MUSNAD, MUSNADILAYHI AND THE EARLY

HISTORY OF ARABIC GRAMMAR:

A RECONSIDERATION*

By RAFAEL TALMON

A. Various opinions about the existence of a subject-predicate concept in Arabic

grammar

Students of the early history of Arabic grammar may note with surprise that, until
now, studies dealing with the possibility of Greek influence on the emergence of that
Islamic scholarly discipline have hardly ever made an issue of the fact that the
Hellenistic grammatical tradition as well as the Greek philosophical legacy differ
significantly from the Arabic grammatical tradition in their conception of the main
parts of the sentence.
Of course, this does not mean that the actual difference in conception was not
acknowledged or discussed. Consider, for example, Weil's observation, formulated
long ago (1915), that the dichotomous division of sentences in the theoretical model
of Arabic syntax into jumla ismiyya vs. jumlafi'liyya has resulted from the absence
of a general sentence concept in that model.1 The unitary pair onoma + rhema
(Syriac: shmd + melltalmemra) appears as quite alien to the dichotomy of
mubtada'+ mabniy 'alayhi ws.fi'I + fail, so typical of the Arab grammarians' model.2
However, neither Weil nor later scholars drew any conclusions from this observation
relevant to the famous controversy over the so-called hypothese grecque.
The revival of scholarly interest during the lastfifteenyears in this alleged peculi-
arity of the Arabic model has so far touched upon the "Greek Connection" only in
an indirect way, as we shall show below. For this reason alone it would be worth-
while to examine the implications of these recent studies of the Arab grammarians'
attitude to the main parts of the sentence, in order to reconsider some long-standing
opinions expressed in favour of the hypothese grecque. Specifically, I am referring
to those studies which deal with the use of the terms musnad and musnad ilayhi in
Arabic grammar. Against the critical observations of Weil and others, some scholars
have defended the existence of a general sentence concept in Arabic grammatical
thinking by referring to the presence of the pair musnad (= m) and musnad ilayhi (+
/n.i.) already in the introductory part of SIbawaihi's Kitab, and by identifying the
two terms as the equivalents of subject + predicate.3

B. Levin's conclusions (1981)

Against this background, Aryeh Levin's study of the grammatical terms musnad,
musnad ilayhi and isriad* in the terminology used by Sibawaihi and the later gram-

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 MVSNAD, MUSNADILA YH1 AND THE EARLY HISTORY OF ARABIC GRAMMAR 209

marians (JAOS 1981) deserves special recognition. In a very detailed study Levin
attempts to prove once and for all that Praetorius's observation (GGA 1895) about
the use of m + m.i. in the Kitab was accurate. Levin systematically deals with three
main questions: (1) How did Slbawaihi define the pair m + m.i. ? (2) Which part was
termed by Slbawaihi m, and which m.i.? (3) What did Slbawaihi consider as the
literal meaning of m and m.i.?
Fundamental in Levin's study is his perception of the difference between
Slbawaihi and the grammarians from the Xth to the XVth century (see below) in the
various aspects of the use of the m + m.i. pair of syntactic terms. In what follows I
shall refer to Levin's perception and present his answers to the three questions
mentioned above:
(1) Levin maintains that, when Slbawaihi used the terms m + m.i. he meant to
emphasize that the parts so termed were both indispensable for the construction of
the sentence and neither of them could exist without the other. As for the later
grammarians see (2) below, and, more specifically (3).
(2) Levin corroborates Praetorius's assumption that Slbawaihi assigned the term m
to the opening member, the mubtadd of the mubtada' + mabniy'alayhi pair, and/z 7
(n. b.: not /5' J7) of the /f / + /« 7/ pair. The later grammarians .whose writings Levin
examines extensively, used a different criterion in their assignment of the terms m +
m.i. to the main parts of the sentence. Levin calls this "the criterion of function",
and explains that according to this criterion the two main pahs were identified as
subject and predicate. SIbawaihi's distinction, contrary to the later one, was based
on "a criterion of position", hence, he grouped together mubtada and/27 under the
term m, while mabniy {'ata l-mubtada') and fail constituted its counterpart, the
m.i., since they occupy the second place in their respective systems.

(3) Levin seems to suggest that the terminology used by the later grammarians shows
greater correlation between the literal meaning and the technical use of the m + m.i.
pair than SIbawaihi's. While both parties recognize the idea of "the leaning of one
thing upon the other" on the level of literal meaning, only the later grammarians
applied this concept in their use of this pair as technical terms. Hence, Levin rightly
observes that, in the writings of the later grammarians, the term isriad did not
denote simply "the relation between the subject and the predicate",5 but "the
assignment of a predicate to a subject". Slbawaihi, on the other hand, emphasized
(so says Levin) the mutual indispensability of m + m.i., i.e. the technical terms, and
thus maintained a clear discrepancy between the literal meaning and the technical
use of this pair, a fact to which we shall return later.
An important observation made by Praetorius and reaffirmed by Levin is that in
SIbawaihi's terminology, on its literal level, the "thing leaning upon another" was
the m.i. (i.e. musnad ila l-musnad), whereas musnad meant literally "the thing upon
which the (other) musnad leans".'

