emotion and words

Upload: yaddayadda

Post on 04-Jun-2018

223 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/13/2019 Emotion and Words

    1/23

    The Distinctiveness of Emotion Concepts: A Comparison between Emotion, Abstract, andConcrete WordsAuthor(s): Jeanette Altarriba and Lisa M. BauerSource: The American Journal of Psychology, Vol. 117, No. 3 (Autumn, 2004), pp. 389-410Published by: University of Illinois PressStable URL: http://www.jstor.org/stable/4149007

    Accessed: 01/02/2010 07:14

    Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

    http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

    you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

    may use content in the JSTOR archive only for your personal, non-commercial use.

    Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

    http://www.jstor.org/action/showPublisher?publisherCode=illinois.

    Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

    page of such transmission.

    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

    of scholarship. For more information about JSTOR, please contact [email protected].

    University of Illinois Pressis collaborating with JSTOR to digitize, preserve and extend access to The

    American Journal of Psychology.

    http://www.jstor.org

    http://www.jstor.org/stable/4149007?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=illinoishttp://www.jstor.org/action/showPublisher?publisherCode=illinoishttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/4149007?origin=JSTOR-pdf
  • 8/13/2019 Emotion and Words

    2/23

    The distinctivenessf emotionconcepts:A comparison etweenemotion,abstract,andconcretewordsJEANETTE ALTARRIBAUniversity at Albany, State University of New YorkLISA M. BAUERUtica College of Syracuse University

    Are the concepts represented by emotion words different from abstract wordsin memory? We examined the distinct characteristics of emotion concepts in 3separate experiments. The first demonstrated that emotion words are betterrecalled than both concrete and abstract words in a free recall task. In the sec-ond experiment, ratings of abstract, concrete, and emotion words were comparedon concreteness, imageability, and context availabilityscales. Results revealed adifference between all 3 word types on each of the 3 scales. The third experimentinvestigated priming in a lexical decision task for homogeneous (abstract-ab-stract and emotion-emotion) and heterogeneous (abstract-emotion and emo-tion-abstract) associated word pairs. Priming occurred only for the homoge-neous and heterogeneous abstract-emotion word pair conditions. Possibleexplanations for these findings are discussed in terms of the circumplex, hier-archical, and semantic activation models. The results are most consistent withthe predictions of the semantic activation model.

    Although the literature on the cognitive processing of emotion words isalmost nonexistent, the evidence indicating the importance of emotion-al self-awareness and emotional intelligence in human development andpsychological well-being is accumulating rapidly (Goleman, 1995). Theawareness of emotions and the ability to recognize, name, and understandthe causes of these feelings are important to overall emotional compe-tence. These abilities have been linked to a lower incidence of depres-sion in both children and adults, the cessation of eating disorders, and areduction in mental illness in recent years (Goleman, 1995). Becauseemotional awareness is important, a logical and necessary step in this areaof research is to discover the fundamental attributes of emotion wordsin human memory.Fundamental attributes of various word types have been examined bystudying word representation and word retrieval. In the study of wordAMERICANJOURNAL OF PSYCHOLOGYFall 2004, Vol. 117, No. 3, pp. 389-410? 2004 by the Board of Trustees of the University of Illinois

  • 8/13/2019 Emotion and Words

    3/23

    390 ALTARRIBA AND BAUERrepresentation ndwordretrieval, robust inding sthat concretewordssuch as bedare rememberedbetter than abstractwords such asfreedom(Schwanenflugel,Harnishfeger,&Stowe,1988). This advantage n theprocessingofconcretewords,often calledaconcretenesseffect,hasbeenreported n avariety f tasks,ncluding exicaldecision(Day,1977;James,1975), pairedassociatelearning,free recall,recognition (Paivio,1971,1986),andcomprehensiontests (Holmes&Langford,1976).Addition-ally,concretenesseffectsexistfor both childrenandadults(Vellutino&Scanlon, 1985). Concreteness effects often are explained in terms ofPaivio's(1971, 1986) dual-codingmodel or bya context availability y-pothesis (Schwanenflugel,Akin,&Luh, 1992).Paivioand his colleaguessuggestedthat the processingadvantageofconcrete words over abstract words has to do with sensory information(Paivio, 1971, 1986; see also Schwanenflugel & Akin, 1994). Accordingto the dual-coding theory, there are two functionally independent yetinterconnected representational systems: a verbal system (the logogensystem) and an imaginal system (the imagen system). The logogen sys-tem is responsible for the representation and processing of linguisticinformation. The imagen systemis a nonverbal systemresponsible for therepresentation and processing of nonverbal information. These repre-sentations are differentially available in memory contingent on the stim-ulus' concreteness. Paivio'sdual-coding theory accounts for concretenesseffects by stating that both concrete and abstract words are representedin the logogen system, but only concrete words are connected to theimagen system.When a concrete word is produced, both the logogen andthe imagen systemsare activated,and they have an additive effect, therebyyielding an advantage for concrete words over abstract words in lexicalprocessing. Therefore, concrete words are remembered better than ab-stract words because the image provides an additional means throughwhich the concrete words can be stored and retrieved.

    A second explanation used to interpret concreteness effects is a con-text availabilityhypothesis, which emphasizes the ease with which a con-text or circumstance in which the word appeared can be recalled(Schwanenflugel et al., 1992). For example, it may be easy to think of acontext for the word bedbut difficult to think of a context for the wordfreedom.According to this hypothesis, concreteness effects arise from adifferential availabilityof contextual information. The contextual infor-mation can be retrieved from prior knowledge or from information inthe stimulus environment. Therefore, when abstract and concrete wordsare presented in isolation it is more difficult to retrieve contextual infor-mation for abstract words than for concrete words, resulting in poorerrecall for abstract words.

