The Voices in our Heads∗

• A great deal of work is devoted to grammatical voice and its syntactic/semantic effects (esp. valency changes, theta assignment, implicit argumenthood, etc.) – But how these voices are syntactically derived, and why they are so cross-linguistically widespread is less well-understood • The “core ” voices (active, middle, and passive) are, within a single clause, in complementary distribution – This strongly indicates a correlation; yet no unified syntactic analysis cur-rently exists, to our knowledge Claim: • The syntax of these voices can be unified under a single functional head, Voice0, which can be defined by only two features. • In this way, voices are not syntactic transformations, and there is no notion of ‘default ’ voice. Analysis: • Voice0 is responsible for any and all syntactic differences distinguishing the grammatical voices. – We accomplish this using only existing syntactic machinery: external ar-gument selection (Kratzer 1996) and predicate fronting (Collins 2005). Consequences: • Finally does away with the notion that non-active voices are transformations on actives – A clause’s voice is simply the result of feature-checking on Voice0 • Greatly constrains the range of derivational possibilities, easing the acquisi-tional burden • Establishes unaccusativity as a voice-related phenomenon ∗We would like to thank the audience of the UCLA Syntax/Semantics Seminar for their remarks, especially Anoop Mahajan for noting (encouragingly) that “nobody in our department really believes what you’re saying.