关键词: 言语测听, 总体信度, P-I曲线, 临界差值

Abstract: Objective To evaluate the reliability of the mandarin lexical neighborhood test(M-LNT), obtain the performance-intensity curve, speech recognition threshold and critical difference, and establish basic curve for the clinical use. Methods Fifty-two normal hearing adults were recruited to evaluate the reliability of M-LNT which consists of four types of list: monosyllable-easy word lists, monosyllable-hard word lists, disyllable-easy word lists and disyllable-hard word lists. Six intensities were chosen from pre-experiments. According to the randomized block design, for every list 3 intensities of which number is odd or even of the six selected ones were presented to subjects. The test sequence is from easy lists to hard. After calculating the scores of every intensity for every list, the Matlab 7.1 software was used to do the P-I curve fitting the scores and calculate the curve's slope, speech recognition threshold and critical difference. Results Statistical analysis showed that there was no significant difference between any two scores of monosyllable-easy word lists, scores of monosyllable-hard word lists, the scores of disyllable-easy word lists and the scores of disyllable-hard word lists(P>0.05). The P-I curves of monosyllable-easy word lists, monosyllable-hard word lists, disyllable-easy word lists and disyllable-hard word lists were analyzed; the speech recognition threshold was respectively 25 dB SPL, 31 dB SPL, 20 dB SPL and 24 dB SPL; the 50% slope was respectively 2.8%/dB SPL, 3.0%/dB SPL, 5.8%/dB SPL and 4.1%/dB SPL; the critical difference at 95% confidence level was respectively ±22.2%, ±26.1%, ±32.7% and ±29.9%. Conclusion After the evaluation of M-LNT, the critical difference at 95% confidence level of four types of lists was respectively ±22.2%, ±26.1%, ±32.7% and ±29.9% which indicate the lists' reliability. The P-I curves were fitted by the scores at different intensities and the speech recognition scores basic curve was established.

Key words: speech test, reliability, P-I curve, critical difference