 3 masc. 










-« -v.: 

2 masc. 

















-dni, -dnu 

Dual 3 masc. 






(a) Sing. 3rd masc. Termination -a, which as final short falls 
away in Hebrew and Aramaic (cf. sect. 74), but is retained 
before suffixed -ni as in Hebrew qHdldni, Syriac qatlan{i), 
and as d before suffixes of the 2nd fem. sing., 3rd masc. sing,, 
and 1st plur. as q^tdldkh, qHdldhu, q^tdldnu. In modern 
Arabic this termination is entirely lost. 

(h) Sing. 3rd fem. Termination -at as for nouns, suggesting 
that the perfect was originally a participle (cf. Assyrian 


perm.). In Hebrew and Aramaic the t is aspirated by the 
preceding vowel (cf. 37) ; and in Hebrew it usually falls when 
not protected by a suffix, as qatHd, qHdldthni ; but rare cases 
of -t retained occur in 'dzlath (Deut. xxxii, 36) and qdrd'dth 
(Deut. xxxi, 29). In Aramaic, as in Hebrew, the addition of 
the suffix causes the reduction of the preceding vowel, and 
thus Syriac q^tal becomes qatlath, where Hebrew shortens to 
a half- vowel qaHd. A similar change takes place in the Arabic 
dialects where the accent in the masc. falls on the vowel 
following the second radical, as in Tripoli masc. kteb, fem. 
kitbet, and so very generally in North Africa and with 
intransitive verbs in Oman, as well as sometimes in Central 
Arabia {dd'fet, but also ahddhat), but in 'Iraq we find both 
ktibet and kiteh{e)t. In Central Arabia the ehsion of the 
preceding vowel is found chiefly when that vowel is -i-, and 
so in Syrian sirih, sirbet, but kdtab, kdtabet, and sometimes in 
Egypt misiket, misket. 

(c) Sing. 2nd masc. -ta, in Abyssinian changed to -ka, 
and so k for t throughout the 2nd person, by the analogy of the 
1st person sing., the original consonants being preserved in 
Assyrian (cf. also note on 1st singular below). In Hebrew 
this becomes -td, and in Aramaic the final vowel is lost except 
before suffixes. The cause of the inserted -d- in the Assyrian 
permansive is unknown. In Arabic dialect this -ta invariably 
becomes -t, in Central Arabia -it, and in 'Iraq sometimes -et. 

(d) Sing. 2nd fem. -ti. The feminine is formed by change of 
vowel ending to -i (cf. 118). In Abyssinian t becomes k 
(cf. above). In Hebrew and Aramaic the final vowel is lost 
except before suffixes. 

(e) Sing. 1st pers. Originally -ku as in Assyrian permansive 
-dku, so Abyssinian, Ge'ez, and Tigrina -ku, Tigre -kU, -ko, 
Amharic -liii (cf. 37) and Mehri -ik, -ek, Soqotra -k ; in all these 
southern languages the k has also affected the 2nd person, but 
Assyrian retains 2nd t, 1st k, which seem to have been the 


original forms. In Arabic dialect -tu becomes -t (Central 
Arabia, Oman, Hadramaut, 'Iraq, Syria, Egypt, and generally 
in North Africa, save where t becomes ts, cf. 176), also -it 
(Central Arabia, Algeria, Nejd, Syria, 'Iraq) and -et (id.). 
In Hebrew -t% or -t the final -i probably is due to the analogy 
of the 1st singular pronominal suffix ; and so Moabite "Tl, 
Phoenician Tl, Jl, and -tin in coranthi (Plant. Poenul.). Aramaic 
-eth with alteration of syllabic constitution to qefleth (cf. 3rd 
sing, fem.), rare -tl (Bib. Aram.), also -it (Targ., T.B., 
Mandaean), -ey (T.B.), and -i (Mandsean). 

(/) Plur. 3rd masc. -U, in Sjn-iac only before suffixes. Arabic 
dialect changes -u to -5 in Central Arabia and Oman, or to 
-aw in Central Arabia, Hadramaut, and 'Iraq. Egypt shows 
-M, but also -um by assimilation to the pronoun. Hebrew 
-u, -un (Isa. xxvi, 11), -um (Isa. liii, 1 ; Deut. viii, 3). 

{g) Plur. 3rd fem. -a or -na (cf. imperfect). In Arabic 
dialect -an in Central Arabia, -en in Oman, Hadramaut, 
'Iraq. In Syria, Egypt, and North Africa this termination is 
obsolete, as is also the case in Hebrew. Bib. Aram, -a, Syriac 
-a before suffixes, also -en by analogy of the suffixed pronoun. 

(h) Plural 2nd masc. -turn. Abyssinian, as usual, u becomes 
e ; Hebrew vowel assimilated to the fem. ; Aramaic either 
u >o, or as in Hebrew -tern ; Syriac -ton, or -tond before 
suffix. Assyrian shows -{d)tunu. Added -u in Abyssinian 
(with preceding consonant doubled), and in Assyrian, sometimes 
in Arabic. Abyssinian shows k as in the singular. In Arabic 
dialect -tu ('Iraq, Central Arabia, Syria, Tripoli, Tunis, Oran, 
South Algeria, Morocco, Malta, and Tlemsen -tsu), -to (Oman, 
Hadramaut), -tum in Spanish Arabic. 

{j) Plur. 2nd fem. -tin, i.e. fem. sing, with plur. -n. Thus 
Arabic dialect -ten (Oman, Hadramaut, 'Iraq), but in all other 
dialects this person is obsolete. So -ten in Hebrew, Aramaic, 
and -ken in Abyssinian (before suffix -kennd, -kd, cf. 856), In 
classical Arabic this person shows vowel -it- by analogy with 


the masculine (cf. personal pronoun) ; the added -na is 
perhaps taken from the imperfect imperative, and accounts 
for the retention of -n after u. Aramaic -ten, and -tena before 

{k) \st plur. -na in Arabic (in dialect -na in Central Arabia, 
'Iraq, Morocco, -ne in Oman, Hadramaut, -na in South 
Algeria, and -nd in Tripoli, Tunis) ; so Abyssinian. Aramaic 
-nd before suffixes, otherwise -n, or -nan by analogy of the 
absolute pronoun. In Hebrew -nu (cf. personal pronoun) 
under the influence of the 3rd masculine plural. 

(I) Dual. The dual appears only in ancient Arabic, where 
the characteristic -a is added to the singular in the 3rd person, 
and to the masculine plural in the 2nd. 

147 (d) Persons of the Imperfect Tense 

The persons of the imperfect are denoted by personal 

prefixes and suffixes denoting gender and number. The 

prefixes are : — 

Arabic Abyssin. Hebrew Aramaic Assyrian 
Sing. 3 masc. . ya- ye- yi- ne- i- 

3 fem. \ 
2m.f. J 

1 . 
Plur. 3 masc. 

3 fem. 

2 . 

1 . 
Dual 3 . 

2 . 

ta- te- ti- te- ta- 

'a- 'e- 'e- 'e- a- 

ya- ye- yi- ne- i- 

ta- te- ti- te- ta- 
na- ne- ni- ne- na- 

(a) ya- : 3rd masc. of all numbers, and 3rd fem. in plural 
and dual. In Syriac ne- (but Bib. Aram., etc., yi-). With 
this we compare I- in the jussive in Tigre liskun, etc., where 
I- is the particle (preposition) " to " (cf. 158) ; so Hebrew 
likroth (Isa. xliv, 14, but cf. Driver, Tenses, 203), Aramaic 
leh^we (Dan. ii, 20), and frequently in T.B., later Hebrew, and 


Mand^an. In older Aramaic this I- appears with the infinitive, 
in later Aramaic as in Tigre with the jussive ; in Bib. Aram, 
only with niH (Ezra iv, 13 ; Dan. ii, 20, 43) ; hence probably 
the n- in Syriac as n often appears for I- in T.B., Mandsean, 
etc. (cf. 28), Ass3n:ian ya->i- (cf. 52 (1)). 

