Developmental changes in the neural influence of sublexical information on semantic processing

Author’s Accepted Manuscript Developmental changes in the neural influence of sublexical information on semantic processing Shu-Hui Lee, James R. Booth, Tai-Li Chou

www.elsevier.com/locate/neuropsychologia

PII: DOI: Reference:

S0028-3932(15)30017-8 http://dx.doi.org/10.1016/j.neuropsychologia.2015.05.001 NSY5579

To appear in: Neuropsychologia Received date: 15 November 2014 Revised date: 24 March 2015 Accepted date: 1 May 2015 Cite this article as: Shu-Hui Lee, James R. Booth and Tai-Li Chou, Developmental changes in the neural influence of sublexical information on semantic processing, Neuropsychologia, http://dx.doi.org/10.1016/j.neuropsychologia.2015.05.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Developmental changes in the neural influence of sublexical information on semantic processing

Shu-Hui Lee1, James R. Booth2, Tai-Li Chou134

1 2

Department of Psychology, National Taiwan University, Taipei, Taiwan

Department of Communication Sciences and Disorders, The University of Texas at Austin, Austin, TX, USA 3 4

Neurobiology and Cognitive Science Center, National Taiwan University, Taipei, Taiwan

Graduate Institute of Brain and Mind Sciences, National Taiwan University, Taipei, Taiwan

Please address all correspondence to:Tai-Li Chou Department of Psychology National Taiwan University No.1, Sec. 4, Roosevelt Road, Taipei, 10617, Taiwan Phone: 886-2-33663082 Fax: 886-2-23631463 E-mail: [email protected]

Abstract Functional magnetic resonance imaging (fMRI) was used to examine the developmental changes in a group of normally developing children (aged 8-12) and adolescents (aged 13-16) during semantic processing. We manipulated association strength (i.e. a global reading unit) and semantic radical (i.e. a local reading unit) to explore the interaction of lexical and sublexical semantic information in making semantic judgments. In the semantic judgment task, two types of stimuli were used: visually–similar (i.e. shared a semantic radical) versus visually–dissimilar (i.e. did not share a semantic radical) character pairs. Participants were asked to indicate if two Chinese characters, arranged according to association strength, were related in meaning. The results showed greater developmental increases in activation in left angular gyrus (BA 39) in the visually-similar compared to the visually-dissimilar pairs for the strong association. There were also greater age-related increases in angular gyrus for the strong compared to weak association in the visually-similar pairs. 1

Both of these results suggest that shared semantics at the sublexical level facilitates the integration of overlapping features at the lexical level in older children. In addition, there was a larger developmental increase in left posterior middle temporal gyrus (BA 21) for the weak compared to strong association in the visually-dissimilar pairs, suggesting conflicting sublexical information placed greater demands on access to lexical representations in the older children. All together, these results suggest that older children are more sensitive to sublexical information when processing lexical representations. Keywords: fMRI; semantic; orthographic similarity; association strength; age

Introduction The most prominent characteristic in Chinese is probably its use of a logographic script. Therefore, the recognition of distinctive orthographic units (i.e. characters) is widely studied (Tsang & Chen, 2009). Emerging evidence of developmental changes in semantic processing mainly focuses on the lexical level (i.e. the whole character), such as semantic association (Chou, Chen, Fan, Chen, & Booth, 2009; Lee, Booth, Chen, & Chou, 2011) and semantic priming (Wang, Dong, Ren, & Yang, 2009; Zhou, Wu, & Shu, 1998). However, the interactive process between lexical and sublexical (i.e. radicals, which are components of a character) semantic information in Chinese is not understood at a neural level. We aimed to explore developmental changes in how the semantic radical interacts with semantic association. Previous morphological studies in English have investigated the mapping from orthography to semantics in word recognition (Morris, Frank, Grainger, & Holcomb, 2007; Pastizzo & Feldman, 2009; Rastle, Davis, Marslen-Wilson, & Tyler, 2000). The abovementioned studies suggest a convergent contribution of both orthographical similarity and semantic relatedness in visual word recognition. However, the relation between form and meaning in English is only partly consistent and admits many exceptions (e.g. –or in traitor and anchor), reflecting the lack of reliability of semantic information at the sublexical level (Seidenberg & Gonnerman, 2000). In contrast, Chinese 2

includes greater semantic information at the sublexical level, showing a more direct mapping between orthography and semantics (Ho, Law, & Ng, 2000). In terms of the orthographic characteristics of Chinese characters, about 80% of Chinese characters are phonetic compound (phonograms) that consist of a semantic radical and a phonetic radical (Zhou, 1978). These semantic radicals may provide a reliable cue to the semantic category of the character. For instance, all characters that contain the semantic radical 金 (/jin1/, metal), such as 銅(/tong2/, copper), 鐵(/tie3/, iron), indicate that the characters are related to the category of metal. In addition, many semantic radicals may not stand alone as a character in Chinese. An example is 湖 (/hwu2/, lake) which is composed of a semantic radical of three dots arranged vertically on the left with an associated meaning of ‘water’. The semantic radical in this example does not correspond to a real character. Regarding the direct mapping between orthography and semantics in alphabetical languages such as English and French, when primes and targets share the same root (i.e. suffix: viewers-view), the activation of the root in the prime facilitates processing of the target via excitatory connections between morphemic representations and the whole word (Diependaele, Duñabeitia, Morris, & Keuleers, 2011; Giraudo & Grainger, 2000, 2001). In addition, previous behavioral studies in Chinese have shown faster reaction time for visually-similar character pairs (i.e., sharing semantic radicals) compared to visually-dissimilar character pairs (i.e., not sharing semantic radicals) in semantic tasks for adult readers (Chen & Allport, 1995; Fang & Zhang, 2009; Leck, Weekes, & Chen, 1995; Williams & Bever, 2010). Aforementioned studies proposed that characters are automatically decomposed into radicals during recognition. Furthermore, interactive models suggested that the semantic radicals access semantic representations independently of and in parallel to the whole characters (Taft, Liu, & Zhu, 1999; Zhou & Marslen-Wilson, 1999, 2000). While there is evidence for inter-connections between semantic radicals and whole characters from normal adults, relevant developmental data are not yet available. In particular, it is not clear how the brain constructs both lexical and sublexical level semantic information into an organized system over age. In order to explore the interaction between both levels of semantic information in Chinese, we 3

systematically manipulated the degree of semantic association between character pairs at the lexical level and orthographic similarity at the sublexical level. In this study, orthographic similarity was defined as characters having the same orthographic component (i.e. semantic radicals) (Zhou & Marslen-Wilson, 1999). In the semantic association task, two types of stimuli were used: visually–similar versus visually–dissimilar character pairs. Based on the mapping from orthography to semantics in Chinese, this experimental design allowed us to examine the interactive process between lexical (i.e. semantic association) and sublexical (i.e. semantic radical) information. Previous research using a semantic judgment task at the lexical level has identified brain regions for processing distantly related pairs (i.e. weak association) in left inferior frontal gyrus (IFG. BA 45, 47) and left middle temporal gyrus (MTG, BA 21) (Booth et al., 2002; Chou et al., 2006a, b, 2009). One of functional roles of the left inferior frontal gyrus has been suggested to result from the difficulty of retrieving/selecting appropriate features and/or inhibiting irrelevant features (Lee et al., 2011; Badre & Wagner, 2007). In support of this, many studies have identified greater inferior frontal activation in semantic tasks with greater retrieval and/or selection demands (Bedny, McGill, & Thompson-Schill, 2008; Kuperberg, Lakshmanan, Greve, & West, 2008; Pexman, Hargreaves, Edwards,

