Producing morphologically complex words: An ERP study with children and adults

ContentslistsavailableatScienceDirect

Developmental

Cognitive

Neuroscience

jo u r n al ho me p ag e :htt p : / / w w w . e l s e v i e r . c o m / l o c a t e / d c n

Producing

morphologically

complex

words:

An

ERP

study

with

children

and

adults

Mary-Jane

Budd

_,

_Silke

_Paulmann

_,

_Christopher

_Barry

_,

_Harald

_Clahsen

b_{PotsdamResearchInstituteforMultilingualism(PRIM),UniversityofPotsdam,Germany}

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received3May2014 Receivedinrevisedform 18November2014 Accepted19November2014 Availableonline26November2014

Keywords: ERPs Morphology Linguistics Languageproduction

a

b

s

t

r

a

c

t

Awidelystudiedmorphologicalphenomenoninpsycholinguisticresearchisthe plurals-inside-compoundseffectinEnglish,whichistheavoidanceofregularpluralmodifierswithin compounds(e.g.,*ratshunter).Thecurrentstudyemploysevent-relatedbrainpotentials (ERPs)toinvestigatetheproductionofplurals-inside-compoundsinchildrenandadults. WespecificallyexaminedtheERPcorrelatesofproducingmorphologicallycomplexwords in8-year-olds,12-year-oldsandadults,byrecordingERPsduringthesilentproductionof compoundswithpluralorsingularmodifiers.Resultsforbothchildrenandadultsrevealed anegativityinresponsetocompoundsproducedfromregularpluralformswhencompared tocompoundsformedfromirregularplurals,indicatingahighlyspecificbrainresponseto asubtlelinguisticcontrast.Althoughchildrenperformedbehaviourallywithanadult-like patterninthetask,wefoundabroaderdistributionandaconsiderablylaterlatencyin children’sbrainpotentialsthaninadults’,indicatingthateveninlatechildhoodthebrain networksinvolvedinlanguageprocessingaresubjecttosubtledevelopmentalchanges.

©2014TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCC BY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction

Whilelanguageproductioninadultspeakershasbeen studied extensively,relatively little isknown aboutthe processesinvolvedinchildren’sproductionofwordsand sentences.Modelsoftheadultspeakerposita sequence of steps during production, from conceptual encoding via lemma selection, morphological and phonological encoding, and finally motor execution and articulation (e.g.,Indefreyand Levelt,2004).Thetemporal sequenc-ing of theseprocesses has beenstudied in detail using behaviouralandneurophysiological measuresleadingto theproposalthatmorphologicalencodingoccursatabout 250–330msafterthestimulus,whichislaterthansemantic

∗ Correspondingauthor.Tel.:+4401206873422. E-mailaddress:[email protected](S.Paulmann).

encoding (175–250ms) but earlier than phonological (330–455ms)encoding(seee.g.,StrijkersandCosta,2011; Janssen et al., 2011). Neurophysiological studies of the time-courseunderlyinglanguageproductionprocessesin childrenareparticularlyrare(Buddetal.,2013),andthe questionofwhetherthetemporalsequencingoflanguage production processes posited for adults also holds for childrenremainslargelyunanswered.Anumberof previ-ousbrain-imagingstudieshave reporteddevelopmental changesofthebrainnetworksinvolvedinlanguage pro-duction.For instance,it hasbeenproposedthatfocused left-lateralizednetworkscontrolling, forexample, silent naming, have emerged from bilateral networks during latechildhoodandadolescence(e.g.,Lidzbaetal.,2011; Evertsetal.,2009),leadingtospeculationsabouthowthese changesinbraindevelopmentcanimpactonlanguage(and otherhighercognitive)processes.Againstthisbackground, thequestionofwhetheradultsandchildrenmakeuseof

http://dx.doi.org/10.1016/j.dcn.2014.11.002

comparablemechanismsforlanguageproductionbecomes evenmore relevant. Thepresent studycontributes new findingsto this under-researchedissue byinvestigating processesinvolvedinchildren’s(incomparisontoadults’) production of morphologically complex words using ERPs.

Thespecificphenomenonwestudiedis the distribu-tionofsingular and pluralmodifiersinsidecompounds. InEnglish,forexample,compoundswithsingular modi-fiersaremoreacceptablethanthosewithpluralones,and amongstthosewithpluralmodifiers,compoundswith reg-ularpluralmodifiersarelessacceptablethanthosewith irregular ones; compare goose/duck feeder,?geese feeder,

*ducks feeder (e.g., Haskell et al., 2003; Cunnings and Clahsen,2007).Theso-calledplurals-in-compoundseffect issubjecttoanumberofsubtlelinguisticconstraints(see below),whichhavebeenwidelystudiedinthe psycholin-guisticliterature,albeitmainlyinbehaviouralexperiments. Plurals-in-compoundshavealsoplayedaprominentrole indebates concerningthenatureofchildren’s language developmentand in thecontroversy between symbolic rule-based versus associative models of language (e.g., Pinker, 1999). Three-year-old children already demon-strated adult-like knowledge of this phenomenon (e.g.,

Gordon, 1985; Alegre and Gordon, 1996), which they

couldnothavelearntfrommodellingthelinguisticinput, butinsteadmay reflectproperties ofthe innately spec-ifiedarchitectureof thelanguagesystem(Pinker,1999: 208); see Haskell et al. (2003) and Ramscar and Dye (2010)for a differentview.Although compounding has been examinedpreviously in ERP studies (e.g., Koester andSchiller,2008;MacGregorandShtyrov,2013),studies oftheplurals-in-compoundseffectusingbrainmeasures arenotavailable.Theneuralcorrelatesofthe plurals-in-compoundseffectandthequestionofwhetherthereare anydevelopmentalchangesinthebrain’sresponsesto con-trastssuchasducksfeedervs.geesefeederhavetherefore remainedunanswered.Thepresentstudyaddressesthese questionsbymeasuringERPsfrom8-to12-year-old chil-drenaswellasadultsduringtheir(silent)productionsof Englishcompoundscontaining(regularvs.irregular)plural modifiers.

