E

Please cite this article in press as: Lin SE, et al., Left-lateralized N170 response to unpronounceable pseudo but not false Chinesecharacters—the key role of orthography, Neuroscience (2011), doi: 10.1016/j.neuroscience.2011.05.071

Neuroscience xx (2011) xxx

d

LEFT-LATERALIZED N170 RESPONSE TO UNPRONOUNCEABLEPSEUDO BUT NOT FALSE CHINESE CHARACTERS—THE KEY ROLE

OF ORTHOGRAPHY

wb(AlcurcetcrCtoe

baanbbtrfacircst

S. E. LIN,a,b H. C. CHEN,c* J. ZHAO,a S. LI,a

S. HEd AND X. C. WENGa,e*aKey Laboratory of Behavioral Science, Institute of Psychology, Chi-nese Academy of Sciences, Beijing 100101, ChinabGraduate University of Chinese Academy of Sciences, Beijing100049, ChinacDepartment of Psychology, The Chinese University of Hong Kong,Shatin, NT, Hong Kong, ChinadDepartment of Psychology, University of Minnesota, MN 55455, USAeCenter for Human Brain Research, Hangzhou Normal University,Hangzhou 310000, China

Abstract—A negative event-related potential (ERP) compo-nent, known as N170, can be readily recorded over the pos-terior left brain region when skilled readers are presentedwith visual words. This left-lateralized word-related N170 hasbeen attributed either to linguistic processes, particularlyphonological processing, or to the role of orthographic reg-ularity, emphasizing a perceptual origin. This debate, how-ever, is difficult to resolve in the context of alphabetic scriptsbecause of the tight relations between orthography and pho-nology. In contrast, Chinese characters have arbitrary map-pings between orthographic and sound forms, making it pos-sible to tease apart these two properties of visual words. Wetherefore addressed this issue by examining ERP responsesto Chinese characters and three types of structurallymatched but unpronounceable stimuli: pseudo-characters,false-characters, and stroke combinations. A content-irrele-vant color matching task was adopted to minimize potentiallydifferent top-down modulations across stimulus types. Re-sults show that, relative to false-characters and stroke com-binations, real- and pseudo-characters evoked greater N170in the left posterior brain region. Critically, despite beingunpronounceable, pseudo-characters produced the sameamplitude and left-lateralized N170, just as real-characters.These results provide strong evidence that orthographyrather than phonology serves as the main driver for the en-hanced and left-lateralized N170 to visual words. © 2011IBRO. Published by Elsevier Ltd. All rights reserved.

Key words: N170, Chinese character, orthography, phonol-ogy.

*Correspondence to: X. C. Weng, Center for Cognition and BrainDisorders, Hangzhou Normal University, Hangzhou 310000, China, orH. C. Chen, Department of Psychology, The Chinese University ofHong Kong, Shatin, NT, Hong Kong, China. Tel: �86-134-6631-5051or �852-2609-6485; fax: �86-10-6483-2070 or �852-2603-5019.

-mail address: wengxc@psych.ac.cn (X. C. Weng) or hcchen@psy.cuhk.edu.hk (H. C. Chen).

eAbbreviations: ERP, event-related potential; VWFA, visual word formarea.

0306-4522/11 $ - see front matter © 2011 IBRO. Published by Elsevier Ltd. All righoi:10.1016/j.neuroscience.2011.05.071

1

Visual objects can evoke a prominent negative event-relatedpotential (ERP) component peaking between 150 and 200ms after stimulus onset, which is often referred to as N170(Bentin et al., 1999; Maurer et al., 2008a; Rossion et al.,2003, 2004; Schendan et al., 1998). In skilled readers, theN170 evoked by visual words is significantly larger than thatevoked by control stimuli such as strings of symbols andforms (e.g., Bentin et al., 1999; Maurer et al., 2005a,b). The

ord-related N170 is typically left-lateralized, contrary to theilateral or right-lateralized N170 evoked by faces or objectsBrem et al., 2005; Dehaene, 1995; Rossion et al., 2003).lthough there is a consensus that the enhanced and left-

ateralized N170 effects for visual words reflect the neuralhange resulting from extensive experience with this partic-lar type of visual category (McCandliss and Noble, 2003), itemains challenging to identify the specific level at which thishange occurs. While many researchers focus on the influ-nce of linguistic processing, suggesting, for example, that

he left-lateralized N170 to visual words reflects automaticonnections between orthographic forms and phonologicalepresentations (Proverbio and Zani, 2003; Maurer and Mc-andliss, 2007), others emphasize a perceptual origin within

he visual system, without additional involvement of phonol-gy and other linguistic processes (Bentin et al., 1999; Wongt al., 2005).

