Séminaire de Recherche en Linguistique

Throughout most of its history, what has distinguished generative syntax from virtually all other approaches to grammar is the hypothesis of the autonomy of syntax (AS), namely the idea that the syntactic rules and principles of a language make no reference to meaning, discourse, or language use. However, AS has become progressively weakened in the past two decades, as is illustrated by the theta-criterion, the idea that c-selection can be derived from s-selection, UTAH and the Hale-Keyser approach, the cartography program, the wh-criterion and the Neg-criterion, the reintroduction of lexical decomposition, and much more. The purpose of this talk is to reassert the ‘classic’ view of AS, which was predominant among generative syntacticians until the 1980s.

I begin with two examples from English: the modal auxiliaries and derived nominalizations. There are profound formal generalizations involving both, which were uncovered many decades ago. But much current work, in attempting to represent structurally subtle semantic generalizations regarding these phenomena, has all but abandoned any attempt to account for their well-known formal properties. I then turn to negation. For any language, the default assumption is now that where we have semantic negation, we have a Neg Phrase projection. I argue that this projection, whose primary motivation is semantic, obscures the formal similarities that we find (differently for different languages) between negative elements and other elements in the particular language.

I conclude with two general remarks. First, I argue that current practice leads to redundancy, by representing in the syntax many semantic generalizations that need to be represented in conceptual structure anyway (the Cinque hierarchy is an example). Second, I note that the arguments for the poverty of the stimulus leading to an innate UG are based almost entirely on the correctness of AS. The rebuttal from cognitive linguists, AI people, et al. is ‘The arguments for UG fall through because Chomsky and his supporters do not realize how isomorphic syntax is to semantics’. Thus the semanticization of syntax poses a potential threat to the entire ‘Chomskyan’ program.