in the Old Dialects of Arabic
Abiel A. Bloch, Berkeley
A central problem in Arabic linguistics is that of the relationship
between the modern Arabic dialects and Classical Arabic. There exists
no unanimity of opinion on this question. However, an agreement should
now exist on one methodological point: The dialects are not to be
considered descendants of Classical Ai-abic but rather its contemporaries
throughout history. This implies that not only Classical Arabic which is
acknowledged to be highly conservative, but also the dialects can
reflect the older stage of development of a given linguistic feature. The
dialects, therefore, must be considered just as important for linguistic
reconstruction as is the Classical language.
In a paper entitled 'The Arabic Koine', Chables Feeoijson some
time ago made the following observation: A number of inflectional
affixes with the vowel a in Classical Arabic have in the modern dialects
reflexes of the vowel i (including zero in cases where i was reduced to
zero).* Out of a couple of such cases we have chosen to discuss here the
one with the greatest linguistic importance : The modern Arabic dialects
have in the preformatives of the imperfect of the simple form a vowel i
as against a in Classical Arabic. Thus, for instance, as against Cl. 'aktubu,
takttibu, yaktubu etc. modern Syrian Arabic has 'dklob, tdktob, ydktob etc.
(with a shwa vowel representing i in this position). In the following,
Arabic imperfect forms will be referred to as i-imperfect and o-imperfect according to the vowel of the preformatives.
Since Ferguson presupposes that Classical Arabic represents the
older stage he consequently concludes that the dialectal i-imperfect
represents a secondary development. As he points out, this i for a can not
be the result ofa general sound change or a morphologically conditioned
change because a is preserved in analogous positions in the dialects^
Ferguson considers the dialectal i-imperfect, together with othe»
features in which the dialects agree against Classical Arabic, a proof ol
his theory of a common, non-Classical origin of the dialects (which h^
calls the 'koine').
Although Ferguson's paper contains quite a number of highly valuably
observations* the argumentation is not conclusive because the idea o}
* Ch. Febguson: The Arabic Koine. Language 35, 1959 pp. 616— 63Q
(p. 621).
' In a forthcoming paper (to appear in the JAOS) the present writer
hopes to discuss some further points in Ferguson's article. ,
The Vowels of the Imperfect Preformatives 23
the existence of an Arabic koine was shown to be unprovable.* In our
special case the explanation is different : The older stage of the dialects
is not reflected by Cl. Arabic with its a-imperfect, but by some old
dialects which show two kinds of imperfect, one with a and another
with i in the preformatives, in agreement with what is known as Barth's
Law.
The Arabic grammarians already observed the existence of a dialectal
' i-imperfect, a phenomenon which they called taltala* According to them,
^ the Eastern dialects in the Najd and Iraq had i-imperfect whereas the
^ Western dialects in the Hijaz and its surroundings had a-imperfect (as
^ Cl. Arabic). We can not rely too much on their statement as to the
s extent of the feature within each of these two major dialect groups
^ because of their tendency to substitute a larger tribal grouping for a
6 smaller one.* For this reason our reference to i- and a-imperfect as an
c 'Eastern' and a 'Western' phenomenon, respectively, is meant only
schematically.
6 The grammarians' statement concerning the East was usually under-
'1 stood by modern dialectologists to mean that these dialects had
^ only i-imperfect. However, whereas this is true for a later period, one of
o the earliest statements, made by Sibawaih (8th Cent.), shows that the
^ East had both kinds of imperfect. It is his statement that sheds a most
important light on the phenomenon under discussion.
* Dealing with the phenomenon in his Kitab (ed. Derenbourg, vol. II,
^> p. 275), Sibawaih adduces for the Eastern dialects a list of verbs in the
5- i-imperfect with their corresponding perfect forms. This list reveals some
51 important facts :
't
1. All the i-imperfects have a as their stem vowel (that is to say, they
all follow the pattern yiqtalu).
te
3^ 2. The corresponding perfects all follow the pattern qatila.
>t 3. The verbs making up this pair (qatilalyiqtalu) have meanings like "to
be or become miserable, to be afraid, to imagine" etc., which indicates
8> a class-meaning usually referred to as 'stative' (corresponding to
Hebrew käbedjyikbad). (Of course, the original meaning of this class
3l is for us no longer transparent, and therefore the term 'stative' is
i<i used here with reference to the form rather than the meaning).