    This theory predicts that with adequate contextual support, abstract

  • 8/13/2019 Emotion and Words

    4/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 391stimuli can be learned as well as concrete stimuli. This prediction wassupported in a paired-associate experiment. Bransford and McCarrell(1974) found that when abstractstimuli were meaningfully related to theresponse pairs, they were recalled as well as concrete stimuli because theresponse pairs were providing the context for the abstract stimuli(Schwanenflugel et al., 1992). Additional support for the context avail-ability hypothesis stems from studies that have found that concretenesseffects disappear when the stimuli are presented in a supportive context(Schwanenflugel et al., 1988; Schwanenflugel & Stowe, 1989).The role of imageability, the extent to which a mental image is evokedby a referent, has also been examined. The basic premise of the dual-coding theory is that concrete words are recalled better than abstractwords because concrete words have an imagen system that abstractwordslack. Therefore, the imageability of various word types plays a role in thepredictions of the dual-coding theory. Additionally, it is possible thatimagery plays a role in the accessibility of contextual information.Schwanenflugel et al. (1992) investigated whether the effects of imagea-bility on recall are independent of the effects of contextual informationon recall. Participants were asked to recall unrelated abstract and con-crete stimuli that were equated on the accessibility of contextual infor-mation. They also investigated the strategies that participants used whenrecalling the stimuli. The results uncovered an effect of imageability in-dependent of context availability.When participantsreported implement-ing imagery, concreteness effects were present regardless of whether thestimuli were equated on context availability.However, when participantsdid not report using imagery to recall the items, concreteness effects werefound only when the stimuli were not equated on context availability.Therefore, it appears that when imagery is perceived as being helpful incompleting a task, it may be used. However, if imagery does not appearto be necessary in completing a task, information that is readily availablefrom prior knowledge will be used.The aforementioned research on concreteness effects indicates thatconcrete and abstract words have been investigated on dimensions suchas concreteness, context availability,and imageability. It is important tonote that in the concreteness effects literature emotion words often areincluded with the abstract stimuli with no justification for this classifica-tion (Chiarello, Senehi, &Nuding, 1987; Eviator,Menn, &Zaidel, 1990;Nelson & Schreiber, 1992). Previous studies have not considered thatemotion words may elicit different activations and may be distinct fromabstract words and that if emotion terms are classified as abstract, theresults may be confounded. It may be that emotion words placed in theabstract category increase or decrease the actual concreteness effects byinfluencing the measures of the abstract category.

  • 8/13/2019 Emotion and Words

    5/23

    392 ALTARRIBA AND BAUERThe aimof the presentseriesof experiments s to investigatewhetheremotionwordsbehaveasabstractwordsbehave(i.e.,lowin concreteness,context availability,nd imagery)or whetherconcepts represented by

    emotionwordshavecharacteristics hat aredistinctlydifferentfromthecharacteristicsof the concepts represented by abstract and concretewords.Thiswasaccomplishedby looking at the results of a free recallexperiment,a ratingstudy,and a primingexperiment.In each case, itwaspredicted hatemotionwordswouldbe responded odifferentlyhanconcrete and abstractwords.Experiment1investigatedwhether the classification f emotionwordsasabstractn previous tudieswasappropriate yconductingasimplefreerecallexperiment.Experimenters Paivio,1971, 1986) have found that

    concretewordswererecalledmoreoften than abstractwords.Ifemotionwordsdo not elicitdifferentcharacteristicshan abstractwords, hen thenumber of emotionwordsrecalledin a free recalltaskshould be no dif-ferent from the number of abstractwords recalled. In addition, bothabstractand emotionwordsshould be recalled less often than concretewords.Experiment1 investigatedwhetheremotionwordshavethe samememorialpropertiesasabstractand concrete words.EXPERIMENTMETHODParticipants

    Sixtyundergraduatetudents, 0 menand30women,rom heUniversitytAlbanyookpartnthisstudyorpartialulfillmentf a course equirement.achparticipantasanativeEnglishpeaker.Materials

    Theabstract,oncrete,andemotionwordswereclassified priori.Wordswhosemeaningdenoteda material bjectwereclassified sconcretewords.Multidimensionalscaling ndfactor nalytictudies uggesthatemotionalx-periencesonsist fbothavalencepleasantrunpleasant)ndan arousallow,medium, rhigh)componentRussell, 980,1991;Russell&Bullock, 985).Therefore,wordswhosemeaningswereaffective nd hadpleasantnessr un-pleasantnessndarousalcomponentswereclassifiedsemotionwords.Wordswhosemeaningseferredosomethingndependentromamaterialbjecthatwerenotclassifiedsanemotionwordweredesignatedsabstract ords.Usingthesecriteria,2words24ofeachof the three ypes)were elected romBleas-dale(1987),Chiarellotal. (1987),Clore,Ortony,ndFoss 1987),NelsonandSchreiber1992),Shaver,chwartz, irson,ndO'Connor1987),andWhissell(1989).

