Effects of lexical factors on word recognition among normal-hearing and hearing-impaired listeners.

An investigation was conducted to examine the effects of lexical difficulty on spoken word recognition among young normal-hearing and middle-aged and older listeners with hearing loss. Two word lists, based on the lexical characteristics of word frequency and neighborhood density and frequency (Neighborhood Activation Model [NAM]), were developed: (1) lexically "easy" words with high word frequency and a low number and frequency of words phonemically similar to the target word and (2) lexically "hard" words with low word frequency and a high number and frequency of words phonemically similar to the target word. Simple and transformed up-down adaptive strategies were used to estimate performance levels at several locations on the performance-intensity functions of the words. The results verified predictions of the NAM and showed that easy words produced more favorable performance levels than hard words at an equal intelligibility. Although the slopes of the performance-intensity function for the hearing-impaired listeners were less steep than those of normal-hearing listeners, the effects of lexical difficulty on performance were similar for both groups.

Examination of the neighborhood activation theory in normal and hearing-impaired listeners.

OBJECTIVE: Experiments were conducted to examine the effects of lexical information on word recognition among normal hearing listeners and individuals with sensorineural hearing loss. The lexical factors of interest were incorporated in the Neighborhood Activation Model (NAM). Central to this model is the concept that words are recognized relationally in the context of other phonemically similar words. NAM suggests that words in the mental lexicon are organized into similarity neighborhoods and the listener is required to select the target word from competing lexical items. Two structural characteristics of similarity neighborhoods that influence word recognition have been identified; "neighborhood density" or the number of phonemically similar words (neighbors) for a particular target item and "neighborhood frequency" or the average frequency of occurrence of all the items within a neighborhood. A third lexical factor, "word frequency" or the frequency of occurrence of a target word in the language, is assumed to optimize the word recognition process by biasing the system toward choosing a high frequency over a low frequency word. DESIGN: Three experiments were performed. In the initial experiments, word recognition for consonant-vowel-consonant (CVC) monosyllables was assessed in young normal hearing listeners by systematically partitioning the items into the eight possible lexical conditions that could be created by two levels of the three lexical factors, word frequency (high and low), neighborhood density (high and low), and average neighborhood frequency (high and low). Neighborhood structure and word frequency were estimated computationally using a large, on-line lexicon-based Webster's Pocket Dictionary. From this program 400 highly familiar, monosyllables were selected and partitioned into eight orthogonal lexical groups (50 words/group). The 400 words were presented randomly to normal hearing listeners in speech-shaped noise (Experiment 1) and "in quiet" (Experiment 2) as well as to an elderly group of listeners with sensorineural hearing loss in the speech-shaped noise (Experiment 3). RESULTS: The results of three experiments verified predictions of NAM in both normal hearing and hearing-impaired listeners. In each experiment, words from low density neighborhoods were recognized more accurately than those from high density neighborhoods. The presence of high frequency neighbors (average neighborhood frequency) produced poorer recognition performance than comparable conditions with low frequency neighbors. Word frequency was found to have a highly significant effect on word recognition. Lexical conditions with high word frequencies produced higher performance scores than conditions with low frequency words. CONCLUSION: The results supported the basic tenets of NAM theory and identified both neighborhood structural properties and word frequency as significant lexical factors affecting word recognition when listening in noise and "in quiet." The results of the third experiment permit extension of NAM theory to individuals with sensorineural hearing loss. Future development of speech recognition tests should allow for the effects of higher level cognitive (lexical) factors on lower level phonemic processing.

Subjective judgements of clarity and intelligibility for filtered stimuli with equivalent speech intelligibility index predictions.

The purpose of this investigation was to determine whether subjective judgments of clarity or intelligibility would be rated equally among conditions in which speech was equated for predicted intelligibility (using the Speech Intelligibility Index, SII) but varied in bandwidth. Twenty listeners with normal hearing rated clarity and intelligibility for sentence material (Hearing In Noise Test) in speech-shaped noise at six paired low- and high-pass filtered conditions in which SII was equated for each pair. For three paired conditions, predicted intelligibility increased as SII increased monotonically (0.3, 0.4, 0.5). In the remaining paired conditions, SII continued to increase monotonically (0.6, 0.7, 0.8) but predicted intelligibility was held at a maximal level (> or = 95%). Predicted intelligibility was estimated from the transfer function relating SII to speech recognition scores determined in preliminary experiments. Differences in ratings between paired low- and high-pass filtered sentences did not reach statistical significance for either clarity or intelligibility, indicating that the spectral differences at equivalent SIIs did not influence the judgments for either of the two dimensions. For conditions in which predicted intelligibility increased, both clarity and intelligibility ratings increased in a similar manner. For conditions in which predicted intelligibility was maximized, intelligibility ratings remained the same statistically across conditions while clarity ratings changed modestly. Although high correlations were observed between clarity and intelligibility ratings, intelligibility ratings were consistently higher than clarity ratings for comparable conditions. The results indicated that listeners with normal hearing produced clarity and intelligibility ratings for the same speech material and experimental conditions that were highly related but differed in magnitude. Caution is required when substituting clarity for intelligibility.

