http://ldx.sagepub.com/Journal of Learning Disabilities

http://ldx.sagepub.com/content/34/6/566The online version of this article can be found at:

 DOI: 10.1177/002221940103400609

2001 34: 566J Learn DisabilJane Joseph, Kimberly Noble and Guinevere Eden

The Neurobiological Basis of Reading  

Published by:

  Hammill Institute on Disabilities

and

http://www.sagepublications.com

can be found at:Journal of Learning DisabilitiesAdditional services and information for     

  http://ldx.sagepub.com/cgi/alertsEmail Alerts:

 

http://ldx.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://ldx.sagepub.com/content/34/6/566.refs.htmlCitations:  

What is This? 

- Nov 1, 2001Version of Record >>

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

The NeurobiologicalBasis of ReadingJane Joseph, Kimberly Noble, and Guinevere Eden

Abstract

The results from studies using positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) in adults havelargely revealed the involvement of left-hemisphere perisylvian areas in the reading process, including extrastriate visual cortex, inferiorparietal regions, superior temporal gyrus, and inferior frontal cortex. Although the recruitment of these regions varies with the partic-ular reading-related task, general networks of regions seem to be uniquely associated with different components of the reading process-For example, visual word form processing is associated with occipital and occipitotemporal sites, whereas reading-relevant phonologicalprocessing has been associated with superior temporal, occipitotemporal and inferior frontal sites of the left hemisphere. Such findingsare evaluated in light of the technical and experimental limitations encountered in functional brain imaging studies, and the implicationsfor pediatric studies are discussed.

n this society, where the quality oflife is in large part determined byliteracy, there has been a growing

interest in understanding readingmechanisms and their development indiverse cultures. Until fairly recently,the neurological basis for reading hadin most part been developed from casestudies of brain lesion patients with se-lective reading defidts. 4s functionalbrain imaging tools have become morewidely accessible, investigations intothe neuroanatomical organization ofreading and language hale becomemore frequent, and the informationgained about reading has become moredetailed. Together, these lines of inves-tigation have led to the development oftheories of the modularity of languageand reading. Today, functional brainimaging has a distinct advantage withits ability to noninvasively reveal in-formation that will contribute to the

study of reading acquisition and de-velopment. Although most of the brainimaging studies of reading to date

have used adults, it is now possible toperform functional brain imagingstudies on children- This information iscrucial for an understanding of read-ing achievement, reading strategies,

and individual variation in masteringthe reading process.This article provides an overview of

the current findings in functional neuro-imaging research regarding the spatiallocalization of reading and reading-related tasks in the brain. Althoughmany of the studies described in thisarticle were designed to study lan-

guage processing in a more generalsense, most involved the presentationof written material, which allows us tomake inferences about the functional

specialization for single word pro-cessing. The various components ofthe reading process that are discussedhave been extracted from existingmodels of reading, thereby providing abroad overview of the neurobiologicalbasis of the reading process. However,we limit this review to neuroimagingstudies of word decoding and do notfully address related processes such ascomprehension, auditory language pro-cessing, working memory, and atten-tion. Although significant progress hasbeen made since the earliest neuro-

imaging studies of the reading pro-cess (Ingvar & Schwartz, 1974), manyquestions remain about the cortical spe-cialization for various cognitive com-

ponents of reading. The present dis-cussion attempts to find a consensuswhere possible. However, we also con-sider the limitations inherent in thevarious neuroimaging techniques aspossible reasons for apparent discrep-andes across studies. The results fromthese adult studies provide a usefulfoundation for the investigation ofreading acquisition in children, whichis currently still in its infancy.

Language and the Brain

Investigators have attributed language-specific processing to the areas thatsurround the St-1~-ian fissure, known asthe perisylvian areas (see Figure 1).One of these areas, the temporopanetalcortex, receives projections containingbut not limited to visual and auditoryinformation-The posterior superior tem-poral gyrus, or Wernicke’s area, hasrepeatedly been associated with a vari-ety of language functions, usually in-volved in comprehension. Wernickeproposed, based on his observations inaphasia patients, that this area was

specialized for auditory word recogni-tion (Wernicke, 1887). Inferior parietal

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

567

FIGURE 1. Lateral view of the left hemisphere of the human brain. For all figuresin this article, the anterior portion of the brain appears on the left, and the superiorsurface appears at the top of the figure. Specific anatomical and functional regions(underlined) implicated in reading and reading-related tasks are indicated with la-bels. The Sylvian fissure is represented by a dotted line along the superior surfaceof the temporal lobe. AG = Angular gyrus; extrastriate = extrastriate visual cortex;IFG = inferior frontal gyrus (includes Broca’s area); INS = insula; ITG = inferior tem-poral gyrus; MFG = middle frontal gyrus; MTG = middle temporal gyrus; SFG =superior frontal gyrus; SMG = supramarginal gyrus; STG = superior temporal gyrus(includes Wernicke’s area). Although the insula (INS) is represented on the lateralsurface of the brain in this figure, this cortical area actually lies buried in the depthsof the Sylvian fissure.

sites, such as the supramarginal gyrusand the angular gyrus, have beenwidely associated with written andspoken language comprehension. De-jerine argued that a disconnection ofthe angular gyrus from the primary vi-sual areas would result in alexia, anddirect damage to this area results inalexia with agraphia (Dejerine, 1892).Consequently, Dejerine proposed theangular gyrus to be the site of writtenlanguage. The frontal lobe is also at-tributed with certain types of languageprocessing. Most notable are the infe-rior frontal gyrus, which includes

Broca’s area, and the dorsolateral pre-frontal cortex. These areas are gen-erally associated with organization,manipulation, and production of lan-guage, as well as grammar and syntax.However, it is probably oversimplisticto describe temporoparietal areas asthose responsible for the reception oflanguage whereas frontal regions areresponsible for expressive language.On the contrary, it is most likely that adistributed network is responsible forfull coherence of the language system.Although the aforementioned left

hemisphere regions are those that are

classically associated with language, theanalogous structures in the right hemi-sphere also appear to play a role. More-over, noncortical structures such as the

cerebellum, the cingulate, and the

parahippocampal region have all beenassociated with various aspects of lan-

guage processing. The functional neu-roimaging studies reviewed here im-plicate a very wide network of regionsassociated with the reading process.These brain regions are referred to

using the anatomical labels providedin Figure 1.

Physiological Basis of PETand fMRl

As proposed by Roy and Sherrington(1890), local cerebral hemodynamics inthe brain are closely linked to localneuronal activity. Positron emission to-mography (PET) allows the quantita-tive measurement of regional cerebralblood flow and blood volume. To as-sess task-related activity, which under-lies sensory and cognitive processes,1’O-water and butanol are commonlyinjected intravenously as freely diffu-sible tracers. The increase of regionalcerebral blood flow (rCBF) in areas ofhigh neuronal activity that are asso-ciated with the performance of a cog-nitive or sensory process can then bemonitored by measuring the abun-

dance of the tracer. This local increaseassociated with the task is typicallycompared to a baseline or resting con-dition that is believed not to engage thetask under study (see Figure 2).

Functional magnetic resonance imag-ing (fMRI) can be achieved because ofthe paramagnetic properties of hemo-globin in its deoxygenated state. A lo-cal blood flow increase introduces more

oxygenated blood than is required bythe tissue, resulting in a relative de-crease in deoxyhemoglobin (Belliveauet al., 1991; Ogawa, Lee, Nayak, & Glynn,1990; Turner, Le Bihan, Moonen, &

Frank, 1991). The relative concentra-tion of deoxygenated hemoglobin rela-tive to oxygenated hemoglobin acts asan endogenous, intravascular, para-

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

568

FIGURE 2. A typical task design for PET and fMRI studies. In each case, twoexperimental tasks (Tasks A and B) and one rest task are used.

magnetic contrast agent b~- affectingthe image intensify.PET and &1RI techniques are similar

in that they are based on the phenom-enon of increased local blood flow dur-

ing increased neuronal activity- To as-sess task-related signal change, bothtypes of experiments usually employtask and control (resting) conditions.The signal changes that are associatedwith these conditions are contrasted to

identify areas of the brain that showunique activation only during the taskcondition (see Figure 2). The accuracyof the resulting statistical maps in de-termining the location of task-relatedactivity is, of course, dependent on thecognitive components involved andthe strategies used during the task andcontrol conditions. As we shall see,

there are a number of ways to go aboutthis for a task such as reading.

