Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural dynamics of stuttered versus fluent speech
Poster Session B, Friday, September 12, 4:30 - 6:00 pm, Field House
This poster is part of the Sandbox Series.
Neeraj Kumar1, Joan Orpella1; 1Georgetown University Medical Center
Stuttering is a neurodevelopmental communication disorder affecting approximately 80 million individuals worldwide, characterized by involuntary disruptions in speech fluency, including sound and syllable repetitions, prolongations, and articulatory blocks [1,2]. Converging evidence links stuttering to dysfunctions in the basal ganglia-thalamocortical (BGTC) loop, a cortico-subcortical network implicated in the initiation and temporal sequencing of speech motor programs [2,3]. However, the precise neurophysiological mechanisms by which BGTC dysfunction leads to speech disfluencies remain poorly understood. Here we investigate the neural dynamics within the BGTC network during stuttered vs. fluent speech via a functional connectivity analysis of magnetoencephalography (MEG) data. While the assessment of neural dynamics using MEG is relatively straightforward, its application to stuttered speech suffers from two main issues. First, it is notoriously difficult to elicit a balanced number of stuttered and fluent trials inside the scanner to allow for meaningful comparisons between these conditions. Second, overt speech production introduces significant muscle and movement-related artifacts, which can be exacerbated during stuttered speech due to common accessory behaviors such as rapid eye blinking and lip tremors. To elicit a balanced number of stuttered and fluent speech, we will use a paradigm previously validated on both adults and children who stutter [4,5,6]. In a nutshell, PWS can predict stuttering on certain words with high accuracy (> 85%); the approach consists of interviewing PWS for words on which they anticipate stuttering and using those words to generate prompts that will be presented during the MEG session, each prompt requiring a response containing one of the anticipated words. To mitigate the effect of speaking-related artifacts, we will use concurrent electromyography and speech audio recordings. Artifact reduction will be achieved using a combination of regression-based techniques and signal space projection methods, enabling the removal of artifacts from the MEG data arising from non-neural sources [7]. MEG data during the speech production task will be acquired from a matched sample of adults who stutter (N=30) and typically fluent controls (N=30). Diagnoses and stuttering severity will be confirmed by a speech-language pathologist using standardized clinical protocols (SSI-4). To assess the temporal dynamics within the BGTC loop during stuttered and fluent speech, we will perform source localization via beamforming to reconstruct neural activity and use phase-based measures in these key speech-related regions to characterize connectivity profiles for stuttered speech and for fluent speech in both controls and PWS. We hypothesize significantly different profiles for stuttered and fluent speech and no differences between the fluent speech of PWS and controls. The direction of putative information flow for each condition will be confirmed using Granger causality. References: 1. Stuttering . https://www.nidcd.nih.gov/health/stuttering ; 2. Alm, P. A. (2004), J Commun Disord ; 3. Chang et al., (2019), Neuroscientist ; 4. Jackson et al., (2019), J Fluency Disord ; 5. Orpella et al., (2024), Neurobiol Lang (Camb) ; 6. Goldfarb et al., (2023), J Speech Lang Hear Res ; 7. Abbasi et al., (2021), Front Neurosci
Topic Areas: Disorders: Developmental, Language Production