SMITH Barbara - Poetry As Fiction

Poetry as Fiction
Author(s): Barbara Herrnstein Smith

Source: New Literary History, Vol. 2, No. 2, Form and Its Alternatives (Winter, 1971), pp.
259-281
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/468602 .
Accessed: 17/06/2014 09:35
Your 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 of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to
New Literary History.
http://www.jstor.org
This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

All use subject to JSTOR Terms and Conditions
Poetryas Fiction
BarbaraHermstein
Smith
p RADOXES makeintriguing titles,but I am not otherwise

of themand intend,by the end of thisarticle,to dissolvethe
fond
onethatentitles it. I meanto do thisbyelaborating theproposi-

tionthatfictiveness
is thecharacteristicquality ofwhat we call "poetry"
whenwe use thetermin thebroadsensebequeathedby Aristotle, i.e.,
to referto thegeneralclassof verbalartworks.My primaryconcern
willbe to developa conception of poetrythatallowsus to distinguish
it fromand relateit to bothnonpoeticdiscourseand otherartforms.
The viewpresented herewas initially,butratherincidentally, proposed
elsewhere.'I have foundthe elaborationof it of continuing interest,
however,especiallysincethe groundsforthosedistinctions and the
natureofthoserelationships remain,to mymind,extremely problem-
aticincontemporary linguisticand aesthetic theory.
Sincemyprocedurein whatfollowsmayseeminitially perplexing,
someprefatory remarks maypreventconfusion.Firstof all, I shallbe
sayinga good deal aboutlanguagein generalbeforeI say anything at
all aboutpoetry.Anytheoryof poetryinevitably, thoughnot always
presupposes
explicitly, a theoryoflanguage.Thus,thosewho have at
varioustimesregardedpoetryas inspired speech,or embellished prose,
orthelanguageofpassion,or"emotive"statements, haveobviously had
somewhat differentnotionsofwhatlanguageis whenit is notpoetry-
e.g.,uninspired speech,plain prose,thelanguageof reason,or "veri-
fiablestatements." Since,moreover, linguistic theoryis nowin a very
volatilestate,no generalpropositions concerning languagecan be of-
feredcasuallyor takenforgranted.(One can oftentellmoreabouta
man's politicsand metaphysics now fromhis viewsof languagethan
one oncecouldfromhisclassor religion.)In anycase,althoughI am
by no meansoffering hereanything thatcould be called a theoryof
language,thefirst sectionofthisarticlewilldevelopsomegeneralob-
servationson nonpoetic or whatI call "natural"discourse, particularly
in thoserespectsthat are mostsignificant in distinguishing it from
I Poetic Closure: A Study of How Poems End (Chicago, 1968), esp. pp.
14-25.

260 NEW LITERARY HISTORY
poetry.The second sectionof the paper will developsome implications

of the conceptionof poetryas mimetic,or what I shall be callingfictive,
discourse.
Althoughthe making of distinctions, and classifications
definitions,
will occupy a good deal of the discussionthroughout, it should become
clear that my ultimateinterestis not in taxonomybut in poetryas an
artform. I am concernedwith how, on what basis, we actually do
identifypoetry,and how that identificationdirectsand modifiesour
experienceand interpretation of a literaryartwork,both as distinct
froma natural utteranceand as relatedto otherartforms.
Finally, I should mentionthat what is presentedhere is actually a
set of extractsfroma largerstudyin progress,and I am consciousof
the fact that many matterstouched upon in what followsdeservecon-
siderablymore attentionthan I have thespace to givethem.
By "natural discourse,"I mean here all utterances- trivialor sub-

lime, ill-wroughtor eloquent,true or false,scientificor passionate-
that can be taken as someone's saying something,somewhere,some-
time: i.e., as the verbal acts of real personson particularoccasionsin
responseto particularsets of circumstances.In stressingall thesepar-
ticularities,I wish to emphasizethat a natural utteranceis an histori-
cal event: like any otherevent,it occupies a specificand unique point
in time and space. A natural utteranceis thus an eventin the same
senseas the Coronationof Elizabeth I on January15, 1559, or the de-
parturethismorningfromAlbany of AlleghenyAirlinesFlight617, or
the fallingof a certainleaf froma certainelm tree. Other eventsmore
or less resemblingthesein various respectsmay occur at othertimesor
in otherplaces, but the eventitself-that coronation,that flight,that
utterance- cannot recur,forit is historically unique.
The point requiresemphasisbecause it reflectsa fundamentaldis-
tinctionthat may be drawn between natural utterancesand certain
otherlinguisticstructures whichare not historicaleventsand which can
be both definedand describedindependentlyof any particularinstance
of occurrence.Dictionaryentries,forexample,or what we referto ab-
stractlyas "the word fire"or "the phraselaw and order"are not them-
selvesparticularevents;theyare, rather,linguisticforms,or the names
of certaintypesor classesof events. And, as such, certainobservations
may be made about them: for example, the morphemicor phonetic

POETRY AS FICTION 26I
features thatdefineall members oftheclass,or thesyntactic rulesgov-

erning their accepted use in Englishsentences, of
or, course,thechar-
acteristic featuresofthecircumstances in whichtheydo occuras part
ofutterances - in otherwords,their"dictionary meanings."Butthese
linguistic forms - words,phrases, etc. - are not themselves historical
eventsunlessoruntiltheyoccuras theverbalresponses ofparticular per-
sonson particularoccasions.Obviously"thewordfire"as a general
classis a verydifferent sortofthingfroma specific utterance, "Fire!",
whichmaywarna manthathislifeis in dangerorsenda bulletspeed-
ingtowardhim,verymuchdependingon theparticular circumstances
inwhichtheutterance occursand to whichitis a response.
A naturalutterance not onlyoccursin a particularset of circum-
-
stances whatis often referred to as itscontext - but is also under-
stoodas beinga responseto those circumstances. In otherwords,the
historical of
"context" an utterance does not merelysurroundit but
occasionsit,brings it into existence. The context ofan utterance, then,
is bestthought or
ofnotsimplyas itsgrossexternal physicalsetting, but
ratheras the totalset of conditions that has in fact determined its
occurrence of
and form.2That totalset conditions, what makes us say
something at a particular timeand also shapesthelinguistic structure
ofourutterance-thespecific wordswe choose,oursyntax, ourintona-
tion,etc.-is likelyto be manifoldand complex no matter how simple
the utterance.Moreover,the totalset of conditions thatdetermines
whatwe say and howwe speakis byno meansconfined to theobjects
and events"spokenabout,"or whatlinguistic theorists of variousper-
suasionsreferto as "referents," "designations," "denotations," or "sig-
nifications."
It is worthnotingthattheexistence ofan objector eventor even,as
we say,an "idea," is nevera sufficient reasonforresponding to it ver-
bally. In other words, the fact that something is true is never a suf-
ficient reason for saying it. If I should be heard to say, "It's five
o'clock,"thereasonsformysayingso wouldclearlyinclude more than
2 Since the term context has been acquiring increased currencyin contemporary
aesthetics and linguistics,I should point out that it is not my intention here to
quarrel with or qualify the sense it bears for other theorists. It might have been
better to discover or devise another term altogether for what I am here defining
and later elaborating,but the alternativesthat presented themselvesseemed just as
likely to create comparable confusions,and I confess to a temperamental loathing
of neologisms. It should also be noted that, in proposing that we view the context
of an utterance not merelyas its physical settingbut as the totalityof its determin-
ants, I am not so much broadening the ordinaryreferenceof the term as affirming
the existence and significanceof a particular relation, namely causality, between a
verbal event and the universe in which it occurs. Defined in termsof that relation-
ship, the "context" of an utterance inevitably refersto somethingmore extensive
than what the common use of the term suggests,but also somethingmore particular.

