Inferences of "will"
Cross-linguistically future tense markers can additionally be used to mark a present inference (Aikhenvald 2004). This interpretation has commonly been analyzed as an epistemic modal (Condoravdi 2002, 2003, Werner 2002, 2006, Kush 2011, Rivero 2014, a.o.). This dissertation provides an in-depth look at English “will” when it receives a present-inference interpretation. It compares this use of “will” with epistemic “must” to show that the present-inference use of future markers cannot simply be analyzed as an epistemic modal.
The dissertation puts forth two main empirical claims about how “will” differs from epistemic “must”. The first builds on the observation that epistemic modals require that the speaker have inferred the truth of the prejacent, rather than concluding it through direct perception (Karttunen 1972, von Fintel and Gillies 2010, Mathewson 2015, a.o.). Just as is observed for “must”, “will” also requires an inference, but it more constrained than “must” in the type of inferences it's compatible with. This is characterized as an incompatibility between “will” and abductive inferences. The comparison between “will” and “must” suggests that epistemic modals carry requirements about the inference they are compatible with, not the evidence. The findings for PT “will” are also replicated for the present-inference use of future tense markers cross-linguistically.
The second claim is about how epistemic modals interact with negation. Epistemic “must” is varied in how it is interpreted relative to negation. For example, it is interpreted above clause-mate negation, but below superordinate negation. In contrast, “will” is always interpreted above negation, regardless of the syntactic configuration it occurs in. These findings are relevant for two reasons: First, “will” seems to constitute a new class of polarity sensitive predicates that is distinct from those discussed previously in the literature. Specifically, I analyze it as a "strict" Neg-Raising Predicate (NRP). This provides new insight on the semantic status of the Excluded Middle inference which has been a topic of debate for traditional NRPs (Batsch 1973, Gajewski 2005, 2007, Romoli 2002, 2003, and Kriz 2015). Second, it adds to a typology of polarity sensitive modals (Iatridou and Zeijlstra 2013, Homer 2015).