1.1. Morphologicalencodingin(adultsand)children

A number of previous experimental studies have comparedprocessesinvolvedintheproductionof morpho-logicallycomplexwordsinchildrenandadults;seeClahsen (2008)forareview.Themostwidelyusedtechniqueisthe speededproductionparadigminwhichparticipantsread bareverbstimuli(e.g.,“collect”)andareaskedtoproducea correspondinginflectedform(e.g.,“collected”),asquickly andasaccuratelyaspossible.Studyingadultspeakersof English,Prasadaetal.(1990)andPradoandUllman(2009) foundsignificantlylongerproductionlatenciesforirregular

past-tenseformswithlowthanforhighfrequencies,but therewasonlyaconsiderablyweakerfrequencyeffectfor theproductionofregularlyinflected(-ed)forms(Prasada etal.,1990;PradoandUllman,2009).Thiscontrastsuggests thatirregularEnglishpasttenseformsareaccessedasfull formsduringproductionwhileregularsarecomposedfrom

theircomponentparts.Thespeededproductiontaskhas alsobeenusedtostudymorphologicalencodinginchildren (e.g.,Clahsenetal.,2004;FleischhauerandClahsen,2012). Testing5-to12-year-oldchildrenindifferentagegroups oninflectedverbformsofGerman,anadvantagefor high-overlow-frequencyirregularformswasfoundinall par-ticipantgroups,paralleltothefindingsontheEnglishpast tense.Thus,abehaviouralmeasure,namelyspeeded pro-ductionlatencies,yieldedadult-likeperformancepatterns in5-to12-year-oldchildren.

The speeded production task has also been applied to examine the plurals-in-compounds effect in English, albeitonlyforadultspeakers.(Buck-Gengleretal.,2004: 459)foundthatwhenparticipantswereprobedwithan irregularpluralformtheytooksignificantlylongerto pro-ducecompoundscontainingsingularnon-headsthanwhen theywereprobedwiththecorrespondingsingularform (e.g.,Tubholdingmiceisa....vs.Crateforcarryingabeadis a...).Still,whenprobedwitharegularpluralform(e.g.,Box fortransportingaxesisan...),therewasnosuchcontrast. Theseresultsindicatethatsingularformsaremoreeasily accessiblefromaregular(thanfromanirregular)plural formduringtheencodingofthecompound.Severalother behavioural studies have used off-line elicitation tasks toexamine children’ssensitivitytomodifierconstraints forcompoundsinproduction(Gordon,1985;Oettingand Rice,1993;vanderLelyandChristian,2000;Clahsenand Almazan,2001;Zukowski,2005).Aconsistentfindingfrom thesestudieswasthatchildrenasyoungasthreeavoided regular plurals inside compounds (e.g., *rats eater), but wereverywillingtoproduceirregularpluralsinside com-pounds (e.g.,miceeater).Despitethereplicabilityofthe plurals-in-compounds effect,theinterpretationof these behaviouralexperimentshasremainedcontroversial;see, for example, Ramscarand Dye (2010)vs. Jaensch et al. (2014).

1.2. Neurophysiologicalmarkersoflanguageproduction

(e.g.,towalk,tofall)andwerethenpromptedtosilently produceeitherthepast-tenseform(e.g.,walked,fell)orthe third-personsingularpresent-tenseform(e.g.,walks,falls). Finally theywerecued toovertlyproduce theinflected word form.ERPsrecorded fromthecuetosilently pro-duce the inflected forms showed a negativity between 300and450msaftercueonsetforregularcomparedto irregularpast-tenseformsinadults.Asimilarbutlonger lastingnegativitywasfoundin10-to12-year-olds.; addi-tionally,children showeda smallpositivity 650–800ms after cue onset. These ERP responses were interpreted assignallingcombinatorialprocessingrequiredby regu-larbutnotirregularpast-tenseforms.Asthisistheonly ERPstudyinvestigating morphologicalencodingin chil-dren,itremainstobeseenwhetherthefindingsreported are replicable and generalizable toother morphological phenomena.

1.3. Thepresentstudy

This study is the first to report results from brain measures during children’s (and adults’) production of compoundswithpluralandsingularmodifiers.InEnglish, compoundsareright-headedwiththeheadnoun denot-ingthekindofobjectthecompoundrefersto.Ahandgun, for example, is a kind of gun, not a kind of hand. A compound’snon-head(ormodifier)istypicallyan unin-flected stem, identical to the singular form of nouns (e.g.,handinhandgun).If,however,thenon-headappears in pluralform, nativeEnglishspeakers ratecompounds withregularpluralmodifiers(e.g.,ducksfeeder)asworse than those with irregular ones (e.g., geese feeder). This contrast has been explained in morphological terms. According to Kiparsky (1982) regular inflectional pro-cesses (e.g.,–s plurals in English) are formed at a later stage than compounding and are therefore unavailable as non-heads of compounds; see Berent and Pinker (2007), Cunnings and Clahsen (2007) and Silva et al. (2013) for experimental evidence and furtherlinguistic background.