U 3 Qf_ ]3_ Cohen: Koini, langues communes et dialectes arabes, Arabica 9,
1962 pp. 119—144, and the first chapter of J. Blau: The Emergence and
lAnguistic Background of Judeo-Arabic, Oxford, 1965 and in Tarbi? 30, 1960
pp. 133—135.
* Ch. Rabin: Ancient West Arabian. London, 1951, pp. 61—63.
* Ch. Rabin op. cit. p. 11.
4. Sibawaih adduces one pair in each class of the strong and the weak verbs (e.g. SaqitajtiSqä, hilnäjnihälu etc.).
This last fact clearly indicates that the 'stative' class represented by
the pair qatilal yiqtalu was still productive in Sibawaib's time.
The correspondence qatilalyiqtalu has one exception, mentioned by
Sibawaih. In the third person the vowel is not i but a, thus ti'lamu,
ni'lamu etc. but ya'lamu. This has to be explained as a dissimilation of
the sequence yi- which is avoided in Arabic* In any case, this does not
invalidate Sibawaih's basic statement because at least a few tribes had
i throughout, including the third person.'
On the other hand, Sibawaih reports (op. cit. p. 276. 5) that the
Eastern dialects had qatala in the perfect to wliich corresponded only
a-imperfect, not i-imperfect.^ To this class we shall refer as the 'active' class (in the same sense as above).
Sibawaih's statements reveal for the Eastern dialects of Arabic what
J. Baeth has proved for Hebrew and Aramaic* and what later received
strongest support from the facts in Ugaritic*' : The coexistence of two
forms of the imperfect, one with i in the preformatives and a in the
stem for the 'stative', and another with a in the preformatives and i or
u in the stem for the 'active'. According to Baeth, the individual lan¬
guages which inherited the phenomenon from Proto-Semitic, at a later
stage gave up the differentiation by standardizing one of the imperfect
forms which then took over the function of the other: In Hebrew and
Aramaic it was *-imperfect, in Arabic a-imperfect.
The existence of the phenomenon in the old Eastern dialects of Arabic
was recognized by Baeth in his historic article** and by a few later
• Cf. H. Fleisch: TraitS de Philologie Arabe, vol. I, Recherches publikes
sous la direction de I'institut de lettres orientales de Beyrouth, tome XVI,
Beyrouth 1961 p. 137.
' Cf. H. Fleisch loc. cit. Ch. Rabin: Archaic Vocalization in Some Bibli¬
cal Hebrew Names, Joumal of Jewish Studies, 1, 1948, p. 26.
« At this point it should be interesting to consider how Sibawaih ex¬
plains the phenomenon in the East. For him the i-imperfect is the result of
an adaptation of the preformative vowel to the vowel of the second root
consonant of the perfect "just as they had o-imperfect whenever the vowel of the second root consonant of the perfect was a" (op. cit. p. 27.5.18). Such
an explanation is typical for the linguistic method of the classical Arabic
grammarians.
» J. Baeth: den Vocalen der Imperfect- Präfixe, ZDMG 48, 1894, p. 4—6.
*° Cf. H. L. Ginsberg, in Tarbi? 4. 1933, p. 182, and in Orientalia 8, 1939.
p. 318; Z. S. Habeis, in JAOS 57, 1937, p. 153; E. Hammbeshaimb : Das
Verbum im Dialekt von Ras Schamra, Kopenhagen 1941, pp. 108, 181; C. H.
Gordon: Ugaritic Textbook, Analecta Orientaha 38, Roma 1965, p. 71.
** By basing himself on scanty and poorly transcribed dialect material
Barth thought that the phenomenon exists even in the modern dialects.
The Vowels of the Imperfect Preformatives 25
semitists,** but it never became generally known in modern Arabic
dialectology. Here the fact was either not recognized, or it was recognized
but misinterpreted. It is the aim of this paper to point at the existence
of the phenomenon and to shed light on the consequences for both
the old and modern dialects as well as for Classical Arabic.
