Predicting and explaining second language speech through individual differences: Insights from testing and extending the Second Language Linguistic Perception model

Abstract

In this talk, I will present five studies aimed at demonstrating that individual native production and perception abilities predict and explain success in L2 sound perception and word recognition. These studies tested and extended the theoretical and computational framework of the Second Language Linguistic Perception (L2LP) model was tested and extended. The model’s main claim states that accurate predictions for L2 success should be based on individual learners’ native production and perception as well as their responsiveness to rapid perception training. In Study 1, we show how differences in how listeners produce native vowel sounds are predictive of their discrimination accuracy when perceiving L2 vowel contrasts. The predictive power of individual native and L2 vowel perception on the learning of novel L2 words are reported in Studies 2 and 3. Study 4 shows that distributional training, based on the frequency distributions of difficult non-native contrasts, leads to short and long term improvement of difficult non-native contrasts and whether this improvement generalizes the contrasts produced in different contexts and by multiples speakers. Finally, Study 5 tests whether the perception training in Study 4 has short and long term impact on the L2 word learning. A discussion of the findings and their implications for the L2LP proposal and that of other models of L2 speech will be presented.