Patterns of cerebral activation during lexical and phonological ...

1847

Braz J Med Biol Res 38(12) 2005

Patterns of cerebral activation during readingBrazilian Journal of Medical and Biological Research (2005) 38: 1847-1856ISSN 0100-879X

Patterns of cerebral activation duringlexical and phonological reading inPortuguese

1Grupo de Neurologia Cognitiva e do Comportamento,Departamento de Neurologia, 2Instituto de Radiologia,3Instituto do Coração, Faculdade de Medicina,Universidade de São Paulo, São Paulo, SP, Brasil

M.L.H. Senaha1,M.G.M. Martin2,

E. Amaro Jr.2,C. Campi3

and P. Caramelli1

Abstract

According to the concepts of cognitive neuropsychology, there aretwo principal routes of reading processing: a lexical route, in whichglobal reading of words occurs and a phonological route, responsiblefor the conversion of the graphemes into their respective phonemes. Inthe present study, functional magnetic resonance imaging (fMRI) wasused to investigate the patterns of cerebral activation in lexical andphonological reading by 13 healthy women with a formal educationallevel greater than 11 years. Participants were submitted to a silentreading task containing three types of stimuli: real words (irregularand foreign words), nonwords and illegitimate graphic stimuli. Anincreased number of activated voxels were identified by fMRI in theword reading (lexical processing) than in the nonword reading (pho-nological processing) task. In word reading, activation was greaterthan for nonwords in the following areas: superior, middle and inferiorfrontal gyri, and bilateral superior temporal gyrus, right cerebellumand the left precentral gyrus, as indicated by fMRI. In the reading ofnonwords, the activation was predominant in the right cerebellum andin the left superior temporal gyrus. The results of the present studysuggest the existence of differences in the patterns of cerebral activa-tion during lexical and phonological reading, with greater involve-ment of the right hemisphere in reading words than nonwords.

CorrespondenceM.L.H. Senaha

Rua Montesquieu, 371/62

Chácara Klabin

04116-190 São Paulo, SP

Brasil

Fax: +55-11-3251-3698

E-mail: [email protected]

M.L.H. Senaha held a scholarship

from FAPESP (No. 99-1067-2)

from 1999 to 2001. M.G.M. Martinreceived financial support from CAPES(No. BEX1341/04-9) from 2004 to

2005.

Received March 3, 2004

Accepted August 5, 2005

Key words• Functional magnetic

resonance imaging• Reading• Lexical processing• Phonological processing• Cerebral activation

Introduction

Systematic studies of reading disturbancesbegan at the end of the 19th century with theinvestigations conducted by the neurologistsWernicke, Charcot and Déjerine. This pe-riod was marked by the search for the neuro-anatomic substrates responsible for the men-tal functions and the characterization of syn-dromes through the description and associa-

tion of symptoms (1,2). Since 1970, with theconcept of cognitive neuropsychology, nu-merous studies have attempted to under-stand the processes involved in several cog-nitive abilities (3-10). More recently, studiesinvolving resources of functional neuroim-aging have started to reintegrate the knowl-edge related to the cognitive functions withthe neural substrate. The exchange of knowl-edge between cognitive neuropsychology and

1848


M.L.H. Senaha et al.

functional neuroimaging has been illustratedin several studies (11-16).

Cognitive neuropsychology aims to un-derstand the function of a normal or dam-aged brain through models of informationprocessing. Cognitive models of readingpostulate the existence of two main routes ofprocessing, a lexical one and a phonologicalone. In the phonological route, also calledperilexical route, the segmentation from thegraphic stimuli (e.g., word) to sublexicalcomponents (e.g., letters) occurs and eachletter/grapheme is converted to its respec-tive sound/phoneme. This type of readingmay occur for low-frequency and low-fa-miliar regular words. This kind of readingprocess is essential for nonword reading, forinstance, “ced”. In the lexical route, the wordis read without the decomposition from thewritten stimuli to sublexical components,the word is read as a whole and the access tothe semantic system may happen. Only pre-viously known words can be read by thelexical route. Frequent and familiar wordstend to be processed by the lexical route andthe appropriate pronunciation of irregularwords like “pint” depends on the lexicalreading. Therefore, the linguistic variablesof the stimuli determine or facilitate the typeof reading processing to be done. Irregularwords must be read by the lexical route, andnonwords by the phonological route.

