- Email: [email protected]
Brain strategies for reading in the second language are determined by the first language
Neuroscience Research 40 (2001) 351– 358 www.elsevier.com/locate/neures
Brain strategies for reading in the second language are determined by the first language Tsutomu Nakada a,b, Yukihiko Fujii a, Ingrid L. Kwee b a
Department of Integrated Neuroscience. Brain Research Institute, Uni6ersity of Niigata, Niigata951 -8585, Japan b Department of Neurology, Uni6ersity of California, Da6is, CA 95616, USA Received 18 December 2000; accepted 9 May 2001
Abstract Brain activation associated with reading was investigated in ten normal Japanese volunteers (five highly literate in both Japanese and English) and ten American native English speakers (five highly literate in both English and Japanese) in order to determine the neuroanatomic substrates employed in reading the first language (L1), and to determine the effect of L1 on the neurosubstrates involved in reading the second language (L2). The study was performed using blood oxygenation level dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) on a high-field (3.0T) system specifically optimized for fMRI. The activation patterns in Japanese subjects reading Japanese (L1) were substantially different from the patterns obtained in American subjects reading English text (L1). The activation patterns reading L2 were virtually identical to the patterns seen when reading L1 in both Japanese and English natives highly literate in both language systems. The results demonstrated that the neuroanatomical substrates underlying the cognitive processing of reading are differentially determined based on the language system. The study further indicates that the cognitive processes for reading in the second language involve the same cortical structures employed for the first language, supporting the hypothesis that the second language represents the cognitive extension of the first language. © 2001 Elsevier Science Ireland Ltd and the Japan Neuroscience Society. All rights reserved. Keywords: Literacy; Language; BOLD; fMRI; High field; Japanese; English
1. Introduction In clear contrast to spoken language, literacy is a cognitive skill which requires specific learning processes. Cross-language differences in literacy acquisition are known to be substantial, even within the alphabetbased European languages. Young children learning to read and write have significantly more difficulties with languages which have inconsistent orthography, such as English, than with languages which have consistent orthography such as Italian (Wanner and Gleitman, 1982; Olson et al., 1994). It is our hypothesis that should differences in written language architecture have significant impact on the physiological acquisition of literacy in the first language then literacy established in an individual’s first language should also have signifi* Corresponding author. Tel.: + 81-25-227-0677; fax: +81-25-2270821. E-mail address: [email protected] (T. Nakada).
cant effect on the acquisition of literacy in the second language. Compared to alphabet-based languages, the Japanese written language system has a substantially different coding architecture consisting of a mixture of both semantic symbols (kanji) and orthographically consistent phonetic symbols (kana). Evidence has now been accumulated indicating that brain activation patterns associated with reading Japanese are substantially different from the patterns associated with reading English. The difference between the languages has been primarily attributed to the cognitive processing for kanji (Miller, 1967; Iwata, 1986; Peterson et al., 1990; Taylor and Taylor, 1995; Pugh et al., 1996; Rumsey et al., 1997; Nakada et al., 1998a,b). Various studies endeavoring to isolate the cortical areas responsible for kanji decoding have appeared in the literature (Yamadori, 1975; Iwata, 1986; Sakurai et al., 1992; Sugishita et al., 1992; Koyama et al., 1998). This unique characteristic of the Japanese written system makes it
0168-0102/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd and the Japan Neuroscience Society. All rights reserved. PII: S 0 1 6 8 - 0 1 0 2 ( 0 1 ) 0 0 2 4 7 - 4
352
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
especially attractive in studying brain strategies in literacy. Our previous studies on musical literacy in right handed Japanese subjects demonstrated that activation associated with reading Japanese text of paragraph length occurred in four distinct areas within the occipital – temporal lobe, namely, the left fusiform gyrus (FG), the left and right lingual gyri (LG), and the areas flanking the left inferior temporal sulcus (ITS) (Nakada et al., 1998a; Nakada, 2000). Should any of these areas indeed subserve the role specific for kanji decoding, the areas could potentially be identified by examining them in contrast to the activation map associated with reading of a written language composed of only phonetic symbols, alphabet, such as English. Furthermore, a comparative study on literacy in Japanese and English in subjects literate in both languages should provide an excellent opportunity to test the aforementioned hypothesis. Accordingly, we investigated brain activation associated with reading in normal volunteers using blood oxygenation level dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) on a high-field (3.0T) system.
2. Materials and methods
2.1. Subjects 2.1.1. Japanese nati6e Ten normal volunteers of Japanese descent (20–29 year old) whose first language was Japanese (henceforth referred to as Japanese native) participated in the study. All subjects had a minimum of 16 years of formal education and were highly literate in their native Japanese. Five out of these ten subjects had shown a level of proficiency in English equivalent to a score of higher than 700 on the United States Scholastic Aptitude Test (SAT) Achievement Test for reading and were considered literate in English for the purpose of this study. These subjects had received formal education in English for at least 10 years, starting in the fifth grade (11 year old or older). All subjects were right handed. Subject handedness was confirmed using the Edinburgh inventory (Oldfield, 1971). 2.1.2. English nati6e Ten normal volunteers (23– 26 year old) whose first language was English (henceforth referred to as English native, five of Western European descent, five, East Asian) participated in the study. All had more than 16 years formal education and were literate in English. Five out of these ten subjects were also literate in Japanese and had studied the language for at least twelve years, two starting in the fourth grade and three in the fifth grade. Their level of Japanese reading
proficiency, confirmed by formal testing, was equivalent to the passing level of the Japanese scholastic aptitude test equivalent to aforementioned SAT. All subjects were right handed. Subject handedness was confirmed using the Edinburgh inventory (Oldfield, 1971).