(6) ta- : 2nd pers. and 3rd fern. sing, in Hebrew also for 
the 3rd fern, plur., by analogy ; but y- in Gen. xxx, 38 ; 
Dan. viii, 22. 

(c) 'a- : Assjrrian a-, 1st sing, only ; cf. 'a in 'an-'a (§ 77). 

(d) na- : Ge'ez, Tigre, we- or 'en- ; Amharic 'enne-. 

(e) Vocalization of the prefixes 

-a- in ancient Arabic, in Assyrian in ta- and a- (1st sing.). 
Also in Arabic and Hebrew before first radical laryngal 
(cf. § 53). 

-e- in Abyssinian (but causative ye'a>yd), Syriac (West), 
Arabic dialect of Morocco, Algeria, and Tripoli. Hebrew in 
1st sing. 

-i- Hebrew save in 1st sing, or before first radical laryngal 
(but ya' >ye'). Syriac (East). Assyrian ya->i- and na- >ni- 
(analogy of 3rd sing.). Also in Arabic dialect of 'Iraq, Syria, 

-u- in all languages in the passive (cf. § 143), intensitive, 
and causative, in Hebrew and Aramaic with change to -o- 
as described in § 48. But Abyssinian causative ye'a- 
(for yii'a-) >yd-. 

In Egypt and Oman we find regular assimilation, as yimsih, 
yuskun, etc. 

(ii) Terminations of the Imperfect 

Arabic Abyssin. Hebrew Aramaic Assyrian 
Sing. 2 fem. . -i{na) -i -i -lin) -i 

Plur. 3, 2 masc. -u{na) -u -u -u{n) -ii 

3, 2 fem. -na -a -nd -an -a 

Dual . . -d{ni) 






pur us 











For the short vowel endings employed to denote moods, 
cf. § 145 above. 

(a) -i, Syriac -in, and Hebrew -in in |''pl'l^ (Ruth ii, 8), 

etc., appears elsewhere as a feminine afformative (cf. 79, 
85, 118). 

(6) -u masc. plur., cf. § 147. 

(c) -a fern. plm-. in Abyssinian, Aramaic, and Assyrian : 
-na in Arabic and Hebrew. 

148 (e) The Imperative 

The types of the imperative are as follow : — 

Arabic Abyssin. 

Sing. masc. uqtul qetel 

fern. uqtuU qetell 

Plur. masc. uqiulu qetelu 

fern. uqtulna qeteld 

The terminations denoting gender and number are the same 

as those already noted in the imperfect (above). As regards 

the stem, the base form is qtul, qtal, qtil, in which the first 

consonant is vocalized by the insertion of a half-vowel in 

Hebrew and Aramaic, by the insertion of the short and vague 

e in Abyssinian, and in Assj^rian by a vowel which assimilates 

to the following stem vowel, whilst in Arabic a prosthetic 

vowel is used, normally i- but u- by assimilation before stem 

vowel -u- (cf. § 66). The vocalization in Hebrew necessarily 

modifies when the syllabic constitution is altered by the 

addition of a vowel termination which opens the medial 

syllable, thus : — 

Sing. fem. qitHt, in pause qHdli. 
Plur. masc. qitH'A „ qHolu. 

149 (f) Secondary Tenses 
(i) The Assyrian Permansive 

This is a new tense obtained in Assyrian by adding personal 
afiormatives to a participial form of the primary theme, and 


in derived themes by analogous treatment of infinitive 
stems. Its meaning lias the inherent idea of duration in the 
state or action of the verb sense. It does not correspond to 
the West Semitic perfect, which in its stem form qatal, etc., 
is represented by the Assyrian present, so called, though it 
has personal afformatives instead of preformatives, and these 
suffixed persons bear a resemblance to those used in the West 
Semitic perfect. These endings are : — 

Singular Plural 

masc. fem. masc. fem. 

pers, 3 . . parts pars-at pars-u pars-a 

2 . . pars-dt{a) pars-dti pars-dtunu — 

1 . . pars-dk{u) pars-dn{u) 

The long -a- which appears before the termination in the 
2nd and 1st persons has not yet been accounted for 

(ii) Syriac Participial Tenses 

In Syriac a present tense is formed by adding enclitic 
pronouns to the participle, and this formation becomes more 
and more popular in later speech until in neo-Syriac it 
practically replaces the older tenses. Other tenses are formed 
on the same lines in the way described below. The pronominal 
forms thus used are : — 

Singular Plural 

masc. fem. masc. fem. 

pers. 3 , . -, -u -i, -at -in, -an 

2 . . -at -at{i) -atun, -tun -atin, -tin 
1 . . -na -nan 

But dialectal differences occur in these forms in neo-Syriac 
(cf. Maclean, Vernac. Syr., p. 84). 

(iii) Introduction of time relations into the tenses 

Although the original Semitic tenses conveyed Aktionsart 
and not temporal relations, we find in later forms the intro- 


duction of the time sense by the addition of particles, etc. 
In Arabic the particle c-J^-w, or its contracted form in the 
prefix ^yt prefixed to the imperfect, is used to express the. 
future, and this already appears in the Qur'an, e.g. ^ J^^ 


" we shall show them ", etc. Later we find the particle >X» , 

properly an emphatic used to lay stress on the certainty of 
an act or state, prefixed to the perfect so as to convey the 
sense of past time, and thus to form a pluperfect, or attached 
to the imperfect so as to form an imperfect " was ", etc. 

Similarly, the verb ju " be " is used with the perfect to 

make a pluperfect, and with the imperfect to give the past 
indefinite " was " sense. 

In the dialects of Syria and Egypt prefixed ba- or ma- 
attached to the imperfect produces the sense of past time, as 

hayuskun "he dwelt". Without i^j.^, ^y, ha- or 7na-, 

the imperfect is generally employed for present or future time. 
In North Africa the present is expressed by ka- or ra- prefixed 
to the imperfect, as kayatub " he is writing ", and in the eastern 
dialects the same sense is often conveyed by prefixed 'amm-. 
In Omani the imperfect acquires a future sense by the use of 
ha-, he-, as hayaktuh " he will write ", and the same particle 
is used in Maltese to express an optative. Omani uses the 

particle ^dd instead of the verb ju to form the pluperfect 

from the perfect ; this particle appears as sing. masc. 'dd, 
fem. ^ddit, plur. masc. 'ado, fem. 'dden with the 3rd person, 
and in the 2nd person sing, 'odt, plur. 'odto. 


In neo-Syriac the older tenses are almost extinct and are 
replaced by the participles, the present being expressed either 
by the participle followed by the personal pronoun as described 
above, or by the substantive verb followed by the verbal noun 
with prefixed he-. The future shows the participle itself 
with the prefix he-, bed-, or 6*-. The perfect has either the 
participle with the auxiliary verb »iOpO " finished ", or the 
participle with V- and the pronominal suffix, a form rare 
save in the dialect of Alqosh (cf. Maclean, Vernac. Syr., p. 82). 
The past imperfect is expressed by the participle with b'- 

and the substantive verb ^ooi. 

