THE PLURAL OF PAUCITY IN ARABIC AND ITS ACTUAL

THE PLURAL OF PAUCITY IN ARABIC AND ITS ACTUAL SCOPE
On two claims by Siibawayhi and al-Farraaÿ
(paper delivered at the 16th Symposium on Arabic Linguistics,
Cambridge, England, 1-2 March, 2002)
Ignacio Ferrando
University of Cadiz
Contents:
1. Introduction
2. PP in Arabic
3. Parallels in Semitic: Akkadian and Tigre
4. The terms for PP in Arabic
5. The Arabic patterns of PP according to Siibawayhi
6. Interest of Siibawayhi’s description of PP
7. Other grammarians’ views
8. Today’s scope
9. The opinion of Šawqii ïayf
10. My own views: figures of Pre-Islamic Poetry
11. Its connection to agreement. al-Farraa’s claim
12. References
1. INTRODUCTION
The Arabic language includes, besides singular, dual and plural (broken and regular) numbers,
something that could be labelled as the fourth number category: the so-called “plural of small
number or paucity”, henceforth PP (jamÁu l-qilla, latin pluralis paucitatis) which has been studied
in detail by Arab grammarians, particularly and most importantly, Siibawayhi (ob. 793), in his
celebrated work al-Kitaab. Numbers are then singular kalb “one dog”, dual kalbaani/kalbayni “two
dogs”, plural of abundance (henceforth PA) kilaab “dogs (more than ten)” and plural of paucity
Êaklub “a few dogs (from three to ten)”. This gives rise to a number of questions. Is it actually
another number category inherited from Proto-Semitic or Proto-Arabic, or simply a subcategory of
later emergence? What use and what distribution does it have in the variety of Arabic described by
Siibawayhi and also in present day Arabic? Does the notion of PP have any effect on the agreement
variation in Classical Arabic (full vs. deflected)? This paper aims to respond to this type of queries
through a critical examination of some Arabic sources, in adittion to several brief considerations of
a more general nature. We shall use early Arabic data to test two separata claims made about the
notion of PP and related matters. The first one, made by Siibawayhi, is that a small set of the broken
plural patterns represent small quantities. The other claim, made by al-Farraa’, is that the variation
between “natural” agreeement (plural attribute with plural subject) and “deflected” agreement
(feminine singular attribute with plural subject) was also based on a distinction of pacucity and
abundance respectively.
2. PP IN ARABIC
The notion of PP is regarded as common and familiar to Arab grammarians, v. gr., the plural of
objects or animates whose number ranges from three to ten. This is actually a natural notion (ten are
the fingers of the hands) handled by many languages. It can be understood without any serious
difficulty as the only and true plural for beings or items conceived as different; thus the narrow
syntactic relationship of iñaafa, where the number bears the case corresponding to its syntactic
function and the being/item is assigned the genitive plural case. This plural can easily be described
as an individualised plural. Starting from eleven, the syntactic relationship between the number and
the object is less narrow, being analysed by the Arabic grammatical tradition as tamyiiz
(specification), thus requiring the accusative singular case. This is the consequence of the
conception of that plural as a collection of objects conceived primarily as a group, that is to say, a
collectivised plural. Hence, PP may be defined as an “individual, specific” plural, whereas PA, on
the other hand, may be treated as a “collective, non-differential” plural1.
What seems to be particular of Arabic is the form chosen to convey the notion of PP. Whereas other
languages have developed a special set of suffixes (like Tigre or Akkadian, as we will see later) or
have recourse to lexicalised quantifiers (English a few, several, French quelques, and also Arabic
baÁñ, biñÁat, nafar, nayyif, dawd, rah’), Arabic makes use of a special form of the broken plural,
i.e., a stem-internal type of plural which is, according to Ratcliffe (1998:242) “simply a process of
plural formation by infixation inherited from Proto-Semitic and Proto-Afroasiatic and analogically
extended”2.
3. PARALLELS IN THE SEMITIC LANGUAGES?
Fleisch (1961:I, 485) indicates that Brockelmann (1982:I, 436-7) points out that in Tigre (a North
Ethiopian living language) the plural of small number is consistently used. The great German
semitist says, in fact, that there is an abundant use of small number plurals as opposed to big
number plurals in Tigre, the latter being obtained by means of the plural of the plural. But the
examples he offers are not confirmed by Palmer’s detailed study, where only one case of plural of
plural or double plural (the author uses the term “big plural”) is attested: nälät pl. nälat pl. pl.
nälatat “kind of deer”. An attentive reading of the work of Palmer (1962:esp. 60-2) reveals,
however, that an opposition does exist between PP and PA, but it is conducted by means of what
Palmer calls “plurals of the diminutive” which must bear a suffix whose gender is the same as the
gender of the singulative. Examples:
gämäl (m. s.)
kis (m. s.)
1
“a camel”
“a bag”
gämmelam (m. p.)
kisetam (m.s.)
“a few camels”
“a few bags”
Fischer (1980) is perhaps one of the most lucid papers dealing with this topic. He links the notion of paucity
to the quality of the plural, supporting his views on comparative evidence, mainly from Akkadian, where a
special form of plural is used to convey a particular plural, apparently irrespective of the items numbered.
2
See a review of this important book in Ferrando (1999)
The corresponding broken plurals are Êägmal, Êäkyas
The corresponding diminutives are gämmelay, kisay.
It is a matter of discussion whether these forms are actually derived from the diminutive, because
the cases of färäs (sg.), färresät/färresay (dim.), Êäfras (pl.) y Êäfresat/Êäfresam (dim. pl.), or ñƏna
(sg.), ñƏnetät/ñƏnetay (dim.), Êäñännit (pl) y Êäñannitam (dim. pl) seem rather to indicate that, at
least for some words, the so-called diminutive plural does not stem, as it is the case with the
examples included in the above figure, from the singular diminutive form. One could suggest
instead that we are faced here with a suffix attached to the broken plural form.
Other examples: dƏmmu “cat” has two diminutives: 1. dƏmmätit pl. dƏmmetat/dƏmmutat “a few
she-cats” y 2. dƏmmutay pl. dƏmmetam/dƏmmutam “a few cats”, and gäzirät (sg.) “an island”,
gäzayƏr “islands” (pl.), gäzirätit (dim.), gäzerat “a few islands” (dim. pl.).
