Speech Discrimination and Hearing Loss Sloping to High Frequencies: A Phonetic Approach

Publisher Website
Google Scholar
Abstract
Phoneme confusions noted in three groups, each of 25 subjects aged 19–27 years, are analysed in order to find reasons for the great variation in speech discrimination in patients with steep high frequency hearing loss. The confusions were made at five sound pressure levels when simulating three degrees of steep, high frequency hearing loss (cut-off values 125 Hz, 250 Hz and 500 Hz, slope 22 dB/octave). The findings lead to the following conclusions: (1) Moderate hearing loss (cut-off 250 Hz, slope 24 dB/oct.) obliterates part of the F, band, thus distorting the spectrum at high intensities, more than is the case with other filtrations. The same phenomenon also distorts the discrimination of consonantal place of articulation, which in Finnish largely depends on transitions in the surrounding vowels. Laterals, tremulants, semivowels and nasals suffer most from this filtration. The features of manner (stops, fricatives), which have nothing to do with F1 show a steep decrease in the incidence of confusion when the sound pressure level is increased. This configuration of the pure-tone audiogram is a highly inconvenient one for the patient. (2) An additional attenuation (cut-off 125 Hz, slope 24 dB/oct.) will increase front-back vowel confusion, the class of spread front vowels /i.e/ (especially /i/) being confused with /y/ more frequently than with other vowels. This filtration produces similar curves for consonantal place and manner confusions. Although severely attenuating, this configuration of the pure-tone audio-gram produces a relatively recognizable spectrum, leaving these intensity relations of the formants intact. (3) With the light filtration (cut-off 500 Hz, slope 24 dB/oct.) those vowels that have an F1 below the cut-off frequency, /i,y, u/, present the greatest number of identification errors almost irrespective of sound pressure level. This means that the amplitude relations of the vowel spectra are distorted, which leads to back vowel identifications from vowel qualities /i, y/. A slight increase in correct responses is seen with increasing sonority, open vowels giving fewest identification errors. The consonant confusions show that this filtration distorts the labial transitions (low F2) more and at the same time makes the velar transitions (relatively high F2) inaudible. The feature peculiar to the nasal identification is the sudden rise in errors at the 80 to 90 level. This derives from the upsetting of the relation of energy distribution between the nasal and neighbouring vowel sounds. The plateau largely results from increasing front-back vowel confusions and confusions between the resonants. The significance of the finding for a patient with high frequency hearing loss and for hearing aid assessment are discussed.