This issue is difficult to resolve in the context of sound-ased scripts, because orthographic stimuli (i.e., wordsnd their variants, such as pseudo- and non-words) mayutomatically activate phonological processing even inon-phonological tasks due to the systematic mappingetween orthographic forms (e.g., letters and letter com-inations) and sounds. In contrast, in logographic Chinese,he mapping between visual and phonological forms iselatively arbitrary, thus allowing for investigation of visualorm effects with minimal involvement of phonology (Chennd Juola, 1982). Interestingly, about 70% of Chineseharacters have a left–right compound structure, consist-

ng of two radicals with one on the left and another on theight. If the two radicals never occur together in a realharacter but each appears at a frequently appeared po-ition in the form of a compound character, this combina-ion is regarded as a pseudo-character (e.g., ), al-

though it has no meaning or phonology of its own. Ifthese two radicals or one of them are placed at unusualpositions, this combination becomes a false-character(e.g., ) (Ho et al., 2003; Yin and Butterworth, 1998).

In the present study, we aimed to examine whether the

nhanced and left-lateralized N170 to visual words is due to

ts reserved.

fitAe

P

F

S

ff c propert

S. Lin et al. / Neuroscience xx (2011) xxx2

either lateralized processing at the perceptual stage or theinvolvement of phonological activation, by taking advantageof the above-mentioned characteristics of the Chinese script.To this end, we constructed four types of structurally matchedstimuli (i.e., all had a horizontally arranged left–right radicalstructure and were formed by 8–12 strokes, but differed intheir pronounceability and orthographic properties; see Table1 for examples), including real-characters, pseudo-charac-ters, false-characters, and stroke combinations.

Real characters were existing compound characters inChinese. Thus, they are pronounceable and meet all or-thographic regularities of the Chinese script. The otherstimuli were constructed on the basis of two orthographicvariables in Chinese (i.e., radical identity and radical posi-tion), both of which have been shown to play a role inprocessing Chinese characters (e.g., Ding et al., 2004;Tsang and Chen, 2009). Specifically, each of the unpro-nounceable stroke combinations (e.g., ) was formed by

two non-existing radicals with random strokes, and thistype of stimuli was constructed to control for low-levelvisual features. Pseudo- and false-characters were formedby existing radicals from real characters, but they differedin radical position: both radicals in a pseudo-characterwere at commonly appeared positions, whereas those in afalse-character were at unusual positions. Importantly,both pseudo- and false-characters were also unpro-nounceable, because all radicals adopted to build thesestimuli were in bound form, in that they alone cannot beused as meaningful characters and must be combined withother radicals to form a character. To further reduce thepossibility of phonological activation and to minimize thedifferential top-down modulations across stimulus types,we designed a delayed color-matching task, in which par-ticipants were asked to make judgments about the color ofthe stimuli only and to ignore the other properties of thestimuli.

In sum, our study was designed to identify the maindeterminant of the left-lateralized word-related N170. Ifphonological processing provides significant contributionsto this ERP component, then we expect that only real

Table 1. Description and examples of different types of stimuli

Type of stimuli Example Pr

Real-character �

seudo-character �

alse-character �

troke combination �

All four stimulus types were matched in structural complexity (i.e. aor stroke combinations which were formed by non-existing radicals, eour types of stimuli differed in their pronounceability and orthographi

characters, but not the unpronounceable stimuli, will pro-

Please cite this article in press as: Lin SE, et al., Left-lateralized N170characters—the key role of orthography, Neuroscience (2011), doi: 10.1

duce a larger and left-lateralized N170, relative to strokecombinations. Alternatively, if this N170 component ismainly induced by orthographic processing with minimalinvolvement of phonology, we expect that not only realcharacters, but also the unpronounceable pseudo- and/orfalse-characters will elicit a larger and left-lateralizedN170. If this is indeed the case, contrasting the results ofthe three types of unpronounceable stimuli will furtherinform us the specific kind of orthographic property (i.e.,radical position or radical identity) in the Chinese script thatis responsible for the N170 response.

