The Acoustic Change Complex in Young Children with Hearing Loss: A Preliminary Study

Abstract

The acoustic change complex (ACC) is a cortical auditory evoked potential elicited in response to a change in an ongoing sound. The ACC may have promise for assessing speech perception in infants and toddlers. In this preliminary study, the ACC was elicited in adults and young children in response to changes in speech stimuli representing vowel height /u/-/a/ and vowel place /u/-/i/ contrasts. The participants were adults with normal hearing (n = 3), children with normal hearing (n = 5), and children with mild to moderately severe bilateral sensorineural hearing loss (n = 5). The children with hearing loss were hearing aid users. The ages ranged from 2 years 3 months to 6 years 3 months for the children and 44 to 55 years for the adults. Robust P1-N1-P2 responses were present for the adults and P1-N2 responses were present for all but the youngest child with hearing loss. The ACC response for the vowel place contrast was less robust than that for the vowel height contrast in one child with substantial hearing loss. The findings from this preliminary study support the conclusion that the ACC can be used successfully to assess auditory resolution in most young children.

Learning Outcomes: As a result of this activity, the participant will be able to (1) describe the utility of the acoustic change complex to assess auditory resolution in children and (2) compare and contrast the acoustic change complex expected in normal hearing adults and children.

Substantial gains have been made in recent years in the early identification and audiological management of young children with hearing loss. Widespread newborn hearing screening followed by more comprehensive assessment and fitting of an auditory sensory device have markedly reduced the adverse consequences of auditory deprivation during the sensitive period for auditory development and language learning. Electrophysiological and specialized behavioral assessment techniques make it possible to measure detection thresholds in infants. Such information is essential for describing the degree and configuration of hearing loss in each ear and for making decisions regarding selection and fitting of a sensory device. Technological advancements in the design of auditory sensory devices, including digital hearing aids and cochlear implants, enable access to many of the acoustic cues of speech that are important for spoken word recognition.

Despite these considerable achievements in early management of hearing loss, clinical measures to assess auditory resolution remain sparse. This is particularly true for children from birth to 3 years of age. The primary issue is the ability of the auditory system to differentiate among the phonemic building blocks of spoken language. For this reason, clinical tools have traditionally used the results of speech-perception tests as indices of resolution. Such measures are needed to help determine whether a child is benefiting from a hearing aid or should be considered for a cochlear implant. At the present time, these decisions tend to be based on unaided and aided detection thresholds, observation, and parent report. Appropriate measures of speech perception in children below 3 years of age would be a valuable additional tool in establishing guidelines for intervention, and for tracking outcomes—at least until the child is of an age to be assessed on more traditional closed- and open-set speech recognition tests.

There is a long history in the development of behavioral speech-perception tests for children. Most are adapted from adult measures of word repetition by modifying vocabulary and response format.¹ In older children, who have had the opportunity for optimal use of hearing in the development of spoken language, performance on appropriate tests of word recognition can reasonably be assumed to reflect underlying auditory resolution. For infants and toddlers with congenital hearing loss, however, the assumption that they have had the opportunity for optimal use of hearing in developing spoken language is untenable. Researchers have, therefore, focused on the detection of a phonemic change in a series of otherwise identical utterances, using developmentally appropriate response tasks such as a reinforced head turn,² conditioned play,³ or imitation.^{4 5} Our previous work on this topic has led to the development of a battery of behavioral speech-pattern contrast tests in which both the stimuli (vowel-consonant-vowel utterances) and the perceptual task remain constant, but the response task is modified according to the age and developmental status of the child.^{6 7}

In one study of speech-pattern contrast perception using the reinforced head turn, we found that infants between the ages of 9 and 21 months, with hearing loss ranging from mild to profound, responded to the vowel height contrast (/u/ versus /a/) with a high degree of confidence.⁸ The vowel place contrast (/u/ versus /i/), by comparison, was found to be sensitive to differing degrees of hearing loss, resulting in a spread of performance. Employing a conditioned-play activity or an imitation task, children with normal hearing demonstrated criterion levels for most vowel and consonant contrasts by ∼36 months of age.^{4 5 9} A spread in outcomes was shown for children with hearing loss, reflecting different degrees of hearing loss and type of sensory device.^{4 9}

Behavioral tests of infants and toddlers are, unfortunately, confounded by numerous developmental and task-related variables. Responses to complex auditory stimuli using behavioral measures not only require contributions from sensory, or subcortical, processes but also from higher-level “global” processes. A consequent limitation to behavioral assessment is that auditory resolution cannot be measured directly but must be inferred from performance. Indeed, between ages 18 and 36 months we have found that such factors as interest, motivation, attention, and compliance often undermine performance on our existing tests. Although positive results can indicate the presence of the necessary capabilities, deficits of performance cannot unequivocally be attributed to deficits in those capabilities.