Henry,

&

Goodyear,

2007;

Snyder,

Feigenson,

&

Thompson-Schill,

2007;

Thompson-Schill, D’Esposito, Aguirre, & Farah, 1997; Thompson-Schill, D’Esposito, & Kan, 1999). Distantly related pairs also produced greater activation in the left MTG (Chou et al., 2009). There is extensive evidence demonstrating that semantic content is represented in this region (Blumenfeld, Booth, & Burman, 2006; Davis, Meunier, & Marslen-Wilson, 2004; Gourvitch et al., 2000; Kable, Kan, Wilson, Thompson-Schill, & Chatterjee, 2005; Noppeney, Friston, & Price, 2003). A review article also proposed that the MTG is a principal site for the storage of representations (Binder, Desai, Graves, & Conant, 2009). For weak association, participants may need extensive access to semantic representations in order to identify overlapping features for distant relationships (Booth, Bebko, Burman, & Bitan, 2007; Huang et al., 2012). In terms of the present study, we wished to determine whether lack of congruent sublexical information for the visually-dissimilar pairs may place greater 4

processing demands on the access to lexical representations and greater engagement of selecting appropriate representations, reflecting greater activation in the MTG and IFG as compared to the visually-similar pairs. In contrast, processing closely related pairs (i.e. strong association) elicited greater activation in left angular gyrus (AG, BA 39), suggesting greater integration of features to determine the semantic relationship between words, as closely related pairs share more overlapping features (Grossman et al., 2003; Koenig et al., 2005). Recent studies have suggested that the AG is related to manipulating semantic knowledge within the current context (Corbett, Jefferies, & Lambon Ralph, 2009; Jefferies & Lambon Ralph, 2006; Whitney, Kirk, O’Sullivan, Lambon Ralph, & Jefferies, 2011). In addition, Binder and Desai (2011) propose that this region is related to capturing similarity structures to define conceptual categories. Researchers suggest the role of this region as “combinatorial semantic processing”, integrating incoming semantic information into a current lexical level, sentential or narrative context (Humphries, Binder, Medler, & Liebenthal, 2006, Humpheries, Binder, Medler, & Liebenthal, 2007; Lau, Philips, & Poeppel, 2008). In the context of the present study, we propose that AG, as a high level convergence zone, may use the shared radical to categorize the semantic knowledge into an organized structure, allowing the overlapping features to be integrated at the lexical level. This integration in visually similar pairs with shared semantic radicals may be especially pronounced for strong association pairs because of the large number of shared features. Evidence from both functional and structural imaging studies is consistent with the notion that a developmental increase of processing conceptual knowledge continues into adolescence and is associated with changes in left IFG, left MTG, and left AG. Adolescents (aged 13-16) are better than children (aged 8-12) at selecting/retrieving relevant features (Adleman et al., 2002; Lamm, Zelazo, & Lewis, 2006), accessing lexical representations (Lee et al., 2011), and integrating conceptual relations (Crone et al., 2006; Thibaut, French, & Vezneva, 2010). The age-related changes in processing conceptual knowledge are thought to be related to semantic selection/retrieval in the IFG, access of lexical representations in MTG, and the integration of semantic information in the angular 5

gyrus (Chou et al., 2009). The changes during adolescence suggest that teenagers are more capable of maintaining conceptual knowledge in an organized state, using a flexible mechanism to retrieve/select relevant information and integrate relationships between these concepts (Blakemore & Choudhury, 2006; Crone et al., 2009). Previous studies have shown developmental increases in the MTG when making semantic judgments are particularly pronounced for the weaker association, whereas these studies have shown developmental increases in the AG are pronounced for the stronger association (Chou et al., 2009; Lee et al., 2011). We extend this work by directly examining the neural correlates of processing lexical and sublexical semantic information for children. In particular, we aimed to determine whether the semantic radicals at the sublexical level would interact with the semantic association at the lexical level, supporting the shared features between two characters to be detected and integrated at the lexical level. We expected that character pairs that shared a semantic radical (i.e. visually-similar pairs) would modulate AG activation because this shared information would allow for greater integration of semantic features at the level of the radical. In addition, the effect of the shared semantic radicals in the AG should increase with age. In contrast, we expected that pairs that did not share a semantic radical (i.e. visually-dissimilar pairs) would modulate developmental differences in the MTG and IFG because the lack of shared information would place greater demands on access to semantic representations and engage greater effort to select the appropriate features and/or inhibiting irrelevant features during judgment. Methods Participants In the visually–similar pairs, a group of 26 native monolingual Chinese children (mean age = 11.6, standard deviation = 2.1, 11 girls) participated. In the visually–dissimilar pairs, another group of 26 native monolingual Chinese children (mean age = 11.8, standard deviation = 1.9, 14 girls) participated. To examine the developmental changes of conceptual knowledge, we divided participants into two age groups: younger (8- to 12-year-old, n = 13), and older (13- to 16-year-old, n 6

= 13). All 52 children were recruited from the Taipei city metropolitan area. The parents were given an informal interview to ensure that their children met the following inclusionary criteria: (1) right-handedness, (2) normal hearing (3) normal or corrected-to-normal vision, (4) free of neurological disease or psychiatric disorders, (5) no history of intelligence, reading, or oral-language deficits, and (6) no learning disability or attention deficit hyperactivity disorder (ADHD). After the administration of the informal interview, informed consent was obtained. The informed consent procedures were approved by the Institutional Review Board at the National Taiwan University Hospital. Standardized intelligence testing was administered, using the Wechsler Intelligence Scale for Children (WISC-III) Chinese version (The Psychological Corporation, 1999). For participants in the visually-similar pairs, participants’ standard scores (mean ± SD) were 114± 11 on the verbal scale and 112 ± 13 on the performance scale. For participants in the visually-dissimilar pairs, participants’ standard scores (mean ± SD) were 110 ± 10 on the verbal scale and 107 ± 14 on the performance scale. In the visually-similar pairs, the younger and older children were matched on verbal IQ (t = 1.41, p > .05) and performance IQ (t = -0.13, p > .05). In the visually-dissimilar pairs, the younger and older children were matched on verbal IQ (t = 1.04, p > .05) and performance IQ (t = 0.77, p > .05). Functional activation tasks The children were given two practice sessions, one outside the scanner and the other in the scanner, to make sure that they understood the task. The practice items were different stimuli than those used in fMRI sessions. All participants achieved at least 80% correct for each condition in both practice sessions. The visually-similar pairs (i.e. shared a semantic radical) included 36 related pairs and 18 unrelated pairs. The 36 related pairs were further divided into 18 strong and 18 weak association pairs, according to their free association values (mean = 0.09, SD = 0.13, ranging from 0.77 to 0.01) (Hue, Kao, Lo, 2005). The visually-dissimilar pairs (i.e. no shared semantic radical) included 36 related pairs and 18 unrelated pairs that were chosen from Chou et al’s (2009) study. The examples for similar and dissimilar pairs are included in Table 1. The 36 related pairs were further 7