But first we have to ask whether we can rely on our source. We know
that the Arabic grammarians quite often falsified and distorted the data
for the sake of schematization or for some other reason. Is it possible
that this is the case with the statement under discussion? We think
that in the present case this is highly improbable. It is very unlikely
that a grammarian's invention or error should coincide with a well
established linguistic law involving rather complicated relations of form
and meaning. A real difficulty, however, seems on first sight to arise for
Sibawaih's rule from another direction. Some non-canonical Koran-
readings reported by commentators of later centuries and declared by
them as Eastern contradict Sibawaih's rule; for instance: ni'budu "we
worship" (Sura 1. 4) for nu'budu and similar forms.** However, these
cases do not invalidate Sibawaih's statement but supplement it. The
system of the twofold imperfect did not live on, but was followed by a
stage in which one form, the i-imperfect, was standardized. The men¬
tioned Koran-readings reflect this stage. They may belong to a post
Sibawaih period which had only i-imperfect. Or, the process of stan
dardization of i-imperfect may have started akeady in the time of Siba¬
waih. If this was the case we should not wonder why he did not mention
these forms: They were stih outside the system of the normal corres¬
pondences. Or, stated differently: In Sibawaih's time each qatala had
its corresponding a-imperfect, whereas i-imperfect was — if at all exi¬
stent in this function — still the irregular (not yet productive) corres¬
ponding imperfect of qatala.
We sum up at this point : Sibawaih's statement proves that Barth's
Law was fully or at least partially operative in the Eastern dialects of
Arabic in the 8th Cent.
We come to the question of the historical evaluation. Chaim Rabin
who recognized the feature in the Eastern dialects, assumed that it
12 Cf. especially Chb. Sabatjw, in ZA 21, 1908, p. 46; H. Fleisch loc. cit.
(n. 6) and in Etudes de phonitique arabe, Melanges de l'universitö Saint
Joseph, tome XXVIII, fasc. 6, Beyrouth 1949—1950, p. 275; Ch. Rabin
loc. cit. n. 7 and n. 14. In this connection mention should be made also of
Th. Nölreke's remark in WZKM 9, 1895, p. 16 n. 1. The article was written
in the year of Babth's discovery (1894) when Nöldeke apparently was still
unaware of it. He mentions Sibawaih's statement, but he does not apply it.
lä Cf. K. VoLLEBS : Volkssprache und Schriftsprache im alten Arabien,
Straßburg 1906, § 26c.
emerged secondarily, namely under the influence of some Canaanite
or Aramaic dialect in Palestine which he supposed to have been in
contact with the Eastern dialects of Arabic.** We can not agree with
this opinion. To maintain a secondary development of such a feature
means nothing else but to maintain that the verbal system of the
Eastern dialects of Arabic reorganized itself after the pattern of a neigh¬
boring dialect. It is not thinkable that verbal systems spread like words
from one language to another. Since Rabin does not accept the primary
character of the feature he necessarily has to apply the same explanation for other languages of the area as well.**
Although the processes of standardization and levelling have effected
Classical Arabic more than any other Semitic language, it still betrays
** Cf. Ch. Rabin loc. cit. (n. 7), and also his article: Tlie Ancient Arabic
dialects and their Relation to Hebrew (in Hebrew), Melilah 2, Manchester
1946, p. 252.
" The assumption of a secondary development of the feature in the
Eastem dialects is for Rabin necessary within his basic theory on the
old Arabic dialects. According to this theory "Classical Arabic is on the whole based on an archaic form of Eastem dialects, and if the differentiation between the prefixes of o-stems and others had been an old-inherited feature
of those dialects, there is no reason why Classical Arabic should not have
taken it over" (Rabin loc. cit. n. 7). The weakness of the argumentation is
in its being based on two assumptions: The Eastern origin of Classical
Arabic and its never having known the preformative differentiation. The
first is Rabin's own theory and the second is contradicted by traces of the
feature in Classical Arabic (see the following above). The whole phenomenon is explained by Rabin in this way: "The use of i in the prefixes of a-imper-
fects (that is. in Rabin's terminology : imperfects with stem-vowel a. My re¬
mark) . . . must have begim some time before 1500 B.C. somewhere in
Syria or Palestine, probably in a Canaanite dialect, and spread to Aramaeans and Arabs who were in contact with Canaanites" (loc. cit.) If the use of i in
the preformatives really spread over this whole area why did it always do
so only in the verbs with a-stems and never in those with i- or w-stems?
This would mean that in all these languages the verbal system reorganized
itself after a foreign pattern, and always in such a way that effected only
a part of the paradigm of the imperfect (verbs with o-stems), whereas the
other part (verbs with i- and M-stems) remained unchanged. Such a spread
of a differentiation within the imperfect over a large area and over very long
periods is most unlikely. Rabin upholds his opinion also in his later and
basic work Ancient West Arabian (loo. cit. n. 4; cf. also his: The Origin of
the Subdivisions of Semitic. Hebrew and Semitic Studies Presented to G. R.
