Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Jun 27.
Published in final edited form as: J Cogn Neurosci. 2010 Oct 20;23(9):2280–2290. doi: 10.1162/jocn.2010.21583

Hebrew brain vs. English brain: Language modulates the way it is processed

Atira S Bick 1,2,*, Gadi Goelman 2, Ram Frost 3
PMCID: PMC3694568  NIHMSID: NIHMS477845  PMID: 20961169

Abstract

Is language processing universal? How do the specific properties of each language influence the way it is processed? In this study we compare the neural correlates of morphological processing in Hebrew – a Semitic language with a rich and systematic morphology, to those revealed in English – an Indo-European language with a linear morphology. Using fMRI we show that while in the bilingual brain both languages involve a common neural circuitry in processing morphological structure, this activation is significantly modulated by the different aspects of language. Whereas in Hebrew morphological processing is independent of semantics, in English morphological activation is clearly modulated by semantic overlap. These findings suggest that the processes involved in reading words are not universal, and therefore impose important constraints on current models of visual word recognition.

Introduction

Is language processing universal? Are there qualitative differences in processing different languages? If so, how do the specific properties of the language modulate the way it is processed? These theoretical questions have been the focus of heated debates in neuro-cognitive research. A large corpus of behavioral studies that has focused on cross-linguistic comparisons has not produced so far unequivocal results. In the present study, we attempt to address these issues by considering imaging data of morphological processing during reading.

As a general approximation, current computational models of reading are implicitly universal, and they do not consider themselves to be language specific. Hence, current opinion posits that although languages differ in the statistical distribution of letters or letter-clusters, printed words in English, Spanish, or French, are assumed to be processed and recognized alike. However, the question whether similar principles of lexical processing are shared by two languages becomes intriguing when the two languages come from different language families, so that they do not share linguistic representations.

In this context, comparing morphological processing of printed words in Semitic and indo-European languages has been drawing significant attention. Morphology concerns the internal structure of words. It enables languages to convey similar meanings by preserving a part of the word’s form (e.g., heal-health). Hence, it is reflected by systematic correlations of form (orthography, phonology) and meaning (semantics). Languages, however, differ in their morphological properties and in how they create morphological complexity. In this study, we focus on Hebrew and English, which represent two different types of linguistic systems. Hebrew, like other Semitic languages, has a rich and systematic morphology where most words are morphologically complex and consist of derivations of tri-consonantal root morphemes (see Frost et al., 1997 for a detailed description of Hebrew morphology). Hebrew words are thus formed by intertwining consonant root morphemes and phonological word-pattern morphemes. English morphology, on the other hand, is characterized by a linear and sequential concatenation of prefixes or suffixes to a base morpheme, which forms multi-morphemic words. In contrast to Hebrew, the base morphemes in most cases constitute word-forms in their own right (such as dark in darkness, or dream in dreamer), and their orthographic and phonological integrity remains intact.

Behavioral studies investigating the relation between morphological processing and semantic and orthographic processing, consistently suggest that reading processes in Hebrew and in English differ qualitatively. The main difference seems to be that in English (as in other Indo-European languages), the processing system is tuned to the word’s linear orthographic structure, whereas in Hebrew the processing system automatically searches for the root letters. This is because words in the Hebrew mental lexicon have been shown to be organized by morphology – i.e. by root families rather than by simple letter sequences (Frost et al., 1997; Frost, 2009). Regarding morphological processing, a significant difference between Hebrew and English concerns the interplay of semantic similarity and the perception of morphological relatedness. Several studies have shown that the extent of semantic overlap between two Hebrew words that share the same root (and hence are morphologically related), does not seem to have any effect on the perception of morphological relatedness (Frost et al., 1997; Frost et al., 2000; Deutsch et al., 2003). In contrast, for English readers, the results seem to be much less consistent and do not present an unequivocal pattern. Some masked-priming studies show that semantically transparent (e.g. farmer-farm), semantically opaque (e.g., department-depart), or pseudo-derivations (e.g., irony-iron) are treated alike, suggesting that, like in Hebrew, semantic overlap between derivations does not play any role in word recognition (see Rastle & Davis, 2008 for a review). However, other studies demonstrate that morphological processing is clearly modulated by semantic considerations (Feldman, 2000; Giraudo & Grainger, 2001; Diependaele et al., 2005). These results in English however, intermix different research paradigms and therefore may tap different processing stages of morphological information.

The difference between Hebrew and English is reflected in imaging studies as well: fMRI results in Hebrew consistently reveal a network involved in morphological processing that is not influenced by the semantic properties of the stimuli. This was found for both explicit (Bick et al., 2008) and implicit processing (Bick et al., 2009). In contrast, fMRI results in English, again, do not present a clear and unequivocal picture. Devlin et al. (2004) found no evidence of morphological processing that could not be explained by semantics or orthography. However, both Bozic et al. (2007) and Gold and Rastle (2007) did find regions showing independent morphological processing, but the locations of these regions were inconsistent. Several event-related potential (ERP) studies have investigated the relation between morphological processing and semantics, but have again reported contradicting results: while some found no effect of semantic transparency (Lavric et al., 2007; Koester & Schiller, 2008), others could not clearly report an effect of morphological relatedness for semantically opaque words (Morris et al., 2007). These contradicting results challenge the view that considers orthographic, phonological, and semantic sublinguistic units to be the main building blocks of language (Seidenberg & McClelland, 1989), and questions whether morphology should be regarded as a distinct level of lexical architecture or a simple interplay of form and meaning.

The present study was designed to investigate whether the morphological structure of the language influences the way it is processed in the brain. More specifically, we were concerned with the following question: do Hebrew-English bilinguals use a common neural network when processing morphology in the two languages? If so, does the network display similar neural patterns of activation when processing morphological information for Hebrew and for English material?