2.2. BOLD-fMRI Studies were performed according to the human research guidelines of the Internal Review Board of the University of Niigata. A General Electric (Waukesha, WI, USA) Signa-3.0T system equipped with an Advanced NMR (ANMR) EPI module was used to perform all the studies. Subjects were asked to silently read for comprehension full sentences written in their native language. Each session consisted of nine 30 s epochs in the boxcar alternative sequences design (five control and four experimental epochs). The four experimental epochs consisted of paragraphs of varying contents written in the respective languages. Stimulus sentences were randomly chosen from a previously constructed data pool based on several standard official language proficiency tests including level 2 Japanese Language Proficiency Test published by the Association of International Education, Japan, and the Japan foundation for Japanese and reading comprehension learning package for the Test of English as a Foreign Language (TOEFL) published by the Educational Testing Service (ETS) for English, respectively. Control stimuli were a non-specific visual experience and comprised various non-language characters oriented horizontally identical to the experimental sentences. Subjects were probed after the imaging session for contextual comprehension of the reading material presented during the fMRI study. Gradient echo echo-planar images (GE-EPI) were obtained using the following parameter settings: FOV 40 cm× 20 cm; matrix 128× 64; slice thickness 5 mm; inter-slice gap 2.5 mm; TR 1 s. Spatial resolution was approximately 3 mm×3 mm ×5 mm. Sessions which showed brain motion exceeding 0.6 mm were re-performed to avoid fictitious activation due to pixel misalignment. fMRI time series data consisting of consecutive EPI images for each slice were analyzed utilizing SPM96 (the Wellcome Department of Cognitive Neurology) (Friston et al., 1995). The data were smoothed using a 5-mm full width at half maximum (FWHM) kernel. Statistical analysis was performed using a delayed (6 s) boxcar model function in the context of the general linear model as employed by SPM96. To minimize effects of physiological noise, a high pass filter and global normalization were applied within the design matrix. Specific effects were tested by applying appropriate linear contrasts to the parameter estimates for each condition, resulting in a t statistic for each and every voxel (Worsley, 1994). These t statistics,
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
which were transformed to Z statistics, constituted an activation map. These images were interpreted by referring to the probabilistic behavior of a Gaussian field. As pointed out by several authors, such univariate transformation would not transform a t field into a Gaussian field unless the degree of freedom of the t statistics was reasonably high. The condition was dealt with on the assumption that the activation map based on Z statistics was a reasonable lattice representation of an underlying continuous Gaussian field, which was conformed using a spatial smoothing process, a 5-mm FWHM kernel described above. fMRI images were presented with contrast between the two conditions specified and showed activated areas which conformed to statistical criteria of significance (P B 0.001). Data were analyzed and reproducibility was confirmed for each subject individually. Anatomical identification of activated areas was performed individually by mapping areas onto the subject’s own anatomical images obtained with identical coordinates. Analysis based on multiple subjects was performed by mapping individually identified activated areas onto the best fitted area of normalized images according to gyrus/sulcus patterns in three dimensional coordinates (Talairach and Tournoux, 1988).
3. Results Representative activation maps (three consecutive slices) of two representative subjects, one whose native language was Japanese (a 29 year old Japanese male neurosurgeon) and one whose native language was English (a 29 year old American male neurologist) is shown in Fig. 1. The former showed prominent activation in the areas flanking the posterior part of the ITS in the left hemisphere (yellow arrow), while the latter showed prominent activation in the LG bilaterally (red arrows). The areas of activation common to both subjects included the left FG (green arrow). Activation maps while subjects read their respective L1, based on group analysis performed on the raw data normalized anatomically to a standardized brain for Japanese and English natives, are shown in Fig. 2. The four cortical areas with significant activation were localized on a standardized brain schema. These cross-sectional figures, indicated by coordinate numbers, corresponded to the respective figures of the Talairach– Tournoux atlas. Table 1 summarizes the results of all of the subjects in the two groups reading their respective native language, Japanese or English. All areas of activation identified in group analysis, except for the auditory association area, were subjected to individual analysis. Auditory association area activation was utilized for fMRI confirmation of subject comprehension of the
353
silent reading task. The anatomy of these cortical areas specified in Table 1 in relation to the standardized brain is shown in Fig. 2. Activation of the left FG occurred with almost identical frequency in native Japanese readers (80%) as in native English readers (90%). Activation in the LG bilaterally occurred with significantly higher frequency in native English (bilateral average 95%) readers than Japanese (bilateral average 15%), while activation in the areas flanking the posterior aspect of the left ITS occurred exclusively in all native Japanese readers. Table 2 summarizes the activation patterns of five Japanese native and five English native subjects literate in both languages reading their respective first (L1) and second (L2) languages. Activation patterns associated with reading L2 were virtually identical to the patterns associated with reading L1 (Fig. 3). The pattern was further confirmed by analyzing activation maps created by statistically subtracting the activation areas between L1 and L2. No significantly activated areas were found in any of the subjects studied. These dual-literate subjects demonstrated both full word and contextual comprehension of the reading material in the second language.
4. Discussion
4.1. Japanese nati6e 6ersus English nati6e The recognition of separate neuroanatomic substrates subserving the cognitive processes involved in reading kanji and kana originally derived from lesion analyses. Subsequently, functional imaging studies, principally positron emission tomography (PET), have provided support for the concept of neuroanatomic separation of kanji/kana decoding processes (Sakurai et al., 1992; Sugishita et al., 1992; Koyama et al., 1998). The overall current findings are consistent with previously published reports. Nevertheless, our use of a paradigm focused on the natural reading process, in sentence form rather than individual words, and the high degree of anatomic resolution of activation areas in individual subjects afforded by the high-field fMRI system used in this study (Nakada et al., 1998a,b) provide two noteworthy findings that shed further insight into the neuronal substrate of reading. First, the reading paradigm in the Japanese language was associated with activation of the cortex flanking the posterior aspect of the left ITS observed exclusively in all native Japanese speakers reading Japanese in this study. Second, there was a significant difference in the incidence of activation of the LG between subjects whose native language was English (95%) versus Japanese (15%) reading their corresponding native language.