Driver in Celebration of his Seventieth Birthday, 20 August 1962, ed. by
D. WrNTON Thomas and W. D. McHabdy, Oxford 1963, p. 109). It is not
within the scope of this paper to discuss Rabin's theory itself (conceming the origin of Cl. Arabic and its relation to the old dialects), but we maintain
that the arguments concerning the imperfect preformatives are an incon¬
clusive element in it.
The Vowels of the Imperfect Preformatives 27
some evidence of a former existence of Barth's Law. The usual corre¬
spondance in Classical Arabic is
qatala yaqtijulu
qatila yaqtalu
Of course, there can be no direct evidence for Barth's Law because
Cl. Arabic has only a-imperfect. However, two facts are revealing: The
pairs show that the imperfects with i and u as the stem-vowel malje up
a group against the imperfect with stem-vowel a. This, together
with the fact that the pairs mostly have 'active' and 'stative' functions,
respectively, supports our assumption.** An old-inherited differentiation
in the preformative-vowels as preserved in the old Eastern dialects
may have been eliminated in Cl. Arabic by a generalization of a-pre-
formatives, just as Hebrew, Aramaic and the later Eastern dialects of
Arabic have generalized i-preformatives.*' The verb 'ihälu "methinks",
the only case of an t-imperfect in Classical Arabic, also suggests an old
i-imperfect with 'stative' function in an earlier stage.** However, the
The function of the second pair, it is true, is not uniform : We may
distinguish on the one hand (according to H. Fleisch: L'Arabe Classique,
Esquisse d'une Structure Linguistique, Recherches publiees sous la direction
de I'institut de lettres orientales de Beyrouth, tome V, Beyrouth 1956.
p. 90, 91) verbs like kabira/yakbaru „devenir vieux", and on the other
hand verbs like rabiha/yarbahu "gagner", dahika/yadhaku "rire". The first group is characterized by Fleisch as "sans agent: verbe de qualitö", the second as "agent Interesse: verbe tr. ou intr." However, important for our
discussion is only the fact that the function of the first pair (according to
Fleisch "agent pur et simple: verbe tr. ou intr.") is clearly different
from the function (or functions) of the second pair. We may, thus, label the
second pair negatively as "non-pure-and-simple actor."
*' According to a source other than Sibawaih, Barth's Law must have
been preserved also by some Western dialect, namely by a part of the
Hudhail, Rabin op. oit. n. 4, § 8aa. The fact is not as well attested as by Siba¬
waih for the East because the source, 'Abü 'Amr (d. 154), does not adduce
examples; but there is no reason to doubt the fact itself. ('Abü 'Amr agrees
with Sibawaih by reporting the feature also for the Eastem dialects. On
the other hand, Sibawaih does not report the feature for the West, probably
because it was present only in that minimal area). The statement is most
valuable because it once more corroborates the primary character of the
feature by attesting it also in the West thus making plausible its former
existance all over the Arabic dialect area. According to the same source,
the other part of the Hudhail had only a-preformatives like the rest of the
Western dialect area. We assume, therefore, that in the other parts of the
West the o-preformatives have been generalized, like in Classical Arabic.
*' The possibility that this verb is a loan-word from an Eastem dialect
must, however, be taken into consideration — although this is minimized
by statements of Arabic grammarians who declare it as "better Arabic"
than the variant 'ahOlu, cf. Fleisch loo. oit. n. 6. Rabin, in accordance with
situation in Cl. Arabie is mucb less clear cut than in the Eastern dialects.
Corresponding pairs like
seem to contradict our assumption. However, the two first pairs are
very rare,** in the third pair yaqtalu mostly goes back to an 'active'
yaqtilu in which i> a because of a gutturaP", and for the last pair there exist a few variants with yaqtalu which seem to point to an old 'stative'.**
These facts seem to indicate that these pairs were not part of the original
system. Nevertheless, even iftheymaybe explained partially as secondary
by assuming thorough schematization, conditioned sound change, etc.,
there still remains uncertainty. Only if we know more about how these
factors really worked in every single case, may we be able to prove the
original existence of Barth's Law also in these cases. Here, also, old
dialectal, less schematized verbal forms may shed light on older stages.
To sum up we return to our starting point: Taking for granted that
Classical Arabic preserved the old stage, Chaeles Feeguson assumed
that i-imperfect was secondary. He attributed the fact to an unprovable
koine. We know now that i-imperfect is not secondary but existed side
by side with a-imperfect, the forms being distributed according to Barth's
Law, and that dialectal i-imperfect is the result of a generalization.**
his theory, assumes secondary development by dissimilation of the pre-
formative-vowel to the following a, op. cit. n. 4 p. 90. However, the fact
that this verb is also attested in the old Western dialects rather seems to
point to an old origin.
