Hong Kong J Psychiatry 2005;15(3):77-81


Evaluation of Bilateral Brain Activation in Chinese Speech

SE Chua, C Cheung, GM McAlonan, IWS Lam, V Cheung, TKW Wong, EKC Chan, F Lieh Mak, KS Tai, LKC Yip, PK McGuire

Dr Siew-Eng Chua, FHKAM (Psychiatry), Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Mr Charlton Cheung, BSc, Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Dr Grainne M. McAlonan, PhD, Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Ms Isabel W.S. Lam, BSc, Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.

Ms Vinci Cheung, MPhil, Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Ms Teresa KW Wong, BEng, Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Mr Eddie Kin-Chui Chan, BEng, Department of Electrical and Electronic Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China.
Prof. Felice Lieh-Mak, MD, FHKAM (Psychiatry), Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Dr Kin-Shing Tai, FHKCR, Department of Radiology, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Mr Lawrence KC Yip, BSc, Department of Radiology, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Dr Philip K. McGuire, PhD, FRCPsych, Institute of Psychiatry, De Crespigny Park, London, United Kingdom.

Address for correspondence: Dr Siew-Eng Chua, Department of Psychiatry, The University of Hong Kong, Queen Mary Hospital, Pokfulam Road, Hong Kong, China.
Tel: (852) 2855 4486; Fax: (852) 2855 1345;
E-mail: sechua@hkucc.hku.hk

Submitted: 15 November 2004; Accepted: 9 January 2006


Objective: To investigate the cerebral hemispheric representation of Chinese subjects employing functional magnetic resonance imaging during verbal fluency tasks.

Patients and Methods: Healthy subjects were presented stimuli of the word categories animals, fruit, occupations. Each had to respond by saying aloud in Chinese any item which belonged to the category. Subjects' speech responses were recorded during functional magnetic resonance imaging. The blood oxygen level-dependent response was averaged and data analysed using statistical parametric mapping.

Results: Verbal fluency tasks spoken in Chinese were associated with bilateral fronto-temporal lobe activation.

Conclusion: These results provide preliminary support for bilateral cerebral hemispheric repre- sentation of Chinese speech.

Key words: Brain mapping, Chinese, Language, Magnetic resonance imaging, functional


The concept of left cerebral hemisphere dominance for language from a young age is well established.1,2 Much of the evidence for left cerebral hemisphere dominance for language is derived from experiments using alphabetic languages. Whether the concept can be regarded as univer- sal can be tested by reference to non-alphabetic language systems, such as Chinese.

Chinese is a logographic language system, composed of characters or strokes positioned within a square. Neuro- imaging experiments using Chinese language have tradi- tionally employed silent reading tasks. These have shown left inferio-frontal and left middle frontal activation when Chinese characters3-5 and sentences6 were read but additional right inferior occipital gyrus activation during visuo-spatial analysis of Chinese characters.7 A study conducted by Tan et al suggests that reading aloud in Chinese is associated with strong bilateral cerebral activation.8 Since reading aloud requires the effective integration of speech and hearing, it is likely that a more widely distributed or bilaterally repre- sented function is needed for Chinese speech.

To investigate this further, the authors selected the ver- bal fluency (VF) task.9,10 In alphabetic languages, VF is known to be associated with strong left cerebral hemisphere activation, activating brain regions which control executive function and semantic processing — the left prefrontal cortex, left superior temporal cortex, and left cingulate cortex.11-13 Thus, if a Chinese VF task activated both hemi- spheres it would provide convincing evidence that the brain representation of executive and semantic processing for Chinese is bilateral. This is a novel and empirical approach since to our knowledge, no previous study has used VF to investigate cerebral lateralisation for Chinese speech.