Extensive research has addressed the question of the multiple representations of words in the bilingual lexical system. Hence, models such as the Revised Hierarchical Model (RHM, Kroll & Stewart, 1994), or the Bilingual Interactive Activation model (BIA, Dijkstra & Van Heuven, 1998), provided a description of how exactly L1 and L2 conceptual and lexical links constantly interact in the bilingual brain, and how automaticity and inhibitory control develop in proficient bilinguals. In general, most of these studies focused on relatively skilled adult bilinguals, showing that successful L2 learning consists of developing new linguistic representations, and that this process requires cognitive support of executive functions (see Kroll & Tokowicz, in press, for an extensive review). Imaging studies have demonstrated the relation between the neural representation of subject’s native language (L1) and languages learned later in life (L2) (e.g., Mahendra et al., 2003; Tan et al., 2003; Chee et al., 2004; Chee, 2006) defining the neural correlates involved in the control of language switching (Rodriguez-Fornells et al., 2002; Abutalebi et al., 2008; Hernandez, 2009). Beyond understanding bilingualism itself, presenting bilinguals with linguistic stimuli in the two languages is used to allow a well balanced comparison between the processing of the two languages. Such was done recently by Lehtonen et al’s who compared processing of inflectional morphology in Finnish-Swedish bilinguals (Lehtonen et al., 2009). They found a fronto-temporal increase in activation when nouns were morphologically complex relative to monomorphemic only in Finnish but not in Swedish. As subject and environment conditions were identical in both languages, the plausible interpretation would be that these differences reflect different cognitive processes involved in processing inflectional morphology in the two languages.

In this study bilingual subjects, highly proficient in both languages from a young age preformed a lexical decision task in both languages. Target words were preceded by primes presented briefly and sub-consciously. Morphological, semantic, and orthographic manipulations were introduced by changing the relations between the target and its primes. Two morphological conditions were used – in one, words were morphologically related but not semantically related, and in the other, the words were both semantically and morphologically related. We chose to use the masked priming paradigm (Forster & Davis, 1984) as it has been extensively used in morphological research, specifically in fMRI studies of morphology (Gold & Rastle, 2007; Bick et al., 2009). Additionally, this method reflects automatic rather than strategic processes, since subjects are unaware of the linguistic manipulation. This allowed us to identify brain regions that were sensitive to morphological relations between words and compare this effect to that caused when words were related to targets only in orthographic form, or in semantic meaning.

Our goal was to compare the activation pattern during processing morphology in Semitic languages and in Indo-European languages, using Hebrew and English. We were interested in identifying the networks involved in morphological processing, to test if both languages use the same network, and to quantify the differences in language processing in the two languages. Specifically, we investigated whether the interaction between processing morphological, semantic and orthographic information is similar in both languages. This would indicate whether each language’s morphological structure influences the way it is processed, or whether printed information is processed alike in alphabetic orthographies, regardless of significant morphological differences.

Methods

Subjects

Twenty seven volunteers participated in this study (twelve male, ages ranging from 22–36, mean age 26.33). All participants were right-handed without any neurological record or reading disorders. All subjects were early bilinguals, exposed to both English and Hebrew before they were eight years old, and reported being highly proficient in both languages at testing. Participants gave written consent before taking part in the study and were paid for participating. Ethical approval was granted by the Hadassah Hebrew University Medical Center.

Magnetic Resonance Protocol

The blood oxygenation level dependent (BOLD) fMRI measurements were performed in a whole-body 3-T, Siemens scanner. BOLD contrast was obtained with a gradient-echo echo-planar imaging sequence and a standard head coil. Functional data was obtained using TR = 2 s, TE = 30 ms, flip angle = 90°, imaging matrix = 80 × 80, FOV = 22×22 cm (inplane resolution = 2.75×2.75 mm) and 30 slices, 3 mm each with 0.5 mm gap between slices. Slices were placed oblique to cover most of the brain. High resolution three-dimensional spoiled gradient echo (SPGR) anatomical sequence was preformed in the same session as functional data.

Stimuli

256 target words were used in the experiment (128 in Hebrew and 128 in English). Each target word was matched with a prime from one of the four conditions: (1) The prime was the morphological base from which the word was derived. In this condition primes were morphologically and semantically related to the targets (MS). (2) The prime was the morphological base from which the word was derived or pseudo-derived and was not semantically related to target. (M) (3) A prime orthographically similar to the target. In Hebrew, this was a three-letter word included in the target, which was not the root of the word. In English, it was a real word, identical to the beginning of the target word, but not followed by a combination of letters used as a morphological suffix in English (O). A prime that was semantically related to the target, but not derived from the same root or base (S). Examples of the stimuli employed in the experiment are presented in Table 1. In each language conditions were balanced as to word length, frequency and orthographic neighborhood density (no significant differences were found).

Table 1.

examples of Hebrew and English stimuli

Hebrew Orth. Trans. Phonetic Trans. Semantic meaning length frequency Semantic relation
Semantic (S) graphic file with name nihms477845t1.jpg cxwk tzxok laughter 4.88 12.9 6.31
graphic file with name nihms477845t2.jpg smx same’ax happy 3 26.47
Morphologic +semantic (MS) graphic file with name nihms477845t3.jpg gyswr gishur bridging 5.09 16.6 6.52
graphic file with name nihms477845t4.jpg gsr gesher bridge 3 44.25
Morphologic (M) graphic file with name nihms477845t5.jpg hxlmh haxlama recovery 4.97 9.41 1.55
graphic file with name nihms477845t6.jpg xlm xalam dreamed 3 34
Orthographic (O) graphic file with name nihms477845t7.jpg mzlg mazleg fork 4.94 14.1 1.22
graphic file with name nihms477845t8.jpg mzl mazal luck 3 33.81
English length frequency Semantic relation
Semantic (S) demand 7.03 34.87 5.91
ask 4.09 113.8
Morphologic +semantic (MS) guilty 6.78 21.39 6.1
guilt 4.31 78.6
Morphologic (M) department 6.91 51.2 1.62
depart 4.22 52.33
Orthographic (O) start 6.78 39.96 1.67
star 4 76.73
Open in a new tab

Since we were constrained to a reasonable length for an imaging experiment, we could not achieve a probability of 0.5 for a “yes” response in the lexical decision task; therefore, we employed 64 (not 128) target nonwords for each language in this experiment. As we were interested in the difference in activation in the different priming conditions for each language, this did not constitute a problem. Half of the nonwords were primed by a theoretically possible morphological root/stem and half were primed by a non-related prime.

Semantic relatedness was assessed by averaging subjective ratings. Fifty native speakers of each language judged the relatedness of each prime-target pair on a 1 (not related) to 7 (strongly related) scale. All pairs considered “semantically related” were ranked 5 and above, and all pairs considered “semantically unrelated” were ranked below 2.5. Mean ranks of semantic relatedness for each condition are presented in Table 1. As it is impossible to find targets with both a semantically related root/stem and a semantically unrelated stem, different targets appeared in the different conditions.