Palmer also mentions a pejorative plural, which consists of adding a suffix whose gender differs
from that of the singulative. So, gäzeram (pl. m.) means “some poor islands”.
Unfortunately, Palmer does not include information on the use and distribution of paucity plurals, as
he concentrates mainly on the morphological profiles and the phonetic processes involved in the
plural formation.
It can be said that the semantic category of PP in Tigre is alive and productive, but the
morphological device is clearly different form that of Arabic. While Tigre forms PP by adding a
suffix mainly to the diminutive form, Arabic prefers either a special broken plural form or the
regular suffixed plural.
Akkadian: Fischer (1980:77), quoting the words of von Soden (1969: § 61 f and i) states that
Akkadian has two plural suffixes: -uu/-ii for a general collective-like plural and -aanu/aani for an
individualised plural, that is to say a PP, although he points out that this variation is lacking in
feminine nouns. Moreover, it is not thoroughly applied in all domains of language. However, the
very question of this -aan suffix in Akkadian is beyond discussion. Some scholars, like von Soden,
op. cit. or Corriente (1971:117), consider it a plural suffix, but others, like Ratcliffe (1998:139), and
Buccellati, in his last summary of Akkadian (1997:76-7) see -aan as a particularising suffix
attached to the singular, the real plural suffix being –uu/ii . Whatever it may be, there is a semantic
difference between šarruu/ii “kings” and šarr-aan-u/ii “particular kings”, which could be related to
the notion of PP, or perhaps to the opposition between an individualised vs. collectivised plural.
This notwithstanding, as Lipiński (1997:251) indicates, further research is needed concerning this
point.
4. THE TERMS FOR PP IN ARABIC
A complete and detailed survey of the variegated PP patterns as described and classified by
Siibawayhi may be found in my previous work Ferrando (2001). A close examination of the actual
scope of PP use in Siibawayhi’s variety of Arabic reveals the following facts:
Siibawayhi does not use of the term jamÁu l-qilla, which is the usual term Arab grammarians came
to employ some centuries after him for Arabic PP, and which is also the most popular term used
nowadays. Instead, Siibawayhi prefers the term adnà l-Áadad, that is to say, “the lower or small
number”, always in contrast to aktaru l-Áadad “the upper or big number”. The verb used to express
the exceeding of the small number is jaawaz “to surpass.”, i.e. idaa jaawazuu bi-hi adnà l-Áadad
(570) “when they surpass in it the small number”, or idaa lam tujaawiz Êadnà l-Áadad (570) “if you
do not surpass the small number”. It was az-Zamaxšarii (ob. 538 h.) who used for the first time the
term jamÁu l-qilla.3
5. PATTERNS FOR ARABIC PP ACCORDING TO SIIBAWAYHI
The following chart provides a summary of the variegated PP patterns collected by Siibawayhi,
including at least one sample of each. Bold letters indicate original (non-repeated) patterns. No data
from collectives, irregular stems, exceptions, frequent alternation cases, or modifications due to
phonetic constraints are included4.
Singular pattern
faÁl
faÁal, faÁil, fiÁal, faÁul,
fuÁul
fuÁal
fiÁil (only one case)
fiÁl
fuÁl
faÁlat, faÁalat
fuÁla
fiÁla
Plural pattern
Samples
kalb-Êaklub, farx-Êafrux
jabal-Êajbaal, katif-ÊaktaafÊ, Áinab- ÊaÁnaab,
Áajuz-ÊaÁjaaz, Áunuq-ÊaÁnaaq
fiÁlaan,ÊafÁaal
”urad-”irdaan, rubaÁ- ÊarbaaÁ
ÊafÁaal
Êibil-Êaabaal
ÊafÁaal, fiÁala,ÊafÁul
™iml-Êa™maal, qird-qirada, diÊb-ÊadÊub
ÊafÁaal
jund-Êajnaad
faÁalaat
jamra-jamaraat, rahaba- ra™abaat
ġurfa-ġurufaat, ġurafaat
fuÁulaat, fuÁalaat
fiÁilaat,
fiÁalaat, kisra-kis(i/a)raat, niÁma-ÊanÁum
fiÁlaat, ÊafÁul (rare)
fiÁaal, faÁaal, fuÁaal, ÊafÁila
™imaar-Êa™mira, zamaan-Êazmina, ġuraabÊaġriba, raġiif-Êarġifa, xaruuf-Êaxrifa
faÁiil, faÁuul
fuÁlà (non elativ)
fuÁlayaat
™ublà-™ublayaat
hamaama-™amaamaat,
fiÁaala,
faÁaala, fiÁaalaat,
faÁaalaat, risaala-risaalaat,
fuÁaala, faÁuula
fuÁaalaat, faÁuulaat
dubaaba-dubaabaat, haluuba-™aluubaat
ÊafÁul
ÊafÁaal
It becomes clear from the above picture that four patterns are definite broken PP patterns (ÊafÁaal,
ÊafÁul, fiÁala, ÊafÁila). Eleven patterns, on the other hand, include the regular plural suffix -aat.
Indeed, on p. 227 of his treatise al-Mufaëëalu fii Áilmi l-luġa we read the following words, included in a
chapter entitled jamÁu l-qillati wa-l-katra: “wa-yanqasimu Êilà jamÁi qillatin wa-jamÁi katra fa-jamÁu lqillati l-Áaìratu fa-maa duunahaa wa-Êamtilatuhu ÊafÁulun ÊafÁaalun ÊafÁilatun fiÁlatun ka-Êaflusin waÊatwaabin wa-Êajribatin wa-ġilma wa-minhu maa jumiÁa bi-l-waawi wa-n-nuuni wa-l-Êalifi wa-t-taaÊ wamaa Áadaa dalika jumuuÁu katra”: “it is divided (the plural) into plural of small number and plural of big
number. The small number plural is for ten and any number below ten. Its patterns are ÊafÁulun ÊafÁaalun
ÊafÁilatun fiÁlatun, like Êaflusin wa-Êatwaabin wa-Êajribatin wa-ġilmatin “money”, “dresses”, “sacks” and
“boys”. Sometimes the plural (that of the small number) is formed with the waaw and the nuun and with the
alif and the taaÊ. And the rest is plural of big number.” See also Fleisch (1961:485).
