Content uploaded by Ruth Litovsky
Author content
All content in this area was uploaded by Ruth Litovsky on Jan 26, 2014
Content may be subject to copyright.
Speech intelligibility and spatial release from masking in young
children
a)
Ruth Y. Litovsky
b)
Waisman Center, University of Wisconsin—Madison, 1500 Highland Avenue, Madison, Wisconsin 53705
共Received 13 August 2003; accepted for publication 27 January 2005兲
Children between the ages of 4 and 7 and adults were tested in free field on speech intelligibility
using a four-alternative forced choice paradigm with spondees. Target speech was presented from
front 共0°兲; speech or modulated speech-shaped-noise competitors were either in front or on the right
共90°兲. Speech reception thresholds were measured adaptively using a three-down/one-up algorithm.
The primary difference between children and adults was seen in elevated thresholds in children in
quiet and in all masked conditions. For both age groups, masking was greater with the speech-noise
versus speech competitor and with two versus one competitor共s兲. Masking was also greater when the
competitors were located in front compared with the right. The amount of masking did not differ
across the two age groups. Spatial release from masking was similar in the two age groups, except
for in the one-speech condition, when it was greater in children than adults. These findings suggest
that, similar to adults, young children are able to utilize spatial and/or head shadow cues to segregate
sounds in noisy environments. The potential utility of the measures used here for studying
hearing-impaired children is also discussed. © 2005 Acoustical Society of America.
关DOI: 10.1121/1.1873913兴
PACS numbers: 43.66.Pn, 43.66.Qp, 43.71.Ft 关AK兴 Pages: 3091–3099
I. INTRODUCTION
Children spend numerous hours every day in complex
auditory environments, such as classrooms, where multiple
sounds that vary in content and direction typically co-occur.
In addition to voices of adults and children, instructional
aids, environmental sounds, and reverberation are standard
aspects of acoustic environments in classrooms. Some work
indicates that children learn best in relatively quiet environ-
ments, and often have difficulty hearing speech in the pres-
ence of distracting sounds 共Crandell, 1993; Yacullo and
Hawkins, 1987; Papso and Blood, 1989兲. Psychophysical
studies in which stimuli were presented over headphones
have shown that, compared with adults, preschool listeners
exhibit poorer attentional selectivity on auditory tasks 共e.g.,
Stellmack et al., 1997; Oh et al., 2001兲 and reduced unmask-
ing for tone detection under dichotic conditions 共Wightman
et al., 2003; Hall et al., 2004兲.
Also under headphones, it has been found that in the
presence of two-talker maskers speech reception thresholds
are higher in children than adults, and for both age groups
thresholds are higher in the presence of two-talker maskers
than with speech-shaped noise maskers 共Hall et al., 2002兲.
Headphone stimulus presentation is limited, however, be-
cause spatial cues that are known to be important for sound
segregation in realistic environments are missing. Studies
with adults have shown that the ability to segregate target
speech from competing speech and/or noise is determined by
a complex set of auditory computations that involve both
monaural and binaural processes 共Hawley et al., 1999, 2004;
Bronkhorst, 2000; Culling et al., 2004兲. Spatial cues in par-
ticular play a key role in facilitating source segregation.
Speech intelligibility improves by up to 12 dB when the
target speech and competing sounds are spatially separated,
resulting in ‘‘spatial release from masking’’ 共Plomp and
Mimpen, 1981; Bronkhorst and Plomp, 1992; Nilsson et al.,
1994; Koehnke and Besing, 1996; Peissig and Kollmeier,
1997; Hawley et al., 1999, 2004; Shinn-Cunningham et al.,
2001; Litovsky et al., 2002兲.
The extent to which children demonstrate spatial release
from masking for speech is poorly understood. Of particular
interest in the present study is the effect of number of
maskers, as well as their content, on the extent to which
young children experience spatial release from masking. In
adult listeners spatial release from masking is especially
large for multiple 共two or more兲 maskers that carry linguistic
content or context 共i.e., speech or reversed speech兲, and rela-
tively small for a single, nonspeech masker such as speech-
shaped noise 关Hawley et al., 共2004兲; see also Bronkhorst
共2000兲 for review兴. The authors of those works have con-
cluded that release from masking as provided by spatial cues
is particularly effective when the auditory environment is
complex. The concept of ‘‘informational masking’’ has been
invoked to explain this phenomenon, whereby, in the pres-
ence of maskers that are harder to ignore, spatial cues be-
come important for sound source segregation. In this case,
maskers that are multiple in number and/or that carry infor-
mation resembling that contained in the target result in
greater spatial release from masking 共e.g., Brungart 2001;
Freyman et al., 2001; Arbogast et al., 2002; Durlach et al.,
2003兲.
Several studies have reported that speech masking in
children depends on the masker type 共Papso and Blood,
a兲
Select portions of these data were presented at the 143rd Meeting of the
Acoustical Society of America, Pittsburgh, PA, and at the 24th Meeting of
the Association for Research in Otolaryngology, Tampa, FL.
b兲
Electronic mail: litovsky@waisman.wisc.edu
3091J. Acoust. Soc. Am. 117 (5), May 2005 0001-4966/2005/117(5)/3091/9/$22.50 © 2005 Acoustical Society of America
1989; Hall et al., 2002, 2004兲. However, the effect of num-
ber and spatial cues, and the possible contribution of these
stimulus parameters to spatial release from masking, remain
poorly understood. Binaural abilities in children are adultlike
on measures of binaural masking level differences 共Nozza
et al., 1988; Moore et al., 1991兲 and minimum audible angle
共Litovsky, 1997兲. Since spatial cues are known to play a key
role in speech understanding for adults, it is important to
understand how young children comprehend speech in real-
istic, multi-source acoustic environments, and the conditions
that enable them to benefit from spatial cues. The research
paradigm used here may ultimately also be useful in evalu-
ating performance of hearing-impaired children. Noisy envi-
ronments are particularly problematic for children with a his-
tory of otitis media 共e.g., Hall et al., 2003; Moore et al.,
2003; Roberts et al., 2004兲 and for hearing aid and cochlear
implant users 共e.g., Dawson et al., 2004; Eisenberg et al.,
2004; Litovsky et al., 2004兲. Because the important task of
hearing speech in noise can be a daily struggle for many of
these children, ultimately their performance on these mea-
sures can assist with diagnosis and fitting strategies.