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 210 MUSNAD, MUSNADILAYH1 AND THE EARLY HISTORY OF ARABIC GRAMMAR

C. Major problems in the reconstruction of the history ofm + m.i.

Although Levin refrains from focusing on chronological aspects and concentrates
mainly on the difference between the mainstream of Arab grammarians and
SIbawaihi in the use of these terms, he nevertheless seems to hold a certain opinion
about the development of the "functional", subject-predicate, concept of m + m.i.
in the circles of Arab grammarians.
Levin observes that al-Mubarrad (floruit IXth century) was the only grammarian
who followed SIbawaihi in ascribing mutual indispensability to m + m.i. rather than
"the assignment of a predicate to a subject". Levin admits that nothing is known
about al-Mubarrad's view of "which part of the sentence is referred to by m and
which one by m.i.".1 Further, he abstains from drawing definite conclusions from the
fact that the first evidence in grammatical sources for the existence of the terms m +
m.i. and the like, with the meaning of "assigning a predicate to a subject", appeared
in the writings of Xth century grammarians.
Yet, it must have been Levin's emphasis on the notion of the mutual indispensa-
bility of the m + m.i. pair which made him believe that these terms meant for
al-Mubarrad exactly what they had meant for SIbawaihi. This seems to be the reason
Levin states that "in the later sources, beginning with the tenth century, we find a
new [emphasis added] concept of the terms m and m.i. fundamentally different from
that of Sibawaihi" .8
On this point Levin's view of the historical process diverges implicitly from the
opinion expressed by Praetorius in his above mentioned study of the terms in the
Kitab. In his review of Jahn's translation of the Kitab he puts forward the assump-
tion that the use of the term m. i. by al-Khalll and SIbawaihi was a deviation from
another, a genuine rendering of the Aristotelian id hupokeimenon (or its Syriac
translation). Praetorius was following Noldeke's suggestion about the origin of the
Arabic grammatical term m.i. He wrote "1st es richtig, dass al-musnad Hay hi
Wiedergabe von to hupokeimenon (subjectum, mellta mettasemanlta) ist
(Noldeke, Lit. Centralbl, 1890 Sp. 1217), so sind SIbawaihi und schon sein Lehrer
Halll ein wenig von dem deutlischern Wege gewichen, den die Spateren weder-
gewonnen haben . . .".9
According to Praetorius's somewhat laconic formulation (qualified by a con-
ditioned acceptance of Noldeke's suggestion about the origin of m.i.), the Arabic
term successively exhibited the following stages:
1. Through an unspecified channel of transmission, Vlllth century Arab gram-
marians received (or created) an Arabic translation of the philosophical notion of
"subject", either in its original (i.e. philosophical or "functional") form or in a
modified manner (i.e. the "positional", non-"functional") as preserved in the
Kitab.
2. Later grammarians did not follow the great masters from Basra. They adopted
the original meaning of the philosophical term in an accurate Arabic rendering,
viz. al-musnad ilayhi, "a thing upon which something leans", being "the subject

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 MUSNAD, MUSNADILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR 211

of a predicate". We are not told how "die Spateren" returned to this original
meaning (see below).
3. Along with the application of the original Aristotelian concept of "subject" in
this grammatical term of the S-N-D root, the later grammarians also changed it
from (Sibawaihi's) musnad to musnad ilayhi, both having the meaning of
"support, a thing upon which something leans".
Now, whereas Praetorius is ready to consider expressly the presence of Greek
influence from the very early stages of Arabic grammar (Ilnd/VIIIth century), Levin
remains strictly in the sphere of this scholarly field and refrains from touching upon
the question of foreign influence. However, by attributing the adjective "new" to
the concept of the function of the terms m + m. i. in the use of grammarians from the
IVth/Xth century on, Levin implicitly takes sides with a well defined group of
scholars who believe that the direct impact of Greek philosophy started only in that
century and accordingly, that the Kifdb of Sibawaihi was impervious to the intro-
duction of any significant elements of this tradition.10 It could be argued now that
the fundamental conflict between Sibawaihi's concept of "position" of the m + m.i.
pair and its later "functional" use corroborates significantly the overall rejection by
the aforementioned group of any Greek influence on early Arabic grammar.
Noldeke's assumption that m.i. ("the thing upon which sth. leans") translates the
Greek hupokeimenon is, of course, far from being certain. Actually, as far as I
know, no Greek term has been found which can claim etymological paternity to m.i.
in either the Organon or its interpretations (this point will be discussed at length in
section F, below).
On the other hand, and on further thought, some of Levin's arguments are not
sufficiently substantiated to support the historical assumption suggested by the
adjective "new" discussed above.
A preliminary counter-argument should emphasize the difficulty created by
Levin's assumption that the correlation between the literal meaning and the
technical use of the m + m.i. pair was preceded by a period in which Arabic
grammatical terminology did not maintain such a correlation at all.
Then, it could be suggested that in so-called "nominal" sentences (jutnal ismiyya)
the literal meaning "a thing upon which sth. leans" and its counterpart ("the thing
which leans upon a (preceding) thing") perfectly suit the pair mubtada and mabniy
('ata l-mubtada) and thus may have had this meaning in Sibawaihi's technical jargon
too.
Levin's (following on Praetorius's) suggestion that, in so-called "verbal"
sentences (J.fi'liyya),Sibawaihi considered the verb as m, in the sense of "the thing
upon which sth. leans" is not proved; it is only supported by implicit pieces of
evidence, as Levin himself admits, which are open to opposite interpretations. Note
that this suggestion played an important role in the formation of Levin's thesis of
Sibawaihi's concept of position allegedly expressed by the m + m.i. pair.
Levin's actual rejection of any correlation between the literal meaning of the m +

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 212 MUSNAD, MUSNADILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR

m.i. pair and their technical (grammatical) use in Sibawaihi's system made him
translate (p. 153) the technical m as "the first indispensable part" and m.i. as "the
second indispensable part". An extreme expression of this rendering is Levin's
translation of the passage in the Kifab ii 61.8: lam yusnad ila musnad "it (had) not
become the second indispensable part of the sentence(!)".
But more important than this criticism of Levin's conception of the use of m + m. i.
in Sibawaihi's Kifab are the data which will be presented below from sources from
the end of the Ilnd/VIIIth century, and which constitute ample evidence for the
existence of a subject-predicate concept for the main parts of the sentence already in
that early period of grammatical thinking. Most important for our discussion is the
existence of evidence on the use of words from the root S-N-D to denote the
subject-predicate relationship, more precisely, to denote the "leaning" of the
predicate upon the subject.

D. The presentation of the new data

I. A note on the existence of subject-predicate terminology in the Kifab

In his study Levin deals also with terms derived from the roots KH—B-R and
H-D-TH, which correspond in their meaning with the "functional" notion of m +
m.i. (and isnad). He mentions the terms al-rnuhaddath_ 'anhu and al-mukhbar 'anhu
as identical with m.i. and al-hadith (anhu) and al-khabar as equivalent to m.n The
detailed references to grammatical treatises which used these corresponding pairs of
terms do not include loci in the Kitab. One might conclude that their absence from
the list of references results from their non-existence in the sense of "subject" and
"predicate" in Sibawaihi's book, similar to the situation regarding m + m.i.u
However, SIbawaihi used the phrase al-muhaddath 'anhu while referring to both
the actor (fa'it) of the verb and the mubtada' P Although this phrase was not used as
a term, its existence in the Kitab indicates that the concept of "subject" and
"predicate" was not alien to Sibawaihi's syntactic system.
Let us turn now to examine the occurrences of the root S-N-D in two early works
written at the end of the second century of Islam: the grammatically oriented Qur I n
commentary Ma'am 'l-Qur'an of Abu Zakariya al-Farra',14 and the philosophical
treatise of Ibn al-Muqaffa" ,15 From the analysis of sixteen occurrences of this root in
these two works, we can conclude that it was used in a semi-technical manner as a
denotation of "underlying/dependent" relations, both in logic and grammar,
sometimes specifically for the indication of the subject-predicate nexus. In what
follows I shall examine these occurrences in detail.

