The Visual Word Form Area: Expertise for Reading in the Fusiform Gyrus

Research Article New Enquiry, Cognition and Behavior

The VWFA Is the Abode of Orthographic Learning When Houses Are Used equally Letters

eNeuro eleven February 2019, 6 (1) ENEURO.0425-17.2019; DOI: https://doi.org/x.1523/ENEURO.0425-17.2019

Abstract

Learning to read specializes a portion of the left mid-fusiform cortex for printed word recognition, the putative visual give-and-take form area (VWFA). This study examined whether a VWFA specialized for English is sufficiently malleable to support learning a perceptually atypical second writing system. The study utilized an artificial orthography, HouseFont, in which business firm images represent English phonemes. House images elicit category-biased activation in a spatially singled-out brain region, the so-chosen parahippocampal place area (PPA). Using house images as messages made it possible to test whether the chapters for learning a second writing system involves neural territory that supports reading in the kickoff writing system, or neural territory tuned for the visual features of the new orthography. Twelve human being adults completed two weeks of training to establish basic HouseFont reading proficiency and underwent functional neuroimaging pre and mail-preparation. Analysis of three functionally defined regions of interest (ROIs), the VWFA, and left and right PPA, constitute significant pre-preparation versus mail service-grooming increases in response to HouseFont words only in the VWFA. Analysis of the relationship between the behavioral and neural data found that activation changes from pre-grooming to post-preparation inside the VWFA predicted HouseFont reading speed. These results demonstrate that learning a new orthography utilizes neural territory previously specialized by the acquisition of a native writing organisation. Further, they suggest VWFA engagement is driven past orthographic functionality and non the visual characteristics of graphemes, which informs the broader debate almost the nature of category-specialized areas in visual association cortex.

• fMRI
• language learning
• left fusiform gyrus
• linguistic bridge business relationship
• reading development

Significance Statement

Fluent reading recruits a portion of the brain known as the visual give-and-take form surface area (VWFA), only it is less well understood how malleable the VWFA remains later on acquiring literacy in a native language. There is also debate nigh the type of visual data the VWFA can process as orthographically meaningful. Nosotros tested whether native English-speaking adults could larn a second, visually atypical writing organization for English and used neuroimaging data to assess the location of whatsoever learning effects. Participants acquired basic reading ability and learning effects were found in the neural territory that underlies English language reading. This suggests that the VWFA remains plastic afterwards initial literacy and is not restricted by the visual features of a writing organization.

Introduction

Acquiring a second language in adulthood is challenging, in part because neural resources get specialized for native language processing (Tan et al., 2003; Hull and Vaid, 2007). This specialization can brand it difficult to use the aforementioned neural tissue to support fluency in a 2d language (Mårtensson et al., 2012; Klein et al., 2014). In this commodity, nosotros examined a related question: to what degree can adults acquire a second writing system for their native linguistic communication? To address this question, nosotros taught adult native English speakers a perceptually atypical bogus orthography for English. We used behavioral and functional magnetic resonance imaging (fMRI) methods to ascertain if their newly learned reading skill involved a region already specialized for reading English, the putative visual word form area (VWFA).

The VWFA is a region in the left fusiform gyrus that preferentially responds to orthographic visual stimuli (Cohen et al., 2002; McCandliss et al., 2003; Cohen and Dehaene, 2004; Szwed et al., 2011; Glezer et al., 2015; but for alternative accounts of the VWFA, see Price and Devlin, 2003; Vogel et al., 2014). This response specialization emerges with the acquisition of literacy (Saygin et al., 2016), even when native linguistic communication literacy is caused in adulthood (Dehaene et al., 2010), suggesting an absence of a "critical" menstruum of plasticity (Bornstein, 1989).

Less is known well-nigh the degree to which the VWFA remains plastic one time it has become specialized to support a native writing organisation, and to what extent its recruitment depends on the perceptual characteristics of a writing system. The widespread acquisition of second language literacy suggests the VWFA can support skilled reading for multiple orthographies (Tschirner, 2016). Nevertheless, this apparent ease may be misleading due to the loftier degree of visual similarity between naturally occurring orthographies (Hirshorn and Fiez, 2014). This visual similarity may reflect the cultural evolution of writing systems to use forms that are optimized for the representational capacities of the VWFA (Dehaene, 2009), in which case, the VWFA may be poorly equipped to respond to a perceptually atypical orthography. Farther, the loftier degree of visual similarity between natural writing systems may allow any literacy-driven specialization of the VWFA to readily transfer to another orthography, thereby overestimating the plasticity of the VWFA for orthographies that are perceptually distant from the native orthography.

A strong test of the VWFA'southward plasticity therefore requires acquisition of a perceptually atypical orthography by an individual whose VWFA has already been specialized by a native orthography. The need to disentangle factors that are intertwined in naturally occurring orthographies motivates the employ of an artificial orthography in the present study. Nosotros build on a previously reported study that used face images as "letters" to represent English phonemes (Moore et al., 2014b). In this previous study, orthographic learning effects were observed in the left mid-fusiform cortex, but there was ambiguity whether these furnishings localized to the VWFA or to tissue specialized for face processing, the left fusiform confront expanse (FFA). Thus, it remains unclear whether orthographic learning effects localize to tissue that is specialized for processing the visual characteristics of the character forms (e.g., words printed with confront letters to the FFA) or whether visual stimulus with orthographic functionality may induce plasticity inside the VWFA, even when it has already been specialized for a perceptually typical native orthography.

To address this question, we trained English speakers to read an artificial orthography in which images of houses represent English phonemes (HouseFont). Nosotros chose houses because they are preferentially candy in a region known equally the parahippocampal identify area (PPA), which is spatially distant from the VWFA. The PPA'due south distinctiveness allows us to identify the neural tissue defended to processing the graphemes of our new orthography. Nosotros employed a localizer browse to functionally place the PPA and VWFA, and pre-preparation and post-training scans to isolate neural changes associated with HouseFont learning. This allowed for a clear examination of whether a VWFA tuned to a native orthography (English) has the flexibility to respond to a 2nd orthography (HouseFont), even when this second orthography uses graphemes that are highly distinctive from those used in the Roman alphabet. If the perceptual characteristics of graphic symbol forms drive the locus of orthographic learning, significant learning effects should be observed in the PPA. Alternatively, if the functional use of visual forms as orthographic symbols drives the locus of orthographic learning, and the neural tissue that supports this learning remains malleable, pregnant learning effects should be observed in the VWFA.

