Electroencephalographic Classification Reveals Atypical Speech Motor Planning in Stuttering Adults

Area of Science:

• Neuroscience
• Speech-Language Pathology
• Computational Neuroscience

Background:

• Stuttering, also known as stammering, is a speech disorder affecting speech fluency.
• Understanding the neural underpinnings of speech motor control is crucial for developing effective interventions.

Purpose of the Study:

• To investigate differences in neural activity related to speech motor planning between adults who stutter (AWS) and adults who do not stutter (ANS).
• To test the hypothesis that AWS exhibit reduced separability in neural signals during speech motor planning tasks.

Main Methods:

• Utilized electroencephalographic (EEG) signals recorded during speaking and silent reading tasks.
• Applied machine learning algorithms, specifically kernel support vector machines, to classify neural activity.
• Employed a holistic approach to analyze cortical surface activity and classifier complexity.

Main Results:

• Classifiers distinguishing speech motor planning from silent reading required more complex decision boundaries for AWS compared to ANS.
• This indicates that EEG signals associated with speech motor planning are less discernible in AWS.

Conclusions:

• AWS demonstrate altered neuronal dynamics affecting speech motor planning, leading to lower inherent separability of neural conditions.
• Findings support the hypothesis of reduced neural signal distinctiveness in AWS and may inform the development of new speech therapies or assistive technologies.