We adapted the silent-production-plus-delayed-vocalization taskfromBuddet al.(2013)tothecurrent study.ThistaskhasanumberofadvantagesoverotherERP productiondesigns.Firstly,itminimizes speech-muscle-relatedartefactsoftheEEGsignal,asERPsaretime-locked tothesilentproductionofagivenwordform.Secondly,the delayed-vocalizationpartofthetaskrequiresparticipants toproduceanovertspokenresponseof thiswordform, whichmaximizesthechancethattheEEGrecordingdoes indeedreflectprocessesinproducingthiswordform.We hypothesize that the constraint against regular plurals insidecompoundsismorphologicalinnatureandengages the same combinatorial mechanism as –ed past-tense formation, namely stem+affix (de)composition. We therefore expect tofind ERPs similartothose obtained by Budd et al. (2013), namely an enhanced negativity forcompoundsintheregularpluralconditionrelativeto theirregular one forboth children and adults. In order tocompareresultstopreviouslyreportedfindings(Budd etal.,2013),westudiedthesameagegroupsofchildren, namely8-to12-year-olds.

2. Methods

2.1. Ethicsstatement

Allparticipantsgaveinformedwrittenconsentbefore completingthestudy,whichwasethicallyapprovedbythe UniversityofEssexEthicalReviewBoard.Inthecaseofthe children,theparentsofthechildrenwhoparticipatedgave informedconsent.

2.2. Participants

Seventy-threeright-handednativespeakersofBritish English were tested, all with normal or corrected-to-normal vision, 20 adults (7 men,mean age 23.6years, range18–31years)and53children.Thirteenchildrenwere excluded;two childrencouldnot performthetask,and 11 had highartefactrates (>70%trialslost)during EEG recording.FollowingBuddetal.(2013),theremaining40 children were divided into two agegroups, 11- to 12-year-oldchildren‘Ch-12(n=18,12boys,meanage:11;07, range:11;01–12;08)and‘8-to9-year-oldchildren‘Ch-8

(n=22,10boys,meanage:9;02,range:8;08–9;07).

2.3. Materials

Theexperimentalconditionsandthecriticalnounstobe usedasnon-headsofcompoundswerethesameasinSilva etal.(2013).Eightnounsthattakeirregularplural inflec-tionswere matchedfor frequency, lengthand meaning witheightnounsthattaketheregularplural.Theseitems were:man/menvs. boy/boys,woman/womenvs.girl/girls, child/children vs. baby/babies, foot/feet vs. hand/hands,

tooth/teethvs.eye/eyes,goose/geesevs.duck/ducks,louse/lice

vs.beetle/beetles,mouse/micevs.rat/rats.Participantswere presentedwiththesewordformsaspotentialnon-heads of noun-noun compounds; this yielded fourconditions, irregularpluralnon-heads(‘Pl-I’),regularpluralnon-heads (‘Pl-R’)andthecorrespondinguninflected‘singular’ non-heads(‘Sg-I’,‘Sg-R’).Eachnon-headnounwaspresented withfivedifferentverbs(e.g.,bite,chase)fromwhich com-poundscontainingdeverbal-eragentiveheadnounscould beformed(e.g.boy(s)/man(men)biterandboy(s)/man(men) chaser).Thisyielded160experimentalitems(40per con-dition).The100filleritemsconsistedofheadnounsthat were‘occupations’(e.g.,vet,teacher,dentist)pairedwith non-headsthatwereeitheradjectives(e.g.,red,green,wet, tall)orpluraliatantumnouns(e.g.,maths,measles).

2.4. Procedure

wordwasthenon-head(e.g.,duck)eitherinitsuninflected (singular)formoritsinflected(plural)form.Presentation orderofthetwowordswasalwaysdifferentfromtheir orderinthecompoundstobeproduced.Thiswasalsothe caseforthefilleritems.Thestimuluswasfollowedbya blankscreen,whichvariedinduration(400,600or800ms). Thepresentationdurationoftheblankscreenwas counter-balancedacrossconditions.Tocuethesilentproductionof thecompound,apictureofacartoonfigurewaspresented for2000ms.

Participantswereinstructedtofirstreadthetopand thenthebottomwordandonlysilentlyproducea‘good sounding’compoundwhenthecueappeared(i.e. partici-pantsunderstoodthattheydidnotsimplyhavetoproduce acompoundfromtheexactwordsdisplayedonscreen).If duringthepracticetrialsparticipantswere,forexample, presentedwiththewordssocksandwasherandthey pro-ducedsockswasher,theexperimenterwouldsaytothem: ‘Thatdoesn’tsoundquiteright.Don’tyouthinkthatsock

washersoundsbetter?’Thiscommentwasthenfollowed

byapresentationscreenonwhichthewordsockwasher

wasdisplayed.ERPsweretime-lockedtotheonsetofthis silentproductioncue.Thesilentproductioncuewasthen followedbya1000mslongpresentationofaloudspeaker picturetocueovertproductionofthecompound.Finally, a blank screen was presented for 1500ms. Trials were

pseudo-randomized and distributedover tenblocks (26 itemseach).Participantsonlysawanitemineitherits unin-flected(singular)orpluralforminanyoneblock.Blocks weredividedbyashortbreakandwerecounterbalanced amongstparticipants.Participantswereaskedtominimize eyeandmusclemovementsduringsilentproduction.The run-time of theexperiment wasapproximately 35min. Oneexperimentalsession(includingEEGsetup)lastedfor approximately105min.

2.5. EEGrecordinganddataanalysis

Theadultparticipants’EEGswererecordedusing Neu-roscan(version4.5)acquisitionsoftware,from64electrode sites (see Fig. 1) according to the international 10–20 system using Ag/AgCl sintered electrodes embedded in an elasticcap(Quik-Cap, NeuromedicalSupplies). Bipo-larhorizontalandverticalelectro-oculograms(EOGs)were recorded for artefact rejection purposes. Epochs were extractedfrom200msbeforetheonsetofthesilent pro-ductioncueupto1000msaftercueonset.Recordingswere referencedonlinetotheleftmastoid.Signalswererecorded continuouslywithanon-lineband-passfilterbetween0.1 and70Hzanddigitizedat500Hz.Electrodeimpedances werekeptbelow5K.Recordingswerere-referenced off-linetotheaverageoftheleftandrightmastoidelectrodes,

Fig.2. ERPeffectsfromchildrenatselectedelectrodesites.TheheadontherightshowsROIsusedinthestatisticalanalysis.

band-passfilteredbetween0.1Hzand30Hz,andbaseline corrected.Forgraphicalillustrationpurposesonly,grand averageERPsaresmoothedwitha7Hzlow-passfilter.For thechildparticipants,theEEGwasrecordedfromonly32 electrodesites(seeFig.2).

