Abstract
The paper studies the degree of grammaticalization of the structures gotta, gonna, wanna and better. The study presumes that the semantics of these structures – more precisely their modal polyfunctionality (i.e. the ability to express deontic and epistemic meaning at the same time) – has an impact on their morphosyntactic properties. Using corpora (predominantly the British National Corpus and the Corpus of Contemporary American English) and web forums, the paper studies in detail the level of independence of gotta, gonna, wanna and better from their respective auxiliaries (have and be) and the development of the operator properties of these structures typical for central modals (i.e. inversion in questions, compatibility with clausal negation and occurrence in elliptical contexts). It demonstrates that gonna and gotta are partially grammaticalized, especially with respect to the independence of their auxiliaries, but they do not syntactically behave as modals. The verb wanna behaves as a modal morphologically but not syntactically. On the other hand, better is grammaticalized to a high degree, and it does demonstrate both the morphology and syntax of central modal verbs.