Neural pathways underlying the production of pitch and rhythm in aphasia.

Singing is a universal human attribute. Previous studies suggest that the ability to produce words through singing can be preserved in poststroke aphasia (PSA) and that this is mainly subserved by the spared parts of the left-lateralized language network. However, it remains unclear to what extent the production of rhythmic-melodic acoustic patterns in singing remains preserved in aphasia and which neural networks and hemisphere(s) are involved in this. In this cross-sectional study, we set out to investigate the structural neural networks underpinning singing production abilities by combining a whole-brain white matter correlational tractography approach together with a comprehensive appraisal of pitch, melodic contour, and rhythm production accuracy during both spontaneous and cued singing in a sample of 45 patients with PSA. Our results indicate that PSA patients have poorer singing accuracy (pitch, melodic contour, and rhythm) than matched healthy controls (N = 33). The network associated with singing accuracy in aphasia was identified in the left hemisphere-dominant dual stream network involved in auditory-motor integration of speech, but also extends to multiple other associative and projection pathways, also in the right hemisphere. The results provide insight into alternative communication methods and therapeutic approaches, leveraging music's inherent structure to aid in language recovery and rehabilitation.