Verb Phrase Ellipsis and Pseudogapping are (really) not the same

Verb Phrase (VP) Ellipsis and Pseudogapping have at times been treated as variants of the same process (Kuno 1981, Jayaseelan 1990, 2001, Lasnik 1995, 1999, Johnson 2001). In both cases something verbal is elided. In VP Ellipsis the verb and all of its arguments are elided, while in Pseudogapping, only the verb (and a subset of its arguments) is elided stranding one of its arguments; in both cases the ellipsis leaves behind an auxiliary. One potential difference between the two constructions has recently been discussed by Merchant (2008) who notes that while sentences with VP Ellipsis can at least tolerate voice mismatches between the antecedent and ellipsis clauses, such mismatches seem categorically unacceptable in comparable Pseudogapped sentences. For Merchant, the purported asymmetry between VP Ellipsis and Pseudogapping was explained by a difference in the height of ellipsis: the elided constituent includes the voice feature on v in Pseudogapping, but not in VP Ellipsis. As a result, voice is subject to syntactic identity for Pseudogapping but not VP Ellipsis. Empirically, then sentences with Pseudogapping should be even more sensitive to voice mismatches than VP Ellipsis. An orthogonal dimension in which VPE and Pseudogapping differ is the dimension of anaphoricity. Kehler (2000) argued that VPE–as opposed to Gapping–is ‘anaphoric’ because it can take place across a discourse and it allows cataphoric reference. By those criteria, Pseudogapping is not anaphoric. This provides an additional difference between VP Ellipsis and Pseudogapping that could be (part of) the source of purported differences between the two ellipsis types. In collaboration with Christina Kim, we tested these claims in two magnitude estimation experiments. The first experiment tested Merchant’s claim that VPE tolerated voice mismatch better than Pseudogapping. The second experiment manipulated the discourse coherence relations between the two clauses, which is hypothesized to interact with anaphoricity such that anaphoric relations (VP Ellipsis) should be more affected than non-anaphoric relations (Pseudogapping). In this talk I will discuss the results of these experiments, with an eye towards relating them to the on-going discussion of verb phrase syntax.