divided into 18 strong and 18 weak association pairs, according to their free association values (mean = 0.12, SD = 0.13, ranging from 0.73 to 0.01) (Hue et al., 2005). In this design, the related pairs in visually-similar pairs shared meaning at both lexical and sublexical level, whereas the related pairs in visually-dissimilar pairs shared meaning only at the lexical level. Because the distribution of these association values was positively skewed, we performed a logarithmic transformation on both visually-similar pairs (mean = 0.81, SD = 0.36, ranging from 1.89 to 0.30) and visually-dissimilar pairs (mean = 0.96, SD = 0.36, ranging from 1.87 to 0.30) to get a symmetrical distribution. In addition, unrelated character pairs were semantically unrelated with zero association values. The participants were instructed to quickly and accurately press with their right hand the yes button to the related pairs and the no button to the unrelated pairs. Experimental procedures For the visually-similar pairs and the visually-dissimilar pairs, the presentation procedures were the same. Two Chinese characters were presented sequentially and the participant had to determine whether the character pair was related in meaning. Trials lasted 4500 ms and consisted of a solid square (500 ms), followed by the first character (800 ms), a 200 ms blank interval, and the second character for 3000 ms. The participant was instructed to make a response during the presentation of the second character. The perceptual control condition had 24 pairs of non-characters. Non-characters were created by replacing radicals of real characters with other radicals that did not form real Chinese characters. Non-characters were larger (50 font size) than real characters (40 font size) in order to encourage participants to perform the task based on the recognition of low-level visual similarity and not on the extraction of semantic information. For the perceptual control condition, trials consisted of a solid square (500 ms), followed by the first non-character (800 ms), a 200 ms blank interval, and the second non-character for 3000 ms. Participants determined whether the pair of stimuli were identical or not by pressing a yes or no button with their right hand. In addition, the non-characters do not share any semantic radicals with the stimuli in the visually-similar and visually-dissimilar pairs. There were also 24 baseline events as “null” trials so 8

that we could better deconvolve the response to the lexical and perceptual trials. The participant was instructed to press a button when a solid square (1300 ms) at the center of the visual field turned to a hollow square (3000 ms) after a blank interval (200 ms). We chose to present data comparing the semantic to the perceptual control conditions to subtract out activation in the primary visual cortex that it is not likely due to semantic processing. Stimulus characteristics For visually-similar and visually-dissimilar pairs, several lexical variables were controlled across the strong association, weak association and unrelated character pairs. First, all Chinese characters were monosyllabic. Second, the first character and the second character did not form a compound word (Huang, 1998; Wu & Liu, 1987). Third, in terms of association strength, t-tests demonstrated that the strong association had larger association values than the weak association for both visually-similar pairs (t = 3.03, p < .05) and visually-dissimilar pairs (t = 4.44, p < .05). Fourth, characters were matched for visual complexity (in terms of strokes per character) across conditions (Hung et al., 2010). Fifth, characters were matched for written frequency (Wu & Liu, 1987), radical frequency (National Language Committee in Ministry of Education, 1998) and written familiarity (Hung, Lee, Chen, & Chou, 2010) for children. Sixth, the correlation of frequency (first or second characters) with association strength was not significant, indicating that association effects should not be due to frequency (Hung et al., 2010). Several lexical variables were also controlled across visually-similar and visually-dissimilar pairs, including association strength, strokes, and written frequency. For the visually-similar pairs, the correlation of the transparency of semantic radicals with association strength was not significant, indicating that association effects should not be due to the transparency of semantic radicals (Hung et al., 2010). The rating for transparency was conducted with fifty undergraduate students who were asked to determine if the meaning of the semantic radical and the character were similar (Hung et al., 2010). Raters were asked to judge each character on a 5-point scale, in which 1 stood for low transparency and 5 stood for high transparency. The rating

9

procedures and the operational definitions were similar to previous developmental studies on semantic transparency of radicals (Chung & Leung, 2008; Ho et al., 2003; Shu & Anderson, 1997). MRI data acquisition Participants lay in the scanner with their head position secured with a specially designed vacuum pillow. An optical response box was placed in the participants’ right hand. The head coil (8-channel head coil) was positioned over the participants’ head (Siemens, Erlangen, Germany). Participants viewed visual stimuli projected onto a screen via a mirror attached to the inside of the head coil. This study adopted an event-related design. All images were acquired using a 3 Tesla Siemens TIM Trio human MRI scanner. Gradient-echo localizer images were acquired to determine the placement of the functional slices. For the functional imaging studies, a susceptibility weighted single-shot EPI (echo planar imaging) method with BOLD (blood oxygenation level-dependent) was used. Functional images were collected parallel to AC-PC plane with interleaved whole brain EPI acquisition from bottom to top. The following scan parameters were used: TE = 24 ms, flip angle = 90o, matrix size = 64  64, field of view = 25.6 cm, slice thickness = 3 mm, number of slices = 34, TR = 2000 ms. In visually-similar pairs, participants performed two functional runs and each run had 116 image volumes (4.0 minutes, 18 strong, 18 weak, and 18 unrelated pairs in total). In visually-dissimilar pairs, each participant performed two functional runs and each run had 136 image volumes (4.7 minutes each, 24 strong, 24 weak, and 24 unrelated pairs in total), For data analyses, we chose 18 strong, 18 weak and 18 unrelated pairs from the visually-dissimilar pairs that were well-matched to the visually-similar pairs so the number of trials was the same across tasks. In addition, a high resolution, T1 weighted 3D image was acquired (TR = 1560 ms, TE = 3.68 ms, flip angle = 15o, matrix size = 256  256, field of view = 25.6 cm, slice thickness = 1 mm, number of slices = 192). The orientation of the 3D image was identical to the functional slices. The task was administered in a pseudorandom order for all subjects, in which the order of strong, weak, unrelated, perceptual, and baseline trials was optimized for event-related design (Burock, Buckner, Woldroff, Rosen, & Dale, 1998). 10

Image analysis Data analysis was performed using SPM2 (Statistical Parametric Mapping). The functional images were corrected for differences in slice-acquisition time to the middle volume and were realigned to the first volume in the scanning session using affine transformations. No participant had more than 3 mm of movement in any plane. Co-registered images were normalized to the MNI (Montreal Neurological Institute) average template (12 linear affine parameters for brain size and position, 8 non-linear iterations and 2  2  2 nonlinear basis functions). Statistical analyses were calculated on the smoothed data (10 mm isotropic Gaussian kernel), with a high pass filter (128 seconds cutoff period) in order to remove low frequency artifacts. Data from each participant was entered into a general linear model using an event-related analysis procedure (Josephs & Henson, 1999). Character pairs were treated as individual events for analysis and modeled using a canonical HRF (Hemodynamic Response Function). There were five event types: strong association, weak association, unrelated, perceptual, and baseline. Only correct responses were included in the analysis. The resulting model coefficients for individual subjects were entered into subsequent second-order random effects analyses in a whole brain analysis. Three kinds of analyses were conducted in the current study. First, in order to determine overall association effects within similar and dissimilar pairs, we compared the strong association and the weak association separately to the perceptual control condition across age groups in a whole brain analysis. All the reported areas of activation were significant using p < .05 FDR (false discovery rate) corrected at the voxel level with a cluster size greater than or equal to 10 voxels. Second, within each association condition, in order to determine age differences between the similar and dissimilar pairs, we conducted analyses of 2 pair (similar, dissimilar) by 2 age (older, younger) ANOVA on the contrast [strong association versus perceptual control] and the contrast [weak association versus perceptual control] separately. In addition, within similar and dissimilar pairs, in order to examine the age differences between the strong versus weak association, we conducted analyses of 2 association (strong, weak) by 2 age (older, younger) ANOVA on the contrast [strong association 11

