Title: Operational opacity at the clausal middlefield

Abstract: Early formulations of Phase Theory posit that each clause consists of two locality domains (phases): the complete clause (CP) and a clause internal domain located roughly around vP. Syntactic evidence for the phasehood of CP comes from two types of phenomena: footprints of successive cyclic movement and operational opacity. In contrast, evidence for a clause-internal phase has primarily come from successive cyclic movement (i.a. Legate 2003, Aldridge 2008, Bennett et al. 2012, van Urk 2015). Operational opacity is not observed at vP, a category notoriously transparent for syntactic relations. This contrast between CP and vP has led some researchers to conclude that only CP is a phase and that intermediate movement through the edge of vP must be explained differently. This talk presents the missing type of evidence, from operational opacity, for the clause-internal phase hypothesis. In particular, I argue that VoiceP in Zimbabwean Ndebele is opaque for A-movement and phi-agreement. A VoiceP-external probe can only access a VoiceP-internal goal if the goal moves to Spec,VoiceP, and not otherwise. The talk additionally motivates the assumption that successive cyclic movement, both A and A-bar, is feature-driven. This is based on the observation that all movement out of VoiceP in Ndebele is blocked when Voice is independently shown to lack a movement-triggering feature. I conclude with general thoughts on why operational opacity is much easier to detect at a CP boundary than at the clausal middlefield.