The concept of 'joy' in Old and Middle English. A semantic analysis

(1)

The concept of 'joy'

in Old and Middle English

A semantic analysis

Malgorzata Fabiszak

Wyzsza Szkola Biznesu

Pila

(2)

INTRODUCTION

The aim ofthis work is to arrive at a description ofthe semantic structure of the concept of ,joy' in OE and ME periods as well as to pinpoint changes that the concept underwent over this time. The concept of 'joy' is here understood as a mental representation ofthe meaning ofa lexicalfield.Inthe presentstudyonly the core members ofthe field will be analysed for OE. These have been identified as bliss, blioe, dream,glcednes,liss, mirhoand wynsumnesse.

The development ofthese words will be traced on intoME,even ifthe words show signs offalling out ofthe field of 'joy', i.e.Iwill in this respectbe adopting a semasiologicalapproachto the change ofmeaning.New words borrowed intoME to designatethe conceptof 'joy', such as cheer, delight, gay (gainess) andjoy, will alsobe investigated,Iwillthus alsobe incorporating an onomasiological perspective. The choiceofthisparticularfieldis based on the suggestion that it is relatively poorly describedin thehistoryofthe Englishlanguage(cfStrite1989on OE,Diller 1992).A number of works have been devoted to other emotion words, 'gloom' in particular (for example: Lochrie1986,Magennis1986).Those studieswhich have dealt with 'joy', however, have either investigated it with a different aim in mind (Ostheeren1964)or focused on other lexical items (Fell1982-83).

The present study consistsoffour chapters and a conclusion.

Chapter One is a review ofmethods and theories in lexical semantics. The review is intendedas an attemptto find a common core in the competing theoriesso that the analysis of the datacan be based on widely accepted, thoughoftenvariously labelledmethods. Thischapterbeginswith a surveyofthebasictenetsoflexical field theory and compositional analysis, devotes some space to those approaches to semanticsstemmingfrom generative transformational grammar, and finishes with a discussionofcognitive semanticsas advocatedby the Berkeleyschool.

ChapterTwopresents a numberof studies devotedto emotions froma number of different perspectives: psychological, linguistic and socio-cultural. The goalof this chapteris to provide a widerbackground formy study. Emotions havebeen aresearch

(4)

2 Malgorzata Fabiszak - The concept of 'joy' Malgorzata Fabiszak - The concept of 'joy' 3

concernin psychologyfor almosta hundredyears. I therefore believethat problems identified within this discipline, as well as attempts at solvingthese problems, can serve as reference points for those studying emotions from a differentperspective. Aninterestin emotions in socialhistory, on the otherhand, is a recent development, but this thrivingresearchfieldcan also contributeto our understanding ofemotions and, in particular, changesin emotions.

The first part of Chapter Three is devoted to an analysis ofOE words for 'joy'. TheMicrofiche Concordance to Old English(henceforthMCO£)serves as a source of data. The second part presents the results ofthe analysis which is an attemptto constructthe scenario for 'joy' in OE andto identifythe metaphorswhich. governit.

Chapter Four contains an analysis of ME terms for 'joy'. This chapter not onlyuses the methodology developed in ChapterThree,but alsoutilisesitsresultsin both the analysis and interpretation of data. The data come from a number of concordances, supplemented withliterarytextsas indicatedin theBibliography. The chapter ends with an outline ofdevelopmentsbetween the OE and ME periods.

The thesis endswith conclusions which summarises the resultsofthe study. The ReferenceSectionconsistsoftwo parts:the sourcesand referencebooks employed, and the literature quoted in the thesis.

CHAPTER ONE

LINGUISTIC APPROACH

TO LEXICAL SEMANTICS.

1.1. Introduction.

Diachronic lexical semantics stemsfrometymology, whichhasbeenan interest oflinguistically oriented philosophers at leastsincePlato.Initsmost commonformit was based on a beliefthatwordshave a truemeaning(theidealin the Platonic sense) and that the role ofa philosopheris to fmd this meaning.Inmodem times the quest for the only truemeaningwas oftenunderstoodas finding the Latinor Greeksource word and tracingits development in otherEuropeanlanguages (MalkieI1993). This model oflinguistic inquiryhas a strongresonancein the folk theoryoflanguageand manifests itselfin vehement public discussion on the deterioration and decay of language, itslexical meanings andgrammatical form. With thedevelopment ofhistorical linguistics, a betterunderstanding ofthe mechanisms oflanguage changediscredited such views as non-scientific. The approach to the study ofthe change ofmeaning also evolved.

1.2. Semantic fields.

Trier (1934) was one ofthe first scholars,working within the framework of

field theory,topointouttheweaknesses of atomistic approaches tomeaning. Studying wordsin isolation leads to a neglect of important generalisations. As a post-Saussurean structuralist,he advocated an analysis ofwords as parts ofa system, aword-jield.

In his influentialstudy ofthe conceptual field ofknowledge and understanding in Middle High Getman, he pointed out that a change in meaning ofone lexicalitem influences the meaningof the relateditemsandthe structure ofthe systemitself The organisation ofa lexicalfield, as conceivedof by Trier, was based onparadigmatic relations. Otherscholars, notably Porzig (1934),understoodsemanticfields as sets

(5)

4 Malgorzata Fabiszak - The concept of 'joy' _{Malgorzata Fabiszak - The concept of 'joy'} ₅

ofwords standing insyntagmatic relations, arranged in terms ofcollocations and selectional restrictions.

Trier has been criticised for making some ungrounded assumptions, for exampleforpostulatingthe mosaic structureof lexicalfields. Such astructur~implies that all the possible meanings within a conceptual field are expressed lexically, or to put it in other words, that the whole semantic space is covered by themeanings oflexical items. This position has been criticised by Lehrer (1974), whop~oved that this is not the case in many lexical fields. Following the results ofBerhn and Kay's (1969) research on colour terms, she suggested that lexical fields consist.of a set ofbasic termsand less clearly delineatedperipheral terms. Componential analysis ofbasic terms determines the dimensions or parametersst~c.mringthe lexical field. Parameters or components weave into the senses ofexistmgterms, but not all possible combinations ofthe components are lexicalised,thu~ cre~ti~g lexical gaps in the field. The lack of lexicalisation is oftend~e t~ extral~ng~Istlc, pragmatic factors oflanguage use. Only ifa need for a new lexical Item ansesillthe

speech community is a lexical gap filled. The drive to internal symmetry or analogy

plays no role in filling the gaps. . .

Apart from the meaning components common to the words~napartlcul~r semanticfield,these words, especiallybasic terms, often share syntacticstructuresill

which they can appear, as illustrated by Zwicky's (1971) analysis

ofmanner-of-speaking verbs. . .

These two characteristic features oflexical fields, I.e.semanticparameters creating lexical gaps, and common syntactic behaviourofwo~d~belonging to :he sanlefield endowfieldtheorywith a certainexplanatory andpredictive power. Lexical gaps help

~o

explain the extension ofmeaning oflexical

item~

and the

inventi~~

and bon-owing ofnew items. The shared syntacticbehaviour predictsthe collocabilityof the new items added to the lexical field.

The theory of semantic fields has been successfully applied in lexicographic research, for example in constructingRoget

s

Thesaurus(1962) and thedI~tlO~~ry of about500Germanverbs,designed as a reference forteachersat Germanumversmes preparingcoursesandtextbooks in Germanas a foreignlanguage(Ballweg-Schramm 1981).Itis also believed that semantic fields understood as sets ofcontrastl~eItems are a necessary device for the organisation of memory, language learning and categorisation (Grandy 1987).

1.3. Componential analysis.

Another approach to lexical semantics, stemming from structural linguistics, and often used in combination with the semantic field framework, iscomponential analysis.Itis based on the assumption that words in a lexicon constitute a structure ofsystematically related elements which themselves are further decomposable.It aims at determining a particular word's place in this structure by delimiting the

distinctive featuresthat recur in the meaning ofrelated items. They are often held in binary orn-tuple oppositionswhich weave into a complex net of paradigmatic relations. Moreover,binaryoppositionis consideredto be "one of the most important principles govemingthe structureoflanguages" (Lyons 1977:271).

