Association rules in usage-based linguistics

Lecture by Kim Ebensgaard Jensen.

Within cognitive linguistics and other functionalist approaches to grammar (and other aspects of language), there is an increased interest in usage-based theorizing and usage-based descriptions. Given that usage is extremely complex and multifaceted, researchers within functional cognitive linguistics and, specifically, construction grammar, have started advocating for the use of methods that allow analysts to address grammatical (and other linguistic) phenomena as multidimensional entities. That is, there is a need for multivariate methods. Now, a number of multivariate methods have already seen use such as generalized linear mixed model tree analysis, behavioral profiles, and hierarchical configural frequency analysis, to mention a few. Very recently, the data-mining technique Association Rules has been added to the inventory of multivariate methods applied within usage-based analysis and description – mainly under the rubric of cognitive linguistics where it has been applied in the study of metaphor variation and grammatical constructions.

This talk addresses the use of Association Rules in usage-based linguistic research and introduces how it can be applied in the study of grammatical (and other linguistic) phenomena as multidimensional entities, offering more complex and integrated pictures, or profiles, of the phenomena in question. While this talk focuses on language, the method originated as a way to identify purchasing patterns in point-of-sale analysis and thus has a broad applicability range and might be of interest in the Digital Humanities more broadly.