Materials and Methods

Participants

14 University of Pittsburgh undergraduate students were originally enrolled in the study. This sample size was selected based on research showing that imaging research tin can achieve ability of roughly lxxx% using a threshold of 0.05 and 12 subjects (Desmond and Glover, 2002), and results for our prior study (Moore et al., 2014b) in which significant differences in the VWFA territory were observed for between-group comparisons (N = xi and N = 12) of the response to a trained versus untrained orthography. I participant dropped out on the second mean solar day of preparation and one dropped out after having completed everything except the post-preparation imaging session. Data from the final sample of 12 individuals (8 female, four male) are reported (G age = xix.17 years, SD = i.nineteen). All participants were recruited from a database of individuals interested in participating in enquiry studies. All study participants were right-handed, native English language speakers, and had no history of 2nd language fluency, hearing or vision issues, learning or reading problems, drug or booze corruption, mental illness, neurologic issues, or contraindications for fMRI. All participants provided informed consent and were compensated for their fourth dimension. All procedures were approved by the institutional review lath of the University of Pittsburgh.

Report overview

The study involved a 2-week preparation protocol to learn HouseFont. Training occurred later two pre-training fMRI sessions and before a post-training fMRI session. The beginning of the pre-grooming fMRI sessions was designed to localize three regions of interest (ROIs): the VWFA and the left and right PPA. The purpose of the second pre-grooming fMRI session was to measure the response to words printed in HouseFont before training. The terminal fMRI session measured the response to HouseFont later on training. Behavioral measures of post-preparation reading skill were as well acquired as role of this last session. Participants were debriefed and paid following the post-training scan. Effigy 1 provides an overview of the written report timeline and the blueprint of specific tasks. Table one summarizes the HouseFont training protocol. Farther details are provided below.

• Download figure
• Open in new tab
• Download powerpoint

Figure i.

Participants completed a localizer scan, a pre-training scan, HouseFont training, and a post-training scan. The images aslope each point on the timeline are examples of the stimuli used for the neuroimaging sessions.

View this tabular array:

• View inline

Tabular array i.

HouseFont training protocol

Pre-training fMRI sessions

Localizer session

Participants started the written report by completing a localizer fMRI session and a bombardment of standardized reading tests. The localizer session was conducted using a Siemens Medical Systems 3T Magnetom TIM Trio scanner with a 32-aqueduct radio frequency roll. High-resolution structural scans were collected using an axial MPRAGE with 192 slices and 1-mm isotropic voxels. Functional data were collected across 29 interleaved slices in the same plane as the structural data (TR = 1500 ms, TE = 25 ms, FOV = 200 mm, FA = 70°).

During functional data acquisition, participants completed a 1-back task with v categories of visual stimuli: (i) houses, (ii) faces, (3) words, (4) letter-strings, and (5) patterns (Fig. one). Following similar localizer protocols used in prior studies (Pull a fast one on et al., 2009; Rossion et al., 2012), stimuli were drawn from sets of 40 exemplars for each of the non-orthographic (houses, faces, and patterns) categories, and sets of 157 exemplars for the orthographic (discussion and letter-cord) categories. The browse consisted of 4 functional runs each lasting 6 min. Every run had a total of 15 blocks (iii of each category, randomly ordered). Blocks consisted of 15 trials, with the stimulus for each trial presented for 200 ms followed past an 800-ms fixation cross. Participants were asked to press a key when they detected a stimulus that repeated the one shown previously (i.e., 1-dorsum). A 1-dorsum target was presented for 12.v% of each block. A 9-s baseline status followed each block. During this baseline, participants attended to a fixation cross at the center of the screen. During each run, the sets of house, face, and pattern stimuli were distributed pseudorandomly within each of the three blocks for each condition. With the exception of ane-back trials, the word and letter-cord stimuli did non repeat. None of the house images used in the localizer task were used as stimuli in the subsequent parts of the study.

Pre-training session

The pre-training browse was completed within a calendar week of the localizer session. For logistic reasons, the scanner, a 3T Siemens Allegra equipped with a standard radio frequency gyre, differed from that used for the localizer session. High-resolution structural scans were nerveless using a sagittal MPRAGE with 192 slices and i-mm isotropic voxels. Functional data were collected beyond 38 interleaved slices (3.125 × 3.125 × iii.2 mm voxels) parallel to the inductive-posterior commissure (TR = 2000 ms, TE = 25 ms, FOV = 200 mm, FA = seventy°).

During the pre-training scan participants passively viewed 140 words printed in HouseFont and an untrained artificial orthography, KoreanFont. KoreanFont is an artificial alphabetic orthography that borrows graphemes from Hangul, the Korean writing organisation, and assigns them to English phonemes. They also saw 16 design displays that were repeated over 140 trials. Word and pattern stimuli were matched for length. Participants completed two runs, which consisted of vii blocks of each stimuli type for a total of 21 blocks. Each cake contained x trials of the same stimulus blazon. For each trial, participants saw i HouseFont or KoreanFont give-and-take or design fix for 1500 ms, followed by 500 ms of a centrally located fixation cross (Fig. ane). They were instructed to attend to the stimuli, but were not asked to perform an overt job. The same set of HouseFont words were presented during the pre-training and post-training sessions; individuals were not exposed to this set of HouseFont words at whatever other fourth dimension.

HouseFont training

HouseFont consists of 35 grapheme-to-phoneme mappings, where each graphic symbol is a particular house paradigm that is used to stand for a single phoneme or (in a few cases) two very like sounds (due east.grand., /ɑ/ in hot and /ɔ/ in ball). All of the house images used for HouseFont were 300 × 300 pixels, normalized, and lightened to a low-cal gray. Participants were trained to read HouseFont across nine sessions, which were cleaved into three phases: house-phoneme mapping (session 1), discussion-level training (sessions ii–five), and story-level training (session six–9). Each training session lasted from i to 2 h. These training phases are summarized.

Session ane: house-phoneme mapping

Participants began their training by learning to map each HouseFont grapheme with a corresponding phoneme using a cocky-paced computer program. The 35 house graphemes were visually presented in random order, and participants pressed a spacebar to hear the corresponding sound later on each grapheme was displayed. Participants completed 5 cycles of the phoneme training, followed past a test of their ability to produce the phoneme associated with each graphic symbol. Participants who achieved <90% accuracy repeated the training. All participants passed in three or fewer attempts.

