The traditional point of view on analyticity implies that truth in virtue only of meaning entails a priori acceptability and vice versa. The argument for this claim is based on the idea that meaning as it concerns truth and meaning as it concerns competence are one and the same thing. In this paper I argue that the extensions of these notions do not coincide. I hold that truth in virtue of meaning— truth for semantic reasons—doesn't imply a priori acceptability, and that a priori reflection based only on knowledge of meaning—in the sense of competence—doesn't necessitate true conclusions.The main consequence of this view concerns conceptual analysis, as it presupposes we have a privileged—incorrigible in the face of empirical evidence—access to non-trivial truths about the world on the basis of mere a priori reflection founded on meaning. If, as I argue, such access is not incorrigible the project of conceptual analysis loses its special epistemological status.