Psychology and Aging
1999, Vol. 14, No. 3, 380-389
Copyright 1999 by the American Psychological Association, Inc.
0882-7974/99/S3.00
Regaining Lost Time: Adult Aging and the Effect of Time Restoration
on Recall of Time-Compressed Speech
Arthur Wingfield, Patricia A. Tun, Christine K. Koh, and Merri J. Rosen
Brandeis University
Two experiments in which time was restored to artificially accelerated (time-compressed) speech are
reported. Experiment 1 showed that although both young and older adults’ recall of the speech benefited
from the restoration of time, time restoration failed to boost the older adults to their baseline levels for
unaltered speech. In Experiment 2, either 100% or 125% of lost time was restored by inserting pauses,
either at linguistic boundaries or at random points within the passages. Experiment 2 showed that the
beneficial effects of time restoration depended on where processing time was inserted, as well as how
much time was restored. Results are interpreted in terms of age-related slowing in speech processing
moderated by preserved linguistic knowledge and short-term conceptual memory.
An age-related slowing in information processing has been
frequently cited as a major source of the cognitive and memory
performance declines seen in adult aging (Salthouse, 1991, 1994).
Given the postulate of perceptual and cognitive slowing with
aging, it would be reasonable to expect that older adults would be
especially hampered in recall of rapid speech. Rapid speech input
would reduce the amount of time available for effective memory
processing, thus making it likely that a slowed processing system
would fall behind in recognition and analysis of the input. Con-
sistent with this postulate are numerous reports showing that older
adults have special difficulty in comprehension and recall of
artificially accelerated time-compressed speech (Gordon-Salant &
Fitzgibbons, 1993; Konkle, Beasley, & Bess, 1977; Riggs, Wing-
field, & Tun, 1993; Stine & Wingfield, 1987; Tun, Wingfield,
Stine, & Mecsas, 1992; Wingfield, Poon, Lombard!, & Lowe,
1985).
Although there is a trend toward a decline in auditory acuity
with age (presbycusis), the available evidence suggests that the
special vulnerability of older adults to time-compressed speech is
a phenomenon independent of peripheral sensory impairment. That
is, older listeners continue to show greater difficulty with time-
compressed speech than do young adults even when young and
older adults are matched for hearing acuity, either by testing young
adults with hearing losses that match those of an older group or by
comparing normally hearing young adults with older adults who
have no audiometric signs of presbycusis (Gordon-Salant &
Fitzgibbons, 1993; Konkle et al., 1977; Luterman, Welsh, & Mel-
rose, 1966; Sticht & Gray, 1969).
Arthur Wingfield, Patricia A. Tun, Christine K. Koh, and Merri J.
Rosen, Department of Psychology and Volen National Center for Complex
Systems, Brandeis University.
This work was supported by National Institutes of Health Grant R37
AG04517 from the National Institute on Aging. We also gratefully ac-
knowledge support from the W. M. Keck Foundation.
Correspondence concerning this article should be addressed to Arthur
Wingfield, Volen National Center for Complex Systems, MS 013, Brandeis
University, Waltham, Massachusetts 02454-9110. Electronic mail may be
sent to wingfield@volen.brandeis.edu.
The question of processing speed is of critical importance in
aging research because even the "slower" rates of ordinary con-
versation represent a very rapid input. Although subject to wide
variability (Miller, Grosjean, & Lomanto, 1984), speech rates in
ordinary conversation typically average between 140 to 180 words
per minute (wpm), and speakers working from a prepared script
easily exceed even these rates (Stine, Wingfield, & Myers, 1990).
That is, the limits on maximum speech rates result primarily from
a speaker’s ability to organize his or her thoughts to convey a
message. Once an utterance has been made, however, the normal
perceptual system can operate far more rapidly than the system for
production. For example, depending on the speech content, young
adults can comprehend and recall speech that has been artificially
accelerated to twice its normal rate, albeit with some effort
(Foulke, 1971; Wingfield, 1975).
Experimental studies of comprehension and recall for rapid
speech typically use speech that has been time compressed by a
computer rather than relying on speakers attempting to speak
rapidly. This is so because even trained speakers’ attempts to speak
very rapidly are accompanied by subtle and uncontrollable
changes in articulatory clarity and in the patterns of linguistically
based pauses and intonational contour (Lane & Grosjean, 1973;
Speer, Wayland, Kjelgaard, & Wingfield, 1994). By contrast,
modern computer-editing compression techniques allow one to
reduce to an equal degree the durations of both the words and the
pauses that occur naturally in speech. This is accomplished by use
of the so-called sampling method of time compression in which
speech is recorded, and small segments are then periodically
deleted from the recording, with the remaining segments then
abutted in time. By removing these segments at a regular rate from
both speech and natural pauses that occur in the speech utterance,
one preserves the relative temporal pattern of speech and silences
of the original, as well as maintaining the original intonation
pattern (Foulke, 1971).
Although there is some inevitable loss of richness of the speech
signal imposed by these procedures, especially for very high rates
of compression (Heiman, Leo, Leighbody, & Bowler, 1986), time-
compressed speech generally sounds quite normal except for its
rate. The decrement in comprehension and recall that occurs with
380

TIME RESTORATION 381
time-compressed speech has been attributed primarily to the loss
of the processing time that ordinarily would be available with
slower speech rates (Chodorow, 1979; Foulke, 1971). Thus, to the
extent that older adults need more processing time than young
adults to recognize and encode speech for memory, it would be
reasonable to expect that older adults would have special difficulty
with time-compressed speech. There is a corollary to this sugges-
tion: If lack of time is an important factor limiting older adults’
effective memory encoding of speech materials, then restoring
time to artificially accelerated speech should raise the performance
level for older adults as well as for young adults.