" Cf. H. Fleisch, op. cit., n. 16, p. 89, n. 1.
2» Cf. H. Fleisch, op. cit., n. 16, p. 89, n. 2.
21 Cf. W. Wright: A Orammar of the Arabic Language, Cambridge 1955,
vol. 1, p. 59B.
22 Other modem semitists usually attributed the i-imperfect to Aramaic
influence, cf. I. Gabbell: Remarks on the Historical Phonology of an Eastern
Mediterranean Arabic Dialect, Word 14, 1958, p. 312. The contact with
Aramaic may, of course, have had some influence on the process of
generalization of the i-imperfect, but it could never have been the
reason for its first occurrence in Arabic.
What has been said above does not imply that a secondary differentiation
of the preformative vowels is altogether impossible. But if such a de¬
velopment has actually taken place, it must have occurred at a very early
stage before the single languages involved had split apart. Since we know
too little about the proto-Semitic verbal system, any suggestion of that
kind necessarily remains a hypothesis; cf., for instance, J. Kurylowicz :
L'apophonie en Simitique, pp. 30, 31. (I am indebted to H. Fleisch for this
reference.)
qatila qatila qatala qatula
yaqtilu yaqtulu yaqtalu yaqtulu
t t
The Vowels of the Imperfect Preformatives 29
Barth's Law was proved to be fully or at least partially operative in the
old Arabic dialects. In Classical Arabic, traces indicate its former exi¬
stence. Thus, Arabic as a whole shares here an archaic West Semitic
phenomenon found also in Ugaritic, Hebrew and Aramaic. The fact
that Barth's Law was found in the dialects not as a fossil left over from
an archaic period, but as a living, operative feature, shows the great
importance of the Arabic dialects for the history of the Semitic languages.
von Wolfdieteich Fischee, Erlangen ^
In allen Sprachgemeinschaften herrschen bestimmte, von Sprache
zu Sprache, oft von Mundart zu Mundart unterschiedliche Artikulations¬
gewohnheiten, die die Lautfolgen in bestimmter Weise gliedern und die
theoretisch möglichen Phonemkombinationen einschränken. Die not¬
wendigen Folgen von Öffnung (Explosion) und Schließung (Implosion)
der Ai'tikulationsorgane beim Sprechen gliedern im Arabischen die Rede
in Artikulationseinheiten, die man herkömmlich als Silben bezeiclmet*.
Die zulässigen Silbenstrukturen, d.h. die gewohnheitsmäßig festgelegten
Folgen von Explosion und Implosion, bilden einen wesentlichen Teil
solcher einschränkender Artüvulationsgewohnlieiten, auf Grund derer
nur einige Typen von Lautfolgen möglich sind.
1. Die Silbenstruktur des Altarabischen
Das Altarabische, wie es in der im Klassisch-Arabischen fixierten
Gestalt vorliegt, weist einen außerordentlich einfachen Silbenbau auf.
Er weist folgende Silbenelemente auf:
< = Explosion
> = Implosion
+ = neutrale Öffnung
Das Altarab. kennt nur zwei Arten von Silben, deren Struktur mit diesen
Elementen folgendermaßen dargestellt werden kann :
1. offene Silbe < +
2. geschlossene Silbe < + > i
Bezüglich der Realisation der Silbenelemente < und + und > durch
bestimmte Phonemklassen gelten die Regeln :
< ist immer Konsonant (x).
' Als Artikulationseinheit läßt sich die „Silbe" nicht allgemein, für alle
Sprachen gültig definieren. In den verschiedenen Sprachen sind sehr unter¬
schiedliche Artikulationsmerkmale für deren Gliederung in Silben fest.stell-
bar. Für das Arabische und wohl auch für das Semitische überhaupt erweist
sich die von F. De Saussuee, Cours de linguistique gdnirale (1922), 79 ff.
verwendete Theorie der Öffnungssilbe unmittelbar geeignet. Sie wurde hier
mit einer kleinen Modifikation, der Einführung eines hinsichtlich Explosion
imd Implosion neutralen Silbenelements (-f), übernommen. Insbesondere
im Neuarabischen zeigt sich, daß die Schallfülle für die Silbenbildung völlig irrelevant ist, sodaß Silbentheorien, die auf diesem Artikulationsmerkmal
aufgebaut sind, zur Beschreibung der Silbe im Arabischen nicht verwendbar
sind.