fMRI of syntactic processing in typically developing children: Structural correlates in the inferior frontal gyrus

(1)

ContentslistsavailableatScienceDirect

Developmental

Cognitive

Neuroscience

jo u r n al h om ep a ge :h t tp : / / w w w . e l s e v i e r . c o m / l o c a t e / d c n

fMRI

of

syntactic

processing

in

typically

developing

children:

Structural

correlates

in

the

inferior

frontal

gyrus

S. Christopher

Nu ˜

nez

a,b

_,

_Mirella

_Dapretto

c,d,e

_,

_Tami

_Katzir

a,b,f

_,

_Ariel

_Starr

a,b

_,

Jennifer

Bramen

a,b,d

_,

_Eric

_Kan

a,b

_,

_Susan

_Bookheimer

d

_,

_Elizabeth

_R.

_Sowell

a,b,e,∗ a_{Developmental}_Cognitive_Neuroimaging_Laboratory,_Department_of_Neurology,_University_of_California,_Los_Angeles,_Los_Angeles,_CA,_USA b_Laboratory_of_Neuro_Imaging,_Department_of_Neurology,_University_of_California,_Los_Angeles,_Los_Angeles,_CA,_USA

c_{Ahmanson-Lovelace}_Brain_Mapping_Center,_University_of_California,_Los_Angeles,_Los_Angeles,_CA,_USA d_UCLA_Department_of_Psychiatry_and_{Biobehavioral}_Sciences,_Los_Angeles,_CA,_USA

e_FPR-UCLA_Center_for_Culture,_Brain,_and_Development,_USA

f_Education_Department,_Department_of_Learning_{Disabilities,}_University_of_Haifa,_Haifa,_Israel

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received18June2010

Receivedinrevisedform19February2011 Accepted22February2011 Keywords: Syntax language Typicaldevelopment Lateralization fMRI Multimodal

a

b

s

t

r

a

c

t

Developmentofsyntacticprocessingwasexaminedtoevaluatematurationalprocesses includingleftlanguagelateralizationfunctionsandincreasedspecializationofbrainregions importantforsyntacticprocessing.Weutilizedmultimodalmethods,includingindicesof brainactivityfromfMRIduringasyntacticprocessingtask,corticalthickness measure-mentsfromstructural MRI,andneuropsychologicalmeasures.Toevaluatehypotheses aboutincreasinglateralizationandspecializationwithdevelopment,weexamined rela-tionshipsbetweencorticalthicknessandmagnitudeandspatialactivationextentwithin theleftinferiorfrontalgyrus(IFG)anditsrighthemispherehomologue.Wepredictedthat increasedactivationintheleftanddecreasedactivationintherightIFGwouldbe associ-atedwithincreasedsyntacticproficiency.Aspredicted,amorematurepatternofincreased thicknessintherightparstriangulariswasassociatedwithdecreasedactivationintensity andextentintherightIFG.Thesefindingssuggestamaturationalshifttowardsdecreased involvementoftherightIFGforsyntacticprocessing.Bettersyntacticskillswere associ-atedwithincreasedactivationintheleftIFGindependentfromage,suggestingincreased specializationoftheleftIFGwithincreasedproficiency.Overall,ourfindingsshow relation-shipsbetweenstructuralandfunctionalneurodevelopmentthatco-occurwithimproved syntacticprocessingincriticallanguageregionsoftheIFGintypicallydevelopingchildren. © 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Bythetimechildrenbegintoattendschoolataround

agesix,theyarecapableofproducinggrammatically

cor-rectandwell-formedsentencesaboutnumeroustopicsin

∗ Correspondingauthorat:DepartmentofNeurology,DavidGeffen SchoolofMedicineatUCLA,710WestwoodPlaza,Reed1-138,LosAngeles, CA90095-1769,USA.Tel.:+13102675116.

E-mailaddress:[email protected](E.R.Sowell).

conversationswithfamilyandfriends,andonthesurface

itmayappearthattheyhavemasteredthesubtletiesof

thesyntacticrulesoftheirnativelanguage(Scott,2004).

Whiletheunderstandingofindividualwordmeaningsand

expansionofvocabularyisessentialtothedevelopment

of further languageskills, the grammatical or syntactic

structure inwhich individual wordsare contained

con-veyessentialinformationaswell.Syntacticcapacityisa

necessaryprerequisitefor sentencecomprehensionthat

allows the listener to understand “who is doing what

towhom”throughgrammaticalrelationshipsinsentence

(2)

elements (Friederici and Kotz, 2003). Early behavioral

studieshavesuggestedthat semanticand syntactic

pro-cessing are interdependent early in childhood (age 5),

and that syntactic processing gradually becomes

inde-pendentthroughchildhood byabout age10 (Friederici,

1983).However,morerecentstudieshaveshown

contin-ued improvement of syntactic skillsthrough childhood,

adolescenceandintoearlyadulthood(ages20–29),

stabi-lizingonlyinmiddleage(Nippoldetal.,2005).Essentially,

thisprotracteddevelopmentisassociatedwithmasteryof

lessfrequent,moredemandinglinguisticstructurestaking

longertoacquirethanstructuresthataremorefrequently

encountered(Leechetal.,2007).Consistentwiththisidea,

behavioralstudies acrossawider age-range oftypically

developingchildren andadolescents showthat children

graduallybecomefasterandmoreaccurateatinterpreting

andproducingcomplexsentences,aswellasindetecting

grammaticalviolationsinthesesentences(Berman,2004;

Nippoldetal.,2005;Wulfecketal.,2004).

Neuroimagingstudiesarebeginningtoshedlight on

thegradualprocessofspecializationandlateralizationof

syntacticdevelopmentinthebrain.Event-relatedpotential

(ERP)studiesinchildrenprovidefurtherevidencethat

syn-tacticprocessingisnotfullyestablishedneurallyinearly

childhoodbutgraduallydevelopstowardadult-like

pro-cessingduringlatechildhood(reviewedinFriederici,2006;

Hahneetal.,2004).fMRIstudieswithadultshavesuggested

thatsemanticandsyntacticprocessingareneurallydistinct

atleastinpart,andthattheleftparsopercularis

(Brod-mannarea44)maybeparticularlyimplicatedinsyntactic

processingwithother portionsof theIFG alsoinvolved

dependingonthe task(reviewedin Bookheimer, 2002;

Friederici,2002).Syntacticdevelopmenthasbeen

inves-tigatedwithfMRIinyoung(ages5–6)typicallydeveloping

children (Brauer and Friederici, 2007), and methods for

assessingsyntactic processing have includedrule

viola-tiontasksandsentencecomprehensiontasks.Ingeneral,

studieswithchildrenhavenotdemonstratedafully

devel-opedlevelofspecializationorlateralizationforsyntactic

processing.Forexample,BrauerandFriederici(2007)

eval-uateddifferencesinfunctionalactivationbetweenchildren

ages5–6andadultsforsemanticandsyntacticaspectsof

sentenceprocessingwitharuleviolationparadigm.They

foundthatadultsdemonstratedsyntax-speciﬁcactivation

inthesuperiortemporalgyrus(STG)andparsopercularis

whilechildren had signiﬁcantoverlap for semanticand

syntacticprocessing,engagingadditionalareasoftheleft

andrightinferiorfrontalgyrus(IFG),whichareknownto

supportresourcedemandingprocesses.

