CLITIC CONSTRUCTIONS

An analysis of French and Romance clitic constructions in simple clauses is proposed reconciling the two dominant approaches on this topic. This analysis treats them as involving both movement and base generation and, as part of a larger research program, assimilates the syntax of clitics to that of other functional heads. Accordingly, a clitic is analyzed as heading its own projection and as licensing in its specifier a particular property of a designated argument agreeing with it in the relevant features (person, number, gender Case...). It is further argued that clitics subdivide in two types. The first type (such as French en or le assimilates to such functional heads as [+wh] complementizers or [+negative] heads licensing certain operator-like properties (e.g. wh or negative quantifiers). We show that (some) Germanic Scrambling is fundamentally similar to these clitic constructions. This leads us to postulate that the operator like property these clitics license is specificity in DPs as has often been proposed but that reciprocally the specificity of some DPS must be licensed in similar "clitic-like" configurations. In essence, this comes down to assimilating pronouns to bare operators (like bare wh-phrases who, what, why), and specific DPs to non bare operators with operator determiners (like which book, what reason..). The second type of clitics such as French lui are not linked to specificity. We suggest that these clitics should be analyzed as pure agreement analogous to AGRO or AGRS, presumably responsible for Case assignment (here Dative, i.e AGRIO).