150 (g) The Participles 

The participles are deverbal noun forms. In the primary 
theme the active participle has the form of qdtil, thus Arabic 

Arli, Abyssinian qdtel, Hebrew /toip, Syriac ^\4^' Assyrian 

pdris. The primary passive participle appears in Hebrew as 
qdtul (7^tOp), Abyssinian qetul. Arabic adds the m- 

preformative and thus produces J^^a^. This form is obsolete 

in Aramaic, where we find *qdtil (^^jl^^), which is a common 
adjectival type, the adjective having displaced the passive 

In the derived stems the participles are formed by the 
addition of the preformative m- to the verb stem (cf. sect. 110 
above) ; in Abyssinian this is vocalized ma-, but in the other 
Semitic languages the vocalization is afiected by the phonetic 
principles already described. The exception to this is the 
participle of the passive in N. (Nif'al) of the Hebrew verb, 
which is derived directly from the perfect without the 
preformative m-, thus 7Dp''J- 


151 (h) The Infinitive 

The infinitive or verbal noun has very many various types ; 
indeed, it can take the form of almost any abstract noun, 
and this is especially true in Arabic. The following infinitive 
forms are the commonest : — 

(1) Primary stem. Assyrian ji^aras, corresponding to 
Hebrew qdtol, construct (ftol. Such a form appears amongst 
others in Arabic, but the commonest is qatl. In Abyssinian 
it is most often qatil, or qatilot. Aramaic prefixes m- (cf. 110) 
and frequently shows suffixed -u, thus meqtal, 7neqtalu. 

(2) Intensitive. Assyrian purrus, so Hebrew qattol, con- 
struct qattel. Aramaic m^qattalu. Arabic has here very 
various forms, but taqtil is one of the commonest. Abyssinian 

(3) Causative. Arabic 'iqtal ; Abyssinian 'aqtelo{t) ; 
Hebrew haqtil, construct haqtel ; Aramaic maqtalu ; Assyrian 

(4) N. passive. Arabic inqital ; Hebrew niqtol, or hiqqatol, 
construct hiqqatel ; Assyrian naprus. 

(iv) Verbs showing Phonetic Changes 

152 (i) Verbs with semi-vowel radical 

A^erbs having a semi- vowel as one of their radicals show the 
phonetic changes described in §§ 51, 52 above, but there are 
certain extensions and restrictions which must be noted in 

(A) Verbs with first radical semi-vowel 

When the first radical is without preformative no change 
takes place save that in Hebrew and Aramaic initial y becomes 
w, as stated in sect. 20 ; as closure after a short vowel in the 
imperfect, etc., the changes are normally those described in 
sect. 51. 


(1) Arabic. — (a) Imperfect and imperative. Verbs with 
initial w elide this first radical in the imperfect-imperative 

before stem vowel i ; thus imperfect yawlidu becomes Xh , 

imperative A) . This applies only to the active forms of the 

Primary stem. 
In the passive the stem vowel is a (cf. sect. 140). As a rule, 

these verbs have stem vowel a in the active perfect, i in the 

imperfect. There are, however, eight verbs having i in 

the perfect and also i in the imperfect (class (5) of sect. 142 

above) ; these are ^"j " trust in ", t_j»jj " inherit ", 
f-jj " refrain from ", * jj " swell ", ^^3 " be firm set ", 
^ J " be in good condition ", (Jj " be near ", and 

f^i " love ". All these having % in the imperfect elide 

the first radical w in the imperfect-imperative. 

A few verbs having a in the imperfect-imperative follow 
the same rule. In these it is presumed that they originally 
had imperfect in ^, but this has been changed to a by the 
influence of a final or medial laryngal. 


Thus f-i^J "let alone", imperfect f- A* ; so t«^J 

" give ", '^3 " fall ", 'J^5 " be wide ", etc. 

(b) Infinitive. Where the first radical w falls away as 
described above, and the imperative is formed from the 
shortened stem, the infinitive is formed from the same stem 
with the addition of the abstract termination -at (cf. sect. 117), 


tlius perfect AJj, imperfect Jlii , imperative -0, infini- 

tive d-U. 

(c) Reflexive of the primary. Assimilation of semi-vowel 
w ox y \jO the reflexive i takes place as described in sect. 30. 

{d) Dialectal peculiarities. In the ancient dialect of Asad 
-aw becomes -1 instead of remaining -aw as usual in classical 

speech, thus \>=-^ instead of /l>-^ ; this is a case of 

abnormal change of w to y, the intermediate Ast-J is also 
found. In the ancient dialect of the Hijaz this -aw became 

-a, and so we have A>-u for the same word. In the modern 
dialect of Tunis, Tripoli, and the town of Damascus wi becomes 

wu and so w, thus original ♦AJJ becomes wilid in these 

dialects, and thence ulid. 

(2) Abyssinian. — (a) Imperfect and imperative. The elision 
of first radical w described in Arabic (la above) takes place 
also in the Ethiopic verbs ivalada " bear child ", warada 
" descend ", wadaya " lie down ", wadaqa " fall ", and some- 
times in waqara " cut down " and wagara " cast out ", the 
shortened imperfect-imperative showing vowel a for I as 
though in double closure ; thus perfect walada, imperfect 
yalad, imperative lad. 

(b) The infinitive in these stems either adds afEormative 
-at or doubles the final radical, thus ledat or ladd. 

(3) Hebrew. — (a) In the imperfect-imperative stem con- 
traction takes place in the verbs "7^^ " bear child ", yi* 
"know", VT "go out", n^* "sit", pT "pour", and 
l^** " descend " (but Hl*^ in Q^ri of Ps. xxx, 4), and some- 
times in l^li'' " possess " and Ip"" " kindle ", thus imperfect 

HT, imperative *l7. 


Similarly with the verb *t7T] " go ", as though 'IT, imperfect 
*zp\ imperative 1| 7, the strong forms being found only in later 
Hebrew and in poetry as *n^nM (Ps. Ixxiii, 9). 

(b) The shortened form appears also in the construct 
infinitive of these verbs as in T\l7, the absolute retaining the 
regular form '2)^'', etc. Regular also is yi"* under the influence 
of the final laryngal. In 1 Sam. iv, 19, we find the assimilation 
-dt>-tt (cf. sect. 23). 

(c) Initial w becomes y unless doubled as in I'^V (Nif, 
imperfect) ; aw becomes (cf. sect. 50) as in l''^in, ^t^iX 
etc., and ay becomes e as in y't^T] (cf. sect. 49). 

(4) Aramaic. — Contraction is less common than in Hebrew, 

and occurs only in li[j "know", ^Lk " sit ", and »r30U " give ", 
this last showing elision of the h in the perfect sing. 3rd masc. 

»C(JL», 2nd A^qia, plur. 3rd masc. »squ, fem. . »«-^ fn.. 1st 

^CTU, the imperfect being obsolete and replaced by ^Aj. 
In the infinitive, where there is preformative m-, there is no 
need of the compensatory addition of -at, thus from »sAj 

we have jui^. 

In the causative initial radical y assimilates to w, whilst 

conversely in the intensitive initial w becomes y, thus causative 

r-^o(, r^Q*, mtensitive ,^, etc. 

• • • 

(5) Assyrian. — Contraction appears very generally in the 
imperative, as sib " sit " from wsb, etc. 

