Unreliability Refigured: Narrative in Literature and Film

Author(s): Gregory Currie

Source: The Journal of Aesthetics and Art Criticism , Winter, 1995, Vol. 53, No. 1
(Winter, 1995), pp. 19-29
Published by: Wiley on behalf of The American Society for Aesthetics

Stable URL: https://www.jstor.org/stable/431733

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/431733?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The American Society for Aesthetics and Wiley are collaborating with JSTOR to digitize,
preserve and extend access to The Journal of Aesthetics and Art Criticism

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
 GREGORY CURRIE

Unreliability Refigured:
Narrative in Literature and Film

As consumers of fiction, we have become ing directly for this hypothesis, what follows
skilled at recognizing unreliable narratives; as will constitute a test of it. For the worth of the
theoreticians, we are less well able to say what hypothesis is directly proportional to its success
constitutes unreliability and how it is detected. in explaining those particular devices which
I aim to improve our understanding of the theo- make up the repertoire of narrative. Unreliability
retical issues. In the process, I hope to show is one of them. The fact that our theory does
four things: well in explaining the mechanisms of unre-
liability tells strongly in its favor.
(1) That narrative unreliability is a concept sep-
arable from the concept of an unreliable
I. FICTIONS MISDESCRIBED
narrator;
(2) That narrative unreliability requires for its
A newspaper article can be unreliable, meaning
explanation the concept of an implied author;
that it misleads us about what actually hap-
(3) That narrative unreliability bears close and
pened, or would mislead us if we found it cred-
interesting connections to the importantly
ible. Being misleading in this sense requires a
distinct concept of an ambiguous narrative;1
disparity between the world as it is and the world
(4) That we can explain the prevalence of cer-
as it is represented to be, in this case by a news-
tain devices in narrative in terms of the ease
paper. Is there then, in the case of the unreliable
or difficulty of the reader's task in figuring
fictional narrative, a disparity between the world
out whether and how those devices are being
of the novel and the claims that someone-
used.
we might very naturally call her a narrator-
Another theme that runs through the discussion makes about that world? That this is always and
is the relation between narration in literary fic- necessarily the explanation of unreliability in
tions and in film. Indeed, the central example narrative is an idea I want to challenge. Before I
of narrative unreliability I shall use is an exam- do, it is worth noting that, even from the point
ple from film. Part of the problem of develop- of view of one who thinks that narrators are
ing an adequate general theory of narrative has always the source of unreliability, the idea of
been the tendency to fix on narrative literature "fictional reality misdescribed" is hardly one
as the central explanatory target, and to apply we can appeal to in explanation of that unre-
the theory so generated to other representa- liability. You may believe that there are fic-
tional modes: a method which often leads to tional worlds.2 But if you do, you must be care-
strained and implausible results. ful not to fall into a quasi-magical mode of
I shall be assuming throughout that the com- explanation, whereby you "explain" what hap-
prehension of narrative is essentially a matter of pens in the fiction by appeal to what happens in
intentional inference; the reader or viewer has the corresponding world. For how does a par-
to infer, on the basis of her reading or viewing, ticular fictional world, w, get to be the fictional
the complex and sometimes covert intentions world of The Good Soldier? Not because of any
that seem to lie behind the words and images straightforward fit between the meaning of the
the work presents. But while I shall not be argu-text and world w; we want to say that, since this

The Journal of Aesthetics and Art Criticism 53:1 Winter 1995

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
20 The Journal of Aesthetics and Art Criticism

