Download Can Age-Related Decline in Speech Understanding Be Explained

J Am Acad Audiol 3: 33-38 (1992)
Can Age-Related Decline in Speech
Understanding Be Explained by Peripheral
Hearing Loss?
James Jerger*
Abstract
Speech audiometric scores were compared across the age range from 50 to 90 years in 137
subjects selected in such a way that average audiometric thresholds were matched across
four age groups . Thus any age-related changes in speech audiometric scores could not be
attributed to age-related differences in peripheral hearing sensitivity. Four speech audiometric measures were studied ; phonemically-balanced words (PB), Speech Perception in Noise
(SPIN) test for both high- and low-predictability sentences, and the Synthetic Sentence
Identification (SSI) test . There was an age trend for all four speech measures but only the
change in SSI (30%) was statistically significant . The argument that the change in SSI can
be explained by subtle changes in the auditory periphery, not reflected in audiometric
thresholds, is weakened by the fact that the change in SSI was greater than the change in
either of the two monosyllabic word tests (PB and SPIN-low) . The argument that the change
in SSI can be explained by concommitant cognitive decline is not supported by correlations
among SSI performance and any of several neuropsychological measures of cognitive
function in the same subjects . Finally, the lack of a significant interactive effect between
hearing sensitivity level and cognitive status does not support a model in which a peripherally
degraded speech signal interacts with a deficit in cognitive function to produce the decline
in speech audiometric scores . We conclude that the observed age-related decline in SSI
performance cannot be satisfactorily explained by peripheral hearing sensitivity loss .
Key Words: Aging, elderly, hearing loss, speech audiometry
he effect of aging on speech audiometric
scores is currently a controversial topic.
T Previous studies showing apparently
age-related declines in performance on various
speech audiometric tests (e .g., Pestalozza and
Shore, 1955 ; Balas,1962 ; Punch and McConnell,
1969 ; Carhart and Nicholls, 1971 ; Smith and
Prather, 1971 ; Rintelmann and Schumaier,
1974 ; Bergman et al, 1976 ; Bess and Townsend,
1977 ; Jerger and Hayes, 1977 ; Bergman, 1980 ;
Arnst, 1982 ; Otto and McCandless, 1982 ;
Duquesnoy, 1983a, 1983b; Dubno et al, 1984 ;
Wingfield et al, 1985 ; Martini et al, 1988) have
been challenged on the basis that degree of
hearing sensitivity loss has not been sufficiently
controlled (CHABA Working Group on Speech
'Division of Audiology, Baylor College of Medicine,
Houston, Texas
Reprint requests ; JamesJerger, 11922Taylorcrest Rd,
Houston, TX 77024
Understanding and Aging [WGSUAj, 1988 ;
Willott, in press) . Since high-frequency sensitivity is known to decline systematically with
age, and since high-frequency sensitivity impacts many speech audiometric measures, it
can be argued that the systematic decline in
speech audiometric scores with age results from
age-related decline in high-frequency hearing
sensitivity . The WGSUA report notes that, although many studies have attempted to control
for degree of audiometric loss, such studies
have typically been flawed in at least three
ways . Firstly, in some studies the comparison
has been made between young subjects and
elderly subjects whose audiometric thresholds
were "within normal limits ." But if "normal
limits" is defined by a screen at 20 dB HL, for
example, it is possible that young adults, often
in the 0 to -10 dB HL range, will have significantly better average hearing levels than elderly subjects, especially in the important high-
Journal of the American Academy of Audiology/Volume 3, Number 1, January 1992
average threshold hearing level at a particular
test frequency (250, 500, 1000, 2000, 3000, or
4000 Hz) did not differ by more than 6 dB across
the four age groups . The relatively smaller
numbers of subjects in the 50 to 65 year group
and in the 76 to 90 year group reflects the
difficulty in finding younger persons with substantial loss and very elderly persons without
substantial loss .
Table 1 summarizes means and standard
deviations of the audiometric thresholds for
right ears across the four age groups . The same
data are presented graphically in Figure 1. All
four age groups showed the presbyacusic audiometric pattern typically associated with aging.
No one of the four age groups showed a consistently poorer pattern of sensitivity loss than any
other age group. Across all four groups the
average pure-tone threshold hearing level (averaged across 500, 1000, and 2000 Hz) of each
individual subject ranged from 5 to 63 dB .
frequency region around 4000 Hz . Secondly,
even though young and elderly subjects are
matched for degree of hearing loss, it may not
be appropriate to compare what may be "apples" and "oranges" in terms of the pathologic
conditions underlying the impaired hearing of
young and elderly persons . Thirdly, if subjects
are selected over a wide range of ages and
degrees of hearing loss, then the effects of age,
hearing loss, and sometimes other relevant
variables are confounded .
