Although prevalence estimates vary with the definition of hearing loss, about one in three adults age 65 and older will have a hearing loss sufficient to impair everyday communication. For centuries, the loss of hearing sensitivity with advancing age has been acknowledged—referred to in the writings, for example, of Hippocrates and Shakespeare. The first clinical descriptions of this phenomenon and the term "presbycusis," however, emerged in the late 1800s (Schacht & Hawkins, 2005).
Today, the progressive loss of hearing sensitivity with advancing age is so well established that an international standard now describes the median and statistical distribution of pure-tone hearing thresholds as a function of age in both men and women (ISO, 2000). Median values calculated from this standard for men are shown in Figure 1 [PDF] for 20-, 40-, 60-, and 80-year-olds. Although systematic differences appear in hearing thresholds between men and women for each age group, the pattern of progressive loss of primarily high-frequency hearing sensitivity illustrated in Figure 1 [PDF] is the same for both genders. Even though the bulk of data from which our knowledge of age-related pure-tone hearing loss was gleaned came from cross-sectional studies of separate groups of individuals differing in age, the general trends were confirmed by a smaller number of longitudinal studies. These longitudinal studies indicate that the rate of hearing loss progression in older adults is about 1 dB per year, slightly faster in the higher frequencies and a little slower at the lower frequencies (e.g., Pearson et al., 1995; Lee et al., 2005).
It is seldom the case, however, that the older patient shows up at the hearing clinic complaining of an inability to hear pure tones. Rather, the chief complaint of older adults is that they can hear speech but can't understand it. This condition is referred to as the "speech understanding" problems of older adults. "Speech understanding" refers to deficits in the closed-set identification or open-set recognition of speech stimuli (typically nonsense syllables, words, or sentences). The term is not intended to include phenomena such as the discrimination of minimally contrasting speech sounds (e.g., "ba" vs. "da") or the comprehension of the meaning of speech stimuli (as in replies to queries such as "What day is it today?").
In 1988 a working group of the Committee on Hearing and Bioacoustics and Biomechanics (CHABA) of the National Research Council published a report on the speech-understanding problems of older adults (CHABA, 1988). The report used a "site of lesion" framework to synthesize the research literature on this topic, with the potential sites of lesion identified as peripheral, central-auditory, or cognitive. Of course, as was noted in the CHABA report, it was possible that more than one site of lesion could underlie the speech-understanding problems of older adults—collectively in describing the group trends and individually in accounting for the problems of a given older adult. This framework has proven useful as a means to organize literature reviews in the past (e.g., Humes, 1996; Humes & Dubno, 2008) and will be employed in this brief overview.
The Peripheral Hypothesis
Although age differences in the outer ear and middle ear of young and old adults have been documented, the consequences of these structural and functional changes on the speech-understanding problems of older adults are considered minimal. As a result, the peripheral hypothesis has focused on age differences in the structure and function of the cochlea; in particular, the impact of the high-frequency sensorineural hearing loss (Figure 1 [PDF]) resulting from pathology in the base of the cochlea. The impact of this peripheral hearing loss on the audibility of various speech sounds is depicted schematically in Figure 2 [PDF] for speech presented at typical conversational levels (60-65 dB SPL).
Clearly, this hearing loss renders many speech sounds—represented simplistically in Figure 2 [PDF] at single points along the sound-level (dB HL) and frequency (Hz) coordinates of the audiogram—inaudible to the older adults. For the most part, high-frequency hearing loss makes it difficult for older adults to hear high-frequency, low-intensity speech sounds, many of which are information-bearing consonants. Thus, words such as "sun," "fun," and "ton" all may be heard as "un," leading to the complaint that speech can be heard, but not understood, by the older adult.
Figure 2 [PDF] illustrates what the CHABA report described as the "simple" form of the peripheral hypothesis—one based primarily on the inaudibility of certain speech sounds. The report also describes the "multiple-distortion" version of the peripheral hypothesis, which attributes the speech-understanding problems of older adults to the effects of inaudibility arising from cochlear pathology, and also identifies the potential negative impact on speech understanding of other aspects of cochlear pathology such as abnormal spectral resolution (i.e., a diminished ability to encode the frequency content of speech sounds).
Beyond the Auditory Periphery
In addition to these two versions of the peripheral hypothesis, the CHABA report identified two potential contributing mechanisms that go beyond the auditory periphery. The first, the central-auditory hypothesis, theorizes age-related
deterioration in the ascending pathways of the auditory portions of the central nervous system—from the cochlear nucleus through the primary auditory cortex. The second, the cognitive hypothesis, conjectures age-related declines in amodal processes at the cortical level, such as information storage and retrieval (memory) and attention.