  • 8/13/2019 Emotion and Words

    6/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 393A list of words was constructed for each of the three word types (abstract,concrete, and emotion). Each list consisted of 20 words, and all three lists werematched in frequency and in length. These lists are presented in Table 1. Addi-

    tionally, four buffer words, two presented at the beginning of each list and twopresented at the end of each list, were used to reduce primacy and recency ef-fects. For the abstract list, the buffers were easy,donor, ravel,and finish. For theconcrete list, the buffers were dragon,penny,card,and pepper. or the emotion list,the buffers were excited, onely, nfatuated,and upset.Therefore, each participantheard 24 words, but only the middle 20 words were included in the scoring.Procedure

    To ensure that the participants understood the task, a 10-word practice listcomposed of all three word typeswas presented before the experimental list. Forboth the practice and the experimental list, each word was presented auditorilyfor 5 s. Participants were told before the presentation of each list that they wereto remember as many words as possible because they would later need to recallthe words.After the presentation of each list, the practice list and one of the threeexperimental lists, participants were asked to write down as many words as theycould remember on a sheet of paper containing 10 blanks for the practice listand 24 blanks for the experimental lists. They were given as much time as theyneeded.

    Table 1. Words used in Experiment 1Emotion words Concrete words Abstract wordsangry machine heightthrilled airplane quenchfurious flag patriotismhappy basket healthgrateful clock capabilityglad apartment aiddepressed scissors humordelighted building wealthdisappointed balloon winhopeful magazine advicediscouraged sock dramaafraid cigar wisdomcheerful dog attitudelove newspaper culturesad castle intellectsurprised girl nonsensemad rope honoranxious elephant chaosjoy canoe obeyannoyed factory heaven

  • 8/13/2019 Emotion and Words

    7/23

    394 ALTARRIBA AND BAUERRESULTSANDDISCUSSION

    Incorrectresponsesand buffer wordswerenot scored.An analysisofvariance(ANOVA)asconducted to comparethe means acrossthe threeconditions.Meannumberof wordsrecalledandstandarddeviations orthe three wordtypesare asfollows.Onaverage,participantsecalled5.70abstractwords (SD= 0.40), 7.75 concrete words (SD= 0.40), and 9.1emotion words (SD= 0.38). The resultsof the ANOVAndicate that thethreemeans arestatistically ifferentfrom eachother,F(2,108) = 19.01,MSE= 3.08, p < .001. Planned comparisonsrevealedthat participantsrecalled more emotion wordsthanabstractor concretewords,followedbyconcretewords,and then byabstractwords(allps< .05). These find-ings replicateearlierresults hat concretewordsarerecalled

    moreoftenthan abstractwords n freerecallexperiments.Additionally,o ourknowl-edge, these resultsdemonstrate or the firsttimethatemotionwordsarebetter recalledthaneither concreteorabstractwords n afree recalltask.The current indingssuggestthat thereareunderlyingdifferences n theprocessingof these three wordtypes.In termsof the dual-coding heoryandthe contextavailabilityypoth-esis mentionedearlier, he current indingsseemtosuggestthatperhapsthere is an underlyingdifferencebetweenthe threewordtypes.Paivio's(1971, 1986) dual-coding heorystatesthatconcretewordsarerecalledbetter thanabstractwordsbecauseconcretewordshaveanimageassoci-atedwiththemthatabstractwords ack.Perhaps motionwordsalsohaveanimage system,allowing hem to be recalledbetterthanabstractwords.Alternatively,n termsof the contextavailabilityypothesis Schwanenflu-gel et al., 1988), perhapsemotionwordshavemorecontextsassociatedwiththemand arerepresentedbya broaderrangeof contextsin whichtheyhave been experienced.The secondexperimentused a ratingtaskto investigatehowemotion,abstract,and concrete wordswere ratedon concreteness,imageability,and context availabilitycales. It waspredictedthatthe emotion wordswould more readilyactivateimages and contexts than abstractwordswould. This hypothesisis based partiallyon previousresearchon theuniversalityf facialexpressions.Several ross-culturaltudieshavefoundthatwhenparticipants erformarecognitionofphotographed motionalexpressionstask at least six emotions (anger,disgust,fear,happiness,sadness,and surprise)havea universalexpression (Ekman,Friesen,&Ellsworth, 972;Izard,1971).This research ndicates hatemotionwordshaveperceptualreferents.Therefore,it is plausiblethatan imagemaybe moreaccessible or emotion wordsthanforabstractwords.Referringto the dual-coding theory,it maybe that both concrete and emotionwords are stored in two representational systems,whereas abstractwords

  • 8/13/2019 Emotion and Words

    8/23

  • 8/13/2019 Emotion and Words

    9/23

    Table 2. Words used in Experiment 2Emotion words Concrete words Abstract wordsangry costume learningcontent cloud jeopardyexcited brush masteryfurious mirror truthmoody machine chancehappy airplane virtuecalm pencil treatscared bible attitudeglad garden quenchlonely scissors facilityanxious asphalt compulsionsecure timber decencysurprised pepper healthalert animal originsorry balloon capablecurious apartment wealthdepressed dentist entrytroubled crutch excusedelighted factory gloryhysterical cards patriotismlove liquor dareupset newspaper faultskeptical jewel capabilitythrilled basket legendhurt orange fictionzealous eagle abilityhopeful building graceuncertain penny methodthankful magazine impressionconcerned poison permissionafraid tower conceptunhappy mouth chaosgrateful movie conquestobsessed cigar essencejoy elephant adviceworried dragon dramamiserable burglar hauntnervous daughter cultureserious fence easyannoyed jungle responseconfident police welfarestupid flute benefitsad castle intellectdisgusted clown countaffectionate nurse theftmad father falseawful crown nonsensecheerful truck heaven

