Content uploaded by James Mahshie
Author content
All content in this area was uploaded by James Mahshie on Nov 13, 2021
Content may be subject to copyright.
JASA Express Lett. 1, 115201 (2021); https://doi.org/10.1121/10.0007057 1, 115201
© 2021 Author(s).
Contrastive stress production by children
with cochlear implants: Accuracy and
acoustic characteristics
Cite as: JASA Express Lett. 1, 115201 (2021); https://doi.org/10.1121/10.0007057
Submitted: 04 June 2021 • Accepted: 12 October 2021 • Published Online: 03 November 2021
James J. Mahshie and Michael D. Larsen
Contrastive stress production by children with cochlear
implants: Accuracy and acoustic characteristics
James J. Mahshie
1,a)
and Michael D. Larsen
2,b)
1
George Washington University, Cochlear Implant Communication Lab, Department of Speech, Language and Hearing
Sciences, Washington, DC 20052, USA
2
Saint Michael’s College, Department of Mathematics and Statistics, Colchester, Vermont 05439, USA
jmahshie@gwu.edu,mlarsen@smcvt.edu
Abstract: The aim of this study was to examine the abilities of eight early-implanted children with cochlear implants (mean
age 7.1 years) to produce contrastive stress and to compare their use of amplitude, duration, and fundamental frequency, to
an age-matched group of children with typical hearing (mean age 6.11 years). A set of 16 utterances were elicited in which the
child was required to stress either an adjective or noun in a short phrase. Although both groups of children produced similar
proportions of utterances with stress patterns perceived by hearing listeners as accurate, they employed different strategies for
achieving contrastive stress. V
C2021 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons
Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
[Editor: Qian-Jie Fu] https://doi.org/10.1121/10.0007057
Received: 4 June 2021 Accepted: 12 October 2021 Published Online: 3 November 2021
1. Introduction
An important aspect of prosody is sentence stress, or the prominence of individual words within a sentence (Meister
et al., 2009). Sentence stress in adults’ speech can be used to show contrastive focus, and can convey significant informa-
tion about the intention of the speaker (Buring, 2012). The production of contrastive stress in adults is the result of alter-
ing three main acoustic elements: amplitude, duration, and fundamental frequency (F0) (Meister et al., 2009).
Children with significant hearing loss have benefited considerably from cochlear implant (CI) technology [for
example, Niparko et al.(2010) and Sharma et al. (2020)]. Cochlear implants (CIs) have provided children with unprece-
dented access to speech, but the devices are limited in the type of acoustic information that is provided to the listener. CIs
divide the speech signal into frequency channels, the output of which is an electrical pulse directly to the spiral ganglion
of the auditory nerve, bypassing the damaged inner hair cells of the cochlea. CIs maintain the essential frequency organi-
zation of the cochlea (lower frequencies near the apex, higher frequency information near the base), thus providing an
approximation of the auditory information that would be provided by an intact cochlea to the auditory nerve.
Typical implant coding strategies in use today provide only limited fine structure information, however, exclud-
ing fundamental frequency (F0) (Geurts and Wouters, 2001) as well as information about other features of speech such as
low frequency phonetic cues (M€
uller et al., 2020). Recent research further suggests that implanted children are only able
to perceive amplitude changes greater than those normally found in speech (Hegarty and Faulkner, 2013).
Although there is evidence that children with CIs (CWCI) can learn to produce certain aspects of speech that
they are less able to perceive through audition [for example, through teaching or visual cues, e.g., Mahshie et al. (2015)
and Barbu (2016)], it is generally acknowledged that the ability to perceive speech features through audition is an impor-
tant precursor to the production of those features. The ability of a child with CIs to learn to produce contrastive stress is
thus particularly intriguing since there is evidence that two of the three parameters typically used to control stress are not
well represented in the signal provided by the implant.