is an unreliable narrative, the text misdescribes narrators of both kinds to count as unreliable,
w. So what extratextual mechanism links the rather different kinds of unreliability attach to
text to this world and no other? More than one each, and it is not altogether easy to state the
answer is possible, but whatever answer is fa- difference between them. But I shall not enquire
vored, the locus of our interest in the nature of more deeply into the difference between these
unreliability must surely be the mechanism it- two kinds of unreliability, because I shall be
self and not the world that mechanism locates. asking: What happens to the concept of unre-
For it is in the workings of that mechanism that liability when there is no narrator? Merely for
we shall find the justification for saying, "The the sake of simplicity, and not because it favors
text is a misdescription of this world, rather my argument, I shall speak only of intradiegetic
than a correct description of that one." Since I narrators in what follows: they are more com-
hold that this mechanism is intention, and the monly the source of narrative unreliability, and
recognition of intention on the part of the audi- their role in opposition to the implied author is
ence, I hold that intention is the key to narrative easier to conceptualize, since it is often very
unreliability. difficult to distinguish between a case where
But that, we may suppose, is a lesson well implied author and extradiegetic narrator are
learned. For the standard account of narrative distinct, and a case where the implied author is
unreliability is one that appeals, exactly, to the simply speaking ironically.
mental economies of agents-though these agents The implied author is epistemically domi-
are typically thought of as hypothetical or imagi- nant over the narrator in this sense: that the
native constructs rather than living beings. The intentions of the implied author determine what
standard account says that narrative unreliabil- is true in the story, while the mental economy
ity is a product of a discrepancy between what of the narrator is thought of simply as a part of
we might call internal and external perspec- the story itself and not as authoritative-not, at
tives. The external perspective is that of the so- least, automatically so. The narrator's role is to
called "implied author," a figure who in a sense tell us what is true in the story, and, like tellers
may herself be fictional or imagined, because in real life, she may have it wrong, or wish to
her mental economy does not necessarily corre- tell us other than what she believes is true. On
spond to that of the actual author, but who is not this model, we perceive narrative unreliability
to be thought of as occupying a position within when we perceive a disparity between the (de-
the work itself. Rather she is conceptualized as termining) intentions of the implied author
the agent responsible for the story qua fiction.3 concerning what is true in the story and the
The internal perspective is that of a narrator; a (reporting) intentions of the narrator concern-
creature who is conceptualized as a product of ing what she would have the reader believe
the work itself, rather than as the work's pro- occurred.4 And because the implied author is
ducer. In a moment I shall describe how the authoritative, this amounts to recognizing a dis-
conflict between these two perspectives creates parity between what is true in the story and the
unreliability. But we need first to note that nar- intentions of the narrator concerning what she
rators come in a variety of kinds, of which two would have the reader believe occurred. This
shall be distinguished at once. view is expressed in a summarizing remark of
The narrator may be internal (or intradiege- Wayne Booth's: "I have called a narrator reli-
tic): a character within the story itself, to be able when he speaks for or acts in accordance
thought of as telling what is in fact fiction as if with the norms of the work (which is to say, the
it were known fact-or as lies or deluded rav- implied author's norms), unreliable when he
ings, but not, anyway, as fiction (then she is does not."5 That, roughly, is how it is in Ford's
internal not only to the work but to the fictional The Good Soldier, in Camus's The Fall, in Ishi-
story that work has to tell). Or she may be an guro's The Remains of the Day, and in many
external (extradiegetic) narrator, who announces other narratives we commonly describe as unre-
herself as telling the story as fiction, but where liable.
there is some reason to think of this narrator's But what of those narratives which are intu-
voice as distinct from and dependent on that of itively unreliable, but where the unreliability is
the imnlied author While it mav he nossihie for not, or at least not obviously, attributable to a

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
Currie Unreliability Refigured: Narrative in Literature and Film 21

narrator? In the literary case you might insist liable one in the sense of Booth.8 But some-
that there always is a narrator to blame for the times narrators are noncontrolling. That occurs
unreliability, however unobvious her presence when the narrator is embedded: when her own
may be (and there are some notably unobvious text is not the text we read but a text described
narrators). This seems rather ad hoc; it is worth or reported in the text we read (which may also
asking whether there is some more elegant so- describe the embedded narrator's telling of it).9
lution to the difficulty. And with film, the idea As I have said, our two distinctions cut across
of a hidden narrator such as would be postu- one another. That gives us four options: fore-
lated in order to save the Boothian definition of grounded/controlling, foregrounded/noncontrol-
unreliability strains the bounds of coherence. It ling, backgrounded/controlling, and backgrounded/
will take a moment to see why. noncontrolling. But it will be seen that there is a
difficulty in the last of these combinations. A
II. THE ASYMMETRY BETWEEN backgrounded narrator is a shadowy figure
LITERATURE AND FILM whose characteristics are hard to identify in
detail, and everything about her that can be
I want now to introduce two further distinctions inferred has to be inferred on the basis of very
between kinds of narrators, and these distinc- tenuous evidence-otherwise she would not
tions will play an important role in the argu- count as backgrounded. In reasoning about the
ment that follows. (Remember: we are ignoring characteristics of such a narrator, as with other
external narrators, and the distinctions I am kinds of evidentially underfunded reasoning,
now making are, for our purposes, distinctions we rely very heavily on default assumptions; if
within the class of internal narrators.) These a decision has to be made about the possession
new distinctions cut across one another. First, I of some characteristic and there is no evidence
draw a distinction between a foregrounded and either way, we tend to favor the answer that is
a backgrounded narrator. A foregrounded nar- simpler or otherwise preferable on a priori
rator is one whose presence is signaled in the grounds. And it is simpler to assume that a
work itself, a backgrounded one is a narrator backgrounded narrator is controlling than that
whose presence has to be inferred (I do not she is not; to suppose that she is not is to see her
claim that there is a sharp boundary between appearing as told about in the text, rather than
these two kinds, but indeterminacy of boundary as the source of the text itself. But if she is told
is, of course, no argument against their distinct- about in the text, there ought to be some evi-
ness).6 The second distinction is one between dence in the text for her existence, which, by
controlling and noncontrolling narrators. Nar- assumption of her backgroundedness, there is
rators are characters within the world of the not. In that case, backgrounded narrators are
fiction who are to be thought of as telling us almost bound to be controlling-it being rela-
facts, or lies, or deluded ravings-but not as tively unproblematic that the agent responsible
telling us a fictional story.7 Narrators tell by for a text might not signal her presence within
making utterances, and we can speak of the text the text itself. And I certainly am not aware of
of that utterance, whether written or not. Now any actual cases of works where we could point
that text-the text of which it is fictional that it with any conviction to a backgrounded but non-
is uttered by the narrator-may coincide with controlling narrator.
the text we are reading when we read the work. Now there is something awkward-indeed,
In that case, we imagine the narrator to be con- something close to incoherence-about the idea
trolling; he or she is the source of the text of a controlling narrator in film. With literature
before us. We know, of course, that the text is it is often natural to imagine that what one is
fictional, and we probably know the identity of reading is a true account of certain events wit-
its real author, but we may think that it is part of nessed or otherwise known about by someone,
the fiction itself that the narrator is the source who then went to the trouble of setting it all
of this text and accept the fiction's implicit or down for us in writing; some of John Buchan's
explicit invitation to imagine exactly that. In adventure stories, we are to imagine, are the
this sense, Watson is a controlling narrator of product of a careful editor who has heard from
the Holmes stories and to some extent an unre- the parties concerned and has created a judi-