versus weak association]. Due to our a priori hypothesis on regions of interest (AG, MTG, and IFG), all reported areas of activation in the interaction effects were significant using a functional mask with a strict threshold (p < .05 FWE, familywise error) corrected with a sphere of 5 mm radius centered on the commonly significant voxels from the within-group analysis (Chen et al., 2013). Third, in order to examine that developmental effects were not due to differences in behavioral performance, we partialled out behavioral performance (i.e. reaction time) for the interaction between pair and age, and the interaction between association and age. We extracted the beta values from peak voxels of brain regions that showed association effects across the similar and dissimilar pairs (i.e. angular gyrus, inferior frontal gyrus, and middle temporal gyrus) for the older group and the younger group. This allowed us to do ANCOVA analysis of age effect, using reaction time as a covariate. Results Behavior performance The overall accuracy and reaction times for the five experimental conditions in the similar and dissimilar tasks are listed in Table 2. Analyses of accuracy within and between pairs A 2 pair (similar, dissimilar) by 2 age (older, younger) by 2 condition (strong, weak) ANOVA on accuracy was performed. This analysis showed no significant main effect of pair, F(1, 48) = 1.13, p = .29. The main effect of age was significant, F(1, 48) = 7.01, p < .05, with the older children being more accurate than the younger children. The main effect of condition was significant, F(1, 48) = 8.82, p < .05, with the strong association being more accurate than the weak association condition. There were no significant interaction effects (ps > .05). In addition, the accuracies for the unrelated, perceptual, and baseline events were not significantly different between age groups (ps > .05). Analyses of reaction times within and between tasks A 2 pair (similar, dissimilar) by 2 age (older, younger) by 2 condition (strong, weak) ANOVA on reaction times was performed. This analysis showed no significant main effect of pair, F(1, 48) = 0.33, p = .57. The main effect of age was significant, F(1, 48) = 10.64, p < .05, with the older 12

children being faster than the younger children. The main effect of condition was significant, F(1, 48) = 10.46, p < .05, with the strong association being faster than the weak association. The interaction of pair by condition was significant, F(1, 48) = 5.18, p < .05. A simple main effect analysis showed faster reaction times for the strong association than for the weak association in the visually-similar pair (t = 4.18, p < .05), but not in the visually-dissimilar pair (ps > .05). There were no other significant interaction results (ps > .05). In addition, the reaction times for the unrelated, perceptual, and baseline events were not significantly different between age groups (ps > .05). Brain activation patterns Overall association effects within similar and dissimilar pairs Table 3 shows greater activation for the strong or the weak association condition versus the perceptual control condition for similar and dissimilar pairs. In the visually-similar pairs, the strong association versus perceptual controls produced greater activation in left inferior frontal gyrus (IFG, BA 45/47), middle temporal gyrus (BA 21) and the angular gyrus (AG, BA 39). In addition, the weak association versus perceptual controls produced greater activation in the AG, IFG, and left middle temporal gyrus (MTG, BA 21). In the visually-dissimilar pairs, the strong association versus perceptual controls produced greater activation in the IFG, AG, and MTG. In addition, the weak association versus perceptual controls produced greater activation in the IFG and MTG. Age effects between pair and between association In order to equate power between age groups and between conditions, we randomly eliminated items from the conditions with higher numbers of correct trials. Table 4 lists the interaction effects. First, for the interaction of age by pair within the strong association, greater activation was found in the AG for older children compared to younger children between similar and dissimilar pairs on the contrast of the strong association versus perceptual controls (Fig. 1a). Second, there was no age by pair interaction effect within the weak association. Because common activations were found in the IFG and MTG for the weak association, we pooled the brain activation across similar and dissimilar pairs to observe age effects. The weak association versus perceptual controls elicited greater 13

activation in the IFG and MTG for older children compared to younger children (Fig. 1b). Third, for the interaction of age by association in the visually-similar pairs, the contrast of the strong association versus weak association elicited greater activation in the AG for older children compared to younger children (Fig. 1c). Fourth, for the interaction of age by association in the visually-dissimilar pairs, the contrast of weak association versus strong association elicited greater activation in the MTG for older children compared to younger children (Fig. 1d). Age effects partialing for behavioral performance We extracted beta values from 3 regions of interest (ROIs) including the AG, MTG, and IFG by a functional mask defined by the overlap activation between similar and dissimilar pairs. Because the peaks were very close across two association conditions for both pairs, we performed ANCOVA analyses with reaction time as a covariate. First, for the interaction of age by pair within the strong association, we performed an age (older, younger) by pair (similar, dissimilar) ANCOVA on the AG. The interaction was significant [F(1, 47) = 4.43, p < .05], reflecting greater intensity for the similar pairs than the dissimilar pairs in older children (t =4.60, p < .05), but not in younger children (p > .05) (Fig 2a). Second, for the age effects across similar and dissimilar pairs within the weak association, we performed one way ANCOVAs for the MTG and IFG. Greater signal intensity was found for older than younger children in the MTG [F(1, 47) = 4.61, p < .05] and in the IFG [F(1, 47) = 5.52, p < .05] (Fig 2b). Third, for the interaction of age by association in the visually-similar pairs, we performed an age (older, younger) by association (strong, weak) ANCOVA on the AG. The interaction was significant [F(1, 47) = 4.57, p < .05], reflecting greater signal intensity for the strong association than the weak association in the older group (t = 3.36, p < .05), but not in the younger group (p > .05) (Fig 2c). Fourth, for the interaction of age by association in the visually-dissimilar pairs, we performed an age (older, younger) by association (strong, weak) ANCOVA on the MTG. The interaction was significant [F(1, 47) = 4.52, p < .05], reflecting greater signal intensity for the weaker association than the stronger association in the older group (t = 2.84, p < .05), but not in the

14

younger group (p > .05) (Fig 2d). Similar effects were obtained when treating age as a continuous factor.

Discussion By orthogonally manipulating semantic association and semantic radicals, this study aimed to explore the interaction of lexical and sublexical information during reading. We examined two groups of Chinese children (8-12 versus 13-16 years of age) by comparing the visually-similar to the visually-dissimilar pairs that involved relatedness judgments to character pairs varying in their semantic association. The results demonstrated greater developmental increases in the angular gyrus (AG, BA 39) in the visually-similar pairs compared to the visually-dissimilar pairs for the strong association. We also found greater age-related increases in the AG for the strong compared to weak association in the visually-similar pairs. Finally, we found developmental increases in left MTG and left IFG for weak association across similar and dissimilar pairs and also left MTG in the contrast of the weak compared to strong association in the visually-dissimilar pairs. The first major finding was that the strong association pairs elicited greater AG activation in older children for the visually-similar pairs than the visually-dissimilar pairs. This region has been suggested to be associated with semantic control, manipulating underlying semantic information within the current context in order to produce task-appropriate responses (Corbett et al., 2009; Jefferies & Lambon Ralph, 2006). It should be noted that the AG (BA 39) is more posterior ([-63, -51, 24] and [-57, -66, 24]) than the supramarginal gyrus (BA 40), which has been indicated as the phonological store for working memory (Buchsbaum & D’Esposito, 2008; Paulesu, Frith, & Frackowiak, 1993; Ravizza, Delgado, Chein, Becker, & Fiez, 2004; Smith, Jonides, & Koeppe, 1996). More specifically, the AG has been suggested to be sensitive to similarity structures that define categories and integrate the incoming semantic information into a larger conceptual unit (Binder & Desai, 2011; Humphries et al., 2006, 2007, Lau et al., 2008). Due to the direct mapping of 15