II. Evidence from grammar: al-Farra', Ma'am'l-Qur'an

In Ma'ani 'l-Qur'an derivatives of the root occur nine times in eight loci (179 bis;

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 MUSNAD, MUSNADILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR 213

83; 119.11 203; 207; 320; 328; 385). In six of them, al-Farra' refers to annexion
constructions which can be classified into two sub-groups:

(a) The first member of the annexion is a temporal entity functioning either as an
adverbial or as a locative predicate:
1119.6: ashhur in the sentence al-hajju ashhur-un ma'lumat "The Hajj (takes place)
in certain months" is in the raf case, but when it bears the meaning of a locative
(temporal) predicate it takes the nasb case. Al-Farra' notes that with this function
"it is more powerful" to use the noun as "leaning upon something", as in the
sentence with the annexion construction (al-hajj-u) ashhur-a s-sayfi "(the
pilgrimage) is in the summer months": . . . al-ism idha kanafima'ria sifa 'aw mahall
qawiya idha usnida ita shay' "when the noun has the meaning of a temporal or
locative entity it is stronger when it leans upon something".
Similarly, see II203, for hin-a "in the time that . . .", which is hin ma'turn musnad
ita lladhi ba'dahu " . . . a definite time marker leaning upon the following
member"; and II 320 with 'awwal-a "first of . . ." which is defined as ghaya "time
border".

(b) In the second group, I could not identify any common feature shared by the
three constructions under discussion: 183,14: the annexion of 'irrian + 2nd per. pi.
m. pronoun (— kurri) in Qur. 2,143 is described as 'asnada (or: 'usnida) 'l-irrian ita l-
ahya mina'l-mu minina "he leaned (the word) 'irrian upon (the pronoun referring
to) the living among the Believers"; see also II 328, 16, where the annexion sawtu
'l-hamir "the voice (in sing.!) of the asses" presents sawt as "leaning upon" the word
hamir.

II385 bottom presents an interesting construction in which the annexion comprises

a present participle followed by a 1st per. sing, pronoun. This pronoun is an
underlying direct object to the participle (mu'tuya. "my donors"). Actually,
al-Farra' emphasizes that the native speakers prefer such an annexion to a
construction in which the present participle exercises a verbal power by governing its
accusatival cornplement, as in mutli'uNi. Here al-Farra' describes the annexion
construction as asnadu fa'il-an . . . ita 'sm-in makni 'anhu "they (viz. the Arabs)
leaned a participle . . . upon a pronoun". It seems safe to assume that in these cases
the verb asnada is an equivalent of adafa as a denoter of the relationship of the two
members constituting the annexion, and it reflects clearly (as adafa does) the
conception of this construction as a specific link between "a base" and "an added"
member, the latter being placed first.
The more important occurrences of S-N-D in Ma'anfl-Qur'an are, of course, the
two loci in which nexal relations between a verb and its agent are discussed.
In 179, 13, al-Farra' analyses the structure with tamyiz (in his terminology: tafsir
"explanation") following a verb, and maintains that it is a transformation of another
verbal structure:

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 214 MUSNAD, MUSNADILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR

daqa dhar'i bihi —» diqtu bihi dhar-an "I tightened my grasp over him, an arm
(grasp)".
This transformation is explained in terms of a shift in the nexal relations between
the verb and its subject (here, from dhar to the 1st pers. sing, enclitic pronoun).
The new nexal relationship is described as: ja'alta 'd-diq musnad-an ilayka "you
made the idea of 'grasping' lean upon you (viz., upon the pronoun - tu) and later
usnida l-fi'l ita' r-rajul "the verb was made to lean upon the person".
The second locus is II207,11, where al-Farra' examines Qur. 21,63 and supports
the reading (bat) fa'alahu kabiruhum "(But this), their chief has done it" as an
alternative to the reading (baT) fa'allahu kabiruhum "(But this), it may be (viz.
la'alla + enc. pron.) their chief (sc. who has done it)." Al-Farra' expresses his
support for the reading with a verbal sentence by saying fa-ja'ala fi'l al-kabir
musnadan ilayhi "and he made the verb of (the agent) kabir lean upon it (viz. upon
the agent)".
To sum up, the root S-N—D was used by al-Farra' in reference to two well-defined
bi-partite syntactic relations: the constituents of an annexion construction, and the
verb with its agent. Literally, it denotes the "leaning" of the first member of each
pair "upon" its counterpart, thus conceiving of the latter as "prior to, more basic
than" the member leaning upon it. Most significant for our issue is the fact that, in
the "verbal" sentences analysed by al-Farra', it is the verb that "leans upon" its
counterpart, not the agent! Finally, it should be noted that the use of derivatives of
S-N—D in Ma'ani l-Qur'an is synonymous with that of derivatives of D-Y—F. This
latter root constitutes the more normative terminology for the presentation of
annexion constructions, although I have not made a statistical count of their
occurrence. One may be misled by the co-appearance of adafa and asnada in II 385
bottom, where the preference of annexion constructions over structures with present
participle + pronoun in the accusative (see above ad. loc.) is explained as:
li-anna'I-Arab Id takhta.ru ilia16 7- idafa idha asnadu fa'ilan . . . "because the
Arabs do not prefer (to use anything) but an annexion construction when they make
a present participle lean upon . . .". It looks obvious that the D-Y—F derivative is
the established term. Yet, someone might conclude that the terminological use of a
S-N—D derivative is shared also by the syntactic relations between verbal nouns and
their accusatival complements. I believe that this is not the case, and that asnada etc.
was confined to the two sets of syntactic relations. We shall bear in mind our duty to
look for an explanation of why these two sets are bound together by S-N—D
derivatives. This will be done after the consideration of further data. In the
meantime, it is important to note the significant fact that D-Y—F is used by al-Farra'
in reference to nexal relations in the "nominal sentence", in which the predicate is a
verbal noun. The metaphorical expressions layluka rid'im and naharuka sa'im "your
night is asleep" and "your day-time is fasting"17 are analysed (syntactically) in terms
of "the acceptability of attribution of the verb (sc. verbal noun?) to the nouns, as if
these nouns (viz. layl, nahar) were the agents" (wa-qad yajuzu an nudifa'l-fi'I

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 MUSNAD, MUSNAD ILAYHl AND THE EARLY HISTORY OF ARABIC GRAMMAR 215

ita l-layl wa-yakuria kal-fa ilaini).

