<p>This article develops the inferentialist idea of discursive practice, construed as “the game of giving and asking for reasons”, with a focus on the relationships between meaning, normativity and knowing, particularly in empirical statements and scientific definitions. The principal claim – following from these considerations – is that, in the inferentialist backdrop, the traditional division between perceptual and discursive knowledge is no longer in effect since all knowledge is discursive in the last resort. This, in turn, involves three indispensable stances, i.e. reliabilism, revisionism and fallibilism, all being characteristic of the normativity of discourse. In the following sections, we first analyze the normative nature of meaning and inference; next, we scrutinize the concept of assertion, commitment, entitlement and scorekeeping; finally, we intertwine normativity with reliabilism, revisionism and fallibilism, taking as an exemplification the case of the definitions of scientific concepts. Particularly, we emphasize the historical changeability of the definitions of acid indicating their context of a discursive and normative practice.</p>