Is construction grammar cognitive?

Abstract

This paper examines the view that usage-based construction grammar is a cognitive theory of language. In the usage-based (or cognitive) strand of constructional work, constructions are typically theorised as mental  representations. A culmination of this view was Adele Goldberg’s (2006) definition of construction, which states two criteria for constructionhood: formal/functional idiosyncrasy and sufficient frequency, which is presumed to lead to a pattern being represented mentally, even redundantly. In this paper, this view is examined against the backdrop of the distinction between mental and social levels of analysis. The paper argues against treating constructions as mental representations by definition, and it also advocates caution in using the notion of construal in characterising constructional meanings. Constructions are argued to be social conventions that function as intersubjective cues for meaning. Specific instances of constructions (i.e. constructs) are produced with the aid of mental representations, but constructions are not necessarily coextensive with these representations. Usage-based construction grammar is a cognitive theory, but only in the sense of constraining individual-level phenomena.