354
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
The results strongly indicate that the cortex flanking the posterior part of the left ITS is the most specific neuroanatomic substrate involved in the decoding pro-
cess of kanji. The findings are furthermore consistent with clinical observations on selective impairment of kanji reading. The present study also indicates that the
Fig. 1. Activation maps (three consecutive slices) of two representative subjects, one whose native language was Japanese (a 29 year old Japanese male neurosurgeon) and one whose native language was English (a 29 year old American male neurologist). The former shows prominent activation in the areas flanking the posterior part of the ITS in the left hemisphere (yellow arrow), while the latter shows prominent activation in the LG bilaterally (red arrows). The areas of activation common to both subjects include the left FG (green arrow). Activation of the auditory association language comprehension area (white arrow) was utilized as fMRI confirmation of subject comprehension of the silent reading task. Table 1 Number of subjects showing significant activation cluster(s) within the specified areas associated with reading their respective first language (Japanese vs. English) Gyrus/sulcus
Left-FG
Left-LG
Right-LG
Left ITS
Brodmann area Talairach coordinates Japanese native English native
18 (−27, −90, −8) 8/10 9/10 NS *
18 (−13, −60, +4) 2/10 10/10 PB0.001 *
18 (+13, −60, +4) 1/10 9/10 PB0.001 *
37 (−51, −60, −1) 10/10 0/10 PB0.001 *
See Fig. 1 FG, fusiform gyrus; LG, lingual gyrus; ITS, inferior temporal sulcus; NS, not significant. * t test.
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
355
Fig. 2. Activation maps while subjects read their respective L1, based on group analysis performed on the raw data normalized anatomically to a standardized brain for Japanese and English natives. In order to eliminate artifacts of the normalization processes due to anatomical variation in individual brain shapes, the data are smoothed using a 20-mm instead of 5 mm full width at half maximum (FWHM) kernel prior to group analysis. Accordingly, the spatial resolution of the activation map is substantially lower compared to the individual activation maps shown in the other figures. Nevertheless, the four cortical areas with significant activation in each group are clearly localized on a standardized brain schema. These cross-sectional figures, indicated by coordinate numbers, correspond to the respective figures of the Talairach – Tournoux atlas. FG, fusiform gyrus; LG, lingual gyrus; and ITS, the posterior part of the inferior temporal sulcus. The results of individual determined activation map are summarized in Table 1.
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
356
Table 2 Number of subjects showing significant activation cluster(s) within the specified areas associated with reading their first (L1) vs. second language (L2) Gyrus/sulcus
Left-FG
Left-LG
Right-LG
Left ITS
Brodmann area Talairach coordinates
18 (−27, −90, −8)
18 (−13, −60, +4)
18 (+13, −60, +4)
37 (−51, −60, −1)
5/5 5/5 5/5 5/5
1/5 1/5 5/5 5/5
0/5 0/5 5/5 5/5
5/5 5/5 0/5 0/5
Japanese native English native
L1 L2 L1 L2
See Fig. 1. FG, fusiform gyrus; LG, lingual gyrus; ITS, inferior temporal sulcus; NS, not significant.
LG are not the universal neuroanatomic substrate for reading. In clear contrast to native English speakers reading English, only a small percentage of native Japanese speakers reading Japanese activated these areas. While each kana symbol, also referred to as syllabograph, represents a syllable equivalent in alphabet-based languages, alphabet-based reading requires the additional earlier cognitive processes of identifying specific letter combinations forming syllables. The LG may be related to the cognitive processes involved in the combination analysis of alphabet.
4.2. First 6ersus second language The present study unequivocally demonstrated that reading in the second language (L2) engages the identical cognitive neuroanatomic substrates employed in reading in the first language (L1). Whereas the pattern of activation associated with reading English by subjects who first acquired literacy in this language system was shown to be clearly distinct from the pattern associated with reading Japanese by subjects who first acquired literacy in Japanese, the activation pattern in those English native or Japanese native subjects who acquired literacy in the second language (L2), Japanese or English, respectively, followed that of L1. These results definitively support our hypothesis that the physiological acquisition of literacy in L1 has significant effect on the acquisition of literacy in L2, even when L1 and L2 utilize dramatically variant coding systems, as is the case for Japanese and English. The findings additionally indicate that, similar to spoken language, reading may exhibit a cognitive ‘accent’. Symbol decoding strategies appears to be determined by the complexity of the written symbolic representation to which the individual is exposed during the critical developmental period. Acquisition of a second spoken language after the primary spoken language predisposes an individual to an audible accent in the second language, while literacy in a second language to a ‘‘reading accent’’.
Our study demonstrates that the cortex flanking the posterior part of the left ITS is the most specific anatomic substrate involved in the post vision processing of kanji. This specificity, however, holds only if kanji represents the subject’s L1 reading system. Thus in subjects whose first reading system is English, the neural processes involved in decoding kanji are subsumed within the distinct neuroanatomic substrates used for decoding an alphabet-based reading system, English in this study. This phenomenon was clearly demonstrated by the five English native subjects reading Japanese. Our results support a system of conservation of cognitive processing for reading in L1 and L2, as opposed to a model of differential spatial localization of activation (Albert and Obler, 1978; Paradis, 1995; Rumsey et al., 1997; Perani et al., 1998; Chee et al., 1999).