Research examining contrastive stress production in children with typical hearing (CWTH) is limited. Patel and
Brayton (2009) examined stress production by children aged 4, 7, and 11 years. Listeners had greater difficulty perceiving
contrastive stress by the 4-year-olds than for either of the older groups, suggesting that fairly stable prosodic control
occurs in CWTH between the age of 4 and 7.
Patel and Grigos (2006) also examined the production of contrastive intonation patterns by 4-, 7-, and 11-year-
old CWTH. They found that the 4-year-old children relied mainly on durational cues while the 7- and 11-year-old chil-
dren tended to use all three cues to produce these intonation patterns.
a)
ORCID: 0000-0002-1254-3518.
b)
ORCID: 0000-0001-8005-0750.
JASA Express Lett. 1(11), 115201 (2021) V
CAuthor(s) 2021. 1, 115201-1
ARTICLE asa.scitation.org/journal/jel
Previous studies comparing contrastive stress production by CWCI and their typical hearing peers have reported
somewhat contradictory findings. Van De Velde et al. (2019) examined 13 CWCI and an age-matched group of CWTH.
They found that in both the production and perception of contrastive stress, the CWCI performed comparably to their
typical hearing peers.
Other research, however, suggests that CWCI have difficulty producing accurate stress patterns in speech [for
example, Lenden and Flipsen (2007),Lyxell et al. (2009), and O’Halpin (2010)]. Lenden and Flipsen examined prosody
and voice characteristics of six CWCI during conversational speech production and found that stress production was the
most notable prosodic challenge for the CWCI—they were able to accurately produce stress for only 61% of the utterances.
Lyxell et al. (2009) examined prosody production of eight CWCI and found that they had significantly poorer performance
on the measures of production prosody at word and phrase level than their hearing peers. O’Halpin’s findings were consis-
tent with these previous studies - the CWCI tended to exhibit poorer ability to signal stress than their hearing peers.
These earlier studies, however, examined children implanted at a later age. Speech and language outcomes tend to be
poorer for children implanted after 3 years of age, although other factors can also play a role in speech outcomes
(Duchesne and Marschark, 2019;Connor et al., 2006). Kalathottukaren et al. (2017) also reported poorer stress production
accuracy for a group of children with hearing loss, although their study examined a combination of both hearing aid users
(N ¼9) and cochlear implant users (N ¼6) making it difficult to draw direct conclusions about stress production in
cochlear implant users.
In addition to the somewhat contradictory findings related to stress production accuracy by CWCI, there is little
information available concerning how these children manage duration, amplitude and F0 to effect contrastive stress and
how their production of stressed vs unstressed utterances compare to CWTH, suggesting the need for additional research
on contrastive stress in these groups of children.
The present study was designed to address the following questions:
(1) Do CWCI and their typically developing peers exhibit similar levels of accuracy when producing contrastive stress?
(2) How do CWCI manage duration, amplitude, and pitch when producing contrastive stress and are variations in these three
production parameters similar for implanted and typically developing children?
2. Procedures
2.1 Participants
Two groups of children who met the following inclusionary and exclusionary criteria were recruited from the metropolitan
Washington DC area; the first group was comprised of eight CWCI, identified at or near birth with severe to profound hearing
loss (see Table 1).TheCWCIwerebetween5.4yearsand8.3yearsofage,withameanageof7.1years.Sixchildrenwere
implanted prior to 16months while the remaining two children were implanted at 24 and 28months of age. No additional dis-
abilities were reported. The duration of usage ranged from 3;2 to 7;8 years with a mean duration of 5;6 years of implant use.
The second group was comprised of an age-matched group of eight CWTH with no reported speech, language,
or hearing issues (mean age, 6.9, range 5.7–8.4 years). A two sample t-test revealed no significant difference in chronologi-
cal age between the two groups of children [t(14) ¼0.24, p ¼0.82, two tailed].
2.2 Data collection procedure
Production of utterance with contrastive stress were elicited using the focus subtest of the Profiling Elements of Prosodic
Systems—Children (PEPS-C) developed by Pepp
e and McCann (2003). In this subtest the children were presented a
Table 1. Demographic and implant characteristics of children with cochlear implants.
