entire region), but contains /d/ in the quantifying phrase /s�b'a d-I-U1ad/
seven sons( 1 980: 1 3 1 ).
2.5 Summary
87
quantification, which the exponents of the other dialects may not do.
The higher frequency of Moroccan /dyal/ is thus partly explained by its greater functional capacity.
Chapter 1 demonstrated that animacy and specification seem to affect definite marking in Moroccan nominal phrases in such a way that Moroccan speakers tend to prefer definite marking on animate and specified nouns. It seems possible that the high frequency of /dyal/
relative to exponents of other dialects may be partly related to the relatively wider use of the definite article in Moroccan speech, since /dyal/ allows the definite article to be retained on one or both nouns in a genitive phrase.
2.5 Summary
This chapter has explored the influence of individuation on two types of noun modification, number agreement and possession. The discussion of number agreement included an exploration of the category of dual in spoken Arabic.
Examination of various phenomena of dual number and agreement led to several conclusions and a hypothesis concerning its historical development. Building on Blanc's distinction between dual and pseudo
dual (1970), I proposed further distinctions between productive and non-productive, and specific and non-specific duals. The category of dual consists of several different kinds of dual: frozen fonns, most of which refer to body parts, a non-specific dual meaning a couple of, and a "new topic" dual that has specific pragmatic functions and is limited to individuated nouns. The productivity of the dual suffix /-en/ on nouns appears to be limited, either to certain classes of words in some cases, or to pragmatic functions in others. Periphrastic dual fonn s appear to be the nonn i n Morocco and Kuwait; and the dual suffix /-ayn/ is not a productive fonn in Morocco. This evidence led to a call for a reevaluation of the history of the dual in spoken Arabic, one that takes into account category-specific developments.
Number agreement patterns in the dialects were shown to be affected by the hierarchy of individuation. The agreement of adjectives, verbs and pronouns tends to be plural if the noun is highly individuated,
i.e., is specific, definite, and/or high in textual or social prominence, an agreement pattern called here individuated plural agreement. By contrast, plural nouns that reflect collective, abstract, non-specific, and/or inanimate entities of low textual prominence tend to attract feminine singular or deflected verb and pronoun agreement (adjectival agreement less so, because adjectives tend to specify, giving the nouns they modify greater individuation). This latter agreement pattern is designated as collective plural agreement.
Similarly, the individuation of a possessive phrase, and particularly the possessor, influences the choice of genitive construction, such that the more individuated the phrase as a whole, or one of its members, the more likely the speaker to express that relationship using a genitive exponent. Formal and sociolinguistic motivating factors for the use of the exponent were also noted.
Within this general shared framework, slight variations occur.
Moroccan speakers seem to have a greater tendency to use plural agreement patterns than speakers of other dialects. This tendency provides an interesting parallel to the high frequency of the genitive exponent /dyal/ in Moroccan, and also the greater use of definite and specifying articles in Moroccan noted in Chapter 1 .
A distinction i s found between the two central dialects on the one hand and the two peripheral dialects on the other in the use of the genitive exponent (as opposed to the construct). Moroccan and Kuwaiti speakers can use genitive exponents to indicate classificatory as well as individuated identity, while Egyptians and Syrians tend to restrict the use of the exponents to individuated phrases. The fact that the Moroccan exponent /dyal/ has a much higher frequency than Kuwaiti /mal/ is probably due in part to the Moroccan tendency to mark individuation in nouns and to its wider syntactic range (/dyal/ also quantifies nouns between two and ten in number). Even so, this parallel between Moroccan and Kuwaiti in the use of genitive exponents bears watching for future developments.
3 RELATIVE CLAUSES
3.0 Introduction
Ferguson (1959:630) includes the definite relative pronoun JilIi/
as one of fifteen features common to most modem Arabic dialects. I Grammars of the dialects concur: Mitchell (1956:57) notes that the definite relative pronoun for Egyptian is /ilIi/; Cowell ( 1964:494) lists for Syrian /halli/ and /yalli/ as variants of /illi/;2 Johnstone ( 1 967) gives filIi! for Kuwaiti; and Harrell (1962: 164) notes /lli/ and /a�/ for Moroccan.
These grammars specify that /ilIi! only relativizes a definite head noun, and that no relative pronoun is used if the head noun is indefinite.
In his Moroccan grammar, Harrell claims that /lli/ in definite relative clauses "indicates specifically that the subordinate clause is a restrictive adjectival modifier" of the head noun ( 1 962: 1 64). This claim is not made for any other dialect, and it is refuted for Moroccan by this counter-example from my data in which /lli/ modifies the already restricted /I-walida/, [my] mother:
M2
�.J � '!� .:.4- .}J 1
ioU I"J I �JlA. .:..:ats kant xarfa l-walida iii fat 'ndi hiyya w 'ammtiwas-she going-out the-mother reI came-she at-me she and aunt-my My mother, who had come to my house, she and my aunt, was going out
This example shows that /ilIi! can modify both restricted and non
restricted nouns in Moroccan, as it can in the other three dialect regions.
IRelativization in formal Arabic utilizes a set of definite relative pronouns inflected for gender and number (see Cantarino 1975iii:162), which contrast with the invariant pronoun lillil used in spoken Arabic.
2It is not clear whether Iyalli/, /hallil and lilli! are free variations or reflect some nuance of meaning. Some of Cowell's examples suggest that Iyallil tends to be restrictive, whereas /hallil appears to be parallel to anaphoric demonstrative /hal in that the reference is to an established or otherwise identifiable referent, and is thus non-restrictive (Cowell 1964:495-8; see also 4.2). Cowell has few examples of lilli/, whereas my northern Syrian texts contain only lilIi/; this discrepancy may reflect regional differences. Cowell prefers to call these particles attributives, since in his view they do not correspond to the English relative pronouns (1964:495).
89
Relativization in spoken Arabic thus seems at first glance to be almost uniform and highly analogous to the rules of literary Arabic in its distinction between definite and indefinite head nouns and clauses, the existence of a definite relative pronoun and the absence of an indefinite one, and the lack of distinction between restrictive and non
restrictive clauses. Formal Arabic and most of the dialects make use of what Keenan and Comrie ( 1 977) call case-marking relativizing strategy;
in Arabic relative clauses, a resumptive pronoun normally marks the syntactic position of the relativized noun.3
However, relativization in the four dialect regions turns out to be more complicated, and more interesting, than this simplified picture suggests. Relative clauses in spoken Arabic include a variety of structures. First, the pronoun lillil can relativize not only definite nouns, but also indefinite nouns, in all four dialect regions. Second, while lillil clauses normally leave a resumptive pronoun in the position of the relativized noun, the resumptive pronoun in the direct object position is optional in many varieties of Arabic.4 Among the four regions examined here, Moroccan speakers in particular regularly omit the resumptive pronoun in direct object position, whereas Egyptian and Syrian speakers regularly reject its omission. My Aleppan data contain lillil clauses in which the head noun of the relative clause stands in construct (li<;lafal) with lillil, rather than the more usual noun-modifier construction in which the head noun is marked definite. At the same time, the function of lillil extends beyond relative clauses. Several dialects use lillil as nominalizer andlor sentential complementizer in certain contexts that will be explored here.
Relativization with particles other than lillil, while limited geographically, is of comparative interest. While non-attributive relative clauses headed by lillil are found in all areas, non-specific, non-attributive relative pronouns Imal and Iminl appear only in certain areas: Imal