In his A Theory of Legal Argumentation, Robert Alexy lists four modes of justification for his rules of practical discourse. Of these, the only one that to him seems to have a true foundationalist capability is Apel’s transcendental-pragmatic mode of justification. But there is another strategy, which Alexy calls 'definitional', based on the concept of language game, that seems to have been employed by Ronald Dworkin, in his 'Objectivity and truth. You’d better believe it', in order to reject moral scepticism. In this paper I will argue that these two modes of justification can be traced back, from a logical and genealogical perspective, to one form of argument, the transcendental argument. But I will also argue that there is a peculiar circularity in the way this kind of argument develops, a circularity possibly ascribable to certain idealising presuppositions.