4
Note that Siibawayhi does not include fiÁla in the list of PP forms, as done by other later grammarians like
az-Zamaxšarii (see the preceding footnote). He speaks rather of fiÁala as one of the distinct PP forms.
3
Although we shall deal with the question of productivity and scope of PP later, a preliminary
observation should be made here: surely enough, PP forms show a restricted distribution and this is
unexpected if the category was productive: Siibawayhi’s broken plural patterns are restricted to
masculine nouns of the shape CvC(v)C and CvCvvC. Feminine nouncs of the shapes CvC(v)Cat
and CvCvvCat use the sound feminine plural as PP. But other singulars, including CaaCiC, do not
have way of making a PP5. Another interesting point to be made is that Siibawayhi acknowledges
the lack of correspondence between a particular form of PP and a regular use of it for a particular
singular pattern. Thus, he says that a given PP form is often used to convey PA meanings and
viceversa. As to the claim of Fischer (1980) based on a proposal by Ferguson (1959) and also
adopted by Ratcliffe (1998) that the plurals of paucity (or at least those with an initial alif) were
phonological variants (vgr. fuÁulun > *a-fuÁul > ÊafÁul, or fiÁaalun > *a-fiÁaal > ÊafÁaal) which
occurred after numbers which ended in taa` marbuu’a (3-10), our data did not provide any evidence
for or against this claim6.
6. INTEREST OF SIIBAWAYHI’S DESCRIPTION OF PP
It is noteworthy that Siibawayhi dedicates a considerable part of his long chapter on the broken
plural to the treatment of the PP. There is a constant concern to discriminate between the forms of
PP and of PA, although exceptions and crossing forms are also noted. One could say that on several
occasions the specific forms of PP, in spite of their scarce functional yield, have allowed
Siibawayhi to complete the paradigms and to assign a different PP form to each singular. However,
it is easy to be overcritical of Siibawayhi’s claims and this is not my intention. On the contrary, I
think considerable credit should be given to what seems, in general, a well-founded intuition. It is
important to bear in mind that Siibawayhi aimed to describe the kalaamu l-Áarab, the language of
the Bedouins or Arab tribes, which was in danger already at his time. He could capture a good deal
of the ancient language peculiarities, such as PP, and put them in order within his impressive
description of Arabic. A tendency to regularise and complete language areas showing empty “cells”
could have led Siibwayhi to extend the notion of PP and to depict it as a more widespread feature
than it was in actual fact. Another thing to be considered is that the systematic presentation of
Siibawayhi includes some shortcomings, because everything points to the fact that the
morphological expression of the traditional concept of PP was already in Old Arabic an archaism
prone to disappear.
To sum up, Siibawayhi stands as a defender of the PP category as a truly component of Arabic, and
we should have in mind that he had a profound knowledge of the Arabic language. Hence the
interest of studying his contentions on this topic.
7. OTHER ARAB GRAMMARIANS’ ATTITUDES BEFORE THE NOTION OF PP
5
It is interesting to note that these observations made by Siibawayhi about the restricted distribution of the PP
patterns, especially `/aCCuC/, /aCCaaC/ and /aCCiCat/ are confirmed by modern distributional studies, such
us those of Murtonen 1964, Levy 1971 and Ratcliffe 1998.
6
This proposal is supported by the middle and modern use of forms like xamas taláf “five thousands”
(Jerusalem dialect) where the –t suffix is incorported into the noun. See on this Levin (2003). Other
interesting forms, bearing a prosthetic alif (xamsat Øa™muur instead of the expected xamsat ™umuur) are
attested in Middle Arabic texts, see Hopkins (1984:3, 109). Although this claim accounts reasonably well for
the derivation of Øaf∂ul and Øaf∂aal respectively from fu∂ul and fi∂aal, other forms remain unexplained.
The overall impression we get from the available sources is that later grammar treatises add little to
the issue. Most grammarians summarise, repeat and take extracts from the comments by
Siibawayhi, often citing and commenting on the very same examples. However, practically no
grammarian gives detailed attention to the issue, at least not to the same extent as Siibawayhi did.
This may indicate that the awareness of PP as an independent feature was gradually weakening with
the passage of time. This notwithstanding, with by no means pretending to offer an exhaustive
study, let us look at some samples of these studies.
a) Al-Mubarrad, (ob. 998) al-Muqtañab, uses, like Siibawayhi, the term Êadnà l-Áadad (I, 131-2).
But he merely offers a sizeable summarised description of the topic, inspired and guided by
Siibawayhi. In addition, one may observe that al-Mubarrad is perhaps less prone than Siibawayhi to
make a clear distinction between PP and PA (for instance, he says nothing about PP for R2w/y
stems). In I, p. 201, he remarks the option of using PP stems for PA meanings.
b) az-Zamaxšarii (ob. 1144). As we have seen before, az-Zamaxšarii seems to be the first Arabic
grammarian to use the terms PP (jamÁu l-qilla) and PA (jamÁu l-katra). Moreover, according to the
study of Schub (1978), he is well aware of the mutual blurring of these two categories, in both
directions. However, he does not overgeneralize to the extent of claiming that there is a complete
blurring producing the fall of the PP and PA categories.
c) Ibn al-ÊAnbaarii (ob. 1183), ÊAsraar: in p. 37 a curious observation is presented: the singular
pattern faÁl, as it is widely used, has a plural pattern ÊafÁul, which is “lighter” (al-Êaxaff), whereas
other patterns like fiÁl, faÁal, faÁul, faÁil, fuÁl, fiÁal, fuÁal, less commonly used, have a plural pattern
ÊafÁaal, which is “heavier” (al-Êatqal), in order to “redress the balance between them” (liyuÁaadiluu baynahumaa). Such explanations of “phonetic-statistical” nature are but intuitive
contentions, and cannot be regarded as logical arguments to explain linguistic phenomena. On p.