  • 8/13/2019 Emotion and Words

    10/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 397theparticipants nderstood hedirections.Theinstructionsoreachof thescalesaresimilar o the instructionsused by previousresearchers Altarriba,Bauer,&Benvenuto, 1999;Campos,1990;Friendly,Franklin,Hoffman,&Rubin, 1982;Gilhooly&Logie,1980;Schwanenflugel t al., 1992).Eachparticipant atedthewordson a singleattribute.RESULTSAND DISCUSSION

    A 3 x 3 ANOVAhat treated the rating scale type condition as a between-subject variable and the word type condition as a within-subject variablewasperformed. There was a significant scale type effect, F(2, 141) = 94.05,MSE= 0.47, p < .001. There was also a significant word type effect, F(2,141) = 977.10, MSE= 320.35, p < .001. A significant interaction betweenscale type and word type was also observed, F(4, 282) = 128.11, MSE=42.00, p < .001, which indicates that there are differences in the ratingsof word types on the different scales.Table 3 presents the mean concreteness, imageability, and contextavailability ratings for the 144 concrete, abstract, and emotion words.Within each scale, reliable differences were found between all three wordtypes (allps

  • 8/13/2019 Emotion and Words

    11/23

    398 ALTARRIBA AND BAUERba et al., 1999).As canbe seen in Table3, the previouspatternof resultsisconsistentwiththesecorrelations.Concretewordshavehigh concrete-nessratings(M=6.3) and alsohigh imagery M=6.6) andcontextavail-ability(M=5.6) ratings.Abstractwordshave ow concreteness(M= 3.3),imagery(M=2.4), andcontextavailabilityM=4.6) ratings.Unlikepre-viousstudies, he presentstudyalso examinedratings or emotionwords.The emotion wordshad low concreteness (M= 3.0) and imagery(M=3.3) ratings.The contextavailabilityating(M=5.0) is a bithigherthanexpected.However,t is important o note thatthe aforementionedcor-relationswerecalculatedon concrete and abstractwordratings.Takentogether,the results of Experiment2 indicatethatemotion wordsacti-vatedifferentpropertiesthan abstractand concretewords.

    Anotheraim of the current researchwasto investigatehow the con-cepts representedby emotion and abstractwords are representedin abroaderneuralnetworkby means of a primingparadigm.At least twomodels have been proposed to describe the mental representationofemotionwords.One model is Russell's(1980, 1989) circumplexmodelof emotions.The circumplexmodel is a structuralmodelwherebyemo-tion labelsarecategorizedon twodimensions,pleasantnessor unpleas-antnessandhigharousalor lowarousal, hereby orminga circular truc-ture.Synonyms end to be closer to one another,whereasantonyms allon the oppositeside of the circle.The correlationbetweenanytwoemo-tionwordsequalsthecosine of the anglebetweenthem. Russell's ircum-plex modelpostulates hatpositiveprimingor facilitation n respondingwilloccurin relation to the distancein whichthe emotion wordsfall inthe model.Shaveret al. (1987) proposeda hierarchicalmodel of emotion words.The emotionwordsbecomehierarchically rganized irstaspositiveandnegativeand then withprototypicalemotions (love, oy, anger,sadness,and fear) at the superordinate eveland membersof these prototypicalemotionsstored at subordinate evels. This model, like the circumplexmodel,predicts hatpositiveprimingorfacilitationwilloccurcontingenton the closeproximity,n termsof distance,of the two emotionwordsofinterest.Although hesetwomodelshaveaddressed he mentalrepresen-tation of emotionwords, heyhavedone so usingexplicitmethods(suchas ratingstudies) and have not addressedthe notion of how emotionwordsarerelated to concreteand abstractwords n termsof mentalrep-resentation.A third model that has been studied implicitly is the semantic networkmodel, which is also called the spreading activation model. This modelproposesthatconceptsare connected to one another basedon seman-tic relatedness (Collins & Loftus, 1975). The more semantically relatedthe two concepts, the shorter the path between them. Activation of one

  • 8/13/2019 Emotion and Words

    12/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 399node, the source node, leads to activation of the next node, which leadsto activation of the next node, and so on. As the activation spreads, thestrength of the activation decreases, so that concepts further from thesource node are less likely to become activated than are nodes that arein close proximity to the source node. Although this model has not beenapplied to the mental representation of emotion words, it is plausible thatemotion words can be represented by the semantic network model.The current research uses a priming paradigm to examine the mentalrepresentation of emotion words. Priming paradigms have been usedextensively to examine the structure of semantic memory. If two wordsare semantically related and stored in close proximity in memory, thenthe presentation of the word prime should facilitate the response to theword target (Neely, 1977). In a lexical decision task participants judgewhether a string of letters is a real word or a pseudoword. Facilitation inresponses is assumed to reflect the degree of semantic relatedness be-tween two concepts. Highly related concepts are stored more closely to-gether. Related concepts also produce fasterresponse times in word prim-ing tasks. To our knowledge, the use of priming paradigms to study themental representation of emotion words in relation to abstract and con-crete wordshas not been reported. However,Bleasdale (1987) has appliedthe priming paradigm to investigate whether lexical processes are func-tionally distinct for concrete and abstractwords. To examine this he cre-ated associated and unrelated word pairs that varied on prime and tar-get concreteness (concrete-concrete [CC], abstract-abstract [AA],concrete-abstract [CA], and abstract-concrete [AC]). The related wordpairswere equated for mean association strength. Bleasdale hypothesizedthat concrete and abstract words have functionally distinct lexical pro-cesses, so homogeneous word pairs (CC and AA) should show primingand heterogeneous word pairs (CA and AC) should not show priming.In Bleasdale's Experiment 3 the word pairswere presented to participantsusing a priming and lexical decision task. The participants were asked torespond to the target as quickly and as accurately as possible. A typicaltrial proceeded as follows. A prime was presented for 177 ms, an inter-stimulus interval (ISI) lasted for 500 ms, and then the target appeared.Bleasdale found that all pair types (AA,CC, AC, and CA) showed signifi-cant priming.Bleasdale (1987) stated that this finding could result from postlexicalprocesses such as postlexical relatedness checking (checking the targetagainst the prime for relatedness after target recognition). To investigatewhether these findings resulted from postlexical processing, Bleasdalemasked the prime and decreased the prime presentation rate in his Ex-periment 4 to eliminate postrecognition processes and permit only au-tomatic processing to occur. Using the same stimuli as in his Experiment