Bendix (1966), following Stem (1931) rejected the idea that words may be studied in isolation. Hence he constructed a set ofnuclear sentences ofthe form "A has B", "A gives B toC",etc. to establish what relations verbs in such sentences create. One ofthe bases ofhis theory was that to describe the meaning ofa sentence is to supply its closest possible paraphrase. Thus he subjected contrastingsentences with the verbs under study to native speaker's judgements. From these he extracted criterialcomponentsofmeaning.Then he devised a set ofconversionrules which, in a quasi-logical form, presented a defmition ofmeaning ofa given word. Finally,he presented the results in a table ofcomponents for the whole domain.

The method ofcomponentialanalysisintegratedwith the fieldtheory has been further refined and tested in numerous semantic studies conducted by the linguists from the Tiibingen School of Structural Semantics. The framework, as originally developed by Coseriu (1962, 1967) under the name oflexematics, is basically paradigmaticand employsdecompositioninto the meaning-differentiating features. A lexical field, according to Coseriu (for an excellent summary see Geckeler 1981,

1988),consistsoflexemes analysable by meansofsemes(minimaldistinctive features organised in oppositions along a dimension characteristic for a single field) and

classemes(units fimctioning across linguistic fields, determining syntagmatic relations, often grammaticalised). The contentofthe whole lexicalfield is expressedby means ofanarchilexeme,a general label determiningthe scope ofthe field,which may but need not be lexicalised in a particular language (Geckeler 1988: 15, Lyons 1977: 296).

Lehrer (1974) pointed to the contribution componential analysis can make to the understanding of the dynamicnature oflanguage.She emphasisedthe difference between general and salient components (similar to the markers and distinguishers

(6)

6 Matgorzata I...."htc"''''oIr-The concept of 'joy' Malgorzata Fabiszak - The concept of 'joy' 7

posited by Katz and Fodor (1963) - see below). Non-universal components of meaning(distingushers) maybe moresalient forthe meaning of a givenwordandit is these that determine the metaphoric extensions of words. "If there is no way of matching these features with anythingin the new field of discourse, there does not seem to be any point in using the words. Moreover,much of the rhetoricalforce of using these terms in slightly non-standard contexts is achievedby associating the normal [salient-MF] component with a newreferent"(Lehrer1974: 116). The ability ofthesecomponents to influence andextendthe meaningof wordsmodifiedby their carriers, without themselves belonging to the specification ofthese words, aligns them with Weinreich's(1966)transferfeatures.

The basic tenets ofcomponentialanalysis treated in a less orthodoxmanner than in the approaches presented above are also present inthe contextual theory oflexical meaningdevelopedby Cruse (1986, 1988).Withinthis frameworkword meaning is analysed in terms ofsemantic traits,which are derivedfrom theactual

andpotential contexts.Actualcontextishereunderstoodasthe data-derived context, whereaspotential context isthe one available to nativespeakers (including linguists), through introspection. The meaningofa word canbe derivedfrom an investigation ofboth its syntagmatic(collocation)and paradigmatic(substitution)affmitiesand is understood as the entirepattern ofa word's contextualrelations.

1.4. The criticism and defence of componential analysis.

Componential analysis in combination with semantic field theory has been utilisedin severalanthropological and linguisticstudiesof, for example, possession verbs(Bendix1966), cooking verbs,beliefpredicates, container words, soundwords (Lehrer1974),judgingverbs (Fillmore 1971), semantic oppositions (Mettinger 1988a and b), and the domain ofhumans (Kleparski 1990).

Despite itswide application the featurebased approachhas been criticisedon thefollowing grounds:

1.Semantic components are indeterminate, andtheir ontological statusvague so that one cannot tell the difference between metalinguistic components and the words of the naturallanguage.

2. Binary oppositions do not allow gradability of concepts as revealed by prototypeeffects.

3. The approachsuggests that certainconceptsaremore complexthan others, which is allegedly not supported bypsycholinguisticexperiments (seebelow).

The suggested solutions to thesecriticisms, however, eitherlookconspicuously

similarto componential analysis (points 1and 2) or do not themselves withstandthe same criticism (point3).

BalmerandBrennenstuhl (1981 a andb) rejectfeature analysis on the grounds that the ontologicalstatusoffeatures is disputableand that discoveringexhaustive lists offeatures both for the meaning ofa single word and for the entire lexicon is problematical. They propose aholistic approach to lexical analysis,that is starting with the analysisof the entire lexicon and breaking it down into smallerparts. The firstdivision theysuggest isthatintoword categories, i.e.nouns, verbsandadjectives, adverbs and prepositions. Then they proceed to break up the verbs into smaller groups on the basis of two semantic relations:similarityandpresupposition.This leadsto the establishment of averbgroup' ,which seems to be nothing more than a group ofsynonymsofa classicalThesaurus, for instance: "verbgroup, e.g.tunjd 1 etw2 (do), durchfiihrenjd 1 etw2 (perform),machenjd 1 etw2 (make),bastelnjd 1 (work at as a hobby), schaffenjd 1 (execute), tatig seinjd 1 (be busy)". The next step is to extract atitle verbwith the most general meaning (cf. a thesaurus entry), so that the remainingverbs in a group can be paraphrasedby means ofthe titleverb with a modifier. Verbgroup together with its title verb creates averbcategory.On the basis ofsimilarity oftitle verbs these can, in tum, be grouped intoverbmodels

(cf.scripts, scenes,etc.). For example, the Individual/Object-Existence Model consists ofthefollowing verbcategories: NotExist-Come IntoExistence-Exist- Vanish-Have Existed (Balmerand Brennenstuhl 1981b: 421-422).

The question arises: what is left of the novelapproachonce it is stripped ofthe new labels? How doesthismethoddifferfromthe commonlexicographic practice of grouping synonymous words in thesauri and paraphrasing them in alphabetical dictionaries?Infactno differenceis detectable.

However, duetribute mustbe paidto theauthors forpointing outcertain general trends in the organisationofverbmodels. They singledout three dimensionsalong which verbmodelsare structured. These are:aktionsart,i.e. the linear arrangement in the process phases;intensityofthe process visualised by means of a hat graph; andinterference(involvement, influence), i.e. degreeofsubjectagentivity.

Nevertheless, as far as methodologyis concerned BalmerandBrennenstuhl's approach seemsjust a notional variant ofthe lexicologicalapproach.Moreover,if we disregard the ontological status of semantic features, it is difficult to see the difference between a classical paraphrase, feature analysis and the Balmer and Brennenstuhl method. Whatthey alldo is describe themeaningof a word in different

I Terms: verbgroup, verbcategory, verbmodel, title verb, hat graph follow the terminology and the spelling

(7)

8 Matzorzata Fabiszak - The concept of 'joy' Malgorzata Fabiszak - The concept of 'joy' 9

metalanguages, one closer to natural language and the other attempting to draw on the artificallanguageoflogic.Inmost cases,however,both would point to the same components ofword meaning.For example:

not commit ourselves t? the philosophical implications adhered to by the author, it ~rovesto be a

:r

ery efficient methodofsemanticanalysis (Fabiszak2000).Moreover, Itclearlyremainsoneofthebetterdeveloped andmorewidelytestedsemantic theories (see the collection Goddard and Wierzbicka 1994).

review of the theory of prototypes see Section 2.2.

The criticismofcompo~enti~lanalysis stemmingfromresearchon prototypes should not be regarded as being aimed at discrediting the model but rather as an important

~sight

that can improvethe existingtheory. Most linguists workingwithin the~elllantic fe~ture fra~eworkincorporate the resultsofRosch's- experiments into their own studies, admit the existence of focus-periphery structure and introduce prototypicality-basedordering oflinguistic features (Lehrer 1990,Lewandowska-Tomaszcz~k 1985, 1993). Nonetheless, it must be stressed that the theory of prototypes IS a theory of conceptualisation and not oflinguisticcoding. The idea ofa prototyp~asth~centre ofa conceptual category is useful for semantic theory in as much as Itprovides a referent, understood as a mental representation and not a real world entity, for the lexical item (Wierzbicka 1985, 1990).Thediffere~cebetween themeaningofa lexicalitem and itsrealword referenthas oftenbeenunderestimated inprot~type ori~nt~dlinguisticinvestigationsand has led to a misrepresentation of selllantI~meanmg In terms offeatures characteristic ofthe referent, which do not necessanly have to~oincidewith those ofa lexical entry (for details see Kastovsky 19~8).Encyclopaedic~owledgeofthe world as perceivedand interpreted by human beI~gsshoul.d not be mistaken for the dictionary meaning ofwords. Definitions of

~eX1calm~anmgusuallyreferto encyclopaedic knowledge, buttheyalsoincludevital

informationon a word's expressive meaning, register affiliation and collocational

J."" ...HJ.~v"JlVH"(Cruse 1988, 1990).