141 there is also an interesting paragraph indicating that PP patterns may replace PA patterns (or
regular plural patterns may appear instead of PA) just because both of them share the meaning of
the plural (li-štiraakihimaa fii l-jamÁ). On the other hand, PA patterns may substitute PP patterns
just because generic items may be used to convey particular items (yajuuzu Êan yunwà bi-lÁumuumi l-xu”uu”). The overall idea is that the plural meaning is common to both PP and PA
(maÁnà l-jamÁi muštarakun fii l-qaliili wa-l-katiir). However, the majority of these explanations
sound like ad hoc or a posteriori contentions. Moreover, they seem to be contradictory in a sense,
for on one hand generic items serve to denote particular items, and on the other hand particular
items may designate generic items because both of them belong to the same global entity. This
would result in a general lack of the specificity for both categories (PP and PA).
d) Ibn ÁAqiil (ob. 1375), Šar™: in a meaningful explanation of the section in the famous Alfiyya of
Ibn Maalik dedicated to PP, he defines (p. 465) PP and PA notions, adding that in a metaphorical or
figured sense either PP and PA may replace each other (wa-yustaÁmalu kullun [minhumaa] fii
mawñiÁi l-Êaaxari majaazaa). Other explanations are thus precluded, by simply accepting an
“irregular” extended use.
To sum up, the borderline between PP y PA is more and more diffuse. Later grammarians are
inclined to offer vague explanations on the reasons for the interference between both of them, or
simply recognise the lack of any explicit difference7, thus supporting the theory of a gradual loss of
discrimination.
8. TODAY’S SCOPE:
In a paper focused on Modern Arabic, Blohm (1994) reviews the actual status of PP, concluding
that there is essentially one plural category, with the use of PP in its traditional sense restricted to a
few fossilised words. He suggests, in addition, another possible use of the morphological
opposition, namely, the discrimination between an abstract and a concrete meaning of the plural
(nufuus “souls” vs. Êanfus “themselves”, wujuuh “faces” vs. Êawjuh “aspects”). While PP patterns
would tend to express abstract meanings, PA patterns would indicate concrete, physical entities.
However, Blohm’s study is quite limited and, as he himself acknowledged, further research is still
needed on this matter.
Drozdík (2001:73) offers another commentary: “The preference of plurals of paucity in combination
with 3-10 cardinals which can no longer be convincingly attested in the extant Classical Arabic
texts can, nevertheless, be illustrated on examples of the following type: ÊarbaÁatu Êajbulin min
jibaali l-janna “four mountains among the mountains of Paradise” or Êanna Áiddata l-bi™aari lmu™ii’ati bi-l-Êarñi xamsatu ab™ur “the number of oceans surrounding the land is five oceans”.
The overall impression8 received with regards to Modern Standard Arabic is that PP forms are
obviously restricted to a few items, the most known of them being šahr PP Êašhur PA šuhuur
“month”, where the two plural forms are perceived as different, not free variation patterns.
Similarly, the noun alf PP Êalaaf PA Êuluuf “thousand” retains the distinction between PP and PA
forms. But other similar plural pairs which were seen as distinct, not free alternative patterns for
Siibawayhi9, seem to be unfamiliar to modern Arabic speakers, who do not perceive any difference
between Êas’ur and su’uur, both plural forms of sa’r “line”, or Êaklub/kilaab plurals of kalb “dog” or
Êab™ur/bi™aar, plurals of ba™r “sea”, to give but three examples. Generally, the PP form is no
longer in use in the modern language, or it is merely perceived as an archaic and/or high-register
variant. Another area in which we still find some interesting data is the agreement between a head
noun and its modifier (mainly an adjective, more rarely a verb). It is, I believe, not very rare to find
expressions like mašaakil kibaar “big problems” or so, where the plural form of the adjective is
involved, as opposed to mašaakil kabiira “big problems”, where the adjective is fem. sg.. I assume,
however, that the writer’s purpose when using these forms is mainly to produce a high or ancient
flavoured form. In other words, these archaic-like devices are not guided by any sense of
discrimination between PP and PA notions.
9. THE OPINION OF ŠAWQƒ ïAYF
In some highly thought-provoking pages, the Egyptian scholar Šawqii ïayf (1990:56-64) questions
whether any of the different plural forms in Arabic (regular, collective and broken) was ever set
7
Sometimes the question is simply avoided. This happens, for example, in otherwise masterpieces of Arabic
philology and grammar, like Ibn F``ris’, a”-”``™ibhh, Ibn Jinnhh&r, Xa”``’i”, as-Suyuu’ii’s al-Muzhir.
8
A detailed study of these facts in modern Arabic prose is beyond the scope of this paper. Hence, we shall
limit ourselves to bring some data and draw some impressionistic inferences.
9
As clearly shown by expressions like tamaaniya ™ijajin ™ajajtuhunna bayta l-llaah “eight pilgrimages I
made to the house of God” (Siibawayhi, Kitaab, I, 178), or ÊarbaÁu ™arakaatin mutawaaliyaatin “three
consecutive vowels” (Ibn al-ÊAnbaarii, Asraar, 234), both of them very well marked to denote a P notion.
aside from the others for the expression of either PP or PA. He begins indicating that the Arabic
Language Academy of Cairo, in its 45th session, stated that both regular and broken plural express
both the small and the big number. After that, he presents some arguments for this statement.
Firstly he examines the regular suffixed plural, and the claim, supported by most Arab grammarians
since Siibawayhi, that it is actually a plural of paucity. ïayf disagrees, believin instead that the
regular plural stands for the expression of PP as well as PA. The arguments put forward are the
following:
1. The regular is previous to the broken plural, in accordance with the view of many Semitists. If
this is true, an ancient stage of the language with no PA at all must be posited, and this is rather
unthinkable.
2. In both the Koran and Pre-Islamic Poetry occurrences of supposed PP unequivocally denoting a
large number (PA) are well attested.
3. The broken plural pattern kakaakii/ik (in Arabic terms ”iiġatu muntahà l-jumuuÁ), which is by no
means a PP pattern, may bear a regular (allegedly PP) suffix (plural of plural).
As to the forms of the broken plural that have been traditionally considered as carriers of the PP
category, ïayf affirms that they serve as much for the PP as for the PA, and that the forms usually
classified within the PA category may also show a PP-like usage.
Faced with this state of affairs, the proposal of ïayf is to get rid of the “annoying” or “irritating”
distinction between PP and PA from the Arabic grammar, since it is an imprecise, unproductive and
non-rigorous distinction. The above mentioned morphological weakness is the reason why a
semantically well defined category doesn't find an appropriate, suitable and regular expression.
ïayf’s arguments aim to demonstrate that there is no clear parallelism between the notion of PP and
its form, that is to say, there is not an independent form of PP for each singular and that, whenever
there is a form, it is not used in an unequivocal way.