Sessions 2–5: word-level training

After a brief refresher on the house-phoneme mapping, participants learned how to read aloud curt words printed in HouseFont. Each session of the word-level training involved reading 400 one- to two-syllable words, which were two to five phonemes in length. The same prepare of 400 words was used in sessions 2–5, with the give-and-take order randomized across sessions. For each trial, participants were encouraged to attempt to read the discussion when it appeared; they had the option to hear any individual phoneme or the unabridged word if necessary. At the terminate of each session, a computer-based, single-discussion-reading examination was administered. Each discussion test consisted of three conditions presented in a block design, with the social club of blocks randomized beyond test sessions: old HouseFont words (words included in word-level training), new HouseFont words, and pronounceable HouseFont non-words. There were twenty trials per condition. A trial consisted of a ane-syllable word that was three to iv phonemes in length. The pronunciation accuracy was scored for each item, and reading latency was measured from the time a word first appeared on the screen to when the participant pressed the space bar to advance to the next discussion.

Sessions six–9: story-level training

In the final preparation phase, participants advanced to reading aloud short stories printed in HouseFont (Fig. two). For each session, participants read ten early reader stories of similar difficulty from the "Now I'm Reading!" series (Gaydos, 2003). The story level increased in difficulty with each successive session. Performance on story reading was measured by words read per minute. At the cease of each session, participants completed a single-word-reading exam identical in design and scoring to those used during word-level grooming.

Post-training behavioral and fMRI session

During the final session (session 10), participants completed behavioral testing to appraise their final HouseFont reading skill and an fMRI session to measure out learning-related changes in the neural response to HouseFont. For the behavioral testing, participants' reading speed and accuracy were assessed using six passages (Form A Stories 1–half-dozen) from the gray oral reading test-4 (GORT-iv; Wiederholt and Bryant, 2001) that were transcribed into HouseFont. Number of words read per minute and number of errors made per discussion were calculated as an index of reading speed and accuracy respectively. The number of errors made per discussion was determined past dividing the number of errors (east.thou., omissions, phoneme substitutions, whole word or part word repetitions, etc.) made past the number of words in each passage. The postal service-preparation scan was completed during session x immediately later assistants of the behavioral tests, using the same scanner and fMRI protocol equally in the pre-preparation scanning session.

fMRI information assay

fMRI data preprocessing

Preprocessing of the fMRI information were completed using the Assay of Functional NeuroImages (AFNI) software package (Cox, 1996). The commencement two brain volumes from the localizer runs and the start brain volume from the pre-preparation and post-training runs were removed to allow for stabilization of the signal. The functional images were piece fourth dimension corrected (3dTshift), and all information were motility corrected (3dvolreg). The information were smoothed using a Gaussian filter prepare to a smoothing kernel of 5.5-mm full width at one-half maximum. Adjacent, the functional images were registered to the skull stripped high-resolution structural images. Images were then transformed into standard Talairach space using a not-linear warping procedure in AFNI to let for group assay (Talairach and Tournoux, 1988). Functional images were scaled to a mean global intensity.

ROIs identification

The central question of this study is whether HouseFont learning is supported by neural tissue specialized by the acquisition of a native (English language) orthography (i.e., territory at or near the VWFA) or tissue that shows selectivity for the perceptual characteristics of the not-native HouseFont orthography (i.eastward., the territory at or about the PPA). To address this question, the data from the localizer session were used to functionally localize a priori ROIs in the left fusiform and bilateral parahippocampal cortices.

Multivariate pattern analysis (MVPA) was used to identify each of the 3 ROIs inside MATLAB using the Princeton Multi-Voxel Pattern Analysis toolbox (Detre et al., 2006). For this analysis, the functional information preprocessing was the same as described in a higher place, with 1 exception: as is common in MVPA, the data were not spatially smoothed (Mur et al., 2009). MVPA has been plant to be more than sensitive to fine grain differences betwixt stimuli (for review, come across Coutanche, 2013). This increased sensitivity allowed united states of america to successfully localize the left fusiform ROI using the hallmark dissimilarity used in early piece of work characterizing the VWFA: words and alphabetic character-strings (Petersen et al., 1990; Cohen et al., 2002; Dehaene et al., 2002). To localize the PPA ROIs, a house and word contrast was used.

For each run, nosotros z scored the pre-candy activity values for each voxel, accounting for the hemodynamic delay by shifting the condition time class by two TRs. A Gaussian naive Bayes (GNB) classifier was trained and tested on the action patterns for the contrasts of involvement (words versus alphabetic character-strings and houses versus words) using a exit-ane-run-out cross-validation process, where each iteration was trained on data from all-but-1 run (e.g., three runs), and tested on data from the held-out run. Classification performance from the iterations was averaged to requite a unmarried accuracy value. The resulting accuracy for the contrasts (where chance is fifty%) was then allocated to the central voxel of a three-voxel radius searchlight sphere, which was moved serially across the brain.

We identified the voxel with summit decoding accurateness for the words versus letter-strings contrast within AFNI's anatomic mask of the left fusiform cortex and for the houses versus words contrast within anatomic masks of the left and right parahippocampal cortex for each subject field. To generate the grouping level ROIs for the VWFA and PPAs, we created a 6-mm radius sphere centered on the location of average peak accuracy across all subjects for the corresponding contrast in each anatomic mask (Table 2).

View this tabular array:

• View inline

Table 2.

Functionally divers ROIs that were practical to the pre-preparation and post-training data

Analysis of behavioral and neural learning effects

Analysis of behavioral learning effects

To exam whether participants showed improvements in HouseFont reading during grooming, reading accuracy and reading speed were assessed for each of the discussion tests. A one-way repeated measures ANOVA was performed on the average reading latency scores for right responses across the eight-word tests to determine whether reading speed changed over the course of grooming.

Analysis of neural preparation furnishings

To test whether participants showed neural changes associated with training (i.e., changes in the neural responses to HouseFont words), the pre-grooming and post-grooming data were modeled using AFNI'southward 3dDeconvolve to estimate the Bold response (boilerplate beta-weight value) for HouseFont and KoreanFont. The motility estimates from preprocessing were included equally regressors of no involvement. So, nosotros compared the resulting t-values for HouseFont and KoreanFont beyond the pre-preparation and post-training sessions, using both an ROI-based and a whole-encephalon (voxel-wise) group analysis.

For the ROI analysis, the VWFA and PPA ROIs identified from the localizer (Table ii) were applied to the pre- and mail-training session data. Using AFNI 3dROIstats, the averaged beta weight value for the voxels within each ROI was obtained for each participant's response to HouseFont and KoreanFont before and after HouseFont training. These values were exported to IBM Statistical Package for the Social Sciences (SPSS) version 25. To determine whether there were training and ROI based differences in HouseFont activation, a ii × two × iii repeated measures ANOVA was performed with orthography (HouseFont, KoreanFont), session (pre-training, post-preparation), and region (VWFA, left PPA, and right PPA) specified as within-subject variables. Information technology was expected that in that location would be a pregnant three-way interaction, which would suggest there was a differential change in HouseFont activation between ROIs that resulted from HouseFont reading training. A significance threshold of p < 0.05 was used, with correction for all violations of normalcy in the data.