orthographic form and semantics in Chinese, the readers may access partial meaning cues from the semantic radicals at the sublexical level. For instance, the character pair 銅(/tong2/, copper) and 鐵 (/tie3/, iron) having the same semantic radical 金 (/jin1/, metal) in the left of the characters, may generate the concept in a superordinate level of the semantic category (i.e. metal). The semantic radical may automatically activate relevant semantic features for the visually-similar pairs (Hung et al., 2010; Tsang & Chen, 2009; Yeh & Li, 2002, 2004; Zhou & Marslen-Wilson, 1999). Our results are also consistent with the interactive model that readers may automatically decompose the embedded radicals from the whole character and map radicals onto their own semantic representations, in parallel to the mapping that relies on the whole character, producing a priming effect on meaning judgments (Law et al., 2005; Zhou & Marslen-Wilson, 1999). In the present study, compared to the visually-dissimilar pairs, the shared semantics at the sublexical level (i.e. semantic radicals) in the visually-similar pairs may facilitate the integration of overlapping features at the lexical level. We also found age-related increases in the AG in the strong association compared to the weak association in the visually-similar pairs. A developmental study has identified age-related changes in the AG for meaning processing in Chinese (Chou et al, 2009). However, this study did not vary orthographic similarity in their design. In the present study, the greater activation in this region for the visually-similar pairs indicates that older children (i.e. adolescents) use the semantic radical to perform semantic integration more effectively in the strong association compared to the weak association. Age-related increases in the sensitivity to semantic radicals have been found in behavioral studies, suggesting that Chinese children start to use the semantic radical to identify the semantic category of a character and to access its meaning at age 6 (Chan & Nunes, 1998). This ability to use semantic radicals further improves between ages 6 and 12 (Ho, Ng, & Ng, 2003; Shu & Anderson, 1997; Yeh, Lin, & Li, 2004). Moreover, fifth graders (aged 12) have been shown to have better ability in using semantic radicals for effective reading than younger children (aged 8-11) presumably as a result of the older children’s more elaborate semantic representations (Ho et al., 16

2003). In the present study, the more structured semantic representations in older children may support their utilization of semantic radicals to guide semantic integration, showing greater activation in the AG for the visually-similar pairs. In the second major finding, a developmental increase was observed in the MTG and IFG in the weak association across similar and dissimilar pairs. Previous English and Chinese studies have also found age-related changes in semantic processing in these two regions (Chou et al., 2006a, 2006b, 2009; Lee et al., 2011). Several studies suggest that the MTG is related to extensive access to lexical representations (Booth et al., 2007; Huang et al., 2012), whereas the IFG is engaged in effortful situations that required selection and/or retrieval (Lee et al., 2011; Badre & Wagner, 2007). Compared to the strong association pairs, the weak association may require extensive access to lexical representations to identify overlapping features for distant relationships and need to select the appropriate representations. Moreover, we found that, compared to the strong association, the weak association produced greater activation over age in the MTG for the visually-dissimilar pairs, demonstrating that lack of congruency between the meaning of the radicals and the meaning of the whole characters may place greater processing demands on the access to lexical representations. Older children may be more sensitive in detecting the congruency between meanings at the sublexical and lexical levels when processing lexical representations. There are several potential limitations of the current study. First, we showed that there were developmental differences in accuracy and reaction time. However, we performed our fMRI analyses only on correct responses and we covaried out differences in reaction times, mitigating the influence of these variables on observed developmental effects. In addition, we showed that there was an interaction between pair (similar, dissimilar) and condition (strong, weak) with the expected effect of best performance for the strong association in the visually-similar pairs. Importantly, however, there were no interactions with age suggesting that developmental brain changes in task or condition effects were not driven by performance differences. Second, some researchers highlight the role of the anterior temporal lobe related to semantic integration (Patterson, Nestor, & Rogers, 2007). In the 17

present study, the lack of anterior temporal activation may be due to susceptibility effects in fMRI (Zhu et al., 2009). Alternatively, numerous studies have found anterior temporal activation that responds preferentially to sentence-level stimuli (Dronkers, Wilkins, Van Valin, Redfern, & Jaeger, 2004; Humphries, Willard, Buchsbaum, & Hickok, 2001; Rogalsky & Hickok, 2009). In the present study, it is possible that anterior temporal lobe may not be significantly activated to stimuli in the character-level. Third, we did not have an independent measure of reading ability. Although standardized verbal and performance IQ scores were matched between age groups in this study, further studies are needed to clarify the relations between age effects and reading abilities. Finally, unequal levels of motivation between younger and older children may affect their task performance. Evidence from neurodevelopmental studies is consistent with the notion of an age-related increase in motivational control and reward processing that is associated with changes in the medial orbital frontal cortex (BA 10, 11) and ventral striatum (Bjork et al., 2010; Geier et al., 2010; Leijenhorst et al., 2010). Their findings suggest that developmental changes in the frontostriatal circuits may account for influences of rewards on task performance (Padmanabhan et al., 2011). However, it should be noted that the age-related increase in the IFG in our study has not been implicated in motivation or reward, but in cognitive control. In conclusion, when comparing the visually-similar pairs with the visually-dissimilar pairs, the strong association pairs produced greater activation in the angular gyrus for older children. In addition, this age effect was more prominent in the strong association compared to the weak association in the visually-similar pairs. Both results indicate that shared semantics at the sublexical level may facilitate the integration of overlapping features at the lexical level in older children. Moreover, we demonstrated developmental increases for the weak association in the middle temporal gyrus for visually dissimilar pairs, reflecting that conflicting information at the sublexical level may result in greater demands on access to lexical representations in older children. More generally, this study shows that older children are more sensitive to sublexical information when reading.

18

Acknowledgements

This research was supported by grants from the National Science Council of Taiwan (NSC 98-2410-H-002-025-MY3, NSC 103-2420-H-002-008, MOST 104-2420-H-002-004) to Tai-Li Chou. This research was also supported by grants from the National Institute of Child Health and Human Development (HD042049) to James R. Booth. This research was supported in part by the Department of Medical Imaging and 3T MRI Lab in National Taiwan University Hospital. References Adleman, N. E., Blasey, C. M., White, C. D., Warsofsky, I. S., Glover, G. H., & Reiss, A. L. (2002). A developmental functional fMRI study of the Stroop color-word task. NeuroImage, 16, 61-75. Badre, D., & Wagner, A. D. (2007). Left ventrolateral prefrontal cortex and the cognitive control of memory. Neuropsychologia, 45, 2883–2901. Bedny, M., McGill, M., & Thompson-Schill, S. L. (2008). Semantic adaptation and competition during word comprehension. Cerebral Cortex, 18, 2574–2585. Binder, J. R., & Desai, R. H. (2011). The neurobiology of semantic memory. Trends in Cognitive Sciences, 15(11), 527-536. Binder, J. R., Desai, R. H., Graves, W. W., & Conant, L. L. (2009). Where is the semantic system? A critical review and meta-analysis of 120 functional neuroimaging studies. Cerebral Cortex, 19, 2767-2796. Bjork, J. M., Smith, A. R., Chen, G., & Hommer, D. W. (2010). Adolescents, adults and rewards: comparing motivational neurocircuitry recruitment using fMRI. PLoS One, 5, e11440. Blakemore, S. J., & Choudhury, S. (2006). Development of the adolescent brain: implications for executive function and social cognition. Journal of Child Psychology and Psychiatry, 47(3), 296–312. Blumenfeld, H. K., Booth, J. R., & Burman, D. D. (2006). Differential prefrontal– temporal neural correlates of semantic processing in children. Brain and Language, 99, 226-235. 19

