- This event has passed.
Hearing and Donuts (Brain and Bagels) Seminar
September 30, 2022 @ 8:30 am - 9:30 am
Parrell (SMAC) and Niziolek (BLAB) Labs, Waisman Center
Topic: Language Categories in the Brain and Their Processing: Mismatch Negativity and Chinese Lexical Tone
Categorization is the foundation of speech perception and spoken word recognition. Mismatch negativity (MMN) is a brain response often used to probe category representation, which is reliably elicited when stimuli are presented in an oddball paradigm (e.g., d-d-d-d-d-d-t, where the last deviant stimulus will elicit MMN; the repeated stimuli are called standards). MMN is sensitive to the physical distance of the standards and the deviant, such that greater physical distance is correlated with larger MMN amplitude. MMN has also been extensively employed to examine the representation of linguistically meaningful categories in the brain, ranging from phonemic contrasts (e.g., /d/ vs. /t/ in English) to syntactic agreement (“They are” vs. “You are”).
Lexical tone is the contrastive use of F0 (determined by the rate of vocal fold vibration and perceived as pitch) to differentiate words. Tonal language is estimated to comprise approximately 50% to 70% of the world’s languages. This talk will present two studies examining the representation and processing of Chinese lexical tone in the brain using EEG (electroencephalography), which has millisecond temporal resolution.
The first study examines whether MMN can index Chinese lexical tone categories. While MMN’s sensitivity to stimuli’s physical distance is well-known, what is less considered in previous language-relevant MMN research is the brain’s ability to detect ad hoc arbitrary rules (e.g., the duration of the current sound determines the frequency of the following sound) and generate MMN. The design of the first study allows us to tease apart three potential MMN generating mechanisms: (1) basic auditory processing, (2) arbitrary rule formation, and (3) linguistic processing. Surprisingly, mixing different Chinese lexical tone categories in the standards can still generate robust MMN, and two contrastive Chinese lexical tones do not always yield MMN. The experimental data are best understood as a synergy of the three MMN generating mechanisms described above. The results are also compared with MMN studies of English stops such as /d/ and /t/ to shed light on how different linguistic features may be differentially presented and processed in the brain.
Speech sound has different surface realizations depending on their context (e.g., the English plural -s can be either /s/ as in cat-s or /z/ as in dog-s). Using the Chinese 3rd tone alternation as a test case (T3 + T3 → T2 + T3; the numbers are tone categories), the second study examines whether the brain can access the underlying tone when the surface realization can be ambiguously interpreted. The presence of tone alternation (i.e., T3+T3) as either standard or deviant eliminates MMN, despite the apparent vowel distinction on the surface (e.g., the comparison between /xuaŋ2 ma3/ and /xai3 ma3/ does not yield MMN), suggesting that the brain can access the underlying representation during online speech perception. A tentative model with three hypothesized underlying processing mechanisms is proposed to explain the observed latency and amplitude differences across conditions.