A

lcbTtitts©

K

1

csmnt

t(

Hf

0d

Neuropsychologia 46 (2008) 1371–1378

Broca’s area plays a role in syntactic processing duringChinese reading comprehension

Suiping Wang a, Zude Zhu a, John X. Zhang b, Zhaoxin Wang c,Zhuangwei Xiao c, Huadong Xiang c, Hsuan-Chih Chen b,∗

a Department of Psychology, South China Normal University, Guangzhou, Chinab Chinese University of Hong Kong, Hong Kong S.A.R., China

c Guangdong Province Key Laboratory of Medical Molecular Imaging,Shantou University, Shantou, China

Received 3 August 2007; received in revised form 11 November 2007; accepted 17 December 2007Available online 28 December 2007

bstract

Event-related functional magnetic resonance imaging (ER-fMRI) was adopted to examine brain activation of syntactic processing in readingogographic Chinese. While fMRI data were obtained, 15 readers of Chinese read individually presented sentences and performed semanticongruency judgments on three kinds of sentences: Congruous sentences (CON), sentences with a semantic violation (SEM), and sentences withoth semantic and syntactic violations (SEM + SYN). The two types of incongruous sentences were matched in the degree of semantic plausibility.hree brain regions were identified showing significantly different levels of percent signal change across the three conditions, including BA44 in

he left inferior frontal gyrus (IFG) and both BA9 and BA10/46 in the left middle frontal gyrus (MFG). Furthermore, the mean percent signal changen the left BA44 observed in the SEM + SYN condition was significantly stronger than that in either the SEM or the CON condition, while the latter

wo conditions were at a similar level, implying an important role of this area in Chinese syntactic processing. These results, in conjunction withhose found in alphabetic scripts, suggest that there are some common neural substrates underlie syntactic processing across distinctive writingystems such as the logographic Chinese and the alphabetic English.

2007 Elsevier Ltd. All rights reserved.

a’s are

RMC2NB2t

eywords: Syntactic processing; fMRI; Chinese reading comprehension; Broc

. Introduction

Reading comprehension involves a variety of complex pro-esses, such as orthographic, phonological, semantic, andyntactic processing. Recently, a specific question attractinguch attention from those researchers who take the cognitive

euroscience approach to reading is whether semantic and syn-actic processes involve distinctive brain structures.

Although there is mixed evidence both for and againsthe inter-dependence of semantic and syntactic processingBookheimer, 2002; Dapretto & Bookheimer, 1999; Friederici,

∗ Corresponding author at: Department of Psychology, Chinese University ofong Kong, Shatin, N.T., Hong Kong. Tel.: +852 2609 6489;

ax: +852 2603 5019.E-mail address: hcchen@psy.cuhk.edu.hk (H.-C. Chen).

tdwt2l(2s

028-3932/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.oi:10.1016/j.neuropsychologia.2007.12.020

a

uschemeyer, Hahne, & Fiebach, 2003; Humphries, Binder,edler, & Liebenthal, 2006; Kaan & Swaab, 2002; Kang,onstable, Gore, & Avrutin, 1999; Keller, Carpenter, & Just,001; Kuperberg et al., 2003, 2000; Newman, Pancheva, Ozawa,eville, & Ullman, 2001; Ni et al., 2000; Roder, Stock, Neville,ien, & Rosler, 2002; Service, Helenius, Maury, & Salmelin,007; Stowe, Haverkort, & Zwarts, 2005; Vigneau et al., 2006),he overall pattern of results on the topic seems to indicatehat, at least partially, syntactic and semantic processes haveissociable neural representations. Briefly, semantic processingas found to be mainly associated with the middle and pos-

erior temporal regions (Bookheimer, 2002; Friederici et al.,003; Kuperberg et al., 2000; Ni et al., 2000), the left angu-

ar gyrus (Humphries et al., 2006), and BA45/47 in the left IFGBookheimer, 2002; Dapretto & Bookheimer, 1999; Hagoort,005; Newman, Just, Keller, Roth, & Carpenter, 2003), whereasyntactic processing was reported to be primarily associated

1 holog

wW&SF2eR22Pesti2r

agsu1sscotEpdsbeAcnsiirott

vbisepmpasss

etu

“

(c

(“

tspCIstpt

wciwnpwAt

“ts

tisv

(tTc

2

2.1. Participants

Fifteen right-handed native Chinese speakers (10 males and 5 females, agedbetween 22 and 25 years, with a mean age of 23.8 years) participated in thisstudy with informed consent. All had normal or corrected-to-normal vision andnone had a history of any psychiatric or neurological disorders.

1 This syntactical violation can possibly be detected in the following ways.

372 S. Wang et al. / Neuropsyc

ith Broca’s area (BA44/45) (Caplan, 2006; Caplan, Alpert, &aters, 1998; Caplan, Alpert, Waters, & Olivieri, 2000; DaprettoBookheimer, 1999; Embick, Marantz, Miyashita, O’Neil, &

akai, 2000; Fiebach, Schlesewsky, Lohmann, von Cramon, &riederici, 2005; Indefrey, Hagoort, Herzog, Seitz, & Brown,001; Just, Carpenter, Keller, Eddy, & Thulborn, 1996; Morot al., 2001; Ni et al., 2000; Stromswold, Caplan, Alpert, &auch, 1996), the left anterior temporal lobe (Friederici et al.,003; Friederici, Meyer, & von Cramon, 2000; Humphries et al.,006; Humphries, Love, Swinney, & Hickok, 2005; Noppeney &rice, 2004), and the left posterior temporal area (e.g., Friedericit al., 2003). Among various topics discussed in the previoustudies, one particularly relevant to the present research is abouthe Broca’s area. Some researchers proposed that this area is crit-cal for syntactic processing (e.g., Caplan, 2006; Embick et al.,000; Ni et al., 2000), though others have reported inconsistentesults (Kuperberg et al., 2003; Newman et al., 2001).