Booth, J. R., Bebko, G., Burman, D. D., & Bitan, T. (2007). Children with reading disorder show modality independent brain abnormalities during semantic tasks. Neuropsychologia, 45(4), 775–783. Booth, J. R., Burman, D. D., Meyer, J. R., Gitelman, D. R., Parrish, T. B., & Mesulam, M. M. (2002). Modality independence of word comprehension. Human Brain Mapping, 16, 251–261. Buchsbaum, B. R., & D’Esposito, M. (2008). The search for the phonological store: from loop to convolution. Journal of Cognitive Neuroscience, 20, 762–778. Burock, M. A., Buckner, R. L., Woldorff, M. G., Rosen, B. R., & Dale, A. M. (1998). Randomized event-related experimental designs allow for extremely rapid presentation rates using functional MRI. NeuroReport, 9, 3735–3739. Chan, L., & Nunes, T. (1998). Children’s understanding of the formal and functional characteristics of written Chinese. Applied Psycholinguistics, 19, 115-131. Chen, P. J., Fan, L. Y., Hwang, T. J., Hwu, H. G., Liu, C. M., & Chou, T. L. (2013). The deficits on a cortical-subcortical loop of meaning processing in schizophrenia. NeuroReport, 24, 147-151. Chen, Y. P., & Allport, D. A. (1995). Attention and lexical decomposition in Chinese word recognition: conjunctions of form and position guide selective attention. Visual Cognition, 2, 235-268. Chou, T. L., Booth, J. R., Bitan, T., Burman, D. D., Bigio, J. D., Cone, N. E., Lu, D., & Cao, F. (2006a). Developmental and skill effects on the neural correlates of semantic processing to visually presented words. Human Brain Mapping, 27, 915–924. Chou, T. L., Booth, J. R., Burman, D. D., Bitan, T., Bigio, J. D., Lu, D., & Cone, N. E. (2006b). Developmental changes in the neural correlates of semantic processing. NeuroImage, 29, 1141–1149. Chou, T. L., Chen, C. W., Fan, L. Y., Chen, S. Y., & Booth, J. R. (2009). Testing for a cultural influence on reading for meaning in the developing brain: The neural basis of semantic processing in Chinese children. Frontiers in Human Neuroscience, 3(27), 1–9.

20

Corbett, F., Jefferies, E., & Lambon Ralph, M. A. (2009). Exploring multimodal semantic control impairments in semantic aphasia: evidence from naturalistic object use. Neuropsychologia, 47, 2721-2731. Crone, E. A., Wendelken, C., Donohue, S., van Leijenhorst, L., & Bunge, S. A. (2006). Neurocognitive development of the ability to manipulate information in working memory. Proceedings of the National Academy of Sciences, 103(24), 9315–9320. Crone, E. A., Wendelken, C., Leijenhorst, L. V., Honomichl, R. D., Christoff, K., & Bunge, S. A. (2009). Neurocognitive development of relational reasoning. Developmental Science, 12(1), 55-66. Davis, M. H., Meunier, F., & Marslen-Wilson, W. D. (2004). Neural responses to morphological, syntactic, and semantic properties of single words: An fMRI study. Brain and Language, 89 (3), 439-449. Diependaele, K., Duñabeitia, J. A., Morris, J., & Keuleers, E. (2011). Fast morphological effects in first and second language word recognition. Journal of Memory and Language, 64, 344-358. Dronkers, N. F., Wilkins, D. P., Van Valin, R., Jr, Redfern, B., & Jaeger, J. (2004). Lesion analysis of the brain areas involved in language comprehension. Cognition, 92, 145–177. Fang, Y. H., & Zhang, J. J. (2009). Asymmetry in naming and categorizing of Chinese words and pictures: Role of semantic radicals. Acta Psychologica Sinica, 41(2), 114-126. Geier, C. F., Terwilliger, R., Teslovich, T., Velanova, K., & Luna, B., (2010). Immaturities in reward processing and its influence on inhibitory control in adolescence. Cerebral Cortex, 20, 1613–1629. Giraudo, H., & Grainger, J. (2000). Effects of prime word frequency and cumulative root frequency in masked morphological priming. Language and Cognitive Processes, 15, 421–444. Giraudo, H., & Grainger, J. (2001). Priming complex words: Evidence for supralexical representation of morphology. Psychonomic Bulletin & Review, 8 (1), 127-131. Gourovitch, M. L., Kirkby, B. S., Goldberg, T. E., Weinberger, D. R., Gold, J. M., Esposito, G., Van Horn, J. D., & Berman, K. F. (2000). A comparison of rCBF patterns during letter and semantic fluency. Neuropsychology, 14(3), 353-360. 21

Grossman, M., Koenig, P., Glosser, G., DeVita, C., Moore, P., Rhee, J., Detre, J., Alsop, D., & Gee, J. (2003). Neural basis for semantic memory difficulty in Alzheimer’s disease: an fMRI study. Brain, 126, 292–311. Ho, C. S., Law, T. P., & Ng, P. M. (2000). The phonologcial deficit hypothesis in Chinese developmental dyslexia. Reading and Writing, 13, 57–79. Ho, C. S. H., Ng, T. T., & Ng, W. K. (2003). A “radical” approach to reading development in Chinese: The role of semantic radicals and phonetic radicals. Journal of Literacy Research, 35, 849-878. Huang, C. R. (1998). Academia Sinica Balanced Corpus (3rd ed.). Taipei. Taiwan: Academia Sinica. Huang, J., Zhu, Z., Zhang, J. X., Wu, M., C., H.-C., W., S. (2012). The role of left inferior frontal gyrus in explicit and implicit semantic processing. Brain Research, 1440, 56-64. Hue, C. W., Kao, C. H., & Lo, M. (2005). Association norms for 600 Chinese characters. Taiwanese Psychological Association. Humphries, C., Binder, J. R., Medler, D. A., & Liebenthal, E. (2006). Syntactic and semantic modulation of neural activity during auditory sentence comprehension. Journal of Cognitive Neuroscience, 18(4), 665–679. Humphries, C, Binder J. R., Medler D. A., & Liebenthal E. (2007). Time course of semantic processes during sentence comprehension: an fMRI study. NeuroImage, 36, 924-932. Humphries, C., Willard, K., Buchsbaum, B., & Hickok, G. (2001). Role of anterior temporal cortex in auditory sentence comprehension: An fMRI study. NeuroReport, 12, 1749–1752. Hung, K. C., Lee, S. H., Chen, S. Y., & Chou, T. L. (2010). Effects of semantic radical and semantic association on semantic processing of Chinese characters for adults and fifth graders. Chinese Journal of Psychology, 52(3), 327-344. Jefferies, E., & Lambon Ralph, M. A. (2006). Semantic impairment in stroke aphasia versus semantic dementia: a case-series comparison. Brain, 129, 2132-2147. Josephs, O., & Henson, R. N. (1999). Event-related functional magnetic resonance imaging: modelling, inference and optimization. Philosophical Transactions of the Royal Society B, 354, 1215–1228. 22

