Cerebral mechanisms for different second language writing systems

Abstract

In this fMRI study, we examined the cerebral processing associated with second language (L2) reading in different writing systems in late L2 learners. To examine the impacts of cross-linguistic differences between the first language (L1) and L2 on learning to read in L2, we employed a bidirectional approach and compared brain activation during single word processing in two groups of late L2 readers: (1) L2 readers of English whose L1 was Japanese (Japanese-L1/English-L2) and (2) L2 readers of Japanese (of syllabic Kana only) whose L1 was English (English-L1/Japanese-L2). During English reading, the L2 readers of English (Japanese-L1/English-L2) exhibited stronger activation in the left superior parietal lobule/supramarginal gyrus, relative to the L1 readers of English (English-L1/Japanese-L2). This is a region considered to be involved in phonological processing. The increased activation in the Japanese-L1/English-L2 group likely reflects the increased cognitive load associated with L2 English reading, possibly because L1 readers of Kana, which has an extremely regular orthography, may need to adjust to the greater phonological demands of the irregular L2 English orthography. In contrast, during Kana reading, the L2 readers of Japanese Kana (English-L1/Japanese-L2) exhibited stronger activation in the lingual gyrus in both the left and right hemispheres compared to the L1 readers of Kana (Japaese-L1/English-L2). This additional activation is likely to reflect the lower level of visual familiarity to the L2 symbols in the English-L1/Japanese-L2 group; Kana symbols are uniquely used only in Japan, whereas Roman alphabetic symbols are seen nearly everywhere. These findings, bolstered by significant relationships between the activation of the identified regions and cognitive competence, suggest that the cerebral mechanisms for L2 reading in late learners depends both on which language is their L1 and which language is to be learnt as their L2. Educational implications of these results are discussed.

Full text

Cerebral mechanisms for different second language writing systems
Maki S. Koyama
a,b,
n
, John F. Stein
a
, Catherine J. Stoodley
c
, Peter C. Hansen
d
a
Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK
b
Center for Molecular and Behavioral Neuroscience, Rutgers University, Newark, USA
c
Department of Psychology, American University, Washington DC, USA
d
School of Psychology, University of Birmingham, Birmingham, UK
article info
Article history:
Received 25 October 2012
Received in revised form
22 July 2013
Accepted 1 August 2013
Available online 11 August 2013
Keywords:
Second language reading
Different writing systems
Orthographic regularity
English
Japanese Kana
abstract
In this fMRI study, we examined the cerebral processing associated with second language (L2) reading in
different writing systems in late L2 learners. To examine the impacts of cross-linguistic differences
between the first language (L1) and L2 on learning to read in L2, we employed a bidirectional approach
and compared brain activation during single word processing in two groups of late L2 readers: (1) L2
readers of English whose L1 was Japanese (Japanese-L1/English-L2) and (2) L2 readers of Japanese (of
syllabic Kana only) whose L1 was English (English-L1/Japanese-L2). During English reading, the L2
readers of English (Japanese-L1/English-L2) exhibited stronger activation in the left superior parietal
lobule/supramarginal gyrus, relative to the L1 readers of English (English-L1/Japanese-L2). This is a
region considered to be involved in phonological processing. The increased activation in the Japanese-L1/
English-L2 group likely reflects the increased cognitive load associated with L2 English reading, possibly
because L1 readers of Kana, which has an extremely regular orthography, may need to adjust to the
greater phonological demands of the irregular L2 English orthography. In contrast, during Kana reading,
the L2 readers of Japanese Kana (English-L1/Japanese-L2) exhibited stronger activation in the lingual
gyrus in both the left and right hemispheres compared to the L1 readers of Kana (Japaese-L1/English-L2).
This additional activation is likely to reflect the lower level of visual familiarity to the L2 symbols in the
English-L1/Japanese-L2 group; Kana symbols are uniquely used only in Japan, whereas Roman alphabetic
symbols are seen nearly everywhere. These findings, bolstered by significant relationships between the
activation of the identified regions and cognitive competence, suggest that the cerebral mechanisms for
L2 reading in late learners depends both on which language is their L1 and which language is to be learnt
as their L2. Educational implications of these results are discussed.
Published by Elsevier Ltd.
1. Introduction
With the advent of globalization, learning to read one or more
second languages (L2) fluently has become increasingly important in
the modern world. In many countries (e.g. the USA and Japan), L2
education does not officially start until the age of 10–12, well beyond
the most receptive period for language acquisition (Au, Knightly, Jun,
&Oh,2002;Johnson & Newport, 1989;Lenneberg, 1967). Hence, late
L2 learners seldom fully master L2 phonology as regards speech
perception and speech production (see review by Bongaerts, 2005;
Flege, 1991;Long, 1990;Newport, 1990). However, unlike listening or
speaking, reading and writing are recent cultural developments
(Lawler, 2001), and thus may require explicit and intensive learning
not only for the first language (L1) but also for L2. Therefore, late L2
learners are only likely to be able to achieve native-like proficiency in
L2 reading after a great deal of effort.
Writing systems differ widely in the way language units are
represented (Bolger, Perfetti, & Schneider, 2005). In reading, it is
crucial to map letters and letter combinations (orthography) onto
their sounds (phonology), but grapheme-phoneme conversion
varies according to orthographic regularity, even when the same
orthography (e.g. the Roman alphabet) is used to represent
different written languages. For example, Italian and English both
use the same Roman alphabet, but the Italian orthography is
considerably more regular/transparent than the English orthogra-
phy: “d”can be read only as /d/ in Italian, but in English, “d”can be
read as /d/ (“bed”), /dʒ/(“procedure”), or not pronounced at all
(“Wednesday”). Consequently, L2 readers of English need to learn
arbitrary orthographic patterns and an additional range of pho-
nological representations present in English words. Thus, for
Italian L1 readers, learning to read English is inevitably more
demanding cognitively than for English L1 readers learning to read
Italian.
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/neuropsychologia
Neuropsychologia
0028-3932/$ - see front matter Published by Elsevier Ltd.
http://dx.doi.org/10.1016/j.neuropsychologia.2013.08.002
n
Corresponding author at: Department of Physiology, Anatomy and Genetics,
University of Oxford, Parks Road, Oxford, OX1 3PT, UK. Tel.: +44 1865 272552.
E-mail address: makisophiakoyama@gmail.com (M.S. Koyama).
Neuropsychologia 51 (2013) 2261–2270

Effects of orthographic regularity are evident in early develop-
mental patterns of word reading. Seymour, Aro, & Erskine (2003)
compared the rate of reading acquisition in age-matched children
from different European countries, and showed that word reading
can be acquired faster by readers of regular orthographies (e.g.,
Italian, Spanish, Finnish) than by readers of the irregular English
orthography. These linguistic differences are consistent with the
psycholoinguistic grain size theory of reading (Ziegler and
Goswami 2005), and indicates that word reading in an irregular
orthography may be cognitively more demanding for reading
beginners, and potentially even more so for late L2 learners.
In addition to orthographic regularity, visual features of sym-
bols can vary widely between different writing systems. For
example, when L1 English readers learn to read an L2 such as
Japanese or Russian, the familiar L1 symbols of the Roman
alphabet are replaced with unfamiliar L2 symbols in the target
language (Japanese Kana or Russian Cyrillic). However, in these
phonographic writing systems, symbols are still fundamentally
mapped to sounds as phonemes and syllables. More radically, in
logographs, such as Japanese Kanji or Chinese Hanzi, no simple
mappings from symbols to sounds exist at all. Therefore, when
visual features of symbols are different between late readers' L1
and L2, successful L2 reading requires them to accommodate to
the additional visual demands of the L2 script (Nelson, Liu, Fiez,
and Perfetti 2009), since applying L1 reading strategies to a new
writing system will be ineffective. That is, phonological decoding
employed in reading alphabetic English words simply cannot be
applied to logographic reading.
A recent fMRI study has highlighted brain regions significantly
more engaged in L2 readers compared to L1 readers (Zhao et al.,
2012). In this study, L2 readers of Chinese, who had L1s with
alphabetic systems (e.g. English), exhibited stronger activation
during logographic word processing in the right visual cortex
(including fusiform gyrus) compared to L1 readers of Chinese. This
L2 group effect (greater activation in L2 readers than L1 readers)
indicates that the L2 readers of Chinese had adapted to the greater
visual demands of the Chinese characters. Clearly, the Roman
alphabet is far less visually complex than the logograph. Even in
skilled logographic L1 readers, the logographic symbols/words
(e.g. Japanese Kanji) activated the right visual cortex more strongly
than non-logographic symbols (e.g. Japanese Kana) (Koyama,
Stein, Stoodley, & Hansen, 2011;Nakamura, Dehaene, Jobert,
Le Bihan, & Kouider, 2005), emphasizing that visual features of
symbols and words exert strong influences on visual networks.
To date, differences in cerebral mechanisms between L1 read-
ers and L2 readers have been mainly investigated by targeting one
writing system as the L2 (e.g., Chee, Tan, & Thiel, 1999;Klein et al.,
2006;Kovelman, Baker, & Petitto, 2007;Leonard et al., 2010;
Marian et al., 2007;Parker et al., 2012;Yokoyama et al., 2006).
However, this approach does not fully allow us to understand the
impact of cross-linguistic differences between L1 and L2 writing
systems on L2 reading in late learners, whose neural networks for
reading have been established based on their L1 reading. Here, we
employed a bidirectional approach and examined patterns of brain
activation in two groups of late L2 readers: (1) L2 readers of
English whose L1 was Japanese; and (2) L2 readers of Japanese (of
syllabic Kana only) whose L1 was English.
For both English and Japanese Kana scripts, the symbols are
visually simple (relative to logographic symbols) and mapped onto
sounds, but these two writing systems are very different in their
level of orthographic regularity: English has an irregular ortho-
graphy, whereas Kana is extremely regular with nearly one-to-one
mapping with only a few exceptions (none of which were used in
this experiment). Hence, we postulated that the orthographic
difference between L1 and L2 writing systems would exert an
impact on the cerebral mechanisms underlying L2 reading only in
the L2 group learning English, because their L2 orthography
(English) was far more irregular than their L1 orthography (Kana).
2. Methods
2.1. Participants
Two groups of late L2 readers participated in this study: fif teen native Japanese
readers who had learnt English as L2 (Japanese-L1/English-L2 group, “J1/E2”group;
mean age7SD ¼29.3 76.4 years) and fourteen native English readers who had
learnt Japanese Kana as L2 (English-L1/Japanese-L2 group, “E1/J2”group; mean
age7SD ¼26.2 75.7 years). They were all right-handed, as measured by the Annett
Handedness Questionnaire (Annett, 1970). Participants reported no history of
psychiatric disorders or learning disability (including dyslexia). A questionnaire
confirmed that no one in either group started learning their L2 language before the
age of 12. Thus, all participants were defined as late L2 readers. In addition, all
participants had L2 experience at university abroad (e.g. on exchange programs) for
at least for 6 months (i.e. English-L2 readers in the UK; Japanese-L2 readers in
Japan). The study was approved by the Oxfordshire Research Ethics Committee.
At the time of the current study (both cognitive testing and fMRI scanning),
participants in the J1/E2 group were either full-time students (N¼5) or exchange
students (N¼10) at universities in the UK, whereas those in the E1/J2 group were
full-time students who were studying Japanese at universities in Oxford.
In addition, all participants in the E1/J2 group had learnt Japanese at university
in Japan (e.g. on exchange programs) at least for 6 months.
2.2. Cognitive measures (Tasks performed outside the scanner)
Single word reading competence was assessed for English by the WRAT-III
(Wilkinson, 1993) and for Japanese Kana by the Kana Word Reading test (Koyama,
Hansen, & Stein, 2008). Additionally, we administered the Kanji Word Reading test,
a measure of word reading in the logographic Japanese script (Koyama et al., 2008).
Nonverbal IQ was measured using the Raven's Advanced Progressive Matrices
(Raven, Raven, & Court, 1998). We administered two short-term memory tests.
Phonological short-term memory was measured by nonword repetition tasks –the
Comprehensive Test of Phonological Processing (CTOPP: Wagner, Torgessen,
& Rashotte, 1999) for English sounds, and nonword repetition in morae (Koyama
et al., 2008) for Japanese sounds. Visual short-term memory was measured by the
Visual Patterns Test (Della Sala, Gray, Baddeley, Allamano, & Wilson, 1999).
It should be noted that the Kana Word Reading, Kanji Word Reading and nonword
repetition tasks in Japanese were not standardized measures. We therefore used raw
scores or percent accuracy for further analysis.
2.3. Tasks performed in the scanner
Participants performed a phonological one-back matching task for both
Japanese Kana words and English words. Fig. 1 illustrates the task paradigm and
the conditions of interest (words printed in lower case English and in Japanese
syllabic Kana). Participants were instructed to press a button with their right index
finger if successively presented words were phonologically identical (“Whenever
you see two words in succession that sound the same, press the button”).
To minimize visual strategies during the phonological one-back matching task,
successive words were also printed in alternating fonts that were significantly
different (regular vs. italic for English words; Mincho vs. Gyosho for Kana words).
This encouraged covert articulation and consequent phonological encoding. All
words were four characters long and represented high frequency nouns, based on
the Amano and Kondo (1999) norms for Japanese Kana words, and frequency
norms by Kucera and Francis (1967) for English words. For Kana words, visual
familiarity ratings were examined to exclude any word that is more commonly
printed in logographic Kanji (see details in Koyama et al., 2011).
The paradigm was a block design with alternating 24 s task blocks and 15 s rest
blocks. In the rest block, a small red fixation point was visible at the center of the
visual display. In the task block, 24 words were presented at a rate of 1/second,
with an onscreen duration of 250 ms and a blank period of 750 ms between words.
Within each task block, 3–5 of the 24 words were phonologically identical and
required a button response. The participants were encouraged to respond as
quickly and accurately as possible. Prior to the scan session, participants performed
a computerized practice run outside the scanner to ensure task familiarity. In order
to prevent word-specific practice effects, the word stimuli used in the practice run
were different from the words used in the in-scanner task.
Even though words were presented in alternating fonts or style for both word
conditions, the possibility that participants employed some degree of visual
matching strategies cannot be entirely ruled out. Hence, the two groups performed
a further control task that was a purely visual one-back matching task involving
visually unfamiliar, unpronounceable, but ecologically valid, Tibetan letter strings.
(Note that no Tibetan symbols similar to symbols present in either English or Kana
were selected). The paradigm applied to this visual task was the same as the
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–22702262

phonological task above, the difference being that the participants were instructed
to attend to the visual features of the presented words during the visual task
(“Whenever you see two words in succession that look the same, press the
button”). Examination of the between-groups results on this task allows us to
identify whether there were any differences between the two groups in the visual
processing of valid, but unfamiliar and unpronounceable orthography.
2.4. MRI data acquisition
Functional and structural images were acquired with a Varian Siemens 3T
scanner at the Oxford Centre for Functional MRI of the Brain (FMRIB). Prior to data
acquisition, an automated shimming algorithm was applied to reduce magnetic
field inhomogeneities (Wilson et al., 2002). For whole brain functional imaging, a
T2
n
-weighted gradient-echo EPI sequence was employed with parameters:
TR¼30 00 ms, TE¼30 ms, flip angle¼901, FOV ¼192 mm
2
, voxel
size¼3"3"3 mm, with 43 slices acquired in axial orientation. The phonological
one-back matching fMRI protocol consisted of 264 volumes and the control visual
one-back matching fMRI protocol consisted of 368 volumes. For reasons of
comparison in other studies (including Koyama et al., 2011), both fMRI protocols
contained additional ‘filler’conditions that are not relevant to the current analysis.
For structural images, a high-resolution T1-weighted scan was acquired (3D
TurboFLASH sequence, TR¼13 ms, TE¼5 ms, TI ¼200 ms, flip angle¼81,
FOV¼265 mm
2
, voxel size¼1"1"1 mm).
2.5. MRI data analysis
Data were analyzed using the FMRIB Software Library (FSL, www.fmrib.ox.ac.
uk/fsl). The initial four dummy volumes were discarded from functional MRI data to
eliminate non-equilibrium effects of magnetization. The following pre-processing
procedures were applied: a high-pass filter cut-off of 40 s, motion correction using
MCFLIRT, regular-up slice-timing correction, and spatial smoothing using a Gaus-
sian spatial filter with kernel size 5 mm full width half maximum. The registration
of functional images for each participant into standard space was carried out using
FMRIB NonLinear Image Registration Tool (FNIRT).
After the pre-processing, statistical analysis at the individual level was performed
for both English and Kana conditions using a general linear model (GLM) with local
autocorrelation correction (FILM prewhitening; Woolrich, Ripley, Brady, & Smith,
2001). At the single subject level for the main phonological one-back matching task,
the two word conditions (English and Kana) were set up as regressors of interest.
To help correct for motion-related artifacts, the six motion correction parameters
estimated with MCFLIRT were included in the model as regressors of no interest.
In addition, in order for the model to best fit the time course of the actual data
acquisition, temporal derivatives of the main conditions were added as separate
regressors and temporal filtering was applied.
Group analysis was performed with random effects analysis using FLAME. Direct
comparisons were made between L1 and L2 groups, with contrasts “L1 group4L2
group”and “L2 group4L1 group”examined for both English and Kana word
conditions using Gaussian Random Field theory thresholding (cluster-corrected for
multiple comparisons, po0.05). Subsequently, to examine pairwise differences
between the groups for each word condition and differences between the word
conditions within each group, the % BOLD signal changes in the significant clusters (as
determined by FSL Featquery) were further investigated via unpaired and paired t-
tests, respectively. Note that the90th percentile measure of % BOLD signal change was
utilized, rather than the mean or maximum value, because it is considered a more
representative measure of a typically active voxel within the ROI (Buck, Singhal, Arora,
Schlitt, & Constable, 2008).
3. Results
3.1. Cognitive measures (Performance outside the scanner)
Table 1 gives a summary of the demographic and cognitive
measures (performed outside the scanner) for each group. It can
be seen that the two groups were matched in age, gender,
intellectual competence, and visual short-term memory.
3.1.1. Word reading tasks
Although each L2 group achieved high-level performance in their
L2 word reading, the accuracy was still higher for L1 group than L2
group. For the WRAT (total number of words ¼44), the mean
accuracy7SD was 37.073.6 in the E1/J2 group and 31.575.5 in
the J1/E2 group (t¼3.2, po0.01). For Kana (total number of
words ¼20), the mean accuracy7SD was 20.070.0 in the J1/E2
group and 18.271.3 in th e E 1/J2 g r oup ( t-test was not performed for
Kana accuracy due to a ceiling effect in both groups). This level of
accuracy for Kana reading (i.e., no errors made by the native Japanese
speakers; a small standard deviation even in the non-native Japanese
speakers) confirms that Kana has an extremely regular orthography.
As with the fluency, the response time was significantly shorter
(i.e. more fluent) in the L1 group than the L2 group. For WRAT, the
mean response time7SD was 62.2 711.1 s in t he E1/J2 g r o u p a n d
76. 4 719.8 s in th e J1 /E 2 group (t¼2.4, po0.05). For Kana, the mean
response time7SD was 35.677. 9 s i n the J 1 /E2 gr oup an d
67.0721.2 s in the E1/J2 group (t¼5.4, po0.01).
For writing systems with regular orthography, reading fluency,
rather than reading accuracy, may be a more sensitive index of
reading proficiency (De Luca, Zeri, Spinelli, & Zoccolotti, 2010).
This can be more evident in Kana, a writing system with extremely
regular orthography (Seki, Kassai, Uchiyama, & Koeda, 2008).
In the current study, the L2 readers of Japanese (i.e., E1/L2 group)
had significantly slower RT for Kana reading relative to the L1
readers of Japanese (i.e., J1/E2 group), suggesting that their
Table 1
Demographic and cognitive measures for each group.
Japanese-L1/
English-L2
English-L1/
Japanese-L2 Group difference
Mean (SD) Mean (SD) t or χ
2
test
Age 29.3 (6.4) 26.2 (5.7) t¼1.4, N.S.
Sex 4 M/11 F 4M/10 F χ
2
¼0.1, N.S.
Raven (max¼36) 29.1 (5.8) 27.3 (4.6) t¼0.9, N.S.
WRAT (max¼44) 31.5 (5.5) 37.0 (3.6) t¼3.2, po0.01
WRAT RT (s) 76.4 (19.8) 62.2 (11.1) t¼2.4, po0.05
Kana (max¼20) 20.0 (0.0) 18.2 (1.3) No statistical test
Kana RT (s) 35.6 (7.9) 67.0 (21.2) t¼5.4, po0.01
Kanji (max¼60) 56.8 (2.5) 21.1 (6.1) t¼20.9, po0.001
NWRepEng (max¼18) 13.3 (2.7) 15.9 (1.5) t¼3.0, po0.01
NWRepJap (max¼40) 32.2 (4.5) 28.5 (3.2) t¼2.5, po0.05
VPT (max¼42) 21.2 (3.9) 23.4 (2.3) t¼1.8, N.S.
L1¼first language, L2 ¼second language, WRAT ¼Wide Range Achievement Test,
M¼male, F¼female, RT ¼response time, NWRepEng ¼nonword repetition in
English, NWRepJap¼nonword repetition in Japanese, VPT ¼Visual Patterns Test,
N.S.¼Not Significant.
Fig. 1. Schematics of phonological and visual one-back matching tasks. The
paradigm used was a block design with alternating 24-s task blocks and 15-s rest
blocks. In each task block, a fixation cross appeared at the center of the visual
display, and then 24 words were presented at a rate of 1/second. The stimulus
duration was 250 ms followed by a 750 ms blank period, during which participants
were guided to press a button when stimuli presented in succession were identical
phonologically (the phonological task) and visually (the visual task). To minimize
visual strategies during the phonological task, successive words were printed in
alternating fonts/styles that were significantly different (Arial regular vs. italic for
English words; Mincho vs. Gyosho for Kana words). The Kana words from the top
mean “friend”,“socks”,“fruits”, and “fruits”.
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–2270 2263

proficiency of L2 Kana reading did not reach the native-level
despite their highly accurate performance. Hence, individuals in
both L2 groups were not equal-bilinguals between L1 and L2, with
their L1 as the more dominant language. With respect to logo-
graphic Kanji (total number of words¼60), the mean accuracy
score was significantly lower in the Japanese-L2 group (mean
score7SD ¼21.1 76.1) than the Japanese-L1 group (mean
score7SD ¼56.8 72.5), with t¼20.9, po0.001. Because the two
groups' accuracy performance on Kanji reading was not compar-
able, Kanji stimuli were not used in the fMRI experiments.
3.1.2. Non-verbal IQ and memory tasks
All participants' scores on the Raven's Advanced Progressive
Matrices were within the normal range (mean score7SD¼29.17
5.8 and 27.374.6 for the J1/E2 and the E1/J2 group, respectively).
This confirmed no inclusion of intellectually impaired participants in
the study. No group differences were observed for either the Raven's
Advanced Progressive Matrices (t¼0.9, p¼0.37) or the Visual Pat-
terns Test (total number of patterns ¼42, the mean accuracy 7SD ¼
21.2 73.9 in the E1/J2 group and 23.472.3 in the J1/E2 group; t¼1.8,
p¼0.08). For the nonword repetition task using English sounds, the
mean score was significantly higher in the E1/J2 group (mean
score7SD ¼15 .9 71. 5) th an th e J 1/ E2 gro up (m ea n s co re 7SD¼
13. 3 72.7), with t¼3.0, po0.01. Similarly, for the nonword repetition
task using Japanese morae, the J1/E2 group mean score was
significantly higher (mean score7SD¼32.274.5) than the E1/J2
group (mean score7SD ¼28.573.2), with t¼2.5, po0.05.
3.2. Task performance in the scanner
For the main phonological one-back matching task, no differ-
ences were observed between L1 and L2 groups in either accuracy
or response time for word reading. More specifically, the mean
accuracy (%)7SD for English words was 97.870.8 in the E1/J2
group and 94.6771.4 in the J1/E2 group (t¼1.98, p¼0.06),
whereas the mean accuracy (%) 7SD for Kana words was
96.370.8 in the J1/E2 group and 95.171.3 in the E1/J2 group
(t¼0.75, p¼0.46). The mean response time (s) 7SD for English
words was 0.5170.01 in the E1/J2 group and 0.52 70.02 in the
J1/E2 group (t¼0.25, p¼0.81), whereas the mean response time
(s) for Kana words was 0.5170.02 in the J1/E2 group and
0.5370.01 in the E1/J2 group (t¼0.61, p¼0.55). These results
suggest that any group differences recorded in brain activation
were not likely to have arisen purely from differences in task
difficulty. For the control visual one-back matching task, no group
difference was observed in either the mean accuracy (%) or mean
response time (s): mean accuracy (%)7SD ¼86.2 710.4 in the
J1/E2 group and 87.479.6 in the E1/J2 group; t¼0.32, p¼0.75;
mean reaction time (s)7SD¼0.68 70.04 in the J1/E2 group and
0.6670.03 in the E1/J2 group; t¼1.51, p¼0.14.
3.3. Brain activation
3.3.1. Whole-brain analysis
During the task of interest in the current study, the phonolo-
gical one-match task, both groups significantly activated brain
regions considered to be parts of the reading network (Pugh et al.,
2000) for both word conditions (Fig. 2,Table 2 for English words,
and Table 3 for Kana words). Despite the similarities in activation
patterns, there was greater activation in the L2 group than in the
L1 group for the L2 word condition (the L2 group effect). More
interestingly, the identified clusters/regions showing the L2 group
effect were located in functionally different networks for the two
groups (Table 4). Specifically, for English words, the left superior
parietal lobule, extending into the supramarginal gyrus (L-SPL/
SMG, peak MNI x¼#35, y¼#52, z¼38, z¼3.1; 83 voxels)
exhibited greater activation in the J1/E2 group than the E1/J2
group (Table 4 and Fig. 3A). For Kana words, the left lingual gyrus,
extending into the lateral occipital complex (L-LG/LOC, peak MNI
x¼#16, y¼#72, z¼12, Z¼4.0; 264 voxels) exhibited greater
activation in the E1/J2 group than the J1/E2 group (Table 4 and
Fig. 4A). Similarly, this L2 group effect for Kana words was
observed in the right lingual gyrus (R-LG, peak MNI x¼8,
y¼#72, z¼12, Z¼4.3, 219 voxels) but this R-LG cluster did not
Fig. 2. Brain regions activated during (A) English word condition and (B) Kana word condition. L1 group and L2 group activated largely overlapping cortical regions within
the known reading network: clusters in red and blue represent activation patterns in the Japanese-L1/English-L2 group and the English-L1/Japanese-L2 group, respectively.
Z42.3, po0.05, corrected (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–22702264

extend into the lateral occipital complex (Table 4 and Fig. 5A).
Of note, there was no L1 group effect: no region was found to be
more activated in L1 group than L2 group for either word
condition.
For the control visual one-back matching task, both groups
showed activation in a wide range of brain regions, compared to
rest (Supplementary Fig. 1 and Supplementary Table 1). However,
there was no significant between-group difference in brain activa-
tion associated with visual processing of unpronounceable Tibetan
letter strings. This is consistent with the behavioral results for
performance accuracy on this task. Hence, any differences
observed in the phonological one-back matching task cannot be
attributed to differences in fundamental visual processing
competency.
3.3.2. Region of interest (ROI) analysis
To further characterize the patterns of brain activation in the
two regions identified from the above GLM analyses as being
associated with the L2 group effect (greater activation in the L2
group than the L1 group), we performed a region of interest
analysis for the L-SPL/SMG (Fig. 3B), L-LG/LOC (Fig. 4B), and R-LG
(Fig. 5B). For each ROI, we compared the BOLD signal change
between the groups for each word condition as well as between
the word conditions within each group, using unpaired and paired
t-tests, respectively.
3.3.2.1. L-SPL/SMG. For the English word condition, the BOLD
signal change was significantly stronger in the J1/E2 group than
the E1/J2 group (t¼2.78, po0.01) confirming the L2 group effect
for the English word condition. There was no group difference in
this region's activation for the Kana word condition: no L2 group
effect for the Kana word condition in the L-SPL/SMG (t¼1.23,
p¼0.23). We also compared the activations between the two word
conditions within each group. Within the J1/E2 group, the
activation was significantly stronger for L2 English words than
their L1 Kana words (t¼2.81, po0.05). However, the E1/J2 group
exhibited no significant difference in the L-SPL/SMG activation
between their L1 and L2 word conditions (t¼1.90, p¼0.08). The
main result here is therefore that the level of this region's
activation was the highest when the J1/E2 group processed
English words.
3.3.2.2. L-LG/LOC. For the Kana word condition, the BOLD signal
change in the L-LG/LOC was significantly stronger in the E1/J2 than
the J1/E2 group (t¼2.18, po0.05) confirming the L2 group effect
for Kana words. No group differences in this region′s activation
were observed for English words (t¼0.94, p¼0.35). Within each
group, the E1/J2 group exhibited greater L-LG/LOC activation for
Kana words than English words (t¼2.71, po0.05) but no
significant L1-L2 difference was observed in the J1/E2 group,
(t¼0.45, p¼0.66). The main finding here is that this region′s
activation was highest when the Japanese-L2 group processed
Kana words. Of note, no significant activation of this region was
reported for either Kana words or English words in either group.
3.3.2.3. R-LG. Similar to L-LG/LOC, the BOLD signal change in the
R-LG was significantly stronger in the E1/J2 group than the J1/E2
group for the Kana word condition (t¼2.11, po0.05). No group
difference in this region's activation was observed for English
words (t¼0.72, p¼0.47). Within the E1/J2 group, the activation
was significantly stronger for L2 Kana words than their L1 English
words (t¼2.50, po0.05). However, the J1/E2 group exhibited no
significant difference in the R-LG activation between their L1 and
L2 word conditions (t¼0.47, p¼0.65). The main finding here is
that this region′s activation was highest when the E1/J2 group
Table 2
Peak MNI coordinates of significant clusters for English words.
