Dissociation of writing processes: functional magnetic resonance imaging during writing of Japanese ideographic characters

Cognitive Brain Research 9 Ž2000. 281–286 www.elsevier.comrlocaterbres

Research report

Dissociation of writing processes: functional magnetic resonance imaging during writing of Japanese ideographic characters Kayako Matsuo a

a,)

, Toshiharu Nakai a , Chikako Kato a,b, Tetsuo Moriya a , Haruo Isoda c , Yasuo Takehara c , Harumi Sakahara c

MRSL, Life Electronics Research Center, Electrotechnical Laboratories, 1497-1 Teragu, Tsukuba, Ibaraki 300-4201, Japan b Toyohashi Sozo College, Toyohashi, Japan c Hamamatsu UniÕersity School of Medicine, Hamamatsu, Japan Accepted 25 January 2000

Abstract Dissociation between copying letters and writing to dictation has been reported in the clinical neuropsychological literature. Functional magnetic resonance imaging ŽfMRI. was conducted in normal volunteers to detect the neurofunctional differences between ‘copying Kanji’, the Japanese ideographic characters, and ‘writing Kanji corresponding to phonological information’. Four tasks were conducted: the copying-Kanji task, the writing-Kanji-corresponding-to-phonogram task, the Kanji-grapheme-puzzle task, and the control task. The right superior parietal lobule was extensively activated during the copying-Kanji task Ža model of the copying letters process. and the Kanji-grapheme-puzzle task. These observations suggested that this area was involved in referring the visual stimuli closely related to the ongoing handwriting movements. On the other hand, Broca’s area, which is crucial for language production, was extensively activated during the writing-Kanji-corresponding-to-phonogram task Ža model of the writing-to-dictation process.. The Kanji-grapheme-puzzle task activated the bilateral border portions between the inferior parietal lobule and the occipital lobe, the left premotor area, and the bilateral supplementary motor area ŽSMA.. Since the Kanji-grapheme-puzzle task involved manipulospatial characteristics, these results suggested cooperation between visuospatial and motor executive functions, which may be extensively utilized in demanding visual language processing. The neurofunctional difference between ‘copying Kanji’ and ‘writing Kanji corresponding to phonogram’ was efficiently demonstrated by this fMRI experiment. q 2000 Elsevier Science B.V. All rights reserved. Keywords: fMRI; Handwriting movement; Visual language; Transcription; Japanese ideographic character; Kanji

1. Introduction This study was performed to examine writing processes. Several neuropsychological reports have described the dissociation between writing to dictation and copying letters w20x. For example, a patient with pure alexia showed preserved ability to write to dictation, but copying letters was very awkward and was accompanied by many trials and errors w11x. In contrast, an aphasic patient with infarction of the left supplementary motor area ŽSMA. could copy words, while writing to dictation was markedly impaired even for single words w13x. In this study, functional magnetic resonance imaging ŽfMRI. was conducted in normal volunteers to distinguish between the neurofunc-

) Corresponding author. Fax: q81-298-69-2152; e-mail: [email protected]

tional procedures involved in copying letters and in writing letters corresponding to phonological information. Tasks were designed to represent these two writing processes using Kanji, the Japanese ideographic characters. Kanji simultaneously provides phonological as well as visual and semantic information in one letter. When a Japanese pure alexic was asked to read a Kanji meaning ‘spear’, he said ‘it is related to samurai or warrior’, although he could not read it aloud correctly w9x. Thus, in contrast with pure alexics among English speakers, Japanese patients with pure alexia Žand a writing disturbance of Kanji, see Ref. w9x. often preserve global semantic grasp for Kanji to some extent. Transcription requires a correspondence between visuospatial analyses of the displayed letters and the ongoing handwriting movements. In this experiment, activation during a Kanji-puzzle task that requires intense visuospatial analyses of Kanji figures was also examined. This task was

0926-6410r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 6 4 1 0 Ž 0 0 . 0 0 0 0 8 - 2

282

K. Matsuo et al.r CognitiÕe Brain Research 9 (2000) 281–286

expected to elucidate the correspondence between the visuospatial and the motor functions more effectively than mere copying.

