ihaff loudness contour test: reliability and effects of ... · 1haff loudness contour test:...

J Am Acad Audiol 8 : 243-256 (1997)

1HAFF Loudness Contour Test : Reliability and Effects of Approach Mode in Normal-Hearing Subjects Randall C. Beattie* Rosalie C . Huynh* Van N. Ngo* Romayn L. Jones*

Abstract

This study investigated the effects of approach mode (ascending, descending, random) on the normal loudness function using the Loudness Contour Test suggested by the Independent Hearing Aid Fitting Forum . We also assessed how increasing the number of trials from one to five affects the short-term reliability of the Loudness Contour Test . Additionally, we exam-ined the relationship between loudness judgments (very soft to uncomfortably loud) for warble tones and loudness judgments for speech . Thirty-one normal-hearing subjects were tested using 500-Hz and 3000-Hz tones and speech (CID W-22 words preceded by a carrier phrase). The results revealed that 5- to 12-dB higher SPLs were found with the descending approach than with the ascending approach . The reliability results suggest that it is not beneficial to present more than one or two trials when administering the Loudness Contour Test . The true sound pressure level (SPL) for a given loudness category will fall within ±10 to 12 dB of the obtained SPL approximately 95 percent of the time . This 20- to 24-dB range is fairly large and it is questionable whether reliability is sufficient to justify obtaining individual loudness measurements . When the relationship between warble tones and speech was investigated, the results revealed an overall standard error of estimate of approximately 8 dB . These data suggest that warble tones are not accurate predictors of the corresponding loudness cate-gories for speech . If the clinician wants to ascertain SPLs that correspond to the various loudness categories for speech, then direct measurement with a speech stimulus appears necessary.

Key Words: Hearing aids, Independent Hearing Aid Fitting Forum, Loudness Contour Test, reliability, warble tones and speech

Abbreviations : ANSI = American National Standards Institute, IHAFF = Independent Hearing Aid Fitting Forum, PTA = average pure-tone hearing threshold level

T

he Independent Hearing Aid Fitting Forum (IHAFF) was formed in 1993 to develop a standardized hearing aid eval-

uation protocol that could be used for fitting linear, nonlinear, and/or programmable hear-ing aids . The result was a document (IHAFF, 1994) that describes a protocol consisting of three components : (1) the Abbreviated Profile of

*Department of Communicative Disorders, California State University, Long Beach, California

Reprint requests : Randall C . Beattie, Department of Communicative Disorders, California State University, Long Beach, Long Beach, CA 90840

Hearing Aid Benefit (APHAB), a self-report inventory to assess communicative difficulty and the benefit derived from amplification (Cox and Alexander, 1995); (2) the Loudness Contour Test, which generates a loudness growth func-tion (Cox, 1995); and (3) the Visual Input/Out-put Locator Algorithm (VIOLA), which is used to select certain electroacoustic characteristics of a hearing aid based on an individual's loud-ness growth function (Cox, 1995). These com-ponents are available as the IHAFF Suite (Van VIiet, 1995) and consist of three computer pro-grams for administration of the APHAB, Loud-ness Contour Test, and VIOLA. The IHAFF group (1994) states that this protocol will serve

Journal of the American Academy of Audiology/Volume 8, Number 4, August 1997

as an excellent starting point for identifying the key electroacoustic needs of the user, and stan-dardized procedures should make it easier to identify those factors that contribute to dissat-isfied hearing aid users. Because the present study was designed to investigate the Loudness Contour Test, the following discussion will focus on that component of the IHAFF protocol .

The IHAFF group (1994) states that an optimal hearing aid fitting will restore normal loudness relationships among speech and envi-ronmental sounds over the range of typical input levels . That is, any hearing aid fitting should include the following goals for amplified stimuli: (1) soft sounds/speech should be audi-ble, (2) environmental sounds/speech that are comfortably loud to normal-hearing listeners also should be comfortable for hearing-impaired listeners, and (3) loud sounds/speech should not be uncomfortable. Because loudness per-ceptions vary widely among individuals, even those with similar hearing losses (Skinner, 1988, p. 155), a loudness growth function must be established for each hearing aid candidate. A standardized loudness perception procedure is recommended to obtain consistent results across clinicians, settings, and test sessions . The IHAFF (1994) procedure recommends pre-senting pulsed, warble tones and having the lis-teners categorize their loudness perceptions on a 7-point scale modified from Hawkins et al (1987) . These categories and their correspond-ing numbers are as follows: (1) very soft, (2) soft, (3) comfortable, but slightly soft, (4) comfortable, (5) comfortable, but slightly loud, (6) loud, but okay, and (7) uncomfortably loud . The IHAFF (1994) group contends that these loudness mea-surements may allow "substantially better" hearing aid fittings because contemporary hear-ing aids enable clinicians to make use of this additional information.

Loudness growth should be assessed at 500 Hz and 3000 Hz because these frequencies ". . .most likely represent the regions of maxi-mum output for the low- and high-frequency bands of multiple-band compression instru-ments, and are also appropriate for most contemporary single, channel broad band instru-ments" (IHAFF, 1994, p. 11). Although at least two frequencies are required, the authors note that additional frequencies will add precision to the loudness function . Insert earphones are rec-ommended to allow ease of conversion to HA-1 (2-cc) electroacoustic measures of hearing aid performance. Either a manual or computer-assisted (IHAFF Suite) presentation method

may be used . The recommended procedure is an ascending approach mode . Starting at threshold, the stimuli (four warble pulses) are increased in 5-dB steps until the uncomfortably loud level (category 7) is reached. Following a practice run, three or four ascending test trials are con-ducted . The median of the values averaged across all three or four trials is taken as the loud-ness percept in each category. The intent of this testing is to define the HA-1 (2-cc) sound pres-sure level (SPL) values for each of the seven loudness categories from very soft to uncom-fortably loud .

An ascending procedure is recommended because " . . .dispensers and patients seem to prefer a procedure that progresses from softer to louder over one that uses random or descend-ing levels" (Cox, 1995, p. 39) . In contrast, the loudness growth procedure recommended by the Resound Corporation (Pluvinage, 1994) employs a random presentation of stimulus lev-els . Moreover, several authors have found that a descending approach yields higher loudness levels than the ascending approach over the middle portion of the loudness range (Woods et al, 1973; Kamm et al, 1978; Ventry and John-son, 1978 ; Berger et al, 1982); however, the effect of approach mode over the entire loudness range has not been systematically studied. Therefore, one purpose of this investigation was to use the Loudness Contour Test to describe the normal loudness function using ascending, descending, and random approach modes. Because the normal loudness function may be used as a baseline for prescribing gain for hearing aid users, it is important to deter-mine if approach mode has a differential effect on loudness growth . As noted above, the IHAFF group (1994) recommends defining the loud-ness growth function with warble tones; 500 Hz and 3000 Hz typically are measured for hear-ing aid selection. Because speech is the major signal of interest for everyday listening, it is important to document the relationship between warble tones and speech for the seven loud-ness categories . Aided speech may also be used to verify several loudness categories (IHAFF, 1994). Additionally, speech intelligibility may be measured unaided and aided at soft, comfort-able, or loud levels with monosyllables pre-ceded by a short phrase such as "Say the word -" (Rupp and Stockdell, 1980 ; Konkle and Rin-telmann, 1983 ; Olsen and Matkin, 1991). In view of the foregoing considerations, the present study was designed to describe the normal loud-ness function for 500 Hz, 3000 Hz, and speech

244

using the IHAFF (1994) instructions and employing ascending, descending, and random approach modes . We also investigated the rela-tionship between warble tones (500 Hz and 3000 Hz) and speech for the seven loudness categories .

