|Title:||Coordination in Optimality Theory|
|Abstract:||This dissertation provides an analysis of coordination within a restricted version of Optimality Theory (OT), operating exclusively with alignment and faithfulness constraints. It is shown that this version of OT, which does not assume the existence of syntactic structure, is capable of handling coordination phenomena that have been problematic for previous structureless accounts, while at the same time also avoids the problems encountered by structure-based theories trying to fit coordination into models developed to account for structures occurring elsewhere.
It is standardly assumed in OT that syntactic structure conforming to X-bar Theory axioms is imposed on all the candidates generated from the input by the structure building component of the grammar. I reject this view and contend that it is by eliminating from the grammar historically inherited axioms about X-bar Theory or even the central notion of phrases that we can unlock the real power of OT: the violable constraints that make up the evaluation system are solely responsible for the grammar and grammatical differences between languages.
I show how such an assumption, originally conceived in Newson (2000b), can predict basic word order patterns and can also account for topicalization as well as coordination in a wide variety of language types. As syntactic OT in general, and alignment OT in particular are relatively novel developments within linguistic research, the basic principles of the theory are still being formed. I show that the originally assumed family of predicate alignment constraints alone is not capable of accounting for a variety of data such as SOV language pattern as well as topicalization and focalization in a variety of languages, and introduce a second family of alignment constraints. Contrary to the predicate alignment constraints, which establish the position of an element with respect to its predicate (or, more generally, functor), first and last constraints sanction the placement of elements to the beginning and the end of the string, respectively. The interaction of these constraints with each another and with the functor alignment and the faithfulness constraints provides a powerful grammar.
The coordinative conjunction is argued to be a functor, taking the predicates of the propositions it conjoins as its arguments. As a functor, the functor constraints originally developed for the predicate-argument relationship are relevant for the conjunction-conjunct alignment relationship as well.
Central to the analysis of ellipsis is the claim that ellipsis is not only characterized by phonological absence, but by syntactic absence as well. Essentially, ellipsis is treated as a semantic phenomenon: it is up to the semantic component of the grammar to reconstruct elided material missing from all other levels of grammatical description. In the framework adopted semantic interpretation is based on the input of the optimality system. Thus the input is the foundation of ellipsis phenomena -- as much as it contains gaps for material that surface as elided. I suggest that semantic reconstruction of missing input material occurs via a higher-order unification process and that the sentence receives interpretation only if this unification process succeeds.