Neuroscience Research 54 (2006) 276–280

www.elsevier.com/locate/neures

The right hemisphere of sleeping infant perceives

sentential prosody
Fumitaka Homae a,b,*, Hama Watanabe a,b, Tamami Nakano b,
Kayo Asakawa a, Gentaro Taga b,a
a
Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Agency,
Kawaguchi Center Building, 4-1-8 Honcho, Kawaguchi-shi, Saitama 332-0012, Japan
b
Graduate School of Education, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Received 26 October 2005; accepted 15 December 2005
Available online 20 January 2006

Abstract
Behavioral studies proposed that prosodic information in speech sounds plays important roles for human infants to acquire their native
languages. Here, we examined the neural basis of prosodic processing in 3-month-old infants. In order to obtain hemodynamic responses with high
signal-to-noise ratio, we used near-infrared optical topography in the infants while they were in quiet sleep. First, we observed bilateral activation
under each of the normal and flattened speech conditions. The flattened speech sound was created by eliminating changes in the pitch contours of
the original utterance. In a direct comparison between the two conditions, the right temporoparietal region showed more prominent activation to
normal speech sounds than to flattened speech sounds. This result demonstrates that the localized region of the right hemisphere in 3-month-old
infant is involved in the processing of pitch contours. Our findings suggest that prosodic processing in the right hemisphere may facilitate the
acquisition of lexical or syntactic knowledge in the early stages of language development.
# 2005 Elsevier Ireland Ltd and the Japan Neuroscience Society. All rights reserved.

Keywords: Infant; Intonation; Language acquisition; Near-infrared optical topography; Prosody; Speech perception

1. Introduction strapping hypothesis, wherein infants could use prosodic

Language acquisition is one of the unique features of human
nature. Human infants begin to process language inputs prior to
the onset of speech production (Jusczyk, 1997; Werker and
Tees, 1999; Kuhl, 2004). Among various types of information
related to language inputs, sentential prosody is considered to
be essential for neonates and infants to acquire their native
languages (Fernald and Kuhl, 1987; Mehler et al., 1988;
Mandel et al., 1994). The attributes of speech prosody are
mainly served by three acoustic parameters of pitch (funda-
mental frequency), duration and amplitude. The information
from these acoustic parameters provides cues to find phrase
boundaries and helps listeners decode syntactic structures of
sentences (Cutler et al., 1997). Taking into consideration the
close correspondence between prosodic and syntactic proper-
ties, Gleitman and Wanner formulated the prosodic boot-
* Corresponding author. Tel.: +81 3 5841 2797; fax: +81 3 5841 1396.
E-mail address: fhomae@p.u-tokyo.ac.jp (F. Homae).
0168-0102/$ – see front matter # 2005 Elsevier Ireland Ltd and the Japan Neuroscience Society. All rights reserved.
doi:10.1016/j.neures.2005.12.006
information to acquire valuable knowledge about the syntactic
organization of their native language (Gleitman and Wanner,
1982). This hypothesis has been supported and developed in
recent behavioral studies on prosodic processing in infants
(Morgan and Demuth, 1996; Jusczyk, 1997).
A critical question here is which cortical region is related to
prosodic processing in infants. Lateralization and localization
in the infant brain for processing sentential prosody has never
been demonstrated; this is partly because measurement tools
that have been applied to infants (e.g., electroencephalogram
(EEG) and magnetoencephalogram (MEG)) (Cheour et al.,
2000; Kujala et al., 2004) are less apt at revealing localization
and detecting cortical responses to stimuli lasting several
seconds. With regard to lateralization in adult, there appear to
be a number of positions in previous clinical and neuroimaging
studies (Baum and Pell, 1999). One major hypothesis is that the
right hemisphere is related to emotional or paralinguistic
prosody. Recent studies have confirmed the involvement of the
right hemisphere in at least some aspects of pitch processing
(Zatorre et al., 1992; Johnsrude et al., 2000; Meyer et al.,
 F. Homae et al. / Neuroscience Research 54 (2006) 276–280 277