The central, linguistic disturbances ofreading can be divided into two groups: thefirst includes patients who predominantlyread using the phonological route throughthe rules of grapheme-phoneme conversion,and the second includes those who read thewords globally, that is, using the lexicalroute. Surface dyslexia belongs to the firstgroup and the deep, phonological and se-mantic dyslexias belong to the second (17-19).

Phonological dyslexia is characterizedby the relative preservation of the lexicalroute with respect to a disturbance of thephonological reading responsible for the

grapheme-phoneme conversion. For this rea-son, the individuals with phonological dys-lexia can read frequent irregular words lexi-cally whereas they are not capable of read-ing nonwords, unknown or low-frequencywords. In contrast, surface dyslexia preservesreading ability by the phonological route,while the lexical route is impaired. The indi-viduals with surface dyslexia read nonwordsand regular words through the grapheme-phoneme conversion route, but they havedifficulties in reading irregular words. Theseindividuals, due to damage to the lexicalmechanisms involved in reading, read theirregular words through the phonologicalroute, applying the grapheme-phoneme con-version rules, and consequently presentingregularizations, i.e., they produce /saδofone/when trying to read the written word <saxo-phone>. The verification of these two differ-ent types of dyslexia is considered to be anexample of double dissociation. Thus, fromthe observation of patients with surface andphonological dyslexias, it is concluded thatthe phonological and lexical routes involvedifferent cognitive mechanisms and it is as-sumed that the neural mechanisms respon-sible for these processes are at least partiallydifferent.

One of the first studies involving readingand functional neuroimaging was conductedby Petersen et al. (11). The authors used apositron emission tomography (PET) tech-nique to determine orthographic effects ineight right-handed individuals. They usedfour types of stimuli: real words (single com-mon nouns), pseudowords (letter strings thatobey the rules of English orthography), un-pronounceable consonant letter strings, andfalse letters and they identified the activa-tion of the lateral extrastriate cortex in thefour groups of stimuli. Words and pseudo-words produced activation of the medialextrastriate cortex on the left. Comparing theactivation in the passive presentation of realwords and pseudowords, the activation ofthe left frontal cortex was verified during the

1849


Patterns of cerebral activation during reading

silent reading of real words, suggesting thatthe prefrontal area on the left could be in-volved in the semantic processing of words.

Rumsey et al. (14), in another PET study,examined the orthographic (lexical) and pho-nological processing of 14 right-handed men.Even in reading aloud tasks as well as inlexical decision, activation was lateralizedto the left hemisphere, including the follow-ing areas: lingual and fusiform gyri, peri-Rolandic cortex, thalamus, and anterior cin-gulate. Reading aloud real words andnonwords activated the left superior tempo-ral gyrus with a more significant activationduring the phonological than the lexical (or-thographic) reading. The authors concludedthat their results supported the connectionistmodels of reading since they did not revealdifferences in the activation during the read-ing of real words and nonwords, and sug-gested that the processing of real words aswell as of nonwords occurs inside a commonneural network.

Different results were obtained in an-other study (20) using functional magneticresonance imaging (fMRI), in which 38 in-dividuals were submitted to four differenttasks involving graphic stimuli. One of thetasks involved phonological processing (non-word rhyme judgment) and another involvedlexical-semantic processing (semantic cat-egory judgment). The researchers observedactivation during phonological processingin the inferior frontal gyrus and temporallobe (with unilateral activation of the infe-rior frontal gyrus in men, but bilateral acti-vation in women). During the lexical-se-mantic processing, the important areas werethe middle and superior temporal gyri ofboth hemispheres.

Considering the Cognitive Neuropsychol-ogy approach regarding the different read-ing processes in patients with brain lesions -the existence of double dissociation (phono-logical dyslexia versus surface dyslexia) sug-gests that the phonological and lexical routesinvolve different cognitive mechanisms and,

probably, different neural substrates are re-sponsible for these processes. The objectiveof the present study was to investigate thepatterns of cerebral activation by fMRI dur-ing lexical and phonological reading in 13healthy females. We hypothesized that thecerebral activation patterns differ accordingto the different types of reading process.