EXPERIMENTAL PROCEDURES

Participants

Fourteen native-Chinese speakers (five males, mean age: 23.4years) participated in the experiment. They were neurologicallyhealthy with normal or corrected-to-normal vision. All were right-handed according to the Edinburgh Handedness Inventory (Old-eld, 1971). The research protocol was approved by the institu-ional review board of the Institute of Psychology of Chinesecademy of Sciences, and informed consent was obtained fromach participant.

Stimuli

There were four types of experimental stimuli; each type of stimuliconsisted of 30 items (see Appendix A). Different types of stimuliwere matched in the average number of strokes (ranging from 8 to12) and structural complexity (all stimuli were formed by twohorizontally arranged radicals). Real characters were at occur-rence frequency of 300–600 per million (the modern Chinesefrequency dictionary, 1985). Pseudo- and false-characters werebuilt with the same set of bound radicals, except that both radicalswere at the commonly appeared positions for pseudo-characters,but at unusual positions for false-characters. Between the twotypes of stimuli, the radicals were matched at the individual stim-ulus level in that exchanging the relative positions of the tworadicals in one formed the other. Each stroke combination wascomposed of two horizontally arranged non-existing radicals(each radical was formed by randomized strokes).

Procedures

Each trial started with a grid presented at the center of computer

ability Radical position Radical

� �

� �

� �

� �

eft–right radical structure and were formed by 8–12 strokes). Exceptlus of the other three types was formed by two existing radicals. The

ies.

onounce

ll had a lach stimu

screen for 200 ms, in one of three following colors: blue, green or

response to unpronounceable pseudo but not false Chinese016/j.neuroscience.2011.05.071

mh

a[Ph

S. Lin et al. / Neuroscience xx (2011) xxx 3

red. A texture noise mask then appeared at the same position for50 ms. After an interval of 700 ms, a target stimulus (one of thefour types), subtended 1.2 visual degrees, was presented in oneof the above colors for 200 ms. After the display of the target andanother blank interval of 600 ms, a five-pointed star was pre-sented at the same position for 200 ms. Participants were in-structed to compare the color of the target stimulus with that of thepreceding grid and to respond by pressing the right or left mousebutton as quickly and accurately as possible immediately after thedisplay of the star signal (Fig. 1). The target stimuli were pre-sented in a pseudo-randomized manner to ensure that any twosuccessively presented targets would not share a common radi-cal.

ERP recording and analysis

Electroencephalographic activity was recorded from a 64-channelNeuroscan version 4.3 system (NeuroScan Inc., EI Paso, TX,USA) with a common vertex reference, and transformed into

Fig. 1. Experiment procedure (see Experimental procedures for de-tails). For interpretation of the references to color in this figure legend,the reader is referred to the Web version of this article.

Fig. 2. Grand averaged ERP waves of four types of stimuli at bilateraFalse: false-characters, Stroke: stroke combinations. Gray boxes in P

N170.

Please cite this article in press as: Lin SE, et al., Left-lateralized N170characters—the key role of orthography, Neuroscience (2011), doi: 10.1

averaged ears references (M1�M2). Signals were amplified witha DC-100Hz band-pass filter and digitized at a 1000 Hz samplingrate. Resistances across all the electrodes were kept below 5 k�.

The acquired data were corrected for eye-movements andblinks, and segmented to epochs of 800 ms post stimulus onsetplus a 100 ms pre-stimulus baseline. Epochs containing incorrectbehavioral responses (less than 1%) or with peak-to-peak differ-ences larger than 50 �V were rejected. The remaining epochswere baseline corrected over a 100 ms pre-stimulus interval andaveraged for each condition and participant, and digitally filteredwith a 0.5–40 Hz band-pass filter. The N170 amplitude for eachstimulus type was computed at channels of PO7-PO8, PO5-PO6,PO3-PO4, CB1-CB2 and O1-O2, selected according to the topo-graphic maps. Time window for the mean amplitude was 140–190ms (Fig. 2), based on visual inspection of our data and theliterature (e.g., Bentin et al., 1999; Rossion et al., 2003; Schendanet al., 1998). Statistic analysis of mean amplitude of N170 wasconducted using repeated measures ANOVA, with two within-subjects variables: stimulus type and hemisphere (Greenhouse-Geisser corrected). Bonferroni correction was used to avoid multi-comparison errors.

RESULTS

Fig. 2 displays grand-averaged ERPs of the selected chan-nels to the four stimulus types. A negative deflection peak-ing around 170 ms was clearly visible for each stimulustype over bilateral occipito-temporal sites. Fig. 3 shows

ean amplitudes of N170 for each stimulus type at eachemisphere.

