New hope for linguistic ersatzism?
Are all truths a priori entailed by the fundamental truths upon which everything else supervenes? If 'entailed' means 'strictly implied', this is trivially true. The more interesting question is: Are all truths deducible from the fundamental truths (deducible, say, in first-order logic) with the help of a priori principles?
If yes, then it seems that Lewis' 'primitive modality' argument against linguistic ersatzism (On the Plurality of Worlds, pp.150-157) fails. Recall: Lewis argues that if you take a very impoverished worldmaking language then even though it will be feasible to specify (syntactically) what it is for a set of sentences to be maximally consistent, it will be infeasible to specify exactly when such a set represents that, e.g., there are talking donkeys. Now if all truths are a priori deducible from fundamental truths, and -- as seems plausible -- fundamental truths are specifiable in a very impoverished language, then we can simply say that a maximal set of such sentences represents that p iff p is a priori deducible from it.
Unfortunately, I find the 'primitive modality' argument quite compelling. So, by modus tollens, I have to conclude that not all truths can be a priori deducible from fundamental truths. Does anyone know whether Lewis himself believes the deducibility claim he attributes to Jackson in 'Tharp's Third Theorem' (Analysis 62/2, 2002)?