Chinese Science BulletinVol. 48No. 23December20032607
Chinese Science Bulletin 2003 Vol. 48 No. 23 2607?2610
Chinese semantic processing
SHAN Baoci1, ZHANG Wutian2, MA Lin3, LI Dejun3,
CAO Bingli4, TANG Yiyuan5, WU Yigen1
& TANG Xiaowei1,4
1. Institute of High Energy Physics and Laboratory of NuclearAnalysis
Techniques, ChineseAcademy of Sciences, Beijing 100039, China;
2. Institute of Psychology, ChineseAcademy of Sciences, Beijing
3. MRI Center of PLAGeneral Hospital, Beijing 100853, China;
4. Physics Department, Zhejiang University, Hangzhou 310027, China;
5. Institute of Neuroinformatics, Dalian University of Technology, Dalian
Correspondence should be addressed to Tang Xiaowei (e-mail: shanbc@
Abstract This study has identified the active cerebral
areas of normal Chinese that are associated with Chinese
semantic processing using functional brain imaging. Accord-
ing to the traditional cognitive theory, semantic processing is
not particularly associated with or affected by input modality.
The functional brain imaging experiments were conducted to
identify the common active areas of two modalities when
subjects perform Chinese semantic tasks through reading
and listening respectively. The result has shown that the
common active areas include left inferior frontal gyrus (BA
44/45), left posterior inferior temporal gyrus (BA37); the
joint area of inferior parietal lobules (BA40) and superior
temporal gyrus, the ventral occipital areas and cerebella of
both hemispheres. It gives important clue to further discern-
ing the roles of different cerebral areas in Chinese semantic
Keywords: Chinese cognition, semantic processing, cerebral, cere-
bellum, functional brain imaging.
Language plays an important role in the people’s
intercommunion. Reading and listening are the main ap-
proaches to achieve language information. Though the two
modalities provide different kinds of information, we can
get the same meaning by reading and by listening. It has
been the crucial question for cognitive neuroscience to
answer how human brain understands language and which
parts of the brain take part in language processing. Identi-
fying the semantic processing cerebral areas is important
not only in cognitive theory, but in clinic practice as well.
For example, the important brain areas, such as speech
center, should be kept away in neurosurgery surgery.
Previous knowledge about lingual areas of brain
came primarily from the studies on brain-injured patients
with language impairment. The lingual areas can be de-
termined according to the injured loci and symptom of
language impairment. However, language processing is a
complex neural process that is associated with several
brain areas. The single patient may be damaged only one
of these areas or the connecting part of two areas, so it can
only provide partial information we wanted. Thus it needs
many variety cases to determine the lingual areas accord-
ing to the language impairment of patients. But it is diffi-
cult to find the appropriate cases. Functional neuroimag-
ing has been an important method to study human brain
function. It has been used in many studies on language
processing. Semantic processing is one of the most popu-
lar tasks in these studies. According to the widely adopted
view of cognition, there exists a semantic-processing net-
work in the brain that is independent of input or output
modalities. So finding out the common active areas by
directly comparing the cerebral areas activated by visual
and auditory semantic tasks is an appropriate method to
identify the semantic processing areas. However, only a
few studies conducted simultaneously visual and auditory
experiments[1,2], most previous studies employed visual or
auditory semantic processing tasks. And what’s more, the
results of these studies are not fully consistent.
Chinese character as an ideographical writing system
may differ from alphabetsin processing. Studies on
Chinese semantic processing may extend our knowledge
about semantic processing. In recent years, there have
been some studies on Chinese neuroimaging[4?9]. But most
of these studies only used visual stimuli. As far as we have
known, there was not any study on functional neuroimag-
ing of Chinese semantic processing to use visual and
auditory stimuli simultaneously. According to the tradi-
tional cognitive theory, semantic processing is not par-
ticularly associated with or affected by input modality.
Our study will identify Chinese semantic processing areas
with the method of searching the common areas activated
by visual and auditory semantic processing using fMRI.