In an attempt to remove task-related and other developmental confounds when assessing auditory resolution in young children, a promising avenue involves implementation of electrophysiological measures. Cortical auditory evoked potentials in response to speech-type stimuli have been reported in children,¹⁰ even as young as 6 months of age.¹¹ Many such potentials can provide insights into the auditory processing of speech. An excellent description and analysis of these potentials has been provided by Stapells.¹²

Our present interest is in cortical response to a change in an ongoing speech sound. There are many reports of responses to change in both speech and nonspeech sounds.^{13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28} In 1998, Ostroff et al observed an auditory cortical potential in response to the acoustic change in the middle of an ongoing consonant-vowel syllable.²⁹ This response appeared to be a combination of responses to the offset of the consonant segment and the onset of the vowel segment. The morphology of the response and its latency with respect to the acoustic change suggest that it was a P1-N1-P2 potential elicited by the consonant-vowel transition. This potential has been referred to in the literature as the acoustic change complex (ACC).³⁰

The ACC occurs between 50 and 300 milliseconds after the acoustic change and is believed to originate in the auditory cortex. Using synthetic speechlike stimuli, adults with normal hearing produce an ACC in response to: (1) small changes of amplitude (around 2 dB) and of spectrum ³¹; (2) periodicity changes in vowel-like stimuli with the same spectral envelope³⁰; and (3) relatively small shifts of second formant frequency (around 40 Hz) in synthetic steady-state vowels.³² Some authors report that the ACC can be detected without contamination by electrical artifacts in listeners with severe to profound hearing loss using cochlear implants and hearing aids.^{33 34 35} The individual ACC has good test–retest reliability in adults and children.^{25 31 36} The question addressed in this preliminary study was whether a reliable ACC can be obtained in very young children, with and without hearing loss, in response to phonemically significant change. It is already known that an N1-P2 response to sound onset can be demonstrated in young children.^{37 38 39 40} The ACC, however, is a response to change in an ongoing stimulus. We needed to know whether this too could be demonstrated in young children.

Efficiency is an important consideration for any clinical test, especially in a pediatric population. Martin and colleagues have investigated methods of optimizing efficiency when measuring the ACC. To compare different strategies, they defined efficiency as the root mean square (rms) signal-to-noise ratio within the response window divided by total testing time.⁴¹ The ACC was elicited by a change in second formant frequency in an otherwise steady-state synthetic vowel—effectively a shift between /u/ and /i/. The highest efficiency was obtained by alternating between /u/ and /i/ every 500 milliseconds and combining responses to the two directions of change. This was the technique used in the present study.

In this preliminary study, we looked for the ACC in response to changes in speech stimuli representing vowel height and vowel place contrasts. We were particularly interested in the vowel place contrast, however, because of its greater reliance on resolution in the region of the second formant. Vowel and consonant place contrast are highly susceptible to the effects of sensorineural hearing loss.^{8 42}

Methods

Subjects

Event-related potentials were recorded from 13 participants in three subject samples: adults with normal hearing (n = 3), children with normal hearing (n = 5), and children with mild to moderately severe bilateral sensorineural hearing loss (n = 5). Evidence of normal hearing (<25-dB hearing level) was documented by audiometric screening from 250 to 8000 Hz. The mean age of the adults with normal hearing was 50 years, ranging from 44 to 59 years. The mean age of the children with normal hearing was 3 years 2 months, ranging from 2 years 8 months to 4 years 8 months. The mean age of the children with bilateral hearing loss was 4 years 9 months, ranging from 2 years 3 months to 6 years 3 months. All five children with bilateral sensorineural hearing loss were bilateral hearing aid users. Individual characteristics for these five children are shown in Table 1.

Table 1. Demographics for the Five Children with Hearing Loss.

				Thresholds (dB Hearing Level)
	Age at test (y, mo)	Age at Hearing Loss ID (mo)	Age Hearing Aids Fit (mo)	Ear	250 Hz	500 Hz	1000 Hz	2000 Hz	4000 Hz	8000 Hz	Etiology
HI1	2, 3	Birth	6	R	30	35	45	45	45	35	Connexin
				L	30	35	40	45	40	40
HI2	4, 10	Birth	6	R	20	40	70	85	100	NR	Ototoxic drugs
				L	20	40	65	70	80	70
HI3	5, 1	6	6	R	DNT	50	65	65	65	DNT	Connexin
				L	85	90	95	90	85	80
HI4	5, 5	Birth	1	R	50	55	70	85	85	DNT	Ototoxic drugs (likely)
				L	60	65	70	90	90	DNT
HI5	6, 3	36	36	R	30	40	55	60	50	35	Connexin
				L	35	40	50	55	60	40

DNT, did not test; L, left; NR, no response; R, right.

