An Electrophysiological Analysis of the Time Course of...

An Electrophysiological Analysis of the Time Course ofConceptual and Syntactic Encoding during Tacit Picture

Naming

Bernadette M Schmitt13 Kolja Schiltz2 Wanda Zaake2 Marta Kutas3and Thomas F Munte24

Abstract

amp A central question in psycholinguistic research is whenvarious types of information involved in speaking (conceptualsemantic syntactic and phonological information) becomeavailable during the speech planning process Competingtheories attempt to distinguish between parallel and serialmodels Here we investigated the relative time courses ofconceptual and syntactic encoding in a tacit picture-namingtask via event-related brain potential (ERP) recordingsParticipants viewed pictures and made dual-choice gono-go

decisions based on conceptual features (whether the depicteditem was heavier or lighter than 500 g) and syntactic features(whether the picturersquo s German name had feminine ormasculine syntactic gender) In support of serial models ofspeech production both the lateralized readiness potential orLRP (related to response preparation) and the N200 (relatedto response inhibition) measures indicated that conceptualprocessing began approximately 80 msec earlier than syntacticprocessing amp

INTRODUCTION

In studying language processing psycholinguists inquireas to rsquo rsquowhatrsquo rsquo sorts of information the brain processesand rsquorsquowhenrsquorsquo this information is processed Within mod-els of speech production it is generally assumed thatgoing from an idea to an utterance involves knowledgeat the conceptual the syntactic and the phonologicallevels of processing Consider for example what takesplace when a speaker wishes to describe a scene where-in a light motor bike is located next to a heavy oneAccording to Jackendoff (1999 see also Levelt Roelofsamp Meyer 1999 Bierwisch amp Schreuder 1992) planningthis utterance involves building a conceptual structurefrom constituents that carry information about the sit-uation the event the state andor the place In thesentence rsquo rsquo the light motor bike is beside the heavymotor bikersquo rsquo for instance the situation is in thersquo rsquopresentrsquo rsquo there is information about objects with prop-erties rsquo rsquo light heavyrsquorsquo and about place (rsquo rsquobesidersquo rsquo) Con-ceptual structure also carries information about therelations between concepts such as the function-argu-ment structure of rsquo rsquobersquo rsquo which in our example maps theobjects and the place into a state (rsquo rsquoxrsquo rsquo is beside rsquo rsquoyrsquo rsquo) Itis to this type of information that we refer when wespeak of the conceptual level Following Jackendoff we

assume that the information about an objectrsquo s weight isencoded as one of its properties at the conceptual level

At the syntactic level information about word classtense number or syntactic gender is accessed and isbuilt into a sentence structure (see Bock amp Levelt 1994for details) Finally at the phonological level informa-tion about the phonological form of a word and asentence becomes accessed including morphologicalsegmental syllabic and prosodic structure (see Levelt1989 Levelt et al 1999 Dell 1986 1988)

These processing levels are considered distinct basedon tip-of-the-tongue phenomenon (for a review seeBrown 1991) investigations of brain-damaged individu-als (Miozzo amp Caramazza 1997 Rapp amp Caramazza1995 Garrett 1992 1995) speech errors (eg Dell1986 1990) and reaction times (Peterson amp Savoy1998 Levelt et al 1991a 1991b Schriefers Meyer ampLevelt 1990) as well as electrophysiological data fromintact individuals (Schmitt Munte amp Kutas 2000 VanTurennout Hagoort amp Brown 1997 1998)

In order to speak fluently these different types ofknowledge about words must be accessed at very highspeeds Thus one central research question is rsquo rsquowhenrsquorsquoinformation about a wordrsquos concepts syntax and pho-nology actually becomes available in real time Behav-ioral data have been taken to suggest that a wordrsquosconceptualsemantic and syntactic properties are re-trieved before its phonological form is available (Peter-son amp Savoy 1998 Dell amp OrsquoSeaghdha 1991 1992

1 Maastricht University 2 University of California San Diego3 Medizinische Hochschule Hannover Otto-von-Guericke-Uni-versitat Magdeburg

copy 2001 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 134 pp 510ndash522

Levelt et al 1991a Schriefers et al 1990) Electro-physiological studies using the lateralized readinesspotential (LRP) and the N200 (a component related toresponse inhibition) have supported this processingsequence Specifically electrophysiological data indicatethat conceptualsemantic information is encoded about90 to 120 msec prior to phonological information(Schmitt et al 2000 Van Turennout et al 1997) andthat syntactic encoding precedes phonological encodingby about 40 msec (Van Turennout et al 1998)

In contrast to the large body of research on therelative time courses of conceptualsemantic versusphonological retrieval and syntactic versus phonologicalretrieval there are no on-line studies of the time coursesof conceptual versus syntactic encoding The theoreticalviews on the temporal processing of prelinguistic con-cepts and their linguistic syntactic realization fall intotwo general categories In a prototypical serial process-ing model prelinguistic conceptual information such asthe abstract features of the intended meaning is pre-sumed to be available prior to syntactic information(Levelt et al 1999 Bock amp Levelt 1994) According tothis view conceptual information is encoded first andserves as the input to the formulator which is respon-sible for syntactic encoding (ie access to word classsyntactic gender)

On an alternative view conceptual and syntacticinformation processing are not serial but rather proceedhand in hand with both types of information availablesimultaneously In Jackendoffrsquo s (1999) formulationalthough syntactic structures and conceptual structuresare generated by independent formation rules there arealso rules interfacing the two An example of such a ruleis one specifying how a syntactic verb-to-object relationmaps to a conceptual action-to-patient relation Forexample parsing the sentence rsquo rsquothe aardvark was bittenby the antrsquo rsquo requires some sort of mapping indicatingthat the sentence object (the ant) is not acted upon butis the actor Of course this would be a different map-ping in the case of an active sentence such as rsquo rsquo the antbites the aardvarkrsquo rsquo wherein the syntactic object is theconceptual patient According to some theories of lan-guage comprehension (see eg Cutler amp Clifton 1999)this syntacticconceptual integration takes place at thesame time that is as the listener is recognizing theword On the assumption that this type of interactiveprocess occurs in language production as well concep-tual and syntactic information would have to be availableat the same time rather than in sequence (see Figure 1for a simplified version of a parallel and a serial process-ing model) The main aim of the present experiment wasto seek evidence for or against the relative timing of theconceptual and syntactic encoding aspects of these twoclasses of language processing models

Previous work has shown the N200 and the LRP to beespecially useful in investigating the time course ofinformation processing The peak latency of the N200

component has proven to be a sensitive index of thetiming of information processing during semantic andphonological encoding (Schmitt et al 2000) Here weuse the N200 and the LRP to examine the relative timecourse of conceptual and syntactic encoding during tacitpicture naming an act we assume to involve planningprocesses similar to those used in overt language pro-duction