Alongwithfunctionalchangesinsyntacticprocessingin

childhood,dynamicchangesinbrainstructureoccur

simul-taneously.Manybrainregionsundergocorticalthinning

throughmaturation,includingthedorsalcorticesoffrontal

and parietal lobes, whereas the IFG (Broca’s area) and

posteriorperisylvian cortices(Wernicke’sarea) undergo

corticalthickening(Shawetal.,2008;Sowelletal.,2004a).

Corticalthinningisthoughttobeassociatedwiththe

prun-ingofunderutilizedsynapses andincreasedmyelination

related to increased efﬁciency in these neural systems

(reviewedinSowelletal.,2004b).Corticalthickeningin

theIFGhasbeenassociatedwithimprovedphonological

processingskillswithinindividualsstudiedlongitudinally

(Lu et al., 2007).In addition,structural brain

asymme-tries in perisylvian brain language regions have been

observedperinatallyandthroughoutchildhood,suggesting

thatbrainlateralizationbeginsearlyindevelopmentand

continueslaterinchildhood(Amuntsetal.,2003;Chietal.,

1977;Shawetal.,2009;Wadaetal.,1975).Thesestructural

changesarelikelyrelatedtofunctionalactivationchanges

inthebrainthathavebeenobservedduringthesameage

range.For example,increasedhemisphericlanguage

lat-eralizationwithincreasedagehasbeendemonstratedin

developmentalfMRIstudies,andthereisevidenceforsome

degreeoflateralizationandspecializationearlyin

child-hoodthatmayvaryamongthosewithdifferentlevelsof

languageskills(Hollandetal.,2001;Szaﬂarskietal.,2006).

Toourknowledge,nostudieshavedirectlyexploredthese

relationshipsusingmultimodalbrainimagingmethodsto

examinedevelopmentalchangesintheneuralcorrelatesof

syntacticprocessingorotherlanguageabilities.

Using an fMRI paradigm previously shown to yield

differentactivationpatternsbetweensemanticand

syn-tactic processing of sentences in adults (Dapretto and

Bookheimer,1999),here,structuralandfunctional

corre-lates ofsyntactic processing wereexaminedin a group

ofschool-agedchildrenandyoungadolescents.The

sen-tencecomprehensionparadigmemployedismorecomplex

than rule violation tasks used in previous studies, and

allowsevaluationofsyntacticprocessinginthiswiderage

rangeofchildren.Inaddition,weevaluatedbrain–behavior

relationshipswithneuropsychologicalmeasures.First,we

hypothesizedthattherewouldbenodifferential

activa-tionbetweenthesyntacticandsemanticconditionsofthis

taskacrosstheentireageragestudied,asbehavioraland

ERPstudiessuggestalateindependentemergenceof

syn-tactic processing, but we predicted that we would see

increasedactivationintheleftIFG,primarilypars

oper-cularisoftheIFG(BA44),forthesyntacticconditionthan

thesemanticconditionforasubsetofolderchildren.We

expectedthatbrainactivationpatternswouldshowleft

lat-eralization withdecreased relianceonrighthemisphere

homologous regions and increasedspecializationof the

leftIFGinchildrenwithbettersyntacticskills,regardless

of chronologicalage.In other words,wepredicted that

inter-individualvariabilityinsyntacticskillwouldbetter

predictactivationpatternsthanagealonesuchthatmore

proﬁcientindividualswouldhaveincreasedleft

specializa-tionandlateralization.Further,wepredictedthatincreased

thickness(observedinolderrelativetoyoungerchildren

inpreviousstudiesintheagerangestudiedhere,Sowell

etal.,2004a,b;Shawetal.,2008)inbilateralIFGwouldbe

associatedwithincreasedmagnitude(zvalues)andspatial

extentofactivationintheleftIFG,anddecreasedactivation

intensityandextentintherightIFG.

2. Materialsandmethods

2.1. Participants

Participantswerehealthyandtypicallydeveloping

chil-dren (n=19) from the Los Angeles region who were

(3)

develop-ment.UsingUCLAIRBapprovedprocedures,participants’

parentsconsentedtoparticipationinthestudy,and

chil-drenprovidedbothverbalandwrittenassent.According

to parental report during a structured background and

developmentalinterview, participantshadnohistory of

medicalorneurologicalconditions(e.g.,lowbirthweight,

epilepsy, head injury with loss of consciousness),

psy-chiatric or behavioral disorders (e.g., attention deﬁcit

hyperactivitydisorder,schizophrenia),developmental

dis-ability(e.g.mentalretardationorautism),developmental

languagedisorders,learningdisabilities,orsigniﬁcant

pre-natalexposuretoteratogenssuchasalcohol.Theywere

allnativeEnglishspeakers.Genderwasdistributedevenly

(females=10),andtheiragerangedfrom7.2to15.8years

withameanageof11.1years(s.d.=2.6).Oneparticipant

wasleft-handed.Inspectionof thisparticipant’s

individ-ualactivationmap didnot suggestthat this participant

wasanoutlier,andthesedatawereretainedintheﬁnal

analysis.Childrenwerealsoadministereda

comprehen-sivebatteryofneuropsychologicalmeasures,includingIQ

testingwiththeWechslerIntelligenceScaleforChildren,

FourthEdition(WISC-IV;Wechsler,2003)andsubtestsof

theClinicalEvaluationofLanguageFundamentals,Fourth

Edition(CELF-4;Semeletal.,2003).TheSentence

Assem-blysubtestoftheCELF-4wasofmostinterestinthecurrent

study,asitisadirectmeasureofsyntacticskillinwhich

examineesareaskedtocreatetwodistinctsentencesusing

jumbledphrasesthatareprovided.TheCELF-4isa

well-validatedmeasure,andtheSentenceAssemblysubtestis

believedtotapsyntacticprocessingasitrequires

applica-tionofsyntacticrules.Onlyrawscores(withpossiblerange

between0and19)wereusedbecauseoflimitednormative

dataacrosstheagegroupstudiedhere.

2.2. Experimentalstimuliandactivationparadigm

DuringfMRI,participantsperformedasentence

com-prehensionandjudgmenttaskinwhichtheywereasked

tolistentopairsofsentencesanddecideiftheyhadthe

samemeaning.Theyindicatedtheirresponsesbypressing

buttonsinthescanner.Unknowntotheparticipants,there

weretwoconditionspresentedinablockeddesign:

seman-ticandsyntactic.Inthesemanticcondition,sentencepairs

wereidentical,andmeaningwasvariedbyreplacingone

wordwithadifferentwordormaintainingthemeaningby

replacingawordwithasynonym.Forexample,children

heardthroughearphonesinthescanner,“Thecarisinthe

garage”and“Theautoisinthegarage.”Inthesyntactic

con-dition,sentenceswereeitherinadifferentform(e.g.,active

vs.passivevoice),orword orderdiffered(e.g.,preposed

vs.postposedprepositionalphrases).Forexample,children

heard“Theschoolissouthofthepark”and“Southofthe

schoolisthepark”andwereaskedifthetwo sentences

meantthe samething. Thenumber of sentencesin the

activeandpassiveformswasthesameinbothconditions,

aswasthenumber ofpreposedand postposedphrases.