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
22 The Journal of Aesthetics and Art Criticism

cious account on paper from their reports; it is Notably, there is no foregrounded narrator in
that account, we imagine, that we are now read- the film, and no evidence, so far as I can see,
ing, and its imagined author counts as a control- for the existence of that very rare bird, the
ling narrator. But what are we to imagine that backgrounded, noncontrolling narrator either.
would be analogous to this in the filmic case- In that case we seem bound to conclude that we
that the person in the know has gone to the have unreliable narrative without a narrator.15
trouble of recreating it all for us on camera, I should say that I am not entirely convinced
spending millions of dollars, employing famous by Wilson's interpretation, and it seems to me
actors and a vast army of technicians? That that some of the evidence that Wilson cites in
seems implausible, especially in cases where support of it-in particular a crucial shot inside
the narrator, if there is one, would most natu- the getaway car-does not in fact support it.
rally be thought of as living in the pre-cine- But Wilson's interpretation does mesh very
matic age. The same argument tells against the finely with certain parts of the film which are
hypothesis that the narrator is a documentary otherwise hard to understand. And if Wilson's
filmmaker who went to the trouble of recording interpretation does not apply in all detail to
the events of which he has knowledge on film at Lang's film, it is not difficult to imagine a film,
the time; it would also seem to leave no room different from Lang's in minor ways, to which
for narrative unreliability, which is after all it does. So we may conclude that it is possible
what we are trying to account for.10 But if con- for there to be unreliable narration in film where
trolling narrators in film are ruled out, and no foregrounded narrator is present-and when
backgrounded narrators are almost inevitably it comes to definitions, possible counterexam-
controlling, we may conclude that backgrounded ples are as telling as actual ones. In that case we
narrators in film are very rare, if possible at really will have to look for another definition of
all.1" In that case, it will not do simply to insist narrative unreliability. Anyway, I shall assume,
on their presence whenever we encounter a film for the sake of the argument, that Wilson is right.
narrative that is unreliable but where there is no How might narrative unreliability occur other
foregrounded narrator. Better to say simply that than as a result of a disparity between the view-
in such a case we have unreliable narrative points of the narrator and the implied author? It
without a narrator.12 can occur, I claim, as a result of there being a
Perhaps the argument just given merely shows certain kind of complex intention on the part of
that the only unreliable film narratives there the implied author. I shall explain. An agent can
can be are those which involve noncontrolling, do something with an intention of the following
and therefore probably foregrounded narrators complex kind: she creates or presents some-
as with Rashomon, Stage Fright, and most of thing which she intends will be taken as evi-
the other filmic narrations we think of as unre- dence of her intentions, and she intends that
liable. The trouble with that conclusion is that superficial evidence will suggest that her inten-
there seems to be a counterexample to it. At any tion was X, whereas a better, more reflective
rate, George Wilson claims to have found one: grasp of the evidence will suggest that her in-
Fritz Lang's You Only Live Once.13 Wilson tention was Y. Trivial example: Frieda compli-
argues that the natural interpretation of the ments Fred on his sophisticated sense of humor.
film, according to which the young man Eddie Her flattery is a little disturbing; this is not the
is innocent of the crime for which he is due to Frieda we know. But now we see: it was all
be executed, is on closer examination not sup- ironic and intended to be recognized, ulti-
ported by, and is in fact at certain crucial points mately, as more of the abuse she usually heaps
undermined by, the film's narration and its upon Fred. Frieda's performance was unreli-
studiedly selective presentation of events. On able, and there may be people, Fred among
Wilson's view, a right interpretation of the film them, who didn't get to the second stage, be-
would have us withhold judgment as to Eddie's cause that took just a little more calculating
guilt or innocence. But that is certainly not than some of us can be relied on to make.
what most viewers of the film have done; gener- That seems to be what is going on in You
ations of critics and lay viewers have accepted Only Live Once; we take the images and sounds
the view that Eddie is an innocent victim.14 that go to make up the film as intended one