An open question is whether results from the studies usinglphabetic materials, can also apply to written Chinese, a logo-raphic writing system drastically different from alphabeticcripts in how meaning and speech are represented, and a scriptsed by nearly a quarter of the world’s population (Chen & Juola,982; Chen & Shu, 2001; Chen & Zhou, 1999). Note that oppo-ite theoretical predictions can be made, depending on whichide one takes in an ongoing universality/specificity debate con-erning the syntax and semantic relationship in Chinese. In fact,ne camp of researchers (e.g., Zhang, 1997a, 1997b) argueshat while syntax and semantics are relatively independent innglish, they are closely inter-related in Chinese. For exam-le, in English, there are explicit devices (e.g., inflectional orerivational morphology) to mark a word’s lexical category oryntactic function. The grammaticality of a sentence can usuallye determined independent of its semantics. In Chinese, how-ver, words do not generally have explicit grammatical markers.lso, there is no transparent correspondence between lexical

ategory and syntactic function—many verbs can be used asouns without any morphological changes. Hence, the syntactictatus of a Chinese word can be highly context-dependent so thatts semantic information is often needed to determine whethert fits the sentence grammatically. In contrast, another camp ofesearchers (e.g., Fan, 2007; Shi, 2000) insists that there are rig-rous syntactic rules in Chinese independent of semantics, evenhough such rules are much more flexible and opaque, relativeo those in English.

To test out the neural implications of the mentioned twoiews, Luke, Liu, Wai, Wan, and Tan (2002) conducted the firstrain imaging study comparing semantic and syntactic process-ng in Chinese and obtained results consistent with the languagepecific position mentioned above. Following the design of Kangt al. (1999), they asked subjects to do semantic or syntacticlausibility judgment task on Chinese phrases, including nor-al, syntactically unacceptable, and semantically unacceptable

hrases. Unacceptable phrases can also be called phrases with

violation. The results showed that the cortical regions for

yntactic analysis coincided with those for semantic analysis,uggesting that there was no unique brain region specific toyntactic processing in Chinese.

Ffiofo

ia 46 (2008) 1371–1378

However, the manipulation of syntactic acceptance in Luket al. (2002) may not be effective enough to reveal syntac-ic processing effects. Indeed, they constructed syntacticallynacceptable materials by reversing ADV-V phrase (e.g.,

”, meaning “left quickly”) to a V-ADV structure

“ ”). Note that, although the ADV-V structure isommonly seen, the V-ADV construct is also used in Chinese

e.g., “ ” and “ ”, meaning “come quickly” andgo quickly”).

More importantly, it is problematic to examine Chinese syn-actic processing by adopting exactly the same logic used intudies with alphabetic stimuli. In fact, Chen (1992, 1999)ointed out that the way semantics and syntax are related inhinese makes it difficult to induce a sheer syntactic violation.

ndeed, any syntactic violation in Chinese will always disruptemantic processing and be accompanied by a semantic viola-ion. To get around this difficulty, Chen designed a violationaradigm, shown to be more appropriate for investigating syn-actic processing in Chinese.

In the present study, adopting Chen’s (1992, 1999) paradigm,e followed up the study of Luke et al. (2002) and examined

ortical activations of syntactic and semantic processing dur-ng the reading of Chinese. Our goal was to test whether thereas any brain region specific to syntactic processing in Chi-ese, as reported in English. The basic manipulation in Chen’saradigm is to change a genuine word (usually a single-characterord) in a sentence to create two types of violation sentences.s shown in Table 1, a control (CON) sentence is both seman-

ically and syntactically appropriate with (a verb, meaning

cook”) being the predicate and (a noun, meaning “porridge”)he object. A semantic violation sentence, abbreviated as a SEMentence, differs from the control sentence by one word (e.g.,

he object noun was changed to another noun , mean-ng “shed”) to maintain its syntactic correctness but to becomeemantically nonsense. For the critical semantic plus syntacticiolation (SEM + SYN) sentence, the object noun in the sentence

) was changed to a verb ( , meaning “remove”) to violatehe syntactic acceptance and to become semantically nonsense.1

hus, this violation paradigm enables us to compare the variousonditions to tease apart semantic and syntactic processes.

. Materials and methods

irst, the readers may find that the second verb is not a proper object for therst verb, which requires a noun object. Alternatively, they may find that a nounbject is missing for the second verb (i.e., sometimes in Chinese, one verb mayollow another verb, provided that the second verb binds with a following nounbject to form a short phrase, which serves as the object of the first verb).

S. Wang et al. / Neuropsychologia 46 (2008) 1371–1378 1373

Table 1Illustration of the three experimental conditions with a sample congruous sentence and the two violation sentences constructed from it

T

2

cvcwcssrvc

u2asFsc(

i(cmdi(o5a(c

Svc

st5a(t1tBww

Mtawaw

cpoCiT5jssbfra

arget characters are underlined.