4.3. Multiple subject analysis of indi6idually determined acti6ation maps 6ersus single acti6ation map determined utilizing multiple subject data The current study has provided another example demonstrating the usefulness of highly reproducible, high resolution functional activation maps for an individual subject, without necessity of applying artificially established standardization of brain anatomy (Nakada et al., 1998a,b; Kwee et al., 1999; Nakada et al., 2000). As a result, activation maps obtained by functionally related, modality unrelated, independently performed paradigms (e.g. Japanese reading and English reading) in individual subject can effectively be compared to determine functionally overlapping, spatially discrete areas in the brain. Such an approach would be difficult, if not impossible, using other methods for activation studies, such as H2O15 PET and fMRI performed on conventional systems. The current study clearly underscores the significant advantage of multiple subject analysis based on individually defined activation maps over studies based on single composite activation maps derived from multiple subjects and/or sessions em-
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
ployed in other modalities of activation studies, a common practice in PET or fMRI performed on conventional systems. The current approach is especially im-
357
portant for comparative studies or studies involve volve patients where activation pattern may vary significantly individual by individual.
Fig. 3. fMRI images associated with reading Japanese and English in ten subjects (five Japanese native and five English native) literate in both languages. Reading L2 shows an activation pattern virtually identical to the pattern observed for L1.
358
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
Acknowledgements The manuscript was presented in part at the 6th Annual Meeting of the International Society for Magnetic Resonance in Medicine. The study was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology (Japan) and Department of Veteran Affairs Research Service (USA).
References Albert, M., Obler, L., 1978. The Bilingual Brain: Neuropsychological and Neurolinguistic Aspects of Bilingualism. Academic Press, New York. Chee, M.W.L., Caplan, D., Soon, C.S., Sriram, N., Tan, E.W.L., Thiel, T., Weekes, B., 1999. Processing of visually presented sentences in Mandarin and English studies with MRI. Neuron 23, 127 – 137. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.B., Frith, C.D., Frackowiak, R.S.J, 1995. Statistical parametric maps in functional imaging: a general linear approach. Hum. Brain Mapp. 2, 189 – 210. Iwata, M., 1986. Neural mechanism of reading and writing in the Japanese language. Funct. Neurol. 1, 43 –52. Koyama, S., Kakigi, R., Hoshiyama, M., Kitamura, Y., 1998. Reading of Japanese kanji (morphograms) and kana (syllabograms): a magnetoencephalographic study. Neuropsychologia 36, 83 – 98. Kwee, I.L., Fujii, Y., Matsuzawa, H., Nakada, T., 1999. Perceptual processing of stereopsis in human: high-field (3.0 T) functional MRI study. Neurology 53, 1599 –1601. Miller, R.A., 1967. The Japanese Language. University of Chicago Press, Chicago, IL. Nakada, T., Fujii, Y., Suzuki, K., Kwee, I.L., 1998a. ‘‘Musical Brain’’ revealed by high-field (3 tesla) functional MRI. NeuroReport 9, 3853 – 3856. Nakada, T., Fujii, Y., Suzuki, K., Kwee, I.L., 1998b. High-field (3.0 T) functional MRI sequential epoch analysis: an example for motor control analysis. Neurosci. Res. 32, 355 – 362. Nakada, T., 2000. Functional MRI of literacy. In: Kato, T. (Ed.), Frontiers of the Mechanisms of Memory and Dementia. Elsevier, Amsterdam, pp. 21 –24.
.
Nakada, T., Suzuki, K., Fujii, Y., Matsuzawa, H., Kwee, I.L., 2000. Independent component-cross correlation-sequential epoch (ICS) analysis of high-field fMRI time series: direct visualization of dual representation of the primary motor cortex in human. Neurosci. Res. 37, 237 – 244. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97 – 113. Olson, R., Forsberg, H., Wise, B., Rack, J., 1994. Measurement of word recognition, orthographic and phonological skills. In: Lyon, G.R. (Ed.), Frames of Reference for the Assessment of Learning Disabilities. Brookes, Baltimore, MD, pp. 243 – 277. Paradis, M., 1995. Aspects of Bilingual Aphasia. Pergamon Press, Oxford. Perani, D., Paulesu, E., Galles, N.S., Dupoux, E., Dehaene, S., Bettinadi, V., Cappa, S.F., Fazio, F., Mehler, J., 1998. The bilingual brain. Proficiency and age of acquisition of the second language. Brain 121, 1841 – 1852. Peterson, S.E., Fox, P.T., Snyder, A.Z., Raichle, M.E., 1990. Activation of extrastriate and frontal cortical areas by visual words and word-like stimuli. Science 249, 1041 – 1044. Pugh, K.R., Shaywitz, B.A., Shaywitz, S.E., Constable, R.T., Skudlarski, P., Fulbright, R.K., Bronen, R.A., Shankweiler, D.P., Katz, L., Fletcher, J.M., Gore, J.C., 1996. Cerebral organization of component processes in reading. Brain 119, 1221 – 1238. Rumsey, J.M., Horwitz, B., Donohue, B.C., Nace, K., Maisog, J.M., Andreason, P., 1997. Phonological and orthographic components of word recognition. A PET r-CBF study. Brain 120, 739 –759. Sakurai, Y., Momose, T., Iwata, M., Watanabe, T., Ishikawa, T., Takeda, K., Kanazawa, I., 1992. Kanji word reading process analyzed by positron emission tomography. NeuroReport 3, 445 – 448. Sugishita, M., Otomo, K., Kabe, S., Yunoki, K., 1992. A critical appraisal of neuropsychological correlates of Japanese ideogram (kanji) and phonogram (kana) reading. Brain 115, 1563 –1585. Talairach, J., Tournoux, P., 1988. Co-planar Stereotaxic Atlas of the Human Brain. Thieme, Stuttgart. Taylor, I., Taylor, M.M., 1995. Writing and literacy in Chinese, Korean and Japanese. John Benjamins Publishing, Philadelphia, PA. Wanner, E., Gleitman, L.R. (Eds.), 1982. Language Acquisition: The State of the Art. Cambridge University Press, Cambridge. Worsley, K.J., 1994. Local maxima and the expected Euler characteristic of excursion sets of X 2, F and t fields. Adv. Appl. Probab. 26, 13 – 42. Yamadori, A., 1975. Ideogram reading in alexia. Brain 98, 231 –238.
