Dissociations in Perceptual Learning Revealed by Adult Age Differences in
Adaptation to Time-Compressed Speech
Jonathan E. Peelle and Arthur Wingfield
Brandeis University
When presented with several time-compressed sentences, young adults’ performance improves with
practice. Such adaptation has not been studied in older adults. To study age-related changes in perceptual
learning, the authors tested young and older adults’ ability to adapt to degraded speech. First, the authors
showed that older adults, when equated for starting accuracy with young adults, adapted at a rate and
magnitude comparable to young adults. However, unlike young adults, older adults failed to transfer this
learning to a different speech rate and did not show additional benefit when practice exceeded 20
sentences. Listeners did not adapt to speech degraded by noise, indicating that adaptation to time-
compressed speech was not attributable to task familiarity. Finally, both young and older adults adapted
to spectrally shifted noise-vocoded speech. The authors conclude that initial perceptual learning is
comparable in young and older adults but maintenance and transfer of this learning decline with age.
Keywords: speech comprehension, perceptual learning, aging, time-compressed speech, frequency-
shifted speech
One of the hallmarks of human speech perception is our ability
to recognize words produced by different speakers, even though
the acoustic representation of any given word differs from person
to person. For example, across different speakers, formant frequen-
cies corresponding to a single vowel category vary extensively
(Peterson & Barney, 1952), and even words produced by the same
speaker can vary in duration depending on previous contextual
information (Shields & Balota, 1991). When listening to everyday
conversations, we are largely unaware of these variations. How-
ever, when processing speech that is more removed from that to
which we are accustomed—be it a significantly different speech
rate, intonation, or accent—we can become acutely conscious of
the difficulties presented. Even so, under these conditions we may
notice that, over time, we are able to understand a speaker who was
unintelligible at the beginning of a conversation or lecture. This
gradual adjustment highlights the flexibility of our perceptual
processing and the remarkable ability of the auditory system to
normalize across a wide range of speech parameters.
One characteristic of speech that fluctuates widely is the rate at
which it is produced. This is true even within a single conversa-
tion, in which speakers frequently alter their speaking rate by
significant amounts (Miller, Grosjean, & Lomanto, 1984). These
variations can, in turn, affect the perceptual criteria applied by
listeners. Miller and Liberman (1979), for example, investigated
the stop-semivowel distinction of /ba/ and /wa/ and found that
deleting the last 216 ms of a vowel caused the /b/–/w/ category
boundary to shift toward a shorter duration. They concluded that
listeners use the overall duration of syllables to interpret transition
duration. Effects of speech rate are also readily apparent with more
complex linguistic stimuli: Regardless of age, listeners find rapid
speech more difficult to comprehend and remember, with this
difficulty amplified for syntactically complex sentences (Gordon-
Salant & Fitzgibbons, 1993; Wingfield, Peelle, & Grossman,
2003). The variability of speech rate in everyday conversation, as
well as the potential perceptual difficulties caused by this variabil-
ity, implies that the ability of listeners to adjust to changes in
speech rate is critical for successful comprehension of spoken
language.
In the experiments presented here, we used a computer program
to artificially compress speech to very rapid rates to study percep-
tual adaptation. The time compression method we used is a vari-
ation of the sampling method, in which small portions are deleted
at regular intervals from both voiced and silent portions of the
speech signal. The remaining portions of the signal are then
abutted in time, which results in a signal that is shorter in duration
than the original but that retains the same pitch and relative
temporal patterning (Foulke, 1971). This technique has several
advantages over studying perceptual adaptation with unaltered
stimuli. First, when speech rate changes are of the magnitude
encountered in everyday conversation, temporal normalization oc-
curs almost immediately, making any adjustment difficult to study.
Second, when speakers attempt to increase their speaking rate
naturally, they tend to distort the normal proportion of speech to
silence (Lane & Grosjean, 1973), which may make the speech less
intelligible when compared with artificial time compression, espe-
Jonathan E. Peelle and Arthur Wingfield, Volen National Center for
Complex Systems, Brandeis University.
We acknowledge support from National Institute on Aging Grants
AG04517 and AG019714 to Arthur Wingfield and National Institute on
Deafness and Other Communicative Disorders Fellowship F31 DC006376
to Jonathan E. Peelle. We also gratefully acknowledge support from the
W. M. Keck Foundation.
We are grateful to Stuart Rosen for his assistance in creating the stimuli
used in Experiment 5. We thank Julie Golomb, Ann Kim, and Sandra
McCoy for their help with data collection.
Correspondence concerning this article should be addressed to Arthur
Wingfield, Volen National Center for Complex Systems, MS 013, Brandeis
University, Waltham, MA 02454-9110. E-mail: wingfield@brandeis.edu
Journal of Experimental Psychology: Copyright 2005 by the American Psychological Association
Human Perception and Performance
2005, Vol. 31, No. 6, 1315–1330
0096-1523/05/$12.00 DOI: 10.1037/0096-1523.31.6.1315
1315

cially at extremely rapid rates (Janse, 2003). The use of computer
compression algorithms to produce speech rates much faster than
what could be spoken naturally allows us to push the perceptual
system to its limit and observe the characteristics and time course
of perceptual adaptation.