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
Currie Unreliability Refigured: Narrative in Literature and Film 23

way. But if we are scrupulous in our examina- has intentions that can be grasped only on a
tion of those images, we find peculiarities, in- significantly deeper level of reflection. The
congruities, and apparently unmotivated ele- warranted conclusion is surely that the implied
ments that start to fall into place when we see author intends us to see, straight off, the moral
that it can be interpreted in another. Their fall- idiocy of the narrator.
ing into place consists in their being seen as That is not to say that the extensions of
intended to suggest that second, less obvious Booth's definitions and mine are disjoint, for it
interpretation. Narratives which are the product is possible for a work to satisfy both of them. In
(or which seem to be the product-remember such a work there is a disparity of outlooks
that it is the implied author who concerns us between the implied author and the internal
here) of this kind of two-tier system of inten- narrator, but the disparity is not obvious, and it
tions constitute a distinctive and especially is only on deeper reflection that we realize that
challenging class of narratives, and I do not it was intended that we find the narrator unreli-
think that they are very well understood. I hope able.17 Then we have an instance of our com-
to change that somewhat in the rest of this plex intention.18 So there is overlap, but not
paper. Before I begin, a methodological remark. sameness, for the extensions of these concepts-
In appealing to the notion of an implied au- mine was introduced, after all, to cover cases
thor here I leave behind those intentional real- that Booth's does not cover-and so the defini-
ists who insist that the work must be interpreted tions characterize different concepts.
in the light of the real author's intentions- But I am afraid I shall not be able to endorse
where those intentions are to be understood as the comfortably ecumenical position that these
"embodied" or "made effective" in the text. definitions are merely different: different but
While there seem to me to be great difficulties equal. I believe that my characterization of unre-
in the realist's position, the present focus of our liability in terms of a complex intention-call it
attention need not be the occasion for a dispute unreliability2-is of greater theoretical and criti-
between us.16 The realist may take over my def- cal interest than the familiar Boothian charac-
inition of unreliability and say that it applies in terization in terms of a disparity of outlook
those cases where a complex intention of the between the narrator and the implied author-
kind I have described is possessed by the real call it unreliability1. There are of course inter-
author and is embodied in the text by her sto- esting cases of unreliability1, but they tend also
rytelling actions. to be cases of unreliability2; they tend to be
cases where the narrator's unreliability is to
III. WHICH DEFINITION? some degree unobvious. We are past the point
where a narrator's unreliability is intrinsically
Defining unreliable narrative in terms of com- interesting, because we are past the point where
plex intentions attributable to an implied author we bring to the work a strong presumption that
allows us to count a narrative as unreliable even narrators will be reliable. Without that pre-
when there is no narrator who we can identify sumption, narratorial unreliability is, of itself,
as the source of unreliability. What, then, are no more significant than the mendacity of a
the relations between this kind of unreliability dramatic speaker: no more significant, that is,
and the cases that are covered by Booth's narra- from the point of view of a theory of narrative.
tor-centered definition? Certainly, some of the But unreliability that is to some degree hidden
cases that are unreliable on Booth's definition is of theoretical interest because its operation
would not be unreliable on mine. In Lardner's depends on delicately balanced inferential strat-
"Haircut," for example, we have an internal egies that the reader must undertake. For the
narrator whose outlook (his "norms" as Booth rest of this essay I want to concentrate on unre-
puts it) is different from, and undermined by, liability2, and on the structure of our inferences
that of the implied author. But this is not a case, to it. For reasons to do with the structure of those
I believe, where we would attribute a complex inferences, unreliability2 is less frequently en-
intention to the implied author. The disparity of countered than its Boothian rival. To see why
outlooks is too obvious in this case for us to be will require the introduction of another kind of
warranted in concluding that the implied author narrative that I want to call ambiguous.

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
 24 The Journal of Aesthetics and Art Criticism

