When is it legitimate to cancel a potential scalar implicature? The roles of the Question Under Discussion and optimal relevance

This paper explores the conditions for legitimate speaker cancellation of a potential scalar implicature, with the broader aim of gaining a better understanding of when and why a scalar inference is derived as part of the speaker’s meaning. It brings under scrutiny Mayol and Castroviejo’s (2013) proposal in terms of a Question Under Discussion Constraint on Speaker Cancellation and shows that despite its merits, it unduly restricts the possibilities available to speakers for felicitous cancellation of a potential scalar implicature. This is because the concept of relevance in terms of focus congruent questions that get partially/completely answered on which the authors rely fails to integrate contextual assumptions that will determine the kind of interpretation that the scalar containing utterance will receive. And also, because in their account the role that focal stress plays is too strongly associated with the triggering of a scalar inference. I show how a more broad-base cognitive pragmatics model like Relevance theory (Sperber and Wilson 1986/1995) avoids the problems that their approach faces by allowing context accessibility to play a key role in the derivation of cognitive effects, and giving prosodic prominence a facilitating role in processing without directly linking it to any specific type of effect. • Draws a neat distinction between Gricean and speaker cancellability. • Discusses predictions and limitations of Question Under Discussion theory. • Dissociates scalar implicature derivation from prosodic prominence. • Offers a Relevance theoretic account of speaker cancellability.