In the present study the task involved a four-alternative
forced-choice 共4AFC兲 word discrimination paradigm. Sub-
jects selected a picture that matched the speech target from
an array of four pictures that appeared on a computer moni-
tor. Other tests such as the HINT-C 共Nilsson et al., 1994兲
may be usable for measuring speech intelligibility in noise in
children as young as 6 years, but are difficult to implement
with younger children. The test protocol described here was
specifically designed to enable the study of speech intelligi-
bility in noise in children as young as 4 years old, an age at
which many children begin to spend a significant number of
hours in noisy environments such as preschool classrooms.
II. METHODS
A. Subjects
A total of 36 volunteer children were recruited from lo-
cal public schools and the general community 共14 males and
22 females兲, and all subjects completed testing on the three
required conditions. Subjects ranged in age from 4.5 to 7.5
years 共average and standard deviation⫽5.5⫾1 years; see also
Table I兲.
1
All were native speakers of English with no known
auditory dysfunction or other cognitive disorders. According
to the parents’ report, none of the children were on medica-
tion or had known illness or ear infections on the day of
testing, and none of the children had a known history of
hearing loss. Total testing time for each listener was approxi-
mately 45 min.
Nine paid adult volunteers, with normal hearing as veri-
fied by standard audiometric testing for frequencies between
250 and 8000 Hz, and English as their first language, were
also tested. Since testing was much less time consuming with
adults than with children, a within-subject design was used
whereby each subject participated in all conditions that per-
tained to the four groups of children.
B. Testing chamber, materials apparatus
Testing was conducted in a single-walled sound booth
共3.6⫻4m兲 with carpeting. This room had a reverberation
time (T
60
)⫽ 250ms and ambient noise levels averaging 35
dB SPL. During testing, subjects were always seated in the
center of the room, with loudspeakers 共Radio Shack Mini-
mus 7兲 placed at 15.24 cm above ear level for children 共ear
level for adults兲 and at a distance of 1.67 m from the center
of the subject’s head. All stimuli were prerecorded, digitized,
and stored on a laptop computer 共Winbook兲. In the one-
competitor conditions, the target and competing sound were
fed to separate channels of a two-channel soundcard 共Digi-
gram VX Pocket兲, amplified 共Crown D-75兲, and presented to
separate loudspeakers. When both target and competitor
were presented from the front position, the speakers were
placed next to one another, with their centers at ⫾2°, with
their medial walls nearly touching. Each loudspeaker sub-
tended 4° in the horizontal dimension, hence strictly speak-
ing, speakers were separated by 4°. In the two-competitor
condition, when both occurred from the front, they were pre-
sented from the same loudspeaker. Target stimulus selection,
level controls, and output as well as response acquisition
were achieved using Matlab. A picture book containing four
target pictures per page was placed on a small table in front
of the subject.
C. Stimuli
Stimuli consisted of target words and competing sen-
tences. Targets comprised a closed set of 25 spondaic words
from CID W-1 obtained from Auditech and spoken by a male
talker. Although a larger set of words is available, the subset
chosen for the present study consisted of words that were
easily represented with a visual illustration and readily rec-
ognized as such during pilot testing of 20, 4 to 5 year-old
children 共a list of the target words used is shown in the
Appendix兲. The root-mean-square levels were equalized for
all target words using Matlab software. The competitors were
either speech or modulated speech-shaped noise. Competing
sentences were taken from the Harvard IEEE list 共Rothauser
et al., 1969兲 and recorded with a female voice. Examples of
sentences are ‘‘Glue the sheet to the dark blue background,’’
TABLE I. List of conditions tested for children 共nine subjects per condition兲.
Group
No. of
competitors
Age range
共years. months⫾SD兲
Competitor
type Conditions
1 1 5.4⫾1.1 Speech Quiet, 1 front, 1 right
2 1 5.6⫾1.2 Speech-noise Quiet, 1 front, 1 right
3 2 5.8⫾1 Speech Quiet, 2 front, 2 right
4 2 5.6⫾1 Speech-noise Quiet, 2 front, 2 right
3092 J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children
‘‘Two blue fish swam in the tank,’’ and ‘‘The meal was
cooked before the bell rang.’’ Ten such sentences were used,
and these were presented in a random order during testing.
Speech-noise was made based on the ten competitor sen-
tences and also played in a random order during testing.
These interferers were filtered to match the long-term spec-
trum of the speech competitors, calculated for each talker
separately. The noise samples were scaled to the same root-
mean-square value and cut to the same length as the match-
ing speech competitor. The envelope was then extracted from
the speech competitor and was used to modulate the noise
tokens, giving the same coarse temporal structure as the
speech. The envelope of running speech was extracted using
a method similar to that described by Festen and Plomp
共1990兲 in which a rectified version of the waveform is low-
pass filtered. A first-order Butterworth low-pass filter was
used with a 3-dB cutoff at 40 Hz.
D. Design
The target words were always presented from the front
共0°兲. Competitors were presented from either front or side
共90°兲. Four groups of children with nine subjects per group
were tested 共see Table I兲. The side condition was always with
competitor共s兲 on the right. Each child subject was randomly
assigned to a group that was tested on one combination of
type 共speech or speech-noise兲 and number 共1or2兲 of com-
petitor共s兲. The subject was then tested on three conditions:
共1兲 quiet: no competitor共s兲, 共2兲 front: target and competitor共s兲
in front, and 共3兲 right: target in front and competitor共s兲 at 90°
on the right; the order of conditions was randomized using a
Latin-square design. For the adult group, testing was con-
ducted in a single 2-h session, with the order of the nine
conditions randomized for each listener.
For each condition one adaptive track was measured.
When two competitors were presented they were of the same
type, but different samples were used for the two sources; in
the two-speech conditions the same female voice was pre-
sented, speaking two different sentences, and in the two-
speech-noise conditions two different segments of the noise
were presented.
E. Familiarization
The present study was not aimed at testing children’s
vocabulary, but rather their speech intelligibility for known
words. The 25 words were selected from the spondee list
after pilot testing indicated that 20, 4 to 5 year-old children
were either familiar with the words or could easily ascertain
their meaning after one presentation. For each of the 25
words a commissioned artist-drawn picture was used to vi-
sually represent the meaning of the word. Prior to testing,
subjects underwent a familiarization session 共approximately
5 min in duration兲 in which they were presented with the
picture-word combinations and tested to insure that they as-
sociated each of the pictures with their intended auditory
target.
