A critique of the model-building approach

Abstract

This paper puts forward the view that generative grammar makes two interrelated errors. One is its endemic practice of making assumptions systematically; the other is the way in which in “formalizing” those assumptions it confuses notation with form. Schoorlemmer's 1995 book on the Russian participial passive is taken as a case in point. S gives an account of the participial passive in Russian which sees it as having perfect-like properties and hence being compatible only with perfective verbs, not with imperfective verbs. Descriptively speaking this is an analysis which I welcome. The major problem is that S “formalizes” her analysis, in which she retains the derivation of passives from actives (the voice analysis) and a rule of NP-movement, within a generative framework. But an analysis of the passive like S's precisely shows that the voice analysis is incorrect, and is to be replaced by the aspectual analysis. Her retention and “formalization” of the incorrect voice analysis exposes the twin errors of the generative method, viz. that of confusing notation with form, and that of systematic ad hoc postulation. Some recent criticisms of the model-building approach in theoretical physics are adumbrated which corroborate our criticism of the model-building approach in theoretical linguistics.