You are on page 1of 15

The Logical Status of Fictional Discourse

Author(s): John R. Searle


Reviewed work(s):
Source: New Literary History, Vol. 6, No. 2, On Narrative and Narratives (Winter, 1975), pp.
319-332
Published by: The Johns Hopkins University Press
Stable URL: http://www.jstor.org/stable/468422 .
Accessed: 08/05/2012 07:39

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
The Logical Statusof FictionalDiscourse

JohnR. Searle

I
BELIEVE THATspeakingor writing in per-
in a languageconsists
formingspeech acts of a quite specifickind called "illocutionary
acts." These include makingstatements,askingquestions,giving
orders, making promises,apologizing, thanking,and so on. I also
believethatthereis a systematicset of relationships
betweenthe mean-
ings of the words and sentenceswe utterand the illocutionaryacts we
performin the utteranceof thosewords and sentences.'
Now foranybodywho holds such a view the existenceof fictional
discourseposes a difficult
problem. We mightput the problemin the
formof a paradox: how can it be both the case that words and other
elementsin a fictionalstoryhave theirordinarymeaningsand yet the
rulesthatattachto thosewordsand otherelementsand determinetheir
meanings are not complied with: how can it be the case in "Little
Red Riding Hood" both that "red" means red and yet that the rules
correlating"red" withred are not in force? This is onlya preliminary
formulationof our questionand we shall have to attack the question
more vigorouslybeforewe can even get a careful formulationof it.
Beforedoing that, however,it is necessaryto make a few elementary
distinctions.
The DistinctionBetween Fiction and Literature: Some worksof fic-
tion are literaryworks,some are not. Nowadays most worksof litera-
ture are fictional,but by no means all worksof literatureare fictional.
Most comic books and jokes are examplesof fictionbut not literature:
In Cold Blood and Armiesof theNightqualifyas literaturebut are not
fictional.Because mostliteraryworksare fictionalit is possibleto con-
fuse a definitionof fictionwith a definitionof literature,but the
existenceof examplesof fictionwhichare notliteratureand of examples

x For an attempt to work out a theory of these relationships,see J. R. Searle,


Speech Acts (Cambridge, 1969), esp. Chs. 3-5.
320 NEW LITERARY HISTORY

of literaturewhich are not fictionalis sufficient to demonstratethat


this is a mistake. And even if therewere no such examples,it would
stillbe a mistakebecause the conceptof literatureis a differentconcept
fromthat of fiction.Thus, forexample, "the Bible as literature"indi-
cates a theologicallyneutralattitude,but "the Bible as fiction"is ten-
dentious.2
In what followsI shall attemptto analyze the conceptof fictionbut
not the concept of literature.Actually,in the same sense in which I
shall be analyzing fiction,I do not believe it is possible to give an
analysisof literature,forthreeinterconnected reasons.
First,thereis no traitor setoftraitswhichall worksof literaturehave
in common and which could constitutethe necessaryand sufficient
conditionsforbeing a work of literature.Literature,to use Wittgen-
stein'sterminology, is a family-resemblance notion.
Secondly, I believe (though will not attemptto demonstratehere)
that "literature"is the name of a set of attitudeswe take toward a
stretchof discourse,not a name of an internalpropertyof the stretchof
discourse,thoughwhywe take the attitudeswe do will of coursebe at
leastin parta functionofthepropertiesofthe discourseand notentirely
arbitrary.Roughlyspeaking,whetheror not a workis literatureis for
the readersto decide, whetheror not it is fictionis for the author to
decide.
Third, the literaryis continuouswith the nonliterary.Not only is
thereno sharp boundary,but thereis not much of a boundaryat all.
Thus Thucydidesand Gibbon wroteworksof historywhichwe may or
may not treatas worksof literature.The SherlockHolmes storiesof
Conan Doyle are clearlyworksof fiction,but it is a matterof judg-
mentwhethertheyshould be regardedas a part of Englishliterature.
The DistinctionBetweenFictional Speech and FigurativeSpeech: It
is clear that just as in fictionalspeech semantic rules are altered or
suspendedin some way we have yetto analyze,so in figurative speech
semanticrulesare alteredor suspendedin some way. But it is equally
clear that what happens in fictionalspeech is quite differentfrom
and independentof figuresof speech. A metaphorcan occur as much
in a workof nonfiction as in a workof fiction.Justto have some jargon
to workwith,let us say thatmetaphoricaluses of expressionsare "non-
literal"and fictionalutterancesare "nonserious."To avoid one obvious
sortof misunderstanding, thisjargon is not meantto implythatwriting
2 There are other senses of "fiction" and "literature" which I will not be dis-
cussing. In one sense "fiction" means falsehood, as in "The defendant's testimony
was a tissue of fictions,"and in one sense "literature" just means printed matter,
as in "The literatureon referentialopacity is quite extensive."
THE LOGICAL STATUS OF FICTIONAL DISCOURSE 321