  • 8/13/2019 Emotion and Words

    13/23

    400 ALTARRIBA AND BAUER3, a typical trialin Bleasdale's Experiment 4 proceeded asfollows.A primewas presented for 16.7 ms, there was an ISI of 32 ms, a mask appearedfor 168 ms, and then the target appeared. The results of his Experiment4 revealed significant priming for the homogeneous word pairs (AA andCC) and significant inhibition for the CAword pair condition. Bleasdalefound that when prime and target word pairs were homogeneous, theyprimed each other; however, when prime and target word pairs wereheterogeneous, they did not prime each other. Therefore, the results ofhis Experiment 4 imply that there are functionally distinct lexical pro-cesses for concrete and abstract words.The current research implements procedures similar to Bleasdale's(1987) in an attempt to establish whether emotion words elicit differentrelationships than abstract words. Concrete words were not included inthis experiment, as Bleasdale has already demonstrated a distinction inprocessing concrete words as compared with abstract words. To do this,we constructed related and unrelated word pairs for abstract-abstract(AA), emotion-emotion (EE), abstract-emotion (AE), and emotion-abstract (EA) conditions. The related word pairswere matched on meanassociation strength across the four pair types. The hypothesis was simi-lar to Bleasdale's. We hypothesized that under conditions of automaticprocessing homogeneous word pairs (AA and EE) would yield primingand heterogeneous word pairs (AE and EA) would not, suggesting thatdifferent lexical processing occurs for emotion and abstract words.Based on previous research conducted by Bleasdale (1987) on abstractand concrete word pairs,we hypothesized that in the current study,prim-ing would occur for the homogeneous word pairs (AA, EE) but not forthe heterogeneous word pairs (AE,EA). These resultswould suggest thatdifferent lexical processes are involved in the coding, storage, and retriev-al of abstract and emotion words.EXPERIMENT 3METHOD

    Prime-targetpairsweregeneratedusingAltarriba t al.'s(1999) collectionofwordassociations.Twentypairseach weregeneratedfor theAA,EE,AE,andEAgroups,fora totalof 80 prime-targetpairs.Themean concreteness oremotionwordswas2.86 (SD= 0.32), and the mean concretenessfor abstractwordswas3.37 (SD=0.61).Thisdifferencewassignificant, (79) =-7.03, p < .001.The meanimageabilityoremotionwordswas3.23 (SD=0.81), and the meanimageabilityforabstractwordswas2.8 (SD=0.90).Thisdifferencewasalsosignificant, (79) =3.27,p < .001. The meanstrengthof associationwas matchedfor the fourtypesof related word pairs:AA, 13%;EE, 13%; AE, 11%;and EA, 11% (all ps > .05).

  • 8/13/2019 Emotion and Words

    14/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 401These mean strengths are highly similar to those used by Bleasdale (1987) in hiscomparison of concrete and abstract words. Each target was then recombinedwith an unrelated prime. Additionally, 80 pronounceable word-nonword pairs,20 pairs in each of the four conditions, were produced from the same pool asthe critical items.Participants

    Eighty University at Albany students participated for fulfillment of a courserequirement. Participants had normal or corrected-to-normal vision and werenative English speakers.Materials and apparatus

    Two lists were created, each containing 40 related pairs (10 pairs in each ofthe four conditions), 40 unrelated pairs (10 pairs in each of the four conditions),and 80 pronounceable word-nonword pairs. Table 4 lists sample stimuli. Theunrelated word pairs in the first list and the related word pairs in the second list

    Table 4. Sample stimuli used in Experiment 3Condition

    Related UnrelatedAbstract-abstract

    easy/hard wide/hardvirtue/truth struggle/truthwisdom/knowledge turn/knowledgeanswer/response effort/responsemastery/expert beside/expertEmotion-emotionhappy/sad interested/saddelighted/thrilled neglected/thrilledaffectionate/caring indecisive/caringglad/pleased isolated/pleasedregret/remorse bewildered/remorseAbstract-emotionbusy/preoccupied apart/preoccupiedhonor/pride join/pridechaos/crazy protected/crazysoft/sensitive grow/sensitivewealth/greed scheme/greedEmotion-abstractaggressive/forceful puzzled/forcefulzealous/over terrible/overindifferent/same selfish/samerage/violence sentimental/violenceconfident/strong hopeless/strong

  • 8/13/2019 Emotion and Words

    15/23

    402 ALTARRIBA AND BAUERappeared in the same serial position across the two lists. The stimuli within thelists were randomized with the exception that no more than three consecutiveword pairs were of the same word pair condition. Each prime word and eachtarget word was presented only once within a list. The practice trial conditionswere proportionally equivalent to those of the experimental trials. This experi-ment was created using Micro Experimental Laboratory (MEL) psychology soft-ware (Schneider, 1988, 1990). The experiment was presented on a crystal scanmonitor interfaced with an IBM-PCcomputer. All primes and targets were pre-sented in white lowercase letters on a black background.Procedure