Thus,bothoverallsyntacticcomplexityandworking

mem-orydemandswerecomparablebetweenbothconditions.

RecordedsentenceswerereadbyanativeEnglishspeaker

witha standardAmericanaccentattherateofone

sen-tencepairevery7.5s.Therewereeighttrials(2sentences

ineachtrial,totalof16sentencesineachblock)ineachof

thetwo60-sblocks,andinitialpresentationofthesemantic

orsyntacticconditionwascounterbalancedevenlyacross

subjects.Ineachcondition,halfofthesentencepairshad

thesamemeaning,andtheotherhalfhadadifferent

mean-ing.A30-s restblockwasplaced beforetheﬁrstblock,

betweenblocks,andafterthelastblock.Stimuliwere

pre-sentedwithMacStim3.2 psychologicalexperimentation

software(CogState,WestMelbourne,Victoria,Australia).

Bothaccuracyandresponsetimewererecorded.

2.3. Dataacquisition

fMRIdatawere acquiredona3 Tesla,Siemens

Alle-gra magnet. A gradient echo EPI sequence was used,

with TE=25ms, 90◦ ﬂip angle, TR=2.5s, slice

thick-ness=3mm/1mmskip,36totalslices,FOV=200mm.The

acquisitionmatrixwas64×64voxels,within-plane

res-olution=3mm×3mm.T2-weightedanatomicalvolumes

were acquired during the same scanning session for

eachsubject,withTE=33ms,90◦ﬂipangle,TR=5s,slice

thickness=3mm/1mmskip,36total slices(wholebrain

coverage),FOV=200mm.Theanatomicalvolumewas

uti-lizedforregistrationofthefunctionaldataforeachsubject,

andfunctionaldatawereregisteredtoanatomicalvolumes

witha6DOFlineartransformation,allowingregistration

tostandardspace(MNI-152space)witha12DOFlinear

transformation.Bothfunctionalandstructuraldatawere

acquiredatalignmenttotheanteriorandposterior

com-missureswhichfacilitatedregistrationtoMNI-152space.

High-resolution T1-weighted structural MRI images

wereacquiredona1.5Tesla,SiemensSonata.Both

func-tionalandstructuralscanswereacquiredonthesameday.

A T-1 weighted3 dimensional (3D) magnetization

pre-paredrapidgradient echo(MPRAGE) protocolwasused

withthefollowingparameters:TR=1900ms,TE=4.38ms,

ﬂipangle=15◦,voxel size=1mm×1mm×1mm,

acqui-sition time=8:08min. Up to four scans per child were

obtainedinordertoimprovethesignal-to-noiseratio.Two

ofthe19participantswithusablefMRIdatadidnothave

useablestructuralscansduetoexcessivemotion.

2.4. Dataprocessing

2.4.1. fMRI

FSL, 4.1 (FMRIB Analysis Group, www.fmrib.ox.

ac.uk/fsl)wasusedtoanalyzeallfMRIdata,anddetailed

assumptionsandmethodsaredescribedelsewhere(Smith

et al., 2004; Woolrich et al., 2009). Preprocessing was

carriedoutwithMCFLIRT(themotioncorrectiontoolin

FSL),toobtain6 parameterestimates ofmotion, which

wereentered intothegenerallinearmodel atthe

indi-vidual subject level as individual explanatory variables

(EVs) witha temporal derivative tostatistically control

for motion. The addition of the temporal derivative to

motionparameterEVseffectivelymodelsrelativemotion,

and time points with very high levels of motion are

down-weightedinthemodel.Atrainedresearchassistant

visuallyinspectedrawfMRIdataforallsubjectstoassess

for signal dropout artifact related toexcessive motion

(4)

excludedbasedontheseinspections.Inaddition,motion

parametervaluesforeachsubjectwereinspected,andin

ordertoensureadequatestatisticalpower,subjectswere

onlyincludedintheanalysisiftheyhad15imagesforeach

conditionwithlessthan2mmofmotionfromthecenter

image.Nosubjectswereexcludedbasedonthiscriterion.

Inordertoincreasethesignaltonoiseratio,allfMRIdata

weresmoothedusinga6mmGaussiankernel.Individual

functionaldatawereregisteredtosubjects’T2-weighted

structuralimagewitha6DOFlinearregistration,utilizing

FMRIB’sLinearImageRegistrationTool,v5.5(FLIRT).For

group analyses, images were normalized into MNI-152

spacewitha12DOFlinearregistrationusingFLIRT.

Uti-lizingadult-basedatlasesforregistrationofchilddatahas

beendemonstratedtobevalidforchildren7yearsofage

andolder(e.g.,Burgundetal.,2002;Kangetal.,2003).

Contrasts at the individual level included

syntac-tic>rest, semantic>rest, semantic>syntactic, and

syn-tactic>semantic. Group means for each contrast were

examined,andZ-statisticimageswerethresholdedusing

clustersdeterminedateitherZ>1.7orZ>2.3.Totheextent

thatwefocusedouranalysesonaprioriregionsofinterest

(ROIs)intheinferiorfrontalgyrus,wedidnotcorrectfor

multiplecomparisonsacrosstheentirebrainvolumefor

mostanalyses.Forthegroupanalysis,meanactivationfor

syntactic>restwastheprimarycontrastexamined,andthe

relationshipbetweenactivation,CELF-4SArawscore,and

age(inmonths)wereexaminedusingthegenerallinear

model.AnROIwascreatedforthegroupmeansyntax>rest

andthesyntax>semanticanalysisfortheolderage

sub-set,usingregionswitha10%probabilityofincludingthe

IFG(regions4,5,6,7,33,41,42)fromtheHarvard-Oxford

proabablisticatlasprovidedinFSL(FMRIBAnalysisGroup,

www.fmrib.ox.ac.uk/fsl).ThisresultedinaliberalROIthat

takesintoaccountvariabilityinIFGlocationacross

individ-ualswhilealsorestrictingtheanalysistofrontallanguage

regionsinsteadofawholebrainanalysis.

2.4.2. sMRI

Preprocessing and deﬁnition of cortical and

subcor-tical gray matter regions on structural images were

conducted in the UCLA Laboratory of Neuro

Imag-ing (LONI) Pipeline Processing Environment (Dinov

et al., 2009; Rex et al., 2003) and using FreeSurfer’s

automated segmentation software (FreeSurfer 4.0.2,

http://surfer.nmr.mgh.harvard.edu), as described in the

workofFischlandDale(Daleetal.,1999;Fischletal.,2002,

1999).Duringpreprocessing,twoMPRAGEacquisitionsfor

eachparticipantwereaveragedtoimprovesignaltonoise

ratios. The resultant imageswere then brain extracted

and gray/white matter boundaries were automatically

delineated. A surface of connectedwhite matter voxels

wasreﬁnedtocreatesubmillimetervoxelresolutionin

thegray/whitematterboundary(Daleetal.,1999;Fischl

etal.,1999).Thegray/whitematterboundarywasthen

deformedoutwardtoestimatethepialsurfacewiththe

following constraints;thesurface needed tobesmooth

andmaintainthenaturaltopologyofthebrain.