Regularly aw, iw become U (cf. 51, 52), but saw>se (cf. 58), 
thus usesab for usawsab ; iy, ay become I or e, as isir for 
aysir and eniq for ayniq. When awa, aya become a and uwa 
becomes u (cf. sect. 52), compensation is made by doubling 
the medial radical, thus awasab appears as assab. In the 
reflexive of the primary (Gt) w assimilates to t, thus awtasab 


becomes attasab, etc. (cf. sect. 30), and in the N. stem 
it becomes Hamza, to which n assimilates, thus anwasah> 
arCasah >a'^asah >a'asah. 

153 (B) Verbs with medial semi-vowel 

(1) Arabic. — In Arabic some medial wly verbs follow the 
regular type of strong forms ; such are (i) verbs which have 

final semi- vowel as well as medial, thus {S_y^ " roast " ; 
(ii) later derivatives such as denominal verbs, verbs specialized 

to express wonder, and elatives such as ij^\ "he speaks 
better than . . . " ; and (iii) stems where the semi- vowel is 

doubled as in conj. ii, v, J^9, etc., and so conj. iii, vi, where 

awa occurs as a dissimilation from awwa, etc. (cf. sect. 137). 
Generally we have the phonetic changes described in §§ 51, 
52. In the imperative the prosthetic vowel is not required, 
as the first consonant is vocalized by a following vowel. In 
the imperative and jussive the vowel resultant from con- 

traction is properly shortened, thus ASt , As, but this 
shortening is commonly neglected in dialect, as Egyptian 
qui " say ". ^ 

In the causative infinitive wd, yd become d after closure, 
but in compensation the stem receives the addition of the 

abstract afiormative -at, thus 43b I . 

In the active participle of the primary stem, dwi, dyi become 
d'i, but this assimilation is often ignored in modern dialect, 

where we hear (jdyil for JiU (Egypt, etc.), the semi-vowel 
assimilating to the following vowel. 


In the infinitive of conj. viii, iijd is maintained, but iwd 

is assimilated to it, so that Jl «,-91 becomes JL^dl . 

(2) Abyssinian. — No shortening or change takes place when 
medial wjij is doubled (cf . Arabic). In the imperfect the me dial 
semi-vowel is retained with inserted a in the indicative, and 
mi, yi are retained in the infinitive, otherwise the phonetic 
changes already described in §§51, 52 are carried out. 
Shortening in double closure is confined to Tigre dialect, 
where o, u, I, in this position become e, thus sd7na for sawama, 
becoming semka in the 2nd person. 

In the causative there are two alternative forms, either (i) 
the phonetic rules already described are observed, or (ii) 
the vowel is shortened, so that o becomes a and u becomes e, 
and thus we may have ^aqoma or 'aqdma, etc. 

(3) Hebrew. — Some verbs preserve the consonantal value 
of a medial semi-vowel, as TlH, Vlil, etc., but most verbs 

- T - T 

of this type show the operation of the phonetic principles 
described in §§ 51, 52. In the jussive the m or * resultant 
from these changes is shortened to o or e, and these in turn 
shorten to o, e, with prefixed Waw, thus D1p\ jussive Dp*, 
with Waw np^\ and so ]^n;, jl^, \y], etc. 

In double closure a, e become a, as Dp, P\t2p ; and o becomes 
o, as JJ^iS, nJi'3. In the intensitive we find the form D^ip, 
passive DDIp, on the analogy of the mediae geminatse 
verbs. In the derived stems verbs with medial semi-vowel 
are influenced by the analogy of verbs with weak final, thus 
in the passive in N. we have niZS^pX in which the -5- before 
the termination shows this influence ; so causative niD''pnj 
etc. Sometimes the same influence appears in the primary 
stems of verbs with medial y, thus from ^'2 we have 2nd 
sing. masc. perfect either Pi}'3, or niJ^'S. In place of this -o- 


we find -e- (from final -y) in the 2nd, 3rd fem. plur. imperfect, 
as njto^pn, etc. In all cases -e- (also from final -y verbs) 
is inserted before a consonantal pronominal suffix, as '•i^\'5'' 

In the present participle of the primary stem we find o 
in accordance with the regular phonetic change described in 
sect. 43c, but a with the semi-vowel changed to Ham^a also 
appears, thus tOI? (Isa. xxv, 7) and also tOX? for la't 
(Judges iv, 21). 

(4) Aramaic— No contraction takes place in medial wjy 
verbs which are also with final semi-vowel. In the intensitive 
stem all medial weak verbs are assimilated to medial y, and 
in the causative they are assimilated to medial w ; this is 
the normal course, but sometimes iv is retained in the 
intensitive, and sometimes we find the form *^QLoao as in 

In the reflexives of the primary and causative in Syriac the 
reflexive t is doubled, except after the personal prefixed ^, 
thus -V^.o/7] etc., by analogy with the roots having initial 
semi-vowel. In the active participle of the primary stem the 
semi- vowel becomes Hamza in the singular in Bib. Aram., 
thus D^5lp (Dan. ii, 31), but yp'^p in the plural (Dan. iii, 3, 
although Keth ^^^{p) ; in Syriac this Hamza is written, 
but is sounded as y, thus »^}o = qdyem. The verb ]- >> some- 
times follows the medial iv type, and sometimes the 
preformative receives compensatory lengthening, and thus 
we find both ]m£d and V»^|li>. 

(5) Assyrian. — Medial w/y verbs never retain the con- 
sonantal value of the semi-vowel, but follow the phonetic 
rules of §§ 51, 52. In the intensitive (D) the medial, though 


doubled, quiesces, and then either (i) the two vowels are left 
and form a long vowel or diphthong, or (ii) the first vowel 
falls away and the second vowel with the preceding semi- 
vowel follow the phonetic rules of quiescence ; thus from 
uqayyis we get either uqais or uqis. But when a vowel is 
added as suffix, the third radical is doubled and the vowel 
shortened, thus utawwiru becomes utlru and thence utlrru. 
Medial wjy verbs in Assyrian appear with any one of the 
three vowels a, i, u, as asdm, adln, atur. 

154 (C) Verbs with final semi-vowels 

(1) Arabic. — Verbs with final w conform to final y type in 
all the derived conjugations ; and in dialect final iv very often 
conforms to final y in the primary conjugation as well. The 
semi- vowels retain their consonantal value in final awa, 
aya, iwa, iya, and uya, but iwa becomes iya, and uya becomes 
uwa, the semi-vowel assimilating to the preceding vowel. 
In double closure aw, ay are retained, but in the dialects of 
Egypt and North Africa aw may become u, and in North 
Africa ay may become e/i before a consonantal suffix, as 
in saqet, nisit. 

In the active participle, -iy-u-, -iy-i become -i-, as in \>- for 

jX^, etc., and in the passive participle -uy-u- becomes 

-iyy-u- as in (c^r* jfor (c-**^, vulgar marmt. 

(2) Abyssinian. — No change is made with final awa, awu, 
ayu, but uw, of course, becomes u and iy >7. In Tigre awa, 
aya become a as in Arabic. In Tigriiia aw becomes 6, and 
ay>e before a consonantal suffix. 

(3) Hebrew. — Final w and final y are confused in one type, 
as -ay always with the vowel a, thus final -aw, -ay become -a, 
with suffixed -u (3rd plur.) -awu, -ayu become -u. 




Before consonantal terminations and suffixes -aw, -ay, -iw, 
-iy become -i in the perfect, -e in the imperfect. The infinitive 
construct adds T\-. 

(4) Aratnaic. — Final w and y are confused, and both confuse 
with final |, which has lost its consonantal value. Unlike 
Hebrew we find that Aramaic has verb stems with vowel a 
and others with i. 