The N200

In a gono-go paradigm an N200 is elicited when aperson withholds a response When an individual isasked to respond to one class of stimuli (go trials) andto withhold responses to another class of stimuli (no-gotrials) the event-related potential (ERP) on no-go (rel-ative to go) trials is characterized by a large negativity(1ndash4 m V) between 100 and 300 msec after stimulus onset(N200) especially over fronto-central sites (ThorpeFize amp Marlot 1996 Sasaki Gemba Nambu amp Matsu-zaki 1993 Gemba amp Sasaki 1989 Kok 1986 Pfeffer-baum Ford Weller amp Kopell 1985 Simson Vaughan ampRitter 1977) It has been suggested that size of the N200is a function of neuronal activity required for rsquo rsquoresponseinhibitionrsquorsquo (Sasaki amp Gemba 1993 Jodo amp Kayama1992)

The presence of an N200 under these circumstancesimplies that the information which is used to determinewhether or not a response is to be given must havebeen analyzed that is was available Thus a researcher

Figure 1 An illustration of parallel processing (after Jackendoff 1999)and serial processing (after Bock amp Levelt 1994) of conceptualsyntactic and phonological processing Arrows indicate informationflow

Schmitt et al 511

can vary the nature of the information on which a gono-go decision is based and use the peak latency of theN200 effect (difference between go and no-go ERP) as anupper estimate of the time by which the specific infor-mation must have been encoded

Schmitt et al (2000) used this logic to examine thetime course of semantic and phonological encoding asparticipants viewed simple line drawings on which theyperformed a dual-choice gono-go task In one condi-tion participants were asked to prepare a leftright-handbutton press contingent on semantic information (iewhether the picture was that of an animal or object) andto make a gono-go decision based on phonologicalinformation (ie press the button [go] whenever thename of the picture starts with a vowel sound do notrespond [no-go] whenever the name starts with a con-sonant sound) In another condition the instructionswere reversed The leftright-hand response was contin-gent on phonological information whereas the gono-goresponse was contingent on semantic informationSchmitt et al observed an enhanced N200 for no-gotrials relative to go trials Moreover when the gono-godecision was based on semantic information the N200effect reached its maximum peak around 380 msec afterpicture onset whereas when the gono-go decision wasbased on phonological information the N200 effectpeaked later (sup1470 msec) Taking the peak latency ofthe N200 effect as the moment in time when sufficientinformation is available to decide whether or not torespond we concluded that semantic information wasavailable about 90 msec earlier than phonological infor-mation This difference was not observed in the reactiontimes on go responses which did not differ statisticallyThese N200 data thus demonstrated that the electro-physiological measure is indeed very sensitive to earlyfine-grained processing differences in this task amongothers

The present investigation targets the temporal courseof the availability of conceptual and syntactic informa-tion during tacit naming in a similar fashion by makingthe gono-go decision contingent on each in turn Bycomparing the relative latencies of the two N200 effectsit will be possible to determine whether conceptualencoding precedes follows or is coincident with syn-tactic encoding

The LRP

The LRP is derived from the readiness potential (RPKornhuber amp Deecke 1965) that precedes voluntaryhand movements In contrast to the RP the LRP isusually computed time-locked to the stimulus to whichthe response is given By averaging the lateralizedactivity to responses made with the left and right hand(given contralateral vs ipsilateral recording) any asym-metry that is not related to response preparation cancelsout The LRP is presumed to reflect the average amount

of lateralization specifically related to the motor prepa-ration of the responding hands with larger amplitudesseen over the motor cortex of the hemisphere contrala-teral to the response (eg Kutas amp Donchin 1974) TheLRP allows researchers to see motor-related brain activ-ity prior to an overt or go response (Mulder WijersBrookhuis Smid amp Mulder 1994 Miller Riehle ampRequin 1992) Perhaps more importantly the LRP alsoreveals motor activation in the rsquo rsquoabsencersquo rsquo of any overtresponse That is the LRP reflects preparation to re-spond even when the response is not executed as in thecase of no-go trials (cf Osman Bashore Coles Donchinamp Meyer 1992) These characteristics make the LRP asuitable brain measure for studying the time course ofinformation encoding such as during speech planningand production (Schmitt et al 2000 Van Turennoutet al 1997 1998)

The Experimental Paradigm

The experiment was carried out in German In an initialpractice session participants were asked to name aseries of pictures in order to make sure that they knewtheir intended names which were then used during therecording session During the ERP recording sessionparticipants were asked to make a dual decision that isclassifying the pictures with respect to both a conceptualand a syntactic feature

The conceptual decision was to determine whetherthe pictured object was heavier or lighter than 500 gThis decision was assumed to involve conceptual activa-tion of the depicted object as well as nonlinguisticinformation in this case related to the weight of theobject in the real world The syntactic decision was todetermine whether the depicted objectrsquo s name hadfeminine (rsquo rsquodiersquo rsquo ) or masculine (rsquo rsquoderrsquo rsquo) syntactic genderNote that in German syntactic gender is independentof biological gender (see Van Berkum 1997 for detailedinformation on gender access)

To ensure that the conceptual and the syntacticdecisions were equally difficult we pretested the partic-ipants with the stimulus pictures from the main experi-ment in two simple reaction time conditions In onecondition participants were asked to make a simplerightleft-hand decision based on conceptual informa-tion of the picture (heavier or lighter than 500 g) In theother condition participants were asked to carry out asimple rightleft-hand decision based on syntactic infor-mation (masculine or feminine syntactic gender) Theorder of the blocks as well as the hand responses (left vsright) was counterbalanced The order of the items wasrandomized The results showed a mean reaction timeof 812 msec (SD = 73 msec) for the conceptual decisionand a mean reaction time of 842 msec (SD = 80 msec)for the syntactic condition A paired sample t testshowed that the two conditions did not differ signifi-cantly from each other t(19) = 118 p = 25

512 Journal of Cognitive Neuroscience Volume 13 Number 4

In the main experiment the instruction was forexample rsquo rsquoPress the left button for a light object andthe right button for a heavy object However respondonly if the picturersquos name has feminine gender and notif it has masculine genderrsquo rsquo Thus depending on theoutcome of these two decisions a response was givenwith either the left or right hand or not at all (see Figure2 for an illustration)

In one condition the responding hand will be con-tingent on conceptual information (rsquo rsquohand = conceptrsquo rsquo)and the gono-go decision will be contingent on syntac-tic information (rsquo rsquogono-go = syntaxrsquorsquo ) In this case thenthe timing of the N200 effect (ie the difference be-tween go and no-go response) provides an upper limiton the moment in time by which the syntactic informa-tion for determining whether or not a response is to begiven must be available In the other condition theresponse contingencies will be reversed that is theresponding hand will be contingent on syntactic infor-mation (rsquo rsquohand = syntaxrsquo rsquo) whereas the gono-go deci-sion will be dependent on conceptual information (rsquo rsquogono-go = conceptrsquo rsquo) In this case then the latency of theN200 effect can be taken as an upper limit on theavailability of conceptual information On a serial ac-count of language production conceptual encoding ispresumed to precede syntactic encoding It would seemthen that information of a conceptual nature would beavailable earlier than information of a syntactic natureregardless of the purpose for which it will be used