In the present paper we have attempted to
re-examine this question by containing peripheral hearing sensitivity within relatively rigidly defined limits as we studied the effect of
age, in the range from 50 to 90 years, on four
speech audiometric measures .
METHOD
Subjects
We evaluated speech audiometric scores
from a subset of 137 elderly subjects, selected
from an original database of 200 elderly subjects with extensive speech audiometric results
(Jerger et al, 1989). All subjects in the parent
database were ambulatory, in the age range
from 50 to 90 years, and in good general health .
Each had volunteered to participate in a study
of hearing and aging. None were clients of our
hospital's audiology service and few expressed
auditory complaints .
The subset of 137 subjects was selected on
the basis of the pure-tone audiometric levels of
the right ear . The objective of this selection
procedure was to identify a group of subjects, in
each of four age ranges, with equivalent frequency-by-frequency average pure-tone hearing threshold levels . The four age groups were
(1) 50 to 65 years, n = 11 ; (2) 66 to 70 years, n =
56 ; (3) 71 to 75 years, n = 54 ; and (4) 76 to 90
years, n =16. There were 80 women and 57 men
in the subset . The 137 subjects were selected
from the larger pool in such a way that the
Table 1
50-65
66-70
71-75
76-90
Details of test selection and administration are available in a previous publication
80
60
40
20
250
250
(10)
(10)
(12)
(9)
(10)
(11)
(10)
(9)
9'
0
Figure 1 Means and standard deviations of audiometric thresholds in four groups of elderly subjects .
500
20
19
21
22
0
CD
4000
1000
Hz
Frequency in
Frequency in Hz
20
19
20
23
U)
w
a
C
m
Means and Standard Deviations`of Audiometric Thresholds,
in dB HL, in Four Age Groups
Age Group
(years)
Procedure
2k
1k
18
22
21
24
(11)
(13)
(11)
(10)
25
31
30
30
4k
3k
(12)
(18)
(15)
(12)
37
42
40
38
(20)
(24)
(18)
(13)
44
45
48
45
(30)
(26)
(18)
(16)
'The standard deviations are in parentheses .
34
414 , ;; 1
Age-Related Decline in Speech Understanding/Jerger
Table 2
Means and Standard Deviations`of Speech Recognition Scores, in Percent Correct,
in Four Age Groups Matched for Audiometric Level
Speech Audiometric Measure
Age Group
(years)
PB
SPIN-low
50-65
66-70
92 .0 (13 .0)
82 .6 (23 .6)
60 .7 (19 .5)
49 .5 (26 .0)
76-90
79 .4 (19 .3)
47 .8 (21 .1)
71-75
81 .5
(26.1)
47 .0
(25.2)
SPIN-high
SS/
97 .2 (6 .1)
88 .7 (17 .2)
84 .8 (23 .0)
94 .5
(6 .9)
80 .5 (21 .9)
72 .2 (26 .1)
88 .5
(10.2)
64 .4
(21 .3)
'The standard deviations are in parentheses .
(Jerger et al, 1989). The present analysis is
based on the results of four speech audiometric
test procedures .
1.
2.
3.
4.
Phonemically balanced word test (PB) .
Synthetic sentence identification test (SSI) .
Speech perception in noise test-high context (SPIN-high).
Speech perception in noise test-low context (SPIN-low) .
All tests, PB, SSI, SPIN-high, and SPINlow, were presented monotically at a sensation
level of 50 dB relative to the subject's threshold
for multi-talker babble . For PB there was no
simultaneous competing message. For SSI the
competing message was the continuous discourse of a single talker, presented at a message-to-competition ratio of 0 dB . For both modes
of the SPIN test the simultaneous competition
was multi-talker babble presented at a signalto-babble ratio of +8 dB .
Throughout the present analysis statistical significance was evaluated at the relatively
stringent alpha level of 0.01. We thought this
significance level appropriate in order to minimize the likelihood that the relatively parsimonious null hypothesis (i .e ., that hearing sensitivity level is a sufficient explanation for agerelated decline in speech understanding) would
be incorrectly rejected in this experiment .
systematically, from a high of 94 percent, in the
youngest age group (50-65 years), to a low of 64
percent in the oldest age group (76-90 years), a
decline of 30 percent. The other three speech
measures also showed trends toward decreased
performance with increasing age, but in none of
these three measures did the decline exceed 15
percent. The change with age was 13 percent for
both PB and SPIN-low, and 9 percent for SPINhigh .