Subjects and Methods

Subjects

Thirteen healthy right-handed womenaged 18-36 years (mean = 25.2 years) par-ticipated in the study. All individuals werereaders of the Portuguese language, with aminimum of 12 years of education and withthe maximum possible mark in the EdinburghHandedness Questionnaire (21). No historyof neurological disease or of learning diffi-culties was reported and the subjects werenot using any drug with action on the centralnervous system. Subjects were selectedthrough the application of a questionnairethat contained questions to determine if theyfulfilled the inclusion criteria mentionedabove. A brief reading task was applied forthe exclusion of non-fluent readers. The studywas approved by the Ethics Committee ofthe University Hospital and the individualsselected gave written informed consent toparticipate.

Tasks

Individuals performed a silent readingtask during the acquisition of the images byfMRI. This silent reading task presented anABC design whose objective was to identifythe activation of lexical and phonologicalreading in the alphabetic system. The ABCdesign corresponds to a task that is com-posed of three different lists. Condition Acorresponded to list 1 (real words - irregularand foreign words), condition B to list 2(nonwords) and condition C to list 3 (non-

1850



readable graphic stimuli; see Table 1 for anexample).

The first list (list 1) was composed of 50real words, half of them consisting of irregu-lar words in Portuguese and the other halfconsisting of foreign words. The wordsranged from four to nine letters in length andall were concrete nouns. The stimuli of thislist containing irregular words and foreignwords, should be read by lexical processing.

The second list (list 2) consisted of 50nonwords written in the alphabetic systemand was elaborated from the stimuli of thefirst list. The consonants, but not the vowels,of each real word were modified, withnonwords with the same structure and syl-labic extension thus being obtained. For in-stance: starting from the irregular word axila,the nonword abida was generated. The read-ing of nonwords should be accomplished byphonological processing, that is, by graph-eme-phoneme conversion processing.

The third list (list 3) had 50 illegitimategraphic stimuli. These stimuli were not read-able because they were formed by pseudo-letters obtained through distortions of theletters. The illegitimate graphic stimuli wereelaborated from the second list, with thelines of the letters being modified and servedas controls for the stimuli of lists 1 and 2.Therefore, in the analysis of the images fromfMRI, these illegitimate graphic stimuli wereused as a baseline condition, containing ele-ments of visual and pre-linguistic processing.

As described above, the silent readingtask resulted from the combination of threedifferent lists, each containing 50 stimuli,involving real words, nonwords and illegiti-mate graphic stimuli. The 50 stimuli of each

list were distributed into five blocks of 10stimuli each. Therefore, the task was dividedinto 15 blocks that were randomly presented.To minimize possible effects from habitua-tion to the type of stimulus, each epochcontained 80% of the stimuli belonging toone condition and 10% of the stimuli be-longing to each of the remaining conditions.

Preparation of the stimuli for the reading task

The graphic stimuli written in the alpha-betic system were typed using the MicrosoftWord program (Microsoft, Seattle, WA,USA), in capital letters of the Arial font type,size 90. All stimuli were transported fromthe Word to the Paint program where theyappeared against a black background. Theillegitimate graphic stimuli generated fromthe nonword letter distortions were elabo-rated in the Paint program using the “eraser”feature to fade part of the letters. Stimuliwere presented as bitmap files to be used bythe Visual Basic program version 4.0 (Mi-crosoft).

The duration of stimulus presentation andthe interval between stimuli were 1000 ms.Stimulus presentation and image acquisitionwere synchronized via a TTL pulse (Zurc &Zurc DataSystems, São Paulo, SP, Brazil)and the stimuli were projected by a multime-dia projector. The subjects visualized thestimuli through a mirror (each letter sizeoccupied 4" degrees in the visual field) at-tached to the head coil, and were instructedto read them silently.

After the silent reading task during theacquisition of the images, in order to verifyif the subjects had adhered to the experi-ment, they were asked if they rememberedthe stimuli they had read. To confirm thecapacity of lexical and phonological read-ing, a post-test of word and nonword readingaloud was applied. All individuals had ad-hered to the experiment and were perfectlycapable of reading through lexical and pho-nological processes.

Table 1. Stimuli used in the silent reading task.

List Type of stimulus Reading processing Examples

1 Words (irregular and foreign words) Lexical AXILA2 Nonword Phonological ABIDA

3 Illegitimate graphic stimuli Without reading

1851



Image acquisition and data processing

The images were obtained in a Horizon1.5 Tesla magnet of the General ElectricMedical Systems (Milwaukee, MN, USA),equipped with a 40 mT/m gradient, using theblood oxygen level-dependent (BOLD) con-trast technique with echo planar imaging.Fifteen slices were collected. The imageswere oriented on the anterior commissure-posterior commissure line, with 7-mm thick,0.7-mm gap slices covering the whole brain.The time of repetition was 2000 ms, the timeto echo was 40 ms, the field of view was 400x 200 mm, with a 90º flip angle and a matrixof 128 x 64 voxels. A total of 2250 imageswere collected for each subject. Analysiswas performed with a personal computerrunning Linux.

