Institutional members access full text with Ovid®

Share this article on:

00003446-199908000-0000100003446_1999_20_279_pittman_combinations_4article< 113_0_17_8 >Ear and HearingCopyright © 1999 Wolters Kluwer Health, Inc. All rights reserved.Volume 20(4)August 1999p 279–289Recognition Performance for Four Combinations of FM System and Hearing Aid Microphone Signals in Adverse Listening Conditions[Articles]Pittman, Andrea L.; Lewis, Dawna E.; Hoover, Brenda M.; Stelmachowicz, Patricia G.Received December 7, 1998; accepted April 23, 1999Address for correspondence: Andrea Pittman, Ph.D., Boys Town National Research Hospital, 555 North 30th Street, Omaha, NE 68131.Boys Town National Research Hospital, Omaha, Nebraska.AbstractObjective: Children with moderate to severe hearing loss routinely use personal frequency modulated (FM) systems in the classroom to improve the signal to noise ratio of teacher‐directed speech with notable success. Attention is now being given to the ability of these children to hear other students via the hearing aid (HA) microphone while using an FM system. As a result, a variety of FM system and HA microphone combinations have been recommended for classroom use. To date, there are no studies regarding the efficacy of these FM/HA combinations. The purpose of this study was to evaluate recognition performance using four FM/HA combinations and to characterize that performance for stimuli received primarily through FM system and HA microphone transmission.Design: Recognition performance for FM system and HA microphone signals was evaluated for two symmetrical and two asymmetrical FM/HA combinations using two commercially available FM systems (one conventional and one FM‐precedence circuit). Eleven children (ages 9 to 12) with moderate to severe sensorineural hearing loss and eight children (ages 10 to 11) with normal hearing served as subjects. The two symmetrical FM/HA combinations included: 1) binaural FM system and HA microphone input using the conventional FM system, and 2) binaural FM and HA input using the FM‐precedence circuit. The conventional FM system was used for the two asymmetrical combinations and included: 1) binaural FM input and monaural HA input, and 2) FM input to one ear and HA input to the other. Stimuli were 33 consonants presented in the form of nonsense syllables. The stimuli were presented through three loudspeakers representing a teacher and two fellow students in a classroom environment. Speech shaped noise was presented through two additional loudspeakers.Results: In general, no statistically significant differences in recognition performance were found between any of the FM/HA combinations. Mean recognition scores for HA microphone transmission (55%) were significantly poorer than those for FM system transmission (75%). As expected, initial consonants were more easily recognized than final consonants via FM system and HA microphone transmission. However, voiceless consonants were more easily recognized than voiced consonants via HA microphone transmission, which was not predicted on the basis of previous research.Conclusions: These results suggest that a certain amount of flexibility is present when choosing an FM/HA combination. However, recognition performance via the HA microphones was consistently poorer than performance via FM transmission. Because relevant material also originates from fellow students (e.g., answering teacher‐directed questions), input via the HAs is often as important as information originating from the teacher. The results suggest that attempts to improve performance for signals transmitted through the HA microphones in a classroom setting would benefit children with hearing loss.The deleterious effects of noise, distance, and reverberation on speech perception, especially for individuals with hearing loss, have been well documented (Finitzo, 1988; Olsen, 1988; Ross, 1986). These effects can be especially devastating to students in a classroom environment. One of the most common methods of reducing these effects in the classroom is the use of individual frequency modulated (FM) systems (Berger & Millin, 1989;Bess & Sinclair, 1985; Blair, 1977; Blair, Myrup, & Viehweg, 1989; Blake, Field, Foster, Plott, & Wertz, 1991; Boothroyd, 1992; Flexer, Wray, & Ireland, 1989; Hawkins, 1984, 1988; Jones, Berg, & Viehweg, 1989; Lewis, 1994; Nábelek, Donahue, & Letowski, 1986; Ross & Giolas, 1971; Smith, McConnell, Walter, & Miller, 1985; Wedmore, 1992). With these systems, the teacher wears a microphone/transmitter that sends a signal via FM radio waves to the student, who wears a receiver coupled to his or her hearing aids in one of a variety of ways. In general, the benefit received from FM systems is a consistent signal level and increased signal to noise ratio that results from placing a microphone close to the teacher’s mouth.With FM systems, the greatest improvement in signal to noise ratio is achieved when the system is worn in the FM‐only mode. In this arrangement, the hearing aid (HA) microphones are disabled, allowing input only from the FM receiver. Hawkins (1984) reported a mean signal to noise ratio advantage of 15.3 dB when comparing monaural FM‐only conditions with conventional HA conditions. When the FM system and HA microphone were active simultaneously with equal outputs, the binaural FM system advantage decreased to 7.4 dB. Clearly, the FM‐only mode of operation provides a significant advantage over the FM system plus HA microphone mode. However, the FM‐only mode precludes two important listening requirements: the ability to monitor the level of one’s own voice and the ability to hear other students who are not wearing the FM microphone. Thus, in a classroom environment, FM systems often are used in the FM system plus HA microphone mode of operation.When selecting and fitting FM systems, it is important to give precedence to the FM system signal while maintaining the audibility of the HA microphone signal. Because children with hearing loss require a 15 to 20 dB signal to noise ratio to adequately perceive speech (Finitzo‐Hieber, 1981; Finitzo‐Hieber & Tillman, 1978; Gengel, 1971; Olsen, 1988), FM system output should exceed the HA microphone output by that amount. The extent to which this can be accomplished is influenced by the design features of the FM system and the HA as well as the degree of hearing loss. For example, in cases of severe hearing loss, it may not be possible to amplify the FM system signal above the HA microphone signal without exceeding comfort levels. The only remaining option is to reduce the HA microphone output with the risk that portions of that signal then may be inaudible. Setting equal outputs for both the HA microphone and FM system signals is also