Participant Gender
Test
Age
(yr-mo)
1st
Implanted
Ear
First CI
Activation
Age (months) First CI Model
Duration of
CI usage
(years;mos)
2nd
Implanted
Ear
Second CI
Activation
Age (months)
Second
CI Model
1 F 7y Right 12 Nucleus-Freedom 6;1 Left 36 Nucleus Freedom
2 M 5y 6m Left 10 Nucleus Freedom 4;10 Right 10 Nucleus Freedom
4 M 5y 5 m Left 28 Nucleus Freedom 3;2 Right 48 Oticon Safari
(Hearing aid)
6 M 8y 4 m Right 13 Nucleus Freedom 7;4 Left 19 Nucleus Freedom
7 M 7y 11 m Left 13 Freedom 6;11 Right 25 Freedom
8 M 7y 8 m Left 24 Nucleus Freedom 5;9 Right 26 Nucleus Freedom
9 F 6y 3 m Right 16 Advanced
Bionics-Harmony
5;0 Left 19 Advanced
Bionics Harmony
10 F 8y 4 m Left 9 Nucleus - Freedom 7;8 Right 13 Nucleus Freedom
ARTICLE asa.scitation.org/journal/jel
JASA Express Lett. 1(11), 115201 (2021) 1, 115201-2
picture and a narrator incorrectly described the picture by providing a counterfactual description. The child was asked to
correct what the narrator has said. For example, the narrator would prompt a picture of a red cow and a soccer ball with
“The blue cow has the ball.” The child is expected to respond with “No, the RED cow has the ball.” Conversely, the narra-
tor might say “The red sheep has the ball,” with the expected response being “No, the red COW has the ball.” There were
a total of 16 stimuli resulting in 8 stressed and 8 unstressed adjectives (color) and nouns (animal). All testing sessions
were audio-video recorded.
2.3 Accuracy determination
A panel of three normal hearing adults listened to the recordings of each of the child’s productions and judged the stress
pattern associated with the target adjective-noun clause. Judges could indicate the following: (i) the adjective was stressed,
(ii) the noun was stressed, (iii) the production was ambiguous re the stressed word. Where there was agreement, the con-
sensus judgements were considered the stressed syllable. The consensus judgement of each utterance was subsequently
compared to the “correct” stress pattern and scored as correct if the produced stress pattern matched the target, and as
incorrect if the produced pattern did not match the target or was ambiguous. The proportion of utterances perceived
as accurately produced were subsequently calculated individually for intended adjective stress and intended noun stress, as
well as for overall perceived accuracy.
2.4 Acoustic measures
The acoustic beginning and end of the vocalic portion of the adjective and noun (color and animal) in each utterance
were located and labeled using PRAAT (Boersma and Weenink, 2019). The duration, mean amplitude and mean F0 of
the vocalic portion of each target word was then measured. To quantify the contrastive use of Duration and
Amplitude in each utterance, we calculated a difference between the adjective and noun to yield D
dur
and D
Inten.
To account for any interspeaker variability in overall F0, we calculated the semitone difference in F0 between the
adjective and noun—D
F0-semi
.
2.5 Statistical analysis
Acoustic factors contributing to accurate productions by CWCI and those with typical hearing were statistically examined
in R(R Core Team 2021). Only utterances that were perceived as having accurate stress patterns were included. The analy-
sis employed a mixed-effects logistic regression analysis in which the difference in F0, duration, and intensity between the
stressed and unstressed words were used to predict the stress pattern in the elicited utterances (adjective or noun).
Random effects were used for children as there were repeated measures of each child. The analysis was conducted
separately for the two groups of children.