As a complementary analysis approach, a whole-encephalon voxel-wise analysis was used to identify pre-preparation versus mail service-training changes in the response to HouseFont without a priori constraints. The computed t values for the HouseFont versus KoreanFont contrast for each participant were contrasted across the pre-preparation versus post-grooming sessions for each voxel using AFNI 3dClustSim, with a significance threshold of p = 0.005 (corrected p = 0.05) and a cluster size threshold of 60 contiguous voxels.

Relationship between behavioral and neural measures

To examine the relationship betwixt behavioral and neural measures of learning, each participant's reading speed score from the final word test was standardized and combined with the standardized reading speed score from the GORT-4. This composite reading speed score was examined using a regression analysis, to make up one's mind whether the pre-training versus postal service-training alter in the estimated BOLD responses within the VWFA ROI accounted for HouseFont reading speed variability.

Because the sample size of the current written report is modest, we performed a similar analysis that combined data from the participants in the current study (N = 12) with data from two participant groups reported past Moore et al. (2014b): 1 grouping that learned an artificial orthography with face up images as letters (FaceFont; N = 12) and one group that learned an artificial orthography with borrowed Korean graphs mapped to English phonemes (KoreanFont; N = 11). For each participant from the Moore et al., study, the last reading speed was calculated in the aforementioned way every bit information technology was for HouseFont, by averaging the z score of the GORT reading speed and the inverse z score of the final word examination reading speed. The imaging data from the Moore et al. written report were acquired using the aforementioned pattern and scanner equally in the current written report, with the exception that only a post-training session was acquired, and instead of viewing HouseFont and KoreanFont words, participants viewed FaceFont and KoreanFont words. Considering the data from the Moore et al., written report were previously analyzed using a dissimilar software package, they were reprocessed using the same methods as in the electric current study.

Next, nosotros used an ROI analysis to extract the boilerplate estimated BOLD response within the VWFA territory for each participant across our three groups (HouseFont-trained, FaceFont-trained, KoreanFont-trained). To avoid biasing the results by using the VWFA ROI identified using data from just the HouseFont participants, we drew on the literature to define an unbiased ROI for this across-grouping analysis. Specifically, nosotros used a coordinate from a recent written report past Lerma-Usabiaga et al. (2018), where real words and consonant strings were assorted to localize a specific VWFA subregion in the middle occipitotemporal sulcus (mOTS) that exhibits lexical-level orthographic selectivity, and which can be distinguished from a more posterior VWFA subregion that is more more often than not responsive to visual word forms (pOTS). The average height coordinate reported by Lerma-Usabiaga et al. (2018) for their mOTS subregion were rounded to the closest whole numbers, transformed into Talairach space, and used as a middle of a six-mm sphere (–42, –57, –iv). Using AFNI 3dROIstats, the averaged beta weight value for the voxels within this mOTS ROI was obtained for each participant's response to their trained orthography during the post-training scan. These values were entered into a regression analysis, along with the orthography learned past the participant, to predict participants' reading speed following training.

Results

Behavioral measures of HouseFont learning

Boilerplate accuracy for trained participants across all of the give-and-take tests performed during training was 90%. This is not surprising, because HouseFont is a transparent orthography and and so once the grapheme-phoneme mappings have been mastered, they can in theory be used to decode English words and pronounceable nonwords with perfect accurateness. For this reason, the focus of the behavioral preparation analyses was reading latency. To test whether participants showed improvements in HouseFont reading over the course of their training, a one-way repeated measures ANOVA was performed on the average reading latency score for correct responses on the eight-discussion tests. Two individuals were missing a single word test and were excluded from the analysis. The Greenhouse-Geisser correction was applied because Mauchly's test of sphericity was not met, p = 0.01. At that place was a significant outcome of examination session F _(2.28,20.48) = 10.47, p = 0.001, which reflects a decrease in reading latencies over the grade of HouseFont grooming. From the first discussion test (session ii) to the last word test (session 9), the average reading latency dropped from 6288 ms (SD = 1963 ms) to 4670 ms (SD = 1126 ms). This 25% reduction in reading latency indicates that participants became more skilled at reading HouseFont across the two weeks of grooming.

Improvements in HouseFont reading were also evident in the context of story reading. Participants maintained a relatively steady rate of reading across story level training (sessions vi–9), although the stories became increasingly more hard across sessions (Fig. three). By the stop of story-level training (session 9), participants were reading an average of 21.85 words per infinitesimal (SD = 2.88). Participants also read half-dozen passages of a standardized reading cess, the GORT, to assess final reading accuracy and speed. On this measure participants attained a hateful fluency of 21.15 (SD = 5.13) words per minute, with a mean fault rate of 2% (SD = 0.02) per word. These proficiency results are similar to those observed for get-go grade children learning English (Hasbrouck and Tindal, 2006).

• Download figure
• Open in new tab
• Download powerpoint

Figure iii.

Stories increased in difficulty over the 4 d of story-level reading, but participants maintained a similar rate of words read per minute. The performance of HouseFont participants on the early on reader training stories was consequent with performances seen for other artificial orthographies, KoreanFont, and FaceFont. KoreanFont and FaceFont information adjusted with permission from Moore et al. (2014b).

Neural measures of HouseFont learning

ROI analysis

A 2 × 2 × 3 repeated measures ANOVA was used to examine the effect of orthography (HouseFont, KoreanFont), session (pre-preparation, postal service-preparation), and region (VWFA, left PPA, and right PPA) on neural activity. This analysis revealed a chief effect of orthography, F _(ane,11) = 97.07, p < 0.001, = 0.90, and region, F _(one.37,22) = 7.97, p = 0.008, = 0.42, with no effect of session, F _(1,11) = 0.eleven, p = 0.749, = 0.01. There was a significant interaction between orthography and region, F _(1.79,22) = 10.41, p = 0.001, = 0.49, and trend level interactions for orthography and session, F _(i,11) = four.32, p = 0.062, = 0.28, and training and region, F _(one.49,22) = 3.20, p = 0.079, = 0.23. Most importantly, the predicted three-way interaction was too pregnant, F _(one.44,22) = half dozen.25, p = 0.016, = 0.36.