Cox (1995) comments that it is desirable to select a procedure providing acceptable test-retest reliability within a reasonable time . Several studies have investigated the effects of immediate (same test session) or short-term (test sessions separated by about 1-14 days) test-retest reliability on the most comfortable listening level (MCL) or loudness discomfort level (LDL). These studies indicate that approx-imately 95 percent of test-retest LDLs vary within 4 to 8 dB (Beattie et al, 1979 ; Edgerton et al, 1980 ; Beattie and Sheffler, 1981) and about 95 percent of test-retest MCLs vary within 8 to 12 dB (Beattie and Culibrk, 1980 ; Christen and Bryne, 1980 ; Berger et al, 1982 ; Beattie and Himes, 1984). Although Ventry and Johnson (1978) and Berger et al (1982) reported some-what better test-retest reliability for MCL using the descending approach than with the ascend-ing approach, this observation has not been con-firmed over a wide range of loudness categories . Moreover, although three or four trials are rec-ommended (IHAFF, 1994), it is not clear how reli-ability of the seven loudness categories is affected by the number of trials for ascending, descend-ing, and random approach modes. Although some evidence indicates that one or two trials are sufficient to establish stable LDLs (Beattie et al, 1979 ; Edgerton et al, 1980), data are not available over the entire loudness function . Therefore, another purpose of this investiga-tion was to assess how increasing the number of trials from one to five (500 Hz, 3000 Hz, and speech) affects the reliability of loudness contour judgments for the ascending, descending, and random approach modes . Short-term reliability, in which the test and retest sessions were sep-arated by 1 to 10 days, was assessed .

METHOD

Subjects

Thirty-one normal-hearing subjects were tested . Prior to both the test and retest, the subjects passed a 15 dB HL (ANSI, 1989) pure-tone screening test at the octave frequencies from 500 Hz to 4000 Hz and had normal tym-panograms (ASHA, 1990) . The test ear was selected randomly.

IHAFF Loudness Contour TestBeattie et al

Instrumentation and Stimuli

An audiometer (Grason-Stadler, Inc., Model 16) was used to present the warble tones (500 Hz and 3000 Hz) and speech (CID W-22 words) to insert earphones (E-A-RTONE, Model 3A). The manual method of stimulus presentation was used for comparing the ascending, descend-ing, and random approach modes because soft-ware for the computer-assisted method is available from IHAFF only for the ascending approach mode . The warble tones had a modu-lation rate of 5 Hz and a frequency deviation of about ±15 percent at 500 Hz and ±5 percent at 3000 Hz (Quest Audiometer Analyzer, Model AA-188). The tones were pulsed on and off 2.5 times per second . The base-to-peak rise-fall times were 50 msec, the plateau was 150 msec, and the interstimulus interval was 150 msec as monitored on a storage oscilloscope (Nicolet, Model CA-1000) . A cassette recording of the Auditec CID W-22 test was used to present the speech stimulus ("Say the word .").

The 500-Hz and 3000-Hz stimuli were cal-ibrated to the SPLs recommended in Appendix G of the ANSI (1989) specifications for audiome-ters . That is, the SPLs corresponding to 0 dB HL were 8.5 dB at 500 Hz and 5.5 dB at 3000 Hz . As suggested by E-A-RTONE (1995), 0 dB HL for speech was calibrated to 16 dB SPL, which is 12.5 dB above the normal 3.5 dB SPL thresh-old at 1000 Hz . The audiometer was calibrated by attaching a foam eartip (E-A-Rlink) with putty to a HA-1 coupler (2 cc) directed to a sound level meter (Quest, Model 1600) . This proce-dure is described in the E-A-RTONE brochure (1995) .

Instructions

As suggested by IHAFF (1994, pp. 14, 29), the following written instructions were pre-sented to the subjects . The subjects were asked to read along with the examiner as the instruc-tions were read aloud.

The purpose of this test is to find your judg-ments of the loudness of different sounds .

You will hear sounds that increase and decrease in volume . You must make a judg-ment about how loud the sounds are. Pre-tend you are listening to the radio at that volume. How loud would it be?


CATEGORIES OF LOUDNESS

7. Uncomfortably loud 6. Loud, but okay 5 . Comfortable, but slightly loud 4. Comfortable 3. Comfortable, but slightly soft 2. Soft 1. Very soft

After each sound, tell me which of these cat-egories best describes the loudness.

Keep in mind that an uncomfortably loud sound is louder than you would ever choose on your radio no matter what mood you are in .

More than one level of loudness may be judged at a given level. For example, sev-eral tones of differing loudness may be judged as comfortable (category 4) .

Every loudness level does not necessarily need to be used in order; it is acceptable to skip a level (e.g., from category 2 [soft] to category 4 [comfortable]) .

Procedure

After reading and explaining the instruc-tions, the subjects were given a sheet of paper that contained the printed loudness categories and the corresponding numbers from 1 to 7. At a given intensity level, the warble tones were pulsed on and off four times, whereas the CID W-22 speech stimulus "Say the word -' was presented once . After the stimuli were presented, the subjects responded by reporting a number from 1 to 7, which corresponded to the perceived loudness category. The order in which the 500 Hz, 3000 Hz, and speech stimuli were presented was randomized .

Prior to testing, a training session was con-ducted (1) to familiarize the subjects with the stimuli and procedures for the test and retest ses-sions (IHAFF, 1994) and (2) to find the uncom-fortably loud level (loudness category 7) . The uncomfortably loud level was used as the start-ing point for the descending approach . Also, it was necessary to know the uncomfortably loud level in order to set the upper level for the ran-dom approach . The stimuli were presented ini-tially at 0 dB HL and then increased in 10-dB steps to 70 dB HL. Thereafter, the stimuli were increased in 5-dB steps until the uncomfortably

loud category was identified (category 7) . The intensity then was decreased by 10 to 20 dB before ascending in 5-dB steps until the uncom-fortably loud level was reported (category 7) . This was labeled trial 1 . The procedure was repeated two more times (trials 2 and 3) . For the descending and random approaches, the uncom-fortably loud level (number 7) was taken as the median of the three training trials . The ascend-ing, descending, and random procedures are described below.