A repeated measure ANOVA analysis of mean N170mplitudes revealed main effects of both stimulus type

F(3,39)�14.05, P�0.001] and hemisphere [F(3,39)�5.65,�0.05], and a significant interaction of stimulus type�emisphere [F(3,39)�4.01, P�0.05].

Post hoc comparisons of the N170 between the leftand right hemispheres indicated a significant left lateral-ization for both real and pseudo-characters (Ps�0.05 for

r electrode sites. Real: Real-characters; Pseudo: Pseudo-characters,PO8 indicate the time window for computing the mean amplitude of

l posterioO7 and

response to unpronounceable pseudo but not false Chinese016/j.neuroscience.2011.05.071

aatcf0a

S. Lin et al. / Neuroscience xx (2011) xxx4

both types), but not for false-characters and stroke com-binations (both Ps�0.05).

We next analyzed stimulus type differences separatelyt each hemisphere. At the left hemisphere, the meanmplitudes of N170 evoked by real- and pseudo-charac-

ers were significantly higher than those evoked by false-haracters and stroke combinations: for real characters vs.alse-characters and stroke combinations, P�0.014 and.001, respectively; for pseudo-characters vs. false-char-cters and stroke combinations, P�0.013 and 0.004, re-

spectively. There was no statistic difference between real-vs. pseudo-characters and between false-characters vs.stroke combinations. At the right hemisphere, there was noN170 difference across real-, pseudo-, and false-charac-ters. However, the mean amplitudes of N170 evoked bythese three types were significantly larger than that evokedby stroke combinations: Ps for real-, pseudo-, and false-characters vs. stroke combinations were 0.01, 0.03, and0.003, respectively.

DISCUSSION

Various hypotheses have been proposed to account forthe enhanced and left-lateralized N170 response to visualwords in skilled readers. The key point of contention iswhether this neurophysiological effect is mainly caused byphonological activation or by visual perceptual expertise.We addressed this issue by comparing N170 responses tofour types of stimuli with a similar structural complexity butdifferent degrees of conformity to the orthography of Chi-nese. In the context of logographic Chinese, we were ableto examine the role of orthographic regularity with minimalinvolvement of phonology. Potential phonological influenceat the lexical level was further reduced by using a content-irrelevant color matching task. Indeed, the processing ofstimulus properties other than color in this paradigm ispresumably even more implicit than that in the commonly

Fig. 3. The mean amplitudes of N170 for the four stimulus types ateach hemisphere. * P�0.05, ** P�0.01. Error bars represent standarderrors.

used one-back repetition detection paradigm. Further-

Please cite this article in press as: Lin SE, et al., Left-lateralized N170characters—the key role of orthography, Neuroscience (2011), doi: 10.1

more, the use of unpronounceable (bound) radicals toconstruct the stimuli to prevent possible sub-lexical pho-nological activation is an additional feature of the presentstudy.

Consistent with previous findings (Compton et al.,1991; Dehaene, 1995; Grossi and Coch, 2005; Mariol etal., 2008; McCandliss and Noble, 2003), we observed asignificantly larger N170 for visual words (i.e., real charac-ters) than for structurally matched control stimuli (i.e.,stroke combinations). Moreover, the N170 to real charac-ters was strongly left-lateralized, which confirms a numberof ERP studies showing that logographic characters alsoproduce a left-lateralized N170 (Cao et al., 2011; Liu andPerfetti, 2003; Liu et al., 2003; Maurer et al., 2008b; Wonget al., 2005). One exception is the study by Kim et al.(2004), which demonstrated bilateral N170 responses toChinese characters. However, participants’ unfamiliaritywith Chinese characters and the use of a semantic task inthat study may induce an additional recruitment of the righthemisphere (Maurer et al., 2010, 2005b). Therefore, ourcurrent findings together with those in the literature sug-gest that the degree of left-lateralization of N170 is depen-dent on the level of expertise with a script rather than thetype of scripts (see Maurer et al., 2008b for more detaileddiscussion).