what time of day it was just then,for at any momentit is a certain

time, but I do not announce the time continuouslythroughthe day.
Perhaps, on thisoccasion, I wished to remindsomeone of an appoint-
ment,or perhapssomeonehad just asked me forthe correcttime. Cer-
tainlythesecircumstanceswere as significant in occasioningmy utter-
ance as that specificone to which my words,"It's fiveo'clock," might
seem exclusivelyto "refer,"namelythe timeof day.
Given any utterablefact or state of affairs,grossor subtle,physical
or psychological- the state of the weather,the color of swans,or my
opinionof thewar - whetheror not I will actuallyutterit, and how I
will utterit, will always depend upon othervariables,i.e., attendant
circumstancesotherthan that fact or state of affairs.These variables
will include, among otherthings,the presenceof a potentiallistener,
my relationshipto him,the natureof the social occasion,the immediate
verbal context (what eitherhe or I have been saying) and, perhaps
mostsignificantly, theconventionsof the linguisticcommunityto which
we bothbelong.
There is no reason to maintain a sharp distinctionbetweenthe sort
of physicaland social variablesjust mentionedand what mightother-
wise be thoughtof as the internal,personal,mental,or psychological
springsof speech. It is obviousthat among the circumstancesthatpro-
voke, occasion, and shape an utteranceare conditionspeculiar to the
speaker's currentstate: his emotions,his feelings,his memories,ex-
pectations,beliefs,and desires. I may say "It's fiveo'clock" partlybe-
cause I am hungryor anxious or bored, and such conditionsmustalso
be recognizedas part of the contextof the utterance. We should note,
moreover,thatthespeaker's"currentstate" is inevitablythe productof
his past as well as his currentexperiences,including,mostsignificantly,
his past verbalexperiences,and thatpart of his psychologicalor mental
condition- and thereforepart of the contextof his utteranceis
how he has learnedto use language.
Althoughwe may, for certain purposes,describe an utteranceex-
clusivelyin termsof its linguisticform(e.g., as a certainconcatination
of lexemes and/or phonemes), a natural utterancecan never be ade-
quatelyspecifiedor describedas an eventexceptin relationto the con-
textin whichit occurred. In otherwords,a verbal event,like any other
event,is individuatedas much by its contextas by its form. Thus, al-
thoughwe could say thattwo men each pullingthe triggerof a gun are
engagedin acts of thesame form,it is clear that Mr. X shootingMr. Y
is not the same eventas Mr. A shootingMr. B, or as Mr. X shooting
Mr. Y again fifteen minuteslater. Similarly,when I say,makingintro-
ductions at a party,"This is my husband," it may not be a unique
eventwithrespectto its linguisticform,but it is certainlynot the same

POETRY AS FICTION 263
eventas some otherwoman sayingit of her husband or, indeed, as my

own saying it on some other occasion, eitherfifteenminuteslater to
some otherguestor even absentmindedly to the same one as before.
Moreover, it is unlikelythat any two natural utteranceswould be
even formallyidenticalif one extendsattentionto the more subtle as-
pects of theirlinguisticform. For although each utterancecould be
transcribedwiththe same symbols,such a transcription preservesonlya
fractionofthetotalphysicalrealityconstituting each utterance,a reality
that would include not only a certainsequence of phonemes,but also
intonationalfeaturessuch as pitchcontours,stress,pacing, and usually
facial expressionsand other gesturesas well. While some linguists
may regard the latteraspects of the utterancewith suspicionand dis-
pute theirstatus as linguisticfeatures,it is neverthelessbecoming in-
creasinglyevidentthat thereis no absolute discontinuity between the
part of an act or event that is called "verbal" and the totalityof that
act or event. In otherwords,a natural utteranceis always continuous
withthe speaker'stotal ongoingbehavior and also continuouswith the
totalworldof naturalevents. The professionallinguist'sor our own or-
dinary descriptionof the utterancereflectsan arbitrarydemarcation
and abstractionfromthefullness,the density,and the spatial,temporal,
and casual continuity of all human action and all eventsin nature.
Most of us would agree that it is impossibleto provide a complete
and exhaustivedescriptionof a nonverbalhistoricaleventsuch as the
Coronationof Elizabeth or the departureof Flight6 I7. What the his-
torianofferswill usuallybe a selectionor abstractionof certainfeatures
of theseeventsat a level thoughtadequate forthe purpose at hand. It
is clear, moreover,that neitheran eye-witness reportnor, if we had it,
even a videotape, would constitutea total record of the event; and
neitherone, of course, would constitutethe event itself. The same
limitsand distinctionsapply to the descriptionsand recordsof verbal
events: Elizabeth's firstspeech to Parliamenton February4th, 1559,
or my farewellsthis morningto my family. No descriptionor record
would be complete,neithera vocal quotation nor a tape-recording, in
eitherof which manyfeaturesof the originaleventwould be lost. The
fact,however,thatverbal eventscan be transcribedin a standardnota-
tional systemoftenseems to obscure for us their similarityto other
events. It is true that orthographyand phoneticnotationallows us to
record or describe natural utteranceswith considerablesubtletyand
specificityof detail through conventionalizedsymbols. Moreover, a
transcription of this kind - i.e., a "text" of the utterance- may be
an adequate descriptionor recordof it formostpurposes. Nevertheless,
we should not confusea copy of that textwith the verbal eventitself,
the historicalact of a particularspeakeron a particularoccasion.

The relation of utterances to texts is of special interestto us

here since, at least in our own culture,we typicallyencounterpoetry
as texts. The relation is extremelycomplex, however,with respect
to both natural and poetic discourse,and, indeed, it is not always
the same relation. I have just been speaking of texts that serve
as recordsor descriptionsof naturalutterances,i.e., inscriptions of ver-
bal eventsthat occurredat some specifictime,such as Elizabeth's first
addressto Parliament. Not all textsbear thisrelationto some natural
utterance. Many texts- personal letters,for example - are not
recordsor descriptionsof utterances,but constituteutterancesthem-
selves,onlyin writtenratherthan vocal form. It is true,of course,that
thereare otherverysignificantaspects to the relationbetweenwriting
and vocal speech, and they are not mutuallyindependentor simply
parallel possibilities.Nevertheless,to the extentthat the writer'sact of
composing and inscribing is an historicallyspecificand unique verbal
event, it is analagous to the speaker's act of emittingthe sounds that
comprisespoken discourse. And thus we may regard the product of
eitheract as a naturalutterance.
In view of the Gutenbergrevolution,the question may arise as to
whetherprinted (or otherwiseduplicated) textscan also be regarded
as natural utterances,and the answerhere is sometimesyes and some-
timesno. A printedtextmay be simplyone of many copies of an in-
scribedrecordof a vocal utterancethat,like Elizabeth's Address,did
occur at some specifictime and place. In this case, the text is not
a natural utterance,but the transcription of one. But a printedwork
may also be a naturalutteranceitselfin writtenform,exactlylike a per-
sonal letter- though the letter,of course, usually exists as only a
single text. It may be initiallydifficult to conceive of a printedwork
as a natural utteranceand thus,by our definitionhere, an historically
unique event. We should recognizehowever,thatno matterhow many
duplicationsof a text are subsequentlyproduced, the writer'sactual
compositionof the linguisticstructurethat constitutesthat text was
and remains an historicallyunique event. ("Unique" here does not
mean unitary,and it is understoodthatthe compositionof the textwill
often consistof numerous "acts" dispersedin time, fromthe initial
jottingsto theultimaterevisions.)
To summarizethesepoints,then: whetheror not a compositionwas
writtento be printed,and no matterhow long it is, or how long it took
to write,and no matterhow remotein timeor space thewriterfromhis
ultimateaudience, or how eloquent its style,or how culturallysignifi-
cant and otherwiseestimableit is, the compositionmuststillbe regarded
as a naturalutteranceso long as it may be taken as the verbal responses
of an historicallyreal person,occasioned and determinedby an his-

