This thesis argues that for a concept of analyticity--and an attendant analytic-synthetic distinction--to be workable, we must acknowledge its essential normative dimension. Indeed, I point out that contemporary debates surrounding so-called metaphysical and epistemic conceptions of analyticity are importantly misguided for failing to see the significance in this point: that analytic sentences express norms, not descriptions. Moreover, I argue that no sentence can--in one and the same use--function both normatively and descriptively. These thoughts are marshalled to show that the positions of Timothy Williamson and Paul Boghossian on analytic truth are jointly untenable, despite appearing to be opposite ends of a debate. In later chapters, I seek to show the positive contribution of these insights. First, I demonstrate how John McDowell's arguments against the Myth of the Given point us both toward seeing analytic truth as a precondition for empirical description and away from Robert Brandom's inferentialist understanding of conceptual content. And second, finally, I show that Mark Wilson's innovative work on conceptual change in various applied fields can complement the approach I endorse; i.e., that normativity and conceptual change interact rather than conflict.