Given the known relation between the N200 and with-holding a response we would expect to see this differ-ence in information availability in an earlier N200 effectwhen the gono-go decision is based on conceptualrelative to syntactic information By the same token ifconceptual and syntactic information were processedsimultaneously then we would expect to see no differ-ence in the timing of their associated N200 effects

A similar logic applies to the LRPs recorded in dual-choice gono-go paradigm albeit with respect to re-sponse preparation rather than response inhibitionprocesses Based on a large literature we assume thatpeople prepare their response with a particular hand assoon as they have some information available aboutwhich it will be and that this motor preparation isreflected in an LRP If the responding hand is contingenton conceptual information and conceptual encodingprecedes syntactic encoding (as assumed by serial pro-cessing accounts of language production) then an LRPshould develop for both go and no-go trials alikeHowever as soon as the syntactic information is en-coded indicating that no response is to be made theLRPs for go and no-go trials should diverge from eachother specifically the LRP for no-go trials should dropback to baseline as preparation to respond is aban-doned In contrast when the response contingenciesare reversed (ie the responding hand is contingent onsyntactic information and the gono-go decision is con-tingent on conceptual information) no LRP shoulddevelop on no-go trials This is expected on a serialaccount because its proponents assume that the earlyavailability of the conceptual information indicating thatno response is to be given would effectively forestall thedevelopment of any preparation By contrast on aparallel processing account partial information aboutwhether or not a response is to be executed would beavailable from the beginning of processing and in bothresponse contingency conditions Thus we would notexpect to see a no-go LRP develop in either condition

Note that by opting to use the dual-choice gono-goparadigm as a means of probing language productionprocesses we are in the position of making inferencesfrom the ERPs prior to button presses rather than overtnaming We nonetheless contend that this type ofexperiment provides data that are relevant to issueswithin language production for several reasons (1)Pictures are the method of choice for activating infor-mation processing during language production in psy-cholinguistic experiments as they are thought to triggerthe same encoding processes as those that occur natu-rally in a speaker putting an abstract idea into words(see Glaser 1992) For example picture primes seman-tically related to target pictures have been found toinhibit naming of the target pictures compared tounrelated picture primes (Tipper et amp Driver 1988)implicating a common semantic representation for theprime and target Moreover in picturendashword interfer-

Figure 2 This illustration of the experimental design shows anexample from the hand = concept condition The response hand iscontingent on conceptual information The gono-go response iscontingent on syntactic information In the hand = syntax conditionthe pictures were the same but the response contingencies werereversed in that case the response hand would be contingent onsyntactic information and the gono-go response would be contingenton conceptual information (see also Appendix B for a detailedexample)

Schmitt et al 513

ence experiments (Levelt et al 1991a) the planned butnot executed naming of pictures has been shown toinfluence lexical decisions of target words either seman-tically or phonologically These interference effectsamong other results have been taken as evidence forthe assumption that preparing to name a picture in-volves the same semantic andor phonological processesas word processing Furthermore (2) since our partic-ipants were trained to name the pictures before themain recording session was conducted we believe thatthey accessed the correct picture name in the mainsession as well Although they did not say the picturersquosname aloud it is generally assumed that internal speechplanning mechanisms (tacit naming) are automaticallyactivated by the presentation of the picture In addition(3) the correctness of the subjectsrsquo responses serves asevidence that they indeed accessed the correct under-lying linguistic information that is the name of thepicture Finally (4) the dual-choice gono-go paradigmwe chose to use in the present study was based on thesame logic and essentially the same paradigm as thatintroduced by Van Turennout et al (1997 1998) forestimating the time course of semantic syntactic andphonological encoding Although they did have partic-ipants directly name pictures on some (filler) trials theirLRP analyses and inferences were not based on datafrom these rsquo rsquonamingrsquorsquo filler trials but rather from buttonpresses on trials in which no directndash but only delayedndashnaming was involved Van Turennout et al argued thattheir results applied to language production proper thatis to intrinsic access of the picturersquos name upon itsappearance needed to perform the required judgmentsWe appeal to the same logic for the LRP and N200 datareported here

RESULTS

Push-Button Reaction Times

The mean reaction times for correct go responses inthe dual-choice gono-go task were 1001 msec (SD =130) for the hand = syntax condition and 1032 msec(SD = 131) for the hand = concept condition F(119)= 263 ns Incorrect responses (time-outs ie reactiontimes longer than 1500 msec and wrong-hand re-sponse) were excluded from the analysis The errorproportions did not differ for the two response con-tingencies hand = concept 101 (SD = 32) hand =syntax 97 (SD = 33) according to a paired sample ttest t(19) = 09 ns

N200 Analysis

We assume that the increased negativity for no-gotrials in comparison to go trials reflects the momentin time by which the relevant information necessary towithhold a response must have been encoded The

time it takes to encode the relevant information mighttherefore be seen in the peak latencies of the N200effects

Of interest was whether the latency characteristics ofthe N200 effects reliably differ for the two responsecontingency conditions (hand = concept vs hand =syntax) The ERP difference waveform rsquorsquono-go minus gorsquo rsquowas calculated for each of the conditions Figure 3 showsgrand average ERPs for 20 participants at midline sites(going from the front to the back of the head) Bothresponse contingency conditions show an N200 effect(see left and middle column of Figure 3) with no-gotrials being more negative than go trials However ascan be seen in the difference waveforms (no-go minusgo) for both conditions the onsets and latencies of theN200 effects are strikingly different (right panel in Figure3) The N200 effect in the rsquo rsquohand = syntax gono-go =conceptrsquo rsquo condition precedes that in the rsquo rsquohand = con-cept gono-go = syntaxrsquorsquo condition

The statistical comparison of the ERP differencewaveforms at four frontal electrodes (Fp12 F34)supported the above description of the results basedon visual inspection For each participant we measuredthe peak latencies of the N200 effect between 200 and700 msec at each frontal electrode for correct trials(sup1200 trials per condition minus errors) For the peaklatencies an ANOVA was carried out with rsquo rsquoresponsepreparationrsquo rsquo (hand = concept vs hand = syntax) andrsquo rsquoelectrodesrsquo rsquo (the four electrodes) as repeated meas-ures factors The main effect of rsquo rsquo response preparationrsquo rsquowas significant F(119) = 1373 p lt 0015 reflecting adifference in peak latencies When the gono-go deci-sion is contingent on conceptual information (as it isthe case for the rsquo rsquohand = syntaxrsquorsquo condition) the meanpeak latency of the N200 effect is 477 msec (SD = 128msec) In contrast when the gono-go decision iscontingent on syntactic information (hand = concept)the mean peak latency of the N200 effect is 550 msec(SD = 118 msec) The mean latency difference (acrossthe four electrode sites) of the two N200 effects is 73msec Neither the main effect rsquo rsquoelectrodesrsquo rsquo nor theinteraction rsquo rsquoelectrodes pound response preparationrsquo rsquo issignificant