Nine right-handed healthy Chinese (6 male and 3
female, age range: 24?37 years old) participated in this
study. All subjects speak mandarin and at least have a col-
lege level education. There are many homophones in Chi-
nese. In order to avoid the affect of homophone? this study
used double-character Chinese words as materials. All the
used Chinese words were common words with a fre-
quency of occurrence no less than 34 per million accord-
ing to the Modern Chinese Frequency Dictionary. Because
there is category-specific effect in semantic processing,
the animal nouns were used as materials in this study. The
experimental paradigm was block-design, which consisted
of two sessions, one was for visual stimuli, and the other
for auditory stimuli. The visual and auditory stimuli were
delivered in separate sessions. There were in total 4 task
blocks (40 s) and 4 control blocks (20 s) in each session,
task and control were alternated and began with control.
In the visual session, the words were shown through a
projector system connected with a PC-computer, which
2608Chinese Science Bulletin Vol. 48No. 23December 2003
controlled the appearing sequence and lasting time of each
word. Each word lasted 2 s. The subjects were asked to
fixate on “+” at the center of visual field in the control
blocks, and to make no judgment. Whereas, in the task
blocks, the subjects were asked to judge whether or not
the words delivered were animal names. Subjects were
asked to press right hand button if the word was an animal
name, and press left hand button if not. The reaction ac-
curacy data were recorded using a custom-made mag-
net-compatible key-press system. If the accuracy was
lower than 80%, the data were eliminated. Each task block
included 10 animal words and 10 non-animal words,
which were presented randomly. In the auditory session,
the experimental paradigm was similar to that in visual
session, except that the subjects did not listen to any mate-
rial in control block. The auditory materials were digitized
at a normal speaking rate by a PC computer, and were
delivered through a modified headphone connected to an
MR-compatible sound transducer.
Images were acquired on a 1.5T GE signa magnetic
resonance imaging system. Blood oxygen level-dependent
(BOLD) images were acquired using a T2
planar imaging (EPI) sequence, with TR=2 s and TE=40
ms. To cover the whole brain, there were 14 slices, each
was 7 mm thick with 1 mm gap between them. The acqui-
sition matrix was 64×64×14. In order to avoid the T1 ef-
fect, stimuli were presented after the functional image
scanned for 8 s. The first four images were deleted when
data were analyzed. Every session lasted 4 min, 120 func-
tional images of entire brain were acquired. The functional
image data were analyzed using SPM99 software (the
Wellcom Department of Cognitive Neurology, Institute of
Neurology, London, UK). The image data were trans-
formed into the Analyze format used by SPM99 using
in-house software. Spatial realignment was performed
prior to further analysis. The realigned image sequences
were spatially normalized to stereotaxic atlas space of
Talairach, and then the normalized images were spa-
tially smoothed with Gaussian kernel of 12 mm×12 mm
×24 mm (FWHM). The statistical analysis on the
smoothed time series images was based on the general
linear model (GLM). Global effects were removed using
proportional scaling. A high-pass filter and a low-pass
filter were used to remove the measurement noise and
physiological noise. Significant changes of task compar-
ing with control were accessed using one-tail t-statistics.
The t-ratios were estimated for each voxel in the image
and formed the statistical parametric maps which showed
activation above the uncorrected height threshold P<0.01.
Finally, the common brain areas activated jointly by visual
and auditory sessions were identified using Mask function
The common significant areas of activation are
summarized in Table 1, which include left inferior frontal
gyrus (BA44/47), the joint area of left inferior parietal
lobules (BA40) and posterior superior temporal gyrus
(BA22), joint area of left fusiform gyrus (BA37) and ven-
tral occipital areas (BA18/19), the joint areas among right
inferior parietal lobules (BA40) and posterior superior
temporal gyrus (BA22) and postcentral gyrus (BA1/2/3),
right ventral occipital areas (BA18/19), and the cerebella
of both hemispheres. These active areas are overlaid on
the 3-dimensional structure image in color mode as shown
in Fig. 1.
Table 1The common areas activated by visual and auditory Chinese
Common areasBA Talairach coordinate (x,y,z) T value
L inferior prefrontal gyrus44
L inferior prefrontal gyrus47
L inferior parietal lobules40
R inferior parietal lobules40
L superior temporal gyrus 22
R transververse temporal gyrus 42
L ventral occipital areas18
R ventral occipital areas18
P=0.01; BA, Brodmann areas.
ture image in color mode.
