Evaluating Unalternative Semantics on a large corpus of naturalistic speech

Abstract

While the experimental turn of the last decade has helped connect formal semantic/pragmatic theories to empirical reality, extending this empirical reach to naturally occurring data has lagged behind. One of the challenges is the lack of large corpora in which the relevant technical notions have been annotated. Nevertheless, some insight may be gained by relying on automatically computed approximations of these notions. We take such an approach to focus realization– the relation between focus, grammar and pitch accents, taking inspiration in particular from Unalternative Semantics.