The lack of interaction between frontal electrodesand response preparation indicates that the two N200effects have a similar scalp distribution that is aregenerated by the same neural populations To seewhether this was indeed the case we compared thescalp distributions of the two N200 effects across allelectrode sites as shown in Figure 4 The scalpdistributions for the conceptual and the syntacticN200 effects look very similar and this similarity wasconfirmed statistically For the analysis a differenceERP (no-go minus go) was calculated for each of thetwo response-conditions for each participant Nextgrand average difference waves were created for eachcondition and peak latencies of the N200 effect were


measured We then calculated the mean amplitudes ofeach individualrsquos difference wave in a time windowplusmn40 around the peak latency in the grand meanseparately for each condition (ie for each participantwe measured the mean amplitude between 420 and500 msec for the concept condition and between 520and 600 msec for the syntax condition) The resultingmeasures were normalized according to McCarthy andWood (1985) and analyzed in an ANOVA with rsquo rsquo re-sponse conditionrsquorsquo (concept vs syntax) and rsquorsquoelectro-desrsquo rsquo (30 sites) as repeated measures factors The maineffect of rsquo rsquoresponse conditionrsquorsquo was not significant Themain effect rsquo rsquoelectrodesrsquo rsquo was significant F(29551) =50 GreenhousendashGeisser corrected p lt 01 Mostimportantly the interaction was not significant indi-cating indistinguishable scalp distributions for the twoN200 effects Also clearly visible from Figure 4 is thefrontal location of the effect comparable to the datareported by Thorpe et al (1996)

LRP Analyses

Four average LRP waveforms for each participant werecalculated (1) hand = concept go = syntax (2) hand =

concept no-go = syntax (3) hand = syntax go =concept and (4) hand = syntax no-go = concept

Figure 5 shows the grand average of go LRPs (toppanel) and no-go LRPs (bottom panel) for the tworesponse conditions hand = concept and hand =syntax at middle central electrode sites (C3 and C4)The typical LRP pattern for go trials was obtained in bothconditions The beginnings of a no-go LRP however wasdiscernible only in the hand = concept condition andrsquo rsquonotrsquo rsquo in the hand = syntax condition The dashedvertical lines in the bottom panel of Figure 5 delimitthis no-go effect

The results of the statistical analysis support thisdescriptive analysis LRPs were quantified by mean am-plitude measures relative to the prestimulus baseline(ndash 200 to 0 msec before picture onset) Their onsetlatency was determined by one-tailed serial t tests be-tween 300 and 800 msec after picture onset against zeromean The t tests were carried out stepwise with a stepsize of 4 msec For each test 40-msec worth of data (ieplusmn20 msec around the measured point) was averagedOnset latency was defined as the point at which fourconsecutive t tests were significantly different from zero(in the same direction)

Figure 3 Grand average ERPs(n = 20) on go and no-go trialsin the hand = concept condi-tion (left column) and the hand= syntax condition (middlecolumn) The ERPs were time-locked to picture onset Bothconditions are associated with afrontal negativity (N200) that ismore negative for no-go thanfor go trials In the right col-umn the rsquo rsquono-go minus gorsquorsquodifference waves (ie the N200effects interpreted as responseinhibition) for the two condi-tions are shown superimposedDisplayed are data from 20participants (200 trials per con-dition per subject minus re-jected trials) over five midlineelectrodes (from the front [Fz]to the back [Pz] of the headsee head icon for electrodepositions) Note that the cali-bration bar shows 5 m V for theleft and middle column and 3m V for the difference waves inthe right column Negative vol-tage is plotted up in this and allsubsequent figures

Schmitt et al 515

The mean onset latency for the go LRPs in the hand =concept condition was 472 msec after picture onset(from that time on all t(19) lt ndash 173 all p lt 05) Themean onset latency for the go LRP in the hand = syntaxcondition was 424 msec after picture onset (from thattime on all t(19) lt ndash 173 all p lt 05)

Similar analyses for the onset of no-go LRPs revealed asignificant divergence from baseline for the hand =

concept condition starting at 452 msec which lastedfor approximately 90 msec until 540 msec after pictureonset (in this time interval all t(19) lt ndash 173 all p lt 05)In contrast no significant divergence from baseline wasobserved for the no-go LRP in the hand = syntaxcondition

The two go LRPs were compared in two ways (1) Wecompared the hand = syntax go LRP and the hand =concept go LRP via a serial paired sample t test andfound no significant differences over the entire record-ing epoch (2) We also carried out a fractional peakanalysis in order to make the LRP onset analysis com-parable to that for the N200 peak First per participantand per condition we defined the onset of the LRP asthe moment at which the go LRP reached the 15 of itspeak (maximum) amplitude This method revealed amean onset latency for the hand = concept go LRP of534 msec (SD = 101 msec) and for the hand = syntax goLRP of 555 msec (SD =82 msec) When compared via apaired sample t test these two go LRP onset latencieswere not significantly different

Figure 4 Scalp distribution of normalized mean voltage amplitudes ofthe two N200 effects The upper panel shows the N200 effect for thegono-go = concept condition (mean amplitudes of the time window420ndash500 msec after stimulus onset) The lower panel shows the N200effect for the gono-go = syntax condition (mean amplitudes of thetime window 520ndash600 msec after stimulus onset) Lighter shadesindicate regions of negative voltage amplitudes in the rsquorsquono-go minusgorsquorsquo difference wave (ie the N200 effect) Darker shades indicateregions of positive voltage amplitudes The electrode locations aremarked by black dots The view is from the top of the head thetriangle (nose) indicates the front of the head

Figure 5 Grand average LRP on go trials (top panel) and no-go trails(bottom panel) in the two dual-choice tasks involving hand prepara-tion contingent on conceptual or on syntactic information (20participants 100 trials per condition minus rejected trials) The LRP istime-locked to picture onset The two dashed vertical lines surroundthe no-go effect in the hand = concept condition


DISCUSSION

In this study we used the high temporal resolution ofelectrophysiological measures to estimate the relativetime courses of conceptual and syntactic encodingduring tacit picture naming The availability of thesetwo kinds of information was measured by means oftwo different ERP componentsndash the N200 and theLRPndash in a dual-choice gono-go decision task Ourresults clearly showed that under these conditionsconceptual information becomes available prior to syn-tactic information

N200

Our N200 results speak to the temporal course ofinformation flow related to response inhibition Specif-ically the N200 effect (rsquo rsquono-go minus gorsquo rsquo ERPs) reflectsan upper limit on the time by which information aboutwhether an actual response needs to be made or with-held must have become available In the extant liter-ature on the N200 (see Introduction) this time istypically captured by the mean peak latency of theN200 effect Here we found that the average peaklatency of the N200 effect was 73 msec earlier whenthe gono-go response was contingent on conceptualinformation than when it was contingent on syntacticinformation