Ascending Approach Mode. The warble tone or speech stimuli were initially presented at 0 dB HL and then increased in 5-dB steps until the uncomfortably loud level (category 7) was sig-naled. The four pulsed tones or one presentation of "Say the word -' were presented at each level. This procedure was repeated five times (trials) .

Descending Approach Mode. The stimuli were initially presented at the uncomfortably loud level (category 7) identified during the prac-tice period (the median level of three ascending trials). If the starting level did not elicit a cat-egory 7 judgment, then the intensity was increased by 5 dB and the descent began again. This procedure was repeated again if the start-ing level did not result in an uncomfortably loud judgment . The intensity was then decreased in 5-dB steps to 0 dB HL. This procedure was repeated five times (trials) .

Random Approach Mode. Stimuli were pre-sented at intensity levels selected randomly in 5-dB steps from 0 dB HL to the subject's uncom-fortably loud level (category 7) . The subjects responded to the stimuli by selecting the per-ceived loudness category from very soft (category 1) to uncomfortably loud (category 7) . If the highest level ascertained from the practice ses-sion did not elicit an uncomfortably loud judg-ment (category 7), then another stimulus 5 dB higher than the initial upper level was ran-domly inserted among the remaining levels . This procedure was repeated five times (trials) .

The order of presentation of approach mode (ascending, descending, random) was random-ized . After one stimulus was completed, train-ing and testing commenced for the second stimulus . Finally, training and testing for the third stimulus was completed.

Retesting was completed 2 to 10 days after the initial test session. The retest procedures were identical to those described above.

246

IHAFF Loudness Contour Test/Beattie et al

RESULTS

Effects of Approach Mode on the Normal Loudness Function

Tables 1(500 Hz), 2 (3000 Hz), and 3 (speech) present means, standard deviations, and ranges in dB SPL corresponding to the seven loudness categories comprising the IHAFF (1994) loudness function . Values are shown for the combined test and retest data for trials 1 through 5 (Tl-T5) because a series of t-tests revealed no statisti-cally significant differences between the test and retest means at any loudness category or approach mode (p > .01) .

The mean data also are presented in Figures 1 (500 Hz), 2 (3000 Hz), and 3 (speech) . A one-way analysis of variance (ANOVA) for repeated measures was conducted at each loudness cat-egory for each stimulus to determine if there was a statistically significant difference among the ascending (A), descending (D), and random (R) approach modes. The results of these ANOVAs are shown in Table 4. For example, D > A = R should be interpreted to mean that the descend-ing approach yielded higher SPLs (p < .01) than the ascending and random approaches . More-over, the ascending and random approaches were not significantly different (p > .01) . Simi-larly, R > D > A means that the random approach yielded significantly larger SPLs than the descending approach, which exhibited larger SPLs than the ascending approach.

Comparison of Figure 1 (500 Hz) and Fig-ure 2 (3000 Hz) reveals little difference between the two warble stimuli with respect to the rela-tionship among approach modes. Therefore, only the 500-Hz data will be described in detail . Examination of Figure 1 and Table 4 reveals sev-eral noteworthy results. First, the descending approach yielded higher SPLs than the ascend-ing approach for all seven loudness categories (p < .01) . Second, although the random approach yielded SPLs similar to those with the ascend-ing approach at the very soft, soft, and com-fortable-soft loudness categories (p > .01), the random approach yielded higher SPLs than the ascending function at the higher loudness cat-egories (p < .01) . Third, the random approach yielded lower SPLs than the descending approach at the very soft through comfortable loudness categories (p < .01), there was no dif-ference at the comfortable-slightly loud cate-gory, and the random approach yielded higher SPLs than the descending approach at the loud-okay and uncomfortably loud categories (p < .01) .

The speech data in Figure 3 show that the descending approach yielded higher SPLs than the ascending approach for all loudness cat-egories (p < .01) . Although the descending approach yielded higher SPLs than the random approach at the very soft through comfortable loudness categories (p < .01), no significant dif-ferences were found at the higher loudness cat-egories (p > .01) . Also, the random approach yielded higher SPLs than the ascending

Table 1 Means, SDs, and Ranges in dB SPL for Loudness Categories Very Soft, Soft, Comfortable-Slightly Soft, Comfortable, Comfortable-Slightly Loud, Loud-Okay, and Uncomfortably Loud

Very Soft Soft

Comfortable- Slightly Soft Comfortable

Comfortable- Slightly Loud Loud

Uncomfortably Loud

500 Hz-Ascending Mean 28.3 51 .2 64 .0 75.3 86.7 96 .0 103.9 SD 6.3 9 .7 10 .1 9 .8 10.2 11 .4 12 .4 Range 30.0 40.0 37 .5 41 .3 47.5 50 .0 55 .0

500 Hz-Descending Mean 36 .8 66.0 76 .8 86 .3 93.1 100.4 108 .1 SD 7 .7 11 .9 10 .1 9 .1 10 .3 9 .5 10.6 Range 40 .0 53.8 42 .5 40.0 43 .7 42.5 47.5

500 Hz Random Mean 26.8 51 .2 66 .8 82 .6 95 .8 104.6 112.0 SD 5.7 10.6 10 .3 9 .9 10 .3 10 .8 11 .1 Range 22 .5 45.0 46 .3 42 .5 45 .0 42 .5 42 .5

Values are shown for 500 Hz at each approach mode (ascending, descending, random) and represent the combined test and retest data for trials 1 to 5.


Table 2 Means, SDs, and Ranges in dB SPL for Loudness Categories Very Soft, Soft, Comfortable-

Slightly Soft, Comfortable, Comfortable-Slightly Loud, Loud-Okay, and Uncomfortably Loud

Very Soft Soft



Uncomfortably Loud

3000 Hz-Ascending Mean 23.7 47.5 60 .1 71 .4 81 .6 89 .2 97 .0

SD 6 .0 9 .7 9 .8 10.2 10 .7 11 .3 11 .7

Range 27.5 27 .5 43 .8 42.5 50 .0 51 .2 51 .2

3000 Hz-Descending 0 33 1 63 72 2 80 .9 87 .9 95 .1 102.8 . Mean

SD 7.4 .

12 .7 .

12 .2 11 .1 10 .8 10 .6 11 .4

Range 28 .8 41 .2 37 .5 38 .8 42 .5 42 .5 42 .5

3000 Hz-Random Mean 23.0 48.8 65.3 79 .3 90.9 99.8 107.4

SD 6.5 10 .9 11 .8 11 .0 9 .9 11 .1 12 .2

Range 30.0 45.0 41 .2 41 .2 37 .5 44 .2 46 .2

Values are shown for 3000 Hz at each approach mode (ascending, descending, random) and represent the combined test and

retest data for trials 1 to 5.

approach at the comfortable-slightly soft through uncomfortably loud categories (p < .01) .