Component Processesof ReadingTo study the neuronal correlates of cog-nition, investigators often aim to teaseapart the components of a cognitiveprocess. The complex process of read-ing involves several distinct compo-nents. Not all models of reading agree

on the number of components, the con-nectiviity among the components (Iscomponent A connected to component tB?), or the flow of information amongcomponents (Is information relayedbetween components in a serial or par-allel fashion? e.g., Nfcclevand, 1979).However, in this article, we provides asurvey of functional neuroimagingstudies that deal with the componentprocesses of reading that are typicallyshared among various models. Specif-ically, the components of the readingprocess that are reviewed in this articleinclude processing of visual- wordforms, lexical orthography, semanticinformation associated with words,lexical and sublexical phonology, andphonological and phonetic encoding,each of which is addressed in a sepa-rate section.

Functional NeuroimagingStudies of Reading

Visual Word Forin ProcessingThis component of reading involves vi-sual analysis of letter and word stimulito distinguish them from other visualpatterns. Letters are translated into ab-

stract representations, which preservethe basic form of the letter despite st~·1-istic differences such as size, font, orscript The tasks that have been used toisolate brain regions involved in visualword form processing have includedpassive viewing (Indefrev et al., 1997;Menard, Kosslvn, Thompson, Alpert,& Rauch, 1996; Petersen, Fox, Posner,.-lintun, & Raichle, 1989; Petersen, Fox,Snyder, & Raichle, 1990; Puce, AUison,~sgani, Gore, & McCarthy, 1996); fea-ture, color, or letter detection (Frith,Kapur, Friston, Liddle, & Frackowiak,1995; Kapur et al., 1994; Price, Wise, &

Frackowiak, 1996); and visual match-ing (Kuriki, Takeuchi, & Hirata, 1998;Pugh et al., 1996). Studies have em-ployed words, pseudowords, letter

strings, false fonts (i.e., letter-lilce stim-uli), or symbols. Figure 3 summarizesthe loci of activation across these stud-ies. Some of these studies have shown

greater activation in the left extrastri-ate cortex than in the right, using stim-uli containing letters (e.g., Petersen et at.,1989; Price, WISer, & Frackowiak, 1996;Puce et al., 1996). Also, letter stringsand faces activate different regions ofvisual cortex (Puce et aL, 1996), and dif-ferent types of visual word forms mayactivate different cortical areas. Peter-sen et aL (1990) compared v, ords, pseu-dowords, and consonant strings to

false font strings. They isolated a re-gion of the left medial extrastriate cor-tex that w as activated bv both wordsand pseudowords but not bv conso-nant strings. Their conclusion was thatthis cortical area is involved in the

recognition of visual w ord forms thatobey English spelling rules. Pugh et aL(1996) have also shown that this regionresponds more to real words than toletter strings when a category judge-ment w as made on real w ords and acase judgement was made on letter

strings. Such findings support the pro-posal that the medial extrastriate cortexresponds preferentially to word forms.Other studies, however, do not sup-

port this proposal. Price, Wise, andFrackowiak (1996) demonstrated ex-trastriate activation using letfier stringscompared to false fonts only during a

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

569

FIGURE 3. Functional loci associated with visual word form processing. 1-Frith et al. (1995); 2-Kapur et al. (1994); 3-Kuriki et al. (1998); 4-Menard et al. (1996); 5-Petersen et al. (1990): 6-Price, Moore, & Frackowiak (1996); 7-Puce et al.(1996); 8-Pugh et al. (1996). The placement of each locus in this figure and in Figures 4 through 7 was determined by Ta-lairach coordinates (Talairach & Tournoux, 1988), Brodmann’s areas (Brodmann. 1909), or gyraUsulcal anatomy provided ineach study. The lateral view reflects loci more than 25 mm from the midline projected to the surface of the brain, whereas themedial view reflects loci less than 25 mm from the midline projected onto the medial plane.

feature detection task and not when

words or pseudowords were com-pared to false fonts or letter strings.Consequently, extrastriate activation

may not be specific to word forms.Likewise, Menard et al. (1996) failed toshow consistent activation of any ex-trastriate visual area when participantspassively viewed real words versus

viewing pictures. More recently, Inde-frey et al. (1997) demonstrated that themedial extrastriate activation associ-

ated with passive viewing of pseudo-word strings was largely due to thelength of the string; no medial extra-striate activation resulted from the

comparison of pseudoword strings tolength-matched false font strings.

In summary, whereas earlier studies

pointed toward the specialization ofleft medial extrastriate cortex for visualword forms that obey English spellingrules, more recent studies have ques-tioned this conclusion. Different corti-cal regions, such as the lingual andfusiform gyri, may be selectively in-volved in visual word form processing(Kuriki et al., 1998; Polk & Farah, 1998).Moreover, it is of great interest for the

understanding of reading acquisitionwhether cortical circuitry is specificallydedicated to the recognition of visualword forms and how that cortical spe-cialization may depend on an individ-ual’s experience with words. Suchquestions are just starting to be ex-plored with functional neuroimaging.

Lexical OrthographyOrthographic units of language are ab-stract representations that indicatewhich letters compose a word and in

which order the letters occur. Develop-ing cognitive tasks that isolate ortho-graphic processing from phonologi-cal or semantic processing is difficult.Some authors have compared singleword reading to reciting a word in re-sponse to false font strings (Howardet al., 1992; Small et al., 1996) to isolatelexical orthography. This task, how-ever, likely involves phonological de-coding and processing as well as auto-matic semantic activation of word

meanings (all of which are discussed insubsequent sections). Although pos-terior temporal and parietal areas were

activated in these studies, these re-

gions may also be involved in phono-logical processing, and their activationmay not necessarily be attributed tolexical orthography per se.Orthographic fluency (also called ver-

bal, letter, or phonemic fluency) is an-other commonly used task. In this task,participants are given a single letter (pre-sented either auditorily or visually)and asked to generate a word or wordsthat begin with that letter. Access to theorthographic lexicon is required be-cause the correct spellings of wordsmust be retrieved. However, phono-logical strategies may also be used byconverting the given visual letter intoa corresponding sound and retrievingwords from the phonological ratherthan the orthographic lexicon (L. Fried-man et al., 1998). Despite notable dif-ferences across studies, nearlv all ofthese orthographic fluency tasks havedemonstrated activation of Broca’s area,as shown in Figure 4 (Benson et al.,1996; L. Friedman et al., 1998; Paulesuet al., 1997; Rueckert et al., 1994). Thesecond most commonly reported areaacross studies using orthographic flu-

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

570

FIGURE 4. Functional loci associated with orthographic processing. 1-F~owers et ai. ll991): 2--u. Fneoman et af. ~996;.3--Frith et al. (1991); 4--Howard et al. (1992): 5-Paulesu et al. (1997); 6-Price et al. (1994); 7---Reucker1 et al. (1994);8-Rumsey, Horwitz, et al. (1997); 9-Small et al. (1996).

ency is the left posterior superior tem-poral gyrus (STG) or BBemicke’s area(Benson et al., 19%; L. Friedman et al.,1998; Rueckert et al., 1994). It is not sur-

prising for these two classically de-fined language areas to participate inorthographic fluency, given the lin-

guistic nature of the task, but it is notentirely dear which components of flu-ency are subserved by which areas.

