Effects of Fast, Slow, and Adaptive Amplitude

J Am Acad Audiol 25:834–847 (2014)

Effects of Fast, Slow, and Adaptive Amplitude Compression on Children’s and Adults’ Perception of Meaningful Acoustic Information DOI: 10.3766/jaaa.25.9.6 Andrea L. Pittman* Ashley J. Pederson* Madalyn A. Rash*

Abstract Background: Fast- and slow-acting amplitude compression parameters have complementary strengths and weaknesses that limit the full benefit of this feature to hearing aid users. Adaptive time constants have been suggested in the literature as a means of optimizing the benefits of amplitude compression. Purpose: The purpose of this study was to compare the effects of three amplitude compression release times (slow, fast, and adaptive) on children’s and adults’ accuracy for categorizing speech and environmental sounds. Research Design: Participants were asked to categorize speech or environmental sounds embedded in short trials containing low-level playground noise. Stimulus trials included either a high-level environmental sound followed by a lower-level speech stimulus (word) or a high-level speech stimulus followed by a lower-level environmental sound. The listeners responded to the second (low-level) stimulus in each trial. The two stimuli overlapped temporally in half of the trials but not in the other half. The stimulus trials were processed to simulate amplitude compression having fast (40 msec), slow (800 msec), or adaptive release times. The adaptive-compression parameters operated in a slow fashion until a sudden increase/decrease in level required a rapid change in gain. Study Sample: Participants were 15 children and 26 adults with hearing loss (HL) as well as 20 children and 21 adults with normal hearing (NH). Data Collection and Analyses: Performance (in % correct) was arcsine transformed and subjected to repeated-measures analysis of variance with pairwise comparisons of significant main effects using Bonferroni adjustments for multiple comparisons. Results: Overall, the performance of listeners with HL was poorer than that of the listeners with NH and performance for the environmental sounds was poorer than for the speech stimuli, particularly for the adults and children with HL. Significant effects of age group, stimulus overlap, and compression speed were observed for the listeners with NH, whereas effects of stimulus overlap and compression speed were found for the listeners with HL. Whereas listeners with NH achieved optimal performance with slow-acting compression, the listeners with HL achieved optimal performance with adaptive compression. Conclusions: Although slow, fast, and adaptive compression affects the acoustic signal in a subtle fashion, amplitude compression significantly affects perception of speech and environmental sounds. Key Words: Amplification, amplitude compression, children, adults

*Arizona State University, Tempe, AZ Andrea Pittman, Department of Speech and Hearing Science, Arizona State University, P.O. Box 0102, Tempe, AZ 85287-0102; E-mail: andrea. [email protected] This work has not been published in any form. This project was funded by a grant from Oticon A/S.

834 Delivered by Ingenta to: Arizona State University IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53

Fast, Slow, and Adaptive Amplitude Compression/Pittman et al

Abbreviations: AI 5 articulation index; ANOVA 5 analysis of variance; DSL 5 Desired Sensation Level; HL 5 hearing loss; NH 5 normal hearing; WDRC 5 wide dynamic range compression

INTRODUCTION

L

ike many advances in hearing aid technology during the last 25 yr, only a few of the benefits of amplitude compression were apparent when it was first introduced. Chief among the initial benefits was the reduced need for the hearing aid user to adjust the volume control as the listening environment changed. When it was also shown that hearing aid users preferred amplitude compression versus linear amplification, the feature became standard in most commercially available devices (Souza, 2002; Walden et al, 2000; Kam and Wong, 1999). The benefits of amplitude compression were later determined to include improved tolerance for loud sounds as well as a higher rate of successful adjustment to hearing aid use (Larson et al, 2000; Johnson et al, 2010). The most popular application of this feature is wide dynamic range compression (WDRC). In simple terms, WDRC is a specific implementation of the dynamic and static properties of amplitude compression in which more gain is applied to low-level sounds that fall at or below the listener’s hearing thresholds whereas less gain is applied to higher-level sounds that may exceed the listener’s comfort. To do this, WDRC involves a number of static and dynamic parameters such as (1) the input level at which gain is no longer applied linearly (i.e., compression kneepoint), (2) the change in input level relative to the change in output level (i.e., compression ratio), and (3) the speed with which the gain parameters are applied (i.e., attack/release time). The first two parameters (compression kneepoint and ratio) have received a fair amount of attention such that their effects are fairly well understood (Souza, 2002; Souza and Turner, 1999; Davies-Venn et al, 2009; Souza et al, 2005). The impact of the dynamic parameters (attack and release time) have received somewhat less attention, although evidence suggests that they may significantly affect successful communication for some hearing aid users (Stone and Moore, 2003; Lunner and SundewallThoreń, 2007; Jenstad and Souza, 2005; Gatehouse et al, 2006). By convention, the dynamic parameters of amplitude compression are subdivided into fast- and slow-acting categories denoting the time it takes for the compression parameters to reach their nominal gain values for any level of input. Most commercially available devices have a fast attack time (,50 msec), although their release times can differ substantially. Release times in fast-acting compressors are generally less than 100 msec, whereas the release times of slow-acting compressors can be between 500 and 2000 msec or more. The goal of fast-acting compression is to reduce the dynamic

range of continuous speech over the short term by operating fast enough to apply gain to low-level phonemes (e.g., voiceless fricatives) and less gain to high-level phonemes (e.g., vowels). The goal of a slow-acting system, on the other hand, is to limit the effects of compression on the short-term variations of speech and instead modulate levels of vocal effort and environmental sounds over the long term. Both of these approaches cause unique problems for the hearing aid user. Slow-acting compression with its longer release times can result in inadequate amplification of low-level input for a substantial period following a loud input. Fast-acting compression can cause two problems: (1) it can distort the acoustic waveform resulting in a somewhat poorer signal quality (Jenstad and Souza, 2005), and (2) it can decrease the signal-to-noise ratio by amplifying low-level background noise during periods of low-level speech (Naylor and Johannesson, 2009). Recent evidence comparing the effects of fast- and slowrelease times on speech perception suggests that adults with lower cognitive abilities have more difficulty with the subtle signal distortion caused by fast-acting compression than adults with better cognitive abilities (Gatehouse et al, 2006; Souza et al, 2000; Jenstad and Souza, 2005, 2007; Cox and Xu, 2010; Lunner and Sundewall-Thoreń, 2007; Rudner et al, 2011). To date, no studies examining the effects of compression speed have been conducted in children with hearing loss (HL) who may be vulnerable to the distortion caused by fast-acting compression while their cognitive skills are developing. Another area that has received little attention is the relationship between the acoustic event that triggers the compression and the acoustic event that the hearing aid user would like to hear. Most research has focused on the effects of compression speed on the speech signal. However, complex listening environments in which amplitude compression would likely occur contain multiple sound sources and levels of acoustic information. For example, walking along a city street may require the hearing aid user to manage environmental sounds (e.g., traffic noise), multitalker babble from other pedestrians, and an interaction with a conversation partner. Any of these sound sources could be of interest to the hearing aid user at any point in time, although the loudest sound source (usually closest) to the hearing aid is likely to trigger compression, even speech. Thus, the manner in which the hearing aid applies compression to both speech and environmental sounds may be quite important to the hearing aid user. Several things have contributed to the limited amount of research in the perception of environmental sounds:


