On the view that was standard up through the 1970's and early 1980's, one of Frege's central innovations was to put language and the theory of meaning, rather than epistemology, at the center of philosophy. It is an odd view, in light of Frege's own characterization of his central project as epistemological - as an attempt settle the question of whether our knowledge of arithmetic (as well as most of mathematics) as a priori or a posteriori, synthetic or analytic. Odder still, given that, on Frege's view, these categories are epistemological: the correct categorization of an item of knowledge is determined by the sort of justification that item requires. The focus of most of my work on Frege has been to make sense of Frege's writings in the light of his own epistemological characterization of his project.
In my most recent work on Frege I address the matter of how exactly we are to understand the work for which Frege is most renowned: his writings about language. This work, I argue, was meant as a contribution to the epistemological project. And to understand it in this way is to see that Frege does not view himself as providing a semantics, either for natural language or for his logically perfect language.
I have also begun a new, related, but non-historical project. The thrust of the project is to argue that our thinking about the semantics of natural language has gone astray because of our failure to appreciate how semantics needs to be constrained by the methodology of the special sciences. My contention is that many philosophers writing today are promulgating views that require the rejection of widely accepted and unobjectionable metholodologies of productive scientific inquiry. The science on which I particularly focus is the science of epidemiology and biostatistics. What I mean to show is that the key to rectifying what is going wrong is to adopt a new Frege-inspired approach to the semantics of natural language.