Several studies have investigated perceptual adjustment to time-
compressed speech in young adults. These studies have consis-
tently shown that when young adults are presented with 10–20
sentences compressed to a very rapid rate, their recall accuracy
increases with practice (Altmann & Young, 1993; Dupoux &
Green, 1997; Mehler et al., 1993; Pallier, Sebastia´n-Galle´s, Du-
poux, Christophe, & Mehler, 1998; Sebastia´n-Galle´s, Dupoux,
Costa, & Mehler, 2000). In addition, adaptation has been observed
in response to artificially synthesized speech (Schwab, Nusbaum,
& Pisoni, 1985) and phonetically legal nonwords (Altmann &
Young, 1993). Young adults also demonstrate a high degree of
improvement in time-compressed speech comprehension in one
language after exposure to time-compressed speech in a second
language with similar phonemes, regardless of whether they un-
derstand the second language (Sebastia´n-Galle´s et al., 2000). The
presence of perceptual adaptation in response to nonwords and
words in a language not understood by the listener strongly sug-
gests that adaptation does not require knowledge of word structure
or meaning and therefore occurs at some prelexical stage of
processing.
Because of the importance of spoken language comprehension
throughout the life span, the continuing ability of the perceptual
system to understand different speakers is essential. The fact that
older adults are able to understand a wide variety of speakers
indicates that they are able to normalize across several speech
parameters, including rate. This observation has been confirmed
experimentally by studies showing that older adults’ recall accu-
racy for speech compressed to between 80% and 65% of its
original duration is essentially equivalent to that for normal speech
(Wingfield et al., 2003; Wingfield, Wayland, & Stine, 1992). By
way of contrast, age-related differences in perceptual learning have
been reported in a number of modalities (e.g., Fernandez-Ruiz,
Hall, Vergara, & Diaz, 2000; Gilbert & Rogers, 1996; Rogers,
Fisk, & Hertzog, 1994), including some evidence indicating that
older adults are less able than young adults to normalize across
various speech parameters (Sommers, 1997). These studies raise
the question of whether perceptual learning in the context of
speech comprehension will be preserved. Studying older adults’
perceptual learning in this context is appealing for two reasons.
First, speech comprehension is an overpracticed task and therefore
presumably is relatively resistant to age-related declines. Second,
it is a behaviorally relevant task, so results of these studies may
have practical consequences for communication practices with
older adults. To date, adaptation to rapid speech has not been
studied in older adults. Thus, the extent to which older adults can
adapt to this rapid sensory input is unknown.
The fact that older adults’ overall competence on linguistic tasks
remains quite good (e.g., Wingfield & Stine-Morrow, 2000) sup-
ports the notion that their perceptual systems are still generally
flexible. However, many peripheral and cognitive changes occur in
normal aging that adversely affect speech comprehension. At a
sensory level, older adults generally have poorer hearing acuity
than do young adults (Morrell, Gordon-Salant, Pearson, Brant, &
Fozard, 1996). Peripheral causes of this decline include loss of
both inner and outer hair cells of the inner ear and decreased blood
flow to important auditory structures, resulting in declines in
frequency discrimination, intensity discrimination, and perfor-
mance on simultaneous masking tasks (Schneider, 1997). In addi-
tion to this peripheral hearing loss, older adults exhibit declines in
temporal auditory processing. For example, compared with young
adults, older adults require a significantly larger gap between
adjacent tones to perceive them as separate (Schneider & Hamstra,
1999; Schneider & Pichora-Fuller, 2001).
Older adults also demonstrate a general age-related decline in
processing speed (Salthouse, 1994, 1996). This limitation would
likely have an especially pronounced effect on spoken language
comprehension: Unlike written language comprehension, in which
the reader has the opportunity to reread material, speech compre-
hension is by its very nature a time-dependent process. From the
acoustic signal, words must be recognized, syntactic structure
determined, and meaning integrated with other constituent units,
all while new information continues to arrive. When sensory input
is rapid, listeners have less time to process the speech signal; this
time constraint should make time-compressed speech especially
difficult for older adults to process, independent of any changes in
auditory acuity.
The effects of these age-related declines are evident even at the
level of word processing. Older adults perform worse than their
younger counterparts on time-compressed word identification
tasks, with this difference accentuated at faster speech rates (Bea-
sley & Maki, 1976; Konkle, Beasley, & Bess, 1977). At the
sentence level, older adults’ comprehension of time-compressed
speech is also differentially impacted relative to young adults’
(Wingfield, 1996; Wingfield et al., 2003). In addition to word
identification, processing connected discourse requires the organi-
zation of language into meaningful units, an operation that is
presumably adversely affected by reductions in the amount of
available processing time. In support of this notion, Wingfield,
Tun, Koh, and Rosen (1999) inserted silent periods at clause
boundaries in passages of time-compressed speech. The silent
periods had the same total length as the total amount of speech
signal deleted, such that overall passage length was kept constant.
Thus, the added silent periods increased the amount of processing
time available to the listener at syntactic boundaries but had no
effect on the degraded sensory input. At a moderate rate of
compression (240 words per minute [wpm]), this additional time
brought both young and older adults back to their baseline levels
for recall accuracy. At a faster rate (300 wpm), however, young
adults returned to baseline, but older adults did not. This illustrates
the fact that although available processing time affects speech
comprehension, perceptual factors still play a large role in speech
intelligibility, particularly for older adults.
Our goal in the current set of studies is to determine whether
auditory perceptual learning of compressed speech sounds is
equivalent in young and older adults. To do so, we investigate
whether older adults can adapt to highly compressed speech and
how this adaptation compares with that seen in young adults. Older
adults’ ability to successfully comprehend spoken language under
a variety of conditions indicates that they are capable of such
adaptation. Conversely, the cognitive declines associated with
normal aging suggest that any improvement might happen over a
longer time scale in older adults or may be absent altogether at
very rapid speech rates.
1316 PEELLE AND WINGFIELD