Individual Differences in the Adoption of Sound Change

Abstract

It is still unclear whether an individual’s adoption of on-going sound change starts in production or in perception, and what the time course of the adoption of sound change is in adult speakers. These issues are investigated by means of a large-scale (106 participants) laboratory study of an on-going vowel shift in Dutch. The shift involves the tense mid vowels /eː,øː,oː/, which are changing into phonologically conditioned upgliding diphthongs, and the original diphthongs /εi,œy,ɔu/, whose nuclei are lowering. These changes are regionally stratified: they have all but completed in the Netherlands, but have not affected the variety of Dutch spoken in neighboring Belgium. The study compares production (word-list reading) and perception (rhyme decision) data from control groups from each country to those of 18 “sociolinguistic migrants”: Belgian individuals who moved to the Netherlands years ago. Data are analyzed using mixed-effects models, considering not just the group level, but also individual differences. Production results show that at the group level, the migrant group is in between the two control groups, but at the individual level it becomes apparent that some migrants have adopted the Netherlandic norms, but others have not. Perception results are similar to the production results at the group level. Individual-level results do not provide a clear picture for the perception data, but the individual differences in perception correlate with those in production. The results agree with and extend previous findings on the role of individual differences in the individual adoption and eventual community propagation of on-going sound change.

How long is ‘a long term’ for sound change?

“…This is done by performing laboratoryphonological experiments with sociolinguistic migrants, in this case young speakers of Flemish Dutch who have moved to the Netherlands to do their university studies there. A previous large-scale cross-sectional study on the aforementioned four sound changes (Voeten, 2020a) has confirmed that, in the long term (years-multiple decades), these changes are indeed adopted by the sociolinguistic migrants studied there. This (eventual) adoption of the nd sound changes by fd sociolinguistic migrants, used here as a model to investigate the individual adoption of community change, presents a case of second-dialect acquisition.…”

“…This has also been found in cases of sound change: Sneller (2018) demonstrates how a diachronically innovative simple allophonic rule can rapidly overtake an older, more abstract and irregular, system as a result of contact between the two systems. Individual and social factors similarly overlap between sda and sound change; those discussed by Siegel (2010) largely overlap with those reported in sound change, such as duration of exposure, cognitive-processing styles, social-network size, and motivation (Yu, 2013;Beddor, 2015;Coetzee, Beddor, Styler & Wissing, 2018;Lev-Ari, 2018, Voeten, 2020a.…”

“…Such single-segment ultra-short-term accommodation has already been established in non-sda contexts (Zellou, Dahan, & Embick, 2017), and would extend the research on short-term accommodation to cross-dialectal linguistic primes (Walker, 2014;Pinget, Kager, & Van de Velde, 2020;Wade, 2020), which together form the basis for the aforementioned prediction. If such accommodation is dependent on the amount of familarity that the sociolinguistic migrants have with the Netherlandic variety (for which evidence in production was already found in Voeten, 2020a), then it would furthermore be expected that this priming effect interacts with the session of the experiment. Thus, the expectation is that the production experiment will replicate the between-groups dialectal differences known from the literature discussed in Section 1.1, but that the Flemish sociolinguistic migrants adapt their realizations in the directions of the Netherlandic controls over the nine months that were measured, as they have been found to do in the long term (years-decades; Voeten, 2020a).…”

“…Next, the same procedure as in Voeten (2020a) was applied: a cluster analysis was run on each random-effect term, using function Mclust from R package mclust based on a one-dimensional variable-variance model. As demonstrated in Voeten (2020a), this provides an empirical test of the extent of any individual differences: if participants behave uniformly with respect to these factors, the cluster analysis will identify a single cluster, whereas if there is a statistically significant pattern in the between-participant variation, multiple clusters will be identified.…”

The problematic concept of "native speaker" in Psycholinguistics: Replacing vague and harmful terminology with inclusive and accurate measures

“…In line with the individual differences approach, the authors argue that this demonstrates how new insight can be gained "when the cross-subject variability is treated as a source of evidence rather than a source of noise" (Tanner et al 2013). One method to analyse individual variation is to use random effect coefficients output by mixed effects model as the response variable, an approach that is increasingly common, for example, in individual-level correlation analyses of speech production and perception patterns (e.g., Pinget et al 2020;Voeten 2020). Overall, this approach has the advantage of moving away from categorical thinking and moving towards understanding underlying factors and mechanisms in language processing, which is abound with meaningful variability (e.g., see Yu and Zellou 2019 for an individual differences approach to phonological processing).…”