An Item-Difficulty Based Speech Discrimination Test

Abstract

In an attempt to improve the differentiating ability of monosyllabic speech discrimination tests, two equivalent, 25-word test recordings with rectilinear distributions of item difficulty were constructed from W-22 Hirsh recordings. These lists and selected half lists of the standard W-22 recordings were then presented to 40 ears of listeners with losses ranging in severity from -10 dB to 60 dB (SRT) at sensation levels of 20, 30, 40, and 50 dB. Scores obtained with the new lists describe a less skewed frequency distribution than those obtained with the W-22 half lists. The new recordings were also more successful in differentiating between subjects with varying levels of sensorineural hearing losses, particularly between normal-hearing listeners and listeners with mild loss, who are usually poorly differentiated by W-22 tests. This was accomplished without resorting to deliberate distortion of stimulus words. These new recordings appear to have clinical usefulness in their ability to more accurately reflect differences in discrimination ability among listeners and particularly in their ability to reveal reduced discrimination in mild sensorineural hearing loss.