Patients and Methods


Ten healthy Chinese volunteers were recruited from the community. All spoke Chinese as their first language. None had a history of neurological disease, alcohol or substance use, electroconvulsive therapy, head injury or loss of con- sciousness, or attendance at a special school. None had any history of ferromagnetic material in situ. All ten had normal hearing. Myopic subjects used plastic and titanium goggles to rectify visual acuity during magnetic resonance imaging (MRI). All gave written informed consent to participate after a full description of the study was provided. Partici- pants were reimbursed for travel expenses only. Approval for the study was obtained from the Queen Mary Hospital Ethics Committee.

Clinical Assessments

On the day of the MRI scan, each subject was assessed by a trained rater using a sociodemographic interview, the Information and Digit Span subtests of the Wechsler Adult Intelligence Scale-Revised Manual14 (Cantonese version, Hong Kong Psychological Society 1989), for handedness,15 and using the Mini-Mental State Examination.16 Subjects generated normal scores on VF task,9 previously validated in Chinese adults.10,17 Subjects named items belonging to the categories of animals, or fruits or occupations. The latter category was chosen rather than "vegetables" since it is a common category and not semantically related to the category fruits.

Experimental Design

All subjects were trained to perform the VF task satisfacto- rily before the scan. A blocked periodic design lasting 180 seconds was employed, meaning a periodic or repeating cycle of A-B-A-B-A-B where A is the active condition and lasted for a 30-second "block" followed by the baseline con- dition B which also lasted for 30 seconds. There were 3 stimuli categories for condition A: A1 = animal, A2 = fruit, A3 = occupation. These were presented in the order A1-B- A2-B-A3-B. Each category was presented at a rate of one every three seconds. Thus, for A1, the characters for animal (動物 ) were presented as 10 cues in succession over 30 seconds. Each time, the subject had to give an overt response of one item belonging to that category (eg, 狗"dog" in Chinese).

The subject was instructed not to repeat any item for a given category. During passive condition B, the instruction "rest" ( 休息) was likewise presented as a cue every three seconds for 30 seconds, and the subject was required to re- peat this word in response to the stimulus. The term was selected because it means no occupation and thus represented a control term for the category A3 occupation. Each subject performed the VF task twice, once with visual presentation of cues, the other with auditory presentation of cues. This was done because processing of speech is a socially dynamic process which depends on tuning into different sensory modalities, especially auditory and visual; eg, the cocktail party phenomenon when one selectively tunes in to immediate stimuli while ignoring all other envi- ronmental interference. Accordingly, this kind of mixed event design permitted investigation of whether VF activates brain regions with a common processing component.6,18 Auditory VF task cues were delivered via headphones and visual VF task cues were back-projected on a screen via a computer. The order of stimulus presentation was counter- balanced across subjects — half the subjects had the audi- tory stimuli first and then the visual stimuli, and vice versa. During the auditory presentation of stimuli, the subject was instructed to close his or her eyes to eliminate extraneous environmental stimuli. In response, the subject was required to give an overt speech response, and this was recorded on an MRI-compatible microphone during the scan.

Magnetic Resonance Imaging Acquisition

MRI brain images were obtained using a 1.5 T supercon- ducting magnet (Signa Horizon Echospeed, General Electric, Milwaukee, USA) and a transmit-receive head coil. The anterior AC-PC line was located using sagittal localiser images. A total of 7 contiguous oblique scans of the brain 10 mm thick were obtained starting from 7 mm below the AC-PC line in order to visualise the inferior and middle frontal gyri, superior and middle temporal gyri, and the cingulate gyrus. Using echo planar imaging gradient echo pulse sequence and blood oxygen-level-dependent technique (in which endogenous oxygenated blood is treated as the imaging contrast medium), functional MRI (fMRI) scans were acquired once every 3 seconds for 180 seconds (60 scans per cycle) giving a total of 420 T2*-weighted images per cycle. In plane resolution was 3 mm; imaging parameters were: TE = 40 ms; TR = 3 s; no. signal averages = 1. Each subject performed the VF task twice, yielding 840 T2*-weighted images per subject. T1-weighted struc- tural MRI data were also acquired at the same location as the T2*-weighted images to facilitate co-registration of fMRI images. The subject's head was restrained within a convex, padded head-rest and Velcro tabs were employed to secure the forehead and chin in order to limit head movement during scanning.

