Training effects on native and novel linguistic categories

Abstract

Categorical perception (e.g., Liberman et al., 1957) and the perceptual magnet effect (Kuhl, 1991) are frequently invoked to help explain how the perceptual system deals with highly variable speech input by warping the perceptual system in favor of native contrasts. Research into L2 acquisition demonstrates that this warping can be overcome and non-native categories can be introduced to listeners with extensive training (e.g., Logan et al., 1991); further studies into novel category acquisition have shown a perceptual magnet effect for nonspeech stimuli (Guenther et al., 1999). This paper addresses native English speakers’ acquisition of a novel category, the velar fricative /x/, by testing the effects of category labeling and discrimination training on subjects sensory trace and context coding auditory modes. The results demonstrate training affects the novel category differently from the native categories, with labeling training creating a perceptual magnetlike effect only in the novel category. Both types of training increased subjects’ abilities to discriminate categories, but mixed testing and training types (i.e., categorical testing with noncategorical training and vice versa) resulted in the greatest improvements.