toricallyreal universe. And this means that most of what we call

"literature"in the generalsense of inscribedcompositionsdoes in fact
consistof natural utterances.This would include worksrangingfrom
Aristotle'sMetaphysicsand Macaulay's HistoryofEngland to an article
in a scientific
journal or an editorialin thismorning'sNew York Times.
These are all as much naturalutterancesas the remarksexchangedbe-
tweenme and a colleaguea fewmomentsago.
There remains,however,one other class of textsthat are neither
natural utterancesin writtenformnor the transcription of natural ut-
terancesthat originallyoccurredin vocal form,and this class consists
of the textsof fictiveutterances,includingmostprominently thosecom-
positionsthatwe otherwisereferto as worksof imaginativeliterature-
poems,tales,dramas,and novels. I shall reservecommenton thesetexts
untillater,in connectionwiththe generaldiscussionof fictivediscourse;
for,as we shall see, fictiveutterancesbear an altogetherdistinctiverela-
tion to theirown textswhen indeed (as is not always the case) such
textsexist.
But we may returnnow fromthe textsto the contextsof natural ut-
terances,and therebyto the crucial questionof meaning and interpre-
tation. A naturalutterancecannotbe exclusivelyidentifiedor described
independentof its context,nor can its meaning be understoodinde-
pendentofthatcontext. Indeed, what we oftenmean by the "meaning"
of an utteranceis its context,i.e., the set of conditionsthat occasioned
itsoccurrenceand determineditsform. The view of meaningproposed
hereis not offeredas an analysisof all the numeroussensesin whichthe
termhas been or could be used, and certainlynot as a solutionto the
ever-proliferating numberof problemsassociated with it in contempo-
rary linguisticsand philosophy. Nevertheless,a casual conceptionof
meaning - which thisis - has much to recommendit, particularly
here, since it permitsus to appreciate betterthe distinctivenature of
poetic discourse and of its "interpretation."Moreover, it is not so
idiosyncraticas may firstappear, for "meaning" in the sense of causes
or determinants will oftenbe found to accommodate or correspondto
familiarusage of theterm.
I must emphasize that I am speaking here of the meaning not of
wordsbut of utterances,a distinctionnot always graspedeven by those
most concernedwiththeseproblems. One may ascertainthe meanings
of those abstractclasses called words by determiningthe conventions
governingtheirusage in the relevantlinguisticcommunity,usuallyby
consultingone's experienceof the language or, when difficulties arise,
eithera dictionary-or an analyticphilosopher. Dictionariesand phi-
losophersare of onlylimitedhelp, however,in ascertainingthe meaning
of particularverbal events. When we speak ordinarilyof the meaning

of a particularutterance-i.e., what someonehas said-we are usually

concerned not with the definitionsof the words that compose it or
even, in a restrictedsense, with what it "refers"to, but ratherwith
why it occurred: the situationand motivesthat produced it, the set
of conditions,"external" and "internal,"physicaland psychological,
that caused the speaker to utterthat statementat that time in that
form-in otherwords,what we are calling here its context.
For example, definitionsand referentsare not what interestJohn
when he asks, "What do you mean?", in responseto his friend'sre-
mark,"You know,I thinkBill is a fool." Pointingto Bill and offering
an analysisof the "concept" of follywill probablynot answerhis ques-
tion. Knowing this,his friendis more likelyto describe certain cir-
cumstances,observationshe has made, impressionshe has had (and
perhaps also his motivesfor articulatingthem at that moment), and
so forth,untilJohnsays,"Oh, well,now I understandwhat you mean,"
meaningthat he has located to his own satisfactionthe reasonsforor
causes of his friend'sremark. The qualificationhere,"to his own satis-
faction,"is an importantone, forit is mostunlikelythat Johnwould in
facthave identified all thedeterminants involved.
We rarely "understand completely"one another's utterances,nor
do we need or seek to do so. Criteriaforthe adequate understanding
of an utterancevarywidely,dependingon the nature of the utterance
and theprimarypurposesand interests of thespeakerand listener.And
although sometimes - for example, in a psychoanalyst's office- one
may probe for increasinglysubtle and obscure determinants,both
speaker and listenerare usuallysatisfiedwith considerablyless than a
total identificationof all of them. It is usually not necessary,and of
courseit is usuallynot possible,forthe listenerto ascertainall the con-
ditionsthat make up the contextof an utterance. It is not necessary
because many of them will be trivialand irrelevantto his concerns.
And it is not possible eitherbecause the speaker's original contextis
remotein time or space, or because many of the springsof speech are
not apparentfromthe immediatecontextor, as we say, are privateor
internalto the speaker. The listeneror audience, therefore, is always
obligedto "interpret"what is said or written.That is, to theextentthat
the listenerhas an interestin thoseunavailable determinants, he must
hypothesize, imagine,or inferthem.
When we read the inscribedutteranceof a friend,such as a letter
fromhim, we may be more aware of interpreting as such than when
we listento him speak,but we do so in both instancesand by the same
process: partlythroughinferences based on what we know of him per-
sonally,but mostlythroughinferencesbased on all our own prior ex-
periences,especiallyour priorexperienceswith language. And, when

he alludes eitherin speech or writingto mattersof which we have no
specificknowledge--e.g., a third person whom we have never met,
a place we have nevervisited- we supplyour ignoranceby an imagi-
nativeprojectionof what we do know generally.It is importantto em-
phasize, however,that these projectionsare attemptsto inferor ap-
proximate actual circumstances,and thus are subject to correction
should our knowledgebecome more specific. ("Oh, you're Charlie's
brother.From what he said, I picturedyou as much older.")
What makes a letterparticularlyinterestingas an utteranceis the
factthat,sinceit lacks the supplementaryinformationusuallyconveyed
to the listenerby intonationand gesturesas well as by shared physical
contexts(we cannot point to thingsin letters),thissortof information
will commonlybe suppliedby thewriterin otherways: by explicitallu-
sions ("As I writethis,I am sittingby my studywindow-you know,
the one that looks out over the back garden, etc.") by graphic substi-
tutesfor intonation(e.g., underlining,punctuation,spacing), and by
moresubtlemodifications of thelanguage itself(e.g., in diction,syntax,
turnsof phrasing,and metaphor). Our syntaxin letters,because it
carriesa greaterburden of informationthan in conversationalspeech,
not onlycan be but mustbe more controlled.To be sure,since we are
oftenmore or less conscious of the generic relation of our lettersto
"literature,"we will employformssuch as archaismsand metaphoric
imagery that would seem pretentiousor otherwiseinappropriatein
conversationalspeech. This, however,does not altogetheraccount for
thefactthatsome of us become,in our letters,ratheruncharacteristical-
ly eloquent and "literary";for,as we shall see, thereare otherreasons
why the linguisticfeaturesof lettersoftenbear an interestingresem-
blance to thosecommonlyassociatedwithpoeticdiscourse,to which,at
last, I am now happy to turn.
II
Poems are not natural utterances,not historicallyunique verbal acts

or events; indeed a poem is not an event at all, and cannot be said
ever to have "occurred" in the usual sense. When we read the textof
a poem or hear it read aloud, our responseto it as a linguisticstructure
is governedby quite special conventions,and it is the understanding
that these conventionsare operatingthat distinguishesthe poem as a
verbal artworkfromnaturaldiscourse. The operationof theseconven-
tionsis mostreadilyapparentin dramaticpoetry,i.e., plays,whereit is