Journal of the American Academy of Audiology/Volume 25, Number 9, 2014

(1) It is more challenging to represent the wider variety of environmental sounds in an experimental paradigm than it is to represent the more narrow range of speech sounds (Shafiro and Gygi, 2004). (2) Individuals with severe-to-profound HLs are more likely than individuals with lesser HL to express concern about their lack of access to environmental sounds (Tyler, 1994; Zhao et al, 1997; Looi and Arnephy, 2010). (3) Environmental sounds are often confused with undesirable noise levels associated with public health concerns (Kim et al, 2012; Smith, 1998; Shepherd et al, 2010), causing them to be viewed as dangerous rather than informative. (4) Humans do not interact with environmental sounds in the same way that they interact with speech (Gygi and Shafiro, 2007). When necessary, listeners respond to environmental sounds by modifying their behavior rather than engaging with the sound source as they might with another individual. Thus, a lack of awareness for environmental sounds may go unnoticed until an individual fails to modify his or her behavior appropriately in an important situation (e.g., not responding to an alerting device, being startled by an approaching person or vehicle). Environmental sound recognition has been studied in related disciplines to evaluate such things as awareness of, and tolerance for, sounds in the environment (Dawson et al, 2004; Thawin et al, 2006; Dockrell and Shield, 2004; Spaulding et al, 2008); pattern recognition in autistic children (van Lancker et al, 1988); and the development of left- versus right-ear dominance in children (Kraft et al, 1995; Kraft, 1982). Fabiani et al (1996) used an environmental-sounds naming paradigm to examine, among other things, the effects of developing and aging cognition. Their experimental groups included typically developing children (ages 5–7, 9– 11, and 14–16 yr) as well as younger (ages 19–34 yr) and older adults (ages 61–88 yr). Using somewhat strict naming criteria for the environmental sounds, they found a significant increase in naming accuracy as a function of age that reached adult-like performance by ages 14–16 yr. Performance decreased significantly for the older adults such that their naming accuracy was equivalent to that of the youngest children. Although the reasons for the poorer performance by the youngest and oldest listeners were not addressed directly, the children may have been less familiar with environmental sounds because of inexperience, whereas the mild to moderate HLs reported for the older adults may have reduced their ability to recognize the sounds. One study examined the effect of hearing aid signal processing (frequency lowering) on listeners’ recognition of environmental sounds (Ranjbar et al, 2008); how-

ever, the participants were children and adults with normal hearing (NH), so the results cannot be applied with confidence to individuals with HL. There are currently no data regarding the effects of amplification on the identification of environmental sounds in children or adults with mild to moderately severe HL. To determine the impact of HL on environmental sound identification, the same listening conditions would need to be applied to both NH and hearing-impaired listeners. Referring back to the auditory scene of the busy city street described earlier, a hearing aid user may need to respond to any number of sounds in the environment either indirectly (i.e., modifying his or her behavior) or directly (i.e., interacting with the sound source). In this environment, both fast- or slow-acting compression may cause problems for the hearing aid user. Specifically, fast-acting compression will effectively reduce the startling effects of sudden and loud sounds (e.g., shout, car horn) but will amplify lower-level street noise such that the effective signal-to-noise ratio will be poorer than the nominal signal-to-noise ratio. Put another way, the listening environment will sound noisier to a hearing aid user than to someone without a hearing aid (Naylor and Johannesson, 2009). On the other hand, slow-acting compression will respond well to sudden high-level sounds, but the longer release time will cause the amplitude of later sounds to be reduced such that the hearing aid user will receive less-thanadequate audibility for a period (i.e., the duration of the release time). To address these complex situations, Moore et al (2010) described a third option in the form of an amplitude compressor with adaptive-time constants. Such a system is “… usually slow-acting but the gain is rapidly reduced by the fast system in response to sudden large increases in sound level. If the increase in sound level lasts for only a short time, then the gain rapidly returns to the value set by the slow-acting system. This provides protection from brief intense transients, with little effect on the long-term gain.” (p. 369). With a device using both slow- and fast-acting compression, the hearing aid user may have a better chance of recognizing lower-level acoustic events (speech or environmental sounds) that occur in close temporal proximity to loud acoustic events. In the present study, the effects of three configurations of amplitude compression (slow-, fast-, and adaptivetime constants) were examined in regard to the perception of meaningful acoustic stimuli in complex listening environments. The purpose of the study was to determine the manner in which the three forms of compression affect children’s and adults’ accuracy for categorizing speech and environmental sounds, particularly when the stimulus is partially masked by another stimulus responsible for triggering the amplitude compression. It was hypothesized that the performance of the adults would exceed that of the



children and that the listeners with NH would perform better than those with HL. It was also hypothesized that performance with the adaptive-time constants would exceed performance for the slow- or fast-time constants. These results were expected to be most apparent for partially masked stimuli in which only a few milliseconds of information is available. METHOD Participants Participants were 21 adults between 21 and 30 yr old (mean age 5 24 yr) with NH and 26 adults between 50 and 78 yr old (mean age 5 66 yr) with permanent HL (Table 1). (Originally, the adults with HL were divided into two age groups (age ranges 50–65 yr and 66–78 yr) in an effort to detect age effects. The groups were combined when it was determined that performance and age were not significantly related.) There were 20 children with NH and 15 children with permanent HL who also participated. Both groups were between 7 and 12 yr old (mean age 5 10 yr). At the time of testing, the outer and middle-ear status of each participant was determined using otoscopy and tympanometry, respectively, and confirmed to be consistent with the existing underlying aural pathologic features. Pure-tone thresholds were also obtained using a standard clinical audiometer and insert earphones. All testing was conducted in a sound-treated booth meeting American National Standards Institute (ANSI) standards for room noise (ANSI, 1999). The participants with NH had pure-tone thresholds of 20 dB HL or less for octave frequencies between 250 and 8000 Hz. Figure 1 shows the average (61 SD) hearing thresholds for the right and left ears of the children and adults with HL in the upper and lower panels, respectively. On average, the participants with HL had mild to moderately severe HLs of similar configurations. The results are shown in units of dB SPL to facilitate the display of the stimulus levels (dashed lines) in a format consistent with that used during hearing aid fitting (discussed below). Stimuli Speech Stimuli The speech stimuli were three lists of 36 common words containing equal numbers of unique items referring to people, foods, and animals [see McFadden and Pittman (2008) and Pittman (2011) for a detailed description of these stimuli]. All of the words were nouns and were within the vocabulary of first-grade children (Moe et al, 1982). The words varied in length from one to three syllables and were spoken by a female talker having a standard American-English dialect. Digitized recordings were made at a sampling rate of

Figure 1. Average right and left hearing thresholds using standard audiometric symbols are shown in units of sound pressure level (dB SPL) as a function of frequency for the children with HL (upper panel) and the adults with HL (lower panel). Error bars represent 61 SD. Also shown is the average level of the stimuli presented to the right and left ears of each listener (dashed lines).

