In this thesis I adopt an antirealist view of language understanding. According to that view we have understanding of a language only when it can be shown to others in common linguistic practice. I examine claims for the kind of knowledge involved in that understanding; claims about what we possess, how this is revealed, and how we may investigate it. By accepting that language use is a cognitive skill, I have to reject accounts of knowledge in purely linguistic, behavioural or social terms. The best way to examine cognitivist claims is by developing a cognitivist theory for the empirical investigation of the mind. I then show that these epistemic considerations bear on the epistemological claims of the antirealist. Further, I show that this way of explaining is quite compatible with the antirealist scruples about public display of our knowledge. It is claimed that empirical cognitive theories are not known to speakers who recognise and reveal their understanding, but explain how they do this. It is then shown why this claim is neither irrelevant to the philosopher's inquiry into meaning, nor posterior to it, nor leads to the replacement of it. A partnership of constructivism and cognitivism is necessary to reflect and explain the epistemic limits to the semantical content of human languages.