268 NEWLITERARY
HISTORY
understoodthat the acts and eventsperformedupon the stage are not
happening but are being representedas happening. When we see a
productionof Hamlet, we do not watch a queen drinkingpoison,but
the enactmentof such an event,which may be said to "occur" only
in beingthusenacted. But among the acts and eventsrepresentedupon
the stage are also verbal ones. As the actor who portraysClaudius
leans forwardand extendshis arm in a gestureof horrorand abortive
warning,thus representing a man leaning forwardand extendinghis
arm, etc., that actor also uttersthe words, "Gertrude,do not drink,"
thus representing a man utteringthosewords. We are not aware here
of any radical discontinuity betweenthe enactmentof a physicalaction
and the enactmentof an utterance- and of course an utterancezs
a physical action, though it has other characteristicsthat sometimes
obscurethatfact.
Most of us would be quite willingto grant the existenceof what
could be called mimeticdiscourse- i.e., the fictiverepresentation of
speech - at least in dramatic poetry. What I would like to suggest,
however,is that all poetrymay be so regarded,that we could conceive
of as mimeticdiscoursenot onlythe representation of speechin drama,
but also lyrics,epics,tales, and novels. Indeed, I wish to proposethat
this,the fictiverepresentation of discourse,is preciselywhat definesthat
class of verbal compositionswe have so much troublenamingand dis-
tinguishing, i.e., "imaginativeliterature"or "poetryin thebroad sense."
The conceptionof poetryas mimeticis, of course,quite ancient,and
modem theoristsdo continueto assertthat literatureis a representa-
tional art. It is by no means clear,however,what or what kindof thing
it is thatthe poem "imitates"or represents.One commonnotionseems
to be thatpoetry,apparentlyon the analogy of painting,somehowrep-
resents"images in words." Or, in view of the existenceof numerous
image-lesspoems and passagesin novels,thatit represents ideas or feel-
ings, eitherthe author's or those of his characters. Or, view of how
in
restrictiveeven thisformulation is, it is sometimes suggestedthatliterary
works,especiallynarrativefictions,representimagined eventsor even
worlds- in, it will solemnlybe added, themediumof language. I will
not attempthere to indicate all the problemsentailedby such sugges-
tions,3forI wish onlyto point out that theyall ignorewhat mightbe
thoughtmost apparent,namelythat what poems do represent"in the
medium of language" is language, or more accurately,speech, human
utterance,discourse. The definitionproposedhere attemptsto close in
on poetryfromtwo directions: one, as it may be distinguishedfrom
3 I have considered the matter elsewhere: see "The New Imagism," Midway
(Winter, 1969), pp. 27-44.

othermimeticartforms, and two, as it may be distinguishedfromother

verbal compositions.As a mimeticartform,what a poem distinctively
and characteristically representsis not images, ideas, feelings,charac-
or
ters,scenes, worlds, but discourse. Poetrydoes, like drama, represent
actionsand events,but exclusivelyverbal ones. And, as a verbal com-
position,a poem is distinctively and characteristicallynot a natural
utterance, but the representationof one.
A poem represents discoursein thesame senseas a play,in itstotality,
representshuman actions and events,or a painting representsvisual
objects. When we speak of the objects representedin or by a painting,
it is understoodthattheyneed not correspondto any particularobjects,
but ratherto an identifiableclass of them. A paintingcan depict a
landscape that exists as a visual object only in the depiction itself.
Thus, when we speak of mimesisor representation in an artwork,we
recognizethat it does not constitutethe imitationor reproductionof
existingobjects or events,but ratherthe fabricationof fictiveobjects
and eventsof which thereare existingor possibleinstancesor types-
whetherthey be rural landscapes, star-crossedlovers,or laments for
dead friends. In otherwords, to say that an artisthas representeda
certainobject or eventis to say that he has constructeda fictivemem-
ber of an identifiableclass of natural (real) objectsor events.
Part of what has obscuredthe relationof poetic mimesisto pictorial
and other kinds of artisticrepresentationare traditionalnotionsthat
identifythe various artformsin terms of their characteristicmedia.
Thus, sound is said to be the medium of music, pigmentthe medium
of painting,and of course words or language the medium of poetry.
The corollaryformula-X (artwork) representsY (object of imita-
tion) in Z (medium)--has created more problemsthan it has illumi-
nated, most conspicuously,perhaps, in regard to music, where art
theorists, underthe presumedobligationto locate the object that music
imitates, have come up with an amazing assortmentof chimeras,from
shapes feelingto statesofbeing. It is anotherproblem,however,that
of
concernsus here. The plasticmaterialsthat are presumablythe media
of the visual arts- pigment,stone,metal,and so forth- do not have
an expressivefunctionindependentof the artworksinto which theyare
fashioned. These materials,moreover,do not in themselvesresemble
the objects and scenesthattheyrepresent.A block of marble is a very
different thingfroma human figure. The correspondingmedium of
poetry, however, language, is not a "raw" material,but itselfa sym-
bolic systemwith expressivefunctionsindependentof its use in art-
works. For thisreason,it has been difficult to conceiveof language as
boththe mediumof an artworkand also what is represented by it.
The difficulty is
here,however, really the traditionalconcept of the

art mediumitself,particularlyits implicitdualism of formand matter.

This dualism- i.e., the notionofthe artmediumas formlessmatter-
not only createsproblemswith regard to poetry (for language is ob-
viouslynot formlessmatter), but it also obscuresthe nature of other
artforms.We could just as readilyand, I think,more fruitfully, think
of the medium of the visual arts not as pigmentand stone but as the
visuallyperceivedpropertiesof matteror, indeed, as the elementsand
dynamicsof visual perceptionitself. And, if we must have a corres-
ponding "medium" for poetry,we would do betterto locate it not
simplyin words or language conceived abstractly,but in the whole
dynamiccomplexof verbal behavior and verbal experience.
But if we are contentto do withoutthe traditionalnotionof the art
medium altogether,we may be betterable to appreciatethe essential
nature of poetic representation and its relationto artisticmimesisgen-
erally. As I suggestedabove, we may conceiveof an artworknot as the
imitation,in some different "matter,"of the "form" of particularob-
jects or eventsalreadyexistingin nature,but as the creationof a fictive
memberof a certainclass of natural objectsor events. Thus, paintings
are fictiveinstancesof what, in nature,are visuallyperceivedobjects.
Musical compositionsare fictiveinstances of acoustically perceived
events,in otherwords designed sounds as distinguishedfromsounds
simply occurringin nature. And poems are fictiveutterances. The
kindsof natural eventsrepresentedin poetryare, of course,quite spe-
cial: utterancesare themselveshuman constructions, and in that sense
"artificial." This should not, however,obscurethe sense in which they
are neverthelessnatural events,like the flightof birds,the fallingof
leaves, and all the particularactions of individualmen moving about
in, and being moved about by, the naturaluniverse.
We can, I think,readily conceive of a-man-walkingas a natural
eventand should be able to conceiveof a-man-talkingas such; for,as
I have alreadysuggested,thereis no real discontinuity betweenverbal
and non-verbalactions. A painting can represent,througha visual
configuration of line and color,a man walkingor a child sleeping,be-
cause such eventsare ordinarilyperceivedprimarilyas visual events.
And although a visual artistcan also representa man talking (one
may think,forexample,of some of Daumier's printsof lawyersin ani-
mated conversation),he cannot representpictoriallythe utteranceit-
self,for speech is not perceived as a visual event-except of course,
when it is in writtenform,a matterto which I will returnlater. But
fornow let us pursuethe exampleof Daumier a bit further.As a visual
artist,he was of course extraordinarily sensitiveto the expressiveand
otherwiseinterestingqualities of the appearances of his fellow crea-
tures: the way theystood and grouped themselvestogether,the "ex-

