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

 
Poster B53, Wednesday, November 8, 3:00 – 4:15 pm, Harborview and Loch Raven Ballrooms

Auditory Cortex Represents Abstract Phonological Features: A Mismatch Negativity Study of English Voicing

Philip Monahan1, Jessamyn Schertz1;1University of Toronto

Understanding how speech sounds are represented in auditory cortex has important implications for neuro-cognitive models of spoken word recognition (Hickok, 2014; Hickok & Poeppel, 2007; Obleser & Eisner, 2009). Neurophysiological evidence for abstract phonological features has been equivocal. While recent EEG/MEG findings provide support for monovalent distinctive features (Eulitz & Lahiri, 2004; Hestvik & Durvasula, 2016; Scharinger, Monahan, & Idsardi, 2016; Schluter, Politzer-Ahles, & Almeida, 2016b), participants were not required to abstract over distinct phonetic implementations of a phonological feature, leaving open the possibility that response patterns were driven by acoustic, not phonological, similarity. We report mismatch negativity (MMN) findings that suggest a representation for the abstract phonological feature [voice] that encompasses acoustically distinct implementations in English: voice onset time (VOT) in stop consonants and laryngeal excitation in fricatives. Moreover, results suggest that only voiceless sounds have a stored feature for voicing. Eighteen native English speakers heard two blocks of naturally-produced consonant-vowel syllables (voiced /ba, da, ga, va, za/, voiceless /pa, ta, ka, fa, sa/), one block with voiced standards and voiceless deviants, and the other with voiceless standards and voiced deviants in an oddball paradigm. Each block also included sham deviants that shared the same voicing as the standard but differed in their vowels (e.g., voiced /vu/, voiceless /po/). EEG recordings were acquired with a 32-channel system (Brain Products GmbH, Germany). Mean ERP amplitudes averaged over fronto-central electrode sites were submitted to a linear mixed effects model with the fixed effects Condition (Standard, Deviant, Sham) and Block (Voiced Standard, Voiceless Standard). Analyses were performed in three time-windows (Early: 100-170ms, Middle: 170-250ms, Late: 250-550ms). Significant Condition by Block interactions were found in all three time windows. Pairwise interaction comparisons were performed to assess the effect of Condition in each block separately. Early window: In the Voiceless Standards block, there was a difference between Standards and Deviants (X2(1)=10.9, p < 0.01). Middle window: In the Voiceless Standards block, there was again a difference between Standards and Deviants (X2(1)=7.78, p < 0.05). In the Voiced Standards block, the Sham condition was different from both Standard (X2(1)=29.37,p<0.001) and Deviant (X2(1)=22.55,p<0.001) conditions. Late window: In the Voiced Standards block, the Sham condition was different from both Standard (X2(1)=10.97,p<0.01) and Deviant (X2(1)=14.69,p<0.001) conditions. The Sham stimuli, which differed from Standards and Deviants in vowel quality, elicited differences only in the Late window. This is expected given that acoustic cues to the vowel emerge later. Standards differed from Deviants, however, in earlier windows, as consonantal information emerges earlier. Given that Standards consisted of tokens with distinct articulatory and phonetic implementations, unified only by their phonological category, the systematic response to Deviants suggests that listeners construct an abstract phonological representation of the standards, including both stops and fricatives. An MMN was observed only when the standard was voiceless, suggesting that only voiceless sounds have a stored feature for voicing (see Lahiri & Reetz (2010)), consistent with recent MMN findings (Hestvik & Durvasula, 2016; Schluter, Politzer-Ahles, & Almeida, 2016a) and primary linguistic research (Avery & Idsardi, 2000).

Topic Area: Phonology and Phonological Working Memory

Back to Poster Schedule