a fictionalnovel or poem is not a seriousactivity,but ratherthat, for


example, if the author of a novel tells us that it is rainingoutside he
isn't seriouslycommittedto the view that it is at the time of writing
actually rainingoutside. It is in this sense that fictionis nonserious.
Some examples: If I now say, "I am writingan articleabout the con-
cept of fiction,"thatremarkis bothseriousand literal. If I say, "Hegel
is a dead horseon the philosophicalmarket,"that remarkis seriousbut
nonliteral. If I say, beginninga story,"Once upon a time therelived
in a farawayKingdom a wise King who had a beautifuldaughter...,"
that remarkis literalbut not serious.
The aim of thispaper is to explorethe difference betweenfictional
and seriousutterances;it is notto explorethedifference betweenfigura-
tive and literal utterances,which is another distinctionquite inde-
pendentofthefirst.
One last remarkbeforewe begin the analysis. Everysubject matter
has its catchphrasesto enable us to stop thinkingbeforewe have got
a solutionto our problems. Justas sociologistsand otherswho ponder
social change findtheycan stop themselvesfromhaving to thinkby
recitingphrasessuch as "the revolutionof risingexpectations,"so it is
easy to stop thinkingabout the logical statusof fictionaldiscourseif we
repeat slogans like "the suspensionof disbelief" or expressionslike
"mimesis." Such notionscontain our problembut not its solution. In
one sense I want to say preciselythat what I do not suspend when
I read a serious writerof nonseriousillocutionssuch as Tolstoy or
Thomas Mann is disbelief.My disbeliefantennaeare much moreacute
for Dostoevskythan they are for the San Francisco Chronicle. In
another sense I do want to say that I "suspend disbelief,"but our
problemis to say exactlyhow and exactlywhy. Plato, accordingto one
common misinterpretation, thoughtthat fictionconsistedof lies. Why
would such a view be wrong?

II

Let us begin by comparingtwo passages chosen at random to illus-


trate the distinctionbetween fictionand nonfiction.The first,non-
fiction,is fromthe New York Times ( 15 December 1972), writtenby
Eileen Shanahan:

Washington,Dec. I4-A group of federal,state,and local government


officials
rejectedtodayPresidentNixon'sidea thatthe federalgovernment
providethe financialaid thatwould permitlocal governments to reduce
property taxes.
322 NEW LITERARY HISTORY

The second is froma novel by Iris Murdoch entitledThe Red and


the Green,whichbegins,
Ten more gloriousdays withouthorses! So thoughtSecond Lieutenant
AndrewChase-Whiterecently commissioned in thedistinguishedregiment
of King EdwardsHorse, as he potteredcontentedly in a gardenon the
ofDublinon a sunnySundayafternoon
outskirts in Aprilnineteen-sixteen.3

The firstthingto notice about both passages is that,with the possible


exceptionof the one word potteredin Miss Murdoch's novel, all of
the occurrencesof the words are quite literal. Both authorsare speak-
ing (writing) literally. What then are the differences?Let us begin
by consideringthe passage fromthe New YorkTimes. Miss Shanahan
is makingan assertion.An assertionis a typeof illocutionaryact that
conformsto certainquite specificsemanticand pragmaticrules. These
are:
(I) The essentialrule: the makerof an assertioncommitshimselfto
the truthof the expressedproposition.
(2) The preparatoryrules: the speakermustbe in a positionto pro-
vide evidenceor reasonsforthe truthof the expressedproposition.
(3) The expressedpropositionmustnot be obviouslytrue to both the
speakerand the hearerin the contextof utterance.
(4) The sincerityrule: the speakercommitshimselfto a beliefin the
truthof the expressedproposition.4
Notice that Miss Shanahan is held responsiblefor complyingwith
all theserules. If she failsto complywithany of them,we shall say that
her assertionis defective.If she failsto meetthe conditionsspecifiedby
the rules,we will say that what she said is false or mistakenor wrong,
or that she didn't have enough evidence for what she said, or that it
was pointlessbecause we all knew it anyhow,or thatshe was lyingbe-
cause she didn't reallybelieveit. Such are the ways that assertionscan
characteristically go wrong,when the speaker fails to live up to the
standardsset by the rules. The rules establishthe internalcanons of
criticismof the utterance.
But now notice that none of these rules apply to the passage from
Miss Murdoch. Her utteranceis not a commitmentto the truthof the