The data were first realigned (22) to mini-mize motion-related artifacts and smoothedwith a Gaussian filter (full-width half-maxi-mum - 7.2 mm). Responses to the experi-mental paradigms were then detected bytime-series analysis using gamma variatefunctions (peak responses at 4 and 8 s) tomodel the BOLD response. The analysiswas set up as follows. First, each experimen-tal condition was convolved separately withthe 4- and 8-s Poisson functions to yield twomodels of the expected hemodynamic re-sponse to that condition. The weighted sumof these two convolutions that gave the bestfit to the time series at each voxel was thencomputed. This weighted sum effectivelyallows voxel-wise variability in time to peakhemodynamic response. Following this fit-ting operation, a ratio of the sum of squaresof deviations from the mean intensity valuedue to the model (fitted time series) dividedby the sum of squares due to the residuals(original time series minus model time se-ries) was calculated. This statistic is calledthe SSQratio. The percent change of theBOLD signal at each voxel was also calcu-lated. This was done by ((fitmax – fitmin)/mean signal intensity) x 100, where fitmax

and fitmin were the maximum and minimumvalues of the fitted response for the timeseries in question. In order to sample thedistribution of the SSQratio under the nullhypothesis that observed SSQratio valueswere not determined by experimental design(with minimal assumptions), the time seriesat each voxel was permuted using a wavelet-based resampling method described in detailby Bullmore et al. (23). This process wasrepeated 10 times at each voxel and the datawere combined over all voxels, resulting in10 permuted parametric maps of the SSQratioon each plane for each participant. The samepermutation strategy was applied to eachvoxel to preserve spatial correlational struc-ture in the data during randomization. Com-bining the randomized data over all voxelsyields the distribution of the SSQratio underthe null hypothesis. A test that any givenvoxel is activated at any required type I errorcan then be carried out by obtaining theappropriate critical value of the SSQratiofrom the null distribution. For example,SSQratio values in the observed data lyingabove the 99th percentile of the null distri-bution have a probability of ≤0.01 under thenull hypothesis. We have shown that thispermutation method gives very good type Ierror control with minimal distributional as-sumptions (23,24).

After these processes, two activationmaps were created for each individual, oneof them comparing real words to illegitimategraphic stimuli (A x C), and the other com-paring nonwords to illegitimate graphicstimuli (B x C). The objective of this com-parison was to study the word and nonwordsilent reading by eliminating (subtracting)the activations of the neural structures re-sponsible for the primary and pre-linguisticvisual processing. Consequently, the A x Cactivation map represents the activation dur-ing the silent reading of words and the B x Cmap represents activation during the silentreading of nonwords.

Each activation map was composed of 15

1852



slices (Figure 1). To identify the anatomicallocalization of the activated areas, each slicewas analyzed visually, determining whichareas were activated using as reference theAtlas of Duvernoy (25). The activated areaswere identified independently using anAtlas by two examiners (MLHS and PC).The points of disagreement between the twoanalyses were discussed by the examiners,who then reached a consensus. Based onthese data, two tables were elaborated, onefor each comparison (A x C and B x C) forthe areas activated in each subject. The num-ber of activated voxels was recorded in allactive clusters resulting from the A x C andB x C comparisons.

Initially we determined the frequency ofactivation in different brain areas for eachindividual in each specific condition. Thisprocedure provided an overview of all pos-sible areas detected in the subjects. Since notall areas were present in all subjects, thefrequency of activation detected in each areaprovides a measure of the commonalities inthe neural correlates of each task found inthe participants. Areas displaying activation

in more than 50% of the individuals werejudged to represent relevant activations. Thisapproach has been used in past publicationsby others, and we believe it represents acompromise between subject variability andgroup-shared activation foci (26,27).

Results

The comparison of the number of acti-vated voxels in all individuals in the condi-tions A x C (words x illegitimate graphicstimuli) and B x C (nonwords x illegitimategraphic stimuli) showed a larger number ofvoxels activated in the first condition, indi-cating a greater activation in word than innonword reading. Furthermore, in both condi-tions the left cerebral hemisphere was moreactivated than the right hemisphere (Table 2).

