Beattie R C
California State University, Long Beach.
J Speech Hear Disord. 1989 Feb;54(1):20-32. doi: 10.1044/jshd.5401.20.
Word recognition functions for Auditec recordings of the CID W-22 stimuli in multitalker noise were obtained using subjects with normal hearing and with mild-to-moderate sensorineural hearing loss. In the first experiment, word recognition functions were generated by varying the signal-to-noise ratio (S/N); whereas in the second experiment, a constant S/N was used and stimulus intensity was varied. The split-half reliability of word recognition scores for the normal-hearing and hearing-impaired groups revealed variability that agreed closely with predictions based on the simple binomial distribution. Therefore, the binomial model appears appropriate for estimating the variability of word recognition scores whether they are obtained in quiet or in a competing background noise. The reliability for threshold (50% point) revealed good stability. The slope of the recognition function was steeper for normal listeners than for the hearing-impaired subjects. Word recognition testing in noise can provide insight into the problems imposed by hearing loss, particularly when evaluating patients with mild hearing loss who exhibit no difficulties with conventional tests. Clinicians should employ a sufficient number of stimuli so that the test is adequately sensitive to differences among listening conditions.
针对在多说话者噪声环境下CID W - 22刺激的Auditec录音，使用听力正常以及轻度至中度感音神经性听力损失的受试者获得了单词识别功能。在第一个实验中，通过改变信噪比（S/N）来生成单词识别功能；而在第二个实验中，使用恒定的S/N并改变刺激强度。正常听力组和听力受损组单词识别分数的分半信度显示出的变异性与基于简单二项分布的预测结果非常吻合。因此，无论单词识别分数是在安静环境还是在有竞争背景噪声的环境中获得，二项式模型似乎都适用于估计其变异性。阈值（50%点）的信度显示出良好的稳定性。正常听力者的识别函数斜率比听力受损受试者的更陡。在噪声环境中进行单词识别测试可以深入了解听力损失所带来的问题，特别是在评估那些在传统测试中没有困难的轻度听力损失患者时。临床医生应采用足够数量的刺激，以便测试对不同听力条件之间的差异具有足够的敏感性。
