Content area
Full text
The Speech in Noise (SIN) test consists of a series of Institute of Electrical and Electronic Engineers sentences presented in a background of four-talker babble at two presentation levels (83 and 53 dB SPL) and four signal-to-noise ratios (SNR) (15, 10, 5, and 0 dB). In this study, the nine lists were tested for equivalency and the test-retest reliability was determined. Twenty listeners with normal hearing and 20 listeners with sloping sensorineural hearing losses served as subjects. Five sentences were presented at each SNR at each level, and five key words in each sentence were scored (100 key words per presentation level). Each key word was scored as correct or incorrect, with errors of plurality scored as half-correct words. For percent-correct scores, Lists 1, 2, and 9 and
Lists 3, 4, and 5 were found to be equivalent for listeners with normal hearing. For both groups of listeners, the test-retest correlations were high, and the critical differences appropriate for sentence material were 10 tol6% at a .95 level of confidence. Because of floor and ceiling effects across the different lists, many subjects with normal hearing did not score as low as the 50% level, and many subjects with hearing loss did not score as high as 50%. Suggestions are offered for alternate scoring in order to obtain a SNR for 50% performance. Future versions of this test should be designed with improved list equivalency and 50% performance capability.
Key Words: speech audiometry, list equivalency, reliability
The Speech in Noise (SIN) test was originally developed as a speech recognition task to be used in an investigation of the effects of compression ratio on speech intelligibility and quality (Fikret-Pasa, 1993). A compact disk version of the test is currently available (Etymotic Research, 1993). A corpus of 72 lists of 10 sentences (Institute of Electrical and Electronic Engineers [IEEE], 1969) was reduced to nine lists of 40 sentences each by Fikret-Pasa (see Appendix). These sentences are presented in a background of four-talker babble (three women and one man), at two levels (83 and 53 dB SPL) and four signal-to-noise ratios (SNRs; 15, 10, 5, and 0 dB). The sentences and the competing noise are recorded on a single channel; as a result, the SNRs are not...