The key finding of the current study is that the left-later-alized N170 was sensitive to an orthographic property, butnot to the phonological manipulation. Specifically, we foundthat unpronounceable pseudo-characters, but not false-char-acters, also produced a left-lateralized N170, and its ampli-tude was as strong as that induced by pronounceable realcharacters. Indeed, detailed analysis of the amplitude ofN170 at channels over the left posterior region showed thatthe amplitude of N170 follows the order: real-characters�pseudo-characters�false-characters�stroke combinations.In other words, within the first 200 ms of stimulus presenta-tion, the left posterior brain seemed to treat pseudo- andreal-characters as one category, and false characters andstroke combinations as another category. Because the criti-cal difference between pseudo- and real-characters lies intheir pronounceability, whereas that between pseudo- andfalse-characters lies only in a specific orthographic property(i.e., radical position), our results strongly suggest that theleft-lateralization of word-related N170 is induced by ortho-graphic processing rather than by phonological activation.

The left lateralized N170 to both real- and pseudo-characters is in agreement with many ERP findings ofperceptual specialization for alphabetic scripts usinggrapheme-phoneme conversion rules. For example, Ben-tin et al. (1999) and Mariol et al. (2008) reported a similarN170 left lateralization to French words and pseudo-words, which was also demonstrated in a study with Italianparticipants (Proverbio et al., 2004). While both Frenchand Italian possess a regular orthography with very con-sistent spelling to sound mappings, a similar finding of thelack of N170 word vs. pseudo-word difference has alsobeen observed in English (Hauk et al., 2006), which pos-sesses an irregular orthography with inconsistent spelling

to sound mappings (but see Maurer et al., 2005a). How-

response to unpronounceable pseudo but not false Chinese016/j.neuroscience.2011.05.071

fitdgSaa

fotopboaaMatwhtcsaamigt

ddWrrefeosehchssbslzliii

S. Lin et al. / Neuroscience xx (2011) xxx 5

ever, in these sound-based alphabetic scripts, it is possiblethat this lateralization effect is a mixed result of orthogra-phy and phonology, as both real- and pseudo-words couldreadily produce phonological activation (Bentin et al.,1999; Simon et al., 2006). On the contrary, in logographicChinese, real characters are thought to induce weaker(Chen and Shu, 2001; Yeh and Li, 2002) and possiblydelayed (beyond the N170 window, Liu and Perfetti, 2003;Perfetti et al., 2005) activation of phonology, because themapping between the graphic and sound forms is arbitrary(Chen and Juola, 1982; Zhou et al., 2009). More impor-tantly, within this orthographic system, we could build aspecific type of stimuli (i.e., pseudo-characters), which areunpronounceable but possess a relatively high degree oforthographic regularity. Therefore, our finding that real-and pseudo-characters produced the same degree of theleft-lateralized N170 effect provides strong evidence thatphonological activation has minimal contribution to theleft-lateralized N170 for visual words.

The left-lateralization effect of pseudo- and real-char-acters is unlikely driven by low-level visual features, be-cause there was no left-lateralization effect in the conditionof false-characters, which were visually very similar to realand pseudo-characters (i.e., all with existing radicals andmatched structural complexity). In addition, the lack ofleft-lateralization of N170 for false-characters also rulesout the possibility of sub-lexical (radical) phonological in-fluence, as false- and pseudo-characters were made of thesame set of unpronounceable radicals.

Our interpretation of the left-lateralization of N170 tovisual words does not necessarily contradict with the well-known phonological mapping hypothesis (Maurer andMcCandliss, 2007). According to this hypothesis, the left-lateralized N170 arises due to increased sensitivity to vi-sual words in the left extra-striate visual region, namely thevisual word form area (VWFA, Cohen et al., 2000), whichis caused by repeated co-activations of the visual systemand the left-lateralized speech network during reading ac-quisition. Thus, phonology does contribute to shaping aword-specialized region—the VWFA—during the processof learning to read visual words. However, in skilled read-ers, the lateralization in processing written words is alreadyestablished in visual areas and phonology does not nec-essarily contribute to the early neurophysiological re-sponse—the N170—during the process of viewing thisclass of stimuli, although activation of anterior inferior fron-tal gyrus and VWFA at the same time window in adults canbe found when more explicit reading tasks were adopted inthe processing of alphabetic scripts (e.g., Cornelissen etal., 2003; Pammer et al., 2004; Wheat et al., 2010).

