Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural representations of speech sounds in children who stutter
Poster Session C, Saturday, September 13, 11:00 am - 12:30 pm, Field House
Ja Young Choi1, Zhe-chen Guo1, Ashley Parker2, Bharath Chandrasekaran1, Amanda Hampton Wray2; 1Northwestern Univeristy, 2University of Pittsburgh
Stuttering is a speech disorder characterized by involuntary repetitions, prolongations and blocks that disrupt the flow of speech. While stuttering is primarily characterized by speech production deficits, previous studies have reported considerable evidence that children and adults who stutter exhibit behavioral and neural patterns of speech perception divergent from fluent controls of the corresponding age group. Here, we hypothesized that neural representation of phonemes may be less distinct in children who stutter than controls of same age. Participants were composed of three groups -- children who stutter (age 7-12 years), their age-matched control peers and adults without speech, language or hearing disorders. As the participants listened to 15 minutes of naturalistic, continuous speech from the public domain audiobook Alice’s Adventures in Wonderland in quiet, we acquired their electroencephalographic (EEG) responses. We analyzed the preprocessed EEG data time-locked to 0 – 500 ms after the onset of phoneme instances (phoneme-related potentials; PRPs) in order to examine cortical representations of phonemes. For each participant, we analyzed the separability of their PRPs among different phonemes and among different manners of articulation (stop, fricative, nasal approximant, vowel) by computing the F-statistic on the PRPs over time. These F-statistics were compared between the groups of participants. Also, we trained neural network classifiers using EEGNet to predict phonemes from the PRPs. Over the duration of the PRPs, the F-statistics were generally lower for children who stutter than the control children and adults, showing that there is a less distinct separation between phonemes as well as between manner categories in the PRPs of children who stutter than fluent peers. Also, the classifiers trained to decode phonemes from the PRPs predicted phonemes less accurately and with more uncertainty in children who stutter than controls, further demonstrating that cortical representations of phonemes are less distinctive in children who stutter. The pattern of group differences in PRPs also varied between the manner categories, not only in terms of the magnitude but also timeframes within the duration of the PRPs when the difference was more pronounced. These differences suggest that the developmental trajectories of how different phonemes are represented in the brain real-time while listening to continuous speech vary between children who stutter and their fluent peers.
Topic Areas: Disorders: Developmental, Speech Perception