(5) Assyrian. — Imperfect shows -u- with final -w, and -i- 
with final -y. 

Paradigms of Verbs with Semi-Vowel Radicals 
(i) Verbs with first radical w 

Arabic Abyssin. Hebrew Aramaic Assyrian 














Imperat. . 



























Imperat. . 























Imperat. . 












Passive in N 







Imperat. . 



(ii) Verbs with medial semi-vowel 
(a) Primary. Medial w. 






Act. Perf. . 











Imperf. . 





























Pass. Perf. 


Imperf. . 






(6) Primary. 

Medial y 


Act. Perf. . 




sdm (sim) 



Imperf. . 























(c) Intensitive. 

Act. Perf. . 

qawwama qawwama 




Imperf. . 















Pass. Perf. 


Imperf. . 










(d) Causative 


Act. Perf. . 




Imperf. . 















Pass. Perf. 



Imperf. . 






(e) Passive in 


Perf. . 



Imperf. . 












(iii) Verbs with final semi-vowel 

(a) Primary. Final w. 

Act. Perf. . tald talawa 

Pass. Perf. 











ih) Primary. Final y. 

Act. Perf. . ramd 

Imperf. . yarml 

Imperat. irmi 









Aramaic Assyrian 
^aqim (SD.) uskdn 







m^qim muskin 
m^qdmu sukun 










Pass. Perf. 

(c) Intensitive. 

Arabic Abyssin. Hebrew Aramaic Assyrian 

ramy- rameyot 




Act. Perf. 







rdme rdm- 

(ramma rammaya 


yuramnii yerammi y^ramme 

(fin. w) yefannu 

rammi rammey ramme 

muramm{in) nframme 

{tarmiyat) rammeyot rammoth 






n ramme urammi 

ramTnd rummi 

m^ramme murammu 
m^rammdyu rummu 

155 (D) Verbs with Initial Hamza 

The principal points to be noted about verbs with initial 
Hamza are (1) the treatment of Hamza following Hamza 
with a short vowel between, a case which occurs in the perfect 
causative and in the 1st singular imperfect primary, etc. ; 
(2) the tendency for these verbs to assimilate to those with 
first radical semi-vowel (cf. 10) ; and (3) the assimilation of 
Hamza to reflexive t. 

(a) Arabic. — Generally 'a'>'d, '*'>% and 'w'>'m. Thus 
causative 'dhadha for 'a'hadha, imperfect 1st sing, primary 
'dhudhu for 'a'hiidhu, causative 'iihidhu for 'u'hidhu, etc. 
The verbs 'akala " eat ", 'amara " command ", and ^ahadha 
" seize " follow the analogy of verbs with 1st radical w/y 
(cf. 149) and lose the first radical in the imperative, thus 
kul, etc. But this does not hold good after conjunctions wa- 
or fa- ; nor is this contraction extended to the imperfect. 
For the assimilation of Hamza to reflexive t cf, sect. 25. 

In Arabic dialect, chiefly in North Africa, there is a tendency 
to assimilate these verbs to final semi-vowel roots, as 3rd fern, 
sing, perfect Met for 'aJcalat, and hence 3rd masc. kid. 


(6) Abyssinian. — Hamza is retained when it is the initial 
of a syllable, but falls with compensatory lengthening when 
a closure (cf. 10) : thus causative perfect 'a'haza, imperfect 
'd'haza (soimded ^ahaza) ; imperfect primary (with inserted 
-a- in the indicative) ya'ahaz. Imperative primary ^ahaz. 

(c) Hebrew. — In the 1st person of the imperfect of the verbs 
^akal " eat ", 'dbadh " perish ", 'dmar " say ", 'dbd " wish ", 
'dfd " cook ", the 'a' becomes 'd and thence 'o, this being 
extended by analogy to the whole of the imperfect yokal, 
etc. ; occasionally similar forms appear in other verbs, and 
especially in 'dJiaz " take ". These analogies, however, are 
confined to the primary stem. 

(d) Aramaic. — Initial Hamza is not sounded in Syriac, 
and the following half-vowel is increased to a short vowel, 
thus 'ehadh for '^hadh. Otherwise the verbs with initial 
Hamza assimilate to those with initial y-, and so become as 
initial w- in the causative. For assimilation to reflexive t 
cf. sect. 25. 

(e) Assyrian. — The assimilations follow the lines described 
in §§ 10, 25, thus a'bat becomes dbbat (primary present), etc. 

156 (E) Verbs with the Assimilation of a Radical 

The assimilation of a radical has already been treated under 
the head of assimilation (cf. Ill, i). Only in the case of first 
radical n- in North Semitic does this lead to further develop- 
ments. As described (cf. 27), n as the closure of a syllable 
assimilates to the following consonant in contact in Hebrew, 
Aramaic, and Assyrian. This will happen in verbs ha\dng 
initial n- in the imperfect of the primary, in the causative, 
etc. Thus Hebrew primary imperfect ymgas becomes 
yiggas, yinpol becomes yippol, etc., and so Aramaic 
neppuq for nenpuq, etc. The noteworthy result is 
that from this imperfect is formed an imperative 
without the first radical, as in some verbs with initial semi- 


vowel (cf. 152), thus imperfect ijiggas (for yingas), imperative 
gas, etc. In Hebrew this shortened imperative is formed 
from imperfects which have vowel -a- or -i- {-e-), but it does 
not appear when the vowel is -o- (-w-), as n^fol, but in Aramaic 
and Assyrian it is extended to all stems. In Hebrew the 
verb \T\1, the only one with imperfect -i-, also ends in -n, 
and this final assimilates to a following suffix in contact 
(cf. 27). 

Hebrew Aramaic Assyrian 

nettel issur iddin 






Imperat. . 







Imperf. . 
Imperat. . 



Passive in n-. 


Imperf. . 
Imperat. . 




157 (F) 

Verbs Mediae Geminatae 

sab usur idin 


(a) Arabic. — (i) When all three syllables have the same 
vowel, the medial vowel falls out and madada becomes madda, 
etc. (cf. 74). Sometimes this is followed by analogy in the 
verbs with ifu after the medial radical, but more often these 
remain uncontracted. 

(ii) When the 2nd and 3rd radicals have vowels but the 
first has none (following), the second throws its vowel back on 
the first, so that yamdudu becomes yamuddu, etc. 

(iii) When the 3rd radical has no vowel there is properly 
no contraction, e.g. madadta, etc., but — 


(1) In tlie jussive and in the imperative tlie analogy of the 
indicative so far operates that the vowel of the 2nd radical 
is frequently dropped, thus ymndud becomes yamudd ; but 
as this leaves a final double without a vowel a supplementary 
vowel is added, a or i after a/i, and a, i, or u after u, thus 
yamudda, yamuddi, yamuddu ; and so the imperative madud > 
mudd > muddu, besides the regular imidud, etc. Persons of the 
imperative, formed directly from the 2nd masc. sing., 
according to this type, diverge from (ii) above, thus 2nd 
fem. sing, may be muddi from mudda, or umdudi from umdud. 

(2) Regularly the 2nd sing, perfect requires madadta, 
but colloquial speech is influenced by the analogy of madda, 
and thus forms 7naddayta, imitating the final -y verbs. 

(iv) Where the medial or final is duplicated (e.g. conj. ii) 
no elision is possible, thus maddada, imdadda, etc. 

