Syntaxe quantitative et expérimentale : objets et méthodes

In this paper, we discuss some issues related to the kind of data typically used in syntactic research. We first propose a critical view of grammaticality judgments, traditionally employed in formal syntax, by highlighting their variability and their gradient dimension. We then present different methodological aspects of quantitative and experimental syntax, by insisting on the complementarity between corpus-based approaches and controlled experimental approaches. We argue in favour of an approach which, in addition to binary grammaticality contrasts, integrates preferential constraints to account for the gradient nature of linguistic data.