Typicality in referent choice: what the description leaves open

Abstract

A description built on a superordinate category term (an animal perched on a branch) does not pick out its members evenly: some come to mind more readily than others. This paper separates three conditions under which such a description meets two candidate referents differing in typicality, according to whether the full description is satisfied by both, by the atypical candidate only, or by the typical candidate only. These conditions support different interpretations of the same choice response and therefore require separate analysis. Forced-choice referent selection is the appropriate primary instrument for comparing them, but choice alone cannot separate a candidate rejected as false from one rejected as a worse fit, so judgment data remain necessary. Picture-based versions add a further confound: the restricting detail must be visually recoverable before it can override anything. An existing image-based study is described as a worked instance. No new results are reported.