3. Results
3.1 Accuracy
A t-test comparing accuracy rates (one measure per child) revealed no significant difference in overall accuracy between
the cochlear implant (M ¼80%, SD ¼14%) and typically hearing group (M ¼88%, SD ¼15%; t(14) ¼–1.09, p ¼0.29, two-
tailed, Cohen’s D ¼0.56, medium effect). A logistic regression mixed effects model predicting accuracy on each trial (fixed
effect ¼hearing status, random effect ¼children) confirmed the similarity of the two groups (p¼0.25). Age was not a sig-
nificant predictor, and its inclusion in the model did not change the result for the effect of hearing status. The odds ratio
for accuracy comparing TH to CI groups is 1.96 (95% interval 0.63, 6.74). An odds ratio below 2.0 can be considered a
small to medium effect (Chen et al., 2010).
Contrastive stress accuracy was further examined separately for the intended stressed adjectives and nouns (see
Table 2). The accuracy rates for the CWTH were similar for both adjectives and nouns—adjectives ¼88% and
nouns ¼87%. The accuracy rates for the CWCI were 86% for adjectives and 73% for nouns. Using the mixed-effects logis-
tic regression model to predict accuracy for hearing status and the word stressed revealed no significant interaction
(p ¼0.25), nor effect for hearing status (p ¼0.23) or word stressed (p ¼0.16).
3.2 Acoustic characteristics of stressed vs unstressed words
The impact of Hearing on the relationship between which word in a sentence (adjective or noun) is stressed and the three
acoustic variables was examined through two logistic mixed effect models employing the three predictor variables for each
utterance: D
Dur
,D
Intensity
, and D
F0-semi
. Each subject, except for one, had 16 trials; one child with typical hearing had 15
trials. Random effects were used for subjects as each subject was involved in multiple trials.
Only productions perceived as accurate (that is, matching the intended stress pattern) were used in this analysis
so the number of observations used in analysis varied by subject, yielding 213 total accurate trials: 111 for the CWTH and
102 for the CWCI (pitch measurements could not be made for one CWCI trial). This resulted in 212 usable trials.
ARTICLE asa.scitation.org/journal/jel
JASA Express Lett. 1(11), 115201 (2021) 1, 115201-3
Shown in Fig. 1are boxplots for D
Dur
,D
Intensity
, and D
F0-semi
. Data are shown separately for each group of chil-
dren, and for the intended stressed adjective vs stressed noun. Based on the way that the difference (D) measures were
derived, a positive value indicates that the adjective measure was greater than the noun measure while negative value indi-
cates that the noun parameter was greater than the adjective parameter.
Both groups tended to produce the stressed adjectives with greater durational differences than that for stressed
nouns. The magnitude of D
Dur
when the adjective was stressed was similar for the two groups—the median difference was
87.7 ms for the CWCI and 87.6 ms for the CWTH. When the noun was stressed, however (e.g., blue COW), both groups
demonstrated limited difference between the adjective and noun duration—the median difference was 14.8 ms for the CI
group and 11.8 ms for the CWTH.
Both groups tended to produce utterances having stressed adjectives with relatively greater intensity than that for
the unstressed noun (median D
Intensity
for CWCI ¼6.17 dB, CWTH ¼4 dB). When the noun was stressed, however, the
amount of intensity difference was less evident for the CWCI (median D
Intensity
¼2 dB) while the CWTH tended to pro-
duce both the adjective and noun with similar D
Intensity
(median ¼0 dB).
Both groups tended to utilize relatively large D
F0-semi
differences to distinguish between stressed words. When
the adjective was stressed, both groups utilized a relatively large D
F0-semi
as reflected in positive semitone differences
(median D
F0-semi
for CWCI ¼2.116; CWTH ¼2.504 semitones). When the noun was stressed, the semitone difference
reflected a higher F0 for the noun than the adjective (median D
F0-semi
for CWCI ¼–0.3866; CWTH ¼–3.166 semitones).
Both groups exhibited considerable variability in the D
F0-semi
measure when the noun was stressed, suggesting different
strategies for accomplishing this distinction.