Spelling judgment tasks are another,perhaps less ambiguous, way to exam-ine access to and functioning of the or-thographic lexicon. In a PET study byFlowers, Wood, and Naylor (1991),participants decided whether audi-torily presented words were four let-ters in length. This task involves accessto information about the number of let-ters in a word, which is stored in the

orthographic lexicon. Flowers et al.

found that behavioral spelling perfor-mance was related to the magnitude ofbrain activation onlv in Wernicke’sarea (left posterior STG). Although thephonological representation of the w ordis undoubtedly used in this task, it isunlikely that the phonological worddescription would contribute signifi-cantly to determining the number ofletters in a word. Nevertheless, pure

isolation of orthographic processingmay not be best achieved with this par-ticular spelling judgment task.

In the spelling judgment task devel-oped by Rumsey and colleagues (Rum-sey, Horwitz, et al., 1997; Rusmev, !B;aœ,et al., 1997), a word and a pseudohomo-phone (e.g., third and thurd) were pre-sented visually, and participants de-ceded which of the two was a realword. This task successfully isolatesorthographic processing from phono-logical processing because the spokenforms of the two stimuli are identical..

Hence, participants can only rely onspelling, not phonology, to make thedecision. This task was compared witha phonological decision task, in whichparticipants decided which of two vis-ually presented pseudowords soundedlike a real word (e.g., jope or lo~nk). Re-gions in the left lingual gyrus (extrastri-ate cortex) were more activated duringthe orthographic than during the pho-nological decision task, but it is alsodear that orthographic and phonologi-cal tasks recruited similar regions ofcortex, differing in magnitude of activa-tion rather than in spatial localization

In summan; studies that have at-

tempted to isolate lexical orthography

from lexical phonology hale impli-cated left temporal, left inferior frontal,and left inferior parietal cortex How-ever, many of the tasks that have pre-sumed purely orthographic processingmay have, in fact, also included phono-logical processing, making it difficultto attribute a brain region or networkof regions to orthographic processing.One conclusion that can be drawnfrom these PET studies is that there isa great deal of overlap in terms of brainregions subserwing the two kinds ofprocessing, even when using tasks thatmore effectively isolate orthographicfrom phonological lexical processing.

Lexical PhonologyThe phonological units of languagespecify the sound structure of wordsand the ordering of the phonemes thatmake up the pronunciation of a word.Access to the phonological lexicon hasbeen studied with tasks that require per-ception and evaluation of the soundstructure of words and letters- The most

frequently used tasks in the neuro-imaging literature have been rhymejudgments using letters (Paulesu et al,,1996; Sergent, Zuck, Levesque, & Mac-

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

571

Donald, 1992) or words (Petersen et al.,1989) and certain lexical decision tasks(Frith, Friston, Liddle, & Frackowiak,1991; Rumsey, Horwitz, et al., 1997).Rhyme generation has also been used(Shaywitz, Pugh, et al., 1995).As summarized in Figure 5, several

perisylvian regions likely participatein phonological processing. The areamost commonly reported across stud-ies is the left posterior STG (includingWernicke’s area). For example, rhymejudgment tasks using letters, which

require participants to decide whethera visually presented letter rhymes witha target letter (e.g., B; Paulesu et al.,1996) or a target sound (e.g., ee; Sergentet al., 1992), result in activation of theleft posterior STG. Further evidence foractivation of the left posterior STG comesfrom Petersen et al.’s (1989) study us-ing rhyme judgment with words andfrom Frith et al.’s (1991) lexical deci-sion tasks using auditorily presentedwords and nonwords.Another commonly reported area

across tasks involving phonologicalprocessing is the left insula. Paulesuet al. (1996) reported left insular acti-vation in a rhyme judgment task us-ing letters. Rumsey, Hortwitz, et al.’s s

(1997) phonological lexical decision taskrequired participants to decide whichof two visually presented nonwordssounded like a real word (e.g., jope orjoak? as described earlier). When thistask was compared to the spellingjudgment task described earlier (e.g.,Which word is a real word, third orthurd?), the left insula, among otherareas, was activated. Left insular acti-

vation was also reported as a major sitefor phonological processing in Shay-witz, Pugh, et al.’s (1995) rhyme gener-ation task that required silent genera-tion of words that rhyme with a giventarget word.

Finally, phonological tasks also in-volve the inferior frontal gyrus or

Broca’s area. The rhyme judgmenttasks using letters (Paulesu et al., 1996;Sergent et al., 1992) and the phonolog-ical lexical decision task used by Rum-sey and colleagues (Rumsey & Eden,1997; Rumsey, Hort~~itz, et al., 1997) re-sulted in activation of Broca’s area. Thetwo rhyme judgment tasks using let-ters also reported involvement of theleft caudate (Paulesu et al., 1996; Ser-

gent et al., 1992).In summary, the neuroimaging evi-

dence suggests that the lexical pho-

nological network appears to be

composed of the posterior superiortemporal gyrus, left insula, and infe-rior frontal cortex. It is important tonote, however, that many other cortical

regions have been reported in thesestudies. Moreover, we have restrictedour definition of phonological tasks tothose that require evaluation of thesound structure of words. Production

tasks, such as word and object naming,are also used to elucidate phonologicalretrieval (Price & Friston, 1997) andhave led to the identification of a basal

posterior temporal region that is re-cruited regardless of the mode of inputof the stimulus to be named. For ex-

ample, this basal temporal region is re-cruited even when blind participantsare reading Braille (Buchel, Price, &

Friston, 1998). Further experimenta-tion is needed to determine which re-

gions participate exclusively in phono-logical aspects of reading and whichregions participate in phonological as-pects of production.

Sublexical PhonologyThis component of reading refers to theprocessing of the submorphemic sound

FIGURE 5. Functional loci associated with lexical phonological processing. 1-Frith et al. (1995); 2-Howard et al. (1992);3-Paulesu et al. (1996); 4-Petersen et al. (1989); 5-Rumsey, Horwitz, et al. (1997); 6-Sergent et al. (1992); 7-Shaywitz,Pugh, et al. (1995).

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

572

units of phonological word descrip-tions, or phonemes. Phonemes alone gen-erally do not have associated semanticinformation, unlike orthographic andphonological word descriptions. How-ever, they can be accessed and manip-ulated separately from words, as inpseudoword reading, phonemic moni-toring, and phoneme deletion. Figure 6summarizes the loci of activation in-volved in sublexical phonology De-monet and colleagues (Demonet et al.,1992; Demonet, Price, Wise, & Frack-

owiak, 1994a, 1994b) used phonemicmonitoring, which requires monitor-ing auditorily presented nonwords forthe phoneme /bee/ where the pho-neme may occur at the beginning of theword (presumed to be simpler) or inthe middle of the word (presumed tobe more difficult). Even more difficultversions of this task involve detectingthe phoneme /bee/ only if the word it-self begins with the phoneme /dee/ orwhen it is preceded by the phoneme/dee/ anywhere in the word- The dif-ferent versions of this task tend to acti-vate left superior and middle temporalregions and areas along the occipito-temporal junction. Other investigatorshave confirmed activation of these

temporal regions using nonword read-

ing (compared with irregular wordreading; Rumsey, Horwitz, et al., 1997)and nonword rhyme judgment (Pughet al., 1996). Both of these nonwordtasks are presumed to be operating at asublexical levels, because nonwordshould not have representations in themental lexicon.

Investigators have also demon-strated the following frontal sites of ac-tivation involved in sublexical phono-logical processing: left inferior frontal(Demonet et al., 1992; Hagoort et al.,1999; Pugh et al., 1996), promoter cortex(Demonet et al., 1994b; Hagoort et al.,1999; Rumsey, Hon,I&, et al., 1997),and left orbital frontal (Pugh et aL,1996). Although the brain regions in-volved in sublexical phonology over-lap a great deal with those involved inlexical phonology, there are some no-table differences (cf. Figures 5 and 6).First, lexical phonology appears to in-vole the insula, whereas sublexical

phonology does not. Second, the tem-poral regions engaged during sublexi-cal phonology include the ocdpitotem-poral junction and appear to be morerestricted to the middle temporal gyri,,whereas those regions in the temporallobes involved in lexical phonology aremore superior. Third, frontal lobe

activations (both lateral and medial)appear to be more extensive duringlexical than during sublexical phono-logical processing. Despite the range oftasks used during sublexical phonol-ogy; induding visual and aural presen-tation paradigms, there is agreementthat sublexical phonology elicits task-related activity in ocdpitotemporal,middle temporal, and inferior frontalregions of the left hemisphere.

