Learning of non-native tonal contrasts with or without tonal context

Abstract

Listen

Translate article

The learning of non-native (Cantonese) tonal categories by native speakers of another tonal language (Mandarin) was investigated under two learning conditions. For listeners in the tonal context (TC) group, learning occurred by listening to target tones within various carrier phrases. Listeners in the no-context group (NC) were trained using target tones produced as single words. Training stimuli were produced by the same speakers for both groups. A high-variability procedure with feedback was used for training. Sixteen listeners (eight in each group) took part in eight sessions over a period of about 1 month. Categorical perception of Cantonese tones was measured before training (following a brief familiarization session) and after training. For both groups, tone perception was tested for one level- and one contour-tone continua using identification and discrimination procedures. Identification performance before and after training did not differ between the two groups for the level-tone continuum, showing evidence of learning of the (non-native) mid-level category. For the contour-tone continuum, the TC group showed better learning of the (non-native) low-rising/low-level contrast, as well as better perceptual distinction between low-rising and high-rising tones. [Work supported by Hong Kong RGC Grant HKU 7224/03H.]