The Role of Polarity in Inferring Acceptance and Rejection in Dialogue

Abstract

We study the role that logical polarity plays in determining the rejection or acceptance function of an utterance in dialogue. We develop a model inspired by recent work on the semantics of negation and polarity particles and test it on annotated data from two spoken dialogue corpora: the Switchboard Corpus and the AMI Meeting Corpus. Our experiments show that taking into account the relative polarity of a proposal under discussion and of its response greatly helps to distinguish rejections from acceptances in both corpora.