Tarskian and Kripkean truth (Q676185)

In Tarski's hierarchy of languages, a successor language contains a (total) truth predicate for its predecessor language. The author outlines an expansion of Tarski's hierarchy into the transfinite, and compares this with Kripke's unramified language containing only one, essentially partial, truth predicate. In the expansion procedure, the author considers questions of constructibility and decidability -- which easily are missing with the abuse of language for the set of its formulas. In explaining the meaning of the truth predicate \(T_\alpha\) in the language \(L_{\alpha+1}\), the author notices that if there is a symbol \(T_\alpha\) for each ordinal \(\alpha\), these symbols cannot be identified with natural numbers. But to give up an identification of expressions with natural numbers would make it senseless to ask whether a set of formulas or a whole language is, or is not, recursive because the notion of recursiveness is applicable only to sets of natural numbers. The author proposes to use codes, i.e., natural numbers, as indices of the truth predicates instead of ordinals. Furthermore, he puts decidability conditions on the indices. It should be decidable whether a given number is an index at all and whether an index is higher than another (and, as an additional condition, that it is decidable from the index notation whether an index refers to an ordinal having a predecessor or to a limit ordinal). This reflects the idea that the hierarchy of languages be somehow constructible. It is shown that the transfinite hierarchy can be embedded into Kripke's minimal fixed point model, and several interesting comparative results on the expressive power of the two approaches to truth are given.