3 Iris Murdoch, The Red and the Green (New York, 1965), p. 3. This and other
examples of fictionused in this article were deliberatelychosen at random, in the
belief that theoriesof language should be able to deal with any text at all and not
just with speciallyselected examples.
4 For a more thorough exposition of these and similar rules, see Searle, ibid.,
Ch. 3-
THE LOGICAL STATUS OF FICTIONAL DISCOURSE 323

propositionthat on a sunny Sunday afternoonin April of nineteen-


sixteena recentlycommissionedlieutenantof an outfitcalled the King
Edwards Horse named Andrew Chase-Whitepotteredin his garden
and thoughtthat he was going to have ten more gloriousdays without
horses. Such a propositionmay or may not be true,but Miss Murdoch
has no commitment whateveras regardsitstruth. Furthermore, as she
is not committedto its truth,she is not committedto being able to
provide evidence for its truth. Again, there may or may not be evi-
dence forthe truthof such a proposition,and she may or may not have
evidence. But all of that is quite irrelevantto her speech act, which
does not commither to the possessionof evidence. Again, since there
is no commitmentto the truthof the propositionthereis no quesion
as to whetherwe are or are not already apprisedof its truth,and she
is not held to be insincereifin factshe does not believeforone moment
that there actually was such a characterthinkingabout horses that
day in Dublin.
Now we come to the crux of our problem: Miss Shanahan is mak-
ing an assertion,and assertionsare definedby the constitutive rulesof
the activityof asserting;but what kind of illocutionaryact can Miss
Murdoch be performing?In particular,how can it be an assertion
since it complieswith none of the rulespeculiar to assertions?If, as I
have claimed, the meaning of the sentenceutteredby Miss Murdoch
is determinedby the linguisticrulesthat attach to the elementsof the
sentence,and if those rules determinethat the literalutteranceof the
sentenceis an assertion,and if, as I have been insisting,she is making
a literalutteranceof the sentence,thensurelyit must be an assertion;
but it can't be an assertionsince it does not complywith those rules
thatare specificto and constitutive of assertions.
Let us begin by consideringone wrong answer to our question,an
answer which some authorshave in fact proposed. Accordingto this
answer,Miss Murdoch or any otherwriterof novelsis not performing
the illocutionaryact of making an assertionbut the illocutionaryact
of tellinga storyor writinga novel. On this theory,newspaper ac-
counts contain one class of illocutionaryacts (statements,assertions,
descriptions,explanations) and fictionalliteraturecontains another
class of illocutionaryacts (writingstories,novels,poems, plays,etc.).
The writeror speakerof fictionhas his own repertoireof illocutionary
acts whichare on all fourswith,but in additionto, the standardillocu-
tionaryacts of asking questions,making requests,making promises,
givingdescriptions, and so on. I believe that thisanalysisis incorrect;
I shall not devote a great deal of space to demonstrating that it is in-
correctbecause I preferto spend the space on presentingan alternative
324 NEW LITERARY HISTORY

