Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Language-specific phonological patterning shapes neural encoding of phonetic categories
Poster Session D, Saturday, September 13, 5:00 - 6:30 pm, Field House
This poster is part of the Sandbox Series.
Chao Han1, Tae-Won Lim2, Say Young Kim2,3, William Idsardi4, Philip Monahan1; 1University of Toronto Scarborough, 2Hanyang University, 3Hanyang Institute for Phonetics and Cognitive Sciences of Language, 4University of Maryland
Linking neurophysiology and linguistic theory requires understanding how representational primitives are encoded in the brain (Embick & Poeppel, 2015). Advances have been made in identifying where the brain integrates acoustic information to encode phonetic categories (Yi et al., 2019), but little is known about how encoding the same phonetic categories differs based on language-specific phonological patterning (Monahan et al., 2022; Mai et al., 2024). We test how the neurophysiological encoding of the flap [ɾ] is shaped by English and Korean phonology, using mismatch negativity (MMN), which is widely used to probe phonetic and phonological categories, and phoneme-related potential (PRP), an ERP time-locked to the onset of phonemes, with greater PRP similarity reflecting closer groupings in feature space (Mesgarani et al., 2014; Khalighinejad et al., 2017). Although [ɾ] is allophonic in both languages, in English, it patterns with [t] as allophones of /t/, contrastive with /l/; whereas in Korean, [ɾ] patterns with [l] as allophones of /ɾ/, contrastive with /t/. We observed an MMN asymmetry consistent with phonological patterning, while PRP similarity aligned with phonetic features. Methods: English speakers (n = 21) participated; data collection from Korean speakers is ongoing. Stimuli were three CV syllables: [ata], [ala], and [aɾa]. The MMN block adopted a roving-standards paradigm, where participants heard randomly presented stimulus trains each consisting of repetitions of either [ala] or [ata] as standards followed by the deviant [aɾa]. This design allows direct comparison of the same deviant in different standard contexts. MMN amplitudes for [aɾa] were computed for both standard contexts by averaging over frontocentral electrodes within the time window (160-324ms) identified by a temporal cluster-based permutation test. In the PRP block, 70 repetitions of each of the three syllables were randomly presented. PRPs across all electrodes from 100 to 400ms were submitted to hierarchical clustering using Euclidean distance to assess neural similarity. Predictions: If phonetic encoding is shaped by language-specific phonological patterning, English listeners should show a larger MMN for [aɾa] following [ala] (phonological contrast) than following [ata] (allophonic variation), and greater PRP-based similarity between [ta] and [ɾa]. For Korean, we expected the reverse MMN asymmetry and a PRP clustering grouping [la] and [ɾa]. Results (English): MMNs were elicited for deviant [aɾa] in both standard contexts, with a significantly larger MMN when [aɾa] followed [ala], consistent with [ɾ] patterning phonologically with [t] in English. Nonetheless, PRP-based clustering showed greater neural similarity between [la] and [ɾa], with [ta] forming a distinct cluster, consistent with findings that PRP tracks phonetic features, particularly manner of articulation (Khalighinejad et al., 2017). The results suggest that language-specific phonological patterning shapes phonetic encoding (as indexed by MMN) but does not neutralize the influence of phonetic features (as indexed by PRP). If this pattern holds, we expect a reversed MMN asymmetry in Korean, with a larger MMN for [aɾa] following [ata] than following [ala], while PRP continues to show a greater neural similarity between [la] and [ɾa]. Together, these findings support the view that different levels of linguistic representation are underpinned by distinct neural processes.
Topic Areas: Speech Perception, Phonology