To examine the three-way interaction and accost our a priori hypothesis that HouseFont-elicited activity in the VWFA would change after training, we ran a divide 2 × two repeated measures ANOVA [orthography (HouseFont, KoreanFont), session [pre-training, mail-preparation] for each region. Within the VWFA in that location was a main effect of orthography, F _(1,11) = 15.23, p = 0.002, = 0.58 and no effect of session, F _(1,11) = 0.86, p = 0.374, = 0.07 (Fig. 4). Critically, yet, at that place was a significant interaction between orthography and session, F _(ane,11) = nine.79, p = 0.010, = 0.47, in the VWFA. Post hoc comparisons of the interaction revealed that the response to KoreanFont decreased across sessions, p = 0.100, while HouseFont evoked greater activation in the post-training session compared to pre-preparation session, p = 0.059. These are the expected results if the HouseFont preparation tuned the VWFA to treat strings of HouseFont images as orthographic information.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

The VWFA showed no master effect for session or orthography, but at that place was a significant interaction of session and orthography. The left and right PPA showed the expected pregnant principal outcome of orthography, no principal effect of preparation, and no significicant interaction between session and orthography. Error confined indicate SE.

In the left PPA, there was an issue of orthography, F _(ane,11) = 55.43, p < 0.001, = 0.83, no effect of session, F _(1,eleven) = 0.47, p = 0.507, = 0.04, and no significant interaction between orthography and session, F _(1,11) = 1.91, p = 0.194, = 0.15. Similarly, in the right PPA, there was an effect of orthography, F _(1,eleven) = 62.12, p < 0.001, = 0.85, no effect of session, F _(1,xi) = 1.31, p = 0.276, = 0.11, and no interaction betwixt orthography and session, F _(i,11) = 0.00, p = 0.993, = 0.00. The expected primary furnishings of orthography and the lack of other effects show that the PPA bilaterally responded more than to HouseFont than KoreanFont and that HouseFont training did not alter this difference.

Whole-brain voxel-wise analysis

To investigate whether HouseFont training altered the response to HouseFont strings in areas outside of the a priori ROIs, a whole-brain voxel-wise analysis was conducted with the pre-preparation and mail-training fMRI information. HouseFont activation was compared to KoreanFont activation in both the pre-training and post-training scans separately. And then, the difference in pre-grooming was compared to the departure in post-training. This comparison yielded x pregnant training effect clusters, 9 of which were negative, indicating more than activation in post-preparation. The ane positive cluster, which was located in the left middle temporal gyrus (BA19), indicates more activation during pre-training (Tabular array three). Several of the clusters are in regions known to be involved in reading (Bolger et al., 2005), including the left inferior frontal gyrus, the left superior parietal lobe, and the left fusiform gyrus. Portions of the left fusiform gyrus training outcome cluster overlapped with the VWFA ROI (Fig. v), which is not surprising given the meaning interaction effect found in the VWFA ROI. No training effect clusters were identified inside the left or correct parahippocampal gyrus.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

VWFA ROI (greenish) identified past the localizer browse (–34, –55, –13), and the learning consequence cluster (blue) identified from the whole-brain voxel-wise analysis of activation for HouseFont versus KoreanFont from pre-training to post-grooming (–40, –49, –x). Red represents the overlap. Coordinates are in Talairach infinite.

View this tabular array:

• View inline

Table three.

Clusters identified by the whole-encephalon voxel-wise assay [trained orthography (HouseFont) vs untrained orthography (KoreanFont), pre-training to postal service-training]

Relationship betwixt behavioral and neural measures of HouseFont learning

To probe the human relationship betwixt neural and behavioral measures of HouseFont learning furnishings, we performed a regression to test the contribution of grooming related activation change in the VWFA to HouseFont reading speed. A HouseFont reading speed score was calculated by averaging the z score of the number of words read per minute on the GORT and the inverse z score (z score multiplied by –ane) of the response time per word on the last discussion exam. The change in activation from pre-training to mail-training in the VWFA did significantly predict reading speed b = iii.34, t ₍₁₀₎ = 3.90, p = 0.003, and it explained a significant proportion of variance in reading speed scores, R ² = 0.60, F _(1,10) = xv.24, p = 0.003 (Fig. 6). Based on these results, we conclude that the VWFA is critical for rapid HouseFont reading.

• Download effigy
• Open in new tab
• Download powerpoint

Figure 6.

Scatter plot of the variance explained past the pre-training to mail service-training change of the VWFA for reading speed. The VWFA change showed a significant positive human relationship with reading speed. Reading speed scores were zilch centered.

We obtained convergent results using information from the HouseFont-trained participants in the electric current study, and the FaceFont-trained and KoreanFont-trained participants previously reported by Moore et al. (2014b). While the three orthographies differ in the graphs they use and in their boilerplate reading speed (Fig. 3), we expected that behavioral measures of reading speed would be significantly predicted by the VWFA activation in response to the trained orthography. We assessed this using a specific VWFA subregion reported in the literature (mOTS; Lerma-Usabiaga et al., 2018) equally an ROI (to avoid biasing our ROI localization to the HouseFont group). The post-training response to the trained orthography inside the mOTS ROI significantly predicted reading speed b = ane.38, t ₍₃₂₎ = 2.82, p = 0.008. On the other hand, which orthography a participant learned (FaceFont, KoreanFont, or HouseFont) did not significantly predict reading speed b = –0.00, t ₍₃₂₎ = –0.01, p = 0.992. These results align with previous reports of FaceFont and KoreanFont learning effects (Moore et al., 2014b) and the findings from HouseFont. Moreover, the significant relationship between the neural and behavioral measures of learning suggest that despite the visual differences in the graphs used, reading speed variation beyond all iii artificial orthographies can be predicted by learning effects seen within the VWFA territory (Fig. 7).

• Download figure
• Open in new tab
• Download powerpoint

Effigy 7.

Scatter plot of the variance in reading speed explained by the response to trained orthography inside the VWFA ROI. The response to the trained orthography showed a meaning positive relationship with reading speed. Reading speed scores were cipher centered beyond all three orthographies.

Discussion

This study tested whether conquering of a perceptually atypical 2nd writing arrangement recruits the same neural tissue already tuned past native-English reading, or whether instead the locus of orthographic learning tracks with the perceptual characteristics of the grapheme forms. More than specifically, we were interested in the presence or absence of artificial orthography (HouseFont) learning effects within three functionally defined areas: an orthographic area (VWFA) within the left mid-fusiform gyrus (Cohen and Dehaene, 2004), and bilateral identify areas (left PPA, right PPA) within the parahippocampal gyri (Epstein and Ward, 2010). We hypothesized that orthographic learning effects would exist observed in either the VWFA or the PPA, only not in both regions. Meaning learning effects were found only within the VWFA, and individual differences in the magnitude of pre-preparation versus postal service-training changes in VWFA activation correlated with differences in HouseFont reading speed. We conclude the VWFA was recruited to support HouseFont literacy acquisition in our adult participants.

