Proceedings of the Annual Meeting of the Cognitive Science Society

What determines whether a novel referent is judged as similar enough to extend an existing term, or different enough to coin a new one? We investigated how nameability and event structure influence verb extension. Participants learned novel words for unlexicalized actions varying in nameability, then saw variants differing in end state, process, or object and chose whether to extend the learned word or select an alternate. Nameabil- ity was the strongest predictor: more nameable actions saw more extension. For event structure, results replicated prior findings—end-state changes resisted extension most, followed by process changes, with object changes permitting the most extension. Extension was the dominant response overall, con- sistent with efficiency pressures favoring reuse of existing vo- cabulary. These findings suggest that lexical decisions reflect a tradeoff between the accessibility of existing representations and the degree of semantic difference introduced by variation.