Kripke's claim that there is a normative relation between meaning and use -- that "meaning is normative" -- has come under considerable criticism in recent years. I defend the claim by arguing for a new interpretation of the "ought" relevant to meaning. Both critics and defenders of the normativity thesis have understood statements about how an expression ought to be used as either prescriptive (indicating that speakers have reason to use the expression in a certain way) or semantic (designating certain uses as correct in a sense explicable in terms of truth or other semantic notions)  I propose an alternative view of the "ought" as conveying the "primitively" normative attitudes speakers must adopt towards their uses if they are to use the expression with understanding.  This yields a conception of the normativity of meaning which resists recent lines of objection.