Measuring Language Change in Diachronic Corpora

This article outlines statistical methods for measuring language change with diachronic corpora. Specifically, it discusses competitive and non-competitive changes, both of which frequently follow S-shaped curves. However, their distinct data structures necessitate different analytical approaches. Using the complementation patterns of English help as an illustration of competitive change, the article presents (mixed-effects) logistic regression as a powerful technique to quantify effects, transitional periods, and potential actuation dates. Non-competitive change is measured with frequencies that are relative to a standardised normalization base for different time points. The article illustrates and discusses advantages as well as disadvantages of such analyses.