POETRY AS FICTION 27I
pressions"on theirfaces,the gesturesof theirhands, and so forth.Had

he also been, as some people are, extraordinarily sensitiveto the expres-
sive and otherwiseinteresting qualities of the speech of his fellowcrea-
tures,he mighthave soughtto representthat too. But how could he do
so? The answerI am suggestinghere is that he could fashiona fictive
representationof speech, i.e., a poem - something,perhaps, like
Browning's"The Bishop Orders his Tomb," which I thinkwe might
recognizeas a verbalcounterpartof a satiricDaumier print: ut pictura
poesis.
The relationof "dramaticmonologues"to dramaticpoetryproperis,
of course,readilyappreciated,and we can see how eithercould be re-
garded as mimeticdiscourse. My claim here,however,is moregeneral,
forwhat is centralto the concept of the poem as a fictiveutteranceis
not thatthe speakeris a "character" distinctfromthe poet, or that the
audience purportedlyaddressed,the emotionsexpressed,and the events
alluded to are fictional,but that the speaking,addressing,expressing
and alluding are themselvesfictiveverbal acts. To be sure, a fictive
utterancewill oftenresemblea possiblenatural utteranceveryclosely,
for the distinctionis not primarilyone of linguisticform. Moreover,
although certain formalfeatures- verse,most notably- oftendo
mark and indeed identifyforthe readerthe fictiveness of an utterance,
the presenceof such featuresare not themselvesthe crux of the dis-
tinction.The distinctionlies, rather,in a set of conventionsshared by
poet and reader,accordingto which certainidentifiablelinguisticstruc-
turesare takento be not the verbal acts theyresemble,but representa-
tionsof such acts. By thisconvention,Keats's ode "To Autumn" and
Shakespeare'ssonnetsare preciselyas fictiveas "The Bishop Orders His
Tomb" or Tennyson's "Ulysses." All of these poems are understood
not as the inscriptions of utterancesactuallyutteredby men who spoke
poetically, but rather as linguisticstructurescomposed by men whom
we call poets because they compose such structures.The statements
in a poem may, of course, resemblequite closelystatementsthat the
poet mighthave trulyand truthfully utteredas an historicalcreaturein
the historical world. Nevertheless,insofar as they are offeredand
recognized as statements in a poem, theyare fictive.To the objection,
"But I know Wordsworth meant what he saysin that poem," we must
reply, "You mean he would have meant themif he had said them,but
he is not saying them." As I shall explain later,we may choose to re-
gard the composition not as a poem but as an historicalutterance,but
thenthe conventionsby virtueof whichitsfictiveness is understoodand
has its appropriateeffectsare no longerin operation.
Anothermattershould, however,be clarifiedat this point. I have
said that novels and tales, as well as lyrics,epics, and dramaticpoems

are also fictiverepresentations of discourse. The fictivenessof prose

fictionis, of course, commonlyacknowledged,but it is more radical
than is sometimessupposed. For not onlyare the charactersand events
narratedin a novel fictional,and not onlyis the narratorwhose voice
relatesthe eventsfictional,but most significantly, so also is the entire
structureof discoursethroughwhich the narrationis presented. In-
deed, as we all know, many novels such as War and Peace allude to
quite real personsand events,a considerationthat has createdtheoreti-
cal problemsfor many literarytheorists.The essentialfictivenessof
novels,however,as of all literaryartworks,is not to be discoveredin
the unrealityof the characters,objects,and eventsalluded to, but in the
unrealityof the alludingsthemselves.In otherwords,in a novel or tale,
it is the act of reportingevents,the act of describingpersonsand refer-
ringto places,thatis fictive.The novel represents theverbal action of a
man reporting, and
describing, referring.
Considerthefollowingtwo passages:
(a) "He was a gentlemanof good familyin Buckinghamshire, and
bornto a fairfortune, and of a mostciviland affabledeportment.In his
entranceintotheworld,he indulgedhimselfall the licensein sportsand
exercisesand companywhichwas used by men of themostjollyconversa-
tion; afterwards he retiredto a more reservedand melancholysociety."
(b) "He had been a memberof the Court of Justice,and died at the
age of forty-five. His fatherhad been an officialwho, afterservingin
variousministries and departments in Petersburg,had made the sortof
careerwhichbringsmen to positionsfromwhichby reasonof longservice
theycannotbe dismissed."
The firstis from the descriptionof John Hampden in Clarendon's
Historyof the Rebellion; the second is fromTolstoi's "Death of Ivan
Ilyitch." (In both, we mightnote, allusionsare made to real places,
Buckinghamshireand Petersburg.) I am suggestinghere that the re-
lationbetweenthe two passagesis thatthe second is a representation of
the kind of thingthe firstreallyis, namely a biography. "The Death
of Ivan Ilyitch"is not thebiographyof a fictionalcharacter,but rather
a fictivebiography. The fictionattachesno more to the narratedfacts
of Ilyitch'slifethan to the factof someone'snarratingthem. Tolstoi is,
if you like, pretendingto be writinga biographywhile actuallyfabri-
catingone.
If we considerliteraturefromthepointofview I am developinghere,
it becomesevidentthatthevariousgenresofliteraryart- forexample,
tales, classicalodes, and lyrics- can oftenbe distinguished fromeach
otheraccordingto what typesof naturaldiscoursetheyrepresent: here,
respectively, anecdotal reportsof past events,public speeches,and more

or less private or personal utterance.4 Poetry itself,as distinctnow

fromnovelsand stories,traditionally representsvariouskindsof spoken
discourse. Certain typesof discourse,however,are themselvestypically
textualinscriptions;i.e., theyexistcharacteristically in writtenand often
in printedform- for example, chronicles,journals,letters,memoirs,
and biographies. And certaingenresof literaryart, roughlywhat we
referto as "prose fiction,"characteristically representsuch varietiesof
inscribeddiscourse.Novels,forexample,a distinctively post-Gutenberg
genre,have typicallybeen representations of chronicles,journals,letters,
memoirs,and biographies. This aspect of prose fictionhas some in-
terestingimplicationsforthe natureof novelsas texts,but theywill be
betterappreciatedafterwe have given some attentionto literarytexts
generally.
A poem - i.e., a fictiveutterance- consistsentirelyof a linguistic
structure,unlikea naturalutterance,which consistsof a linguisticevent
occurringin an historicalcontext.In a non-literate culture,e.g., among
NorthwestIndian tribes,thelinguisticstructurethatwould be identified
as thatsong or storyis preservedand duplicated,if at all, onlyin being
rememberedand recited. But in a literateculture,the identityof the
poem may be preservedand reproducedthrougha standardnotational
system,i.e., in a writtentext. The text of a poem, however,bears a
quite special relationto the utteranceof which it is presumablyan in-
scribed counterpart.For it is neithera transcriptionof an utterance
that actuallyoccurredat some specificpriortime,like Elizabeth's first
speech to Parliament,nor is it a natural utterancein writtenform,like
a personalletter. It is, rather,likethescoreof a musicalcompositionor
the scriptof a play, i.e., formalspecificationsforthe physicalproduc-
tion of certainevents. The text of the poem tells us, in otherwords,
how to produce the verbal act it represents.This is evidentenough for
a playscript,which directsthe performer'sverbal actions along with
other more obviouslyphysicalactions: e.g., "enter," "exit," "is stab-
bed," "falls,"says "I am dead, Horatio; wretchedQueen, adieu." But
thisis trueof any poetic text,i.e. of any verbal artworkthat represents
spoken ratherthan writtendiscourse. The textof a novel mustbe re-
garded somewhatdifferently, as I will explainbelow. But, allowingfor
thisexception, the text of any poem is to be interpreted, in the firstin-
stance,as, in effect,a score or stage directionsfor the performanceof a
purelyverbal act that exists in
only being thus performed.A poem is
neverspoken, not even by the poet himself. It is always re-cited;for
whatever its relationto words the poet could have spoken,it has, as a
4 "Private or personal utterance" may be extended to include not only overt

but interiorspeech. The representationof the latter, particularlyin romantic and
modern lyrics,is discussed in Poetic Closure, pp. 139-50.