III. Evidence from philosophy: Ibn al-Muqaffa', K. al-Mantiq
Ibn al-Maqaffa"s treatise on logic must have been written during the second half
of the Ilnd/VIIIth century (see note 14 above). As a translation (often rather
loose)18 of Categoriae, De Interpretation and Analytica Priora it uses S-N—D
derivatives six times, four of them as finite verbs (asnadahu, yusnadu(2),
yastanidu), all in his Qatughuryus, and twice as a participle, the loosely idiomatic
musnad ilayhi "leaning upon (it)" in De Int. (Arabic Faryar Manis, a rendering of
the Greek peri hermeneias). Because of their special relevance to our subject these
two occurrences will be discussed at some length in what follows:
In the definition of "verb" 19 (chap. 3 of the Aristotelian original, according to the
modern division), Ibn al-Muqaffa' paraphrases the Greek (16b 7) . . . kai estin aei
ton kath heterou legomenon semeion, ". . .it indicates always that something is said
or asserted of something"20 by wa-la yakuna ilia mahmulan 'ata ghayrihi musnadan
ilayhi2* "and it occurs only as carried on something else, leaning upon it." Then, in a
concluding key-sentence referring to the function of "verb" as contrasted with that
of its subject, the Greek (16b 10) (kai aei ton kath heterou legomenon semeion estin,)
hoion ton kath hupokeimenou e en hupokeimeno "(Then, a verb was an indication
of something asserted of something;) I mean of a something predicated of a subject
or found present in it" is paraphrased in Ibn al-Muqaffa''s translation by: wa- 'anna
'l-harf la yakunu ilia musnadan ilayhi (wajiban . . . 'aw 'aradan . . .)22 "and (viz.
besides its consignificatio (of the reference to time)) that the verb occurs only as
leaning upon it (i.e. upon the ism) (permanently . . . or temporarily)".

It is clear from these two occurrences of musnad ita that Ibn al-Muqaffa'
considered the root S-N—D as an appropriate technical expression for the
designation of the logical-syntactical relations of the predicate with its subject. Its
double occurrence in the same passage supports the assumption that for our
translator, whatever his Vorlage may have been, the use of musnad was not a chance
selection of a common word.
Significant as this passage is for the history of syntactic terminology in Arabic
grammar, it seems advisable to note here that, in spite of this discovery, we cannot
conclude that for Sibawaihi the verb (fi'l) was not the part of the sentence upon
which the agent (fa it) "leans". What we can do (and should do) is claim that the use
of S-N-D derivatives for the indication of the relations of "predicate" with
"subject" was well-known in logical and grammatical circles of Muslim scholars
already in Sibawaihi's own time, as we learn from the data extracted from the
treatises of Ibn al-Muqaffa' and al-Farra'. Grammarians of the Xth century must
have continued a tradition already known in earlier (the earliest?) stages of Arabic
grammatical thinking, in which "subject" and "predicate", a central pair of
logico-grammatical terms, had been introduced inter alia by the use of S-N-D
derivatives.

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 216 MUSNAD, MUSNAD ILAYH1 AND THE EARLY HISTORY OF ARABIC GRAMMAR

IV Further reflections on the double designation of S-N-D derivatives for

predication and annexion relations
The fact that the use of musnad etc. in late VIIIth century writings can now be
connected with a subject-predicate relationship is important in the attempt to
establish a rational history for the relations between the grammatical use and the
literal meaning. The logical tradition in Arabic, witnessed already in Ibn
al-Muqaffa''s work, usually conceives of "subject" as being a base upon which the
"predicate" is carried (mahmul 'ala).2i I have expressed the opinion (Sec. C, above)
that in Sibawaihi's Kitab this literal meaning was also part of the technical
conception of the relevant syntactic terminology, as is clearly demonstrable in the
case of ii 61,8: lam yusnad ila m.
But how did the same terminology come to be used for two different syntactic
constructions in al-Farra''s technical parlance, when this grammarian used S-N—D
and D-Y—F derivatives for the description of both predication and annexion?
Far from being able to give a definite answer to this question, I believe that the
four occurrences of the verbal forms asnada etc. and istanada in Ibn al-Muqaffa''s
translation of Categoriae are meaningful for the question. After specifying the ten
categories, within the fourfold classification of "things there are" (i.e. 'amm
"general", khass "accident", 'ayn "substantive" and 'arad "individual") our
translator discusses the following apparent contradiction: according to the assertion
that the 'arad alone is predicated of the 'ayn, how should we interpret the fact that
some substances which are parts of more general substances can be predicated of
them? Does this mean that the same object can be both a substance and an accident,
which leads to an inevitable self-contradiction?24. Ibn al-Muqaffa' describes the
feature (or rather potential) of being predicated of something in terms of asnadahu
ita ghayrihi "he (sc. any speaker of the language) leaned it (viz. the predicated
entity) upon something else" etc. He uses this term for predication of both types: the
normal one in which an accident is predicated of a substance, and the one under }
discussion, of substances which are parts of other, more general, substances. Then, 3
when he presents Aristotle's explanation, according to which "the part of a I
substance is a substance, not because it is part of a substance (but because it is a j
substance of a certain "thing")", he reaffirms that, from the point of view of its |
relationship with the more general substance, this part belongs to the category (no. |
4) termed pros ti, which Ibn al-Muqaffa' terms mudaf.25 1
To sum up, the data collected from Ibn al-Muqaffa''s translation of Categoriae j
confirm our conclusion that the use of the phrase musnad ila in the discussion of j
relations between the predicate and its subject in Faryar Manls reflects the j
contemporary usage of an established terminology. Further, we realize that S-N-D
and D—Y—F derivatives were essential for the discussion of logical relationships in
two pairs of entities which share some basic common features.
It might be that this close affinity, demonstrated in the old VHIth century
translation, has to do with the somewhat accidential use of adafa etc. by al-Farra'

