3.1 Relativization of Indefinite-Specific Nouns with /illi/ 91 something, anything occurs only in my Moroccan data and Iminl someone, anyone only in Syrian.s The existence of non-specific relative pronouns provides an important parallel to the existence of an indefinite
specific article {§il in these two dialects ( 1.4). Moroccan speakers also have at their disposal an additional relativizing strategy in which oblique objects may be relativized with pronoun lia§! in a word-order type relative clause. The final section documents examples of a relative pronoun Idl and variants in the speech of an elderly woman originally from rural northern Morocco.
How do speakers of these dialects manage this range of relative structures? A comparison of the contexts and types of nouns that appear in these various clauses reveals that the degree of individuation of the head noun affects the strategy used to relativize it. The more individuated the noun, the greater the tendency to use the standard lilli!
clause with resumptive pronoun. Temporal nouns, lower on the scale of individuation, have greater tendency to be relativized without resumptive pronouns. The use of lilIiI as nominalizer or sentential complementizer often occurs when a highly individuated noun appears anywhere in the preceding clause, suggesting that high individuation affects not only the noun phrase but also the sentence as a whole.6
This chapter will examine these relative structures and other clauses involving lillil in the four dialect areas.
3.1 Relativization of Indefinite-Specific Nouns with filIiI
The grammars of Cowell 1964 (Syrian), Harrell 1962 (Moroccan), Mitchell 1956 (Egyptian) and Holes 1990 (Gulf) all state that, in the construction of relative clauses, definite head nouns require the relative pronoun lillil, whereas indefinite head nouns do not admit it. Examples of definite relative clauses from each 'dialect region:
�ese two pronouns are reflexes of fonnal Arabic relatives /mat and /man/.
Wright defines these relative pronouns as "either definite or indefinite"
(1898ii:319), but a better description according to the analysis proposed here would be "partly specified."
6Cf. Hopper and Thompson's (1980) theory of transitivity, which argues for a symbiotic relationship among various sentence constituents such that features of "high transitivity," e.g., nominal salience, agency, perfective verbal aspect, tend to co-occur in highly transitive clauses.
M l O �
& .J...U. �lSLa � I ..:.o.,...,J1 Hliit
lli rna kayn� ':mdu qima the-fish rei neg there-is at-it value The fish that has no valueE3 ��I eJl.l.. � I �I
ig-gU illi �alic ig-gidid
the-generation rei coming-up the new The new generation that's growing up
S5 �I.J,J l,)"'i IJ '�
lS�l;.� .J I '�I
il-lJayat il 'i�t[h]a bi-'amerka lJayat dalas w daynasti the-life rei lived-I-it in-America life of-Dallas and Dynasty The life [ lived in America [was} the life of "Dallas" and "Dynasty"
K I LA4,J ..r'L....i �i � I <:,jLA
",....�
silliti mii hagi lli 'abi 'asafir wiyyaha
group-my neg this rei I-want I-travel with-her
This is not my group [offriends} that [ want to travel with Examples of indefinite relative clauses include:
MIO � I .) I I
....'�J
&:!lS c.F'l.:.J,Jl.&.
'awd tani kayn rnza kaylb::lsha r-raz::ll
then second there-is turban indic-he-wears-it the-man Moreover, there is a turban that the man wears
EI ..:.04-6. � �
j� .�l.&.cayza gihaz yitsaggil 'aleh lJagat
wants-f machine be-recorded on-it things
She wants a machine that things can be recorded on S2 � J=...
.�I I",J� ..:.o� �
fi ban at biqiilu 'eh, mitil baCc;l[h]a
there-is girls they-say yeah, like other-of-it There are girls who say okay, that's alright K2 �y.A..H
La I,)" U � I
.�,y.A..HI,)" U �I
'aku nas yinjiin, 'aku nas rna yir<}iin
there-are people they-agree there-are people neg they-agree There are people who accept [this}, there are people who do not accept
3.1 Reiativization of Indefinite-Specific Nouns with fillil
93
Based on this description and these examples, then, the use of the definite relative /ilIi/ appears to be controlled by the definiteness of the head noun of the clause. However, data from all dialects contain examples in which an indefinite head noun is relativized using the definite /ilIi/. Such data constitute clear violations of the "rules" for relative clause formation. While it might be argued that such sentences represent performance errors on the part of the speaker, the repeated occurrence of this structure should not be dismissed without further investigation. The traditional view that a head noun "controls" the syntax of a modifying clause cannot account for a range of naturally occurring data in which the structure of the modifying clause is at odds with the definiteness of the head noun. Rather, both the marking of the head noun and the marking of the relative clause are controlled by the speaker, who can combine definite and indefinite markings to indicate a range of specificity.
All four dialects allow indefinite head nouns to be relativized with the definite relative marker /ilIi/ (or Syrian variant /yalIi/). A Moroccan informant supplied the following example (based on a sentence cited by Harrell 1962: 1 65):
M3 �4j.e � .}J I
�.,.."J.. � bgIt tumubii IIi t�m�i mnyan wanted-I car reI it-goes goodI want a car that will run good (elicited)
In the following Egyptian sentence, the indefinite noun /tamsiliyya/
serial is modified by a clause headed by the defmite relative /(i)lli/:
E4 J�
� ..,u l �.H� I � 1".:.ts: .}J 1 (:'/10;
�n�4 �
fi tamsiliyya ilIi Unu biygibfiha fi t-tilivizyon illi hiyya bit)nl l.tabakit ya <umda?there-is serial reI were-they indic-they-bring-it in the-TV reI she she-says was-necessary-she 0 mayor?
There's a serial that they used to show on 1V that says, Is it really necessary, Mayor?
To test for performance error, this sentence was repeated to two other Egyptian speakers, and both found it perfectly acceptable. Subsequent
observation of Egyptian speakers has confirmed that similar examples occur with some regularity.
The individuation hierarchy provides a principled explanation for this syntactic behavior. In the example given above, the speaker appears to have in mind a specific serial program. He introduces this program using the indefinite form because he assumes that it is unfamiliar to or unidentifiable by his interlocutors, but at the same time, the use of /illil implies a particular, and presumably identifiable, serial. The noun in this case is thus not entirely indefinite, but rather falls in the indefinite
specific range. Lacking a syntactic indefinite-specific marker, however, the speaker utilizes a combination of indefinite and definite markings across the clause to represent the specific identity of the serial.
Cowell notes that "some [Syrian] speakers occasionally use /yalli/
to introduce clauses that are attributive to an indefinite term" ( 1964:499).7 Cowell has no explanation for this infraction other than that "some speakers occasionally" do this ( 1 964:499); to leave it at that would suggest tha�as many native speakers of Arabic believe-colloquial has no rules. The answer here, as above, seems to lie in the indefmite-specific quality of the relativized nouns. Similar to the Egyptian example given above, the head noun in Cowell's data appears to fall in the indefinite
specific range; here, the speaker clearly has a specific entity in mind ( 1 964:499):
S
.�I �fi wal:lde yalli b;)tzakkar[h]a [rna ';)sm[h]a
[there-is one rei indic-I-remember-it in-it name-its]
There's one I remember that has her name in it
The inability of the definite/indefinite marking dichotomy to predict the occurrence/absence of the relative pronoun /illi/ shows that specificity is a feature that can extend beyond the noun to the noun phrase as a whole, affecting the interaction of nouns with modifiers. In the next example from Kuwait, the speaker uses the definite /illi/ to relativize the indefinite noun /bnayyal girl. The speaker combines