It is worth noting that earlier studies have shown dif-erent patterns of the left-lateralization of N170 using var-ous types of orthographic stimuli, ranging from single let-ers, consonant strings, pseudo-words, to real words inifferent tasks and scripts with different degrees of ortho-raphic depth (Bentin et al., 1999; Hauk et al., 2006;pironelli and Angrilli, 2007; Maurer et al., 2005a; Mitrand Coch, 2009; Proverbio and Adorni, 2009; Sereno et

l., 1998). One plausible explanation is that both stimulus

Please cite this article in press as: Lin SE, et al., Left-lateralized N170characters—the key role of orthography, Neuroscience (2011), doi: 10.1

actors (e.g., the baseline used and the orthographic depthf a particular script) and task demand could contribute tohe different findings. For example, Maurer et al. (2005a)bserved an N170 laterality difference between words andseudowords in relatively deep orthographies like English,ut not in shallow orthographies like German, while manythers have failed to detect such a difference (Sereno etl., 1998; Hauk et al., 2006). Because Sereno et al. (1998)nd Hauk et al. (2006) used the lexical decision task, whileaurer et al. (2005a) adopted the one-back detection par-digm, it is possible that task modulation is responsible forhe inconsistent findings about N170 laterality for pseudo-ords in relatively deep orthographies. Our current study,owever, does not support this account, because we foundhat even in a color matchingtask, which is more implicitompared with the one-back repetition task, there was noignificant difference in N170 lateralization between realnd pseudo characters. One tentative explanation is that,s we have discussed, the N170 recorded in our studyainly reflects orthographic processing, whereas the N170

n other studies with alphabetic stimuli additionally en-ages phonological processing, which may be more proneo task modulation. This possibility deserves further study.

In addition to examining the lateralization of N170 toifferent types of stimuli, it is also interesting to see theistinct patterns of N170 response in each hemisphere.hile N170 in the left hemisphere was sensitive to orthog-

aphy, our results also showed that N170 in the posterioright hemisphere clearly differentiated between stimuli withxisting radicals (i.e., real-, pseudo-, and false-characters)rom those without (i.e., stroke combinations): the formerlicited a larger N170. This could be due to the rightccipitotemporal region’s processing of familiar visualhapes (i.e., existing radicals are familiar shapes but non-xisting ones are not). This result suggests that the rightemisphere is also involved in the processing of Chineseharacters, but its role is different from that of the leftemisphere (see, e.g., Liu et al., 2003, for related discus-ion). Moreover, our result is also consistent with a long-tanding postulation that the right extra-striate region mighte involved in a process of feature detection at an earlytage of visual word recognition, such as the processing of

etter identity (Dehaene et al., 2005; Coltheart, 1983; Su-uki et al., 1998), and corroborates a recent neuropsycho-

ogical study in a pure alexia patient with combined lesionsn the left occipitotemporal region and the splenium, show-ng that the right hemisphere was indeed involved in rad-cal processing (Shan et al., 2010).

CONCLUSION

Using real Chinese characters and three types of speciallydesigned, unpronounceable stimuli with different ortho-graphic properties, we observed a robust left-lateralizedN170 effect for both real- and pseudo-characters, but notfor false characters. Importantly, the lateralized N170 ef-fect was modulated by an orthographic manipulation (i.e.,radical position), but not by phonology. These results thus

provide strong evidence that orthographic rather than pho-

response to unpronounceable pseudo but not false Chinese016/j.neuroscience.2011.05.071

S. Lin et al. / Neuroscience xx (2011) xxx6

nological properties are behind the enhanced and left-lateralized N170 to visual words. Future ERP studies com-bined with functional magnetic imaging will more preciselydetermine the spatio-temporal sequence for the computa-tion and representation of the orthographic and phonolog-ical information of visual words.

Acknowledgments—This work was supported by the ChineseMinistry of Science and Technology (grant no. 2007CB512300),the National Science Foundation of China (grant no. 31070905),the Chinese Academy of Sciences, the National Science Founda-tion (NSF BCS-0818588), and a departmental grant from theChinese University of Hong Kong.

REFERENCES

Bentin S, Mouchetant-Rostaing Y, Giard MH, Echallier JF, Pernier J(1999) ERP manifestations of processing printed words at differentpsycholinguistic levels: time course and scalp distribution. J CognNeurosci 11:235–260.

Brem S, Lang-Dullenkopf A, Maurer U, Halder P, Bucher K, BrandeisD (2005) Neurophysiological signs of rapidly emerging visual ex-pertise for symbol strings. Neuroreport 16:45–48.

Cao XH, Li S, Zhao J, Lin SE, Weng XC (2011) Left-lateralized earlyneurophysiological response for Chinese characters in young pri-mary school children. Neurosci Lett 492:165–169.

Chen HC, Juola JF (1982) Dimensions of lexical coding in Chinese andEnglish. Mem Cogn 10:216–224.