Effects of Approach Mode on Short-Term Test-Retest Reliability

To assess test-retest reliability, standard errors of measurement (Guilford and Fruchter, 1973 ; Mehrens and Lehmann, 1973) were cal-culated for one trial (T1) to five trials (T1-T5) at each loudness category, stimulus, and

approach mode . To evaluate the effects of the

number of trials on reliability, the standard errors were collapsed across categories, stimuli, and approach mode for T1 to T1-T5. These data are shown in Figure 4 and indicate that the

standard error decreased from 5.9 dB for T1 to

4.9 dB for T1-T5. A one-way ANOVA for repeated measures revealed a statistically significant dif-

ference among means (F [4,2481 = 38.5, p < .01) .

Tukey's post hoc test (Linton and Gallo, 1975)

indicated that T1 was significantly larger than

Table 3 Means, SDs, and Ranges in dB SPL for Loudness Categories Very Soft, Soft, Comfortable-

Slightly Soft, Comfortable, Comfortable-Slightly Loud, Loud-Okay, and Uncomfortably Loud

Very Soft Soft



Uncomfortably Loud

Speech-Ascending Mean 28 .2 42.9 53 .1 67 .6 81 .8 93 .8 102 .1

SD 6.9 9 .1 9 .7 11 .4 12 .2 13 .2 14 .1

Range 25.0 31 .2 35.0 50 .0 53.7 63 .7 72 .5

Speech-Descending Mean 30.6 50 .7 64 .5 78.2 90.8 97.9 106.6

SD 6.7 10.8 10 .8 10 .2 9 .5 9 .9 11 .1

Range 26.3 46.3 42.5 40 .0 40.0 38 .8 42 .5

Speech-Random Mean 28 .2 45 .2 59 .5 75.6 89 .9 99 .9 107 .7

SD 6 .7 9 .8 11 .0 9 .9 10.2 10.5 11 .6

Range 27.5 41 .2 47 .5 40 .0 46 .2 43.8 45.0

Values are shown for speech at each approach mode (ascending, descending, random) and represent the combined test and retest

data for trials 1 to 5 .

248

IHAFF Loudness Contour Test/Beattie et al

I Ascending + Descending -*-Random

Comfortable

r Comfortable, Slightly Soft

INTENSITY IN dB SPL (500 Hz)

Figure 1 Mean SPLs for 500 Hz corresponding to the seven loudness categories comprising the IHAFF (1994) loudness function . Values are shown for the ascending, descending, and random approach modes and represent the combined test and retest data for trials 1 through 5 (T1-T5).

the other four conditions, and T1-T2 was sig-nificantly larger than T1-T4 and T1-T5 (p < .01).

In view of the small differences among trials (<-1 dB), the standard error data were collapsed across all trials (Tl to T1-T5) to summarize the effects of approach mode and stimulus on reliability. These computations yielded similar

I Ascending t Descending 4 -Random

N 7 ~ Uncomfortably : Loud ; " }', ~ 7

¢ 6 ~ Loud,' O . K; . . ~6

O U W 5 ~ Comfortable, Slightly Loud _ ; " * .X ~ 5

Q U 4 Comfortable . "

*:

. ~ 4

N N W

Comfortable, Slightly Soft 3

Z 3

D 2 Soft s .

~2 n J l r very aon wr -11

0 0 0 10 20 30 40 50 60 70 80 90 100110

INTENSITY IN dB SPL (Speech)

Figure 3 Mean SPLs for speech corresponding to the seven loudness categories comprising the IHAFF (1994) loudness function . Values are shown for the ascending, descending, and random approach modes and represent the combined test and retest data for trials 1 through 5 (Tl-T5).

standard errors across all conditions (4.4-5 .7 dB), and there was no consistent trend among stim-uli or approaches . Thus, the data were collapsed across all conditions and yielded an overall stan-dard error of measurement of 5.3 dB .

To assess the effects of loudness categories on test-retest reliability, standard errors were

Ascending -*-Descending -*-R andom

Table 4 300 Hz, and

ANOVA/Tukey Speech at Eac

Results at h of Seven

500 Hz, Loudness

Categories

Stimulus 0) 7 Uncomfortably Loud w Loudness 2 6 Loud, CA, 6 Category 500 Hz 3000 Hz Speech O O W 5 Comfortable, Slightly Loud 5 Very soft D>R=A D>R=A D>R=A H a U 4 Comfortable k 4 Soft D>R=A D>R=A D>R=A

N i Comfortable- D>R=A D>R>A D>R>A N W 3 Comfortable, Slightly Soft 3 slightly soft Z Comfortable D>R>A D=R>A D>R>A a 2 Soft

_'4_ A Comfortable- D = R > A D = R > A D = R > A

O J 1 - Very Soft 1 slightly loud

Loud-okay R>D>A R>D>A D=R>A 0 Uncomfortably R > D > A R > D > A D = R > A

0 10 20 30 40 50 60 70 80 90 100110 loud INTENSITY IN dB SPL (3000 Hz)

Figure 2 Mean SPLs for 3000 Hz corresponding to the seven loudness categories comprising the IHAFF (1994) loudness function . Values are shown for the ascend-ing, descending, and random approach modes and rep-resent the combined test and retest data for trials 1 through 5 (Tl-T5).

A one-way ANOVA was conducted at 500 Hz, 3000 Hz, and speech at each of the seven loudness categories (very soft to uncomfortably loud) to determine if there were statistically significant differences (p < .01) among the ascending (A), descending (D), and random (R) approach modes. Because these analyses were statistically significant, Tukey's post hoc test was used to identify statistically significant comparisons between means (p < .01) and the results of these analyses are shown.


ALL STIMULI, APPROACH MODES, AND LOUDNESS CATEGORIES COMBINED

Figure 4 Standard errors of measurement in dB are shown for trial 1 (Tl), trials 1 and 2 (Tl-T2), trials 1-3 (Tl-T3), trials 1-4 (Tl-T4), and trials 1-5 (Tl-T5) . The data represent combined values from all stimuli (500 Hz, 3000 Hz, speech), approach modes (ascending, descending, random), and loudness categories (very soft-uncomfortably loud) .

collapsed across trials, approaches, and stimuli. These data showed that the standard error was smaller (p < .01) for the very soft loudness cat-egory (4.0 dB) than for the other six categories, which ranged from 5.1 dB to 6.1 dB .

Several investigators have assessed relia-bility by computing the magnitude of test-retest differences (Ventry and Johnson, 1978; Berger et al, 1982 ; Beattie and Boyd, 1986; Sammeth et al, 1989). Therefore, for comparative pur-poses, we calculated the standard deviation of test-retest differences for one trial to five trials at each loudness category, stimulus, and approach mode . To assess the effects of the num-ber of trials on reliability, the standard deviations were collapsed across categories, stimuli, and approach mode for T1 to T1-T5 . These data revealed that the standard deviation decreased from 5 .7 dB for T1 to approximately 5 dB for T1-T2 through T1-T5 (p < .01) .