poem, no initial historicaloccurrence. What the poet composes as a

textis not a verbal act but rathera linguisticstructurethat becomes,
throughbeing read or recited,the representation of a verbal act.
As I pointedout above, worksof prose fictionare characteristically
representations not of spokenbut of inscribedutterances,and forthis
reason the textsof novels are, interestingly enough, closer to pictures
than to musical scores. What the text of Richardson'sClarissa repre-
sents is not the speech of certain charactersbut a collectionof their
letters;what David Copperfieldrepresentsis not the spoken reminis-
censes of a man, but his autobiography. Each novel itself,i.e., the
marksprintedon its pages and, if you like,the pages themselves,plus
coversand binding,is a depictionof- a fictiveinstanceof- a kindof
book. Indeed, in view of itsthree-dimensionality, the copy of the novel
we hold in our hands could be conceived of as a sculpture,where the
sculptorhas not satisfiedhimselfin representing the grossphysicaland
visual qualities of a book, but has soughtto representthe verytext of
one. But, ratherthan complicatematters,we may at least agree that
what the textof a novel represents is, precisely,a text.
In what follows,I shall be speakingagain of poetryin the narrower
sense,i.e., as representations ofspokendiscourse,usuallyin verse. Some
of the pointsI shall be makingwould requirea somewhatdifferent or
additional formulationwith respect to novels or representationsof
writtendiscourse,but I will not have the space hereto develop them.
Althougha poem, unlike a natural utterance,consistsentirelyof a
linguisticstructure, we obviouslydo not respondto poemsas pure forms
or merelyas organizationsof sound,any morethan we respondto plays
as purelyformalstructuresof movementor to traditionalpaintingsas
pure configurations of line and color. For each of theseis understood
to be a representationalartform,and the spectatorreadily infersa
meaningor context-though a fictionalone - forthe objects,actions,
and events represented. The curtain rises on Hamlet, and we
see a human figureblowing his fingersand stamping his feet on a
dimlylit stage. Beforea word is uttered,we have already inferredat
least a cold nightas the contextforhis speech. We read or hear recited
a sonnetby Shakespeare: "To me, fairfriend,you never can be old
...," and no matterhow littlewe know about William Shakespeareof
Stratfordand the various earls withwhom he may have been intimate,
we immediatelybegin to create for those words a plausible and
appropriate context: at the minimum, a speaker addressingsome
other person whom he regards as fair and, in some sense, as his
friend. All our experienceswith language and the contextsin which
men speak not only enable us to make this inferencebut reallyoblige
us to make it.

Throughoutour livesas verbal creatures,we have learnedto respond

to linguisticstructures in a certainway: namelyto interpret theirmean-
ings, i.e., to infertheir contexts from theirforms. The effectsof poetry
as a representational artformdepend upon the strengthof our habitual
tendencyto infercontextsfromverbal structures.We should note that
Milton,in Paradise Lost, does not createEve or Eden; what he creates,
rather,are statementsabout "Eve" and "Eden" that lead the reader
to create a woman and a place - in order,as it were,to providerefer-
ents for those statements. Other representationalartformsdepend
for their effectsupon comparable tendenciesin the spectator: illu-
sionist painting, for example, depends upon fundamentalhabits of
visual perceptionto transforma configuration of lines and colorson a
flatsurfaceinto the appearance of a three-dimensional scene or object.
It is only because of perceptualconditioningproduced by our exper-
iences in the natural visual world that we can see, as a cow grazing
in the distance,what is actuallyonlya fewbrushstrokes of color on the
upper part of a canvas. This of
process interpretive filling-inor per-
ceptual inference is very similar to the processby which we infer,from
a fewlinesin a poem, a richcontextof motives,feelings,and situations.
"To me, fair friend,you never can be old . .. ." Nine small words
that summonup forus a man, his consciousnessof the pathos of muta-
bility,and his impulseto denyitshold upon his friend.
Thus, although a poem is a fictiveutterancewithout a real and
particularhistoricalcontext,its characteristic effectis to createits own
contextor, more accurately,to inviteand enable the readerto create a
plausible contextforit. And what we mean when we speak of inter-
pretinga poem is, in large measure,preciselythis processof inference,
conjecture,and indeed creation of contexts.5 But these contexts-
i.e., "meanings" - that we half create and half perceive can be no
more than "plausible," forthe poem is a fictiveutteranceand its con-
textscan be neitherdiscoverednor verifiedin natureor history.As we
saw earlier,when we interpreta naturalutterance,we seekto ascertain
its real historicaldeterminants, the contextthat did in fact occasion its
occurrenceand form. However complex and elusivethat context,it is
neverthelesshistoricallydeterminateand particular. The contextof a
fictiveutterance,however,is historicallyindeterminate.This is not to
5 I should emphasize that I am not specificallyreferringhere to those formally

and publicly articulated "interpretations"of poetrythat we associate with academic
or professionalcriticism,but ratherto the informaland oftenenough private activi-
ties of the reader as such, or what we might otherwisespeak of as his response to
or experience of the poem. Of course much formalcriticismis an extensionof these
informal activities, but the very fact that professional critics are offeringpublic
statementsentails other concerns and responsibilities,and I am not presuminghere
either to limit or to account for them. See, however,fn. 7 and p. 280, below.

saythatwe mustregardthepoemas an anonymous giftdroppedfrom

theEmpyreanor ignorethefactthatit was composedby a realman
at a particulartimeand place. It is to say,rather,thatwe mustdis-
tinguishbetween thepoet'sact of composing thepoemand theverbal
act thatthe poem represents, just as we would distinguish William
Shakespeare's act in composing Hamlet and the acts of the Princeof
Denmarkrepresented in theplay. Shakespeare the
composed play, let
us say,in 1603,but in whatyeardid HamletkillClaudius? In one
sense,he killsClaudiuseverytimethe playis performed, whetherin
1603 or 1970; but in anothersense the slayingof Claudius is an act
thatneverdid,neverwill,and nevercan occurin thehistorical world.
It can onlybe represented as occurring.The composition of theplay,
then,was an historicallydeterminate event,buttheeventsrepresented
in the play are historically
indeterminate. This means,amongother
things,that when we ask why Hamlet abuses Opheliain thenunnery-
scene,we do not expectto find theanswer in anyhistorical
particulars
of thelifeof WilliamShakespeare or of the circumstances
thatocca-
sionedhiscomposition oftheplay. Knowledgeoftheseparticulars and
circumstances of
may, course,help us account forwhyShakespeare
wrotea playin whicha characternamedHamletabusesa character
namedOphelia,butthatis an altogether differentquestion.To under-
standwhyHamletabusesOphelia,thereadermustinferfrom,on the
one hand,thelinguistic structure of theplayand, on theotherhand,
everything he knows about the world of menand therelationof their
actsto theirsituations
and motives, a plausiblesetofmotives and situa-
tionsforthatact.
Similarlyfora sonnetbyShakespeare, say87, whichbegins: "Fare-
well,thouarttoodearformypossessing,/And likeenoughthouknowst
thy estimate..." To interpretit as a poem, to understandwhy the
speakeris saying"farewell"in suchbittertonestosomeoneuponwhom
he thoughtto have some claimsof love,the readerwill not require
anyparticulars concerning Shakespeare'sprivatelife: theidentity and
moralcharacter ofwhatever youngmenhe knewat thetime,thespe-
cificincidentsofpersonalbetrayalfromwhichhe mayhavesuffered, or
hisopinionof himself as a lover. Whatthereaderdoes require is the
capacityto conceiveofthekindofsituation thatmightlead a man to
feelthus and speakthus, and thereader can developthatcapacityonly
outofhisownexperiences withmen,theirsituations, theirfeelings,and
their
especially language.
The interpretation of a poemas an historical utterance may serve
thespecialpurposes oftheliteraryhistorianorbiographer, butitis like-
lyto appearshallow,reductive, or "literal-minded" to thede-
precisely
the contextof the poem to historicalparticularsand
greethatit restricts

