DOI: 10.3366/cor.2026.0363 ISSN: 1749-5032
Continuous keyness? Identifying words and features that correspond with continuous extra-linguistic variables
Jesse Egbert, Elizabeth Hanks, Douglas BiberKeyword and key feature analyses provide researchers with a rank-ordered list of linguistic items based on a dichotomous comparison between a target corpus and a reference corpus. But what if we want to rank words and linguistic features based on the strength of their relationship with a ‘continuous’ (as opposed to a dichotomous) extra-linguistic variable? We propose new methods that allow us to extend ‘keyness’ to continuous extra-linguistic variables. We discuss steps for conducting these analyses, including appropriate adaptations in the case of non-prevalent linguistic items and short texts. We then demonstrate how these methods can be successfully applied to rank ordering words and features in two case studies.