22.05 kHz and edited to create individual, normalized audio files (Adobe Audition, CS5.5). The appendix contains the speech stimuli sorted alphabetically by category. It should be noted that the relative difficulty of the stimulus lists is not known. Environmental Sound Stimuli The environmental sounds also comprised three sets of 36 unique sounds representing biologic, mechanical, and animal sound sources. The sounds were drawn from multiple sources including several websites (www.findsounds. com, www.sound-ideas.com, www.hollywoodedge.com, sound-effects-library.com), those available through a royalty-free sound-effects feature in Adobe Audition (CS5.5), and those generously shared by Dr. Valeriy Shafiro in the Auditory Research Laboratory at Rush University Medical Center, Chicago, IL. All files were converted to wave file format having a single (mono) channel with a sampling rate of 22.05 kHz and 32-bit resolution and normalized. The appendix contains the complete list of sounds in each category sorted alphabetically. As with the speech stimuli, the relative difficulty of the stimulus lists is not known.



half (lower panel). The expected effects of these parameters on performance are described below.

Acoustic Competitor Playground noise was used as an acoustic competitor to the speech and environmental sounds. This competitor was obtained from Adobe Audition (CS5.5) and contained sounds consistent with a neighborhood park or recreation area where children are playing. Stimulus Trials Each trial contained a speech stimulus, an environmental sound, and the playground noise as depicted in the upper and lower panels of Figure 2. The upper waveform (trigger) represents a high-level stimulus designed to trigger the amplitude compression. The middle waveform (target) represents the stimulus to which the listener was instructed to respond. The lower waveform represents a low-level playground noise that was presented continuously throughout each 1.5–3 sec trial. The target was presented at a conversational level of 65 dB SPL, whereas the presentation level of the trigger and the playground noise was 6 dB above and 4 dB below that of the target, respectively. Thus, the target, triggering, and playground stimuli were presented at 65, 71, and 61 dB SPL, respectively. Put another way, the target and trigger stimuli were 10 and 4 dB above the playground noise, respectively. Figure 2 also shows the two temporal arrangements of the triggering and target stimuli. In half of the trials, the two stimuli were separated by z400 msec (upper panel), whereas in the remaining trials the two stimuli overlapped by roughly

Figure 2. Schematic of the complex stimuli for the nonoverlapping and overlapping trials.

Amplitude Compression Speed The stimuli were processed digitally by engineers at Oticon A/S using conventional parameters for slowacting (attack time z20 msec, release time z800 msec) and fast-acting (attack time z10 msec, release time z40 msec) compression, as well as a custom adaptivecompression strategy specific to this manufacturer’s devices (i.e., SpeechGuard). The adaptive compression strategy uses two internal level estimators (one for increases in input level and one for decreases in input level) as well as a difference detector. The difference detector tracks the magnitude of the level changes and computes new time constants accordingly. When there are no large changes in the input level, longer time constants are applied (.800 msec). When a large increase in input level occurs, short-time constants are applied, which causes an instantaneous, short-term reduction in gain. When a large decrease in input level is detected, slow-time constants are again applied. This allows the system to quickly follow increases and decreases in input level. A nominal compression ratio of 3:1, a knee point of 40 dB, and 4 channels having crossover frequencies at 625, 1562.5, and 3125 Hz were used in all three compression schemes for an overall stimulus level of 65 dB SPL. The nominal compression ratio was limited to 3:1 to impose some, but not extensive, distortion to the amplitude variations between vowels and consonants (Hornsby and Ricketts, 2001; Boike and Souza, 2000), as would be expected in a fast-acting system. No frequency shaping or maximum power output (OSPL90) parameters were applied during digital processing. Figure 3 shows the effects of the compression parameters on the environmental and speech stimuli. Each panel shows the envelope of the original waveform (dashed line) and the envelope of the compressed waveform (solid line). In this example, the stimuli were a birdcall overlapping with the word “lifeguard.” The amplitude of the second segment of the birdcall is twice that of the first, and the lowest amplitude occurs during the fricative “f ” in the word “lifeguard.” The solid line represents the envelope of the waveform after the slow-, fast-, and adaptive-compression parameters were applied and are shown in the upper, middle, and lower panels, respectively. All three compression parameters effectively lowered the amplitude of the birdcall, but the gain applied to the word “lifeguard” differed substantially. As expected, slow compression maintained the amplitude variations within the waveform but decreased the level overall. Fast compression amplified the playground noise and the target but reduced the amplitude variations within the waveform and decreased the overall signal-to-noise ratio. Adaptive compression amplified



Figure 4. Normalized amplitude envelopes of the fast (black), slow (red), and adaptive (blue) compression stimuli.

Figure 3. Normalized amplitude envelopes of the original (dashed line) and compressed stimuli (solid line) for the slow (upper panel), fast (middle panel), and adaptive (lower panel) compression conditions.

the target and preserved the amplitude variations within the waveform. Figure 4 shows the compressed stimuli displayed together in one panel such that the relative amplitude of each segment of the waveforms is revealed. Referring back to Figure 2, the nonoverlapping and overlapping arrangement of the stimuli were designed to reveal the listener’s perception of partial and wholetarget stimuli with respect to the unique characteristics of the three amplitude compression speeds. That is, the same compression effects were expected for both configurations of the stimuli, but the listeners received only a partial stimulus in the overlapping condition. Performance in the overlapping condition was expected to reveal the compression condition that best preserved the integrity of the target stimulus. Stimulus Presentation The stimuli were presented binaurally under earphones (Sennheiser 25D) having a flat frequency response to

10 kHz. The presentation level of the target stimuli was set to a conversational level (65 dB SPL) for the listeners with NH. Custom laboratory software was used to shape the long-term spectra of the stimuli to accommodate the hearing thresholds of each listener with HL. Amplification targets for average conversational speech were calculated using the Desired Sensation Level (DSL) v5.0 fitting algorithm (Seewald et al, 1997). The age, transducer, and hearing thresholds of each participant were entered into custom DSL v5.0 laboratory software provided by members of the Child Amplification Laboratory at the University of Western Ontario. Figure 1 shows the average frequency shaping (dashed lines) relative to the hearing threshold of the children and adults in the upper and lower panels, respectively. The hearing thresholds were obtained under earphones (insert or supra-aural) and converted to dB SPL relative to the appropriate coupler (2 cm3 or 6 cm3, respectively). The audibility of the frequency-shaped stimuli was calculated relative to the listener’s hearing thresholds using the following formula, which is a modified version of the speech intelligibility index (ANSI, 1997): AI5

18 1 X ½Li hi W i 30 i51

ð1Þ

In this formula, i is the center frequency of each thirdoctave band between 0.1 and 8 kHz, u is the hearing threshold in dB SPL (interpolation was applied at interoctave frequencies), and L is the level of the amplified speech and environmental sounds (in dB SPL). These values were taken from the long-term spectrum of the stimulus transduced by a supra-aural earphone (Sennheiser 25D) and measured in a 6 cm3 coupler. The long-term spectrum was composed of 18 third-octave bands analyzed with a 40 msec Hanning window (50% overlap). (The long-term spectra of the speech and environmental sounds differed by ,3 dB at frequencies between 0.2 and 8 kHz.) Sensation levels (L-u) across the thirdoctave bands were restricted to a range of 0–30 dB and multiplied by the importance of each band (W) using weights for average speech (ANSI, 1997), and summed.



