What is Poetics?Author(s): Stein Haugom OlsenReviewed work(s):Source: The Philosophical Quarterly, Vol. 26, No. 105 (Oct., 1976), pp. 338-351Published by: Wiley-Blackwell for The Philosophical QuarterlyStable URL: http://www.jstor.org/stable/2218864 .Accessed: 23/09/2012 13:38Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact email@example.com.. Wiley-Blackwell and The Philosophical Quarterly are collaborating with JSTOR to digitize, preserve and extend access to The Philosophical Quarterly.http://www.jstor.org
338 WHAT IS POETICS? BY STEIN HAUGOMOLSEN I The conceptionof poetics as an objective and systematic, indeed, or,even a "scientific"study of literature has gained wide currency amongtheorists and criticssince the last war. Primarily this has been due to the ofinfluence the "semantic"poeticswhichfollowed upon the New Criticism,but morerecently structuralist the poeticsin France has developedsimilarideas. Structuralism semanticpoeticsdo indeed draw theirideas from andverydifferent sources,but theysharea commonassumption whichmay be cancalled the axiom of objectivity.This assumption be formulated roughlyas follows: literary the workis a piece ofdiscourse text) possessing (a certaincharacteristics whichmake it what it is: a literary work. As a piece of dis-courseit is accessibleto all the speakersof the language;its qualities canbe observedand classified interested by observers, and if, in a particularcase, thereis dispute about what these qualities are, it can be settledbyreference the textitself. Thus a systematic to (thestructuralists "scien- saytific")studyof literary worksis possible;a studywhichwillultimately leadto a fullunderstanding the qualities whichmake a text into a literary ofwork. The text is accordedan object-like status. In the semantic theory itis labelled "artefact". In scientific(structuralist) poetics,textsare "pheno-mena" to be studiedscientifically. The axiom of objectivity manifests itselfdifferently the two different intheories.The pointofdeparture the scientific for poeticsis that all discourseis structured.This structuring goes beyondthe rule-governed combinationof wordsinto sentences. Largerpieces of discoursecan also be seen to bestructured.In particular,the sequences of sentencesor texts which we asrecognize literary workshave characteristic structural properties.Theseproperties analogousto the structural are properties of sentences, and they aconstitute higher-order languagewithits own unitsand grammar.Poeticsidentifies unitsof this languageand describes rules of combination. the theIt becomesin this way a "linguistics" this higher-order of language. For example, describing the structure a plot, it is possible to use ofcategories analogousto those of noun,adjective and verb used in linguisticdescription (the episodedescribed fromThe Decameron): is Prenonsun exemple qui nous permettra dillustrer "parties du ces discours"narratif.Peronnelle re9oitson amanten labsencedu mari, pauvre ma9on. Mais un jour celui-cirentre bonneheure. Peron- de nelle cache lamant dans un tonneau;le mari une foisentre,elle lui dit que quelquun voulait acheterle tonneau et que ce quelquun
WHAT IS POETICS? 339 est maintenant trainde lexaminer. Le marila croitet se rejouit en de la vente. I1 va racier le tonneau pour le nettoyer; pendant ce temps,lamant fait lamour a Peronnellequi a passe sa tete et ses bras dans louverture tonneauet la ainsi bouche (VII, 2). du Peronnelle, lamant et le mari sont les agents de cette histoire. Tous les trois sont des noms propresnarratifs, bien que les deux derniers soientpas nommes;nous pouvonsles designer ne par X, Y et Z. Les mots de lamant et du mari nous indiquentde plus un certainetat (cest la legalitede la relationavec Peronnelle est ici qui en cause); ils fonctionnent donc commedes adjectifs.Ces adjectifs decriventlequilibreinitial: Peronnelleest 1epouse du ma9on, elle na pas le droitde fairelamour avec dautreshommes.1 In the furtheranalysisthe theorist identifies "verbs" in the plot which twostructure the action; and he can conclude "Ainsi lanalyse du recit nouspermetdisolerdes unitesformelles presentent analogiesfrappantes qui desavec les partiesdu discours:nompropre, verbe,adjectif".2The formal unitsin questionhereare found thetext. Theirpresence a givenon whichthe in istheorist can build. Higher-order