Scrapbooking

Mondini et al 2008 NL

Description
Mondini et al 2008 NL
Categories
Published
of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
   Available online at www.sciencedirect.com Neuroscience Letters 430 (2008) 48–53 Mass and Count nouns activate different brain regions:An ERP study on early components Sara Mondini a , b , ∗ , Alessandro Angrilli a , c , Patrizia Bisiacchi a ,Chiara Spironelli a , Katia Marinelli a , Carlo Semenza d a  Department of General Psychology, University of Padova, Via Venezia 8, 35131 Padova, Italy b Casa di Cura Figlie di S. Camillo, Via F. Filzi, 24, Cremona, Italy c  Neuroscience Istitute of CNR, Via G. Colombo 3, 35121 Padova, Italy d  Department of Psychology, University of Trieste, Via S. Anastasio, 12, 34134 Trieste, Italy Received 21 June 2007; received in revised form 4 September 2007; accepted 15 October 2007 Abstract In the present study, event related brain potentials (ERPs) showed that, in an implicit Lexical decision task in which participants had to decidewhetherawordorapseudowordwaspresented,averyearlydistinctionbetweenMassandCountnounswasfoundat160msafterwordonset(N150).Mass nouns elicited greater left-lateralization over frontal locations while Count nouns were more lateralized in the left occipito-parietal sites. Inthe 430–490ms interval activity and lateralization shifted to anterior sites and a different distribution was found between Mass nouns, Count nounsand Pseudowords. Mass nouns showed greater left-lateralization both in anterior and posterior regions, whereas Count nouns showed relativelyless left-lateralization especially over frontal cortex. Results point to a functional distinction between Mass and Count nouns as indicated by thevery early automatic N150 difference between the two categories. Count nouns involved left visual associative regions that are typically relevantfor object recognition and categorization. Mass nouns, instead, required the activation of more widely spread out linguistic networks that includedalso left frontal sites, a result that indicates a more difficult and engaging automatic retrieval and an extended cortical representation of these nouns.© 2007 Elsevier Ireland Ltd. All rights reserved. Keywords:  Lexical task; N150; Recognition potential; Lateralization; Evoked potentials The semantic system distinguishes between compact, enduringthings and the stuff of which they are constituted. The formerare counted, the latter is measured. At the lexical level mostlanguages reflect this distinction in the categories of Count andMass nouns. Count nouns, like “dog”, “table” or “hammer”,apply to perceptual entities that in combination do not yieldanother entity of the same kind [16]. Mass nouns, like “water”, “sand” or “oatmeal”, denote instead, monadic entities whoseboundaries are perceptually inaccessible: the samples to whichthey are applied are taken as constituting another sample of thesame kind when in combination.In languages like English or Italian, the lexical categoriesof Mass and Count nouns are also distinguished by syntactic ∗ Corresponding author at: Department of General Psychology, University of Padova, Via Venezia 8, 35131 Padova, Italy. Tel.: +39 049 827 6641;fax: +39 049 827 6600.  E-mail address:  sara.mondini@unipd.it (S. Mondini). properties.Cardinalnumeralsandquasi-cardinalnumerals(e.g.,“several”) modify Count nouns but never Mass nouns. More-over,indefinitequantifierslike“little”and“much”modifyMassnouns, never Count nouns, whereas “few” and “many” modifyCount nouns, never Mass nouns. Count nouns admit a morpho-logicalcontrastbetweensingularandplural;Massnounsdonot,beingusuallysingular.Massnounswithsingularmorphologydonot tolerate the indefinite article, whereas singular Count nounsdo.ThepresentstudyaddressesthequestionofwhetherMassandCountnounsareprocesseddifferentlyinthebrainwithparticularconcern to the lexical–semantic aspects. A categorical organi-zation of noun processing has been shown to be supported inthe brain in many instances [14,22,27,30]. The most convincing findings distinguishing Mass and Count nouns in neuropsycho-logical investigations have so far concerned morphosyntacticaspects. Thus, a patient whose grammar was otherwise perfectshowed, as a consequence of brain damage, an isolated deficitin the use of the grammatical properties of Mass nouns [28]. 