Articulation functions were generated on a normal-hearing population with the Auditec of St. Louis cassette recordings of the NU-6 and CID W-22 speech discrimination tests. Both tests were similar and yielded slopes of about 4.4%/dB. Each gave a speech discrimination score of approximately 95% at 32 dB SL. Speech reception thresholds were obtained with monitored live voice and yielded good test-retest consistency. Speech thresholds were about 9 dB better than the ANSI (1969) specifications.

Using three speakers and 75 subjects, a 7.90dB range in the relative intelligibility of the Central Institute for the Deaf (CID) spondees was found when they were presented via monitored live voice. This range is consistent with previous research and is considered unnecessarily large and detrimental to the efficiency and precision of the speech reception threshold (SRT) test. The exclusive use of 18 words that have a mean range of 1.5 dB is suggested. The slope of the 3l-item spondee articulation function is 12 percent per decibel and is slightly steeper than previous research using the CID W-1 recorded spondees. The slope for the suggested 18-item test is also about 12 percent per decibel. For the 36-item test the mean speech detection threshold (SDT) was 6.6 dB sound pressure level (SPL) and the mean SRT was 14.2 db SPL. The 7.6-dB difference between the two speech measures is consistent with previous research but the SRT is about 6 dB better than the 20 dB SPL recommended by ANSI 1969. For the suggested 18-item test, the mean SRT was 14.5 dB SPL, which was 7.9 dB above SDT. When the 18-item test is used, average normal hearing can be expected to be about 5.5 dB better than specified by ANSI 1969.

No abstract