Video Recordings

It is a truism that context determines meaning, but problems arise in deciding what counts as a “meaning” and what counts as relevant context. In this talk, I will show how painstaking analysis of evidence from large corpora enables us to recognize contextual patterns. I argue that words in isolation do not have meaning, they have meaning potential. A meaning potential is not a checklist of possible meanings; it is a complex linguistic Gestalt. Different contexts activate different components of a word’s meaning potential, at varying levels of delicacy. In Corpus Pattern Analysis (CPA), normal usage is distinguished from “exploitations” of normal usage such as ellipsis, metonymy, simile, and dynamic metaphor. Observing this distinction, a checklist of normal contexts can be derived from corpus evidence for each predicator in a language, with the consequence that a distinctive meaning can be attached to each pattern. A pattern consists of a predicator, its syntagmatic structure, its arguments with semantic values, and (occasionally) subargumental cues.