L3 acquisition of phonological variation: Schwa and (non) sensitivity to phonotactic constraints in Norwegian learners of French

Abstract

According to Bayley and Regan (2004), the mastery of phonological variation forms an integrated part of the competence of the post-L1 speaker who aims a near-native production. In French, two phonological variables are acquired rather late, i.e. liaison and schwa. Both involve deletion of word boundaries which complicates lexical recognition, as well as sensitivity to register and modality. The two phenomena do however vary in that liaison - with inter-vocalic consonant realization - doesn't entail additional difficulty for production, while schwa, when absent, creates consonant sequences often not perceived and difficult to produce. While post-L1 acquisition of liaison is rather extensively studied, schwa is not, which according to Hannahs (2007) might reflect the complexity of the influencing factors. We know however, from L1 studies, that phonotactic constraints, on the syllabic as well as the segmental level, do influence schwa alternation (for an overview, see Andreassen, 2013), and in this paper, we present a first, detailed study of Norwegian learners of French, which aims to understand the importance of phonotactics during post-L1 acquisition.

The study is based on conversational data from two corpora of Norwegian learners of French, 16 informants from Tromsø, proficiency level A2, and 8 informants from Oslo, proficiency level B1/B2. Schwa behavior is extracted using the newly developed IPFC schwa coding system (Isely et al., 2017), of which a pilot has already been tested on parts of the Norwegian learner data by Andreassen and Lyche (2016). While the latter, initial analysis indicates a gradual sensibility towards phonotactic context, other factors such as frequency and access to written representations are observed as playing an important role. In this paper, we aim to identify an acquisitional path by going in detail into the phonotactic context of schwa and how it behaves, as well as the realization of the secondary cluster in the case of schwa absence. To detect L1 influence, we take into account L1 phonotactic as well as prosodic constraints. To determine sensibility towards phonological variation in general, we take into account previous analyses of liaison in the same corpora (Andreassen & Lyche, 2015).