IV. AMBIGUITY AND UNRELIABILITY that case we would be doing the work a favor it
does not deserve by calling it ambiguous.
An ambiguous narrative is one which does not But the case of narrative incompetence is the
enable us to answer all the significant questions clue to solving our problem. When a question
which arise concerning the story. Significance arises which is not answered in the narrative, but
is an important condition here, since no narra- where we ascribe the nonanswer to incompe-
tion can possibly provide complete information tence, we do not think that a question has been
about the characters and events it describes. deliberately raised by the implied author, and
When is a question significant? One answer is deliberately left unanswered; we think, exactly,
this: when it is a question that members of the that either the raising or the failure to answer
audience are (normally) inclined to ask con- were due to some oversight or other failure of
cerning that narrative. But this will not do. execution. So I propose the following as the cri-
Sometimes there are questions we as audience terion of when a narration is ambiguous: when it
are inclined to ask at the end of the work and raises a question in the viewer's mind which it
which the work does not answer, but which fails to answer, and where the raising and the
would not be grounds for calling it ambiguous. nonanswering seem to have been intentional.
Many people are inclined to wonder what will This proposal gives the result that Gone with
happen to Rhett and Scarlet at the end of Gone the Wind is not ambiguous; while readers and
with the Wind (as the recent and long-awaited viewers may wonder about the future of Rhett
sequel indicates). But this would not be grounds and Scarlet, and the makers may have expected
for saying that the narrative (either the book or that they would wonder about it, and while all
the film) is ambiguous in the sense I am inter- this may be common knowledge between audi-
ested in here. Questions about the continuation ence and maker, the question does not seem to
or noncontinuation of relationships are ones we be intentionally raised and intentionally left un-
are almost always inclined to ask-at least they answered by the narrative.19
arise fleetingly in our minds-at the end of the It is easy at the level of theory to see the
work. This proposal is going to make too many differences between unreliable and ambiguous
narratives ambiguous. narratives. But it is not always easy to say
A proposal with a similar defect has it that which kind a particular work belongs to. Is
the narrative is ambiguous if it is a narrative Rashomon an ambiguous or an unreliable nar-
that leads us to expect an answer to a question ration? If there figure within it embedded nar-
when in fact it does not provide an answer; rators who are unreliable, we may grant that it
though this would at least explain the intuition is unreliable in the sense of Booth (unreliabil-
that Gone with the Wind is not ambiguous, since ity1). It would be ambiguous if it left it an open
that narrative does not lead us to expect that an question which of the conflicting accounts is
answer will be given to the question "What true. But is it unreliable in the sense that I
happens to them after the narrative breaks off?" defined above (unreliability2), viz., that it is
But this proposal is neither necessary nor suffi- possible to detect in its making the influence of
cient for ambiguity: the narrative might make it a complex intention of the kind I have described
clear right from the start that a certain question in connection with You Only Live Once? That
is not going to be answered (in the case of a would be so if we thought of Rashomon this
film that might require a voice-over to be con- way: as intended, first, to suggest to us that the
vincing). So we are not led to expect an answer, problem is to decide which account is true, and
and indeed the question may not be answered second. to suggest on deeper reflection the rela-
by the narrative, yet the question may be one tivity of truth and, in consequence, the falsity of
such that, in not answering it, the narrative our first question's presupposition, that there is
takes on the intuitive character of ambiguity. So a right answer. For some of us, the easy relativ-
satisfaction of the proposed criterion is not nec- ism of the last option is too banal to be a plaus-
essary for ambiguity. And a question may arise ible candidate for interpretation, but this may
to which we expect an answer, but where we be just an indication that there is sometimes no
put down failure to provide an answer to incom- neutral perspective from which to choose be-
petence in the construction of the narrative. In tween ambiguity and unreliability, a situation

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
Currie Unreliability Refigured: Narrative in Literature and Film 25

we sometimes experience with other interpre- subtlety and complexity of the reasoning that the
tive choices.20 audience will have to go through to cover the
But while ambiguity and unreliability are gap, and the less likely it is that they will suc-
distinct interpretive options, they are compati- ceed. And trying to raise the probability of suc-
ble, not merely in the sense that there is some- cess in such a case by reinforcing the clues at
times no principled choice between them, but in level two may simply undermine the whole proj-
the stronger sense that a single interpretation of ect by making the inference to level two more
the work may require the application of both. obvious and natural than that to level one.
We might, for instance, take Rashomon as unre- It will generally be the case that there is a
liable2 in this way: that at first glance the greater distance, in this sense, between the two
options are between the explicit accounts of the levels if at one level we are given, say, a yes
various narrators, while on reflection we see answer, and at the other a no answer, than there
that there is another option-the relativistic will be if at one level we are given an answer
one-and that the story is ambiguous between (yes or no) and at the other we are told that no
those collected at the first round and this one. (I answer is forthcoming. In the first case one has
would count that as only marginally less banal to persuade the audience to abandon a position
than straightforwardly opting for relativism, and adopt the opposite one; in the second one
but it might still be the best thing we can come has merely to do the first of those two things.
up with.) On that view, Rashomon is both am- So the second is an easier thing to do than the
biguous and unreliable2. first. And if Wilson is right, the second of these
Note that You Only Live Once is, on Wilson's two things is what Lang (or his implied surro-
account, both ambiguous and unreliable2. At gate) seeks to do with You Only Live Once.
first the question: "Is Eddie guilty of murder?" Ambiguity is also probably the best thing that
seems to be answered by the narrative. But we advocates of the delusional interpretation of
see, on closer inspection, that it is left open by The Turn of the Screw can hope for. Reflection
it. In that case might there be some internal doesn't show that there are no ghosts; at best it
connection between unreliability2 and ambi- shows that another hypothesis does about as
guity? I believe there is, though the connection well as the supernatural one when it comes to
is not a straightforward logical one. It is not that explaining the text.
unreliability2 necessitates ambiguity; rather, un- When an unreliable narrative is one that
reliability2 is an easier effect to achieve when it seems, superficially, to close a certain issue but
goes with ambiguity than when it does not. The reveals on reflection that the question is left
reason is this: it is easier to persuade the reader open, as in You Only Live Once, let us say that
or viewer that a question has been answered we have a "transition to openness." Consider a
when it has not than it is to persuade her that a transition in the opposite direction-a "transi-
question has been answered one way when in tion to closure"-where the narrative seems,
fact it has been answered in another. The task superficially, to leave a certain issue open but is
of the author of an unreliable2 narration is a seen on reflection to answer the question one
difficult one. It is to set clues at two levels: at way or another. Would a transition to closure be
level one where the clues are more obvious but easier or more difficult to effect than a transi-
only superficially persuasive, and at level two tion to openness? Taking into account only
where they are less obvious but more weighty what I have called epistemic distance suggests
when reflected upon. But the degree of diffi- that it would be neither more nor less difficult,
culty of the task varies from case to case, and since distance is symmetrical-the distance be-
one determinant of its degree of difficulty is tween two things is independent of the order in
what we might call the epistemic distance be- which they are taken. However, I think there
tween the two levels; increase the distance and are grounds for saying that it would, other
you increase the difficulty. By "distance" I mean things being equal, be more difficult to effect a
the disparity between what you want to convey transition to closure than a transition to open-
as a first impression and what you want the audi- ness. The problem, in creating an unreliable
ence to catch on to on further reflection. The narration, is to suggest one hypothesis by means
greater the distance in this sense, the greater the of more obvious but ultimately less convincing