10. MY OWN VIEWS AND FIGURES
All this is true, much more for modern phases than for old phases of Arabic. But it should not be
forgotten that we are dealing with a primarily semantic category, and that its morphological
independent device is but partial. It can be said, in short, that the category of PP is barely productive
in Arabic (less in Modern than in Old Arabic), but there is no need to suspect an “invention” of
grammarians. On the contrary, it makes good sense to suppose that the intuition of Siibawayhi
together with his detailed description of PP forms respond to an observation of the actual facts of
the language, although these facts may be archaic or prone to archaism.
What about the ancient language? To the best of my knowledge, no specific studies exist on this
topic, i.e., the scope and extent of PP in, let us say, Pre-Islamic Poetry. As for the Koran, some hints
pointing in this direction may be found in traditional grammar and tafsiir sources, like the above
cited MaÁaanii l-qurÊaani by al-Farraa’. Modern studies also treat this and other related topics, but
mostly in a rather brief way, for example in as-SaamarraaÊii (1954).
Statistical evidence from the MuÁallaqaat:
As a preliminary survey of the ancient Arabic corpus which will be completed in the near future
with surveys of other poets and writers, let us now have a look at a few masterpieces of Pre-Islamic
Poetry, the muÁallaqaat. These poems’ language represents the formal Arabic of Pre-Islamic times,
a unified koiné language modelled for poetry and formal occasions. It is of course well
representative of the language Siibawayhi is actually describing. The language of poetry, however,
may be somewhat archaic in its structures and forms, for it tends to preserve usually older stages of
the language.
In order to examine a look at a possible variation between Central Arabic poets and Eastern Arabia
poets10 (no single muÁallaqa from Western Arabia does really exist), we have selected two poets for
each area:
1. Zuhayr ibn Abii Sulmà (dead about 615), from the Bedouin tribe of Dubyaan (Central Arabia),
2. al-©aarit ibn ©illiza (dead about 580), from the tribe of Bakr (Central Arabia)
3. ÁAntara ibn Šaddaad (dead about 615), from the tribe of ÁAbs (Eastern Arabia)
4. ÁAmr ibn Kultuum (dead about 600), from the tribe of Taġlib (Eastern Arabia).
Our purpose is to check out plural forms in these poems to make sure that PP plays a role and is
different from other plurals. It is indeed true that the choice of a PP or a PA pattern in a given
context could reflect to some extent the demands of meter11. But, together with Belnap and Gee
(1994:129), we consider that the poets’ plural pattern usage is doubtlessly representative of their
language, for there are no substantial discrepancies between one poet or another in the following
figures:
PA-A
PP-P
PA-P
PP-A
NDF
DF
Zuhayr
10
13
2
0
23
2
al-©aarit
12
7
3
3
19
6
ÁAntara
7
4
6
6
11
12
ÁAmr
17
20
6
16
37
22
Total
46
44
17
25
90
42
Fig.1. Occurrences of PP and PA forms in the above indicated MuÁallaqaat
where:
PA-A = PA forms within a context of Abundance
PP-P = PP forms within a context of Paucity
PA-P = PA forms within a context of P
10
See Frolov (2000.224-38), where clear attested differences in the choice of meters by the different groups
of Pre-Islamic poets are discussed, according to geographical distribution. On this basis, it would not be
surprising to find some differences in agreement patterns or in the use of PP/PA patterns between Eastern,
Central and Western poets.
11
I’d like to thank the suggestions and remarks made at this respect by Professor G. Bohas at the presentation
of this paper in the ALS Symposium in Cambridge.
PP-A = PP forms within a context of A
DF = Total of non expected or deviated forms
NDF = Total of expected or non-deviated forms
PA-A
PP-P
PA-P
PP-A
NDF
DF
Zuhayr
20%
26%
4%
0%
92%
8%
al-©aarit
21.43%
12.50%
5.36%
5.36%
76%
22%
ÁAntara
17.95%
10.26%
15.39%
15.39%
47,8%
42,2%
ÁAmr
12.69%
14.92%
4.48%
11.94%
62,7%
37,3%
Total
16.49%
15.77%
6.09%
8.96%
69,62%
30,48%
Fig. 2. Percentages of PP and PA forms in our corpus
Observations:
1. It is at times very difficult to establish to some degree of certainty if a given context is one of P or
A. These figures should then be treated as preliminary, for they are far from being conclusive. In
fact, different angles of interpretation and some ambiguity may indeed produce a margin of error in
our figures, but I think that no substantial modifications are to be expected after a more refined
examination of the texts.
2. I consider particularised or individualized contexts closer to P categories. In the case of the
modifiers of a head-noun (adjectives, pronouns and verbs), the feminine plural forms, clearly used
to denote separate items, are included in the P category, whereas feminine singular ones, bearing a
collective, general meaning, are placed within the A category12.
3. All forms of plural, broken and regular, are included, even f.s. adjectives qualifying a plural noun
(this is known as deflected agreement, Zuhayr 43 buyuutan katiiratan “many houses” and Zuhayr
49 Êumuurin katiiratin “many things”), as these forms could be interpreted form the point of view
of agreement facts as the plural pattern of the adjective. Pronouns and verbs are also taken into
account.
Cases where a PP or PA form agrees with its semantic value:
1. Zuhayr 9: anmaa’ Áitaaq “ancient saddle-rugs”, where the context clearly refers to a few items.
2. Zuhayr 36: buyuutan katiratan “many houses”, a clearly A context.
5. Zuhayr 49: fii Êumuurin katiiratin “in many respects, things”, a clearly A context.
4. al-©aarit 48: Êaayaatun talaatun fii kullihinna l-qañaaÊu “three clear-cut proofs, signs”, where
the context is clearly P and both the noun plural and the pronoun are PP.
5. al-©aarit 61: wa-Êataynaahum bitisÁati Êamlaakin kiraamin “we brought them nine noble kings”,
a clear P context where the noun Êamlaak is a quite rare paucity plural pattern for malik “king” (the
usual form is PA muluuk).
6. ÁAmr ibn Kultuum 77: Áalaynaa l-bayñu wa-l-yalabu l-yamaanii / wa-Êasyaafun yaqumna wayan™aniinaa “we have the Yemenite helmet and shield, and swords which are either straight or
curved (bent)”, where the poet alluded to particular swords, being both the noun and the two verbs
typical P forms.
