Source code for a draft of a new paper with Meredith Tamminga and Aaron Ecay is available on github. We use generalized additive models to simultaneously estimate the effect of priming and style-shifting on the use of DH-stopping variants: Check this out! vs. Check dis out!

Here’s the abstract:

Intraspeaker variation is typically characterized by repetitiveness in variant choice, but there are multiple possible sources of such repetitiveness. We distinguish two types of clustering: sequential dependence, which we associate with priming, and baseline deflection, which we associate with style-shifting. We argue that because both priming and style-shifting are likely to be at play in producing observed quantitative patterns, it is desirable to adopt quantitative models that can simultaneously estimate sequential dependence and baseline deflection as distinct sources of temporal clustering. We propose the use of Generalized Additive Models (GAMs) for this purpose. We test the use of GAMs on a case study of DH-stopping in Philadelphia English sociolinguistic interviews and find that the resulting parameter estimates are reasonable given basic expectations about how style and priming should be distributed across individuals. We advocate for the adoption of this and similar new tools to advance the integration of psycholinguistic, sociolinguistic, and corpus linguistic insights in the study of intraspeaker variation.