After the Implant What cochlear implants do for children’s language performance is revolutionary. But the device can’t do all the work. It takes concentrated, sustained post-implant intervention to truly realize the miracle. Features
Free
Features  |   March 01, 2015
Courtesy of Advanced Bionics LLC
After the Implant
Author Notes
  • Susan Nittrouer, PhD, is professor and director of research in the Department of Otolaryngology-Head and Neck Surgery at Ohio State University Medical Center. She is an affiliate of ASHA Special Interest Group 9, Hearing and Hearing Disorders in Childhood. Susan.Nittrouer@osumc.edu
    Susan Nittrouer, PhD, is professor and director of research in the Department of Otolaryngology-Head and Neck Surgery at Ohio State University Medical Center. She is an affiliate of ASHA Special Interest Group 9, Hearing and Hearing Disorders in Childhood. Susan.Nittrouer@osumc.edu×
Article Information
Development / Hearing Aids, Cochlear Implants & Assistive Technology / Features
Features   |   March 01, 2015
After the Implant
The ASHA Leader, March 2015, Vol. 20, 44-49. doi:10.1044/leader.FTR2.20032015.44
The ASHA Leader, March 2015, Vol. 20, 44-49. doi:10.1044/leader.FTR2.20032015.44
When I listen to parents recount their experiences surrounding their child’s severe-to-profound hearing loss, they invariably mention the relief they felt when learning about cochlear implants. Up to that point, these parents describe feeling confused and anxious about their child’s future.
Not that many years ago, their concern would have been well-founded. When I started my career as a teacher of students who are deaf at Clarke School in Northampton, Massachusetts, parents relinquished their children to the school at the young age of 3, in hopes that the school’s intensive language environment would help their child develop spoken language. Though often successful, the process was painstaking. And it required children to live away from home for most of their childhood.
Then, in the latter half of the 20th century, two technological advances significantly altered prospects that these children could develop spoken language well enough to participate in mainstream educational settings. First were new tools that recorded biological signals evoked by sensory inputs. This innovation gave rise to newborn hearing screening, which allowed intervention to start earlier than ever before. The second was the cochlear implant, which—for the first time—gave children with severe-to-profound hearing loss access to sensory input through auditory means.
Also bolstering the ability of children who are deaf to acquire spoken language is the arsenal of intervention techniques refined by teachers and clinicians such as Arthur Boothroyd, Helen Beebe, Dan Ling, Jean Moog and Doreen Pollack. Through them, we recognize the power of generative language as the driving force for language acquisition, and know how to apply methods of recasting and expansion to enhance that language. We appreciate that syntax emerges in parallel to lexical knowledge, rather than as an addendum to it.
So, we can now identify children with significant hearing loss soon after birth, give them access to sensory input, and provide state-of-the art early intervention—treatments enabling them to leave early intervention programs with language skills in the “normal” range. But as promising as that term sounds, is it enough? Is there more we could be doing? In short, the answer is yes, absolutely—and here’s why.
Language performance
Children with hearing loss who receive CIs leave early intervention programs with language skills that are, on average, one standard deviation below the skill level of an average child with normal hearing. That means that on standardized test instruments with means of 100, the typical child with CIs achieves a score of 85. It also means that half the children with CIs receive scores better than 85, which is usually defined as “within normal limits.” These outcomes mark great improvement over what was possible before CIs were available, but we must conclude that they do not represent the level of performance that most of these children would achieve in the absence of hearing loss.
Along with an explanation of effect size, the box below shows mean performance at kindergarten for two groups of children participating in an ongoing longitudinal study by our research team: one group with normal hearing and one group born with severe-to-profound hearing loss and implanted with CIs, generally before age 2. These groups were well-matched on demographic variables such as socioeconomic status and nonverbal cognitive function. On every measure, the mean score for children with CIs was close to one standard deviation below that of children with normal hearing. This situation exists in spite of our best efforts.
What’s the problem?
As helpful as CIs are, there is one thing they cannot provide: good frequency resolution. CIs operate by dividing the speech spectrum into a number of channels, each of which signals amplitude changes across time within that specific frequency range. Although there may be close to 20 channels in an implant array, that number is far fewer than in a normal-functioning cochlea, and the spread of excitation along the basilar membrane further reduces the effective number of channels.
The resulting lack of frequency resolution would be expected to have a disproportionately greater effect on phonologic development than on lexical or syntactic knowledge, because it is possible to recover lexical form from acoustic signals that lack detail. Less-detailed lexical representation can still be used to help parse a sentence. However, poor frequency resolution makes it very difficult to recognize the internal phonemic structure of words.
As a result, children with CIs would be expected to have difficulty acquiring phonemic sensitivity—a prediction supported by the data shown in the box below. These data indicate that effect size for the final consonant choice task is much larger than for any of the other language measures. Final consonant choice is the only measure that required sensitivity specifically to phonemic structure. It is especially striking that children with CIs performed more similarly to the children with normal hearing on the syllable-counting task (effect size was much smaller). Syllable structure can be retrieved based on amplitude structure in the acoustic signal, and that structure is well-preserved in CI processing.
Image provided courtesy of Cochlear Americas, © 2014 Cochlear Americas.
What to do
Without question, CIs have improved outcomes for children with severe-to-profound hearing loss. Thanks to this technology, these children have made it to the ballpark of typical language. But we have work to do to help them maximize their potential, and there are a number of ways we can enhance their treatments:
  • Boost signal quality. Given that a primary source of continued deficit resides in the CI’s lack of spectral resolution, we must work to strengthen the signal. One variable that greatly affects outcomes in our longitudinal study is whether children continue using a hearing aid for some time after receiving a first CI. Those who’ve had a period of electric-acoustic (bimodal) stimulation show long-lasting, positive effects. In particular, their sensitivity to phonemic structure is much greater. On the final consonant choice task (see box on page 47), children who wore hearing aids for at least a year after receiving their first CIs got 19 percent of the 48 items correct; children without that period of electric-acoustic stimulation got just 8 percent correct. Neither age of implantation nor pre-implant auditory thresholds could account for this finding.