12
But see the interpretation of this very facts according to the second claim by al-Farraa’ analysed therein.
Cases of disagreement:
1. al-©aarit 13: wa-’iraaqin min xalfihinna ’iraaqun / saaqi’aatun Êalwat bihaa ”-”a™raaÊu , where
a PP pronoun -hinna alternates with a PA pronoun -haa both referring to the same thing.
2. al-©aarit 73: wa-tamaanuuna min tamiimin biÊaydii- / him rimaa™un ”uduuruhunna l-qañaaÊu
“eighty of Tamiim people with lances in their hands, whose edges are the death”, where a f.s. P
form –hunna does appear instead of the expected PA form -haa.
3. ÁAntara 12: fiiha tnataani wa-ÊarbaÁuuna ™aluubatan suudan “there are in it forty-two milkgiving black she-camels”, where suudan, a broken pl. form (P form) appears instead of the
expected PP form sawdaaÊ, in sharp contrast with the previous large number.
4. ÁAmr ibn Kultuum 25: wa-Êayyaamin lanaa ġurrin ’iwaalin / Áa”aynaa l-malka fiihaa Êan
nadiinaa “Oh, that long and glorious days we refused to serve the king in”, where we find two
broken pl. adjectives ’iwaal and ġurr (P forms) and a sg. suffixed pronoun -haa (A form) both
referring to the noun “days”, in a context close to Q.
5. ÁAmr ibn Kultuum 83: waritnaahunna Áan ÊaabaaÊi ”idqin / wa-nuurituhaa Êidaa mutnaa
baniinaa “we inherited them (horses) from our good fathers and we will leave them at the time of
our death to our sons”, where the first pronoun referring to the horses (in bold) is f.pl. (P form)
whereas the second is f.sg. (A form) without any apparent semantic variation.
6. ÁAmr ibn Kultuum 92: wa-s-suyuufu musalsalaatun “being the swords unsheathed”, where a PA
noun is followed by a clear PP f.pl. adjective.
7. ÁAmr ibn Kultuum 94: Êidaa qubabun bi-Êab’a™ihaa buniinaa “when the tents were raised in the
plain”, where a PA noun is followed by a clear PP f.pl. verb.
Some conclusions
1. Broadly speaking, the discrimination between the notions of Abundance and Paucity seems to be
mantained in our corpus only to a certain extent. It should be pointed out that the percentage of
cases where there is simply no way of making a distinction between PP and PA surpases half of the
total plural or plural-like forms, according to our statistics (52.69% of the total sample). This
numbes speak for themselves: the rate of nouns accepting variation in the plural is less than half of
the total number of nouns. There is little room for the choice between PP and PA forms, but it is
perhaps more than “insignificant”. Within these forms in which variation could be theoretically
attested, the rate of agreement form-meaning (PP-P and PA-A) is higher (69,62%) than that of lack
of agreement (30,48%), thus suggesting that a tendency to parallel form and meaning has to be
acknowleged. If we consider the data from Central Arabia poets as against the Eastern poets, the
result is that Central poets tend to use more coincidental forms (84%) than Eastern ones (55,25%).
One is tempted before this state of matters to suggest that Eastern poets represent a more innovating
type of language, in which the trend towards the decay of the PP category was more advanced,
whereas Central poets are more conservative at this point13. But, once again, I should stress that
these contentions, based as they are on provisional statistics, are only preliminary.
It would be of interest to compare these figures with others obtained through the examination of
later poetic corpus, as partially made by Belnap and Gee (1994:127, 132), in order to know if, as
13
In this respect, it is useful to take into account that Frolov (2000) has convincingly stated that Eastern poets
make a more innovating use of metrical patterns, whereas Central and Western poets are more conservative.
This is probably because of the belonging of Eastern poets to the al-âiira court, famous for its contact with
Persian civilization and for its more innovating air.
expected, crossing forms tend to be more numerous the more we move towards later times, thus
implying that the A-P distinction was gradually disappearing along with the evolution of Arabic.
2. These findings and inferences come to confirm that Siibawayhi’s insights are not devoid of value
and significance. PP was indeed an archaic feature already at Siibawayhi’s time. But it is well
known that he was actually describing an older stage of the language, in which the Plural of Paucity
could has been treated as an integral part of Arabic, and this is the way Siibawayhi analyzed it,
despite some shortcomings. PP was, I believe, still productive at Pre-Islamic times, but there are
many indications to suggest an old tendency towards its fall. This supports, by the way, a wellknown contention recently developed by Owens (2001:420-5) in the sense that linguistic variation
in the first stages of the history of Arabic was probably higher and more widespread than Arab and
Western scholars usually acknowledge. It is not difficult to imagine that some regional variation in
the field of plural assignment was responsible for the attested differences, although we are far from
being able to prove such contentions14.
11. PP’S POSSIBLE CONNECTION TO AGREEMENT FACTS: AL-FARRAA’S CLAIM
In the work of the well-known Arab grammarian al-Farraa’, MaÁaanii l-qurÊaan, p. 435, we read a
very stimulating comment on the differences in the plural agreement distribution. These are his very
words:
wa-kadaalika kalaamu l-Áarabi li-maa bayna t-talaatati Êilà l-Áašrati taqaal: li-talaatati
layaalin xalawna wa-talaatatu Êayyaamin xalawna Êilà l-Áašra. fa-Êidaa juzta l-Áašrata
qaaluu: xalat wa-mañat wa-yaquluuna li-ma bayna t-talaatati Êilà l-Áašrati «hunna» wa«haaÊulaaÊ» fa-Êidaa juzta l-Áašrata qaaluu «hiya» wa-«haadihi» Êiraadatan Êan tuÁrafa
simatu l-qaliili Áani l-katiir
“and also in the Bedouins language, for three to ten items, they say: «three nights or three
days remain [to the end of the month]» (fem. pl. verb form), and so until ten. But if you
surpass ten, they say: «it remains, it has gone» (fem. sing. verb form). From three to ten items
they shay «they», «these» (fem. pl. pronouns) whereas if you surpass ten, they say «she»,
«this» (fem. sg. pronouns) in order to make the distinction between the quality of small
number from that of big number”.