Semantic ProcessingIhe semantic system, for the presentpurposes, refers to conceptual knowl-edge about the referents of words, in-cluding the category to which an objector entity belongs, its associated attrib-utes and functions, and other associ--ated concepts. Although a number offunctional imaging studies have ex-plored the semantic system in and of it-self, the present discussion is inter-ested only in access to the semanticsystem from the written word.

Investigators have often used cate-gory judgment tasks, in which the par-tidpant decides whether a written wordis a member of a certain semantic cate-

gory, such as animals or tools (De-monet et aL, 1992; Kapur et aL, 1994;

FIGURE 6. Functional tod associated with sublexicaJ phonotogtcai processing. 1-Deffx)net et a]. (1992): 2-Demonet et al.(1994a): 3-Demonet et a]. ( t 994b); 4--Pugh et a]. (1996); ~-Rumsey, Horwitz, et ai. (1997).

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

573

Petersen et al., 1989; Price, Moore,

Humphreys, & Wise, 1997; Pugh et al.,1996), or whether two stimuli of differ-ent parts of speech are semanticallyrelated (e.g., apple and eat; Warburtonet al., 1996; Wise et al., 1991). In thesetasks, the participant must attend tothe meaning of the stimulus and deter-mine whether its referent is a member

of the designated category. Categoryjudgment tasks have been associatedwith activation in superior (Demonetet al., 1992), middle (Demonet et al.,1992; Price et al., 1997), and inferior(Demonet et al., 1992; Pugh et al., 1996)temporal gyri, as well as in the tem-poral pole (Price et al., 1997) and thesupramarginal gyrus (Demonet et al.,1992), as shown in Figure 7. However,there have been PET studies in which

category judgment tasks did not acti-vate the temporal lobe at all, but rathershowed activation in the left frontal

lobe (Petersen et al., 1989) including theleft inferior frontal cortex (Posner & Pa-

vese, 1998). Similar semantic monitor-

ing tasks, such as verb-noun compari-son, have also activated both temporaland frontal lobe structures (Warburtonet al., 1996; Wise et al., 1991).

In other studies, semantic generationtasks have been used in which partici-

pants generate novel semantic coordi-nates for the stimuli presented. For in-stance, in a verb generation task, theparticipant is presented with a noun(e.g., chair) and must then generate averb appropriate to that noun (e.g., sit;Fiez & Petersen, 1993; Fitz Gerald et al.,1997; Petersen et al., 1989; Snyder, Ab-dullaev, Posner, & Raichel, 1995; War-burton et al., 1996; Wise et al., 1991).Similarly, verbal fluency tasks, in

which participants must respond withas many exemplars of the stimuluscategory as possible, also require thegeneration of novel, conceptually re-lated items (Frith et al., 1991; Shaywitz,Pugh, et al., 1995). When comparing averb generation task to rest, Wise et al.(1991) found activation in the superiortemporal lobe, the left inferior and leftmiddle frontal lobes, and the supple-mentary motor area. However, whenPetersen et al. (1989) compared verbgeneration to the repetition of singlenouns, they found activation in the leftanterior inferior frontal cortex, the an-terior cingulate, and the right inferiorlateral cerebellum, but not in the tem-

poral lobes. It is possible that no tem-poral activation was found in such acomparison because access to semanticknowledge was subtracted out in this

comparison (Petersen et al.,1989; Shay-witz, Pugh, et al., 1995; Wise et al.,1991). In contrast, when Warburton et al.(1996) compared the verb generationtask to the aforementioned verb-noun

comparison task, they continued to

find both frontal and temporal regionsof activation. Again, access to semanticknowledge was expected to be pres-ent in both tasks and thereby cancelledout with the subtraction, but this didnot occur. These conflicting results areprobably due to the use of differentcontrol conditions across studies, a

point that is addressed more exten-

sively in the discussion.In summary, both category judg-

ment and semantic generation usingwritten words tend to activate wide re-

gions of temporal and frontal cortex(see Figure 7). Although there are no-table exceptions, category judgment hasmore often been found to result in ac-

tivation of temporal structures, whereassemantic generation frequently leadsto frontal activation.

Phonological and PhoneticEncoding and ArticulationVocal word production involves a se-ries of complex processes that are be-

FIGURE 7. Functional loci associated with semantic processing. 1-Demonet et al. (1992); 2-Kapur et al. (1994);3-Petersen et al. (1989); 4-Price et al. (1994); 5-Price et al. (1997); 6-Pugh et al. (1996); 7-Shaywitz, Pugh, et al.(1995); 8-Warburton et al. (1996); 9-Wise et al. (1991).

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

574

yond the scope of the present article i

(for more in-depth treatments of this ]

topic, see Blumstein, 1995; Levelt, 1

1989). Although the present focus is on ]

word decoding rather than word pro- I

duction, the encoding of a phonologi-cal representation into an articulatoryprogram for overt speech is briefly re-viewed here. Overt word productionhas been studied primarily with PETbut not with fMRI. One reason for this

is a technical limitation-articulation is

associated with motion, and motion is

detrimental to MRI data quality.Using PET, Rumsey, Horwitz, et al.

(1997) demonstrated that the covert orovert nature of the response stronglypredicts the locus of activation in thebrain. In their study, pronunciationtasks activated the superior temporalgyus bilaterally, but decision-makingtasks (responding yes or no with a but-ton press) preferentially activated theleft inferior frontal cortex. These find-

ings are consistent with other studies

comparing word pronunciation withsilent reading (Bookheimer, Zeffiro, Blax-ton, Gaillard, & Theodore, 1995; Price

et al., 1994). An important confound toconsider when comparing oral andsilent reading is that oral reading mayactivate superior temporal regionsclose to the primary auditory cortex be-cause individuals are hearing the soundof their own voices. Consequently, theSTG activation may merely reflect au-dition’ feedback from the task rather

than access to phonological codes andexecution of articulatory programs.E%idence against this proposal is thatwhen reading aloud is compared toword articulation in response to a

meaningless stimulus, STG activationremains (Moore & Price, 1999). Ad-

dressing such confounds is critical for

illuminating cortical specialization forovert word production.

Functional Brain Imaging ofAcquired Reading Disorders

Acquired reading disorder, or alexia,mav result from a variety of postdevel-opmental forms of brain damage, suchas cortical insult, tumor, stroke, or de-

mentia. Reading may be disrupted inmultiple ways, leading to a variety of ]

behavioral deficits depending on the ’i

location of neural dysfunction (for a

complete reviews of lesion studies, seeR. B. Friedman, Ween, & Albert, 1993).

Although the benefits of examining ac-quired disorders to elucidate the com-ponents of typical reading are limited(Snowling, Bryant, & Hulme, 1996), it

is useful to examine some of the neural

structures that may be functionally im-

paired in alexia. Two types of alexia arepresented here: surface alexia and

phonological alexia.The acquired reading disorder known

as surface alexia results in a patient hav-ing greater difficulty in reading wordswith exceptional spelling-sound corre-spondences (e.g., gene) than words withtypical spelling-sound correspondences(e.g., green). The appearance of a sur-face alexia is possible evidence for reli-ance on a sublexical grapheme-phonemeroute in oral reading. Although anatom-ical lesions producing surface alexiaare diverse, thev tend to involves the

left temporoparietal cortex while spar-ing the occipital cortex (see Patterson,~farshall, & Coltheart, 1985). Func-tional neuroimaging data are in accordwith this site being the locus of im-

pairment. Parkin (1993) used PET to re-veal regional h~-pometabolism at restrather than examine relative neural ac-

tivation during a specific task. Parkinshowed that persons with surface alexia

displayed regional h~-pometabolism inall three left temporal gyri as w ell as inthe left posterior frontal gyrus.