Experimental Setup

Stimulus presentation and recording of responses were implemented with Presentation software (http://www.neurobs.com/presentation). All words were visually presented via an LCD projector onto a tangent screen located inside the scanner, in front of the subject. Subjects viewed the screen through a tilted mirror. Behavioral performance was assessed during the fMRI scan using a response box.

Experimental Design

Subjects were asked if a visually presented letter string was an existing word in Hebrew/English and their response was monitored. The experiment consisted of eight runs (6:30 min each), containing one Hebrew and one English block and included four repetitions of the six conditions (4 of words and 2 of non-words, in each language) per run. The order of the runs was balanced between subjects.

Blocks were preceded by a notice of the upcoming language, and by two practice trials, which were removed from the analyses. Conforming with the forward-masking paradigm (Forster & Davis, 1984), trials consisted of a sequence of three events: a 500 ms forward mask (########), followed by a prime presented for 33 ms, replaced by the target which remained on the screen for 2000 ms. Participants were told that a series of hash marks would precede targets but no mention was made of the primes. The different trial types (6 conditions) were presented in pseudorandom order, with a fixation cross (+) presented during a variable inter-trial interval to enable fMRI jittering in an event-related design. The inter-trial interval range and pseudo-random ordering were customized for the present design using the optseq2 program (http://surfer.nmr.mgh.harvard.edu/optseq/), in order to achieve optimal experimental efficiency (Dale, 1999).

Data Analysis

A behavioral log file was used to identify trials with null or incorrect responses. Errors and responses longer than 2500 ms were discarded from the behavioral or imaging data analysis. Similarly, trials in which prime presentation included timing errors beyond 5 ms were excluded from further analysis as well.

Error rate (including both errors and null responses) was calculated for each subject. Within each subject, reaction times that were outside a range of 2 standard deviations were replaced with the appropriate cutoff value. This standard procedure was used to minimize the effect of outliers. Corrected reaction times were used for ANOVA.

Imaging data analysis was performed using the BrainVoyager Qx software package (Brain Innovation, Maastricht, The Netherlands, 2000). Prior to the statistical analysis, the raw data were examined for motion and signal artifacts. Head motion correction and high-pass temporal filtering in the frequency domain (3 cycles/total scan time) were applied in order to remove drifts and to improve the signal-to-noise ratio. Significant motion was found maximum movement of 6 pixels), but as the motion was not time locked to experimental conditions no significant motion related artifact was identified and no cut off was used to discard this data. The complete data set was transformed into Talairach space (Talairach & Tournoux, 1988), Z-normalized, and concatenated.

Changes in BOLD contrast associated with the different conditions were assessed on a pixel-by-pixel basis, using the general linear model (Friston et al., 1995) with the standard hemodynamic response function (Boynton et al., 1996). Group analyses were performed using random-effect analysis. In order to correct for multiple comparisons, cluster-size threshold, based on Monte Carlo simulation (p<0.05) was used. Morphologically related areas in the different languages were defined to be the areas showing repetition suppression, that is, a reduction in activation in the two morphological conditions relative to the orthographic and semantic condition, and additionally showing an increase in signal during the orthographic and semantic conditions relative to the resting condition (([O+S]>[MS+M])&([O+S]>0)). This was done in order to eliminate noise and include only areas involved in processing the different aspects of language.

Functional images were incorporated into the three-dimensional data sets through trilinear interpolation. The statistical parametric maps were overlaid on a cortical inflated map of a representative subject. The inflated maps were reconstructed from the T1-weighted 3D images. The procedure included segmentation of the white matter, using a grow-region function, the smooth covering of a sphere around the segmented region, and the expansion of the reconstructed white matter into the gray matter.

Functional regions of interest (ROIs) corresponding to morphological conditions were defined in the lIFG and lMFG for each subject individually, using the same contrast used for defining morphological related areas in group analysis (p<0.05, not corrected). This contrast allowed us to investigate the influence of semantic transparency on morphological processing in regions involved in morphological processing. Note that the definition of the ROIs is independent of the contrast between the morphological conditions. Activated voxels, located within 25mm of the multi-subject activity center, were defined as morphological ROIs. Such ROIs were found in 24 subjects in the MFG and in 23 subjects in the IFG. A similar method was used when processing data from each language separately: we contrasted the two morphological conditions with the other two language conditions, in conjunction with an increase in activation in the orthographic and semantic conditions in each language. In this analysis we included also the occipital regions in the left hemisphere to allow a fuller investigation of the regions in the left hemisphere involved in morphological processing in both languages. Activated voxels located within 25mm of that language’s multi-subject activity center were defined as ROIs. Such ROIs were found for Hebrew in 21 subjects in the MFG and in 23 subjects in the IFG, and for English in 25 subjects in the MFG and in 23 subjects in the IFG. In these ROIs, hemodynamic responses associated with the different conditions were estimated using deconvolution analysis (Glover, 1999). From these hemodynamic responses, averaged bar histograms for the different tasks were calculated (t=2–12 sec) and subjected to statistical analysis using paired two tailed t-tests and paired ANOVA.

Results

Behavioral Results

The mean accuracy and reaction time (RT) are shown in Table 2. Overall, subjects displayed adequate performance (average error rate for all priming conditions: 3.37±4.64%). Further statistical analysis was done on reaction times for correct trials only. ANOVA of RTs identified a main effect of language (F(1,26) = 28.8, p<0.001) and morphology (F(1,26) = 28.6, p<0.001). No significant effect of semantics (F(1,26) <1.0) was found nor any interaction between language and semantics (F(1,26) < 1.0) or language and morphology (F(1,26) <1.0), although the interaction between morphology and semantics was nearly significant (F(1,26) = 4.01, p < 0.056). The faster RTs for words in Hebrew probably reflect some proficiency differences between languages: most subjects defined themselves as Hebrew dominant.

Table 2.

Behavioral results, mean response latencies and standard errors.

Hebrew English
S MS M O S MS M O
% error 2.6±3.1 1.06±1.5 1.39±2.5 1.62±2.5 2.35±3 1.79±2.3 5.66±5.5 6.94±5.6
Reaction time (ms) 745±23 715±19 719±22 740±20 858±30 812±26 827±30 847±31
Open in a new tab

Imaging Results

Maps were calculated to identify voxels involved in unique morphological processing in both Hebrew and English. As priming usually creates repetition suppression we expected a reduction in signal in the relative areas. Areas involved in morphological processing in each language were defined as those areas where activation in both morphological conditions was reduced relative to the semantic and orthographic conditions, and activation in the semantic and orthographic conditions was higher than the resting condition (p<0.001). In both languages, the left inferior frontal gyrus (lIFG), the left middle frontal gyrus (lMFG), and regions in the inferior temporal lobe and occipital lobe were significantly modulated by morphological priming (see Figure 1 and Table 3). As can be seen in Figure 1, the networks involved in morphological processing in Hebrew and in English highly overlap.