The common active areas overlaid on the 3-dimensional struc-
Chinese Science BulletinVol. 48 No. 23December 20032609
The most active areas of this study were consistent
with the results of former studies. In these common active
areas, left inferior frontal gyrus (BA44/47) was the classic
lingual area, Broca’s area and its adjacent area. The stud-
ies on brain damageand functional neuroimaging[8,10]
have all indicated that these areas were closely related to
language-processing although there exist different opin-
ions on its function in language processing. The left infe-
rior frontal gyrus was traditionally believed to have “ex-
pressive” or “output” functions. This kind of view mainly
came from the studies on brain injured patients, because
damage of this area (main part at Broca’s area) will cause
ataxic aphasia. However, our study and several other re-
ports[4,6?8]found the activation of left inferior frontal
gyrus under the condition of none of speech production.
Recently, some researchers have suggested that it serves
as “execute” function, such as choosing, comparing,
judging and retrieving.
The left posterior superior temporal gyrus and its ad-
jacent area was another common active area, which was
the second traditional language area, Wernicker area. Pre-
vious studies have shown that the damage of this area will
bring sensory aphasia. In recent years, many functional
neuroimaging experiments have revealed that this area
plays an important role in semantic processing[14,15], it is
activated not only by visual semantic taskbut also by
auditory semantic task. In addition, this area was com-
monly activated in the visual and auditory functional
The left posterior inferior temporal gyrus and its ad-
jacent areas, fusiform gyrus and inferior occipital areas,
were frequently mentioned in the recent language experi-
ments. The study on brain injured patients showed that left
posterior inferior temporal gyrus is related to semantic
impairment. Price et al.indicated that this area was a
crucial area for semantic processing. Recent experiments
of functional neuroimaging[20,21]and ERPhave also
indicated that this area is probably related to semantic
processing, because it was activated by visual semantic
processing task[15,23,24], auditory semantic processing
task, and picture semantic cognition task. In the vis-
ual and auditory language experiments, Booth et al.
found that this area and adjacent left posterior middle
temporal gyrus were activated. They suggested that these
areas were responsible for semantic processing. Left pos-
terior inferior temporal gyrus and its adjacent areas was an
important area not only for English and other alphabetic
language processing, but also for Chinese processing[2,3].
In addition, fusiform gyrus has strong relativity with pa-
rietal-temporal area. The left posterior inferior temporal
gyrus and its adjacent areas were called the third language
area following “Broca area” and “Wernicker area”.
The activation of right supramarginal gyrus and pos-
terior temporal lobe is consistent with the result of former
functional neuroimaging study on Chinese processing[4?6].
However, the activation of this area is seldom found in
semantic processing experiments of English and other
alphabetic languages. It needs to be further verified if this
area is a specific Chinese processing area.
The activation of cerebellum was also found in the
other studies on the Chinese[5?7]and other language[6,8]
processing. Now, the view that cerebellum was concerned
with cognition has been accepted. But the function of
cerebellum was still not clear in language processing. It
was considered to serve as assistant function in semantic
processing. Considering that there were both semantic
processing and action of pressing button in our experiment,
we could not judge that either semantic processing or ac-
tion of pressing button activated cerebellum.
The activation of the ventral part of occipital lobe
(BA18/19) was not predicted. It was reasonable that these
areas were activated by visual task, but the reason for
these areas activated by auditory task was not very clear.
We inferred that this was the result of subjects imaging the
word form when they heard a Chinese word. Previous
study has found that the ventral part of occipital lobe is
related to imaging, so we guess that the activation of
this area is not caused by semantic processing. It needs
further exploring why this area is activated by auditory
Research Program of China (Grant No. G1999054000).
This work was supported by the National Basic
1. Chee, M., O’Craven, K. M., Bergida, R. et al., Auditory and visual
word processing studied with fMRI, Human Brain Mapping, 1999,
2. Booth, J. R., Burman, D. D., Meyer, J. R. et al., Modality inde-
pendence of word comprehension, Human Brain Mapping, 2002,
3. Rozin, P., Poritsky, S., Sotsky, R., American children with reading
problems can easily learn to read English represented by Chinese
characters, Science, 1971, 171: 1264?1267.
4. Chee, M., Tan, E., Thiel, T., Mandarin and English single word
processing studied with functional magnetic resonance imaging, J.
Neurosci., 1999, 19: 3050?3056.
5. Tan, L. H., Spinks, J. A., Gao, J. H. et al., Brain activation in the
processing of Chinese characters and words: A functional MRI
study, Human Brain Mapping, 2000, 10: 16?27.