Table 3 shows the areas that were acti-vated in more than 50% of the individuals inthe conditions A x C and B x C in the rightand left hemispheres. A larger number ofactivated common areas can be observed inthe silent reading of words than of nonwordsin both hemispheres. In the first condition,i.e., word silent reading, a mirror correspon-dence of activations was observed when com-paring the left and right hemispheres. Thefollowing areas were activated: superior,middle and inferior frontal gyri and the su-perior temporal gyrus in both cerebral hemi-spheres. The cerebellum was activated in theright hemisphere and the precentral gyrus inthe left. In the condition B x C, the rightcerebellum and the left superior temporalgyrus were activated in more than 50% ofthe individuals.

Discussion

Differences in the cerebral activationpatterns of lexical and phonological reading

The activation patterns detected in sub-jects during word and nonword reading weredifferent. In the word reading task, we iden-

Figure 1. Brodmann areas (BA) are shown in circles for two different analysis from the samesubject: A, areas detected when comparing real words (lexical processing) versus illegiti-mate graphic stimuli; B, areas detected when comparing nonwords (phonological process-ing) versus illegitimate graphic stimuli. Cer = cerebellum.

1853



tified ten different activated areas in morethan 50% of the individuals, five of them inthe left hemisphere and five in the right. Inthe reading of nonwords, only two activatedareas were found in more than 50% of theindividuals. During word reading, a similarpattern of activation was observed in the leftand right hemispheres. In other words, of thefive areas activated in each hemisphere, fourwere activated bilaterally: superior, middleand inferior frontal gyri, and superior tem-poral gyrus. In addition to these areas, theright cerebellum and the left pre-central gy-rus were also activated. In the reading ofnonwords, only the right cerebellum and theleft superior temporal gyrus were activated.The mirror pattern of activation found inword reading was not observed in nonwordreading. The areas activated during the read-ing of nonwords, the right cerebellum andthe left superior temporal gyrus, were alsoactivated during the reading of words.

Petersen et al. (11) and Pugh et al. (20)have also reported different patterns of cere-bral activation when comparing the mechan-isms involved in word and nonword reading.Petersen et al. (11), in a study with PET,observed that silent reading of words acti-vated the left frontal cortex. The medialextrastriate visual cortex in the left hemi-sphere was activated during the silent read-ing of both words and nonwords. In thestudy of Pugh et al. (20), using fMRI, 38individuals completed a judgment task ofphonological processing (nonword rhymejudgment) and lexical-semantic processing(semantic category judgment). The research-ers observed activation during phonologicalprocessing in the inferior frontal gyrus andin the temporal lobe (with lateralization ofthe activation of the inferior frontal gyrus tothe left in men, and bilateral activation inwomen) and activation of the superior andmiddle temporal gyri in both hemispheresduring the lexical-semantic processing. Incontrast, in a PET study on 14 right-handedmen, Rumsey et al. (14) did not find differ-

ences in activation during tasks of lexicaland phonological reading. The reading ofirregular words as well as of nonwords acti-vated the following areas: left lingual andleft fusiform gyri, cerebellum, superior tem-poral gyrus bilaterally, left thalamus, peri-Rolandic cortex (pre- and post-central gyri)bilaterally, anterior/mid cingulate gyrus onthe left, and left insula. Price et al. (13), inanother study with PET and reading of realwords, found activation during silent read-ing in the following areas: medial extrastri-ate cortex on the left, middle temporal cortexon the left, middle frontal cortex on the left,and middle temporal cortex and inferior fron-tal cortex on the right.

The comparison of our data with thosereported by other researchers should be madewith caution. One of the reasons for this isrelated to the use of different tasks to probe

Table 2. Number of voxels activated in the word and nonword silent reading.

Condition A x C Condition B x CWords Nonwords

Right cerebral hemisphere 794 voxels 446 voxelsLeft cerebral hemisphere 874 voxels 526 voxels

Condition A = list 1; condition B = list 2; condition C = list 3. See Table 1 for explanationof lists.

Table 3. Areas activated in more than 50% of the individuals during the silent wordand nonword reading in the right and left hemispheres.