The EEGdata wasprocessed with EEGLAB(Delorme

and Makeig, 2004). To remove typical muscle and eye

movementartefacts,anindependentcomponentanalysis (ICA)algorithm(Infomax)wasappliedtothedata. Addi-tionally,trialswithartefactswerevisuallyidentifiedand removed.Trialsforwhichaparticipant’sovertproduction wasinappropriate werenot includedin the ERP analy-sis;theseincludedincorrectcompoundswithunreduced regularpluralnon-heads(e.g.,ducksfeeder),single-word responses(e.g.,feeder)andresponsescontainingalexical itemthatwasnotavailablefromthevisualstimulus(e.g.,

eaterinsteadoffeeder).

Overt behavioural responses were analysed using generalizedlinearmixedmodels(lme4functioninR),with respecttowhetherthecandidatenon-headnounformwas maintainedorreducedinsidethespokencompound.For theERPdata,timewindowsofinterestformeanamplitude quantificationwereidentifiedbasedonvisualinspection andalsobasedonpreviouslyreportedfindings(Buddetal., 2013).Fortheadultdata,visualinspectionrevealedthat

effectsbetween300and450mswereprimarilyvisibleat frontalelectrode siteswithapossibledominanceinthe righthemisphere. Thus, two ROIs wereanalysed which included electrodes from left and right frontal (LF/RF) electrode sites. Midline (ML) electrodes were analysed separately(seeFig.1)tokeepthenumberof electrodes constantineachROI.Forthetwochildagegroups,visual inspectionindicatedcontrastsmostprominentlybetween 800 and 900ms, without any strong hemispheric or regionalspecialization.Thus,electrodesfromleftandright frontalsites(LF/RF),leftand rightparietalsites(LP/RP), aswellasmidlineelectrodes(ML)wereanalysedforthis timewindow(SeeFig.2).

3. Results

3.1. Behaviouralresponses

errorswereexcludedfromanyfurtheranalyses.For the remainingdata,wecalculatedproportionsofproductions inwhichtheformofthestimulusnounpresentedto partic-ipantsasacandidateforacompound-internalmodifierwas maintainedintheircompoundproductions(againstcases inwhichitwaschanged)forthethreetypesofcandidate nounforms(singular,irregularplural,regularplural)and theadultandthechildgroups.

Thecompoundresponsesrevealedthatwhenpresented withanuninflectedsingular form(e.g.,duck/goose), this wasalmostalwaysincludedasacompound-internal mod-ifier, irrespective of whether the noun’s corresponding pluralform wasregularor irregular (adults: 99.8%, Ch-12:95.6%,Ch-8:98.8%). Whenthe nounto beincluded asacompound-internalmodifierwasanirregularplural form,it wasstill mostly maintainedinside compounds, butless sothan baresingular forms(adults:91.1%; Ch-12:85.2%;Ch-8:82.3%).A clearcontrast wasfoundfor nounspresentedasregularpluralformswhichwererarely maintainedincompoundsbutinsteadmostlyreducedto bare singulars (adults: 12.5%;Ch-12: 4.3%; Ch-8: 5.7%). To examine these data statistically, we performed an ANOVA on the compound responses with the factors Condition(Singular,IrregularPlural, RegularPlural)and Group(Adults,Ch-12,Ch-8),whichrevealedmaineffects of Group (F(2, 57)=15.3, p<.001) and Condition (F(2, 114)=814.6,p<.001),butnoreliableinteractionof Con-dition by Group (F(4,114)=1.649, p=.167). Subsequent pairwise contrasts showed that irregular plural forms weresignificantlylessoftenmaintainedinsidecompounds thanbaresingularformsinboththeadultandthechild groups(adults:ˇ=4.55,SE=1.05,z=4.34;Ch-12:ˇ=1.26, SE=0.21, z=5.98;Ch-8: ˇ=2.69,SE=0.31, z=8.72). Fur-thermore, regular plural forms were significantly less often maintained in compounds than irregular plural forms(adults:ˇ=5.75,SE=0.26,z=21.44;Ch-12:ˇ=4.99, SE=0.22, z=22.38; Ch-8: ˇ=4.44, SE=0.18, z=24.70). These results suggest the same pattern of behavioural responses for adults and children, with irregular plu-rals less likely to be used than uninflected (singular) forms inside compounds, and regular –s plurals rarely occurringasacompound’snon-head.Itshouldbenoted that thesebehavioural data shouldbe interpreted with caution given the ceiling effects observed in the adult data.

3.2. ERPdata

Meanamplitudes wereextractedforeachparticipant ateach electrode site. After datacleaning, 81%of trials wereincludedinthestatisticalanalysis.Onaverage, simi-larnumbersoftrialswereincludedforthedifferenttypes ofnon-head,baresingulars(adults:84%,ch-12:85%, ch-8:79%), regular plurals (adults: 73%, ch-12: 76%, ch-8: 76%),and irregularplurals (adults:85%,ch-12: 84%, ch-8:81%).Thegrand-averageERPwaveformscanbeseenin Figs.1and2.