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
26 The Journal of Aesthetics and Art Criticism

evidence, and to suggest another by means of entity resembling a human being. As for the implied
less obvious but more convincing evidence. The author, this construction adds nothing to our under-
difficulty is to ensure that the more convincing standing of filmic narration. No trait we could assign
evidence will in fact be less obvious, without to an implied author of a film could not more simply
having it disappear entirely from view. And that be ascribed to the narration itself: it sometimes sup-
difficulty will be the greater, the stronger the presses information, it often restricts our knowledge,
hypothesis that it is evidence for. After all, it it generates curiosity, it creates a tone, and so on. To
takes more evidence to get us to believe a strong give every film a narrator or implied author is to
conclusion than a weak one. It's not hard to indulge in an anthropomorphic fiction ... . [Filmic]
convince me that either Oswald shot Kennedy narration is better understood as the organization of a
or someone else did; it's much harder to con- set of cues for the construction of a story. This presup-
vince me that someone else did. So, other things poses a perceiver, but not any sender, of a message.23
being equal, one will have to provide stronger
evidence to support a definite conclusion than to Bordwell may be right to say that when watch-
support a mere ambiguity. But then the transition ing a film, we are seldom aware of being told
to closure requires stronger evidence at the level something by a human being. But that is no
of the less obvious, and that kind of transition is argument against the dependence of interpreta-
going to be more difficult to effect than is a tion on the idea of an intentional agent as sender.
transition to openness. On that assumption, we The mechanisms whereby we arrive at the in-
would expect to find transitions to openness terpretation of films and other works are no
more frequently in literature and other narrative more likely to be continually present to con-
forms than transitions to closure. And that, I sciousness in their operation than are the mech-
believe, is exactly what we do find.21 anisms of, say, arithmetical calculation. And it
is unclear how some of the functions that Bord-
V. IMPLIED AUTHOR AND NARRATOR well assigns to narration could be accounted for
outside the scope of assumptions we make about
I have been arguing that narrative unreliability a sender; he speaks, for instance, of a narration
in literature and film can occur in the absence which "suppresses information." Without re-
of a narrator, but not in the absence of an course to the idea of intention, you can speak of
implied author. There are two kinds of theories a system that fails to deliver all the information
that clash with that idea: theories that deny the you want, but not of a system that suppresses
necessity of an implied author, and theories that information. And, of course, the idea of suppres-
assert the necessity of a narrator. Such theories sion (rather than just of informational incom-
have been advocated by, respectively, David pleteness) is essential to an adequate description
Bordwell and Seymour Chatman. of filmic narration and our reaction to it; we
David Bordwell has argued that in film we feel, in some cases, that we are being deliber-
have narration without a "sender." Part of the ately deprived of information (as when we see
way I can agree with him. He argues, as I have only the hands of the murderer), that we are
argued, that narrators in film must be embed- being deprived of it for some dramatic or emo-
ded: they are "invariably swallowed up in the tional purpose, that our expectations are being
overall narrational process of the film, which played with. None of this would make sense
they do not produce."22 But I have argued that unless we understood the narration as some-
while narrators are optional elements in film thing communicated to us by someone.24 In
(and noncontrolling when present), the implied particular, unreliable narratives of the kind
author is not dispensable, that the interpretation exemplified in You Only Live Once depend, for
of film is crucially dependent on our seeing the their interpretation, on our perception of a cer-
images and other elements of the film as the tain kind of complex intention on the implied
products of intention. Here we disagree: maker's part.
But while it is an error to dispense with the
... literary theory may be justified in looking for a notion of an intelligence that communicates the
speaking voice or narrator. But in watching films we story to us, we must not confuse the need for
are seldom aware of being told something by an such an intelligence with the requirement that

