This paper aims at specifying the complex links which two major and polemically related 18th-century linguistic theories James Harris' universal grammar in Hermes (1751) and John Horne Tooke's system of etymology in the Diversions of Purley (1786, 1804) bear to empiricism. It describes both the ideologicalethical determining factors of the theories and the epistemological consequences dependent upon their respective philosophical orientation (Harris using classical Greek philosophy against empiricism, Tooke criticizing Locke's semantics along Hobbesian lines). The effects within the linguistic theories (...) are examined through a comparison of the theories of determination which follow from divergent theses concerning abstraction. The analysis proposed in this paper exemplifies once more the historical question of the exact location of the compatibility/incompatibility between empiricist and non-empiricist linguistic theories. (shrink)