The results from this written report converge with Moore et al. (2014b), who also observed training-related increases in the VWFA territory when participants learned 1 of two artificial alphabets for English language: FaceFont, in which confront images were used as letters, and KoreanFont, in which letters were borrowed from the Korean alphabet and mapped to English phonemes. Taken together, the results from the current study and Moore et al. (2014b) point toward three principles of VWFA role: (one) learning a new alphabetic orthography uses VWFA tissue already specialized past conquering of English literacy, (2) orthographies with a wide range of visual forms can induce neural plasticity in the VWFA, (3) the laterality of the VWFA is influenced by the mapping principles of an orthography.

New orthographic learning uses the same tissue as English

The HouseFont training furnishings demonstrate that the VWFA in native English language speakers was modified past HouseFont learning. Similarly, Moore et al. (2014b) found a left-lateralized grooming consequence for FaceFont in the vicinity of the VWFA. Yet, they could not conclusively assign FaceFont learning to the aforementioned territory that supports English language reading for ii reasons. First, a putative left homolog of the right-lateralized face processing expanse (Kanwisher et al., 1997) falls in close proximity to the VWFA (Nestor et al., 2013). Consequently, the locus of observed FaceFont learning effects could arguably reflect the employ of neural tissue specialized for confront or orthographic processing. Second, Moore et al. (2014b) did non localize the response to printed English in their participants, then they were unable to directly compare the functional response to English and FaceFont. The present report circumvented these problems by using firm graphs associated with category-specific activation in tissue that is spatially distant from the VWFA and by functionally localizing the VWFA earlier HouseFont training.

While we aspect the change in HouseFont activation within the VWFA to orthographic learning, culling accounts warrant consideration. It is possible that repetitive exposure to a small prepare of visual images could be sufficient to increase the VWFA response to the frequently experienced images. We cannot completely discount this possibility because none of our studies have involved a control group with similar exposure to the paradigm sets in a not-literacy context. However, we favor the idea that the activation changes in the VWFA are related to literacy conquering. This is because the regions in which activation increased were selective, the learning furnishings in the fusiform gyrus correlate with reading (Fig. six; Moore et al., 2014b), and the connectivity of the VWFA is suited for visual-phonological mapping (Alvarez and Fiez, 2018).

It is also of import to remember that imaging is a correlational, rather than a causal, method. It is possible that office or all of the increased VWFA activation following training could be from accessing the English orthographic representations of the HouseFont words. If this were the example, it could hateful the VWFA is not necessary for accurate HouseFont reading, simply rather is activated as a past-product of accurately decoding the HouseFont word. Nosotros took extra care to ensure that HouseFont graphemes were never equated with an English language grapheme and no English appeared during the grooming phase. Additionally, prior piece of work with artificial orthographies found that a patient with acquired alexia was unable to learn a small set of face-phoneme pairings simply was able to learn face-syllable pairings (Moore et al., 2014a). This finding suggests that the VWFA territory is critical rather collateral to learning an bogus alphabetic orthography.

Visual and encephalon constraints on orthographic learning

Our findings likewise demonstrate that at that place is considerable flexibility in the type of visual forms that can serve as letters of an alphabet. This is not a trivial point, as this observed flexibility is counter to some theories of how the brain and reading shape 1 another. About notably, Dehaene (2009; p 184) conjectured that orthographies have culturally evolved to exist visually similar to each other because they are forced to conform to the abilities of the available neural tissue. As part of this argument, Dehaene specifically suggested that both face and house images are avoided most entirely by writing systems because the VWFA, which supports skilled reading, is not the preferred processing area for this kind of visual information (Dehaene, 2009). The findings of this report, and those of Moore et al. (2014b), challenge this thought, because they show that participants tin can readily obtain basic reading proficiency for an orthography with perceptually atypical forms (house or face images).

I potentially of import caveat is that individuals tend to read FaceFont and HouseFont more slowly than an artificial orthography fabricated of more than typical graphs (KoreanFont; Fig. 3). This could reflect intrinsic limitations, such as those posited by Dehaene (2009). Alternatively, information technology could reflect differences in the visual complication and discriminability of faces and houses, as compared to the simpler and higher-contrast letter forms in KoreanFont, or that tissue tuned for printed English might better transfer this tuning to a visually similar orthography (due east.g., KoreanFont) as compared to a visually dissimilar (e.g., FaceFont, HouseFont) orthography. Transfer effects also might occur for other characteristics of an orthography, such every bit its grouping of graph elements (such equally the dots in Standard arabic words; Abadzi, 2012). This transfer effect hypothesis could be tested by comparing the learning of artificial orthographies in which graphemes are borrowed from natural orthographies varying in perceptual distance from a reader's native orthography. For example, we might predict native English speakers would read an artificial orthography with Korean graphemes more rapidly than 1 with Arabic graphemes because Korean letters are more than visually similar to English language letters.

Despite baseline differences in reading speed, similar rates of learning are found across HouseFont, FaceFont, and KoreanFont (Fig. 3) and there is no prove of a learning plateau across six weeks of preparation (Martin et al., 2018). Taken together, these results support Moore et al. (2014b)'s determination that tuning of the VWFA for English creates a "perceptual bottleneck" that slows the visual discrimination of a perceptually atypical second orthography, without preventing accurate reading and fluency gains with continued reading experience. In sum, the weight of evidence suggests that learnable orthographies are not constrained past the encephalon, but instead that feel with an orthography shapes the brain.

Laterality effects in orthographic learning

Finally, our results demonstrate that alphabetic orthographic learning recruits left-lateralized brain regions, regardless of the perceptual characteristics of the orthography. In the whole-encephalon voxel-wise analysis, a strong design of left-lateralized regions showed HouseFont preparation furnishings (Table three), and a like fix of regions showed preparation effects in FaceFont (unpublished findings). Most notably, both the electric current study and Moore et al. (2014b) plant preparation effects in the left fusiform gyrus. The lack of a training effect in the right fusiform gyrus in Moore et al. (2014b) is particularly hit as confront processing has been associated with right-lateralized visual processing (Kanwisher et al., 1997; Grill-Spector et al., 2004).

