It seems to be clear what the truth-bearers are in Tarski's semantic definition of truth given in (1933).¹ They are sentences. The predicate "is true' is affirmed in the metalanguage of the sentences in the object language, i.e. the language in which we talk about things. Then the first question is: do the metalinguistic expressions which tell us about sentences univocally define the notion of a sentence? The second question is: why should sentences and not, for example, judgements, play the role of truth-bearers?