If we follow the same line of reasoning this would mean that buyuut kabiira (f.sg. adjective) was
the choice for large numbers whereas buyuut kibaar (pl. adjective) was the choice expected for
small numbers. Other scholars15 mentioned this contention of al-Farraa’, but, as far as I know, none
have developed an integrated theory on account of these and similar statements. Could this
discriminating usage be one of the reasons behind the old variation between strict and deflected
agreement in verbs, adjectives and pronouns? One could add to the study of the transition from an
14
However, it should be emphasized that other domains of language clearly show a good deal of variation,
like lexical units or phonetics, as convincingly stated by many scholars, starting from Rabin (1951). The very
same Siibawayhi deal with some variation in his book, according to Levin (1999).
15
See, for example, the brief comment of Hopkins (1984:144): “In C(lassical) A(rabic) when inanimate plural
nouns do not exceed the number of ten the concord may take place in the feminine plural rather than the
feminine singular”. The use of “may” indicates here the assumption of this agreement distribution as an
optional, non categorical feature.
almost free variation (Old Arabic) to a near categorical deflected agreement (Modern Standard
Arabic), as analysed By Belnap and Gee (1994), that perhaps the distinction between the notions of
PP and PA justifies the adoption of strict agreement for expressing a few individualised items and
the deflected agreement for larger collectivised groups. Hence, the lack of semantic discrimination
could well have led to the decline of the morphological opposition. Note that the pattern of
discrimination between both types of plural agreement does not correlate Siibawayhi’s criteria on
the use of PP as opposed to PA. According to al-Farraa’, sg. f. pronouns and verbs are to be
considered as PA forms, whereas the plural counterparts of these pronouns and verbs stand for PP
agreement. This contention may be safely extended to attributes (mainly adjectives, but also relative
clauses). We should, on this basis, carefully separate the actual facts of agreement between nouns
and attributes, as explained by al-Farraa’, from pure morphological shapes related to isolated nouns,
as taken by Siibawayhi. Although both perspectives could appear at a first glance not easy to
reconcile, it may be suggested that al-Farraa’ deals with the agreement variation f.sg vs. pl, whereas
Siibawayhi focuses solely on the variation within pl. (not sg.) patterns.
I firmly believe that a careful research carried out on old pre-classical texts, paying attention to the
meaning provided by context, would result in a clearer picture of the reasons for the apparently
unmotivated variation on agreement facts (natural vs. deflected). Until more detailed surveys are
undertaken, we can limit ourselves to the following, brief outline.
What about agreement frequencies in our corpus? Let’s have a look at these figures:
Pl. V
Sg. V
Zuhayr
3
2
al-©aarit
0
2
ÁAntara
ÁAmr
5
1
Total
9
0
17
5
Figure 3. Frequencies of strict and deflected agreement (verbs)
where Pl.V stands for plural verb forms, that is to say P (individualized) forms, whereas Sg. V
indicates singular verb forms = P forms
Pl. Ad.
Sg. Ad.
Zuhayr
5
2
al-©aarit
2
2
ÁAntara
ÁAmr
3
0
Total
10
1
20
5
Figure 4. Frequencies of strict and deflected agreement (adjectives)
where Pl.Ad stands for plural adjective forms, that is to say P (individualised) forms, whereas Sg.
Ad. indicates singular adjective forms = P forms
Pl. Pr.
Sg. Pr.
Zuhayr
0
0
al-©aarit
3
0
ÁAntara
ÁAmr
0
0
Figure 5. Frequencies of strict and deflected agreement (pronouns)
Total
3
2
6
2
where Pl.Pr stands for plural pronouns, that is to say P (individualized) forms, whereas Sg. Pr
indicates singular pronouns = P forms
Although our samples are far from being great enough to draw valid statements, they suggest that P
forms dominate the scene. This agrees with Belnap and Gee’s (1994)16 findings: full agreement is
the system preferred by Pre-Islamic poets, with the spread of deflected agreement rather scarce.
Moreover, it seems that the passage from full to deflected agreement is faster and earlier in
pronouns than in verbs, whereas adjectives constitute the more conservative area in this respect,
showing more reluctance to abandon paucity (here plural) patterns. Once again, these preliminary
results mostly agree with Belnap and Gee’s (1994:132) observation in the sense that “adjectival
agreement shows the highest percentages of strict (plural) agreement and is supposedly the most
conservative agreement locus”.
With regards to the correspondence between the agreement forms and the general meaning of the
verses in which they are included, three more charts may be of interest here:
PP-P
PA-A
PP-A
PA-P
Zuhayr
3
0
0
2
al-©aarit
0
0
0
2
ÁAntara
ÁAmr
2
0
1
1
Total
7
0
2
0
12
0
3
5
Figure 6. Verbs agreement
PP-P= Paucity verbs (plural forms) within Paucity context
AP-A= Abundance verbs (singular forms) within Abundance context
PP-A= Paucity verbs (plural forms) within Abundance context
AP-A= Abundance verbs (singular forms) within Paucity context
PP-P
PA-A
PP-A
PA-P
Zuhayr
5
2
0
0
al-©aarit
2
2
0
0
ÁAntara
ÁAmr
0
0
2
0
Total
5
0
5
1
12
4
7
1
Figure 7 Adjectives agreement
PP-P= Paucity adjectives (plural forms) within Paucity context
AP-A= Abundance adjectives (singular forms) within Abundance context
PP-A= Paucity adjectives (plural forms) within Abundance context
AP-A= Abundance adjectives (singular forms) within Paucity context
PP-P
PA-A
16
Zuhayr
0
0
See their tables on p. 127 and 133.
al-©aarit
2
0
ÁAntara
1
0
ÁAmr
Total
1
1
4
1
PP-A
PA-P
0
0
1
0
0
0
2
1
3
1
Figure 8. Pronouns agreement
PP-P= Paucity verbs (plural forms) within Paucity context
AP-A= Abundance verbs (singular forms) within Abundance context
PP-A= Paucity verbs (plural forms) within Abundance context
AP-A= Abundance verbs (singular forms) within Paucity context
It is clear from these figures, although the sample is small, that correspondence between form and
meaning is higher in adjectives (16 correspondent forms against 8 crossing forms), then in verbs
(12/8) and lower in pronouns (5/4).
To sum up, little can be affirmed concerning agreement. We can draw only a hypothesis: the
distribution of PP-PA forms according to the context is one of the primary criteria to be handled
when trying to explain the apparently unmotivated free variation of agreement patterns as discussed
by Belnap and Gee (1994). Further studies could confirm this hypothesis through a critical
examination of later Arabic sources, in order to assess the evolution of PP meaning and forms.
