The Empirical Structure of Pejorative Categories: Group Slurs, Trait Pejoratives, and Particularistic Insults

The philosophical literature has long debated whether trait-targeting pejoratives (e.g., whore, wino, bitch) and group-targeting slurs (e.g., nigger, faggot) form distinct categories, or whether the boundary between them reduces to extra-linguistic factors. We report five experimental studies comparing particularistic insults, trait/behaviour pejoratives, and group slurs (8 tokens per class). Studies 1--3 measure scalar dimensions (offensiveness, tabooness, dehumanisation); Studies 4--5 measure a categorical property: whether hearers take an utterance to target the named individual alone or also others beyond them. The three classes pattern as distinct on every measure (group slurs highest, particularistic insults lowest). At the token level, the categorical and scalar measures classify the lexicon differently, coming apart precisely at the items the trait/group debate treats as boundary cases. The pattern is not straightforwardly captured by current theoretical accounts. We propose two routes that can accommodate it: a pragmatic framework combining the entitativity of the targeted kind with the social-historical situation of its members as two partially independent dimensions, and a revised semantic-encoding mechanism on which the expressive content of a slur is a generic negative attitude whose target is resolved contextually, with entitativity biasing resolution toward the group reading. We argue that the pragmatic route is the more parsimonious.