Abstract

This article explores the basic assumptions of Generative Lexicon Theory (GL) and the implications for the general theory of the linguistic sign that arise from the generative mechanisms “selective binding,” “co-composition,” and “type coercion.” The article focuses on the assumption underlying GL that interpretation and polysemy are part of lexical structure. It is shown that encoded lexical meaning and inferred non-lexical knowledge cannot be clearly distinguished in GL. In order to be consistent, GL must also be supplemented by a theory of “normal language use” and be able to account for semantic underspecification in a semiotically coherent way.