Brain strategies for reading in the second language are determined by the first language Tsutomu Nakada a,b, Yukihiko Fujii a, Ingrid L. Kwee b a
Department of Integrated Neuroscience. Brain Research Institute, Uni6ersity of Niigata, Niigata951 -8585, Japan b Department of Neurology, Uni6ersity of California, Da6is, CA 95616, USA Received 18 December 2000; accepted 9 May 2001
Abstract Brain activation associated with reading was investigated in ten normal Japanese volunteers (five highly literate in both Japanese and English) and ten American native English speakers (five highly literate in both English and Japanese) in order to determine the neuroanatomic substrates employed in reading the first language (L1), and to determine the effect of L1 on the neurosubstrates involved in reading the second language (L2). The study was performed using blood oxygenation level dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) on a high-field (3.0T) system specifically optimized for fMRI. The activation patterns in Japanese subjects reading Japanese (L1) were substantially different from the patterns obtained in American subjects reading English text (L1). The activation patterns reading L2 were virtually identical to the patterns seen when reading L1 in both Japanese and English natives highly literate in both language systems. The results demonstrated that the neuroanatomical substrates underlying the cognitive processing of reading are differentially determined based on the language system. The study further indicates that the cognitive processes for reading in the second language involve the same cortical structures employed for the first language, supporting the hypothesis that the second language represents the cognitive extension of the first language. © 2001 Elsevier Science Ireland Ltd and the Japan Neuroscience Society. All rights reserved. Keywords: Literacy; Language; BOLD; fMRI; High field; Japanese; English
1. Introduction In clear contrast to spoken language, literacy is a cognitive skill which requires specific learning processes. Cross-language differences in literacy acquisition are known to be substantial, even within the alphabetbased European languages. Young children learning to read and write have significantly more difficulties with languages which have inconsistent orthography, such as English, than with languages which have consistent orthography such as Italian (Wanner and Gleitman, 1982; Olson et al., 1994). It is our hypothesis that should differences in written language architecture have significant impact on the physiological acquisition of literacy in the first language then literacy established in an individual’s first language should also have signifi* Corresponding author. Tel.: + 81-25-227-0677; fax: +81-25-2270821. E-mail address: [email protected] (T. Nakada).
cant effect on the acquisition of literacy in the second language. Compared to alphabet-based languages, the Japanese written language system has a substantially different coding architecture consisting of a mixture of both semantic symbols (kanji) and orthographically consistent phonetic symbols (kana). Evidence has now been accumulated indicating that brain activation patterns associated with reading Japanese are substantially different from the patterns associated with reading English. The difference between the languages has been primarily attributed to the cognitive processing for kanji (Miller, 1967; Iwata, 1986; Peterson et al., 1990; Taylor and Taylor, 1995; Pugh et al., 1996; Rumsey et al., 1997; Nakada et al., 1998a,b). Various studies endeavoring to isolate the cortical areas responsible for kanji decoding have appeared in the literature (Yamadori, 1975; Iwata, 1986; Sakurai et al., 1992; Sugishita et al., 1992; Koyama et al., 1998). This unique characteristic of the Japanese written system makes it
0168-0102/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd and the Japan Neuroscience Society. All rights reserved. PII: S 0 1 6 8 - 0 1 0 2 ( 0 1 ) 0 0 2 4 7 - 4
352
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
especially attractive in studying brain strategies in literacy. Our previous studies on musical literacy in right handed Japanese subjects demonstrated that activation associated with reading Japanese text of paragraph length occurred in four distinct areas within the occipital – temporal lobe, namely, the left fusiform gyrus (FG), the left and right lingual gyri (LG), and the areas flanking the left inferior temporal sulcus (ITS) (Nakada et al., 1998a; Nakada, 2000). Should any of these areas indeed subserve the role specific for kanji decoding, the areas could potentially be identified by examining them in contrast to the activation map associated with reading of a written language composed of only phonetic symbols, alphabet, such as English. Furthermore, a comparative study on literacy in Japanese and English in subjects literate in both languages should provide an excellent opportunity to test the aforementioned hypothesis. Accordingly, we investigated brain activation associated with reading in normal volunteers using blood oxygenation level dependent (BOLD) contrast functional magnetic resonance imaging (fMRI) on a high-field (3.0T) system.
2. Materials and methods
2.1. Subjects 2.1.1. Japanese nati6e Ten normal volunteers of Japanese descent (20–29 year old) whose first language was Japanese (henceforth referred to as Japanese native) participated in the study. All subjects had a minimum of 16 years of formal education and were highly literate in their native Japanese. Five out of these ten subjects had shown a level of proficiency in English equivalent to a score of higher than 700 on the United States Scholastic Aptitude Test (SAT) Achievement Test for reading and were considered literate in English for the purpose of this study. These subjects had received formal education in English for at least 10 years, starting in the fifth grade (11 year old or older). All subjects were right handed. Subject handedness was confirmed using the Edinburgh inventory (Oldfield, 1971). 2.1.2. English nati6e Ten normal volunteers (23– 26 year old) whose first language was English (henceforth referred to as English native, five of Western European descent, five, East Asian) participated in the study. All had more than 16 years formal education and were literate in English. Five out of these ten subjects were also literate in Japanese and had studied the language for at least twelve years, two starting in the fourth grade and three in the fifth grade. Their level of Japanese reading
proficiency, confirmed by formal testing, was equivalent to the passing level of the Japanese scholastic aptitude test equivalent to aforementioned SAT. All subjects were right handed. Subject handedness was confirmed using the Edinburgh inventory (Oldfield, 1971).
