2025/10/21 by Elena Sheard, Jennifer Hay, Robert Fromont +2 · 1 voice
Computer Science · #Library Science and Information Systems
paper · doi:10.16995/labphon.19054
openalex publication_date 2025/10/21 · openalex created_date 2025/10/22 · openalex updated_date 2026/06/26
This is an accepted article with a DOI pre-assigned that is not yet published.Clusters of New Zealand English monophthongs systematically covary across speakers, over time and across corpora (Brand et al., 2021; Hurring et al., In press), meaning information about how a speaker realises one vowel also gives us information about how they produce the other vowels in a given cluster. There is also evidence that listeners perceptually differentiate between speakers based on their realisations of one cluster that captures vowels undergoing change in New Zealand English, across which speakers are ‘leaders’ or ‘laggers’ (Sheard et al., In prep). Here, we ask whether listeners socially evaluate speakers with distinct covarying vowel patterns differently. We conduct a free classification task in which 127 New Zealanders made, and labelled, groups of speakers who they thought sounded similar, based on audio stimuli from 38 older New Zealand women. We apply Multidimensional Scaling analysis to create a two-dimensional perceptual space of the stimuli and identify distinct groups of speakers who are characterised by (a) shared aspects in speech production, and (b) how listeners label them. The analysis reveals that listeners do socially evaluate leaders and laggers differently, with leaders associated with rurality and lower socioeconomic status. However, listeners use a wide range of strategies to group speakers, and the broader speech context mediates the salience of the covarying vowels.