Chen HC, Shu H (2001) Lexical activation during the recognition ofChinese characters: evidence against early phonological activa-tion. Psychon Bull Rev 8:511–518.

Cohen L, Dehaene S, Naccache L, Lehéricy S, Dehaene-Lambertz G,Hénaff MA, Michel F (2000) The visual word form area: spatial andtemporal characterization of an initial stage of reading in normalsubjects and posterior split-brain patients. Brain 123:291–307.

Coltheart M (1983) The right hemisphere and disorders of reading. In:Functions of the right hemisphere (Young A, ed), pp 173–201.London: Academic Press.

Compton PE, Grossenbacher P, Posner MI, Tucker DM (1991) Acognitive anatomical approach to attention in lexical access. JCogn Neurosci 3:304–312.

Cornelissen P, Tarkiainen A, Helenius P, Salmelin R (2003) Corticaleffects of shifting letter position in letter strings of varying length. JCogn Neurosci 15:731–746.

Dehaene S (1995) Electrophysiological evidence for category-specificword-processing in the normal human brain. Neuroreport 6:2153–2157.

Dehaene S, Cohen L, Sigman M, Vinckier F (2005) The neural codefor written words: a proposal. Trends Cogn Sci 9:335–341.

Ding GS, Peng DL, Taft M (2004) The nature of the mental represen-tation of radicals in Chinese: a priming study. J Exp Psychol LearnMem Cogn 30:530–539.

Grossi G, Coch D (2005) Automatic word form processing in maskedpriming: an ERP study. Psychophysiology 42:343–355.

Hauk O, Patterson K, Woollams A, Watling L, Pulvermuller F, RogersTT (2006) [Q:] When would you prefer a SOSSAGE to a SAU-SAGE? [A:] At about 100 msec. ERP correlates of orthographictypicality and lexicality in written word recognition. J Cogn Neurosci18:818–832.

Ho CSH, Ng TT, Ng WK (2003) A “radical” approach to readingdevelopment in Chinese: the role of semantic radicals and phoneticradicals. J Lit Res 35:849–878.

Kim KH, Yoon HW, Park HW (2004) Spatiotemporal brain activationpattern during word/picture perception by native Koreans. Neu-

roreport 15:1099–1103.

Please cite this article in press as: Lin SE, et al., Left-lateralized N170characters—the key role of orthography, Neuroscience (2011), doi: 10.1

Liu Y, Perfetti CA (2003) The time course of brain activity in readingEnglish and Chinese: an ERP study of Chinese bilinguals. HumBrain Mapp 18:167–175.

Liu Y, Perfetti CA, Hart L (2003) ERP evidence for the time course ofgraphic, phonological, and semantic information in Chinese mean-ing and pronunciation decisions. J Exp Psychol Learn Mem Cogn29:1231–1247.

Mariol M, Jacques C, Schelstraete MA, Rossion B (2008) The speedof orthographic processing during lexical decision: electrophysio-logical evidence for independent coding of letter identity and letterposition in visual word recognition. J Cogn Neurosci 20:1283–1299.

Maurer U, Blau VC, Yoncheva YN, McCandliss BD (2010) Devel-opment of visual expertise for reading: rapid emergence ofvisual familiarity for an artificial script. Dev Neuropsychol35:404 – 422.

Maurer U, Brandeis D, McCandliss B (2005a) Fast visual specializa-tion for reading in English revealed by the topography of the N170ERP response. Behav Brain Funct 1:13.

Maurer U, Brem S, Bucher K, Brandeis D (2005b) Emerging neuro-physiological specialization for letter strings. J Cogn Neurosci17:1532–1552.

Maurer U, McCandliss BD (2007) The development of visual expertise forwords: the contribution of electrophysiology. In: Single-word reading:biological and behavioural perspectives (Grigorenko EL, Naples AJ,eds), pp 43–64. Mahwah, NJ: Lawrence Erlbaum Associates.

Maurer U, Rossion B, McCandliss BD (2008a) Category specificityin early perception: face and word N170 responses differ in bothlateralization and habituation properties. Front Hum Neurosci2:7.

Maurer U, Zevin JD, McCandliss BD (2008b) Left-lateralized N170effects of visual expertise in reading: evidence from Japanesesyllabic and logographic scripts. J Cogn Neurosci20:1878 –1891.

McCandliss BD, Noble KG (2003) The development of reading impair-ment: a cognitive neuroscience model. Ment Retard Dev DisabilRes Rev 9:196–205.

