1 Dysrhythmic Speech is a Characteristic of Developmental Stuttering in Adults: A Quantitative Analysis Using Duration- and Interval-Based Rhythm Metrics

Purpose: Aberrant speech rhythm has previously been identified as a hallmark of fluency disorders, including stuttering. However, evidence of dysrhythmic speech in adults who stutter (AWS) has largely been limited to qualitative research. Here, we conduct a quantitative analysis of speech rhythm in AWS and non-stuttering peers. We also investigate potential differences in speech rhythm as a likely marker of articulatory stability in speech preceding moments of stuttering-like disfluency versus not preceding moments of stuttering-like disfluency. We hope to establish rhythm of speech as a diagnostic criterion for the assessment of stuttering. Methods: Interval and duration-based rhythm metrics (nPVI, rPVI, %V, ΔC, ΔV, VarcoC, VarcoV) were computed for speech produced by 16 self-identified AWS and 16 adults who do not stutter (AWNS). Acoustic speech data were recorded in reading and spontaneous speaking tasks. Following previous literature, we hypothesized that the non-stuttered utterances of AWS would be less rhythmic than those of AWNS. Furthermore, it was expected that speech rhythm would be disrupted immediately preceding the production of stuttering-like disfluency in AWS. We conducted statistical tests to determine differences in speech rhythm metrics between groups (i.e., AWS and AWNS) and utterance types (i.e., preceding stuttering-like disfluency and not preceding stuttering-like disfluency). Additionally, a classification model was trained to predict AWS vs. AWNS identification based on speech rhythmicity. Results: Multiple metrics revealed reduced speech rhythmicity in AWS compared to AWNS for both non-stuttered oral reading and spontaneous speech tasks. Further, speech rhythmicity in AWS was significantly reduced immediately preceding stuttering-like disfluency compared to not preceding stuttering-like disfluency in both the reading and spontaneous speech tasks. The classification model was able to discriminate between AWS and AWNS, with 71.3% precision and 70.1% recall. Conclusion: We present quantitative evidence of reduced speech rhythm as a characteristic of developmental stuttering. AWS exhibited relatively less rhythmic speech overall and particularly immediately preceding a stuttering-like disfluency. This reduction in the temporal stability of speech may be interpreted as a sign of imminent breakdown in articulatory coordination. The potential significance of rhythm metrics to speech science and clinical diagnostics is discussed.