Subjective judgments of speech clarity measured by paired comparisons and category rating.

OBJECTIVE: The purpose of this study was to compare listeners' subjective judgments of speech clarity via paired comparisons and category rating using stimulus conditions that varied in the relative spacing between stimulus items, producing either a wide or narrow range of performance. DESIGN: Subjective judgments of speech clarity were measured via paired comparisons and category rating in 12 normal-hearing (Experiment 1) and eight hearing-impaired adults (Experiment 2). Sentences processed by six band-pass filters that increased monotonically in Articulation Index (AI) estimates constituted the stimuli to be judged. Using subsets of three filters from the group of six, subjective judgments were additionally obtained for stimulus conditions in which the performance ranges were wide (large differences in AI) and narrow (small differences in AI). RESULTS: Speech clarity judgments obtained by paired comparisons and category rating were highly related to the AI estimates both for normal-hearing and hearing-impaired subjects. When the performance range was wide, both methods provided similar judgments for the normal-hearing subjects. For the hearing-impaired subjects, paired comparisons were more sensitive than category rating. When the performance range was narrow, paired comparisons were more sensitive than category rating in differentiating between filters for both groups of subjects. This difference was less obvious for the normal-hearing subjects when paired comparison data were converted to a scale comparable to the category ratings. Large between-subject variability was evident for the hearing-impaired subjects on the psychophysical scaling procedures, most notably for category rating. CONCLUSIONS: When judging the clarity among stimulus items where performance varied over a wide range, both category rating and paired comparisons provided comparable judgments for normal-hearing listeners. For conditions in which perceptual differences between stimulus items were restricted either by the choice of conditions or by the effects of sensorineural hearing loss, the method of paired comparisons was the more sensitive procedure.

Report of the Eriksholm Workshop on auditory deprivation and acclimatization.

The terminology used in studies documenting changes in auditory performance following fitting of hearing aids has been diverse. Definitions for the auditory deprivation effect and auditory acclimatization are offered as a first step in rationalization. Two statements summarize current knowledge concerning auditory deprivation effects and auditory acclimatization, as well as considering the potential implications for research, field trial and clinical practice applications. Potential areas for future research are identified.

Comparison of probe insertion methods on estimates of ear canal SPL.

Real-ear sound pressure levels (SPLs) were compared among three methods used for positioning a probe microphone in the ear canal. The probe insertion techniques included (1) an acoustic method that incorporates use of the quarter-wave anti-resonance property of the ear to determine acoustically the location of the probe tube relative to the eardrum; (2) a constant insertion depth method (25 mm from the intratragal notch); and (3) the earmold +5-mm method, which places the probe 5 mm beyond the tip of the earmold, thereby avoiding problems associated with the transition region where sound exits from the bore of the earmold into the larger ear canal. Measurements were obtained at 32 test frequencies in 24 adults with normal middle ear impedance. Results indicated that the SPLs measured by the acoustic method were modestly higher than those measured by the other two methods. This result was most evident in subjects with long ear canals (> 25 mm) and at high test frequencies (3.0 to 6.3 kHz).

Reliability and sensitivity of paired comparisons and category rating in children.

Children's subjective judgments of speech clarity using the methods of paired comparisons and category rating were evaluated in this investigation. Eighty children with normal hearing between the ages of 4 and 8 years judged the clarity of sentences that were systematically bandpass-filtered using conditions that increased intelligibility as estimated by the Articulation Index. Subjects were classified into four age groups (4-, 5-, 6-, and 7-8-year age groups) with 20 subjects per group. With use of materials and training methods suitable for children, judgments were obtained via the two psychophysical procedures (10 subjects per age group for each procedure). Results indicated that children 5 years of age and older were able to make reliable clarity judgments using either procedure; however, the method of paired comparisons was more sensitive than category rating in detecting differences between the bandpass-filtered conditions.

