Inter- and intra-L1 differences in L2 speech perception

In a perception experiment, L1 Mandarin and L1 Japanese novice learners of Korean classified non-tense /s/or tense /s*/initial Korean CV tokens. A mixed effects logistic regression model with acoustic cues as predictor variables was built for each L1 group, and each individual’s regression coefficients were interpreted to be the cue weighting used in identifying Korean /s/ and /s*/. We propose that the weighting of L2 perceptual cues is influenced by the weighting of the same cues in the L1 perception of acousticaly similar contrasts, but that intra-L1 individual variation is great enough that the expected inter-L1 differences may appear less well defined.