Phonological alexia is defined as an in-ability to read pronounceable nonwords,or pseudow-ords, while the reading ofboth regular and exceptional real wordsis preserved. Patients with this disor-der must use semantics to translate or-

thographic representations into phono-logical output, because the lexical

route is impaired. Words without se-mantic representation (i.e., pseudo-words) or with very little semantic

representation (i.e., functors and low-

frequency words) therefore pose a dif-ficulri- for these patients. Phonologicalalexia has been linked to lesions of the

superior temporal lobe and the angu-lar and supramarginal gyri (Bub, Black,Ho«’ell, & Kertesz, 1987; Derouesne &

Beauvois, 1985; R- B. Friedman & Kohn,

1990). Small, Flores, and ~loll (1998) in-

vestigated the brain activation in an ac-quired phonological alexia patient (inthis case with a large left frontotempo-ral lesion) during oral reading usingfMRI. During the experimental condi-tion, the patient was asked to performa reading task, which when comparedwith a control faLse-font task, primarilyactivated the left angular gyrus. Also,the lingual gyrus, the lateral premotorarea, the supplementary motor area, theprecuneus, the middle temporal gyrus,and the prefrontal cortex all showed alesser degree of activation. After engag-ing in therapy designed to teach the

patient a set of rules for translatinggraphemes to phonemes, the patientwas able to read pseudowords and func-tors. When retested using the samefMRI paradigm, the reading task pref-erentiallv activated the left lingual gy-rus, with lesser activation occurring inthe middle temporal gyrus, the lateralpromoter area, the cuneus, me precu-neus, and the precentral gyrus, and muchless activation than in the initial scan in

me angular gyurs. The authors proposedthat during grapheme-phoneme con-version, the patient learned to visuallydecompose and convert orthographyinto phonology in order to read aloud.It is possible that functional reorgani-zation of the brain followed therapy,such that circuits involving occipitalareas, including the lingual gyrus, wereunmasked to allow reading via thegrapheme-phoneme route. These ond-ings are consistent with task-related

signal changes observed in typicalreading populations, N hich assign thetemporoparietal areas to orthographicand the occipitotemporal areas to

phonological components of reading(described earlier and summarized in

Figures 3 through 6).

SummaryFunctional neuroimaging data that

support the various components of the

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

575

single word reading process have beenpresented. Figures 3, 4, 5, and 6 showthe loci of activation across a numberof different studies for each componentof reading. In general, left-hemisphereperisylvian regions are engaged acrossmany of the subcomponents. Morespecifically, visual word form process-ing tends to engage posterior corticalregions, primarily in the occipital andoccipitotemporal cortex. Orthographicprocessing primarily involves poste-rior temporal, inferior parietal, and in-ferior frontal regions. Lexical phono-logical, sublexical phonological, andsemantic components recruit large re-gions of temporal and inferior frontalcortex. These are, of course, general-izations. As Poeppel (1996) has pointedout in his analysis of PET studies ofphonological processing, and as is evi-dent here, there is a great deal of vari-

ability across studies of brain and lan-guage relationships. The extant bodyof research on brain-behavior relation-

ships within the reading domain needsto be substantiated and disambiguatedwith future research.

Discussion

PET and fMRI have enhanced our

knowledge about how specific compo-nents of reading map onto the brain.As these techniques become more re-fined and technologically advanced,our knowledge of the brain structuresthat underlie reading will also ad-

vance. Improvements in experimentaldesign and postprocessing of imagedata will help to improve the reliabil-ity of functional imaging data, allow-ing more precise and definitive conclu-sions about the neural underpinningsof the reading process. The followingdiscussion evaluates the reviewed lit-

erature in light of the limitations andpitfalls of using the existing imag-ing techniques. These limitations arebroadly classified into the followingcategories: technical limitations, exper-imental design issues, pediatric func-tional neuroimaging, and cognitiveassumptions.

Technical Limitations

Most of the studies presented in this re-view have used the older functional

imaging technique PET. The better spa-tial information gained with fMRI mayhelp future studies in isolating sepa-rate but spatially proximal brain re-gions for tasks such as orthographicand phonological processing, whichappear to overlap a great deal in termsof functional anatomv (e.g., Rumsey,Horwitz, et al., 1997).

y

In fMRI, the ultimate temporal limi-tation is the duration of the hemody-namic response, which is in the rangeof seconds; hence, the temporal resolu-tion of fMRI (and PET) is poor. Other

techniques with exquisite temporalinformation, such as evoked poten-tials (EPs) or magnetoencephalography(MEG), might be more useful for someof these questions. There are ongoingefforts to use techniques with highertemporal resolution in conjunctionwith fMRI in the same individuals per-forming the same tasks (Guy, ffytche,Brovelli, & Chumillas, 1999). This way,the strengths of each technique can becombined to provides good temporaland spatial information to elucidatebrain mechanisms involved in readingand other cognitive functions.Only partial brain volumes may

have been collected in a number ofthese studies (depending on the equip-ment available and time constraints ofthe study), further hindering our abil-ity to make comparisons across stud-ies. For example, the cerebellum wasoften not included during data acqui-sition in many of these studies, yet instudies that have collected whole brain

volumes, it appears that the cerebel-lum is often active during reading-related tasks (e.g., Rumsey, Horwitz,et al., 1997).fMRI studies of reading aloud have

rarely been conducted because of theresulting artifacts introduced by headand jaw movements. Alternating dataacquisition with task execution can cir-cumvent this problem (Eden, Joseph,Brown, Brown, & Zeffiro, 1999). Withthis approach, the jaw movement oc-

curs in a period different from the dataacquisition interval. Because of the de-lay of the hemodvnamic response, thesignal is captured in the delayed ac-quisition period. Future developmentsin data acquisition techniques will in-evitably improve neuroimaging stud-ies of reading and address other issuessuch as noise contamination and inter-and intrasubject reliability.

Experimental DesignThe choice of control task varies enor-

mously across the studies reviewedhere. As alluded to in earlier sections,the control task that is contrasted to the

experimental task can have a tremen-dous impact on the cortical regionsthat appear to be activated by a task.Statistical contrasts that compare an ex-

perimental task such as reading wordsto a rest task such as visual fixation, for

example, will not be as selective ascomparing to a control task (e.g., pseu-doword reading) that shares many ofthe same cognitive components as theexperimental task. Interpreting resultsfrom functional imaging studies, there-fore, requires an understanding of thecontrol task that is used in statisticalcontrasts.

Another consideration in functional

imaging studies of reading is the stim-ulus presentation rate. Blood flow andMRI signal changes can be modulatedby the presentation rate and exposureduration of visual and auditory stim-uli. Using PET, Price et al. (1992)showed that primary auditory cortexand middle STG were linearly modu-lated by the presentation rate of audi-tory speech stimuli, such that a fasterpresentation rate led to increases inblood flow. This linear relationship,however, did not hold for Wernicke’sarea. In the visual modality using PET,Price, Moore, and Frackowiak (1996)showed that when reading wordsaloud, blood flow in early visual areasincreases with increased presentationrate and longer exposure durations.Such relationships have also been ex-plored with fMRI. Binder et al. (1994)showed that the MRI signal increases

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

576

in a nonlinear fashion with respect tothe rate of presentation of speech stim-uli. However, in contrast with Priceet al.’s (1992) work using PET, this re-lationship was not area-dependent. Toensure that the differences betweenPET and &00 with respect to presen-tation rate were not due to differencesin design, Rees et al. (1997) comparedsuch relationships in PET and fMRIdirectly. Using passive listening to

words, they showed that responses inprimary auditory cortex linearly in-

creased with presentation rate in-

creases for PET, but that the relation-

ship was not linear for &00 responses.All of these studies show that the pre-sentation rate of visual and auditorystimuli can significantly modulate cor-tical responses for both PET and f~l~iRl.In general, increasing the presentationrate leads to increases in blood flow

measures or’.~iRI signal changes in pri-mary sensory areas.The functional neuroimaging stud-

ies of reading to date have often usedvery small samples of participants. Fur-thermore, the samples are rarely bal-anced for gender. Because some stud-ies investigating reading-related taskshave suggested that there are differ-ences between men and women (Shav-witz, Shaywitz, et al., 1995), the compo-sition of the sample should be carefullyconsidered when evaluating results

from these studies.