ineffective. In a study of 13 teenagers with severe to profound hearing loss, Boothroyd and Iglehart (1998) found that the FM system advantage was virtually eliminated for equal FM system and HA microphone outputs under typical use conditions (80 dB SPL input to the FM microphone and 65 dB SPL input to the HA microphone). For this reason, the American Speech‐Language‐Hearing Association revised guidelines (1 ) suggest that, whenever possible, the SPL provided by the FM portion of the system should exceed that provided by the HA microphone by 10 dB. This recommendation represents a compromise between an ideal FM‐to‐HA advantage (15 to 20 dB) and what is realistic given current FM‐system technology.In classroom situations where new material often is generated by students (e.g., answering teacher‐directed questions), audibility of signals transmitted via the HA microphone becomes as important as those transmitted via the FM system signal. At least four restrictions prevent optimal audibility of both the HA microphone and FM system signals. First, the HA microphone signal must be set to a lower level than the FM system signal for the reasons described above. Second, coupling personal hearing aids to an FM system often is restricted by the options available within the hearing aids (e.g., hearing aids that have only microphone‐telecoil‐off options), which may prevent simultaneous FM system and HA microphone input. Third, with the advent of FM receivers that are contained within or easily attached to a hearing aid (audio boot), financial restrictions may limit the use of FM system input to one ear. Finally, in situations in which the signal to noise ratio is poor, HA microphone input to one ear and FM system input to both ears may be necessary to maintain a favorable FM system advantage. These and other restrictions have resulted in four common FM/HA combinations: 1) FM plus HA to both ears simultaneously, 2) FM plus HA to both ears simultaneously using an FM system that gives precedence to the FM signal, 3) FM plus HA to one ear and FM only to the other, and 4) FM only to one ear and HA only to the other ear.Although each of the above strategies has been recommended for a variety of academic environments, there have been no systematic studies of the efficacy of each of these arrangements. The primary purpose of this study was to compare the overall effectiveness of these four strategies with respect to listeners’ ability to hear and understand the main talker via the FM microphone as well as to hear and understand others via the hearing aid in a classroom setting. The secondary goal was to evaluate and characterize performance differences for signals transmitted via the FM system and HA microphones.MethodsParticipantsEleven children with moderate to severe hearing losses and eight children with normal hearing participated in this study. Pure‐tone thresholds at octave frequencies between 0.25 and 8 kHz and word‐recognition scores were obtained bilaterally for the children with hearing loss (Table 1). Hearing thresholds for the children with normal hearing were screened at 20 dB HL bilaterally at the same octave frequencies. All children in the hearing loss group were experienced hearing aid users, were mainstreamed in either public or private elementary schools, used oral communication as their primary method of communication, and had used FM systems in the classroom. The normally hearing children participated without the use of amplification or FM devices. Each child was paid for participation in this study.Table 1. Pure tone averages (PTA) in dB HL and CID W‐22 word‐recognition scores (percent correct) for the children with hearing impairment.Test MaterialsFour subtests from the Nonsense Syllable Test (Resnick, Dubno, Hoffnung, & Levitt, 1975) were used as test stimuli. The syllables in each subtest were produced in the /[α]/ vowel context and were: 1) eight initial voiced consonants (/b/, /d/, /g/, /m/, /n/, /đ/, /v/, and /z/), 2) nine initial voiceless consonants (/t∫/, /f/, /h/, /k/, /p/, /s/, /∫/, /t/, and /θ/), 3) nine final voiced consonants (/b/, /d/, /g/, /m/, /n/, /v/, /z/, /η/, and /đ/), and 4) seven final voiceless consonants (/f/, /k/, /p/, /s/, /t/, /∫/, and /θ/). With the exception of the phonemes /t∫/, /h/, and /η/, all consonants occurred in both the initial and final position. Test stimuli were spoken by a male speaker and digitized at a sampling rate of 20 kHz. A carrier phrase, “ready,” was spoken by another male speaker and digitized in the same manner.Classroom Setup and InstrumentationAll testing was conducted in a 9 × 9 meter classroom with a reverberation time of 600 msec.* The child was seated at a table in the center of five loudspeakers as illustrated in Figure 1. A single‐ cone loudspeaker representing the teacher was positioned directly in front of the child. Two loudspeakers were placed behind and to each side of the child to simulate students seated one row behind and one row to the side of the child. To represent classroom noise, speech‐weighted noise was delivered through two loudspeakers positioned to each side of the child.The calculation of critical distances was complicated by irregularities in the height of the ceiling but was at least 3.86 meters, placing the child within the critical distance of each of the five loudspeakers.Figure 1. Schematic diagram showing the loudspeaker placement relative to the child. All signal levels are indicated relative to the child’s position, except for the level at the FM microphone, which was positioned 20 cm from the front loudspeaker.An FM microphone was positioned 20 cm in front of the front loudspeaker.† Long‐term average speech levels at this distance were adjusted to 80 dB SPL with a flat frequency response to simulate the levels commonly found at a teacher’s lavaliere microphone position (Hawkins, 1984; Hawkins & Schum, 1985; Lybarger, 1981; Picard & LeFrancois, 1986; Turner & Holte, 1985). At the child’s position, the front loudspeaker then was adjusted to achieve a speech level of 66 dB SPL, corresponding to the level of a teacher’s voice at 2 meters (Pearsons, Bennett, & Fidell, 1977). This level was intended to simulate the raised vocal levels of a teacher in a typical classroom measured at the position of a child seated in the first or second row. Speech‐weighted noise presented from the side loudspeakers was adjusted to 60 dB SPL at the child’s position, thus yielding a teacher‐to‐noise ratio of +6 dB at the unaided ear (Finitzo‐Hieber & Tillman, 1978). When amplified by the FM systems, teacher‐to‐noise ratios were between 15 and 20 dB. Speech levels from the two loudspeakers representing the students were adjusted to 59 dB SPL at the child’s position. These levels approximated