6. Tan, L. H., Liu, H. L., Perfetti, C. A. et al., The neural system un-
derlying Chinese logograph reading, Neuroimage, 2001, 13: 836?
7. Tan, L. H., Feng, C. M., Fox, P. T. et al., An fMRI study with writ-
ten Chinese, Neuroreport, 2001, 12: 83?88.
8. Luck, K. K., Liu, H. L., Wai, Y. Y. et al., Functional anatomy of
syntactic and semantic processing in language comprehension,
Human Brain Mapping, 2002, 16: 133?145.
9. Xiang, H., Lin, C., Ma, X. et al., Involvement of the cerebellum in
REPORTS Download full-text
2610 Chinese Science BulletinVol. 48 No. 23December 2003
semantic discrimination: An fMRI study, Human Brain Mapping,
2003, 18: 208?214.
10. Noppency, U., Price, C. J., Functional imaging of the semantic sys-
tem: Retrieval of sensory-experienced and verbally learned knowl-
edge, Brain and Language, 2003, 84: 120?133.
11. Talairach, J., Tournoux, P., Co-planar Stereotaxic Atlas of the Hu-
man Brain, New York: Thieme, 1988, 1?110.
12. Swick, D., Role of left inferior prefrontal cortex in retrieval of se-
mantic, Cognitive Brain Research, 1998, 7: 143?157.
13. Cabeza, R., Nyberg, L., Imaging cognition ?: An empirical re-
view of 275 PET and fMRI studies, J. Cogn. Neurosci., 2000, 12: 1
14. Demonet, J., Chollet, F., Ramsay, S. et al., The anatomy of phono-
logical and semantic processing in normal subjects, Brain, 1992,
15. Grossman, M., Koenig, P., DeVita, C. et al., Neural representation
of verb meaning: An fMRI study, Human Brain Mapping, 2002, 15:
16. Perani, D., Cappa, S. F., Schnur, T. et al., The neural correlates of
verb and noun processing: A PET study, Brain, 1999, 122: 2337?
17. Newman, S. D., Twieg, D., Difference in auditory of words and
pseudowords: An fMRI study, Human Brain Mapping, 2001, 14: 39
18. Alexander, M. P., Hiltbrunner, B., Fischer, R. S., Distributed
anatomy of transcortical sensory aphasia, Archives of Neurology,
1989, 46: 885?892.
19. Price, C. J., The functional anatomy of word comprehension and
production, Trends in Cognitive Sciences, 1998, 2: 281?288.
20. Murtha, S., Chertkow, H., Beauregard, M. et al., The neural sub-
strate of picture naming, J. Cogn. Neuroscience, 1999, 11: 399?
21. Buchel, C., Price, C., Friston, K., A multimodal language region in
the ventral visual pathway, Nature, 1998, 394: 274?276.
22. Hinojosa, J. A., Martin-Loeches, M., Rubia, F. J., Event-related
potentials and semantics: An overview and an integrative proposal,
Brain and Language, 2001, 78: 128?139.
23. Devlin, J. T., Russell, R. P., Davis, M. H. et al., Is there an ana-
tomical basis for category-specificity? Semantic memory studies in
PET and fMRI, Neuropsychologia, 2002, 40: 54?75.
24. Thompson-Schill, S. L, Aguirre, G. K., D’Esposito, M. et al., A
neural basis for category and modality specificity of semantic
knowledge, Neuropsychologia, 1999, 37: 671?676.
25. Vandenberghe, R., Price, C., Wise, R. et al., Functional anatomy of
a common semantic system for words and pictures, Nature, 1996,
26. Horwitz, B., Rumsey, J. M., Donohue, B. C., Functional connec-
tivity of the angular gyrus in normal reading and dyslexia, Proc.
Natl.Acad. Sci. USA, 1998, 95: 8939?8944.
27. Luders, H., Lesser, R. P., Hahn, J. et al., Basal temporal language
area, Brain, 1991, 114: 743?754.
28. O’Craven, K. M., Kanwisher, N., Mental imagery of faces and
places activates corresponding stimulus-specific brain regions, J.
Cogn. Neuroscience, 2000, 12: 1013?1023.
(Received June 6, 2003; accepted October 9, 2003)