UCLA

This dissertation examines verbal number agreement with disjunctive subjects. Whereas disjunctions, like indefinites, are known to freely take existential scope above other sentential operators — even out of environments from which scope-taking is typically blocked — I show that scope readings are more limited in sentences where disjoined singular subjects control plural agreement. In particular, I argue that sg-sg disjunctions must take existential scope within a polar question or beneath a downward-entailing or quantificational operator. I call this pattern Ivlieva’s Generalization based on Ivlieva (2012, 2013). While Ivlieva’s findings conflict with other existing accounts of disjunctive agreement, I present data — primarily from English, but also from Spanish, German, and Romanian — that strengthen the case that plural agreement with disjunctive subjects is semantically restricted. The cross-linguistic convergence of evidence indicates that the restricted distribution of plural disjunctive agreement owes to more general properties of disjunction and agreement rather than language-internal factors. I show that disjoined singular subjects that control singular verbal agreement, as well as disjoined plural subjects that control plural verbal agreement, display none of the same semantic restrictions. That agreement morphology correlates with semantic environment may seem surprising under a feature-copying view of agreement, but I show that the pattern becomes more familiar once we compare the distribution of plural-agreeing disjunction to that of other semantically marked indefinites, such as negative polarity items and epistemic indefinites. Though the empirical signature of each class of words is slightly different, considering plural-agreeing disjunction within the larger typology of indefinites (e.g., Haspelmath 1997) makes its distribution less anomalous. Compared to these other marked indefinites, the difference with plural-agreeing disjunction is that its distributional restrictions correlate with a reflex external to the nominal (namely, the verbal agreement) rather than with a lexical alternation within the nominal itself. I also show that the semantic environments where plural-agreeing disjunction can take existential scope strongly resemble the environments that block cross-sentential anaphora to indefinites. In light of these cross-domain connections, I present a formal account using tools from Giannakidou (1998, 2011)’s analysis of NPIs and Kamp and Reyle (1993)’s Discourse Representation Theory (DRT). In my account, the morpheme that is responsible for plural agreement with disjunction also restricts how the disjunction can existentially quantify over the disjunctive alternatives. Finally, I reconcile Ivlieva’s Generalization with past studies reporting that verbal agreement with disjoined subjects is simply unpredictable, rather than sensitive to semantic environment, as I argue it is. These previous studies primarily examine sentences where disjunction appears in matrix clauses without any overt quantification, and tend not to control for pragmatic context. I show that controlling for context allows us to better predict when singular and plural disjunctive agreement will be possible. Further, I argue that in some contexts, covert modality can provide the licensing environment that plural disjunctive agreement requires.