average speech levels from a distance of 1 meter (Cox & Moore, 1988) and resulted in a −1 dB student‐to‐noise ratio for unaided ears as well as for amplification via HA microphone input.Two different personal FM systems (Phonak Microvox and Phonic Ear Solaris) were used in conjunction with behind‐the‐ear hearing aids (Phonak Pico‐Forte C‐2) that had been modified to allow FM system only and FM system plus HA microphone input. For each hearing‐impaired child, target frequency/gain characteristics were determined using the Desired Sensation Level (V. 4.1) procedure (Seewald, Cornelisse, Ramji, Sinclair, Moodie, & Jamieson, Reference Note 2 ). The overall goal of each fitting was to allow audibility of speech via the HA microphone without compromising the FM system advantage. For the conventional FM system (Phonak Microvox), a 10 dB FM system advantage was achieved by setting the FM system output to 5 dB above target and the HA microphone output to 5 dB below target. Both the FM system and HA microphone settings for the FM‐precedence system (Phonic Ear Solaris) were set to Desired Sensation Level targets because the FM‐precedence feature of this system automatically reduces the HA microphone output by 10 dB when activated by an FM system signal. Figure 2 illustrates the frequency/gain characteristics of the conventional and FM precedence systems in panels a and b, respectively. An advantage to the FM precedence system is that incoming signals to the FM or HA microphones are delivered at optimal sensation levels with precedence given to the FM system signal when input from the FM microphone is 72 dB or higher. The output levels for these systems were confirmed in a standard 2 cm3 coupler using a 70 dB composite weighted noise input to the HA microphone and an 80 dB composite weighted noise input to the FM microphone (ASHA, Reference Note 1; Seewald et al., Reference Note 2 ).Figure 2. Hearing aid frequency response curves for the conventional and FM precedence systems. Panels A and B show the Desired Sensation Level target values of the FM and HA signals for the conventional and FM precedence systems, respectively.FM/HA CombinationsFour commonly used combinations that allow for both FM system and HA microphone input were chosen for this study. Schematic diagrams of these combinations are shown in Figure 3. Two symmetrical (S1 and S2) and two asymmetrical combinations (A1 and A2) are depicted in the upper and lower panels, respectively. In the symmetrical combinations, the FM system plus HA microphones were active binaurally using either the conventional FM system (S1) or the FM‐precedence system (S2). In the asymmetrical conditions, the conventional FM system was configured to deliver an FM system plus HA microphone signal to one ear and an FM system signal to the other ear (condition A1) and an FM system signal to one ear and a HA microphone signal only to the other ear (condition A2). An FM‐only combination was not included because it is rarely used in a typical classroom and the advantages of FM‐only in single‐speaker situations have been clearly demonstrated (Hawkins, 1984).Figure 3. Schematic representation of each FM combination. S1 and S2 illustrate the symmetrical combinations, and A1 and A2 illustrate the asymmetrical combinations.ProcedureEach child was seated at a table positioned near the center of the five loudspeakers (see Fig. 1). Using a personal computer and mouse, the children were asked to select the correct response from seven to nine possible choices (depending on the subtest). Written representations as well as pictures containing each nonsense syllable were provided for each response option. Before testing, each child was given a practice test to familiarize him or her with the test stimuli and response format.Before the start of each trial, the child was informed of the source of the speech stimuli (from the front, left, or right loudspeakers). Speech‐weighted noise was presented throughout each trial. To further simulate a classroom environment, each child was encouraged to make any necessary adjustments in head position (without moving his or her chair) to optimize audibility for stimuli originating from the rear. The carrier phrase “ready” was delivered through the front loudspeaker 1000 msec before each nonsense syllable to allow sufficient time for a head turn. This pause between the end of the carrier phrase and the test items simulated the timing of a teacher calling on a student and then waiting for a response. The length of the pause was a potential problem for the FM‐precedence circuit; however, the recovery time was sufficiently brief for the circuit to return to target settings and receive HA microphone input.‡For the children with hearing loss, there were 10 experimental conditions. Stimuli were presented from three loudspeaker locations (front, right, and left) for the two asymmetrical FM/HA combinations (2 × 3). Only two loudspeaker locations were used for the two symmetrical FM/HA combinations (front and right or front and left) because recognition scores most likely would be equivalent for both side loudspeaker locations (2 × 2). For the children with normal hearing, only two loudspeaker locations (front and right or front and left) were necessary to evaluate recognition in an unaided state. The subtests were presented in the same order for each trial: initial voiced, initial voiceless, final voiced, and final voiceless consonants. Within each subtest, nonsense syllables were presented randomly three times. The order of FM/HA combinations and loudspeaker locations were randomized across participants.Results and DiscussionRecognition Performance and FM/HA CombinationThe primary purpose of this study was to determine which of the FM/HA combinations resulted in better perception of the teacher via the FM system as well as the other students via the HA microphone. Recognition results for the front, left, and right loudspeaker conditions were converted to rationalized arcsine units§ for analyses (Studebaker, 1985). Recognition scores for the left and right loudspeaker conditions then were grouped as on‐side or off‐side according to the position of each relative to the HA microphone worn by the child. A 1‐way analysis of variance (ANOVA) revealed a significant difference in recognition performance between the three loudspeaker positions (front, on‐side, and off‐side) [F(2) = 139.28; p < 0.01]. Post hoc analyses revealed a significant difference between the front and on‐side and the front and off‐side loudspeaker positions but failed to show a significant difference between the on‐side and off‐side loudspeaker positions. These results suggest that recognition performance was not negatively influenced when the HA microphone was