Condition A x C Condition B x CWords Nonwords

Right hemisphere Superior frontal gyrus (10) Cerebellum (9)Medium frontal gyrus (10)

Cerebellum (9)Inferior frontal gyrus (8)

Superior temporal gyrus (7)

Left hemisphere Superior frontal gyrus (8) Superior temporal gyrus (7)Medium frontal gyrus (7)Inferior frontal gyrus (7)

Superior temporal gyrus (7)Precentral gyrus (7)

The numbers in parentheses refer to the number of subjects of the 13 in the study whopresented activation in the area. Condition A = list 1; condition B = list 2; condition C =list 3. See Table 1 for explanation of lists.

1854



the same function. In the investigations men-tioned above, the activation during readingwas confirmed through similar but not iden-tical tasks, i.e., silent reading, reading aloud,lexical decision, and judgment. Probably,the execution of these tasks involves similarand different mechanisms. In this way, thecerebral areas activated during the perfor-mance of these tasks should be different.Even the sensitivity of the paradigm caninfluence the data. Price et al. (13) varied theduration of stimulus presentation (150 and981 ms) in tasks of silent reading, readingaloud and lexical decisions and showed thatthe time of exhibition, significantly influ-enced the correlation between blood flowand word reading. The divergence of theresults obtained in the different studies is animportant issue to be considered in studiesthat involve functional neuroimaging andmental functions and should be addressedby additional studies in the field.

Right cerebral hemisphere and word reading

An interesting aspect of the results ob-tained in the present study is the mirrorcorrespondence of the areas that were acti-vated. This pattern was only observed for thesilent reading by the lexical route, that is, forthe processing of irregular and foreign words.From these findings the following questioncould be raised: does the right cerebral hemi-sphere play a role in lexical reading?

In reality this issue has already been dis-cussed in the literature for some years (28-30). Some researchers defend the “right hemi-sphere hypothesis”, which proposes that theright cerebral hemisphere plays a role in theorthographic and lexical processes of read-ing, permitting the processing of highimaginability and high frequency writtenwords. This hypothesis is supported by casestudies of left hemispherectomy, split-braincases and patients with deep dyslexia andpure alexia.

Patterson et al. (31) described the case of

patient NI, who, after a normal childhood,developed refractory epilepsy, needing thesurgical removal of the left cerebral hemi-sphere. After the hemispherectomy, NI pre-sented reading residual abilities characteris-tic of deep dyslexia. She was capable ofreading some concrete and high frequencywords, making some semantic mistakes.Zaidel and Peters (32) described two split-brain patients who were able to associatewith objects the written word presented tothe right hemisphere. Coslett and Saffran(33) reported four patients with pure alexiawhose performance in tasks of lexical deci-sion and semantic categorization was con-sistent with the right hemisphere hypothesis.Some investigators relate the residual read-ing abilities of patients with deep dyslexia tothe orthographic and semantic aspects ofreading by the right hemisphere. In this con-dition, the patients present an extensive le-sion of the left hemisphere and are able toread some frequent words through lexicalprocessing, which would be an indicationthat in these patients with deep dyslexiareading is being accomplished by the righthemisphere.

Coslett and Monsul (29) conducted a studyusing transcranial magnetic stimulation(TMS) in a case of pure alexia (JG) to test the“right hemisphere hypothesis”. The authorscompared the reading performance of JG inthree different situations: without TMS, withTMS on the right parietal lobe and with TMSin the same area on the left. Having as a basethe performance in reading without TMS,the authors compared the results of the influ-ence of TMS on the right and the left parietallobe and verified that the reading perfor-mance of JG was impaired only when theright cerebral hemisphere was stimulated,supporting the idea that the residual readingof that patient was due to a linguistic abilityof the non-dominant hemisphere.

The results obtained in the present studyregarding the mirror pattern of activation inthe left and right cerebral hemispheres dur-

1855



ing real word reading support this idea thatthe right cerebral hemisphere has a role inlexical reading.

The role of the cerebellum

In this study, the right cerebellar hemi-sphere was activated in both the word andnonword reading tasks in more than half thesubjects. These data suggest the participa-tion of the cerebellum in reading, corrobo-rating the findings from the study of Fulbrightet al. (34). The contribution of the cerebel-lum to cognitive processing was discussedby these researchers, who studied the cer-ebellar activation by fMRI during readingand concluded that the cerebellum is en-gaged not only in the ability of phonologicassembly but also in semantic processing.

