RESUMEN
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.