Unfortunately, a 6 cm3 coupler-to-real-ear transform does not exist, so the resulting articulation index (AI) values are a close approximation to real-ear sound pressure levels. AI values were calculated for each ear of each individual and averaged to provide one AI value per listener. These values range from 0–1 with 0 indicating that the stimuli were not audible (i.e., they were below threshold), whereas an AI value of 1 indicates that the stimuli were fully audible. Because no data are available regarding the audibility required to optimize environmental sounds, amplification parameters were set to maximize the audibility of speech. Those values were referenced to Stelmachowicz et al (2000), who examined the relationship between speech perception and audibility for both children and adults with NH as well as children with HL. Their results showed that by age 8 yr, perception scores of at least 90% can be expected for AI values of 0.5. In the present study, the AI values calculated for the children and adults with HL were 0.86 (SD 5 0.12) and 0.77 (SD 5 0.15), respectively. These AI values were considered to be sufficient to optimize performance such that the subtle acoustic effects of amplitude compression could be detected. Procedures Listeners participated in a total of 6 test conditions (3 compression conditions 3 2 target stimuli) having 36 trials each. The experimental task was to correctly identify as many (a) speech stimuli or (b) environmental sound stimuli as possible under different test conditions. Performance (in % correct) was calculated for each participant in each test condition. For three of the test conditions, an environmental sound triggered the compression, whereas in the other three listening conditions, a speech stimulus triggered the compression. Half of the 36 trials in each condition contained overlapping stimuli whereas the stimuli did not overlap in the other half of the trials. The overlapping and nonoverlapping trials were randomly presented within each listening condition so that the temporal pattern of the trigger and target stimuli would not become familiar to the listeners and aid their performance. The listeners were instructed to ignore the triggering stimulus (the first sound heard) and indicate the category to which the target stimulus belonged (the second sound heard). A visual aid was provided to remind the listeners of the response categories (i.e., person/food/ animal, person/animal/thing). No trials contained pairs of waveforms from the same categories, although one or two interesting combinations of words and environmental sounds slipped through the quality-control process before data collection (horse neighing followed by the word “cowboy”; rat squeaking followed by the word “lawyer”). Listeners were given the option to say, “I don’t

know” or to guess if they were unable to respond with confidence. This effectively reduced the rate of chance performance to 25% or less. The order of the six listening conditions was counterbalanced across participants to distribute learning effects equally across the conditions. One of the challenges in adult-child research is finding a task for both children and adults that is not too easy or too difficult for either group. The problem is further confounded when the groups include listeners with normal and impaired hearing. To moderate the performance of each group, the listeners were also asked to perform a visual task while responding to the auditory stimuli. At the start of each session, the listener was handed an 8-1/2 3 11-inch page containing 20 rows of patterns of simple shapes (3 of the patterns are shown in Figure 5). The listener was instructed to solve the pattern and circle the shape that would occur next from the choices provided at the end of each row. Approximately two pages of patterns were completed per condition. The visual task served as a competitor to the auditory stimuli and prevented performance (particularly for the listeners with NH) from reaching a ceiling in all listening conditions. Also, the visual task occupied the attention of the listeners such that reinforcement for correct or incorrect responses to the auditory stimuli was not expected or provided. Previous research has shown that performance for similar visual tasks is not affected by the difficulty of the auditory tasks in children with NH and HL (Pittman, 2011). Therefore, the visual task was used as a competitor to the auditory task only and was not scored. In accordance with the policies of the Internal Review Board at Arizona State University, informed consent was obtained from the adult listeners before testing. Likewise, parental consent and child assent were obtained for the children. Testing required no more than 2 hr, and each listener was paid for his or her participation. RESULTS

T

he data for each listening condition were first separated into overlapping and nonoverlapping trials and were then reduced to a percent correct score. These values were then arc-sine transformed to equalize the variance across the range of scores (Studebaker, 1985) and were averaged across groups and conditions. Analyses included an examination of the results for the listeners with NH to determine the effects of age and

Figure 5. Example of the visual pattern-completion task.



stimulus condition. The overall performance of the listeners with NH and HL was then compared to identify effects of HL. Finally, the performance of the listeners with HL was examined in greater detail to determine the effects of compression and overlap for each type of stimulus. This approach provided a framework with which to assess the performance of the listeners with HL relative to the optimal performance of the listeners with NH so that reasonable expectations may be made regarding the degree to which performance was affected by HL and stimulus condition. HL versus NH Figure 6 shows the results for the speech stimuli in the upper panel and for the environmental sounds in the lower panel for the listeners with NH. The hatched-and-filled bars represent performance in the nonoverlapping and overlapping conditions, respectively. Each set of bars represents a different amplitude compression condition. As expected, performance was poorer for the overlapping stimuli, and the children performed more poorly than the adults. These observations were examined via repeated-measures analysis of variance (ANOVA) with age (adults, children) as the between-

Figure 6. Average performance of the listeners with NH for the speech and environmental sound categorization task shown as a function of age group. The parameter in each panel is compression and overlapping stimulus condition.

participant factor and overlap (nonoverlapping, overlapping), compression speed (slow, fast, adaptive), and target stimuli (speech, environmental sounds) as the within-participant factors. Significant main effects of age [F(1,39) 5 49.9, p , 0.001, h2 5 0.562, b 5 1], overlap [F(1,39) 5 178, p , 0.001, h2 5 0.821, b 5 1], speed [F(1,78) 5 3.353, p 5 0.04, h2 5 0.79, b 5 0.617], and stimuli [F(1,39) 5 101, p , 0.001, h2 5 0.723, b 5 1) were revealed. Also, significant speed 3 overlap [F(2,78) 5 15.6, p , 0.001, h2 5 0.286, b 5 0.99] and stimuli 3 age [F(1,39) 5 5.9, p 5 0.02, h2 5 0.132, b 5 0.659] interactions were found. All other interactions were not significant. These results confirm that, for the listeners with NH, performance was significantly affected by the age of the listener (adults . children), the degree to which the stimulus was masked by the trigger (nonoverlapping . overlapping), and the type of stimuli (speech . environmental sounds). As for the significant main effect of compression speed, pairwise comparisons (with Bonferroni adjustment) revealed that performance was significantly higher for the slow compression condition relative to fast and adaptive compression. Figure 7 shows the results for the listeners with HL using the same convention as in Figure 6 and shows that overall, the performance of the listeners with HL was poorer than their NH counterparts. Separate univariate

Figure 7. Average performance for the listeners with HL using the same convention as in Figure 6.