like structures these are not always foundon the surfaceof texts. They may have to be reconstructed fromthe datawhichthe text offers. The data, however, givenin the same way as the aredata forotherscientific theoriesare given. The theoryis "empirical". The centralinsightwhichthe semantictheorydevelopsin some detailis based on a distinction betweenthe primary and secondarymeaningsofwordsand sentences. The notionof primary meaningis taken forgrantedin thetheory.It is the way in whichone wouldnormally understand word. aA wordmayhave severalprimary meanings whichmakeit useful different intypesof context. It also has secondary meanings.These are the associationswhichit evokes; what it suggests a receiver virtueof its connections to byto certaintypesofobjects,events, or situations, linguistic frames.Secondarymeaningis also called "connotation"or "impliedmeaning",and it is heldto be one of the semanticproperties a term. Like primary of meaningitattaches to the word and will be discoverableby a competent speakerofthe language. In ordinarylanguage such primarymeaningsand connotations a ofwordas may lead to misunderstanding are irrelevant the messageare or tosuppressed. Ordinary languageis "transparent": is used to attain a goal it isand attention concentrated thisgoal and neveron the linguistic on means.Literary discoursediffersfrom ordinary languageby making ofsecondary usemeaning. Primarymeaningswhich are not requiredby the contextandconnotations allowed to come into play and enrichthe meaningof the arewords and sentencesemployed. Ambiguities and paradoxes both of singletermsand of wholephrasesare used in literature give languagewhat has to 1Tzvetan Todorov, "La grammaire du recit" (1968), in Tzvetan Todorov, Poetiquede la prose (Paris, 1971), p. 122. 2Ibid.
340 STEIN HAUGOM OLSENbeen called "semanticdensity". To bringout the secondary meanings a ofpassage one uses the techniquecalled "explication".Here is an exampleofhow a theorist deals withsomelinesfrom Macbeth: Besides,thisDuncan Hath bornehis faculties meek,hath been so So clearin his greatoffice,that his virtues Will plead like angelstrumpet-tongud against The deep damnation his takingoff; of And pity,like a naked new-born babe, Striding blast,or heavens cherubim, the horsd Upon the sightless of couriers the air, Shall blowthe horrid deed in everyeye, That tearsshall drownthe wind. Pity is like the naked babe, the most sensitiveand helplessthing; yet, almost as soon as the comparison announced,the symbolof is weakness begins to turn into a symbol of strength; the babe, for thoughnewborn, pictured "Striding blast" like an elemental is as the force-like "heavens cherubim".... The finaland climacticap- pearanceofthe babe symbolmerges the contradictory all elements of the symbol.For, withMacduffs statement about his birth, naked the babe risesbefore Macbethas not onlythe future that eludes calcula- tion,but as avenging angel as well.3Here "pity" is described ambiguous, as the involving ideas of bothhelpless-ness and power. The ambiguity builton the secondary is meanings con- ornotations whichBrookstakes to be the properties the expressions of nakednew-born babe, striding horsd upon the the blast, heavens cherubim,sightless couriers the air. Thus we have a classiccase ofsemantic of density,ambiguity giving riseto ambiguity. Scientificpoetics accepts structuralpatterns as given; the semantictheory takes the givenfactsto be secondary meanings phrasesand words. ofIn both cases it is a question of an unarguedassumptionthat these areobjectiveproperties texts. Whentheymeetwithobjections, response of theof these theories is to producearguments the "look-and-see"type,i.e., ofmoreand moreempirical data are collectedwhichwill show that literaturedoes have secondary meaningsor poetic structures.The basic questionofthe appropriateness and adequacy of this type of "look-and-see"response areis not discussedat all. It is simplyaccepted that these features thereto be observed. II Scientificpoetics and the more puristversionsof the semantictheory asdo not look upon themselves just anotherliterary theory. They do notsimply attempt to answer the traditionalproblemsof poetics; they alsowant to definea disciplinewhichhas its own methodsand poses its own 3Cleanth Brooks, The Well WroughtUrn (New York, 1947), p. 45.