2. Materials and methods 2.1. Subjects and stimulus presentation Twelve right-handed native Japanese volunteers participated in this study Žfour females, eight males, aged 25–41.. All subjects gave their informed consent, and the protocol was approved by the institutional review board. Sixty-seven Kanji were selected from the Japanese elementary school education list. Their frequencies of appearance in Japanese newspapers were within the rank 1000 w15x. Each selected Kanji could be divided into two or three parts, each of which alone was also a complete Kanji ŽRef. w16x, p. 9.. A black-colored stimulus with a white background was displayed through a goggle-type video system ŽResonance Technology, Van Nuys, CA.. Visual activities of myopic volunteers were adjusted with a built-in eyesight corrective system. 2.2. Paradigm One trial consisted of four conditions ŽFig. 1.: copying-Kanji task ŽCOPY. — a model of the copying

Fig. 1. The three Kanji-generating tasks and the control task. COPY: a Kanji was displayed, and the subjects transcribed it with their right index finger. PHONO: a pronunciation was displayed in Hiragana script ŽJapanese phonographic characters., and the subjects wrote the Kanji corresponding to the pronunciation. PUZZLE: parts of a Kanji were displayed and the subjects rearranged them into the proper form for writing. CONT: a circle was displayed and was traced by the subjects. The character described in display of COPY means ‘‘desire’’ in English. Characters in display of PHONO are pronounced Žnozomi.. The three characters in display of PUZZLE mean ‘‘king’’, ‘‘perish’’ and ‘‘moon’’ in English, respectively.

letters process. A Kanji from the list was presented on the display, and the subjects transcribed the Kanji with their right forefinger onto the surface of the MRI stretcher; writing-Kanji-corresponding-to-phonogram task ŽPHONO. — a model of the writing-to-dictation process. A pronunciation of a Kanji written in Hiragana script ŽJapanese phonographic characters. was displayed, and the subjects wrote down the corresponding Kanji. Although PHONO was not exactly the same as writing to dictation for Kanji, it shared functionally equivalent characteristics with the process because handwriting movements for Kanji were organized by referring to phonological information. In the Japanese writing system w9,16x, the phonological information for Kanji can be represented visually by the Hiragana script. This procedure has the advantage of avoiding activation in the auditory areas. Some listed Kanji shared the same pronunciation with other Kanji despite their difference in meaning as well as in their written forms. Arbitrary selection of the Kanji corresponding to the pronunciation was allowed in this experiment; Kanji-grapheme-puzzle task ŽPUZZLE. — two or three parts of a Kanji were presented simultaneously in random order in the horizontal direction, and the subjects rearranged them to the proper form for writing. PUZZLE included the transcription process to the extent that all parts of the Kanji to be written were displayed visually. PUZZLE, however, demanded additional visuospatial mental operation of Kanji figures in comparison with COPY. Also, subjects had to search for the suitable Kanji from memory. All these demands and the unfamiliarity of PUZZLE to the average subjects required some effort w24x; control task ŽCONT. — subjects repeated tracing a circle on the stretcher surface while a circle was continuously displayed. Four task and rest cycles Ž30 s each. were repeated for each condition. In the rest period, subjects were presented with a blank display. The fMRI signal intensity of task periods was statistically compared with that of the rest periods. Three Kanji or circles were displayed successively, each for 10 s continuously in the 30-s task period. The subjects repeated tracing Kanji or the circle during all task periods. Thus, the quantities of the physical finger movements during the task periods were adjusted to be almost equivalent among the four tasks. The 67 Kanji selected were divided into three groups. To avoid learning effects, each of the three groups was assigned to one of the three Kanji-generating tasks in a pseudorandom order, so that each Kanji appeared only once in the whole trial. The four conditions were also carried out in a pseudorandom order across the subjects to avoid order effects. 2.3. Data acquisition All studies were performed on a 1.5-T MRI scanner ŽSigna, General Electric, Milwaukee, WI.. The three-dimensional Ž3D. whole brain dataset was obtained for

