Postersession 3
Poster #: 117
Topic: Speech and language (incl. deficits)
Friday, Sep 11, 2015
1st floor

Changes in MMR amplitude reflect fast phonetic learning in adult listeners

Kateřina Chládková1 & Paola Escudero2

1Amsterdam Center for Language and Communication, University of Amsterdam, Amsterdam, Netherlands
2MARCS Institute, University of Western Sydney, Surry Hills, Australia

This study examined whether phonetic learning occurs automatically after a short exposure to a statistical distribution of speech sounds. We assessed the amplitude of the mismatch response (MMR) before and after 9 minutes of unattended distributional training. During training, twenty Spanish listeners were exposed to a bimodal distribution of sounds on the first-formant (F1) dimension: half of them listened to a distribution with a trough at low F1 values, and the other half with a trough at high F1 values (“low-boundary” and “high-boundary” groups, respectively). A pre-attentive oddball paradigm was presented before and after training: the Standard had an F1 value representative of the Spanish /i-e/ boundary, and two Deviants had values of Spanish /i/ “i-Deviant”, and /e/ “e-Deviant”. (Note that in training the Standard, i-Deviant, and e-Deviant all had the same probability of occurrence.) If listeners learned from the bimodal distributions they heard in training, the trained boundary locations should affect their pre-attentive discrimination of stimuli at post-test. Specifically, for the low-boundary group, Standard and e-Deviant should be perceived as one category and i-Deviant as the oddball, while for the high-boundary group, e-Deviant should be the oddball. Difference waves were computed as post-test minus pre-test responses to physically identical stimuli, and MMR quantified as the average absolute amplitude between 100 and 200ms after stimulus-onset. As predicted, i-Deviant yielded larger MMR than e-Deviant for the low-boundary group, and vice versa for the high-boundary group. This demonstrates that adults successfully learn to discriminate non-native vowel contrasts after short distributional training.