WHAT IS POETICS? 341 questions;theythus necessarily limitthe theorists concerns prescribing byboth what are relevantproblemsand the appropriate methodsforsolvingthem. Naturally,to accept the axiom of objectivitywill have serious con-sequencesforthe conception poeticswhichemerges.If literary of discourseis distinguished from other types of discourseeither by local semanticfeatures by structural or properties whichare objectivelyobservable, thenthe task of the theorist will be to describeand classify these features andtheirinternalrelationships.The structuralist be engagedin mapping willsemiological structures onto literary works. And semantic will theorists tryto show in what ways language can be semantically dense. However,themostimportant consequence the axiom of objectivity of concerns what thetheorist could notbe doingif he wantedto keep within literary theory. Hecould not concern himself withthe relationship betweenliterary worksandthe world,betweenliterary worksand theirauthorsor literary worksandtheirreaders. If he did, he would then be taking a step outside literarytheoryand into psychology, sociology, history ideas, etc.). A poetics or (ofbased on the axiom of objectivity will have no tools fordealingwiththesetypesofrelationships, indeed,willit recognize nor, any questionsconcerningthemas fallingwithinpoetics. This consequence bound to bringpoetics isintotrouble. For whileit is true that a literary workis a type of whichthe tokensare physicalobjects (the copiesof the text) and whichcan therefore some inrespects be characterized an object,it is impossible ignore factthat as to thethe conceptof a literary workalso involvesassumptions about value. Togive something the title"literary work"is to place it within regionof our aconcerns whichhas an established claim on our timeand attention.It is tosee it as attempting yield a certainpay-off to whichis conventionally ex-pected of texts classed in this way. The notionof pay-off can, of course,only be understood termsof a relationship in betweenliterary worksandthe aims and purposesof a group of people. Literatureas an institutionof culture founded the perception is on that it is valuable. The achievementof thisvalue is dependent upon therebeingperformances be appreciated; toand performances createdby men and judged by them. It is on this areinterplay betweencreation and judgement that the conceptof literary valuedepends. To cut off literary the workfrom readersand authorsis to cut itofffrom any meaningful function can fulfil.To treatit as an objectrather itthan as a performance of may help to gain objectivity description but itmeans losing the chance of characterizing aestheticdimensionof the thework. Sever the connection betweenliterature and value and you makethe conceptof a "literary work"uninteresting. One must suspect,then,that the axiom of objectivity radicallyin- isadequate as a basis fora conception poetics. I shalltryin thenextsection ofto showthat this is the case.
342 STEIN HAUGOM OLSEN III If "the actual objectsof poeticsare the particular regularities that occur in literary texts and that determine the specific of effects poetry",4 then therearises a problemof relevance. The question"Given a literary work, which features patterns relevant thestatusofthistextas a literary and are to work?" an requires answer before objectiveanalysiscan getoff ground. the the Some features relevantto this status and othersare not. How are we are to choose a description whichincludesonly the relevantfeatures?If one acceptsthe axiom of objectivity answeris that no decisionis necessary the because the structures given. One does actuallyknowwhat "particular are regularities" give poetic effects and these are the intuitions poetics must build on: A meaningful investigation that attempts describethis systemPS to [poeticstructure] mustthus growout of the effects and judgements that come about through the maximallyadequate understanding of a poetictext.5 The questionof how one arrivesat a "maximallyadequate understanding"of a poetic text, and what constitutes such an understanding, not taken is to be a questionof poetics. And, furthermore, is regardedas a question it whichwill have no bearingon the possible conclusions a poetic theory. of Interpretation is givena place in literary activity butit is seenas theoreticallyinnocuous. It is given the status roughlyof pre-theoretical observations whichmay be givendirection once theory introduced whichwill still is butremainobservations. There are in actual fact wide disagreements about the correctinter-pretation of mostliterary works. This alone is reasonenoughto rejecttheassumptionthat poetic structuresare given throughshared intuitions.Interpretative judgements are based on arguments and can be challengedthrough arguments.If an investigation into poeticstructure to be based ison interpretative judgements, thenthe questionof how suchjudgements canbe evaluatedmustbe brought theforefront literary to of theory, thusshifting toits whole focus fromobjectivelygiven structures the process of inter-pretation. Otherwise the theorymust resignitselfto producing arbitraryresults. Perhaps,to get out of the cornerinto whichhe has boxed himself byinsisting that he relieson interpretative judgements, scientific the theoristwill now make the following move: thoughinterpretative judgements con-cerning the same workare widelydifferent, may say, it remainsa fact hethat all these interpretative judgements rest on sets of structural patternsand regularities whichare independent any one interpretation which of andcan be described by the theorist. ". . . une description scientifique doit 4Manfred Bierwisch, "Poetics and Linguistics", in Donald C. Freeman (ed.), Linguis-tics and LiteraryStyle (New York, 1970), pp. 98-9. Originallypublished as "Poetik undLinguistik", in Helmut Kreuzer and Rul Gunzenhauser (edd.), Mathematikund Dich-tung (Munich, 1965, 1967). 5Bierwisch,p. 109.