Figure 1.

Figure 1

Open in a new tab

Regions involved in Morphological processing in Hebrew (blue) and in English (red). Regions of overlap are marked in purple.

Table 3.

Regions showing a morphologically related repitition suppresion

Hebrew BA Mean x Mean y Mean z Z at peak voxel Cluster size (mm)
Left Occipital-temporal 18, 19, 37 −33 −80 −6 5.57 1835
Left Calcarine Sulcus 17, 18 −17 −92 −6 4.67 750
Left MFG 6, 9 −35 7 29 5.52 431
Left IFG 13 −27 12 11 5.34 1646
Left Precentral Gyrus, Left MFG 24, 6 −21 −15 45 5.18 444
Left Medial Frontal Gyrus 6, −8 3 53 5.03 560
Right Cingulate Gyrus, 24, 32 14 9 39 4.79 1151
Right IFG, Right Insula 13 33 22 13 5.59 1023
Right Occipital-Temporal regions 17, 18, 19 30 −84 0 6.05 5705
Right Cingulate 23 17 −29 31 4.88 567
Right Insula 21 2 7 4.83 368
White matter 28 −6 30 5.44 561
−24 −18 30 5.45 423
ENGLISH BA Mean x Mean y Mean z Z at peak voxel Cluster size (mm)
Left MFG, Left IFG 44, 9, 13, 6 −43 8 18 5.99 765
Left IFG 13 −27 22 11 5.17 914
Left Inferior Parietal Lobe 2, 40 −45 −28 39 5.43 318
Left Visual regions 18, 19 −30 −82 2 5.98 2046
Left Fusiform Gyrus 20, 37 −31 −43 −17 5.7 793
Left MFG 6 −22 −10 49 5.53 398
Left Cingulate Gyrus, Left Medial Frontal Gyrus 24, 32, 6 −4 3 51 5.81 2009
Right Visual regions 18, 19, 28 −76 3 5.4 3607
Right Insula, Right IFG 13 29 21 10 5 621
Right Cingulate Gyrus, 24, 32 9 15 31 5.48 482
Right Fusiform Gyrus, 37, 40 −50 −10 4.91 497
Open in a new tab

In order to directly compare the activation between conditions in the different languages, we defined ROIs in the regions we previously found to be involved in morphological processing in Hebrew (Bick et al., 2009) – in the lIFG and lMFG by identifying a common morphological network for both languages (see methods). In all these regions, activation was reduced when targets were primed by morphologically related words compared to the orthographic and semantic conditions. This effect was found in both Hebrew and English for both semantically related (two tailed paired contrast, IFG: Hebrew p < 0.001; English p < 0.001; MFG: Hebrew p < 0.001; English p < 0.001) and semantically unrelated pairs (IFG: Hebrew p < 0.001; English p < 0.001; MFG: Hebrew p<0.001; English p< 0.001). It is important to notice that in these ROIs, no significant differences were found between the semantic and orthographic conditions in Hebrew or in English, (two tailed paired t-test, IFG: Hebrew t(22) = 0.81 p < 0.42; English t(22) = 0.24 p < 0.82; MFG: Hebrew t(23) = 1.69 p < 0.1; English t(23) = 1.16 p < 0.87), ruling out the possibility that these regions are specifically involved in semantic or in orthographic processing.

We now come to the main question of our investigation, that is, is morphological processing in the two languages similar? Significant differences were found between Hebrew and English in the modulation of semantics in morphological processing. While in Hebrew, in both areas, there was no significant difference between the semantically transparent (MS) and semantically opaque (M) morphological conditions (two tailed paired t-test, IFG: t(22) = 0.98 p < 0.33; MFG: t(23) = 0.97 p < 0.34), in English, activation in the two morphological conditions differed significantly, with a significant decrease of activation when prime and targets were both morphologically and semantically related (IFG: t(22) = 4.5 p<0.005, MFG: t(23) = 2.54 p<0.05). This differential effect of repetition suppression in the two languages was significant (IFG: F(1,22) = 6.046 p < 0.022; MFG: F (1,23) = 6.066 p <0.022).

To further examine this cross-linguistic difference and avoid possible confounds between the two languages, we repeated the analysis using ROIs defined for each language individually. The procedure was similar to that used for both languages together, and this time we included the regions significantly activated in the occipital region as well. Results in frontal regions showed the same pattern as before – in both languages, in all relevant areas, a significant morphological effect was found regardless of semantic transparency (p<0.001). However, similar to the findings obtained when ROIs were defined for both languages together, a significant modulation by semantics was found in English but not in Hebrew: i.e., while in Hebrew the morphological conditions did not differ (IFG: t(22) = 0.53, p < 0.6; MFG: t(20) = 0.88 p < 0.39), in English semantic overlap between morphologically related primes and targets resulted in repetition suppression, and activation was significantly reduced relative to when primes and targets were not semantically related (IFG: t(24) = 3.13 p<0.005; MFG: t(19) = 2.34. p<0.05).

Occipital regions in both languages included two ROIs – one occipital region and one occipito-temporal (see table 3). In the occipital regions involved in early visual processing, the pattern of activation was similar to that found in the frontal regions. In Hebrew, a weak morphological effect was found in the semantically opaque condition relative to the semantic condition (p<0.03) and to the orthographic condition (p<0.01). When words were semantically transparent, there was a significant reduction in signal relative to the orthographic condition (p<0.02), but not relative to the semantic condition (t(18) = 1.72, p = 0.1). Similar to the frontal regions, no difference was found between the morphological conditions (t(18) = 1.19, p = 0.26). The similarity between the occipital and frontal regions is also apparent in English. Reduction was significant for both morphological conditions (p<0.001), and a significant difference was found between them (t(16) = 2.16, p<0.05). As these regions are involved in processing low level visual information we assume this effect reflects top-down modulation caused by attention and does not reflect morphological processing.

