Abstract

We reconstruct “the classical transformational theory” of Chomsky, and fit it into the structuralist theory of science. We describe both the formal and the empirical features of this classical account, so that one basic hypothesis of this theory – where central notions are used – can be formulated, and in which Chomsky’s “classical” distinction between surface structure and deep structure is clarified. In the empirical claim of this theory are words, sentences and high-structured entities in an inseparable way intertwined. We claim that the formal structure of a natural language is not approximately the same as that of an empirical theory in general. We clarify two special points which affect the structure of the notion of an empirical theory, namely: the delineation of intended applications and the fit between data and models. We hold that the concept of the empirical claim for a linguistic theory should be generalized in comparison with the “standard” structuralist approach.