suggeststhat the meaningsof the poem are to be located exclusively

in an historicallydeterminatecontext. For example, a recent editor
of Shakespeare'sSonnets prefixesthe followingnote to Sonnet o07 as
part of his runningcommentaryon what was happening in Shakes-
peare's personal life at the verymomenthe was writingthe poems:
"Shakespeare had just escaped fromthe danger of his Company's in-
volvementwith the Essex rebellionand . . . the Queen, furiouswith
PembrokeforfatheringMary Fitton'schild and refusingto marryher,
had sentPembroketo jail ... ."6 Then comes the sonnet:
Not mineownfears,northepropheticsoul
Of thewideworlddreamingon thingsto come,
Can yetthelease ofmytruelovecontrol,
Supposedas forfeitto a confineddoom....
Forfeit,indeed, to a confineddoom, if interpretedas this editor sug-

gests. But forthe readerwho regardsthe sonnetas a poem, thissortof
interpretation is absurd not onlybecause its foundationsin historyare,
in fact,quite dubious,but because the invocationof particularsof this
kind - even if theywere accurate - have no greaterclaim to con-
stitutingthe "real" meaning of the poem than an interpretation that
infersfromit and providesforit an appropriatelyrich,subtle,and co-
herent context of human feelings,quite independent of Pembroke,
Mary Fitton,particularjails and particularrebellions.
I pointedout earlierthat personallettersoftenexhibitcharacteristics
thatwe associatewithpoeticdiscourse,a "literariness"thatis produced,
for example, by unusuallywell-controlledsyntax,precisionof diction,
elaboratenessor specificity of descriptions,
imagery,allusion,and meta-
phor. Since a letterwill be read in a contextboth temporallyand spa-
tiallyremotefromthatin whichit was composed,the writermustpro-
vide exclusivelythroughitslinguisticstructurethesortof supplementary
informationthat is otherwise,in a spoken utterance,providedby the
physicalcontextsharedby speakerand listenerand also by thespeaker's
intonationand gestures.The letterwriter, in otherwords,must exploit
all the expressivepossibilitiesof language itselfto enable his reader to
inferand reconstruct properlythe meaningsand contextof his original
utterance.
The poet is obviouslyoperatingunderthe same limitations, but even
more so. He mustconveyto his readersnot onlya contextremotefrom
themin space and time,but one that neverexistedin historyor nature,
and consistsentirelyof what the reader will be enabled to construct
(rather than reconstruct)fromthe verbal formof the poem. More-
6 The Sonnets, Songs, and Poems of Shakespeare, ed. Oscar James Campbell
(New York and Toronto, I964), p. 136.

over,the poet mustsuggest,throughthe possibilitiesof language alone,

not only a plausible and interestingcontextthat has no independent
reality,but also the experiences,attitudes,and feelings- and, indeed,
the identityof a speakerwho has no otherexistenceand of whom
the readerhas no otherknowledge. Finally,especiallysince the textof
the poem will functionas the scriptforits futureperformance(and by
recitersotherthan the poet), it must specifyor directits own vocal
realization,includingits pacing and otherintonationalfeatures.
The poet will, therefore,in the verbal structurehe composes,be
strainingto the limitall the expressiveresourcesof language. And, be-
yond that limit,he will sometimesdevise new ones. But what are
sometimesspoken of as "poetic devices" (and we may include here
rhythmor meter) are really the potentiallyexpressivefeatures of
natural discourse. Tropes and figures,distortionsof idiomaticsyntax,
departuresfromidiomaticdiction,imageryand allusion-all theseare
certainlynot restrictedto poetic discourse; nor can theybe taken as
the distinctivecharacteristics of poetic language. They are not what
definespoetrybut are, rather,entailedby what does defineit, namely
itsfictiveness.
Because a poem does not reflectbut create the contextin which its
meaningsare located, its linguisticstructuremust carryan extraordi-
naryburden. Poetic language seems- and indeed is - richer,more
"suggestive,""connotative,"and "evocative" than natural language
preciselybecause and to the extentthatit requiresthe readerto cooper-
ate in thecreationof itsmeanings. In our efforts to interpretthe poem,
to constructthe contextof human situationsand motivesit demands in
orderthat its meaningsbe realized,we will draw upon all our experi-
ences of theworld and wordsof men. Indeed, the activityof interpret-
ing poetryoften becomes the occasion for our recognitionand ac-
knowledgementof otherwiseinaccessiblefeelingsand in a sense, our
own otherwiseunknowableknowledge. The richerand more extensive
our experiencesand feelings- or as we say, "the morewe bringto the
poem"-the more significanceit can have for us, which is why, of
course,subsequentreadingsof a poem "reveal" more meanings. The
language of a poem seems characteristically "concentrated"because it
allows for such an extraordinaryand continuous expansivenessof
meaning,not confinedto itsown finiteand particulardeterminants, but
drawing on all we know that we can relateto it. The language of the
poem continues to mean as long as we have meanings to provide for
it. Its meaningsare exhaustedonly at the limitsof the reader's own
experienceand imagination.
But now I should addressmyselfto the suspicionthat the view of in-
terpretationdeveloped here leads one directlyinto the camp of the