Stimulus

Stimuli consisted of quasi-synthetic vowels created by iterating single cycles taken from tokens of /u/, /a/, and /i/ spoken by the talker used for our earlier behavioral studies of contrast perception. Fundamental frequency was maintained at a constant 230 Hz and amplitudes were adjusted to the same rms value. Onsets and offsets of the vowels were tapered over 40 milliseconds using raised cosine functions. Vowels were then concatenated with a 40-millisecond overlap. The stimulus alternated between two vowels (/u/ and /a/ for the vowel height contrast and /u/ and /i/ for the vowel place contrast). The time between alternations was 500 milliseconds, with no silent period. Sampling triggers were placed at the midpoints of the transitions. Although we examined both the vowel height and vowel place (fronting) contrasts, our main interest was in vowel place, which taps into an acoustic cue important for higher-frequency spectral processing.

EEG Recordings

The ACCs were measured using a Neuroscan system and a 32-channel Synamp amplifier (NeuroScan Compumedics System, El Paso, TX). Surface electrodes were placed according to the International 10–20 system.⁴³ The active electrode was placed on the Cz site and the reference electrode was placed on the mastoid. The electrode placed on the forehead served as the ground. Vertical eye movements and eye blinks were monitored via electrodes placed immediately above and below the right eye. With the exception of the two youngest children, all electrode impedances were below 5000 Ohms.

During data acquisition, the continuous electroencephalogram (EEG) was digitized (sampling rate = 1000 Hz), amplified (gain = 1000), and band-pass filtered (0.15 to 100 Hz). An eye blink reduction algorithm was applied to the continuous EEG.⁴⁴ Epochs were extracted from the continuous EEG, starting 100 milliseconds before a change and ending 400 milliseconds after the reverse change. Epochs in which voltage changes exceed 100 microvolts were rejected from further analysis. Each epoch was baseline corrected and the data averaged by individual subject, stimulus, and group. A single recording consisted of ∼350 to 500 epochs.

Procedure

Testing was performed in an Industrial Acoustic Company (IAC) double-walled sound-attenuating booth. Children were seated in a high chair watching a video with the sound turned off. The parent and a test assistant were in the booth with the child in most instances. Adults were seated in a regular chair. Stimuli were routed from the Neuroscan through the sound booth to a Grason Stadler loudspeaker (Grason Stadler, Eden Prarie, MN), which was stationed at 45-degree azimuth from the subject. The stimulus level was 65 decibels, A-weighted (dBA). The children with hearing loss were tested with their hearing aids activated and set to everyday use. Also included were trials for two children with their hearing aids turned off.

Results

Displayed in Fig. 1 are the group mean waveforms in response to changes in vowel place (the /u/-/i/ contrast) for the three subject samples. The group means confirm that the P1-N1-P2 response from the adults and the P1-N2 response from the two groups of children (normal hearing and hearing loss) are robust and consistent with the expected waveforms and latencies for these age groups.

Figure 1.

Group mean acoustic change complex (ACC) responses for /u/-/i/ (vowel place contrast). P1-N1-P2 responses for three adults with normal hearing are on top. Below are group mean ACC responses (P1-N2) for five children with normal hearing (NH) and five children with hearing loss (HI). Hearing aids were activated during data collection.

Group mean and individual waveforms for the five children with hearing loss are shown in Fig. 2 in response to the changes of vowel height (the /u/-/a/ contrast). Clearly observable P1 responses at around 114 milliseconds followed by distinct negativity (N2) at ∼230 milliseconds were present for all but the youngest child (subject HI1, 27 months of age). Electrode impedances for this child were difficult to maintain below 5000 Ohms due to difficulty applying the surface electrodes, which may have accounted for the poor response.

Figure 2.

Individual acoustic change complex (P1-N2) and group mean (thick black line) responses for /u/-/a/ (vowel height contrast) for five children with hearing loss. Hearing aids were activated during data collection.

Group mean and individual results for the vowel place contrast are shown in Fig. 3 for the five children with the normal hearing (left panel) and the five children with hearing loss with hearing aids activated (right panel). Robust P1-N2 responses were present for each of the children with normal hearing, although the responses were noisy for the two youngest children (subjects NH1 and NH2). For the children with hearing loss, P1-N2 responses were evident for all but the youngest child (subject HI1) with high impedances. Absence of a response is noteworthy because subject HI1 had a mild to moderate hearing loss and the vowel place contrast should have been accessible with hearing aids activated. The response for subject HI3 was much less robust than that of subjects HI2, HI4, and HI5.

Figure 3.

Group mean and individual responses for /u/-/i/ (vowel place contrast) for five children with normal hearing (left panel) and five children with hearing loss (right panel).