K. Matsuo et al.r CognitiÕe Brain Research 9 (2000) 281–286

detailed assignment of the anatomical structure by a sequence of inversion recovery prepared fast spoiled gradient recalled acquisition in the steady state. The parameters were: repetition time ŽTR., 11.2 ms; echo time ŽTE., minimum; inversion time ŽTI., 300 ms; field of view ŽFOV., 24 cm; 256 = 160 matrix in the axial plane; slice thickness, 1.5 mm; 124 images. A single-shot gradient recalled echo-based echo-planer imaging ŽEPI. sequence was used for the functional studies. The parameters were: TR, 3000 ms; TE, 60 ms; flip angle ŽFA., 908; FOV, 24 cm; 64 = 64 matrix; slice thickness, 8 mm; slice interval, 4 mm; six axial images were obtained covering from the parietal region to the inferior frontal gyrus; 80 framesrslice were obtained.

283

2.4. Data analysis The fMRI data were analyzed according to the general linear model using SPM96 ŽWellcome Department of Cognitive Neurology, London, UK. w3,4x. The analyses were performed on each slice separately. The dataset of each slice was realigned to the first image to remove image drift, and smoothened for 2D convolution. The Gaussian filter width ŽFWHM. was set at twice the voxel size Ž7.5 mm.. The response function was specified to the half-sine wave. A high pass filter Žthe cut-off period, 120 s. and a global normalization ŽANCOVA. were applied w3,4x. The activation maps were generated by the t-statistics ŽSPM Z 4. with a height threshold of 0.001 and an extent threshold of

Fig. 2. Activation maps of typical subjects. In subject A, the right SPL activation was greater in COPY and PUZZLE than CONT and PHONO. At a lower slice level in subject B, extensive activation in the border between parietal and occipital lobes in the right hemisphere was observed under PUZZLE. In subject C, the left inferoposterior prefrontal activation was extensive under PHONO. SM1, the primary sensorimotor area; PMA, the premotor area; SMA, the supplementary motor area; SPL, the superior parietal lobule; AIPL, the anterior part of the inferior parietal lobule; IPrO, the border portion between the inferior parietal lobule and the occipital lobe; IFG, the inferoposterior prefrontal area.

K. Matsuo et al.r CognitiÕe Brain Research 9 (2000) 281–286

284

The activated pixels in each region of interest ŽROI. were counted, and ANOVA of repeated-measured designs was performed for comparison among the four conditions. The Student–Newman–Keul’s test was used as the posthoc test for ANOVA. The gyri and sulci were confirmed individually using a 3D image display system ŽAdvantage Windows, GE. and the ROIs were determined with reference to a standard stereotaxic atlas w26x.

3. Results

Fig. 3. Averages and the standard deviations Žerror bars. of activated pixel number in each area. The vertical axis shows the average number of activated pixels.

0.5. The obtained statistical maps were superimposed onto the mean image of each EPI time series.

All subjects reported successful performance in all tasks. Some subjects showed bilateral and the others unilateral activation in the following foci ŽANOVA results are given in parentheses. ŽFigs. 2 and 3.: the primary sensorimotor area ŽSM1, n.s. for bilateral.; PMA Žthe superior part of the precentral gyrus and sulcus, F3,33 s 9.486, p - 0.001 for left, n.s. for right.; SMA Ž F s 3.018, p - 0.05 for left, F s 3.338, p - 0.05 for right.; the superior parietal lobule ŽSPL, the parietal cortex above the intraparietal sulcus, F s 7.369, p - 0.001 for left, F s 5.312, p - 0.01 for right.; the anterior part of the inferior parietal lobule ŽAIPL, the postcentral sulcus in the inferior parietal area, the supramarginal gyrus, and the anterior part of the intraparietal sulcus, n.s. for bilateral., the border portion between the inferior parietal lobule and the occipital lobe ŽIPrO, the posterior part of the intraparietal sulcus, the angular gyrus, and the adjacent part of the occipital lobe, F s 11.267, p - 0.001 for left, F s 15.062, p - 0.001 for right.; the inferoposterior prefrontal area ŽIFG, Broca’s area and the homologous area in the right hemisphere, including the inferior part of the precentral sulcus, F s 3.004, p - 0.05 for left, n.s. for right.; and the superior occipital lobe ŽSO, in the scan volume, F s 13.643, p 0.001 for left, F s 6.967, p - 0.001 for right.. Results of the post-hoc comparison are summarized in Table 1. The results can be summarized as follows. CONT activated smaller areas than the other three Kanji generating tasks. COPY and PUZZLE induced more extensive activa-

Table 1 Differences in number of activated pixels among the tasks Notes: n.s.s non-significant; U p - 0.05; UU p - 0.01; UUU p - 0.001. ‘‘S’’ in the post-hoc comparison indicates that a significant difference was detected between the tasks at the level of p - 0.05 by the Student–Neuman–Keul’s test. SM1

PMA

Left

Right

ANOVA Significant levels

n.s.

n.s.