Proceduresfortheautomaticquantiﬁcationofgray

mat-terthicknessforavarietyofbrainstructuresaredetailed

byFischlandcolleagues(Fischletal.,2002),andinclude

the hypothesis-driven regions investigated here within

theIFG(leftandrightParsOpercularis,ParsTriangularis,

ParsOrbitalis).Brieﬂy,thicknessofthecorticalribbonwas

computedforeachsubjectacrossbothhemispheres.The

thicknessisdeﬁnedastheshortestdistancebetweenthe

whiteandpialsurfacemodels,whichprovideestimatesof

submillimeterdifferences.Allsubjectswerealignedtoa

commonsurfacetemplateusingahigh-resolution

surface-based averaging technique that aligned cortical folding

patterns.Gyralbasedregionsofinterest(ROIs;described

in Desikanetal.,2006)ona standardbrainprovided in

theFreeSurferpackage(2007standard)werethenmapped

backtoeachsubject’snativespaceT1-weightedimage.The

corticalthicknesssurfacevalueswerethenconvertedinto

a3DvolumebasedontheFreeSurfer-generatedwhite

mat-tersurface(i.e.eachvoxelinthenewvolumewasassigned

thethicknessvalueforcortexatthat x,y, zcoordinate).

WhiteandgraysegmentationsforeachoftheIFGROIswere

combinedtoensurethatallthicknessvaluevoxelswere

capturedbytheROI(seeFig.1).Finally,meanthickness

valuesofcorticalgraymatterineachFreeSurfer-generated

ROIwerecalculated.

2.4.3. fMRIactivationvaluesforROIs

ActivationimagesfromFSLwerealignedtotheMPRAGE

images for each subject using the following steps: (1)

T2-weighted high-resolutionstructuralimagesfromthe

functionalrunwerealignedtotheT1-weightedMPRAGE

image used inthe FreeSurfer analysesdescribed above,

witha full-afﬁneregistrationinFSL’sFLIRTprogram.(2)

The resultant transformation ﬁle was then applied to

thevolumeforthethresholdedactivationcontrast

(syn-tax>rest) to bring theactivation map into T1 MPRAGE

space.ActivationresultsfromFSLwereutilizedtoobtain

mean activation and extent of activation measures in

the FreeSurfer cortical ROIs (i.e. IFG). The number of

suprathresholdactivationvoxelsthatfellintoeachROIwas

thensummedtogetthe“extentofactivation”foreach

sub-ject(seeFig.1).Meanactivationintensityofsuprathreshold

voxels(Zvalues)inthesameROIswasalsocalculated.

Data processing procedures for ROI-based cortical

thickness and activation measures(mean intensity and

activationvolume)areillustratedinFig.1.Allactivation

measureswereoutputintabularformatandmergedwith

thicknessmeasuresandneurobehavioraldataforeach

sub-ject.ThistablewasthenexportedintoSPSS17forstatistical

analysis, including correlation analyseswith Bonferroni

correctionformultiplecomparisons.

3. Results

3.1. Neuropsychologicaltesting

MeanfullscaleIQwas112.7(s.d.=14.7,range88–147),

andmeanverbalcomprehensionindexscore,acomposite

measureofoverallverbalabilities,was116.0(s.d.=18.4,

range83–148).MeanrawscorefortheCELF-4SAwas12.0

(s.d.=5.4)witharangebetween2and19.Thecorrelation

betweenCELF-4SArawscoresand agewasmoderately

(5)

Fig.1.Summaryofmethodsforobtainingstructural(i.e.segmentation,corticalthickness)andfunctional(segmentation-basedROImeanactivationand volume)measurements.TheT1-weightedimageisprocessedthroughFreeSurfer’srecon-allcommandtoobtainathicknesssurfacemapandsegmentation labelsofanatomicalregions.Thicknessvalueswereﬁrstconvertedintoa3Dvolumethatfollowsthecontoursofthewhitemattersurface,andthenthe ROIswereappliedtogetaveragethicknessvalueswithintheIFG.Forthickness,acombinationofbothwhiteandgraymattersegmentationsforeach ROIwasusedtoensurethatallvaluesinthewhitesurface-basedthicknessribbonwerecaptured.FSL-processedactivationmapswereﬁrstregisteredto T1-weightedactivationmapsandthentheROIswereappliedtotheresultanttransformedimagetogetaveragethicknessandextentvalueswithintheIFG ROIs.GraymattersegmentationsforeachROIwereusedtocapturevaluesonthecorticalsurface.

3.2. Taskbehavior

Mean accuracy for the semantic condition was 83%

(s.d.=0.11),andforthesyntacticconditionmeanaccuracy

was88%(s.d.=0.11).Accuracydidnotdiffersigniﬁcantly

between conditions (t=_−1.72, p=0.10). Mean response

time for the semantic condition (5.1s, s.d.=0.36) and

syntactic condition (4.9s, s.d.=0.45) did not differ

sig-niﬁcantly(t=1.37,p=0.19).Meansyntacticaccuracywas

not correlated with age (r=0.24, p=0.32), but there

was a trend level age correlation with mean semantic

accuracy (r=0.43, p=0.07).Response time wasnot

cor-relatedwithageineitherthesyntactic(r=0.22,p=0.38)

or semantic (r=−0.18, p=0.47) condition. Correlations

between in-scanner task performance and

neuropsy-chological test results did not indicate any signiﬁcant

relationships.

3.3. fMRI

Group mean activation for the syntax>rest contrast

includedsuprathreshold(uncorrected)activationin

bilat-eralprimaryauditorycortex,lefttemperoparietalcortex,

leftsuperiorfrontalgyrus(SFG),leftmiddlefrontalgyrus

(MFG),leftinferiorfrontalgyrus(IFG),leftsupplementary

motorarea(SMA),andrightcerebellum(seeFig.2).Results

aresummarizedinatablewithpeakcoordinatesand

z-valuesforregionsshowingactivationinthesyntax>rest

analysis(Table1).An ROI (prefrontal languageregions)

analysis(corrected,z>2.3,p=0.05)includedgroupmean

activationindiscrete andcontinuousleftprefrontal

lan-guageregions,includingleftIFG(seeFig.3).

Asnoted previously,differencesbetweenthe

seman-tic and syntactic conditions were not predicted across

(6)

Fig.2. Syntaxtask:groupaverageactivationmaps.Lateralandmedialwholebrainmapswithactivationonthesurfaceofanindividualbrainregistered toMNI-152space.Axialsectionsdisplayinggroupaverageactivationmapsduringthesyntaxvs.restcontrast.Regionsofsigniﬁcantactivation(Z>2.3and auncorrectedclustersigniﬁcancethresholdofp=0.05)forthesyntax>restcontrast.ZvaluesindicatezcoordinateinMNI-152space.

previousbehavioral andERP studiesthat suggesta late

independentemergence of syntactic processing in

chil-dren.Usingamediansplit(medianage=10.7)toidentify

oldersubjectsinthesample(n=10),anROIanalysis

indi-cateddifferentialactivationbetweenthetwoconditionsin

thissubsetofoldersubjects(seeFig.4),withthe

syntac-ticconditionshowingincreasedactivationrelativetothe

semanticconditioninleftIFG(uncorrected,z>1.7),

specif-icallyparsopercularisoftheIFG(BA44).Thesubgroupof

youngerchildrendidnothaveanydifferentialactivation

Fig.3.SyntaxtaskROIanalysis:groupaverageactivationmaps.Axialsectionsdisplayinggroupaverageactivationmapsduringthesyntaxvs.restcontrast. RegionsofsigniﬁcantactivationintheleftIFG(Z>2.3andacorrectedclustersigniﬁcancethresholdofp=0.05)forthesyntax>restcontrast.Zvalues indicatezcoordinateinMNI-152space.