F. Speech reception threshold estimation
The test involved a single interval 4AFC discrimination
procedure. On each trial, the child viewed a set of four pic-
tures from the set of 25 picture-word matches. A word
matching one of the pictures was randomly selected and pre-
sented from the front speaker. A leading phrase such as
‘‘Point to the picture of the...’’ or ‘‘Where is the...’’ preceded
each target word. The child was asked to select the picture
matching the heard word, and to guess if not sure or if the
word was not audible. The randomization process ensured
that for every subject, on average, all 25 words were selected
an equal number of times. The experimenter entered the
child’s response into the computer. Following correct re-
sponses, feedback was provided in the form of 3-s musical
clips from popular children’s music. Approximately 20 clips
were digitized and stored on the computer, and randomly
selected on correct-feedback trials. Following incorrect re-
sponses, feedback was provided in the form of a brief phrase
such as ‘‘Let’s try another one’’ or ‘‘That must have been
difficult.’’ Five such phrases were digitized and stored on the
computer, and randomly selected on incorrect-feedback tri-
als.
An adaptive tracking method was used to vary the level
of the target signal, such that correct responses result in level
decrement and incorrect responses result in level increment.
The algorithm includes the following rules: 共1兲 Level is ini-
tially reduced in steps of 8 dB, until the first incorrect re-
sponse. 共2兲 Following the first incorrect response a three-
down/one-up rule is used, whereby level is decremented
following three consecutive correct responses and level is
incremented following a single incorrect response. 共3兲 Fol-
lowing each reversal the step size is halved. 共4兲 The mini-
mum step size is 2 dB. 共5兲 A step size that has been used
twice in a row in the same direction is doubled. For instance,
if the level was decreased from 40 to 36 共step⫽4兲 and then
again from 36 to 32 共step⫽4兲, continued decrease in level
would result in the next level being 24 共step⫽8兲. 共6兲 After
three consecutive incorrect responses a ‘‘probe’’ trial is pre-
sented at the original level of 60 dB. If the probe results in a
correct response the algorithm resumes at the last trial before
the probe was presented. If more than three consecutive
probes are required, testing is terminated and the subject’s
data are not included in the final sample. 共7兲 Testing is ter-
minated following five reversals.
For each subject, speech-reception-thresholds 共SRTs兲
were measured for each condition. At the start of each SRT
measurement, the level of the target was initially 60 dB SPL.
When competitors were present 共non-quiet conditions兲, the
level of each competitor was fixed at 60 dB SPL, such that
the overall level of the competitors was increased by ap-
proximately 3 dB when two competitors were presented
compared with the one-competitor conditions. Thus, the
adaptive track began with a signal-to-noise ratio of 0 dB in
the one-competitor cases and ⫺3 dB in the two-competitor
cases.
Results were analyzed using a constrained maximum-
likelihood method of parameter estimation outlined by Wich-
mann and Hill 共2001a, b兲. All the data from each experimen-
tal run for each participant were fit to a logistic function.
3093J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children
Thresholds were calculated by taking the inverse of the func-
tion at a specific probability level. In our 4AFC task, using
an adaptive three-down/one-up procedure, the lower bound
of the psychometric function was fixed at the level of chance
performance, 0.25, and the threshold level corresponded to
the point on the psychometric function where performance
was approximately 79.4% correct. Biased estimates of
threshold can occur. Bias can be introduced by the sampling
scheme used and lapses in listener attention. Wichmann and
Hill 共2001a, b兲 demonstrated that bias associated with lapses
was easily overcome by introducing a highly constrained pa-
rameter to control the upper bound of the psychometric func-
tion. This approach was used to assess our data. The upper
bound of the psychometric function was constrained within a
narrow range 共0.06兲 as suggested by Wichmann and Hill
共2001b兲. As the authors suggest, under some circumstances,
bias introduced by the sampling scheme may be more prob-
lematic to avoid even when a hundred trials are obtained per
level visited. The possibility of biased threshold estimates
due to our sampling scheme was assessed by comparing the
thresholds obtained using the constrained maximum-
likelihood method with traditional threshold estimates based
on the last three reversals in each experimental run. A re-
peated measured t-test on quiet thresholds for the 36 children
tested revealed no statistically significant difference
between the estimated threshold values obtained using the
ML approach versus the traditional approach
关
t(35)⫽1.37,
p⬎ 0.05, two tailed兴.
III. RESULTS
SRTs were statistically analyzed for the children groups
using a mixed-design analysis of variance 共ANOVA兲 with
two between-subjects variables 共number of competitors,
competitor type兲 and one within-subjects variable 共condi-
tion兲. Significant main effects of number
关
F(1,32)⫽4.05;
p⬍ 0.05
兴
and condition
关
F(2,32)⫽119.57, p⬍0.0001
兴
were found, but there was no effect of type. Significant in-
teractions were found for condition with number
关
F(2,64)
⫽ 66.50; p⬍0.03
兴
and condition with type 关F共2,64兲
⫽162.01; p⬍0.001兴. Scheffe’s posthoc contrasts 共signifi-
cance value p⬍ 0.05) showed that SRTs in quiet were sig-
nificantly lower than SRTs in either front or right. Children
tested with two competitors had significantly higher SRTs
than those tested with one competitor for the front and right
conditions 共further comparisons between front and right are
described below with regard to spatial release from mask-
ing兲. Finally, for reasons that are not clear, SRTs on the quiet
conditions were lower in the two speech-noise groups than in
the groups tested with the speech competitors. Adult
data were analyzed with a one-way ANOVA for the nine
conditions, which revealed a significant main effect
关
F(8,8)⫽3.77; p⬍0.05
兴
. Scheffe’s posthoc contrasts
关
F(8,8);Fp⬍ 0.01
兴
revealed that quiet SRTs were lower than
SRTs on all other conditions. Child and adult SRTs were
compared with independent t-tests for each of the nine con-
ditions; since the quiet condition was tested for each of the
child groups, a total of 12 comparisons were conducted. The
Bonferroni adjustment for multiple comparisons as described
by Uitenbroek 共1997兲 was applied 共df⫽16, criterion of t
⬎3.34 and p⬍ 0.004). Significant differences were found for
all 12 comparisons, suggesting that adults’ SRTs were lower
than those of children for all conditions tested.
