The effect of the speech task characteristics on perceptual judgment of mild to moderate dysphonia: a methodological study.

Objective: to study the differences of perceptual ratings of mild and moderate dysphonia related to the speech task, and their impact on intrarater and interrater reliabilities. Patients and Methods: Voice recordings of fifteen outpatients with mild or moderate dysphonia related to laryngopharyngeal reflux were presented to 6 female experienced judges blinded to the clinical state of patient recordings. From these, the GRBASI evaluations were performed on connected speech and sustained vowel of the pretreatment voice recordings and absolute agreement, and bot intrarater and interrater reliabilities were assessed. Results: The average GRBASI scores were significantly worse when performed on sustained vowel. Intrarater reliability substantially varied according to the judge and the task. Good interrater reliability was broadly found for the evaluations of all GRBASI components irrespective to the speech task. Concerning agreement, we only found absolute agreement between judges for G and R items assessed on text. Conclusion: Average grade of perceptual voice impairment, intrarater reliability and agreement vary according to the speech task