Search

Speech-recognition tests are a routine component of the clinical hearing evaluation. The most common type of test uses recorded monosyllabic words presented in quiet. The interpretation of test scores relies on an understanding of the variance of repeated tests. Confidence intervals are useful for determining if two scores are significantly different or if the difference is due to the variability of test scores. Because the response to each test item is binary, either correct or incorrect, the binomial distribution has been used to estimate confidence intervals. This method requires that test scores be independent. If the scores are not independent, the binomial distribution will not accurately estimate the variance of repeated scores. A previously published dataset with repeated scores from normal-hearing and hearing-impaired listeners was used to derive confidence intervals from actual test scores in contrast to the predicted confidence intervals in earlier reports. This analysis indicates that confidence intervals predicted by the binomial distribution substantially overestimate the variance of repeated scores resulting in erroneously broad confidence intervals. High correlations were found for repeated scores, indicating that scores are not independent. The interdependence of repeated scores invalidates confidence intervals predicted by the binomial distribution. Confidence intervals and confidence levels for repeated measures were determined empirically from measured test scores to assist in interpreting differences between repeat scores.

Two experiments document the reliability of a speech discrimination test designed for use with an adaptive procedure to measure thresholds for sentences and words. Stimulus words were categorized as either high usage or low usage based on published indices of word usage, and also as either high confusability or low confusability depending on phonetic similarity to other words in the lexicon. The first experiment employed 20 young normal‐hearing adults, each presented with two 20‐item lists of sentences from each of the four lexical categories. One week later, the testing sequence was repeated. A five‐factor ANOVA was employed using word usage, word confusability, list sample, and test–retest delay; an adaptive procedure, converging determined sentence thresholds in dB level. Test–retest thresholds, both immediate and delayed, were within 2 dB for all lists. There was a 1.5‐dB difference between high‐ and low‐usage words. A second experiment examined thresholds for groups of hearing‐impaired individuals categorized by audiometric configuration and speech discrimination scores. Each subject group was presented two sentence lists varying in lexical characteristics based on the Neighborhood Activation Model. The difference between lexical items varied significantly with speech discrimination scores and audiometric configuration of the listeners. [Work supported by Veteran Affairs R&D Contract V621P‐3664.]

The detection, loudness, and discrimination performances of subjects with normal hearing were assessed using 3 five-component tonal complexes (1020-1100 Hz) with crest factors of 1.8, 2.6, and 3.2. The 3 five-component tonal complexes were: (1) equally detectable in broadband noise when presented at equal rms amplitude; (2) perceived as equally loud when presented at essentially equal rms amplitude; and (3) discriminable from one another when presented at equal loudness. The present data indicate that listeners can make discriminations on the basis of amplitude variations.

Two investigations were conducted (i) to determine a reliable procedure for obtaining loudness discomfort levels (LDL) and (ii) to observe the effects of frequency on the LDL. Three psychophysical methods were compared in the measurement of the LDL—a programmed method of constant stimuli, the method of adjustment, and a tracking method. The method of constants provided the most reliable judgments over six trials for a 1000‐Hz stimulus. The LDL was then measured utilizing the method of constants for pure tones at octave frequencies from 125 to 4000 Hz, and for a wide and narrow band of noise. The LDLs (in dB SPL) were highest at low frequencies (through 500 Hz) and relatively constant for 1000, 2000, 4000 Hz, and the wide‐band noise. The relatively high LDLs in the low frequencies are explained on the basis of difference in the intensity level for low frequencies measured in a conventional 6 cm3 coupler and that observed in the ear canal when a supra‐aural earphone is fitted on the head. When a correction is made for this difference the LDL results across frequency compared closely with results from loudness‐contour studies.

Monaural masked thresholds for clicks were established on three listeners. Masking functions across selected Δt's (the time disparities between the masker and click) were determined for four masking conditions in which the 70 dB SPL masker was a 500‐msec burst of random noise. These four conditions were (1) simultaneous masking, (2) forward masking (FM) at Δt's from 1 to 250 msec, (3) backward masking (BM) at Δt's from −1 to −250 msec, and (4) forward‐backward masking at selected Δt's between two 500‐msec noise bursts. In this last instance, the interval between bursts was varied to achieve seven interburst intervals (IBIs) ranging from 25 to 500 msec. Simultaneous masking was 3–5 dB less at the boundaries of the noise burst than it was in the middle of the noise burst. The functions for FM and for BM were not inversely symmetrical with one another, although both showed less threshold shift (TS) the further displaced in time the click was from the masker. The masking that appeared when the click was presented between two noise bursts was greater immediately adjacent to a noise burst than near the center of the IBI. During the longer IBIs (500–300 msec), the configuration of masking was equivalent to that of FM followed by BM, with slight excess appearing only in the interval 20–40 msec before the second noise burst. The configurations of FM and BM were preserved during the shorter IBIs (200–50 msec), but the TSs were greater than either type would have produced if operating alone. These TSs became larger as the IBI was shortened, so that the brief IBIs were permeated with large amounts of masking even in the middle of the IBI. This cumulation of masking was so great when the IBI was reduced to 25 msec that the masking function lost the configuration characterizing the longer IBIs. The TSs during shorter IBIs exceeded the sum of the residual effective powers of FM and BM.