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 ML«M4D,Mtm4D/£/lY///ANDTHEEARLYHISTORYOFARABICGRAMMAR 217
26
and later grammarians in the description of predication; on the other hand, it
might also explain al-Farra''s more occasional choice of asnada etc. in his discussion
of annexion relationships. However, it is by no means my intention, with the
presentation of the data from Ibn al-Muqaffa''s treatise and the formulation of the
subsequent suggestion concerning al-Farra''s terminology, to cover up the diffi-
culties in tracing the developments in the use of S-N-D and D-Y—F in early Arab
grammatical circles.
E. Remarks on the use of m/m.i by SIbawaihi in the light of the new data

A strong argument in Levin's thesis about the essence of m/m.i. according to

SIbawaihi and al-Mubarrad, in contrast to later grammarians, is the fact that in his
Kitab SIbawaihi used this pair whenever attention to the interdependence of the
main parts of the sentence was called for.27 In the later sources we find a different
attitude to this pair of terms. Leaving aside the other aspects of alleged difference, it
seems that Levin would argue that these later sources evince a shift from the notion
of interdependence to that of dependence ("leaning upon") of the one part
(according to them, the predicate) on the other. This argument fits well into the
other elements of Levin's thesis about the "two essential views of m and m.i." in
medieval Arabic grammar.
Now, in the previous sections I have expressed an opinion about two other
elements of Levin's thesis, namely, that we are unable to decide what part of the
fi'l-fa'il pair in the Kitab is m (= "the thing upon which the other m leans") and that
the grammatical usage of m/m.i. was not wholly impervious to the literal meaning of
the terms. The last objection was even supported by the use of the phrase musnad ita
and similar derivatives in a semi-terminological manner by other Vlllth century
scholars. Hence, on a deeper analysis of the collected data it seems to be a too
far-reaching conclusion to speak of "two essential views" about m and m.i., even
about "the ancient view" vs. "the later sources". Yet, it is impossible to ignore the
fact that, in spite of expectations, it is not the literal meaning which is reflected in the
usage of the pair (in its three occurrences), but rather their indispensability and
interdependence, as is mentioned above.
This interesting fact undoubtedly deserves an explanation. Again, the scantiness
of the data hardly leaves room for any well-founded explanation. However, I
suggest that the relevance of the following fact to our subject be seriously
considered:
In three of the four occurrences of derivatives of the S-N—D root, the two terms
m/m.i. appear as a terminological pair (Chapters 3, 117, 132). In all these
occurrences they are presented as a superordinate pair, superimposed upon the two
existing regularly used pairs oimubtada + mabniy (ala l-mubtada) and fi'l + fail.
It is conceivable that the rare appearance of the mlm.i. pair in the Kitab, solely as
a superordinate pair of syntactic terms, has to do with the fact that the notion of
fe interdependence between the two main parts of the sentence, which is expressed

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 218 MUSNAD, MUSNADILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR

twice by this pair, plays a negligible part in the syntactic chapters of the Kiiab.28
F. A note on the growth of the mlm.i. terms and foreign influence
In an earlier part of this article (Section C, above), I expressed my opinion against
Noldeke's suggestion that the term rn.i. "a thing upon which sth. leans" originated
as the rendering (Uebersetzung) of hupokeimenon. Moreover, I stated that I do not
know of any similar Greek notion of which this Arabic term could have been
calqued: My search for a foreign origin to musnad/musnad ilayhi etc. focused so far
on the following sources:
a) The Aristotelian origin of the relevant chapters in Categoriae and De
Interpretatione, namely those parallel to the passages in Ibn al-Muqaffa''s
translation in which S-N—D derivatives appear, and others which deal with similar
ideas.29
b) Greek commentaries on the relevant chapters.30
c) Syriac translations of the relevant chapters and grammatical treatises .31
The fact that neither the Greek sources, nor the Syriac works show any trace of
this notion of "leaning" suggests that it is possible that the origins of this notion
should be looked for in the domain of Arabic scientific thought. Note that the term
mahmul too does not have a foreign cognate which can be identified as its direct
origin (note 23, above).
Now, although the derivatives of S-N-D are closely related with the concept of
"subject" and "predicate" in the works of al-Farra' and Ibn al-Muqaffa', it seems
unnecessary to draw the conclusion that Arabic grammar borrowed musnad etc.
from Arabic treatises on philosophy. Again, we should notice that the paucity of
textual evidence and even of reliable information about the history and mutual
relations of the extant material is a major obstacle to determining whether the
process was as suggested above or, on the contrary, that musnad etc. came into
being in grammatical circles and then passed into Ibn al-Muqaffa''s translation of
logic.32
One last point should be made about the terminological development of the root
S-N-D in the later stages of Arabic grammar. Despite the heavy restrictions
imposed by the paucity of early documentation it seems right to say that the late
terminology of musnad etc. was more systematic than the early one. The three early
sources used S-N—D derivatives for the most part non-terminologically, e.g., as
finite verbs and participles with verbal meaning. Even in Slbawaihi's pair, the m.i.
includes an anaphoric pronoun, and therefore requires that its counterpart should
precede it. It seems that the shift identified by Praetorius and Levin from the
personal meaning of the pronoun in the early terminology of m.i. (viz. "leaning
upon it") to an impersonal pronoun in the use of the later grammarians (viz. "a thing
upon which sth. else leans") amounts to a historical process of systematization of
this syntactic pair of terms, probably by analogy with the closely related pair
mudaflmudaf ilayhi.