In the occipital-temporal (OT) region the pattern is more complex. In Hebrew the results are similar to those found in the frontal lobe; activation in each of the morphologically related conditions was reduced relative to the other conditions (p<0.005), and no significant difference was found between the morphological conditions (t(16) = 0.62 p = 0.54). In English, on the other hand, similarly to the frontal lobe there was a significant morphological effect for each of the morphological conditions (p<0.001), but unlike the frontal lobe no significant difference between morphological conditions could be found (t(24) = 0.86 p = 0.39).

This difference between Hebrew and English is clearly seen in figure 3: in this figure we indicate those areas that show a morphological effect (obtained by contrasting both morphological conditions with the semantic and orthographic conditions) together with a difference between the two morphological conditions. As seen in the figure, large areas, similar in location to those identified previously to be involved in morphological processing (p<0.05, cluster size corrected to larger than 756mm), were obtained only for processing in English. In Hebrew, on the other hand, no significant activation was found, suggesting again that morphological activation was not modulated by semantic transparency (p<0.05 cluster size corrected. At lenient cluster of 108mm, small clusters were found in the OT).

Figure 3.

Figure 3

Open in a new tab

Areas showing both a morphological priming effect and a significant difference between the two morphological conditions in English. In Hebrew this contrast resulted in no significant cluster.

One issue to consider is the level of English proficiency of our participants. This proficiency was assessed through self-report, and our behavioral results demonstrate that although our participants were all highly proficient in English, they were nevertheless on the average Hebrew dominant. The question at hand is whether language dominance has influenced our results. Our sample included 12 participants who were English dominant or fully balanced bilinguals. We therefore examined the results in the ROIs in this subgroup. Our analyses revealed an identical pattern as the full sample of participants: In both IFG and MFG a significant reduction in activation was found for both morphological conditions (IFG: p<0.002; MFG: p<0.01). However, while in Hebrew no difference was found between the two morphological conditions (IFG: t(9) = 1.89 p=0.09, MFG: t(10) = 0.6, p=0.56), in English a significant difference was found (IFG, t(9) = 2.44, p<0.05; MFG: t(10) = 2.4, P<0.05). These findings suggest that language dominance was did not effect our findings.

Discussion

In this paper we show how the structural properties of language shape the processing of morphological information in the brain. Focusing on morphological priming, we demonstrate that while the same network is involved in processing morphology in languages such as English and Hebrew, clear cross-linguistic differences emerge in how semantic information modulates morphologically related repetition suppression in the two languages. These findings are in line with an ecological view of language processing (Frost, 2009), and impose important constraints on all current models of visual word recognition. We will now discuss our main findings and their implications for current neurobiological models of language.

First, in contrast to the findings reported by Devlin et al. (Devlin et al., 2004), the present imaging results clearly demonstrates the independent role of morphological processing. This was shown in both Hebrew and in English, two languages that have very different morphological systems. Moreover, our findings suggest that the same network is involved in processing morphological information in Semitic and in Indo-European languages: in both languages the same areas showed a significant decrease in activation in the morphological conditions, when contrasted with the semantic and the orthographic conditions.

Thus, three main areas seem to be implicated in processing morphological information: the lIFG, and the lMFG and lOT. Our present results in the lIFG converge with our previous fMRI results in Hebrew (Bick et al., 2009), and also coincide with imaging results in English. For example, Bozic et al. found morphological related activation within the inferior frontal lobe, independent of semantics or orthography (Bozic et al., 2007). The IFG has been found to be connected to morphological processing in various languages such as English (Longe et al., 2007), German (Longoni et al., 2005), Finnish (Lehtonen et al., 2006), and Spanish (Hernandez et al., 2004). These studies compared simple vs. complex words (Laine et al., 1999), the processing of inflected verbs and nouns (Tyler et al., 2004), and the processing of regular vs. irregular verbs (Beretta et al., 2003) as well as gender judgments (Heim et al., 2005). The lMFG was shown to be involved in morphological processing in Hebrew (Bick et al., 2008; Bick et al., 2009), and in morpho-syntactic processing in English (Shapiro et al., 2005; Tyler et al., 2005), Italian (Marangolo et al., 2006) and artificial syntax (Forkstam et al., 2006). For a detailed discussion of the role of these areas in language processing in general and specifically in morphological processing see (Bick et al., 2008; Bick et al., 2009).

The regions in the left OT showed a similar repetition suppression effect for both morphological conditions on both languages. The OT is known to be involved in form processing (Cohen et al., 2000; Dehaene et al., 2001; Cohen et al., 2002; Dehaene et al., 2002). Specifically, this region was activated by explicit morphological processing (Bick et al., 2008) but activation could not be differentiated from activation during orthographic or phonological processing. Similar findings were reported by Devlin et al. (2004), and in a revised study Gold et al (Gold & Rastle, 2007) found in the OT a region showing morphologically repetition suppression that could not be explained but semantic or orthographic factors. Our results showing a morphological effect that is not influenced by semantics strengthen Gold’s conclusions that this region is involved in the structural level of morphological processing that is mainly morph-orthographic (see Rastle & Davis, 2008 for a detailed review). At this level, in both languages morphological processing is blind to semantic consideration.

Cross-linguistic differences in the brain

While the same areas seem to be implicated in morphological processing in the two languages, clear cross-linguistic differences emerge in what determines activation in these areas. Our main interest centered on how semantic information modulates morphological activation. Our results seem straightforward: while in Hebrew, morphological processing was unaffected by the semantic properties of the stimulus, in English, modulation of morphological activation by semantic processing was observed. As no semantic adaptation was found in these areas, this effect seems to result from morphological processing per-se. Although our experimental design did not assess directly semantic priming effects (an unrelated condition was not included), the semantic modulation of early morphological effects in English is in accordance with ERP studies demonstrating involvement of semantics at early stages of reading, commonly associated with orthographic and morphological encoding of English words (Penolazzi et al., 2007). They are also consistent with behavioral results showing that early morphological processing in English is morpho-semantic (e.g., Feldman et al., 2009). However, they seem inconsistent with the common view supported by a large number of studies suggesting that morphological analysis of English words is initially morpho-orthographic and hence blind to semantic considerations (see Rastle & Davis, 2008 for a review). Interestingly, event-related potential (ERP) studies also show contradictory results regarding the dependence of the morphological effect on semantics: while some found no effect of semantic transparency (Lavric et al., 2007; Koester & Schiller, 2008), others could not clearly report an effect for semantically opaque morphologically related words (Morris et al., 2007).