(v) The above general principles hold good where the vowels 
are short, although regular forms also occur, as sakaka " to 
be knock-kneed ", qatata " be curly ", etc. Sometimes they 
are extended by analogy to cases where the vowels are long, 
but here the strong forms are commoner. 

(6) Abyssinian. — For the most part these verbs are treated 
as regular in Abyssinian ; only in the perfect, and, 
occasionally, in the imperfect-imperative, of verbs with e 
(for iju) do we find contraction, thus hamma for hamema 
" he was ill ", yenaddu for yenadedu " they burn ". 

(c) Hebrew. — (i) As in Arabic the medial loses a short vowel 
as sdbb for sdbdb ; and as the first syllable is now closed, there 
is no reason for lengthening the vowel, and where the 3rd 
radical is without a vowel the duplicate fails so that sdbb 
becomes sab, but sdbbu, etc. Imperfect 2nd fem. plur. 
tisbobnd becomes Vsubbend or tissobnd (cf. for this d<aw 
with inserted w, on the analogy of final semi-vowel stems, 
the inserted y in colloquial Arabic described above). The 


stem -suhb- in the former of these reproduces the original 

(ii) When the doubled medial is followed by a consonant 
suffix a connecting vowel is inserted : (1) before personal 
endings in the imperfect this is -5-, thus sabboilid, n^sabhothd, 
etc. (2) Before fern. plur. -nd of the imperfect-imperative it is 
-ey-. (3) Before pronominal suffixes attached to the 
imperfect-imperative we find -e- as in y^suhheni. 

(iii) WTien the medial vowel falls away and leaves the first 
radical with w or I this accented becomes o, e, as usual, e.g. 

(iv) When the short vowel is left in an open syllable 
preceding the accent it is necessarily either lengthened or 
the syllable is closed by doubling the following consonant, 
thus yi- (original ya-) in ydsob or yissob (cf. 47), 

(v) In these verbs Hebrew shows differences of vocalization 
in the imperfect, thus ydsob, with waw tvayydsob, suffix 

(vi) We find the regular intensitive in use, hillel, etc., but 
it is more commonly replaced by the qatal form, as sobeb. 

(d) Aramaic. — (i) In the perfect when the first radical has 
only a half-vowel and the second has a full vowel, the latter 
throws back its vowel to the former, the resulting double 
being treated as a single consonant, thus k^faf becomes kaf (for 
kaff), and so k^faft{a) becomes kaft{a), but keffath, etc., remain. 

(ii) In the imperfect the vocalization is the same as in the 
strong verb, but the second radical assimilates to the first, 
with which it is in immediate contact, thus 7iekfof becomes 
nekkof. Based on this analogy the imperative appears as 
kof. By analogy this treatment is also extended to the 
causative, and so 'akfef becomes 'akkef, etc. 

(iii) In the active participle we have kd'ef, as though perfect 
kaf represented a medial w/y stem. 



(e) Assyrian. — These verbs are regular in Assyrian, except 
only in the permansive, where we find sail, sall-at, etc., like 
the perfect in West Semitic. Dissimilation often occurs in 
the intensitive (cf. 32). 

Arabic Abyssinian Hebrew Aramaic 

(1) Primary. 



Act. Perf. 





2 masc. sing. 





Imperf. . 


, yajirru 








. farr 



Pass. Perf. 

. furra 

Imperf. . 

. yufarru 
. mafrur 



(2) Causative. 


Act. Perf. . 

. ^afarra 




2 masc. sing 





Imperf. . 

. yufirru 




3 fem. plur. 

. 'afrir 
. mufirr 












Pass. Perf. 



Imperf. . 

. yufarru 
. mufarr 






(3) Primary Refit 



Act. Perf. . 

, iftarra 



Imperf. . 

. yaftarru 




Abyssinian Hebrew Aramaic 

(4) Causative Reflexive. 


Imperf. . 

(5) Passive in N. 

Imperf. . 

(6) Intensitive. 

Act. Perf. . 

Imperf. . 
Pass. Perf. 

Imperf. . 




(cf. 32) 







158 (a) The Prepositions 

The prepositions generally govern the genitive case. In 
some instances these prepositions are noun stems originally 
construct (cf. 127), followed by the annexed genitive, as 
tahta " under ", accusative of taht-un " the lower part ", 
but we are not entitled to say that this was so with 
aU prepositions. There are certain prepositions of simpler 
type which show no evidence of having ever been other than 
particles prefixed to nouns. The preposition may govern the 
genitive of a noun or of its equivalent, i.e. noun clause or 
sentence, and in this latter position is followed by the relative 
pronoun (md) or particle introducing the subordinate clause. 
But the relative is often added to the preposition before a 
noun and does not, as a rule, affect its meaning. 

The principal prepositions are : — 

(i) " in ", denoting place, time, instrumental " with ", 
price, cause, etc., and " by " in oaths. 

ba-, Arabic hi-, Abyssinian ba-, baeda " in the hands of 
. . . ", Hebrew ¥-, ba-, bi-, and so Aramaic. 

In combination fl (Arabic for bi-fi " in the mouth of "), 
albild (Arabic), bHo (Hebrew) " in not ", i.e. " without ". 

Akin to this is : — 

bai/n " between ", Arabic bayna, Tigre, Tigrifia bayna, 
Ge'ez usually with ba- as babayna ; Hebrew ben ; Aramaic 
baynath or ben, bayn, beth ; Urmi bin, 'embJn ; Assyrian 
(ina) birit. 


ina " in ", Assyrian, and Abyssinian 'enta, Tigre 'et (for 
'ent). In all its uses this ina corresponds to West Semitic 
ba-, etc. 

(ii) " to ", denoting dative (recipient, owner, etc.), purpose, 
direction, towards, etc. 

la, Arabic la-, more often li- ; Abyssinian la- ; Hebrew 
la-, li-, l^- ; Aramaic /*-, etc., as Hebrew — same root as the 

'ild: "towards", "up to, until"; Arabic 'ild (/^Jl), 

Hebrew, Aiamaic 'el {7^, \>]). In Arabic J is not used 

in the sense of " towards " or " to " of place, but /_J I is 

employed exclusively in this meaning. 

ana : " to ", " towards " in Assyrian, equivalent to both the 

Combinations of la-, etc. 

Arabic (_^A) " at the hands of ", and so " near ", etc. 

Hebrew ''^Sh "at the face", i.e. "near", Arabic ^, 

i.e. bi-fh " in ", and Hebrew ^3? " near " (1 Kings xvii, 1 ; 
Hos. X, 12). 

Hebrew '•^y? " before the eyes of . . . ", " in the presence 
of ... " (Gen. xlvii, 19 ; Exod. iv, 30), Syriac .^^ 

Hebrew |y^7 " for the reason of ", i.e. " because ". 

(iii) " on ", " over ", of place, etc. 

'aid : Arabic /Jip, Hebrew Aramaic 7^, ^\l replacing 

7 in later Hebrew (Esther iii, 9 ; Job xxxiii, 23) ; Assyrian 
al, aid, all. 
(iv) " Uke ". 


ha : Arabic Jca-, Hebrew Jca-, ki-, k^-, and so Aramaic. 
Also in Arabic as kama, Abyssinian kmna, Hebrew IDS 

Aramaic y,], lios], Assyrian kl, klma. 
(v) " from ". 

min : Arabic ^a , Hebrew Vl2, Aramaic V'h, _Ld, 
Abyssinian 'am, 'amna. Denoting " from ", in Arabic as 


ex contrasted with ob (/%.£■), and the difference " than " 

after comparative. Only in later Arabic dialect and 
Abyssinian and Syriac for the agent. 

an : Arabic ^c- " from, by ". 