3.2.1. Adults

MeanERPamplitudesextractedfromthe300–450ms time-windowwerestatisticallyanalysedwitharepeated

measuresANOVAfor Number(Pluralvs. Singular), Reg-ularity (Regular vs. Irregular), and ‘Regions of Interest’ (ROIs)aswithin-subjectsfactors.Forthefrontalelectrode sites, this analysisrevealed a significant main effect of Regularity(F(1,19)=7.26,MSE=1.56,p<.05)anda Num-ber×Regularity interaction (F(1,19)=4.19, MSE=3.17,

p=.05).Follow-upanalysesbyNumberrevealeda signif-icant Regularity effectwithin theplurals (F(1,19)=7.06, MSE=3.48,p<.05),butnotforthesingulars(F(1,19)=0.03, MSE=1.25,p=.86).Theseresultsshowthatthesourceof the Number×Regularity interactionin the frontal elec-trode sites is due to more negative-going amplitudes for compounds produced fromregular plural forms (as candidatecompound-internalmodifiers)relativetothose producedfromirregularpluralforms,withoutany corre-spondingcontrastinthesingularcondition.Nosignificant effects were found at ML electrode-sites. These results show that adults showa bilateral, frontally distributed negativitywhen aregular (relativetoanirregular) plu-ralformwaspresentedasacandidatenon-headelement for the (silent) production of a compound. This effect is visually illustrated in the left hand panel of Fig. 1. The right-hand panel of Fig. 1 compares the wave-forms for the singular conditions, for nouns (e.g. rat) thathaveregularplurals(‘singular-R’)versusnouns(e.g.

mouse)thathaveirregularplurals(‘singular-I’), illustrat-ing that this comparison indeed shows little difference betweenthetwowaveforms,in linewiththestatistical results.

3.2.2. Children

Regardingthechilddata,considerfirsttheleftsideof Fig. 2which shows that between800and 900msafter thecueonset,ERPsforcompoundsproducedfromregular pluralforms(ascandidatecompound-internalmodifiers) indicatebroadlydistributedmorenegative-going ampli-tudesthanERPs tocompoundsproducedfromirregular pluralforms.TherightsideofFig.2compareswaveforms in response tothesingular-R and singular-Iconditions; visualinspectiondoesnotindicateanyclearcontrastfor this comparison.AnANOVAincludingNumber, Regular-ity and ROI as within-subjects factors and Age Group (‘Ch-8,‘Ch-12)asabetween-participantsfactorwas car-ried out for mean amplitudes extracted between 800 and 900ms after cue onset. This analysis revealed an interactionofNumberxRegularitythatwasapproaching significance(F(1,38)=3.68,MSE=24.35,p=.06)andwhich wasnotqualifiedbyROI.Follow-upanalysesbyNumber showedasignificantRegularityeffectwithintheplurals (F(1,38)=6.32,MSE=29.59,p<.05)butnotwithinthe sin-gulars (F(1,38)=0.0, MSE=33.66, p=.9).This contrast is due to more negative-going amplitudes for compounds producedfromregularpluralformsrelativetothose pro-ducedfromirregularpluralforms,andnocorresponding contrastfor thesingularformsof thenounstested(e.g.

children’sovertcompoundresponses,bothintheregular andtheirregularpluralcondition,hadanuninflected(i.e. reduced)non-head.Noothereffectswerefoundsignificant. Theseresultsshowalate globallydistributednegativity forboth agegroupsofchildren whenaregular(relative toanirregular)pluralformwaspresentedasacandidate non-head elementfor the(silent)production ofa com-pound.

4. Discussion

Theplurals-in-compoundseffecthasreceived consid-erable attention in the psycholinguistic literature and has played a crucial role in more general debates in cognitivescienceconcerningtheroleofgrammarin lan-guageprocessinganddevelopment(e.g.,McClellandand

Patterson, 2002; Pinker and Ullman, 2002). The

cur-rentstudyreports thefirst ERPproduction studyofthe plurals-in-compounds effect in 8- to 12-year-old chil-drenincomparison toadults.Wefoundthat thebrains of both children and adults appear tohonour the sub-tle linguisticconstraints that governthedistribution of compound-internalmodifiers.Specifically,theconstraint againstregularpluralsinsidelexicalcompoundswas sig-nalledbyanenhancednegativityforthe(silent)production ofcompoundsfromregular(comparedtoirregular) plu-ral forms as non-heads. This negativity was found for bothadultsandforchildrenacrosstheagerangetested. Thebehaviouralresultsalsoseemedparallelforthetwo groups,indicatingthat8-to12-year-oldsaresensitiveto theconstraintinthesamewayasadults,thoughinteraction effectsmighthavebeenmaskedbyceilingeffectsobserved for adults. We alsoobserveda number of developmen-talchanges,particularlyintheEEGdata.Firstly,children’s electrophysiological markers wereconsiderably delayed (by∼450msrelativetoadults).Secondly, thenegativity wasfronto-centrallydistributed(andslightlymore right-lateralized)foradults,buthadamuchbroaderdistribution forchildren.

4.1. Temporalsequencingofmorphologicalencoding

TheadultERPfindingsarebroadlyconsistentwiththe temporalsequencingproposedinmodelsofspeech pro-duction(e.g.,IndefreyandLevelt,2004).Specifically,itis hypothesizedthat(inpicturenamingtasks) morpholog-ical encoding occurs around 250–330ms afterstimulus presentationwithanassumedaveragenaminglatencyof 600ms.The natureof oursilent production taskis dif-ferentfromastandardword-namingexperimentinthat wepresentedparticipantswithwrittenwordpromptsfor producingcompounds,ratherthanwithpictures. Further-more,thecompoundstobeproducedwererelativelylong wordsandthuslikely toyieldlongerlatencies.The tim-ing of our adult ERP data is also in line with Koester andSchiller(2008),whoreportedmorphological encod-ingtooccurataround350msafterthestartofconceptual processing,andwithBuddetal.(2013),whoreported mor-phological encoding effects in adults between 350 and 500msaftertheproductioncue.Moreover,intra-cranial EEGrecordings(Sahinetal.,2009)alsoobtainedneural

signalmodulationataround320msaftercueonset. Cru-cially,electrophysiologicalsignalsduring(silent)language productionappear tobereproducible fordifferent mor-phologicalphenomenaandacrossavarietyoftasksand designs.

