口头复述与书写反应方式对普通话言语识别率的影响

Impact of Repeat Versus Written Response on Mandarin Speech Recognition Scoring

目的观察口头复述与书写反应方式在普通话单音节小词表言语识别率测试中产生的误差大小,探讨不同反应方式对普通话言语测试的影响。方法在6种给声强度(20、15、10、5、0、-5dBHL)下针对6张普通话单音节小词表测试12例正常听力受试者的言语识别率。受试者以口头复述并同时书写记录所听到的测试内容,测试者以听觉监测记录受试者反应。通过对比受试者和测试者的记录,获得在不同声强级下不同反应方式对单音节词言语识别率得分的误差大小。结果口头复述与书写反应记录的言语识别率得分高度相关(P<0.001)。72对言语识别率测试记录中,有19对(26.4%)不一致。口头复述与书写反应的误差范围为-5.0%～10%。在给声强度为5dBHL时口头复述与书写反应的言语识别率误差有统计学意义(P<0.05),而在给声强度为20、15、10、0dBHL时误差无统计学意义。给声强度为-5dBHL时,口头复述与书写反应完全一致。结论不同反应方式影响言语测试结果,建议在开放项列言语测试中,受试者在复述测试内容时,同时进行书写记录,并与测试者听觉监测记录对照,以减少误差的产生,提高测试信度。

Objective To observe differences between repeat and written response in speech recognition tests of short Mandarin monosyllable lists. Methods 12 normal--hearing subjects were tested by 6 short monosyllabic lists. The subjects should repeat and write down the contents of the tests at the same time, the tester monitors au- rally and records the subjects＇ reactions. The speech recognition scoring error under different responses at different presentation levels could be obtained by comparing the records of the subjects with the tester. Results The speech recognition scoring of repeat and written response were highly correlated with each other （P〈0.01）. However, 19 （26.4%） out of 72 pairs of test records were inconsistent. The error range of them was from -5% to 10%. When the presentation level was 5 dB HL, the difference among the speech recognition scoring was significant statistically, but not significant when the presentation level at 20, 15, 10 and 0 dB HL. Conclusion The test results of speech audiometry are impacted by different respond methods. In the open--set test, it＇s supposed that the subjects should make notes while repeating the contents of the tests, then compare with the aural monitoring records of the tester to keep the reliability of the test.