Theeffect ofthetheoryofprototypes on lexical semantics is oftenoverestimated,

",,",,-I+h.~striking similaritiesbetween prototype theory and field theory overlooked.

can be demonstrated by a comparison ofCoseriu's lexematics and prototype SenlantIcsas follows.

Althou~ ~he th~ory ofprototypesclaimsto examinethemeaningofconcepts,

cases It IS their lexic.al realisations that serve as inputs in psychological ~:q)e,nLrrll\Xl·ti::'l.Hence, the baSIC level concepts and their superordinates posited by . andarchilexemes, witha hyponymic relation obtaining

~he unIty.oft~e semanticcategory/fieldis determinedby its internal

which organise ItS structure and attributes/featuresand account for the contentofa word (for prototype theory see Russel and Mehrebian 1977

_,

(a) 'mother' 1. Your mother is the woman who gave birth to you.(Collins

Dictionary)

(b)'mother' [female],[parent]

(c) 'mother' person giving birth (a possible definition within Balmer and Brennenstuhl framework)

A more viable solution is the one developed by Wierzbicka (1992, 1996) who suggested that semantic components reflectsemantic universals (primitives)

common to all humankind. The first step

in

semantic research should then be to produce a set of such primitives through cross-linguistic research. TheNatural Semantic Metalanguage (NSM) as worked out by Wierzbicka and her colleagues is very appealing,but has certaindrawbacks.The list ofthe proposed universalshas grown from 14 (1972) to over 50 (1996). They are not mutually independent and it isdifficult to ascertain theirpsychological reality. Moreover, in defining abstract entities, Wierzbickaallowsthe use of featureswhich are not includedin the list, for economy and ease ofdescription, on the assumption that these components are defmable meansofprimitives.Interms

to employthe higher levelitems.However,if we do so, our semanticanalysis longer meet the requirement ofbeing based on universals and will notn'++O,"-rY'llli"'h

from classicalcomponential analysis, whose only metalinguistic commitmentis the defining features should exhibit a higher frequency ofoccurrence, most determinedonly by a linguist's introspection, than the definedterm.

Nevertheless,ifwe treat NSM as a tool for the description ofmeaning and where 'woman' of the firstdefmition= 'female' ofthe second= 'person' ofthe third, and 'who gave birth to you'= 'parent' = 'giving birth' . Ofcourse, these defmitions highlightslightlydifferentaspectsofthe word meaning.The firstis the most detailed one, but disregards the relationships between the defmed word and other words in the lexicon;the seconddoes not differentiatethe role ofthe female as opposedto the male parent,but placesthe word fmnly withinthe network ofotherlexicalitems; and the third gives the highest superordinate ofthe defmed word thus disregarding the component 'female', but emphasising thelink: withallthe otherwordsdenoting human beings. These nuances ofmetalinguistic description, however, are too small to be regarded as a solution to the problem ofthe ontological status ofsemantic features.

(8)

10 Malgorzata Fabiszak - The concept of 'joy' _{Malgorzata Fabiszak - The concept of 'joy'} ₁₁

for lexematicssee Geckeler1988).In fact, attributes or elements used as prompting cuesin psychological experiments arenothingmorethan semantic features (Pullman 1983,Lipka1986).Their psychological reality is the same as that ofany meaning component, that is theyworkthroughtheir association withnaturallanguage words. This fact is used by Cruse(1986)in his contextual approach to lexical meaning to circumvent theproblemoftheontological status of semantic features. He emphasises that the meaningof a word manifestsitselfin its relations to otherwords and should bedescribedin termsoftheserelations withoutcommitting himselfto any statement on whatmeaningreallyis.

Thethirdcriticism of componential analysis, i.e., thatitpresupposes thatcertain concepts are more complex than others, which, allegedly, is not borne out by psychologicalresearch,has been put forward by Chierchiaand McConnel-Ginnet (1990: 360-368).They advance the Carnapian meaning postulate approach and argue for it as a more adequatetheory on the basis ofthe acquisitionof meaningby children. Theirlineof reasoning runs as follows:

Children acquire words likekill and die long before they learn words like' cause' and

'become' .... it is striking that what we have analysed as the relatively more complex items semantically are apparently more directly salient for children. (...) Even at a less abstract level we find the order of the acquisition frequently opposite to what relative complexity of proposed translations might predict. As Fodor (1987) p.161 observes 'Children know about fathers long before they know about males and parents ....'( ...) The decompositional approach suggests that the simple concept of 'father' that children first acquire is different from, and perhaps ultimately replaced by, the complex one that they later built from 'male' and 'parent'. In contrast, the meaning postulate approach need only suppose an enrichment ofIPC [Intensional Predicate Calculus -MF] to include basic expressions like 'male' and 'parent' and the concepts they designate along with meaning postulates connecting these later acquired concepts to one another (Chierchia and McConnel-Ginnet 1990: 363).

The acquisition argument above does not seem to have more bearing on the psychological plausibility of componential analysis thanit doesonmeaningpostulate approach.Itis quite obvious that the "simple conceptof/atherthat children first acquire is differentfrom ...the complex one that they laterbuild." In fact the whole conceptual system of a child is different from that which evolves later,be it word meaning, syntactic structures or more general cognitive skills (Gopnik 1983). However, if the original conceptis abandonedor merelydeveloped is not of concern

for decompositional analysis, which,by defInition, doesnot accountfor a changein meaning, whetherfromthediachronic perspective ofa language orthedevelopmental perspective ofanindividual language user, butconcentrates ondetermining themeaning of the words synchronically (Coseriu 1967). Consequently, the developmental argument disputes one ofthe basic assumptions upon which the theory offeature analysisis founded, but cannot be regarded as a challengeto its methodology.

1.5. Semantics and generative grammar.

Semanticists coming from the Chomskyan tradition of generative syntax attemptedto devise a semantictheory which proposed projectionrules generating themeaning of sentences andphrases fromthemeaning oflexicalitems asrepresented in the dictionaryof a particularlanguage (InterpretiveSemantics- Katz and Fodor 1963).Theirworkoften involved criticism ofthephilosophical implications ofprevious approaches. Katz(1972),for instance,rejects the ideational, the referentialand the behaviouraltheories ofmeaning as inadequate.Fodor(1977)disputesthe value of reductionist definitions as he believes that "the reduction of one kind ofthing to anothercanbe veryrevealing, but there is no guaranteethat such a reductionwill be possiblefor anyparticulargivenkind ofthing" (Fodor1977: 43).Theproblemwith such an argument is that any type of science is by its very virtue a hypothesis that humans, endowed with cognitive abilities, construct about the external world. Hypotheses as such can onlybe expectedto hold for somepart of realityand,unless they are highly general,do not attemptto make claimsaboutevery"particularkind ofthing".