Mitra P, Coch D (2009) A masked priming ERP study of letter pro-cessing using single letters and false fonts. Cogn Affect BehavNeurosci 9:216–228.

Modern chinese frequency dictionary (1985). Beijing: Beijing Lan-guage Institute Press. In Chinese.

Oldfield RC (1971) Assessment and analysis of handedness: Edin-burgh inventory. Neuropsychologia 9:97–113.

Pammer K, Hansen PC, Kringelbach ML, Holliday I, Barnes G, Hill-ebrand A, Singh KD, Cornelissen PL (2004) Visual word recogni-tion: the first half second. Neuroimage 22:1819–1825.

Perfetti CA, Liu Y, Tan LH (2005) The lexical constituency model:some implications of research on Chinese for general theories ofreading. Psychol Rev 112:43–59.

Proverbio AM, Adorni R (2009) C1 and P1 visual responses to wordsare enhanced by attention to orthographic vs. lexical properties.Neurosci Lett 463:228–233.

Proverbio AM, Vecchi L, Zani A (2004) From orthography to phonetics:ERP measures of grapheme-to-phoneme conversion mechanismsin reading. J Cogn Neurosci 16:301–317.

Proverbio AM, Zani A (2003) Time course of brain activation duringgraphemic/phonologic processing in reading: an ERP study. BrainLang 87:412–420.

Rossion B, Joyce CA, Cottrell GW, Tarr MJ (2003) Early lateralizationand orientation tuning for face, word, and object processing in thevisual cortex. Neuroimage 20:1609–1624.

Rossion B, Kung CC, Tarr MJ (2004) Visual expertise with nonfaceobjects leads to competition with the early perceptual processing offaces in the human occipitotemporal cortex. Proc Natl Acad Sci U

S A 101:14521–14526.

response to unpronounceable pseudo but not false Chinese016/j.neuroscience.2011.05.071

S. Lin et al. / Neuroscience xx (2011) xxx 7

Schendan HE, Ganis G, Kutas M (1998) Neurophysiological evidencefor visual perceptual categorization of words and faces within 150ms. Psychophysiology 35:240–251.

Sereno SC, Rayner K, Posner MI (1998) Establishing a time-line ofword recognition: evidence from eye movements and event-relatedpotentials. Neuroreport 9:2195–2200.

Shan CL, Zhu RJ, Xu MW, Luo BY, Weng XC (2010) Implicit readingin Chinese pure alexia. Brain Lang 114:147–156.

Simon G, Bernard C, Lalonde R, Rebaï M (2006) Orthographic trans-parency and grapheme-phoneme conversion: an ERP study inArabic and French readers. Brain Res 1104:141–152.

Spironelli C, Angrilli A (2007) Influence of phonological, semantic andorthographic tasks on the early linguistic components N150 andN350. Int J Psychophysiol 64:190–198.

Suzuki K, Yamadori A, Endo K, Fujii T, Ezura M, Takahashi A (1998)

Dissociation of letter and picture naming resulting from callosaldisconnection. Neurology 51:1390–1394.

Please cite this article in press as: Lin SE, et al., Left-lateralized N170characters—the key role of orthography, Neuroscience (2011), doi: 10.1

Tsang YK, Chen HC (2009) Do position-general radicals have a role toplay in processing Chinese characters? Lang Cogn Process24:947–966.

Wheat KL, Cornelissen PL, Frost SJ, Hansen PC (2010) During visualword recognition, phonology is accessed within 100 ms and maybe mediated by a speech production code: evidence from magne-toencephalography. J Neurosci 30:5229–5233.

Wong AC, Gauthier I, Woroch B, DeBuse C, Curran T (2005) An earlyelectrophysiological response associated with expertise in letterperception. Cogn Affect Behav Neurosci 5:306–318.

Yeh SL, Li JL (2002) Role of structure and component in judgments ofvisual similarity of Chinese characters. J Exp Psychol Hum PerceptPerform 28:933–947.

Yin WG, Butterworth B (1998) Chinese pure alexia. Aphasiology12:65–76.

Zhou XL, Ye Z, Cheung H, Chen HC (2009) Processing the Chineselanguage: an introduction. Lang Cogn Process 24:929–946.

APPENDIX: A. THE COMPLETE LIST OF STIMULI

(Accepted 27 May 2011)

response to unpronounceable pseudo but not false Chinese016/j.neuroscience.2011.05.071