A mechanistically informed framework of hierarchical temporal speech organization for objective differential assessment of motor speech disorders in neurodegenerative diseases.

Introduction: Speech is organized into discrete utterances and pauses in accordance with underlying linguistic structure, temporal coordination, prosodic organization, and respiratory demands. Temporal speech characteristics, including rhythmic and pausing patterns, are governed by multiple neurophysiological mechanisms and provide valuable insights for the assessment and management of motor speech disorders. This study employed a mechanistically informed analytic framework of hierarchical temporal speech organization to derive interpretable, objective measures for assessing rhythmic and pausing disturbances in neurodegenerative diseases. Methods: Orofacial kinematic and acoustic recordings were obtained from two neurodegenerative disease groups-amyotrophic lateral sclerosis (ALS) and Parkinson's disease (PD)-along with a neurologically healthy control (HC) group, during a passage reading task. Using automated analytic procedures, kinematic rhythm measures (movement-based) and acoustic rhythm measures (sound-based) were derived to characterize hierarchical rhythmic modulation of articulatory movements and critical-band acoustic modulation envelopes at prosodic, syllabic, and sub-syllabic levels. Pause measures were derived to characterize within- and between-sentence pausing patterns. Rhythmic and pausing patterns were compared across groups. The acoustically derived rhythm and pause measures were subsequently subjected to (1) discriminant analyses for multiclass classification among the ALS, PD, and HC groups, and (2) regularized regression models to evaluate their associations with functional outcomes. Results: Both neurodegenerative disease cohorts exhibited a reorganization of hierarchical rhythmic modulation characterized by reduced prosodic-level modulation and increased syllabic-level modulation. Such changes were, however, driven primarily by articulatory impairment in ALS and by respiratory-laryngeal impairment in PD. In addition, both cohorts showed reduced regularity of intra-syllabic temporal organization and altered within-sentence pausing patterns, whereas the ALS cohort exhibited additional changes in between-sentence pausing. Measures capturing these rhythmic and pausing disturbances demonstrated promising classification performance (mean accuracy = 0.77; mean area under the curve [AUC] = 0.84) and meaningful associations with functional speech decline across diseases. Discussion: The rhythm and pause measures captured disease-specific subclinical changes in the physiological substrates underlying rhythmic and pausing disturbances in ALS and PD, demonstrating potential as clinically applicable, objective markers to enable more accurate differential diagnosis, targeted intervention, and measurement-based care for neurodegenerative motor speech disorders.