2004). The prosodic processing in the right hemisphere is
supposed to interact with syntactic processing at the phrasal or
sentential level (Baum and Dwivedi, 2003; Herrmann et al.,
2003). In contrast, the involvement of the left hemisphere in
the processing of linguistic prosody including lexical tones
is still under debate (Wong, 2002). The left-lateralized
activation was reported when adult subjects were required
both prosodic and linguistic processing, suggesting that
activation in the left hemisphere is task dependent (Gandour
et al., 2004). Based on previous findings on prosodic pro-
cessing, we expect that several-month-old infants have some
neural mechanisms for processing sentential prosody. If the
right hemisphere of infants is related to prosodic processing,
we can consider that the neural mechanisms affect cortical
regions involved in other aspects of language processing as
shown in adults.
In the present study, we used a near-infrared optical
topography (OT) (Maki et al., 1995), which is an effective
method to examine the cortical hemodynamic responses in
infants to stimuli lasting several seconds (Taga et al., 2003).
Here, we investigated the cortical activation of 3-month-old
infants; behavioral studies have reported that at least 2- or 4-
month-old infants are sensitive to sentential prosody (Fernald
and Kuhl, 1987; Mandel et al., 1994). The infants were
studied while they were in quiet sleep. In recent studies,
neural responses of sleeping infants have been reported
by using auditory mismatch negativity (MMN) (Cheour
et al., 2000). Adult subjects were also examined by using
functional magnetic resonance imaging (fMRI), and the
activation pattern for speech sounds in the bilateral temporal
regions was similar in wakefulness and sleep (Portas et al.,
2000). Thus, we considered that the OT recordings during
quiet sleep can detect cortical responses to auditory stimuli,
and further that the OT recordings under the sleeping state
provide long time and motion-free data with high signal-to-
noise ratio. In the present study, we focused on the variation
in the levels of oxyhemoglobin ([oxy-Hb]) in the OT data,
and identified the cortical regions that were responsive
to each of the normal and flattened speech conditions
(Fig. 1A; see Section 2) by performing t-tests for each
measurement channel. We further performed direct com-
parisons between the two conditions to identify the cortical
regions that are related to the processing of pitch contours in
sentences.
2. Materials and methods
2.1. Subjects
Twenty-one full-term healthy Japanese infants participated in the present
study (girls: 12, boys: 9; mean age: 107.6 days; range: 92–118 days). An
additional 10 infants were studied, but excluded from the analysis due to
their awakening during the experiment (n = 5), head movements producing
large motion artifacts in the signals (n = 3) and experimental error (n = 2).
Informed consent was obtained from the parents of the infants prior to the
initiation of the experiments. The study was approved by the Department of
Physical and Health Education, Graduate School of Education (University of
Tokyo).

