Constraints on Disjunctive Sentences

Abstract

Hurford's Constraint (HC) states that a disjunction A or B is infelicitous when its disjuncts are in an entailment relation. Singh (2006, 2008) argues that HC must be modified in two ways: (i) HC is to be checked incrementally at the basic meaning of each disjunct to the right before it can be strengthened by implicature, and (ii) HC requires inconsistency, not non-entailment, between disjuncts. These modifications are, however, empirical generalizations entirely drawn from linguistic data, and do not provide explanations why they should be so, as Singh (2006) himself admits. The aim of this paper is to provide explanations to Singh's generalizations. I argue that Singh's two generalizations can be explained under the assumption that a disjunction has the property of a correction construction. First, inconsistency between a corrective claim and its antecedent in a correction construction is almost a truism; otherwise, it is not a correction. Second, I propose that a corrective claim is an argument in that it is a reason advanced for the falsity of its antecedent, and show that only asserted, not implicated, meaning is qualified as an argument. It follows then that when inconsistency is checked between disjuncts, only the asserted meaning of the second disjunct counts, which corresponds to a corrective claim in a correction construction