Withrespecttoitsfunctionalsignificance,Buddetal. (2013)interpretedthenegativityas signalling combina-torialprocessing requiredby regular(butnot irregular) past-tenseforms.We proposethesameaccountfor the typesofcompoundsexaminedinthecurrentstudy. Con-siderthetwokindsofpluralformpresentedasnon-head elements,forexample,(feed)ducks/geese.The morpholog-icalconstraintbansregularpluralsfromappearinginside compounds.Consequently,the–saffixofaregularplural formsuchasducksneedstobestrippedoffbeforebeing enteredasanon-headintotheformationofacompound. Thisprocessofaffix strippingengagesthesame combi-natorialmechanismas–edpast-tense formation,namely stem+affix(de)composition,henceelicitingsimilarbrain responses.

ThecurrentERPresultsareinlinewithfindingsfrom comprehensionandofflinejudgmentstudieson plurals-inside-compounds in adult native speakers of English. Violations of the constraint against compound-internal regular plurals disrupt both written and spoken lan-guage comprehension. Compoundswith regular plurals yielded the longest first fixation and gaze durations in aneye-movementduringreadingexperiment(Cunnings and Clahsen, 2007), indicating that the –s plural con-straintaffects earlystages ofprocessing.Earlyeffectsof thisconstraintwerealsofoundinaneye-movement dur-inglisteningexperiment(Silvaetal.,2013).Furthermore, findings from acceptability judgment tasks (Berent and Pinker,2007;Jaenschetal.,2014)revealedthat morpho-logicalpluralsaredisfavouredinsidecompounds,butnot non-head elements with/s/or/z/sibilant-final codas that merely sound like regular plurals, indicating that the plurals-in-compounds effect is morphological in nature and cannot be accounted for by superficial phonolog-ical properties of the non-head nouns (contra Haskell

et al., 2003; Ramscar and Dye, 2010). The proposed

functional interpretation of the enhanced negativity as signallingmorphologicalstructureisconsistentwiththese findings.

4.2. Developmentalchangesfromchildtoadult

adults(Männel andFriederici,2008).Thirdly,astudyof morphologicalprocessinginlanguagecomprehensionhas alsoreporteddelayedERPeffectsforchildrenwhen com-paredtoadults(Clahsenetal.,2007).Ifweassumeasimilar delayforoursilentproductiontaskasforpicturenaming, atemporaladjustmentoftheobservedERPsforchildrenis required,theresultofwhichputsthenegativityobtained fortheproduction ofcompoundsintoroughly thesame time-frameas for adults. ThattheERPs have a delayed onsetin children suggeststhatthecompound encoding processisfairlyestablishedalbeitslowerinchildren.This interpretation is supported by the current behavioural data which showed that 8- to 12-year-olds produced adult-likecompoundforms.Whileadelayedvocalisation paradigmoffersmanyadvantageswhenworkingwith chil-dren(by,forexample,minimizingmusclemovementsand byallowingchildrentoperformthetaskin theirtime), resultsfromthedelayedvocalisationtaskconcerning time-courserelated information need tobe interpreted with caution.Thus,futurestudiesapplyingdifferentparadigms (including immediate production) are required to fur-thertestthetime-courseofmorphologicalprocessingin children.

However,theenhancednegativityweobtainedwasnot onlydelayed,butalsohadabroaderdistributionin chil-drenthaninadults.Itwasobservedatbothfronto-central andparietalelectrode-sitesbilaterally,whereasinadults thenegativitywasmorelocalized,withthelargesteffect visibleatfronto-centralelectrodesites.Clearly,ERPshave aratherlowspatialresolution;however,topographical dif-ferencesofERPsignaturesaregenerallyacceptedtoreflect neuralsourcedifferences(c.f.Luck,2014).Inotherwords,if ERPpatternsdifferwithregardtobothtemporalandspatial attributes,itislikelythattheneuralsource(s)underlying theeffectsareatleastpartiallydifferent.Thelatter inter-pretationgoeswellwithreports intheliteraturewhich suggestthatthebrainsystemsunderlyinghighercognitive functionsincludinglanguagearestillfine-tunedeven dur-ingadolescence(seeCaseyetal.,2005;Blakemore,2012 forreviews).Specifically,bothcross-sectionaland longitu-dinalfunctionalimagingstudieshavehighlightedthatthe developingbrainundergoessignificantstructuraland func-tionalchangesoffronto-temporalcorticesinchildrenaged 8–12years(i.e.,theagegroupsinvestigatedhere) gener-atedbyacombinationofneuronpruning,myelination,as wellasvascular,neuronalandglialdensitychanges.These developmentalneuroanatomicalchangesarearguedtobe particularlyimportantforgrowingandenhancingefficient neuronalcommunication(Caseyetal.,2005).Hence,the observeddelayedanddifferently distributedERPeffects mayreflect slowerandlessefficientabilities in produc-ingmorphologicallycomplex wordsin children than in adults.

It is noteworthy that despite the observed ERP dif-ferences, children seemed to produce the same overt behaviouralresponsesasadultsinourexperiment.Itis pos-siblethatgroupdifferencesmightnothavebeenobserved becauseofhighresponseaccuracy(i.e.,ceilingeffectscan maskinteractions).Itcouldalsobearguedthatthis discrep-ancyisduetochildrensometimesself-correctinginitially erroneous responses. This is unlikely, however, firstly

becausetheexperimenterencouragedparticipantsnotto change their initialresponse and secondly becausethe samepatternobtainedinthecurrentstudy,viz.adult-like offlineperformancepairedwithdistinctERPresponsesfor children,hasbeenreportedinERPcomprehensionstudies testingmorpho-syntacticandmorphologicalphenomena (Hahneetal.,2004;Clahsenetal.,2007).Thelatterresults havebeentakentoindicatethatdespitesimilaroffline per-formanceadult-likecomprehensionmechanismstaketime todevelop.Our resultsindicatethatthis alsoappliesto processesinvolved intheproductionofmorphologically complexwords.