HouseFont, FaceFont, and KoreanFont differ visually, only share the same alphabetic mapping principle. To clarify whether the principle of left-lateralization holds truthful for non-alphabetic orthographies, we turn to Hirshorn et al. (2016)'due south Faceabary training study in which face images represented English syllables. The written report found Faceabary training effects in both the left and right mid-fusiform gyrus, with more than bilateral patterns of activation correlated with higher Faceabary reading fluency. In dissimilarity, Hirshorn et al. (2016) found a potent pattern of left-lateralization exterior of the fusiform gyrus when comparing pre-training to post-preparation activation for Faceabary, which is consistent with results from both the current written report and Moore et al. (2014b). This leads us to conclude that a key commuter of left-lateralized fusiform gyrus recruitment is whether an orthography implements an alphabetic mapping principle, while a broader left-lateralized reading network is recruited irrespective of an orthography's mapping principle.

Conclusions

The current report found that adult acquisition of a perceptually atypical alphabetic orthography induced left-lateralized neural plasticity in the VWFA. We conclude that the VWFA remains highly malleable in adulthood. Further, our results, in combination with other work, indicate that the localization of orthographic learning to the VWFA is driven by orthographic functionality rather than the visual characteristics of a script, while the lateralization of the VWFA is influenced by the mapping principles of a script.

Acknowledgments

Acknowledgements: We thank Paul Brendel, Robert Schwartz, Brandon Carlos, and members of the Fiez Lab for their assistance and helpful discussions.

Footnotes

• The authors declare no competing financial interests.

• This work was supported by the National Institutes of Health Grant R01 HD060388.

This is an open-admission article distributed under the terms of the Artistic Eatables Attribution 4.0 International license, which permits unrestricted utilize, distribution and reproduction in whatsoever medium provided that the original piece of work is properly attributed.

References

1. ↵

   Abadzi H (2012) Tin can adults become fluent readers in newly learned scripts? Educ Res Int 2012:8. doi:10.1155/2012/710785

2. ↵

   Alvarez TA , Fiez JA (2018) Current perspectives on the cerebellum and reading evolution. Neurosci Biobehav Rev 92:55–66. doi:ten.1016/j.neubiorev.2018.05.006 pmid:29730484

3. ↵

   Bolger DJ , Perfetti CA , Schneider W (2005) Cross‐cultural effect on the brain revisited: universal structures plus writing arrangement variation. Hum Brain Mapp 25:92–104. doi:10.1002/hbm.20124 pmid:15846818

4. ↵

   Bornstein MH (1989) Sensitive periods in development: structural characteristics and causal interpretations. Psychol Bull 105:179–197. doi:10.1037/0033-2909.105.2.179 pmid:2648441

5. ↵

   Cohen L , Dehaene S (2004) Specialization inside the ventral stream: the case for the visual word form area. Neuroimage 22:466–476. doi:10.1016/j.neuroimage.2003.12.049

6. ↵

   Cohen Fifty , Lehéricy S , Chochon F , Lemer C , Rivaud S , Dehaene S (2002) Language‐specific tuning of visual cortex? Functional backdrop of the visual word form area. Brain 125:1054–1069. doi:10.1093/brain/awf094 pmid:11960895

7. ↵

   Coutanche MN (2013) Distinguishing multi-voxel patterns and mean activation: why, how, and what does it tell us? Cogn Affect Behav Neurosci xiii:667–673. doi:10.3758/s13415-013-0186-2 pmid:23857415

8. ↵

   Cox RW (1996) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 29:162–173. doi:10.1006/cbmr.1996.0014 pmid:8812068

9. ↵

   Dehaene Southward (2009) Reading in the brain: the new scientific discipline of how we read. New York: Penguin Books.

10. ↵

    Dehaene S , Le Clec'H G , Poline JB , Le Bihan D , Cohen 50 (2002) The visual word form surface area: a prelexical representation of visual words in the fusiform gyrus. Neuroreport 13:321–325. doi:10.1097/00001756-200203040-00015 pmid:11930131

11. ↵

    Dehaene S , Pegado F , Braga LW , Ventura P , Filho GN , Jobert A , Cohen 50 (2010) How learning to read changes the cortical networks for vision and language. Scientific discipline 330:1359–1364. doi:10.1126/scientific discipline.1194140 pmid:21071632

12. ↵

    Desmond JE , Glover GH (2002) Estimating sample size in functional MRI (fMRI) neuroimaging studies: statistical power analyses. J Neurosci Methods 118:115–128. doi:ten.1016/S0165-0270(02)00121-viii pmid:12204303

13. ↵

    Detre GJ , Polyn SM , Moore CD , Natu VS , Vocaliser BD , Cohen JD , Haxby JV , Norman KA (2006) The multi-voxel pattern analysis (MVPA) toolbox. .

14. ↵

    Epstein R , Ward E (2010) How reliable are visual context effects in the parahippocampal identify area? Cereb Cortex xx:294–303. doi:ten.1093/cercor/bhp099

15. ↵

    Fox CJ , Iaria G , Barton JJ (2009) Defining the face processing network: optimization of the functional localizer in fMRI. Hum Brain Mapp xxx:1637–1651. doi:ten.1002/hbm.20630 pmid:18661501

16. ↵

    Gaydos N (2003) Playful pals. Norwalk, CT: Innovative Kids.

17. ↵

    Glezer LS , Kim J , Rule J , Jiang 10 , Riesenhuber M (2015) Adding words to the brain's visual dictionary: novel word learning selectively sharpens orthographic representations in the VWFA. J Neurosci 35:4965–4972. doi:ten.1523/JNEUROSCI.4031-xiv.2015 pmid:25810526

18. ↵

    Grill-Spector M , Knouf Northward , Kanwisher N (2004) The fusiform face up expanse subserves face perception, not generic within-category identification. Nat Neurosci 7:555–562. doi:ten.1038/nn1224 pmid:15077112

19. ↵

    Hasbrouck J , Tindal GA (2006) Oral reading fluency norms: a valuable assessment tool for reading teachers. Read Teach 59:636–644. doi:10.1598/RT.59.7.three

20. ↵

    Hirshorn EA , Fiez JA (2014) Using bogus orthographies for studying cross-linguistic differences in the cognitive and neural profiles of reading. J Neurolinguistics 31:69–85. doi:ten.1016/j.jneuroling.2014.06.006 pmid:28280288

21. ↵

    Hirshorn EA , Wrencher A , Durisko C , Moore MW , Fiez JA (2016) Fusiform gyrus laterality in writing systems with unlike mapping principles: an artificial orthography training study. J Cogn Neurosci 28:882–894. doi:10.1162/jocn_a_00940 pmid:26918586

