Voice quality ratings: Empirical evidence for a new model.

Abstract

This model of voice quality ratings suggests that context effects may cause drift in listeners’ ratings by influencing the internal standard against which judgments are made. To test this hypothesis, we asked 12 clinicians to judge the roughness of 22 synthetic stimuli using two scales: a traditional five-point equal-appearing interval (EAI) scale and a scale with explicit anchor stimuli (also synthetic) for each scale point. The stimulus set included a relatively large number of near-normal voices; we predicted that this would produce an increase in the perceived roughness of moderately rough stimuli over time for the EAI ratings (by influencing an unstable internal standard), but not for the explicitly anchored protocol. As predicted, ratings on the unanchored EAI scale, but not the anchored scale, drifted significantly within a listening session in the direction expected. Ratings using the anchored protocol were also significantly more reliable than those gathered using the unanchored paradigm. These results are consistent with our model, and suggest that explicitly anchored protocols for voice quality evaluation might improve both research and clinical practice.