Finally, a similar pattern of ERP results to the one reportedherewasobtainedbyBuddetal.(2013),astudy employingthesilentproductionparadigmtoexaminepast tense formationin children and adults.In both studies, childrenandadultsrespondedwithanenhanced negativ-itytoregular(whencomparedtoirregular)inflection,but withalateronsetlatencyandamorebroadlydistributed effectforchildrenthanfor adults.Takentogether,these findings lend support tothe assumptionthat thebrain mechanismsunderlyingtheproductionofmorphologically complex wordsmaynot yetbefully adult-likein 8-to 12-year-oldchildren.Specifically,itseemsasifprocessing is slower and less efficient(delayed negativity) aswell aspossiblymoreresource-demandinginchildrenthanin adults.

4.3. Formpriminginthesilentproductiontask?

Anunexpectedresultfromthecurrentstudywarrants commenting.Severalpreviousbehaviouralstudiesof plu-ralsin compoundshavefoundthatirregularplurals are clearlydis-preferredascompound-internalmodifiers rel-ative tobare (singular) forms;seeJaensch et al.(2014) for a recentreview. It istruethat in thecurrentstudy, singularmodifiersweresignificantlymorecommonthan irregularpluralonesinparticipants’overtproductions,for both childrenandadults,butstill irregularplural modi-fiersweremostoftennotreducedtosingularformsinside compounds. This admittedly unexpected finding could, however,bedue toa form-primingeffect(Ramscarand Dye, 2010). In our experiment, participants were con-fronted with a singular or plural form of a noun as a potential compound modifier immediately prior to the compound-productiontask.Thisrecentpresentationofa candidatemodifierformcouldhavebiasedparticipantsto maintainthepresentednounforminsidethecompound,

amorphologicalconstraintagainstcompound-internal–s

pluralsasthesourceforthereportedERPandbehavioural responses.

5. Conclusion

Thisstudyilluminatesthebrainmechanismsinvolved in children’s (as well adults’) language production, by determiningthebrainresponsesobtainedfromthe(silent) production of morphologically complex word forms. Althoughpluralsincompoundshavefeaturedprominently in studies of child language development, the present study isthefirst toreportresultsfrom children’sbrain measures for this phenomenon. Our main findingis an enhanced negativityforcompoundsproduced from reg-ularpluralnon-headsin 8- to12-year-oldchildren and inadults,whichhadalater onsetlatencyandabroader scalpdistributionforchildrenthanforadults.Weinterpret theseERPseffectsasreflecting thecombinatorial mech-anisms involved in producing morphologically complex wordforms.Weconcludethatthe silent-production-plus-delayed-vocalizationtechniqueusedhereisaviableERP paradigmtostudybrainprocessesinvolvedin morpholog-icalprocessinginbothchildrenandadults.

Conflictofinterest

Theauthorsdeclarethattheyhavenoconflictof inter-est.

Acknowledgements

This project has been supported by a grant to H.C. and C.B.fromtheLeverhulmeTrust(F/00213/U)and an Alexander-von-Humboldt Professorship to H.C. We are gratefultoallparticipantsandtheschoolsthatallowedus tocollectdatafromtheirstudents.WealsothankRoger DeeblefortechnicalsupportandAnnaJessenforhelpwith statisticalanalysesinR.

References

Alegre,M.,Gordon,P.,1996.Redratseaterexposesrecursioninchildren’s wordformation.Cognition60,65–82.

Berent,I.,Pinker,S.,2007.Thedislikeofregularpluralsincompounds: phonologicalfamiliarityormorphologicalconstraint?Ment.Lex.2, 129–181.

Blakemore,S-J.,2012.Developmentofthesocialbraininadolescence.J. R.Soc.Med.105,111–116.

Buck-Gengler,C.,Menn,L.,Healy,A.,2004.What‘micetrap’tellsusabout thementallexicon.BrainLang.90,453–464.

Budd,M.J.,Paulmann,S.,Barry,C.,Clahsen,H.,2013.Brainpotentials dur-inglanguageproductioninchildrenandadults:anERPstudyofthe Englishpasttense.BrainLang.127,345–355.

Casey,B.J.,Tottenham,N.,Liston,C.,Durston,S.,2005.Imagingthe devel-opingbrain:whathavewelearnedaboutcognitivedevelopment. TrendsCogn.Sci.9,104–110.

Clahsen,H.,Almazan,M.,2001.Compoundingandinflectioninlanguage impairment:evidencefromWilliamsSyndrome(andSLI).Lingua111, 729–757.

Clahsen,H.,2008.Behavioralmethodsforinvestigatingmorphological andsyntacticprocessinginchildren.In:Sekerina,I.,Fernández,E., Clahsen,H.(Eds.),DevelopmentalPsycholinguistics:OnlineMethods inChildren’sLanguageProcessing.JohnBenjamins,Amsterdam,pp. 1–27.

Clahsen,H.,Hadler,M.,Weyerts,H.,2004.Speededproductionofinflected wordsinchildrenandadults.J.ChildLang.31,683–712.

Clahsen,H.,Lueck,M.,Hahne, A.,2007. Howchildrenprocess over-regularizations:evidencefromevent-relatedbrainpotentials.J.Child Lang.34,601–622.