3.3 Acoustic factors contributing to production of contrastive stress in children with typical hearing
The impact of hearing status on the relationship between which word in a sentence (adjective or noun) is stressed and the
three acoustic variables was examined through separate logistic mixed effect models. In the typical hearing group, D
Dur
,
D
Intensity
, and D
F0-semi
were all significant predictors of whether the adjective or noun was perceived as stressed. Table 3
presents coefficients in the logistic regression models. Effect size is the odds ratio for a one standard deviation increase in
the predictor variable; for CWTH the effects for difference in duration and intensity are moderate and for pitch is very
large.
The predicted probability that the adjective or noun is stressed can be computed for each trial and used to clas-
sify a trial as likely to have stress on the adjective or noun. When the prediction is compared to the perceived judgement
of the trial, the concurrence rate is 90%: (48 þ52)/111. When added to the model, age was not a significant predictor
(p ¼0.55) and did not alter results appreciably.
3.4 Acoustic factors contributing to the production of contrastive stress in children with cochlear implants
A similar direct logistic regression was performed to assess the impact of duration, amplitude, and F0 differences on the
production of a stressed or unstressed word by the cochlear implant group. Again, the dependent variable was the location
Table 2. Duration, F0, and Intensity difference measures between adjective and noun for accurately stressed utterances produced by CI and
TH groups.
Adjective stressed Noun stressed
Hearing status Mean Std. Deviation Mean Std. Deviation
CWCI 86% 35.0% 73% 44.5%
CWTH 88% 33.3% 87% 33.6%
Total 87% 34.1% 80% 39.9%
Fig. 1. Boxplots of difference values for predictor variables. Shown for CWCI (blue) and CWTH (black) are duration, F0, and semitone differ-
ences between the adjective and noun. Plots are shown separately for stressed adjectives and nouns.
ARTICLE asa.scitation.org/journal/jel
JASA Express Lett. 1(11), 115201 (2021) 1, 115201-4
of the stressed syllable (adjective or noun) and the model contained three independent predictor variables—the difference
in duration, mean amplitude, and the semitone difference in mean F0 between the adjective and noun.
1
In the cochlear implant group, Duration and Pitch are significant predictors of a stressed adjective (versus the
noun) (see Table 2). Difference in Duration and in Pitch are statistically significant predictors, but D
Intensity
is not. Effect
sizes are very large, except for D
Intensity
, which is small. When the predicted probability was compared to the perceived
judgement of the trial, the concurrence rate is 86%: (42 þ45)/101. When added to the model, age was not a significant
predictor (p ¼0.69) and did not alter results appreciably.
4. Discussion
Earlier research has presented a somewhat conflicting picture of contrastive stress production in CWCI [for example, see
the differences among Van de Velde et al (2019),Lenden and Flipsen (2007),O’Halpin (2010), and Kalathottukaren et al.
(2017)]. For example, the participants of Lenden and Flipsen (2007) were younger than those in the present study, with
less experience with their devices. O’Halpin (2010) examined children whose ages ranged from 5;9 to 17;10. The group
had considerably less implant experience that the participants in the present study and only 5/17 CWCI were implanted
before 3 years of age.
The present findings showed a degree of similarity between the cochlear implant and typically hearing groups in
both the overall production accuracy of stressed vs unstressed words, and the stress of adjectives and nouns in the elicited
production task employed. Van De Velde et al. (2019) also reported that their CWCI had production accuracy similar to
that of their hearing peers. Earlier implant age may have played a role in their findings—ten of the 13 children examined
were implanted before 3 years of age, a factor that may account for the similarity between this study and the present
findings.
The finding of similar levels of accuracy between the CWCI and their typically hearing peers is consistent with
some previous research examining intonation patterns (Barbu, 2016), but is inconsistent with other research examining F0
distinctions for questions and statements (Peng et al., 2008) and tones (Zhou et al., 2013). There are, however, important
differences between the earlier-implanted CI group in the present study, and the range of implant age, duration of use,
concurrent amplification use, and other participant attributes that might account for the differences observed.