The standard deviations were collapsed across all trials (T1 to T1-T5) to summarize the effects of approach mode and stimulus on reli-ability. These analyses revealed no consistent trend among stimuli or approach modes . Thus, the data were collapsed across all conditions and yielded an overall standard deviation of test-retest differences of 5.1 dB .

To assess the effects of loudness categories on test-retest reliability, standard deviations were collapsed across trials, approaches, and stimuli. These analyses revealed that the stan-dard deviation for the very soft loudness category (3.8 dB) was smaller (p < .01) than the other six loudness categories (5.1 dB to 5.6 dB).

Stimulus Effects on Loudness Judgments

Mean SPLs for 500 Hz, 3000 Hz, and speech are shown at each of the seven loudness cat-egories in Figures 5 (ascending), 6 (descend-ing), and 7 (random) . These mean data were taken from Tables 1, 2, and 3 and are rounded to the nearest decibel. A series of one-way ANOVAs for repeated measures was conducted at each loudness category for the ascending, descending, and random approach modes. With one exception, the comfortable-slightly loud cat-egory for the ascending approach mode, these ANOVAs were statistically significant (p < .01) . Therefore, Tukey's post hoc test was used to identify statistically significant mean differences (p < .01) . The results of these analyses are shown in Table 5, where 5 = 500 Hz, 3 = 3000 Hz, and S = speech . For example, 5 > 3; 5 = S; 3 < S should be interpreted to mean that the SPL for 500 Hz was greater than the SPL for 3000 Hz (p < .01), that no statistically significant SPL differences were found between 500 Hz and speech, and that the SPL for 3000 Hz was less than the SPL for speech (p < .01) . Although examination of Table 5 reveals no consistent mean differences across all loudness categories for the ascending, descending, or random approach modes, four observations were noted for all approaches . First, the 500-Hz stimulus yielded 7 to 12 dB higher SPLs than the speech stimulus at the soft, comfortable-slightly soft, and comfortable cat-egories . Second, no statistically significantly differences were found between 3000 Hz and

ASCENDING APPROACH MODE

V-SOFT SOFT COM-SS COM COM-SL LOUD ULOUD

Figure 5 Mean SPLs for 500 Hz (5), 3000 Hz (3), and speech (S) are shown for each of the seven loudness cat-egories : very soft (V-SOFT), soft, comfortable-slightly soft (COM-SS), comfortable (COM), comfortable-slightly loud (COM-SL), loud, and uncomfortably loud (ULOUD) . The ascending approach mode was used .

250


DE SCENDING APPROACH MODE RANDOM APPROACH MODE

120 120 112

110 toe 107 110 - - 105- - 197168 03- 109 00100ME N 100 . . . . . 98 93 . . . -,95 - 100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .96 . . . . .

__91 09190 86

77

. . . . . . . . .-6663

51

72

64

8178

5 3 S 5 3 S 5 3 S 5 3 S 5 3 S 5 3 S 5 3 S .


Figure 6 Mean SPLs for 500 Hz (5), 3000 Hz (3), and speech (S) are shown for each of the seven loudness cat-egories : very soft (V-SOFT), soft, comfortable-slightly soft (COM-SS), comfortable (COM), comfortable-slightly loud (COM-SL), loud, and uncomfortably loud (ULOUD) . The descending approach mode was used .

speech at the comfortable-slightly loud, loud-okay, and uncomfortably loud categories (p > .01) . Third, 500 Hz yielded about 6-dB higher SPLs than 3000 Hz at the loud-okay and uncom-fortably loud categories (p < .01) . Fourth, 500 Hz and 3000 Hz yielded SPLs that were not statis-tically significant (p > .01) at the soft and com-fortable-slightly soft categories .

Relationship between Warble Tones and Speech

The relationship between the speech and warble tone stimuli for the seven loudness cat-egories was investigated by calculating Pearson product-moment correlation coefficients (r) and standard errors of estimate . Table 6 presents these data for the ascending, descending, and

90 83

80 . . . . .7976. . .

70 F . . . . . . . . . . . . . . . . . . .67.65 . . . . .

49 j

40 . . . ,-

30

20

10

O

27 . .20. . 23

59

53S 53S 53S 53S 63S 53S 53S .


Figure 7 Mean SPLs for 500 Hz (5), 3000 Hz (3), and speech (S) are shown for each of the seven loudness cat-egories : very soft (V SOFT), soft, comfortable-slightly soft (COM-SS), comfortable (COM), comfortable-slightly loud (COM-SL), loud, and uncomfortably loud (ULOUD) . The random approach mode was used .

random approach modes. When the data were averaged across all loudness categories and approach modes, the standard error was 7.4 dB for the 500 Hz/speech relationship and 8.7 dB for the 3000 Hz/speech relationship .

DISCUSSION

Effects of Approach Mode on the Normal Loudness Function

Higher SPLs were found with the descend-ing approach than with the ascending approach for all conditions . These differences, when aver-aged across all three test stimuli, varied from about 5 dB for the loud-okay and uncomfort-ably loud categories to approximately 12 dB for

Table 5 ANOVA/Tukey Results at 500 Hz, 3000 Hz, and Speech at Each of Seven Loudness Categories for the Ascending, Descending, and Random Approach Modes

Ascending Descending Random

Very soft 5>3;5=S;3<S 5>3;5>S;3=S 5>3;5=S;3<S Soft 5=3;5>S;3>S 5=3;5>S;3>S 5=3;5>S;3=S Comfortable-slightly soft 5 = 3 ; 5 > S ; 3 > S 5=3;5>S;3>S 5=3;5>S;3>S Comfortable 5 = 3 ; 5 > S ; 3 = S 5>3;5>S;3=S 5=3;5>S;3>S Comfortable-slightly loud 5 = 3 ; 5 = S ; 3 = S 5>3;5=S;3=S 5>3;5>S;3=S Loud-okay 5 > 3 ; 5 = S ; 3 = S 5>3;5=S;3=S 5>3;5>S;3=S Uncomfortably loud 5 > 3 ; 5 = S ; 3 = S 5>3;5=S;3=S 5>3;5=S;3=S

A one-way ANOVA was conducted at each of the seven loudness categories (very soft to uncomfortably loud) to determine if there were statistically significant differences (p < .01) among the 500 Hz (5), 3000 Hz (3), and speech (S) stimuli . With one exception, these analyses were statistically significant . Therefore, Tukeys post hoc test was used to identify statistically significant comparisons between means (p < .01) and the results of these analyses are shown.


Table 6 Relationship between 500 Hz and Speech and 3000 Hz and Speech for Seven Loudness Categories for the Ascending, Descending, and Random Approach Modes.

