Disfluency in spasmodic dysphonia: a multivariate analysis

This study examined visual analog scaling (VAS) judgments of disfluency by normal listeners in response to oral reading by speakers with spasmodic dysphonia (SD) and by nondysphonic controls, as well as the variables of frequency of occurrence of disfluencies, speaking rate, number of reading errors, and temporal acoustic measures of interword interval duration and articulation time. MANOVA yielded statistically significant differences between SD and control speakers for all variables except reading errors. Although no significant fluency-related differences were observed in terms of type of vocal spasm or voice tremor, significant differences in disfluency measures were obtained for clinical ratings of severity of dysphonia. Greater dysphonia severity ratings were associated with decreased fluency, but milder ratings were not necessarily associated with disfluency. Stepwise multiple regression analysis demonstrated that frequency of disfluency occurrence, speaking rate, and reading errors accounted for more than three fourths of the variability in VAS judgments of disfluency. Findings suggest that although disfluency is not a defining feature of SD, it does contribute significantly to the overall clinical impression of severity of the disorder.