Bargraphswithstandarderrorsforsignificantmain
effectsoftheERPanalysis.
Grand-meanvoltagesareplottedforelectrodesat
peaksintheERPdistributionatthecorresponding
latencies.
A)Maineffectsoftypicality.
B)Maineffectsoflexicality.
ERPtopographiesofearlymaineffects:
Mapswereinterpolatedandprojectedonatwo-
dimensionalplane.Thep-valuesbasedonpaired2-
tailedt-testsareshownontheleft,whilethe
correspondingERPamplitudesarepresentedonthe
right.F:Front;R:Right.
A)MaineffectofTypicalityat100ms.
B)MaineffectsofLexicalityat210msand240ms.
InteractionbetweenLexicality,Typicalityand
Lateralityat158ms:
A)Grand-meanvoltagesareplottedforselected
posteriorelectrodesasafunctionofelectrode,lexicality
andtypicality.
B)Statisticalparametricmaps(top)andERP
topographies(bottom)fortheTypicalitycontrast,
separatelyforwords(left)andpseudowords(right).
Sourceestimates
computedonthegrand-meanERPdisplayedfortheleft
andtherighthemisphere,respectively.Redandblue
coloursdistinguishthedirectionofeffectsaccordingto
labelswithineachfigure.
A)
B)
A)
B)
Pseudowords
Typical-Atypical
RMSforIndividualConditions
0
0.51
1.52
2.5
-100
0100
200
300
400
500
600
700
Latency(ms)
RMS(uV)
TypicalWords
AtypicalWords
TypicalPseudowords
AtypicalPseudowords
Introduction
ProcedureandMethods
Results
Results
Results
Conclusion
Cognitionand
BrainSciencesUnit
1MRCCognitionandBrainSciencesUnit,Cambridge,UK.
[Q:]WhenwouldyoupreferaSOSSAGEtoaSAUSAGE?
[A:]Atabout100ms.
ERPcorrelatesoforthographictypicalityandlexicalityinwrittenwordrecognition
HaukO1,PattersonK1,WoollamsA1,WatlingL1,PulvermüllerF1,RogersTT1,2
2DepartmentofPsychology,UniversityofWisconsin,Madison,USA
Writtenwordrecognitionreliesondifferentsources
ofinformation,whichcanberoughlyclassifiedintotwo
kinds[1,2,3]:
i)surfaceproperties(e.g.wordlengthorfrequencyof
lettercombinations)
ii)lexico-semanticproperties(e.g.wordfrequencyor
imageability).
Thisstudyaimedatdissociatingthetimecoursesand
neuralsubstratesoftheseprocesses.
Usingaspeededlexicaldecision(LD)task,event-
relatedpotentials(ERP)andminimumnorm
(MN)
sourceestimates,weinvestigatedearlyspatio-
temporalaspectsofcorticalactivationelicitedby
wordsandpseudowordsthatvariedintheir
orthographictypicality,i.e.inthefrequencyoftheir
componentletterpairs(bigrams)andtriplets
(trigrams).
TaskandStimuli
14subjectswereconfrontedoneachtrialwithaword
orapseudoword,andaskedtopressonebuttonwith
theindexfingeroftheirlefthandforwordsandwiththe
middlefingerofthesamehandforpseudowords.
Quartetsofstimuliweregeneratedsuchthatthe
surface-orthographicrelationshipbetweentypical
words(e.g.drew)andatypicalpseudowords(driew)
wasidenticaltotheonebetweenatypicalwords(view)
andtypicalpseudowords(vew)(seetablebelow).The
crucialmanipulationwasthat,inonepairofeach
matchedquartet,thewordwaslessorthographically
typical(asmeasuredbybigramandtrigram
frequencies)thanitspartner.Eachcategory
comprised50items.Thesewerematchedforlength,
frequency,RT/accuracy(typicalvs.atypical),and
numberofpseudo-homophones.
ERPanalysisandsourceestimation
63EEGand2EOGchannels(Neuroscan).Bandpass
1-20Hz.Averagereference.Baseline100ms.Artefact
threshold100uV.Onlytrialswithcorrectresponses
wereanalysed.