Pediatric Functional

Neuroimaging

Very few functional neuroimagingstudies have been conducted with chil-

dren, mainly because PET requires theapplication of radioactive material.

Due to its noninvasive nature, fMRI

can be used to study cognitive functionin children. In early pediatric studies,fMRI was used to map language dom-inance in children with partial epilepsy.(Hertz-Pannier et al., 1997). The resultswere in agreement with intracarotidamobarbital testing performed for

presurgical evaluation. Recent studieshave used fMRI to investigate nonlan-guage cognitive tasks in children and

compare them to adults to yield de-velopmental information (Casey et al.,1997; Thomas et al., 1999). Ongoingstudies using fMRI to investigate lan-guage processing in typical volunteersage 5 years and older dearly demon-strate that children can tolerate theMRI environment for experimentaland clinical purposes. Figure 8 showsan example of a 12-year-old child read-ing aloud versus visual- fixation, asstudied with BOO in our laboratory.The activation pattern is similar to thatseen in adults. The fact that multiplemeasurements can be made on thesame individuals with MRI allows for

tracking developmental changes onboth short- and long-term scales. Oneadditional advantage of &00 is the

higher spatial resolution gained in

these images. However, the quality ofthis finer resolution is highly depen-dent on the participant keeping his or

her head steady throughout the scan.Head movement is more noticeable inchildren compared to adults, and if thetotal amount of head motion exceeds a

certain limit, artificial signal changeswill be observed, resulting in data arti-facts. Current solutions to the problemof excessive head motion are to pro-vide pre-experimental training in

keeping the child’s head steady, usinghead restraining devices, and applyingmotion correction routines prior to

data analysis. W ith the continued re-finement and development of theseprocedures, pediatric functional brainimaging for cognitive research is be-coming more feasible.

Cognitive AssumptionsInterpretations of the reviewed studiescan be complicated bv misattributing acognitive function to a given task For

FIGURE 8. Functional 100 assoc4ated with ~;-’9’¿ ..~;.:. reading compa;~ :Gfixation baseline in a 12-year-old child. The most anterior region of activation is inthe inferior frontal gyrus and frontal opercutum (i.e., the region of the frontal cortexthat lies superior to the Sylvian fissure). The midde region of activation is in theprecentral gyrus, in the most inferior portion of motor cortex. The posterior regionof activation is in posterior middle terr~poral gyrus and superior temporal gyrus.

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

577

instance, because word generation isprobably more demanding than wordrepetition, it is difficult to say whetherincreased activation in the former task

is due to its semantic requirements orsimply to increased effort. Studies

comparing novel and overpracticedtask-related activation have shown sig-nificant shifts in the location of activa-

tion loci (Raichle et al., 1994).Another difficulty is devising a rest-

ing state task that does not involvecognitive processing of some sort. Forinstance, Binder et al. (1999) have pro-posed that it is incorrect to assume thata so-called &dquo;resting state&dquo; condition ofan experiment is truly a state of neuralinactivity (see Figure 2). Thev testedthis by comparing a resting state, a

tone perception task, and a semanticretrieval task. They found a network ofleft-hemispheric cortical regions thatwas more active during rest than dur-ing the perceptual task, but that hadequal amounts of activity during therest and the semantic task. These areas

included a number of areas that feature

prominently in the studies of readingdescribed in this review. The investiga-tors maintained that the semantic task

and the rest condition both engagedgeneral conceptual processing; there-fore, these conceptual processing areaswere seen in the rest-tone comparisonand in the semantic-tone comparison,but not in the semantic-rest compari-son. If processes involved in semanticretrieval and manipulation are en-gaged during conscious resting states,then studies designed to detect or ma-nipulate these processes may lack sen-sitivity if a resting condition is used asa control comparison.

Conclusion

In conclusion, PET and fMRI are pow-erful tools for studying the brain-behavior relationships that underlie

the reading process. The existing stud-ies in adults have laid the groundworkfor understanding single word decod-ing ; however, future studies will enrichand expand these past efforts as tech-

niques, experimental design, and dataanalysis become more refined. Prom-ising future avenues of investigationinclude brain-behavior relationshipsthat underlie the integration of singleword decoding and comprehension oftext, language dependencies in brain ac-tivation during reading and language,and reading acquisition in children.

ABOUT THE AUTHORS

Jane Joseph, PhD, is an assistant professor atthe University of Kentucky Medical Center. Herresearch interests include functional brain

imaging studies of visual cognition and lan-gauge. Kimberly Noble, BSc, is currently inthe MD/PhD program at the University ofPennsylvania. She is interested in developmen-tal and acquired reading disorders. GuinevereEden, Dphil, is director of the Center for theStudy of Learning at Georgetown UniversityMedical Center. Her research interests include

functional brain imaging studies of reading andreading disability and the neurophysiolog-icnlbasis of reading remediation. Address: Guine-vere Eden, Center for the Study of Learning,Georgetown University Medical Center, Build-ing D, Suite 150, 4000 Reservoir Road, Wash-ington, DC 20007 (e-mail: eden@georgetoum.edu).

AUTHORS’ NOTE

Support for this work uxrs provided by theNational Institute of Child Health and HumanDevelopment, National Institute of Health.

REFERENCES

Belliveau, J. W., Kennedy, D. N., McKinstry,R. C., Buchbinder, B. R., Weisskoff, R. M.,Cohen, M. S., Vevea, J. M., Brady, T. J., &

Rosen, B. R. (1991). Functional mappingof the human visual cortex by magneticresonance imaging. Science, 254, 716-719.

Benson, R. R., Logan, W. J., Cosgrove, G. R.,Cole, A. J., Jiang, H., LeSueur, L. L., Buch-binder, B. R, Rosen, B. R, & Caviness,V S., Jr. (1996). Functional MRI localiza-tion of language in a 9-year-old child. TheCanadian Journal of Neurological Sciences,23, 213-219.

Binder, J. R., Rao, S. M., Hammeke, T. A.,Frost, J. A., Bandettini, P. A., & Hyde, J. S.(1994). Effects of stimulus rate on signalresponse during functional magneticresonance imaging of auditory cortex.

Cognitive Brain Research, 2, 31-38.

Binder, J. R., Frost, J. A., Hammeke, T. A.,Bellgowan, P. S. F., Rao, S. M., & Cox,R. W. (1999). Conceptual processing dur-ing the conscious resting state: A func-tional MRI study. Journal of Cognitive Neu-roscience, 11, 80-95.

Blumstein, S. E. (1995). The neurobiology ofthe sound structure of language. In M. S.Gazzaniga (Ed.), The cognitive neuro-

sciences (pp. 915-929). Cambridge, MA:MIT Press.

Bookheimer, S. Y., Zeffiro, T. A., Blaxton, T.,Gaillard, W., & Theodore, W. (1995). Re-

gional cerebral blood flow during objectnaming and word reading. Human BrainMapping, 3, 93-106.

Brodmann, K. (1909). Vergleichende Lokalisa-tionslehre der Grosshirnrinde in ihren Prin-

zipien dargestellt auf Grund des Zellenbaues[Comparative topography of the cerebralcortex based on principles of cytoarchi-tectonics]. Leipzig, Germany: Barth.

Bub, D., Black, S. E., Howell, J., & Kertesz,A. (1987). Speech output processes andreading. In M. Coltheart, G. Sartori, &R. Job (Eds.), The cognitive neuropsychologyof language (pp. 79-110). Hillsdale, NJ:Erlbaum.