Nevertheless, when facedwith the questionofthe defInition oflexicalitems, semanticists had to yield to an atomistic (and hence reductionist) approach. For KatzandFodor(1963)themeaning ofwordscanbe analysed intosmaller components andrepresented bymarkersanddistinguishersstrikingly similarto semantic features (foran illuminating discussion on similarities betweenmarkersand distiguishers on theone hand and classemes and semes on the other see Lyons1977: 327).Semantic markers are awardeda universalstatus(see alsoWierzbicka'sNSM). Katz's(1972) most cherishedexampleis that of 'bachelor' analysedas

bachelor:(Human)(Male)(Adult) [whohas never married]

(9)

Another attempt to describe lexical meaning withintheframework ofgenerative semantics was that ofFillmore (1968) who developed a theory ofcase relations.It aims at definingthe semanticstructureof verbs in terms of the case relationsof their arguments. These relations determine sentence syntax structure and explain the combination ofverbs and their arguments. They also characteriserelations among verbs,illproviding a particularly neatmethodof relatingcausative verbsto theirnon-causativecounterparts. Their specifications differonly in terms of which case must be selected as their subject. Fillmore suggested six case categories: Agentive, Instrumental, Dative, Factitive, Locative, Objective. The major contribution of this approachis a more completespecificationof the meaning ofverbs. However, it was criticised by Chomsky(1970)as not abstract enoughandnot preserving meaning relations.

Gruber (1965) formulated a theory of thematic relations, which was based on the assumption thatthe spatiotemporal domainis the underlyingfoundation of all human reasoning.Itis the conceptual construct in tenus of which more abstract conceptsare generated. Accordingto this theory all verbs canbe analysedby means of thematic roles such as Agent, Patient, Theme, Source, Path, Goal. This approach has been further developed by Jackendoff (1983, 1990) and Grimshaw (1990)3 among others.

Jackendoff develops a full theory of semantics, starting with philosophical assumptionsand fmishingwith detailed specificationsof word meanings. For him, word meaning is a part of a conceptual structure together with its phonological and syntactic specifications. The conceptualstructureis connectedwithperceptual and motor modules of the mind, so that a lexical entry for physical objects also includes a 3D (three-dimensional) model of the object as put forward by Marr (1982), and lexicalentriesfor verbs would include a scenario determiningwhat the action looks like and what it feels like to perform it. The major advantageof the 3D model is that: "[it] eliminate[s] the need for a plethora ofobjectionable conceptual featuresillfavourof geometric representation ...[whichremains] invisible to syntax." (Jackendoff1990: 33-34). Simultaneously, "referential linking is not all there is to word meaning" (Jackendoff1992: 59).

Toposit intermodulelinks between conceptualstructuresand perceptualand motor modalitiesis a very elegantsolutionto the problemof the defmitionof natural kinds,but cannotbe appliedin specification of meanings of abstractterms.Abstract terms, and emotion tenus in particular, do not have objective reference and hence cannot be stored as 3D models. In most cases they cannot be determined purely on

3See Section 2.4.5.

the basis of emotional display, so the link between the cognitive module and the perceptualmodule(fordecoding emotional displays) andwiththemotormodule(for encodingthem) is not sufficient for the understandingof these terms. They can only be understood and learned in terms ofshared and differentiating properties with other concepts.

Nevertheless, the intermodule links can produce some interestingresults in definingemotion concepts. Ifwe assumethat a lexical entryfor an emotionconcept includes linksto representations ofthe mostcommonperceptions andmotoractivities related to experiencingemotions,then they can account for the dilemmas linguists encounter in determiningwhether a particularword belongs to the field ofemotion terms or not. For example, Omondi (1997) observed thatillDhoulo it is difficultto judge ifa given use of a word refers to emotion, physiological change within the organismthat the emotionmighthave caused,or to the actionwhichis the culturally accepted means of expressing the emotion. Ifall three aspects are unified in one conceptual structure thenit is onlynaturalthatlexicalitemsusedilloneofthe domains might be extended to the other related domain, through metaphoric or metonymic extension.

Jackendoff(1992) postulates the existence ofbasic conceptual categories, such as Object, Event, Place, Action, Property and Amount. These coincide with thesemanticprimitivessuggestedby Wierzbickafor her NSM. For instance, Object canbe translated as SOMETHING,Event as HAPPEN, Place as HERE, Action as

DO, Propertyas GOOD,BAD, BIG, SMALL andAmount as ONE, TWO,~

SOME, ALL. The factthatsemanticists workingwithindifferent frameworks arrived atsimilarresultsstrengthens theirconclusions. However, Jackendoffemphasises that theproposed categoriesand thematic roles should not be regarded aspsychological universals,but rather as "relational notions defined structurally over conceptual structure, with a status precisely comparable to that ofthe notions of Subject and Objectillmany syntactic theories" (Jackendoff1990: 47).In thisrespectthesenotions share the same status as the semantic features of classical componential analysis whichclaim no psychologicalrealitybut are purely functional and intralinguisticill

nature (Kastovsky 1988,Mettinger 1988a and b).

Conceptual constituents, according to Jackendoff, combineilla rule-governed fashionto generate word, phrase and, ultimately,sentence meanings. The rules are integrated in Preference Rule Systems, which account for prototypical effects. "Preference Rule Systems statethatthe decompositional features of a conceptcreate ahierarchical structure withthe stereotypical instance occupying thetopmostposition withthe lessstereotypical instances gradedillan orderlyfashion belowit" (Jackendoff 1990: 37). He also notesthat "fuzzinessof categorisation may be somewhatstrange

(10)

in a strictly digital conception ofmental computation, but slightly less strange in a system that contains the possibility of continuously graded strengths of criterion application" (JackendoffI992: 8).

Indefining verbmeaning Jackendoff concentrates on the formal specification of verbs' thematic roles, their selectional restrictions and their mappings to

argument structure.Thematic roles can be divided into two tiers: thethematic tier

and theaction tier.The thematic tier deals with motion and location and the action tier is responsible for Actor-Patient relations. Selectional restrictions determineverb arguments and "shouldnot be regarded as a conceptualconditionon the insertionof a verb. Rather, it is apart ofthe verb meaning" (JackendoffI990: 53). Thematic roles oftheLexical Conceptual Structure (LCS) are eventually linked to syntactic structuresin the syntactic module.

Inhis conceptual theoryoflexicalmeaning,Jackendoffdrawson the fmdings offield theory.He stressesthat "to determine the boundaries ofexplanation rather than mere description, it is necessary to consider the behaviour ofeach individual verb within a context ofthe overall class to which it belongs. The class as a whole and its place in the larger system must be examined in detail in order to layout a place ofa verb within the system" (JackendoffI990: 123).

Conceptual Semanticsalso shares certain featureswithCognitive Semantics.

They both attempt to describe human cognition and categorisation in terms of a

representational theory ofconceptsand believethat spatialconceptsare the primary basis for the development of abstract concepts. They differ in that Conceptual Semanticsadheres to theautonomy ofsyntax principleand is committed to the

formalisationof the theory.

Another attemptto create a satisfactorysemantictheory,developedfrom the Chomskyan tradition, is thegenerative lexicontheory proposed by Pustejovsky (1996). He stresses the importance ofthe link between meaning and structure and maintains that lexicalmeaning cannot be successfullyanalysedwithout taking into account syntactic structure. This basic assumption underlies the structure ofthe generative lexiconwhich, he suggests, consists of at leastfourlevels ofrepresentation:

argument structurespecifyingthe type and number ofarguments ofa given lexical item;event structure(reminiscent ofaktionsart)determining whether the lexical item belongsto STATE, PROCESSor TRANSITION;qualia structure ,describing an item's predication in terms of FORMAL, CONSTITUTIVE, TELIC and AGENTIVE roles; andlexical inheritance structure (lis),relating every semantic

4For examples see Section 2.4.5.

5Qualia structure is a term introduced by Pustejovsky to cover the propositional component of word meaning.

structure to othersemantic structures anddetermining itsposition intheglobal structure of the lexicon (Pustejovsky 1996: 58,61).Qualia structuresandlisare the level of representation which hasbeentraditionally investigated insemantics, whereas argument and event structure have usually been dealt with by syntax. Pustejovsky (1996) concentrates on explicatingthe organisationof qualia structure.He stressesthat not all qualiaroleshave to be specifiedfor everylexicalitem,And he derivestheirroots fromAristotle's notionof modes of explanation. Pustejovsky(1996: 85-86)proposes thefollowing valuesfor each qualerole:

1. CONSTITUTIVE: the relation between an object and its constituents or proper parts.

i. Material ii. Weight

iii. Parts and component elements

2. FORMAL: That which distinguishes the object within a larger domain. i. Orientation ii. Magnitude iii. Shape iv. Dimensionality v. Colour vi. Position

3. TELIC: Purpose and function of the object

i. Purpose that an agent has in performing an act.

ii. Built-in function or aim which specifies certain activities.