Monaural intelligibility functions for spondaic and monosyllabic words and identification functions for synthetic sentences were obtained for 14 experimental conditions of masking by random noise. The noise was presented continuously and interrupted or modulated (−14‐dB interburst ratio) at rates of 1, 10, and 100 interruptions per second. The intensity levels of the maskers were 50 and 90 dB SPL. Results demonstrated that (1) masking effectiveness decreased for spondees and sentences as the interruption rate decreased; (2) the greatest release from masking for monosyllabic words was found during the 10‐ips condition at favorable S/N ratios; however, at the more unfavorable S/N ratios, greater release from masking was observed during the 1‐ips condition; (3) greater release from masking occurred for interrupted than for comparable modulated noise conditions. The possible effects of the interaction between the type of speech material and forward and backward masking during the silent interval between noise bursts is discussed.

The changes in masking for spondee words that result from varying both the level and the interruption rate of a white‐noise masker were studied with 14 normal hearing subjects and 14 subjects with sensorineural hearing loss diagnosed as resulting from cochlear otosclerosis. The masker was a white noise and was presented monaurally at either 90 dB sound‐pressure level (SPL) or 30 dB sensation level (SL). It was either continuous or was pulsed at rates of 1, 10, or 100/sec with 50% duty cycle. During the burst‐off half of each cycle, the noise was either dropped 14 dB in level or was fully interrupted. The masked speech reception threshold (SRT) was not improved re the masked SRT in continuous noise when the 30 dB SL masker was pulsed 100 times/sec. Reduction in masking was observed under all other circumstances of cycling the noise. This reduction was more pronounced when the noise was completely interrupted rather than only modulated by 14 dB, when the masker was at its higher level (90 dB SPL as opposed ...

Hartman et al. [J. Acoust. Soc. Am. 79, 1915–1925 (1986)] presented data on discrimination of tonal complexes differing only in number of spectral components within a fixed bandwidth. The Hartmann et al. data suggested that discrimination could be based upon perception of temporal fluctuations in the intensity levels of the sounds, or upon the detail of the spectral structure of the complexes. The present experiment assessed discrimination of tonal complexes that differed in only the phase relationships among five fixed frequency components. The phase relationships established a range of crest factors (peak/rms amplitude) for the complexes. When presented at equal loudness, the tonal complexes were easily discriminable by a musically trained subject, and were discriminable by all subjects with training. The data from the current study support a model of discrimination based only upon the temporal fluctuations in the complex envelopes. Discrimination between tonal complexes can be made for complexes differing either in rate or depth of such fluctuations, as long as the fluctuations are within the limits of temporal integration of the car.

Previous studies have shown that the alternation of speech between ears does not affect the recognition performance of normal listeners when nonshadowing response techniques are used. In the present study, phoneme‐tied alternation was employed with CVC monosyllabic words. In this signal paradigm, the temporal sequence of the words was maintained with the initial consonant segment presented to one ear, the vowel segment presented to the other ear, and the final consonant presented to the ear that received the initial consonant. The segmentation was accomplished digitally. Psychometric functions on 16 young adult subjects were generated for the segmented materials under monaural (nonalternating) and binaural (alternating) listening conditions. Future testing of patients with auditory dysfunction is planned using the alternated segmented words as a possible test of brain stem integration. [Work supported by VA.]

Thresholds for homophasic and antiphasic multicomponent tonal complexes were measured as a function of bandwidth in broadband noise to replicate and to extend the classic masking experiment [G. Gassler, Acoustica 4, 408–414 (1954)]. Gassler reported that masked thresholds for tonal complexes were nearly constant for bandwidths less than a nominal “critical bandwidth,” beyond which thresholds increased 3 dB/oct. Three experiments were conducted in broadband noise employing (a) homophasic signals similar to Gassler's (1 to 20 sinusoids spaced 20 Hz apart extending downward from 1100 Hz); (b) homophasic and antiphasic signals composed of 1 to 23 sinusoids centered at 500, 1000, and 2000 Hz; and (c) homophasic and antiphasic signals centered at 500 Hz with bandwidths to 800 Hz (71 sinusoids). The slope of the threshold/signal bandwidth function beyond 100 Hz was < dB/oct at each center frequency, ranging from 0.3 dB/oct (2000 Hz) to 1.9 dB/oct (1000 Hz). The data indicate that the integration of signal energy continued beyond the so‐called “critical band” estimates reported by Gassler. There were no substantial differences between the slopes of the homophasic and antiphasic conditions at 500, 1000, and 2000 Hz. [Work supported by VA Medical Research Service.]

Two studies have been conducted to measure the loudness discomfort levels (LDLs) for subjects with normal hearing. First, the LDL for a 1000‐Hz stimulus was measured utilizing the method of constant stimuli, the method of adjustment, and a tracking method. The method of constants provided the most reliable judgments among subjects over six trials. In the subsequent study, LDLs were measured utilizing the method of constants for pure tones at octave frequencies from 125 through 4000 Hz and for two bands of random noise. LDLs (in dB SPL) varied as a function of the stimulus employed. LDLs were highest at low frequencies (through 500 Hz) and relatively constant for 1000, 2000, and 4000 Hz and for the wide‐band noise.