Allison Merrill and Dr. Richard Harris, Audiology & Speech-Language Pathology
Whenever we perform a comprehensive hearing evaluation we measure thresholds for the patient for puretones and also for speech. Measurement of puretone hearing thresholds in individuals whose native language is not English can be very difficult. Measurement of puretone thresholds in isolation is inadequate. It is very important to obtain some measure of the ability to hear and to understand speech.
One of the measures of the ability to hear speech is the Speech Reception Threshold (SRT). The SRT is defined as the lowest intensity at which a patient can correctly repeat 50% of the words given to them. To measure the SRT it is best to use recorded speech materials to insure that the test- retest variability is at a minimum for intra- and inter-clinic measurement.
Appropriate stimuli are essential when measuring the SRT. The ideal stimuli are words that have steep performance-intensity (PI) functions. Words that are homogeneous with respect to audibility are preferred. Steep PI functions are those where intelligibility increase rapidly with small changes in intensity.
The purpose of the present investigation was to record, evaluate, and produce a compact disc containing homogeneous Brazilian Portuguese words for use in measurement of the SRT in individuals whose native language is Portuguese. Eighty-nine Portuguese trisyllabic words (for SRT testing) and 271 Portuguese bisyllabic words (for speech discrimination testing) were digitally recorded and evaluated. All of the words were among the 1500 most frequently used Portuguese words based upon a corpus containing over 6 million words. All stimuli were recorded using a male and female native Brazilian Portuguese talker.
Seven normally hearing native Portuguese speaking subjects participated in the initial study. All subjects had normal hearing (15 dB HL or better) at all octave and mid-octave frequencies from 125 to 8000 Hz. In addition, all subjects had normal middle ear mobility and compliance as measured by a tympanometer. Acoustic reflexes were also present at normal levels.
After the initial evaluation of the 89 Portuguese trisyllabic words were selected and digitally edited to produce a homogeneous list of Portuguese trisyllabic SRT materials, 30 trisyllabic words were selected from among the 89 trisyllabic words. They were selected because of their uniform steep PI function slopes. The intensity of these 30 trisyllabic words was then digitally edited to yield a 50% correct intensity value at 0 dB HL for all words for both talkers.
PI functions are being obtained for each of the 271 bisyllabic Portuguese words at intensities of –4 to +32 dB HL in 4 dB increments. The 71 bisyllabic words with the poorest speech recognition will be eliminated. The remaining 200 words will be ranked in order of difficulty and them randomly assigned to four lists in blocks of four words each. Data for the bisyllabic PI functions is still being gathered and will be analyzed at a later date.