account,but by way of illustrating its incorrectness I want to mention


a serious difficultywhich anyone who wished to presentsuch an ac-
count would face. In generalthe illocutionaryact (or acts) performed
in the utteranceof the sentenceis a functionof the meaning of the
sentence. We know, for example, that an utteranceof the sentence
"John can run the mile" is a performanceof one kind of illocutionary
act, and thatan utteranceof the sentence"Can Johnrun the mile?" is
a performanceof anotherkind of illocutionaryact, because we know
that the indicativesentenceformmeans somethingdifferent fromthe
interrogative sentence form. But now if the sentencesin a work of
fictionwere used to performsome completelydifferent speech acts from
those determinedby their literal meaning,they would have to have
some othermeaning. Anyonetherefore who wishesto claim thatfiction
contains differentillocutionaryacts from nonfictionis committedto
the view that words do not have theirnormal meaningsin worksof
fiction. That view is at least prima facie an impossibleview since if
it were trueit would be impossibleforanyoneto understanda workof
fictionwithoutlearninga new set of meaningsfor all the words and
otherelementscontainedin the workof fiction,and since any sentence
whatevercan occur in a work of fiction,in order to have the ability
to read any workof fiction,a speakerof the language would have to
learn the language all over again, since everysentencein the language
would have both a fictionaland a nonfictionalmeaning. I can think
of variousways that a defenderof the view under considerationmight
meet theseobjections,but as theyare all as unplausibleas the original
thesisthat fictioncontainssome whollynew categoryof illocutionary
acts,I shallnotpursuethemhere.
Back to Miss Murdoch. If she is not performingthe illocutionary
act of writinga novel because thereis no such illocutionaryact, what
exactlyis she doing in the quoted passage? The answerseems to me
obvious,thoughnoteasyto stateprecisely.She is pretending, one could
say, to make an or
assertion, acting as if she were an
making assertion,
or going throughthe motionsof makingan assertion,or imitatingthe
making of an assertion. I place no great store by any of these verb
phrases,but let us go to work on "pretend," as it is as good as any.
When I say that Miss Murdoch is pretendingto make an assertion,it is
crucial to distinguishtwo quite different sensesof "pretend." In one
sense of "pretend,"to pretendto be or to do somethingthatone is not
doing is to engage in a formof deception,but in the second sense of
"pretend," to pretendto do or be somethingis to engage in a per-
formancewhich is as if one were doing or being the thingand is with-
out any intentto deceive. If I pretendto be Nixon in orderto fool the
THE LOGICAL STATUS OF FICTIONAL DISCOURSE 325

Secret Service into lettingme into the White House, I am pretending


in thefirstsense; if I pretendto be Nixon as part of a game of charades,
it is pretendingin the second sense. Now in the fictionaluse of words,
it is pretendingin thesecond sensewhichis in question. Miss Murdoch
is engaging in a nondeceptivepseudoperformancewhich constitutes
pretendingto recountto us a seriesof events. So my firstconclusion
is this: the authorof a workof fictionpretendsto performa seriesof
illocutionaryacts, normallyof the representative type.5
Now pretend is an intentionalverb: that is, it is one of those
verbswhich containthe conceptof intentionbuilt into it. One cannot
trulybe said to have pretendedto do somethingunless one intended
to pretendto do it. So our firstconclusionleads immediatelyto our
second conclusion: the identifying criterionforwhetheror not a text
is a work of fictionmustof necessitylie in the illocutionaryintentions
of the author. There is no textual property,syntacticalor semantic,
that will identifya textas a workof fiction.What makes it a work of
fictionis, so to speak, the illocutionarystance that the author takes
toward it, and that stance is a matterof the complex illocutionary
intentionsthat the authorhas when he writesor otherwisecomposesit.
There used to be a school of literarycriticswho thoughtone should
not considerthe intentionsof the author when examininga work of
fiction. Perhaps there is some level of intentionat which this extra-
ordinaryview is plausible; perhapsone should not consideran author's
ulteriormotiveswhen analyzinghis work,but at the most basic level
it is absurd to suppose a criticcan completelyignorethe intentionsof
the author,since even so much as to identifya textas a novel,a poem,
or even as a text is already to make a claim about the author's in-
tentions.
So far I have pointedout that an authorof fictionpretendsto per-
formillocutionaryacts whichhe is not in factperforming.But now the
question forcesitselfupon us as to what makes this peculiar formof
pretensepossible. It is afterall an odd, peculiar, and amazing fact
about human language thatit allows the possibility of fictionat all. Yet
we all have no difficulty in recognizingand understandingworks of
fiction.How is such a thingpossible?
In our discussionof Miss Shanahan's passage in the New York
Times, we specifieda set of rules,compliance with which makes her

