Content area
Full text
Several studies demonstrate that thresholds for the individual CID W- 1 spondaic words peaked at 0 vu are not equivalent. The purpose of this study was to equate the spondaic word thresholds psychometrically. Two studies were performed on 2 groups of 20 listeners with normal hearing. In Experiment 1, psychometric functions were established for the 36 spondaic words spoken by a male (original recording) and female speaker. Based on the threshold data from Experiment 1, the words spoken by the female speaker were adjusted digitally in level to produce equal thresholds (equal intelligibility). In Experiment 2, psychometric functions then were established for the 36 spondaic words adjusted in level. The mean thresholds for the 2 experiments were the same (0.5 dB HL; ANSI, 1996), but the standard deviations for the word thresholds in Experiment 2 (0.7 dB) were significantly smaller than the standard deviations in Experiment 1 (1.6 dB). Both versions of the spondaic words spoken by the female speaker are included on the Speech Recognition and Identification Materials (Disc 2.0) compact disc.
KEY WORDS: spondaic-word threshold, equivalency, speech perception, auditory perception, compact disc, psychometric function, homogeneity, rms
When establishing a pure-tone threshold, all parameters of the signal are the same on each trial (presentation), except the level of the signal that is the variable. Even in the level domain, pure tones from trial to trial presented at the same level are in fact at the same level. The spondaic words typically used to establish thresholds for speech do not have the same qualities of homogeneity that are characteristic of pure tones. Each of the 36 CID W-1 spondaic words has different spectral, temporal, and level characteristics. The benchmark characteristic for spondaic words has been that both syllables of the word peak at 0 vu, which for presentation and recording purposes is a convenient reference point. As Hirsh et al. (1952) and numerous subsequent studies have demonstrated, spondaic words that peak at 0 vu are not homogeneous with respect to the thresholds for the individual words (Bilger, Matthies, Meyer, & Griffiths, 1998; Bowling & Elpern, 1961; Cambron, Wilson, & Shanks, 1991; Curry & Cox, 1966; Young, Dudley, & Gunter, 1982). Improving the homogeneity of the thresholds for the individual spondaic words would produce...