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 MUSNAD, MUSNADILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR 219

NOTES

* I wish to thank Professor Gideon Goldenberg of the Hebrew University of Jerusalem for his
wise guidance and invaluable comments on former drafts of this article.
1
Weil (1915), 385, later surveyed by Bravmann, 46. Bravmann notes that until his days it
was "an almost general opinion" among scholars that the Arab grammarians were mainly
interested in "the external marks of linguistic phenomena". In his review of Weil's article,
Weiss (1917), p. 130 points out that the absence of a sentence concept (Satzbildungslehre)
within the syntactic theory was also characteristic of Greek grammar. He quotes from
Steinthal's Geschichte der Sprachwissenschaft bei den Griechen und Romern about
Apollonios Diaskolos: "Apollonios fragt nicht wie wird der Satze gebaut und welches sind die
Elemente des Satzes? Sondern nur: wie verbinden sich die Worter in Satze? Daher fehlt ihm
jede Kategorie fur Satzverhaltnisse; er weiss nichts von Subjekt und Objekt, Pradikat und
Attribut. Statt dieser erscheinen nur Nominativ und Akkusativ, Verbum, Transition d.h.
Wortverhaltnisse''.
2
See the citations from the orientalist literature quoted by Ayoub-Bohas pp. 36-38 (also
Fleisch (1957), p. 153 f; idem (1961) p. 24 f.), in which this "peculiarity" of the Arab gram-
matical model is mainly criticized.
3
Mehiri (al-Mahiri) (1966) p. 39; idem (1973) p. 362. Versteegh (1977) p. 71 is rather vague.
Rundgren (1976), p. 137 seems to believe that Sibawaihi considered only mubtadd + khabar as
the two parts of isnad (subject + predicate). According to Rundgren it was al-Zamakhshari
who attributed the same isnad relations to ft'I + fail as well. See also Frank (1981), p. 271.
4
The term isnad does not appear in the Kitab, a fact that Levin knows well.
5
For details, see Levin p. 157, n. 113.
6
For Praetorius's conclusion and Levin's further considerations in favour of it, see Levin, p.
150f. A. Levin kindly showed me the material now available in the newly published Kitab
al-'Ayn, in which the material quoted by the Lisan al-'Arab is fully documented (vol. VIII, pp.
228-29).
7
Levin, p. 162.
9
Op. cit. p. 151.
9
Praetorius (GGA 1895) p. 711. Noldeke's words (loc. cit) are also worthy of citing:
"(Dagegen) gehort hierher der allgemeine Name des Subject al-musnad ilayhi, eine sehr gute
Uebersetzung von hupokeimenon, Dariiber, dass die arabische Grammatik diesen grundbegriff
wenig hervorhebt, da sie dem Subjecte des Nominal und des Verbalsatzes besondere Namen
giebt, hat M. [=A. Merx, on whose Historia artis grammaticae apud Syros Noldeke wrote this
review- R.T.] (S.147) ganz iibersehen, dass sie ihn doch wirklich kennt."
10
The up-to-date history of the rejection of Greek influence on the early growth of Arabic
grammar, including the period of the Kitab, is documented in a few recent publications, e.g.
Sezgin (GAS IX) p. 3f; Talmon (al-Karmil 1984) pp. 45-53. It is difficult to refer to any
particular study in which it is asserted straightforwardly that only in the Xth century was
Greek logic first introduced in Arabic grammatical writings! Gaetje (1971) explicitly rejects
Aristotelian influence on the tripartite division of speech units in the Kitab (pp .4—5), but it is less
certain (pp. 8-9) whether he considers al-Farabl's teaching as the source of philosophical
influence on later grammarians. Versteegh (1977) holds that the translated Aristotelian writings
started to influence grammarians from the Xth century onwards, but he still considers the vue
diffuse of introducing Greek influence into Arabic grammar as its ancient antecedent. Even
Troupeau (1983, p. 143) formulates his general view about the philosophical orientation of Ibn
al-Sarraj's Kitab al-Usul as carefully as "a marque un tourant dans la production grammaticale
arabedelV/Xsiecle."
11
Levin (1981) Section VI, p. 161f.
12
Note that the sources quoted by Levin are all from the Xth century on! However, in his
article in ZAL (1985) Levin (p. 118 n. 2) mentioned that the terms al-muhaddath 'anhu (viz. the
subject) and al-hadlth or hadlth 'anhu (viz. the predicate) were already used in all types of

