Linguistic generalization in L 2 consonant identification accuracy : a preliminary report

Cross-language perception of phonetic features was investigated via an experiment in which native speakers of Korean and English identified speech sounds varying across voicing (voiced vs. voiceless), place of articulation (labial vs. coronal), and manner of articulation (stop vs. fricative) features as well as prosodic context (syllable initial vs. syllable final). Because Korean has no anterior non-sibilant fricatives and exhibits voicing and manner neutralization in syllable codas, the identification task served as a test of the effects of native language phonological patterns on generalization in the perception of non-native features. While some features (e.g., voicing) were identified fairly accurately and generalized across other features (e.g., manner of articulation), complex patterns of interaction between the experimental factors were also observed (e.g., coronal fricatives in coda position were almost all identified as voiceless, whereas labial stops in coda position were identified equally often as voiced or voiceless regardless of their actual voicing specification). These results are discussed in terms of previous approaches to second language speech perception. Limitations of the experimental protocol are discussed, and directions for future research are briefly outlined.