Van De Velde et al. suggested that because they found stress production accuracy to be similar between CWCI
and CWTH, that the two groups likely employ similar approaches to producing accurate stress in this task. To better
understand how CWCI and their hearing peers produce the distinction between stressed and unstressed words, acoustic
measures were examined, and regression analyses were conducted.
Comparison of the duration and F0 characteristics of stressed and unstressed words revealed similarities between
the two groups of children. Both groups tended to produce stressed words with longer durations and higher F0 than was
observed in the unstressed words in an utterance. This was the case regardless of whether the adjective or noun was being
stressed. The regression findings, however, suggest, differences between the CWCI and the typical hearing children in the
role of intensity adjustments for production of contrastive stress. The typical hearing group tended to produce stressed
adjectives with little intensity difference between the adjective and noun, while the cochlear implant group, tended to con-
trast amplitude for a stressed adjective by reducing the amplitude of the noun, a pattern that differed from that of their
typical hearing peers.
The regression analysis revealed that the CWTH tended to use F0, duration, and intensity to produce a contrast.
This is consistent with the finding of Patel and Grigos (2006) that CWTH in this age range tend to use all three parame-
ters (F0, duration and intensity) to realize another prosodic contrast—yes/no questions and statements. The CWCI group,
however, did not systematically use intensity to distinguish between stressed adjectives and nouns.
It is generally acknowledged that auditory perception of speech is a necessary prerequisite for accurate produc-
tion (Lowenstein and Nittrouer, 2019). The use of F0 for conveying contrastive stress by the cochlear implant group might
at first appear somewhat contrary to expectations since current CIs do not directly code fundamental frequency informa-
tion. While implants can simulate the frequency-to-place mapping performed by the normal cochlea, this is limited by the
Table 3. Fixed-effects coefficients, standard errors, p-values and effect size (odds ratio or OR for one standard deviation increase in predictor)
for predicting Stress in TH and CI subjects.
CWTH CWCI
Estimate Standard Error p-value Effect size (OR) Estimate Standard Error p-value Effect size (OR)
Intercept 0.93 0.46 0.044 0.62 0.45 0.17
Difference in Duration 6.11 2.14 0.004 2.87 12.23 3.51 <0.001 8.22
Difference in Intensity 0.19 0.08 0.019 3.25 0.07 0.07 0.326 1.53
Difference in Pitch 0.48 0.11 <0.001 10.96 0.47 0.14 <0.001 10.61
ARTICLE asa.scitation.org/journal/jel
JASA Express Lett. 1(11), 115201 (2021) 1, 115201-5
number of electrodes, the insertion depth, and the interactions of electrical current by adjacent electrodes. Much less
robust pitch information is available, however, through temporal fluctuations in the spectral envelope or in the pulse rate
being transmitted through the device (Mehta and Oxenham, 2017). This suggests that earlier implanted pediatric implant
users may be able to develop adequate pitch production performance through the limited pitch information available from
the implant.
The reduced role of amplitude variation for producing contrastive stress is likely related to the inherent limita-
tions of CIs in providing amplitude detail. Hegarty and Faulkner (Hegarty and Faulkner, 2013) reported that the majority
of the nine CWCI that they studied were only able to perceive amplitude changes greater than those typically found in
speech. A similar limitation was reported by Saki et al. (2015) who examined amplitude discrimination in adult cochlear
implant users. They reported poorer amplitude discrimination between 500 Hz and 4 KHz by CI users than that exhibited
by typically hearing listeners. These findings suggest that the amplitude differences involved in contrastive stress may not
be accessible to CWCI.
An alternate explanation concerns the relationship between production and perception. It is typically argued that
perception must precede production. Lowenstein and Nittrouer (2019) suggest that the quality of the input signal con-
strains the speech production capacities of young children. They conclude that the production problems of children with
hearing loss can be explained to some extent by the degradation in the signal they hear. While the present findings do not
refute this notion, the do suggest that experience with both production and perception likely play a role in the eventual
accuracy of contrastive stress.