In contrast to English, in Hebrew, semantics and morphology are clearly orthogonal factors. Behavioral experiments using masked priming consistently reveal that morphological effects are not modulated by the semantic relation between primes and targets (Frost et al., 1997; Frost et al., 2000). These findings have been reflected in parallel imaging studies independent processing system (Bick et al., 2008; Bick et al., 2009). The results of the present study clearly demonstrate that in contrast to English, the neural circuitry involved in processing morphological information in Hebrew simply does not consider the semantic overlap between primes and targets if they are morphological derivations.

The cross-linguistic differences in brain activation revealed in Hebrew vs. English stem from the structural properties of the two languages. The structure of Semitic languages such as Hebrew is based on embedding roots into phonological word patterns so that almost all words are morphologically complex. Since the Hebrew lexical system is organized by root morphemes, the reader (and speaker) of Hebrew considers root information as a primary step in word recognition. The process of extracting root consonants from print or speech is a structural process blind to semantic considerations. Indeed, awareness to the root morpheme and the ability to manipulate it can be found in children from a very early age (Berman, 1982; Ravid & Malenky, 2001). In English, on the other hand, stems are words on their own, so that most words are not morphologically complex. Morphological complexity is created by appending prefixes and suffixes to base words, however, prefixes and suffixes are represented by letter clusters that often do not represent morphemes (ER in darker vs. ER in corner, or IM in impatient vs. IM in impress). Since there are no a-priory rules that systematically predict when a given letter cluster is a morphemic unit, morphological relatedness is in fact determined and hence modulated by semantic considerations.

This is the essence of the ecological view of language processing (Frost, 2009). Our imaging results indeed demonstrate that the specific linguistic properties of a given language modulate the way it is read and processed. Consequently, the neural circuitry involved in identifying words is designed to optimally process the specific properties of the linguistic environment of the reader and speaker. Hence, although some aspects of language are universal, the specific characteristics of each language must be considered in neurocognitive models of language processing. This ecological view has important implications for adapting methods of linguistic remediation or reading instruction methods from one linguistic system to another.

Most subjects participating in the experiment were bilinguals exposed to both languages at an early age. The similarity between our results for Hebrew-English bilinguals and our previous results with Hebrew speakers (Bick et al., 2009) demonstrate that even when two languages, such as Hebrew and English, are acquired in early childhood, the processing system develops for each language, given its linguistic structure. However, since results regarding English monolinguals are inconsistent, it is unclear whether our findings in English obtained with our bilingual subjects were influenced by the dominant role of morphology in their other language – Hebrew. The question of the possible interactions between the two languages in the bilingual brain has been attracting much attention in recent years (for some examples Chee, 2009), and additional research is necessary in order to address this question. It is important to note that if such interaction does exist, it may be that the differences between morphological processing in Semitic languages and Indo-European languages are even greater than those revealed in this study.

In conclusion, our findings support an ecological theory of language processing. In both languages we found morphological processing that is independent of orthography and semantics, and identical brain circuitry is involved in processing morphological information. However, the structural properties of the language modulate the way it is eventually processed. Further fine-grained research is necessary to account for the possible influence of one linguistic system on the other in the bilingual brain, to gain better understanding of how more than one language is processed in the brain.

Figure 2.

Figure 2

Open in a new tab

Activation areas showing a morphological effect in both languages. ROIs were defined as regions showing an increase in activation in the orthographic and semantic conditions, and in addition a decrease in activation during the morphological conditions relative to orthographic and semantic conditions. For each ROI, the average beta weights 2–12 sec after presentation of stimuli was calculated for the different conditions (morphologically and semantically related prime – light pink, morphologically and semantically unrelated prime – dark pink, semantically related prime - green, orthographically related prime – blue. Hebrew is solid blocks, English in the squared blocks). Activation in the following regions in the left hemisphere is presented: a: IFG b: MFG.

Table 4.

regions where morphologically related repetition suppression in Hebrew and English overlapped

Hebrew-English Overlap Mean x Mean y Mean z size (mm)
Left MFG −37 12 25 47
Left IFG −30 20 15 179
Left visual regions −31 −82 −3 578
Right IFG 30 23 14 185
Left SMA −7 3 53 183
Right visual regions 32 −82 1 1057
Open in a new tab

Acknowledgments

This work was supported by the Center for Complexity Science, Israel, the Israel Science Foundation (grant award #1334/08), and by the National Institute of Child Health and Human Development (Grant HD-01994 awarded to Haskins Laboratories). We are indebted to Ken Pugh, Einar Mencl and Steve Frost for their contribution in conducting this research, and to Leon Deouel for his insightful comments.