placed on the opposite side of the head from the source of the speech stimuli. Although this finding was somewhat surprising, it probably occurred because the children were allowed to turn their heads. For all subsequent data analyses, recognition results for the off‐side loudspeaker position were omitted.Figure 4 shows the means and standard deviations of the speech recognition results in percent correct as a function of FM/HA combination for the children with hearing loss. For comparative purposes, data from the children with normal hearing are shown at the right side of the figure. In general, differences in recognition performance across the four FM/HA combinations were small. Recognition performance for both groups of children was highest for the front loudspeaker position and decreased for the side loudspeaker position. Mean percent correct scores (with standard deviations in parentheses) for FM/HA combinations S1, S2, A1, and A2 were 73.67 (14.16), 72.25 (12.23), 77.99 (11.21), 74.57 (13.03), and 50.91 (14.43), 54.68 (14.92), 50.84 (14.11), 55.18 (15.42) for the front and side loudspeaker conditions, respectively. A 2‐way ANOVA was performed using the within‐subjects factors of FM/HA combination (S1, S2, A1, or A2) and loudspeaker position (front or side). This analysis revealed a significant main effect for loudspeaker position [F(1, 20) = 209.71; p < 0.01] but not for FM/HA combination.∥ No interaction between FM/HA combination and loudspeaker position was found. The significantly poorer recognition performance via HA microphone transmission (side loudspeaker) compared with FM system transmission (front loudspeaker) was expected due to the lower level and poorer signal to noise ratio in the HA microphone condition. Differences in recognition performance also were expected across the symmetrical and asymmetrical FM/HA combinations; however, none were found. These results suggest that, under the conditions used in this study, a binaural HA combination did not provide an advantage over a monaural HA combination.Figure 4. Mean recognition performance in percent correct for FM system and HA microphone transmission as a function of FM/HA combination. S1, S2, A1, and A2 indicate the symmetrical and asymmetrical combinations depicted in Figure 2. Recognition performance for children with normal hearing (NH) is shown at the right. Error bars indicate 1 SD.Differences in recognition performance between the FM‐precedence and conventional FM system also were expected due to differences in signal processing, but none were found. Recall that for the FM‐precedence circuit, the FM system and HA microphone outputs were set to target values, whereas, for the conventional FM system, the FM system and HA microphone outputs were set to +5 and −5 dB, respectively, relative to target values (see Fig. 2). The conventional FM system provided a 5 dB higher FM system signal, and the FM‐precedence system provided a 5 dB higher HA microphone signal. For this reason, higher recognition scores were expected via HA microphone transmission for the FM‐precedence circuit and via FM system transmission for the conventional FM system. These results suggest that an increase of 5 dB for either the FM system or HA microphone signal was not large enough to improve recognition performance. It is possible that the output levels produced by both FM system and HA microphone transmission were already sufficiently high to maximize recognition performance for these children.Compared with the children with hearing loss, the children with normal hearing exhibited higher recognition scores for both the front and side loudspeaker conditions (see Fig. 4). Mean percent correct scores (with standard deviations in parentheses) for the front and side loudspeaker conditions were 89.89 (8.50) and 83.41 (8.65), respectively. To determine whether these scores differed significantly from those of the hearing‐impaired children, data were collapsed across FM/HA combinations and a 2‐way ANOVA was performed using the between‐subjects factors of hearing status (normal hearing or hearing loss) and loudspeaker position (front or side). Both main effects were significant (hearing status: F(1, 36) = 174.70; p < 0.01; loudspeaker position: F(1, 36) 67.78; p < 0.01) as was their interaction [F(1, 36) = 11.43; p = 0.001]. Thus, children with hearing loss exhibited significantly lower recognition scores for the front (FM) and side (HA) loudspeaker conditions overall, with a greater decrement for the side loudspeaker position. For these children, recognition scores via FM system transmission were likely limited by their underlying speech recognition abilities (see Table 1). These results suggest that during student‐to‐teacher and student‐to‐student interactions, substantially more information would be missed by the children with hearing loss compared with their normally hearing peers.To determine whether individual differences across FM/HA combinations were present, a normal approximation of the binomial distribution was used to provide an estimate of the standard errors of means and mean differences as described by Thornton and Raffin (1978). Ninety‐five percent critical differences were calculated for the highest scores obtained by each child in the four FM/HA combinations and then were compared with the remaining three scores. Separate calculations were made for the front and side loudspeaker conditions. Table 2 lists the individual scores for the four FM/HA combinations by loudspeaker condition for each subject. In general, the group effects discussed above represent the individual performances for all but three of the hearing‐impaired children. Subjects 1 and 9 showed significantly higher recognition scores for one of the FM/HA combinations in both the front and side loudspeaker conditions. However, the FM/HA combinations resulting in the significantly higher score were not the same for the two loudspeaker conditions, suggesting that one FM/HA combination would not be sufficient to maximize speech perception via the FM system and HA microphone for these two children. Subject 10, on the other hand, showed significantly higher performance for one FM/HA combination in the side loudspeaker condition. For this child, that FM/HA combination would have been optimal for receiving both FM system and HA microphone signals. Because recognition performance may be significantly poorer with one FM/HA combination for some children, it may be wise to try different FM/HA combinations whenever a child appears to be performing poorly with his or her current FM/HA combination.Table 2. Recognition scores (in percent correct) for FM/HA combinations (S1, S2, A1, A2) and loudspeaker positions (front and side) by subject. Numbers in bold