5 The representativeclass of illocutions includes statements,assertions, descrip-


tions, characterizations,identifications,explanations, and numerous others. For an
explanation of this and related notions see Searle, "A Classificationof Illocutionary
Acts," Minnesota Studies in the Philosophy of Language, ed. K. Gunderson, forth-
coming.
326 NEWLITERARY
HISTORY

utterancea (sincere and nondefective)assertion. I findit useful to


thinkof these rules as rules correlatingwords (or sentences) to the
world. Think of them as verticalrules that establishconnectionsbe-
tween language and reality. Now what makes fictionpossible,I sug-
gest,is a set of extralinguistic,
nonsemanticconventionsthat break the
connectionbetweenwords and the world establishedby the rulesmen-
tioned earlier. Think of the conventionsof fictionaldiscourseas a set
of horizontalconventionsthatbreak the connectionsestablishedby the
verticalrules. They suspend the normal requirementsestablishedby
these rules. Such horizontalconventionsare not meaning rules; they
are not part of the speaker'ssemanticcompetence. Accordingly,they
do not alter or change the meaningsof any of the words or otherele-
mentsof the language. What theydo ratheris enable the speakerto
use wordswiththeirliteralmeaningswithoutundertakingthe commit-
mentsthat are normallyrequiredby those meanings. My thirdcon-
clusionthenis this: the pretendedillocutionswhich constitutea work
of fictionare made possibleby the existenceof a set of conventions
which suspendthe normaloperationof the rulesrelatingillocutionary
acts and the world. In thissense,to use Wittgenstein's jargon, telling
storiesreallyis a separate language game; to be played it requiresa
separate set of conventions,thoughtheseconventionsare not meaning
rules; and the language game is not on all fourswith illocutionary
language games,but is parasiticon them.
This point will perhapsbe clearerif we contrastfictionwith lies. I
thinkWittgenstein was wrong when he said that lyingis a language
game that has to be learned like any other.6 I thinkthis is mistaken
because lyingconsistsin violatingone of the regulativerules on the
performanceof speech acts, and any regulativerule at all contains
withinit the notion of a violation. Since the rule defineswhat con-
stitutesa violation,it is not firstnecessaryto learn to followthe rule
and thenlearn a separatepracticeof breakingthe rule. But in contrast,
fictionis much more sophisticatedthan lying. To someone who did
not understandthe separate conventionsof fiction,it would seem that
fictionis merelylying. What distinguishesfictionfrom lies is the
existenceof a separatesetof conventionswhichenablesthe authorto go
throughthe motionsof makingstatementswhich he knowsto be not
true even thoughhe has no intentionto deceive.
We have discussedthe question of what makes it possible for an
author to use words literallyand yet not be committedin accordance
with the rulesthat attach to the literalmeaningof those words. Any

6 Wittgenstein,Philosophical Investigations (Oxford, 1953), par. 249.


THE LOGICAL STATUS OF FICTIONAL DISCOURSE 327

answerto thatquestionforcesthe nextquestionupon us: what are the


mechanismsby which the authorinvokesthe horizontalconventions-
what proceduresdoes he follow? If, as I have said, the authordoes not
actuallyperformillocutionaryacts but ony pretendsto, how is the pre-
tense performed?It is a general featureof the concept of pretending
that one can pretendto performa higherorder or complex action by
actually performinglower order or less complex actions which are
constitutiveparts of the higher order or complex action. Thus, for
example, one can pretendto hit someone by actuallymakingthe arm
and fistmovementsthat are characteristicof hittingsomeone. The
hittingis pretended,but the movementof the arm and fistis real.
Similarly,childrenpretendto drive a stationarycar by actuallysitting
in the driver'sseat, movingthe steeringwheel, pushingthe gear shift
lever,and so on. The same principleapplies to the writingof fiction.
The author pretendsto performillocutionaryacts by way of actually
uttering(writing) sentences. In the terminologyof Speech Acts, the
illocutionaryact is pretended,but the utteranceact is real. In Austin's
terminology, the author pretendsto performillocutionaryacts by way
of actuallyperforming phoneticand phatic acts. The utteranceacts in
fictionare indistinguishablefromthe utteranceacts of seriousdiscourse,
and it is for that reason that there is no textual propertythat will
identifya stretchof discourseas a workof fiction.It is the performance
of the utteranceact with the intentionof invokingthe horizontalcon-
ventionsthatconstitutes the pretendedperformanceof the illocutionary
act.
The fourthconclusionof thissection,then,is a developmentof the
third: the pretendedperformancesof illocutionaryacts which con-
stitutethe writingof a work of fictionconsistin actually performing
utteranceacts withtheintentionof invokingthe horizontalconventions
that suspend the normal illocutionarycommitmentsof the utterances.
These pointswill be clearerifwe considertwo special cases of fiction,
first-person narrativesand theatricalplays. I have said that in the
standard third-person narrativeof the type exemplifiedby Miss Mur-
doch's novel, the author pretendsto performillocutionaryacts. But
now considerthe followingpassage fromSherlock Holmes:
It was in theyear'95 thata combinationof events,intowhichI need not
enter,caused Mr. SherlockHolmes and myselfto spend some weeksin
one of our great universitytowns,and it was duringthistime that the
small but instructiveadventurewhich I am about to relate befell us.7

