This dissertation argues that contrast is crucial to verb phrase ellipsis (VPE): VPE must be contained in a phrase E that differs in meaning from an antecedent A. Previous research has required A to be a focus alternative to E (e.g. Rooth 1992b, Heim 1997, Fox 2000, Takahashi & Fox 2005). In the simplest case, A and E can have the same meaning, which might seem desirable from the perspective of identity. The thesis here, unsurprising if ellipsis has something in common with focus (Rooth 1992a), is that A must be a proper alternative to E (cf. Griffiths 2019). This contrast requirement comes to light from studying the interaction of VPE with triviality, symmetry, and competition from sluicing. The sameness that is inherent to triviality, created by symmetry, and in the vicinity in MaxElide effects reveals the explanatory role that contrast plays in ellipsis licensing.
After chapter one introduces ellipsis, identity and contrast, chapter two establishes the contrast requirement, starting from the observation that ellipsis is ungrammatical in tautologous conditionals; e.g. "If John is wrong, then he is *(wrong)." Further data involving discourse antecedents, negation, and intensionality favour a theory that attributes the problem to contrast failure in ellipsis licensing over one that attributes the problem to triviality (cf. Gajewski 2002, 2009). Chapter three adds VPE with symmetrical predicates to the literature on ellipsis mismatches; e.g. "John wanted to dance with Mary, but she didn’t want to (dance with him)." Symmetry creates semantic identity, overcoming non-identity in form, but can also lead to contrast failures. VPE with predicates like 'kiss' further shows that alternative-hood is enforced only in one direction between antecedent and ellipsis (Rooth 1992b, Fox 2000) rather than both (Griffiths 2019; cf. Merchant 2001). Chapter four critically surveys theories of MaxElide effects; e.g. "John ate something, but I don’t know what (*he did)" (Schuyler 2001). The effect is standardly explained in terms of competition, with VPE ungrammatical for losing to sluicing (Merchant 2008b). Competition theories crucially allow equality of A and E (esp. Takahashi & Fox 2005), placing them in conflict with this dissertation’s emphasis on contrast. The chapter addresses this point of tension by offering critique of competition theories, and evaluating the prospects for an alternative account in terms of contrast (Griffiths 2019). By way of conclusion, chapter five considers further respects in which focus and contrast do, and do not, interact with ellipsis. The discussion encompasses Winkler’s (2005 et seq.) typology of ellipsis in terms of contrast, reciprocal ellipsis mismatches with 'each other', noun phrase ellipsis, tense, 'why NOT', voice mismatches, question-answer congruence, and 'only'.