Fig. 1. Measurement of cortical activation under the normal and flattened conditions. (A) An example of pitch contours of a sentence. (B) Arrangement of
measurement channels over the frontal and temporoparietal cortices. (C) Averaged time course of [oxy-Hb] changes among all the channels under the two conditions.
The zero point of time was set at the onset of stimulus presentation. (D) Cortical activation under the normal (blue) and flattened (red) conditions. Large- and medium-
filled circles indicate channels that surpassed p < 0.0001 and 0.001, respectively. Small filled circles ( p < 0.01) are shown for the purpose of display.
278 F. Homae et al. / Neuroscience Research 54 (2006) 276–280
2.2. Stimuli
The stimuli consisted of 36 Japanese sentences, which were used in a
previous behavioral study (Nazzi et al., 1998). Two female Japanese speakers
recorded each sentence (16 bit, 22,025 Hz). The mean duration of the stimuli
was 4.1 s (range: 3.1–4.9 s). These recorded utterances were presented under a
normal condition (Fig. 1A). We used the Praat software (Institute of Phonetic
Sciences, University of Amsterdam) to obtain the stimuli in a flattened con-
dition from the stimuli used in the normal condition. In the flattened condition, a
contour of pitch in each utterance was replaced with the mean value of the pitch
of the utterance in the normal condition (Fig. 1A). This process eliminated the
intonation of sentences while preserving formant and spectral information (e.g.,
syllabic information) (Meyer et al., 2004). Thus, adult Japanese speakers could
recognize every word of all the sentences and could understand every sentence.
The duration and intensity of each utterance were maintained in both the
conditions.
2.3. Stimulus presentation
We conducted the experiments in a sound-attenuated room. The infants
were held in an experimenter’s arms, and they were studied while they were
in the quiet sleep state. They were almost motionless and slept soundly
throughout the experimental sessions. The speech sounds were presented at
a maximum amplitude of 65 dB by using a BOSE MMS-1 speaker system,
which was placed in front of the infant. We presented 12 sentences per
condition for each infant. The presentation of sentences was counterba-
lanced among the infants in order to use all the sentences under both
conditions. To reduce the habituation effects of auditory stimuli on cortical
activation, we employed an event-related paradigm with variable inter-
stimulus intervals (12, 13, 14, or 15 s). No sound was presented during
the inter-stimulus intervals.
2.4. OT recordings
We used two near-infrared OT instruments (ETG-100, Hitachi Medical
Corporation, Tokyo, Japan), one for each hemisphere. We used the same
recording parameters that were applied to 2- to 4-month-old infants in our
previous study (Taga et al., 2003), except for the intensity of illumination
(0.6 mW). We placed the measurement channels over the frontal and tempor-
oparietal areas of each hemisphere (Fig. 1B). The measured region in each
hemisphere was correctly positioned by using landmarks of external auditory
pores and the vertex in each infant; the columns of measurement channel 15
(CH15) and CH20 were set on a coronal line through the external auditory
pores. The right CH20 was set near the T4 position of the international 10–20
system of electrode placement. The right CH15 was located below the midpoint
between T4 and C4. This point in adult subjects might be projected around the
sylvian fissure (Okamoto et al., 2004).
2.5. Data analysis
In the present study, we focused on the variation of [oxy-Hb], which estimates
changes in the regional cerebral blood oxygenation during brain activation. In
each individual set of [oxy-Hb] data, we initially extracted data blocks from the
time course data. Each data block ranged from 0.5 s prior to stimulus onset to
14.5 s after the onset. By detecting rapid changes in [oxy-Hb], we eliminated the
data block that showed a low signal-to-noise ratio due to obstruction by hair and
those that included movement artifacts. Following these screening processes,
subject data that contained a minimum of five good data blocks for each condition