Limitations and perspectives

As in other studies that use fMRI, one ofthe limitations of the present study was re-lated to the small number of subjects evalu-ated. We tried to minimize this problem byselecting right-handed female volunteers witha high educational level to obtain a morehomogeneous sample.

Another limitation refers to individualdifferences in the performance of cognitivetasks. The acquisition of reading, differentlyfrom the acquisition of the oral language, isnot an innate ability of humans. Since read-ing is a culturally biased function it mightsuffer a larger inter-individual variabilitythan the other innate and primary cognitiveabilities. In the present study, we chose thesilent reading task to determine the patternof cerebral activation because of our interestin examining the complete process of read-ing from the input of visual information tophonological codification, without the re-quirement of speech articulation that couldprovoke the appearance of artifacts in thefMRI images. Some studies of reading useother tasks, such as lexical decision and

semantic categorization of written words,that test some isolated mechanisms involvedin reading. Leff et al. (35) consider silentreading to be a natural function, differentfrom tasks of lexical decision or of choice,that imply the activation of other associatedmental processes difficult to control for eachindividual. If, on the one hand, it seems tohave advantages over the choice of the silentreading task, it also has disadvantages. Whenusing the silent reading task during imageacquisition by fMRI we cannot be certain ofthe volunteer’s collaboration at the momentof scanning. In the present study, to confirmthat the silent reading task was actually per-formed, soon after the experiment, individu-als had to mention some stimuli that hadbeen presented.

Some studies have investigated cerebralactivation using a limited number of slices,without a global vision of the brain. In ourstudy, the 15 slices covered the whole brain.This type of technical approach became avail-able recently. It facilitates not only the obser-vation of the activation of areas of interest, butalso the inspection of all the areas of the brain.

In spite of the limitations and controver-sies generated from the cerebral activationmaps related to the cognitive abilities, stud-ies with the use of fMRI contribute and willcontinue to contribute to the understandingof the cognitive and neural mechanisms re-lated to the mental functions. fMRI, togetherwith other experimental approaches usingadvanced techniques like PET and TMS, aswell as lesion studies, should provide rel-evant information about reading processes.

According to our hypothesis, the resultsreported here concerning brain activationdemonstrate differences in the activationpatterns associated with lexical and phono-logical reading. However, we believe it isstill too early to know if these differentpatterns correspond to two independent neu-ral networks, one responsible for the lexicalprocessing of reading and the other for thephonological processing.

1856



Acknowledgments

We wish to thank Antônio Cesário Mon-teiro da Cruz Junior for providing the elec-

tronic instruments necessary for synchro-nizing stimulus presentation and data acqui-sition, and for the network configuration.

References

1. Hécaen H & Albert ML (1977). Human Neuropsychology. Wiley-Interscience, New York.

2. Black SE & Behrmann M (1994). Localization in alexia. In: Kertesz A(Editor), Localization and Neuroimaging in Neuropsychology. Aca-demic Press, San Diego, CA, USA, 331-376.

3. Lecours AR, Delgado AP & Pimenta MAM (1993). Distúrbiosadquiridos da leitura e da escrita. In: Mansur LL & Rodrigues N(Editors), Temas em Neurolingüística. Tec Art, São Paulo, SP,Brazil, 45-62.

4. McCarthy RA & Warrington EK (1990). Cognitive Neuropsychology:A Clinical Introduction. Academic Press, San Diego, CA, USA.

5. Lesser R & Milroy L (1993). Linguistics and Aphasia: Psycholinguis-tic and Pragmatic Aspects of Intervention. Longman, New York, 52-80.

6. Ellis AW & Young AW (1988). Human Cognitive Neuropsychology.Lawrence Erlbaum, London, UK.

7. Price CJ (2000). Functional imaging studies of aphasia. In: MazziottaJC, Toga AW & Frackowiak RS (Editors), Brain Mapping: the Disor-ders. Academic Press, San Diego, CA, USA, 181-200.

8. Basso A (2000). The aphasias: fall and renaissance of the neuro-logical model? Brain and Language, 71: 15-17.

9. Marshall JC & Newcombe M (1973). Patterns of paralexia: apsycholinguistic approach. Journal of Psycholinguistic Research, 2:175-200.

10. Caramelli P, Parente MAMP, Hosogi ML et al. (1994). Unexpectedreading dissociation in a Brazilian “nisei” with crossed aphasia.Behavioural Neurology, 7: 165-170.