It is important to note that this difference in the peaklatencies of the conceptual and syntactic N200 effectscannot be attributed to a difference in the difficultybetween the two decisions Our pretest data showedthat the explicit task of conceptual categorization (iedeciding whether an object weighs more or less than500 g) took about the same amount of time as theexplicit task of syntactic categorization (ie decidingwhether an object has masculine or feminine gender) ina simple reaction time task This result suggests that thetwo tasks we used in the dual-choice gono-go paradigmwere of equal difficulty Accordingly it cannot be arguedthat the conceptual and syntactic processes start inparallel but one finishes earlier (as reflected in an earlierN200) because it is an easier decision Rather we main-tain that the observed difference in the latency of theN200 effects reflects serial access to conceptual andsyntactic information during tacit picture-naming pro-cessing

The N200 effects in our dual-choice gono-go taskare relatively late (sup1477ndash550 msec after stimulus on-set) compared to those reported in simple gono-goparadigms We can only speculate here about potentialreasons for this difference Thorpe et al (1996) forexample obtained an N200 effect based on visualprocessing in the simple gono-go task around 150msec after stimulus onset and thus concluded thatthe decision was achieved in under 150 msec Themedian reaction time for go responses in the Thorpe

et al study was 445 msec approximately 300 mseclater than the peak N200 latency Simple reactiontimes for the conceptual and syntactic analyses inthe current study were around 800 msec Both ofthese processes thus seem to be more difficult thanthe visual processing in the Thorpe et al study andnot surprisingly in combination yielded somewhatlonger dual-choice go responses (sup11000 msec) Natu-rally conceptual and syntactic encoding can only takeplace after at least some initial visual processing sowe expected at least somewhat longer reaction timesthan Thorpe et al It is also possible that the relativelycomplex dual-choice task we used placed some addi-tional nonspeech related cognitive load on the par-ticipants If this was the case some constant delaywould be added to the processing time for a trial andwe could not take the actual latencies of the N200effects as absolute values for the speech inherentprocesses However this would not negate a relativecomparison between the N200 peak latencies in thetwo response conditions which clearly shows an ear-lier N200 for conceptually than syntactically baseddecisions

LRP

In the present study the LRP was assessed in additionto the N200 as a means of tracking the time coursesof conceptual and syntactic encoding (according tothe same experimental logic as in Schmitt et al 2000Van Turennout et al 1997 1998) As detailed in theIntroduction development of a no-go LRP in the hand= concept condition but not in the hand = syntaxcondition was considered crucial for a serial process-ing account for conceptual and syntactic encodingduring tacit naming In fact this is exactly what wefound In the hand = concept condition a modestno-go LRPs developed for approximately 88 msec (iebetween 452 and 540 msec after picture onset) Wesuggest that during this interval the conceptual in-formation served to prepare one of the hands torespond unaffected by syntactic information whichapparently was not yet available for the gono-godecision However the syntactic information necessaryto favor either a go or a no-go decision seems to havebecome available after approximately 88 msec or so(540 msec after picture onset) as reflected in thereturn of the no-go LRP to the baseline The observedno-go LRP effect while small (sup112 m V) is about thesame size as those reported by Van Turennout et al(1997 1998)

Taken together the LRP and N200 data reveal a similarpattern of relative timing of conceptual and syntacticprocessing during tacit naming Both the N200 latencydata and LRP data suggest that conceptual informationbecomes available ahead of syntactic information by

Schmitt et al 517

about 73ndash88 msec As outlined in the Introduction dataof this kind are relevant to the time course of languageproduction because they index information access dur-ing intrinsic picture name processing Granted this linkthe present data support any class of models of speechproduction wherein conceptual encoding begins priorto syntactic encoding these include serial processingmodels (Levelt et al 1991a 1999) as well as variousmodels of cascaded processing (Peterson amp Savoy 1998Dell amp OrsquoSeaghdha 1992) At the same time these datareject parallel processing models such as the oneproposed by Jackendoff (1999)

Traditionally the N200 and LRP are considered to bedistinct ERP components where components are de-fined in terms of their neural generators For examplethey differ in their distributions across the scalp the LRPis not present at frontal sites and the N200 is notpresent at central sites (see Figures 3 and 4) That saidwe do not rule out the possibility of a functional depend-ency between the two insofar as they reflect responseinhibition and response preparation processes for ex-ample along the lines proposed by Fuster (1997) Fusterargued that following early visual processing in theparietal and inferotemporal cortex the processed infor-mation is evaluated by the prefrontal cortex Our frontalN200 effect may reflect this evaluation According toFuster once evaluated the information is sent by theprefrontal cortex to the motor cortex It is this activitythat is presumably reflected in the LRP We should notforget however that the N200 effect is a differencebetween ERPs elicited by go and no-go events whereasthe go LRP reflects preparation for a movement that iseventually executed Thus the exact nature of therelationship between these two ERP components andtheir associated neural generators awaits further inves-tigation

In summary two different ERP components were usedto investigate the temporal course of conceptual andsyntactic encoding during language processing Specifi-cally we used the LRP component (related to responsepreparation) and the N200 effect (related to responseinhibition) to monitor on-line picture processing Bothbrain waves clearly show an earlier onset of conceptuallybased compared to syntactically based processing Inso-far as our task involves tacit picture naming (as is typicallyassumed) the results are relevant to theories of languageproduction as they empirically support a serial or cas-caded processing of conceptual and syntactic encodingrather than a continuous interaction of the two

METHODS

Subjects

Twenty native speakers of German participated in theexperiment (eight women 22 to 30 years of age mean25) All participants were right-handed neurologically

healthy and had normal or corrected-to-normal visionSix other participants were excluded because of exces-sive artifacts (more than 30 of the trials) and technicalfailures

Stimuli

The stimuli were 120 simple black-on-white line draw-ings consisting of two conceptual categories 60 objectsand animals that were heavier than 500 g (eg thepicture of a cow) and 60 objects and animals that werelighter than 500 g (eg the picture of an ant) In each ofthe two categories the names of half of the items hadfeminine syntactic gender (eg rsquo rsquodieFEMKuhrsquo rsquo [the cow])and the other half with masculine syntactic gender (egrsquo rsquoderMASC Eselrsquo rsquo [the monkey]) Twenty pictures wereused as practice stimuli The remaining 100 picturesserved as the targets (25 pictures in each conceptualand syntactic condition see Appendix A) Pictures sub-tended about 88 of visual angle in height and 88 in width(viewing distance 110 m) The depicted items werematched on word frequency using the CELEX database(Baayen Piepenbrock amp Van Rijn 1993) The meanlemma frequency (and standard deviation) per millionwas 557 (58) for the heavymasculine category 588 (80)for the heavyfeminine category 626 (83) for the lightmasculine category and 455 (48) for the lightfemininecategory Statistically these four groups did not differ asshown by an ANOVA on frequency with rsquo rsquogrouprsquo rsquo asfactor F(396) = 028

Design

Each participant received eight different instruction setsIn four of the instruction sets the responding hand wascontingent on conceptual information and the gono-godecision was contingent on syntactic information withleft- and right-hand go and no-go responses counter-balanced for each item For the other four instructionsets the response contingencies were reversed theresponding hand was contingent on syntactic informa-tion and the gono-go decision was dependent onconceptual information again counterbalanced for left-and right-hand gono-go responses for each item Ap-pendix B illustrates the eight different instruction setsand provides an example of the different responses tothe same item in this case the picture of an ant Eachpicture was presented eight times to each participantthat is once per condition The order of conditions wasrandomized across participants

