Abstract
To quantitatively evaluate hypernasality, pronunciation of the Japanese vowel /i/ by eighteen patients diagnosed as having cleft palate or congenital velopharyngeal incompetence and seventeen normal subjects (controls) was analyzed acoustically by cepstrum analysis. Spectrum envelopes obtained by the cepstrum method were evaluated every 1/3 octave to obtain the mean level of each band.
Twenty listeners evaluated twenty-six speech samples for hypernasality on a five interval scale of 0 to 4.
Relations between the first perceptual factor of hypernasality and the levels of 1/3 octave bands wer e examined. The results were as follows:
1. Spectral characteristics of hypernasality were the raising of the level between the first and second formants and reduction of the level in the second and the third fomant regions.
2. Two factors were obtained by factor analysis of the judged scores. The first factor, which accounted for 75% of the total variance, was the consensus perception of hypernasality. The second factor, whi ch accounted for 7%, was the individual listeners perception of hypernasality differences.
3. Two kinds of average levels of three 1/3 octave bands were highly correlated with the first perceptual factor of hypernasality. One was the mean level of the 2/3 octave, the 1 octave, and the 4/3 octa ve above the first formant. The other was the mean level of the 3 octave, the 10/3 octave, and the 11/3 octave above the first formant.
