A probabilistic, question-based approach to additivity

Many analyses of additive particles such as too, also, and either have claimed that any such particle requires a contextually salient antecedent sentence that is a focus alternative of its prejacent or is a partial answer to the Question Under Discussion addressed by the prejacent. There is, however, a previously unstudied use—the argument-building use—which occurs without such an antecedent. This paper proposes an analysis of too that accounts for the argument-building use and unifies it with the canonical additive use. The central claim is that too is felicitous if and only if its antecedent can be taken to answer a relevant question such that the conjunction of too’s antecedent and prejacent evidences some answer to that question more strongly than the antecedent does alone. This analysis relies on the notion of a “resolution” from Inquisitive Semantics and the treatment of context update as Bayesian inference, as is done in the Rational Speech Act framework. The treatment of too developed here provides the basis for a new approach to additivity that can be extended to other additive expressions.