Neural Processing of Speech Signals in Children Who Stutter

点击翻译按钮获取中文摘要

PROJECT SUMMARY/ABSTRACT Developmental stuttering is a dynamic, multifactorial neurodevelopmental disorder characterized by unintended disruptions in fluent speech production. Speech planning and production rely on intact speech sound processing, which helps develop and maintain internal speech sound models. Unstable internal speech sound models, which regulate motor signals in the speech motor articulatory network (SMAN), may contribute to disfluent speech in children who stutter (CWS). In concert with frontoparietal attention network, SMAN also modulates attention to phonetic/syllabic information in speech, particularly in difficult listening conditions. CWS often perform worse on speech processing tasks than fluent peers, especially on more challenging tasks, potentially due to inefficiencies in these auxiliary networks. However, the underlying causes of speech processing deficits in CWS remain unclear. A mechanistic understanding of speech sound processing will facilitate future development of neurobiologically informed stuttering interventions that target the specific neural deficits in CWS. The current proposal extends previous findings of atypical speech sound processing in CWS. Combining the complementary expertise of a cross-disciplinary team of investigators, the current project will evaluate the integrity of neural processes underlying speech sound encoding and the ways in which these processes are modulated by task demands using multimodal neuroimaging and systems-level computational modeling approaches. Aim 1 will measure electroencephalography (EEG) in 150 CWS and 150 fluent peers, aged 7-15 years, while children complete four tasks of varying difficulty: A) a syllable identification task (/ba/ vs /da/) in quiet; B) a continuous speech narrative comprehension task in quiet; and C & D) complex speech encoding tasks with syllables and continuous speech presented simultaneously, with attention directed either toward syllables (C) or toward the narrative (D). Directly comparing neural responses elicited in simpler and more complex listening conditions (A/C, B/D) and responses to the same stimuli when attended vs. ignored (C/D) is critical for characterizing effects of task demands on speech sound processing. State-of-the-art machine-learning approaches for EEG will enable simultaneous extraction of temporally precise neural representations of fast and slow temporal fluctuations in speech in the transformation from acoustic to syllable representations. Aim 2 will leverage functional MRI (fMRI) to assess multiple neural systems underlying speech sound processing in CWS. Employing the same tasks in the same participants as Aim 1 will allow for quantifying neural activations and representations in auditory, SMAN, and attention networks during simple and complex speech tasks. Aim 3 will develop a systems-level computational model of speech sound processing in CWS. The model, based on combined EEG and fMRI data, will simulate how interactions between neural networks mediate task performance across listening conditions. This project will provide a mechanistic understanding of speech sound processing in CWS and a unique, curated, open access, multimodal neuroimaging dataset that will be a lasting resource for the field of stuttering.