How to analyze linguistic change using mixed models, Growth Curve Analysis and Generalized Additive Modeling

TLDR

Empirical language evolution studies involve temporal change, and nonlinear change with nested dependencies (e.g., within dyads or chains) is best analyzed using mixed models, GCA, and GAMs. The tutorial introduces mixed models, GCA, and GAMs for analyzing linguistic change. It provides recommendations for model‑fitting choices when applying these methods. Annotated R scripts in the supplementary data serve as a springboard for readers’ analyses.

Abstract

When doing empirical studies in the field of language evolution, change over time is an inherent dimension. This tutorial introduces readers to mixed models, Growth Curve Analysis (GCA) and Generalized Additive Models (GAMs). These approaches are ideal for analyzing nonlinear change over time where there are nested dependencies, such as time points within dyad (in repeated interaction experiments) or time points within chain (in iterated learning experiments). In addition, the tutorial gives recommendations for choices about model fitting. Annotated scripts in the online Supplementary Data provide the reader with R code to serve as a springboard for the reader's own analyses.