Abstract
The intelligibility of Japanese vowel was investigated in normal-hearing listeners using stimuli whose spectra were intended to mimic the effects of reduced frequency selectivity. The vowel stimuli used in the experiment were constructed with systematically flattened spectra in the region of the first-third formants. Vowel identification tests were conducted in both noise-free and noise-added conditions. The difference between these two conditions was examined as a function of the formant bandwidths. Noise condition caused statistically significant reductions of mean percent correct identification as compared to quiet condition with broadening formant bandwidths. In quiet condition, the vowels /a/ and /o/ evidenced the poorer discrimination, compared to the vowels /i/, /e/, and /u/. Less frequency separation of the first and second formants seemed largely responsible for the reduced identification in these vowels. In noise condition, vowels were primarily confused where the proximity of first formant is concerned. Vowel errors were similar to those of the listeners demonstrating high-frequency hearing loss. These results suggested that the ear's abnormal selectivity in frequency may be of major influence on the vowel identification in the presence of interfering sounds.