4. AGENTIVE: Factors involved in the origin or "bringing about" of an object. i. Creator

ii. Artefact iii. Natural Kind iv. Causal Chain.

This explication can be the target ofmuch criticism. First ofall, as the very definitions of therolessuggestthroughtheuse of theword 'object' ratherthan 'lexical item', it is difficultto see that qualia roles are indeed a part ofa specificationofthe meaning of a word and not a description of its referent. Even if we assume that Pustejovsky rejects the division between the dictionary and the encyclopaedic meaning, the valueshe puts forwardhave littleto do with a folk theory of the world, andinmanycases woulddemanda significant scientific knowledge fromthelanguage user. Ofcourse, Pustejovsky admits that not all the values have to be defined for

(11)

16 Malgorzata Fabiszak - The concept of 'joy' _{Malgorzata Fabiszak The concept of 'joy'} ₁₇

eachlexical conceptualparadigm (lcp),but it remains unclear whether the values remain unspecified when they do not apply,or when they are difficultto determine due to a lack ofknowledge on the part ofthe speaker.

1.6. Cognitive semantics.

Thetheoryof cognitive semantics willbe discussed hereasithasbeendeveloped by Langacker(t983, 1991) and also by Lakoffand Johnson (1980), Lakoff(1987) and Johnson(1987). Both approaches stressthe experiential basisofhumancognition and reject objectivism. They describe meaning in terms of cognitive structures determined by interaction between a human being and the world but not identical with it. Human cognition influences human perception, hence rendering objective description ofthe 'teal' world impossible. Thus the referentiallinkoflexicalmeaning is deniedany significance.

Langackerianspacegrammar, laterrenamed cognitive grammar, stemsfrom dissatisfactionwith the existingmodels ofsemantic structure.The criterialattribute model is found inadequate,as it cannot account for the fact that there is no featureof a categoryallmembersshare,andthatthereexistmemberswhich lackcertaincriterial properties. Spacegrammarattempts toreconcile thefmdings ofcognitive psychologists (mainly Rosch) with the descriptionofhuman linguisticcapacities, which shouldbe regarded as an aspectofhuman behaviourin general.As Langackerputs it: "We can furtherobservemany capabilities figuring in linguisticperformancehave farbroader behavioural significance: memory, problemsolvingability, planningstrategies, stylistic and aestheticjudgement, and so on. Here too it seems plausible to regard linguistic phenomena as integral facets ofa broader psychological matrix" (1983, I: 99).

Space grammaris alsobased on a criticismofgenerativegrammar, as it denies the autonomyof syntaxhypothesis. Langacker believesthatit ispossible to concentrate on the phonetic shape ofan utterance and ignore the semantics and vice versa, but a naive speaker is unable to isolate the syntactic structure from its meaning and pronunciation(1983, I: 35).Moreover, lexicon and grammar are claimedto be based on conventional imagery. The imageryis groundedin theuniversalhumanperceptual experience,mainly focusing on spacerelations (hence the name: space grammar),

This assumption is supportedby research in developmental psychology. The forefather ofthe discipline, Piaget(1970), claimedthathuman cognitive development is secondaryto motordevelopment, andit is throughactiveinteraction withthe objects in the world that human conceptualisationdevelops. Although the order ofconcept development and the assumptionthat the human brain is atabularasa at birth have

been questioned and discredited, the contemporary theories retain Piaget's major claim in its weaker form: "whatever innate componentwe invoke it becomes part of ourbiological potential onlythrough the interaction withtheenvironment" (Karmiloff-Smith1992: 10)andrepresentations develop through "interaction withboththeexternal enviromnentand [thebrain's] internalenvironment."

Inhis attemptto graspthe meaningof linguistic expressions, Langackerrejects neat formalism and advocates an eclectic approach. One ofhis assumptions is that rules ofdifferent degrees ofgeneralisation and lists ofexpressions (even ifthey are regular and can be expressed by the rules) coexist in human semantic structure. He alsobelieves thatlinguistic structures canretaincomponentiality eveniftheyareusually treated as unitary entries (Langacker 1983,1: 76-98). Space grammar posits the existence ofthree basic types oflinguistic structure: semantic, phonological and symbolic,the lattercombiningthe other two.

Semantic structureis understood as being a realisation ofcognitive structure regulatedby the requirementsoflinguisticconventions. The meaning ofa linguistic expression is not determined by the situations obtaining in the real world, but by human conceptualisationand interpretation ofthese situations. The differentiation between encyclopaedic and dictionary meaning is not retained (for arguments supporting this approach see Kovacses 1989),hence the meaning ofa lexical item can be related to the entire conceptual system. This assumption cannot be borne out in practice, thus, for descriptive convenience, lexical items are defined by only a portionoflinguistically relevantconceptualstructure, delineation ofwhich is a matter ofdegree. Consequently, word meaning can be represented in relation to one or more domains constitutinga word's matrix. For example, 'knuckle' can be defined in reference to domain 1: [finger], domain 2: [process ofbending], domain 3: [part! whole relation] (Langacker 1983, II: 55).

Accordingto Lakoff and Johnson (1980),later developedby Johnson (1987), human cognitionhas experientialbases. The embodied imagination, which plays a crucialrole in ourunderstanding oftheworldis commonto allhumans,as it originates from their everyday basic experience oftheir bodies, and is a source ofparallels in the semanticmappingsofthe world. Meaning ofwords is defined in terms ofmental concepts and the structure ofthe vocabulary reflects human conceptual structure. Lakoffdescribes the latter in terms ofIdealised Cognitive Models: "each ICM is a complex structured whole, a gestalt which uses four kinds ofstructure principles: propositional structure..., image-schematic structure..., metaphoric ...[and]metonymic mappings..." (Lakoff1987: 68).He alsomakes suggestions concerning the principles oflinguistic organisation, which consists ofthe distinction between central and peripheralmembers, withbasic levelobjectsat its centre, conventional mentalimages

(12)

18 Malgorzata Fabiszak - The concept of 'joy'

Malgorzata Fabiszak - The concept of 'joy' ₁₉

and knowledge aboutthem,imageschematransformations, metonymyas appliedto mentalimageryaswell as domains ofexperience and,finally, metaphors, whichmap domains to other domains (Lakoff 1987: 110)6 .

1.7. Conclusion.

The review ofthe literature on lexical semantics suggests that lexical items should be studied in context, as their meanings reside in their paradigmatic and syntagmatic relations. Words remain parts of a system ofinterrelated items. The systemmay be viewed as a semanticfield, a semanticrepresentation of a conceptual domain, or a network ofconcepts. These structuresexhibitprototypical effects, so thattheymay be differentiated intodifferent levelsof subordination. Themostsalient terms constitute the basic level and can be regarded as the central members of a category.The less typical concepts are regarded as the peripheral members ofthe category. The meaningofa singlelexicalitem canbe describedin termsof meaning components. The recurrent components shared by the most basic terms structure the lexical field. Themembers ofthe samefieldalsosharesimilar syntactic behaviour. New itemsintroduced intothe fieldwillfollowthe syntactic behaviourofthe central members. The most salientfeatures ofmeaningdetermine metaphoric extensions of a given item. Argument structure and thematic roles ofverb arguments should be treated as integralparts oftheir meaning (they determinecollocationalrestrictions, i.e. syntagmatic relations). Themeaning of a lexical itemcanberegarded inconnection withthe entireconceptual system, thoughin a semantic analysis it ismoreplausible to restrictthe investigationto the most immediaterelations. The expressivemeaning, register, style and dialectof a word influenceits collocational restrictions and hence contribute to itsmeaning.

6An example of an analysis of the concept of 'anger' in American English willbe provided in Section 2.3.

