To reveal the acoustic characteristics associated with hypernasality and to
ascertain their correlation to the severity of hypernasality, 30 speech sa
mples produced by 15 maxillectomy patients were acoustically analysed with
and without an obturator prosthesis in place. The isolated, sustained Japan
ese vowel /i/ was used as the stimulus for acoustic measurement and percept
ual judgment to evaluate the severity of hypernasality. Normalized 1/3-octa
ve spectral analysis demonstrated the spectral characteristics of hypernasa
lity, as a rise in amplitude between the first and second formants around t
he 1 kHz region, and a reduction in amplitude of the frequencies higher tha
n the second formant. High correlation was shown between the perceptual rat
ings and the predicted values derived from stepwise regression analysis.