Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Pre-attentive phonological encoding does not track phonetic cue weighting
Poster Session B, Friday, September 12, 4:30 - 6:00 pm, Field House
Chao Han1, Katie Ma1, Yoonjung Kang1, Philip Monahan1; 1University of Toronto Scarborough
Understanding how the brain encodes phonological categories requires insight into how it weighs different phonetic cues. For instance, voicing contrasts in English stops (e.g., /g/-/k/) are primarily cued by voice onset time (VOT), with fundamental frequency (F0) as a secondary cue (Kapnoula et al., 2017). While larger mismatch negativity (MMN) responses have been linked to pre-attentive encoding of phonological–as opposed to phonetic–contrasts (Kazanina et al., 2006) and to more robust phonological category encoding (Díaz et al., 2008), it remains unclear whether a) secondary cues pre-attentively engage in phonological encoding, and b) MMN responses reflect participant-level variation in cue-weighting differences. Thus, we conducted a behavioural task followed by an MMN experiment to address these questions. Methods: English syllables were resynthesized to vary between the dorsal stop [ɡa] and [ka] along VOT (short: 5, 8, 11ms; long: 47, 50, 53ms) and F0 (low: 74, 79, 83Hz; high: 132, 140, 148Hz). English speakers (n = 37) were assigned to either the VOT group (i.e., varying short and long VOT; fixed low F0) or the F0 group (i.e., varying low and high F0; fixed short VOT). Participants first completed a two-alternative forced-choice identification task, categorizing stimuli as [ga] or [ka]. The MMN experiment used the same stimuli, along with glottal [ha] syllables that varied in aspiration (“VOT”, short: 80, 83, 86; long: 122, 125, 128ms) and F0 (phonetic values identical to dorsal stimuli). Each participant completed two oddball blocks for both dorsal and glottal stimuli: one with short VOT or low F0 as standard and long VOT or high F0 as deviant, and another with the reverse configuration. We expected the VOT group to show an MMN in dorsal conditions (phonological contrast). If F0 also supports pre-attentive voicing contrasts, it should elicit a larger MMN in dorsal (phonological) than glottal (phonetic) conditions. If MMN reflects individual cue weighting, amplitudes should correlate with behavioral cue weights. Results: Behavioural results were submitted to a mixed-effects logistic regression. Coefficients for VOT and F0 were extracted for individual participants to index cue weights. Significant VOT and F0 coefficients confirmed that both cues can be utilized for the voicing contrast. The MMN amplitude was computed using temporal cluster-based permutation tests and averaged across frontocentral electrodes. In the VOT group, a significant MMN emerged only for short VOT deviants in the dorsal condition, consistent with featural accounts (Lahiri & Reetz, 2002; Hestvik & Durvasula, 2016; Monahan, 2018). In contrast, the F0 group elicited a significant MMN in both dorsal and glottal conditions with comparable amplitudes, suggesting that F0 was not engaged in pre-attentive encoding of voicing contrast. No significant correlations were observed between MMN amplitudes and individual cue weights across all conditions. Conclusion: Secondary phonetic cues can support phonological contrasts in behavioral categorization but are not engaged pre-attentively. Moreover, individual differences in cue weighting are not captured at the level of the MMN, which aligns with previous findings that fine-grained phonetic details degrade along the subcortical-cortical auditory pathway (Toscano et al., 2010; Ou et al., 2023).
Topic Areas: Speech Perception, Phonology