Limiting gender

Abstract

Grammatical gender features, which are seen as having both a semantic and arbitrary form, have been argued to embody the interpretable-uninterpretable distinction: semantic gender is interpretable, and arbitrary gender is uninterpretable (Kramer 2014, 2015). Using French as a case study, Chapter 5 argues that all gender features (even those which have been seen as arbitrary) are necessarily interpretable at the LF interface. Even if a feature does not contribute compositionally to the meaning of a structure, it must be visible to provide the context for interpretation. This leads to an argument for the abandonment of the interpretable-uninterpretable distinction in the representation of features. Instead, the analysis contends that the mechanism of interpretation is responsible for differences in the semantic contribution of features: both heads and sub-structures can be taken as input to the interpretive mechanism. The interpretation of heads leads to compositional meaning, and the interpretation of sub-structures to non-compositional meaning. The system has the consequence of simplifying restrictions on gender specification such that they are solely linked to the availability of a semantic interpretation, rather than to a combination of phonological and semantic licensing conditions.