There is evidence in CWTH that production can exceed perceptual abilities (Chen, 2014). There is also evidence
that through continued maturation and training, the ability to detect amplitude differences can improve (Saki et al., 2015).
With improved discrimination skills, it is reasonable to predict that production distinctions might ensue. Additional
research is needed examining CWCIs’ development of prosody skills perception, and production of contrastive stress.
The present findings are limited. The children were from middle class families with higher levels of maternal
education, a factor shown to correspond to more successful outcomes in CWCI. The relatively early age of implant may
also explain why the accuracy was comparable to that of the CWTH. Research has shown that early age of implant can
have a positive impact on speech outcomes [for example, Damm et al. (2019)].
Finally, the present study examined a relatively small number of children. The sample size was comparable to
previous research [for example, Lenden and Flipsen (2007) and Lyxell et al. (2009)]. Moreover, the present findings
revealed a moderate to strong effect size supporting the statistical findings. Research examining a larger number of chil-
dren from more diverse background will undoubtedly add to our understanding of stress production in CWCI.
Acknowledgements
This work was conducted under a grant from the National Institute on Disability, Independent Living, and Rehabilitation
Research (NIDILRR Grant No. H133G120272).We are grateful to the parents and children who participated in this project.
Thanks also to the various students who were involved with this project, particularly Kimberley Preminger whose undergrad
honor’s thesis contributed to this project.
References and links
1
Age of implant (p ¼0.14), and duration of implant use (p ¼0.82) were not significant predictors and so were not included in the model.
Barbu, I. (2016). “Listener ratings and acoustic characteristics of intonation contours produced by children with cochlear implants and
children with normal hearing,” Master’s thesis, George Washington University, Washington, DC.
Boersma, P., and Weenink, D. (2019). “Praat:Doing phonetics by computer,” version 6.1.08.
Buring, D. (2012). “Focus and intonation,” in Focus and Intonation (Routladge, London), pp. 103–115.
Chen, A. (2014). “Production-comprehension (A)symmetry: Individual differences in the acquisition of prosodic focus-marking,”
Proceedings of the International Conference on Speech Prosody, pp. 423–427.
Chen, H., Cohen, P., and Chen, S. (2010). “How big is a big odds ratio? Interpreting the magnitudes of odds ratios in epidemiological
studies,” Commun. Stat.: Simul. Comput. 39(4), 860–864.
Connor, C. M. D., Craig, H. K., Raudenbush, S. W., Heavner, K., and Zwolan, T. A. (2006). “The age at which young deaf children
receive cochlear implants and their vocabulary and speech-production growth: Is there an added value for early implantation?,” Ear
Hear. 27(6), 628–644.
Damm, S. A., Sis, J. L., Kulkarni, A. M., and Chatterjee, M. (2019). “How vocal emotions produced by children with cochlear implants
are perceived by their hearing peers,” J. Speech Lang. Hear. Res. 62(10), 3728–3740.
Duchesne, L., and Marschark, M. (2019). “Effects of age at cochlear implantation on vocabulary and grammar: A review of the evidence,”
Am. J. Speech Lang. Pathol. 28, 1673–1691.
Geurts, L., and Wouters, J. (2001). “Coding of the fundamental frequency in continuous interleaved sampling processors for cochlear
implants,” J. Acoust. Soc. Am. 109(2), 713–726.
Hegarty, L., and Faulkner, A. (2013). “The perception of stress and intonation in children with a cochlear implant and a hearing aid,”
Cochlear Implants Int. 14, 35–40.
ARTICLE asa.scitation.org/journal/jel
JASA Express Lett. 1(11), 115201 (2021) 1, 115201-6
Kalathottukaren, R. T., Purdy, S. C., and Ballard, E. (2017). “Prosody perception and production in children with hearing loss and age-
and gender-matched controls,” J. Am. Acad. Audiol. 28(4), 283–294.
Lenden, J. M., and Flipsen, P. (2007). “Prosody and voice characteristics of children with cochlear implants,” J. Commun. Disorders
40(1), 66–81.