Buchel, C., Price, C., & Friston, K. (1998). Amulitmodal language region in the ven-tral visual pathway. Nature, 394, 274-277.

Casey, B. J., Trainor, R. J., Orendi, J. L., Schu-bert, A. B., Nystrom, L. E., Giedd, J. N.,Castellanos, F. X., Haxby, J. V, Noll, D. C.,Cohen, J. D., Forman, S. D., Dahl, R. E., &

Rapoport, J. L. (1997). A developmentalfunctional MRI study of prefrontal acti-vation during performance of a go-no-gotask. Journal of Cognitive Neuroscience, 9,835-847.

Dejerine, J. (1892). Contribution a l’étude

anatomo-pathologique et clinique desdifferentes varietés de cécité verbale

[Contribution to the anatomopathologi-cal and clinical study of various forms ofworld blindness]. Mémoires de la Societé

Biologique, 4, 61-90.Demonet, J. F., Chollet, F., Ramsay, S.,Cardebat, D., Nespoulous, J. L., Wise, R.,Rascol, A., & Frackowiak, R. (1992). Theanatomy of phonological and semanticprocessing in normal subjects. Brain, 115,1753-1768.

Demonet, J. F., Price, C., Wise, R., & Frack-

owiak, R. S. (1994a). Differential activa-tion of right and left posterior sylvian re-gions by semantic and phonologicaltasks: A positron-emission tomographystudy in normal human subjects. Neuro-science Letters, 182, 25-28.

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

578

Demonet, J. F., Price, C., Wise, R, & Frack-owiak, R S. J. (1994b). A PET study ofcognitive strategies in normal subjectsduring language tasks: Influence of pho-netic ambiguity and sequence processingon phoneme monitoring. Brain, 117,671-682.

Derouesne, J., & Beauvois, M. F. (1985). The"phonemic" stage in the nonlexical read-ing process: Evidence from a case of pho-nological alexia. In K. E. Patterson, J. C.Marshall, & M. Coltheart (Eds.), Surfacedyslexia. London: Erlbaum.

Eden, G. F., Joseph, J. E., Brown, H. E.,Brown, C. P, & Zeffiro, T. A. (1999). Uti-lizing the hemodynamic delay and dis-persion to detect fMRI signal changewithout auditory interference: The be-havior interleaved gradients technique.Magnetic Resonance in Medicine, 41, 13-20.

Fiez, J. A., & Petersen, S. E. (1993). PET aspart of an interdisciplinary approach tounderstanding processes involved in

reading. Psychological Science, 4, 287-292.Fitz Gerald, D. B., Cosgrove, G. R, Ronner,

S., Jiang, H., Buchbinder, B. R, Belliveau,J. W., Rosen, B. R, & Benson, R R (1997).Location of language in the cortex: A com-parison between functional MR imagingand electrocortical stimulation. American

journal of Neuroradiology, 18, 1529-1539.Flowers, D. L, Wood, F. B., & Naylor, C. E.

(1991). Regional cerebral blood flow corre-lates of language processes in reading dis-ability. Archives of Neurology, 48, 637-643.

Friedman, L, Kenny, J. T., Wise, A. L, Wu,D., Stuve, T. A., Miller, D. A., Jesberger,J. A., & Lewin, J. S. (1998). Brain activa-tion during silent word generation eval-uated with functional MRI. Brain and Lan-

guage, 64, 231-256.Friedman, R B., & Kohn, S. E. (1990). Im-

paired activation of the phonological lex-icon: Effects upon oral reading. Brain andLanguage, 38, 278-297.

Friedman, R B., Ween, J. E., & Albert, M. L(1993). Alexia. In K. M. Heilman & E. Val-enstein (Eds.), Clinical neuropsychology(3rd ed., pp. 37-62). New York: OxfordUniversity Press.

Frith, C. D., Friston, K. J., Liddle, P. F., &

Frackowiak, R. S. J. (1991). A PET study ofword finding. Neuropsychologia, 29,1137-1148.

Frith, C. D., Kapur, N., Friston, K. J., Liddle,P. F., & Frackowiak, R S. J. (1995). Re-gional cerebral activity associated withthe incidental processing of pseudo-words. Human Brain Mapping, 3,153-160.

Guy, C. N., ffytche, D. H., Brovelli, A., &

Chumillas, J. (1999). fMRI and EEG re-sponses to periodic visual stimulation.NeuroImage, 10, 125-148.

Hagoort, P., Indefrey, P., Brown, C., Herzog,H., Steinmetz, H., & Seitz, R. J. (1999). Theneural circuitry involved in the readingof German words and pseudowords: APET study. Journal of Cognitive Neuro-science, 11, 383-398.

Hertz-Pannier, L, Gaillard, W. D., Mott,S. H., Cuenod, C. A., Bookheimer, S. Y,Weinstein, S., Conry, J., Papero, P. H.,Schiff, S. J., Le Bihan, D., & Theodore,W. H. (1997). Noninvasive assessment of

language dominance in children andadolescents with functional MRI: A pre-liminary study. Neurology, 48, 1003-1012

Howard, D., Patterson, K., Wise, R, Brown,W. D., Friston, K., Weiller, C., & Frack-

owiak, R (1992). The cortical localizationof the lexicons. Brain, 115, 1769-1782.

Indefrey, P, Kleinschmidt, A., Merboldt,K-D., Kruger, G., Brown, C., Hagoort, P,& Frahm, J. (1997). Equivalent responsesto lexical and nonlexical visual stimuli in

occipital cortex: A functional magneticresonance imaging study. NeuroImage, 5,78-81.

Ingvar, D. H., & Schwartz, M. S. (1974).Blood flow patterns induced in the dom-inant hemisphere by speech and reading.Brain, 97, 273-288.

Kapur, S., Rose, R., Liddle, P. F., Zipursky,R B., Brown, G. M., Stuss, D., Houle, S.,& Tulving, E (1994). The role of the left pre-frontal cortex in verbal processing: Se-mantic processing or willed action. Neuro-

Report, 5, 2193-2196.Kuriki, S., Takeuchi, F., & Hirata, Y. (1998).

Neural processing of words in the humanextrastriate cortex. Cognitive Brain Re-search, 6, 193-203.

Levelt, W. J. M. (1989). Speaking. From inten-tion to articulation. Cambridge, MA: MITPress.

McClelland, J. L (1979). On the time rela-tions of mental processes: An examina-tion of systems of processes in cascade.Psychological Review, 86, 287-330.

Menard, M. T., Kosslyn, S. M., Thompson,W. L., Alpert, N. M., & Rauch, S. L. (1996).Encoding words and pictures: A positronemission tomography study. Neuropsy-chologia, 34, 185-194.

Moore, C. J., & Price, C. J. (1999). Three dis-tinct ventral occipitotemporal regions forreading and object naming. NeuroImage, 10, 181-192.

Ogawa, S., Lee, T. M., Nayak, A. S., &Glynn, P. (1990). Oxygenation-sensitivecontrast in magnetic resonance imagingof rodent brain at high magnetic fields.Magnetic Resonance in Medicine, 14, 68-78.

Parkin, A. J. (1993). Progressive aphasiawithout dementia—A clinical and cogni-tive neuropsychological analysis. Brainand Language, 44, 201-220.

Patterson, K. E., Marshall, J., & Coltheart,M. (Eds.). (1985). Surface dyslexia. Lon-don: Erlbaum.

Paulesu, E., Frith, C., Snowling, M., Gal-lagher, A., Morton, J., Frackowiak, R. S. J.,& Frith, C. (1996). Is developmental dys-lexia a disconnection syndrome ? Evidencefrom PET scanning. Brain, 119, 143-157.

Paulesu, E., Goldacre, B., Scifo, P., Cappa,S. F., Gilardi, M. C., Castiglioni, D. P., &

Fazio, F. (1997). Functional heterogeneityof left inferior frontal cortex as revealed

by fMRI. NeuroReport, 8, 2011-2016.Petersen, S. E., Fox, P. T., Posner, M. L, Min-

tun, M., & Raichle, M. E. (1989). Positronemission tomographic studies of the pro-cessing of single words. Journal of Cogni-tive Neuroscience, 1, 153-170.