Magnetic Resonance Imaging Processing and Analysis

Data were analysed using Statistical Parametric Mapping (SPM-99, Wellcome Department of Cognitive Neurology, London, UK) using the general linear model19,20 running in MatLab 6.0 (Mathworks Inc.,Sherborn, Massachusetts, USA) on a SUN Ultrasparc 10 workstation. The first four images were discarded to avoid saturation effects. To cor- rect for motion, all remaining data were spatially realigned within SPM-99, using a least-squares approach, to estimate a six-parameter rigid body transformation for each data set. Thus, for each subject, all T2* images were successfully realigned with the middle volume of the imaging session and motion was less than 2 mm. Normalisation to standard MNI (Montreal Neurological Institute) space within SPM- 99 was performed, whereby the standard space is based on a template T1-weighted dataset closely corresponding to Talairach space. An isotropic Gaussian filter kernel having a full-width half maximum of 14 mm was used to smooth the data in order to increase signal-to-noise. Statistical para- metric maps were generated using the general linear model21 to characterise regionally specific effects in the imaging data. The term in the model was the active task condition (VF task).

A boxcar reference waveform convolved with a kernel that approximates the haemodynamic response curve was used to test specific hypotheses, resulting in a t value at each voxel. Within SPM, each SPM {t} statistic is trans- formed to the unit normal distribution to give an SPM {Z} statistic. For each subject, brain images from the VF task performed using visual and auditory presentation were com- bined using a standard approach6,22 to permit brain regions of activation common to both modalities of stimulus presen- tation to be identified. This was achieved using a cognitive conjunction analysis, because this permits the identification of areas of common activation associated with a common processing component. The null hypothesis was that at any given voxel, no common brain areas are involved in the performance of VF. Group analysis was performed based on voxels which achieved statistically significant activation at a threshold of p < 0.05 (corrected for multiple compari- sons), and 150-voxel spatial extent threshold applied for cluster size at p < 0.05 (corrected for multiple comparisons). The MNI coordinates in standard parametric space were automatically converted by means of a standard conversion algorithm to Talairach and Tournoux coordinates.


Inspection of movement files of 4 subjects showed greater than 2 mm movement during the scan, so the data from these subjects was excluded from analysis. The final sample com- prised 6 subjects (3 men and 3 women), all right-handed, mean age 37 years (SD 9.9), years of education 9 years (SD 3.1), verbal IQ 100, mean Mini-Mental State Examination16 score 29/30.

Verbal Fluency Performance

Subjects performed VF task equally proficiently, and the mean number of items generated for each category was 19- 20 per minute, which is accepted as normal.10 There was no significant difference between the total number of items generated in the active condition compared to the passive condition. In addition, the total number of Chinese charac- ters in each condition, which can potentially give rise to jaw movement, revealed no significant difference.

Brain Regions Activated During Verbal Fluency

Bilateral fronto-temporal cortices were strongly activated during VF task. Strongest activation was noted in the right inferior frontal gyrus (BA 47), right middle temporal gyrus (BA 39), and right superior temporal gyrus (BA 22). The left middle frontal gyrus (BA 10) and left middle temporal gyrus (BA 39) also showed significant activation (Figure 1). The height and extent of clusters reported were signifi- cant at the level of p < 0.05 corrected (Table 1). Excluding female subjects did not substantially alter these findings.