2.2. BOLD-fMRI Studies were performed according to the human research guidelines of the Internal Review Board of the University of Niigata. A General Electric (Waukesha, WI, USA) Signa-3.0T system equipped with an Advanced NMR (ANMR) EPI module was used to perform all the studies. Subjects were asked to silently read for comprehension full sentences written in their native language. Each session consisted of nine 30 s epochs in the boxcar alternative sequences design (five control and four experimental epochs). The four experimental epochs consisted of paragraphs of varying contents written in the respective languages. Stimulus sentences were randomly chosen from a previously constructed data pool based on several standard official language proficiency tests including level 2 Japanese Language Proficiency Test published by the Association of International Education, Japan, and the Japan foundation for Japanese and reading comprehension learning package for the Test of English as a Foreign Language (TOEFL) published by the Educational Testing Service (ETS) for English, respectively. Control stimuli were a non-specific visual experience and comprised various non-language characters oriented horizontally identical to the experimental sentences. Subjects were probed after the imaging session for contextual comprehension of the reading material presented during the fMRI study. Gradient echo echo-planar images (GE-EPI) were obtained using the following parameter settings: FOV 40 cm× 20 cm; matrix 128× 64; slice thickness 5 mm; inter-slice gap 2.5 mm; TR 1 s. Spatial resolution was approximately 3 mm×3 mm ×5 mm. Sessions which showed brain motion exceeding 0.6 mm were re-performed to avoid fictitious activation due to pixel misalignment. fMRI time series data consisting of consecutive EPI images for each slice were analyzed utilizing SPM96 (the Wellcome Department of Cognitive Neurology) (Friston et al., 1995). The data were smoothed using a 5-mm full width at half maximum (FWHM) kernel. Statistical analysis was performed using a delayed (6 s) boxcar model function in the context of the general linear model as employed by SPM96. To minimize effects of physiological noise, a high pass filter and global normalization were applied within the design matrix. Specific effects were tested by applying appropriate linear contrasts to the parameter estimates for each condition, resulting in a t statistic for each and every voxel (Worsley, 1994). These t statistics,
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
which were transformed to Z statistics, constituted an activation map. These images were interpreted by referring to the probabilistic behavior of a Gaussian field. As pointed out by several authors, such univariate transformation would not transform a t field into a Gaussian field unless the degree of freedom of the t statistics was reasonably high. The condition was dealt with on the assumption that the activation map based on Z statistics was a reasonable lattice representation of an underlying continuous Gaussian field, which was conformed using a spatial smoothing process, a 5-mm FWHM kernel described above. fMRI images were presented with contrast between the two conditions specified and showed activated areas which conformed to statistical criteria of significance (P B 0.001). Data were analyzed and reproducibility was confirmed for each subject individually. Anatomical identification of activated areas was performed individually by mapping areas onto the subject’s own anatomical images obtained with identical coordinates. Analysis based on multiple subjects was performed by mapping individually identified activated areas onto the best fitted area of normalized images according to gyrus/sulcus patterns in three dimensional coordinates (Talairach and Tournoux, 1988).
3. Results Representative activation maps (three consecutive slices) of two representative subjects, one whose native language was Japanese (a 29 year old Japanese male neurosurgeon) and one whose native language was English (a 29 year old American male neurologist) is shown in Fig. 1. The former showed prominent activation in the areas flanking the posterior part of the ITS in the left hemisphere (yellow arrow), while the latter showed prominent activation in the LG bilaterally (red arrows). The areas of activation common to both subjects included the left FG (green arrow). Activation maps while subjects read their respective L1, based on group analysis performed on the raw data normalized anatomically to a standardized brain for Japanese and English natives, are shown in Fig. 2. The four cortical areas with significant activation were localized on a standardized brain schema. These cross-sectional figures, indicated by coordinate numbers, corresponded to the respective figures of the Talairach– Tournoux atlas. Table 1 summarizes the results of all of the subjects in the two groups reading their respective native language, Japanese or English. All areas of activation identified in group analysis, except for the auditory association area, were subjected to individual analysis. Auditory association area activation was utilized for fMRI confirmation of subject comprehension of the
353
silent reading task. The anatomy of these cortical areas specified in Table 1 in relation to the standardized brain is shown in Fig. 2. Activation of the left FG occurred with almost identical frequency in native Japanese readers (80%) as in native English readers (90%). Activation in the LG bilaterally occurred with significantly higher frequency in native English (bilateral average 95%) readers than Japanese (bilateral average 15%), while activation in the areas flanking the posterior aspect of the left ITS occurred exclusively in all native Japanese readers. Table 2 summarizes the activation patterns of five Japanese native and five English native subjects literate in both languages reading their respective first (L1) and second (L2) languages. Activation patterns associated with reading L2 were virtually identical to the patterns associated with reading L1 (Fig. 3). The pattern was further confirmed by analyzing activation maps created by statistically subtracting the activation areas between L1 and L2. No significantly activated areas were found in any of the subjects studied. These dual-literate subjects demonstrated both full word and contextual comprehension of the reading material in the second language.
4. Discussion
4.1. Japanese nati6e 6ersus English nati6e The recognition of separate neuroanatomic substrates subserving the cognitive processes involved in reading kanji and kana originally derived from lesion analyses. Subsequently, functional imaging studies, principally positron emission tomography (PET), have provided support for the concept of neuroanatomic separation of kanji/kana decoding processes (Sakurai et al., 1992; Sugishita et al., 1992; Koyama et al., 1998). The overall current findings are consistent with previously published reports. Nevertheless, our use of a paradigm focused on the natural reading process, in sentence form rather than individual words, and the high degree of anatomic resolution of activation areas in individual subjects afforded by the high-field fMRI system used in this study (Nakada et al., 1998a,b) provide two noteworthy findings that shed further insight into the neuronal substrate of reading. First, the reading paradigm in the Japanese language was associated with activation of the cortex flanking the posterior aspect of the left ITS observed exclusively in all native Japanese speakers reading Japanese in this study. Second, there was a significant difference in the incidence of activation of the LG between subjects whose native language was English (95%) versus Japanese (15%) reading their corresponding native language.
