Reliability of individual differences in degraded speech perception

Abstract

Listeners' speech perception abilities vary extensively in challenging listening conditions. There is little evidence as to whether this variability is a result of true, stable individual differences or just variability arising from measurement error. This study examines listeners' word recognition abilities across multiple sessions and a variety of degraded speech tasks (noise-vocoded, time-compressed, and speech in babble noise). Participants transcribed isolated single syllable words presented in all three degradation types and repeated these tasks (with different words) on a separate day. Correlations of transcription accuracy demonstrate that individual differences in performance are reliable across sessions. In addition, performance on all three degradation types was correlated. These results suggest that differences in performance on degraded speech perception tasks for normal hearing listeners are robust and that there are underlying factors that promote the ability to understand degraded speech regardless of the specific manner of degradation. Uncovering these general performance factors may provide insight into the salient performance variance observed in listeners with hearing impairment.