Training to ignore vs. training to attend: The distribution of selective attention in the acquisition of a foreign phonetic contrast

Maria V Kondaurova, Purdue University

Abstract

Perceptual learning of speech sounds in both first and second language acquisition can be understood in terms of the operation of two cognitive processes, cue enhancement and cue inhibition. Cue enhancement means learning to attend to those acoustic cues which characterize a speech contrast for native listeners and cue inhibition means learning to ignore those cues that do not. The present study investigates the role of these two processes in the perceptual learning of American English tense and lax front unrounded vowels ([i] vs. [ I]) by native Spanish listeners. Recent studies have demonstrated that many adult native Spanish listeners have difficulty distinguishing these vowels which differ along two acoustic dimensions: spectrum (vowel quality, related mainly to the first three formant frequencies) and duration (vowel length). While native English listeners rely predominantly on spectrum with vowel duration playing a secondary role, native Spanish listeners have been shown to rely exclusively or primarily on vowel duration in the perception of this contrast. The study compares the effectiveness of three auditory training methods, adaptive training, inhibition training and training with a natural distribution of spectral and duration cues, with respect to their ability to redirect native Spanish listeners' attention from duration to spectrum. While adaptive training has been shown to be successful in enhancing attention towards primary cues that define a non-native phonetic contrast, inhibition training has been less systematically studied especially with respect to vowel contrasts. The results of the study demonstrated that listeners in all training groups increased their reliance on spectrum and simultaneously decreased their attention to duration. However, there were differences in the effectiveness of training methods: Inhibition and adaptive training methods were more successful than natural distribution training in the reduction of attention to duration and the increase of attention to spectral properties both in the training sets and as generalized to a different phonetic context. The adaptive training method was also more successful than either inhibition or natural distribution training for developing between category distinctiveness as shown by the results of a pairwise discrimination task. Overall, the study suggests that phonetic learning can be characterized in terms of two cognitive processes, enhancement and inhibition that function to shift selective attention simultaneously between the multiple separable dimensions available for categorization. Both processes are found to be equally effective in changing cue weighting schemes in perceptual learning of non-native speech sounds. The results suggest that cue-specific training (whether enhancing or inhibitory) maybe an effective tool in the development of better training methods for second language learners and hearing-impaired listeners.

Degree

Ph.D.

Advisors

Francis, Purdue University.

Subject Area

Linguistics

COinS