354
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
The results strongly indicate that the cortex flanking the posterior part of the left ITS is the most specific neuroanatomic substrate involved in the decoding pro-
cess of kanji. The findings are furthermore consistent with clinical observations on selective impairment of kanji reading. The present study also indicates that the
Fig. 1. Activation maps (three consecutive slices) of two representative subjects, one whose native language was Japanese (a 29 year old Japanese male neurosurgeon) and one whose native language was English (a 29 year old American male neurologist). The former shows prominent activation in the areas flanking the posterior part of the ITS in the left hemisphere (yellow arrow), while the latter shows prominent activation in the LG bilaterally (red arrows). The areas of activation common to both subjects include the left FG (green arrow). Activation of the auditory association language comprehension area (white arrow) was utilized as fMRI confirmation of subject comprehension of the silent reading task. Table 1 Number of subjects showing significant activation cluster(s) within the specified areas associated with reading their respective first language (Japanese vs. English) Gyrus/sulcus
Left-FG
Left-LG
Right-LG
Left ITS
Brodmann area Talairach coordinates Japanese native English native
18 (−27, −90, −8) 8/10 9/10 NS *
18 (−13, −60, +4) 2/10 10/10 PB0.001 *
18 (+13, −60, +4) 1/10 9/10 PB0.001 *
37 (−51, −60, −1) 10/10 0/10 PB0.001 *
See Fig. 1 FG, fusiform gyrus; LG, lingual gyrus; ITS, inferior temporal sulcus; NS, not significant. * t test.
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
355
Fig. 2. Activation maps while subjects read their respective L1, based on group analysis performed on the raw data normalized anatomically to a standardized brain for Japanese and English natives. In order to eliminate artifacts of the normalization processes due to anatomical variation in individual brain shapes, the data are smoothed using a 20-mm instead of 5 mm full width at half maximum (FWHM) kernel prior to group analysis. Accordingly, the spatial resolution of the activation map is substantially lower compared to the individual activation maps shown in the other figures. Nevertheless, the four cortical areas with significant activation in each group are clearly localized on a standardized brain schema. These cross-sectional figures, indicated by coordinate numbers, correspond to the respective figures of the Talairach – Tournoux atlas. FG, fusiform gyrus; LG, lingual gyrus; and ITS, the posterior part of the inferior temporal sulcus. The results of individual determined activation map are summarized in Table 1.
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
356
Table 2 Number of subjects showing significant activation cluster(s) within the specified areas associated with reading their first (L1) vs. second language (L2) Gyrus/sulcus
Left-FG
Left-LG
Right-LG
Left ITS
Brodmann area Talairach coordinates
18 (−27, −90, −8)
18 (−13, −60, +4)
18 (+13, −60, +4)
37 (−51, −60, −1)
5/5 5/5 5/5 5/5
1/5 1/5 5/5 5/5
0/5 0/5 5/5 5/5
5/5 5/5 0/5 0/5
Japanese native English native
L1 L2 L1 L2
See Fig. 1. FG, fusiform gyrus; LG, lingual gyrus; ITS, inferior temporal sulcus; NS, not significant.
LG are not the universal neuroanatomic substrate for reading. In clear contrast to native English speakers reading English, only a small percentage of native Japanese speakers reading Japanese activated these areas. While each kana symbol, also referred to as syllabograph, represents a syllable equivalent in alphabet-based languages, alphabet-based reading requires the additional earlier cognitive processes of identifying specific letter combinations forming syllables. The LG may be related to the cognitive processes involved in the combination analysis of alphabet.
4.2. First 6ersus second language The present study unequivocally demonstrated that reading in the second language (L2) engages the identical cognitive neuroanatomic substrates employed in reading in the first language (L1). Whereas the pattern of activation associated with reading English by subjects who first acquired literacy in this language system was shown to be clearly distinct from the pattern associated with reading Japanese by subjects who first acquired literacy in Japanese, the activation pattern in those English native or Japanese native subjects who acquired literacy in the second language (L2), Japanese or English, respectively, followed that of L1. These results definitively support our hypothesis that the physiological acquisition of literacy in L1 has significant effect on the acquisition of literacy in L2, even when L1 and L2 utilize dramatically variant coding systems, as is the case for Japanese and English. The findings additionally indicate that, similar to spoken language, reading may exhibit a cognitive ‘accent’. Symbol decoding strategies appears to be determined by the complexity of the written symbolic representation to which the individual is exposed during the critical developmental period. Acquisition of a second spoken language after the primary spoken language predisposes an individual to an audible accent in the second language, while literacy in a second language to a ‘‘reading accent’’.
Our study demonstrates that the cortex flanking the posterior part of the left ITS is the most specific anatomic substrate involved in the post vision processing of kanji. This specificity, however, holds only if kanji represents the subject’s L1 reading system. Thus in subjects whose first reading system is English, the neural processes involved in decoding kanji are subsumed within the distinct neuroanatomic substrates used for decoding an alphabet-based reading system, English in this study. This phenomenon was clearly demonstrated by the five English native subjects reading Japanese. Our results support a system of conservation of cognitive processing for reading in L1 and L2, as opposed to a model of differential spatial localization of activation (Albert and Obler, 1978; Paradis, 1995; Rumsey et al., 1997; Perani et al., 1998; Chee et al., 1999).
4.3. Multiple subject analysis of indi6idually determined acti6ation maps 6ersus single acti6ation map determined utilizing multiple subject data The current study has provided another example demonstrating the usefulness of highly reproducible, high resolution functional activation maps for an individual subject, without necessity of applying artificially established standardization of brain anatomy (Nakada et al., 1998a,b; Kwee et al., 1999; Nakada et al., 2000). As a result, activation maps obtained by functionally related, modality unrelated, independently performed paradigms (e.g. Japanese reading and English reading) in individual subject can effectively be compared to determine functionally overlapping, spatially discrete areas in the brain. Such an approach would be difficult, if not impossible, using other methods for activation studies, such as H2O15 PET and fMRI performed on conventional systems. The current study clearly underscores the significant advantage of multiple subject analysis based on individually defined activation maps over studies based on single composite activation maps derived from multiple subjects and/or sessions em-
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
ployed in other modalities of activation studies, a common practice in PET or fMRI performed on conventional systems. The current approach is especially im-
357
portant for comparative studies or studies involve volve patients where activation pattern may vary significantly individual by individual.