.2. Stimuli and procedure

As shown in Table 1, there were three conditions of trials, the congruousondition, the semantic violation condition, and the semantic and syntacticiolation condition, abbreviated as the CON, the SEM, and the SEM + SYNonditions, respectively. Sixty-three grammatically correct Chinese sentencesere used in the CON condition. Each of these sentences was modified by

hanging one target character in it to create two incongruous versions: oneemantic violation version for the SEM condition, and the other version with bothemantic and syntactic violations for the SEM + SYN condition. Each participanteceived all the 63 sentences, 21 sentences in the CON format, 21 in the SEMersion, and 21 in the SEM + SYN version, following proper across-subjectsounterbalancing.

The sentences were so constructed that the target characters were always gen-ine words. To eliminate ambiguity of word category (Rodd, Davis, & Johnsrude,005), the characters chosen for the CON and the SEM conditions can only serves a noun in Chinese and that chosen for the SEM + SYN condition can onlyerve as a verb. These target characters were of high frequency (Modern Chineserequency Dictionary, 1986) and closely matched across the three conditions introke number, word frequency and concreteness. In addition, the three targetharacters associated with each sentence were all of the same spatial structuree.g., left–right, or top–bottom).

The mean stroke number was 9.97 (S.D. = 2.73) for the target charactersn the CON condition, 10.05 (S.D. = 2.68) in the SEM condition, and 9.94S.D. = 3.14) in the SEM + SYN condition, with no statistical difference acrossonditions (F < 1). Also, the mean word frequency, in units of occurrence perillion, was 71.89 (S.D. = 106.45) for the target characters in the CON con-

ition, 71.77 (S.D. = 107.45) in the SEM condition, and 71.37 (S.D. = 104.01)n the SEM + SYN condition, with no statistical difference across conditionsF < 1). Furthermore, the mean word concreteness (rated by a separate groupf 24 participants on a 5-point scale, ranging from 1 = “extremely abstract” to= “extremely concrete and vivid”) was 4.67 (S.D. = 0.43) for the target char-cters in the CON condition, 4.68 (S.D. = 0.40) in the SEM condition, and 4.65S.D. = 0.38) in the SEM + SYN condition, with no statistical difference across

onditions (F (2, 46) = 1.22, p > 0.1).

Given that there was a semantic violation in both the SEM and theEM + SYN conditions, we first conducted a rating study to confirm that suchiolation was comparable across the two conditions and consequently not aonfounding factor when we directly compare the two conditions to isolate the

2

is

yntactic component. A separate group of 54 participants were recruited fromhe same participant pool to rate the semantic plausibility of all sentences on a-point scale (ranging from 1 = “extremely unacceptable” to 5 = “fully accept-ble”). The average scores were 3.91 (S.D. = 0.67), 2.31 (S.D. = 0.70), and 2.33S.D. = 0.65) for the CON, SEM, and SEM + SYN sentences, respectively. Thehree conditions significantly differed in terms of semantic plausibility (F (2,06) = 102.47, p < 0.001). Pair-wise t-tests indicated that while the two viola-ion conditions were significantly lower than the CON condition (p < 0.001 withonferroni correction), they had similar semantic plausibility scores (p > 0.5ith Bonferroni correction), suggesting that the extent of semantic violationere matched across the two conditions.

With several practice trials outside, all participants completed four functionalRI runs inside the scanner. Fourteen filler sentences were used in the first run

o familiarize participants with the experimental procedure inside the scannernd the related data were excluded from data analyses. The remaining three runsere the test ones, each containing seven CON sentences, seven SEM sentences,

nd seven SEM + SYN sentences, pseudo-randomly intermixed. All the test runsere 6 min and 12 s long, producing 186 time points (TR = 2 s).

The stimuli were programmed with the E-Prime software package (Psy-hology Software Tools, Pittsburgh, PA). In each trial, a complete sentence wasresented (i.e., all words from the sentence were presented at once) centeredn the screen with a MRI-compatible goggle system (Resonance Technologyompany, Los Angeles, CA) until a response was made. A fixation was shown

n the first and the last 30 s for passive viewing, serving as resting baseline.here were 24 trials (21 experimental and 3 filler sentences) in the remainingmin and 12 s period. Participants were asked to make a semantic congruency

udgment by pressing a left or a right button. Their response would turn off theentence display and turn on a fixation for a variable interval before the next trialtarted. Following typical event-related designs (Xiao et al., 2005), the intervaletween onsets of two consecutive sentences was pseudo-randomized to varyrom 10 to 16 s with a mean value of 13 s. Both reaction time and accuracy wereecorded. Trials with reaction time longer than 5 s were excluded from datanalyses.

.3. Image acquisition and data analyses

The scanning was conducted on a 1.5 T Philips MRI scanner at the Med-cal College of Shantou University, using a standard headcoil. Twenty axiallices covering the whole brain were acquired with a T2*-weighted gradient-

1 hologia 46 (2008) 1371–1378

eager(sfws

f

Fam&fAaetwesc

igtvafwt

arlw

cp

Table 2Mean reaction time and accuracy for all three conditions

Reaction time (ms) Accuracy (%)

Mean S.D. Mean S.D.

