Modelling frequency in a chain shift

Posted on March 25, 2014 by

Next Tuesday (1st April), Susanne Wagner of the University of Oxford will be presenting joint work with Matthias Hofmann (Chemnitz) on Modelling frequency in a chain shift – questions, suggestions, and (some) answers.

The talk is part of the Langwidge Sandwidge series and takes place at 1pm in Samuel Alexander A202. You’d be an April Fool not to come! Here’s the abstract:

When linguists talk of frequency effects, they are generally referring to the fact that variants of the same variable do not behave uniformly across the lexicon. Interestingly, (high) frequency can both advance and block changes in progress, which makes generalisations concerning “the” effect of frequency difficult. Moreover, several studies have shown recently that frequency effects disappear once a random effect for word (and speaker) is included. Modelling frequency poses a number of challenges: should frequency criteria be established locally or globally? How should frequency be represented numerically (e.g. log vs. absolute N)? Should frequency be included as a linear or categorical variable? In this talk, we will address these questions by drawing on data evidencing an ongoing sound change – the Canadian Shift. We discuss strategies for different modelling options and also explore different methods for establishing thresholds and cutoff points for the variable frequency. With the help of exploratory data analysis and “qualitative data mining”, we will show that modelling frequency has to include a number of carefully chosen steps in order to “make the most” out of any frequency effects.