0304-3940/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.neulet.2007.10.020  S. Mondini et al. / Neuroscience Letters 430 (2008) 48–53  49 An ERP investigation [32] demonstrated that, in a paradigmon sentence plausibility in which Mass and Count nouns wereused, only Count nouns elicited frontal negativity. The outcomewas explained by the authors as due to grammatical rather thanconceptual–semanticprocesses,andwasrepresentedbyaN400-like component. Instead, in a single-word reading task requiringexplicit semantic classification of Mass and Count nouns, Bisi-acchi et al. [4] found an early wave peaking at about 140msover left frontal sites during Count noun processing, and a bilat-eral potential, spread across hemispheres, during Mass nounprocessing.However,processingsuchwidecategoriesasMassandCountnouns is likely to be distributed in the left hemisphere in largelyoverlapping regions [28]. Repeated investigations involving a considerable number of aphasic patients found only one par-ticipant with a huge left hemisphere lesion who presented aconsistent dissociation (Count worse than Mass) in naming thetwo categories [29].Studying the time course of electrical events may be a goodstrategytocaptureaneurologicallyimplementedprocessingdis-tinction between Mass and Count nouns at the lexical–semanticlevel. Within this framework, recent studies carried out in orderto investigate reading processes have pointed to a specific earlyevoked component corresponding to automatic visual recog-nition of words [2,5,8,10,15,21,24–26,33]. All these studies found a negative early component, peaking about 130–170msafter stimulus onset, which was characterized by a significantposterior left asymmetry only for word-like stimuli. Bilateralnegativity was observed instead for other visual patterns, suchas faces, objects, and symbol strings. In view of these findings,Dehaeneetal.[11]proposedthattheleftfusiformgyruscontains a Visual Word Form Area (VWFA) which computes the prelex-ical representation of a word as an ordered sequence of abstractletters regardless of size, font, case [18,31] and of position inthe visual field [8]. The left peak of activity of the recognition potential has been demonstrated to be also independent froma linguistic task as compared to a non-linguistic visual featureprocess [31].The present study aimed to identify, by means of evokedpotentials, the cortical correlates underlying the automatic wordprocessing of Mass and Count nouns and Pseudowords, usinga simple lexical decision task based on word/non-word judg-ment. The categorization process in Mass/Count nouns wasthus expected to occur in an implicit way, as participants wererequired to differentiate words from pseudowords. In particular,in line with past findings [4], Mass nouns were expected to acti- vateaneuralnetworkdistributedacrosslinguisticareas,whereasCount nouns would involve more restricted neural networkswithin the left hemisphere.Eighteenright-handednativeItalian-speakingstudents(meanage=24.0 years, S.D.=2.5, range=20–28, eleven females andseven males), were administered a lexical decision task whichrequired a motor response. None of the participants had beentreated for any neurological or psychiatric disorder, nor wasunder any medication.The experimental design was approved by the local EthicsCommittee,andsubjectsparticipatedintheelectrophysiologicalsession after giving their written informed consent according tothe Declaration of Helsinki.EEG was recorded continuously throughout the task bymeans of two SynAmps amplifiers (NeuroScan Labs, Ster-ling, USA) and NeuroScan 4.1 software. Sampling rate wasset to 500Hz and band-pass filters were dc-100Hz. Eventrelated potentials were recorded through 38 tin electrodes(i.e., from left to right hemisphere, Io1–Nz–Io2; Fp1–Fpz–Fp2;F9–F7–F3–Fz–F4–F8–F10; FT7–FC3–FCz–FC4–FT8; T7–C3–Cz–C4–T8; M1–TP7–CP3–CPz–CP4–TP8-M2; P7–P3–Pz–P4–P8; O1–Oz–O2 [20]), 31 placed on standard EEG coor- dinates of an elastic cap (ElectroCap) and the other 7 appliedbelow each eye (Io1–Io2), on the two external canthii (F9–F10),on Nasion (Nz) and on mastoids (M1–M2). All EEG channelswere on-line referred to Cz.Stimuli included 50 Mass and 50 Count nouns as well as100 Pseudowords presented one at a time on a computer screen.Pseudowords were derived from the stimuli, by changing onesyllableandinaccordancewithItalianeuphonicrules.TheMassand Count nouns used as stimuli were balanced for phonolog-ical complexity, word frequency [12] and familiarity that waspreviously tested on a five-point scale on a sample of 50 uni-versity students (mean of Mass nouns=3.5, S.D.