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 220 MUSNAD, MUSNADILA YHI AND THE EARLY HISTORY OF ARABIC GRAMMAR

sentences, since the Illrd/IXth century. Levin is most probably referring to the existence of this
pair of terms in Mubarrad's Muqtadab although he does not specify any IXth century source.
13
Cf. ch. 9, i 10, 18 (ed. Derenbourg) for the definition of the verb and its actor:
fa-'I-asma l-muhaddath 'anha wa-l-amthila dalila 'ata ma mada wa-nid lam yamdi mina
'l-muhaddath bihi 'an al-'asma'. Further, in i 145. 17 the structure a-Tamimiy-Un (variant of
-An) marrat-an wa-Qaysiy-Un (var. of -An) 'ukhra, Sibawaihi justifies this variant by saying:
wal-raf jayyid li-'annahu 'l-muhaddath 'anhu wa l-mustafham. The noun Tamlmly, being a
predicate, is equivalent to the concealed 2nd per. sing, anta, the subject in the underlying
structure (viz. 'a-anta Tamimly-Un). Less certain is the idiomatic status of al-muhaddath 'anhu
in i 725,18.
14
Although this book is a Quran commentary, its author emphasized mainly grammatical
interpretations. Frequently, his lengthy treatment of linguistic structures offers modern scholars
access to the theory and practice of the Kufan grammarians. Previous analyses of the gram-
matical material in this work are mentioned in Diem's bibliography (1981).
15
On Ibn al-Muqaffa''s life and works, see EH (byF. Gabrieli). On the identity of its writer,
probably 'Abdallah b. al-Muqaffa''s son, see Kraus (1934). Zimmermann (c. 1972) studied this
work, using the Beirut MS., in studying certain aspects of the history of Islamic philosophy. Ibn
al-Muqaffa''s work was published in 1978 by Daneshpazhuh. G. Troupeau (1981) is the only
historian of Arabic grammar who has so far made use of this treatise in the study of the history of
this scholarly field. From his comparison of the terminology in the Kitab with Ibn al-Muqaffa 's
translation of the first chapters of De Interpretatione he concludes that there was no connection
between these two^works whatsoever.
16
In the text 'ata. I am indebted to M. Carter for suggesting this correction.
17
This grammatical topic is somewhat close to the ittisa fl 'l-katam etc. discussed by
Sibawaihi in the Kitab chapter 42 and to other items collected by De Goeje in W. Wright, A
Grammar of the A rabic Language, 3rd ed. Cambridge, 1967(repr.).ii 26 Rem.b; also, cf. Jahn's
note 12 to chapter 42 of the Kitab, where the translator quotes from al-Sirafi's commentary
another example of ittisa': al-nahar mubsir.
18
See on this point Zimmermann, n. 37 (p. 542); G. Furlani (1926), p. 210, specifically on the
translation of Categoriae. Also, Kraus, p. 5.
" Harf in Ibn al-Muqaffa's terminology, which is also used by al-Kindi. See Zimmermann
pp.530-31.
20
The text and translation are quoted from Cook's edition, pp. 118-19.
21
See K. al-Mantiq. p. 28,1.12.
22
ibid. 1.15.
23
Note, however, the etymology suggested by Zimmermann, p. 534. Although he notes the
development of its use from logicians to translators, this scholar seems to have ruled out the
metaphor I suggest for the pair Mawdu-mahmul. Otherwise, I cannot understand why he
emphasizes (n. 51, in p. 544) that "[t]he term mahmul was taken literally by al-Kindi, who used
the active participle as a correlative term instead of mawdu'.".
24
The original text in Categoriae seems to be 3a29. In Ibn al-Muqaffa''s Qatughuryus it is
found on p. 11,13 ff.
25
See Qatughuryus, p. 11,19 and passim; the category itself is first presented and defined on
p. 10. See, however, end of note 32, below.
26
See Jahn n 1 to chap. 3, Goldenberg (1985).
27
Levin uses the term "indispensable" in his description of the role played by the two terms in
the sentence. However, this term combines in Levin's usage two different meanings: inter-
dependence between the two constituents of the pair, and their indispensability in creating
self-sufficient stretches of speech (kalam, wrongly translated by many scholars, including Levin,
"sentence". See Talmon, ZDMG [forthcoming]).
28
See the Kitab i 110,6; 343,19 for the two other occurrences of the notion.
29
Esp. Categoriae, chapters l-5;7;De!nt., chapters 1 ^ .
30
Commentaries by Philoponos on Categoriae (GAG XIII ad la20-; 3a29);Stephanus and
Ammonius on De Int. (English translation and commentary by Arens).
31
Logic: Paulus Persa (ed. Land), esp. p. 10,14 ff (Lat. trans, p. 11,13 ff) i p. 15,21 ff (Lat.
trans, p. 17, 4 ff). Translations of Categoriae: George, Bishop of the Arabs (ed. Gottheil);

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 MUSNAD, MUSNADILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR 221

Sergius of ReS 'Ayna in the Italian translation of G. Furlani (1922). I am grateful to Prof. Lidia
Bettini for sending me a xerox of this article; cf. De Int.: Probha (ed. Hoffmann). In grammar:
Jacob of Edessa, To fez Memilla (ed. Wright) esp. pp. 79-80 and the Syriac translation of
Dionysius Trax (in A. Merx, Historia, p. 49 ff).
32
Zimmermann, n. 48 (p. 544) considers the origin of the mlm.i. pair to be in grammar, and
the use of the form musnad in Ibn al-Muqaffa''s Faryar Manls as an indication that grammatical
terminology was known to educated persons outside grammatical circles at quite an early stage.
On the other hand it might be assumed that the use of S-N-D derivatives in logic preceded
their employment in grammar on the ground of their interchangeability with derivatives of
D-Y—F. Not a few studies have referred to the problematic status of the notions "noun" and
"verb", "subject" and "predicate" in Aristotle's system; cf., for instance, Ackrill's note on De
Int. 16 b 6: . . . "It may disturb us to find Aristotle saying that a statement consists of a name and
a verb, because this terminology suggests a confusion of logical with grammatical analysis. But
this is not a confusion imported by the translation; logic and grammar are, in fact, not clearly
distinguished in Aristotle's discussion". Also, cf. Steinthal I pp. 247-48 (also p. 239, for the
definition of rhema) and Arens (1969), p. 15. Thus, it can be maintained with reason that since
logic does not draw a clear distinction between such linguistic relations as subject and predicate,
an annexed language-unit and one to which another is annexed (viz. mudaf ilayhi) or an
attribute and its head, this interchangeability can be expected to have affected grammatical
circles which came under the influence of logical terminology and even translators of logical
works who, in their turn, were somewhat familiar with the work and technical jargon of these
grammarians.
BIBLIOGRAPHY
Ackrill, J. L. Aristotle's Categories and De Interpretatione (translated and noted). Oxford, 1963.
Arens, H. Sprachwissenschaft: Der Gang ihrer Entwicklung von der Antike bis zur Gegenwart.
Freiburg/Munchen, 1969.
Arens, H. Aristotle's Theory of Language and its Tradition. Studies in the History of Linguistics.
Vol. 29. Amsterdam and Philadelphia, 1984.
Ayoub, G. and Bohas, G. "Les grammairiens arabes, la phrase nominale et le bon sens" in: The
History of Linguistics in the Near East (ed. C. H. M. Versteegh et al.), Amsterdam and
Philadelphia, 1983. pp. 31-48.
Bravmann, M. M. "On the achievements of medieval Arab linguists in the light of modern
research" (in Hebrew). Tarbiz 17 (1945-46), pp. 43-64.
Carter, M. G. "Les origines de la grammaire arabe", REI40 (1972), pp. 69-97.
Carter, M. G. "Sarf et Hilaf: contribution a l'histoire de la grammaire arabe", Arabica, 20
(1973) pp. 292-304.
Cook, H. P. (ed. and tr.). Aristotle, The Categories; On Interpretation. Loeb Classical Library,
London and Cambridge, Massachusetts, 1962.
Diem, W. "Bibliographie sekundarliteratur zur einheimischen arabischen Grammatikschrei-
bung". Historiographia Linguistica VIII:2/3 (1981), pp. 431-86.
El (2nd ed.): Encyclopaedia of Islam. New edition. Leiden, 1960.
Al-Farra', Abu Zakariya, Kitab Ma anil-Quran. Cairo, 1955.
Fleisch, H. "Esquisse d'un historique de la grammaire arabe." Arabica A (1957); pp. 1-22.
Fleisch, H. Traite de philologie arabe. I. Preliminaires, phonetique, morphologie nominale.
Beyrouth, 1961.
Fliigel, G. Die grammatischen Schulen der Araber. Erste Abtheilung: Die Schulen von Basra
und Kufa und die gemischte Schule. Leipzig, 1862.
Frank, R. M. "Meanings are spoken in many ways: the earlier Arab grammarians." Le Museon
94 (1981), pp.259-319.
Furlani, G. "Sul trattato di Sergio di Resh'ayna circa le Categorie". Rivista trimestriale distudi
filosoficiereligiosi 3 (1922), pp. 135-72.
Furlani, G. "Di una presunta versione araba di alcuni scritti di Porfirio e di Aristotele". Rendiconti
dell'Accademia Nazionale deilincei, Classe di Sc.more., stor. e filol. 2 (1926), pp. 205-13.
Gaetje, H. "Die Gliederung der sprachlischen Zeichen nach al-Farabl" Der Islam, 47 (1971),
pp.1-24.

