Interaction of markedness and transfer of /r/ in L1 Thai learners acquiring L1 Thai and L2 English

Abstract

In Thai, prevocalic /r/ shows a range of possible variants. The two standard variants are the prestigious trill [r] and the flap [ɾ], but some speakers may also use the lateral [l], or, in an initial cluster, /r/-cluster deletion [Ø]. In this dissertation, I argue that the phonological modification of rhotics in favour of lateral [l] and deletion [Ø] can be explained by considering the preference for the unmarkedness. Features or structures are distinguished as marked or unmarked features. The lateral [l] and deleted [Ø] variants are considered relatively unmarked features in comparison with rhotics. The primary aim of the dissertation is to investigate which variable strategies Thai speakers employ when speaking English, alongside their native language. The rhotics in these two languages differ and, in terms of markedness, the English approximant [ɹ] is comparatively more marked than the rhotics in Thai. This study is an overview of L2 phonology in Thai liquids, and this is supplemented by a variationist analysis. It examines the speech production of prevocalic English /r/ by L2 Thai leaners of English from the perspectives of markedness and transfer. It investigates if the Markedness Differential Hypothesis (MDH) proposed by Eckman (1977) is able to explain the errors subjects make in learning L2 English. In the variationist analysis, the quantitative paradigm was adopted to specify which linguistic and extralinguistic factors were of statistical significance in relation to the merger of /r/ and /l/ in L1 Thai and L2 English. Speech data were subjected to mixed logistic regression analysis with syllable stress, preceding segment, age, gender, speech style, and proficiency as based on IELTS speaking score as independent variables along with speakers and words as random effects. Data were collected from ten native speakers of Thai. The results show that the MDH could predict difficulties in L2 acquisition and which features were more likely to be transferred to L2 English, the results in terms of the relationship between the degrees of difficulty and degrees of markedness did not support the MDH. In the variationist analysis, the best-fit model shows that preceding segment, age group, and speech style influence the choice of variants in terms of the merger of /r/ and /l/ in both the L1 and L2. Gender is a significant factor in L1 Thai but is found to be insignificant in L2 English. With /r/-/l/ merger in Thai, the declining use of the standard Thai flap [ɾ] and trill [r] in formal speech in the present study seems to follow the trend indicated in previous studies, ii but the decline of /r/-/l/ merger in informal context is unexpected, given that the existing literature suggests that the frequency of /r/-/l/ merger in Thai is increasing.