Binding and control have long been at the heart of syntactic theory. However, it has long been known that simple syntactic approaches fall short of capturing the full range of relevant phenomena. This talk will consider two problem cases found in English. The first is binding within locative prepositional phrases, as in (1), where both a reflexive and personal pronoun can covary with the sentence subject. The second is so-called control shift, illustrated in (2a), where the implicit subject of passive complement clause takes a different controller than the one that is usually required (cf. 2b).
 Michelei set a glass next to heri / herselfi.
a. Michelei persuaded her mother [PROi to be allowed to stay up late].
b. Michelei persuaded Bobk [PROk,*i to stay up late].
I will make the case that, far from falling at the fringes of linguistic theory, these phenomena are central to understanding the nature of binding and control. Based on experimental evidence, I argue that both PP-internal binding and control shift favor theories that appeal to thematic roles. More specifically, I suggest that English pronoun choice depends on the identification of distinct thematic roles within a single event description, while control relations depend on the matching of thematic roles in matrix and embedded event descriptions. Insofar as thematic roles are part of the interface between syntax and semantics, binding and control must be understood at this interface, too.