Figure 1 shows group means 共⫾SD兲 for masking 共dif-
ferences between masked and quiet SRTs兲. For each subject
masking amounts for front and right were obtained by sub-
tracting quiet SRTs from front and right SRTs, respectively.
To place the masking values into context, average 共⫾SD兲
SRTs for all groups and conditions are listed in Table II.
Statistical analyses on the amount of masking for the child
groups were conducted with a three-way mixed-design
ANOVA treating condition 共front minus quiet, right minus
quiet兲 as the within-subjects variable and competitor type
and number as the between-subjects variables. A significant
effect of condition
关
F(1,32)⫽29.13; p⬍0.0001
兴
suggests
FIG. 1. Average 共⫾SD, dB SPL兲 differences between speech reception
thresholds 共SRTs兲 in the masked and quiet conditions. Data are plotted for
front 共top panels兲 and right 共bottom panels兲 conditions, for children 共left
panels兲 and adults 共right panels兲. Each panel compares difference values for
the speech and speech-noise competitors when the number of competitor共s兲
was either one 共black bars兲 or two 共gray bars兲.
TABLE II. Mean 共⫾SD兲 speech reception thresholds 共in dB SPL兲
a兲
Group Quiet Front Right
Children
1 speech 26.02共3.81兲 41.81 共6.31兲 36.64 共6.48兲
2 speech 27.32共5.25兲 47.75 共6.30兲 40.33 共6.29兲
1 speech-noise 23.25共5.56兲 44.37 共6.50兲 40.13 共3.89兲
2 speech-noise 21.45共3.3兲 48.01 共2.07兲 44.41 共7.18兲
Adults
1 speech 3.84共3.18兲 16.71 共5.66兲 16.86 共3.84兲
2 speech 23.35 共4.41兲 20.43 共4.01兲
1 speech-noise 27.39 共5.28兲 22.25 共4.82兲
2 speech-noise 32.82 共4.40兲 27.60 共8.65兲
a兲
It is important to recall that each child was tested on three conditions
共quiet, front, right兲 for one masker type, and that each adult was tested on
all nine conditions, hence only one entry in Table II for adult quiet thresh-
olds.
3094 J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children
that masking in the front minus quiet condition was higher
than in right minus quiet. Significant effects of type
关
F(1,32)⫽15.51; p⬍0.0001
兴
and number
关
F(1,32)
⫽ 6.95; p⬍0.013
兴
further suggest that masking was greater
for two competitors than one, and greater for the speech-
noise competitor compared with speech. There were no sig-
nificant interactions. For the adult subjects, a three-way re-
peated measures ANOVA 共condition⫻type⫻number兲
suggested, similar to the children, that masking was greater
in the front versus right conditions
关
F(1,8)⫽27.72;
p⬍ 0.001
兴
, greater with speech-noise than speech
关
F(1,8)
⫽ 30.72; p⬍0.001
兴
and greater for two compared with one
competitor
关
F(1,8)⫽16.71; p⬍0.004
兴
. Masking data for
child and adult groups were compared with independent
t-tests for each competitor location/type/number combina-
tion, and the Bonferroni correction for eight comparisons
was applied 共Uitenbroek, 1997兲. None of the comparisons
yielded a significant difference in masking between the child
and adult groups, and none of the interactions were signifi-
cant.
Spatial release from masking was defined as the differ-
ence between front masking 共front minus quiet兲 and right
masking 共right minus quiet兲. Figure 2 shows individual
points for right minus quiet plotted versus front minus quiet
for all subjects and conditions tested. If no spatial release
from masking occurred, the points would be expected to fall
along the diagonal. Points falling below the diagonal would
be indicative of spatial release from masking. Alternatively,
points falling above the diagonal would represent cases in
which thresholds were higher when the competitors were on
the right rather than in front. The majority of individual data
points in Fig. 2 are below the diagonal, and average points
for all but one group are also indicative of spatial release
from masking.
Figure 3 summarizes the findings for spatial release
from masking. For children, group average values are be-
tween 3.6 and 7.5 dB; the overall average for all 36 children
is 5.25 dB. For adults, group averages range from 0 to 5.2 dB
with an overall average of 3.34 dB. Children’s data were
analyzed with a two-way between-subjects ANOVA 共type
⫻number兲, revealing no significant main effects or interac-
tions. This lack of an effect may not surprising given the
large intersubject variability, which is notable in Fig. 3共A兲;
while some children had spatial release from masking values
greater than 10 dB, other children had values near 0, and a
small number had negative values. Adult data were analyzed
with a two-way repeated measures ANOVA 共type⫻number兲,
also revealing no significant effects or interactions. Finally,
FIG. 2. Masking amounts 共differences
between masked and quiet thresholds兲
for the Right minus Quiet conditions
are plotted vs. Front minus Quiet con-
ditions. Panels 共A兲 and 共C兲 show data
for children and adults, respectively;
each symbol denotes data from an in-
dividual subject, and the four different
symbols refer to the type/number com-
bination of competitor共s兲. The diago-
nal lines denote equality between the
two variables. Panels 共B兲 and 共D兲
show average group data from 共A兲 and
共C兲, respectively, for the four condi-
tions tested.
3095J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children
to compare spatial release from masking for children and
adults independent t-tests were conducted for each type/
number combination, with the Bonferroni correction for four
contrasts applied 共Uitenbroek, 1997兲. The only significant
difference between groups was for the one-speech competi-
tor condition, in which the average spatial release from
masking in adults is 0, compared with an average value of
5.7 for the child group.
IV. DISCUSSION
Speech intelligibility in quiet and in the presence of
competing sounds and the ability to benefit from spatial
separation of the speech and competitor共s兲 were investigated
in children and adults. Although extensively studied in
adults, to date this area of research has been minimal in
children. This study may therefore be helpful towards im-
proving our understanding of children’s ability to hear and
learn in noisy and reverberant environments, especially
given that such abilities are known to be compromised com-
pared with abilities measured under quiet condition 共e.g.,
ANSI, 2002; Yacullo and Hawkins, 1987; Knecht et al.,
2002兲. The results can be summarized as follows: 共1兲 Adults’
SRTs were lower than those of the children for all conditions.
共2兲 For both age groups masking was significantly greater
with speech-noise than with speech and with two competi-
tors compared with one. 共3兲 The amount of masking did not
differ across the two age groups. 共4兲 The amount of spatial
release from masking was similar for children and adults on
all but one condition. 共5兲 The number or type of competitor
did not affect the size of spatial release from masking for
either age group.