WHAT IS POETICS? 343pouvoir rendrecompte de toutes les lecturescoh6rentes possibles. Sanspour autant enoncerexplicitement chaque lecture, les elle d6finit conditionsde chacune."6 A scientific poeticsis not basedon interpretative judgementsof different degreesof competence exceptin the sensethat it triesto explaintheir source. The theorist"can and must explicate those consciously orunconsciously followed regularitiesthat lead to the understanding poetic ofstructure to a judgement poeticality".7 and of Theseregularities, theorist thewill claim, can be describedin a consistent, coherentand comprehensivevocabularywhichclearlyidentifies properties the structural the of patternsand regularities. And it is the task of "scientific" poetics to develop suchdescriptions. In fact, this move leaves the scientific theoryno betteroffthan if it onwereto base itself interpretative judgements.For whilean interpretationcannotcreatethe factson whichit is based, it does pick out whichfactsareartistically relevant and makeclearin whatwaytheyare relevant creating bya description thesefacts. Thus the artistically of relevant structuralproper-ties describedin an interpretation become the productof that particularinterpretation. There is no given structure facts in a work which are ofartistically relevant except as they are given throughan interpretation.This recreative aspect ofthe literaryactivitycan be observedat everystageof interpretation it is possible to illustrateit by choosingalmost at andrandomfromthe vast criticalcorpuswhichhas arisenin connection withevery branchof Westernliterature.Note, forexample,how Auden in thefollowing argument, whichI quote in extenso, identifies patternof facts afromOthello connecting by themin a description and how the descriptioncreatesthe pattern a from set ofpreviously unconnected "facts"in the work. In Othello, thanksto Iagos manipulations, Cassio and Desdemona behave in a way whichwould make it not altogether unreasonable forOthelloto suspectthat theywerein love witheach other, but the timefactor of rulesout the possibility adultery havingbeen actually committed.Some criticshave taken the double timein the play to be merelya dramaturgical device forspeedingthe action whichthe audience in the theatrewill never notice. I believe,however,that Shakespearemeant the audienceto noticeit as, in The Merchant of Venice,he meant them to notice the discrepancy betweenBelmont timeand Venicetime. for If Othellohad simplybeen jealous of the feelings Cassio he imagined Desdemona to have, he would have been sane enough, guilty worstofa lack oftrustin his wife. But Othellois not merely at jealous of feelings which mightexist; he demands proofof an act whichcould not have taken place, and the effect him of believing on in this physicalimpossibility goes far beyondwishing kill her: it to is not onlyhis wifewho has betrayed himbut the wholeuniverse; life has becomemeaningless, occupationis gone. his 6Fran9ois Rastier, "Syst6matique des isotopies", in A. J. Greimas (ed.), Essais desemiotiquepoetique (Paris, 1972), p. 96. 7Manfred Bierwisch, op. cit., p. 108.
344 STEIN HAUGOM OLSEN This reactionmightbe expectedif Othelloand Desdemona were a pair like Romeo and Julietor Antonyand Cleopatrawhose love was an all-absorbing Tristan-Isolde kind of passion,but Shakespeare takes care to inform that it was not. us When Othelloasks leave to take Desdemonawithhimto Cyprus, the he stresses spiritual element his love. of I therefore it not beg To please the palate of my appetite; Nor to complywithheat,the youngaffects In me defunct, propersatisfaction; and But to be freeand bounteousofhermind. Though the imageryin whichhe expresseshis jealousy is sexual- what other kind of imagerycould he use--Othellos marriageis important him less as a sexual relationship to than as a symbolof beingloved and acceptedin the Venetiancommunity. The monster in his own mind too hideous to be shownis the fear he has so far repressed that he is only valued forhis social usefulness the City. to But forhis occupationhe would be treatedas a black barbarian.8In the first part of this argument Auden remarks upon the fact that criticshave simplynot consideredthe double time-scheme Othello have a in toproper as artisticfunction distinguished froma merelydramaturgical one.He thengoes on to place the doubletime-scheme together withother featuresof the play, showing how thesefactstogether forma pattern. At the sametime he rejects certaindescriptions whichwould not allow the patterntobe formed (Othellois not "simplyjealous", he is not simply guilty "a lack ofof trustin his wife"); and he puts forward otherdescriptions whichmakethe double time-scheme instrumental bringing the particular in out natureof Othellosjealousy. In the last part of the argument supports first he hisdescription, whichassigneda place to the doubletime-scheme a within largerpattern, showing by how further factsin the play fallintothe same pattern.Againhe achieveshisendby offering certain and descriptions rejecting others.The quoted passage is described havingthe function makingclear the as ofspiritual natureof Othelloslove. The description Othelloslove as sexual ofis rejected spiteofhisuse ofsexualimagery describe and it is described in to it,insteadas beinga craving trust. All the incidents for and linguisticfeaturesreferred by Auden are to be foundin Othello, they are identified to but as anconstituting artistically relevantstructuralpatternonly throughthisinterpretative description. The movement all interpretative in judgement like this: fromhypo- isthesisto the evidenceor patternwhichsupports hypothesis.Such facts the aas do not form pattern in whichcan be described thehypothesis simply areignored. Different interpretations may, and do, identify differentpatternsin the same text, though,of course, there will be overlappingbetweeninterpretations this respect. But overlapping in does not mean that the 8W. H. Auden, "The Joker in the Pack", in W. H. Auden, The Dyers Hand andOtherEssays (London, 1963), pp. 265-6.