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
Currie Unreliability Refigured: Narrative in Literature and Film 27

every story have a narrator. Seymour Chatman is real within the fiction itself. They are thus
tells us that the idea that stories, or some of attributable to the film's implied maker. What
them, might not be communicated by a narrator Chatman would have the narrator do can be
leads to a conclusion that "contradicts both done by the implied maker, at considerably
logic and common sense"-"that narratives lower cost to common sense and its educated
just appear unannounced."25 We can agree that cousin, theory.
no narrative "just appears"; the question is
whether its appearance requires a narrator VI. CONCLUSION

rather than simply an implied author. Chatman,

who distinguishes narrators from authors, both I have suggested that there are cases-certainly
real and implied, thinks it does: the narrator is possible and possibly actual-of a kind of nar-
'someone or something in the text who or rative unreliability not covered by the standard
which is conceived of as presenting (or trans- account. I have given a general characterization
mitting) the set of signs that constitutes it." But of this kind, and suggested how it might super-
there is certainly no violation of logic, and sede the more familiar, Boothian, kind of unre-
probably none of common sense, when we deny liability. I also described what I call ambiguous
that every text contains such a being. The narrative, and suggested that there is a close
implied author is responsible for the story, what connection between this and the kind of unreli-
she intends to be true in the story is true in it, able narrative I defined-a connection forged
and the text she writes (or the film she makes) by the difficulty readers face in making the in-
is our guide to what she does intend. She may ferences to the implied author's intentions that
intend it to be true in that story that someone are necessary if unreliability is to be detected. I
other than her is telling it, perhaps as known have argued, finally, that unreliability in narra-
fact. In that case we have a narrator. But she tive makes no sense without appeal to the con-
may not intend this; she may simply intend to cept of an implied author, but that the concept
tell a tale in which it is fictional that this and of a narrator is required only by one kind of
that occurred, but not fictional that anyone is unreliability. The implied author, we may say, is
telling that it occurred. (Of course she herself an absolute presupposition of unreliability, the
tells the story, but her doing so does not make it narrator a merely conditional one.27
fictional that she, or anyone, does so.) Where in
all this are there violations of logic and com- GREGORY CURRIE

mon sense? Department of Philosophy

While the view Chatman rejects is unprob- Flinders University
lematic, his own theory faces problems of a Adelaide
kind to which our discussion has rendered us Australia
sensitive. "Only the narrator can be unreli-
able," he tells us, and the duplicitous flashback INTERNET: [email protected]

in Stage Fright is the product of the charac-

ter Johnny who "is 'responsible' for the lying
images and sounds that we see and hear."26 But 1. My terminology only partly coincides with that of
other writers on narrative, and some of the deviations are
Johnny, like the other characters, exists within
noted and discussed in these notes. Sometimes I find their
the story, and it is no part of that story that he terminology inappropriate; sometimes I find that their ter-
produced and edited cinematic images in order minology marks distinctions in the wrong places. Anyway,
to convince his fictional fellows (and us?) of there is not much uniformity of usage evident in writing on
narrative.
his innocence-anyway a transparently self-
2. If commitment to them is nothing more than a com-
defeating enterprise. (Chatman's scare-quotes mitment to sets of propositions, I believe in them myself.
indicate an unease about this notion of respon- 3. So the implied author, as I use that notion, is always
sibility, but they do nothing to solve the prob- "extradiegetic" in Genette's sense (Gerard Genette, Narra-
lem.) Rather, the deceptive images and their tive Discourse, English translation [Oxford: Blackwell,
1972]). See text immediately below.
juxtaposition must be thought of as representa-
4. So narrator and implied author occupy quite distinct
tions of Johnny's account, though we begin by functional roles, and we should not attribute properties of
taking them also to be representations of what the one to the other. Narrators may be (but need not be)

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
 28 The Journal of Aesthetics and Art Criticism