indicate significantly higher scores relative to the lower scores obtained (indicated by asterisks) for each subject in each loudspeaker condition.Performance CharacteristicsA secondary purpose of this study was to evaluate and characterize performance differences for stimuli transmitted via the FM system and HA microphones for the children with hearing impairment. Collapsed across FM/HA combinations and loudspeaker conditions, average recognition performance for consonant position (initial and final) and consonant voicing (voiced and voiceless) was computed, and these results are displayed in Figure 5 a. Initial consonants were recognized more easily than final consonants, and voiceless consonants were recognized more easily than voiced consonants. A 2‐way ANOVA revealed significant main effects for both consonant position [F(1, 20) = 34.91; p < 0.01] and consonant voicing [F(1, 20) = 5.88; p = 0.016] with no consonant position × voicing interaction. Significantly higher recognition scores for initial consonants are consistent with the results of previous studies (Dubno, Dirks, & Langhofer, 1982; Dubno & Levitt, 1981). The higher recognition scores for voiceless compared with voiced consonants were surprising because previous studies in both normally hearing and hearing‐impaired individuals have found that the predominantly high‐frequency, voiceless consonants would be more difficult to recognize in noise than the predominantly low‐frequency, voiced consonants (Dubno et al., 1982; Dubno & Levitt, 1981; Miller & Nicely, 1955). In fact, Miller and Nicely (1955) concluded that voicing and nasality were more salient features in the recognition of phonemes than frication and duration when the speech stimuli were low‐pass filtered or presented in broadband noise. In this study, the voiced/voiceless relationship was reversed.Figure 5. Recognition performance (in percent correct) for consonant position and voicing averaged across all FM/HA combinations. Panel A shows mean recognition scores for the front (FM) and side (HA) loudspeaker positions as a function of position and voicing. Panel B shows mean recognition performance for position (initial and final) and voicing (voiced and voiceless) as a function of FM system and HA microphone input.To investigate these findings further, average recognition performance for consonant position and consonant voicing by loudspeaker position are displayed in Figure 5b. As expected, higher scores were achieved for the front (FM) loudspeaker position compared with the side (HA) loudspeaker position for both consonant position and consonant voicing. Initial consonants were recognized more readily than final consonants for both loudspeaker positions. The recognition of voiced and voiceless consonants through the front (FM) loudspeaker was nearly equivalent, whereas higher recognition scores were achieved for the voiceless consonants through the side (HA) loudspeaker. A 3‐way ANOVA using the within‐subjects factors of consonant voicing (voiced or voiceless), consonant position (initial or final), and loudspeaker position (front or side) revealed significant main effects of voicing [F(1, 20) = 10.17; p < 0.01], consonant position [F(1, 20) = 63.61; p < 0.01], and loudspeaker position [F(1, 20) = 261.52; p < 0.01], as well as a voicing × loudspeaker position interaction [F(1, 20) = 17.98; p < 0.01]. These results indicate that significantly better recognition occurred for: 1) the front loudspeaker condition (FM system transmission) overall; 2) initial consonants for both the front (FM) and side (HA) loudspeaker conditions; and 3) voiceless consonants for the side loudspeaker condition (HA microphone transmission).The reason for the better recognition of voiceless consonants via the side loudspeaker condition is unclear. It is unlikely that the difference in signal to noise ratios between the two conditions (\f20 dB for FM and −1 dB for HA) would have produced the observed differences in voicing perception. If anything, the poorer signal to noise ratio was expected to result in a larger decrement in performance for the voiceless consonants. In Figure 6, mean recognition scores for each consonant are shown for the two loudspeaker positions. The left‐hand panels show the voiced and voiceless consonant pairs arranged by manner (stops and fricatives) and place of articulation (front, middle, back). In the right‐hand panels, the three voiced and voiceless consonants that had no counterparts are displayed. In general, lower recognition scores were achieved for the side loudspeaker position (HA microphone transmission—lower panel) compared with the front loudspeaker position (FM system transmission—upper panel), presumably because of the poorer signal to noise ratio. Overall, there was increased recognition for five of the six voiceless consonants relative to their unvoiced cognates. These higher scores occurred for the /p/ and /f/ for both loudspeaker conditions and for /t/, /s/, and /θ/ for the side loudspeaker condition. The performance for individual phonemes suggests that no one place‐ or manner‐of‐production feature was responsible for the higher voiceless recognition scores for the side loudspeaker condition.Figure 6. Recognition performance (in percent correct) for voiced and voiceless consonant pairs. The top and bottom panels illustrate recognition scores for the front (FM) and side (HA) loudspeaker conditions, respectively. Consonants are paired in voiced/voiceless cognates and arranged by manner and place of articulation (front to back).General DiscussionThe overall goal of this study was to determine whether any particular FM/HA combination can provide superior benefit to children with hearing loss in a simulated classroom environment. Results revealed that the four FM/HA combinations tested yielded recognition scores that were statistically the same with differences of only four to six percentage points. Somewhat surprisingly, scores for the two asymmetric combinations were equivalent to the two symmetric combinations. Likewise, no differences were found between the FM‐precedence system and the conventional FM system. These results suggest that, at least for children with moderate to moderately severe hearing losses, there may be considerable flexibility when choosing an FM/HA combination. In those cases in which an asymmetrical combination may be necessary (e.g., financial or instrument limitations), these results suggest that recognition will not be significantly altered in a classroom, as simulated here. This result was likely due to the head‐turning strategies adopted by each child when attempting to recognize the speech stimuli from off‐side loudspeaker positions. Caution should be used, however, when attempting to extend the