Congruous condition 2409 425 94.0 7.7Semantic violation condition 2482 660 96.2 4.1S

3

3

fd

3

stiv

wtsrc

Fr

374 S. Wang et al. / Neuropsyc

cho echo-planar imaging (EPI) pulse sequence (TR = 2 s, TE = 45 ms, flipngle = 90◦, FOV = 23 cm × 23 cm, matrix = 64 × 64, slice thickness = 6 mm, noap). Coplanar anatomical images were acquired with a T1-weighted spincho pulse sequence (matrix = 256 × 256, TR = 204 ms, TE = 14 ms). A high-esolution 3d volume was also acquired with spoiled gradient recalled echoSPGR) sequence for functional overlay and transformation. Participants layupine inside the scanner, wearing earphones and goggles specially designedor the MR environment, and holding a button box. Their head was restrainedith padding. They were told to keep their head still until they went out of the

canner and to have a short rest with eyes closed during the inter-runs interval.The behavioral data were analyzed using an analysis of variance (ANOVA)

or repeated measures with post hoc analyses Bonferroni corrected.The imaging data were analyzed using the AFNI package (Cox, 1996).

unctional datasets were preprocessed to remove linear drifts, outliers, motionrtifacts, and then spatial smoothed (FWHM = 5, Gaussian kernel) and nor-alized. After that, the data were transformed to Talairach space (TalairachTournoux, 1988) with 1 mm resolution. Individual hemo-dynamic responses

or each condition were estimated with the deconvolution program provided byFNI, using the six motion correction parameters (x, y, z translations, roll, pitch,

nd yaw) as covariates to remove residual motion artifacts. Following typicalvent-related designs similar to those in our previous research (Xiao et al., 2005),he onset of the stimuli or sentence in each trial marked the onset of a trial event,ith the three different types of sentences corresponding to three different trial

vents. For each participant, the deconvolution analysis produced three percentignal change in each voxel, one for each of the CON, SEM, and SEM + SYNonditions.

The percent signal changes were then submitted to the 3dANOVA programn AFNI for a voxel-wised random effect group analysis. The result of thisroup analysis was an F-map which had an F statistic for each voxel indicatinghe extent to which the condition factor modulated the brain activity in thatoxel. An activation map was obtained through filtering this F-map by withn inclusive mask and thresholded at p = 0.005 (uncorrected). To guard againstalse activations for multiple comparisons, the activation map was then clusteredith a minimal voxel size of 404 mm3, with which corrected p would be equaled

o 0.05 as calculated by Monte Carlo simulation.The inclusive mask was created as follows. An activation map was first cre-

ted for each condition by comparing the percent signal change in that conditionelative to the baseline. The map was converted to a binary mask map using a

iberal threshold (uncorrected p < 0.05). All three maps, one for each condition,ere then combined with a logical “OR” to make the inclusive mask.

Once the statistically activated regions were identified, a mean percent signalhange was then computed for each region and all participants to be used forair-wise post hoc comparisons across the three conditions.

iCt

ig. 1. Brain regions revealed by the ANOVA. Talairach Z coordinates are shown foright. Color bar represents p values. The brain regions are labeled as follows: 1, left B

emantic and syntacticviolation condition

2461 585 93.0 10.3

. Results

.1. Behavioral results

As shown in Table 2, no reliable significant differences wereound for either accuracy or reaction time across the three con-itions (Fs < 1).

.2. Imaging results

As shown in Fig. 1, the ANOVA identified three brain regionshowing significant different levels of activation across condi-ions, including the left BA44 in IFG, the left BA9 and BA10/46n MFG (p < 0.05, corrected). Detailed information of these acti-ations is provided in Table 3.

For each of three activated regions, the percent signal changeas calculated and is shown in Fig. 2. For the left BA44 area,

he SEM + SYN condition showed significantly greater signaltrength than either of the CON or the SEM conditions, but noeliable signal difference was found between the CON and SEMonditions.

For the left BA9 and BA10/46 areas, the signal was signif-cantly greater for the two violation conditions relative to theON condition, but no reliable difference was found between

he SEM and SEM + SYN conditions.

each axial slice on the left and X coordinate shown for the sagittal slice on theA44; 2, left BA9; 3, left BA10/46.

S. Wang et al. / Neuropsychologia 46 (2008) 1371–1378 1375

Table 3Detail information for brain regions activated in the ANOVA

Brain region Brodmann’s area Volume (mm3) Talairach coordinates

x y z

Left inferior frontal gyrus BA44 419 −45 10 18Left middle frontal gyrus BA9 2667 −45 12 32Left middle frontal gyrus BA10/46 406

Note. Coordinates specify the peak voxel in a cluster according to the Talairach and To

Fig. 2. Percent signal change (%) in all experimental conditions for activatedbrain regions revealed in the ANOVA. CON for the congruous condition, SEMft*

4

4r

sgcrlttcfswft

ttcvts

diSrn

tas

bpstptc

utttiwF

fasa211JTfsniC

ai

or the semantic violation condition, and SEM + SYN for the semantic and syn-actic violation condition. The post hoc comparisons were Bonferroni corrected.** for p < 0.001, ** for p < 0.01.

. Discussion

.1. Cortical regions contribute to syntactic processing ineading Chinese

Among the few regions identified in the AVOVA to be sen-itive to the experimental manipulation, the left BA44 showedreater activity in the SEM + SYN condition than in the SEMondition when the percentage of BOLD signal change in thisegion was compared across the two conditions directly. Fol-owing our design logic, this increased activity was attributedo the additional syntactic processing in the SEM + SYN condi-ion, beyond the processes shared by the SEM + SYN and SEMonditions. Such result cannot be attributed to complexity dif-erences in sentence structure, as all conditions involved theame set of sentences that differed only on target charactersith matched stimulus characteristics. The absence of any per-

ormance difference from the behavioral data also indicates thathe task difficulty was properly matched across conditions.