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626
 222 MUSNAD, MVSNAD ILAYHI AND THE EARLY HISTORY OF ARABIC GRAMMAR
Goldenberg, G. "Verbal category and the Hebrew verb" (in Hebrew). Language Studies. Ed.
M. Bar-Asher. Jerusalem, 1985. pp. 295-348. English Summary XL-XLII.
Gottheil, R. J. H. "The Syriac versions of the Categories of Aristotle". Hebraica 9 (1892-93),
pp.166-215.
Hoffmann, G. De Hermeneuticis apud Syros Aristoteleis. Dissertation Inauguralis. Berlin,
1868.
Ibn al-Muqaffa', al-Mantiq (Logic) and Ibn Birhiz, Hudud al-Mantiq (Definitions of Logic).
Ed. M. T. Daneshpazhuh. Tehran, 1978.
Jahn, G. Sibawaihi's Buck uber die Grammatik. 2 Bde., Berlin, 1895-1900 (Reprinted
, Hildesheim,1969).
Kraus, P. "Zu ibn al-Muqaffa"'. RSO14 (1933-34), pp. 1-20.
Levin, A. "The grammatical terms al-musnad, al-musnad ilayhi and al-isnad". JAOS 101
(1981), pp. 145-65.
Levin, A. "The distinction between nominal and verbal sentences according to the Arab
grammarians". ZAL 15 (1985), pp. 118-27.
Mahiri al-, A. "Al-jumla fi nazar al-nuhah al-'Arab". Hawliyyat al-Jami'a al-Tunisiyya 3
(1966), pp. 35^6.
Mahiri al-, 'A. (A. Mehiri), Les theories grammaticales d'lbnJinnf. Tunis, 1973.
Merx, E. Historia artis grammaticae apud Syros. Leipzig, 1889.
Al-Mubarrad, 'Abu l-'Abbas, Al-Muqtadab. Ed. M. A. 'Udayma, 1965-66.
Noldeke, Th. Review of G. Fliigel: Die grammatische Schulen derAraber, in: GGA, 1863, pp.
2034-40.
Noldeke, Th. Review of A. Merx: Historia artis grammaticae apud Syros in: Literarisch.es
Centralblatt35 (1890), pp. 1215-20.
Paulus Persa, 'Al Maktebanuta melila de-Aristotelis Filosofa. Anecdota Syriaca IV. ed. J. P. N.
Land, (Leiden, 1875), pp. (Syr.) 1-32.
Philoponi (olim Ammonii) in Aristotelis Categorias Commentarium. Commentaria in
Aristotelem Graeca XIII, i-2. Ed. G. Busse. Berlin, 1898. (= CAG).
Praetorius, F. Review of Jahn's translation and commentary on Sibawaihi's al-Kitab. GGA
(1894), pp. 705^5.
Rundgren, F. "Uber den griechischen Einfluss auf die arabisch Nationalgrammatik". Ada
Societatis Linguisticae Upsaliensis. N.S. 2:5 (1976), pp. 119-44.
Sezgin, F. Geschichte des arabischen Schrifttums, IX. Leiden: E. J. Brill, 1967,1984.
Sibawaihi, Le Livre de Sibawaihi; traite de grammaire arabe . . . par H. Derenbourg. 1.1981;
II. 1885. Paris (Reprints, Hildesheim 1970).
Steinthal, H. Geschichte der Sprachwissenschaft bei den Griechen und Romern. New York,
1971 (Berlin 1890?).
Talmon, R. "Grammatical thinking before Sibawaihi's Kitab: a study in the history of Arabic
grammatical terminology" (in Arabic). al-Karmil5 (1984), pp. 37-53.
Talmon, R."Al-kalam ma kana muktafiyan bi-nafsihl wa-huwa 1-gumla." A study in the history
of sentence-concept and the Sibawaihian legacy in Arabic grammar". ZDMG 137
(forthcoming).
Troupeau, G. "La logique d'lbn al-Muqaffa' et les origines de la grammaire arabe". Arabica 28
(1981), pp.242-50.
Troupeau, G. "Les 'paries orationis' dans le Kitab al-Usul d'lbn al-Sarraj". The History of
Linguistics in the Near East, ed. C. H. M. Versteegh et al. Amsterdam/Philadelphia, 1983.
Versteegh, C. H. M. Greek Elements in Arabic Linguistic Thinking. Leiden, 1977.
Weil, G. "Zum Verstandniss der Methode der moslemischen Grammatiker". Festschrift
Eduard Sachau. Berlin, 1915, pp. 380-92.
Weiss, J. Review of G. Weil (ed). Festschrift Eduard Sachau zum siebzigsten Geburtstage,
gewidmet von Freunden und Schulem, in derem Namen herausgegeben von Gotthold Weil.
Berlin: Verlage von GeorgReimer, 1915. In: Der Islam! (1917), pp. 123-136.
Zimmermann, F. W. "Some observations on al-Farabi and logical tradition". Islamic
Philosophy and Classical Tradition. Essays presented by his friends and pupils to Richard
Walzer on his seventieth birthday. Editors: S. M. Stern, Albert Hourani and Vivian
Brown (Oxford), Cassirer, (c. 1972).

Downloaded from https://www.cambridge.org/core. Bibiliothèque interuniversitaire de Santé, on 24 Dec 2019 at 00:32:59, subject to
the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0035869X00140626