Use of the Connected Speech Test (CST) with hearing-impaired listeners

Two studies were performed in which hearing-impaired subjects responded to the Connected Speech Test (CST). In experiment 1, 40 subjects, divided into four groups according to extent and configuration of hearing loss, responded to the CST version 1 (CSTv1). This version of the test consisted of 57 passages of connected speech: 48 test passages and 9 practice passages. It was developed on the basis of data for normal-hearing listeners. Performance of hearing-impaired listeners for the CSTv1 revealed that, although the passages were equal in average intelligibility for normal hearers, they were not equally intelligible for hearing-impaired persons. Based on results of data analyses, the 57 passages were reconstituted into 28 pairs of passages: 24 test pairs and 4 practice pairs. The pairs were equal in average intelligibility for both normal and hearing-impaired listeners. This form of the test was named the CST version 2 (CSTv2). In experiment 2, an additional 23 hearing-impaired subjects responded to the CSTv2. Critical differences and the slope of the signal to babble ratio (SBR) function were determined for the CSTv2 for hearing-impaired listeners. When two CSTv2 pairs were used per score, the 95% critical difference for hearing-impaired subjects was about 15.5 rationalized arcsine units (rau). The mean SBR function slope for hearing-impaired listeners was 8.5 rau/dB. Comparing the critical difference with the SBR function slope, it may be seen that, for hearing-impaired listeners, differences in intelligibility equivalent to a 2 dB change in SBR can be detected with CST scores based on mean performance across two passage pairs.