were used. Thus, the number of subjects differed among the measurement
channels. For example, all the 21 infants satisfied the criterion for the right
CH16, while 19 infants satisfied the criterion for the left CH16. Next, we corrected
the baseline using linear fitting to mean signal of the first 11 time points (1.1 s) and
that of the last 11 time points in each data block. By averaging the signal changes
over data blocks for each subject under each condition, we obtained the hemo-
dynamic responses at each measurement channel.
To identify the activated regions under each condition, we considered the
individual data as random effects and performed two-tailed t-tests for each
channel. The t-tests were first performed against baseline (statistical threshold
was set at p < 0.001 in order to take into account multiple comparisons among
48 channels). Then, we compared the cortical activation under the normal
condition and those under the flattened condition (two-tailed paired t-test) for
each channel that showed significant activation under the normal condition.
3. Results
In order to quantify event-related hemodynamic responses to
speech sounds, we first determined an averaged time course
among all the measurement channels under the two conditions
and calculated the time to peak and latency in the time course.
The maximum increase was attained at 5.9 s with the signal
change of 0.0293 mM
mm (Fig. 1C). This response returned to
the baseline at approximately 13 s. We compared this averaged
time course with a canonical hemodynamic response function
of adults (Friston et al., 1998) (Wellcome Department of
Imaging Neuroscience; http://www.fil.ion.ucl.ac.uk/spm/). The
correlation coefficient between the hemodynamic response of
adults and infants was larger than 0.9, which was highly
significant ( p < 0.0001). This result indicates that the sleeping
infants show event-related hemodynamic responses similar to
the adult responses. The amount of signal changes between 4.0
and 7.6 s in the averaged time course was larger than the double
standard deviation among the time points (S.D., 0.0108). Thus,
we used mean signal changes in this time window in the
following statistical analyses.
Next, we investigated the cortical regions that were
responsive to each of the normal or flattened speech sounds.
We found that 3-month-old infants showed bilateral activation
to normal utterances (Fig. 1D). The activation in the bilateral
temporoparietal cortex was notable, and the foremost
measurement channels in the frontal cortex showed additional
responses. These bilateral regions also showed activation in
response to the flattened utterances (Fig. 1D). The bilateral
activation patterns were consistent with the results of the
previous studies on infants (Dehaene-Lambertz et al., 2002,
2004) and adults (Petersen et al., 1988; Price, 2000; Homae
et al., 2002); all of these studies reported that the both
hemispheres are involved in processing of speech sounds.
In order to evaluate the difference between the normal and
flattened conditions, we directly compared the signal changes
under the two conditions. The comparison (normal condition–
flattened condition) revealed that the measurement channels
that showed significant activation were localized in the right
temporoparietal region (Fig. 2A). The most significant
activation was found in the right CH16. Although the left
and the right CH16 showed hemodynamic responses in both the
conditions (Fig. 2B and C), a significant difference between the
two conditions was observed only in the right CH16 (Fig. 2D;
paired t-test, left: n = 19, p > 0.1; right: n = 21, p < 0.005).
This result was confirmed at the individual subject level. Of the
21 subjects, 15 showed greater activation in the right CH16
under the normal condition in comparison with the flattened
condition (binomial test, p = 0.013). In contrast, of the 19
subjects, only 10 showed greater activation in the left CH16
under the normal condition ( p > 0.3). These results indicate
that the right temporoparietal region is more sensitive to normal
 F. Homae et al. / Neuroscience Research 54 (2006) 276–280 279