* • 

mundhu : Arabic X^a, >Xa " from, since " ; Abyssinian 
'emze ; Hebrew T^tt. 

T " 

istu : Assyrian istu, ultu " from ". 
(vi) " with ". 

ma' : Arabic >t4, m ; Hebrew D^ ; Aramaic ,i>Qi ; 
Assyrian ema. 

Hnda : Arabic Ale, accusative of Ale " side " ; Hebrew 

(vii) " under, below ". 

tahta : Arabic d*^ ; Hebrew T^HJ^ ; Aramaic A»2 ; 

Abyssinian hatte, hante (cf. sect. 75). 
sapal, Assyrian. 

(viii) " to, up to ". 

hatta : Arabic /c^>-, cs^^t in the dialect of Hudhail ; 
Hebrew 1]} ; Aramaic ^ ; Assyrian adi. 


^eska : Abyssinian. 
(Lx) "after". 

ha^da : Arabic A«.) ; Hebrew 1^3. 

159 (b) Prepositions governing clauses and sentences 

These prepositions may govern nouns or sentences which 
take the place of nouns. Thus we may say " he came to the 
town for plunder ", in which the preposition " for " governs 
the noun " plunder ", or " he came to the town because he 
desired to plunder it ", in which in Semitic the sentence " he 
desired to plunder it " is treated as a noun governed by the 
same preposition " for " followed by one of the particles used 

to introduce the subordinate sentence (j), J)1, etc.) 

or by the neuter relative (L.«, etc.). European dictionaries 

give lyi as meaning " because ", but, in fact, it is 

simply the preposition Q " for " governing a sentence 

introduced by the particle jl. Thus we get such forms as 

Arabic jlj , Hebrew p, Aramaic ^, ^p\ " as if ", 

Arabic l_^ , Hebrew 1^3, Onq. NtoJp, Abyssinian 

Jcama, Assyrian kiTua for " like as . , . ", and so with the other 
expressions used to introduce causal, etc., sentences which 
are simply the prepositions governing dependent clauses. 
In modern Arabic, and in Hebrew and Aramaic, however, 
such clauses are most often introduced by nouns in apposition 
governed by the same prepositions, thus causal " because " 
is introduced by hisdbah " by the reason ", or wm sabah 
"from the reason ..." ('Iraq), likawn or min kawn (Syria, 
'Iraq) from kawn " state, existence ", 'ala hdtir or fi hdtir 


(Algeria, etc.), based on the colloquial use oiyJ[>- " sake ", 
and 'ala §dn (Egypt) ; so Hebrew jy* " reason ", used for 
" because " (1 Kings iii, 11, etc.), or VO*h " on account of " 

followed by the relative or by the particle ''2 lJ) ) or 
j3 (jU " thus "), as ];h, l^-'?^ " therefore " ; Aramaic 

7 7 Tt 

**1^, fD, p ^io " wherefore ", etc. In Assyrian similarly 
the denominal preposition may or may not be followed by 
the relative, as aMu or assu sa " because ". 

160 (c) The Exclamatory Particles 

(i) Particles attracting attention. 

Such are the particles used to introduce the vocative, as 

Arabic u , 1 , b I , vlat I , Abyssinian 'o, or suffixed -a 

(Amharic Jioy, Tigre wo). In Hebrew no such particle occurs, 

unless we so regard the precative XI Aramaic has o|, .o]. 
Assyrian suffixes -a, -md, and -dmd/i. 

(ii) Particles directing attention, as Arabic jl, Hebrew 

tn^ Tliiri, Aramaic o], .], ](j\ ; Abyssinian 'ohd, or nd- 

with pronominal suffix. But these tend to produce verbal 
roots, as n^H " behold ", developing into the imperative 
*' look ", or are themselves imperatives in origin. 

(iii) Other particles, such as those (a) expressing aversion 

as Arabic t-51, c_SI, Abyssinian 'enbi, Hebrew H/vn, 

T • T 

Syriac ^oa ; (h) of lamentation as Arabic Ij, (_^JJ, Ljbl J, 


etc. Abyssinian wayleya, Hebrew "*iX, H^IK, T\^, etc., and 
Aramaic ^o, ]16, o], Ol] sliould be included, and even possibly 
those used as calls to animals, etc., though but little 
grammatical information can be obtained from these inter- 
jections (cf. demonstrative roots in §§ 89-9G). 

161 (d) The Negative Particles 

(i) la : Arabic V. In modern Arabic jI remains in general 

use for the absolute " no ", and in the form if j ; otherwise 
in Northern Arabic it is practically extinct, but survives in 
full force in the Arabic of South Arabia, Mehri, etc., as ?a, Id. 

Hebrew J^S, KlS, '?«. 

Aramaic N7, |]^ Q^. 
Assyrian Id, ul. 

Compounded ^, "without", Hebrew ^s*S!l (cf. J^SS), 
Aramaic ]Jj, Assyrian halil. With the impersonal verb, 
Hebrew (tJ'''N7), Aramaic ZuIL " is not ", Arabic ^J^^, 
Arabic ^) for jl i ; and ^3. 

(ii) hal : Arabic J.) , Hebrew 72, ^^3, vl? (Job xxxviii, 41), 

and other compounds, Aramaic ^5>^, - 'Vo 

(iii) ma : Arabic U , Abyssinian enclitic -m. This is the 
usual negative particle in later Northern Arabic, used even 

in prohibitions, with suflSxed -s (Morocco -si) from ^^ a 
thing ", as md tegis en-nahdrda " do not come to-day ", etc. 



(iv) 'in : Arabic j 1 ,^ Abyssinian 'en, in such forms as 
'en-dd'i " I do not know ", 'enbala " without ", etc., Hebrew 
V^, rX, used also as negative substantive verb ; '•J^ in N'^i"''i< 
(Job xxii, 30). 

(v) 'e, '% : Abyssinian 'e, '*, Tigrina 'ay, Assyrian ai, e. 

162 (e) Interrogative Sentences 

A sentence becomes interrogative (a) by a particular 
intonation of the voice, or (6) by the use of an interrogative 
particle, or an interrogative adverb or pronoun. 

The interrogative particles are : — 

(1) Arabic — I (negative ^1), Hebrew H, H, H (negative 

(2) Arabic JJti, ^A, ^ I , these last two with the perfect to 
chide neglect and with the imperfect to incite to performance ; 
Hebrew 711 in Deut. xxxii, 6 (according to the "'^^^Tini). 

(3) Abyssinian -hu, -nu, Assyrian -u. Thus Abyssinian 
yemaslak-nu " does it seem to you ? " sdnu, sohii " is it ? " etc. 

Interrogative adverbs. 

(!) "how?" Arabic <_ij, Egyptian dialect (_^ljl ; 
Abyssinian 'efo. Hebrew Aramaic 1133, ]ir>n " like what ? " 
n!D"'i^ and nS^i^ (cf. Abyssinian). Assyrian menu, minu/i. 


jidL>- L^-^iC' 1-_!<J jj-*."!' /j3 [y " there is no soul but over 

it is a keeper" (Qur. Ixxxvi, 4). 



(2) " why ? " Arabic 1] , i.e. " for what ? " Egyptian dialect 
'ay, lay, 'ala sdn 'ay (for 'ay cf. interrogative pronoun, 
sect. 106). Hebrew DD'?, V}1p, ri^S, n^'Sj;, Aramaic 

(3) " when ? " Arabic ^Z* ; Abyssinian fmze " what 
time ? " so Hebrew T\f2, Aramaic wiASo]. 