Table 1. Demographic Characteristics of the Children and Adults with HL including Gender, Age (in yr), and Hearing Thresholds (in dB HL) Right Ear #

Left Ear

M/F

Age

0.25

0.5

1

2

4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

M M M F F F F M M F F F M F M M M M F F F M F M F F F M M F F

7 7 9 9 9 11 11 12 12 12 9 72 65 68 59 60 62 64 67 69 69 76 78 61 63 76 60 62 64 50 70

40 40 45 5 25 25 10 10 25 5 30 30 25 35 20 55 30 35 60 5 25 40 25 20 15 20 70 15 20 45 40

40 50 55 20 30 35 15 25 35 10 35 30 35 35 15 65 35 25 60 10 25 55 20 20 20 45 60 15 25 50 50

45 65 65 50 40 50 10 40 45 35 40 45 40 45 10 65 45 35 65 35 45 75 30 30 50 45 50 15 50 65 55

55 65 90 65 35 55 45 50 45 50 55 55 60 45 10 70 60 55 60 55 55 75 55 45 55 50 50 30 75 65 55

50 65 70 45 30 45 60 55 35 45 45 65 95 55 70 80 60 60 65 55 65 80 65 55 45 45 60 50 70 60 55

32 33 34 35 36 37 38 39 40 41

F F M M M M M F F F

9 10 12 12 60 54 70 68 78 75

40 5 10 35 35 20 35 10 20 25

45 5 20 40 45 20 35 10 15 20

50 0 40 50 50 25 25 15 25 40

50 5 55 50 45 40 25 20 20 55

50 25 50 55 60 60 40 35 20 60

8

0.25

0.5

Hearing Device

1

2

4

8

40 35 40 45 60 40 50 60 55 45 55 70 10 10 20 50 30 25 35 30 45 25 45 50 55 10 5 10 70 10 25 40 50 25 35 45 45 5 10 35 35 30 35 50 55 20 20 45 NR 25 35 45 60 35 35 35 75 20 25 15 85 60 70 70 50 35 50 40 85 80 65 65 90 60 60 60 50 10 15 25 70 30 40 45 70 55 75 75 75 10 15 25 50 20 15 25 45 20 25 45 65 25 35 45 75 45 40 40 50 20 20 25 70 20 20 35 55 40 45 55 65 45 55 55 Not Using Amplification 45 35 40 50 45 5 0 5 70 10 15 25 40 30 35 50 45 35 45 50 85 20 20 25 50 25 25 25 45 10 10 15 65 20 15 15 65 20 25 35

55 65 70 65 30 50 45 80 40 40 55 65 50 50 25 65 75 55 60 50 50 80 35 55 55 45 30 35 55 75 50

50 60 80 45 25 65 55 70 40 45 50 65 75 55 75 80 65 80 60 55 80 75 55 55 50 55 60 40 60 70 60

40 65 65 10 25 65 60 75 50 45 40 55 90 60 70 80 60 NR 85 60 75 75 70 50 50 65 60 40 70 60 80

45 0 40 55 45 35 35 15 15 55

50 30 50 50 60 55 45 50 20 70

45 55 65 45 50 65 55 40 65 75

Make

Model

Style

O P P P P P P P P S W CC I MT O O O O O O P P P RS RS RS R ST ST W W

Safari P Solana SP Exelia Art P Nios S H20 III Cassia M H20 Solana Micro P Nios Micro III Nios H20 V Solana Micro P Destiny 1200 AK-9 KSHA03 IIQ43 Seneca Agil Pro Alta P VPC-I Acto Pro unknown Syncro V2 Audeo Smart V Ambra Micro P Versata M Pulse Forza unknown RX34 VQ820 VQ720 Mind 440 Senso Diva

BTE BTE BTE BTE BTE BTE RITA ITE BTE BTE BTE RIC ITE BTE RIC BTE ITE RIC BTE BTE RIC BTE BTE RITA RIC RIC ITE RIC RIC ITE BTE

Note: Also shown is the make, model, and style of hearing devices for those who used amplification. BTE 5 behind the ear; CC 5 Costco; I 5 Interon; ITE 5 in the ear; MT 5 MicroTech; NR 5 no response; O 5 Oticon; P 5 Phonak; R 5 Rexton; RIC 5 receiver in the canal; RITA 5 receiver in the aid; RS 5 ReSound; S 5 Starkey; ST 5 SeboTek; W 5 Widex.

ANOVAs for the speech and environmental sounds were performed to compare the overall performance of the NH and HL groups. For each analyses, hearing group (NH, HL), age (adults, children), and overlap (nonoverlapping, overlapping) were entered as fixed factors. For the speech stimuli, significant main effects were found for hearing group [F (1,484) 5 455, p , 0.001], age [F(1,484) 5 16.9, p , 0.001], and overlap [F(1,484) 5

338, p , 0.001], as well as significant hearing group 3 age [F(1,484) 5 34.7, p , 0.001] and hearing group 3 overlap [F(1,484) 5 9.6, p 5 0.002] interactions. Not significant were the age 3 overlap and age 3 hearing 3 overlap interactions. For the environmental sounds, significant main effects were found for hearing group [F(1,484) 5 436, p , 0.001] and overlap [F(1,484) 5 116, p , 0.001] but not for age [F(1,484) 5 3.8, p 5 0.053]. A significant



hearing group 3 age [F(1,484) 5 34.7, p , 0.001] interaction was found with all other interactions being nonsignificant. These analyses revealed that (1) the performance of the listeners with HL was significantly poorer than that of the listeners with NH, (2) the children performed more poorly than the adults with the speech stimuli but not the environmental sounds, and (3) all groups performed more poorly in the overlapping condition. The group 3 age interaction for both the speech stimuli and environmental sounds indicates that the largest difference in performance occurred between the adults with NH and the adults with HL. Overall, these analyses indicate that nearly every stimulus and age parameter served to reduce the performance of the listeners with HL relative to their NH counterparts. HL versus Compression Speed Four repeated-measures ANOVAs were conducted to examine the interaction between HL and compression speed for each stimulus and overlap condition. In each analysis, age (adults, children) and compression speed (slow, fast, and adaptive) were entered as between- and within-participant factors, respectively. For significant effects of compression speed, post hoc analyses were conducted with Bonferroni adjustment for multiple comparisons. Starting with the nonoverlapping speech stimuli (hatched bars in the upper panel), a significant main effect for compression speed was observed [F(2,78) 5 14.278, p , 0.001, h2 5 0.268, b 5 0.998] but not for age [F(1,39) 5 7.14, p 5 0.403]. Nor was a speed 3 age interaction revealed [F(2,78) 5 1.722, p 5 0.185]. Pairwise comparisons revealed a significant difference between the slow and fast compression speeds (I-J 5 0.162, p 5 0.012). These results indicate that performance was equally good in both the slow and adaptive compression conditions when the stimuli were not masked by the trigger and that the children with HL performed as well as their adult counterparts. Similar main effects and interactions were found for the overlapping condition (filled bars in the upper panel). A significant main effect of compression speed [F(2,78) 5 27.048, p , 0.001, h2 5 0.410, b 5 1] was found but no effect of age [F(1,39) 5 0.006, p 5 0.940) or significant age 3 compression speed [F(2,78) 5 1.937, p 5 0.151] interaction. Pairwise comparisons of compression speed revealed significantly better performance for adaptive compression compared with both slow (I-J 5 0.193, p , 0.001) and fast (I-J 5 0.199, p , 0.001) compression. These results revealed significantly higher performance with the adaptive compression when the target stimulus was partially masked by the triggering stimulus. Also, performance in the slow and fast compression conditions was equally poor for both the children and adults.