In order to evaluate the statistical significance of these trends we first transformed all
percent correct scores to their arc sine equivalents, in order to correct for skewness, then
submitted the transformed data to a two-factor,
split-plot analysis of variance . There was one
between-subjects factor (age group) and one
within-subjects factor (speech measure) . This
initial multivariate analysis showed a significant overall age-group effect (p = .0066) and a
significant interaction between age group and
speech measure (p = .0034) . We carried out,
therefore, individual one-factor (age group)
analyses of variance separately for each of the
RESULTS
F
or each speech audiometric measure in
each age group we computed the mean and
standard deviation of the performance score
obtained on the right ear. These values are
presented in Table 2 . In addition, the means
are presented graphically in Figure 2 . All four
of the speech measures examined showed some
decline with age, but the synthetic sentence
identification test (SSI) showed the largest
change . Average SSI performance decreased
Figure 2 Means and standard errors of four speech
recognition scores in four groups of elderly subjects .
35
Journal of the American Academy of Audiology/Volume 3, Number 1, January 1992
four speech measures . These analyses revealed
that the age trend was significant only for SST
(p = .0001) . For the other three speech measures, PB, SPIN-low, and SPIN-high, probabilities of alpha error exceeded 0 .01 . Thus the
significant interaction between age group and
speech measure, noted in the two-factor analysis, could be accounted for by the fact that SST
showed a significantly greater age effect than
either of the other three speech measures . Further post-hoc analyses were then carried out,
comparing all pairs of means within a given
test . Significance was evaluated at an alpha
error level of 0.01, based on Scheffe's S procedure . For SSI this post-hoc analysis revealed
that, of the six possible pairs ofmeans, two were
significantly different and four were not . Significant differences were noted when the 70 to
90 year age group was compared with the 66 to
70 year age group (p = .0051) and with the 50 to
65 year age group (p = .0003). In the case of PB,
SPIN-low, and SPIN-high, there were no significant differences among any of the six pairs
of means .
DISCUSSION
T
he present results clearly show an agerelated change in performance on the synthetic sentence identification (SSI) test . The
average SSI score in this sample of elderly
subjects declined by 30 percent over the age
range from 50 to 90 years. Can this decline be
explained by age-related changes in hearing
sensitivity? The present results argue against
such an explanation.
Hearing sensitivity was controlled in the
present sample, to the extent that such control
is possible, by equating distributions of audiometric threshold levels for pure tones, including the important frequencies of 3000 and 4000
Hz (Kryter, 1983, 1988), across age groups . It
can be argued, however, that aging may be
accompanied by subtle changes in cochlear function not manifest in the pure-tone threshold
levels . One might hypothesize, for example,
that age-related changes in frequency resolution (cf. Patterson et al, 1982), not reflected in
threshold sensitivity, might impact the suprathreshold coding ofcomplex signals to an extent
sufficient to affect SSI scores . The present data
do not, however, provide support for such a
hypothesis . If, for example, frequency resolution becomes poorer with age, then such a
change in cochlear function should have maximal effect on tests like PB and SPIN-low, since
36
spectral resolution is presumably more important for monosyllabic word recognition than for
the identification of whole sentences . But neither PB nor SPIN-low showed the kinds of age
effects to be expected from progressively moreimpaired spectral resolution,
One might, of course, hypothesize other, as
yet undiscovered, age-related changes in
cochlear function, which spare pure-tone threshold sensitivity, yet have the potential to impact
speech recognition scores like SSI. But a similar argument can be marshaled against all such
hypotheses . Previous studies of speech audiometric scores for different types of speech materials in subjects with high-frequency hearing
loss (cf. Jerger and Hayes, 1977) suggest that
changes in cochlear status can be expected to
have a greater effect on single-syllable word
recognition than on tasks like SSI. But the
present results show exactly the opposite effect :
greater change in SSI than in tasks like PB and
SPIN-low . It is difficult, therefore, to reconcile
the present results with the hypothesis that the
age-related decline in SSI can be explained by
changes at the level of the cochlea.
How, then, can we explain the systematic
declines in SSI scores with age? It has been
suggested (Olsho et al, 1985 ; CHABA, 1988)
that such effects might be explained by concomitant decline in one or more of the cognitive
abilities important for carrying out speech recognition tasks. Arguing against this possibility,
however, is the fact that our previous correlational analysis (Jerger et al, 1991), of the database from which this sample was drawn, demonstrated only a limited predictive relation
between the SSI scores and any of several
neuropsychological tests probing a variety of
cognitive dimensions .
