The equivocation of form and notation in generative grammar

Generative grammar is a misguided approach which makes two interrelated errors. Firstly, the generativists seem to believe that you can do science by assumption or postulation alone; the assumptions are never tested, because that is not the point. Secondly, the generativists seem to believe that notations are forms. They ‘formalize’ their assumptions in a notation, in the belief that by doing so they are producing formal theories which are scientific and explanatory. In fact, all they are doing is expressing the figments of their imagination in their own ad hoc notation, thus engaging in a process which is pseudo-scientific and trivial. The above critique is illustrated by the passive in English. It is shown that various problems raised by a voice analysis of the passive – deriving passives from an underlying active – are explained by analysing the passive as an aspect. Both these descriptive accounts of the passive are shown to be formal and explanatory, with notation and generative style assumptions playing no role in them whatsoever. Three generative accounts of the passive are then presented, all of which formalise the voice analysis, each with its own idiosyncratic assumptions, each with its own ad hoc notation, a notation consisting of symbols borrowed from mathematics, computer science, and logic, combined with a mixture of abbreviations, mnemonics, and drawings. It is argued that the generative accounts are notational variants of the voice analysis, an analysis which has been shown to be incorrect. It is further argued that the way to correct the voice analysis is not to construct ‘formal models’, but to seek a solution within descriptive linguistics, of which the aspect analysis is an example. The paper concludes with a brief mention of ‘the method of lexical exceptions’ as an example of a reinvigorated data oriented, descriptive and Saussurean linguistics.