WHAT IS POETICS? 345patterns questioncould be identified artistic in as structures independently of any interpretation. This argument also worksthe otherway round: whenthe criticidentifies patternand assignsto it no description a whichties itto someinterpretative hypothesis about the work, thenhe has not describeda poetic or an artistic structure, simplyan arbitrarily but chosenone. Theonly way he can prove that he has singledout an artisticstructure by isbringing together the collectionof facts he wants to identify a pattern asundera description whichwilltie it to an interpretative hypothesis concern-ing the artistic purpose of the workin question. To reinforce argument the logical primacyof interpretation the for it awill be illuminating consider case of structural to description whichis notgovernedby an interpretative hypothesis and whichis therefore criticallyuninteresting, givingno insightwhatever into the artisticproperties the ofworkwhichit describes. The following passage is the firstparagraphin alinguistic analysis ofthreeprosepassagesfrom threedifferent novelswrittenby JohnBraine,Dylan Thomas,and AngusWilson: (a) Nominalgroups. In DT [Dylan Thomas],all 49 nominalgroups have lexicalitemas head: thereare no pronouns othergrammatical or heads. Oftheseonly11 have any lexicalmodification qualification, or and of a total of 5 lexical modifiers only "empty" has the value "epithet" in the groupstructure.By contrast JB [JohnBraine], in whichhas 36 nominalgroupsof which4 have grammatical heads, of the remaining withlexical heads 16 have modifier qualifier 32 or (or both) and 22 have deictics. Likewisein AW [AngusWilson],with 37 nominalgroupsof which9 have grammatical heads, 12 of the 28 with lexical heads are lexically modified qualifiedand 15 have or deictics. The DT passage is a heap of mainlysimplenominalgroups (that is, ones consisting a noun only),with also some heapingof of clauses; in AW and JB we have the compound nominalgroupas the centreof attention.All thisis obvious;but the factthat it is obvious does not excuse us fromstatingit accurately. Nor is it usefulto countitemsor patterns a without linguistic analysisto identify what is to be counted.9In additionto this,two otherdescriptive categories employed, first are theconcerning "lexical sets" and the second"cohesion". The difficulty this description that it gives reallyno clue to the with iscriticalpurposeof the comparison.It makes no judgementas to whetheror not the identified patterns mightpossiblyhave an artistic function.Thereasonforthisis that one cannotpinpoint artistic function through any sortof tabulation. Linguisticfacts and patternsby themselves,as they aredescribed here,are just linguistic, not artistic and facts. Thereis no clue inthis collectionof facts to what competent judgementcan be made whichwill make us see them as part of an artisticpatternor about what poeticeffecttheycan justifiably said to contribute This is whythe descrip- be to. 9M. A. K. Halliday, "Descriptive Linguistics in Literary Studies", in G. I. Duthie(ed.), English Studies Today (Edinburgh, 1964). Reprinted in Donald C. Freeman (ed.),op. cit., pp. 64-5.