If neither hearing sensitivity status alone,
nor cognitive function alone, can explain the
phenomenon of age-related decline in speech
recognition scores, then perhaps the observed
decline results from an interaction between the
two factors. Such an interactive model has,
indeed, been proposed by Willot (in press). He
suggests that what have been interpreted as
changes in central auditory function may, instead, be reflections of the interaction between
a speech signal degraded by peripheral pathology and a deficit in cognitive function. The
stress placed on an impaired cognitive system
by the peripherally degraded speech signal
might adversely affect performance on speech
audiometric tasks to an extent not predicted
from either deficit alone. To investigate this
104, s h .
,:10i
Age-Related Decline in Speech Understanding/Jerger
Table 3
Cognitive, Hearing Loss, and Combined Effects, in Percent Correct,
on the Four Speech Measures
Speech Measure
Effect
Cognitive Effect
Hearing Loss Effect
Cog +HL
Actual Combined Effect
Interactive Effect
Notation
PB
SSI
SPIN-high
SPIN-low
A
B
A+B
C
C - (A + B)
2
32
34
28
-6
6
27
33
37
4
1
20
21
19
-2
-2
28
26
32
6
possibility we divided our total group in two
ways ; first, according to degree of peripheral
hearing sensitivity loss (PTA; average of both
ears at 500, 1000, 2000, and 4000 Hz), then
according to a cognitive measure of speed of
mental processing, the Digit Symbol subtest of
the WechslerAdult Intelligence Scale-Revised
(Wechsler, 1977). This is a complex measure of
information processing and speed of manual
response . Its relevance for speech audiometric
procedures (Jerger et al, 1991) derives from its
sensitivity as a measure of speed of mental
processing. On each measure, hearing loss (HL)
and digit symbol (DS) score, we divided the total
group at the 50th percentile into the best half
and the worst half. Thus, HL, = 50 percent with
least loss (PTA < 26 dB); HL 2 = 50 percent with
most loss (PTA ? 26 dB); DS, = 50 percent with
best digit symbol score (Raw Score > 6) ; and DS2
= 50 percent with poorest digit symbol score
(Raw Score <_ 6) .
Next we formed four subgroups as follow :
I = HL, and DSI; II = HL, and DS2; III = HL2 and
DSI; IV = HL2 and DS2.
We then computed the average speech score
for each of these four subgroups on each of the
four measures of speech understanding. We
reasoned that the following relations should
hold :
A = (I - II) = cognitive effect ; i.e ., difference
in average speech score between best half and
worst half on Digit Symbol test with degree of
hearing loss held constant .
B = (I - 111) = hearing loss effect ; i.e .,
difference in average speech score between half
with least loss and half with most loss, with
Digit Symbol score held constant.
interactive model would predict that C > (A+ B) .
The actual values of A, B, C, (A + B), and [C - (A
+ B)] for the present data are summarized in
Table 3 . The final row, labeled [C - (A + B)], is
the interactive effect . For SSI the interactive
effect was + 4 percent. The other three speech
measures showed similarly small effects. Indeed, for two measures (PB and SPIN-high) the
effect was in the negative direction . Averaged
across all four speech measures the interactive
effect was + 0.5 percent. These results fail to
provide strong support for an interactive model
of hearing loss and cognitive effects.
In summary, we have shown a progressive,
age-related decline in performance on the SSI
test that cannot be explained by age-related
change in hearing sensitivity loss . Furthermore, this decline was demonstrated, not between two groups (one young and the other
elderly), but across four age sub-groups within
the elderly age range. After reviewing a number
of alternative explanations, including changes
in the auditory periphery not reflected in puretone thresholds, cognitive status, and a possible
interaction between hearing sensitivity level
and cognitive function, we conclude that none
provides a satisfactory explanation for the observed changes in speech understanding.
Acknowledgment . This research was supported by research grant #AG05680 from the National Institute of
Aging, National Institutes of Health, U.S . Public Health
Service.
Portions of this paper were presented at the 14th
Danavox Symposium, Faaborg, Denmark, September 2528,1990.
REFERENCES
C = (I - IV) = actual combined effect ; i.e .,
difference in average speech score between half
with least loss and best Digit Symbol score and
half with most loss and poorest Digit Symbol
score.
Arnst D. (1982) . Staggered spondaic word test performance in a group of older adults : a preliminary report . Ear
Hear 3:118-123 .
