In this talk, which reports joint research with Esther Torrego (UMass/Boston), I argue that the properties of DP traditionally called "case" are in fact uninterpretable instances of tense features. In the first part, I argue at length that the property of DP traditionally called "nominative case" is in fact an uninterpretable instance of a tense feature ("uT"). This proposal, in the context of a general economy condition on the Agree relation, makes sense of the fact that the uninterpretable T feature of C that is deleted by T-to-C movement in English non-subject matrix wh-questions is apparently also deleted by nominative wh in subject matrix wh-questions (What did Mary buy? [T-to-C] vs. Who bought the book? [no T-to-C]). This proposal extends to that-trace effects, as well as to certain other effects attributed to the ECP in earlier work. This portion of the talk summarizes material available in greater detail as Pesetsky and Torrego 2000 (pdf).

The second part of the talk offers several reasons to think that accusative, prepositional and oblique cases may also be instances of uT on DP -- i.e. that all instances of case are instances of tense features. I suggest a variant of the "split VP" structure of Travis, Kratzer, Koizumi and others, in which a Tense (more specifically, Tense/Location) projection intervenes between v and VP, and suggest that it is this lower instance of T that deletes uT on an accusative DP in a manner consistent with our results concerning nominative case. The structure proposed, in which v takes a lower TP as its complement, makes the clausal semantics of complements to intensional predicates like want unsurprising (McCawley 1974; Larson et al. 1997; Endo et al. 1999), and leads us to expect a variety of interactions between tense, aspect, and object case that appear to be found. Certain restrictions on extraction from within the VP appear, on this view, to be effects involving the lower T analogous to the that-trace effect.