Petersen, S. E., Fox, P T., Snyder, A. Z., &Raichle, M. E. (1990). Activation of extra-striate and frontal cortical areas by visualwords and word-like stimuli. Science, 249,1041-1044.

Poeppel, D. (1996). A critical review of PETstudies of phonological processing. Brain and Language, 55, 317-351.

Polk, T. A., & Farah, M. J. (1998). The neuraldevelopment and organization of letterrecognition: Evidence from functionalneuroimaging, computational modeling,and behavioral studies. Proceedings of theNational Academy of Sciences of the UnitedStates of America, 95, 847-852.

Price, C. J., & Friston, K. J. (1997). Cognitiveconjunctions: A new approach to brain acti-vation experiments. NeuroImage, 5, 261-270.

Price, C. J., Moore, C. J., & Frackowiak, R. S.(1996). The effect of varying stimulus rateand duration on brain activity duringreading. NeuroImage, 3, 40-52.

Price, C. J., Moore, C. J., Humphreys, G. W.,& Wise, R. J. S. (1997). Segregating se-mantic from phonological processes dur-ing reading. journal of Cognitive Neuro-science, 9, 727-733.

Price, C. J., Wise, R. J., & Frackowiak, R. S.(1996). Demonstrating the implicit pro-cessing of visually presented words andpseudowords. Cerebral Cortex, 6, 62-70.

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from

579

Price, C., Wise, R., Ramsay, S., Friston, K.,Howard, D., Patterson, K., & Frackowiak,R. (1992). Regional response differenceswithin the human auditory cortex whenlistening to words. Neuroscience Letters,146, 179-182.

Price, C. J., Wise, R. J. S., Watson, J. D. G.,Patterson, K., Howard, D., & Frackowiak,R. S. J. (1994). Brain activity during read-ing: The effects of exposure duration andtask. Brain, 117, 1255-1269.

Puce, A., Allison, T., Asgari, M., Gore, J. C., & McCarthy, G. (1996). Differential sensi-tivity of human visual cortex to faces, let-terstrings, and textures: A functional

magnetic resonance imaging study. TheJournal of Neuroscience, 16, 5205-5215.

Pugh, K. R., Shaywitz, B. A., Shaywitz, S. E.,Constable, R. T., Skudlarski, P., Fulbright,R. K., Bronen, R A., Shankweiler, D. P, Katz,L., Fletcher, J. M., & Gore, J. C. (1996).Cerebral organization of component pro-cesses in reading. Brain, 119, 1221-1238.

Raichle, M. E., Fiez, J. A., Videen, T. O.,MacLeod, A. M., Pardo, J. V, Fox, P. T., &

Petersen, S. E. (1994). Practice-related

changes in human brain functional anat-omy during nonmotor learning. CerebralCortex, 4, 8-26.

Rees, G., Howseman, A., Josephs, O., Frith,C. D., Friston, K. J., Frackowiak, R. S. J., &

Turner, R. (1997). Characterizing the rela-tionship between BOLD contrast and re-gional cerebral blood flow measurementsby varying the stimulus presentationrate. NeuroImage, 6, 270-278.

Roy, C. S., & Sherrington, C. S. (1890). Onthe regulation of the blood supply of thebrain. Journal of Physiology (London), 11,85-105.

Rueckert, L., Appollonio, I., Grafman, J.,Jezzard, P., Johnson, R., Jr., Le Bihan, D.,& Turner, R. (1994). Magnetic resonanceimaging functional activation of left

frontal cortex during covert word pro-duction. Journal of Neuroimaging, 4, 67-70.

Rumsey, J. M., & Eden, G. F. (1997). Func-tional neuroimaging of developmentaldyslexia: Regional cerebral blood flow indyslexic men. In B. Shapiro, P. J. Accardo,& A. J. Capute (Eds.), Specific reading dis-ability: A view of the spectrum (pp. 35-62).Timonium, MD: York Press.

Rumsey, J. M., Horwitz, B., Donohue, B. C.,Nace, K., Maisog, J. M., & Andreason, P.

(1997). Phonologic and orthographic com-ponents of word recognition: A PET-rCBFstudy. Brain, 120, 739-759.

Rumsey, J. M., Nace, K., Donohue, B., Wise,D., Maisog, J. M., & Andreason, P. (1997).A positron emission tomographic studyof impaired word recognition and phono-logical processing in dyslexic men. Ar-chives of Neurology, 54, 562-73.

Sergent, J., Zuck, E., Levesque, M., & Mac-

Donald, B. (1992). Positron emission

tomography study of letter and objectprocessing: Empirical findings and meth-odological considerations. Cerebral Cor-

tex, 2, 68-80.

Shaywitz, B. A., Pugh, K. R., Constable,R. T., Shaywitz, S. E., Bronen, R. A., Ful-bright, R. K., Shankweiler, D. P., Katz, L.,Fletcher, J. M., Skudlarski, P., & Gore, J. C.(1995). Localization of semantic pro-cessing using functional magnetic reso-nance imaging. Human Brain Mapping, 2,149-158.

Shaywitz, B. A., Shaywitz, S. E., Pugh, K. R.,Constable, R. T., Skudlarski, P., Fulbright,R. K., Bronen, R. A., Fletcher, J. M., Shank-weiler, D. P., Katz, L., & Gore, J. C. (1995).Sex differences in the functional organi-zation of the brain for language. Nature,373, 607-609.

Small, S. L., Flores, D. K., & Noll, D. C.(1998). Different neural circuits subserve

reading before and after therapy for ac-quired dyslexia. Brain and Language, 62,298-308.

Small, S. L., Noll, D. C., Perfetti, C. A.,Hlustik, P., Wellington, R., & Schneider,

W. (1996). Localizing the lexicon for read-ing aloud: Replication of a PET studyusing fMRI. NeuroReport, 7, 961-965.

Snowling, M. J., Bryant, P. E., & Hulme, C.

(1996). Theoretical and methodologicalpitfalls in making comparisons betweendevelopmental and acquired dyslexia:Some comments on A. Castles & M. Colt-

heart (1993). Reading and Writing: An In-terdisciplinary Journal, 8, 443-451.

Snyder, A. Z., Abdullaev, Y. G., Posner,M. I., & Raichle, M. E. (1995). Scalp elec-trical potentials reflect regional cerebralblood flow responses during processingof written words. Proceedings of the Na-tional Academy of Sciences of the UnitedStates of America, 92, 1689-1693.

Talairach, J., & Tournoux, P. (1988). Co-planarstereotaxic atlas of the human brain (M. Ray-port, Trans.). New York: Thieme.

Thomas, K. M., King, S. W., Franzen, P. L.,Welsh, T. F., Berkowitz, A. L., Noll, D. C.,Birmaher, V., & Casey, B. J. (1999). A de-velopmental functional MRI study ofspatial working memory. NeuroImage, 10,327-338.

Turner, R., Le Bihan, D., Moonen, C. T., De-

spres, D., & Frank, J. (1991). Echo-planartime course of a cat brain oxygenationchanges. Magnetic Resonance in Medicine,22, 159-166.

Warburton, E., Wise, R. J. S., Price, C. J.,Weiller, C., Hadar, U., Ramsay, S., &

Frackowiak, S. J. (1996). Noun and verbretrieval by normal subjects: Studies withPET. Brain, 119, 159-179.

Wernicke, C. (1887). Der aphasische Symp-tomencomplex [The symptom complex ofaphasia]. Breslau: Cohn und Weigart.

Wise, R., Chollet, F., Hadar, U., Friston, K.,Hoffner, E., & Frackowiak, R. (1991). Dis-tribution of cortical neural networks in-volved in word comprehension and wordretrieval. Brain, 114, 1803-1817.

at COLUMBIA UNIV on October 19, 2011ldx.sagepub.comDownloaded from