346 STEIN HAUGOM OLSENtion is critically A uninteresting. description like this must necessarily begeneratedand limitedby linguistic interests theyare defined as withinthe academicdiscipline linguistics, thereis no reasonto believethat this of and type of description to should enable the theorist pick out artistically func-tionalpatterns. Now, the above description givenby a well-known is linguistician withafull command of a coherentdescriptivevocabulary made to serve in ageneral descriptionof language; and who furthermore sticks closely topatterns identifiable withthe tools oflinguistics who does not introduce andcovert interpretative judgementsin order to make his description seemrelevant the contextofliterary in studies. This is whythe above descriptionso glaringly shows the absurdityof the claim that artisticstructures aregiven in a piece of literary discoursebecause such discourseis structureddifferently othertypesof language. And in all fairness Halliday one from tomust say that he does not make any extravagantclaims forlinguistics inliterary studies. However,when literary theorists the scientific of schooldescribepoetic structures, theyveryoftenimplythat this description on isa par with the type of description given by Halliday, while they reallyintroducecovertinterpretative judgementsto make the description criti-cally interesting. Consideragain the description given of the storyfromThe Decameron quoted above in the first section. This description of a ismoresophisticated kindthan that givenby Halliday. Beforethe structuraldescription even startedTodorov has alreadyservedup to the readera isselectivesummary the story, of the interpretative implications whichhe ofcontinues develop throughout so-calledstructural to the description.Thisinterpretative movemakesthedescription interesting for literary the student,but his interest disappointed. For the initialinterpretative is move is notcarried beyondthe summary description the passage and no justification offorthe interpretative description is offered. the same timeit is claimed Atthat this is a description formalunitsfound in the plot. This double ofmove of initiating interpretation thenclaiming it the statusof a an and forstraight description createsthe appearance of objectivitywhile seeminglyretaining claimto critical the interest the description. factit achieves for Inneither criticalinterest objectivity. nor It is clear that the same problem of arisesforthe description secondarymeaning as for structural patterns. Granting, for the sake of argument,the assumptionthat secondarymeaningsactually belong to a term or asentenceand are not generatedby the context,there still must be someway of tellingwhichsecondary meanings relevantin the context. One arecannot just go ahead and describeall the secondarymeaningsof all theexpressions a piece of discourseand expect to have identified poetic in itsfeatures,for among all the possible secondarymeaningswhich may bedescribed only some will be relevantin the particular contextin whichtheexpression occurs. To see this,it is instructive compareCleanthBrooks to
WHAT IS POETICS? 347 "description" the passage fromMacbeth of (quoted above) witha "descrip-tion" providedby Helen Gardner. Gardnerfirst takes issue with Brooks assignment of the connotations powerful and avenging angel to the word cherubim. These, she says, are just Brooks personalassociations. If thetermcherubim any connotations all, it is thoseofpassive,contemplat- has ating, beautiful, innocent.This new set of secondarymeaningsrequires adifferent description the passage from of that whichBrooksgives it: The final image ofthewinddropping therainbeginsis thetermina- as tion of the whole sequence of ideas and images. It is to this close that theyhurry.The passage ends withtears stilling blast. The the finalcondemnation the deed is not that the doer of it will meet of withpunishment, even that the doer of it will stand condemned; not but that even indignation the murderwill be swallowedup in at universal pityfor victim.The wholeworld the willknow,and knowing it willnot cursebut weep. The babe, naked and new-born, most the helplessof all things, the cherubim, innocent and beautiful, call out the pity and the love by whichMacbethis judged. It is not terror of heavens vengeancewhich makes him pause; but the terrorof moral isolation.10 How can one decide which set of associationsconnectedto the termcherubim the correct is one?The hollowness the "look-and-see" of argumentis apparent a case likethis,whentwocritics in offercompeting interpretationsof the same term in the same context. No attemptto call to mind theconnotations the termwill have any effect of sincethe criticsarguefortherelevanceof different connotations. This problem cannotbe solvedby refer-ence to further we factsabout the termitself. Therefore need some reasonforpreferring description one to the other. The naturalcourseis to invokethe artistic purposeof the workand tryto find out whichconnotations relevantto this purpose. Brooks may areinsistthat the connotations mentions properties the term, he are of but this ifis uninteresting he cannotoffer someargument show that they contri- tobute to the artistic natureof the work. In fact,the wholeof Brooksessayon "the nakedbabe" is an attempt use the quotedpassage as a convenient topoint "of entryinto the largersymbolswhichdominatethe play".1 BothBrooks and Gardnertry to establish their descriptionsby relating thepassage to what Gardnercalls the "imaginative centre"12of the play. ForGardner this centreis to be foundin the visionthat the murder Duncanofplaces Macbethoutsidethe feeling pity,one ofthe strongest, of profoundestand most distinctively human feelings, and thus places Macbeth outsidehumanity itself: It is the judgement the humanheartthat Macbethfearshere,and of the punishment whichthe speechforeshadows not that he will be is cut down by Macduff, that havingmurdered own humanity but his 10Helen Gardner, The Business of Criticism (London, 1959), pp. 59-60. 11Brooks,op. cit., p. 30. 12Gardner, cit., p. 62. op.