References

  1. Abutalebi J, Annoni JM, Zimine I, Pegna AJ, Seghier ML, Lee-Jahnke H, Lazeyras F, Cappa SF, Khateb A. Language control and lexical competition in bilinguals: an event-related FMRI study. Cereb Cortex. 2008;18:1496–1505. doi: 10.1093/cercor/bhm182. [DOI] [PubMed] [Google Scholar]
  2. Beretta A, Campbell C, Carr TH, Huang J, Schmitt LM, Christianson K, Cao Y. An ER-fMRI investigation of morphological inflection in German reveals that the brain makes a distinction between regular and irregular forms. Brain Lang. 2003;85:67–92. doi: 10.1016/s0093-934x(02)00560-6. [DOI] [PubMed] [Google Scholar]
  3. Berman RA. Verb-pattern alternation: the interface of morphology, syntax, and semantics in Hebrew child language. J Child Lang. 1982;9:169–191. doi: 10.1017/s030500090000369x. [DOI] [PubMed] [Google Scholar]
  4. Bick A, Goelman G, Frost R. Neural correlates of morphological processes in Hebrew. J Cogn Neurosci. 2008;20:406–420. doi: 10.1162/jocn.2008.20028. [DOI] [PubMed] [Google Scholar]
  5. Bick AS, Frost R, Goelman G. Imaging Implicit Morphological Processing: Evidence from Hebrew. J Cogn Neurosci. 2009 doi: 10.1162/jocn.2009.21357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Boynton GM, Engel SA, Glover GH, Heeger DJ. Linear systems analysis of functional magnetic resonance imaging in human V1. J Neurosci. 1996;16:4207–4221. doi: 10.1523/JNEUROSCI.16-13-04207.1996. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bozic M, Marslen-Wilson WD, Stamatakis EA, Davis MH, Tyler LK. Differentiating morphology, form, and meaning: neural correlates of morphological complexity. J Cogn Neurosci. 2007;19:1464–1475. doi: 10.1162/jocn.2007.19.9.1464. [DOI] [PubMed] [Google Scholar]
  8. Chee MW. Dissociating language and word meaning in the bilingual brain. Trends Cogn Sci. 2006;10:527–529. doi: 10.1016/j.tics.2006.09.009. [DOI] [PubMed] [Google Scholar]
  9. Chee MW. fMR-adaptation and the bilingual brain. Brain Lang. 2009;109:75–79. doi: 10.1016/j.bandl.2008.06.004. [DOI] [PubMed] [Google Scholar]
  10. Chee MW, Soon CS, Lee HL, Pallier C. Left insula activation: a marker for language attainment in bilinguals. Proc Natl Acad Sci U S A. 2004;101:15265–15270. doi: 10.1073/pnas.0403703101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cohen L, Dehaene S, Naccache L, Lehericy S, Dehaene-Lambertz G, Henaff MA, Michel F. The visual word form area: spatial and temporal characterization of an initial stage of reading in normal subjects and posterior split-brain patients. Brain. 2000;123(Pt 2):291–307. doi: 10.1093/brain/123.2.291. [DOI] [PubMed] [Google Scholar]
  12. Cohen L, Lehericy S, Chochon F, Lemer C, Rivaud S, Dehaene S. Language-specific tuning of visual cortex? Functional properties of the Visual Word Form Area. Brain. 2002;125:1054–1069. doi: 10.1093/brain/awf094. [DOI] [PubMed] [Google Scholar]
  13. Dale AM. Optimal experimental design for event-related fMRI. Hum Brain Mapp. 1999;8:109–114. doi: 10.1002/(SICI)1097-0193(1999)8:2/3&#x0003c;109::AID-HBM7&#x0003e;3.0.CO;2-W. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dehaene S, Le Clec HG, Poline JB, Le Bihan D, Cohen L. The visual word form area: a prelexical representation of visual words in the fusiform gyrus. Neuroreport. 2002;13:321–325. doi: 10.1097/00001756-200203040-00015. [DOI] [PubMed] [Google Scholar]
  15. Dehaene S, Naccache L, Cohen L, Bihan DL, Mangin JF, Poline JB, Riviere D. Cerebral mechanisms of word masking and unconscious repetition priming. Nat Neurosci. 2001;4:752–758. doi: 10.1038/89551. [DOI] [PubMed] [Google Scholar]
  16. Deutsch A, Frost R, Pelleg S, Pollatsek A, Rayner K. Early morphological effects in reading: evidence from parafoveal preview benefit in Hebrew. Psychon Bull Rev. 2003;10:415–422. doi: 10.3758/bf03196500. [DOI] [PubMed] [Google Scholar]
  17. Devlin JT, Jamison HL, Matthews PM, Gonnerman LM. Morphology and the internal structure of words. Proc Natl Acad Sci U S A. 2004;101:14984–14988. doi: 10.1073/pnas.0403766101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Diependaele K, Sandra D, Grainger J. Masked cross-modal morphological priming: Unravelling morpho-orthographic and morpho-semantic influences in early word recognition. Language and Cognitive Processes. 2005;20:75–114. [Google Scholar]
  19. Dijkstra AFJ, Van Heuven WJB. The BIA-model and bilingual word recognition. In: Grainger J, Jacobs AM, editors. Localist connectionist approaches to human cognition. Mahwah, NJ: Erlbaum; 1998. pp. 189–225. [Google Scholar]
  20. Feldman LB. Are morphological effects distinguishable from the effects of shared meaning and shared form? J Exp Psychol Learn Mem Cogn. 2000;26:1431–1444. doi: 10.1037//0278-7393.26.6.1431. [DOI] [PubMed] [Google Scholar]
  21. Feldman LB, O’Connor PA, Del Prado Martin FM. Early morphological processing is morphosemantic and not simply morpho-orthographic: a violation of form-then-meaning accounts of word recognition. Psychon Bull Rev. 2009;16:684–691. doi: 10.3758/PBR.16.4.684. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Forkstam C, Hagoort P, Fernandez G, Ingvar M, Petersson KM. Neural correlates of artificial syntactic structure classification. Neuroimage. 2006;32:956–967. doi: 10.1016/j.neuroimage.2006.03.057. [DOI] [PubMed] [Google Scholar]
  23. Forster KI, Davis C. Repetition priming and frequency attenuation in lexical access. Journal of Experimental Psychology: Learning, Memory, and Cognition. 1984;10:680–698. [Google Scholar]
  24. Friston KJ, Frith CD, Turner R, Frackowiak RS. Characterizing evoked hemodynamics with fMRI. Neuroimage. 1995;2:157–165. doi: 10.1006/nimg.1995.1018. [DOI] [PubMed] [Google Scholar]
  25. Frost R. Reading in Hebrew vs. Reading in English: Is there a qualitative difference? In: Pugh KMP, editor. How Children Learn To Read: Current Issues and New Directions in the Integration of Cognition, Neurobiology and Genetics of Reading and Dyslexia Research and Practice. New York: Psychology Press; 2009. pp. 235–253. [Google Scholar]
  26. Frost R, Deutsch A, Gilboa O, Tannenbaum M, Marslen-Wilson W. Morphological priming: dissociation of phonological, semantic, and morphological factors. Mem Cognit. 2000;28:1277–1288. doi: 10.3758/bf03211828. [DOI] [PubMed] [Google Scholar]
