Towards computational detection of metaphoric change in language evolution via word embeddings

Abstract

Metaphoric meaning extension is a central process in language evolution and historical language change. Here, we propose a computational method to further investigate the processes of metaphoric extension central to language evolution. Namely, we use word embeddings to computationally detect metaphoric change. We illustrate this method using corpus data from a less-resourced language, Polish, and present a case study of metaphoric change in the domain of technology-related expressions.