Binding Phenomena Within A Reductionist Theory of Grammatical Dependencies

Abstract

This thesis investigates the implications of binding phenomena for the development of a reductionist theory of grammatical dependencies. The starting point is the analysis of binding and control in Hornstein (2001, 2009). A number of revisions are made to this framework in order to develop a simpler and empirically more successful account of binding phenomena.

The major development is the rejection of economy-based accounts of Condition B effects. It is argued that Condition B effects derive directly from an anti-locality constraint on A-movement. Competition between different dependency types is crucial to the analysis, but is formulated in terms of a heavily revised version of Reinhart's (2006) "No Sneaking" principle, rather than in terms of a simple economy preference for local over non-local dependencies. In contrast to Reinhart's No Sneaking, the condition presented here ("Keeping Up Appearances") has a phonologically rather than semantically specified comparison set.

A key claim of the thesis is that the morphology of pronouns and reflexives is of little direct grammatical import. It is argued that much of the complexity of the contemporary binding literature derives from the attempt to capture the distribution of pronouns and reflexives in largely, or purely, syntactic and semantic terms. The analysis presented in this dissertation assigns a larger role to language-specific "spellout" rules, and to general pragmatic/interpretative principles governing the choice between competing morphemes. Thus, a core assumption of binding theory from LGB onwards is rejected: there is no syntactic theory which accounts for the distribution of pronouns and reflexives. Rather, there is a core theory of grammatical dependencies which must be conjoined with with phonological, morphological and pragmatic principles to yield the distributional facts in any given language.

In this respect, the approach of the thesis is strictly non-lexicalist: there are no special lexical items which trigger certain kinds of grammatical dependency. All non-strictly-local grammatical dependencies are formed via A- or A-chains, and copies in these chains are pronounced according to a mix of universal principles and language-specific rules. The broader goal of the thesis is to further the prospects for a "reductionist" approach to grammatical dependencies along these lines.

The most detailed empirical component of the thesis is an investigation of the problem posed by binding out of prepositional phrases. Even in a framework incorporating sideward movement, the apparent lack of c-command in this configuration poses a problem. Chapter 3 attempts to revive a variant of the traditional "reanalysis" account of binding out of PP. This segues into an investigation of certain properties of pseudopassivization and preposition stranding.

The analyses in this thesis are stated within an informal syntactic framework. However, in order to investigate the precise implications of a particular economy condition, Merge over Move, a partial formalization of this framework is developed in chapter 4. This permits the economy condition to be stated precisely, and in a manner which does not have adverse implications for computational complexity.