Loudness Category

Stimulus

X-Y

Ascending

r SE.,

Descending

r SEesr

Random

r SEesr Very soft 500 Hz-Speech 0.83 3.90 0 .72 4 .72 0 .75 4.55

3000 Hz-Speech 0.47 6 .10 0.27 6.59 0 .57 5.63 Soft 500 Hz-Speech 0.70 6 .65 0.75 7.23 0.71 6.98

3000 Hz-Speech 0 .60 7 .42 0.44 9.99 0.67 7.44 Comfortable-slightly soft 500 Hz-Speech 0.61 7 .75 0.80 6.62 0.77 7.16

3000 Hz-Speech 0.52 8 .41 0.55 9.23 0.66 8.38 Comfortable 500 Hz-Speech 0 .64 8.84 0.63 8.10 0.84 5.52

3000 Hz-Speech 0.12 11 .46 0.49 9.08 0.66 7.58 Comfortable-slightly loud 500 Hz-Speech 0.57 10 .17 0.59 7.75 0.71 7.32

3000 Hz-Speech -0 .06 12 .37 0.59 7.77 0.60 8.32 Loud-okay 500 Hz-Speech 0.65 10.19 0.64 7.78 0.72 7.41

3000 Hz-Speech 0.55 11 .22 0.56 8.38 0.62 8.42 Uncomfortably loud 500 Hz-Speech 0.67 10.67 0.67 8.33 0.71 8.26

3000 Hz-Speech 0.61 11 .36 0.62 8.84 0.65 8.99 Combined 500 Hz-Speech 7.42 7.22 6.74

3000 Hz-Speech 8.71 8.55 7.82 Correlation coefficients (r) and standard errors of estimate (SE_) are shown for the relationship between 500 Hz (X) and speech (Y)

and between 3000 Hz (X) and speech (Y) for seven loudness categories (very soft, soft, comfortable-slightly soft, comfortable, comfortable-slightly loud, loud-okay, and uncomfortably loud). The standard errors also were combined across all loudness categories . Results are shown for the ascending, descending, and random approach modes.

the soft, comfortable-soft, and comfortable cat-egories. These data are consistent with research comparing ascending and descending MCLs for pure tones and speech (Conn et al, 1972 ; Woods et al, 1973 ; Kamm et al, 1978 ; Ventry and John-son, 1978 ; Berger et al, 1980, 1982). Neverthe-less, these studies do not agree on the magnitude of ascending-descending differences, which range from about 15 dB (Woods et al, 1973; Kamm et al, 1978 ; Berger et al, 1982) to 39 dB (Berger et al, 1980). It is important to note that our pro-cedures differ from some investigations (Ventry and Johnson, 1978 ; Berger et al, 1982) in that we used a training period exposing our subjects to a wide range of SPLs . This training probably minimized the ascending-descending differences observed in the present study.

Ventry and Johnson (1978) suggest that the ascending-descending differences may occur because the subject is not allowed to explore levels beyond the level at which the signal is first judged comfortable . Similarly, Berger et al (1980) speculated that ascending-descending differ-ences may occur because the loudness judgment is made with respect to the beginning level, not realizing that a lower or higher level would be more comfortable. If these explanations are cor-rect, neither the ascending nor descending approaches may be the method of choice for selecting loudness levels that are typical of everyday listening. This is because SPLs often

vary randomly during daily activities, and because the hearing aid user can alter the gain of a hearing aid to an optimal level. The random approach, in which sounds vary over a wide range of levels, may be more representative of the way acoustic stimuli are experienced in everyday situations . Moreover, a method of adjustment or up-down adaptive approach, in which the subject can search thoroughly around various loudness categories (e.g., soft, comfort-able, and uncomfortable), may enable the selec-tion of SPLs more typical of everyday listening.

It is also of interest to determine which approach mode results in SPLs that optimize speech recognition. Evidence suggests that whereas normal-hearing subjects attain maxi-mum speech recognition scores for monosyllabic words at soft or comfortable-soft levels, hearing-impaired subjects typically require higher loud-ness levels (comfortable-loud or loud-okay) to attain maximum speech recognition scores (Pos-ner and Ventry, 1977 ; Ventry and Johnson, 1978 ; Dirks et al, 1981 ; Beattie and Warren, 1982 ; Beattie and Zipp, 1990). Compared to the ascend-ing approach, the descending or random ap-proaches yield Higher SPLs, which may result in higher word recognition scores . Moreover, Ventry and Johnson (1978) state that the descending approach may result in a wider dynamic range and, thus, better speech intelli-gibility than the ascending approach .

252


One of the assumptions of the IHAFF (1994) procedure is that speech should be amplified so that SPLs that are comfortable for normal lis-teners are also comfortable for hearing-impaired subjects . As noted above, however, whereas nor-mal listeners obtain maximum word recogni-tion scores at a comfortable ldudness level, this may not be the case with hearing-impaired sub-jects. That is, higher word recognition scores may be obtained at higher SPLs . Additionally, the MCL may be substantially higher in the presence of background noise if the listener can adjust the signal independent of the noise (e.g ., when listening to a telephone, radio, or TV) (Beattie, 1982). Perhaps the goal should be to amplify speech to a point that maximizes speech recognition . If so, it is likely that higher gain/out-put will be required than suggested by the IHAFF (1994) procedure, at least over some portion of the loudness function . We recognize that there are many interactive variables affect-ing these questions. Systematic investigation is required using hearing-impaired subjects hav-ing a wide range of hearing losses and using a representative sample of amplifying systems.

The present study indicates that approach mode has a substantial effect on the SPLs cor-responding to loudness levels ranging from very soft to uncomfortably loud . As noted above, these data are consistent with previous research inves-tigating the effects of approach mode on the MCL. The extent to which these observations apply to hearing-impaired subjects is under investigation. Moreover, whether or not these approach mode differences result in recom-mended gain differences must be investigated . That is, provided the same approach mode is employed to obtain normative and hearing-impaired data, the loudness function for hear-ing-impaired subjects may be shifted to higher intensities by an equal amount for the various approach modes. Although this speculation must be subjected to systematic investigation, it is rea-sonable to employ the same approach mode with hearing-impaired subjects who were used to establish the normative data .

Effects of Approach Mode on Short-Term Test-Retest Reliability

The standard error of measurement data, when collapsed across all conditions (see Fig. 4), revealed that increasing the number of trials from one (T1) to five (T1-T5) resulted in a grad-ual reduction in the standard error from 5.9 dB to 4.9 dB . Although some of these differences

attained statistical significance, they are very small and probably not clinically significant. The largest change in adjacent trials was 0.6 dB, observed from T1 (5 .9 dB) to T1-T2 (5.3 dB). Increasing the number of trials from two (TI-T2) to five (T1-T5) decreased the standard error by only 0.4 dB. Similarly, the standard deviation of test-retest differences decreased less than 1 dB from T1 (5 .7 dB) to T1-T5 (4.9 dB), and the greatest decrease (0 .6 dB) was from T1 to T1-T2. These data suggest that one or two trials yield loudness judgments that are nearly as reliable as when three, four, or five trials are obtained . This conclusion is consistent with research inves-tigating the effects of the number of trials on the LDL (Beattie et al, 1979 ; Edgerton et al, 1980) . Compared to one trial, obtaining two trials is advantageous because the clinician can exam-ine the data from individual subjects to assess whether unusually large differences are pre-sent between the two trials . That is, if differences exceed -10 to 12 dB (2 SDs), the clinician may reinstruct the subject and then begin the test again.