11. Petersen SE, Fox PT, Snyder AZ et al. (1990). Activation of extra-striate and frontal cortical areas by visual words and word likestimuli. Science, 249: 1041-1044.

12. Small SL, Flores DK & Noll DC (1998). Different neural circuitssubserve reading before and after therapy for acquired dyslexia.Brain and Language, 62: 298-308.

13. Price CJ, Wise RJ, Watson DG et al. (1994). Brain activity duringreading. The effects of exposure duration and task. Brain, 117:1255-1269.

14. Rumsey JM, Horwitz B, Donohue BC et al. (1997). Phonological andortographic components of word recognition: a PET-rCBF study.Brain, 120: 739-759.

15. Henson RNA, Burgess N & Frith CD (2000). Recoding, storage,rehearsal and grouping in verbal short-term memory: an fMRI study.Neuropsychologia, 38: 426-440.

16. Price CJ, Gorno-Tempini ML, Graham KS et al. (2003). Normal andpathological reading: converging data from lesion and imaging stud-ies. Neuroimage, 20: S30-S41.

17. Coltheart M, Patterson KE & Marshall JC (Editors) (1980). DeepDyslexia. Routledge, Londres, UK.

18. Coltheart M, Masterson J, Byng S et al. (1983). Surface dyslexia.Quarterly Journal of Experimental Psychology, 35A: 469-495.

19. Lecours AR & Parente MAMP (1997). Dislexia: Implicações doSistema de Escrita do Português. Artes Médicas, Porto Alegre, RS,Brazil.

20. Pugh KR, Shaywitz BA, Shaywitz SE et al. (1996). Cerebral organi-zation of component process in reading. Brain, 119: 1221-1238.

21. Oldfield RC (1971). The assessment and analysis of handedness:the Edinburgh Inventory. Neuropsychologia, 9: 97-113.

22. Bullmore E, Brammer MJ, Rabe-Hesketh S et al. (1999). Methodsfor diagnosis and treatment of stimulus-correlated motion in genericbrain activation studies using fMRI. Human Brain Mapping, 7: 38-48.

23. Bullmore E, Long C, Suckling J et al. (2001). Colored noise andcomputational inference in neurophysiological (fMRI) time seriesanalysis: resampling methods in time and wavelet domains. HumanBrain Mapping, 12: 61-78.

24. Bullmore E, Fadili J, Breakspear M et al. (2003). Wavelets andstatistical analysis of functional magnetic resonance images of thehuman brain. Statistical Methods in Medical Research, 12: 375-399.

25. Duvernoy HM (1991). The Human Brain-Surface, Three-Dimen-sional Sectional Anatomy and MRI. Springer-Verlag, Vienna, Aus-tria.

26. Uchida I, Kikyo H, Nakajima K et al. (1999). Activation of lateralextra-striate areas during orthographic processing of Japanese char-acters studied with fMRI. Neuroimage, 9: 208-215.

27. Wood AG, Harvey AS, Wellard RM et al. (2004). Language cortexactivation in normal children. Neurology, 63: 1035-1044.

28. Coltheart M (2000). Deep dyslexia is right-hemisphere reading.Brain and Language, 71: 299-309.

29. Coslett HB & Monsul N (1994). Reading with the right hemisphere:evidence from transcranial magnetic stimulation. Brain and Lan-guage, 46: 198-211.

30. Ellis AW (1995). Leitura, Escrita e Dislexia: Uma Análise Cognitiva.Artes Médicas, Porto Alegre, RS, Brazil.

31. Patterson K, Vargha-Khadem F & Polkey C (1988). Reading withone hemisphere. Brain, 112: 39-63.

32. Zaidel E & Peters AM (1981). Phonological encoding and ideo-graphic reading by the disconnected right hemisphere: two casestudies. Brain and Language, 14: 205-234.

33. Coslett HB & Saffran EM (1989). Evidence for preserved reading in“pure alexia”. Brain, 113: 327-329.

34. Fulbright RK, Jenner AR, Mencl WE et al. (1999). The cerebellum’srole in reading: a functional MR imaging study. American Journal ofNeuroradiology, 20: 1925-1930.

35. Leff AP, Crewes H, Plant GT et al. (2001). The functional anatomy ofsingle-word reading in patients with hemianopic and pure alexia.Brain, 124: 510-521.

Patterns of cerebral activation during lexical and phonological ...

Documents

Transcript of Patterns of cerebral activation during lexical and phonological ...