In addition, the semantic plausibility rating results assuredhat the brain regions identified from the comparison betweenhe two violation conditions can be attributed to syntactic pro-

essing without being confounded by the extent of semanticiolation. This leaves open a possibility that such regions, orhe left BA44 as found in the present study, may contribute toemantic processing to the same extent in the two violation con-

ip(p

−44 42 12

urnoux brain atlas. p value was set to 0.05, corrected for multiple comparisons.

itions. However, the result that the BA44 activation was similarn two specific conditions, one with semantic violation (i.e., theEM condition) and the other without (i.e., the CON condition),ejected this possibility and indicated that the left BA44 doesot have a strong role to play in semantic processing.

Together, the activation pattern in the left BA44 across allhe three conditions provides clear evidence to suggest that thisrea plays an important role in syntactic processing in Chineseentence reading.

Note, however, that the left BA44 may be not the onlyrain area for syntactic processing in Chinese. This is because,resumably, our participants were needed to perform certainyntactic operations in order to comprehend the stimuli in bothhe SEM and SEM + SYN conditions. Hence, those syntacticrocesses equally involved in the two violation conditions andheir associated brain regions would not be revealed from theomparison between them.

Key to our interest, the finding that there was a brain regionniquely involved in syntactic processing indicates either thathe networks for semantic and syntactic processing are not iden-ical but differentiated, or that the neural substrates for the twoypes of processing are at least partially dissociable, whichs consistent with what has been observed in many studiesith alphabetic stimuli (e.g., Dapretto & Bookheimer, 1999;riederici et al., 2003; Kang et al., 1999; Ni et al., 2000).

What is particularly interesting is that the area we identifiedor syntactic processing in Chinese, the left BA44, is the samerea implicated in syntactic processing which was found in manytudies from European languages (e.g., Caplan, 2006; Caplan etl., 1998, 2000; Dapretto & Bookheimer, 1999; Embick et al.,000; Fiebach et al., 2005; Indefrey et al., 2001; Just et al.,996; Moro et al., 2001; Ni et al., 2000; Stromswold et al.,996; but see Kuperberg et al., 2003; Newman et al., 2001) andapanese (Sakai, Homae, & Hashimoto, 2003; Sakai, Noguchi,akeuchi, & Watanabe, 2002). Hence, combined with the resultsrom many previous studies in alphabetic scripts, the presenttudy provides evidence showing that there are some commoneural substrates for syntactic processing across distinctive writ-ng systems, such as the alphabetic English and the logographichinese.

Our success in disclosing differentiate patterns of corticalctivation between syntactic and semantic processing in read-ng Chinese which has not been shown in previous research

s due largely to our adoption of Chen’s (1999) violationaradigm, which took into account linguistic features in Chinesee.g., the difficulty in creating a syntactic violation inde-endent of a semantic violation at the lexical level in the

1 holog

lclcPwati

4

BFr2pssdSwZicp

smeatiotba

iCiparrtmffiloanov

pgc

4

bcgSas

id(mwtbroprcc

swtpJ2isstitMB

5

fpraes

376 S. Wang et al. / Neuropsyc

anguage). However, this violation paradigm itself is not spe-ific to Chinese but can actually be implemented in many otheranguages. For example, in a similar way as in Table 1, onean construct English sentences with a SEM violation (e.g.,eter ate books) or a SEM + SYN violation (e.g., Peter aterite), although it is not really necessary for researchers to

dopt this paradigm for studies with English stimuli wherehey can easily manipulate syntactic and semantic violationsndependently.

.2. The left BA44 and syntactic processing

There are several possible mechanisms through which the leftA44 may contribute to syntactic processing in reading Chinese.or example, it has been suggested that the left BA44 activationeflects a certain disruption in syntactic processing (Moro et al.,001; Ni et al., 2000; Sakai et al., 2003), such as in building localhrasal structure. Specifically, the target word in the SEM + SYNentence as shown in Table 1, being a verb, does not make a plau-ible phrase with the preceding verb, and therefore a syntacticisruption is formed. The idea that the syntactic disruption in theEM + SYN condition could relate to phrase structure buildingas suggested by a previous ERP study (Ye, Luo, Friederici, &hou, 2006) using the same violation paradigm. In that study,

t was found that the SEM + SYN sentences elicited an ELANomponent which has been known to be correlated with localhrasal structure building.

Another possibility is that the left BA44 activation reflectsyntactic unification. Hagoort (2005), for instance, proposed aemory, unification, and control (MUC) model in which differ-

nt types of syntactic, phonological, and semantic informationre retrieved from memory and unified into global representa-ions. In this model, the left BA44 was suggested to be involvedn the unification of syntactic information. Following the logicf the MUC model, the SEM + SYN condition might requirehe unification of syntactic and semantic information, over andeyond that for semantic information alone, and might hencectivate the left BA44 with a higher level of activation.