348 STEIN HAUGOM OLSEN he will enter into a world of appalling loneliness,of meaningless activity, unlovedhimself, unable to love.13 and For his part,Brooksfinds that "the babe signifies future the whichMacbeth would controland cannot control".14In theirdifferent ways these hypo- theses providea generalization the importof the cherubim of passage for the play as a whole,and thus each makes an attemptto integrate in an it overall artisticvision. Gardner supports "reading"she givesof the termcherubim the also by a reference historical, to non-linguistic facts: Dionysius the Areopagite,who established the hierarchyof the angels, sourceofthepopularangelology theMiddleAges,ranked the of the cherubimamong the higherorders,as angels of the presence. They stood about the throne,contemplating gloryof God, not the active,as werethe lowerorders, fulfil will on earth.15 to hisSince the questionof historical evidencein literary criticism receiveda has lot of attention, is worthobserving it herethat thereis no special problemconcerning relevance thistypeofinformation. the of Exactly the same typeof problemarises in connection with the relevanceof connotations any ofsort. Onlyif one has alreadyacceptedthe axiom of objectivity will one seehistorical information of a different as orderfrom that of connotations. An interpretative judgement is not the resultof a mechanicaldecodingprocedure whichcan be applied to any given structural facts. The judge-mentitselfis recreative and involvesan imaginative leap whichconsistsinidentifying it a patternby assigning artistic function.Interpretative judge-mentsare the sole means a critichas of identifying artisticfeatures the ofa literary work. This means that evaluation of interpretative judgements bemust ultimately comparative, since thereis no way of challenging aninterpretative judgement except through anothersuch judgement.A wide-spread comparative activitylike this will sanctionthe use of certainroughand readypredicates whichcan be used in a given case to givean "absolute"judgement, but thisshouldnotlead thetheorist think to thatthesepredicateshave any meaning except within framework the comparative the of activity.Evaluation of interpretative judgementsbeing ultimatelya comparativeactivity, is logically it impossible that thereshouldbe a finalinterpretationof a work. It is always possiblethat even an extremely good interpretationshould one day be challengedand replacedby a completely different one.It is therefore logicallyimpossible that thereshouldbe any set of structuralproperties secondarymeaningswhichare theproperties a work. The or ofaxiom of objectivity musttherefore rejectedand one can expectto solve beno problems substancein literary of theory acceptingstructural by featuresand secondary meanings as "given" facts. 30p. cit., p. 61. 40p. cit., p. 42. 10Op.cit., p. 56.
WHAT IS POETICS? 349 IV The ultimatebasis on whicha readersappreciation a literary of work restsis not objectively given secondarymeaningsand structural patterns. It restson a method(interpretation) assigning of artisticrelevanceto parts of a workidentified through the method. Thoughthe examplesabove have all been of one particulartype of interpretation, far I have made no so attempt to characterize method. Indeed, to describeit theoretically, the to define aims and methods(beyondsayingthat it is concerned its withartistic significance), would be an attemptto workout a particular literary theory, a task beyondmy presentconcerns. In this articleI have considered not so much the inadequacy of two theoriesof literature the inadequacy of as the commonconceptof poeticsthey advocate. In line withthis argument I shall now tryto use the notionofinterpretation establisha requirement to whichany theoryof literature must fulfil.The requirement concernsthe place in literarytheoryof the concept of intention which was dismissed from poeticsby the theories based on the axiom of objectivity. Every interpretative hypothesis may be understood an answerto a as question "What are the reasons for the presenceof . . . in the work?", wherethe blank is to be filledby a description a part of the work. In of this actual criticism question receivesdifferent formulations, theonegiven but above is adequate formypresent purpose. An answerto this question cannot give just any reason, forliterary interpretation requires that the hypothesis mustultimately defended reference artistic be by to or significance purpose. To approacha text as a literary workis to examineit forits artistic signifi- cance. This means that the interpretative questionmust be answeredbyreference something to beyondthe elementof the workidentified the bydescription in the blank. It is not enoughto describe element having the as such and such properties, it is always possibleto ask fora justification forforseeingthese as artisticproperties.In some typesof