=1.9; meanof Count nouns=3.3, S.D.=1.8). Both Mass and Count nounswereselectedfromasampleofconcretewords.Noneofthestim-uli had multiple meanings. Participants had to judge whether aword or a non-word was presented on the computer screen bypressingtwodifferentkeyswiththeindexorthemiddlefingerof the left hand. Each stimulus was presented until one of the twokeys was pressed. The inter-stimulus interval varied randomlybetween 1500 and 2000ms. Reaction times (RT) and error rateswere measured.Data were off-line re-referenced to average reference andwere transformed into epochs organized from − 200 to +500msaround stimuli onset. Epochs were baseline ( − 100ms to 0) andeye artifacts corrected and visually inspected for residual arti-facts exceeding a threshold of  ± 40  V. Then, all accepted trials(mean: 51.5%) were averaged for each condition and for everyparticipant. In line with past literature [2,10,24,31] and on thebasisofvisualinspectionofgrandaveragewaveforms,twotimewindowswerechosenfordataanalysis:thefirst150–170msandthe second 430–490ms following stimulus onset. For statisticalpurposes, starting from the regions of interest [31], electrodes were clustered into four quadrants: left anterior (Fp1–F7–FT7),right anterior (homologous to left), left posterior (O1–P7–TP7),right posterior (homologous to left). This clustering allowed athree-way repeated-measure ANOVA design with the follow-ing factors: Stimulus (Mass nouns  versus  Count nouns  versus Pseudowords), Gradient (Anterior  versus  Posterior) and Later-ality (Left hemisphere  versus  Right hemisphere). Tukey HSDtest was used for  post hoc  statistics and the Greenhouse-Geissercorrection was applied when necessary.In order to clarify and better contrast the cortical activationacross hemispheres, in addition to post-hoc tests, effect sizesbetween left and right locations were calculated [7,23].The analysis on behavioral data showed that participantswere significantly slower with Pseudowords (649ms) than with  50  S. Mondini et al. / Neuroscience Letters 430 (2008) 48–53 Fig. 1. Grand average of event related potentials comparing the stimuli: Mass nouns in blue, Count nouns in red and Pseudowords in green. Negativity is depictedupward. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.) eitherMass(585ms)orCount(587ms)nouns( F  (2,38)=19.24,  p <0.001). However, no RT difference was found betweenMass and Count nouns. The ANOVA performed on errorrates ( F  (2,38)=7.48,  p <0.01) revealed significant differencesbetween Pseudowords (3.83%, error percentage) and both Mass(1.25%;  p <0.01) and Count (1.60%;  p <0.05) nouns. However,as for RT, no difference was found between Mass and Countnouns.Evoked potentials showed an early difference (150–170msafter stimulus onset) between the two noun categories. Fig. 1displaysgrand-averagedwaveformsobtainedduringthecorticalelaboration of all linguistic stimuli.Fig. 2A shows spline maps of the word recognition timeinterval, corresponding to the automatic stimulus processingin posterior cortical sites: Mass nouns clearly elicited smallerposterior negativity than Count nouns and Pseudowords. Thisfinding was confirmed by the significant three-way Stimu-lus × Gradient × Lateralityinteraction( F  (2,34)=4.02,  p <0.05,GG  ε =0.86), which revealed in the post-hoc tests a significantleft/rightasymmetryforallstimulibothatanterior(all  p s<0.05)and posterior sites (all  p s<0.001, Fig. 2B). The greatest left- lateralization was found at posterior sites for Count nouns: thiscategory,atleftposteriorlocations,elicitedasignificantlylargernegativity with respect to