Berkeley Papers in Formal Linguistics

This paper is an opinionated survey of issues and perspectives in current models of Agree, understood as a single abstract grammatical operation common to all syntactic long-distance dependencies. I begin with a brief introduction to Chomsky's 2000, 2001 foundational work on Agree. I then review three strands of literature that have in notable ways chipped away at the conceptual foundations of that work in the course of improving the cross-linguistic empirical adequacy of the theory. These center on valuation and relativized probing, in section 3; defaults and failure to value, in section 4; and the question of whether goals must be made “active” by uninterpretable features, in section 5. In section 6, I review an ongoing debate about the directionality of Agree in light of the issues raised for uninterpretable features in sections 3-5. The paper concludes with a presentation of what I see as a way forward for the theory of Agree: the interaction/satisfaction theory, which provides a new conceptual grounding for Agree that in various respects makes sense of the empirical landscape uncovered by the past two decades of intensive research on this topic.