In a comment on my post Functionalism and Meaning, I worked on rebutting an argument to the effect that in semantics the only important theoretical terms are those figuring in T-sentences and not those figuring in derivations of such sentences. But consider the following.
For the moment, assume something like Hempel's view of nomological explanation, in which there are laws and bridge principles in the explanans and a statement describing some phenomenon in the explanandum. Now, on the analogy we're setting up, T-sentences are like the explanandum, the axioms of the theory are like the governing laws, while the rules of derivation (ensuring that the derivation is canonical) are like the bridge principles. One of Hempel's points about bridge principles is that they link the theoretical terms with terms designating observables. Can we view the canonical derivation procedure as like that, and also view the axioms as theoretical hypotheses? Doing so turns notions like "reference" and "denotation" into theoretical posits, but preserves the empirical nature of the t-sentences.
Many well know the problems with Hempel's view of theory confirmation, but can all of them be applied to the view of our "semantic" theory case? I wanted to get this up as I've been thinking of it lately. More will follow.