  • Provide the visual signal. New brain imaging techniques reveal that the central nervous system operates most efficiently when it can integrate information from across sensory modalities. Where speech is concerned, having an audio-visual signal should help children discover word-internal phonemic structure. Children with CIs should always be able to see the talker.

  • Get children talking to improve speech perception. Brain imaging also reveals that speech is primarily processed in the region of the cortex responsible for motor control, highlighting the connection between speech production and perception. Thus, the ability to produce speech should enhance children’s ability to perceive speech, and our longitudinal data support this prediction. Bottom line: Design clinical activities to get children talking.

  • Use whole structures. We see the most positive outcomes when we focus language intervention on structures longer than the word. Most obviously, this approach helps children discover syntactic patterns, especially word order. In addition, focusing on long sequences of language elements helps children discover the recurrent acoustic patterns that define individual words. Always encourage children with CIs to use complete syntactic structures.

  • Provide direct language instruction. Giving students direct grammar instruction has fallen out of favor, except perhaps with second-language learners. This hands-off approach may work for children acquiring a first language naturally, but those with limited auditory input need more direct instruction in grammatical rules. Similarly, children with CIs will likely benefit from concentrated phonological awareness training.

Language performance in children with CIs is, on average, below where it would likely be without hearing loss. Opportunities remain to improve intervention, both during early intervention and through the elementary grades.

We’ve come a long way since my days teaching students who are deaf: Early intervention and CIs have improved prospects that children born with severe-to-profound hearing loss can develop the language skills needed to function in mainstream settings. Nonetheless, language performance in these children is, on average, below where it would likely be without hearing loss. Opportunities remain to improve intervention, both during early intervention and through the elementary grades.
Interpreting the Language Performance Data

In statistical terms, we use effect size to compare performance across groups, and effect size is related to standard deviation. When outcome data from a group generate a normal curve, 34 percent of the data points are between the mean (which is at the 50th percentile) and plus or minus one standard deviation. Consequently, a child who gets a score of one standard deviation below the mean of the group is performing at the 16th percentile. In sample statistics, the difference between means of two groups is described in terms of standard deviations. An effect size of one indicates that the mean for one group is at the 16th percentile of performance for the other group.

 In statistical terms, we use effect size to compare performance across groups, and effect size is related to standard deviation. When outcome data from a group generate a normal curve, 34 percent of the data points are between the mean (which is at the 50th percentile) and plus or minus one standard deviation. Consequently, a child who gets a score of one standard deviation below the mean of the group is performing at the 16th percentile. In sample statistics, the difference between means of two groups is described in terms of standard deviations. An effect size of one indicates that the mean for one group is at the 16th percentile of performance for the other group.
Interpreting the Language Performance Data

In statistical terms, we use effect size to compare performance across groups, and effect size is related to standard deviation. When outcome data from a group generate a normal curve, 34 percent of the data points are between the mean (which is at the 50th percentile) and plus or minus one standard deviation. Consequently, a child who gets a score of one standard deviation below the mean of the group is performing at the 16th percentile. In sample statistics, the difference between means of two groups is described in terms of standard deviations. An effect size of one indicates that the mean for one group is at the 16th percentile of performance for the other group.

×
Interpreting the Language Performance Data

In statistical terms, we use effect size to compare performance across groups, and effect size is related to standard deviation. When outcome data from a group generate a normal curve, 34 percent of the data points are between the mean (which is at the 50th percentile) and plus or minus one standard deviation. Consequently, a child who gets a score of one standard deviation below the mean of the group is performing at the 16th percentile. In sample statistics, the difference between means of two groups is described in terms of standard deviations. An effect size of one indicates that the mean for one group is at the 16th percentile of performance for the other group.

 In statistical terms, we use effect size to compare performance across groups, and effect size is related to standard deviation. When outcome data from a group generate a normal curve, 34 percent of the data points are between the mean (which is at the 50th percentile) and plus or minus one standard deviation. Consequently, a child who gets a score of one standard deviation below the mean of the group is performing at the 16th percentile. In sample statistics, the difference between means of two groups is described in terms of standard deviations. An effect size of one indicates that the mean for one group is at the 16th percentile of performance for the other group.
Interpreting the Language Performance Data

In statistical terms, we use effect size to compare performance across groups, and effect size is related to standard deviation. When outcome data from a group generate a normal curve, 34 percent of the data points are between the mean (which is at the 50th percentile) and plus or minus one standard deviation. Consequently, a child who gets a score of one standard deviation below the mean of the group is performing at the 16th percentile. In sample statistics, the difference between means of two groups is described in terms of standard deviations. An effect size of one indicates that the mean for one group is at the 16th percentile of performance for the other group.

×
0 Comments
Submit a Comment
Submit A Comment
Name
Comment Title
Comment


This feature is available to Subscribers Only
Sign In or Create an Account ×
FROM THIS ISSUE
March 2015
Volume 20, Issue 3