Agreement in relative clauses and the theory of phrase structure : a study of Standard Arabic
This dissertation is primarily concerned with the structure of relative clauses in Standard Arabic (SA) within the framework of phrase structure proposed in Kayne (1994). According to Kayne, relative clauses are assigned the structure [D CP] where D is base-generated externally and takes CP as complement. This study emphasises that D in SA originates within the relative clause and moves with its NP complement to SpecCP. The moved DP enters into Spec-head agreement with the complementizer since both the head and Spec carry the same [c]-features, in addition to [+defj and, sometimes, [+Case]. The subject trace in main clauses, which I assume to be a null resumptive pronoun, is properly governed by the complementizer which agrees both with the antecedent and the verb. Object extraction, in some cases, requires an overt resumptive pronoun. In such cases, we adopt the analysis that treats resumptive pronouns as spell-out traces. We have proposed that in cases where a resumptive pronoun appears, the "head" moves to SpecCP but its [c]-features remain in the extraction site. When a gap appears, the "head" and its features move together to SpecCP. Subject relativization from ? an-embedded clauses is different from subject relativization from ? anna-embedded clauses. The former takes place from the postverbal position to avoid the Empty Category Principle (ECP). We adopt the Split-CP hypothesis (Rizzi 1997) and assume that subject extraction in ? annaembedded clauses takes place from SpecTop. The extraction site is obligatorily filled with a resumptive pronoun. The object, too, can be extracted from SpecTop or from its base-position. In both cases, an obligatory resumptive pronoun occupies the extraction site. We have shown that the intermediate CP is not a proper landing site for the extracted subject or object. The reason is that the head of the intermediate CP does not bear the features of the antecedent and therefore movement to the Spec of the intermediate CP is not legitimate. We have proposed that free relatives also involve movement to SpecCP. This proposal is based on the fact that 11adhii can be overt only if DP occupies its Spec position. We therefore have proposed that features of a null DP must occupy the Spec position in this type of relatives. Thus Spec-head agreement in these relatives is also realised. Reduced (participial) relatives are analysed as full relatives and therefore are assigned the same structure apart from the fact that they contain a functional head which I call partA° (Participle Affix). We have proposed that the morpheme preceding the participle is a reduced form of the complementizer lladhii. This assumption is supported by the fact that it can only be preceded by a DP whether overt or null. Here again we have Spec-head agreement as in full relative clauses.