I watched this lecture by Amie Thomasson about some topics in Metametaphysics. After sketching some of the methodological problems of contemporary metaphysics, she puts forward her own methodological view. She borrows the distinction of internal and external ontological questions from Carnap. Internal are those questions, that can be answered from within a certain vocabulary. So for instance in mathematics, the answer to the question “Are there numbers?” is trivial because mathematics just uses a vocabulary which contains numbers. What a philosopher can do here is just descriptive theorising. How are the concepts used?
If we ask something like “But do numbers really exist?” we pose external questions. These are for Carnap only pragmatic questions. Which vocabulary should we use for which purpose? This is a normative question for Thomasson. Her examples are mostly from practical philosophy, though. How should we use the concept of a human being or of live if we talk about abortion, for example. I think she is right in such cases. There we are talking about how we should use certain concepts.
But are there really no ontological questions that are not trivial and not pragmatic? When David Lewis is suggesting that for a semantics of modal logic we need to talk about possible worlds, you can see that as a suggestion to use a certain vocabulary. But he does not only do that. He also suggests that this will only work, if we take the ontological commitments to possible worlds seriously. If we need possible world talk for modal logic and if we quantify over possible worlds, we should believe that there are such possible worlds.
Of course, in the first step, he is only suggesting a specific language for a specific purpose. But then he argues, that we should have ontological commitments to things (i.e. possible worlds), that we did not have before. And here he is not describing the use of concepts, he is also not suggesting a special language, he is making claims about which things exist.