(4) " where ? " Arabic /j.t 1 , /j.* I ^-) I ; Abyssinian 'ayte ; 

Hebrew Aramaic "TJ"*J^, nb''^{, ^^^, QD^l ; Assyrian ai. 
For the interrogative pronouns cf . § § 102-6. 

163 (£) The Conditional Particles 

The conditional particles introducing the protasis or " if " 
clause are of two kinds — (i) those expressing " if " which 
introduces an uncertainty, or negative " if not ", as in the 
sentence "if he asks me for a dinar I will give it to him ", 
and (ii) those expressing " if " introducing a statement 
known or believed to be untrue, as "if he had asked me for a 
dinar I would have given it ", implying the fact that he did 
not ask. 

(i) " If " introducing uncertainty. 

Arabic Abyssinian Hebrew Aramaic Assyrian 

jj ( j*) 




(later) p 




■« or, 


(ii) " If " introducing statement known or believed to 
be untrue. 

Arabic Abyssinian Hebrew Aramaic 

"} sola 'b,h'^ ^ 

^ ■^ ^ .^ 

V^, 1) ^ )hb (Jdg. xiv, 18) 

Aramaic does not use oL alone, but only in the combination 

Assyrian lu is used in the sense of " or ", and so does not 
correspond with Arabic jJ, etc. 



Absolute Pronoun, 139. 
Abyssinian, 22-3, etc. 
Accent, 93. 
Accusative, 203. 
Adverbial Accusative, 203. 
Afformatives to Nouns, 187, etc. 
Ablame, 9. 
Aktionsart, 19, 235. 
Amharic, 23. 
Annexation, 198. 
Arabic, 16, etc. 

— loan words from Syriac, 87. 
Aramaic, 13, etc. 

Arimi, 9. 

Aspiration, 88. 

Assimilation of Consonants, 68, etc. 

— of Vowels, 126-7. 

Babylonian, 7, etc. 
Berber, 212, 220, 228. 
Bishari, 214< 228. 
Broken Plurals, 193. 


Canaan, 9. 

Cases, 195. 

Causative, 218. 

Comparison of Adjectives, 206. 

Conative, 217. 

Conditional Particles, 275. 

Construct, 198, 205. 

Contraction of Vowels, 116. 

" Covered " Consonants, 56, etc. 


Demonstrative Pronouns, 158, etc. 
Dentals, 29, 53, etc. 
Determination, 204. 
Dialects of Arabic, 18, 20-1. 
Diphthongs, 113-16. 
Dissimilation of Consonants, 81, etc. 
— of Vowels, 127, etc. 
Dual, 195. 


Egyptian, 5, 139, 140, 176, 183, 212, 

214, 220, 228. 
Elision, 135. 
Emphatic, 30. 
Energetic, 238, 240. 
Exclamatory Particles, 272. 

Final Group of Consonants, 133. 


Galilsean Dialect, 47. 
Ge'ez, 1, 22-3. 
Gender, 191. 


Half-vowel, 92. 
— inserted, 132. 
Hamitic, 5. 
Hamza, 30, etc. 
Haplology, 135. 
Hebrew, 11-12. 
Hijaz, 18. 
Hims Dialect, 82. 

Imale, 95. 

Imperative, 246. 

Imperfect Tense, 244. 

Inception, Group of Consonants, 129. 

Indicative, 238, 239. 

Infinitive, 250. 

Informatives, 186. 

Inserted Vowel, 129. 

Intensitive, 211-17. 

Interrogative Pronouns, 173. 

— Forms, 274. 

Ismam, 91. 

Jussive, 238-9. 



Labials, 30. 

Language, how transmitted, 3. 

Laryngals, 29, etc. 

Lateral Consonants, 63. 

Long Vowels, 94-103. 


Mandaean, 16. 
Medial Group, 132. 
Metathesis, 137. 
Moabite, 12. 
Moods, 238, etc. 


Nabatoean, 13. 
Negative Particles, 273. 
Nejd, 18. 
Nouns, 175. 
Number, 192. 

Palatals, 29, 50, etc. 
Participles, 249. 
Participial Tenses, 247. 
Passive, by Vowel Change, 233. 

— with n-, 223. 
Perfect Tense, 234-5, 241. 
Permansive, 246. 
Phoenician, 6, 10, 56. 
Plural, 192. 

— Formatives, 193. 
Preformatives to Nouns, 180. 
Propositions, 269, etc. 
Primary Theme, 211. 
Pronouns, Personal, 139. 
Prosthetic Vowel, 129. 


Reflexive, 225. 

Relative Pronoun, 172. 

Relative used to denote Genitive, 200. 


Samaritan, 14. 
Semitic, 1, 2, 4, 6. 
Semi-vowels, 30, etc. 
Sentence defined, 24. 
Short Vowels, 103-13. 
Sibilants, 29, 53, etc. 
Sonants, 30. 
Subjunctive, 238, 239. 
Suffixed Pronoun, 149, etc. 
Sumerian, 7-9, 50. 
Syllable, 24, etc. 
Syriac, 14. 

Targums, 14. 

Tenses, 19, 229, 230, 231, 234, etc. 

Themes of the Verb, 209. 

Tigre, 23. 

Tigrina, 23. 

Time Relations in Tenses, 247. 

Transcription, 27-9. 

Velars, 29, 48, etc. 

Verb, 208. 

Vocabulary distinct from Language, 4. 

Voiced Consonants, 30. 

Voiceless Consonants, 30. 

Vowels modified, 120. 

— influenced by Consonants, 120. 

Zinjirli Inscription, 56. 







. 237 


. 235 


. 236 


. 235-6 


. 237 





. 236 















































Judges . 

. xiv. 

18 . 





15 . 





9 . 





Ruth . 

• ii, 

8 . 




1 Sam. . 

• i. 

6 . 




20 . 





19 . 



amar. vers.) 152 


3 . 





8 . 






19 . 



. 222 


5 . 






15 . 





2 Sam. . 


5 . 





1 Kings 

• ii. 

36 . 



60 (S 



11 . 





2 Kings 

• iv. 

25 . 






14 . 



. 236 

1 Chron. 

. xiii. 

12 . 




. 218 

2 Chron. 

. XX, 

35 . 



. 245 


. iv. 

13 . 




. 190 




. 237 





15 . 




. 269 

16 . 





3,9 . 





17 . 




. 218 


2 . 




. 209 

14 . 




. 237 


17 . 





26 . 




. 236 

Esther . 

. iii. 

9 . 




. 153 

Job . . 

• i. 

5 . 




. 219 


11 . 




. 227 


3 . 






15 . 






30 . 




. 213 


6 . 






23 . 



25 . 





41 . 







8 . 






4 . 






6 . 






11 . 





9 . 





. 274 


4 . 



9 . 





3 . 

4 . 




. 203-4 


12 . 




. 165 


6 . 












8 . 



3 . 



26 . 



5 . 


Ecclea. . . 


6 . 



2 . 



16 . 


Song, of Sol. 


6 . 


Isa. . 








2 . 



5 . 



6 . 


17 . 



7 . 






1 . 



10 . 



13 . 



2 . 



17 . 





Jer. . . . 























25 . 


Lam. . . 

















. 206 






. 107 




. 216 


. 256 


. 125 


. 245 



. 222 


. 256 











. 182-3 





. 110 


. 214 



. 125 




. 222 





. 245 






. 160 



. 109 


. 179 






. 201 


MAY 41992