22. ↵

    Hull R , Vaid J (2007) Bilingual language lateralization: a meta-analytic tale of 2 hemispheres. Neuropsychologia 45:1987–2008. doi:10.1016/j.neuropsychologia.2007.03.002 pmid:17433384

23. ↵

    Kanwisher North , McDermott J , Chun MM (1997) The fusiform face area: a module in homo extrastriate cortex specialized for face perception. J Neurosci 17:4302–4311. doi:10.1523/JNEUROSCI.17-11-04302.1997 pmid:9151747

24. ↵

    Klein D , Mok K , Chen J-K , Watkins KE (2014) Age of linguistic communication learning shapes brain construction: a cortical thickness study of bilingual and monolingual individuals. Encephalon Lang 131:20–24. doi:10.1016/j.bandl.2013.05.014 pmid:23819901

25. ↵

    Lerma-Usabiaga G , Carreiras M , Paz-Alonso PM (2018) Converging bear witness for functional and structural segregation within the left ventral occipitotemporal cortex in reading. Proc Natl Acad Sci United states 115:E9981–E9990. doi:10.1073/pnas.1803003115 pmid:30224475

26. ↵

    Mårtensson J , Eriksson J , Bodammer NC , Lindgren Thousand , Johansson M , Nyberg L , Lövdén Thou (2012) Growth of language-related brain areas later foreign linguistic communication learning. Neuroimage 63:240–244. doi:10.1016/j.neuroimage.2012.06.043 pmid:22750568

27. ↵

    Martin L , Hirshorn EA , Durisko C , Moore MW , Schwartz R , Zheng Y , Fiez JA (2018) Do adults acquire a second orthography using their native reading network? J Neurolinguistics. .

28. ↵

    McCandliss BD , Cohen L , Dehaene Due south (2003) The visual word form area: expertise for reading in the fusiform gyrus. Trends Cogn Sci 7:293–299. doi:10.1016/S1364-6613(03)00134-vii pmid:12860187

29. ↵

    Moore MW , Brendel PC , Fiez JA (2014a) Reading faces: investigating the apply of a novel face up-based orthography in acquired alexia. Encephalon Lang 129:seven–thirteen. doi:ten.1016/j.bandl.2013.11.005 pmid:24463310

30. ↵

    Moore MW , Durisko C , Perfetti CA , Fiez JA (2014b) Learning to read an alphabet of human being faces produces left-lateralized training effects in the fusiform gyrus. J Cogn Neurosci 26:896–913. doi:10.1162/jocn_a_00506 pmid:24168219

31. ↵

    Mur M , Bandettini PA , Kriegeskorte Due north (2009) Revealing representational content with pattern-information fMRI—an introductory guide. Soc Cogn Affect Neurosci 4:101–109. doi:10.1093/scan/nsn044 pmid:19151374

32. ↵

    Nestor A , Behrmann M , Plaut DC (2013) The neural ground of visual discussion form processing: a multivariate investigation. Cereb Cortex 23:1673–1684. doi:10.1093/cercor/bhs158 pmid:22693338

33. ↵

    Petersen SE , Play a joke on PT , Snyder AZ , Raichle ME (1990) Activation of extrastriate and frontal cortical areas by visual words and word-like stimuli. Science 249:1041–1044. doi:ten.1126/science.2396097 pmid:2396097

34. ↵

    Price CJ , Devlin JT (2003) The myth of the visual word form area. Neuroimage nineteen:473–481. doi:10.1016/S1053-8119(03)00084-three pmid:12880781

35. ↵

    Rossion B , Hanseeuw B , Dricot L (2012) Defining face up perception areas in the human brain: a large-scale factorial fMRI face localizer analysis. Encephalon Cogn 79:138–157. doi:10.1016/j.bandc.2012.01.001 pmid:22330606

36. ↵

    Saygin ZM , Osher DE , Norton ES , Youssoufian DA , Beach SD , Feather J , Gaab Northward , Gabrieli JD , Kanwisher N (2016) Connectivity precedes function in the development of the visual discussion form area. Nat Neurosci 19:1250–1255. doi:ten.1038/nn.4354 pmid:27500407

37. ↵

    Szwed M , Dehaene S , Kleinschmidt A , Eger E , Valabrègue R , Amadon A , Cohen L (2011) Specialization for written words over objects in the visual cortex. Neuroimage 56:330–344. doi:10.1016/j.neuroimage.2011.01.073 pmid:21296170

38. ↵

    Talairach J , Tournoux P (1988) Co-planar stereotaxic atlas of the man encephalon. 3-Dimensional proportional system: an approach to cerebral imaging. New York: Thieme.

39. ↵

    Tan LH , Spinks JA , Feng CM , Siok WT , Perfetti CA , Xiong J , Gao JH (2003) Neural systems of 2nd linguistic communication reading are shaped by native linguistic communication. Hum Brain Mapp 18:158–166. doi:10.1002/hbm.10089 pmid:12599273

40. ↵

    Tschirner E (2016) Listening and reading proficiency levels of higher students. Foreign Lang Ann 49:201–223. doi:10.1111/flan.12198

41. ↵

    Vogel A , Petersen S , Schlaggar B (2014) The VWFA: it's not just for words anymore. Forepart Hum Neurosci eight:88. doi:x.3389/fnhum.2014.00088 pmid:24688462

42. ↵

    Wiederholt JL , Bryant BR (2001) Gray oral reading tests: GORT-four. Austin, TX: Pro-Ed.

Synthesis

Reviewing Editor: Philippe Tobler, University of Zurich

Decisions are customarily a result of the Reviewing Editor and the peer reviewers coming together and discussing their recommendations until a consensus is reached. When revisions are invited, a fact-based synthesis statement explaining their decision and outlining what is needed to set up a revision will exist listed beneath. Note: If this manuscript was transferred from JNeurosci and a conclusion was made to have the manuscript without peer review, a cursory argument to this effect will instead be what is listed below.

Thank you for a responsive revision. A couple of minor points I noticed:

Figure 2 may exist slightly disruptive for non-specialists in that the same firm symbol appears to exist used for superficially different elements in the English language (e.thousand., the 're' in 'hither' and the 'er' in 'father'). A clarifying explanation would be helpful.

Lines 304/305 include an instance of 'a the'. Please pick i or the other.

The Visual Word Form Area: Expertise for Reading in the Fusiform Gyrus

Source: https://www.eneuro.org/content/6/1/ENEURO.0425-17.2019

Share :