I thought... that the fact that in characterizing truth for a language it is necessary to put words into relations with objects was enough to give some grip for the idea of correspondence; but this now seems to me a mistake. The mistake is in a way only a misnomer, but terminological infelicities have a way of breeding conceptual confusion, and so it is here. Correspondence theories have always been conceived as providing an explanation or analysis of truth, and this, a Tarski-style theory of truth, certainly does not do.