A. SRTs and masking amount
The primary age difference was that of higher SRTs in
children than adults, in quiet and in all masked conditions.
This age effect is consistent with existing developmental
psychoacoustic literature, which has shown that children
ages 4 to 7 typically have higher tone detection thresholds
compared with adults 共e.g., Buss et al., 1999; Oh et al.,
2001兲. Similarly, recognition of spondee words such as those
used here in temporally modulated noise has been shown to
produce higher thresholds in 5 to 10 year-old children than in
adults 共Hall et al., 2002兲.
The age effect found here can be attributed to a combi-
nation of peripheral and central mechanisms. Peripherally,
frequency resolution is highly similar to that of adults by 5
years of age 共Allen et al., 1989; Hall and Grose, 1991;
Veloso et al., 1990兲. However, young children appear to in-
tegrate auditory information over a greater number of audi-
tory channels than adults, suggesting that their ability to ex-
tract auditory cues, and in the present study to identify target
words at low signal levels, is likely to be still developing
共e.g., Hall et al., 1997; Buss et al., 1999; Hartley et al.,
2000; Oh et al., 2001兲. Immaturity of central auditory pro-
cesses and the adoption of listening strategies that are non-
optimal or less efficient than adults 共Allen and Wightman,
1994; Lutfi et al., 2003兲 may have also affected SRTs. Fi-
nally, differences in thresholds may represent age-related dif-
ferences in the ability to take advantage of hearing partial
word segments and to ‘‘fill in’’ the remainder of the target
word. Anecdotal reports from adults suggest that they relied
heavily on this strategy at low signal levels. The ability to
adopt this strategy can most likely be attributed to adults’
having more experience and better-developed language
skills, including the ability to parse phonetic, semantic, and
lexical aspects of speech 共Fletcher and MacWhinney, 1995兲.
Of interest is the lack of an age effect for the amount of
masking. Previous studies have typically shown that adults
experience reduced masking compared with children 共e.g.,
Buss et al., 1999; Oh et al., 2001; Papso and Blood, 1989;
Hall et al., 2002兲. Although this explanation may not be en-
tirely satisfying, the lack of an age-related masking effect
may be attributed to the task itself. In the current study, using
the 4AFC task, quiet thresholds were extremely low in
adults. In contrast, adults tested on the same measure using
identical stimuli, but with a 25AFC did not reveal such low
FIG. 3. Spatial release from masking values are shown for children and
adults in panels 共A兲 and 共B兲, respectively. Each panel shows values grouped
by competitor type/number condition 共on the x-axis labels SP and Sp-Ns
refer to the speech and speech-noise conditions, respectively兲. Individual
values appear in gray circles, and group averages 共⫾SD兲 are shown in black
circles. When necessary to avoid overlap of data points, in some cases there
was a slight shifting along the x axis.
3096 J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children
SRTs in quiet, but continued to show lower masked SRTs.
The amount of masking in the 25AFC task was therefore
lower in adults than children 共Johnstone and Litovsky, 2005兲.
When increasing task difficulty for adults, a more realistic
story with regard to age-related masking differences may
emerge, suggesting the importance of equating for difficulty
of the task when comparing perceptual abilities across age
groups.
B. Competitor type
SRTs did not differ for the two types of competitors for
children, but were higher with speech-noise than speech for
the adults, which may be in part due to greater statistical
power in the adult within-subjects comparisons. For both age
groups, masking was greater with speech-noise than speech.
These findings are consistent with other findings in adults in
a one-masker paradigm, whereby greater amounts of mask-
ing were reported in the presence of speech-noise compared
with speech 共e.g., Hawley et al., 2004兲. This has been attrib-
uted to greater amounts of overlap in the energies of the
speech-noise masker and the target, resulting in the reduction
of F0 discrimination. However, in previous work, as the
number of maskers increased, speech became a more potent
masker, an explanation involving informational masking and
linguistic interference from multiple speech maskers was in-
voked to account for the increased interference from speech
共e.g., Bronkhorst, 2000; Hawley et al., 2004兲. Here, there
was no interaction of type and number of competitors, which
may be explained by stimulus differences across studies.
Studies such as those of Hawley et al. 共2004兲 typically use
male voices for both the target and competitors, whereas here
the target was a male voice and the competitor was spoken
by a female. The differences in voice pitch, quality, and on-
going F0 differences provided a robust cue for source segre-
gation in the presence of speech competitors, regardless of
the number of competitors. The speech-noise competitor,
having momentary dips in amplitude but no ongoing changes
in frequency, served as a more potent masker whose effect
was greater than that of speech. With same-gender competi-
tors it is highly likely that speech would have produced
masking at least as great, if not larger than the speech-noise
competitor 共e.g., Brungart et al., 2001兲. Finally, the differ-
ences in masking amounts for the child groups may be ac-
counted for by the fact that, for reasons that are not entirely
clear, but probably due to random variation within the popu-
lation, SRTs on the quiet conditions were lower in the two
speech-noise
´
groups than in the groups tested with the
speech competitors.
C. Number of competitors
For both children and adults, masking was significantly
greater for two compared with one competitor共s兲, and the
interactions of number with location 共front versus right兲 were
not significant. Averaged over all competitor types and num-
bers, the addition of a second competing sound resulted in
increased masking of 4.7 dB for children and 4.8 for adults.
Two interpretations can be considered here. First, in the pres-
ence of competitors with envelope modulations such as those
used here, listeners may be better able to take advantage of
the modulations and ‘‘listen in the gaps’’ in the presence of a
single competitor. As a second competitor is added the signal
contains fewer gaps, thereby decreasing opportunities of
‘‘gap listening’’ 共e.g., Festen and Plomp, 1990; Hawley
et al., 2004兲. Second, consider the possible role of ‘‘informa-
tional’’ masking. In recent years this term has been used
extensively in the auditory literature to explain masking phe-
nomena that cannot be attributed solely to peripheral audi-
tory mechanisms 共e.g., Neff and Green, 1987; Lutfi, 1990;
Kidd et al., 2003兲. In the speech intelligibility literature, one
of the conditions under which informational masking has
been thought to occur is when the addition of a second
masker elevates thresholds by more than the 3 dB expected
simply from the added energy in the presence of a second
masker 共e.g., Brungart et al., 2001; Hawley et al., 2004;
Durlach et al., 2003兲. This threshold elevation may result
from the increased complexity of the listening environment,
possibly due to uncertainty on the part of the listener as to
what aspects of the stimulus to ignore and what aspects to
pay attention to. Although difficult to evaluate numerically,
this component of masking may have been present here to
some extent, and more direct tests of the effect in children
would be important to pursue in future studies.