Figure 4 displays the current unaided audiograms and the ACC responses for subjects HI3 and HI4, assessed in two conditions: hearing aids activated (vowel height and place) and hearing aids turned off (vowel place only). The ACC responses with the hearing aid off condition were markedly reduced (subject HI4) or absent (subject HI3). In the hearing aid–activated condition, subject HI4 produced robust responses to both vowel height and vowel place contrasts. Subject HI3, in contrast, produced a strong P1 response for vowel height and a much a weaker P1 response to vowel place. Note, also, that the N2 component for subject HI3 showed prolonged negativity for both vowel contrasts.

Figure 4.

Acoustic change complex (ACC) responses and unaided audiograms for two children with hearing loss. The ACC responses are shown for the vowel height contrast (/u/-/a/, thin line) and the vowel place contrast (/u/-/i/, thick line) with hearing aids activated. The black dotted line is the response with the hearing aids turned off for the vowel place contrast (/u/-/i/). HL, hearing level.

Discussion

These data support the conclusion that the ACC can be measured reliably in young children with normal hearing and children with hearing loss, at least down to the age of 2 years. It remains to be seen whether this 2-year age boundary can be lowered further. The three vowels used in this study represent points on the vowel triangle and, therefore, involve large acoustic and phonemic changes. These findings demonstrate that the ACC can be elicited by spectral changes exemplifying contrasts of vowel height and vowel place. Future studies will need to assess finer vowel changes. It also may be informative to examine the ACC in response to voicing and place-of-articulation changes in sustained consonants.

At this stage, the possibility of contribution from loudness change cannot be excluded. The vowels were matched for equal rms amplitude but this does not guarantee equal loudness. Listener thresholds, loudness growth functions, and hearing aid gain curves would all contribute to differences of perceived loudness for the three vowels. Further work will be needed to separate the effects of spectral and amplitude change.

For the three adults with normal hearing, the latencies for P1-N1-P2 waveforms were consistent with results from other studies.^{25 29 30 31 41} The waveform latencies for the two groups of children also were similar to the late cortical potentials following stimulus onset reported in the literature for young children.^{37 38 39 40 45 46 47 48}

One of the two children with hearing loss (subject HI3) is of particular interest because of the weak P1 response with prolonged negativity on the vowel place contrast. Regrettably, we were unable to confirm whether the vowel place response was present or absent on behavioral testing because the child was undergoing cochlear implant surgery relatively soon after the ACC was recorded. Nevertheless, the somewhat poor morphology to the ACC for vowel place might suggest poor access to second formant acoustic information with hearing aids.

The ultimate goal of this project is to determine whether the ACC will be a valid means of assessing auditory resolution in children 18 to 36 months of age. This was the age range in which assessment on behavioral tests was found to be particularly challenging because of nonauditory factors that influence outcomes. In this study, it is encouraging that three children with normal hearing (subjects NH1, NH2, NH3) and one child with hearing loss (subject HI1) were between 25 and 31 months of age. Subject NH2, however, displayed a noisy response and the response for subject HI1 was absent, despite this child having only a mild to moderate hearing loss. The findings for subjects NH2 and HI1 highlight the challenges involved in assessing this age group, regardless of hearing loss. Measurement of the ACC requires the child to be alert, quiet, and sit still long enough to place the electrodes, measure impedances, and deliver the stimuli. It remains to be seen whether the rather poor waveforms obtained from these two cases reflect their young age or the relative lack of investigator experience in conducting electrophysiological tests in young children.

This final point also highlights some of the caveats and concerns that were recently highlighted in a special issue on hearing aids and the brain.⁴⁹ The absence of the traditional P1-N1-P2 response, alone or in the ACC condition, may result from the signal processing alterations introduced by the hearing aid that are still poorly understood. Such examples include the confounds of altered signal-to-noise ratio,^{50 51} alterations to the rise time of the stimuli,⁵² and the need for optimal stimulus and recording paradigms to ensure the recording montages and stimulus rates used for different age groups are defined.⁵³

Conclusion

The findings of this preliminary study do support the notion that the ACC can be used to assess auditory resolution in most young children, at least at the level of broad vowel contrasts. Use of the ACC, compared with traditional P1-N1-P2 testing evoked by short duration stimuli, enables the clinician scientist to quantify not only the neural detection of stimulus onset but also discriminative responses as well. Furthermore, these findings verify that the presence of hearing loss and hearing aids does not represent a barrier; however, the contribution of hearing aid signal processing on evoked neural activity is yet to be fully defined. Thus, although the ACC may one day help guide decisions about sensory and habilitative management in very young children with hearing loss, there are still unresolved issues that are being explored before the ACC can be implemented into clinical practice.