An important, general conclusion is that the notion of PP was firmly rooted in the minds of early
Arab linguists. But the fact that two of the most prominent Arab grammarians, Siibawayhi and alFarraa’, deal with the notion from different, even in some respect contradictory perspectives,
suggest once again that PP as opposed to PA represent an archaic-like notion which is not met by
morphological devices except for a few patterns. In other words, PP may be seen as an optional
feature at the disposal of Arabic writers, but not, as far as we know, a living productive notion.
12. REFERENCES
12. a. Arabic sources:
al-FarraaÊ, MaÁaanii l-qurÊaan, ed. A. Y. Najaatii y M. Á. an-Najjaar (1980, 2ª), El Cairo.
Ibn al-ÊAnbaarii, ÊAsraaru l-Áarabiyya, ed. C.F. Seybold (1886), Leiden.
Ibn ÁAqiil, Šar™u bni Áaqiilin Áalà ÊAlfiyyati bni Maalik, ed. Q. Š. ar-RifaaÁii (1987), Beirut.
Ibn Faaris, a”-“aa™ibiyyu fii fiqhi l-luġa, ed. M. El-Chouémi (1964), Beirut.
Ibn Jinnii, al-Xa”aaÊi”, ed. M.Á. an-Najjaar (1952-6), El Cairo, 3 vols.
al-Mubarrad, Kitaabu l-Muqtañab, ed. M. Á. l-Xaaliq (1966-9), El Cairo, 4 vols.
Siibawayhi, Kitaabu Siibawayhi ÊAbii Bišrin Áamri bni Áutmaana bni Qunbur, ed. Á. S. M. Haaruun
(1977, 2ª); El Cairo, 5 vols.
as-Suyuu’ii, al-Muzhiru fii Áuluumi l-luġati wa-taariixi Êašhari n-nu™aat, ed. Beirut (s.f.), 2 vols.
as-Suyuu’ii, al-ÊIqtiraa™u fii Áilmi Êu”uuli n-na™w, ed. Alepo (1983, 2ª).
az-Zajjaajii, Kitaabu l-jumali fii n-na™w, ed. Á. T. al-âamad (1984), Beirut.
az-Zamaxšarii, al-Mufa””alu fii Áilmi l-luġa, ed. M. as.SuÁaydii (1990), Beirut.
12.b. Modern studies:
R.K. Belnap & J. Gee (1994), “Classical Arabic in contact: The transition to near categorical
agreement patterns”, in Perspectives on Arabic Linguistics VI, 121-49.
D. Blohm (1994), “Ist der Wenigkeitsplurale im Modernen Hocharabisch eine grammatische
Kategorie?”, in Zeitschrift der deutschen morgenländischen Gesellschaft Supp. 10, 192200.
C. Brockelmann (1982) (first ed. 1908-1913, Berlín), Grundriss der vergleichenden Grammatik der
semitischen Sprachen, 2 vols., Hildesheim.
G. Buccellati (1997); “Akkadian”, in R. Hetzron (ed.); The Semitic languages. London-New York,
69-99.
F. Corriente (1971), Problemática de la pluralidad en semítico: el plural fracto, Madrid.
Š. ïayf (1990), Taysiiraat luġawiyya, Cairo.
L. Drozdík (2001), Modern Written Arabic. Al-Áarabiyyatu l-muÁaa”ira, Bratislava.
I. Ferrando (1999), “El plural fracto en semítico: nuevas perspectivas [a review article of Ratcliffe
1998]”, in Estudios de dialectología norteafricana y andalusí 4, 7-23.
I. Ferrando (2001), “Siibawayhi y el concepto de jamÁu l-qilla”, Al-Qan’ara XXII/2 (2001), 271297.
C.A. Ferguson (1959), “The Arabic Koine”, Language 35 (1959), 616-630.
W. Fischer (1980), “Die arabische Pluralbildung”, in Zeitschrift für Arabische Linguistik 5, 70-88.
H. Fleisch (1961), Traité de Philologie Arabe, I: Préliminaires, phonétique, morphologie nominale,
Beirut.
D. Frolov (2000), Classical Arabic Verse. Hisotry and Theory of ÁAruuñ, Leiden.
P. Fronzaroli (1974); review of Corriente (1971), in Journal of Semitic Studies, 275-284
S. Hopkins (1984), Studies in the Grammar of Early Arabic, New York.
A. Levin (1999), “The first book of Arabic Dialectology: Siibawayhi’s Al-Kitaab”, in Jerusalem
Studies in Arabic and Islam 23, 208-220.
A. Levin (2003), “From Suffix to Prefix: Some Synchronic and Diachronbic Aspects Concerning
the t- Prefix Preceeding the Counted Nouns from three to ten”, in AIDA 5th Conference
Proceedings, Cadiz, September 2002, ed. by I. Ferrando & J.J. Sánchez Sandoval, Cadiz.
M.M. Levy (1971), The Plural of the Noun in Modern Standard Arabic, Michigan (PhD.
Dissertation).
E. Lipiński (1997), Semitic languages: outline of a comparative grammar, Louvain muÁallaqaa.
A. Murtonen (1964), Broken Plurals: The Origins and Development of the System, Leiden.
J. Owens (2001), “Arabic Sociolinguistics”, in Arabica 48, 419-69.
F.R. Palmer (1962), The morphology of the Tigre noun, Oxford.
Ch. Rabin (1951), Ancient West Arabian, London.
R. Ratcliffe (1998), The ‘broken’ plural problem in Arabic and comparative Semitic. Allomorphy
and analogy in non-concatenative morphology, Amsterdam-Philadelfia.
as-SaamarraaÊii, I. (1954); “Statistiques des pluriels internes dans le Coran”, in Bulletin de la
Societé de Linguistique de Paris L, xxviii and next pages (quoted in Fronzaroli (1974:280).
M.B. Schub (1978), “Plurals of paucity and abundance”, in Arabica 25/2, 204-5.
W. von Soden (1969), Grundriss der Akkadischen Grammatik, Roma.
A. Zaborski (1991), “Biconsonantal roots and triconsonantal root variation in Semitic. Solutions
and prospects”, In Semitic Studies in honour of Wolf Leslau II, 1675-1703.