Phi-Syntax: A Theory of Agreement (2003)

This thesis explores the formal mechanisms by which agreement relations are derived in syntax. The empirical focus is on some difficult cases of verb-NP agreement (Georgian, Algonquian and others) which present serious locality paradoxes when analyzed using standard mechanisms of syntactic agreement. The agreement systems in question have typically been taken to be symptomatic of a fundamental mismatch between syntax and morphology, however I treat them as evidence for a relativized formulation of conditions on Agree, under which they are in fact systematically isomorphic with well-defined derivational syntactic sequences. The claim is thus that surface complexity in these paradigms is best described in terms that are characteristically syntactic: long-distance dependencies, downwards locality/minimality, intervention effects and cyclicity (defined over syntactic operations). The analysis relies crucially on a fine-grained approach to formal phi (?), where ? encode the grammatical properties of nominals that typically enter into agreement: person, number, gender. Three aspects of ? feature theory are developed: (A) the representation of implicational relations between ? features; (B) feature-theoretic formulations of the syntactic operations that enter into the establishment of ? agreement relations; (C) the notion of ? feature (de)activation, and its interaction with syntactic projection, and cyclicity. The theory of agreement that emerges allows for a unification of complex agreement systems — a diverse and robust group — with their simpler, more familiar counterparts