Mass nouns (  p <0.001). At anteriorsites,significantrelativegreaternegativityofthelefthemisphereforMassnouns(Fig.2B,blueline) versus Countnouns(  p <0.01,red line) and Pseudowords (  p <0.05, green line) was found.Fig.2Cshowstheleftlateralviewofsplinemapscorrespond-ing to the late time window, about 450ms after stimulus onset.Mass nouns elicited the greatest cortical negativity and later-alization at the level of left frontal sites whereas Pseudowordselicited reduced left frontal negativity and more bilateral acti-vation. In the analysis performed on this late time window(430–490ms) a main effect of Laterality ( F  (1,17)=16.06,  p <0.001) revealed higher negativity of left compared withright electrodes ( − 1.60  V  versus  +0.16  V, respectively). Themost interesting effect, however, was the three-way Stimu-lus × Gradient × Lateralityinteraction( F  (2,34)=3.36,  p <0.05,GG ε =0.99).Onfrontallocations,allstimulievokedsignificantgreater left than right negativity (  p <0.001), within this pattern,however, Mass nouns elicited greater negativity with respect toPseudowords (  p <0.01, Fig. 2D). In the right hemisphere, instead, both Mass and Countnouns showed significant greater positivity than Pseudowords(  p <0.01). Also at posterior clusters all stimuli were consis-tently left-lateralized (  p <0.001 for both classes of nouns and  p <0.01 for Pseudowords), and showed significant greater neg-ativity evoked by Mass and Count nouns at the left posteriorquadrant in comparison with Pseudowords (  p <0.05).An effect size analysis of the described comparisons (later-alization in frontal sites) found high Cohen’s  d   values for bothMass and Count nouns ( d  =0.82 and  d  =0.83 respectively) anda medium value for Pseudowords ( d  =0.49), showing signifi-cantly greater left-lateralized activation during word rather thanduring Pseudoword elaborations.The aim of the present work was to study the early automaticprocessing of Mass and Count nouns. ERPs showed that differ-ent cortical networks support the processing of Mass and Countnouns from the first automatic phases of word elaboration.Other neuropsychological methods like the clinical–anatomical  S. Mondini et al. / Neuroscience Letters 430 (2008) 48–53  51Fig. 2. Spline maps (A and C) and three-way Stimulus × Gradient × Laterality interactions (B and D) of early N150 (150–170ms, upper row) and late components(430–490ms, lower row). Cortical negativity is represented in blue, positivity in red. All amplitudes are expressed in microvolts. (For interpretation of the referencesto color in this figure legend, the reader is referred to the web version of the article.) correlation did not produce the same result: indeed, studieson aphasia (e.g., ref. [28]) have not yet demonstrated a double dissociation in the retrieval of the two categories, and only arelative Count noun deficit [29] has been found. One expla-nation for the lack of evidence in anatomical studies is that, if cortical networks specialized in Mass and Count nouns largelyoverlap, the lesion of language areas typically damaged inaphasic patients may affect the processing of both categories of nouns. For the same reason, a difference in cortical distributionbetween Mass and Count nouns could emerge, instead, only inintact networks and may be best captured by tapping the timecourse of their processing in specific tasks.In the present experiment, both RT and error rates did notdiscriminate between Mass and Count nouns showing that theimplicit categorization did not vary between the two classes of nouns. In a Semantic Categorization task in which an explicitMass/Count judgment was required, Bisiacchi et al. [4] f ound, instead,asignificantRTdifferencebetweenthesamesetofMassand Count nouns: Mass noun categorization was significantlydelayed compared with Count nouns. Explicit categorizationrequires complex evaluation and active representation of sev-eral features of the stimulus in the working memory. For thisreason, the explicit categorization task of Bisiacchi et al. [4]was more engaging and therefore more effective in showingnoun differences than the present implicit decision (in explicitcategorization, RT were around 