results of this study to a real classroom situation. First, the head‐turning strategies adopted by the children are not controllable (predictable) and may not be used consistently over the course of an entire school day. Second, binaural HA microphone input may be necessary to perceive speech from a greater distance than that used in this study. For example, the level of the speech stimuli from the side loudspeakers was adjusted to simulate the speech levels of a student seated 1 meter away. Increasing the distance likely will increase the difficulty of speech recognition for which binaural amplification may prove to be advantageous. Finally, there may be individual differences in some children’s ability to benefit from different FM/HA combinations. The mean group data failed to show significant differences, but significant differences (Thornton & Raffin, 1978) were observed across FM/HA combinations for a few individuals. In the absence of objective tests to differentiate performance across FM/HA combinations, subjective evaluation of a child’s performance (by both the teacher and the child) may be necessary to determine when an FM/HA combination is less than optimal. The nature of the problems experienced by the child, the classroom configuration, and the available hardware then may dictate the next course of action.Although an FM‐only condition was not included in this study, recognition performance for nonsense syllables with a 10 dB FM system advantage resulted in scores between 72 and 78%, which are comparable with recognition scores for the CID W‐22 words presented in quiet under earphones (see Table 1). These findings support the notion that the combined use of FM system and HA microphone transmission under the conditions used in this study (10 dB separation between FM and HA output) did not degrade recognition of the FM system signal substantially. Note, however, that recognition scores via HA microphone transmission were consistently poor. Regardless of the FM/HA combination in this study, mean recognition decreased to 50% for signals from the side (HA) loudspeaker at a distance of 1 meter. This observation suggests that student‐to‐student communication from 1 meter via linear HA microphone transmission is less than adequate in noise and would be expected to degrade further for students seated at a greater distance and/or outside of the direct field.The better recognition of voiceless consonants compared with voiced consonants via HA microphone transmission was perplexing. In previous studies of voiced/voiceless contrasts, speech stimuli were presented monaurally through an earphone to normally hearing and hearing‐impaired individuals (Dubno et al., 1982; Dubno & Levitt, 1981; Miller & Nicely, 1955). The speech stimuli in the present study were delivered from an off‐axis loudspeaker position to personal hearing aids. It is possible that significant differences in perception exist between speech that is presented in the sound field and speech presented under earphones. However, additional research is necessary to determine whether these differences are due to the amplification devices, the position of the loudspeaker, or a combination of the two.ConclusionsThe results of this study suggest that a certain amount of flexibility is available when choosing an FM/HA combination for children with moderate to severe sensorineural hearing loss. On average, the children performed equally well with the two symmetrical and two asymmetrical FM/HA combinations used in this study. Likewise, equal average performance was achieved for the FM‐precedence and conventional FM systems. Individual differences for three of the children were present; however, performance with one FM/HA combination was found to significantly exceed that of the other combinations for only one child. Thus, in individual cases where performance is less than optimal with one FM/HA combination, changing to another FM/HA combination may be beneficial.Recognition of the speech stimuli via FM system input was at or near maximum performance, indicating that the addition of noise to the HA microphone did not substantially degrade recognition of the FM system signal. Recognition via the HA microphone, however, was consistently poor. On average, performance of the children with hearing loss was about 30 percentage points poorer than that of the children with normal hearing for speech at a distance of 1 meter. Even poorer performance would be expected when students are seated at greater distances or listening at more deleterious signal to noise ratios. Further research in this area might include the advantages of FM‐precedence circuitry in children with severe hearing loss, recognition performance through FM systems coupled to nonlinear hearing aids, and further study of the recognition characteristics of speech presented from off‐axis loudspeaker positions.Acknowledgments:The authors acknowledge Reinier Kortekaas for his many helpful comments on earlier versions of this manuscript and Steve Neely for his assistance with reverberation time measures.This work was supported by a grant from Phonak Inc.FOOTNOTES* The reverberation time of the room was determined by a reverse‐time energy procedure (Scroeder, 1965). An impulse response of the room was measured by synchronous averaging of a wide‐band stimulus. The squared value of the impulse response was integrated in reverse time to produce a decay curve. A 60 dB reverberation time was estimated by extrapolating the linear portion for the decay curve, which occurred immediately after the initial direct sound. [Context Link]†Empirical tests using a one‐third octave band analysis determined that a distance of 20 cm from the single‐coned front loudspeaker yielded a speech spectrum not influenced by near field effects (Lewis, Feigin, Karasek, & Stelmachowicz, 1991). [Context Link]‡The manufacturer’s specifications for the FM‐precedence circuit reported a “dwell time” of 1000 msec after activation with a release time of 2 msec. Once activated, the system will continue to suppress HA microphone output for 1000 msec and then return to the original gain if no further input is received from the FM microphone. The stimuli were presented 1000 msec after the carrier phrase, leaving a 2 msec overlap between the release from FM precedence and the onset of each test item. [Context Link]§Transforming percent correct scores into arcsine units expresses a range of recognition scores with homogenous variances. The added variation introduced by Studebaker (1985) approximates the original percent correct values before transformation within the range of 20 to 80%. [Context Link]∥The small differences in recognition performance between the four front and four side loudspeaker