(7)

Table1

CoordinatesinMNIspaceandpeakactivationvaluesforthesyntacticcondition>restcontrast.

Syntacticcondition>Rest

Anatomicalregions x y z z-Stat.

Parietallobe L −64 −38 20 2.99

L −64 −40 20 2.76

Superiorparietallobe L −48 −34 56 2.38

L −46 −30 56 1.95

Supplementarymotorarea L −8 6 58 4.18

L −6 4 58 3.88

R 2 8 58 2.62

R 2 4 58 2.38

Superiorfrontalgyrus L −4 12 52 3.42

L −6 14 52 3.20

R 2 12 52 2.82

R 4 16 52 1.90

Middlefrontalgyrus L −54 14 34 2.88

L −52 16 36 2.70

Inferiorfrontalgyrus L −44 14 16 3.83

L −50 12 16 3.50 R 44 22 16 3.38 R 46 16 16 3.29 Parsopercularis L −52 14 22 3.74 R 46 14 22 3.83 Parstriangularis L −54 26 8 2.59 R 50 20 4 2.24 Parsorbitalis L −54 26 −10 3.28 R 52 18 −10 2.53 Cerebellum L −40 −76 −26 3.42 R 42 −62 −26 4.68 R 40 −64 −26 3.95

Figure4.Syntax>semanticconditionsROIanalysis:conditioncontrastactivationmaps.Axialsectionsdisplayingconditioncontrastmapsforthe syn-tax>semanticcontrast.Regionsofsigniﬁcantactivation(Z>1.7andanuncorrectedclustersigniﬁcancethresholdofp=0.05)forthesyntax>restcontrast. ZvaluesindicatezcoordinateinMNI-152space.

betweenthesyntacticandsemanticconditions(resultsnot

shown).Therewerenostatisticallysigniﬁcantdifferences

inthesemantic>restcondition(resultsnotshown),andno

speciﬁchypothesesweremadeaboutthiscondition.

3.4. fMRIandsyntacticskillperformance

Simultaneous multiple regression analyses that

includedCELF-IVSArawandageasexplanatoryvariables

revealedsigniﬁcantrelationshipsbetweentestscoresand

activationinlefthemisphereSFG,MFG,andIFG

indepen-dentof age.Toexamine thecontributionof individuals’

dataontheregressionanalysesandtoruleoutspurious

effectspossiblycausedbyoutliers,activationvalues(Z)at

aspeciﬁcpointwithintheleftIFG(MNI-152coordinates

x=-54,y=12,z=16)wereplottedagainstage-residualized

values for CELF-4 Sentence Assembly raw scores (see

Fig.5).

Correlation analyses between activation in the left

hemispherelanguageregionsandother

neuropsychologi-calmeasures(FSIQandVIQ)didnotyieldsigniﬁcantresults,

suggestingeitherthatactivationintheleftIFGisnotrelated

tooverallintellectualfunctioning(FSIQ)oroverallverbal

abilities(VCI),orthatwelackedstatisticalpowertoﬁnd

theseeffects.

3.5. Corticalthicknessandactivationcorrelations

Correlationsbetweencorticalthicknessinbilateralpars

triangularisandactivationintensityandextentinthepars

opercularis,parstriangularisandparsorbitalisare

(8)

Fig.5.AxialmapsillustrateoverlappingvaluesintheIFGforvoxelsthatweremoreactiveduringsyntaxthanrest(fromFig.2),and/orvoxelswhere activa-tionduringsyntaxvs.restwerecorrelatedwithperformanceontheCELF-4sentenceassembly(SA)subtest(pink=groupmeanactivation,yellow=voxels whereactivationwascorrelatedwithCELF-4SAperformanceindependentfromage,red=overlappingregions).Yellowarrowindicateslocationofvalues (Z)intheIFG(x=−54,y=12,z=16inMNI-152space)plottedagainstage-residualizedvaluesforCELF-4SArawscoresinthegraph.Onlysigniﬁcantvoxels withintheIFGareshown;IFGmaskappliedtotheimage.

ofthisanalysis,astwoparticipantsdidnothaveuseable

structuralimaging dataas a resultof excessivemotion.

WhenusingastrictBonferonnicorrection(p<0.05)across

the24statisticaltests(p<0.002),thenegativecorrelation

betweenactivationintensityinrightparsorbitalisand

cor-ticalthicknessinrightparstriangulariswastheonlytest

tosurvive(r=−0.724,p=0.001).Nevertheless,other

cor-relationsofconsiderableinterestarediscussedgivenour

apriorihypotheses.Aspredicted,meancorticalthickness

in the right parstriangularis wasnegatively correlated

withactivationintensity(r=−0.724,p=0.001)and

activa-tionextent(r=−0.676,p=0.003)intherightparsorbitalis,

activation intensity (r=-0.613, p=0.009) and activation

extent(r=−0.577,p=0.015)intherightparsopercularis,

andactivationintensity(r=−0.616,p=0.009)intheright

parstriangularis,withatrendlevelcorrelationin

activa-tionextentforthisregion(r=−0.464,p=0.061).Inother

words,increasedthicknessina subregionoftheRHIFG

wasassociatedwithdecreasedactivationinallRHIFG

sub-regions.SeeFig.6forplotsofrighthemisphere(RH)pars

Table2

Correlationsbetweenparstriangularis(PT)corticalthicknessandintensityandvolumeofactivation(extent)duringsyntacticprocessing(syntax>rest)in leftandrighthemisphereinferiorfrontalgyrusregionsofinterest.

ActivationIntensity(z-values) VolumeofActivation(extent)

Left Right Left Right

r p r p r p r p RightPTthickness Parsopercularis −0.471 0.056 −0.613 0.009 −0.407 0.105 −0.577 0.015 Parstriangularis −0.232 0.371 −0.616 0.009 0.078 0.765 −0.464 0.061 Parsorbitalis −0.347 0.172 −0.724 0.001 −0.246 0.342 −0.676 0.003 LeftPTthickness Parsopercularis −0.262 0.311 −0.452 0.068 −0.242 0.350 −0.386 0.126 Parstriangularis −0.206 0.427 −0.452 0.068 0.024 0.926 −0.453 0.068 Parsorbitalis −0.188 0.470 −0.456 0.066 −0.123 0.637 −0.474 0.055

(9)

Fig.6. RHparstriangularismeanthickness(mm)plottedagainstmeanactivation(z)andactivationvolume(extent)inthreeRHROIsoftheIFG(pars orbitalis,parsopercularis,andparstriangularis)foreachindividual.

triangularismeanthickness(mm)andmeanactivationand

activationvolume(extent)inthethreeRHROIsoftheIFG.