Procedure

Participants were tested individually while seated in asoundproof chamber in front of a computer screenThey were first familiarized with the pictures during apractice block wherein each picture was shown with its


name printed below it for 4 sec In a second practiceblock the pictures were repeated without their namesand the participants were asked to name them aloudndashas fast and as accurately as possible This procedureguaranteed that each participant knew and used theintended names of the pictures during the experimentalrun Then the reaction time pretest was carried out asdescribed in the Introduction

During the recording session the participants did notname the pictures aloud but rather performed a dual-choice gono-go task Because there were eight differentinstructions each participant carried out eight differentdual-choice tasks one per experimental condition (seeAppendix B)

Each condition began with 40 practice trials (10pictures of each conceptualsyntactic category) fol-lowed by 100 experimental trials (25 pictures for eachconceptualsyntactic category presented in two blocksof 50 trials) Between blocks there was a short breakEach block started with three warm-up trials that wereexcluded from further analysis The order of the in-structions was randomized across participants and thesequence of pictures was randomized in every block andfor every participant Each experimental block lastedabout 5 min The entire experiment lasted about 2 hr

A trial began with the presentation of a fixation pointin the middle of a high-resolution 21-in computerscreen After a randomized interval of 1500 to 3000

Appendix A List of the Names of the Pictures Used in the Experiment

Feminine Syntactic Gender Masculine Syntactic Gender

lt 500 g gt 500 g lt 500 g gt 500 g

Ameise (ant) Achse (axle) Apfel (apple) Adler (eagle)

Banane (banana) Ampel (traffic light) Ball (ball) Affe (monkey)

Biene (bee) Ananas (pineapple) Ballon (balloon) Anker (anchor)

Birne (pear) Antilope (antelope) Bugel (hanger) Bar (bear)

Blume (flower) Ente (duck) Drachen (kite) Brunnen (well)

Brezel (pretzel) Fahre (ferry) Fisch (fish) Bus (bus)

Brille (glasses) Giraffe (giraffe) Frosch (frog) Elch (moose)

Dose (tin) Gitarre (guitar) Fuller (pen) Elefant (elephant)

Feder (feather) Gondel (gondola) Gurtel (belt) Esel (donkey)

Fledermaus (bat) Hyane (hyena) Hefter (stapler) Flamingo (flamingo)

Fliege (fly) Kanone (canon) Igel (hedgehog) Gorilla (gorilla)

Kaulquappe (tadpole) Katze (cat) Kamm (comb) Hai (shark)

Kerze (candle) Kommode (chest of drawers) Knopf (button) Hirsch (deer)

Kette (necklace) Kuh (cow) Kreisel (top) Koffer (suitcase)

Krabbe (crab) Lampe (lamp) Loffel (spoon) Lowe (lion)

Libelle (dragonfly) Leiter (ladder) Pfeil (arrow) Mantel (coat)

Maus (mouse) Palme (palm tree) Pilz (mushroom) Papagei (parrot)

Mucke (mosquito) Pyramide (pyramid) Rasierer (razor) Pelikan (pelican)

Pfeife (pipe) Sage (saw) Reisverschluss (zip) Pfau (peacock)

Raupe (caterpillar) Schaukel (swing) Schirm (umbrella) Reifen (tire)

Schere (scissors) Schlange (snake) Schlager (bat) Schwan (swan)

Schnecke (snail) Taube (pigeon) Schlussel (key) Strauss (ostrich)

Schwalbe (swallow) Torte (cake) Schuh (shoe) Tiger (tiger)

Spinne (spider) Trompete (trumpet) Trichter (funnel) Wal (whale)

Spritze (syringe) Ziege (goat) Wurm (worm) Wolf (wolf)

Schmitt et al 519

msec the picture was presented in the center of thescreen for 1500 msec The picture was immediatelyreplaced by the fixation point indicating the start of anew trial

Participants were instructed to rest their arms andhands on the elbow rest of the armchair and to holdtheir thumbs on the left and right response button Ongo trials participants responded by pressing one of thetwo buttons as quickly as possible On no-go trials theydid not press any of the buttons Participants wereinstructed not to speak blink or move their eyes whilethe picture was on the screen

Apparatus and Recordings

Push-button response latencies were measured frompicture onset with the time-out point (the moment intime after which responses were registered as missing)set at 1500 msec Time-outs and errors that is wrongresponses were excluded from further analyses TheEEG was recorded from 30 scalp sites using tin electro-des mounted in an electrode cap with reference electro-des placed at the mastoids Signals were collected usingthe left mastoid electrode as a reference and re-refer-enced off-line to the mean of the activity at the twomastoid processes Blinks and vertical eye movementswere monitored by electrodes placed on the right andleft lower orbital ridge also referred to the left mastoidLateral eye movements were monitored by a bipolarmontage using two electrodes placed on the right andleft external canthus The eye movements were re-corded in order to allow for later off-line rejectionElectrode impedance was kept below 5 klaquo for theEEG and eye-movement recording

Signals were amplified with a bandpass from 0 to 100Hz and digitized at 250 Hz Averages were obtained for1048-msec epochs including a 100-msec prestimulusbaseline period for the N200 component and for the2048-msec epoch (baseline 200 msec) for the LRP Trialsof correct responses were visually inspected Trials con-taminated by eye movements or amplifier blocking with-in the critical time window were rejected from averagingby a computer program using individualized rejectioncriteria The number of rejections did not reliably differfor the two response contingency conditions On aver-age 19 of the trials in the hand = concept conditionand 17 of the trials in the hand = syntax were excludedfrom further analysis (including ERP artifacts and incor-rect responses)

Both the LRP and the N200 were calculated for allelectrode sites For the LRP analysis only those electrodesites close to C40 and C30 were analyzed further as theseyield the largest RPs for hand movements (eg Kutas ampDonchin 1974) For the N200 ERP peak analysis onlyfrontal electrode sites were investigated as for thesethe N200 effect is the largest (Thorpe et al 1996) Thescalp distributions of the two N200 effects were analyzedfor all 30 electrodes

Acknowledgments

The research reported in this article was supported by apostdoctoral fellowship from the San Diego McDonnell-PewCognitive Neuroscience Institute and a grant from the GermanAcademic Exchange Service (DAAD) to BMS DFG grantMU13117-1 to TFM and grants HD22614 and AG08313 toMK We thank Antoni Rodriguez-Fornels for his comments onprevious versions of the article and Henk Jansma for hisgraphical assistance

Appendix B Illustration of the Eight Different Instructions for One Sample Picture of an lsquolsquoAntrsquorsquo

Eight different instructions (1ndash4 hand = concept5ndash8 hand = syntax)

Response type for the picture rsquo rsquoAmeise (die) = antrsquo rsquo(lt 500 g feminine syntactic gender)