Fig. 3. fMRI images associated with reading Japanese and English in ten subjects (five Japanese native and five English native) literate in both languages. Reading L2 shows an activation pattern virtually identical to the pattern observed for L1.
358
T. Nakada et al. / Neuroscience Research 40 (2001) 351–358
Acknowledgements The manuscript was presented in part at the 6th Annual Meeting of the International Society for Magnetic Resonance in Medicine. The study was supported by grants from the Ministry of Education, Culture, Sports, Science, and Technology (Japan) and Department of Veteran Affairs Research Service (USA).
References Albert, M., Obler, L., 1978. The Bilingual Brain: Neuropsychological and Neurolinguistic Aspects of Bilingualism. Academic Press, New York. Chee, M.W.L., Caplan, D., Soon, C.S., Sriram, N., Tan, E.W.L., Thiel, T., Weekes, B., 1999. Processing of visually presented sentences in Mandarin and English studies with MRI. Neuron 23, 127 – 137. Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.B., Frith, C.D., Frackowiak, R.S.J, 1995. Statistical parametric maps in functional imaging: a general linear approach. Hum. Brain Mapp. 2, 189 – 210. Iwata, M., 1986. Neural mechanism of reading and writing in the Japanese language. Funct. Neurol. 1, 43 –52. Koyama, S., Kakigi, R., Hoshiyama, M., Kitamura, Y., 1998. Reading of Japanese kanji (morphograms) and kana (syllabograms): a magnetoencephalographic study. Neuropsychologia 36, 83 – 98. Kwee, I.L., Fujii, Y., Matsuzawa, H., Nakada, T., 1999. Perceptual processing of stereopsis in human: high-field (3.0 T) functional MRI study. Neurology 53, 1599 –1601. Miller, R.A., 1967. The Japanese Language. University of Chicago Press, Chicago, IL. Nakada, T., Fujii, Y., Suzuki, K., Kwee, I.L., 1998a. ‘‘Musical Brain’’ revealed by high-field (3 tesla) functional MRI. NeuroReport 9, 3853 – 3856. Nakada, T., Fujii, Y., Suzuki, K., Kwee, I.L., 1998b. High-field (3.0 T) functional MRI sequential epoch analysis: an example for motor control analysis. Neurosci. Res. 32, 355 – 362. Nakada, T., 2000. Functional MRI of literacy. In: Kato, T. (Ed.), Frontiers of the Mechanisms of Memory and Dementia. Elsevier, Amsterdam, pp. 21 –24.
.
Nakada, T., Suzuki, K., Fujii, Y., Matsuzawa, H., Kwee, I.L., 2000. Independent component-cross correlation-sequential epoch (ICS) analysis of high-field fMRI time series: direct visualization of dual representation of the primary motor cortex in human. Neurosci. Res. 37, 237 – 244. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97 – 113. Olson, R., Forsberg, H., Wise, B., Rack, J., 1994. Measurement of word recognition, orthographic and phonological skills. In: Lyon, G.R. (Ed.), Frames of Reference for the Assessment of Learning Disabilities. Brookes, Baltimore, MD, pp. 243 – 277. Paradis, M., 1995. Aspects of Bilingual Aphasia. Pergamon Press, Oxford. Perani, D., Paulesu, E., Galles, N.S., Dupoux, E., Dehaene, S., Bettinadi, V., Cappa, S.F., Fazio, F., Mehler, J., 1998. The bilingual brain. Proficiency and age of acquisition of the second language. Brain 121, 1841 – 1852. Peterson, S.E., Fox, P.T., Snyder, A.Z., Raichle, M.E., 1990. Activation of extrastriate and frontal cortical areas by visual words and word-like stimuli. Science 249, 1041 – 1044. Pugh, K.R., Shaywitz, B.A., Shaywitz, S.E., Constable, R.T., Skudlarski, P., Fulbright, R.K., Bronen, R.A., Shankweiler, D.P., Katz, L., Fletcher, J.M., Gore, J.C., 1996. Cerebral organization of component processes in reading. Brain 119, 1221 – 1238. Rumsey, J.M., Horwitz, B., Donohue, B.C., Nace, K., Maisog, J.M., Andreason, P., 1997. Phonological and orthographic components of word recognition. A PET r-CBF study. Brain 120, 739 –759. Sakurai, Y., Momose, T., Iwata, M., Watanabe, T., Ishikawa, T., Takeda, K., Kanazawa, I., 1992. Kanji word reading process analyzed by positron emission tomography. NeuroReport 3, 445 – 448. Sugishita, M., Otomo, K., Kabe, S., Yunoki, K., 1992. A critical appraisal of neuropsychological correlates of Japanese ideogram (kanji) and phonogram (kana) reading. Brain 115, 1563 –1585. Talairach, J., Tournoux, P., 1988. Co-planar Stereotaxic Atlas of the Human Brain. Thieme, Stuttgart. Taylor, I., Taylor, M.M., 1995. Writing and literacy in Chinese, Korean and Japanese. John Benjamins Publishing, Philadelphia, PA. Wanner, E., Gleitman, L.R. (Eds.), 1982. Language Acquisition: The State of the Art. Cambridge University Press, Cambridge. Worsley, K.J., 1994. Local maxima and the expected Euler characteristic of excursion sets of X 2, F and t fields. Adv. Appl. Probab. 26, 13 – 42. Yamadori, A., 1975. Ideogram reading in alexia. Brain 98, 231 –238.