Who is afraid of Alfred Tarski?
Tarski proved the indefinability of truth for a first-order (FO) language that uses traditional logic in the same language. Accordingly, his T-schema cannot be generalized into a truth-definition. The reason turns out to be circularity. The quantifiers in an attempted definiens would depend on the quantifiers characterizing the definiendum, and cannot be freed from this dependence by means of traditional logic. This flaw is corrected in independence-friendly (IF) logic, which therefore admits a missing truth predicate e.g. in the form of the existence of Skolem functions. The validity of this definition is equivalent to the axiom of choice, which is seen to be a FO logical principle. Truth is thus definable for a language in the same language if it is flexible enough to express every kind of independence. Cases in point include IF FO languages and presumably our colloquial language.