7 A. Conan Doyle, The Complete Sherlock Holmes (Garden City, N. Y., 1932),
II, 596.
328 NEW LITERARY HISTORY

In thispassage Sir Arthuris not simplypretendingto make assertions,


but he is pretendingto be John Watson, M.D., retiredofficerof the
Afghancampaign makingassertionsabout his friendSherlockHolmes.
That is, in first-person
narratives,the authoroftenpretendsto be some-
one else making assertions.
Dramatic texts provide us with an interestingspecial case of the
thesisI have been arguingin this paper. Here it is not so much the
author who is doing the pretendingbut the charactersin the actual
performance.That is, the textof the play will consistof some pseudo-
assertions,but it will for the most part consistof a series of serious
directionsto the actorsas to how theyare to pretendto make assertions
and to performotheractions. The actor pretendsto be someoneother
than he actually is, and he pretendsto performthe speech acts and
otheracts of that character. The playwrightrepresents the actual and
pretendedactionsand the speechesof the actors,but the playwright's
performancein writingthe text of the play is ratherlike writinga
recipeforpretensethanengagingin a formof pretenseitself.A fictional
storyis a pretendedrepresentation of a stateof affairs;but a play,that
is, a play as performed,is not a pretendedrepresentation of a state of
affairsbut the pretendedstate of affairsitself,the actorspretendto be
the characters. In that sense the author of the play is not in general
pretendingto make assertions;he is givingdirectionsas to how to enact
a pretensewhichthe actorsthenfollow. Considerthe followingpassage
fromGalsworthy'sThe SilverBox:
Act I, Scene I. The curtainriseson the Barthwick'sdiningroom,large,
modern,and well furnished;thewindowcurtainsdrawn. Electriclightis
burning.On the largerounddiningtable is set out a traywithwhiskey,
a syphon,and a silvercigarettebox. It is past midnight.A fumblingis
heard outsidethe door. It is opened suddenly;JackBarthwickseemsto
fall into the room. .
Jack: Hello! I've got homeall ri--- (Defiantly.)8

It is instructiveto compare this passage with Miss Murdoch's. Mur-


doch, I have claimed,tellsus a story;in orderto do that,she pretends
to make a seriesof assertionsabout people in Dublin in 1916. What
we visualize when we read the passage is a man potteringabout his
garden thinkingabout horses. But when Galsworthywriteshis play,
he does not give us a seriesof pretendedassertionsabout a play. He
gives us a seriesof directionsas to how thingsare actuallyto happen
on stage when the play is performed.When we read the passage from

8 JohnGalsworthy,RepresentativePlays (New York, 1924), p. 3.


THE LOGICAL STATUS OF FICTIONAL DISCOURSE 329

Galsworthywe visualizea stage,the curtainrises,the stage is furnished


like a diningroom,and so on. That is, it seemsto me the illocutionary
forceof the textof a play is like the illocutionaryforceof a recipe for
baking a cake. It is a set of instructionsfor how to do something,
namely,how to performthe play. The elementof pretenseentersat
the level of the performance: the actors pretendto be the members
of the Barthwickfamilydoing such-and-suchthingsand having such-
and-suchfeelings.