Therewasonlyatrendlevelnegativecorrelationbetween

rightparstriangularismeanthicknessandactivation

inten-sity(r=−0.471,p=0.056)andextent(r=−0.407,p=0.105)

in the left pars opercularis with no other correlations

between rightthickness and left activation/extent.Also

consistentwithourpredictions,thoughonlyattrendlevel

signiﬁcance,werenegativecorrelationsobservedbetween

lefthemisphereparstriangularisthicknessandactivation

intensityandextentinallRHIFGsubregions(seeTable2).

Correlations betweencorticalthicknessin parsorbitalis

and parsopercularisandactivationextentandintensity

werenotsigniﬁcantforleftorrighthemisphereinferior

frontalregionsofinterest.

In order to rule out correlations between activation

and structure that couldsimplyreﬂect general changes

incognitionwithbrainmaturation,additionalcorrelation

analysesweredonebetweenbilateralthicknessand

acti-vation/extent in anotherROI with groupactivation and

locatedoutsidetheIFG,thesuperiorfrontalarea.Results

didnotindicateanysigniﬁcantortrendlevelcorrelations

(resultsnotshown),suggestingthatcorrelationsintheIFG

arenotsolelyrelatedtooverallbrainmaturation.

4. Discussion

The resultsof the current study revealrelationships

betweeninter-individualvariabilityinbrainmorphology

andfunctionalbrainactivitysuggestingincreasing

special-izationandlateralizationoflanguagefunctionsintypically

developingchildren.Speciﬁcally,wefoundnegative

corre-lationsbetweencorticalthicknessandactivationintensity

andextentintherightIFGduringasyntacticprocessing

task.Previousstructuralimagingstudieshaveshownthat

cortexintheperisylvianregionsofthefrontaland

tempo-rallobesincreaseinthicknessduringtheagerangestudied

here(Shawetal.,2008;Sowelletal.,2004a).Althoughwe

cannotmakeassumptions aboutcausality,resultsofthe

currentstudysuggestthatchildrenwiththickercortexin

therightIFGrelylessonrighthemispherelanguageareas

toprocessverballinguisticinformationandrelymoreon

lefthemispherestructures forsyntactic processingthan

childrenwiththinnercortexinthesamebrainregions.

Fur-ther,wereportpositivecorrelationsbetweenactivationin

theleftIFGduringasyntacticprocessingandproﬁciency

ofsyntacticskillsindependentfromage,suggestingthat

increased syntactic proﬁciency impacts brain activation

duringsentence-levellanguageprocessing.

Activationpatternsforthesyntacticcomponentofthe

taskinthecurrentstudywereconsistentwiththe

temporo-frontalnetwork known to beassociated withsyntactic

processing (Friederici and Kotz,2003), but as predicted

basedonearlierbehavioralstudies,thedegreeofbrain

spe-cializationforsyntacticprocessingdoesnotappeartobe

fullyestablishedinschoolagedchildrenandearly

adoles-cents(Nippoldetal.,2005).Supportingaqualitativeshiftin

syntacticprocessingataroundage10(Leechetal.,2007),

adirectcontrastbetweenthesyntacticandsemantictask

showeddifferentialactivationintheleftIFGforthe

syn-tacticconditionforanoldersubsetofchildren(overage

10.7)butnotacrossthewholeagerangedstudied(7–15).

Overallsyntacticcomplexity,workingmemorydemands,

andgeneraltaskdemands(i.e.,“dobothsentenceshave

thesamemeaning?”)werecomparablebetweenboth

con-ditions,andthuswebelievedifferencesinactivationcanbe

attributedtosyntacticprocessing.Signiﬁcantvariabilityin

localizationoflanguagetaskshasbeenreported(Burton

et al., 2001), and it was not expected that the current

studywould demonstratea speciﬁcsyntacticprocessing

region in children acrossthe whole agerange studied.

Rather,we (Luet al., 2009), and others (Gaillard et al.,

2000;Holland et al., 2001;Schlaggar et al.,2002)have

speculatedthat inter-individualvariability in activation

of inferior frontal regions would relate to brain

struc-turalmaturationprocessesthatincludecorticalthickening

in IFG and perisylvian brain–languageregions

(10)

thickness–activationcorrelationsprovidesupportforthis

notion.Bothageandproﬁciencywererelatedtoactivation

intheIFG,andthissuggeststheimportanceofboth

phys-icalmaturityandcontinuedexposuretolanguageinbrain

development.Therelationshipwithproﬁciencyonthe

syn-tacticmeasureandbrainactivationpatterns,which was

statisticallyindependentfromage,is importantbecause

it suggests that proﬁciency on a measure of syntactic

processinginvolvesdynamicinter-relationshipsbetween

brainstructureandfunctionthroughdevelopment.Inother

words,childrenwithbettersyntacticprocessingskillshave

different,perhapsmoreleft-lateralized,structure-function

brain relationshipsthan children of thesame age with

poorersyntacticprocessingskills.

Alongwiththeparsopercularis,theparstriangularis

isbelievedtobeimportantforlanguagecomprehension

and it has alsobeen implicated in syntactic processing

(Bookheimer,2002).Recentstudieshaveshownthat

elec-tricalstimulationinoneregionoftheIFGresultsinactivity

inotherregions,andahighdegreeofconnectivitybetween

thedifferentregions of theIFG hasbeendemonstrated

althoughthefunctionalsigniﬁcanceisunknown(Greenlee

etal.,2007).Itisunclearfromthecurrentstudywhy

thick-nessintheparstriangularishadthestrongestrelationships

withneuralactivityinotherpartsoftheIFG.Futurestudies

wouldbeneﬁtfromfocusingontheroleofthepars

tri-angularisin generaland insyntactic processing,aswell

asitsstructuralandfunctionaltrajectoryindevelopment.

Researchershavealsosuggestedthattheinvolvementof

theIFGinsentencecomprehensionmayreﬂectamore

gen-eralizedcognitivecontrolmechanism(Novicketal.,2005),

andfuturestudieswouldbeneﬁtfromincludingcognitive

controlasavariableofinterestwhenstudyinglanguage

processingandtherole oftheIFG.Futurestudiesmight

alsoincludelargersamplesizessothatadditionalstatistical

analysesthatassesstheroleofthisvariablewhiledirectly

comparingdifferences(e.g.,ANCOVA)couldbeutilized.

Resultsofthis studyareconsistentwithother

devel-opmentallanguageprocessingstudiesthatindicatestrong

languagelateralizationinchildrenoverage6witha

contin-uingprocessofregionalspecializationrelatedtomaturity

(Gaillardetal.,2002,2001;Hollandetal.,2001;Schlaggar etal.,2002).Hollandetal.(2001)found,thatlateralization

changesweremorecloselytiedtotheperiodofacquisition

oflanguagetasksthantogeneralmaturation.Moreover,the

largestchangesinlateralizationoccurredforskillssuchas

verbgenerationthatareacquiredoveralongerperiodof

development.Thelateralizationshiftsinourstudysupport

novelbehavioralstudiesthatindicateincreased

indepen-denceofsyntacticprocessingthroughage10andcontinued

developmentof syntacticskills throughearlyadulthood

(Friederici,1983;Nippoldetal.,2005).