  27. Frost R, Forster KI, Deutsch A. What can we learn from the morphology of Hebrew? A masked-priming investigation of morphological representation. J Exp Psychol Learn Mem Cogn. 1997;23:829–856. doi: 10.1037//0278-7393.23.4.829. [DOI] [PubMed] [Google Scholar]
  28. Giraudo H, Grainger J. Priming complex words: evidence for supralexical representation of morphology. Psychon Bull Rev. 2001;8:127–131. doi: 10.3758/bf03196148. [DOI] [PubMed] [Google Scholar]
  29. Glover GH. Deconvolution of impulse response in event-related BOLD fMRI. Neuroimage. 1999;9:416–429. doi: 10.1006/nimg.1998.0419. [DOI] [PubMed] [Google Scholar]
  30. Gold BT, Rastle K. Neural correlates of morphological decomposition during visual word recognition. J Cogn Neurosci. 2007;19:1983–1993. doi: 10.1162/jocn.2007.19.12.1983. [DOI] [PubMed] [Google Scholar]
  31. Heim S, Alter K, Friederici AD. A dual-route account for access to grammatical gender: evidence from functional MRI. Anat Embryol (Berl) 2005;210:473–483. doi: 10.1007/s00429-005-0032-6. [DOI] [PubMed] [Google Scholar]
  32. Hernandez AE. Language switching in the bilingual brain: what’s next? Brain Lang. 2009;109:133–140. doi: 10.1016/j.bandl.2008.12.005. [DOI] [PubMed] [Google Scholar]
  33. Hernandez AE, Kotz SA, Hofmann J, Valentin VV, Dapretto M, Bookheimer SY. The neural correlates of grammatical gender decisions in Spanish. Neuroreport. 2004;15:863–866. doi: 10.1097/00001756-200404090-00026. [DOI] [PubMed] [Google Scholar]
  34. Koester D, Schiller NO. Morphological priming in overt language production: Electrophysiological evidence from Dutch. Neuroimage. 2008 doi: 10.1016/j.neuroimage.2008.06.043. [DOI] [PubMed] [Google Scholar]
  35. Kroll JF, Stewart E. Category interference in translation and picture naming: evidence for asymmetric connections between bilingual memory representations. journal of Memory and Language. 1994;33:149–174. [Google Scholar]
  36. Kroll JF, Tokowicz N. Models of bilingual representation and processing. In: Kroll JF, De Groot AMB, editors. Handbook of Bilingualism: Psycholinguistic Approaches. New York: Oxford University Press; in press. [Google Scholar]
  37. Laine M, Rinne JO, Krause BJ, Teras M, Sipila H. Left hemisphere activation during processing of morphologically complex word forms in adults. Neurosci Lett. 1999;271:85–88. doi: 10.1016/s0304-3940(99)00527-3. [DOI] [PubMed] [Google Scholar]
  38. Lavric A, Clapp A, Rastle K. ERP evidence of morphological analysis from orthography: a masked priming study. J Cogn Neurosci. 2007;19:866–877. doi: 10.1162/jocn.2007.19.5.866. [DOI] [PubMed] [Google Scholar]
  39. Lehtonen M, Vorobyev V, Soveri A, Hugdahl K, Tuokkola T, Laine M. Language-specific activations in the brain: Evidence from inflectional processing in bilinguals. Journal of Neurolinguistics. 2009;22:495–513. [Google Scholar]
  40. Lehtonen M, Vorobyev VA, Hugdahl K, Tuokkola T, Laine M. Neural correlates of morphological decomposition in a morphologically rich language: An fMRI study. Brain Lang. 2006 doi: 10.1016/j.bandl.2006.04.011. [DOI] [PubMed] [Google Scholar]
  41. Longe O, Randall B, Stamatakis EA, Tyler LK. Grammatical categories in the brain: the role of morphological structure. Cereb Cortex. 2007;17:1812–1820. doi: 10.1093/cercor/bhl099. [DOI] [PubMed] [Google Scholar]
  42. Longoni F, Grande M, Hendrich V, Kastrau F, Huber W. An fMRI study on conceptual, grammatical, and morpho-phonological processing. Brain Cogn. 2005;57:131–134. doi: 10.1016/j.bandc.2004.08.032. [DOI] [PubMed] [Google Scholar]
  43. Mahendra N, Plante E, Magloire J, Milman L, Trouard TP. fMRI variability and the localization of languages in the bilingual brain. Neuroreport. 2003;14:1225–1228. doi: 10.1097/00001756-200307010-00007. [DOI] [PubMed] [Google Scholar]
  44. Marangolo P, Piras F, Galati G, Burani C. Functional anatomy of derivational morphology. Cortex. 2006;42:1093–1106. doi: 10.1016/s0010-9452(08)70221-1. [DOI] [PubMed] [Google Scholar]
  45. Morris J, Frank T, Grainger J, Holcomb PJ. Semantic transparency and masked morphological priming: an ERP investigation. Psychophysiology. 2007;44:506–521. doi: 10.1111/j.1469-8986.2007.00538.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Penolazzi B, Hauk O, Pulvermuller F. Early semantic context integration and lexical access as revealed by event-related brain potentials. Biol Psychol. 2007;74:374–388. doi: 10.1016/j.biopsycho.2006.09.008. [DOI] [PubMed] [Google Scholar]
  47. Rastle K, Davis M. Morphological decomposition based on the analysis of orthography. Language and Cognitive Processes. 2008;23:942–971. [Google Scholar]
  48. Ravid D, Malenky D. Awareness of linear and nonlinear morphology in Hebrew: a developmental study. First Language. 2001;21:25–56. [Google Scholar]
  49. Rodriguez-Fornells A, Rotte M, Heinze HJ, Nosselt T, Munte TF. Brain potential and functional MRI evidence for how to handle two languages with one brain. Nature. 2002;415:1026–1029. doi: 10.1038/4151026a. [DOI] [PubMed] [Google Scholar]
  50. Seidenberg MS, McClelland JL. A distributed, developmental model of word recognition and naming. Psychol Rev. 1989;96:523–568. doi: 10.1037/0033-295x.96.4.523. [DOI] [PubMed] [Google Scholar]
  51. Shapiro KA, Mottaghy FM, Schiller NO, Poeppel TD, Fluss MO, Muller HW, Caramazza A, Krause BJ. Dissociating neural correlates for nouns and verbs. Neuroimage. 2005;24:1058–1067. doi: 10.1016/j.neuroimage.2004.10.015. [DOI] [PubMed] [Google Scholar]
  52. Talairach J, Tournoux P. Co-planar stereotaxic atlas of the human brain. New York: Thieme; 1988. [Google Scholar]
  53. Tan LH, Spinks JA, Feng CM, Siok WT, Perfetti CA, Xiong J, Fox PT, Gao JH. Neural systems of second language reading are shaped by native language. Hum Brain Mapp. 2003;18:158–166. doi: 10.1002/hbm.10089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Tyler LK, Bright P, Fletcher P, Stamatakis EA. Neural processing of nouns and verbs: the role of inflectional morphology. Neuropsychologia. 2004;42:512–523. doi: 10.1016/j.neuropsychologia.2003.10.001. [DOI] [PubMed] [Google Scholar]
  55. Tyler LK, Stamatakis EA, Post B, Randall B, Marslen-Wilson W. Temporal and frontal systems in speech comprehension: an fMRI study of past tense processing. Neuropsychologia. 2005;43:1963–1974. doi: 10.1016/j.neuropsychologia.2005.03.008. [DOI] [PubMed] [Google Scholar]

RESOURCES