Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Cortical Tracking of Speech Meter during Natural Storytelling
Poster Session C, Saturday, September 13, 11:00 am - 12:30 pm, Field House
Cyrille Magne1, Camila Zugarramurdi2, Eleonora Beier3, Katsuaki Kojima4, Stephanie Powell1, Jonathan Liu5, Kristin Davis1, Kylie Korsnack6, Brett Myers7, Miriam Lense5, Srishti Nayak5, Reyna Gordon5, Yulia Oganian8; 1Middle Tennessee State University, 2Universidad de la República, 3University of California, Davis, 4University of Cincinnati College of Medicine, 5Vanderbilt University Medical Center, 6University of Richmond, 7University of Utah, 8University of Tübingen
Syllabic stress provides critical acoustic and pragmatic cues during speech comprehension. Moreover, the alternation of stressed and unstressed syllables defines speech meter, which is known to facilitate speech recognition. Acoustically, in English, stress is defined by patterns of intensity, pitch, and duration. Although it has been hypothesized that stressed syllables may drive neural entrainment to the speech signal, empirical evidence for the effects of stress and meter on neural responses to speech is scarce. Recent work has revealed that the human speech cortex tracks syllables by responding to moments of rapid increase in speech intensity – peakRate events – which occur close to vowel onsets. Based on this acoustically defined marker of syllable processing, here we examine how stress and metric regularity influence 1) evoked neural responses to syllables and 2) oscillatory phase-locking to peakRate events. We recorded EEG while participants (n=26) listened to short children’s stories. Participants heard one story that followed an irregular, speech-like meter typical of everyday speech, and another story that followed a regular meter with alternating stressed and unstressed syllables (x - -). Using temporal response function (TRF) modeling, we isolated the effects of syllabic stress and metric regularity on neural responses to peakRate events. Both stress and meter modulated early (within 50-150 ms) cortical responses to peakRate events. Specifically, stress enhanced neural responses to stressed syllables as compared to unstressed syllables, suggesting that stress is identified early during auditory processing. Moreover, while early responses to louder syllables were enhanced overall, this effect was attenuated in the metrically regular condition, suggesting a reduced reliance on acoustic cues in favor of processing based on top-down metric expectations. Furthermore, we found that neural responses to unstressed syllables of larger intensity were temporally more extended (300-400ms) than for stressed syllables of similar intensity. In the time-frequency domain, neural phase-locking around 2 Hz (corresponding to stress rate in the stimuli) was stronger for stressed than for unstressed syllables. Such phase-locking patterns have been previously interpreted as endogenous oscillatory entrainment. However, an explicit model of evoked neural responses showed that this phase-locking pattern reflects the frequency and shape of evoked responses. This result directly shows that delta-phase locking to stressed syllables does not necessarily reflect oscillatory entrainment. It also provides a novel approach bridging between analyses of evoked responses and phase-locking to continuous speech. Overall, our findings reveal that neural responses to individual syllables integrate acoustic cues with perceived stress patterns and metric expectation. These neural dynamics may underlie the perceptual benefits of metrically regular speech, such as poetry.
Topic Areas: Prosody, Speech Perception