One way to test this hypothesis would be to insert or lengthen
silent pauses between words or between phrases in an amount
equal to the time lost by the time compression of the speech signal.
This general principle was demonstrated with young adults using
time-compressed and time-restored digit lists, in which especially
long silent intervals were added between each of the compressed
digits (Aaronson, Markowitz, & Shapiro, 1971). Also studying
young adults, Overmann (1971) demonstrated the effect with con-
nected speech by first time compressing speech passages and then
inserting silent periods between the phrases and sentences. This
time-restoration procedure significantly improved comprehension
relative to performance for time-compressed speech without the
insertion of these silent pauses.
Overmann (1971) interpreted her results in terms of a limited
capacity processing system that can process only so much input
per unit time. It is important to note, however, that Overmann’s
claim was that time restoration improved performance, not that
performance would necessarily recover to the same level as that
for unaltered speech. Indeed, even for her young adult participants,
Overmann found that although insertion of restored time signifi-
cantly improved comprehension, it did not bring performance back
to precompression levels when speech rates exceeded 250 wpm.
In the case of older adults, studies of time compression and time
restoration have produced a somewhat mixed picture. In the liter-
ature, there are certainly many reports of older adults claiming that
rapid speech is difficult to follow as compared with slower speech
(e.g., Obler, Fein, Nicholas, & Albert, 1991). On the other hand,
when actually tested by slowing speech, these claims are not
always accompanied by actual improvement in comprehension or
recall performance (Schmitt & Carroll, 1985; Schmitt & Moore,
1989). Schmitt and McCroskey (1981) used "time expansion" to
examine comprehension of sentences by older adults relative to an
initial speaking rate of 175 wpm. In this technique, speech rate is
slowed using the same technology as for time compression, except
that expansion is achieved by reiteratively resampling the speech
so as to extend the duration of the speech signal. Unlike Over-
mann’s (1971) method of inserting silent intervals selectively
following sentences and clauses, Schmitt and McCroskey’s expan-
sion procedure increased the durations of all of the speech ele-
ments by proportionally equivalent amounts, to include the dura-
tions of syllables, words, and silent periods. These investigators
found that expansion of the speech signal to 140% of original
playing time aided comprehension but that no further improvement
was observed when durations were increased to 180% of original
playing time.
In a similar study in which an old-old group (over 75 years of
age) was included, Schmitt (1983) reported a slight but nonsignif-
icant improvement when speech was expanded from a nominal
normal rate of 175 wpm to 140% of the original speaking time,
combined with a nonsignificant decline in performance when the
speech was further expanded to 180% of the original time. A
similar mixed picture appeared when effectiveness of time expan-
sion of spoken sentences was studied in several patients with
Alzheimer’s disease who differed in cognitive efficacy and work-
ing memory (Small, Andersen, & Kempler, 1997).
This mixed picture of the uncertain benefits of expanding
speech utterances beyond their original duration should not be
surprising for two reasons. The first reason relates to the attempt to
allow more processing time by means of a uniform expansion of
all elements of the speech signal. Although this procedure might
seem reasonable, it can have the negative consequence of tempo-
rally stretching syllabic durations beyond the size of a "perceptual
window" (a syllable-size acoustic-perceptual buffer) needed for
effective syllable identification. This stretching could thus inter-
fere with the recognition process at the syllable and word levels
(Huggins, 1975; Massaro, 1974; see also the review by Mattys,
1997).
The second reason for these mixed results rests on the likelihood
that the amount of time restored to a rapid speech signal is less
important than where this time is restored. That is, to the extent
that speech is processed in linguistic units such as sentences and
clauses, one would expect restoration of processing time to be
beneficial only when it is made available to the listener at these
critical points. There is considerable evidence, for example, that a
listener’s choice of where to allocate processing time in listening
to speech is sensitive to the linguistic structure of the speech
passage. This has been shown with the techniques of spontaneous
segmentation (Wingfield & Butterworth, 1984) and self-paced
listening (Ferreira, Henderson, Anes, Weeks, & McFarlane, 1996).
Both techniques have shown that young and older adults are
similarly drawn to clause and sentence boundaries in speech pro-
cessing (Titone, Prentice, & Wingfield, 1996; Wingfield, Lahar, &
Stine, 1989).
In contrast to the mixed findings arising from use of uniform
time expansion, clear results were obtained by G. Cohen (1979)
with natural speech that had been recorded either at 120 or 200
wpm. In this well-known study, both young and older adults
showed a significant advantage in answering fact and inference
questions for passages spoken at the slower speech rate. G. Co-
hen’s experiment does not indicate what parameters the speaker
there may have attempted to adjust in order to speak more slowly
or more rapidly. It is known, however, that the natural tendency of
speakers asked to speak slowly is to increase the relative lengths of
pauses at linguistic boundaries (Lane & Grosjean, 1973; Speer et
al., 1994). It may thus be that the effectiveness of the slower
speech rate in G. Cohen’s study was due to the way the extra time
was distributed across the speech signal, in contrast to the time-
expansion studies in which linguistic boundaries were given no
special treatment.
In support of this likelihood, Gordon-Salant and Fitzgibbons
(1997) showed that older adults as a group did not benefit from
silent periods inserted between each and every word in sentences.
Indeed, using interword silent intervals ranging up to 1,600 ms,
Gordon-Salant and Fitzgibbons found lower recall accuracy than
occurred when silent intervals were not inserted. These authors
suggested that their insertion of pauses between words may have
altered the normal prosodic pattern of the speech, thus counteract-