Fig. 2. Comparisons between the normal and flattened conditions. (A) A direct comparison of the normal condition–flattened condition. Large- and small-filled
circles indicate measurement channels that surpassed p < 0.005 and 0.05, respectively. The arrow indicates the right CH16. (B) The averaged time course of [oxy-Hb]
changes of the left CH16. (C) The averaged time course of [oxy-Hb] changes of the right CH16. (D) The mean [oxy-Hb] changes of CH16. The blue and red bars
indicate the mean signal changes under the normal and flattened conditions, respectively. Error bars show the standard error among the subjects.

utterances as compared to flattened utterances, and demonstrate
a functional differentiation between the right and the left
temporoparietal regions, at least at the age of 3 months. In the
reversed comparison (flattened condition–normal condition),
no significant activation was observed in any of the regions.
4. Discussion
In the present study, we elucidated that the right
temporoparietal region in 3-month-old infants was sensitive
to normal speech, which includes pitch information. The right-
lateralized activation demonstrates that sleeping 3-month-old
infants do not merely receive speech sounds at the primary
sensory level, but also proceed to analyze acoustic parameters
in utterances. We used a single set of sentences in both the
conditions; therefore, phonological information, lexico-seman-
tics and syntactic structures of stimuli in the flattened condition
were identical to those in the normal condition. Therefore, we
conclude that the right temporoparietal region in 3-month-old
infants is related to the processing of pitch contours. If prosodic
information is critical in obtaining lexico-semantic knowledge
and setting syntactic parameters, as proposed in the prosodic
bootstrapping hypothesis (Gleitman and Wanner, 1982;
Morgan and Demuth, 1996; Jusczyk, 1997), prosodic proces-
sing in this region of 3-month-old infants may be the first step of
language acquisition. The processing in 3-month-old infants
may also form the foundation of the following steps observed in
the latter half of the first year such as speech segmentation and
responding to syntactic/prosodic boundaries in a sentence.
Moreover, we found that 3-month-old infants showed
cortical activation to speech sounds in both the left and the
right hemispheres. In previous studies on speech processing in
infants, there appears to be two views regarding the underlying
neural mechanisms. One view is that the cortical regions related
to the processing of speech sounds are lateralized to the left
hemisphere since birth. Pena et al. reported significant
responses to forward speech only in the left temporal regions
of neonates (Pena et al., 2003). The other view is that both the
left and the right hemispheres are responsive to speech sounds
(Dehaene-Lambertz et al., 2002). An event-related potentials
study on a neonate with a left sylvian infarct suggested that both
the hemispheres contribute to phoneme perception during the
first months of life (Dehaene-Lambertz et al., 2004). Moreover,
clinical studies revealed that focal brain injuries not only in the
left but also in the right hemisphere affect early lexical
development (Thal et al., 1991). Our findings supported the
latter view in that the bilateral regions showed responses under
both conditions.
The results of the present study regarding the involvement of
the bilateral regions in speech processing and the right
lateralization in prosodic processing are in accordance with the
recent models of functional differentiation between the
hemispheres in adults. A difference in the two hemispheres
has been proposed at the level of extracting speech-relevant
information in which the right hemisphere may be sensitive to
slowly changing signals such as pitch shifts in prosody, while
the left hemisphere may be sensitive to rapidly changing signals
such as formant transitions in stop consonants (Poeppel, 2003).
In a dual pathway model of sentence processing (Friederici and
Alter, 2004), it is proposed that there exist two pathways in the
cerebral cortex: (1) a pathway related to the processing of
syntax/lexico-semantics in the left hemisphere and (2) a
pathway related to the processing of prosody in the right
hemisphere. The dynamic interactions between the two
hemispheres are assumed when both syntactic and prosodic
information in auditory sentences is processed in the left and
the right hemisphere, respectively. Taken together, the results of
the present study clarify that prosodic processing in 3-month-
old infants is subserved by the right temporoparietal region,
which may facilitate the left hemisphere to process lexical or
syntactic information in the course of language acquisition.
Acknowledgments
We would like to thank M. Fujiwara and T. Ishizuka for
developing probes of OT for measurement in infants, and N.
Okabe and R. Momoi for their administrative assistance.
References
Baum, S.R., Dwivedi, V.D., 2003. Sensitivity to prosodic structure in left- and
right-hemisphere-damaged individuals. Brain Lang. 87, 278–289.
280 F. Homae et al. / Neuroscience Research 54 (2006) 276–280
Baum, S.R., Pell, M.D., 1999. The neural bases of prosody: insights from lesion
studies and neuroimaging. Aphasiology 13, 581–608.