III

The analysisof the precedingsection,if it is correct,should help us


to solve some of the traditionalpuzzles about the ontologyof a workof
fiction.Suppose I say: "There neverexisteda Mrs. SherlockHolmes
because Holmes nevergot married,but theredid exista Mrs. Watson
because Watson did get married,though Mrs. Watson died not long
aftertheirmarriage." Is what I have said true or false,or lackingin
truthvalue, or what? In order to answerwe need to distinguishnot
only betweenseriousdiscourseand fictionaldiscourse,as I have been
doing,but also to distinguishboth of thesefromseriousdiscourseabout
fiction. Taken as a piece of serious discourse,the above passage is
certainlynot truebecause none of thesepeople (Watson, Holmes, Mrs.
Watson) ever existed. But taken as a piece of discourseabout fiction,
the above statementis true because it accuratelyreportsthe marital
historiesof the two fictionalcharactersHolmes and Watson. It is not
itselfa piece of fictionbecause I am not the author of the works of
fictionin question. Holmes and Watson neverexistedat all, which is
not of courseto denythat theyexistin fictionand can be talked about
as such.
Taken as a statementabout fiction,the above utteranceconforms
to the constitutiverules of statement-making.Notice, for example,
that I can verifythe above statementby referenceto the works of
Conan Doyle. But thereis no questionof Conan Doyle being able to
verifywhat he saysabout SherlockHolmes and Watson when he writes
the stories,because he does not make any statementsabout them,he
only pretendsto. Because the author has created thesefictionalchar-
acters,we on the otherhand can make truestatementsabout them as
fictionalcharacters.
But how is it possiblefor an authorto "create" fictionalcharacters
out of thinair, as it were? To answerthislet us go back to the passage
fromIris Murdoch. The second sentencebegins,"So thoughtSecond
330 NEWLITERARY
HISTORY

LieutenantAndrewChase-White." Now in thispassage Murdoch uses


a propername, a paradigm-referring expression.Justas in the whole
sentenceshe pretendsto make an assertion,in thispassage she pretends
to refer(anotherspeech act). One of the conditionson the successful
performanceof the speech act of referenceis that theremust exist an
object that the speaker is referringto. Thus by pretendingto refer
she pretendsthat thereis an object to be referredto. To the extent
that we share in the pretense,we will also pretend that there is a
lieutenantnamed AndrewChase-Whitelivingin Dublin in I916. It is
the pretendedreferencewhich createsthe fictionalcharacterand the
shared pretensewhich enables us to talk about the character in the
manner of the passage about Sherlock Holmes quoted above. The
logical structureof all this is complicated,but it is not opaque. By
pretendingto referto (and recountthe adventuresof) a person,Miss
Murdoch createsa fictionalcharacter. Notice that she does not really
referto a fictionalcharacterbecause therewas no such antecedently
existingcharacter;rather,by pretendingto referto a personshe creates
a fictionalperson. Now once thatfictionalcharacterhas been created,
we who are standingoutside the fictionalstorycan really referto a
fictionalperson. Notice that in the passage about Sherlock Holmes
above, I really referredto a fictionalcharacter (e.g., my utterance
satisfiesthe rules of reference). I did not pretendto referto a real
SherlockHolmes; I reallyreferredto the fictionalSherlockHolmes.
Another interestingfeatureof fictionalreferenceis that normally
not all of the referencesin a work of fictionwill be pretendedacts
of referring;some will be real referencesas in the passage fromMiss
Murdoch where she refersto Dublin, or in Sherlock Holmes when
Conan Doyle refersto London, or in the passage quoted when he makes
a veiled referenceto eitherOxford or Cambridge but doesn't tell us
which ("one of our greatuniversity towns"). Most fictionalstoriescon-
tain nonfictionalelements:along withthe pretendedreferences to Sher-
lock Holmes and Watson,thereare in SherlockHolmes real references
to London and BakerStreetand PaddingtonStation; again, in War and
Peace, thestoryof Pierreand Natasha is a fictionalstoryabout fictional
characters,but the Russia of War and Peace is the real Russia, and
the war against Napolean is the real war against the real Napolean.
What is the test for what is fictionaland what isn't? The answer is
providedby our discussionof the differences betweenMiss Murdoch's
novel and Miss Shanahan's article in the New York Times. The
testforwhat the author is committedto is what countsas a mistake.
If thereneverdid exista Nixon, Miss Shanahan (and the restof us)
are mistaken. But if there never did exist an Andrew Chase-White,
THE LOGICAL STATUS OF FICTIONAL DISCOURSE 331