literary theory thereasonsto be given in answerto almost everyinterpretative questionaresought not only outsidethe identified element, but outsidethe workitself.Thus Dickens, defending the prefaceto Bleak House his description in ofthe Courtof Chancery, says "everything in set forth thesepages concerning isthe Courtof Chancery substantially true,and withinthe truth". Other,moresophisticated, theories requirereference the worldoutsidethe work toonly at somepoint. Aristotle goes to sometroublein the Poeticsto describewhat features plot a good tragedymusthave, and he brings reference of into the worldoutsideonlyby setting an ultimategoal of the artistic up con-struction:". .. the poets job is to producepleasure springing frompityand fearvia mimesis"(1453b 12). Theories thissophisticated of typeconstruethe answerto the interpretative questionas an attemptto assign a placewithina largerstructure, postulatedthroughan overall interpretation ofthe work,to the elementidentified throughthe question. Thus when acriticsays about the openingof Twelfth Nightthat Orsinosfamousfirst
350 STEIN HAUGOM OLSEN speech- If musicbe the foodof love, play on; Give me excessof it- together with what followsin the briefopeningscene, reveals his attitudein love as a blend of sentiment and artifice,true dedication and elaborate self-centredness; is at once an eloquent statement it a and, by implication, criticism the plays courtly of romantic theme16 he gives, first, description a part of the play ("Orsinos famousfirst a of speech . . . togetherwith what followsin the openingscene"); next the reason forits presenceby re-describing part ("reveals his attitudein this love as a blendof sentiment and artifice,truededicationand elaborateself- and centredness"); finally reasonforthe presence the features a of identified in the seconddescription giventhrough new description is a whichconnects themto the themeofthe play ("it is at once an eloquentstatement and, by a of implication, criticism the plays courtly romantic theme"). The method of answeringan interpretative question by relatingthe identified element otherpartsofthe workthrough seriesof descriptions to a of ascendingorder,subsuming moreand moreparts of the work,is widely accepted as illuminating criticism.There are, indeed,good reasonsboth of a logical and of a historical kind to thinkthat this is what constitutes an illuminating literaryexplanation. But thisfactshouldnot blindthe theorist to a further questionwhichthe criticis not bound to ask, but whichthe theorist cannotleave unanswered.Having reachedan illuminating overall interpretation a work,attributing of certainproperties it, one may still to ask "What is the reason why these properties in the work?". When arethis questionis repeated at this level, it is not a criticalquestion but a questionabout the nature of artistic significance: why do these properties, ascribed to the work by this interpretation, make the work artisticallyinteresting? The reasonwhythe answerto this questionhas to make some referenceto the worldoutsidethe workhas alreadybeen suggested.Artistic signifi-cance is a value. As a value it mustnecessarily understood be withreferenceto humanpurposesand ends. So the reasonswhichjustify the assumptionthat certainproperties a workare artistically of interesting must have todo withthe way in whichthese properties relatedto certainpurposes areand ends. This is a purelyconceptualpoint. Nothingneed be said abouthow theseends and purposesare to be specified.That is the task of literarytheory. What literary theoryis not freeto do is to disguiseor ignorethefactthat it is involvedin specifying purposesand ends. The questionwhy arecertainfeatures artistically interesting to be answered termsof the is inpurposes they can be seenas serving.Everyinterpretative question, evenifitcan be answeredby a structural description or a description "meaning", ofis in the end made significant only because it asks how an elementserves 16Derek Traversi, An Approach to Shakespeare, Vol. I: Henry VI to TwelfthNight(London, 1968), p. 303.
WHAT IS POETICS? 351 athe intendedartisticpurpose. To understand literary workis to under-stand a goal-directed effortmade by a creatingintellect. The subject ofliterary theory the natureofthe effort the goal. This conclusion is and doesnot rule out any of the traditional theoriesof literature, it does place butliterary theoryunderan obligationwhichobjectivistpoeticshas soughttoavoid. Poeticsmustdeal withthe role ofintention and assignto it a properplace in literaryunderstanding. Sincethewholenotionofliterary interpreta-tion restson the conceptsof purposeand ends, it is impossible exorcize tothe problemof intention from literary theory. Intentionhas a place in theunderstanding and appreciation the literary of work; for the problem literarytheory is to assignit the correctplace.University Bergen of