Lowenstein, J. H., and Nittrouer, S. (2019). “Perception-production links in children’s speech,” J. Speech Lang. Hear. Res. 62(4),
853–867.
Lyxell, B., Wass, M., Sahl
en, B., Samuelsson, C., Asker-
Arnason, L., Ibertsson, T., M€
aki-Torkko, E., Larsby, B., and H€
allgren, M. (2009).
“Cognitive development, reading and prosodic skills in children with cochlear implants,” Scand. J. Psychol. 50(5), 463–474.
Mahshie, J., Core, C., and Larsen, M. (2015). “Auditory perception and production of speech feature contrasts by pediatric implant
users,” Ear Hearing 36(6), 653–663.
Meister, H., Landwehr, M., Pyschny, V., Walger, M., and Wedel, H. V. (2009). “The perception of prosody and speaker gender in
normal-hearing listeners and cochlear implant recipients,” Int. J. Audiol. 48(1), 38–48.
Mehta, A. H., and Oxenham, A. J. (2017). “Vocoder simulations explain complex pitch perception limitations experienced by cochlear
implant users,” J. Assoc. Res. Otolaryngology 18(6), 789–802.
M€uller, V., Kl€unter, H. D., F€urstenberg, D., Walger, M., and Lang-Roth, R. (2020). “Comparison of the effects of two cochlear implant
fine structure coding strategies on speech perception,” AJA 29, 226–235.
Niparko, J. K., Tobey, E. A., Thal, D. J., Eisenberg, L. S., Wang, N. Y., Quittner, A. L., and Fink, N. E. (2010). “Spoken language develop-
ment in children following cochlear implantation,” J. Am. Medical Assoc. 303(15), 1498–1506.
O’Halpin, R. (2010). “The perception and production of stress and intonation by children with cochlear implants,” Ph.D. dissertation,
University of London, London.
Patel, R., and Brayton, J. T. (2009). “Identifying prosodic contrasts in utterances produced by 4-, 7-, and 11-year-old children,” J. Speech
Lang. Hear. Res. 52, 790–801.
Patel, R., and Grigos, M. I. (2006). “Acoustic characterization of the question-statement contrast in 4, 7 and 11 year-old children,”
Speech Commun. 48(10), 1308–1318.
Peng, S. C., Tomblin, J. B., and Turner, C. W. (2008). “Production and perception of speech intonation in pediatric cochlear implant
recipients and individuals with normal hearing,” Ear Hear. 29(3), 336–351.
Pepp
e, S., and McCann, J. (2003). “Assessing intonation and prosody in children with atypical language development: The PEPS-C test
and the revised version,” Clin. Ling. Phon. 17(4–5), 345–354.
R Core Team (2021). “R: A language and environment for statistical computing,” R Foundation for Statistical Computing, Vienna,
Austria, https://www.R-project.org/ (Last viewed 10/12/2021).
Saki, N., Karimi, M., Mirahmadi, S., Rostami, M. R., and Nikakhlagh, S. (2015). “Evaluation of loudness perception performance in
cochlear implant users,” Biomed. Pharmacol. J. 8, 79–83.
Sharma, S. D., Cushing, S. L., Papsin, B. C., and Gordon, K. A. (2020). “Hearing and speech benefits of cochlear implantation in children:
A review of the literature,” Int. J. Pediatric Otorhinolaryng. 133, 109984.
Van De Velde, D. J., Schiller, N. O., Levelt, C. C., Van Heuven, V. J., Beers, M., Briaire, J. J., and Frijns, J. H. M. (2019). “Prosody percep-
tion and production by children with cochlear implants,” J. Child Lang. 46(1), 111–141.
Zhou, N., Huang, J., Chen, X., and Xu, L. (2013). “Relationship between tone perception and production in prelingually deafened chil-
dren with cochlear implants,” Otol. Neurotol. 34(3), 499–506.
ARTICLE asa.scitation.org/journal/jel
JASA Express Lett. 1(11), 115201 (2021) 1, 115201-7