Cunnings,I.,Clahsen,H.,2007.Thetime-courseofmorphological con-straints:evidencefromeye-movementsduringreading.Cognition 104,467–494.

Delorme,A.,Makeig,S.,2004.EEGLAB:Anopensourcetoolboxforanalysis ofsingle-trialEEGdynamics.J.Neurosci.Methods134,9–21. Everts,R.,Lidzba,K.,Wilke,M.,Kiefer,C.,Mordasini,M.,Schroth,G.,

etal.,2009.Strengtheningoflateralityofverbalandvisuospatial functionsduringchildhoodandadolescence.Hum.BrainMapp.30, 473–483.

Fleischhauer,E.,Clahsen,H.,2012.Generatinginflectedwordformsinreal time:evaluatingtheroleofage,frequency,andworkingmemory.In: Biller,A.,Chung,E.,Kimball,A.(Eds.),Proceedingsofthe36thannual BostonUniversityConferenceonLanguageDevelopment.Cascadilla Press,Somerville,MA,pp.164–176,1.

Ganushchak,L.Y.,Chrisstoffels,I.K.,Schiller,N.O.,2011.Theuseof elec-troencephalographyinlanguageproductionresearch:areview.Front. Psychol.2,208.

Gordon,P.,1985.Levelorderinginlexicaldevelopment.Cognition21, 73–93.

Hahne,A.,Eckstein,K.,Friederici,A.D.,2004.Brainsignaturesofsyntactic andsemanticprocessesduringchildren’slanguagedevelopment.J. Cogn.Neurosci.16,1302–1318.

Haskell,T.R.,MacDonald,M.C.,Seidenberg,M.S.,2003.Language learn-ingandinnateness:someimplicationsofcompoundsresearch.Cogn. Psychol.47,119–163.

Indefrey,P.,Levelt,W.J.M.,2004.Thespatialandtemporalsignaturesof wordproductioncomponents.Cognition92,101–144.

Jaensch,C.,Heyer,V.,Gordon,P.,Clahsen,H.,2014.Whatpluralsand com-poundsrevealaboutconstraintsinwordformation.Lang.Acquis.21, 319–338.

Janssen,M.,Carreiras,M.,Barber,H.A.,2011.Electrophysiologicaleffect ofsemanticcontextinpictureandwordnaming.NeuroImage57, 1243–1250.

Jerger,S.,Lai,L.,Marchman,V.,2002.Picturenamingbychildrenwith hearingloss.I.Effectofsemantically-relatedauditorydistractors.J. Am.Acad.Audiol.13,463–477.

Kiparsky,P.,1982.Fromcyclicphonologytolexicalphonology.In:vander Hulst,H.,Smith,N.(Eds.),TheStructureofPhonological Representa-tions(part1).Foris,Dordrecht,pp.131–175.

Koester,D.,Schiller,N.O.,2008.Morphologicalpriminginovertlanguage production:electrophysiologicalevidencefromDutch.Neuroimage 42,1622–1630.

Lidzba,K.,Schwilling,E.,Grodd,W.,Krageloh-Mann,I.,Wilke,M.,2011. Languagecomprehensionvs.languageproduction:ageeffectsonfMRI activation.BrainLang.119,6–15.

Luck,S.J.,2014.AnIntroductiontotheEvent-RelatedPotentialTechnique. MITPress,Cambridge,MA.

MacGregor,L.J.,Shtyrov,Y.,2013.Multipleroutesforcompoundword processing in the brain: evidence from EEG. Brain Lang. 126, 217–229.

Männel,C.,Friederici,A.,2008.Event-relatedbrainpotentialsasa win-dowtochildren’slanguageprocessing:fromsyllablestosentences. In:Sekerina,I.,Fernández,E.,Clahsen,H.(Eds.),Developmental Psy-cholinguistics:OnlineMethodsinChildren’sLanguageProcessing. JohnBenjamins,Amsterdam,pp.29–72.

McClelland,J.,Patterson,K.,2002.Rulesorconnectionsinpast-tense inflections:whatdoestheevidenceruleout?TrendsCogn.Sci.6, 465–472.

Oetting,J.B.,Rice,M.L.,1993.Pluralacquisitioninchildrenwithspecific languageimpairment.J.SpeechHear.Res.36,1236–1248.

Pinker,S.,1999.WordsandRules:TheIngredientsofLanguage.Harper Perennial,NewYork.

Pinker,S.,Ullman,M.T.,2002.Thepastandfutureofthepasttense.Trends Cogn.Sci.6,456–463.

Prado,E.,Ullman,M.,2009.Canimageabilityhelpusdrawthelinebetween storageandcomposition?J.Exp.Psychol.:Learn.,Mem.,Cogn.35, 849–866.

Prasada,S.,Pinker,S.,Snyder,W.,1990.Someevidencethatirregularforms areretrievedfrommemorybutregularformsarerule-generated. PaperPresentedatthe31stAnnualMeetingofthePsychonomic Soci-ety.NewOrleans,LA.

Ramscar,M.,Dye,M.,2010. Learninglanguagefromtheinput: why innateconstraintscan’texplainnouncompounding.Cogn.Psychol.62, 1–40.

Silva,R.,Gerth,S.,Clahsen,H.,2013.Morphologicalconstraintsin chil-dren’sspokenlanguage comprehension:a visualworld studyof pluralsinsidecompoundsinEnglish.Cognition129,457–469. Strijkers,K.,Costa,A.,2011.Ridingthelexicalspeedway:acriticalreview

onthetimecourseoflexicalselectioninspeechproduction.Front. Lang.Sci.,http://dx.doi.org/10.3389/fpsyg.2011.00356.

vanderLely,H.K.J.,Christian,V.,2000.Lexicalwordformationin gram-maticalSLIchildren:agrammar-specificorinput-processingdeficit? Cognition75,33–63.