Hypotheses about increased cortical thickness and

increasedactivationintensity/extentinthelefthemisphere

wereonlypartiallysupportedbyourdata.Somedegree

ofleftlateralizationforlanguageisknowntobepresent

very early (Dehaene-Lambertz et al., 2002, 2006), and

itisconceivable thatlefthemispherenetworksarewell

establishedearlywithmaintenanceof righthemisphere

language in place in order to maintain neural

redun-dancy.Withincreaseddevelopment,asthebrainbecomes

moreefﬁcient,itsredundancyandpotentialforplasticity

decreases.Furthermore,theremaybephysiologicallimits

ontheextentofintensityofactivationintheleftIFGthat

havealreadybeenmetintheagerangestudied.

Thisstudyutilized multiplemethods,including

func-tionalMRI,structuralMRI,andneuropsychological

mea-surement, to demonstrate relationships between

func-tional andstructuralbrainmaturationand the

develop-ment of language processing skills. Here, we illustrate

theindependentand prolongeddevelopmentof

syntac-ticprocessinginbrainactivationandstructurethroughout

childhood that had previously been shown only with

behavioralstudies.Furthermore,ourresultssuggestthat

corticalactivationisrelatedtobothageandproﬁciency

with syntactic processing. Finally, the present ﬁndings

illustratetheongoingstructuraldevelopmentofthebrain

throughcorticalthinningandthickeninginchildhoodand

itsrelationshipwithhowthebrainfunctions.

Acknowledgements

This research was supportedby the following

orga-nizations:NationalInstituteofChildHealthand Human

Development(R01HD053893),NationalInstituteonDrug

Abuse(R01DA017831),NationalInstitutesofAlcoholism

and AlcoholAbuse(U01AA017122),NationalCenterfor

ResearchResources(U54RR021813),andMarchofDimes

(6-FY2008-50). Portions of the analysis were also

con-ducted using the UCLA Laboratory of Neuro Imaging

Pipeline processingenvironment. Theauthorsreportno

conﬂictsofinterest.

References

Amunts,K.,Schleicher,A.,Ditterich,A.,Zilles,K.,2003.Broca’sregion: cytoarchitectonicasymmetryanddevelopmentalchanges.J.Comp. Neurol.465,72–89.

Berman,R.A.,2004.Betweenemergenceandmastery:thelong develop-mentalrouteoflanguageacquisition.In:Berman,R.A.(Ed.),Language DevelopmentacrossChildhoodand Adolescence.JohnBenjamins PublishingCompany,Amsterdam/Philadelphia,pp.9–34.

Bookheimer,S.,2002.FunctionalMRIoflanguage:newapproachesto understandingthecorticalorganizationofsemanticprocessing.Annu. Rev.Neurosci.25,151–188.

Brauer,J.,Friederici,A.D.,2007.Functionalneuralnetworksofsemantic andsyntacticprocessesinthedevelopingbrain.J.Cogn.Neurosci.19, 1609–1623.

Burgund,E.D.,Kang,H.C.,Kelly,J.E.,Buckner,R.L.,Snyder,A.Z.,Petersen, S.E.,Schlaggar,B.L.,2002.Thefeasibilityofacommonstereotactic spaceforchildrenandadultsinfMRIstudiesofdevelopment. Neu-roimage17,184–200.

Burton,M.W.,Noll,D.C.,Small,S.L.,2001.Theanatomyofauditoryword processing:individualvariability.BrainLang.77,119–131. Chi, J.G., Dooling, E.C.,Gilles, F.H., 1977. Left-right asymmetries of

thetemporalspeechareasofthehumanfetus.Arch.Neurol.34, 346–348.

Dale,A.M.,Fischl,B.,Sereno,M.I.,1999.Corticalsurface-basedanalysis.I. Segmentationandsurfacereconstruction.Neuroimage9,179–194. Dapretto,M.,Bookheimer,S.Y.,1999.Formandcontent:dissociating

syn-taxandsemanticsinsentencecomprehension.Neuron24,427–432. Dehaene-Lambertz, G., Dehaene, S., Hertz-Pannier, L., 2002. Func-tionalneuroimagingofspeechperceptionininfants.Science298, 2013–2015.

Dehaene-Lambertz,G.,Hertz-Pannier,L.,Dubois,J.,2006.Natureand nur-tureinlanguageacquisition:anatomicalandfunctionalbrain-imaging studiesininfants.TrendsNeurosci.29,367–373.

Desikan,R.S.,Segonne,F.,Fischl,B.,Quinn,B.T.,Dickerson,B.C.,Blacker, D.,Buckner,R.L.,Dale,A.M.,Maguire,R.P.,Hyman,B.T.,Albert,M.S., Killiany,R.J.,2006.Anautomatedlabelingsystemforsubdividingthe

(11)

humancerebralcortexonMRIscansintogyralbasedregionsof inter-est.Neuroimage31,968–980.

Dinov,I.D.,VanHorn,J.D.,Lozev,K.M.,Magsipoc,R.,Petrosyan,P.,Liu, Z.,Mackenzie-Graham,A.,Eggert,P.,Parker,D.S.,Toga,A.W.,2009. Efﬁcient,distributedandinteractiveneuroimagingdataanalysisusing theLONIpipeline.Front.Neuroinform.3,22.

Fischl,B.,Salat,D.H.,Busa,E.,Albert,M.,Dieterich,M.,Haselgrove,C., vanderKouwe,A.,Killiany,R.,Kennedy,D.,Klaveness,S.,Montillo, A.,Makris,N.,Rosen,B.,Dale,A.M.,2002.Wholebrainsegmentation: automatedlabelingofneuroanatomicalstructuresinthehumanbrain. Neuron33,341–355.

Fischl,B.,Sereno,M.I.,Dale,A.M.,1999.Corticalsurface-basedanalysis. II.Inﬂation,ﬂattening,andasurface-basedcoordinatesystem. Neu-roimage9,195–207.

Friederici,A.D.,1983.Children’ssensitivitytofunctionwordsduring sen-tencecomprehension.Linguistics21,717–739.

Friederici,A.D.,2002.Towardsaneuralbasisofauditorysentence pro-cessing.TrendsCogn.Sci.6,78–84.

Friederici,A.D.,2006.Theneuralbasisoflanguagedevelopmentandits impairment.Neuron52,941–952.

Friederici,A.D.,Kotz,S.A.,2003.Thebrainbasisofsyntacticprocesses: functionalimagingandlesion studies.Neuroimage 20(Suppl1), S8–S17.

Gaillard,W.D.,Ahmad,Z.,Balsamo,L.,Sachs,B.,Xu,B.,2002.Language networksunderlyingauditorycomprehensioninyoungchildren iden-tiﬁedwithfMRI.Ann.Neurol.52,S114.