CHAPTER TWO

APPROACHES TO EMOTIONS

2.1. Introduction.

In this chapter we shall first discuss developments in the definition of the concept of emotion. In the overview ofW. James's work the main questions, whichare stillbeing investigated withinthe psychological frameworktodaywill be indicated. These are:

- the statusofemotionsand their categorisationas psychological or social phenomena;

- the universality vs. culture-specificity ofemotion, and relatedto it: - primitivevs. secondaryemotions.

These problems will be discussed in detail in a section devoted to emotions within generalpsychology.

Section 2.4. will then deal with the treatment of the topic of emotions in linguistics. Herewe willconcentrate on lexical semantics andomitmorphopragmatic (e.g. Dressler-Barbaresi 1994) and pragmatic (e.g. Kryk-Kastovsky 1997) approaches to emotions, sincethey arebeyondthe scopeofthe presentwork. Finally, thechapterwill closewithan accountofrecentinvestigations intoemotions fromthe pointofview ofsocialhistory,

2.2. The concept of emotion: Its definition and structure.

William James (1913) addresses all the points outlined above. Many ofhis statements areveryinsightful andpresuppose the. fuzziness oftheconcept of emotion. James does not draw a strict line between emotions and other cognitiveprocesses. For example, he defines "belief' as a psychic state akin to emotion (1913: 283-285).He also concedesthat the borderlinebetween emotions,on the one hand, and whathe callsinstincts, on the other, is fuzzyas well:

(13)

20 Malgorzata Fabiszak The concept of 'joy'

In speaking ofthe instincts it was impossible to keep them separate from the emotional excitements that go with them. (...)

Instinctive reactions and emotional expressions thus shade imperceptibly into each other. Every object that excites an instinct excites an emotion as well (James 1913: 442).

withina physiological theoryof emotions:

... Our mental life is knit up with our corporeal frame, in the strictest sense oftheterm, Rapture, love ambition, indignation and pride, considered as feelings, are fruits of the same soil with the grossest bodily sensations of pleasure and of pain (James 1913: 467).

21

James classifies romantic love, parental love, jealousy, modesty, shame, emulation or rivalry, pugnacity, anger,resentment, sympathy and fear as instincts. For him, fear and anger, for instance, are instincts ofselfpreservation in the material sense, and he places them in a single categorywith hunting, the acquisitive, home-constructing andtool-home-constructing instincts (James 1913: 308).Inaddition, he stresses thatnot only is it pointless totryto establish sharpdistinctions betweenemotions and other psychological processes, but also"the internalshadingsof emotionalfeelings ... merge endlessly into each other" (James: 1913:448).

Perhaps those emotionswhich are alsotermed instinctsshouldbe viewed, in presentdayterminology, asbeingbothprimary/primitive anduniversalemotions, as the following passageon the "instinctof modesty"indicates:

... Everywhere reserve must inspire some respect, and ... Persons who suffer every liberty are persons whom others disregard. Not to be like such people, then, would be one of the first resolutions, suggested by social self-consciousness to a child of nature just emerging from the unreflective state (James 1913: 437) [emphasis mine-MF].

James himselfuses the term 'primitive emotion' and instantiates it with self-complacency, self-satisfaction, rage and pain (James 1913:307).

A certaincontradiction arises when James stresses thatthe onlypossiblevalid classification of emotions is one based on an analysis oftheircausesratherthan their expression, while at the same time he also states that a determinant ofa primitive emotionis its physiognomical, or as we would say today, facial, expression.

Thisbringsus to the defmitionof emotion, Accordingto James,emotions can be excitedby objects, in a general, ontological, sense of the word,which arethyfoci of instinctive(bothinnateand acquired) interestfor a given speciessuch as the prey of the speciesor itsfood, enemies, sexualmates andyoung. The triggers of emotions can be either related to those four basic groups or to preservation and self-satisfactionin a wider sense.

Finally, James attempts to define emotions (inhisownterms "coarser emotions")

The'coarseremotions' are distinguishedfrom the' subtleremotions', which he definesas "the moral,intellectual and aesthetic feelings" (James 1913: 468), and whichdo notfmdanybodilyexpression, but are"genuinely cerebral forms ofpleasure and displeasure" (James 1913: 468).Withthismentionofthe moralaspectof certain emotions, he introduces the problem,often addressedin present day emotionology, of the moralbases ofemotionalevaluationofsituations.

Despitetheinsistence on calling his definition "physiological", Jamespointsto the factthat emotions arebothbodilyand mentalin theirnature. Otherpsychologists, such as Wundt (1912/1924) and Solomon (1977), regard emotions as basically mentalphenomena. Some,notably Watson(1919),consideremotionas behaviour, or,eventually, as a strictly physiological activity(Wenger 1950).

These atomisticapproacheswere soon discreditedas not allowinga holistic approachto emotional experience, and led to all inclusive defmitions suchas Izard's (1972: 51): "... emotion is a complex process that has neurophysiological,motor-expressive and phenomenological aspects." Unfortunately this is too general and allowsoneto labelallhuman experience: 'emotion'.

On the otherhand, empiricallyorientedpsychologists, disillusioned with the lack ofprogress in establishing an acceptable definition, resorted to operational definitions, equating emotions withgalvanic skinresponse (GSR) andsimilar concepts, which are,however, of no use in lexicalsemantics.

Those who persisted in a conceptual analysis of the nature of emotions discovered thatemotions canbe studiedon threebipolarscales: pleasure-displeasure, dominance-submissiveness and the degree of arousal (Kurcz 1976; Russel and Mehrebian 1977). Thesewriters claimthat the three dimensions can be regardedas "necessary and sufficient features to describe a large variety of emotional states" (Russel and Mehrebian 1977: 291). They also believe that the three scales are independent ofone another. This is disputed by Krzeszowski (1997: 180ft) who stresses that the most primmyand salient scale is the positive-negative (good-bad)

(14)

In the first study, friendliness and social affection... both contained small, negative arousal components, whereas in the second study, "friendly" and "affectionate"... both contained high arousal components.

Withthe discovery of prototypicality effects (Rosch 1973aandb, 1975, 1978), psychological research into the nature of emotionsbegan to utilise the new theory, Fehr and Russel (1984), for example,arguedfor a prototype-baseddefmitionof the concept ofemotion,In a series ofpsychological experiments they showed that the

23 Malgorzata Fabiszak - The concent

...t1~e·conceptof emotion [appears] at the topmost, or superordinate, level. At the middle level might appear such familiar types of emotion as anger, fear, love happiness, sadness and the like ... [and] less prototypical emotions as pride, envy, courage, lust. Each middle level category might be further subdivided at a subordinate level: anger subdivided into wrath, annoyance, rage fury, and indignation; fear subdivided into apprehension, panic dread, and alarm; love divided into filial love, romantic love ... (FehrandRusseI1984: 467).

concepthas fuzzyboundaries and exhibitsprototypicalstructure:

They further arguethat"...membership in the concept ofemotion is a matter ofdegree ... and no sharp boundary separates members from non-members" (Fehr and Russel 1984: 464). Hence, as with all gradable concepts, the lexical field of emotionmay not be analysedexhaustively, sincelanguageusers will disagreeon the membershipof non-prototypical examples.

FehrandRusselalsoclaimthattheideaofprimary/basic/fundamental emotions whichrecursin much of the psychological literature may originate in the existence of prototypical emotions at the basic level. Still, despite their significantsaliency in a . given social community, they are culture specific. Furthermore, they may vary in

degree ofbasicness/prototypicalityand the group membership.

According to Fehr and Russel (1984), and Bullock and Russel (1986), the categorisation and labelling of novel emotions as representative ofa given type is performed on the basis of their resemblance to idealised concepts (what Abelson 1981, calls scripts; Lakoff 1987, Idealised Cognitive Models; Langacker 1983, abstractdomains or,in his earlierworks functionalassemblies; and othersvariously referto as frame, scenario, schemata). Theseidealised concepts consist ofprototypical causes,appraisals of situations, physiological reactions, feelings, facialexpressions, actions, and consequences (Bullock and Russel 1986: 210). The source ofthese conceptsis the folktheoryof emotionsadheredto by the membersof a givenspeech community. Naturally, thereisindividual variation inwhatstimulus triggers anemotion and what emotionaldisplaythe bearer ofthe emotionwill employ,

Emotional display, aswellastheconceptualisation andcategorisation ofemotion, likeanysocially shapedphenomenon, is culture-specific (Geertz 1959; Lofland1985; Lutz 1988; Heelas 1996; Omondi 1997). Moreover, some researchers question even the universalityof the emotion concept as such, since there are languagesthat do not have a cover term for the field, or to put it in a different way, they do not recognisemental statesas separatefi..om physiological reactionsor socialbehaviour (Lutz 1988 on Ifaluk; Omondi 1997 on Dhoulo). Specific terms for emotions are

Russel andMehrebiansolvethe problemby concludingthat friendliness is an emotionfor which the degreeofarousalis not perceivedas a determinantandhence can occur with both high and low degree of arousal.