Similar results were again observed for the environmental sounds in the nonoverlapping stimuli (hatched bars in the lower panel). A significant main effect of compression speed [F(2,78) 5 10.163, p , 0.001, h2 5 0.207, b 5 0.983] was revealed but not for age [F(1,39) 5 7.20, p 5 0.401]. For these stimuli, however, no speed 3 age interaction was revealed [F(2,78) 5 1.916, p 5 0.154]. Pairwise comparisons revealed significantly better performance for adaptive compression compared with the slow (I-J 5 0.120, p 5 0.048) and fast (I-J 5 0.210, p , 0.001) compression speeds. Thus, performance was best in the adaptive compression condition when the stimuli did not overlap. As for the overlapping stimuli (filled bars in the lower panel), no main effects of compression speed [F(2,78) 5 1.706, p 5 0.188] or age [F(1,39) 5 0.354, p 5 0.555] were found. However, a significant age 3 compression speed interaction [F(2,78) 5 4.726, p 5 0.012, h2 5 0.108, b 5 0.775] was revealed. Pairwise comparisons failed to reveal significant differences between the compression conditions, indicating that the interaction was likely the result of the difference in performance between groups for the fast compression condition. That is, the adults and children performed similarly in the slow and adaptive compression conditions but not in the fast compression condition. Relationship between Speech and Environmental Sound Performance It was of interest to determine the relationship between the listener’s performance for speech and environmental sound categorization. Shafiro (2008) and other authors have argued that the acoustic and semantic properties of speech and environmental sounds can be exploited to further understand the difficulties that a listener with HL might be experiencing with auditory perception (Fabiani et al, 1996; Gygi and Shafiro, 2007). Table 2 shows the Pearson correlation coefficients relating categorization of words to that of environmental sounds for each group within each compression speed and overlap condition. For example, performance for the words in the nonoverlapping, slow-compression condition was compared with performance for the environmental sounds in the same overlap and compression condition. Asterisks indicate significant one-tailed correlations (p , 0.05). A few correlations were revealed for some groups across conditions; however, the most striking results were those of the adults with HL. Listeners who were able to categorize words well also appeared to do well with environmental sounds, whereas those who did poorly with words also tended to do poorly with the environmental sounds. The relationship between speech and environmental sound performance was significant in every condition for the adults with HL whereas it occurred only in a few conditions for all other groups.



Table 2. Pearson Correlation Coefficients Relating Performance for Word Categorization to Performance for Environmental Sounds within Each Processing Speed Nonoverlapping Age Group Adults Adults Children Children

Overlapping

Hearing Group

Slow

Fast

Adaptive

Slow

Fast

Adaptive

NH HL NH HL

–0.22 0.53* 0.51* 0.43

–0.01 0.64* 0.26 0.48*

0.39* 0.58* 0.29 0.50*

–0.10 0.56* 0.44* 0.13

–0.07 0.49* 0.20 0.42

0.09 0.59* 0.27 0.25

*Significant one-tailed correlations (p , 0.05).

DISCUSSION

T

he goal of this study was to examine the impact of amplitude compression on children’s and adults’ ability to categorize words and environmental sounds in complex listening environments. Three forms of amplitude compression (fast, slow, and adaptive) were simulated in complex acoustic trials containing a triggering stimulus, a target stimulus, and a low-level playground noise. The trigger preceded the target or overlapped with it such that the temporal effects of compression were expected to be most deleterious to the perception of the partial stimuli. Overall, the performance of the listeners with HL was poorer than that of the listeners with NH with unique patterns of performance emerging with respect to age, stimulus overlap, and compression speed for both groups. For example, significant effects of age were observed for the listeners with NH but not for the listeners with HL. Typically, children with HL demonstrate the most difficulty with auditory tasks; however, in this study their performance was equivalent to that of the adults with HL. This unusual finding suggests that either the children with HL performed better than expected or the adults performed more poorly. It is possible that the children were better suited to the dual-task nature of the paradigm than the adults; however, it is also possible that the cognitive skills of some of the adults with HL were in decline, resulting in performance similar to that of children with developing cognitive skills. This finding is supported by the significant correlations between performance for the speech and environmental sounds for the adults with HL in every listening condition. We know that adults with HL who have poor cognitive skills are less able to tolerate the subtle distortions to speech that occur with fast compression (Gatehouse et al, 2006; Souza et al, 2000; Jenstad and Souza, 2005, 2007; Cox and Xu, 2010; Lunner and Sundewall-Thoreń, 2007), and in fact, the adults with HL demonstrated their poorest performance in the fast compression condition. Given that significant correlations were found in every listening condition, it is possible that the effects of declining cognitive skills affected performance in each condition. However, few

correlations between the perception of speech and environmental sounds were found for the children with HL, suggesting that the mechanisms that govern these processes may differ in children relative to adults and are indeed a question for further research. Another example of the differences between the NH and HL groups was that the listeners with NH performed best with slow compression whereas the listeners with HL performed best with adaptive compression. These results suggest that the subtle effects of compression speed interact with healthy and impaired ears in a unique fashion. Why the listeners with NH favored slow-acting compression is unknown and would also be an interesting subject for further research. In terms of the practical application of these results, it is important to bear in mind that, although significant, the performance of the listeners with NH varied by only a few percentage points across compression conditions whereas larger variations in performance were observed for the listeners with HL (as much as 20 percentage points). For these listeners, the different types of subtle distortion caused by slow and fast compression reduced their performance to similarly low levels. These findings are consistent with previous studies showing little differences in speech perception between fast- and slowcompression conditions (Moore et al, 2004, 2010; Shi and Doherty, 2008). Not only does this suggest that listeners with HL would benefit from amplification with adaptive compression, the results support the argument that amplification decisions for adults and children with HL should not be based on the results of listeners with NH. Finally, in the nonoverlapping condition, the performance of the listeners with HL was nearly as good as that of the listeners with NH for the speech stimuli but not for the environmental sounds. Although it is fairly clear that similar cognitive mechanisms are involved in the perception of speech and other acoustic stimuli (Fowler and Rosenblum, 1990), the results of the present study suggest that speech may be prioritized among individuals with HL. That is, adults and children with HL may consider environmental sounds as a source of unwanted noise that may prevent them from interacting with these sounds accurately and



