Abstract
This squib discusses empirical challenges incurred by assuming cognitive reality as a defining feature of constructions and the constructional network, as done in most usage-based, cognitive construction grammar approaches. Specifically, it zooms in on the methodological challenges in identifying cognitively plausible constructions in historical data, in particular when taking a highly exploratory, bottom-up approach with very little pre-selection or pre-analysis. I illustrate this issue with the example of a current project on PPs in the history of English, and the various functions these have in combination with verbs (from prototypical adjuncts to complements). I argue that the constraints of historical data make it necessary to find different, new ways to determine which abstractions and distinctions are likely to have been represented in minds of historical language users, and to furthermore identify changes in constructional networks over time.
Zehentner, Eva