Motion Event Typology Meets Computational Modelling

Speaking involves a process of selecting words and constructions which correspond to the thoughts that one wishes to express. Different languages place different constraints on this selection process, corresponding to their typological characteristics. In this paper we show how Swedish, French and Thai preferentially express motion events differently and suggest a theoretical framework with grounds in cognitive science to account for these differences. We proceed to describe an experimental computer program, Verbalizer, which implements aspects of our framework, showing how a single conceptual representation of a motion event can be realized in multiple ways, both within and between the languages. Our research indicates how typological and computational linguistics can benefit from one another.