effectively (Banerjee, 2011). In fact, informal interactions with several of the adults with HL at the conclusion of testing suggest that they avoid environmental sounds whenever possible to optimize their speech perception. Such a strategy may be detrimental to both children and adults with HL in that it may limit their social, educational, and vocational interactions as well as their ability to dynamically interact with their environment. Efforts to make environmental sounds accessible to individuals with HL may be of significant benefit to them. Limitations of the Present Study There were a few limitations of the present study. First, and most important, the results should be generalized to hearing aid users with caution. Because the three compression conditions examined could not be generated using a single, wearable device, it was necessary to simulate them through digital processing. Thus, the listeners with HL did not receive amplification with compression characteristics consistent with current multichannel, multistage compression devices. Also, the stimuli do not represent all complex listening environments. Thus, the results confirm the potential for benefit from devices using adaptive compression parameters compared with devices using fast or slow compression parameters. Second, the stimuli were not counterbalanced across compression conditions. The complexity of the stimuli required 216 trials to be created by combining 432 stimuli. To counterbalance the stimuli across compression conditions, a total of 1,296 stimuli would have been required, which was not feasible during the development of the study. It is possible that list effects were present (one list being easier than the others); however, little evidence of such effects can be observed in the results. For example, the better word categorization of the listeners with HL in the adaptive compression condition was not observed in the listeners with NH for the same stimuli. Likewise, the better word categorization for the slow compression condition by the listeners with NH was not observed for the listeners with HL. Whether or not list effects are present, the complexity of the stimuli alone warrants replication of the results with similar or related stimuli to confirm, and possibly identify, further benefits of adaptive compression. Third, the source of the words and environmental sounds differed substantively. That is, a single sound source (talker) was used for the speech stimuli whereas the environmental sounds were gathered from multiple sources. Also, the speech stimuli were recorded in quiet but were presented in noise and lacked the vocal effort that would be expected in such surroundings (Pittman and Wiley, 2001). It is recommended that future research in this area include speech samples from a wide range of talkers produced in the same or similar listening environments.

Finally, the categorization task for the environmental sounds was uncharacteristic of how a listener would respond to these sounds. A clever experiment might require a listener to respond dynamically to both speech and environmental sounds in some fashion. CONCLUSIONS

A

mplitude compression speed imposes subtle alterations to the acoustic signal that can affect the performance of both children and adults with HL. Although fast, slow, and adaptive compression parameters were simulated in the present study, the results suggest that the performance of listeners with HL may be reduced similarly when fast or slow amplitude compression is used. On the other hand, listeners with HL may derive significant benefit from hearing instruments that use adaptive amplitude compression, especially in complex listening environments. Acknowledgments. The authors thank the staff of the Pediatric Amplification Laboratory for their assistance with the data collection (Rachel Henrickson, Tia Mulrooney, and Amanda Willman), Fares El-Azm and Jan-Mark de Haan for their help with stimulus preparation, and Thomas Behrens and Kamilla Angelo for sharing their expertise in this area of hearing aid research. Appreciation also goes to the participants who shared their precious time with us during this project.

REFERENCES American National Standards Institute. (1997). Methods for Calculation of the Speech Intelligibility Index (ANSI S3.5-1997). New York, NY: American National Standards Institute. American National Standards Institute. (1999) Maximum Permissible Ambient Noise Levels for Audiometric Test Rooms (ANSI S3.1 1-1999). New York, NY: American National Standards Institute. Banerjee S. (2011) Hearing aids in the real world: typical automatic behavior of expansion, directionality, and noise management. J Am Acad Audiol 22(1):34–48. Boike KT, Souza PE. (2000) Effect of compression ratio on speech recognition and speech-quality ratings with wide dynamic range compression amplification. J Speech Lang Hear Res 43(2): 456–468. Cox RM, Xu J. (2010) Short and long compression release times: speech understanding, real-world preferences, and association with cognitive ability. J Am Acad Audiol 21(2):121–138. Davies-Venn E, Souza P, Brennan M, Stecker GC. (2009) Effects of audibility and multichannel wide dynamic range compression on consonant recognition for listeners with severe hearing loss. Ear Hear 30(5):494–504. Dawson PW, Decker JA, Psarros CE. (2004) Optimizing dynamic range in children using the nucleus cochlear implant. Ear Hear 25 (3):230–241. Dockrell JE, Shield B. (2004) Children’s perceptions of their acoustic environment at school and at home. J Acoust Soc Am 115(6): 2964–2973.



Fabiani M, Kazmerski VA, Cycowicz YM, Friedman D. (1996) Naming norms for brief environmental sounds: effects of age and dementia. Psychophysiology 33(4):462–475.

Moore BC, Stainsby TH, Alcańtara JI, Ku¨hnel V. (2004) The effect on speech intelligibility of varying compression time constants in a digital hearing aid. Int J Audiol 43(7):399–409.

Fowler CA, Rosenblum LD. (1990) Duplex perception: a comparison of monosyllables and slamming doors. J Exp Psychol Hum Percept Perform 16(4):742–754.

Naylor G, Johannesson RB. (2009) Long-term signal-to-noise ratio at the input and output of amplitude-compression systems. J Am Acad Audiol 20(3):161–171.

Gatehouse S, Naylor G, Elberling C. (2006) Linear and nonlinear hearing aid fittings–2. Patterns of candidature. Int J Audiol 45(3): 153–171.

Pittman A. (2011) Children’s performance in complex listening conditions: effects of hearing loss and digital noise reduction. J Speech Lang Hear Res 54(4):1224–1239.

Gygi B, Shafiro V. (2007) General functions and specific applications of environmental sound research. Front Biosci 12: 3152–3166.

Pittman AL, Wiley TL. (2001) Recognition of speech produced in noise. J Speech Lang Hear Res 44(3):487–496.

Hornsby BW, Ricketts TA. (2001) The effects of compression ratio, signal-to-noise ratio, and level on speech recognition in normalhearing listeners. J Acoust Soc Am 109(6):2964–2973. Jenstad LM, Souza PE. (2005) Quantifying the effect of compression hearing aid release time on speech acoustics and intelligibility. J Speech Lang Hear Res 48(3):651–667. Jenstad LM, Souza PE. (2007) Temporal envelope changes of compression and speech rate: combined effects on recognition for older adults. J Speech Lang Hear Res 50(5):1123–1138. Johnson JA, Cox RM, Alexander GC. (2010) Development of APHAB norms for WDRC hearing aids and comparisons with original norms. Ear Hear 31(1):47–55. Kam AC, Wong LL. (1999) Comparison of performance with wide dynamic range compression and linear amplification. J Am Acad Audiol 10(8):445–457. Kim M, Chang SI, Seong JC, et al. (2012) Road traffic noise: annoyance, sleep disturbance, and public health implications. Am J Prev Med 43(4):353–360.

Ranjbar P, Borg E, Philipson L, Stranneby D. (2008) Auditive identification of signal-processed environmental sounds: monitoring the environment. Int J Audiol 47(12):724–736. Rudner M, Ro¨nnberg J, Lunner T. (2011) Working memory supports listening in noise for persons with hearing impairment. J Am Acad Audiol 22(3):156–167. Seewald RC, Cornelisse LE, Ramji KV, Sinclair ST, Moodie KS, Jamieson DG. (1997). DSL v4.1 for Windows: A software Implementation of the Desired Sensation Level (DSL[i/o]) Method for Fitting Linear Gain and Wide-Dynamic-Range Compression Hearing Instruments. User’s manual. (Version 4.1) [Computer software]. London, ON: The University of Western Ontario Hearing Health Care Research Unit. Shafiro V. (2008) Development of a large-item environmental sound test and the effects of short-term training with spectrally-degraded stimuli. Ear Hear 29(5):775–790. Shafiro V, Gygi B. (2004) How to select stimuli for environmental sound research and where to find them. Behav Res Methods Instrum Comput 36(4):590–598.

