The Openness of Natural Languages

Paul M. Postal, New York University e-mail: [email protected] Preface It might seem plausible to the non-specialist who thinks about natural language (NL) that a given NL, NLx, permits one to report linguistic performances, both performances of NLx elements and those of NLs distinct from NLx. By ‘reporting linguistic performances’ I refer to nothing more arcane than forming statements like ‘Amanda just shouted ‘where’s my baby?’’ It might also seem to a non-specialist that NLx permits one to do descriptive linguistics, not only the descriptive linguistics of NLx, but that of other distinct NLs. By ‘doing descriptive linguistics’ I mean nothing more exotic than forming sentences like ‘The German word for ‘air force’ is ‘Luftwaffe’’. But while these non-specialist assumptions might seem not only plausible but self-evidently true, modern linguistics in its dominant instantiation called generative grammar, in fact denies both these claims. Of course, it does this only implicitly and most advocates of generative grammar may be unaware that its doctrines taken literally preclude what any non-specialist would assume possible. Readers not easily accepting this conclusion will find their skepticism addressed in what follows, for a major goal of this study is to justify in detail the claim that generative grammar has the evidently intolerable implications just mentioned.