This paper presents the first description of VF task during Chinese speech associated with bilateral fronto-temporal activation. Classically, the VF task has been regarded as a robust left fronto-temporal hemisphere task, whether or not verbal responses are produced silently or aloud.23,24 The majority of Chinese language tasks evaluated to date have involved silent reading or generating Chinese words and these have also yielded strong left prefrontal activation.25,26 Thus, the bilateral cerebral representation observed in this study is not explained by the use of Chinese per se. It also cannot be accounted for by the auditory and visual nature of VF task stimuli, because these are known to activate left fronto-temporal gyri predominantly.6 It is interesting to note that Tan and coworkers reported that reading Chinese words aloud resulted in bilateral cerebral activation involving the left infero-middle frontal cortex, left motor cortex, right infero-frontal gyri, bilateral anterior superior temporal areas, and the anterior cingulate cortex.27 Tan and colleagues suggest that as with other languages, speaking in Chinese not only recruits the left fronto-temporal cortex but also a distributed brain network incorporating the right cerebral hemisphere. Our preliminary finding of bilateral fronto-temporal activa- tion during Chinese VF task provides additional support for their proposition. It also is consistent with neurophysiological evidence from magneto-encephalography suggesting that words spoken in Chinese result in more bilateral cerebral activity compared with English or Spanish.28 Indeed, process- ing and learning tones, which is essential for Chinese speech, has been shown to involve activation of the right middle and inferior frontal gyrus, respectively.29

A limitation of this study is that the final sample size was small due to high data attrition, a particular disadvan- tage of overt language tasks. Accordingly, these findings are preliminary. The conjunction analysis approach at- tempted to minimise the possibility of Type I errors. This permitted the exploration of core brain regions which are common to both auditory and visual speech processing,22 a method superior to traditional cognitive subtraction analy- sis since it can combine any number of tasks, all of which differ only by one component. Finally, our subjects were all ethnic Chinese, a possible confounding factor.

It is plausible that learning a language with complex tonal and orthographic rules may result in a widely distributed set of brain language systems and our study supports the position that bilateral cerebral activation may be important to Chinese speech. This pattern is likely to be observable from a young age, as the brain shows a striking capacity to adapt to vocal cues very early in life.30 Future research will focus on how brain maturational development determines which brain regions are involved in orthographic and tonal processing of Chinese language.


This is to the authors' knowledge the first study to use fMRI and a VF task to examine Chinese speech. It provides preliminary evidence of bilateral fronto-temporal lobe activation, suggesting that in keeping with other languages, speaking in Chinese recruits not only the left fronto-temporal cortex but also a distributed brain network involving the right cerebral hemisphere.


This study was supported by the Wellcome Trust, UK and the Committee for Research and Conference Grants, The University of Hong Kong. The authors would also like to thank Professor Tan Li-Hai for his helpful comments on the manuscript.