Cheour, M., Leppanen, P.H.T., Kraus, N., 2000. Mismatch negativity (MMN) as
a tool for investigating auditory discrimination and sensory memory in
infants and children. Clin. Neurophysiol. 111, 4–16.
Cutler, A., Dahan, D., vanDonselaar, W., 1997. Prosody in the comprehension
of spoken language: a literature review. Lang. Speech 40, 141–201.
Dehaene-Lambertz, G., Dehaene, S., Hertz-Pannier, L., 2002. Functional
neuroimaging of speech perception in infants. Science 298, 2013–2015.
Dehaene-Lambertz, G., Pena, M., Christophe, A., Landrieu, P., 2004. Phoneme
perception in a neonate with a left sylvian infarct. Brain Lang. 88, 26–
38.
Fernald, A., Kuhl, P., 1987. Acoustic determinants of infant preference for
motherese speech. Infant Behav. Dev. 10, 279–293.
Friederici, A.D., Alter, K., 2004. Lateralization of auditory language functions:
a dynamic dual pathway model. Brain Lang. 89, 267–276.
Friston, K.J., Fletcher, P., Josephs, O., Holmes, A., Rugg, M.D., Turner, R.,
1998. Event-related fMRI: characterizing differential responses. Neuro-
image 7, 30–40.
Gandour, J., Tong, Y.X., Wong, D., Talavage, T., Dzemidzic, M., Xu, Y.S., Li,
X.J., Lowe, M., 2004. Hemispheric roles in the perception of speech
prosody. Neuroimage 23, 344–357.
Gleitman, L.R., Wanner, E., 1982. The state of the state of the art. In: Wanner,
E., Gleitman, L.R. (Eds.), Language Acquisition: The State of the Art.
Cambridge University Press, Cambridge, UK, pp. 3–48.
Herrmann, C.S., Friederici, A.D., Oertel, U., Maess, B., Hahne, A., Alter, K.,
2003. The brain generates its own sentence melody: a Gestalt phenomenon
in speech perception. Brain Lang. 85, 396–401.
Homae, F., Hashimoto, R., Nakajima, K., Miyashita, Y., Sakai, K.L., 2002.
From perception to sentence comprehension: the convergence of auditory
and visual information of language in the left inferior frontal cortex.
Neuroimage 16, 883–900.
Johnsrude, I.S., Penhune, V.B., Zatorre, R.J., 2000. Functional specificity in the
right human auditory cortex for perceiving pitch direction. Brain 123, 155–
163.
Jusczyk, P.W., 1997. The Discovery of Spoken Language. MIT Press, Cam-
bridge, MA.
Kuhl, P.K., 2004. Early language acquisition: cracking the speech code. Nat.
Rev. Neurosci. 5, 831–843.
Kujala, A., Huotilainen, M., Hotakainen, M., Lennes, M., Parkkonen, L.,
Fellman, V., Naatanen, R., 2004. Speech-sound discrimination in neonates
as measured with MEG. Neuroreport 15, 2089–2092.
Maki, A., Yamashita, Y., Ito, Y., Watanabe, E., Mayanagi, Y., Koizumi, H.,
1995. Spatial and temporal analysis of human motor activity using non-
invasive NIR topography. Med. Phys. 22, 1997–2005.
Mandel, D.R., Jusczyk, P.W., Nelson, D.G.K., 1994. Does sentential prosody help
infants organize and remember speech information. Cognition 53, 155–180.
Mehler, J., Jusczyk, P., Lambertz, G., Halsted, N., Bertoncini, J., Amieltison, C.,
1988. A precursor of language acquisition in young infants. Cognition 29,
143–178.
Meyer, M., Steinhauer, K., Alter, K., Friederici, A.D., von Cramon, D.Y., 2004.
Brain activity varies with modulation of dynamic pitch variance in sentence
melody. Brain Lang. 89, 277–289.
Morgan, J.L., Demuth, K., 1996. Signal to Syntax. Erlbaum, Mahwah, NJ.
Nazzi, T., Bertoncini, J., Mehler, J., 1998. Language discrimination by new-
borns: toward an understanding of the role of rhythm. J. Exp. Psychol. Hum.
Percept. Perform. 24, 756–766.
Okamoto, M., Dan, H., Sakamoto, K., Takeo, K., Shimizu, K., Kohno, S., Oda,
I., Isobe, S., Suzuki, T., Kohyama, K., Dan, I., 2004. Three-dimensional
probabilistic anatomical cranio–cerebral correlation via the international
10–20 system oriented for transcranial functional brain mapping. Neuro-
image 21, 99–111.
Pena, M., Maki, A., Kovacic, D., Dehaene-Lambertz, G., Koizumi, H., Bouquet,
F., Mehler, J., 2003. Sounds and silence: an optical topography study of
language recognition at birth. Proc. Natl. Acad. Sci. U.S.A. 100, 11702–
11705.
Petersen, S.E., Fox, P.T., Posner, M.I., Mintun, M., Raichle, M.E., 1988.
Positron emission tomographic studies of the cortical anatomy of single-
word processing. Nature 331, 585–589.
Poeppel, D., 2003. The analysis of speech in different temporal integration
windows: cerebral lateralization as ’asymmetric sampling in time’. Speech
Commun. 41, 245–255.
Portas, C.M., Krakow, K., Allen, P., Josephs, O., Armony, J.L., Frith, C.D.,
2000. Auditory processing across the sleep–wake cycle: simultaneous EEG
and fMRI monitoring in humans. Neuron 28, 991–999.
Price, C.J., 2000. The anatomy of language: contributions from functional
neuroimaging. J. Anat. 197, 335–359.
Taga, G., Asakawa, K., Maki, A., Konishi, Y., Koizumi, H., 2003. Brain imaging
in awake infants by near-infrared optical topography. Proc. Natl. Acad. Sci.
U.S.A. 100, 10722–10727.
Thal, D.J., Marchman, V., Stiles, J., Aram, D., Trauner, D., Nass, R., Bates, E.,
1991. Early lexical development in children with focal brain injury. Brain
Lang. 40, 491–527.
Werker, J.F., Tees, R.C., 1999. Influences on infant speech processing: toward a
new synthesis. Annu. Rev. Psychol. 50, 509–535.
Wong, P.C.M., 2002. Hemispheric specialization of linguistic pitch patterns.
Brain Res. Bull. 59, 83–95.
Zatorre, R.J., Evans, A.C., Meyer, E., Gjedde, A., 1992. Lateralization
of phonetic and pitch discrimination in speech processing. Science 256,
846–849.