Abstract: Objective To assess the test-retest reliability of monosyllable word short lists in Mandarin Speech Test Materials(MSTMs). Methods The sets of monosyllable word short lists were used to test the speech perception score among sixteen normal-hearing subjects utilizing the Chinese Speech Intelligent Audiometric System, which was developed by the Institute of Biomedical Engineering, Beijing Tongren Hospital, Capital Medical University. The stimulus level was fixed at 12 dB HL. The second round was carried out after an interval of 6 to 35 days. The Pearson correlation coefficient r was calculated, significance level of the difference between test and retest was examined by Student's t test, and critical difference(CD) for 95% confidence interval(CI) was analyzed after "rationalized" arcsine transform with SPSS 19.0. Results Speech perception scores in tests and retests were significantly correlated with a Pearson's r value of 0.748(P<0.01) for the short lists. Standard deviation turned out to be 11.5%, thus the resulting value of CD was 22.5. Conclusion The intervention should be considered effective once the number of correct response presented by a subject is no less than 5 in a short list.

Key words: speech audiometry, speech recognition, test-retest reliability, monosyllable