Reliability was very similar for the ascend-ing, descending, and random approach modes. When the data were collapsed across all condi-tions, standard errors were 5.3 dB for the ascend-ing approach, 4.9 dB for the descending approach, and 5.6 dB for the random approach . Similarly, the standard deviation of test-retest differences was about 5 dB for all approach modes . These data suggest that the ascending, descending, and random approaches are equally reliable . These results are consistent with pre-vious investigations that have compared the reliability of ascending and descending approach modes for ascertaining MCLs (Ventry and John-son, 1978; Berger et al, 1982).

The standard errors were similar for the three stimuli. When the data were collapsed across all conditions, the standard error was 5.3 dB for 500 Hz, 5 .6 dB for 3000 Hz, and 4.9 dB for speech . Likewise, the standard deviation of test-retest differences was about 5 dB for all stimuli. The observed standard errors of mea-surement are larger than the 2.4- to 4.0-dB standard errors observed by Beattie and Boyd (1986) for pure-tone (250-6000 Hz) and speech LDLs . However, these authors investigated immediate (same session) test-retest reliability, whereas our retest data were obtained on sep-arate days . Our standard deviation data are similar to the results obtained in several stud-ies investigating immediate or short-term test-retest reliability for MCL or LDL (Ventry


et al, 1971 ; Berger et al, 1980 ; Beattie and Shef-fler, 1981 ; Beattie and Himes, 1984 ; Walker et al, 1984 ; Sammeth et al, 1989) . That is, these investigators reported that approximately 95 percent of subjects exhibit test-retest differ-ences within 4 to 10 dB for the LDL and within 8 to 12 dB for the MCL.

Somewhat smaller standard errors of mea-surement were observed for the very soft cate-gory (4.0 dB) than for the other categories (5.1-6.1 dB). For the very soft through uncom-fortably loud categories, the present study sug-gests that the true SPL for a given subject will fall within ± 10 to 12 dB of the obtained SPL approximately 95 percent of the time . This 20-to 24-dB range is fairly large and it is debatable whether reliability is sufficient to justify obtain-ing individual loudness measurements on sub-jects with normal or near-normal thresholds . Christen and Byrne (1980) state that MCL measurements are of questionable value for selecting hearing aid gain because these mea-surements showed considerable intersubject and intrasubject variability. They concluded that it is unjustified to advocate MCL proce-dures in preference to threshold procedures unless data are presented clearly demonstrat-ing that MCL procedures are superior to thresh-old procedures . Christen and Byrne (1980) comment that threshold procedures are applic-able to a wider range of cases (e.g ., very young children) than loudness procedures . Moreover, loudness procedures are more time consuming than threshold procedures (Humes and Halling, 1994). An alternative threshold-based proce-dure has been described by Killion and Fikret-Pasa (1993), who recommend gains for soft, average, and loud inputs (Gitles and Niquette, 1995).

Stimulus Effects on Loudness Judgments

We observed mean uncomfortably loud lev-els ranging from 97 to 112 dB SPL across all stimuli and approach modes. Moreover, for a given approach mode, the maximum differ-ence between 500 Hz and 3000 Hz was 7 dB and the maximum difference between the warble tones and speech was 5 dB . These data are consistent with previous research investigating LDLs using tones and/or speech in normal-hearing subjects (Hood and Poole, 1966 ; McCan-dless and Miller, 1972 ; Morgan et al, 1974; Dudich et al, 1975 ; Denenberg and Altshuler, 1976 ; Beattie et al, 1979 ; Ritter et al, 1979 ; Hawkins, 1980).

Our mean comfortable levels ranged from 68 to 75 dB SPL for the ascending approach, from 78 to 86 dB SPL for the descending approach, and from 76 to 83 dB SPL for the random approach . The 500-Hz stimulus yielded com-fortable loudness levels that were 7 to 8 dB higher than the speech stimulus . These SPLs are consistent with previous research investigating MCLs using tones and/or speech in normal-hearing subjects (Ventry et al, 1971 ; Woods et al, 1973 ; Dirks and Kamm, 1976 ; Beattie, 1982; Dirks and Morgan, 1983). Dirks and Morgan (1983) comment that a wideband stimulus such as speech is expected to elicit lower SPLs than tonal stimuli because loudness increases as stimulus bandwidth increases.

Relationship between Warble Tones and Speech

The relationship between warble tones and speech was investigated with the standard error of estimate, which reflects how accurately loud-ness categories for speech can be predicted from the corresponding data for warble tones. An overall standard error of estimate of approxi-mately 8 dB was observed when the data were collapsed across stimuli, approach modes, and loudness categories . That is, the 95 percent con-fidence interval (two standard errors) for pre-dicting speech levels from warble tone levels encompasses a range of -} 16 dB . It is evident that warble tones are not accurate predictors of the corresponding loudness categories for speech . If the clinician wants to ascertain SPLs corre-sponding to the various loudness categories for speech, then direct measurement with a speech stimulus appears necessary. Our data are con-sistent with the results of Beattie and Boyd (1986) . They investigated the relationship between pure tones (500-6000 Hz) and speech (CID Test W-22) LDLs on a group of 50 hearing-impaired subjects and found an overall stan-dard error of about 9.5 dB .

Humes and Halling (1994, p. 20) recently concluded that ". . .there have been no studies conducted that demonstrate that hearing aid wearers fit with the more time-consuming suprathreshold-based methods are more satis-fied or successful hearing aid wearers than those fit with the shorter threshold-based methods." More specifically, studies are required compar-ing the suprathreshold-based IHAFF (1994) pro-cedure to threshold-based procedures such as those proposed by Killion and Fikret-Pasa (1993) and Pluvinage (1994) .

254

1HAFF Loudness Contour Test/Beattie et al

REFERENCES

American National Standards Institute . (1989) . American National Standard Specification for Audiometers. (ANSI S3.6-1989) . New York : ANSI.

American Speech-Language-Hearing Association . (1990) . Guidelines for screening for hearing impairments and middle ear disorders. ASHA, 32 (Suppl 2):17-24 .

Beattie RC . (1982) . Comfortable speech listening levels : Effects of competition and instructions . Am J Otol 3: 192-199.

Beattie RC, Boyd RL . (1986) . Relationship between pure-tone and speech loudness discomfort levels among hearing-impaired subjects . J Speech Hear Disord 51 : 120-125.

Dirks D, Kamm C. (1976) . Psychometric functions for loudness discomfort and most comfortable loudness levels . J Speech Hear Res 19:613-627 .

Dirks DD, Kamm C, Dubno JR, Velde TM . (1981) . Speech recognition performance at loudness discomfort level . Scand Audiol 10:239-246 .

Dirks DD, Morgan DE . (1983) . Measures of discomfort and most comfortable loudness- In : Konkle DF, Rintelman WF, eds. Principles of Speech Audiometry . Baltimore: University Park Press, 203-229.