conditions (6 and 4 percentage points, respectively) were not considered clinically significant. With the subjects recruited for this study (11), an effect size of 1.5 times the standard deviation (about 20 percentage points) could be detected with a power of 0.82 (1‐β). [Context Link]ReferencesBerger, K., & Millin, J. (1989). Amplification/assistive devices for the hearing impaired. In R. Schow & M. Nerbonne (Eds.), Introduction to aural rehabilitation (pp. 31–80). Austin, TX: Pro-Ed. [Context Link]Bess, F., & Sinclair, J. S. (1985). Amplification systems used in education. In J. Katz (Ed.), Handbook of clinical audiology (pp. 970–985). Baltimore: Williams and Wilkins. [Context Link]Blair, J. (1977). Effects of amplification, speechreading, and classroom environments on reception of speech. Volta Review, 79, 443–449. [Context Link]Blair, J., Myrup, C., & Viehweg, S. (1989). Comparison of the listening effectiveness of hard-of-hearing children using three types of amplification. Educational Audiology Monograph, 1, 48–55. [Context Link]Blake, R., Field, B., Foster, C., Plott, F., & Wertz, P. (1991). Effect of FM auditory trainers on attending behaviors of learning-disabled children. Language, Speech, and Hearing Services in the Schools, 22, 111–114. [Context Link]Boothroyd, A. (1992). The FM wireless link: An invisible microphone cable. In M. Ross (Ed.), FM auditory training systems: Characteristics, selection, and use (pp. 1–19). Timonium, MD: York Press. [Context Link]Boothroyd, A., & Iglehart, F. (1998). Experiments with classroom FM amplification. Ear and Hearing, 19, 202–217. [CrossRef] [Full Text] [Medline Link] [Context Link]Cox, R. M., & Moore, J. N. (1988). Composite speech spectrum for hearing aid gain prescriptions. Journal of Speech and Hearing Research, 31, 102–107. [Medline Link] [Context Link]Dubno, J. R., Dirks, D. D., & Langhofer, L. R. (1982). Evaluation of hearing-impaired listeners using a nonsense-syllable test. II. Syllable recognition and consonant confusion patterns. Journal of Speech and Hearing Research, 25, 141–148. [Medline Link] [Context Link]Dubno, J. R., & Levitt, H. (1981). Predicting consonant confusions from acoustic analysis. Journal of the Acoustical Society of America, 69, 249–261. [CrossRef] [Medline Link] [Context Link]Finitzo, T. (1988). Classroom acoustics. In R. Roeser & R. Downs (Eds.), Auditory disorders in school children: Identification, remediation, 2nd ed. (pp. 221–233). New York: Thieme-Stratton. [Context Link]Finitzo-Hieber, T. (1981). Classroom acoustics. In R. J. Roeser & M. P. Downs (Eds.), Auditory disorders in school children: The law, identification and remediation (pp. 250–262). New York: Thieme-Stratton. [Context Link]Finitzo-Hieber, T., & Tillman, T. (1978). Room acoustics effects on monosyllabic word discrimination ability for normal and hearing-impaired children. Journal of Speech and Hearing Research, 21, 440–458. [Medline Link] [Context Link]Flexer, C., Wray, D., & Ireland, J. (1989). Preferential seating is not enough: Issues in classroom management of hearing impaired students. Language, Speech, and Hearing Services in the Schools, 20, 11–21. [Context Link]Gengel, R. (1971). Acceptable signal to noise ratio for aided speech discrimination by the hearing impaired. Journal of Auditory Research, 11, 219–222. [Medline Link] [Context Link]Hawkins, D. B. (1984). Comparisons of speech recognition in noise by mildly-to-moderately hearing-impaired children using hearing aids and FM systems. Journal of Speech and Hearing Disorders, 49, 409–418. [CrossRef] [Medline Link] [Context Link]Hawkins, D. (1988). Options in classroom amplification. In F. Bess (Ed.), Hearing impairment in children (pp. 253–265). Parkton, MD: York Press, Inc. [Context Link]Hawkins, D., & Schum, D. (1985). Some effects of FM-system coupling on hearing aid characteristics. Journal of Speech and Hearing Disorders, 50, 132–141. [CrossRef] [Medline Link] [Context Link]Jones, J., Berg, F., & Viehweg, S. (1989). Close, distant and sound field overhead listening in kindergarten classrooms. Educational Audiology Monograph, 1, 56–65. [Context Link]Lewis, D. (1994). Assistive devices for classroom listening: FM systems. American Journal of Audiology, 3, 70–83. [CrossRef] [Context Link]Lewis, D., Feigin, J., Karasek, A., & Stelmachowicz, P. (1991). Evaluation and assessment of FM systems. Ear and Hearing, 12, 268–280. [CrossRef] [Full Text] [Medline Link] [Context Link]Lybarger, S. (1981). Standard acoustical measurements on auditory training devices. In F. Bess, B. Freeman, & J. S. Sinclair (Eds.), Amplification in education (pp. 305–315). Washington, DC: Alexander Graham Bell Association for the Deaf. [Context Link]Miller, G. A., & Nicely, P. E. (1955). An analysis of perceptual confusions among some English consonants. Journal of the Acoustical Society of America, 27, 338–352. [CrossRef] [Medline Link] [Context Link]Nábelek, A. K., Donahue, A. M., & Letowski, T. R. (1986). Comparison of amplification systems in a classroom. Journal of Rehabilitation Research and Development, 23, 41–52. [Medline Link] [Context Link]Olsen, W. O. (1988). Classroom acoustics for hearing-impaired children. In F. H. Bess (Ed.), Hearing impairment in childhood (pp. 266–277). Parkton, MD: York Press. [Context Link]Pearsons, K., Bennett, R., & Fidell, S. (1977). Speech levels in various noise environments. Project report on contract 68 01–2466. Washington, DC: U.S. Environmental Protection Agency. [Context Link]Picard, M., & LeFrancois, J. (1986). Speech perception through FM auditory trainers in noise and reverberation. Journal of Rehabilitation Research and Development, 23, 53–62. [Medline Link] [Context Link]Resnick, S., Dubno, J. R., Hoffnung, S., & Levitt, H. (1975). Phoneme errors on a nonsense syllable test. Journal of the Acoustical Society of America, 58, S114. [CrossRef] [Context Link]Ross, M. (1986). Classroom amplification. In W. Hodgson (Ed.), Hearing aid assessment and use in audiologic habilitation (pp. 231–265). Baltimore: Williams and Wilkins. [Context Link]Ross, M., & Giolas, T. G. (1971). Effect of three classroom listening conditions on speech intelligibility. American Annals of the Deaf, 116, 580–584. [Medline Link] [Context Link]Scroeder, M. R. (1965). New method of measuring reverberation time. Journal of the Acoustical Society of America, 37, 409–412. [CrossRef] [Context Link]Smith, D., McConnell, J., Walter, T., & Miller, S. (1985). Effect of using an auditory trainer on the attentional, language, and social behaviors of autistic children. Journal of Autism and Developmental Disorders, 15, 285–302. [CrossRef] [Medline Link] [Context Link]Studebaker, G. A. (1985). A “rationalized” arcsine transform. Journal of Speech and Hearing Research, 28, 