Miss Murdoch is not mistaken.Again, if SherlockHolmes and Watson


go from Baker Street to Paddington Station by a route which is
geographicallyimpossible,we will know that Conan Doyle blundered
even thoughhe has not blunderedif thereneverwas a veteranof the
Afghancampaign answeringto the descriptionof JohnWatson, M.D.
In part, certainfictionalgenresare definedby the nonfictionalcom-
mitmentsinvolvedin the workof fiction.The difference, say, between
naturalisticnovels,fairystories,worksof sciencefiction,and surrealistic
storiesis in part definedby the extentof the author'scommitmentto
representactual facts,eitherspecificfacts about places like London
and Dublin and Russia or general facts about what it is possible for
people to do and what the world is like. For example,if BillyPilgrim
makes a tripto the invisibleplanet Tralfamadorein a microsecond,we
can accept thatbecause it is consistentwiththe science fictionelement
of SlaughterhouseFive, but if we finda text where SherlockHolmes
does the same thing,we will know at the veryleast that that textis in-
consistentwiththe corpusof the originalnine volumesof the Sherlock
Holmes stories.
Theoristsof literatureare prone to make vague remarksabout how
the authorcreatesa fictionalworld,a worldof the novel,or some such.
I thinkwe are now in a positionto make sense of those remarks. By
pretendingto referto people and to recounteventsabout them, the
author createsfictionalcharactersand events. In the case of realistic
or naturalisticfiction,the author will referto real places and events
intermingling thesereferences withthe fictionalreferences,
thusmaking
it possible to treat the fictionalstoryas an extensionof our existing
knowledge. The author will establishwith the reader a set of under-
standingsabout how far the horizontalconventionsof fictionbreak
theverticalconnectionsof seriousspeech. To the extentthatthe author
is consistentwith the conventionshe has invoked or (in the case of
revolutionaryformsof literature)the conventionshe has established,
he will remainwithinthe conventions.As far as the possibilityof the
ontology is concerned, anythinggoes: the author can create any
characteror eventhe likes. As far as the acceptabilityof the ontology
is concerned,coherenceis a crucial consideration.However, thereis
no universalcriterionfor coherence: what counts as coherencein a
work of science fictionwill not count as coherence in a work of
naturalism.What countsas coherencewill be in part a functionof the
contractbetweenauthor and reader about the horizontalconventions.
Sometimesthe author of a fictionalstorywill insertutterancesin
the storywhich are not fictionaland not part of the story. To take
a famous example, Tolstoy begins Anna Karenina with the sentence
332 NEW LITERARY HISTORY

"Happy familiesare all happy in the same way, unhappy families


unhappy in theirseparate, different ways." That, I take it, is not a
fictionalbut a seriousutterance. It is a genuineassertion.It is part of
the novel but not part of the fictionalstory. When Nabokov at the
beginningof Ada deliberatelymisquotesTolstoy,saying,"All happy
familiesare more or less dissimilar; all unhappy ones more or less
alike," he is indirectly
contradicting(and pokingfun at) Tolstoy.Both
of theseare genuineassertions,thoughNabokov's is made by an ironic
misquotationof Tolstoy. Such examples compel us to make a final
distinction,that betweena work of fictionand fictionaldiscourse. A
work of fictionneed not consistentirelyof, and in general will not
consistentirelyof,fictionaldiscourse.

IV
The preceding analysis leaves one crucial question unanswered:
whybother? That is, whydo we attachsuch importanceand effortto
texts which contain largelypretendedspeech acts? The reader who
has followedmy argumentthis far will not be surprisedto hear that
I do not thinkthereis any simpleor even singleanswerto that ques-
tion. Part of the answer would have to do with the crucial role,
usuallyunderestimated, that imaginationplays in human life,and the
equally crucial role that shared productsof the imaginationplay in
human social life. And one aspect of the role that such productsplay
derivesfromthe fact that serious (i.e., nonfictional)speech acts can
be conveyedby fictionaltexts,even though the conveyedspeech act
is not representedin the text. Almost any importantwork of fiction
conveysa "message" or "messages" which are conveyedby the text
but are not in the text. Only in such children'sstoriesas contain the
concluding"and the moral of the storyis ..." or in tiresomelydidactic
authorssuch as Tolstoy do we get an explicit representation of the
seriousspeech acts which it is the point (or the main point) of the
fictionaltextto convey. Literarycriticshave explained on an ad hoc
and particularisticbasis how the author conveysa seriousspeech act
throughthe performanceof the pretendedspeech acts whichconstitute
the workof fiction,but thereis as yetno generaltheoryof the mechan-
isms by which such serious illocutionaryintentionsare conveyedby
pretendedillocutions.
UNIVERSITY OF CALIFORNIA,
BERKELEY

You might also like