Gaillard,W.D., Hertz-Pannier,L.,Mott,S.H.,Barnett,A.S.,LeBihan,D., Theodore,W.H.,2000.Functionalanatomyofcognitivedevelopment: fMRIofverbalﬂuencyinchildrenandadults.Neurology54,180–185. Gaillard,W.D.,Pugliese,M.,Grandin,C.B.,Braniecki,S.H.,Kondapaneni,P., Hunter,K.,Xu,B.,Petrella,J.R.,Balsamo,L.,Basso,G.,2001.Cortical localizationofreadinginnormalchildren:anfMRIlanguagestudy. Neurology57,47–54.

Greenlee,J.D.,Oya,H.,Kawasaki,H.,Volkov,I.O.,Severson3rd,M.A., Howard3rd,M.A.,Brugge,J.F.,2007.Functionalconnectionswithin thehumaninferiorfrontalgyrus.J.Comp.Neurol.503,550–559. Hahne,A.,Eckstein,K.,Friederici,A.D.,2004.Brainsignaturesofsyntactic

andsemanticprocessesduringchildren’slanguagedevelopment.J. Cogn.Neurosci.16,1302–1318.

Holland,S.K.,Plante,E.,WeberByars,A.,Strawsburg,R.H.,Schmithorst, V.J.,BallJr.,W.S.,2001.NormalfMRIbrainactivationpatternsin chil-drenperformingaverbgenerationtask.Neuroimage14,837–843. Kang,H.C.,Burgund,E.D.,Lugar,H.M.,Petersen,S.E.,Schlaggar,B.L.,2003.

Comparisonoffunctionalactivationfociinchildrenandadultsusing acommonstereotacticspace.Neuroimage19,16–28.

Leech,R.,Aydelott,J.,Symons,G.,Carnevale,J.,Dick,F.,2007.The devel-opmentofsentenceinterpretation:effectsofperceptual,attentional andsemanticinterference.Dev.Sci.10,794–813.

Lu,L.,Leonard,C.,Thompson,P.,Kan,E.,Jolley,J.,Welcome,S.,Toga,A., Sowell,E.,2007.Normaldevelopmentalchangesininferiorfrontal graymatterareassociatedwithimprovementinphonological pro-cessing:alongitudinalMRIanalysis.Cereb.Cortex17,1092–1099. Lu,L.H.,Dapretto,M.,O’Hare,E.D.,Kan,E.,McCourt,S.T.,Thompson,

P.M.,Toga,A.W.,Bookheimer,S.Y.,Sowell,E.R.,2009.Relationships betweenbrainactivationandbrainstructureinnormallydeveloping children.Cereb.Cortex19,2595–2604.

Nippold,M.A.,Hesketh,L.J.,Duthie,J.K.,Mansﬁeld,T.C.,2005. Conversa-tionalversusexpositorydiscourse:astudyofsyntacticdevelopment inchildren,adolescents,andadults.J.SpeechLang.Hear.Res.48, 1048–1064.

Novick,J.M.,Trueswell,J.C.,Thompson-Schill,S.L.,2005.Cognitive con-trolandparsing:reexaminingtheroleofBroca’sareainsentence comprehension.Cogn.AffectBehav.Neurosci.5,263–281. Rex,D.E.,Ma,J.Q.,Toga,A.W.,2003.TheLONIpipelineprocessing

envi-ronment.Neuroimage19,1033–1048.

Schlaggar,B.L., Brown,T.T., Lugar,H.M.,Visscher,K.M.,Miezin,F.M., Petersen,S.E.,2002.Functionalneuroanatomicaldifferencesbetween adultsandschool-agechildrenintheprocessingofsinglewords. Sci-ence296,1476–1479.

Scott, C.M.,2004. Syntacticability inchildren andadolescentswith languageandlearningdisabilities.In:Berman,R.A.(Ed.),Language DevelopmentacrossChildhood andAdolescence.JohnBenjamins PublishingCo,Amsterdam,pp.111–133.

Semel,E.,Wiig,E.H.,Secord,W.A.,2003.ClinicalEvaluationofLanguage Fundamentals,fourthed.(CELF-4).HarcourtAssessment,Inc.,San Antonio,TX.

Shaw,P.,Kabani,N.J.,Lerch,J.P.,Eckstrand,K.,Lenroot,R.,Gogtay,N., Greenstein,D.,Clasen,L.,Evans,A.,Rapoport,J.L.,Giedd,J.N.,Wise,S.P., 2008.Neurodevelopmentaltrajectoriesofthehumancerebralcortex. J.Neurosci.28,3586–3594.

Shaw,P.,Lalonde,F.,Lepage,C.,Rabin,C.,Eckstrand,K.,Sharp,W., Green-stein,D.,Evans,A.,Giedd,J.N.,Rapoport,J.,2009.Developmentof corticalasymmetryintypicallydevelopingchildrenandits disrup-tioninattention-deﬁcit/hyperactivitydisorder.Arch.Gen.Psychiatry 66,888–896.

Smith,S.M.,Jenkinson,M.,Woolrich,M.W.,Beckmann,C.F.,Behrens, T.E.J.,Johansen-Berg,H.,Bannister,P.R.,DeLuca,M.,Drobnjak,I., Flit-ney,D.E.,Niazy,R.,Saunders,J.,Vickers,J.,Zhang,Y.,DeStefano,N., Brady,J.M.,Matthews,P.M.,2004.Advancesinfunctionaland struc-turalMRimageanalysisandimplementationasFSL.Neuroimage23, 208–219.

Sowell,E.R.,Thompson,P.M.,Leonard,C.M.,Welcome,S.E.,Kan,E.,Toga, A.W.,2004a.Longitudinalmappingofcorticalthicknessandbrain growthinnormalchildren.J.Neurosci.24,8223–8231.

Sowell,E.R.,Thompson,P.M.,Toga,A.W.,2004b.Mappingchangesin thehuman cortexthroughoutthespanoflife.Neuroscientist10, 372–392.

Szaﬂarski,J.P.,Schmithorst,V.J.,Altaye,M.,Byars,A.W.,Ret,J.,Plante, E.,Holland,S.K.,2006.Alongitudinalfunctionalmagneticresonance imagingstudyoflanguagedevelopmentinchildren5to11yearsold. Ann.Neurol.59,796–807.

Wada,J.A.,Clarke,R.,Hamm,A.,1975.Cerebralhemisphericasymmetry inhumans.Corticalspeechzonesin100adultsand100infantbrains. Arch.Neurol.32,239–246.

Wechsler,D.,2003.TheWechslerIntelligenceScaleforChildren,fourth ed.ThePsychologicalCorporation,SanAntonio,TX.

Woolrich,M.W.,Jbabdi,S.,Patenaude,B.,Chappell,M.,Makni,S.,Behrens, T.,Beckmann,C.,Jenkinson,M.,Smith,S.M.,2009.Bayesiananalysis ofneuroimagingdatainFSL.NeuroImage45,S173–S186.

Wulfeck,B.,Bates,E.,Krupa-Kwiatkowski,M.,Saltzman,D.,2004. Gram-maticalitysensitivityinchildrenwithearlyfocalbraininjuryand childrenwithspeciﬁclanguageimpairment.BrainLang.88,215–228.