English words
Japanese-L1/English-L2 group English-L1/Japanese-L2 group
xyzZ
score
vox # xyzZ
score
vox #
SMA/AC #2 4 52 5.30 1148 10 14 40 4.11 608
L-IFG/PCG #44 8 26 4.24 4486 #42 2 26 5.13 1632
R-IFG/PCG 48 6 32 4.08 3631 48 8 24 4.51 1201
L-IPL/SPL #24 #62 48 4.38 1260 #26 #68 44 3.54 1023
L-SMG
n
#42 #38 44 3.82 #44 #44 40 3.19
R-IPL/SPL 30 #66 4 4 4.14 2005 34 #62 50 3.78 1221
R-SMG
n
44 #42 38 3.29
L-FFG/LOC #40 #52 #22 7.78 2676 #38 #66 #14 4.26 2583
R-FFG/LOC 44 #68 #18 5.63 2139 40 #60 #18 4.13 1719
Cerebellum 8 #76 #30 4.82 896
L1¼first language, L2 ¼second language, L ¼left, R¼right, SMA/AC¼supplemen-
tary motor area/anterior cingulate, IFG/PCG¼inferior frontal gyrus/precentral
gyrus, IPL/SPL¼inferior parietal lobule/superior parietal lobule, SMG ¼supramar-
ginal gyrus, FFG/LOC¼fusiform gyrus/lateral occipital complex, vox¼voxel; Z42.3,
po0.05, corrected.
n
sub-cluster.
Table 3
Peak MNI coordinates of significant activation clusters for Kana words.
Kana words
Japanese-L1/English-L2 group English-L1/Japanese-L2 group
xyzZ
score
vox # xyzZ
score
vox #
SMA/AC #2 0 56 4.62 1015 #4#2 58 4.33 1186
L-IFG/PCG #46 4 30 4.39 3697 #48 4 32 4.23 4384
R-IFG/PCG 46 4 28 4.09 1073 48 10 24 3.87 3222
L-IPL/SPL #24 #66 48 4.70 1454 #30 #62 40 4.70 2528
L-SMG
n
#42 #52 46 5.79 #36 #48 40 3.46
R-IPL/SPL 32 #68 50 4.46 1170 30 #62 4 4 3.91 1314
R-SMG
n
L-FFG/LOC #44 #76 #18 5.30 2410 #40 #72 #18 4.17 3792
R-FFG/LOC 44 #74 #18 4.69 1660 42 #62 #10 4.16 4424
Cerebellum #8#76 #30 3.94 760 #6#78 #38 4.48 1896
L1¼first language, L2 ¼second language, L ¼left, R¼right, SMA/AC¼supplemen-
tary motor area/anterior cingulate, IFG/PCG¼inferior frontal gyrus/precentral
gyrus, IPL/SPL¼inferior parietal lobule/superior parietal lobule, SMG ¼supramar-
ginal gyrus, FFG/LOC¼fusiform gyrus/lateral occipital complex, vox¼voxel; Z42.3,
po0.05, corrected.
n
Sub-cluster.
Table 4
Peak MNI coordinates of significant differences between the L1 and L2 groups.
L2 Group4L1 Group
English words Kana words
xyzZscore vox # xyzZscore vox #
L-SPL/SMG #35 #52 38 3.2 83
L-LG #16 #72 12 4.0 264
L-LOC
n
#40 #84 6 3.3
R-LG 8 #72 12 4.3 219
L1¼first language, L2 ¼second language, L ¼left, R ¼right, SPL/SMG¼superior
parietal lobule/supramarginal gyrus, LG¼lingual gyrus, LOC¼lateral occipital
complex.
n
Sub-region of L-LG, Z42.3, po0.05, corrected.
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–2270 2265

processed Kana words. Similar to L-LG/LOC, no significant
activation of this region was reported for either Kana words or
English words in either group.
3.4. Brain-behavior relationships
We examined whether the individual differences in literacy
and cognitive abilities measured outside the scanner were corre-
lated with the BOLD signal change in these regions/clusters (L-SPL/
SMG, L-LG/LOC, and R-LG). Each region's activity was plotted as a
function of competence in word reading (both response time and
accuracy for the WRAT; only the response time for Kana Word
Reading), phonological short-term memory (the accuracy of the
NWRep in either English or Japanese), and visual short-term
memory (accuracy in the Visual Patterns Test). The response time,
rather than the accuracy, was used to represent word reading
competence because there was no variability in the accuracy
(ceiling effect) for Kana Word Reading in the J1/E2 group.
The BOLD signal changes in L-SPL/SMG for the English word
condition exhibited a negative correlation with WRAT response
time (R
2
¼0.38, po0.05 in Fig. 3C) and a positive correlation with
WRAT accuracy (R
2
¼0.50, po0.01) as well as the NWRep accuracy
(R
2
¼0.41, po0.01 in Fig. 3D) in the J1/E2 group. That is, the J1/E2
readers who had shorter response times (i.e. more fluent reading)
and higher accuracy for English word reading tended to exhibit
stronger L-SPL/SMG activation for L2 English words. Similarly, the
J1/E2 readers who had higher phonological short-term memory
for nonwords in English tended to exhibit stronger L-SPL/SMG
activation for L2 English words. Such brain-behavior relationships
were not seen in the E1/J2 group. Note that the E1/J2 group
showed a trend (statistically non-significant) for the opposite
brain-behavioral pattern, a negative relationship for phonological
short-term memory. A measure of visual short-term memory, the
VPT, was not a significant predictor of L-LPL/SMG activation in
either group (Fig. 3E).
The BOLD signal changes in L-LG/LOC and R-LG during Kana
reading correlated positively with the Kana Word Reading
response time in the E1/J2 group: L-LG/LOC, R
2
¼0.32, po0.05 in
Fig. 4C; R-LG, R
2
¼0.57, po0.01 in Fig. 5C. Similarly, they positively
correlated with the VPT accuracy in the E1/J2 group: L-LG/LOC,
R
2
¼0.49, po0.01 in Fig. 4E; R-LG, R
2
¼0.47, po0.01 in Fig. 5E.
Thus, the E1/J2 readers with longer response times (i.e. less fluent
reading) for Kana word reading and those who had higher visual
short-term memory tended to exhibit stronger bilateral LG activa-
tion for L2 Kana words. Such significant brain-behavior relation-
ships were seen only for visual short-term memory in the J1/E2
group (R
2
¼0.30, po0.05 for L-LG/LOC; R
2
¼0.31, po0.50 for
R-LG). A measure of phonological short-term memory, the NWRep
in Japanese, was not significantly correlated with either L-LG/LOC
or R-LG activation in either group (Figs. 4 and 5D).
3.5. A secondary analysis
To attempt to further clarify the activation pattern of the
lingual gyrus, which showed the L2 group effect for the Kana
condition but did not show significant activation during either the
Kana or English condition, we performed a secondary region of
interest analysis. Both L-LG/LOC and R-LOC were examined during
Fig. 3. The L2 group effect for English words during the phonological one-back matching task. (A) The left superior parietal lobule/supramarginal gyrus (L-SPL/SMG),
exhibiting greater activation in the English-L2 group than the English-L1 group during the English word condition, Z42.3, po0.05, corrected. (B) BOLD signal changes in the
L-SPL/SMG for each word condition for each group. (C) The correlation between the L-SPL/SMG activation and response time (RT) of English word reading, measured by the
wide range achievement test (WRAT). (D) The correlation between the L-SPL/SMG activation and accuracy in the English nonword repetition (NWRep) task. (E) The
correlation between the L-SPL/SMG activation and accuracy in the Visual Patterns Test (VPT). Graph bars/dots in red and blue represent Japanese-L1/English-L2 group and
the English-L1/Japanese-L2 group, respectively.
n
po0.05,
nn
po0.01, n.s.¼not significant, s¼seconds (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article).
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–22702266

the phonological and visual tasks (i.e., visually unfamiliar and
unpronounceable Tibetan letter strings). In the J1/E2 group, the
activation in L-LG/LOC was highest in the Tibetan control condi-
tion (relative to Kana, t¼4.21, po0.01; relative to English, t¼3.76,
po0.01) and there was no difference in L-LG/LOC activation
between L1 Kana and L2 English. This pattern was not observed
in R-LG where the activation did not differ among the three
conditions. In the E1/J2 group, the activation in L-LG/LOC was
significantly higher for the Tibetan control condition than the L1
English condition (t¼2.81, po0.05) but not the L2 Kana condition.
No significant group difference was noted between the L1 English
and the L2 Kana conditions. Similar to L-LG/LOC, R-LG showed no
difference among the three conditions in either group.
4. Discussion
We compared cerebral activity in response to single word
reading of alphabetic English and syllabic Japanese Kana between
L1 and L2 groups. During L2 word reading, each L2 group recruited
brain regions similar to those employed by the corresponding L1
group. More importantly, however, we found different patterns
associated with the L2 group effect (greater activation in the L2
group than the L1 group) depending on the writing system being
learned as L2. Specifically, the Japanese-L1/English-L2 (J1/E2)
group exhibited stronger activation in the left superior parietal
lobule/supramarignal gyrus (L-SPL/SMG) during reading of English
words, whereas the English-L1/Japanese-L2 (E1/J2) group exhib-
ited stronger activation in the lingual gyrus in both left and right
hemispheres (L-LG and R-LG) during reading of Kana words. These
results suggest that different cortical areas are more strongly
recruited in late L2 readers than L1 readers depending on the
particular qualities of the language/writing system being learnt as
L2. Considering that both L2 groups were non-equal bilingual
groups (i.e. L2 is the less dominant or weaker language of the two),
the increased activation associated with the L2 group effect is
likely to reflect increased cognitive loads to achieve L2 reading in
the L2 groups.
4.1. The L2 group effect for English words
The L2 group effect during the English word condition was
observed in L-SPL/SMG, encompassing superior parietal lobule and
supramarginal gyrus, which are considered to be critical for
working memory performance in both visual and auditory/pho-
nological modalities (Koenigs, Barbey, Postle, & Grafman, 2009;
Wager & Smith, 2003) and for phonological storage (Paulesu, Frith,
& Frackowiak, 1993), respectively. Thus, the increased L-SPL/SMG
activation may reflect increased cognitive loads on phonological
processing for reading in L2 English, the less dominant language,
in the J1/E2 group.
As shown in the brain-behavior relationships, in the English-L2
group (J1/E2 group), stronger L-SPL/SMG activation was associated
with faster/more fluent reading of English words and greater
phonological short-term memory capacity. This result may indi-
cate that the J1/E2 readers accommodated to the additional
phonological demands of irregular L2 English orthography and
relied on phonological short-term memory for achieving L2 read-
ing. In contrast, the E1/J2 group showed the opposite trend (albeit
statistically non-significant) in the relationship between L-SPL/
Fig. 4. The L2 groupeffect for Kana words during the phonological one-back matching task –left lingual gyrus/lateral occipital complex (L-LG/LOC). (A) L-LG/LOC exhibitinggreater
activation in the Japanese-L2 group than the Japanese-L1 group during the Kana word condition, Z42.3, po0.05, corrected.(B) BOLD signal changes in the L-LG/LOC for eachword
condition for each group. (C) The correlation between the L-LG/LOC activation and response time (RT) of Kana word reading as measured by the Kana Word Reading test. (D) The
correlation between the L-LG/LOC activation and accuracy in the Japanese nonword repetition task (NWRep). (E) The correlation between the L-LG/LOC activation and accuracy in
the Visual Patterns Test (VPT). Graph bars/dots in red and blue represent Japanese-L1/English-L2 group and the English-L1/Japanese-L2 group, respectively.
n
po0.05,
nn
po0.01, n.s.¼not significant, s¼seconds (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–2270 2267

SMG activation and phonological short-term memory (i.e. the
stronger the L-SPL/SMG activation, the lower the phonological
short-term memory). This observation is consistent with Prat,
Keller, & Just (2007), in which high-capacity readers were more
efficient, defined by the combination of higher performance and
lower frontal and occipital activation. Therefore, the pattern of
cerebral efficiency in the J1/E2 readers, at least in regions involved
in phonological processing, may be different from the E1/J2 read-
ers, presumably because L2 reading in English requires learners
whose L1 has a regular orthography to accommodate to greater
phonological demands inherent in irregular English orthography.
The L-SPL/SMG activation was greater in the J1/E2 group than
the E1/J2 group only for the English word condition, but not for
the Kana word condition. When comparing its activation within
each group, for the J1/E2 group the L-SPL/SMG's activation was
greater for their L2 English words than their L1 Kana words. This is
consistent with a previous fMRI result in Japanese skilled readers
who learnt English as L2 (Buchweitz, Mason, Hasegawa, & Just,
2009). However, in the E1/J2 group, the L1–L2 difference in L-SPL/
SMG activation was not statistically significant, yet it appears that
L-SPL/SMG was more strongly activated for L2 Kana words than L1
English words (note: there was a non-statistically significant trend
in this group). These results from both L2 groups suggest that the
less dominant language (L2s in both groups) is generally more
phonological demanding, but the additional L-SPL/SMG activation
required for L2 reading is higher particularly when an L2 writing
system requires readers to accommodate to higher phonological
demands relative to their L1 writing system.
Given that the observed L2 group effect on L-SPL/SMG reflects
greater requirement for integrating orthographic and phonological
representations in English words in readers with regular L1
orthography, a lesser level of phonological demands may have
been required for L2 Kana words in the E1/J2 group (thus
explaining why there is no observed L2 group effect in L-SPL/
SMG during the Kana word condition). This is because E1/J2
readers' phonological skills, which enable reading in irregular
English orthography, may be sufficient enough to read words in
regular orthographies. This interpretation is consistent with the
psycholinguistic grain size theory of reading development, which
proposes that different grain sizes of orthographic representations
(fine-grained exemplified by English, coarse-grained exemplified
by Spanish and Finnish) result in the development of different
reading strategies (Ziegler & Goswami, 2005).
4.2. The L2 group effect for Kana words
The L2 group effect during the Kana word condition was seen in
the lingual gyrus in both the left and right hemispheres (L-LG/LOC
and R-LG). When looking at within-group activation in these
clusters, both L-LG/LOC and R-LG exhibited greater activation for
L2 Kana reading than L1 English reading within the E1/J2 group,
but such a L1–L2 difference was absent in the J1/L2 group. The
activation of each L-LG/LOC and R-LG for Kana words was
associated with higher visual short-term memory skills in the
E1/J2 group, indicating greater visual processing involved in Kana
reading in this group. However, a greater involvement of LG
functions may not necessarily be efficient for L2 reading, given
that stronger LG activation was associated with slower or less
fluent reading of Kana words in the E1/J2 group.
Fig. 5. The L2 group effect for Kana words during the phonological one-back matching task –right lingual gyrus (R-LG). (A) The right lingual gyrus (R-LG), exhibiting greater
activation in the Japanese-L2 group than the Japanese-L1 group during the Kana word condition, Z42.3, po0.05, corrected. (B) BOLD signal changes in the R-LG for each word
condition for each group. (C) The correlation between the R-LG activation and response time (RT) of Kana word reading as measured by the Kana Word Reading task. (D) The
correlation between the R-LG activation and accuracy in the Japanese nonword repetition (NWRep) task. (E) Thecorrelation between the R-LG activation and accuracy in the Visual
Patterns Test (VPT). Graph bars/dots in red and blue represent the Japanese-L1/English-L2 group and the English-L1/Japanese-L2 group, respectively.
n
po0.05,
nn
po0.01, n.s.¼not significant, s¼seconds (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–22702268

This brain-behavior relationship is consistent with Uta Frith′s
three-stage theory of word recognition (1985), proposing that
reading beginners (i.e., less skilled readers) recognize symbols and
words visually (i.e., the logographic stage) before shifting to more
efficient stages (e.g., the orthographic stage where grapheme-
phoneme correspondence plays a key role). Although the E1/J2
readers in the current study were neither reading beginners nor
performing worse than the J1/E2 readers (i.e. no group difference
in accuracy) during the task inside the scanner, they were
undoubtedly less fluent readers of Kana words than the J1/E2
readers, as evidenced by their longer response times for Kana
word reading outside the scanner.
In the fMRI literature, the lingual gyrus is not considered as a
core region involved in visual short-term memory (Christophel,
Hebart, & Haynes, 2012;Haxby, Petit, Ungerleider, & Courtney,
2000; see review by Smith & Jonides, 1997), but is often activated
by visual short-term memory tasks, particularly those involving
maintenance (Harrison & Tong, 2009;Konstantinou, Bahrami,
Rees, & Lavie, 2012;Sneve, Alnaes, Endestad, Greenlee, &
Magnussen, 2012). During word reading, the lingual gyrus is
believed to be specifically responsive to the physical length of
stimuli (i.e., letter-strings) (Indefrey et al. 1997;Mechelli,
Humphreys, Mayall, Olson, & Price, 2000), and to visual familiarity
(i.e., greater activation for false fonts than words) (Tagamets,
Novick, Chalmers, & Friedman, 2000). Given that visual word
length was matched between Kana and English words in the
current study, it is likely that the stronger LG activation reflects the
greater visual processing demands associated with the lower
visual familiarity with Kana words in the E1/J2 group.
This assumption, focusing on a close relationship between LG
activation and visual familiarity, is supported by the secondary
analysis (Supplementary Fig. 2). In the E1/J2 group, the L-LG/LOC
activation for the visually unfamiliar and unpronounceable Tibetan
strings was higher only relative to their L1 English words (not
relative to their L2 Kana words). In contrast, the J1/E2 group
exhibited significantly greater L-LG activation for Tibetan letter
strings relative to both their L1 Kana and L2 English words. These
results led us to interpret that visual features of Kana symbols
may remain unfamiliar (or may not become entirely familiar) to
the E1/J2 group.
If our interpretation is true enough to result in the increased LG
activation in the E1/J2 group, an obvious question arising here is
why this L2 group effect on LG was present only in the E1/J2 group
but not in the J1/E2 group. A possible answer is due to the un-
equal level of visual exposure to L2 symbols between the two
groups. Kana symbols, visually distinct from the Roman alphabets
used in English orthography (Fig. 1), are unique to Japanese
orthography and thus rarely seen outside Japan. In contrast,
Roman alphabetic symbols are widely used, and indeed are seen
everywhere in Japan (e.g. in commercial logos, advertising, videos
and across the world wide web). Hence, the level of visual
familiarity to L2 symbols may be much stronger in the J1/E2
group (i.e. Roman alphabets in L2) than the E1/J2 group (i.e. Kana
symbols in L2).
However, it should be emphasized here that mere visual
perception of Roman alphabets is unlikely to enable children to
systematically learn how to phonologically decode words, parti-
cularly in irregular English orthography (Manolitsis, Georgiou,
Stephenson, & Parrila, 2009). Instead, it is more likely that earlier
visual exposure to Roman alphabets in the J1/E2 group increases
the visual familiarity of Roman alphabetic symbols/words, which
may facilitate visual word recognition when they read L2 English
words (i.e., strings of alphabetic symbols/letters). Further research
is required to determine if L2 readers of Japanese with early
exposure to Kana symbols, unlike our participants who started
learning to read Kana during adolescence, show attenuated LG
activation during Kana word reading. This would further clarify
the role of the LG in L2 word reading, particularly considering that
both groups, even the E1/J2 group (which showed the L2 group
effect for Kana), exhibited low/non-statistically significant LG
activation for both word conditions.
5. Conclusion and educational implications
We demonstrate here for the first time that cerebral correlates
of increased cognitive loads to achieve L2 reading are dependent
upon linguistic differences between L1 and L2, and that different
brain regions play an important role in L2 reading in different
writing systems, at least in late L2 readers. In other words,
L2 reading depends both on which language is the learners' L1
and which language is to be learnt as their L2. Late L2 readers of
English, whose L1 writing systems have a more regular orthogra-
phy (e.g. Japanese Kana, Italian, Spanish), may need to meet the
greater phonological demands of the irregular orthography inher-
ent in English. This suggests that reading programs designed for
improving phonological and orthographic skills in English reading
(e.g. Orton Gillingham, www.ortonacademy.org) can be effective
for the majority of late L2 readers of English, considering that
many writing systems have more regular orthographies than
English. In contrast, late L2 readers of Japanese, whose L1 uses
the Roman alphabet, need to adapt to the greater visual demands
of L2 symbols. For these late L2 learners (e.g. learning Kana,
Arabic), familiarization with the visual features of the L2 symbols
is an important initial step. Consequently, language programs that
place a heavy early emphasis on accurate visual recognition ought
to be more effective in the long run. Overall, successful teaching of
L2 reading, particularly for late adult learners, should take into
account the important differences between L1 and L2 scripts.
Acknowledgment
This study was part of the doctoral research of M.S.K., which
was conducted at the University of Oxford, and supported by an
Oxford–Kobe scholarship and the Dyslexia Research Trust. We also
acknowledge the cooperation of Oxford Centre for Functional
Magnetic Resonance Imaging of the Brain (FMRIB), supporting
data collection.
Appendix A. Supporting information
Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.neuropsychologia.
2013.08.002.
References
Amano, S., & Kondo, T. (1999). NTT database series Nihongo-no Goitokusei (Lexical
properties of Japanese). Tokyo: Sanseido.
Annett, M. (1970). A classification of hand preference by association analysis. British
Journal of Psychology,61, 303–321.
Au, T. K., Knightly, L. M., Jun, S. A., & Oh, J. S. (2002). Overhearing a language during
childhood. Psychological Science,13, 238–243.
Bolger, D. J., Perfetti, C. A., & Schneider, W. (2005). Cross-cultural effect on the brain
revisited: universal structures plus writing system variation. Human Brain
Mapping,25, 92–104.
Bongaerts, T. (2005). Introduction: Ultimate attainment and the critical period
hypothesis for second language acquisition.International Review of Applied
Linguistics in Language Teaching, 43, 259.
Buchweitz, A., Mason, R. A., Hasegawa, M., & Just, M. A. (2009). Japanese and
English sentence reading comprehension and writing systems: An fMRI study
of first and second language effects on brain activation. Bilingualism,12,
141–151.
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–2270 2269

Buck, R., Singhal, H., Arora, J., Schlitt, H., & Constable, R. T. (2008). Detecting change
in BOLD signal between sessions for atlas-based anatomical ROIs. NeuroImage,
40,1157–1165.
Chee, M. W., Tan, E. W., & Thiel, T. (1999). Mandarin and English single word
processing studied with functional magnetic resonance imaging. Journal of
Neuroscience,19, 3050–3056.
Christophel, T. B., Hebart, M. N., & Haynes, J. D. (2012). Decoding the contents of
visual short-term memory from human visual and parietal cortex. The Journal of
Neuroscience: the Official Journal of the Society for Neuroscience,32,
12983–12989.
De Luca, M., Zeri, F., Spinelli, D., & Zoccolotti, P. (2010). The acquisition of reading
fluency in an orthographically transparent language (Italian): an eye movement
longitudinal study. Medical Science Monitor,16, SC1–7.
Della Sala, S., Gray, C., Baddeley, A., Allamano, N., & Wilson, L. (1999). Pattern span:
a tool for unwelding visuo-spatial memory. Neuropsychologia,37, 1189–1199.
Flege, J. E. (1991). Age of learning affects the authenticity of voice-onset time (VOT)
in stop consonants produced in a second language. The Journal of the Acoustical
Society of America,89, 395–411.
Harrison, S. A., & Tong, F. (2009). Decoding reveals the contents of visual working
memory in early visual areas. Nature,458, 632–635.
Haxby, J. V., Petit, L., Ungerleider, L. G., & Courtney, S. M. (20 00). Distinguishing the
functional roles of multiple regions in distributed neural systems for visual
working memory. NeuroImage,11, 380–391.
Indefrey, P., Kleinschmidt, A., Merboldt, K. D., Kruger, G., Brown, C., Hagoort, P., et al.
(1997). Equivalent responses to lexical and nonlexical visual stimuli in occipital
cortex: A functional magnetic resonance imaging study. NeuroImage,5, 78–81.
Johnson, J. S., & Newport, E. L. (1989). Critical period effects in second language
learning: The influence of maturational state on the acquisition of English as a
second language. Cognitive Psychology,21, 60–99.
Klein, D., Zatorre, R. J., Chen, J. K., Milner, B., Crane, J., Belin, P., et al. (2006). Bilingual
brain organization: a functional magnetic resonance adaptation study. Neuro-
Image,31, 366–375.
Koenigs, M., Barbey, A. K., Postle, B. R., & Grafman, J. (2009). Superior parietal cortex
is critical for the manipulation of information in working memory. Journal of
Neuroscience,29, 14980–14986.
Konstantinou, N., Bahrami, B., Rees, G., & Lavie, N. (2012). Visual short-term
memory load reduces retinotopic cortex response to contrast. Journal of
Cognitive Neuroscience.,24(11), 2199–2210.
Kovelman, I., Baker, S. A., & Petitto, L.-A. (2007). Bilingual and monolingual brains
compared: A functional magnetic resonance imaging investigation of syntactic
processing and a possible “neural signature”of bilingualism. Journal of Cognitive
Neuroscience,20, 153–169.
Koyama, M. S., Hansen, P. C., & Stein, J. F. (2008). Logographic Kanji versus
phonographic Kana in literacy acquisition: how important are visual and
phonological skills? Annals of the New York Academy of Sciences,1145,41–55.
Koyama, M. S., Stein, J. F., Stoodley, C. J., & Hansen, P. C. (2011). Functional MRI
evidence for the importance of visual short-term memory in logographic
reading. The European Journal of Neuroscience,33, 539–548.
Kucera, H., & Francis, W. N. (1967). Computational analysis of present-day American
English. Providence: Brown University Press.
Lawler, A. (2001). Archaeology. Writing gets a rewrite. Science,292, 2418–2420.
Lenneberg, E. H. (1967). Biological foundations of language. New York: Wiley
(Chapter Chapter).
Leonard, M. K., Brown, T. T., Travis, K. E., Gharapetian, L., Hagler, D. J., Jr., Dale, A. M.,
et al. (2010). Spatiotemporal dynamics of bilingual word processing. Neuro-
Image,49, 3286–3294.
Long, M. (1990). Maturational constraints on language development. Studies in
Second Language Acquisition,12, 251–285.
Manolitsis, G., Georgiou, G., Stephenson, K., & Parrila, R. (2009). Beginning to read
across languages varying in orthographic consistency: Comparing the effects of
non-cognitive and cognitive predictors. Learning and Instruction,19, 466–480.
Marian, V., Shildkrot, Y., Blumenfeld, H. K., Kaushanskaya, M., Faroqi-Shah, Y., &
Hirsch, J. (2007). Cortical activation during word processing in late bilinguals:
Similarities and differences as revealed by functional magnetic resonance
imaging. Journal of Clinical and Experimental Neuropsychology,29, 247–265.
Mechelli, A., Humphreys, G. W., Mayall, K., Olson, A., & Price, C. J. (2000).
Differential effects of word length and visual contrast in the fusiform and
lingual gyri during reading. Proceedings. Biological sciences/The Royal Society,
267, 1909–1913.
Nakamura, K., Dehaene, S., Jobert, A., Le Bihan, D., & Kouider, S. (2005). Subliminal
convergence of Kanji and Kana words: further evidence for functional parcella-
tion of the posterior temporal cortex in visual word perception. Journal of
Cognitive Neuroscience,17, 954–968.
Nelson, J. R., Liu, Y., Fiez, J., & Perfetti, C. A. (2009). Assimilation and accommodation
patterns in ventral occipitotemporal cortex in learning a second writing system.
Human Brain Mapping,30, 810–820.
Newport, E. L. (1990). Maturational constrains on language learning. Cognitive
Science,14,11–28.
Parker, J., Green, D. W., Grogan, A., Pliatsikas, C., Filippopolitis, K., Ali, N., et al.
(2012). Where, when and why brain activation differs for bilinguals and
monolinguals during picture naming and reading aloud. Cerebral Cortex,22,
892–902.
Paulesu, E., Frith, C. D., & Frackowiak, R. S. (1993). The neural correlates of the verbal
component of working memory. Nature,362, 342–345.
Prat, C. S., Keller, T. A., & Just, M. A. (2007). Individual differences in sentence
comprehension: a functional magnetic resonance imaging investigation of
syntactic and lexical processing demands. Journal of Cognitive Neuroscience,
19, 1950–1963.
Pugh, K. R., Mencl, W. E., Jenner, A. R., Katz, L., Frost, S. J., Lee, J. R., et al. (2000).
Functional neuroimaging studies of reading and reading disability (develop-
mental dyslexia). Mental Retardation and Developmental Disabilities Research
Reviews,6, 207–213.
Raven, J., Raven, J. C., & Court, J. H. (1998). Manual for Raven's progressive matrices
and vocabulary scales. Section 1: General overview. San Antonio, TX: Harcourt
Assessment.
Seki, A., Kassai, K., Uchiyama, H., & Koeda, T. (2008). Reading ability and
phonological awareness in Japanese children with dyslexia. Brain Development,
30, 179–188.
Seymour, P. H., Aro, M., & Erskine, J. M. (2003). Foundation literacy acquisition in
European orthographies. British Journal of Psychology,94, 143–174.
Smith, E. E., & Jonides, J. (1997). Working memory: A view from neuroimaging.
Cognitive Psychology,33,5–42.
Sneve, M. H., Alnaes, D., Endestad, T., Greenlee, M. W., & Magnussen, S. (2012).
Visual short-term memory: Activity supporting encoding and maintenance in
retinotopic visual cortex. NeuroImage,63,166–178.
Tagamets, M. A., Novick, J. M., Chalmers, M. L., & Friedman, R. B. (2000).
A parametric approach to orthographic processing in the brain: An fMRI study.
Journal of Cognitive Neuroscience,12, 281–297.
Wager, T. D., & Smith, E. E. (2003). Neuroimaging studies of working memory: a
meta-analysis. Cognitive, Affective, and Behavioral Neuroscience,3, 255–274.
Wagner, R., Torgessen, J., & Rashotte, C. (1999). The comprehensive test of phonolo-
gical processing examiner's manual. Austin, TX: ProEd.
Wilkinson, G. S. (1993). Wide range achievement test 3 (WRAT3). Psychological
Corporation.
Wilson, J. L., Jenkinson, M., de Araujo, I., Kringelbach, M. L., Rolls, E. T., & Jezzard, P.
(2002). Fast, fully automated global and local magnetic field optimization for
fMRI of the human brain. NeuroImage,17, 967–976.
Woolrich, M. W., Ripley, B. D., Brady, M., & Smith, S. M. (2001). Temporal
autocorrelation in univariate linear modeling of FMRI data. NeuroImage,14,
1370 –1386.
Yokoyama, S., Okamoto, H., Miyamoto, T., Yoshimoto, K., Kim, J., Iwata, K., et al.
(2006). Cortical activation in the processing of passive sentences in L1 and L2:
an fMRI study. NeuroImage,30,570–579.
Zhao, J., Li, Q. L., Wang, J. J., Yang, Y., Deng, Y., & Bi, H. Y. (2012). Neural basis of
phonological processing in second language reading: an fMRI study of Chinese
regularity effect. NeuroImage,60,419–425.
Ziegler, J. C., & Goswami, U. (2005). Reading acquisition, developmental dyslexia,
and skilled reading across languages: A psycholinguistic grain size theory.
Psychological Bulletin,131,3–29.
M.S. Koyama et al. / Neuropsychologia 51 (2013) 2261–22702270