Kable, J. W., Kan, I. P., Wilson, A., Thompson-Schill, S .L., & Chatterjee, A. (2005). Conceptual representations of action in the lateral temporal cortex. Journal of Cognitive Neuroscience, 17(12), 1855-1870. Koenig, P., Smith, E. E., Glosser, G., DeVita, C., Moore, P., McMillan, C., Gee, J., & Grossman, M. (2005). The neural basis for novel semantic categorization. NeuroImage, 24, 369–383. Kuperberg, G. R., Lakshmanan, B. M., Greve, D. N., & West, W. C. (2008). Task and semantic relationship influence both the polarity and localization of hemodynamic modulation during lexico-semantic processing. Human Brain Mapping, 29, 544–561. Lamm, C., Zelazo, P. D., & Lewis, M. D. (2006). Neural correlates of cognitive control in childhood and adolescence: disentangling the contributions of age and executive function. Neuropsychologia, 44, 2139–2148. Lau, E. F, Phillips, C., & Poeppel, D. (2008). A cortical network for semantics: (de)constructing the N400. Nature Review Neuroscience, 9, 920–933. Law, S. P., Yeung, O., Wong, W., & Chiu, K. M. Y. (2005). Processing of semantic radicals in writing Chinese characters: data from a Chinese dysgraphic patient. Cognitive Neuropsychology, 22 (7), 885-903. Leck, K. J., Weekes, B. S., & Chen, M. J. (1995). Visual and phonological pathways to lexicon: evidence from Chinese readers. Memory and Cognition, 23, 468-476. Lee, S. H., Booth, J. R., Chen, S. Y., & Chou, T. L. (2011). Developmental changes in the inferior frontal cortex for selecting semantic representations. Developmental Cognitive Neurosicence, 1, 338–350. Leijenhorst, L. V., Zanolie, K., Van Meel, C. S., Westenberg, P. M., Rombouts, S. A. R. B., & Crone, E. A. (2010). What motivates the adolescent? Brain regions mediating reward sensitivity across adolescence. Cerebral Cortex, 20, 61-69. Morris, J., Frank, T., Grainger, J., & Holcomb, P. J. (2007). Semantic transparency and masked morphological priming: An ERP investigation. Psychophysiology, 44, 506-521. 23

National Language Committee in Ministry of Education. (1998). A Report of 1998 Survey on Frequent Words Usage. Taipei, Taiwan: Ministry of Education. Noppeney, U., Friston, K. J., & Price, C. J. (2003). Effects of visual deprivation on the organization of the semantic system. Brain, 126(7), 1620-1627. Padmanabhan, A., Geier, C. F., Ordaz, S. J., Teslovich, T. Luna, B. (2011). Developmental changes in brain function underlying the influence of reward processing on inhibitory control. Developmental Cognitive Neuroscience, 1, 517-529. Patterson, K., Nestor, P. J., & Rogers, T. T. (2007). Where do you know what you know? The representation of semantic knowledge in the human brain. Nature Review, 8, 976-987. Paulesu, E., Frith, C. D., & Frackowiak, R. S. J. (1993). The neural correlates of the verbal component of working memory. Nature, 362, 342–345. Pastizzo, M. J., & Feldman, L. B. (2009). Multiple dimensions of relatedness among words: Conjoint effects of form and meaning in word recognition. Mental Lexicon, 4(1), 1–25. Pexman, P. M., Hargreaves, I. S., Edwards, J. D., Henry, L. C., & Goodyear, B. G. (2007). Neural correlates of concreteness in semantic categorization. Journal of Cognitive Neuroscience, 19(8), 1407–1419. Rastle, K., Davis, M. H., Marslen-Wilson, W. D., & Tyler, L. K. (2000). Morphological and semantic effects in visual word recognition: A time-course study. Language & Cognitive Processes, 15, 507–537. Ravizza, S. M., Delgado, M. R., Chein, J. M., Becker, J. T., & Fiez, J. A. (2004). Functional dissociations within the inferior parietal cortex in verbal working memory. NeuroImage, 22, 562–573. Rogalsky, C., & Hickok, G. (2009). The role of Broca’s area in sentence comprehension. Journal of Cognitive Neuroscience, 23, 1664–1680. Seidenberg, M. S., & Gonnerman, L. M. (2000). Explaining derivational morphology as the convergence of codes. Trends in Cognitive Neuroscience, 4(9), 353-361. 24

Shu, H., & Anderson, R. C. (1997). Role of radical awareness in the character and word acquisition of Chinese children. Reading Research Quaterly, 32(1), 78–89. Smith, E. E., Jonides, J., & Koeppe, R. A. (1996). Dissociating verbal and spatial working memory using PET. Cerebral Cortex, 6, 11–20. Snyder, H. R., Feigenson, K., & Thompson-Schill, S. L. (2007). Prefrontal cortical response to conflict during semantic and phonological tasks. Journal of Cognitive Neuroscience, 19(5), 761–775. Taft, M., Liu, Y., & Zhu, X. (1999). Morphemic processing in reading Chinese. In J. Wang, A. Inhoff, & H.-C. Chen (Eds.), Reading Chinese script: A cognitive analysis (pp. 91–113). Hillsdale, NJ: Lawrence Erlbaum Associates Inc. Thibaut, J. P., French, P., & Vezneva, M. (2010). The development of analogy making in children: Cognitive load and executive functions. Journal of Experimental Child Psychology, 106, 1-19. Thompson-Schill, S. L., D’Esposito, M., Aguirre, G. K., & Farah, M. J. (1997). Role of left inferior prefrontal cortex in retrieval of semantic knowledge: a reevaluation. Proceedings of the National Academy of Sciences USA, 94, 14792–14797. Thompson-Schill, S. L., D’Esposito, M., & Kan, I. P. (1999). Effects of repetition and competition on activity in left prefrontal cortex during word generation. Neuron, 23, 513–522. Tsang, Y. K., & Chen, H. C. (2009). Do position-general radicals have a role to play in processing Chinese characters? Language and Cognitive Processes, 24(7), 947-966. Wang, S. Dong, X., Ren, Y., & Yang, Y. (2009). The development of semantic priming effect in childhood: an event-related potential study. NeuroReport, 20, 574-578. Whitney, C., Kirk, M., O’Sullivan, J., Lambon Ralph, M. A., & Jefferies, E. (2011). The neural organization of semantic control: TMS evidence for a distributed network in left inferior frontal and posterior middle temporal gyrus. Cerebral Cortex, 21(5), 1066-1075. Williams, C., & Bever, T. (2010). Chinese character decoding: a semantic bias? Reading and Writing, 23, 589–605.