Kraft RH. (1982) Relationship of ear specialization to degree of task difficulty, sex, and lateral preference. Percept Mot Skills 54(3):703–714.

Shepherd D, Welch D, Dirks KN, Mathews R. (2010) Exploring the relationship between noise sensitivity, annoyance and healthrelated quality of life in a sample of adults exposed to environmental noise. Int J Environ Res Public Health 7(10):3579–3594.

Kraft RH, Harper LV, Nickel LD. (1995) Stability and ear advantage for dichotically presented environmental sound stimuli: development during early childhood. Percept Mot Skills 80(2): 611–624.

Shi LF, Doherty KA. (2008) Subjective and objective effects of fast and slow compression on the perception of reverberant speech in listeners with hearing loss. J Speech Lang Hear Res 51(5): 1328–1340.

Larson VD, Williams DW, Henderson WG, et al.; NIDCD/VA Hearing Aid Clinical Trial Group. (2000) Efficacy of 3 commonly used hearing aid circuits: A crossover trial. JAMA 284(14): 1806–1813.

Smith AW. (1998) The World Health Organisation and the prevention of deafness and hearing impairment caused by noise. Noise Health 1(1):6–12.

Looi V, Arnephy J. (2010) Environmental sound perception of cochlear implant users. Cochlear Implants Int 11(4):203–227. Lunner T, Sundewall-Thoreń E. (2007) Interactions between cognition, compression, and listening conditions: effects on speech-innoise performance in a two-channel hearing aid. J Am Acad Audiol 18(7):604–617. McFadden B, Pittman A. (2008) Effect of minimal hearing loss on children’s ability to multitask in quiet and in noise. Lang Speech Hear Serv Sch 39(3):342–351.

Souza PE. (2002) Effects of compression on speech acoustics, intelligibility, and sound quality. Trends Hear 6(4):131–165. Souza PE, Jenstad LM, Folino R. (2005) Using multichannel widedynamic range compression in severely hearing-impaired listeners: effects on speech recognition and quality. Ear Hear 26(2): 120–131. Souza PE, Turner CW. (1999) Quantifying the contribution of audibility to recognition of compression-amplified speech. Ear Hear 20(1):12–20.

Moe AH, Hopkins CJ, Rush RT. (1982) The Vocabulary of First Grade Children. Springfield, IL: Thomas Publishers.

Souza PE, Yueh B, Sarubbi M, Loovis CF. (2000) Fitting hearing aids with the Articulation Index: impact on hearing aid effectiveness. J Rehabil Res Dev 37(4):473–481.

Moore BC, Fu¨llgrabe C, Stone MA. (2010) Effect of spatial separation, extended bandwidth, and compression speed on intelligibility in a competing-speech task. J Acoust Soc Am 128(1): 360–371.

Spaulding TJ, Plante E, Vance R. (2008) Sustained selective attention skills of preschool children with specific language impairment: evidence for separate attentional capacities. J Speech Lang Hear Res 51(1):16–34.



Stelmachowicz PG, Hoover BM, Lewis DE, Kortekaas RW, Pittman AL. (2000) The relation between stimulus context, speech audibility, and perception for normal-hearing and hearing-impaired children. J Speech Lang Hear Res 43(4):902–914. Stone MA, Moore BC. (2003) Effect of the speed of a single-channel dynamic range compressor on intelligibility in a competing speech task. J Acoust Soc Am 114(2):1023–1034.

Tyler RS. (1994) Advantages of disadvantages expected and reported by cochlear implant patients. Am J Otol 15(4):523–531. van Lancker D, Cornelius C, Kreiman J, Tonick I, Tanguay P, Schulman ML. (1988) Recognition of environmental sounds in autistic children. J Am Acad Child Adolesc Psychiatry 27(4): 423–427.

Studebaker GA. (1985) A “rationalized” arcsine transform. J Speech Hear Res 28(3):455–462.

Walden BE, Surr RK, Cord MT, Edwards B, Olson L. (2000) Comparison of benefits provided by different hearing aid technologies. J Am Acad Audiol 11(10):540–560.

Thawin C, Kanchanalarp C, Lertsukprasert K, Cheewaruangroj W, Khantapasuantara K, Ruencharoen S. (2006) Auditory performance of cochlear implant children aged 2-5 years. J Med Assoc Thai 89(11):1923–1927.

Zhao F, Stephens SD, Sim SW, Meredith R. (1997) The use of qualitative questionnaires in patients having and being considered for cochlear implants. Clin Otolaryngol Allied Sci 22(3): 254–259.

Appendix. Speech and Environmental Sounds Stimuli by Category Speech

Environmental Sounds

Animal

Food

Person

Mechanical

Biological

Animal

alligator bat bear buzzard camel cat chipmunk cow crocodile deer dolphin donkey eagle elephant fox frog giraffe goldfish gorilla hamster horse lion mouse octopus owl penguin porcupine rabbit raccoon seal squirrel toad walrus weasel whale zebra

apple bacon banana cabbage cake candy carrots cheese cherry chocolate corn cracker cream donut fudge grape hamburger hotdog jelly lollipop macaroni meatloaf milk noodle oatmeal pancake pineapple pizza popcorn popsicle potato sausage soup strawberry sugar tomato

astronaut baby boy brother burglar captain carpenter clown cousin cowboy dentist detective doctor father friend girl girlfriend grandma grandpa janitor king lifeguard mailman man pilot president priest prince queen reporter robber sailor sister uncle wife woman

alarm1 alarm2 camera1 camera2 carhonk1 carhonk2 door1 door2 doorbell garbage can helicopter1 helicopter2 jar opening lawnmower phone1 phone2 pop1 pop2 pop3 power tool1 power tool2 power tool3 school bell scrape1 scrape2 scrape3 siren1 siren2 squeak1 squeak2 squeak3 tool1 tool2 train trash lid1 trash lid2

blow nose1 blow nose2 burp1 burp2 burp3 clear throat cough1 cough2 cough3 cough4 cry1 cry2 footsteps gargle gasp1 gasp2 gasp3 grunt1 grunt2 grunt3 hiccups kiss1 kiss2 laugh1 laugh2 laugh3 scream1 scream2 sigh slurp sneeze1 sneeze2 sneeze3 whistle1 whistle2 yawn

alligator bear1 bear2 bear3 bear4 bird1 bird2 bird3 bird4 bird5 cat1 cat2 chicken1 chicken2 cow dog1 dog2 donkey frog1 frog2 horse1 horse2 horse3 lamb1 lamb2 monkey1 monkey2 monkey3 monkey4 owl1 owl2 owl3 pig rat seal sheep
