Overhauling Collostructional Analysis: Towards More Descriptive Simplicity and More Explanatory Adequacy

In this paper, I make two sets of suggestions of how collostructional analysis can be updated. One set of suggestions involves simplifying the analysis for descriptive/exploratory purposes while at the same time enriching it with bootstrapped confidence intervals. The other set of suggestions involves the idea that we should move away from a single kind of association measure for theoretical/exploratory purposes and instead quantify collostructional attraction as a tuple of, minimally, three ideally orthogonal dimensions, namely frequency, association, and dispersion, because only this kind of analysis will be able to address all the dimensions that are relevant to cognitive/usage-based approaches to constructions. In addition, I end with a (renewed) plea to take the notion of construction more seriously: Rather than looking at associations of constructions to forms, which many studies have basically amounted to, I would like us to ‘go back to’ looking at associations of constructions to constructions, i.e. to take the meaning/functional pole of constructions more seriously again and include sense/function in all kinds of collostructional analyses more.