We further reasoned that an additive model
would predict that C = (A + B), whereas an
Bergman M. (1980) . Aging and the Perception of Speech .
Baltimore: University Park Press.
Balas R. (1962) . Results of the Staggered Spondaic Word
Test with an OldAge Population . Master's Thesis : Northern Illinois University .
37
3, Number 1, January 1992
Journal of the American Academy of Audiology/Volume
Bergman M, Blumenfeld V, Cascardo D, Dash B, Levitt H,
Margulies M. (1976) . Age-related decrement in hearing
for speech . Sampling and longitudinal studies . JGerontol
31 :533-538 .
Bess F, Townsend T. (1977) . Word discrimination for
listeners with flat sensorineural hearing losses . J,Speech
Hear Disord 42 :232-237 .
Carhart R, Nicholls S. (1971) . Perceptual masking in
elderly women. Asha 13 :535 .
Committee on Hearing, Bioacoustics and Biomechanics
(CHABA), Working Group on Speech Understanding and
Aging. (1988) . Speech understanding and aging. JAcoust
Soc Am 83 :859-893 .
Martini A, Bovo R, Agnolitto M, DaCol M, Drusian A,
in
Liddeo M, Morra B. (1988) . Dichotic performance
.
elderly Italians with Italian stop consonant-vowel stimuli
Audiology 27 :1-7 .
Olsho L, Harkins S, Lenhardt M. (1985) . Aging and the
auditory system . In : Birren J, Schaie KW, eds. Handbook
of the Psychology of" Aging. 2nd Ed . New York : Van
Nordstrand, Reinhold, 332-377 .
site of
Otto W, McCandless G. (1982) . Aging and auditory
:110-117
.
.
Ear
Hear
3
lesion
Patterson R, Nimmo-Smith 1, Weber D, Milroy R. (1982) .
The deterioration of hearing with age: frequency selectivity, the critical ratio, the audiogram, and speech threshold. J Acoust Soc Am 72 :1788-1803 .
Dubno J, Dirks D, Morgan D. (1984) . Effects of age and
mild hearing loss on speech recognition in noise. JAcoust
Soc Am 76 :87-96 .
Pestalozza G, Shore J. (1955) . Clinical evaluation of
presbycusis on the basis of different tests of auditory
function . Laryngoscope 65 :1136-1163 .
Duquesnoy A. (1983a). Effect of a single interfering noise
or speech source upon the binaural sentence intelligibility
of aged persons. J Acoust Soc Am 74 :739-749 .
Punch J, McConnell F. (1969) . The speech discrimination
function of elderly adults . JAud Res 9:156-166 .
Duquesnoy A. (1983b) . The intelligibility of sentences in
quiet and in noise in aged listeners. J Acoust Soc Am
74:1136-1144 .
Jerger J, Hayes D. (1977) . Diagnostic speech audiometry .
Arch Otolaryngol 103:216-222 .
Jerger J, Jerger S, Oliver T, Pirozzolo F. (1989) . Speech
understanding in the elderly. Ear Hear 10 :79-89 .
Jerger J, Jerger S, Pirozzolo F. (1991) . Correlational
analysis of speech audiometric scores, hearing loss, age,
and cognitive abilities in the elderly. Ear Hear 12 :103109.
Kryter K. (1983) . Presbycusis, sociocusis and nosocusis. J
Acoust Soc Am 73 :1897-1916 .
Kryter K. (1988) . Definitions: Presbycusis, Sociocusis,
and Nosocusis . Presented at Conference on the Aging
Ear, St Louis, MO .
on
Rintelmann W, Schumaier D. (1974) . Five experiments
speech discrimination utilizing CNC monosyllables (N .U .
Auditory Test No . 6) . Experiment III: Factors affecting
speech discrimination in a clinical setting: List equivaRes
lence, hearing loss, and phonemic regression . JAud
2:Suppl :12-15 .
Smith R, Prather W. (1971) . Phoneme discrimination in
.J
older persons under varying signal-to-noise conditions
.
Res
14
:630-638
Speech Hear
Wechsler D. (1977) . Wechsler Adult Intelligence ScaleRevised. New York : Psychological Corp .
Willott J. (in press) . Aging and the Auditory System :
Anatomy, Physiology, and Psychophysics . San Diego:
Singular .
Wingfield A, Poon L, Lombardi L, Lowe D. (1985) . Speed
rate,
of processing in normal aging: effects of speech
linguistic structure, and processing time . J Gerontol
40 :579-585 .