Speech recognition in amplitude-modulated noise of listeners with normal and listeners with impaired hearing.

The effect of amplitude-modulated (AM) noise on speech recognition in listeners with normal and impaired hearing was investigated in two experiments. In the first experiment nonsense syllables were presented in high-pass steady-state or AM noise to determine whether the release from masking in AM noise relative to steady-state noise was significantly different between normal-hearing and hearing-impaired subjects when the two groups listened under equivalent masker conditions. The normal-hearing subjects were tested in the experimental noise under two conditions: (a) in a spectrally shaped broadband noise that produced pure tone thresholds equivalent to those of the hearing-impaired subjects, and (b) without the spectrally shaped broadband noise. The release from masking in AM noise was significantly greater for the normal-hearing group than for either the hearing-impaired or masked normal-hearing groups. In the second experiment, normal-hearing and hearing-impaired subjects identified nonsense syllables in isolation and target words in sentences in steady-state or AM noise adjusted to approximate the spectral shape and gain of a hearing aid prescription. The release from masking was significantly less for the subjects with impaired hearing. These data suggest that hearing-impaired listeners obtain less release from masking in AM noise than do normal-hearing listeners even when both the speech and noise are presented at levels that are above threshold over much of the speech frequency range.

Effects of probe insertion depth on real ear measurements.

The purpose of this article is to provide an overall summary of the role probe insertion depth has on real ear measurements, and to compare the real ear sound pressure level measured by a probe microphone system, using three methods for positioning the probe in an ear canal. The probe insertion techniques that were compared included: (1) an acoustic method that incorporates use of the quarter-wave antiresonance property of the ear to determine acoustically the location of the probe tube relative to the eardrum in an individual ear; (2) a constant insertion depth method (25 mm from the intratragal notch); and (3) the earmold +5 mm method, which places the probe 5 mm beyond the tip of the individual's earmold in the canal, thereby avoiding problems associated with the transition region, where sound exits from the bore of the earmold into the larger ear canal. Measurements were obtained for each method at 32 frequencies in the unoccluded ears of 17 subjects. Results indicated that the sound pressure levels measured by the acoustic method were significantly larger than those measured by the other two methods. This result was most evident in subjects with long ear canals (> 25 mm) and at high test frequencies (4.0 to 6.3 kHz). For subjects with short or average length ear canals, the three methods provided essentially equivalent results.

Preferred frequency response for two- and three-channel amplification systems.

The purpose of these investigations was to compare the preferred frequency-gain responses obtained from two- and three-channel amplification systems. The current experiments were limited to a linear system in which the crossover frequency dividing the channels was systematically varied. The subjects for the experiment were nine individuals with mild to moderately severe sensorineural hearing loss with various audiometric configurations. The subjects listened to continuous discourse, in noise, via a computer-controlled digital master hearing aid containing two real-time data acquisition processors. Initially, a modified simplex procedure was used to obtain preferred frequency-gain responses using several different crossover frequencies. A round-robin procedure was then conducted in which each preferred response from the simplex was compared with every other preferred response. The frequency-gain responses chosen most often for the two- and three-channel systems were compared. The results showed no significant differences between the preferred frequency-gain response for the two- versus the three-channel system. In addition, the preferred response chosen most often was not consistently observed at the same crossover frequency for all subjects, with the exception of those with steeply sloping hearing loss who chose 1,120 Hz as the first or second preference for the two-channel system. The round-robin results were rank-ordered according to the number of times each frequency-gain response was chosen. In general, subjects chose several frequency-gain responses at various crossover frequencies, which were not significantly different from each other statistically. The results of a final experiment suggested that physical similarities in the preferred responses chosen at the various crossover frequencies played a role in the rank-ordering of the preference judgments obtained in the original investigation.

Frequency-importance functions for words in high- and low-context sentences.

The relative importance and absolute contributions of various spectral regions to speech intelligibility under conditions of either neutral or predictable sentential context were examined. Specifically, the frequency-importance functions for a set of monosyllabic words embedded in a highly predictive sentence context versus a sentence with little predictive information were developed using Articulation Index (AI) methods. Forty-two young normal-hearing adults heard sentences presented at signal-to-noise ratios from -8 to +14 dB in a noise shaped to conform to the peak spectrum of the speech. Results indicated only slight differences in 1/3-octave importance functions due to differences in semantic context, although the crossovers differed by a constant 180 Hz. Methodological and theoretical aspects of parameter estimation in the AI model are discussed. The results suggest that semantic context, as defined by these conditions, may alter frequency-importance relationships in addition to the dynamic range over which intelligibility rises.