PeaksinRMScurveswereselected,andamplitudesat
peakelectrodessubjectedto2-by-2ANOVAs.
Furthermore,p-valuedistributionsarepresentedas
post-hoctestsandtotestforspecificityofeffects.
MNsourceestimates[4]werecomputedongrand-
meandatabeforesubtraction.OnlyvalueswithSNRs
largerthan2areshown.
TimecourseofERPdata:
A)Root-Mean-SquareandERPcurvesforindividual
conditions.Latenciesselectedforanalysisaremarked
fortheRMScurves.
B)Voltagecurvesforselectedelectrodes.
TheearliestERPeffectoftypicalitywasobservedat
~100ms(atypical>typical),andtheearliesteffectof
lexicalityat~200ms(pseudowords>words).Themain
corticalsourcesofbotheffectswerelocalisedinleft
inferiortemporalcortex.Thetwofactorsinteracted
significantlyat~150ms,withatypicalityeffectfor
wordsbutnotpseudowords.Typicalwordsactivated
perisylvianregionsbilaterally,whereasatypicalwords
elicitedstrongersourcecurrentsinleftanterior
inferiortemporalcortex:thelatterregionisthemain
siteofatrophyinpatientswithsemanticdementia,
whohaveparticulardifficultyrecognisingatypicalwords
[1].Ourdatasuggestdistinctbutinteractive
processingstagesinwrittenwordrecognition,with
theinteractionreflectingintegrationofinformationfrom
theearlierform-basedsystemandlaterlexico-semantic
processes.
-1.5-1
-0.50
0.51
1.52
2.53
100ms
284ms
Typical
Atypical
A)Typicality
ERP(µV)
-5-4-3-2-10123
210ms
240ms
512ms
552ms
Words
Pseudowords
B)Lexicality
ERP(µV)
B)LexicalityWords-Pseudowords
p>0.05
p=0.01
p>0.05
p=0.01
0.9µV
-0.9µV
0.8µV
-0.8µV
240ms
210ms
F
R
-6-4-20246
P7
P5
P3
P1
Pz
P2
P4
P6
P8
Electrode
Voltage(uV)
TypicalWords
AtypicalWords
-6-4-20246
P7
P5
P3
P1
Pz
P2
P4
P6
P8
Electrode
Voltage(uV)
TypicalPseudowords
AtypicalPseudowords
1.2µV
-1.2µV
Typical-Atypical
Typical-Atypical
p>0.05
p=0.001
p=0.01
0.9µV
-0.9µV
p>0.05
p=0.001
p=0.01
F
R
TypicalAtypical
Word
Drew
View
Pseudoword
Vew
Driew
p>0.05
p=0.01
0.7µV
-0.7µV
F
R
A)TypicalityTypical-Atypical
References
[1]Rogers,T.T.,LambonRalph,M.A.,Hodges,J.R.,Patterson,K.(2004).
Naturalselection:Theimpactofsemanticimpairmentonlexicalandobject
decision.CognitiveNeuropsychology,21,331-352.
[2]Pulvermüller,F.(2001).Brainreflectionsofwordsandtheirmeaning.Trendsin
CognitiveSciences,5(12),517-524.
[3]Hauk,O.,Pulvermüller,F.(2004).Effectsofwordlengthandfrequencyonthe
humanevent-relatedpotential.ClinicalNeurophysiology,115(5),1090-1103.
[4]Hauk,O.(2004).Keepitsimple:acaseforusingclassicalminimumnorm
estimationintheanalysisofEEGandMEGdata.Neuroimage,21(4),1612-1621.
Contact:[email protected]@wisc.edu
TypicalWords-AtypicalWords
158ms
Words-Pseudowords
210ms
Typical-Atypical
100ms
Front
Back
+0.21
-0.21
+0.12
-0.12
nA/cm2
+0.16
-0.16
nA/cm2
Back
Front
nA/cm2
TypicalWords
TypicalPseudowords
AtypicalWordsAtypicalPseudowords
Words
Typical-Atypical
100ms