1 Press left if gt 500 g press right if lt 500 gpress only if its name has feminine syntactic gender

concept right syntax go

2 Press right if gt 500 g press left if lt 500 gpress only if its name has feminine syntactic gender

concept left syntax go

3 Press left if gt 500 g press right if lt 500 gpress only if its name has masculine syntactic gender

concept right syntax no-go

4 Press right if gt 500 g press left if lt 500 gpress only if its name has masculine syntactic gender

concept left syntax no-go

5 Press right if feminine press left if masculinepress only if lt 500 g

syntax right concept go

6 Press right if masculine press left if femininepress only if lt 500 g

syntax left concept go

7 Press left if masculine press right if femininepress only if gt 500 g

syntax right concept no-go

8 Press left if feminine press right if masculinepress only if gt 500 g

syntax left concept no-go


Reprint requests should be sent to Bernadette SchmittDepartment of Neurocognition Faculty of Psychology Maas-tricht University PO Box 616 6200 MD Maastricht Nether-lands e-mail bschmittpsychologyunimaasnl

REFERENCES

Baayen R H Piepenbrock R amp Van Rijn H (1993) TheCELEX lexical database Philadelphia Linguistic Data Con-sortium University of Pennsylvania

Bierwisch M amp Schreuder R (1992) From concepts to lexicalitems Cognition 42 23ndash60

Bock K amp Levelt W (1994) Language production Gram-matical encoding In M A Gernsbacher (Ed) Handbook ofpsycholinguistics (pp 945ndash984) San Diego AcademicPress

Brown A S (1991) A review of the tip-of-the-tongue experi-ence Psychological Bulletin 109 204ndash223

Cutler A amp Clifton C (1999) Comprehending spoken lan-guage A blueprint of the listener In C Brown amp P Hagoort(Eds) The neurocognition of language (pp 132ndash166)Oxford University Press

Dell G S (1986) A spreading-activation theory of retrieval insentence production Psychological Review 93 283ndash321

Dell G S (1988) The retrieval of phonological forms in pro-duction Tests of predictions from a connectionist modelJournal of Memory and Language 27 124ndash142

Dell G S (1990) Effects of frequency and vocabulary type onphonological speech errors Language and Cognitive Pro-cesses 5 313ndash349

Dell G S amp OrsquoSeaghdha P G (1991) Mediated and con-vergent lexical priming in language production A com-ment on Levelt et al (1991) Psychological Review 98604ndash614

Dell G S amp OrsquoSeaghdha P G (1992) Stages of lexical accessin language production Cognition 42 287ndash314

Fuster J M (1997) The prefrontal cortex Anatomy physiol-ogy and neurophysiology of the frontal lobe PhiladelphiaLippincott-Raven

Garrett M (1992) Disorders of lexical selection Cognition42 143ndash180

Garrett M F (1995) The structure of language processingNeuropsychological evidence In M S Gazzaniga (Ed) Thecognitive neurosciences (pp 881ndash899) Cambridge MITPress

Gemba H amp Sasaki K (1989) Potential related to no-goreaction of gono-go hand movement task with colordiscrimination in human Neuroscience Letters 101 262ndash268

Glaser W R (1992) Picture naming Cognition 42 61ndash105Jackendoff R (1999) The representational structures of the

language faculty and their interactions In C Brown amp PHagoort (Eds) The neurocognition of language (pp 37ndash79) Oxford University Press

Jodo E amp Kayama Y (1992) Relation of a negative ERPcomponent to response inhibition in a gono-go task Elec-troencephalography and Clinical Neurophysiology 82477ndash482

Kok A (1986) Effects of degradation of visual stimuli oncomponents of the event-related potential (ERP) in gonogoreaction tasks Biological Psychology 23 21ndash38

Kornhuber H H amp Deecke L (1965) Hirnpotentia-landerungen bei Willkurbewegungen und passiven Bewe-gungen des Menschen Bereitschaftsndash Potential undreafferente Potentiale [Brain potential changes associatedwith voluntary and passive movements in humans Readi-ness potential and reafferent potentials] Pflugers Archive284 1ndash17

Kutas M amp Donchin E (1974) Studies of squeezing Hand-edness responding hand response force and asymmetry ofreadiness potential Science 186 545ndash548

Levelt W J M (1989) Speaking From intention to articu-lation Cambridge MIT Press

Levelt W J M Roelofs A amp Meyer A S (1999) A theory oflexical access in speech production Behavioral and BrainSciences 22 1ndash75

Levelt W J M Schriefers H Vorberg D Meyer A SPechmann T amp Havinga J (1991a) The time course oflexical access in speech production A study of picturenaming Psychological Review 98 122ndash142

Levelt W J M Schriefers H Vorberg D Meyer A SPechmann T amp Havinga J (1991b) Normal and deviantlexical processing Reply to Dell and OrsquoSeaghdha (1991)Psychological Review 98 615ndash618

McCarthy G amp Wood C C (1985) Scalp distribution of eventrelated potentials An ambiguity associated with analysis ofvariance models Electroencephalography and ClinicalNeurophysiology 62 203ndash208

Miller J Riehle A amp Requin J (1992) Effects of preliminaryperceptual output on neuronal activity of the primary motorcortex Journal of Experimental Psychology Human Per-ception and Performance 18 1121ndash1138

Miozzo M amp Caramazza A (1997) On knowing the auxiliaryof a verb that cannot be named Evidence for the indepen-dence of grammatical and phonological aspects of lexicalknowledge Journal of Cognitive Neurophysiology 9 160ndash166

Mulder G Wijers A A Brookhuis K A Smid H G O M ampMulder L J M (1994) Selective visual attention Selectivecueing selective cognitive processing and selective re-sponse processing In H-J Heinze T F Munte amp G RMangun (Eds) Cognitive electrophysiology (pp 26ndash80)Boston Birkhauser

Osman A Bashore T R Coles M G H Donchin E ampMeyer D E (1992) On the transmission of partial infor-mation Inferences from movement-related brain potentialsJournal of Experimental Psychology Human Perceptionand Performance 18 217ndash232

Peterson R R amp Savoy P (1998) Lexical selection and pho-nological encoding during language production Evidencefor cascaded processing Journal of Experimental Psychol-ogy Learning Memory and Cognition 24 539ndash557

Pfefferbaum A Ford J M Weller B J amp Kopell B S (1985)ERPs to response production and inhibition Electroence-phalography and Clinical Neurophysiology 60 423ndash434

Rapp B C amp Caramazza A (1995) Disorders of lexical pro-cessing and the lexicon In M S Gazzaniga (Ed) The cog-nitive neurosciences (pp 901ndash913) Cambridge MIT Press

Sasaki K amp Gemba H (1993) Prefrontal cortex in the orga-nization and control of voluntary movement In T Ono L RSquire M E Raichle D I Perrett amp M Fukuda (Eds)Brain mechanisms of perception and memory From neu-ron to behavior (pp 473ndash496) New York Oxford UniversityPress

Sasaki K Gemba H Nambu A amp Matsuzaki R (1993) No-go activity in the frontal association cortex of human sub-jects Neuroscience Research 18 249ndash252