Associations among frequency and temporal resolution and consonant recognition for hearing-impaired listeners.

To examine the association between frequency and temporal resolution and speech recognition, auditory filter parameters, rates of masking decay, and stop-consonant recognition were determined for 9 normal-hearing and 24 hearing-impaired subjects. Speech-presentation levels were selected for each subject based on articulation index (AI) predictions. Results suggest that auditory filter widths and dynamic ranges were strongly correlated with pure-tone threshold, while time constants were not. Stop-consonant recognition scores for most hearing-impaired listeners were not significantly poorer than predicted by the AI model. Furthermore, differences between observed recognition scores and those predicted by the AI were poorly correlated with derived psychophysical parameter values, suggesting that measures of frequency and temporal resolution and speech recognition may appear to be associated primarily because of their dependence on auditory threshold.

Closed-set effects in consonant confusion patterns.

Invariance of error patterns in confusion matrices of varying dimensions were examined. Normal-hearing young adults were presented closed-set arrangements of digitized syllable tokens, spoken by 1 male and 1 female talker, and selected from a set of 14 consonants (stops and fricatives). Each consonant was paired with the vowel/a/ in a vowel-consonant format and presented at three intensity levels. Patterns of errors among voiceless stops and among voiced fricatives were dependent on the set of alternatives. Voiceless fricatives and voiced stops were not significantly affected by the number of response alternatives. Speaker differences, individual differences among listeners, and implications relating to the generalization of confusion data collected in small closed-set arrangements are discussed.

Auditory filter characteristics and consonant recognition for hearing-impaired listeners.

To examine the association between frequency resolution and speech recognition, auditory filter parameters and stop-consonant recognition were determined for 9 normal-hearing and 24 hearing-impaired subjects. In an earlier investigation, the relationship between stop-consonant recognition and the articulation index (AI) had been established on normal-hearing listeners. Based on AI predictions, speech-presentation levels for each subject in this experiment were selected to obtain a wide range of recognition scores. This strategy provides a method of interpreting speech-recognition performance among listeners who vary in magnitude and configuration of hearing loss by assuming that conditions which yield equal audible spectra will result in equivalent performance. It was reasoned that an association between frequency resolution and consonant recognition may be more appropriately estimated if hearing-impaired listeners' performance was measured under conditions that assured equivalent audibility of the speech stimuli. Derived auditory filter parameters indicated that filter widths and dynamic ranges were strongly associated with threshold. Stop-consonant recognition scores for most hearing-impaired listeners were not significantly poorer than predicted by the AI model. Furthermore, differences between observed recognition scores and those predicted by the AI were not associated with auditory filter characteristics, suggesting that frequency resolution and speech recognition may appear to be associated primarily because both are degraded by threshold elevation.

Interactive factors in consonant confusion patterns.

Confusion patterns among English consonants were examined using log-linear modeling techniques to assess the influence of low-pass filtering, shaped noise, presentation level, and consonant position. Ten normal-hearing listeners were presented consonant-vowel (CV) and vowel-consonant (VC) syllables containing the vowel /a/. Stimuli were presented in quiet and in noise, and were either filtered or broadband. The noise was shaped such that the effective signal level in each 1/3 octave band was equivalent in quiet and noise listening conditions. Three presentation levels were analyzed corresponding to the overall rms level of the combined speech stimuli. Error patterns were affected significantly by presentation level, filtering, and consonant position as a complex interaction. The effect of filtering was dependent on presentation level and consonant position. The effects stemming from the noise were less pronounced. Specific confusions responsible for these effects were isolated, and an acoustical interaction is suggested, stressing the spectral characteristics of the signals and their modification by presentation level and filtering.

Stop-consonant recognition for normal-hearing listeners and listeners with high-frequency hearing loss. I: The contribution of selected frequency regions.