    Participants read instructions indicating that they would be participating in alexical decision task in which a fixation point, a prime, a mask,and a targetwouldappear. Their task was to indicate whether the target was a word or a nonwordby pressing either the m key or the z key, respectively. Participants were in-formed that both speed and accuracy were important. A typical trial proceededas follows. First, a fixation point ( + )appeared in the center of the computerscreen for 1,500 ms as a pretrial warning. Next, a prime word appeared in thecenter of the screen for 125 ms. Immediately after the prime, a mask of # sappeared. The mask appeared in the center of the screen for 50 ms and wasmatched in length to the longest word in the experiment. Immediately after thepresentation of the mask a target appeared for 1,500 ms. The participants had2,000 ms to indicate whether the target was a word or a nonword before the nexttrial began. Participants were shown 160 experimental trials and were given abrief rest period between the first and second block. A typical session lasted ap-proximately 25 min.

    RESULTSAND DISCUSSIONParticipant reaction times were excluded from the analysis if they wereless than 200 ms or greater than 1,500 ms. A 2 (prime) x 2 (target) x 2(relationship) repeated-measures analysis of variance was conducted.

    There was a significant prime x target x relationship interaction, F (1,79) = 5.41, MSE= 4,117.71, p < .05. There was also a significant main ef-fect of target, F(1, 79) = 15.85, MSE= 3,715.94, p < .001 and a significantmain effect of relationship, F(1, 79) = 26.98, MSE= 5,157.97, p < .001.Emotion words were responded to more slowly than abstractwords (755ms vs. 737 ms). Also, unrelated word pairs resulted in longer responselatencies than related word pairs (762 ms vs. 732 ms).Planned comparisons revealed significant facilitation and overall prim-ing in the homogeneous word pair conditions (AA and EE), and in oneof the heterogeneous word pair conditions (AE;ps < .05). However, nosignificant priming wasfound in the other heterogeneous word pair con-dition (EA; p > .05). Table 5 displays the mean reaction times, priming,and percentage error for Experiment 3.

  • 8/13/2019 Emotion and Words

    16/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 403Table 5. Experiment 3 mean reaction times (ms),priming, and percentage error

    Prime-target relationshipRelated Unrelated Priming effect

    Abstract-abstractM 708 754 +46*Error 2.0% 2.0%Emotion-emotionM 740 776 +36*Error 3.0% 4.0%Emotion-abstractM 738 749 +11Error 3.0% 2.0%Abstract-emotionM 743 767 +24*Error 2.0% 3.0%*p < .05.

    Error analysisA 2 (prime) x 2 (target) x 2 (relationship) analysis of variance on theerror data was conducted. There was a main effect of target, F(1, 79) =8.11, MSE= 0.00, p < .01. There were more errors when the targets wereemotion words (M = 0.03) than when the targets were abstract words (M =0.02). Further investigation showed that there were no speed-accuracytrade-offs because participants were both slower to respond and less ac-curate for word pairs with emotion targets. Therefore, no further analy-ses on the error data were conducted.Because the prime was constrained in this experiment (i.e., a mask wasused), these findings suggest that the significant facilitation and overallpriming in the homogeneous word pair conditions reflects automaticprocessing rather than strategic processing. However, the lack of prim-ing for the heterogeneous word pair condition (EA) suggests that asym-metric priming is occurring. There are severalwaysin which these resultscan be interpreted. It may be that abstract and emotion words possessdifferent attributes or are mentally represented differently.

    GENERALDISCUSSIONThe experiments presented here were designed to investigate wheth-er emotion words possess distinct attributes. Specifically, Experiment 1demonstrated that emotion words are more readily recalled from a list

  • 8/13/2019 Emotion and Words

    17/23

  • 8/13/2019 Emotion and Words

    18/23

  • 8/13/2019 Emotion and Words

    19/23

    406 ALTARRIBA AND BAUERtion that can be spread from a source. Anderson (1974) found that themore paths are associated with a source concept, the more difficult it isto retrieve the information along any one of these paths. The strongerthe path between the source node and an associated concept, the moreactivation the associated concept receives from the source node and sub-sequently the easier it is to retrieve information along this path.In Experiment 3 the strength of the word pairs was held constant.However, it is possible that emotion words have more paths than abstractwords. Emotion words may have more synonyms and antonyms thanabstract words. For example, the word happyhas synonyms such as delight-ed,pleased,ecstatic,and glad and an antonym sad.An abstract word suchas hourdoes not have as many synonyms and antonyms. Additionally,emotion words are rated higher than abstractwords on imageability andcontext availability.The higher rating on context availabilitymayindicatethat emotion words are associated with more contexts than abstractwords.The number of paths departing from a concept node for concrete,abstract, and emotion words was investigated in a normative study con-ducted by the authors (Altarribaet al., 1999). In this study, participantswere instructed to write beside a stimulus word the first word that cameto mind that was meaningfully related to the stimulus word. The num-ber of associations generated for each stimulus was tabulated. The meannumber of associations for each word type was then calculated. On aver-age, emotion words had the greatest number of associations (M = 23.48),followed by abstract words (M= 21.40) and then concrete words (M=17.34). Planned comparisons revealed that the mean number of associ-ations for all of the word typeswere significantlydifferent from each other(all ps < .05). These results are consistent with the aforementioned inter-pretation of the observed priming effects.Although there was no intention to directly test either the hierarchicalmodel of emotions (Shaver et al., 1987) or the circumplex model of emo-tions (Russell, 1980, 1989), the resultsof Experiment 3 do not support thepredictions of either model. The hierarchical model predicts that wordsin a cluster should prime each other more than words across clusters. Thecircumplex model predicts that emotion words stored closer togethershould show more priming than words stored further apart. Both the hi-erarchicaland the circumplex model would predict thatthe emotion wordshappy nd delighted ould show more priming than the words happy nd sadbecause happyand delightedre located much closer (and in the same sub-ordinate level in the subcategory labeled cheerfulness) than happy nd sad(which are located in separate clusters). The hierarchical model mayevenpredict that there should be no priming for the words happyand sad be-cause the word happy s located in a subordinate category in a differentcluster than the word sad,and the word happys on the positive side of themultilevel hierarchy,whereas the word sad is on the negative side. Because