Furthermore, the left BA44 activation may also reflectncreased cognitive load on verbal working memory. Indeed,aplan and Waters (1999) put forward a separate-sentence-

nterpretation-resource (SSIR) hypothesis in which theyostulated a type of working memory resource specific forssigning syntactic structure and related online processing. Thisesource, independent from the general verbal working memoryesources, was also proposed to reside in the Broca’s area. Inhe present study, the two violation conditions were carefully

atched in both stimulus property and task difficulty (as seenrom the behavioral data). Hence, they are not supposed to dif-er in general working memory load. The differential activationn the left BA44 across these two conditions might thus moreikely due to different loads on the specific type of working mem-ry for syntactic processing, like the one suggested by Caplan

nd Waters. In other words, the participants in our study mighteed to keep more syntactic information for later assignmentf syntactic structure once they encountered with the syntacticiolation in the SEM + SYN condition.

toar

ia 46 (2008) 1371–1378

However, we cannot base on the present results or those inrevious studies to verify the various mentioned mechanisms,iven the limited imaging research on Chinese syntactic pro-essing. Further research is needed to explore this issue.

.3. The left BA9 and BA10/46

For the left BA9 and BA10/46, the other two regions identifiedy the ANOVA, the activations in the SEM and the SEM + SYNonditions, while comparable to each other, were significantlyreater than those in the CON condition. Since the SEM andEM + SYN conditions all involved semantic processing whenviolation was detected, such activations can be attributed to

emantic processing.It is, however, possible that the BA9 and BA10/46 activation

n the two violations conditions, relative to that in the CON con-ition, reflected certain processes specific to violation in generale.g., the detection of violation and its resolution in workingemory), but not to semantic processing per se. Consistentith this possibility, Hagoort (2005) discovered evidence for

he involvement of BA9 and BA46 in attention control in ver-al tasks. Also, Petrides, Alivisatos, Meyer, & Evans (1993)eported the involvement of BA46 in verbal working mem-ry tasks. However, among previous studies using the violationaradigm, few had discussed the activation of these two corticalegions in various violation conditions, relative to the controlondition. Hence, it remains to be determined the precise impli-ations of BA9 and BA10/46 activations.

Note that although seldom reported in studies with alphabetictimuli, the left BA9 has been consistently found and associatedith semantic processing in Chinese studies at both the charac-

er/word level and the phrasal level, in addition to visuo-spatialrocessing (e.g., Booth et al., 2006; Luke et al., 2002; Mo, Liu,in, & Yang, 2005; Siok, Perfetti, Jin, & Tan, 2004; Tan et al.,000, 2001). These results suggest that the semantic processingn Chinese may involve some script-specific neural substrates,uch as the left BA9, different from those observed in alphabeticcripts. In contrast, the left BA46 has been found for seman-ic retrieval in Chinese (Mo et al., 2005), semantic processingn English at the sentence level (Ni et al., 2000), and seman-ic encoding and retrieval in English (Tulving, Kapur, Craik,

oscovitch, & Houle, 1994). Hence, the activation in the leftA46 may be less specific to Chinese than that in the left BA9.

. Conclusion

The present study shows that the Broca’s area was activatedor syntactic violation detection, when semantic factors wereroperly controlled. This result indicates that this area plays aole in syntactic processing in Chinese reading comprehension,conclusion consistent with many studies in alphabetic scripts

mphasizing Broca’s area in syntactic processing. We thereforeuggest that syntactic processing across distinctive writing sys-

ems may engage similar neural substrates. However, our resultf stronger activation of the left BA9 in semantic processinglso suggests that script-specific neural activations may appear ineading Chinese. Hence, the present findings reaffirm the impor-

holog

td

A

NNtDRta(SM

R

B

B

CC

C

C

C

C

C

C

C

C

D

E

F

F

F

F

H

H

H

I

J

K

K

K

K

K

L

M

M

M

N

N

N

N

P

R

S. Wang et al. / Neuropsyc

ance of cross-script research to reveal both the similarities andifferences between different writing systems.

cknowledgements

This research was supported by grants from the Nationalatural Science Foundation of China (30670700), Guangdongatural Science Foundation (06200524, 06025045), as well as

he Foundation for the Author of National Excellent Doctoralissertation of PR China (200208) to Suiping Wang, from theesearch Grants Council of the Hong Kong Special Adminis-

rative Region, China (CUHK4142/04H) to Hsuan-Chih Chen,nd from the National Natural Science Foundation of China30670702) to John X. Zhang. We thank Qiulin Wu, Linfa Wu,haoxing Chen in the Guangdong Province Key Laboratory ofedical Molecular Imaging for their helps in data collection.

eferences

ookheimer, S. (2002). Functional MRI of language: New approaches to under-standing the cortical organization of semantic processing. Annual Review ofNeuroscience, 25, 151–188.

ooth, J. R., Lu, D., Burman, D. D., Chou, T. L., Jin, Z., Peng, D. L., et al.(2006). Specialization of phonological and semantic processing in Chineseword reading. Brain Research, 1071, 197–207.

aplan, D. (2006). Why is Broca’s area involved in syntax? Cortex, 42, 469–471.aplan, D., Alpert, N., & Waters, G. (1998). Effects of syntactic structure and

propositional number on patterns of regional cerebral blood flow. Journal ofCognitive Neuroscience, 10, 541–552.

aplan, D., Alpert, N., Waters, G., & Olivieri, A. (2000). Activation of Broca’sarea by syntactic processing under conditions of concurrent articulation.Human Brain Mapping, 9, 65–71.