The purpose of this study is to specify the contribution of certain frequency regions to consonant place perception for normal-hearing listeners and listeners with high-frequency hearing loss, and to characterize the differences in stop-consonant place perception among these listeners. Stop-consonant recognition and error patterns were examined at various speech-presentation levels and under conditions of low- and high-pass filtering. Subjects included 18 normal-hearing listeners and a homogeneous group of 10 young, hearing-impaired individuals with high-frequency sensorineural hearing loss. Differential filtering effects on consonant place perception were consistent with the spectral composition of acoustic cues. Differences in consonant recognition and error patterns between normal-hearing and hearing-impaired listeners were observed when the stimulus bandwidth included regions of threshold elevation for the hearing-impaired listeners. Thus place-perception differences among listeners are, for the most part, associated with stimulus bandwidths corresponding to regions of hearing loss.

Stop-consonant recognition for normal-hearing listeners and listeners with high-frequency hearing loss. II: Articulation index predictions.

Articulation index (AI) theory was used to evaluate stop-consonant recognition of normal-hearing listeners and listeners with high-frequency hearing loss. From results reported in a companion article [Dubno et al., J. Acoust. Soc. Am. 85, 347-354 (1989)], a transfer function relating the AI to stop-consonant recognition was established, and a frequency importance function was determined for the nine stop-consonant-vowel syllables used as test stimuli. The calculations included the rms and peak levels of the speech that had been measured in 1/3 octave bands; the internal noise was estimated from the thresholds for each subject. The AI model was then used to predict performance for the hearing-impaired listeners. A majority of the AI predictions for the hearing-impaired subjects fell within +/- 2 standard deviations of the normal-hearing listeners' results. However, as observed in previous data, the AI tended to overestimate performance of the hearing-impaired listeners. The accuracy of the predictions decreased with the magnitude of high-frequency hearing loss. Thus, with the exception of performance for listeners with severe high-frequency hearing loss, the results suggest that poorer speech recognition among hearing-impaired listeners results from reduced audibility within critical spectral regions of the speech stimuli.

Basic acoustic considerations of ear canal probe measurements.

This publication contains a review of several acoustic investigations in which the effects of probe location on real-ear gain were examined through theoretical models based on acoustic properties of the average human ear and ear simulator studies. The results of these investigations are used to demonstrate the effect of standing waves and eardrum impedance on probe measurements made in the ear canal. Investigations were also conducted in the sound field with a KEMAR manikin. A commercial probe microphone system was used to measure the SPL and real-ear gain at various locations with the KEMAR ear canal. The results emphasize the critical effect of probe location on absolute or relative ear canal measurements and indicate the necessity to establish clinical procedures for probe measurements based on relevant acoustic principles.

Effects of hearing loss on utilization of short-duration spectral cues in stop consonant recognition.

The purpose of this experiment was to evaluate the utilization of short-term spectral cues for recognition of initial plosive consonants (/b,d,g/) by normal-hearing and by hearing-impaired listeners differing in audiometric configuration. Recognition scores were obtained for these consonants paired with three vowels (/a,i,u/) while systematically reducing the duration (300 to 10 ms) of the synthetic consonant-vowel syllables. Results from 10 normal-hearing and 15 hearing-impaired listeners suggest that audiometric configuration interacts in a complex manner with the identification of short-duration stimuli. For consonants paired with the vowels /a/ and /u/, performance deteriorated as the slope of the audiometric configuration increased. The one exception to this result was a subject who had significantly elevated pure-tone thresholds relative to the other hearing-impaired subjects. Despite the changes in the shape of the onset spectral cues imposed by hearing loss, with increasing duration, consonant recognition in the /a/ and /u/ context for most hearing-impaired subjects eventually approached that of the normal-hearing listeners. In contrast, scores for consonants paired with /i/ were poor for a majority of hearing-impaired listeners for stimuli of all durations.

Recognition of nonsense syllables by hearing-impaired listeners and by noise-masked normal hearers.

In the present study, speech-recognition performance was measured in four hearing-impaired subjects and twelve normal hearers. The normal hearers were divided into four groups of three subjects each. Speech-recognition testing for the normal hearers was accomplished in a background of spectrally shaped noise in which the noise was shaped to produce masked thresholds identical to the quiet thresholds of one of the hearing-impaired subjects. The question addressed in this study is whether normal hearers with a hearing loss simulated through a shaped masking noise demonstrate speech-recognition difficulties similar to those of listeners with actual hearing impairment. Regarding overall percent-correct scores, the results indicated that two of the four hearing-impaired subjects performed better than their corresponding subgroup of noise-masked normal hearers, whereas the other two impaired listeners performed like the noise-masked normal listeners. A gross analysis of the types of errors made suggested that subjects with actual and simulated losses frequently made different types of errors.