Schmitt B M Munte T F amp Kutas M (2000) Electrophy-siological estimates of the time course of semantic andphonological encoding during implicit picture naming Psy-chophysiology 37 473ndash484

Schriefers H Meyer A S amp Levelt W J M (1990) Exploringthe time course of lexical access in language productionPicturendashword interference studies Journal of Memory andLanguage 29 86ndash102

Simson R Vaughan H G Ritter W (1977) The scalp topo-

Schmitt et al 521

graphy of potentials in auditory and visual gonogo tasksElectroencephalography and Clinical Neurophysiology 43864ndash875

Thorpe S Fize D amp Marlot C (1996) Speed of processing inthe human visual system Nature 381 520ndash522

Tipper S P T amp Driver J (1988) Negative priming betweenpictures and words in a selective attention task Evidence forsemantic processing of ignored stimuli Memory and Cog-nition 16 64ndash70

Van Berkum J J A (1997) Syntactic processes in speech

production The retrieval of grammatical gender Cognition64 115ndash152

Van Turennout M Hagoort P amp Brown C M (1997) Elec-trophysiological evidence on the time course of semanticand phonological processes in speech production Journalof Experimental Psychology Learning Memory and Cog-nition 23 787ndash806

Van Turennout M Hagoort P amp Brown C M (1998) Brainactivity during speaking From syntax to phonology in 40milliseconds Science 280 572ndash574


Levelt et al 1991a Schriefers et al 1990) Electro-physiological studies using the lateralized readinesspotential (LRP) and the N200 (a component related toresponse inhibition) have supported this processingsequence Specifically electrophysiological data indicatethat conceptualsemantic information is encoded about90 to 120 msec prior to phonological information(Schmitt et al 2000 Van Turennout et al 1997) andthat syntactic encoding precedes phonological encodingby about 40 msec (Van Turennout et al 1998)

In contrast to the large body of research on therelative time courses of conceptualsemantic versusphonological retrieval and syntactic versus phonologicalretrieval there are no on-line studies of the time coursesof conceptual versus syntactic encoding The theoreticalviews on the temporal processing of prelinguistic con-cepts and their linguistic syntactic realization fall intotwo general categories In a prototypical serial process-ing model prelinguistic conceptual information such asthe abstract features of the intended meaning is pre-sumed to be available prior to syntactic information(Levelt et al 1999 Bock amp Levelt 1994) According tothis view conceptual information is encoded first andserves as the input to the formulator which is respon-sible for syntactic encoding (ie access to word classsyntactic gender)

On an alternative view conceptual and syntacticinformation processing are not serial but rather proceedhand in hand with both types of information availablesimultaneously In Jackendoffrsquo s (1999) formulationalthough syntactic structures and conceptual structuresare generated by independent formation rules there arealso rules interfacing the two An example of such a ruleis one specifying how a syntactic verb-to-object relationmaps to a conceptual action-to-patient relation Forexample parsing the sentence rsquo rsquothe aardvark was bittenby the antrsquo rsquo requires some sort of mapping indicatingthat the sentence object (the ant) is not acted upon butis the actor Of course this would be a different map-ping in the case of an active sentence such as rsquo rsquo the antbites the aardvarkrsquo rsquo wherein the syntactic object is theconceptual patient According to some theories of lan-guage comprehension (see eg Cutler amp Clifton 1999)this syntacticconceptual integration takes place at thesame time that is as the listener is recognizing theword On the assumption that this type of interactiveprocess occurs in language production as well concep-tual and syntactic information would have to be availableat the same time rather than in sequence (see Figure 1for a simplified version of a parallel and a serial process-ing model) The main aim of the present experiment wasto seek evidence for or against the relative timing of theconceptual and syntactic encoding aspects of these twoclasses of language processing models

Previous work has shown the N200 and the LRP to beespecially useful in investigating the time course ofinformation processing The peak latency of the N200

component has proven to be a sensitive index of thetiming of information processing during semantic andphonological encoding (Schmitt et al 2000) Here weuse the N200 and the LRP to examine the relative timecourse of conceptual and syntactic encoding during tacitpicture naming an act we assume to involve planningprocesses similar to those used in overt language pro-duction

The N200

In a gono-go paradigm an N200 is elicited when aperson withholds a response When an individual isasked to respond to one class of stimuli (go trials) andto withhold responses to another class of stimuli (no-gotrials) the event-related potential (ERP) on no-go (rel-ative to go) trials is characterized by a large negativity(1ndash4 m V) between 100 and 300 msec after stimulus onset(N200) especially over fronto-central sites (ThorpeFize amp Marlot 1996 Sasaki Gemba Nambu amp Matsu-zaki 1993 Gemba amp Sasaki 1989 Kok 1986 Pfeffer-baum Ford Weller amp Kopell 1985 Simson Vaughan ampRitter 1977) It has been suggested that size of the N200is a function of neuronal activity required for rsquo rsquoresponseinhibitionrsquorsquo (Sasaki amp Gemba 1993 Jodo amp Kayama1992)

The presence of an N200 under these circumstancesimplies that the information which is used to determinewhether or not a response is to be given must havebeen analyzed that is was available Thus a researcher

Figure 1 An illustration of parallel processing (after Jackendoff 1999)and serial processing (after Bock amp Levelt 1994) of conceptualsyntactic and phonological processing Arrows indicate informationflow

Schmitt et al 511




The LRP














Schmitt et al 513


RESULTS



N200 Analysis








LRP Analyses






Schmitt et al 515








DISCUSSION


N200





LRP



Schmitt et al 517




METHODS

Subjects



Stimuli


Design


Procedure









lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521












The LRP














Schmitt et al 513


RESULTS



N200 Analysis








LRP Analyses






Schmitt et al 515








DISCUSSION


N200





LRP



Schmitt et al 517




METHODS

Subjects



Stimuli


Design


Procedure









lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521















Schmitt et al 513


RESULTS



N200 Analysis








LRP Analyses






Schmitt et al 515








DISCUSSION


N200





LRP



Schmitt et al 517




METHODS

Subjects



Stimuli


Design


Procedure









lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521










RESULTS



N200 Analysis








LRP Analyses






Schmitt et al 515








DISCUSSION


N200





LRP



Schmitt et al 517




METHODS

Subjects



Stimuli


Design


Procedure









lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521










LRP Analyses






Schmitt et al 515








DISCUSSION


N200





LRP



Schmitt et al 517




METHODS

Subjects



Stimuli


Design


Procedure









lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521
















DISCUSSION


N200





LRP



Schmitt et al 517




METHODS

Subjects



Stimuli


Design


Procedure









lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521












METHODS

Subjects



Stimuli


Design


Procedure









lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521















lt 500 g gt 500 g lt 500 g gt 500 g


























Schmitt et al 519







Acknowledgments























REFERENCES






































Schmitt et al 521















Acknowledgments























REFERENCES






































Schmitt et al 521










REFERENCES






































Schmitt et al 521









An Electrophysiological Analysis of the Time Course of...

Documents

Transcript of An Electrophysiological Analysis of the Time Course of...