aplan, D., & Waters, G. S. (1999). Verbal working memory and sentencecomprehension. Behavioral and Brain Sciences, 22, 77–94.

hen, H.-C. (1992). Reading comprehension in Chinese: Implications fromcharacter reading times. In H.-C. Chen & O. J. L. Tzeng (Eds.), Languageprocessing in Chinese (pp. 175–205). North-Holland: Amsterdam.

hen, H.-C. (1999). How do readers of Chinese process words during readingfor comprehension? In J. Wang, A. W. Inhoff, & H.-C. Chen (Eds.), ReadingChinese script: A cognitive analysis (pp. 257–278). Mahwah, NJ: LawrenceErlbaum Associates Inc.

hen, H.-C., & Juola, J. F. (1982). Dimensions of lexical coding in Chinese andEnglish. Memory and Cognition, 10, 216–224.

hen, H.-C., & Shu, H. (2001). Lexical activation during the recognition ofChinese characters: Evidence against early phonological activation. Psy-chonomic Bulletin and Review, 8, 511–518.

hen, H.-C., & Zhou, X. (1999). Processing East Asian languages: An intro-duction. Language and Cognitive Processes, 14, 425–428.

ox, R. W. (1996). AFNI: Software for analysis and visualization of functionalmagnetic resonance neuroimages. Computers and Biomedical Research, 29,162–173.

apretto, M., & Bookheimer, S. Y. (1999). Form and content: Dissociatingsyntax and semantics in sentence comprehension. Neuron, 24, 427–432.

mbick, D., Marantz, A., Miyashita, Y., O’Neil, W., & Sakai, K. L. (2000).A syntactic specialization for Broca’s area. Proceedings of the NationalAcademy of Sciences of the United States of America, 97, 6150–6154.

an, X. (2007). Regularity and flexibility of grammatical constructions. ChineseLanguage Learning (Hanyu Xuexi in Chinese pinyin), 158, 3–11.

iebach, C. J., Schlesewsky, M., Lohmann, G., von Cramon, D. Y., & Friederici,A. D. (2005). Revisiting the role of Broca’s area in sentence processing: Syn-

tactic integration versus syntactic working memory. Human Brain Mapping,24, 79–91.

riederici, A. D., Meyer, M., & von Cramon, D. Y. (2000). Auditory languagecomprehension: An event-related fMRI study on the processing of syntacticand lexical information. Brain and Language, 75, 289–300.

R

ia 46 (2008) 1371–1378 1377

riederici, A. D., Ruschemeyer, S. A., Hahne, A., & Fiebach, C. J. (2003). Therole of left inferior frontal and superior temporal cortex in sentence compre-hension: Localizing syntactic and semantic processes. Cerebral Cortex, 13,170–177.

agoort, P. (2005). On Broca, brain, and binding: A new framework. Trends inCognitive Sciences, 9, 416–423.

umphries, C., Binder, J. R., Medler, D. A., & Liebenthal, E. (2006). Syn-tactic and semantic modulation of neural activity during auditory sentencecomprehension. Journal of Cognitive Neuroscience, 18, 665–679.

umphries, C., Love, T., Swinney, D., & Hickok, G. (2005). Response of anteriortemporal cortex to syntactic and prosodic manipulations during sentenceprocessing. Human Brain Mapping, 26, 128–138.

ndefrey, P., Hagoort, P., Herzog, H., Seitz, R. J., & Brown, C. M. (2001). Syn-tactic processing in left prefrontal cortex is independent of lexical meaning.NeuroImage, 14, 546–555.

ust, M. A., Carpenter, P. A., Keller, T. A., Eddy, W. F., & Thulborn, K. R.(1996). Brain activation modulated by sentence comprehension. Science,274, 114–116.

aan, E., & Swaab, T. Y. (2002). The brain circuitry of syntactic comprehension.Trends in Cognitive Sciences, 6, 350–356.

ang, A. M., Constable, R. T., Gore, J. C., & Avrutin, S. (1999). An event-relatedfMRI study of implicit phrase-level syntactic and semantic processing. Neu-roImage, 10, 555–561.

eller, T. A., Carpenter, P. A., & Just, M. A. (2001). The neural bases of sentencecomprehension: A fMRI examination of syntactic and lexical processing.Cerebral Cortex, 11, 223–237.

uperberg, G. R., Holcomb, P. J., Sitnikova, T., Greve, D., Dale, A. M., & Caplan,D. (2003). Distinct patterns of neural modulation during the processing ofconceptual and syntactic anomalies. Journal of Cognitive Neuroscience, 15,272–293.

uperberg, G. R., McGuire, P. K., Bullmore, E. T., Brammer, M. J., Rabe-Hesketh, S., Wright, I. C., et al. (2000). Common and distinct neuralsubstrates for pragmatic, semantic, and syntactic processing of spoken sen-tences: An fMRI study. Journal of Cognitive Neuroscience, 12, 321–341.

uke, K. K., Liu, H. L., Wai, Y. Y., Wan, Y. L., & Tan, L. H. (2002). Functionalanatomy of syntactic and semantic processing in language comprehension.Human Brain Mapping, 16, 133–145.

o, L., Liu, H. L., Jin, H., & Yang, Y. L. (2005). Brain activation during seman-tic judgment of Chinese sentences: A functional MRI study. Human BrainMapping, 24, 305–312.