  1. Broca, P. Remarques sur le siège de la faculté du langage articulé, suivies d'une observation d'aphémie (perte de la parole). Bulletin de la Société Anatomique (1861;36:330-357. Source: Harrington A. "Unfinished Business: Models of Laterality in the Nineteenth Century". Chp 1 at page 7. In: Davidson RJ, Hugdahl K, editors. Brain asymmetry. Cambridge, Massachusetts: The MIT Press; 1995.
  2. Gaillard WD, Sachs BC, Whitnah JR, et al. Developmental aspects of language processing: fMRI of verbal fluency in children and adults. Hum Brain Mapp 2003;18:176-185.
  3. Tan LH, Spinks JA, Gao JH, et al. Brain activation in the processing of Chinese characters and words: a functional MRI study. Hum Brain Mapp 2000;10:16-27.
  4. Tan LH, Liu HL, Perfetti CA, Spinks JA, Fox PT, Gao JH. The neural system underlying Chinese logograph reading. Neuroimage 2001;13: 836-846.
  5. Tzeng OJ, Hung DL, Cotton B. Visual interalisation effect in reading Chinese characters. Nature 1979;282:499-501.
  6. Chee MW, O'Craven KM, Bergida R, Rosen BR, Savoy RL. Auditory and visual word processing studied with fMRI. Hum Brain Mapp 1999; 7:15-28.
  7. Tan LH, Spinks JA, Gao JH, et al. Brain activation in the processing of Chinese characters and words: a functional MRI study. Hum Brain Mapp 2000;10:16-27.
  8. Tan LH, Feng CM, Fox PT, et al. An fMRI study with written Chinese. Neuroreport 2001;12:83-88.
  9. Benton AL, Hamsher K deS. Multilingual Aphasia Examination. Iowa City, Iowa: AJA Associates; 2004.
  10. Chan RC, Wong M, Chen EY, Lam LC. Semantic categorisation and verbal fluency performance in a community population in Hong Kong: a preliminary report. Hong Kong J Psychiatry 2003;13:14-20.
  11. Frith CD, Friston KJ, Liddle PF, Frackowiak RS. A PET study of word finding. Neuropsychologia 1991;29:1137-1148.
  12. Phelps EA, Hyder F, Blamire AM, Shulman RG. FMRI of the prefron- tal cortex during overt verbal fluency. Neuroreport 1997;8:561-565.
  13. Schlosser R, Hutchinson M, Joseffer S, et al. Functional magnetic reso- nance imaging of human brain activity in a verbal fluency task. J Neurol Neurosurg Psychiatry 1998;64:492-498.
  14. Wechsler, D. Wechsler Adult Intelligence Scale-Revised Manual. New York: The Psychological Corporation; 1981.
  15. Annett M. A classification of hand preference by association analysis. Br J Psychol 1970;61:303-321.
  16. Folstein MF, Folstein SE, McHugh PR. "Mini-mental State". A practi- cal method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198.
  17. Chiu HF, Chan CK, Lam LC, et al. The modified FULD Verbal Flu- ency Test: a validation study in Hong Kong. J Gerontol B Psychol Sci Soc Sci 1997;52:247-250.
  18. Price CJ, Friston KJ. Cognitive conjunction: a new approach to brain activation experiments. Neuroimage 1997;5:261-270.
  19. Friston KJ, Frith CD, Liddle PF, Frackowiak RS. Comparing func- tional (PET) images: the assessment of significant change. J Cereb Blood Flow Metab 1991;11:690-699.
  20. Friston KJ, Ashburner J, Frith CD, Poline J-B, Heather JD, Frackowiak RS. Spatial registration and normalization of images. Hum Brain Mapp 1995;2:165-189.
  21. Friston KJ, Frith CD, Liddle PF, et al. Comparing functional (PET) images: the assessment of significant change. J Cereb Blood Flow Metab 1991;11:690-699.
  22. Price CJ, Friston KJ. Cognitive conjunction: a new approach to brain activation experiments. Neuroimage 1997;5:261-270.
  23. Phelps EA, Hyder F, Blamire AM, et al. FMRI of the prefrontal cortex during overt verbal fluency. Neuroreport 1997;8:561-565.
  24. Yetkin FZ, Hammeke TA, Swanson SJ, et al. A comparison of func- tional MR activation patterns during silent and audible language tasks. AJNR Am J Neuroradiol 1995;16:1087-1092.
  25. Siok WT, Jin Z, Fletcher P, Tan LH. Distinct brain regions associated with syllable and phoneme. Hum Brain Mapp 2003;18:201-207.
  26. Chee MW, Tan EW, Thiel T. Mandarin and English single word pro- cessing studied with functional magnetic resonance imaging. J Neurosci 1999;19:3050-3056.
  27. Tan LH, Feng CM, Fox PT, Gao JH. An fMRI study with written Chinese. Neuroreport 2001;12:83-88.
  28. Valaki CE, Maestu F, Simos PG, et al. Cortical organization for recep- tive language functions in Chinese, English and Spanish: a cross-lin- guistic MEG study. Neuropsychologia 2004;42:967-979.
  29. Wang Y, Sereno JA, Jongman A, Hirsch J. fMRI evidence for cortical modification during learning of Mandarin lexical tone. J Cogn Neurosci 2003;15:1019-1027.
  30. Poremba A, Malloy M, Saunders RC, Carson RE, Herscovitch P, Mishkin M. Species-specific calls evoke asymmetric activity in the monkey's temporal poles. Nature 2004;427:448-451.
View My Stats