You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

Poster D61, Thursday, November 9, 6:15 – 7:30 pm, Harborview and Loch Raven Ballrooms

Cross-linguistic differences in MMN asymmetry: Voicing underspecification in Japanese

Yasuaki Shinohara1, Arild Hestvik2, Rinus Verdonschot1, Karthik Durvasula3, Hiromu Sakai1;1Waseda University, 2University of Delaware, 3Michigan State University

Underspecification has been argued to lead to asymmetric Mismatch Negativity (MMN) [1]. In English, a larger MMN is observed when a deviant /d/ is perceived in the context of standard /t/ with varying voice onset time (VOT), compared to when a deviant /t/ is perceived in the context of /d/ with varying VOT [2]. Assuming English to have laryngeal specification for voiceless stops but not for voiced stops, the acoustic /d/-series have no distinctive phonological feature in the memory trace, and therefore cause no conflict with a phonetic deviant /t/, eliciting reduced or no MMN. If this asymmetry arises from language-specific underspecification, then an opposite featural specification is predicted to exhibit the opposite MMN asymmetry. This would be predicted for Japanese, where voiced stops are argued to be laryngeally specified, with voiceless stops unspecified [3]. Recently, an alternative explanation of universal marking of the [SPREAD GLOTTIS] feature has been suggested [4], which predicts that Japanese should exhibit the same asymmetry as English. In the present study, we examined which of these two predictions was correct. Forty-four monolingual Japanese speakers aged 18-24 years participated in a behavioral experiment followed by an ERP experiment. In the behavioral experiment, subjects identified the same stimuli used in [2], continuously varying in VOT as being either /dæ/ or /tæ/, and exhibited a categorical /d/-/t/ phoneme boundary around 35ms. Consequently, four /dæ/ (VOT 10, 15, 20, 25ms) and four /tæ/ (VOT 45, 50, 55, 60ms) tokens were selected. In the ERP experiment, these four VOT variations for each phoneme were presented so that subjects formed their phonemic memory trace for the standard stimuli and used phonetic memory trace for the deviant stimuli. Subjects were randomly assigned to one of two block-order groups (in one order, /tæ/ was the deviant in the first half of the experiment; in the other, /dæ/ was the first deviant stimulus). The results demonstrated that there was a significant MMN effect; deviant tokens had lower amplitude than standards. In addition, Japanese speakers’ MMN for /t/ was significantly larger than that for /d/, when /t/ was presented as deviant in the first half of the ERP experiment. This effect was not observed when /d/ was presented as deviant in the first half. This means that underspecification in Japanese elicits an asymmetric MMN in the opposite direction from English. This indicates that Hestvik and Durvasula's result is not due to universal markedness, contrary to [4]. The MMN asymmetry thus provides neurobiological evidence for language specific pattern of underspecification. References [1] Eulitz, C., and Lahiri, A. (2004). Neurobiological Evidence for Abstract Phonological Representations in the Mental Lexicon during Speech Recognition. Journal of Cognitive Neuroscience. [2] Hestvik, A., and Durvasula, K. (2016). Neurobiological evidence for voicing underspecification in English. Brain and Language. [3] Mester, A., and Ito, J. (1989). Feature Predictability and Underspecification: Palatal Prosody in Japanese Mimetics. Language. [4] Schluter, K. T. et al., (2017). Laryngeal Features Are Phonetically Abstract: Mismatch Negativity Evidence from Arabic, English, and Russian. Frontiers in Psychology.

Topic Area: Perception: Speech Perception and Audiovisual Integration

Back to Poster Schedule