Revisiting the context-update potential of negation at the level of speaker utterance choice

Abstract

Theories of context-update potential have typically treated negative utterances as simply removing worlds from the context where the corresponding positive-polarity sentence is true, predicting no differences between the update potential of truth-conditionally equivalent positive- and negative-polarity sentences. However, any broadly Gricean pragmatics predicts, via the maxim of Manner, that marked forms, like negative sentences, generate inferences not generated by corresponding less-marked forms, like positive sentences. This paper revisits the question of whether negative polarity differs from positive polarity in its context-update potential, and if so, what accounts for such an asymmetry. We provide empirical and modeling evidence showing that negative-polarity sentences have context-update profiles distinct from positive-polarity sentences, which we argue is motivated by the greater pragmatic reasoning negation elicits about speaker’s utterance choice.