The effect of training modality and instruction type on non-native speech learning

Description

Learning to speak a new language requires the ability to perceive and produce a set of new sounds. Despite a general agreement that there is a link between perception and production in L2 speech learning, the nature and directionality of this link remain unclear [1]. On the one hand, studies show that perception improves production, and it is either assumed that the two modalities are related [2], or that they share representations and cognitive resources during an overlapping learning process [1]. On the other hand, empirical evidence suggests that good production abilities can exist without fine-tuned perception [3] and, conversely, speakers can easily perceive different L2 speech sounds without being able to produce them [4]. Outcomes in L2 speech learning may also vary depending on the instruction type (implicit vs. explicit). Although empirical research has shown that implicit instruction is often enough for adult learners to “pick up” a new language [5], a recent meta-analysis on L2 pronunciation suggests that explicit instruction tends to
result in greater learning gains [6]. In this study, we investigate the effect of training modality (perception vs.
production) and instruction type (implicit vs. explicit) on the acquisition of non-native phonological categories.
The linguistic focus in our study is on the mid/low vowel contrast /e/-/ɛ/, which exists in European Portuguese, our target language, but is absent in the native language of our participants. In our study, 152 speakers of Southern British English without prior knowledge of Portuguese were randomly assigned to four conditions (each n = 38): (a) implicit perceptual training, (b) explicit perceptual training, (c) implicit production training, and (d) explicit production training. On day 1, participants completed two pretest tasks - an AX discrimination task and a delayed repetition task - to measure their ability to perceive and produce the target contrasts. They also completed the Language and Social Background Questionnaire (LSBQ; [7]). On days 2, 3, and 4, participants received three phonetic training sessions, each lasting approximately 45 minutes and administered over consecutive days via Gorilla experiment builder (www.gorilla.sc). The perceptual training groups (a and b) were trained on perception by means of an AX discrimination task with feedback. The production training groups (c and d) were only trained in production with a delayed repetition task, also with feedback. The difference between the implicit groups (a and c) and the explicit ones (b and d) concerns the provision of explicit instruction to groups (b) and (d) about the articulation of the target contrast /e/-/ɛ/ prior to each phonetic training session. On day 5, all participants completed two posttest tasks to assess whether the training had impacted on their ability to perceive and produce the target contrast. We used the same tasks as in the pre-tests.
Data collection is ongoing, and our analysis below focuses only on the perception pre- and post-tests, since production test data is currently being transcribed. The analysis of our current results, with 36 participants, revealed a significant effect of test (pretest vs posttest, χ2 (1) = 22.718, p < .001), confirming that three phonetic training sessions over three days improved perceptual discrimination of the /e/-/ɛ/ contrast. The interaction between test and training group was not significant (χ2 (3) = 5.2148, p = .157), indicating that the gains from pre- to posttest between groups do not depend on training modality or instruction type. This could result, however, from the small sample size of the preliminary analyses. With our planned sample, we expect that the explicit training groups will outperform the implicit training groups, and the production training groups will improve more than the perception training groups. Our results will contribute to the perception-production link debate, by investigating the interaction between training modality and instruction type in the earliest stages of L2 speech learning.

Period	23 Oct 2023
Event title	Encontro Nacional da Associação Portuguesa de Linguística
Event type	Conference
Conference number	40
Location	Ponta DelgadaShow on map
Degree of Recognition	National