D. Spatial release from masking
Measures of spatial release from masking did not statis-
tically differ across age groups, nor were there effects of
competitor type and number. The only effect was the lack of
spatial release from masking in the one-speech condition in
adults, compared with 5.7 dB in children. The adult data
differ from other free field studies in adults, in which spatial
release from masking for speech was reported to be at least 3
dB for a single competing talker and as high as 12 dB for
multiple talkers 共Bronkhorst, 2000; Hawley et al., 2004兲.
The lack of release from masking found here with the one-
speech competitor is likely due to the nature of the task and
stimuli; the use of a fairly easy 4AFC task in combination
with different-gender talkers for the target and competitor
most likely created a relatively simple listening situation for
adults.
Spatial cues are thought to be especially useful in chal-
lenging conditions when nonspatial cues are difficult to ac-
cess 共Peissig and Kollmeier, 1997; Bronkhorst, 2000;
Durlach et al., 2003; Freyman et al., 2004兲. In the adult
group tested here, spatial cues were beneficial in the condi-
tions that created greater amounts of front masking 共two-
speech, one-speech-noise and two-speech-noise兲. The lack of
a location effect in the one-speech condition is likely due to
the general ease of listening to spondees when the competitor
consists of a single, different-gender talker. In that condition,
spatial cues did not help to reduce masking in the right con-
dition, since masking was already relatively small in that
condition. In contrast with adults, in children the one-speech
front condition did present a challenging situation, probably
because children are less able to take advantage of the
different-gender competitor to hear the target speech. Thus,
spatial cues were indeed relevant to the children so as to
produce a robust improvement in the right condition com-
3097J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children
pared with the front. These findings suggest that, while tasks
that are more complex, using sentence material and/or same-
gender stimuli may be more appropriate for measuring spa-
tial release from masking in adults, the task used here is a
good tool for measuring the ability of young children to ne-
gotiate complex auditory environments.
The finding that, overall, spatial release from masking in
children is similar to that in adults is consistent with work
showing that preschool-age children perform similar to
adults on measures of binaural masking level differences
共Nozza et al., 1988; Moore et al., 1991兲 and minimum au-
dible angle 关Litovsky 共1997兲; for review see Litovsky and
Ashmead 共1997兲兴. This finding implies that for a simple
closed-set task young children are able to utilize spatial
and/or head shadow cues to the same extent as adults in
order to segregate sounds in noisy environments. That is not
to say that children would be expected to perform similar to
adults on all measures of speech intelligibility in noise.
Given recent findings that children exhibit poorer attentional
selectivity on auditory tasks 共e.g., Oh et al., 2001兲, and re-
duced unmasking for tone detection under dichotic condi-
tions 共Wightman et al., 2003; Hall et al., 2004兲, the possibil-
ity remains that age differences would be seen under more
demanding conditions, such as an open-set test or with same-
gender target and competitors. Those differences, however,
would not be attributable to age-dependent binaural abilities,
but rather to other central processes such as auditory atten-
tion.
E. Conclusions
Young children require higher signal levels than adults
to identify spondees in a simple 4AFC task, and these age-
related differences may be mediated by both peripheral and
central auditory processes. The fact that young children can
benefit from spatial separation of the target speech and com-
peting sources suggests that in a complex acoustic environ-
ment, such as a noisy classroom, they might find it easier to
attain information if the source of interest is spatially segre-
gated from noise sources. Although, the extent to which this
is true with real-world sounds may depend on duration, com-
plexity and type of sounds, and the demand on attentional
resources that various sounds may require. Finally, the test
used here 共developed by Litovsky, 2003兲 is designed to also
be used in pediatric clinical settings where young children
are often fitted with hearing aids or cochlear implants, with
little knowledge about the efficacy of the fittings in noisy
environments. This test may offer a way to evaluate the abili-
ties in children with hearing aids and cochlear implants to
function in noisy environments, and may, for example, be
useful in assessing the extent to which children obtain a ben-
efit from bilateral fitting strategies 共Litovsky et al., 2004兲.
ACKNOWLEDGMENTS
The author is grateful to Aarti Dalal and Gerald Ng for
assistance with programming and data collection, and to Patti
Johnstone and Shelly Godar for helping with data analysis.
The author is also grateful to Dr. Joseph Hall for initially
suggesting the use of spondees in a forced choice paradigm,
and to Dr. Adelbert Bronkhorst and an anonymous reviewer
for helpful suggestions during the review process. This work
was supported by NIDCD 共Grant Nos. DC00100 and
DC0055469兲, National Organization for Hearing Research,
and the Deafness Research Foundation. Portions of the data
were collected while R. Litovsky was at Boston University,
Hearing Research Center.
APPENDIX: LIST OF SPONDEE WORDS USED IN THE
PRESENT EXPERIMENT
Hotdog
Ice Cream
Birdnest
Cowboy
Dollhouse
Barnyard
Scarecrow
Railroad
Sidewalk
Rainbow
Cupcake
Birthday
Airplane
Eyebrow
Shoelace
Toothbrush
Hairbrush
Highchair
Necktie
Playground
Football
Baseball
Bluejay
Bathtub
Bedroom
1
The lower limit of 4.5 years is slightly conservative, and was based on pilot
testing which suggested that by that age all children were familiar with the
majority of the target words. The upper limit of 7.5 is somewhat smaller
than the 10-year limit used in a number of other works 共e.g., Oh et al.,
2001; Hall et al., 2002兲, but similar to that used in studies on auditory
attention in young children, in which there do not appear to be develop-
mental effects within the age range 共e.g., Stellmack et al., 1997; Oh et al.,
2001兲.
Allen, P., and Wightman, F. 共1994兲. ‘‘Psychometric functions for children’s
detection of tones in noise,’’ J. Speech Hear. Res. 37, 205–215.
Allen, P., Wightman, F., Kistler, D., and Dolan, T. 共1989兲. ‘‘Frequency reso-
lution in children,’’ J. Speech Hear. Res. 32, 317–322.