455–462. [Medline Link] [Context Link]Thornton, A. R., & Raffin, M. J. M. (1978). Speech-discrimination scores modeled as a binomial variable. Journal of Speech and Hearing Research, 21, 507–518. [Medline Link] [Context Link]Turner, C., & Holte, L. (1985). Evaluation of FM amplification systems. Hearing Instruments, 36, 6–12, 56. [Context Link]Wedmore, S. (1992). Auditory trainers help distractible students become “working” listeners. Advance for Speech-Language Pathologists and Audiologists, 2 (11), 13. [Context Link]Reference Notes1. American Speech-Language-Hearing Association (in review). Guidelines for fitting and monitoring FM systems. [Context Link]2. Seewald, R. C., Cornelisse, L. E., Ramji, K. V., Sinclair, S. T., Moodie, K. S., & Jamieson, D. G. (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. London, Ontario, Canada: Hearing Healthcare Research Unit, University of Western Ontario. [Context Link]ovid.com:/bib/ovftdb/00003446-199908000-0000100003446_1998_19_202_boothroyd_amplification_|00003446-199908000-00001#xpointer(id(R7-1))|11065213||ovftdb|00003446-199806000-00004SL0000344619981920211065213P71[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100003446_1998_19_202_boothroyd_amplification_|00003446-199908000-00001#xpointer(id(R7-1))|11065404||ovftdb|00003446-199806000-00004SL0000344619981920211065404P71[Full Text]00003446-199806000-00004ovid.com:/bib/ovftdb/00003446-199908000-0000100003446_1998_19_202_boothroyd_amplification_|00003446-199908000-00001#xpointer(id(R7-1))|11065405||ovftdb|00003446-199806000-00004SL0000344619981920211065405P71[Medline Link]9657595ovid.com:/bib/ovftdb/00003446-199908000-0000100005334_1988_31_102_cox_prescriptions_|00003446-199908000-00001#xpointer(id(R8-1))|11065405||ovftdb|SL0000533419883110211065405P72[Medline Link]3352247ovid.com:/bib/ovftdb/00003446-199908000-0000100005334_1982_25_141_dubno_recognition_|00003446-199908000-00001#xpointer(id(R9-1))|11065405||ovftdb|SL0000533419822514111065405P73[Medline Link]7087417ovid.com:/bib/ovftdb/00003446-199908000-0000100004470_1981_69_249_dubno_predicting_|00003446-199908000-00001#xpointer(id(R10-1))|11065213||ovftdb|SL0000447019816924911065213P74[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100004470_1981_69_249_dubno_predicting_|00003446-199908000-00001#xpointer(id(R10-1))|11065405||ovftdb|SL0000447019816924911065405P74[Medline Link]7217523ovid.com:/bib/ovftdb/00003446-199908000-0000100005334_1978_21_440_finitzo_discrimination_|00003446-199908000-00001#xpointer(id(R13-1))|11065405||ovftdb|SL0000533419782144011065405P77[Medline Link]713515ovid.com:/bib/ovftdb/00003446-199908000-0000100004587_1971_11_219_gengel_discrimination_|00003446-199908000-00001#xpointer(id(R15-1))|11065405||ovftdb|SL0000458719711121911065405P79[Medline Link]ovid.com:/bib/ovftdb/00003446-199908000-0000100005333_1984_49_409_hawkins_comparisons_|00003446-199908000-00001#xpointer(id(R16-1))|11065213||ovftdb|SL0000533319844940911065213P80[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100005333_1984_49_409_hawkins_comparisons_|00003446-199908000-00001#xpointer(id(R16-1))|11065405||ovftdb|SL0000533319844940911065405P80[Medline Link]6503247ovid.com:/bib/ovftdb/00003446-199908000-0000100005333_1985_50_132_hawkins_characteristics_|00003446-199908000-00001#xpointer(id(R18-1))|11065213||ovftdb|SL0000533319855013211065213P82[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100005333_1985_50_132_hawkins_characteristics_|00003446-199908000-00001#xpointer(id(R18-1))|11065405||ovftdb|SL0000533319855013211065405P82[Medline Link]3990259ovid.com:/bib/ovftdb/00003446-199908000-0000100046825_1994_3_70_lewis_assistive_|00003446-199908000-00001#xpointer(id(R20-1))|11065213||ovftdb|SL00046825199437011065213P84[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100003446_1991_12_268_lewis_evaluation_|00003446-199908000-00001#xpointer(id(R21-1))|11065213||ovftdb|00003446-199108000-00006SL0000344619911226811065213P85[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100003446_1991_12_268_lewis_evaluation_|00003446-199908000-00001#xpointer(id(R21-1))|11065404||ovftdb|00003446-199108000-00006SL0000344619911226811065404P85[Full Text]00003446-199108000-00006ovid.com:/bib/ovftdb/00003446-199908000-0000100003446_1991_12_268_lewis_evaluation_|00003446-199908000-00001#xpointer(id(R21-1))|11065405||ovftdb|00003446-199108000-00006SL0000344619911226811065405P85[Medline Link]1783229ovid.com:/bib/ovftdb/00003446-199908000-0000100004470_1955_27_338_miller_perceptual_|00003446-199908000-00001#xpointer(id(R23-1))|11065213||ovftdb|SL0000447019552733811065213P87[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100004470_1955_27_338_miller_perceptual_|00003446-199908000-00001#xpointer(id(R23-1))|11065405||ovftdb|SL0000447019552733811065405P87[Medline Link]ovid.com:/bib/ovftdb/00003446-199908000-0000100005247_1986_23_41_nabelek_amplification_|00003446-199908000-00001#xpointer(id(R24-1))|11065405||ovftdb|SL000052471986234111065405P88[Medline Link]3958997ovid.com:/bib/ovftdb/00003446-199908000-0000100005247_1986_23_53_picard_reverberation_|00003446-199908000-00001#xpointer(id(R27-1))|11065405||ovftdb|SL000052471986235311065405P91[Medline Link]3958998ovid.com:/bib/ovftdb/00003446-199908000-0000100004470_1975_58_s114_resnick_nonsense_|00003446-199908000-00001#xpointer(id(R28-1))|11065213||ovftdb|SL00004470197558s11411065213P92[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100000241_1971_116_580_ross_intelligibility_|00003446-199908000-00001#xpointer(id(R30-1))|11065405||ovftdb|SL00000241197111658011065405P94[Medline Link]5132418ovid.com:/bib/ovftdb/00003446-199908000-0000100004470_1965_37_409_scroeder_reverberation_|00003446-199908000-00001#xpointer(id(R31-1))|11065213||ovftdb|SL0000447019653740911065213P95[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100004591_1985_15_285_smith_attentional_|00003446-199908000-00001#xpointer(id(R32-1))|11065213||ovftdb|SL0000459119851528511065213P96[CrossRef]ovid.com:/bib/ovftdb/00003446-199908000-0000100004591_1985_15_285_smith_attentional_|00003446-199908000-00001#xpointer(id(R32-1))|11065405||ovftdb|SL0000459119851528511065405P96[Medline Link]4030662ovid.com:/bib/ovftdb/00003446-199908000-0000100005334_1985_28_455_studebaker_rationalized_|00003446-199908000-00001#xpointer(id(R33-1))|11065405||ovftdb|SL0000533419852845511065405P97[Medline Link]4046587ovid.com:/bib/ovftdb/00003446-199908000-0000100005334_1978_21_507_thornton_discrimination_|00003446-199908000-00001#xpointer(id(R34-1))|11065405||ovftdb|SL0000533419782150711065405P98[Medline Link]713519Recognition Performance for Four Combinations of FM System and Hearing Aid Microphone Signals in Adverse Listening ConditionsPittman, Andrea L.; Lewis, Dawna E.; Hoover, Brenda M.; Stelmachowicz, Patricia G.Articles420