Rational Speech Acts meet exhaustification Modeling free choice disjunction and related inferences

Abstract

We derive inferences of free choice disjunctions by combining grammatical models based on covert exhaustification (Fox 2007) with Gricean reasoning as implemented in the game-theoretic Rational Speech Acts framework (Frank & Goodman 2012). Our approach synthesizes insights from both frameworks and overcomes their respective limitations. The account leverages semantic uncertainty and ambiguity introduced by covert exhaustivity operators, enabling speakers to prevent unwanted inferences about denied permissions. Pragmatic reasoning then helps listeners navigate these ambiguities to infer that each disjunct is permissible without the other. This unified model explains the stability of free choice inferences, accounts for the defeasibility of associated inferences about whether both disjuncts together are permitted or forbidden, avoids overgeneration problems that arise in grammatical accounts, and explains why even fully informed Gricean speakers can prefer uttering disjunctions over more informative disjuncts.

EARLY ACCESS