Surprisingly, a similar discrepancy appeared withthepleasure-displeasure scale. RusselandMehrebian, however, decidedto dismiss "aggression", the problemterm, as an emotionterm,ratherthan questionthe validityof the scale. Takinginto account that they analysed 151 terms,this exclusionof the problematical item seemsa sound decision.

scaleandthat all othersare derivedfromit.He pointsoutthat dominant, high arousal and pleasure can be generalised as positive; submissiveness, low arousal and displeasure asnegative. He convincingly argues fortheprimacy ofthepositive-negative scale throughout his book (cf. Inchaurralde 1997, who also supports this view). Thesetwo conflicting stands canbe reconciled withinthe framework ofLangackerian cognitive grannnar. In developing histheoryof conceptual domains, Langacker (1983, I: 78) claimsthat although somestructures areinnateand serveas inputintocognitive operations, the results of these operations i.e. the secondary structures can in time become internalised to an extent that they are no longer perceived as consisting of functions on input structures, but as non-divisible entities, thusbecoming"primary" secondarystructures. (This view is also supportedby neuropsychological research. Luria (1966) reports that new experiences reshape the human brain and nervous system.)In this way the three scalesused forthe descriptionof emotionscanbe seen as mutually uncorrelated despite being developmentally derived from the basic axiological scale.

The three dimensions can be used to structurethe semantic field of emotion, althoughthey must be appliedwith caution.Apparently,the degree ofarousaldoes not have to be inherent in an emotion, but can change, depending on contextual factors, as indicated by the discrepancy in the results of the two studies conducted by Russel and Mehrebian (1977: 290):

(15)

24 Malgorzata Fabiszak - The concept of 'joy' Malgorzata Fabiszak - The concept of 'j oy' 25

embedded in a language culture, and reflect the moral and ethical values of the community. (Heelas 1996;Steams and Knapp 1996;Krzeszowski 1997;Omondi 1997).

However, linguists working withinthe fieldtheoryof semantics combined with componential analysis have suggestedthat the generalterm for a fieldmay,but need not be, lexicalised in a language. For example, Coseriu (1967) pointed out that an archilexeme is not always realised lexically, and Lyons (1977), when devising an abstract model of a hierarchically organised vocabulary, left the beginning node unlabelledto indicatethat: "the vocabulary, and indeedanyparticularpart ofit,may be structuredhierarchically from a pointwhich itselfis not associated with an actual lexeme" (Lyons 1977: 297). The lack of a label for a lexical field, then, does not necessarily deny its existence,as the unity ofthe field is determinedby its internal dimensions, a notionusedbothinlexematics (Geckeler 1988) andcognitive psychology (Russel and Mehrebian 1977),which organise its structure and its semes/features which, in tum, accountfor the non scalar content ofwords.

2.3. Emotionsingeneral psychology.

The 1980s and 1990s have witnessed a new surge ofinterest in emotions. Some ofthe main issues addressed by the researchers follow the tradition going back to James, although the emphasis has oftenbeen shiftedto lesswell-investigated problems. For example, the question about the existence ofuniversal emotions is now often rephrased as a question concerning what exactly universal means, and what degrees or types of universalitycan be expected ofemotions.

Secondly, withthe developmentofcognitivesciences, anotherpopular issue has become the question ofthe degree, to which emotions are mental, and hence partly consciousphenomena,as opposed to purely instinctivereactions developed by the species to speed up reaction times and avoid the time-consuming mental processing ofstimuli.

The finalissuementioned in this sectionwillbe the development ofemotional behaviourin children. Thisdiscussion willaim at determining to what degreeone can talk ofthe acquisition of a sociallyshaped system as opposedto the developmentof an innate system.Ananswerto this questionmay shed some light on the othertwo.

2.3.1. Universality vs, culture-specificity of emotions.

One of the questions psychologists have been trying to answer is whether basic(universal, primitive) emotions exist. Theissueraises muchdebate. Averill (1994a

andb) claims thatthe question itselfdoesnot havemuchvalidity, as it is not emotions themselves thatmightbebasicbuttheconcepts thataidourunderstanding ofemotional states. He points out that ifwe take prototypical emotions as the basic ones, their basicness shouldonlybe understood as the basisfor identifying new membersofthe category and not endowing these specific emotions with any fundamental import (thus anger, fear and grief should not be regarded as of more weight for the understanding of humanemotional systemthanpride,hope and guilt). He alsonotes that concepts which may be termed basic at a more general level ofcategorisation (e.g. anger as a label for all aggressivebehaviour) cease to display this feature at a more specific level (anger as opposed to anxiety, rage, ire,furyand exasperation). The use of the same label at two different levels ofclassificationcauses confusion when the issue ofbasicness is addressed, hence it is beneficial to dispose of the notionof basicnessaltogether. The lastargument Averill puts forward is basedon the factthat emotions canbe viewedas complexphenomenaconsisting of severallevels of description suchasbiological, socialandpsychological. Differentemotions canbe consideredbasic depending on the perspectiveone employs.This choice,however, is arbitrary anddoesnot havemuchvalidity. Toput it succinctly, Averill believes that "because an emotionalconcept is basic within some classificationscheme, it does not followthatthe corresponding emotionis somehowmore fundamental thanother emotions" (Averill 1994a: 7).Whatthisapproach emphasises isthe disparity between emotions as such and the human construction of emotions.Itfollows from this observationthat hypotheses about conceptualisation and the linguisticlabellingof emotions may not hold for the descriptions ofemotion per se undertaken from, for example,the physiological perspective. Averill (1994c)restrictsthe role of biology to providinghumanswith a capacityfor emotions, but the realisation or instantiation of emotion is culture-driven. He draws a striking parallel between language and emotion:

The capacity for language is a biological adaptation; but any specific language (English, say, or Chinese) is the product ofsocial, not biological evolution. Similarly, emotions presume biological capacities, but any specific emotion (anger, say, or

liget) is the product of social, not biological evolution (Averill1994c: 269).

Scherer (1994a) emphasises thefactthattheverbal labelling ofemotions reflects only the consciouspart ofthe emotion system and may focus on one ofits aspects, for example, the evaluation of the situation, expression, or actiontendencies. He also notesthatverballabelling playsan important communicative andevaluative function in socialinteraction.

(16)

Shweder (1994)avoidsgivinga straightforward answertothe questionof the existence of basicemotions, stressing thatpresentresearchis not sufficient to address theissue. Heviews emotions ascomplex narrative structures whichmakeouremotional experience meaningful andhelpindividuals to understandtheirpositionintheworld. He suggests thatdifferent cultures may sharecertain universal components ofemotion scripts, suchastherelation betweentheegoanditsenvironment, withsubcomponents suchas statuslossand status gain, success and failure, protection andthreat, blocking ofthegoal, andnovelty; bodilyreactions, forinstance fatigue andagitation; behavioural strategies like attack, withdrawal, hiding,confession, repair. The ideaof emotions as narrative structures hasbeen further developed in linguistic researchby,forexample, Bamberg (1997a and b, see 2.4).

Levenson (1994) alsoaddresses theproblem oftheuniversal vs.culture-specific status of emotions. He advocates a biocultural model. According to this model emotionsarise from interactionbetween the organism and its external and internal environment. In the firststage,the stimulusis appraisedin relationto the organism's well-being, long-term plansandimmediate goals, and a prototypical emotion concept is identified. At stage two, the prototype determines response strategies. Cultural learning influencesboth the appraisalofthe situation and the choice ofexpressive behaviour.