omniscient, but implied authors never are; their perspective damning indictment of the injustice, prejudice and brutality
on the story is not one of knowledge but of determining that can be directed, in the name of justice, against an ex-
choice. criminal."
5. Wayne Booth, The Rhetoric of Fiction, 2nd ed. (New 15. Perhaps you think it analytic that narrative must have
York: Viking Books, 1983), pp. 158-159, emphasis in the a narrator (as does Sarah Kozloff: "Because narrative films
original, and in the first edition thereof, 1961. One mislead- are narrative, someone must be narrating," Invisible Sto-
ing feature of this remark is the implication that narrative rytellers: Voice-Over in American Fiction Film [University
unreliability is always and exclusively a matter of value of California Press, 1988], p. 115, quoted approvingly in
(Booth's "norms"), which is certainly not the case, as many Seymour Chatman, Coming to Terms [Cornell University
of Booth's examples attest. Booth's definition is taken over, Press, 1990], p. 133). But then you simply object to my
more or less, by Seymour Chatman, Story and Discourse terminology, and I could avoid the objection by using
(Cornell University Press, 1978), p. 233, and is repeated in another term to refer to the vehicle of narratorless storytell-
Gerald Prince, A Dictionary of Narratology (University of ing. Consider it done.
Nebraska Press, 1987), p. 101. 16. See my "Interpretation and Objectivity," Mind 102
6. Foregrounding a narrator may not require an explicit (1993): 413-428.
statement in the text of the work that there is such a narrator, 17. In that case we have a "seductive" unreliable narra-
though that is certainly one way to achieve foregrounding. tor; the classic case is The Turn of the Screw. Where Booth's
There are various ways that stop short of explicit statement in definition applies and mine does not, we have an unseduc-
which texts make features of the stories they tell obvious. tive one (see James Phelan, "Narrative Discourse, Literary
7. This, of course, is a simplification, since there are Character and Ideology," in James Phelan, ed., Reading
stories within stories where a narrator internal at one level Narrative [Ohio State University Press, 1989], p. 137).
is external at another; in these stories it is fictional that the Sometimes cases of unobviously unreliable narratives are
narrator is telling us that it is fictional that ... Inclusion of described as "ambiguous" (e.g., by Shlomith Rimmon-
such cases into our present taxonomy would further compli- Kenan, Narrative Fiction [London: Methuen, 1983], p. 103),
cate an already complex structure, and ignoring them will but I wish to use this term for another purpose. See below,
not affect the argument. Section IV
8. F. K. Stanzel supposes that all "first-person" narrators 18. We should not forget that all unreliability must ulti-
(in my terms: internal narrators) are necessarily unreliable, mately be traceable to the intentions of the implied author;
because of the limitations on their knowledge (A Theory of we see the narrator as unreliable because we think that
Narrative, trans. Charlotte Goedsche [Cambridge: Cam- words have been put into her mouth by the implied author
bridge University Press, 1984], p. 89). But failure to be so as to signal the narrator's unreliability.
omniscient is one thing and failure to be reliable another. 19. There may be other grounds for saying that Gone
Perhaps the thought here is that a non-omniscient narrator with the Wind is ambiguous in the sense favored here. The
could not be certain of the truth of any of his beliefs. But it film narratives used by David Bordwell and Kristin
is an error of the Cartesian tradition to suppose that lack of Thompson to illustrate ambiguity (Day of Wrath and Last
certainty translates into unreliability. Year at Marianbad) would count as ambiguous on my defi-
9. The term "embedded narrator" is sometimes used to nition. Bordwell and Thompson associate ambiguity
refer to any character-narrator. This strikes me as mislead- closely with causality (Film Art [Reading, MA: Addison-
ing usage; a character-narrator who is controlling in my Wesley, 1980], p. 250). Their idea seems to be that the work
sense is not necessarily embedded in the story. He tells the is ambiguous to the extent that the causes or effects of
story, but he does not tell of his own telling. See, e.g., narrative elements are unclear. Since most narrative events,
Wallace Martin, Recent Theories of Narrative (Cornell Uni- like most events in real life, have many distinct partial
versity Press, 1986), p. 135. causes and many distinct effects, we shall need to distin-
10. Chatman, I think, sees the difficulty here. See his guish the significant from the non-significant causes and
discussion of the famous "lying flashback" in Hitchcock's effects. My proposal above can be read as doing that.
Stage Fright in Story and Discourse, p. 237. But in more 20. See my "Interpretation and Objectivity." An interest-
recent writings Chatman describes this case in much more ing case of the relation between unreliability and ambiguity
problematic terms; see below, text to note 26. is Jack Clayton's The Innocents, a film version of The Turn
11. I think I agree with Christian Metz here: "the explicit of the Screw. One difficulty the filmmakers had to contend
enunciators in the film are always embedded," "The Im- with was that a significant proportion of the film's audience
personal Enunciation, or the Site of Film," New Literary would bring with them their knowledge of the unreliability
History 22 (1991): 747-772, (p. 768). in James's story, which would make it virtually impossible
12. A referee suggested that there is another argument for the film to achieve the same effect; the audience would
against narrators in film: narrators must be utterers, and the
be primed for the discovery of the higher level clues from
audio-visual representation of film is not utterance. But we the start. As I understand the film, their solution, intel-
Griceans have no problem taking "utterance" in a broad ligently enough, was to give the film an ambiguous rather
sense that includes, for example, showing people pictures than an unreliable narrative.
and making gestures. Filmic narrative is just utterance of a 21. George Wilson argues that Ford's The Searchers
complex kind. employs what I have called a transition to closure (Narra-
13. George M. Wilson, Narration in Light (Johns Hop- tion in Light, pp. 46 ff.). But I find Wilson less persuasive
kins University Press, 1986). on this than on You Only Live Once. For more on this, see
14. Recently shown here on a local TV station, Lang's my Imagination and the Image, (Cambridge: Cambridge
film was advertised with a quotation from William Farr: "a University Press, forthcoming).

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms
Currie Unreliability Refigured: Narrative in Literature and Film 29

22. David Bordwell, Narration in the Fiction Film (Uni- truth in terms, not of the beliefs of a narrator, but the
versity of Wisconsin Press, 1985), p. 60, emphasis in the intentions of an implied author. This is well argued in Alex
original. Byrne, "Truth in Fiction: The Story Continued," Australa-
23. Ibid., p. 62. sian Journal of Philosophy 71 (1993): 24-35. See also my
24. See Seymour Chatman, Coming to Terms, chap. 8 for "Interpretation and Objectivity."
criticism of Bordwell on this point. 26. Ibid., p. 132.
25. Seymour Chatman, Coming to Terms, p. 116. For 27. Thanks to Jerrold Levinson and Paisley Livingston
somewhat different reasons I argued that same view in The for discussion, and to the suggestions of a referee. Apolo-
Nature of Fiction (New York: Cambridge University Press, gies to Mark Johnstone for having stolen (part of) his title.
1990), chap. 2. I now think that we can explain fictional

This content downloaded from

188.32.34.57 on Thu, 01 Dec 2022 16:14:15 UTC
All use subject to https://about.jstor.org/terms