Gradience is the idea that grammaticality is more than a binary notion, whereby a string simply belongs to a language or not. Indeed, when applied to the study of natural language, grammaticality may be presented as a gradient notion, whereby even an ill-formed utterance still "more or less" belongs to the language. For example, the utterance "Me are the chair of my department" is not grammatical, but can still convey useful information with its partial grammaticality.

What I am interested in is to come up with a formal model where accounting for such a degree of grammaticality of an utterance is made possible.

In a first part of this talk I will present an overview of different existing models and I'll discuss their respective suitability for modelling gradience. Then in a second part I would also like to present and discuss a couple of ideas that I have in order to account for gradience, and also where I stand in the process of specifying such a model in the context of a computational framework inspired from Construction Grammar and Property Grammars.

Please send comments or queries about this web site to jpprost@ics.mq.edu.au