subjectivists:thosewho, scorningthe revelations of literaryhistory

and scholarship, wouldmaintainthatall and anymeaningsof a poem
are essentially"personal"and equallyvalid. This is not,however, the
In
case. speaking of thecontexts createdor projectedbythereader,I
have repeatedly usedthetermplausible;and althoughI have seemed
to be saying"If themeaningfits, wearit,"I havealso impliedthatthe
meaningmustfit.Thisfitness and plausibility relateto verysignificant
constraints on interpretation
thatarethemselves amongtheconventions
of fictivediscourse.Thoughtheseconstraints differ in manyrespects
fromthoseinvolvedin ourinterpretations ofnaturaldiscourse, theyare
nevertheless substantial;and althoughthereare inevitably groundsfor
argument in determining themforindividualpoems,theyare never-
thelessrelativelyobjective.
The poet,in composing thepoem,will have made certainassump-
tionsregarding his audience,specificallythattheyare membersof a
sharedlinguistic and culturalcommunity, and thusable and willingto
abide by relevantlinguistic, cultural,and indeedliterary conventions.
To thedegreethatourinterpretations of a poemare ignorantofthose
assumptions or violatetheseconventions, we are notthatpoem'saudi-
ence,and whatever usewe maybe makingofit,we are notresponding
toitas whatitis.
Althougha poemis a representation ofdiscourse, we can understand
it, infer for our
meanings it, onlythrough priorexperiences withthe
sortof thingit does represent, namely natural utterances in historical
contexts.The poetassumes, therefore,thathisreaderhas a knowledge
ofthelanguagerepresented by the poemand thelinguistic conventions
thatgoverntherelation ofan utterance toitsmeanings in thatlanguage.
However,as we all know,linguistic convention can hardlybe separated
fromculturalconvention. The readerwhoencounters theword"God"
in a poembya seventeenth-century Englishman is notfreetointerpret it
as thedeityoftheMuslimsor Hopi Indians,anymorethanhe is freeto
interpret a paintingofthecoronation of Elizabethas thecrowning of
theQueen ofSiam.Furthermore, thepoetwillassumethathisreaders
are capableofidentifying his composition as one of a kind- a genre
- ofartwork, and therefore ofinterpreting it in relationto thosegen-
eric and artisticconventions thatoperatedforhim in composingit.
Thusthereaderwhois quiteunfamiliar withtheformsand traditional
functions ofthemasque,and mistakes Comusforthescriptofan ordi-
narytheatricalcomedy,will obviouslybe interpreting it improperly.
We shouldnoteherethatthe poet'sassumptions are not to be con-
fusedwith his intentions.Whereas the latter- his intentions- are
specific,personal,and can onlybe surmizedor hypothesized,
the former

- his assumptions-aregeneral,communal,and can be reasonably

determined oratleastsought. 7
A finalobservation shouldbe made hereregarding theview of in-
terpretation I have been -
developing and thatbywayofconcluding
thisarticle.To recognizea poemas mimeticratherthannaturaldis-
course,as a verbalartworkratherthanan eventin nature,is to ac-
knowledge it as theproductofa humandesignin accordwithcertain
valuedeffects.I have notdiscussedheretheverycrucialquestionof
thedistinction between theeffectsorfunctions offictiveand naturaldis-
coursebecauseit is a questionthatinvolvessubstantial problemsin
linguisticas well as poetictheory, and could not be dealtwith briefly.
We should,however, at leastacknowledge thefactthatpartoftheef-
fectof a poem,as distinct froma naturalutterance, derivesfromthe
reader'sawareness ofthepoetstanding, as it were,behindthepoemas
itscreaterand artificer. This awarenessis also commonly in
reflected
ourinterpretations, foramongthemeanings we seekforand inferfrom
a poemare thosethat,in Aristotelian terms,mightbe calleditsfinal
causes: i.e.,themotives orintentions,
thegoverning design,ofthepoet
as an artist,distinctfromeithera naturalspeakeror thefictive speaker
ofa poem. Thus,we can interpret Hamlet'sabuseofOpheliabothin
termsof a plausibleset of humanmotivesprojectedforHamletand
in termsofa plausiblesetofartistic motives projectedforShakespeare;
7 The linguistic,cultural,and generic constraintson interpretationalluded to here
are, of course, what much professionalcriticism(or "philology" in the broad sense)
is directed toward establishing. And to the extent that it is engaged in determining
the existenceand nature of such assumptionsand conventionson the basis of histori-
cal and publicly accessible data, criticismis a cognitivelyrespectable enterpriseissu-
ing in objectivelyverifiableand indeed cumulative knowledge (granting the prob-
abilistic nature of "verifiability"in regard to historical facts and the inevitable
grounds for uncertaintyand controversyregarding their relevance to individual
poems). It would be well, however,to recognize the distinctionbetween this enter-
prise, which is more or less continuous with that of the cultural historian,and the
aspect of professionalcriticismmentioned in fn. 5, i.e., the public articulation and
elaboration of the critic's experience as the audience of an artwork.Both are com-
monly spoken of as "interpretation"and, of course, both frequentlyappear in con-
junction, but claims that may be made for the one cannot be made for the other,
and their functionsand value are distinctivelydifferent.The meanings of a work
that a philological "interpretation"seeks to establish are those that the poem bears
in relation to the historical universe in which it was composed, and are themselves
historical and determinate; but the meanings that the poem has by virtue of its
characteristicsas a fictiveutterance are historicallyindeterminateand thus cannot
be the object of objective or cumulative knowledge, though we may for various
reasons find their "interpretation"by individual readers interestingand valuable.
It might be added that each of these types of interpretationmay, in turn, be dis-
tinguishedfromthose interpretationsbrieflyalluded to below as the reader's (and,
when publicly elaborated, the critic's) hypotheses and inferences concerning the
poet's governingartisticdesign. The meanings of "interpretation"are no less mul-
tiple than the meanings of "meaning."

and the same sortof doubleinterpretation could be offeredforany

poem.
This double aspectof interpretation reflectsa morefundamental
doubleness in thenatureofpoetry, indeedtheduplicity ofartitself.As
we view the canvas,the myriadspotsof paint assumethe guiseof
naturalobjectsin the visual world,but we are nevertheless always
half-conscious ofthemas spotsofpaint.As we watchtheplay,thestage
recedesand the personalidentities of the actorsyieldto thoseof the
fictions
whomtheyportray, butwhen,at thefinalcurtain, we clap our
hands,it is not Hamlet whom we are applauding, but theperformers
and the playwright himself.The illusionsof art are neverdelusions.
The artwork interests,
impresses, and movesus bothas thethingrepre-
sentedand as the representing itself: as the actionsand passionsof
PrinceHamletand as theachievement ofWilliamShakespeare, as the
speechof men-and as thepoet's fictions.
BENNINGTON COLLEGE


SMITH Barbara - Poetry As Fiction

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

SMITH Barbara - Poetry As Fiction

Diunggah oleh

Hak Cipta:

Format Tersedia

Poetry as Fiction

Author(s): Barbara Herrnstein Smith

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

p RADOXES makeintriguing titles,but I am not otherwise

onethatentitles it. I meanto do thisbyelaborating theproposi-

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

poetry.The second sectionof the paper will developsome implications

By "natural discourse,"I mean here all utterances- trivialor sub-

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

features thatdefineall members oftheclass,or thesyntactic rulesgov-

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

what time of day it was just then,for at any momentit is a certain

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

eventas some otherwoman sayingit of her husband or, indeed, as my

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

The relation of utterances to texts is of special interestto us

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

toricallyreal universe. And this means that most of what we call

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

of a particularutterance-i.e., what someonehas said-we are usually

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

Poems are not natural utterances,not historicallyunique verbal acts

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

othermimeticartforms, and two, as it may be distinguishedfromother

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

art mediumitself,particularlyits implicitdualism of formand matter.

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

pressions"on theirfaces,the gesturesof theirhands, and so forth.Had

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

are also fictiverepresentations of discourse. The fictivenessof prose

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

or less private or personal utterance.4 Poetry itself,as distinctnow

4 "Private or personal utterance" may be extended to include not only overt

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

poem, no initial historicaloccurrence. What the poet composes as a

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

Throughoutour livesas verbal creatures,we have learnedto respond

5 I should emphasize that I am not specificallyreferringhere to those formally

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

saythatwe mustregardthepoemas an anonymous giftdroppedfrom

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

suggeststhat the meaningsof the poem are to be located exclusively

Forfeit,indeed, to a confineddoom, if interpretedas this editor sug-

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

over,the poet mustsuggest,throughthe possibilitiesof language alone,

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

subjectivists:thosewho, scorningthe revelations of literaryhistory

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

- his assumptions-aregeneral,communal,and can be reasonably

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

and the same sortof doubleinterpretation could be offeredforany

This content downloaded from 194.29.185.230 on Tue, 17 Jun 2014 09:35:10 AM

Anda mungkin juga menyukai