25

Wu, J. T., & Liu, I. M. (1987). Exploring the phonetic and semantic features of Chinese words. Taiwan National Science Council, Technical Report NSC75-0301-H002-024. Yeh, S. L., & Li, J. J. (2002). Role of structure and component in judgments of visual similarity of Chinese characters. Journal of Experimental Psychology: Human Perception and Performance, 28, 933–947. Yeh, S. L., & Li, J. L. (2004). Sublexical processing in visual recognition of Chinese characters: Evidence from repetition blindness for subcharacter components. Brain and Language, 88, 47–53. Yeh, S. L., Lin, Y. H., & Li, J. L. (2004). Role of character structure in judgments of visual similarity of Chinese characters for children in elementary school. Journal of Education and Psychology, 27, 95–118. Zhou, X. L., & Marslen-Wilson, W. (1999). The nature of sublexical processing in reading Chinese characters. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 819–837. Zhou, X., & Marslen-Wilson, W. (2000). Lexical representation of compound words: Cross-linguistic evidence. Psychologia, 42, 47–66. Zhou, X. L., Wu, N. N., & Shu, H. (1998). The relative time course of semantic and phonological activation in reading Chinese: Evidence from child development. Psychological Science, 21, 498-501. Zhou, Y. G. (1978). To what extent are the "phonetics" of present-day Chinese characters still phonetic. Zhongguo Yuwen, 146, 172-177. Zhu, Z., Zhang, J. X., Wang, S., Xiao, Z., Huang, J., & Chen, H.-C. (2009). Involvement of left inferior frontal gyrus in sentence-level semantic integration. NeuroImage, 47, 756–763. Figure Captions Figure 1. Developmental changes in the visually-similar and visually-dissimilar pairs. (a) The contrast of strong association versus perceptual controls produced greater activation in left angular gyrus (AG, BA 39) in the visually-similar pairs than in the visually-dissimilar pairs for the older than for the younger group. (b) Across similar and dissimilar pairs, the contrast of the weak association 26

versus perceptual controls produced greater activation in left inferior frontal gyrus (IFG, BA 47) and in the middle temporal gyrus (MTG, BA 21) for the older than for the younger group. (c) In the visually-similar pairs, the contrast of the strong association versus weak association elicited greater activation in left angular gyrus for the older than for the younger group. (d) In the visually-dissimilar pairs, the contrast of the weak versus strong association elicited greater activation in left middle temporal gyrus for the older than for the younger group. Bar plots of the signal intensity are underneath brain maps. Figure 2. Age differences in signal intensity for the visually similar and the visually dissimilar pairs. (a) For both similar and dissimilar pairs, beta values were taken from the contrast [strong condition versus the perceptual controls] for AG. Greater intensity was found for the similar pairs than the dissimilar pairs in older children, but not in younger children. (b) Across similar and dissimilar pairs, beta values were taken from the contrast of [weak association versus the perceptual controls] for the MTG and IFG. Greater signal intensity was found for older than younger children in the MTG and in the IFG. (C) For the AG activation in the similar pairs, beta values were compared between the strong association, the weak association, and unrelated condition separately to the perceptual controls. Greater signal intensity was found for the strong association than the weak association in the older group, but not in the younger group. (d) For the MTG activation in the dissimilar pairs, beta values were compared between the strong association, the weak association, and unrelated condition separately to the perceptual controls. Greater signal intensity was found for the weak association than the strong association in the older group, but not in the younger group.

Table 1 Examples for each condition in similar and dissimilar pairs.

Visually-similar

Strong association 蹦 (hop) 跳 (jump)

Weak association Unrelated 泣 (cry) 淚 (tears) 肥 待 (fat) (wait) 27

Strong association Weak association Visually-dissimilar 線 (thread) 針 (needle) 季 (season) 年 (year)

Unrelated 斑 改 (spot) (change)

Table 2. The overall accuracy and reaction times for the five experimental conditions in the similar and dissimilar tasks. Age group

Strong

Weak

Unrelated

Perceptual

Baseline

ACC (%)

92 (9)

85 (11)

97 (4)

99 (1)

97 (5)

RT (ms)

890 ( 90)

975 (91)

981 (189)

685 (192)

699 (174)

ACC (%)

84 (8)

81 (12)

94 (4)

98 (3)

99 (2)

RT (ms)

1063 (203)

1156 (213)

1092 (195)

755 (123)

723 (106)

ACC (%)

88 (9)

82 (11)

98 (4)

99 (1)

99 (4)

RT (ms)

959 (194)

972 (192)

959 (220)

631 (125)

621 (157)

ACC (%)

81 (14)

76 (11)

95 (7)

96 (6)

98 (3)

RT (ms)

1129 (291)

1146 ( 250)

1039 (224)

723 (159)

665 (175)

Older Similar Younger

Older Dissimilar Younger

Note. Standard deviations are in parentheses.

Table 3 Brain activation in the visually-similar and visually-dissimilar pairs. Pairs Similar

Contrasts

Regions

H

BA

z-test

voxels

x

y

z

Strong - perceptual

Angular gyrus

L

39

5.28

266

-57

-51

24

Inferior frontal gyrus

L

45

4.84

692

-48

27

6

Inferior frontal gyrus

L

47

-45

27

0

28

Weak - perceptual

Dissimilar Strong - perceptual

Weak - perceptual

Precuneus

L

7

4.04

153

-24

-63

39

Middle temporal gyrus

L

21

3.51

22

-48

-54

-9

Thalamus

L

-

3.81

82

-6

-18

12

Inferior frontal gyrus

L

47

4.76

752

-45

24

-6

Inferior frontal gyrus

L

45

-51

18

15

Angular gyrus

L

39

-57

-57

24

Middle temporal gyrus

L

21

-57

-48

0

Superior frontal gyrus

L

6

4.10

121

-3

12

60

Inferior frontal gyrus

L

47

5.57

4382

-48

21

0

Middle temporal gyrus

L

21

4.45

497

-57

-54

-3

Angular gyrus

L

39

-57

-69

21

Parahippocampus

R

-

3.73

235

18

-48

-6

Cuneus

R

18

2.93

18

36

-93

18

Cuneus

L

18

2.87

27

-21

-90

24

Inferior frontal gyrus

L

47

5.04

1464

-42

27

0

Inferior frontal gyrus

L

45

-51

24

21

Middle temporal gyrus

L

21

4.24

364

-51

-54

3

Superior frontal gyrus

L

8

4.11

635

-6

15

54

Angular gyrus

L

39

4.04

66

-48

-51

27

Inferior frontal gyrus

R

47

3.89

752

33

21

-3

4.69

228

Note. H: hemisphere, L: left, R: right, BA: Brodmann’s Area. Coordinates of activation peak(s) within a region based on a z-test are given in the MNI stereotactic space (x, y, z). Voxels: number of voxels in cluster at p < .05 FDR (false discovery rate) corrected, only clusters greater than or equal to 10 are presented. 29

Table 4 Interaction effects: direct comparisons between older and younger children Age

Pairs

Condition

Similar

Strong

vs.

vs.

Dissimilar

perceptual Weak

Regions

H

BA

z-test

voxels

x

y

z

Angular gyrus

L

39

3.54

20

-57

-51

30

Middle temporal gyrus

L

21

4.71

32

-48

-45

-6

Inferior frontal gyrus

L

47

3.04

15

-51

18

0

Angular gyrus

L

39

3.81

26

-63

-51

30

Middle temporal gyrus

L

21

3.79

13

-45

-45

-9

Combined vs. Older

pairs perceptual

vs. Strong Younger Similar

vs. weak Weak

Dissimilar

vs. strong

Note. See Table 3 note. Voxels: number of voxels in cluster at p < .05 FWE corrected with a sphere of 5 mm radius centered on the commonly significant voxels from the within-group analysis, only clusters greater than or equal to 10 are presented.

Highlights 1.

We examined how the brain constructs both lexical and sublexical level semantic information into an organized system over age. 30

2.

There were greater age-related increases in left angular gyrus for visually-similar compared to the visually-dissimilar task for the strong association pairs.

3.

There were age-related increases in left middle temporal gyrus for the weak compared to strong association pairs in the visually-dissimilar task.

4.

Shared semantics at the sublexical level facilitates the semantic integration at the lexical level in the older children.

31

32