Estimating the acoustic reflex threshold from wideband measures of reflectance, admittance, and power.

OBJECTIVE: A method was developed to estimate the contralateral acoustic reflex threshold using shifts in wideband energy reflectance, admittance magnitude and power. DESIGN: In the first experiment contralateral reflex thresholds for a noise activator were estimated on three adult participants using reflectance, admittance and power measurements at frequencies from 250 to 8000 Hz. The reflex threshold was defined using a magnitude and a correlation technique, both having the property of examining the pattern of the reflex-induced shift across a fairly broad frequency range (250 to 2000 Hz). In the second experiment, the magnitude method was modified to include an F test for the comparison of the magnitude of reflex-induced shifts in reflectance, admittance and power relative to response differences in a no-activator baseline condition. Data from four additional participants then were analyzed across a broader frequency range using a method that combined magnitude and correlation methods of estimating reflex thresholds. RESULTS: Acoustic reflex thresholds were obtained using reflectance, admittance and power-level measures in all subjects in both experiments. Individual reflex threshold estimates were as much as 24 dB lower than with the clinical system, with an average of approximately 14 dB lower for the three participants in the first experiment, and approximately 18 dB lower for the four participants in the second experiment. CONCLUSIONS: Wideband measures of reflectance, admittance and power were successfully used to estimate acoustic reflex thresholds in seven participants. A reflex threshold test was devised based on the magnitude of the response shift in the presence of a contralateral activator, and the similarity of the response shift spectra across frequency between successive activator levels. Across all participants in the study, the new test yielded a more sensitive measure of the acoustic reflex threshold than the clinical method. This finding has both clinical and theoretical implications for the study of the acoustic reflex.

Practice and list effects on the synthetic sentence identification test in young and elderly listeners.

A high prevalence of central auditory processing disorders (CAPD) has been reported in the elderly based on scores for the Synthetic Sentence Identification Test (SSI). This study examined practice effects and list differences on the Auditec version of the SSI with an ipsilateral competing message in 30 young and 20 elderly participants. Results for the younger participants suggested that a minimum of one practice list should be used with the SSI, whereas results for the elderly participants suggested that a minimum of three practice lists should be used. The binomial model of A. R. Thornton and M. J. M. Raffin (1978) was suggested for use in determining the required number of lists needed to obtain reliable SSI performance. Significant list differences were found with the Auditec version of the SSI. It was recommended to use only Lists A, C, D, E, G, and I for consistent results across lists. Practice effects and list differences should be accounted for when using the SSI for the assessment of CAPD in elderly and young listeners.

Acoustic reflex detection using wide-band acoustic reflectance, admittance, and power measurements.

The measurement of the acoustic reflex threshold is a basic component of the diagnostic audiological test battery that may subject patients to potentially harmful sound pressures. A wide-band acoustic impedance and reflectance system (D. H. Keefe, R. Ling, & J. C. Bulen, 1992) was investigated as a means of obtaining reflex thresholds at a reduced level and as a means of providing a more complete characterization of the reflex than current clinical methods provide. Reflex thresholds obtained by measuring changes in wide-band admittance, reflectance, and power were at least 8 dB lower than those obtained with the standard clinical technique. These reflex-induced changes were accounted for by a simple oscillator model of the middle ear, assuming that the acoustic reflex results in an increase in stiffness. The results support further investigation of reflectance-based measures of the acoustic reflex as a clinical tool and as a means of studying the reflex mechanism.

Dichotic beats of mistuned consonances.

The beats of mistuned consonances (BMCs) result from the presentation of two sinusoids at frequencies slightly mistuned from a ratio of small integers. Several studies have suggested that the source of dichotic BMCs is an interaction within a binaural critical band. In one case the mechanism has been explained as an aural harmonic of the low-frequency tone (f1) creating binaural beats with the high-frequency tone (f2). The other explanation involves a binaural cross correlation between the excitation pattern of f1 and the contralateral f2--occurring within the binaural critical band centered at f2. This study examined the detection of dichotic BMCs for the octave and fifth. In one experiment with the octave, narrow-band noise centered at f2 was presented to one ear along with f1. The other ear was presented with f2. The noise was used to prevent interactions in the binaural critical band centered at f2. Dichotic BMCs were still detected under these conditions, suggesting that binaural interaction within a critical band does not explain the effect. Localization effects were also observed under this masking condition for phase reversals of tuned dichotic octave stimuli. These findings suggest a new theory of dichotic BMCs as a between-channel phase effect. The modified weighted-image model of localization [Stern and Trahiotis, in Auditory Physiology and Perception, edited by Y. Cazals, L. Demany, and K. Horner (Pergamon, Oxford, 1992), pp. 547-554] was used to provide an explanation of the between-channel mechanism.

The effect of automatic gain control in hearing-impaired listeners with different dynamic ranges.

The purpose of this study was to evaluate the effect of Automatic Gain Control (AGC) on the speech perception of hearing-impaired listeners with varying degrees of reduced dynamic range. Linearly amplified and AGC recordings of the CUNY-Nonsense Syllable Test (CUNY-NST) were constructed in quiet and in noise. They were presented to 30 subjects with flat to gradually sloping sensorineural hearing loss separated into three groups of ten by mean width of dynamic range for puretones. Analysis of the CUNY-NST scores revealed that: (1) AGC significantly improved speech perception in quiet for the subjects with narrower dynamic ranges; (2) the improvement in speech perception with AGC in quiet was dependent upon the manner of articulation of the stimulus; (3) consonant confusions in quiet for AGC were more often due to place errors compared to the linear condition which may be a reflection of a reduction in distinctive feature errors; and (4) linear amplification resulted in higher CUNY-NST scores than did AGC in the presence of back-ground noise.

Distinctive feature scoring of the California Consonant Test.

The purpose of this study was to evaluate a distinctive feature (DF) scoring technique for List 1 of the California Consonant Test (CCT) for the purpose of improving test reliability. The CCT was presented to 50 subjects with high-frequency sensorineural hearing loss (HFSNHL). The DF scoring technique resulted in a split-half correlation of .86 compared to .78 for standard percentage-correct scoring. A partial correlation of .50 was found for DF scoring with the effect of standard scoring held constant, indicating that DF scoring accounted for 25% of the variance in scores unaccounted for by standard scoring. DF scoring also led to smaller percentage differences between the two halves of the CCT (p less than .0005). An analysis of individual feature error patterns suggests that DF scoring may provide a more effective means of differentiating among listeners with HFSNHL than traditional scoring.

Test-retest reliability of a distinctive feature difference test for hearing aid evaluation.

A short closed-set test of speech discrimination using consonants frequently confused by the hearing impaired and scored on the basis of distinctive features was developed and examined for reliability. In experiment I the Distinctive Feature Difference (DFD) Test and the CID W-22 Test were low-pass filtered and presented in a background of speech noise to 21 subjects with normal hearing. The DFD displayed superior test-retest reliability compared with the CID Auditory Test W-22 as indicated by small S.D.s, narrow range of individual test scores, and a lack of significant difference between means. In experiment II, the unfiltered DFD Test was presented in noise to 20 subjects with sensorineural hearing loss. Results similar to those of experiment I were obtained, indicating good test-retest reliability for the DFD Test with hearing-impaired subjects.