Extending perceptual assimilation model to L3 phonological acquisition

ABSTRACT

The scarcity of research on speech perception among multilingual speakers precludes a full understanding of phonological acquisition in the third language (L3). In this controlled case study, we investigate L3 phonological acquisition in the perceptual domain and test the predictions of Perceptual Assimilation Model- L2 (Best & Tyler, 2007) adopted for multilingual learners. We employed an AX discrimination task, testing categorical discrimination of Polish sibilants, and a cross-linguistic similarity task, testing perceptual distance between Polish, English and German vowels. We examined L3 Polish perception in 10 multilinguals (aged 14) with L1 German and L2 English who differed in terms of language status (heritage vs. non-heritage). Their perception tasks performance was analysed for accuracy and reaction time. The cross-linguistic similarity task demonstrated that multilinguals assimilate some L3 sounds to both L1 and L2 categories, with a preference for the latter. In the majority of cases multilinguals make a distinction between similar L1, L2 and L3 sounds. The AX results showed that even beginner L3 learners distinguish highly similar L3 sibilant pairs. Our data suggests that the PAM-L2 model could also be extended to L3 acquisition; however, beginner L3 learners seem more likely to perceive subtle acoustic differences in novel phonological contrasts.

KEYWORDS:

• Multilingualism
• speech perception
• L3 acquisition
• Polish phonological categories
• heritage speakers
• L2/L3 learners

Acknowledgements

The authors would like to thank Ulrike Gut for her support in the design of the study and valuable feedback on an earlier version of the article. We would also like to thank the reviewers for most helpful comments.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Magdalena Wrembel http://orcid.org/0000-0001-7804-3199

Marta Marecka http://orcid.org/0000-0003-2873-9395

Romana Kopečková http://orcid.org/0000-0001-5647-9784

Notes

1 It should be born in mind that research into multilingual acquisition, due to its inherent complexity, is inevitably based on smaller samples than is usually the case in L2 studies.

2 It was a conscious choice of the experimenters the stimuli were recorded by one speaker per language, in case of the L1 German and L2 English, and two speakers for L3 Polish. Not including more voices was dictated by two reasons – a pragmatic one and a theoretical one. The first reason is that allowing for more voices per language would extend considerably the length of the task, which was already quite time-consuming and tiring for the young participants. The second reason is that such a task (e.g. with 3 different speakers for ABX discrimination task) would be too complex and cognitively challenging given the young age of the participants and a relative difficulty of perceptual tests administered (cf. Strange & Shafer, 2008 p. 162).