| Summary: |
The order in which French clitic pronouns occur is highly idiosyncratic. Students of French have to learn to disrupt the normal Subject-Verb-Object pattern of declarative sentences, and to insert object pronouns directly before the verb. Moreover, the preverbal alignment of these clitics does not always reflect the left-to-right order of postverbal phrases to which the pronouns correspond. There are additionally certain puzzling incompatibilities among clitics. The purpose of this thesis is to offer an explanation for preverbal clitic order, and not to account for impermissible clitic sequences. The problems posed by French pronouns are approached within the framework of generative grammar, especially as exemplified in Kayne's (1975) French Syntax. Certain later developments in this model have been incorporated into the arguments presented here, while certain others have been rejected as being non-explanatory when applied to the question of clitic ordering. General proposals designed to regulate the order of application of transformational rules or to delimit their domain are examined to determine their relevance to French pronouns. Proposals specific to the grammar of French are also investigated. It will be shown that none of the explanations offered to date is entirely satisfactory: supposedly universal constraints on rules fail to align clitics correctly, and language-particular claims require convoluted or otherwise improbable sets of rules, make certain dubious assumptions, or simply do not make the right predictions. The claim made here is that important differences in the underlying structure of French pronouns motivate their preverbal order. Elaborate clitic 'templates' like the one advocated by Perlmutter (1971) can therefore be rejected in favour of a single surface filter. No new mechanisms are needed, as clitic order follows from the interaction of a simple clitic movement rule constrained by the very general A-over-A Principle and a filter limiting the surface position of It has been pointed out to me that there are two remaining problems in the analysis of clitic order presented in chapter seven claim that the A-over A Principle can successfully distinguish underlying Pro forms to which a subject clitic is attached from those first and second person Pro forms which I argue are generated without these subject clitics is not above suspicion. Whether the Pro mentioned in the structural description of the rale of Clitic Placement operates on a syntactic category Pro or on an NP marked with a +PRO feature, it is not clear that the structure [NP'-SCL] of page 366 is affected in the desired manner(i.e, NOT subject to Clitic Placement), because the higher NP containing NP' and SCL may well not itself be a Pro constituent. The second problem concerns clitic ordering in cases of non-reflexive idiomatic clitics combined with other clitics having a post-verbal source. My set of rules will not generate, for example, sequences like "1L m'en veut". Further refinements of the notion "idiomatic clitic"will be required to obviate this lacuna. |
