Within-Speaker Vowel Covariation

In 2021, I joined the Marsden funded project Towards an improved theory of language change as a Post Doctoral Fellow at the New Zealand Institute of Language, Brain and Behaviour. The project’s principal investigator is Jen Hay, with Lynn Clark and Kevin Watson as associate investigators.

One overarching theme in the project is the development of quantitiative techniques for investigating whole systems of linguistic variables rather than individual variables.

Before I joined the team, the previous Post Doctoral Fellow, James Brand, led a study into across speaker vowel space variation in New Zealand English (NZE) using the Origins of New Zealand English (ONZE) corpus. This study uses Principal Component Analysis, applied to by-speaker random intercepts from GAMM models, in order to discern structural patterns of variation in the NZE vowel space across a long time course (with speakers both from the mid-19th Century up until the 1980s). The result of this was a publication in the Journal of Phonetics, along with significant online supplementary materials and a Shiny interactive.

My primary work on the project has been to develop these techniques and apply them to within speaker variation. This answers a need in the current literature on style in sociolinguistics, where it is thought that multiple linguistic variables should pattern together as speakers take on different styles, but quantitative methods for discerning and exploring any such patterns are at an early stage of development.

I am lead author on two papers under review within this project:

1. A methods paper on the use of PCA to investigate systems of vowels.
2. An investigation of the effect of relative amplitude variation on speaker vowel spaces within QuakeBox recordings.

Upon acceptance, the associated code and data will be made available through public github repositories.