  • 8/13/2019 Emotion and Words

    20/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 407priming occurs for antonyms (as was observed in Experiment 3), a semanticnetwork model of emotion appears to be more useful in the representa-tion of emotion than either of the other two models. The semantic networkmodel predicts facilitation for emotion words that are highly associatedregardless of where in the hierarchical model they are located and theirposition on the pleasantness-unpleasantness and high-low arousal dimen-sions of the circumplex model.In conclusion, the present data show that emotion words are morememorable and more readily recalled than concrete and abstract words.The results also indicate that the concepts represented by emotion wordsactivate different levels of concreteness, imageability, and context avail-ability relative to both abstract and concrete words. Specifically, the re-sults indicate that concepts represented by emotion words are more im-ageable and are easier to think of a context for than abstract words butare less concrete than abstract words. They are less imageable, less con-crete, and less likely to activate a context than concrete words. Addition-ally, when operating under automatic lexical processing, emotion wordsand abstract words prime associated words of the same type (AA and EE),and abstract words prime emotion words (AE), but emotion words do notprime abstract words (EA). These results are consistent with the predic-tions of the semantic activation model and are inconsistent with the pre-dictions of the circumplex and hierarchical network models. Taken to-gether, these results indicate that the concepts represented by emotionwords are characteristically different from and activate different relation-ships than abstract words.

    NotesWe would like to thank Allyse Scher for her assistance with data collection. Weare also grateful for the comments and suggestions ofJames H. Neely and Doug-las L. Nelson on an earlier version of this manuscript. Portions of this paper werepresented at the 37th Annual Meeting of the Psychonomic Society, Chicago, IL.This work was funded through a faculty research award granted to J.A., andportions of these data were submitted in partial fulfillment of the master of artsdegree in psychology by L.M.B.Correspondence about this article should be sent toJeanette Altarriba,Depart-ment of Psychology, SS 369, University at Albany, State University of New York,Albany,NY 12222 (e-mail:[email protected]). Received for publication April 4,2003; revision received September 1, 2003.ReferencesAltarriba,J., Bauer, L. M., & Benvenuto, C. (1999). Concreteness, context-avail-

    ability,and imageability ratings and word associations for abstract, concrete,and emotion words. BehaviorResearchMethods, nstruments,& Computers,1,578-602.

  • 8/13/2019 Emotion and Words

    21/23

    408 ALTARRIBA AND BAUERAnderson, J. R. (1974). Retrieval of prepositional information from long-termmemory. CognitivePsychology, , 451-474.Benjafield, J., & Muckenheim, R. (1989). Dates of entry and measures of imag-

    ery, concreteness, goodness, and familiarity for 1,046 words sampled fromthe OxfordEnglishDictionary.BehaviorResearchMethods, nstruments,and Com-puters, 21, 31-52.Bleasdale, E A. (1987). Concreteness-dependent associative priming: Separatelexical organization for concrete and abstract words.JournalofExperimentalPsychology: earning, Memory,and Cognition,13, 582-594.Bransford,J. D., & McCarrell, N. S. (1974). A sketch of a cognitive approach tocomprehension: Some thoughts on what it means to comprehend. In W.Weiner & D. Palermo (Eds.), Cognition nd thesymbolicrocessespp. 189-230).Hillsdale, NJ:Erlbaum.

    Campos, A. (1990). Concreteness, imagery, emotionality, and interest values ofwords when meaning is controlled. Perceptual nd MotorSkills, 71, 603-610.Chiarello, C., Senehi, J., & Nuding, S. (1987). Semantic priming with abstractand concrete words: Differential asymmetry may be postlexical. Brain andLanguage,31, 43-60.Christian,J., Bickley, W., Tarka, M., & Clayton, K. (1978). Measures of free re-call of 900 English nouns: Correlations with imagery, concreteness, mean-ingfulness, and frequency. Memory& Cognition,6, 379-390.Clore, G. L., Ortony, A., & Foss, M.A. (1987). The psychological foundations ofthe affective lexicon. Journal ofPersonality nd SocialPsychology,3, 751-766.

    Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semanticprocessing. PsychologicalReview,82, 407-428.Day,J. (1977). Right-hemisphere language processing in normal right-handers.JournalofExperimentalsychology: umanPerceptionndPerformance,, 518-528.de Groot, A. M. B. (1989). Representational aspects of word imageability andword frequency as assessed through word association. Journal ofExperimen-tal Psychology: earning,Memory,and Cognition,15, 824-845.de Groot, A. M. B. (1992). Determinants of word translation. Journal ofExperi-mentalPsychology: earning, Memory,and Cognition,18, 1001-1018.de Groot, A. M. B., Dannenburg, L., & van Hell,J. G. (1994). Forwardand back-wardword translation bybilinguals.JournalofMemory ndLanguage,33, 600-629.Ekman, P., Friesen, W.V., &Ellsworth, P. (1972). Emotion n thehumanface. Elms-ford, NY:Pergamon.Eviator,Z., Menn, L., &Zaidel, E. (1990). Concreteness: Nouns, verbs, and hemi-

    spheres. Cortex,26, 611-624.Friendly, M., Franklin, P. E., Hoffman, D. M., & Rubin, D. C. (1982). The Tor-onto word pool: Norms for imagery, concreteness, orthographic variables,and grammatical usage for 1080 words. BehaviorResearchMethodsandInstru-

    mentation, 14, 375-399.Gilhooly, K.J., &Logie, R. H. (1980). Age-of-acquisition, imagery, concreteness,familiarity, and ambiguity measures for 1,944 words. BehaviorResearchMeth-ods and Instrumentation,12, 395-427.Goleman, D. (1995). Emotional ntelligence.New York: Bantam.

  • 8/13/2019 Emotion and Words

    22/23

    EMOTION, CONCRETE, AND ABSTRACT WORDS 409Holmes, V.M., &Langford,J. (1976). Comprehension and recall of abstract andconcrete sentences.Journal ofVerbalLearningand VerbalBehavior, 5, 559-566.Izard, C. E. (1971). Theface ofemotion.New York:Appleton-Century-Crofts.James, C. T. (1975). The role of semantic information in lexical decisions. Jour-nal ofExperimentalsychology: umanPerceptionndPerformance,04, 130-136.Kroll,J. F.,&Merves,J. S. (1986). Lexical access for concrete and abstractwords.

    JournalofExperimentalsychology: earning,Memory, nd Cognition,12, 92-107.Kucera,J. F., & Francis, W. N. (1967). Computational nalysisof present-dayAmeri-canEnglish.Providence, RI: Brown University.Neely,J. H. (1977). Semantic priming and retrieval from lexical memory: Rolesof inhibitionless spreading activation and limited capacity attention.Journalof Experimental sychology:General,106, 226-254.Nelson, D. L., & Schreiber, T. A. (1992). Word concreteness and word structureas independent determinants of recall. JournalofMemoryand Language,31,237-260.

    Paivio, A. U. (1971). Imageryand verbalprocesses.New York: Holt, Rinehart &Winston.Paivio, A. U. (1986). Mental representations:A dual coding approach.New York:Oxford University Press.Rubin, D. C. (1980). Fifty-one properties of 125 words: A unit analysis of verbalbehavior.Journal of VerbalLearningand VerbalBehavior,19, 736-755.Rubin, D. C., & Friendly, M. (1986). Predicting which words get recalled: Mea-sures of free recall, availability,goodness, emotionality, and pronounceabilityfor 925 nouns. Memory& Cognition,14, 79-94.Russell, J. A. (1980). A circumplex model of affect. Journal ofPersonalityand So-cialPsychology, 9, 1161-1178.Russell,J. A. (1989). Measures of emotion. In R. Plutchik & H. Kellerman (Eds.),Emotion:Theory, esearch, nd experienceVol. 4, pp. 83-113). San Diego, CA:Academic Press.Russell, J. A. (1991). Culture and the categorization of emotions. PsychologicalBulletin, 110, 426-450.Russell, J. A., & Bullock, M. (1985). Multidimensional scaling of emotional fa-cial expressions: Similarities from preschoolers to adults. Journal ofPerson-alityand SocialPsychology, 8, 1290-1298.Schneider, W. (1988). Micro Experimental Laboratory:An integrated system forthe IBM-PCcompatibles. BehaviorResearchMethods, nstrumentation,& Com-puters,20, 206-217.Schneider, W. (1990). MELuser'sguide:Computerechniquesor real timeexperimen-tation.Pittsburgh: Psychology Software Tools.Schwanenflugel, P.J., &Akin, C. (1994). Developmental trends in lexical deci-sions for abstract and concrete words. ReadingResearchQuarterly, 9, 251-263.Schwanenflugel, P.J., Akin, C., & Luh, W. (1992). Context availability and therecall of abstract and concrete words. Memory& Cognition,20, 96-104.Schwanenflugel, P.J., Harnishfeger, K. K., & Stowe, R. W. (1988). Context avail-abilityand lexical decisions for abstract and concrete words.Journalof Memoryand Language,27, 499-520.

  • 8/13/2019 Emotion and Words

    23/23

    410 ALTARRIBA AND BAUERSchwanenflugel, P.J., & Stowe, R. W. (1989). Context availability and the pro-cessing of abstract and concrete words in sentences. ReadingResearchQuar-

    terly,24, 114-126.Shaver, P., Schwartz,J., Kirson, D., & O'Connor, C. (1987). Emotion knowledge:Further explorations of a prototype approach. Journal ofPersonalityand So-cial Psychology, 2, 1061-1086.Toglia, M. P., & Battig, W. F. (1978). Handbookof semanticwordnorms.Hillsdale,NJ: Erlbaum.Vellutino, F. R., & Scanlon, D. M. (1985). Free recall of concrete and abstractwords in poor and normal readers. Journal of ExperimentalChildPsychology,39, 363-380.Whissell, C. M. (1989). The dictionary of affect in language. In R. Plutchik &H.Kellerman (Eds.), Emotion:Theory,research, nd experienceVol. 4, pp. 113-

    111). New York:Academic Press.