odern Chinese Frequency Dictionary (in Chinese). (1986). Beijing: BeijingLanguage Institute Publisher.

oro, A., Tettamanti, M., Perani, D., Donati, C., Cappa, S. F., & Fazio, F.(2001). Syntax and the brain: Disentangling grammar by selective anomalies.NeuroImage, 13, 110–118.

ewman, S. D., Just, M. A., Keller, T. A., Roth, J., & Carpenter, P. A. (2003).Differential effects of syntactic and semantic processing on subregions ofBroca’s area. Cognitive Brain Research, 16, 297–307.

ewman, A. J., Pancheva, R., Ozawa, K., Neville, H. J., & Ullman, M. T. (2001).An event-related fMRI study of syntactic and semantic violations. Journalof Psycholinguistic Research, 30, 339–364.

i, W., Constable, R. T., Mencl, W. E., Pugh, K. R., Fulbright, R. K., Shaywitz,S. E., et al. (2000). An event-related neuroimaging study distinguishing formand content in sentence processing. Journal of Cognitive Neuroscience, 12,120–133.

oppeney, U., & Price, C. J. (2004). An fMRI study of syntactic adaptation.Journal of Cognitive Neuroscience, 16, 702–713.

etrides, M., Alivisatos, B., Meyer, E., & Evans, A. C. (1993). Functional activa-tion of the human frontal cortex during the performance of verbal workingmemory tasks. Proceedings of the National Academy of Sciences of theUnited States of America, 90, 878–882.

odd, J. M., Davis, M. H., & Johnsrude, I. S. (2005). The neural mechanismsof speech comprehension: fMRI studies of semantic ambiguity. Cerebral

Cortex, 15, 1261–1269.

oder, B., Stock, O., Neville, H., Bien, S., & Rosler, F. (2002). Brain activationmodulated by the comprehension of normal and pseudo-word sentences ofdifferent processing demands: A functional magnetic resonance imagingstudy. NeuroImage, 15, 1003–1014.

1 holog

S

S

S

S

S

S

S

T

T

T

T

V

X

Y

potentials. Brain Research, 1071, 186–196.

378 S. Wang et al. / Neuropsyc

akai, K. L., Homae, F., & Hashimoto, R. (2003). Sentence processing isuniquely human. Neuroscience Research, 46, 273–279.

akai, K. L., Noguchi, Y., Takeuchi, T., & Watanabe, E. (2002). Selective primingof syntactic processing by event-related transcranial magnetic stimulation ofBroca’s area. Neuron, 35, 1177–1182.

ervice, E., Helenius, P., Maury, S., & Salmelin, R. (2007). Localization of syn-tactic and semantic brain responses using magnetoencephalography. Journalof Cognitive Neuroscience, 19, 1193–1205.

hi, D. X. (2000). The flexibility of Chinese syntax and the theory of syn-tax. Contemporary Linguistics (Dangdai Yuyanxue in Chinese pinyin), 2,18–26.

iok, W. T., Perfetti, C. A., Jin, Z., & Tan, L. H. (2004). Biological abnormalityof impaired reading is constrained by culture. Nature, 431, 71–76.

towe, L. A., Haverkort, M., & Zwarts, F. (2005). Rethinking the neurologicalbasis of language. Lingua, 115, 997–1042.

tromswold, K., Caplan, D., Alpert, N., & Rauch, S. (1996). Localization ofsyntactic comprehension by positron emission tomography. Brain and Lan-guage, 52, 452–473.

alairach, J., & Tournoux, P. (1988). Co-planar stereotaxic atlas of the humanbrain. New York: Thieme Medical Publishers.

an, L. H., Liu, H. L., Perfetti, C. A., Spinks, J. A., Fox, P. T., & Gao, J. H. (2001).The neural system underlying Chinese logograph reading. NeuroImage, 13,836–846.

Z

Z

ia 46 (2008) 1371–1378

an, L. H., Spinks, J. A., Gao, J. H., Liu, H. L., Perfetti, C. A., Xiong,J., et al. (2000). Brain activation in the processing of Chinese charac-ters and words: A functional MRI study. Human Brain Mapping, 10,16–27.

ulving, E., Kapur, S., Craik, F. I. M., Moscovitch, M., & Houle, S. (1994).Hemispheric encoding/retrieval asymmetry in episodic memory: Positronemission tomography findings. Proceedings of the National Academy ofSciences of the United States of America, 91, 2016–2020.

igneau, M., Beaucousin, V., Herve, P. Y., Duffau, H., Crivello, F., Houde, O.,et al. (2006). Meta-analyzing left hemisphere language areas: Phonology,semantics, and sentence processing. NeuroImage, 30, 1414–1432.

iao, Z., Zhang, J. X., Wang, X., Wu, R., Hu, X., Weng, X., et al. (2005).Differential activity in left inferior frontal gyrus for pseudowords and realwords: An event-related fMRI study on auditory lexical decision. HumanBrain Mapping, 25, 212–221.

e, Z., Luo, Y. J., Friederici, A. D., & Zhou, X. (2006). Semantic and syntacticprocessing in Chinese sentence comprehension: Evidence from event-related

hang, L. (1997a). What is parataxis? Chinese Language Learning (Hanyu Xuexiin Chinese pinyin), 97, 58–61.

hang, L. (1997b). Issue on semantic categories. Chinese Language Learning(Hanyu Xuexi in Chinese pinyin), 100, 8–13.