American National Standards Institute 共2002兲. ‘‘Standard for acoustical
characteristics of classrooms in the United States,’’ ANSI—S12.60.
Arbogast, T. L., Mason, C. R., and Kidd, G. 共2002兲. ‘‘The effect of spatial
separation on informational and energetic masking of speech,’’ J. Acoust.
Soc. Am. 112, 2086–2098.
Bronkhorst, A. 共2000兲. ‘‘The cocktail party phenomenon: A review of re-
search on speech intelligibility in multiple-talker conditions,’’Acta. Acust.
Acust. 86, 117–128.
Bronkhorst, A. W., and Plomp, R. 共1992兲. ‘‘Effect of multiple speechlike
maskers on binaural speech recognition in normal and impaired hearing,’’
J. Acoust. Soc. Am. 92, 3132–3139.
Brungart, D. S., Simpson, B. D., Ericson, M. A., and Scott, K. R. 共2001兲.
‘‘Informational and energetic masking effects in the perception of multiple
talkers,’’ J. Acoust. Soc. Am. 110, 2527–2538.
3098 J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children
Buss, E., Hall, III, J. W., Grose, J. H., and Dev, M. B. 共1999兲. ‘‘Development
of adult-like performance in backward, simultaneous, and forward mask-
ing,’’ J. Speech Lang. Hear. Res. 42, 844–849.
Crandell, C. C. 共1993兲. ‘‘Speech recognition in noise by children with mini-
mal degrees of sensorineural hearing loss,’’ Ear Hear. 14, 210–216.
Culling, J. F., Hawley, M. L., and Litovsky, R. Y. 共2004兲. ‘‘The role of
head-induced interaural time and level differences in the speech reception
threshold for multiple interfering sound sources,’’ J.Acoust. Soc. Am. 116,
1057–1065.
Dawson, P. W., Decker, J. A., and Psarros, C. E. 共2004兲. ‘‘Optimizing dy-
namic range in children using the nucleus cochlear implant,’’ Ear Hear. 25,
230–241.
Durlach, N. I., Mason, C. R., Shinn-Cunningham, B. G., Arbogast, T. L.,
Colburn, H. S., and Kidd, Jr., G. 共2003兲. ‘‘Informational masking: coun-
teracting the effects of stimulus uncertainty by decreasing target-masker
similarity,’’ J. Acoust. Soc. Am. 114, 368–379.
Eisenberg, L. S., Kirk, K. I., Martinez, A. S., Ying, E. A., and Miyamoto, R.
T. 共2004兲. ‘‘Communication abilities of children with aided residual hear-
ing: comparison with cochlear implant users,’’ Arch. Otolaryngol. Head
Neck Surg. 130, 563–569.
Festen, J. M., and Plomp, R. 共1990兲. ‘‘Effects of fluctuating noise and inter-
fering speech on the speech-reception threshold for impaired and normal
hearing,’’ J. Acoust. Soc. Am. 88, 1725–1736.
Fletcher, P., and MacWhinney, B. 共1995兲. Handbook of Child Language
共Blackwell, Oxford, UK兲.
Freyman, R. L., Balakrishnan, U., and Helfer, K. S. 共2001兲. ‘‘Spatial release
from informational masking in speech recognition,’’ J. Acoust. Soc. Am.
109, 2112–2122.
Freyman, R. L., Balakrishnan, U., and Helfer, K. S. 共2004兲. ‘‘Effect of
number of masking talkers and auditory priming on informational masking
in speech recognition,’’ J. Acoust. Soc. Am. 115, 2246–2256.
Hall, J. W., Buss, E., Grose, J. H., and Dev, M. B. 共2004兲. ‘‘Developmental
effects in the masking-level difference,’’ J. Speech Lang. Hear. Res. 47,
13–20.
Hall, III, J. W., and Grose, J. H. 共1991兲. ‘‘Notched-noise measures of fre-
quency selectivity in adults and children using fixed-masker-level and
fixed-signal-level presentation,’’ J. Speech Hear. Res. 34, 651–60.
Hall, III, J. W., Grose, J. H., and Dev, M. B. 共1997兲. ‘‘Auditory development
in complex tasks of comodulation masking release,’’ J. Speech Lang. Hear.
Res. 40, 946–954.
Hall, III, J. W., Grose, J. H., Buss, E., and Dev, M. B. 共2002兲. ‘‘Spondee
recognition in a two-talker masker and a speech-shaped noise masker in
adults and children,’’ Ear Hear. 23, 159–165.
Hall, III, J. W., Grose, J. H., Buss, E., Dev, M. B., Drake, A. F., and Pills-
bury, H. C. 共2003兲. ‘‘The effect of otitis media with effusion on perceptual
masking,’’ Arch. Otolaryngol. Head Neck Surg. 129, 1056–1062.
Hartley, D. E., Wright, B. A., Hogan, S. C., and Moore, D. R. 共2000兲.
‘‘Age-related improvements in auditory backward and simultaneous mask-
ing in 6- to 10-year-old children,’’ J. Speech Lang. Hear. Res. 43,1402–
1415.
Hawley, M. L., Litovsky, R. Y., and Colburn, H. S. 共1999兲. ‘‘Speech intel-
ligibility and localization in complex environments,’’ J. Acoust. Soc. Am.
105, 3436–3448.
Hawley, M. L., Litovsky, R. Y., and Culling, J. F. 共2004兲. ‘‘The benefit of
binaural hearing in a cocktail party: Effect of location and type of inter-
ferer,’’ J. Acoust. Soc. Am. 115, 833–843.
Johnstone, P., and Litovsky, R. Y. 共2005兲. ‘‘Speech intelligibility and spatial
release from masking in children and adults for various types of interfer-
ing sounds,’’ J. Acoust. Soc. Am. 共in press兲.
Kidd, Jr., G., Mason, C. R., and Richards, V. M. 共2003兲. ‘‘Multiple bursts,
multiple looks, and stream coherence in the release from informational
masking,’’ J. Acoust. Soc. Am. 114, 2835–2845.
Koehnke, J., and Besing, J. M. 共1996兲. ‘‘A procedure note for testing speech
intelligibility in a virtual listening environment,’’ Ear Hear. 17, 211–217.
Knecht, H. A., Nelson, P. B., Whitelaw, G. M., and Feth, L. L. 共2002兲.
‘‘Background noise levels and reverberation times in unoccupied class-
rooms: predictions and measurements,’’ Am. J. Audiol. 11, 65–71.
