resembling those cited here (1928: 3 1 1 ).
3.1 Relativization of 1ndefinite-Specific Nouns with lillil
95
these two markings, one indefinite and the other definite, to indicate the existence of some particular girl.
K2
J ... . L.:u L:.., - . J..)J
.�
�ndawwir-la bnayya IIi tnasib-la we-seek-for-him girl reI she-suits-him We look for a girl for him that will suit him
This construction may be contrasted to the (normative) alternate, /bnayya tnasib-la/, which would mean a/any girl that would suit him, the identity of such a girl being less specific.
In a number of examples of this type, the head noun is human, including /bnayya/ girl in the previous Kuwaiti example and /rayyal/
man in the next:
K3 �
L.� �.,...!.
1;1 1 �� I J�..) w�
t... � -".J�.,...!.
. .
�I
� �'IJ
� �'I
�ufi 10 -- kil rna yikun rayyal illi y1:tibb ii-mara �ufi binti rna yihimma la 1:taCi hala wala yihimma 1:ta�i 'al:tad
see-f if -- every nom it-be man reI he-loves the-wife see-f daughter
my neg it-concerns-him talk of-family-his nor it-concerns-him talk of-anyone
Look, if -- whenever there is a man who loves [his] wife, see, my daughter, neither the words of his family nor anyone else's words concern him
Similarly, the following Syrian example contains the indefinite human noun /nas/ people relativized with /yalli/ (Cowell 1964:499):
S ...u"J.I I�I � U"L:. � � wlS ..:..J.,JI dl�
b-hadak �l-wa)�t Un fi ktIr nas y�lli stagallu I-maw'ef
[in-that the-time was-it there-is many people reI exploited-they the-situation]
At that time there were a lot of people who took advantage of the situation
Since humanness is one of the most important features in the hierarchy of individuation, the tendency of this mixed marking to occur with human nouns lends further support to the theory that the individuation of a head noun can affect the syntax of relative clauses.
One might then ask why speakers only occasionally make use of this pattern, since indefinite nouns described by relative clauses by definition have a degree of specificity. In the following example, the speaker mentions an American friend who arrived recently, using an indefinite relative clause (Cowell 1 964:497).
S
.l� 1 t ..L:!� 40 1�lS�i J-:!� <!� �
fi 'andi �adi' 'amerkani '�za zdid 'al-�lad
[there-is at-me friend American came-he new to-the-country]
1 have an American friend who has just recently come to this country
The speaker obviously has a particular person in mind, and the combined features of specificity and animacy might be expected to "push" the speaker towards using /illi/ as an indication of the heightened specificity of the noun. However, native informants much prefer the indefinite clause to a definite clause introduced by /illi/. The following sentence was deemed unlikely to occur:
S
.l� 1 t ..L:!� 40 1� I �lS�i J-:!� <!� � "
*fi 'andi �adi' 'amerkani iii '�za Mid 'al-'blM
*there-is at-me friend American rei came-he new to-the-country
*1 have an American friend who has just recently come to this country
The best explanation for these seemingly paradoxical data appears to lie in the pragmatic role of the noun itself. In this sentence, the noun /�adi' 'amerkani/ an Americanfriend represents an entity being introduced for the first time. The speaker needs to establish both the new identity and the topical role of the person, and therefore must mark the noun as indefinite. Pragmatic and discourse roles thus also play a role in determining the use of definite and indefinite markings in relative clauses.
Relativization also provides further evidence that nouns marked with new-topic article /wal:tid/ (Moroccan /wal:t�d 1-/; see 1 .5) fall within the indefinite range of the definiteness continuum. In all dialects, nouns marked with this article are modified as indefinite nouns (Harrell specifies this rule for Moroccan, 1 962: 165). In the following, /wal:t�d
�r-rz�l/ one foot heads an indefinite relative clause /fih buz�llOm/ it has
3.1 Reiativization of Indefinite-Specific Nouns with lillil
97
arthritis. Here, the absence of the defInite Illi/ "balances" the specifying article Iwal,t�d l/ sc that the combination of syntactic markings does not overly "weight" the noun with defIniteness.
M9 r.,jj-,";,
�.:Jw,6.
�.,J' oAA.'.J 4..JdJ.J c,t!"i
�JU
qal liyya 'azi dlak-Iu wal,tad r-dal J:ta�ak f"lh buzallum
said-he to-me come rub for-him one the-foot may-it-avoid-you in-it arthritis
He told me, come and massage for him a leg that had-may you be spared-arthritis
In contrast, Moroccan nouns with indefInite-specifIc article �i/
are commonly relativized with /Ilil (Harrell 1 962: 165). Harrell cites an example containing �i tumubIl/ a car ( 1962: 1 65):
M
u _..)A � .L
. • I.F"'" - ' • ..::....a.L bgU si tumubIl lli tam�i mazyan [wanted-I si car reI it-goes good]I want a car that will run good
More difficult to explain is the following Kuwaiti example, in which clearly non-specifIc lay by/ anything is modifIed using /illi/:
K2
ay say 'indie muskila umm aJ:tmad iQna J:ta.� vll' �'!' .&:!...,...6. �, �, r' us..:.... �� 4
rin. 'ay say illi tabIn�'!'
any thing at-you problem Umm Ahmed we present. any thing reI you-want
Any kind of problem you have, Umm A hmed, we're here. Any thing that you want
The probable explanation for the use of defInite /illi/ here is that the speaker wishes to give prominence to lay by/ anything, to stress that every single request will be granted.
Speakers from Egypt, Syria, and Kuwait thus make use of a combination of defInite and indefInite markings to indicate a higher degree of specifIcity or individuation in head nouns of relative clauses.
However, pragmatic and discourse factors seem to mitigate against the widespread use of this combination.
Classical Arabic grammars stipulate that relative clauses that modify a definite head noun must be headed by a definite relative
particle, yet Wright cites the following Classical examples which violate this rule ( 1 898ii:3 1 8):8
• • L..
rna yanbagi li-r-rajuli yu�bihuka
that which beseems the man who is like thee
,
. Jl.L...i
� t...-JI _ Jka-ma!ali l-1:1imari yal;1milu 'asJaran like the ass which carries books .)L...)
I vl
�.H �lSka-I-jamn yuc,ia'u fi r-ramadi
like the coal which is put among the ashes
The grammaticality of these sentences rests on an interpretation of each head noun as indicating "not a particular individual (animate or inanimate), but any individual bearing the name" (Wright 1 898ii: 3 1 8).
Thus these definite nouns are somehow "less definite" than others, or more properly less individuated, and the syntax of the relative clause that modifies these nouns reflects that status.
The mirrored relationship between the dialect examples on one hand and Wright's Classical data on the other is worth noting: in the dialects, the noun is indefinite and the descriptive clause carries defmite marking, whereas in Classical Arabic, the noun carries definite marking while the absence of the relative pronoun indicates the non-specific of the referent. Both varieties of Arabic use similar strategies in combining definite and indefinite markings to weight partly definite, partly specific nouns in relative clauses, except that Classical Arabic marks definite non-specific while the dialects, in general, mark indefinite-specific.9
In either case, the specificity of the head noun affects the syntax of the entire clause. The use of the definite relative pronoun lillil in relativizing an indefinite head noun cannot be predicted or explained