|dc.description.abstract||Over the last two decades, there has been an increasing interest in third language (L3) phonology (Cabrelli Amaro, 2012; Cabrelli Amaro & Wrembel, 2016). However, research so far has been primarily directed at specific stages of learning and has independently explored either production (e.g., Llama, Cardoso, & Collins, 2010; Llama & Cardoso, 2018; Tremblay, 2007; Wrembel, 2011, 2014; Wunder, 2011) or perception (e.g., Onishi, 2016; Wrembel, Marecka, & Kopečková, 2019). Almost none of the studies has examined the role of individual differences, including language aptitude, in L3 phonological development. To fill in the gap in the literature, the present study examines the role of cross-linguistic influence and language aptitude in both perception and production of L3 Spanish labial stops, /p/ and /b/. Cross-sectional data were collected from 40 Korean-English bilinguals at different proficiency levels of L3 Spanish. The dependent variables consist of VOT and F0 values as well as cross-language discrimination accuracy and identification tendency coupled with goodness-of-fit ratings.
Findings demonstrated that L3 learners, regardless of proficiency level, were quite accurate at distinguishing, specifically, the Spanish /p/-English /p/ and the Spanish /b/-Korean lenis /p/ pair; however, this ability did not translate into accurate production among low-proficiency learners. Rather, for VOT, cross-linguistic influence from L2 English affected the production of both target sounds, with effects that faded away with increasing L3 proficiency, albeit slowly in the case of Spanish /b/, possibly the result of markedness effects. For F0, no correlations were found with L3 proficiency, in general. Regarding the relationship between perception and production, significant correlations were observed in some cases, yet more direct links between perception and production were identified among intermediate and advanced learners.
Language aptitude appeared to facilitate both the perception and production of the target sounds. Specifically, a moderation analysis revealed a more substantial effect of phonemic coding ability on production at lower levels of L3 proficiency, as predicted in Skehan (2016), and corroborates previous findings (e.g., for L2 grammar, Li, 2015; for segmental learning, Saito, 2019; for feedback noticing, Yilmaz & Koylu, 2016).||