Throughout its long history, the project of a science of grammar has always been an inherently philosophical one, in which the study of grammar was taken to have special epistemological significance. I ask why 20th and 21st century inquiry into Universal Grammar (UG) has largely lost this dimension, a fact that I argue is partially responsible for the prevailing controversy around UG, relating to its formulation, scope, and biological basis. This paper argues for a re-conceptualization of this program along ways that answer these criticisms and are more in line with a pre-modern conception of a science of grammar, bringing out its philosophical significance. Central to the new conception is the role of grammar in giving human thought a species-specific and uniquely linguistic structural format, consistent with the recent claim that UG primarily constrains the Language of Thought, yet different from it in other ways, such as in stressing the role of grammar in establishing a system of deictic reference, which depends on the externalization of grammar in some physical medium. Although UG on the new conception is not the study of linguistic variation but the study of "sapiens"-specific mode of thought, it cannot be dissolved into the study of thought (or semantics) as such: for the mode of thought in question uniquely takes a grammatical format and grammatical meaning is unavailable either pre-linguistically or lexically.