Ekman (1994) observes that there might exist certain universal antecedent eventseliciting emotions, and suggests the following cause-emotion pairs:

An actual or threat ofhann for fear. The loss of an object to which one was attached for sadness. An event that is either unexpected or contrary to one's expectations for surprise. Something that is repulsive, to the senses or to one's beliefs, for disgust. Disapproving or feeling morally superior to someone for contempt. For anger we suggest five antecedents: frustration resulting from interference with one's activity; a physical threat; an insult; seeing someone do something that violates one's values; another person's anger directed at oneself. For happiness we suggested four antecedents: sensory pleasure; excitement; praise; reliefwhen something unpleasant has ceased (Ekman 1994: 146-147).

Ekmandoesnot say, however, thatactualinstantiations ofstimuli foremotions may differfromculture to culture, sincethe valueof objects, beliefs, morality, andthe defmition ofunpleasantness, whichallappearamongthetriggers, areculture specific. This may questionthe validityof the proposed, allegedlyuniversal, stimuli. Scherer (1994b),unlike Ekman, phrases his claim for universality in more cautious terms: "Emotion antecedentsituations arebothuniversal- with respectto many structural

characteristics - andculturally specific - due to differences in values, practices, history, interactionpatterns, demography, climate, economy, and social structure" (Scherer 1994b: 175,his italics). WhereasEllsworth(1994a)notesthat whatmight be cross-culturallyuniversal is whether the stimulus evokes a positive or negative response and not a more specific emotion. This view is supported by research in developmental psychology which suggeststhat infantsdo not differentiate between complexemotioncategories, but canonlycategorise emotions asnegative orpositive (Dunn 1994).

2.3.2. Emotions as cognitive phenomena.

Theissue ofthelinkbetween emotion andcognition hasalso metwithsignificant interest. Clore (1994) defines emotion as a mental state, so that for him the role cognition plays inemotion issubstantial bydefinition. Thisdefinition, however, excludes conceptssuch as 'abandoned' since this refersto an externalsituationratherthan to a mental state, 'tired' as referringto a bodily state,and 'faithful' as not referringto a state at all. This approach is in an apparent contrast with that of Scherer (1994a), who stresses that in the process of conceptualisation only one of the aspects of emotion, for example the evaluation ofthe situation, may be the target ofverbal labelling. Yet, as evaluationis a mental activity, Scherer's defmitioncan in fact be viewedas complementarytothatofClore, whoclaims that"goodexamples ofemotion terms do not refer directlyto events,bodily reactions,feelings, or behaviour, but to mental events in relevance to all these feelings" (1994: 184). Ortony, Clore and Collins(1988)develop a Cognitive Modelof Emotions, in whichevents, actions and objectsare appraised in termsofone's goals,standardsand attitudes and resultin an emotion. According to Clore (1994: 187),

All emotions can be seen as differentiated forms of ... [the] three general affective reactions (1)being pleased or displeased at the outcome ofevents that are appraised as desirable or undesirablefor one

s

goals, (2) approving or

disapproving ofthe actions ofagents appraised as praiseworthy or blameworthy with respect to one

s

standards, and (3)liking or disliking the attributes ofobjects appraised as appealing or unappealing with respect to one

s

tastes or attitudes

(hisitalics).

This modelunderscores the factthat emotionis a part of a largerinformation-processingsystem.

(17)

28 Malzorzata Fabiszak - The concept of 'joy' _{Malgorzata Fabiszak - The conceot of 'joy'} ₂₉

producingemotion, but allowalsofor instinct-based emotional reactions. Cognition isnecessary to assign a label to theemotion and"therealisation of anameundoubtedly changes thefeeling, simplifying andclarifying" (Ellsworth 1994b: 193).

2.3.3. The development of emotion concepts in children.

Developmental psychology hasalsocontributedto theunderstanding ofemotion. The earlyresearchers suggestedthat the expressionofemotionis innate and can be observed already in very young infants. This hypothesis has met with subsequent qualification (Drum 1994): although infants dohavequite anarray offacial expressions, they are used indiscriminately, so that all the categorisationthat infants seem to be capableofis distinguishing betweenpositiveandnegativestimuli, whereasspecific emotions appear as a result of their social development. Infants as young as 10 months recognise and interprettheir mother's facial expressionto guide their own reactions (Klinnert, Campos, Sorce, Emde and Svejda 1983; Bodor 1997). They use the technique of social referencing as a help in deciding on their behaviour (approachor withdrawal) when exposedto a new situation.Itseems,however,that all that infantscan actually inferis the negativevs. positivefactor? .

Inthe processof socialisation human infantslearnhow to organise, categorise and interpretemotions of the selfand other, andbecomean adultofa specific culture (Geertz 1959; Bullock and Russel 1986). Children between 2 and 4 years of age startto associate expressions withthecontext inwhichtheyoccurand, simultaneously, acquire the language of emotion.Gradually they impose a temporal dimension on eventsandeventually construct generalised scripts forthe categorisation of emotional behaviour(socialdimension) andemotional experience (psychological dimension).

Camras (1994)suggeststhat two aspects ofthe emotionprocess change as a resultof a child'spsychological development: theappraisal oftheantecedent situation, and expressive behaviour, both of which are culture-dependent. Only with the development of these do a child's emotions become more discrete and clearly interpretable. Thelanguage of emotion facilitates thecategorisation of emotional states and allows a child to participatein the shared cultural conceptsofemotion oftheir particularworld" (Dunn 1994: 355).

The development of emotion concepts is not finite.Itdoes not lead to an acquisition of a fixed taxonomy of emotion, but should be viewed as a constant process of conceptualisation and labelling of human experience. Yet linguistic expressions of emotionsshould not be viewed as mere labels of inner states. They

7On the basicness of the positive-negative axiological scale see also Krzeszowski 1997.

allow evaluation ofthe emotional situation and enable the bearers ofemotion to controltheir reactions and expressthem in a sociallymeaningfulway. As suchthey shouldbe understoodas performativedevicesin the Austiniansense (Bodor 1997; Bamberg 1997a).Inthis context it must be stressedthat the causes ofan emotional display must be transparent to the decoder to reinforce the correct interpretation (Bodor 1997;Oller and Wiltshire 1997).

2.4. The linguistic approach to emotions

Most linguistic research on emotions, with the notable exception ofDiller (1994), has been devoted to modem data. Wierzbicka (1988a and b, 1992, 1993, 1996)has carriedout large amount ofcross culturalresearch into emotions as well as presented a coherent system for the description of selected emotion words in English. Bamberg (1997a andb),dissatisfied withtheNatural Semantic Metalanguage approach developed by Wierzbicka, has suggested a social-constructionist approach to emotions, which,however, bears considerable similarity to her proposals. Lakoff andKovecses (1987),later includedin Lakoff (1987),have examinedthe meaning oftheconcept of'anger' inAmerican English withinthetheory ofcognitive linguistics.

The fact that emotion verbs require an Experiencer theme has attracted the attentionof generativists, who have researchedthe syntactic behaviourofwhatthey callpsych-verbs in greatdetail. From their work onlythat by Jackendoff(1990) and Grimshaw(1990)willbe presentedbelow.

2.4.1. Emotions in historical linguistics.

Diller(1994)hasmade an attemptto capture themeaningofangerandrelated words in Chaucerusing a schema developedby the cultural anthropologistWhite. Diller(1996)develops a scenariofor the consistent studyofemotionterms in which he emphasisesthe importance ofthe cause of emotion,the expression ofemotion, and the social aspect of emotion (is it individualor sharedby others?).He also tags them with positiveor negativevalue. His analysisaccentuates the consequences of variationin the distribution oftermsacrossliterarygenres.

2.4.2. Emotions within Natural Semantic Metalanguage.

Wierzbicka (1992, 1993, 1996) has also carried out research into emotion terms. She rejectsthe claimthat there arebasicemotionsand emphasises ratherthat they are culturally embedded. She suggests that eventhe emotionalinterpretation of

The concept of 'joy' in Old and Middle English. A semantic analysis