Litovsky, R. 共1997兲. ‘‘Developmental changes in the precedence effect: Es-
timates of Minimal Audible Angle,’’ J. Acoust. Soc. Am. 102, 1739–1745.
Litovsky, R. 共2003兲. ‘‘Method and system for rapid and reliable testing of
speech intelligibility in children,’’ U.S. Patent No. 6,584,440.
Litovsky, R., and Ashmead, D. 共1997兲. ‘‘Developmental aspects of binaural
and spatial hearing,’’ in Binaural and Spatial Hearing, edited by R. H.
Gilkey and T. R. Anderson 共Earlbaum, Hillsdale, NJ兲, pp. 571–592.
Litovsky, R. Y., Fligor, B., and Tramo, M. 共2002兲. ‘‘Functional role of the
human inferior colliculus in binaural hearing,’’ Hear. Res. 165, 177–188.
Litovsky, R. Y., Parkinson, A.,Arcaroli, J., Peters, R., Lake, J., Johnstone, P.,
and Yu, G. 共2004兲. ‘‘Bilateral cochlear implants in adults and children,’’
Arch. Otolaryngol. Head Neck Surg. 130, 648–655.
Lutfi, R. A. 共1990兲. ‘‘How much masking is informational masking?’’ J.
Acoust. Soc. Am. 88, 2607–2610.
Lutfi, R. A., Kistler, D. J., Oh, E. L., Wightman, F. L., and Callahan, M. R.
共2003兲. ‘‘One factor underlies individual differences in auditory informa-
tional masking within and across age groups,’’ Percept. Psychophys. 65,
396–406.
Moore, D. R., Hutchings, M., and Meyer, S. 共1991兲. ‘‘Binaural masking
level differences in children with a history of otitis media,’’Audiology 30,
91–101.
Moore, D. R., Hartley, D. E., and Hogan, S. C. 共2003兲. ‘‘Effects of otitis
media with effusion 共OME兲 on central auditory function,’’ Int. J. Pediatr.
Otorhinolaryngol. 67, S63–S67.
Neff, D. L., and Green, D. M. 共1987兲. ‘‘Masking produced by spectral un-
certainty with multicomponent maskers,’’ Percept. Psychophys. 41,409–
415.
Nilsson, M., Soli, S. D., and Sullivan, J. A. 共1994兲. ‘‘Development of the
Hearing in Noise Test for the measurement of speech reception thresholds
in quiet and in noise,’’ J. Acoust. Soc. Am. 95, 1085–1099.
Nozza, R. J., Wagner, E. F., and Crandell, M. A. 共1988兲. ‘‘Binaural release
from masking for a speech sound in infants, preschoolers, and adults,’’ J.
Speech Hear. Res. 31, 212–218.
Oh, E. L., Wightman, F., and Lutfi, R. A. 共2001兲. ‘‘Children’s detection of
pure-tone signals with random multiple maskers,’’ J. Acoust. Soc. Am.
109, 2888–2895.
Papso, C. F., and Blood, I. M. 共1989兲. ‘‘Word recognition skills of children
and adults in background noise,’’ Ear Hear. 10, 337–338.
Peissig, J., and Kollmeier, B. 共1997兲. ‘‘Directivity of binaural noise reduc-
tion in spatial multiple noise-source arrangements for normal and impaired
listeners,’’ J. Acoust. Soc. Am. 101, 1660–1670.
Plomp, R., and Mimpen, A. M. 共1981兲. ‘‘Effect of the orientation of the
speaker’s head and the azimuth on a noise source on the speech reception
threshold for sentences,’’ Acustica 48, 325–328.
Roberts, J., Hunter, L., Gravel, J., Rosenfeld, R., Berman, S., Haggard, M.,
Hall, III, J., Lannon, C., Moore, D., Vernon-Feagans, L., and Wallace, I.
共2004兲. ‘‘Otitis media, hearing loss, and language learning: controversies
and current research,’’ J. Dev. Behav. Pediatr. 25, 110–122.
Rothauser, E. H., Chapman, W. D., Guttman, N., Nordby, K. S., Silbiger, H.
R., Urbanek, G. E., and Weinstock, M. 共1969兲. ‘‘IEEE Recommended
Practice for Speech Quality Measurements,’’ IEEE Trans. Audio Electroa-
coust. 17, 227–246.
Shinn-Cunningham, B. G., Schickler, J., Kopco, N., and Litovsky, R. 共2001兲.
‘‘Spatial unmasking of nearby speech sources in a simulated anechoic
environment,’’ J. Acoust. Soc. Am. 110, 1118–1129.
Stellmack, M. A., Willihnganz, M. S., Wightman, F. L., and Lutfi, R. A.
共1997兲. ‘‘Spectral weights in level discrimination by preschool children:
analytic listening conditions,’’ J. Acoust. Soc. Am. 101, 2811–2821.
Uitenbroek, D. G. 共1997兲. ‘‘SISA Binomial,’’ Southampton: D. G. Uiten-
broek. Retrieved 1 January, 2004, from the World Wide Web: http://
home.clara.net/sisa/binomial.htm.
Veloso, K., Hall, III, J. W., and Grose, J. H. 共1990兲. ‘‘Frequency selectivity
and comodulation masking release in adults and in 6-year-old children,’’ J.
Speech Hear. Res. 33, 96–102.
Wichmann, F. A., and Hill, N. J. 共2001a兲. ‘‘The psychometric function: I.
Fitting, sampling, and goodness of fit,’’ Percept. Psychophys. 63,1293–
1313.
Wichmann, F. A., and Hill, N. J. 共2001b兲. ‘‘The psychometric function: II.
Bootstrap-based confidence intervals and sampling,’’ Percept. Psychophys.
63, 1314–1329.
Wightman, F. L., Callahan, M. R., Lutfi, R. A., Kistler, D. J., and Oh, E.
共2003兲. ‘‘Children’s detection of pure-tone signals: informational masking
with contralateral maskers,’’ J. Acoust. Soc. Am. 113, 3297–3305.
Yacullo, W. S., and Hawkins, D. B. 共1987兲. ‘‘Speech recognition in noise
and reverberation by school-age children,’’ Audiology 26, 235–246.
3099J. Acoust. Soc. Am., Vol. 117, No. 5, May 2005 Ruth Y. Litovsky: Speech intelligibility in young children