The effects of linguistic experience as revealed by behavioral and neuromagnetic measures: A cross-language study of phonetic perception by normal adult Japanese and American listeners

Speech perception is indelibly marked by an individual's language experience. Kuhl (2000a,b) theorized that early language learning involves a strong statistical mapping process that leads to a neurally committed system, warping perception in service of the first language. But the nature of this “neural commitment” is not well understood. Specifically, what are the essential attributes of the neural commitment to the native language? At what level, where and when do the neural activities in the brain show a difference in encoding native vs. nonnative speech?; This dissertation aimed to address these questions by using traditional behavioral measures and modern neuromagnetic measures integrated with magnetic resonance imaging (MRI). The two cross-language studies used both linguistic and acoustic controls. The stimuli included a /ra-la/ contrast (native to American listeners, and nonnative to Japanese listeners), a /ba-wa/ contrast (native to both populations), and sinewave analogs of the /ra-la/ contrast. Behavioral experiments employed identification and discrimination tasks, and the magnetoencephalography (MEG) experiments used the passive oddball paradigm in two conditions that correspond to discrimination and categorization at the preattentive level.; The results showed that listeners do not perceive difficult nonnative speech contrasts as accurately and efficiently as native speech contrasts. There are at least four neural indexes. First, there is a bilateral reduction in the mismatch response (an automatic change detector), which is more notable at the preattentive categorization condition. Second, there is not a loss of auditory sensitivity to the acoustic features of the sounds; rather linguistic experience alters the neural encoding of basic acoustic properties very early in the left hemisphere. Third, nonnative speech processing involves bilaterally stronger and longer dipole activities in two brain regions, the association auditory cortex and the inferior parietal region. Fourth, processing nonnative speech is similar to processing nonprototype speech sounds, which is harder to categorize and shows a less focal pattern of neural activities as indexed by the number of dipole clusters in spatial distribution. These results not only provide a general characterization of the “neural commitment” to the first language but also point to potential ways to change the system.