Normative generics: Against semantic polysemy

Generic sentences are sometimes characterized as normative or descriptive. Descriptive generics make generalized claims about things: dogs bark, birds fly, doughnuts have holes. Normative generics do something more complicated; they seem to communicate how things should be: boys don't cry, children are seen and not heard, friends don't let friends drive drunk. The latter set of sentences express something like the speaker's endorsement that the predicated terms match up with the kind terms. Sarah-Jane Leslie posits a semantic view of normative generics on which the subject term is polysemous between a normative and a descriptive reading. I argue that this cannot be right, and show how a Gricean implicature view can accommodate everyday normative generics in a way that Leslie's polysemous view cannot. An upshot of my argument is skepticism about drawing semantic conclusions from dual character concepts.