Post-hoc comparison CONT vs. COPY CONT vs. PHONO CONT vs. PUZZLE COPY vs. PHONO COPY vs. PUZZLE PHONE vs. PUZZLE

– – – – – –

– – – – – –

Left UUU

S S S – S S

SMA Right n.s.

– – – – – –

SPL

AIPL

Left

Right

Left

Right

U

U

UUU

UU

– – S – – –

– – S – – S

S S S – – –

S – S S – S

IPrO

Left

Right

n.s.

n.s.

– – – – – –

– – – – – –

IFG

Left

Right

Left

UUU

UUU

U

S – S – S S

S – S – S S

– S – – – –

SO Right n.s.

– – – – – –

Left

Right

UUU

UUU

S S S – S S

– S S – S S

K. Matsuo et al.r CognitiÕe Brain Research 9 (2000) 281–286

tion in right SPL Ž p - 0.01. and bilateral IPrO Ž p 0.001. than PHONO. PUZZLE induced a greater increase in activation in bilateral IPrO than COPY Ž p - 0.001.. PHONO induced extensive activation in left IFG, i.e. around Broca’s area Ž p - 0.05..

4. Discussion 4.1. Difference between the pathways of copying and writing letters corresponding to phonograms The lack of significant differences in the activated pixels in contralateral SM1 among the four conditions ŽFig. 3, Table 1. suggested that the quantities of the physical output of finger movements did not differ markedly. This may have been due to the task procedure in which subjects continued finger movements during all task periods. The left SPL involvement in handwriting, in general, was suggested by these results because extensive activation in this area was observed during the three Kanji-generating tasks and the numbers of activated pixels were not significantly different among the three tasks ŽFig. 3, Table 1.. The left parietal areas are known to be crucial for reading and writing processes, and impairment of these areas can induce agraphia and dysgraphia w14x. Thus, the writing procedures shared with the three Kanji-generating tasks might induce extensive activation in the left SPL. A neurofunctional difference between copying-Kanji and writing- Kanji corresponding to phonographic characters was observed in the right SPL. The right SPL was significantly less activated during PHONO than during COPY and PUZZLE Ž p - 0.01, Fig. 3, Table 1., although the trajectories of the physical handwriting movements were virtually equivalent among the three tasks ŽFig. 1.. Visual references for ongoing handwriting movements may induce this right SPL activation, because a whole Kanji ŽCOPY. or parts of a Kanji ŽPUZZLE. were displayed during the activation. On the other hand, the handwriting during PHONO was only indirectly related to the externally displayed visual stimuli because the pronunciations in phonograms ŽHiragana. were displayed instead of the Kanji figures themselves ŽFig. 1.. Although PUZZLE did not display the exact Kanji to be written but parts of Kanji, subjects might visually refer to each Kanji part according to the order for writing. This study suggested that right Žipsilateral. SPL, as well as the dominant hemisphere, may play a role in the copying-Kanji procedure. This finding is in accordance with bilateral SPL involvement in the proprioceptive reference system during voluntary movements in the extrapersonal space w22x, since handwriting is also such a movement. In contrast to the copying-Kanji procedure, the augmented activation in the left IFG in PHONO Ž p - 0.05. is considered to represent another neural pathway that may