Dudich T, Keiser M, Keith R. (1975) . Some relationships between loudness and the acoustic reflex . Impedance Newsletters 4:12-15 .

E-A-RTONE. (1995) . Instructions for the Use of E-A-RTONE 3A Insert Earphones. Indianapolis, IN : Cabot Safety Corporation.

Beattie RC, Culibrk J. (1980). Effects of a competing mes-sage on the speech comfortable loudness level for two instructional sets . Ear Hear 1:242-248 .

Beattie RC, Edgerton BJ, Gager DW (1979). Effects of speech materials on the loudness discomfort level. J Speech Hear Disord 44:435-458 .

Beattie RC, Himes BE . (1984) . Comfortable loudness levels for speech : Effects of signal-to-noise ratios and instructions . JAudit Res 24:213-229 .

Beattie RC, Sheffler MV (1981) . Test-retest stability and effects of pyschophysical methods on the speech loudness discomfort level. Audiology 20:143-156 .

Beattie RC, Warren VG . (1982) . Relationships among speech threshold, loudness discomfort, comfortable loud-ness, and PB Max in the elderly hearing impaired . Am J Otol 3:353-358 .

Beattie RC, Zipp JA . (1990) . Range of intensities yield-ing PB Max and the threshold for monosyllabic words for hearing-impaired subjects . J Speech Hear Disord 55 : 417-426.

Berger KW, Harrison LD, Monack N, Ferren GE . (1980) . Comfortable loudness judgments for discrete frequency signals. JAudit Res 20:119-127 .

Berger KW, Varvvas DM, Vottero D. (1982) . Test-retest reliability of MCLs, UCLs, and ARTs. Hear Instrum 33(5): 16,19.

Christen R, Byrne D. (1980) . Variability of MCL mea-surements: Significance for hearing aid selection. Aust J Audiol 2:10-18 .

Conn M, Ventry IM, Woods RW. (1972) . Pure-tone aver-age and spondee threshold relationships in simulated hearing loss . J Audit Res 12:234-239 .

Cox RM. (1995) . Using loudness data for hearing aid selection : The IHAFF approach . Hear J 48(2) :10, 39-44 .

Cox RM, Alexander GC. (1995) . The Abbreviated Profile of Hearing Aid Benefit . Ear Hear 16:176-186 .

Denenberg LJ, Altshuler MW (1976). The clinical relation-ship between acoustic reflexes and loudness perception . J Am Audit Soc 2 :79-82 .

Edgerton BJ, Beattie RC, Wides JW. (1980). Loudness discomfort levels of hearing-impaired listeners using speech material . Ear Hear 1:206-210 .

Gitles TC, Niquette PT. (1995) . FIG6 in ten . Hear Rev 2 (Nov/Dec) :28, 30 .

Guilford JP, Fruchter B. (1973). Fundamental Statistics in Psychology and Education. 5th ed. New York : McGraw-Hill .

Hawkins DB . (1980) . The effect of signal type on the Loudness Discomfort Level. Ear Hear 1:38-41 .

Hawkins DB, Walden BE, Montgomery A, Prosek RA . (1987) . Description and validation of an LDL procedure designed to select SSPL90 . Ear Hear 8:162-169 .

Hood J, Poole J. (1966) . Tolerable limit of loudness : Its clinical and physiological significance. JAcoust Soc Am 40:47-53 .

Humes LE, Halling DC . (1994) . Overview, rationale, and comparison of suprathreshold-based gain prescription methods. In : Valente M, ed . Strategies for Selecting and Verifying Hearing Aid Fittings . New York : Thieme Medical Publishing.

Independent Hearing Aid Fitting Forum [IHAFF] . (1994) . A Comprehensive HearingAid Fitting Protocol. Unpublished manuscript .

Kamm C, Dirks D, Mickey R. (1978) . Effect of sen-sorineural hearing loss on loudness discomfort level. J Speech Hear Res 21:668-681 .

Killion MC, Fikret-Pasa S. (1993) . The three types of sen-sorineural hearing loss : Loudness and intelligibility considerations . Hear J 46:11, 31-36.

Konkle DF, Rintelmann WF, eds. (1983) . Principles of Speech Audiometry . Baltimore: University Park Press.

Linton M, Gallo Jr., PS . (1975) . The Practical Statistician : Simplified Handbook of Statistics . Moneterey, CA : Brooks/Cole Publishing Company.

McCandless GA, Miller DL . (1972) . Loudness discomfort and hearing aids . Natl Hear Aid J 26:7, 28, 32 .

Mehrens WA, Lehmann IJ . (1973) . Measurement and Evaluation in Education and Psychology. New York : Holt, Rinehart & Winston.


Morgan DE, Wilson RH, Dirks DD . (1974) . Loudness dis-comfort level: Selected methods and stimuli. J Acoust Soc Am 56:577-581 .

Olsen WO, Matkin ND. (1991) . Speech audiometry . In: Rintelmann WF, ed . Hearing Assessment. Austin, TX : Pro-Ed .

Pluvinage V. (1994) . Rationale and development of the Resound System . In : Sandlin RE, ed . Understanding Digitally Programmable Hearing Aids. Boston : Allyn and Bacon, 15-39.

to speech stimuli in the hearing impaired . Ear Hear 10 : 94-100 .

Skinner MW (1988) . Hearing Aid Evaluation . Englewood Cliffs, NJ : Prentice Hall .

Van Vliet D . (1995) . Determining loudness contour judg-ments . Hear Instrum 46(3):30 .

Ventry IM, Johnson JI . (1978) . Evaluation of a clinical method for measuring comfortable loudness for speech . 43:149-159 .

Posner J, Ventry IM . (1977) . Relationships between comfortable loudness levels for speech and speech dis-crimination in sensorineural hearing loss . J Speech Hear Disord 42:370-375 .

Ritter R, Johnson RM, Northern JL . (1979) . The contro-versial relationship between loudness discomfort levels and acoustic reflex thresholds. JAm Audit Soc 4:123-131 .

Rupp RR, Stockdell KG, eds . (1980) . Speech Protocols in Audiology. New York: Grune & Stratton.

Sammeth CA, Birman M, Hecox KE . (1989) . Variability of most comfortable and uncomfortable loudness levels

Ventry EM, Woods RW Rubin M, Hill W. (1971). Most comfortable loudness for pure tones, noise, and speech . JAcoust Soc Am 49:1805-1813 .

Walker G, Dillon H, Byrne D, Christen R . (1984) . The use of loudness discomfort levels for selecting the maxi-mum output of hearing aids . Aust JAudiol 6:23-32 .

Woods RW Ventry IM, Gatling W. (1973) . Effect of ascend-ing and descending measurement methods on comfortable loudness levels for pure tones . J Acoust Soc Am 54 : 205-206 .

ihaff loudness contour test: reliability and effects of ... · 1haff loudness contour test:...

Documents