1000ms while in implicit lex-ical decision RT were around 580ms), which instead requiredonly recognition of the word.Unlike behavioral results, ERP analyses revealed a cleardifference in Mass/Count processing at the very early stages(150–170ms), namely in a time window related to auto-maticwordprocessing[2,5,8,10,15,21,24–26,33].Inthepresent experiment, the recognition potential recorded over posteriorsiteswasleft-lateralizedduringbothwordandPseudowordpro-cessing.Thisisinlinewithpreviousstudieswhichfailedtoshowsignificant differences between the two classes of stimuli at thisearly stage of language elaboration (e.g., refs. [2,26]). However, unlike other studies, for the first time the present investigationhighlighted an early automatic difference during the implicitprocessing of Mass/Count nouns. In fact, Count words elicitedsignificantly greater negativity-activation over posterior sites of thelefthemisphereincomparisontoMassnouns,whichinsteadshowed significantly greater negativity than Count nouns on theleft anterior regions. In the later 430–490ms interval, the threesamples of stimuli showed a more evident cortical differentia-tion.Theearlyeffectsuggeststhatthesetwolexical–semanticcat-egories activate partially overlapping neural networks that aremainly left-lateralized; however, within the left hemisphere thetwo classes of nouns showed a different activation along theantero-posterior gradient. From the analysis of the effect size itcan be further argued that, overall, the Mass nouns activated alinguistic network more largely distributed within the left hemi-sphere, including both anterior and posterior left regions. TheCount nouns and Pseudowords, instead, showed clear-cut peaksof activity restricted to the left posterior cortex.  52  S. Mondini et al. / Neuroscience Letters 430 (2008) 48–53 These findings are in agreement with the idea that visualrepresentations of concrete objects, such as tools, are mainlystoredinposteriormiddleandinferiortemporalgyri[9].Asharp image of abstract entities would be more difficult to generateand process and would need the recruitment of frontal corticalregions too [9].The results of the present study may appear to be at variancewith a previously reported ERP experiment in which frontalnegativity has been found for Count compared with Mass nouns[32]. The two studies, however, differ in several methodologi-cal aspects (not including the issue of the reference site whichwas different in the previous investigation) thus it is difficult tomake a direct comparison. It is important, nevertheless, to iden-tify the main factors that do support such differences. First, inSteinhaueretal.[32]themaintaskconsistedinsentencereading and a high number of sentences included syntactic violations asfillers. Both the electrophysiological [1,13] and neuropsycho- logical [3] literature indicate that the left frontal cortex plays an importantroleindetectingsyntacticerrors.InSteinhaueretal.’swork  [32], the critical stimuli were embedded in sentences, thus involvingbothsyntacticandsemanticprocessingofMass/Countnouns. The authors interpreted the main differences found overfrontalsitesasaconsequenceofsyntacticfactors,thuschalleng-ing the previous semantic explanation. A further variable, likelytoinfluenceshiftingofactivitytofrontalsites,istheselectionof stimuli.Inthepresentstudy,bothMassandCountnounswereallconcrete words. Instead, Steinhauer et al.’s list of critical stimuliseemed to include also abstract Mass/Count nouns (e.g., “infor-mation”and“explanation”,ascanbeevincedfromthetextofthepaper (p. 1000) and also from Table 1 (p. 1001) [32]). Actually, abstract words do not provide an immediate lexical represen-tation, are associated with slower RT than concrete words andmay require elaboration strategies shifted towards frontal areas.Finally, in order to investigate lexical–semantic aspects, Stein-hauer et al. [32] analyzed a later time interval (i.e., 300–600ms, supposed to correspond to the N400 component).Bisiacchi et al.’s investigation [4] used the same linguisticstimuli as the present study in a different paradigm requiringan explicit judgment of Mass/Count categorization. The earlyN150 evoked by Count