285

be related to the writing-to-dictation procedure. The left IFG, i.e. Broca’s area, has been shown to be associated with language production w7,18,19,23x. PHONO required conversion from phonograms to a Kanji figure, whereas other tasks did not. Subjects had to retain phonological information to convert the phonograms into Kanji. Since Broca’s area is involved in the phonological loop of the verbal working memory w25x, the extensive activation in this area during PHONO may reflect the function of the phonological retention for conversion. Further studies are needed because the lateral prefrontal cortex including the left IFG also mediates other cognitive processes w1x. Although infarction of the left SMA w10,17x was reported to induce impairment of writing to dictation w13x, there were no significant differences in SMA activation between COPY and PHONO ŽTable 1.. 4.2. Manipulospatial characteristics of the Kanjigrapheme-puzzle task Visuospatial processing of letters may activate bilateral IPrO during PUZZLE ŽFig. 3, Table 1.. The deficit of left IPrO is known to induce lexical agraphia — impairment in the spelling of irregularly spelled words Že.g. ‘friend’. w21x. Since patients with lexical agraphia can spell regularly spelled words fairly well, this area seems to be related to access to visual representations of the whole words or complicated characters such as Kanji. On the other hand, the right angular gyrus was suggested to be a relay from the right visual cortex to the left angular gyrus via the callosum in letter processing w5x. Moreover, a Japanese patient with right hemisphere impairment was reported to demonstrate jargonagraphia, combining existing radicals or parts of Kanji into meaningless characters w8x. Thus, the right, as well as the left, IPrO may be related to the visuospatial processing of letters. Left PMA and bilateral SMA also demonstrated the most extensive activation during PUZZLE ŽFig. 3, Table 1. w12x. Extensive activation during PUZZLE both in the higher motor areas and in the IPrO suggested the cooperation between visuospatial and motor executive functions in demanding letter processing. A previous behavioral study revealed that a task, which was similar to PUZZLE but in that subjects responded verbally, induced unconscious forefinger movements in 98% of Japanese subjects w24x. These unconscious forefinger movements are often observed among Japanese subjects when they try to recall Kanji or foreign words w2,6x. This phenomenon also reminds us that patients with pure alexia can recognize and read letters by kinesthetic facilitation, by which they move their hand along with spelling w5,11x. In copying letters, recognition of letters precedes handwriting movements. The above observations suggested that handwriting movements may facilitate recognition of letters. Activation in IPrO, SMA and PMA during PUZZLE suggested this paradoxical manipulospatial relationship between hand-

286

K. Matsuo et al.r CognitiÕe Brain Research 9 (2000) 281–286

writing movements and the visuospatial processing of letters. In summary, the copying-Kanji process was characterized by SPL activation, and the process of writing Kanji corresponding to phonological information appeared to involve Broca’s area. The Kanji-grapheme-puzzle task suggested a connection between visuospatial and motor executive functions in the processing of letters by activation of the parieto-occipital area and the higher motor areas Žthe SMA and the premotor area.. Acknowledgements This work was supported by a Domestic Research Fellowship of Japan Science and Technology, and by a Sasakawa Scientific Research Grant from the Japan Science Society. We gratefully thank our colleagues from the Department of Radiology, Hamamatsu University School of Medicine. We also express our gratitude to Prof. Toshio Inui and Dr. Shigeki Tanaka of Kyoto University and Dr. Tomohisa Okada of National Institute for Physiological Sciences for valuable discussion and suggestions. References w1x M. D’Esposito, D. Ballard, G.K. Aguirre, E. Zarahn, Human prefrontal cortex is not specific for working memory: a functional MRI study, Neuroimage 8 Ž1998. 274–282. w2x Y. Endo, The role of a motoric aspect of representation: spontaneous writing-like behavior in Japanese, in: M.M. Gruneberg, P.E. Morris, R.N. Sykes ŽEds.., Clinical and Educational Implication, Practical Aspects of Memory: Current Research and Issues, Vol. 2, Wiley, New York, 1988, pp. 459–463. w3x R.S.J. Frackowiak, K.J. Friston, C.D. Frith, R.J. Dolan, J.C. Mazziotta, Human Brain Function, Academic Press, San Diego, 1997. w4x K.J. Friston, Statistical parametric mapping and other analyses of functional imaging data, in: A.W. Toga, J.C. Mazziotta ŽEds.., Brain Mapping: The Methods, Academic Press, San Diego, 1996, pp. 363–386. w5x N. Geschwind, Disconnection syndromes in animals and man, Brain 88 Ž1965. 237–294. w6x S. Hasumi, Anti-Japanese Linguistics, Chikuma Shobo, Tokyo, 1977, pp. 276–297, in Japanese. w7x R.M. Hinke, X. Hu, A.E. Stillman, S.-G. Kim, H. Merkle, R. Salmi, K. Ugurbil, Functional magnetic resonance imaging of Broca’s area during internal speech, NeuroReport 4 Ž1993. 675–678.

w8x N. Ihori, T. Kashiwagi, A. Kashiwagi, H. Tanabe, Jargonagraphia in Kanji and Kana in a Japanese-crossed Wernicke’s aphasic, Brain Lang. 47 Ž1994. 197–213. w9x M. Iwata, Kanji versus Kana: neuropsychological correlates of the Japanese writing system, Trends Neurosci. 7 Ž1984. 290–293. w10x S. Jonas, The supplementary motor region and speech emission, J. Commun. Disord. 14 Ž1981. 349–373. w11x T. Kashiwagi, A. Kashiwagi, Recovery process of a Japanese alexic without agraphia, Aphasiology 3 Ž1989. 75–91. w12x C. Kato, H. Isoda, Y. Takehara, K. Matsuo, T. Moriya, T. Nakai, Involvement of motor cortices in retrieval of kanji studied by functional MRI, NeuroReport 10 Ž1999. 1335–1339. w13x J.C. Masdeu, W.C. Schoene, H. Funkenstein, Aphasia following infarction of the left supplementary motor area, Neurology 28 Ž1978. 1220–1223. w14x R.A. McCarthy, E.K. Warrington, Cognitive Neuropsychology: A Clinical Introduction, Academic Press, San Diego, CA, 1990, pp. 241–261. w15x The National Language Research Institute, A Study of Uses of Chinese Characters in Modern Newspapers, Shuei Shuppan, Tokyo, 1976, pp. 67–121, in Japanese. w16x M. Paradis, H. Hagiwara, N. Hildebrandt, Neurolinguistic Aspects of the Japanese Writing System, Academic Press Japan, Tokyo, 1985. w17x W. Penfield, K. Welch, The supplementary motor area of the cerebral cortex, Arch. Neurol. Psychiatry 66 Ž1951. 289–317. w18x S.E. Petersen, P.T. Fox, M.I. Posner, M. Mintun, M.E. Raichle, Positron emission tomographic studies of the cortical anatomy of single-word processing, Nature 331 Ž1988. 585–589. w19x M. Petrides, B. Alivisatos, E. Meyer, A.C. Evans, Functional activation of the human frontal cortex during the performance of verbal working memory tasks, Proc. Natl. Acad. Sci. U. S. A. 90 Ž1993. 878–882. w20x D.P. Roeltgen, Localization of lesions in agraphia, in: A. Kertesz ŽEd.., Localization and Neuroimaging in Neuropsychology, Academic Press, San Diego, CA, 1994, pp. 377–405. w21x D.P. Roeltgen, K.M. Heilman, Lexical agraphia, Brain 107 Ž1984. 811–827. w22x P.E. Roland, E. Skinhoj, N.A. Lassen, B. Larsen, Different cortical areas in man in the organization of voluntary movements in extrapersonal space, J. Neurophysiol. 43 Ž1980. 137–150. w23x Y. Sakurai, T. Momose, M. Iwata, T. Watanabe, T. Ishikawa, K. Takeda, I. Kanazawa, Kanji word reading process analysed by positron emission tomography, NeuroReport 3 Ž1992. 445–448. w24x M. Sasaki, A. Watanabe, An experimental study of spontaneous writing-like behavior Ž‘Kusho’. in Japanese, Jpn. J. Educ. Psychol. 31 Ž1983. 1–10, in Japanese. w25x E.E. Smith, J. Jonides, C. Marshuetz, R.A. Koeppe, Components of verbal working memory: evidence from neuroimaging, Proc. Natl. Acad. Sci. U. S. A. 95 Ž1998. 876–882. w26x J. Talairach, P. Tournoux, Co-Planar Stereotaxic Atlas of the Human Brain: 3-Dimensional Proportional System: An Approach to Cerebral Imaging, Thieme, Stuttgart, 1988.