Although both of these hypotheses are "central" or higher-level than the peripheral hypothesis, they differ from each other in at least two important ways. First, the central-auditory hypothesis includes brainstem and cortical structures, whereas the cognitive hypothesis is operational at the cortical level only. Second, the central-auditory hypothesis is modality-specific and plays a role only for the auditory presentation of speech stimuli, whereas the cognitive hypothesis affects mechanisms such as memory, attention, and linguistic knowledge that are not specific to auditory input and may involve many regions of the brain.
Researchers have pursued a variety of approaches in attempting to sort out the factors underlying the speech-understanding problems of older adults. One common approach has been to make between-group comparisons of speech-understanding performance. Probably the most powerful research design within this general approach has been a 2 x 2 factorial design, in which two levels of age (young and old) are combined with two levels of hearing status (normal and impaired).
This design requires a total of four groups of participants:
- Younger adults with normal hearing
- Younger adults with hearing impairment
- Older adults with normal hearing
- Older adults with hearing impairment
In such research designs, it is important that the ages of the two younger groups (typically 18-30 years of age) are equivalent and the ages of the two older groups (typically 60-80 years of age) are also equivalent. This goal is often fairly easy to achieve. However, it is also important that the two groups with hearing impairment have equivalent hearing loss and the two with normal hearing have identical (and normal) hearing. This goal is much more challenging to achieve, especially to find a sufficient number of older adults with normal hearing who have hearing thresholds identical to those of younger adults with normal hearing.
Probably the research group that has pursued this 2 x 2 factorial group design most to explore the speech-understanding problems of older adults has been Sandra Gordon-Salant and Peter Fitzgibbons at the University of Maryland (see review by Gordon-Salant, 2005). Figure 3 [PDF] depicts two patterns of hypothetical results obtained by this research group, approximating the findings across several research studies since the early 1990s. The illustration in the top panel of Figure 3 [PDF] shows a pattern in the group data that reveals a main effect of hearing status on speech-understanding performance, but no effect of age and no interaction with age.
Notice, for instance, that both groups with impaired hearing—young and old alike—perform the same, and both perform worse than the two groups with normal hearing. Most often this pattern of results has been observed for speech stimuli presented in quiet or in steady-state noise and with speech levels ranging from slightly less (e.g., 55 dB SPL) to considerably greater (e.g., 90 dB SPL) than conversational-speech levels. Because hearing status, but not age, had a significant effect, this pattern of findings generally supports the peripheral hypothesis.
The hypothetical group data in the lower panel of Figure 3 [PDF] reveal a second pattern of observations reported frequently by Gordon-Salant and Fitzgibbons. In this case, mean performance varies across groups, with the young adults with normal hearing performing the best and the older adults with impaired hearing performing the worst. The mean results for the other two groups—young adults with impaired hearing and older adults with normal hearing—lie somewhere in between. This pattern generally reflects significant main effects of hearing status and aging, and may also represent the presence of an interaction between these two variables, depending on the details of the results from a particular study.
The pattern of results depicted in the lower panel of Figure 3 [PDF] has been seen most often by Gordon-Salant and Fitzgibbons (and other research groups) when one of three conditions is present:
- The speech stimulus is temporally accelerated through time compression or is temporally interrupted
- A competing speech stimulus, or a competing noise with speech-like fluctuations in amplitude, is introduced (instead of competing steady-state noise)
- The task complexity is increased in various ways
This pattern of results suggests that an age-related variable other than peripheral hearing loss is needed to account for these results. This variable could be age-related decline in central-auditory or cognitive processing, although the source of these difficulties has seldom been the point of emphasis in these studies.
Another common approach used by researchers studying the speech-understanding problems of older adults focuses on individual data (e.g., van Rooij, Plomp & Orlebeke, 1989; Jerger et al., 1989, 1991; Dubno & Dirks, 1993; Humes et al., 1994; Humes, 2007; George et al., 2007). In most of these studies, some type of correlational or regression analysis is performed between each of several potential predictor variables and one or more measures of speech-understanding performance.
Examples of hypothetical results using this approach, representative of the general trends observed across several studies, are provided in Figure 4 [PDF] as a series of four scatterplots. In each scatterplot, the y-axis represents a measure of speech-understanding performance, such as the percentage of words correctly recognized, and the data points are the results for each of the listeners in the study. The x-axis for the two scatterplots on the left is the average amount of high-frequency hearing loss; for the two scatterplots on the right, the x-axis is age. The top panels show results that are representative of many studies for speech-understanding measures obtained in quiet or in a background of steady-state noise.
Notice, in this case, that speech-understanding performance declines steadily with increasing amounts of high-frequency hearing loss (top left panel), but that age alone has no systematic effect on the data (top right panel). This pattern of findings would be consistent with the peripheral hypothesis. The pattern of data is reversed in the lower two panels of Figure 4 [PDF], which represent listening conditions involving the understanding of temporally accelerated or interrupted speech, or speech with competing speech in the background. This pattern of results would be consistent with an age-related decline of some type, other than peripheral hearing loss. Whether it represents a central-auditory or cognitive decline, however, cannot be ascertained without the measurement of additional variables.
"Central-Auditory" Processing Deficits
What measures are needed to substantiate a central-auditory or cognitive component as the age-related factor underlying the speech-understanding problems of older adults? It has been challenging to develop reliable and valid measures for central-auditory processing that can be used with older adults (Humes, 2008). Two main difficulties
surround the validity of the central-auditory measures. First, most of the tests historically used with older adults to document central-auditory deficits involve speech stimuli. We have already noted the strong influence of high-frequency hearing loss on many measures of speech-understanding performance in older adults. Because many older adults have hearing impairment—especially those who show up at the clinic—the presence of peripheral hearing loss will likely confound measures of central-auditory processing based on speech understanding. Secondly, the pioneering work of Jim Jerger and colleagues (e.g., Jerger et al., 1989, 1991) observed that as many as 50% of the older adults diagnosed with central-auditory processing problems also had cognitive-processing deficits. This additional factor also has a negative impact on the validity of measures of central-auditory processing obtained from younger adults.
It is not surprising that considerable overlap can exist between central-auditory and cognitive processing deficits in older adults. Consider, for example, one common "central auditory" measure used frequently with older adults—dichotic speech, which involves the concurrent presentation of different speech materials (nonsense syllables, words, digits, or sentences) to the right and left ears. For example, for a word-based test, the listener may be presented with "racehorse" in the left ear simultaneous with the presentation of "streetcar" to the right ear. (There are many variations of this basic dichotic task, including ones in which only portions of each word, such as "horse" in "racehorse" and "street" in "streetcar" overlap in time.)
A central-auditory component is clearly present in this task, with concurrent speech stimuli presented to opposite ears competing for representation in the same higher centers of the auditory central nervous system. The task also includes, however, an attention component. Depending on the specific details of the task—whether, for example, the listener is asked to report the word presented to the left or right ear, or the words to both ears, and whether the listener is given this information before or after stimulus presentation—various types of attention resources may be required to perform the task.
Imagine that the listener in this example is prompted to report the word presented to the left ear after the stimuli have been presented. To get the correct response ("racehorse"), the listener must attend to the input at both ears and hold this input in memory until prompted with "left ear," then respond with "racehorse." Clearly, cognitive processes, such as divided attention and memory (both storage and retrieval), can affect performance. These same cognitive factors very possibly would limit performance for analogous dichoptic visual text presentations of these same stimuli. Further, the relative contributions of cognitive processing may be even greater when the speech stimuli have been degraded at the periphery by the presence of cochlear pathology. Again, this potential confound between
central-auditory and cognitive processing compromises the validity of such measures of "central-auditory" processing in older adults.
The cognitive hypothesis for the speech-understanding problems of older adults has been studied much less frequently by audiologists, and we are still learning about the relative contributions of cognitive factors. In fact, two distinct groups of researchers have been interested in the speech-understanding problems of older adults. One group, comprising mainly psychologists and cognitive scientists, has viewed the speech-understanding problems of older adults primarily as a "top-down" processing problem—one driven largely by age-related changes in cognitive processes such as attention and memory (Schneider & Pichora-Fuller, 2000). Research has sought to identify older adults' specific cognitive-processing problems that underlie their speech-understanding problems, often ignoring the contributions of peripheral or potential central-auditory factors.
The second group, comprising mainly audiologists and hearing scientists, has viewed the speech-understanding problems of older adults as primarily a "bottom-up" processing problem—one determined largely by cochlear pathology present in many older adults (Schneider & Pichora-Fuller, 2000). In this case, research has been directed at measures of hearing sensitivity and auditory processing with little attention paid to the potential influence of cognitive factors on speech-understanding performance.
If further progress is to occur in comprehending the speech-understanding problems of older adults, interdisciplinary collaboration will be critical. This point was emphasized in an excellent review chapter by Schneider and Pichora-Fuller (2000). To facilitate such collaboration and to bridge the gap between researchers working on the same problem from these two perspectives, the first International and Interdisciplinary Research Conference on Aging and Speech Communication was held at Indiana University in October 2005. The second such conference was held in October 2007 and a third is planned for October 2009. At the first two conferences, each research group gained a better appreciation of the other group's perspective and ways in which their research could be modified to accommodate that perspective.
As research in this area continues to develop, it is hoped that the relative contributions of peripheral, central-auditory, and cognitive factors to the speech-understanding problems of older adults will be better identified. Once these factors are identified, and reliable and valid clinical tests of each factor developed, it should be possible to offer better treatment for the speech-understanding problems of older adults. For example, the prognosis for benefit from hearing aids may be much better for an older adult whose speech-understanding problems are primarily attributable to peripheral factors than the prognosis for another individual whose difficulties are attributable largely to central-auditory or cognitive factors. In the latter case, interventions that go beyond delivery of amplification, such as counseling or training programs, may prove to be helpful.