nouns showed significant negativity inthe left compared to the right frontal sites, whereas Mass nounsshowed activity distributed in both hemispheres. As alreadymentioned, Bisiacchi et al.’s [4] explicit task required morecomplex frontally sustained strategies. Notwithstanding largeparadigm differences, however, the basic finding is confirmed.Count nouns elicited linguistic networks more confined to theleft hemisphere. Depending on the task, Mass nouns activated,instead,largeneuralnetworksdistributedacrosshemispheres[4]or,asthepresentexperimentshowed,distributedatbothanteriorand posterior sites within the left hemisphere.The present study is the first to show an automatic classi-fication of different categories of nouns at very early stages.Several recent investigations failed to demonstrate clear differ-entactivationsoflinguisticareasduringvisuallexical–semanticprocessing, such as open-class  versus  closed-class or color-related  versus  form-related or concrete  versus  abstract wordcategorizations [6,17,20]. In their recent study, Moscoso del PradoMart´ınetal.[19]f oundthatcolor-relatedandform-related words equally activated the early automatic word recognitioncomponent(N150),butwereselectivelydifferentiatedrelativelylater beyond 200ms after stimulus onset.Consistent with the present study, Mart´ın-Loeches et al. [17]demonstratedthatduringsemanticprocessingofconcrete versus abstract words, small effects emerged in a delayed recognitionpotential that peaked in the 240–296ms interval. The time win-dowisrelativelydelayed(100ms)incomparisonwiththequotedliterature on the N150 component. This delay in recognitionpotential onset may have occurred because the authors useda very particular and complex experimental paradigm (rapidstream stimulation) in which background stimuli were mixedwith concrete and abstract words, orthographically legal andnon-legal non-words or control stimuli (fragments of letters).Alsoinconsiderationofaphasiologicalstudies,allthesefind-ings taken together seem to suggest that the representation of Mass and Count nouns may be stored in largely overlappingregions(whichmaybeexpectedasthetwocategoriesincludeallcommon nouns), but may require different and, to some extent,separately located processing.TheparticularrelationfoundhereforMassnounswithfrontalcortex activation shows how a semantic explanation is entirelysufficient to account for the Mass/Count effect in the left frontallobe. Unlike Mass nouns, Count nouns have been found for thefirst time to be mainly processed in the left posterior areas, in aregion where concrete Count nouns may need to be integratedwith their corresponding visual representations.Given the failure of past research to find an early discrimi-nation between other classes of stimuli believed to be naturallyembedded in the human neural system, the present data indicatethat the strongest neurobiological basis, among all investigatedlexicalcategories,mightberoutedinearlyMass/Countdiscrimi-nation,whichisnotsosurprisinginlightoftheuniversalfeaturesunderlying this basic lexical distinction. Acknowledgements This work has been supported by grants from MIUR toCarloSemenza,PatriziaBisiacchiandAlessandroAngrilli,fromNUMBRA to Carlo Semenza. Conflict of interest statement  : We declare that we have noconflict of interest. References [1] A. Angrilli, B. Penolazzi, F. Vespignani, M. De Vincenzi, R. Job, L. Cic-carelli, D. Palomba, L. Stegagno, Cortical brain responses to semanticincongruity and syntactic violation in Italian language: an event-relatedpotential study, Neurosci. Lett. 322 (2002) 5–8.[2] S. Bentin, Y. Muochetant-Rostaing, M.H. Giard, J.F. Echallier, J. Pernier,ERPmanifestationsofprocessingprintedwordsatdifferentpsycholinguis-tic levels: time course and scalp distribution, J. Cogn. Neurosci. 11 (1999)235–260.[3] R.E.Berndt,Sentenceproduction,in:B.Rapp(Ed.),TheHandbookofCog-nitive Neuropsychology, Psychology Press, Michigan, 2000, pp. 375–396.[4] P. Bisiacchi, S. Mondini, A. Angrilli, K. Marinelli, C. Semenza, Mass andcount nouns show distinct EEG cortical processes during semantic task,Brain Lang. 95 (2005) 98–99.
Search
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks