Reconsidering Unreliability: Fallible and Untrustworthy Narrators

Author(s): Greta Olson

Source: Narrative, Vol. 11, No. 1 (Jan., 2003), pp. 93-109
Published by: Ohio State University Press
Stable URL: http://www.jstor.org/stable/20107302
Accessed: 17-01-2017 18:41 UTC

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.

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

Ohio State University Press is collaborating with JSTOR to digitize, preserve and extend access to
Narrative

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Greta Olson

Reconsidering Unreliability:
Fallible and Untrustworthy
Narrators
INTRODUCTION

Why do we fail to trust some narrators, and why do the tales other narrators tell
strike us as incomplete? How do the phenomena of untrustworthy and fallible narra
tion function within fictional texts, and how do readers respond to these kinds of nar
ration? In this essay I will address these questions by reviewing Wayne Booth's
introduction of the term unreliable narrator and his explication of unreliable narra
tion as a function of irony, since this formulation remains the leading model for un
reliable narration. I will then describe how Booth's text-immanent model of narrator
unreliability has been criticized by Ansgar Niinning for disregarding the reader's role
in the perception of reliability and for relying on the insufficiently defined concept of
the implied author. Niinning updates Booth's work with a cognitive theory of unreli
ability that rests on the reader's values and her sense that a discrepancy exists be
tween the narrator's statements and perceptions and other information given by the
text. Niinning, I argue, overstates his case and ignores the structural similarities be
tween his and Booth's models. Both models have a tripartite structure that consists of
(1) a reader who recognizes a dichotomy between (2) the personalized narrator's per
ceptions and expressions and (3) those of the implied author (or the textual signals).
Finally, I offer an update of Booth's model by making his implicit differentiation be
tween fallible and untrustworthy narrators explicit. Drawing on new research on un
reliability, I suggest that these two types of narrators elicit different responses in
readers and are best described using scales for fallibility and trustworthiness.1

Greta Olson is Visiting Professor for Literature in the North America Studies Program at the Univer
sity of Bonn. She has published work on Martin Heidegger's ethics, eating disorders in literature, and key
holes as liminal spaces in eighteenth-century novels, and is now working on a book on representations of
criminals' bodies in English literature from Shakespeare to Conrad.

NARRATIVE, Vol. 11, No. 1 (January 2003)

Copyright 2003 by The Ohio State University

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
94 Greta Olson

BOOTH'S MODEL

Booth first gave readers a handle on how to think about narrators like Dosto
evski's Underground Man. Insisting that he suffers from liver disease while admit
ting that he cannot locate or identify his pain, the Underground Man contradicts
himself so much that the reader cannot take his words at face value: "I am a sick
man. ... I am a spiteful man. I am a most unpleasant man. I think my liver is dis
eased. Then again, I don't know a thing about my illness; I'm not even sure what
hurts" (3).
Booth defines as unreliable those narrators who articulate values and percep
tions that differ from those of the implied author. The latter term was developed by
Booth to circumvent problems of naively biographical readings of texts in which, for
instance, J. Alfred Prufrock's lack of agency was attributed to the same qualities in
the writer who gave him form, T. S. Eliot.2 As Booth comments, we cannot call
Voltaire when we want to question him about the right interpretation of Candide
(Irony 11).
Booth understands narrator unreliability to be a function of irony. Irony pro
vides the formal means by which distance is created between the views, actions, and
voice of the unreliable narrator and those of the implied author. As the following pas
sage shows, Booth's descriptions of irony may be read as further explications of the
concept of unreliable narration:

All of the great uses of unreliable narration depend for their success on far more
subtle effects than merely flattering the reader or making him work. Whenever
an author conveys to his reader an unspoken point, he creates a sense of collu
sion against all those, whether in the story or out of it, who do not get that point.
Irony is always thus in part a device for excluding as well as for including, and
those who are included, those who happen to have the necessary information to
grasp the irony, cannot but derive at least a part of their pleasure from a sense
that others are excluded. In the irony with which we are concerned, the speaker
is himself the butt of the ironic point. The author and reader are secretly in col
lusion, behind the speaker's back, agreeing upon the standard by which he is
found wanting. (Fiction 304)

Booth applies a communicative model to reading fiction here. This model allows for
secret communion between the "postulated reader" and the implied author (177). By
emphasizing the "unspoken," Booth anticipates work on conversational implicature
(by, for instance, Grice) and irony (Sperber and Wilson)?that is why we understand
"Is that your coat on the floor?" to mean "I want you to pick it up now" or perceive
"Lovely weather today" as a relevant comment on torrential rain. Detecting irony
and narrator unreliability comprises an interpretive strategy that involves reading
against the grain of the text and assuming one understands the unspoken message be
yond the literal one. Booth's emphasis on the pleasures of exclusion suggests that the
reader and implied author belong to an in-group that shares values, judgments, and
meanings from which the unreliable narrator is ousted. Those who grasp irony and

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 95

detect unreliability share the insider joke and enjoy having survived the initiation rit
ual the text appears to require.3
On the one hand, the complicity Booth describes between the implied author
and the postulated reader suggests that the implied author sends a message ("author
conveys") through the fictional medium, which the reader then receives.4 On the
other hand, Booth displays awareness that the recognition of the narrator's inconsis
tencies does not actually occur as between two persons on an intratextual level. The
implied author does not point her finger at the unreliable narrator or wink at the
reader. Rather, this illustrative analogy is used to stress the reading sophistication
that detecting unreliability requires.
Booth's detailed description of how irony functions in fiction bespeaks much
more a reception-oriented model of text interpretation. In A Rhetoric of Irony Booth
describes four steps for how a savvy reader recognizes irony in fiction and, in many
cases, determines unreliability: the reader first rejects a literal semantic understand
ing of the text and recognizes "some incongruity among the words or between the
words and something else that he knows" (10). For instance, a careful reader may
wonder at the face value of Victor Frankenstein's professions of love for Elizabeth
and respect for his mother's deathbed wish that he marry his "cousin" as soon as pos
sible, when he departs soon after her death to take up his fatal studies in Ingolstadt.
Next the reader tries out alternative interpretations. For example, the reader decides
that it is necessary to look critically at Moll Flanders's statements about how the
threat of poverty repeatedly led her to steal after her return from America. Careful
reading reveals that Moll actually has enough money to maintain herself without
working in London for a number of years (Goetsch 281-86). The reader then makes
a decision about the implied author's probable intentions, asking: "how were these
words meant to convey a message other than their intrinsic meaning?" Finally, a non
literal meaning is settled upon, one that most sophisticated readers?"those who
have the necessary information"?would agree upon and unsophisticated readers
would not.
One reads literally until textual markers and indications force one to revise
one's interpretation. Weighing the totality of textual information, one concludes that
the narrator's words are incongruous or incomplete. These textual signals include (1)
paratextual elements, as in titles such as Thomas Mann's Felix Krull, Confidence
Man and in epigraphs; (2) direct warnings that the narrator should not be confused
with the author, as in Nabokov's postscript to Lolita in which he distances himself
from the sexual preferences of Humbert Humbert5; (3) obvious grammatical, stylis
tic, or historical mistakes on the part of the narrator; (4) conflicts between fictional
facts; (5) and discrepancies between the values asserted in the work and those of the
author in other contexts (Irony 47-86).
Given that textual signals of irony are marked, an objective standpoint exists
from which insightful readers will judge irony and reliability. Hence unreliable nar
ration is a textual constant. Furthermore, Booth contends that unreliable narrators are
consistently unreliable: once they have revealed themselves to be unreliable, they do
not suddenly become infallible or conform to values otherwise asserted in the work.6
Booth provides a starting point for creating a typology of unreliable narrators. No

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
96 Greta Olson

tably, he uses the words "unreliable," "untrustworthy," "inconscience" (uncon

scious), and "fallible" to describe the narrators he wants to characterize:

I have called a narrator reliable when he speaks for or acts in accordance with
the norms of the work (which is to say, the implied author's norms), unreliable
when he does not.

If [the narrator] is discovered to be untrustworthy, then the total effect of the

work he relays to us is transformed.

It is most often a matter of what James calls inconscience', the narrator is mis
taken, or he believes himself to have qualities which the author denies him.

Sometimes it is almost impossible to infer whether or to what degree a narrator

is fallible. (Fiction 158-60)7

The above definitions demonstrate that Booth envisages different types of unreliabil
ity. "Unreliable" and "untrustworthy" suggest that the narrator deviates from the
general normative standards implicit in the text. For this reason the narrator cannot
be trusted on a personal level. By contrast, "inconscience" and "fallible" imply that
the narrator makes mistakes about how she perceives herself or her fictional world.
The first terms concern the narrator's qualities as a person and the second her ability
to perceive and report accurately. Booth's mention of degrees of potential fallibility
shows that he conceives of reliability and unreliability as well as fallibility and infal
libility as being interrelated rather than diametrically opposed.8

N?NNING'S MODEL
Ansgar N?nning's recently published monograph, Unreliable Narration, takes
issue with Booth's definition of unreliability.9 Niinning argues that the formulaic
adoption of the definition of the unreliable narrator in glossaries and dictionaries of
literary terms such as M. H. Abrams's has led to imprecise usage, a tendency to treat
reliability and unreliability as binary opposites, and a lack of attention to how unre
liability functions.
According to Niinning, the unexamined adoption of Booth's model of unrelia
bility has resulted in narrator reliability being regarded as a text-immanent issue that
rests solely on the distance between the implied author and the narrator (both intra
textual entities). For instance, Seymour Chatman's explication of Booth's model of
unreliability postulates two levels of communication, one between the implied au
thor and the implied reader and the other between the narrator and the narratee. Yet,
in N?nning's eyes, Chatman fails to say how this communication works or how the
implied reader knows to interpret the narrator as being unreliable (Unreliable 14).
Furthermore, the coupling of unreliable narration with the "notoriously indefinite or
even indefinable implied author" causes obfuscation (16). In N?nning's view the

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 97

vague, pseudo-anthropomorphic concept of the implied author serves as a repository

for all the open questions about the relationship between the author and the reader
but fails to answer any of them.10
N?nning contends that critics use the concept of the implied author to project
their own values onto texts: first, they treat the narrator as a real person, whose lack
of reliability they perceive as a personal, moral failure or, at least, a limitation. This
leads critics to overemphasize mad monologists like the Underground Man, due to
these figures' easily identifiable moral failings. Second, critics understand them
selves to be in a personal relation to a hypothetical "man behind the curtain," a Wiz
ard-of-Oz incarnation of the author, who shares their general values as well as their
judgments of the narrator. N?nning points out that when critics like William Riggan
state that narrators' lack of reliability can be recognized on the basis of "the unac
ceptability of th[eir] moral philosophy in terms of normal moral standards or basic
common sense and human decency" (Riggan 36; qtd. in Unreliable 21), they assume
the normative validity and universality of their "moral philosophy." Yet notions such
as "human decency" are highly culturally dependent. Hence, N?nning contests, crit
ics abuse the notion of the implied author to justify their subjective value judgments
about the narrator's character.
N?nning finds evidence for his argument that attributing unreliability is solely a
function of reader reception in Vera N?nning's study of Oliver Goldsmith's The
Vicar of Wakefield. In her study?one chapter of the monograph on unreliability?
Vera N?nning posits that the reception of unreliability is historically variable: The
Vicar of Wakefield's narrator Primrose was interpreted as sympathetic and trustwor
thy by two centuries of readers, only to have his reliability questioned during recent
decades. Historically varied responses to Primrose's reliability serve as proof for
Ansgar N?nning's argument that narrator unreliability is not a stable objective qual
ity but a function of reader reception. Readers disagree about which narrators are un
reliable. For instance, according to N?nning, a male pederast will not find Nabokov's
Humbert Humbert unreliable because his values are not in discord with the narrator's
(Unreliable 25). Hence ascribing unreliability is a strategy for reading texts rather
than a text-immanent phenomenon. The divergence between the reader's worldview
and that of the narrator?rather than inconsistencies between the narrator and the im
plied author?causes the reader to classify the narrator as unreliable.
Now comes the paradoxical moment in N?nning's argument. While reasoning
that attributing unreliability results from text reception and is not a text-immanent
phenomenon, N?nning nonetheless wishes to clear up the confusion surrounding un
reliable narrators by enumerating specific textual markers that signal them. N?n
ning's list of textual signals is intended to demonstrate that the narrator appears
unreliable to real historically- and culturally-embedded readers (versus Booth's pos
tulated one). This is, however, problematic. For if detecting unreliability functions as
a quality of individual reader response, how can stable textual signals exist to typify
the phenomenon of unreliability?
N?nning's list of textual signals includes: (1) the narrator's explicit contradic
tions and other discrepancies in the narrative discourse; (2) discrepancies between
the narrator's statements and actions; (3) divergences between the narrator's descrip

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 98 Greta Olson

tion of herself and other characters' descriptions of her; (4) contradictions between
the narrator's explicit comments on other characters and her implicit characterization
of herself or the narrator's involuntary exposure of herself; (5) contradictions be
tween the narrator's account of events and her explanations and interpretations of the
same, as well as contradictions between the story and discourse; (6) other characters'
corrective verbal remarks or body signals; (7) multiperspectival arrangements of
events and contrasts between various versions of the same events; (8) an accumula
tion of remarks relating to the self as well as linguistic signals denoting expressive
ness and subjectivity; (9) an accumulation of direct addresses to the reader and
conscious attempts to direct the reader's sympathy; (10) syntactic signals denoting
the narrator's high level of emotional involvement, including exclamations, ellipses,
repetitions, etc.; (11) explicit, self-referential, metanarrative discussions of the narra
tor's believability; (12) an admitted lack of reliability, memory gaps, and comments
on cognitive limitations; (13) a confessed or situation-related prejudice; (14) para
textual signals such as titles, subtitles, and prefaces (adapted from Unreliable
27-28).u
N?nning summarizes cognitive theories of reading to describe how individuals
ascribe unreliability to narrators. To begin with, the reader notes textually evident
discrepancies between the narrator's actions or telling of events and other versions of
the narration or of the narrator. The reader then relates these discrepancies to other
frames of experience. According to theories of "naturalization" (Culler; Fludernik,
Fictions and 'Natural'Narratology), readers relate what they read to ordinary human
actions, motivations, and behavioral scripts. They impose their expectations about
how texts should work and how people tell stories onto the text in order to make
sense of it.12 A part of this process of fitting the text into one's worldview is identify
ing the narrator (if there is a clearly identifiable one) and deciding what sort of per
son that narrator is on the basis of one's referential frames.
Note that the steps of this process differ only in one respect from the detection
of unreliability as Booth describes it. In both cases the reader experiences a moment
of anagnorisis or recognition that literal interpretation fails: (1) the reader notes in
consistencies on the narrator's part, (2) and then makes sense of the initial conun
drum by relating it to other world/literary experience or to what the implied author
actually meant to say; (3) another reading is decided upon; (4) however, in N?n
ning's model the recognition of unreliability is questioned or legitimized with refer
ence to the reader's cultural and individual referential frames that make this
attribution possible. Bowing to the epistemological uncertainty of our age, N?nning
recognizes that every reading is limited and situational, and every reader is poten
tially unreliable. By contrast Booth's model assumes that the classification of a nar
rator as unreliable can be verified when the implied author's irony is recognized as
intended and stable.
The process of attributing unreliability to narrators is highly similar in Booth
and N?nning's models. The following schematic representation reveals that the ele
ments of which their models consist are also nearly identical:

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 99

Booth's Model N?nning's Model

Reader Reader

Narrator Persona-Implied Author Na

(recourse to the values of the implied author

In both models three points of view coe

implied author or the fictional world c
(3) the reader, who responds to and ma
a narrator can only be identified as unr
value, opinion, perception, and forms o
plied author or the totality of textual
models is where the authority to judge
gives authority to an implied author w
tions of reliability can be addressed, w
validity of subjective reader response.
ning's model allows for a potentially li
dividual readers' responses to the te
singular textual whole that is created f
by the implied author.
Importantly, both models treat narr
does so more questioningly, as he ment
onto literary phenomena (Unreliable 21
and N?nning base their models on narra
tell. The jury remains out on whether h
To develop a typology of unreliable na
we characterize narrators like people. R
seems justified in this case, since chara
thy," "unconscious," and "fallible" are
narrators as unreliable, readers treat t
implicit theories of personality as well
text they read (Wall 30). When approach
it is clear that readers detect unreliabi
signs of irregularity are noted, and the
ary schemata of unreliability. However
nologists but somewhat untrustworthy
these types of narrators differs, as do
unreliable. This leads readers to make d
and untrustworthy narrators, as I will

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 100 Greta Olson

MODIFYING THE EXISTING MODELS

I want to suggest an amplification of Booth's model of unreliability that will

lead to clarifying our understanding of how unreliability functions and why readers
respond to this phenomenon in different ways. I see Booth's implicit distinction be
tween fallible narrators and untrustworthy ones as highly pertinent.16 Key here is the
insight that untrustworthiness and fallibility occasion varying reader reactions and
strategies. Other theorists who have worked to differentiate between forms of unreli
ability include Lanser (Narrative Act), Hof, Cohn ("Discordant"), and Phelan and
Martin, who describe six types of unreliability, as well as Fludernik ("Defining").
Importantly, all recent models of unreliability, including those of N?nning, Flud
ernik, Phelan and Martin, and Yacobi, stress that narrators cannot be neatly divided
into the categories of unreliable or reliable. Rather, as Phelan and Martin state it:
"narrators exist along a wide spectrum from reliability to unreliability" (96), and
they can become more or less reliable during the course of the stories they tell.
Monika Fludernik has proposed a tripartite model of unreliability. According to
her analysis, narrators can be unreliable because of their "factual inaccuracy," their
"lack of objectivity" as first-person narrators, or their "ideological unreliability"
("Defining" 76-77). In the first case the narrator may lie consciously or may not
have access to accurate information; in the second case, the narrator is subject to the
"epistemological restrictions" (76) typical of homodiegetic narration or may be
morally unreliable (for instance, racist). In the third case the narrator's actions and
evaluations may be consistent to herself but discordant with the "world view" of the
reading audience (77). While its differentiation between lack of factual information,
bias, and ideological disagreement as causes for unreliability is clarifying, Flud
ernik's tripartite model does not distinguish between limitations inherent to ho
modiegetic storytelling and unreliability per se. Furthermore, the first category of
unreliability does not make a distinction between narrators who reliably convey all
of the information they have access to and those who deliberately withhold it. Such
narrators, I think, invite differing attributions about their behavior and different cate
gorizations.
By contrast, Phelan and Martin describe two different ways in which readers
may detect unreliability (and, implicitly, two different general types of unreliable
narrators). They posit an implied author behind the narrator and an "authorial audi
ence" that is "the author's ideal audience" not dissimilar from Booth's postulated
reader (108). Audience members either "reject" the narrator's face-value account and
"reconstruct" another more satisfying story, or they "supplement the account" of the
narrator with information he does not give them (92). Using The Remains of the
Day's Stevens as an example of a narrator who vacillates in how reliably he per
ceives his own and others' motivations and actions as well as how much he chooses
and is able to report fictional events, the authors describe six types of unreliability.
The first three types of unreliability are grouped together on the basis of how the
reader responds to them, namely by replacing the narrator's story with a less contra
dictory account of fictional events, and the second three on the reader's need to am
plify on the narrator's incomplete tale. Regarding the first group, narrators may

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 101

falsely report fictional events ("misreporting"), or make mistakes of perception

("misreading"), or falsely evaluate events ("misregarding"). In the second group nar
rators may evidence unreliability in their not telling enough about what is happening
("underreporting"), their failing to grasp events completely ("underreading"), or
their making incomplete value judgments ("underregarding"). Phelan and Martin's
categorization of six types of unreliability is based on (1) the axes of the narrator's
faulty factual, ethical, and epistemological evaluations, and (2) on the reader's re
sponse to these evaluations.
Taking the lead from Booth's implicit typology of narrators, I differentiate be
tween fallible and untrustworthy narrators and follow Phelan and Martin in distin
guishing between reading strategies necessary for making sense of various types of
unreliability. Before describing the difference between narrator fallibility and un
trustworthiness, it appears necessary to consider the general limitations of ho
modiegetic narrators. As inhabitants of their textual worlds, these narrators cannot
have metatextual, omniscient knowledge.17 As Stanzel writes, it is the particular
quality of "personalized narrators" to demonstrate to us the "biased nature of our ex
perience of reality" (11). Homodiegetic narrators are subject to the epistemological
uncertainty of lived experience. Yet they are not necessarily unreliable. Often mys
tery and detective novels feature narrators who conform to the normative values of
their narratives. Such narrators reliably report on whatever informational puzzles
they are currently piecing together. Of course, they do not and cannot provide their
readers with vital pieces of the puzzle until they themselves have found them out,
typically towards the end of the narrative. Consider Walter Hartright, whose narra
tion begins and ends Wilkie Collins's The Lady in White, or Philip Marlowe in Ray
mond Chandler's The Big Sleep. An impoverished gentleman and drawing teacher,
Hartright?the name "heart" "right" says it all?is rewarded for his subdued suffer
ing during the engagement, marriage, and seeming death of his beloved Laura Fair
lie as well as for his persistent efforts to solve the mystery of the lady whose figure
gives the novel its title. The novel ends with the villains' deaths and Hartright's mar
riage to Laura, the birth of their son, and their inheritance of Limmeridge House and
a considerable fortune. Similarly, Marlowe unravels the story of Rusty Regan's dis
appearance in a series of caustic stops and starts, but completely faithfully. Before
the denouement in which he tells Vivian Regan how she paid to have her husband's
murder by her little sister hushed up so as to let her father die peacefully, Marlowe
accurately characterizes his narrative style: "I'm not joking, and if I seem to talk in
circles, it just seems that way. It all ties together?everything" (214). Both Hartright
and Marlowe have restricted visions of the action. Yet because both report with fac
tual accuracy and are in accord with the values of their narratives, I regard them both
as reliable.
By contrast, fallible narrators do not reliably report on narrative events because
they are mistaken about their judgments or perceptions or are biased. Fallible narra
tors' perceptions can be impaired because they are children with limited education or
experience, as in Huckleberry Finn', or, as in the case of Marlow from Lord Jim, their
reports can seem insufficient because their sources of information are biased and in
complete. We can imagine circumstances in which fallible narrators like Huck and

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 102 Greta Olson

Marlow might tell their tales completely and reliably. Applying a longstanding in
sight from social psychology about how individuals make attributions about the rea
sons for people's behavior, I believe that readers regard the mistakes of fallible
narrators as being situationally motivated. That is, external circumstances appear to
cause the narrator's misperceptions rather than inherent characteristics. Readers may
justify the failings of fallible narrators?just as they would tend to justify their own
similar mistakes?on the basis of circumstances that impede them rather than on
their intellectual or ethical deficiencies.18
Conversely, untrustworthy narrators strike us as being dispositionally unreli
able. The inconsistencies these narrators demonstrate appear to be caused by in
grained behavioral traits or some current self-interest. Moll Flanders demonstrates a
consistent tendency to equivocate morally; the Underground Man appears mentally
unstable during the entire course of the tale he relates. We surmise that other narra
tors could behave differently (more reliably) in the same narrative situation and that
untrustworthiness is a distinct characteristic of the narrator. Hence, the reactions un
trustworthy narrators elicit in readers differ significantly from those in response to
fallible ones. What the narrator says will be greeted by skepticism and rapidly
amended when it is inconsistent.
To illustrate the difference between narrator fallibility and untrustworthiness I
want to analyze two cases of each.19 Huck Finn represents a highly fallible narrator
and Lord Jim's Marlow a slightly fallible one. Tellingly, Booth uses the term "incon
science" to describe Huck's fallibility: "the narrator claims to be naturally wicked
while the author silently praises his virtues behind his back" (Fiction 159). Although
Huck is smart as a whip and eminently likable, his perceptions are nonetheless mis
taken because of his age, his superstitions, and his simply not knowing pertinent
facts, as well as his as yet literal understanding of the shallow moral norms he has
been exposed to by Miss Watson and her ilk.
A clear case of fallibility concerns the climactic scene in which Huck decides to
"steal Jim out of slavery again" (209). He condemns himself for having been raised
so wickedly that he is not able to do what he knows to be "right" in the eyes of his
white townsfolk, to return Jim to his owner. Yet the larger narrative clearly suggests
a normative standard that fully confirms Huck's action and derides the conventional
mores of behavior he chides himself for failing to conform to: Jim is consistently
portrayed as heroic, if uneducated, and represents, perhaps, the only morally sound
adult male figure in the novel. At the end of the novel he comes out of hiding to help
nurse Tom Sawyer back to health, and this act results in his being taken captive
again. Similarly, Jim parents Huck by allowing him to sleep through his shifts as
lookout on the raft and teaches Huck ethical standards by telling him that it is "trash"
to try to fool one's friend into thinking that a series of traumatic events was only a
dream (90).
A lesser case of fallibility occurs when Huck describes how the widow prays
before meals at the very beginning of the novel. When he complains that "you had to
wait for the widow to tuck down her head and grumble a little over the victuals,
though there warn't really anything the matter with them" (11), Huck, on the one
hand, completely misperceives the nature of the religious ritual. Yet, on the other

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 103

hand, he provides an ironic commentary on the way many individuals actually do

pray before meals.20
Along the range of fallible homodiegetic narrators, Huck Finn appears more
fallible than Marlow. In Lord Jim, Marlow depends on letters, an incomplete text,
highly prejudiced witnesses such as Gentleman Brown (who hates Jim and con
tributed to his death), and Jim's traumatized lover Jewel to piece together the story of
how Jim died. Again Marlow's incomplete narration of the Patusan episode appears
to be caused by the circumstance that he was not with Jim at the time of his death
rather than by any internal motivation to conceal narrative events from the reader.
The reader senses that Marlow's perceptions are convincing in themselves, even
though his mental inclusion of Jim as "one of us" and his disappointment at Jim's
having abandoned the sinking Patna certainly color the quality of his storytelling.
Yet the information he has access to is questionable. Indeed, much of the drama of
Lord Jim stems from Marlow's faltering efforts to read the character of the young
man he studies and tells his audience about. Furthermore, one may reason that had
Marlow witnessed Jim's death, he could have given his dinner audience and the
reader a much more complete and conventionally satisfying account of it.
To determine how fallible a narrator is the reader must ask to what extent the
narrator mistakes the information he has access to and the perceptions he has. Does
the narrator make these mistakes consistently, and can we imagine circumstances in
which the narrator would report infallibly? If we can imagine a grown up Huckle
berry who will know that the widow's "grumbling" is saying grace, and that his de
cision to tear up his letter to Miss Watson was the right one, if we can imagine a
Marlow who, with more information and reflection, could narrate the inchoate mean
ing of Jim's life and death with more clarity, we have good reason to regard the Huck
and the Marlow we encounter as fallible rather than untrustworthy narrators.
To illustrate untrustworthy narration, the speaker of Edgar Allan Poe's "The
Tell-Tale Heart" will serve as an example of a highly untrustworthy narrator and
Moll Flanders as a somewhat untrustworthy one. Like the Underground Man, the
speaker of "The Tell-Tale Heart" begins his tale with a number of inconsistencies.
Poe's narrator insists that his addressee has accused him of madness (always a sure
sign of mental stability): "True!?nervous?very, very dreadfully nervous I had
been and am; but why will you say that I am mad? The disease had sharpened my
senses?not destroyed?not dulled them" (792). To make sense of this narration the
reader will be quick to attribute mental instability and untrustworthiness to its
source. The narrator will be diagnosed with pathological untrustworthiness, and the
reader will choose the therapeutic strategy of reading against the grain.
However, many other narrators appear only marginally unreliable. Moll Flan
ders has been alternately identified as the eloquent speaker of London's poor or the
object of Daniel Defoe's moral derision. At one point Moll describes her second
foray into property crime when she steals a young girl's necklace: Moll first blames
poverty for making her need to steal from a child (note that some critics question
Moll's math); she then regrets that she was insufficiently reformed during her years
as a prosperous married woman. She then projects her desire to steal onto her
"prompter, like a true devil," who puts her up to taking the necklace (151). Finally,

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 104 Greta Olson

she reasons that the child's vain mother is at fault for having dressed the child in such
an expensive bauble, and that the family maid is guilty of having not supervised the
child properly by going off to flirt instead. The parents, Moll believes, will learn a
valuable lesson through having had their child's necklace stolen. During the course
of this passage Moll alternately styles herself as a victim, a fallen sinner, and an am
bassador of morality. She makes the weaknesses of others?the mother's vanity, the
maid's amorousness, the devil's prompting?responsible for her crime and not her
self. Clearly, Moll's narrative demands that the reader undertake several interpretive
moves to make sense of her contradictions. Repeated episodes of moral equivocation
like this one invite the reader to attribute untrustworthiness to Moll as a constant be
havioral trait.
At one end of the spectrum, untrustworthy narrators contradict themselves im
mediately or announce outright that they are insane. At the other end, readers are re
quired to do more "detective" work to determine whether a narrator is trustworthy or
not, and critics remain divided about how to characterize the storyteller (Fludernik,
"Defining" 78). The pertinent question is to what extent the narrator is exposed, or
whether she exposes herself, as dispositionally untrustworthy. Readers attribute in
ternal inconsistency and self-contradiction to narrators they judge to be lacking in
trustworthiness. We predict that they will continue to contradict themselves and take
on a reading strategy that questions and revises all that they say.
To my mind the separation of narrators into untrustworthy or fallible applies for
all narrators traditionally labeled unreliable. Supporting this thesis is Phelan's state
ment in a recent defense of the need for an implied author in accounts of unreliabil
ity: "Narrators . . . can be unreliable in two different ways, either by falling short or
by distorting. Narration that falls short is reliable up to a point; narration that distorts
is simply unreliable" ("Can Readers" 6). However, it is also possible for narrators to
move from being fallible to being untrustworthy in the course of a narration.
Nonetheless, I suspect that readers will, like Booth, prefer making more straightfor
ward attributions of fallibility or untrustworthiness.

CONCLUSION

An analysis of Booth and N?nning's models of unreliable narration shows that

unreliability can only be determined where there are three points of view. The reader
judges the narrator's unreliability based on textual signals, and then moves beyond a
literal reading of the text and attributes fallibility or untrustworthiness to the narrator.
Suspending disbelief, the reader attributes personal qualities of fallibility and trust
worthiness to narrators just as she makes attributions about individuals in other con
texts. Narrators, my analysis has hopefully shown, differ as to whether and to what
degree they are fallible or untrustworthy. Depending on their perceived deficiencies,
narrators elicit different responses and require varying reading strategies. When nar
rators are untrustworthy, their accounts have to be altered in order to make sense of
their discrepancies. Fallible narrators by contrast make individual mistakes or leave
open informational gaps that need to be filled in. Untrustworthy narrators meet with

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 105

our skepticism about their characters, whereas fallible narrators are more likely to be
excused for their failures to deliver on the informational goods.
N?nning's singling out reader response as the sole basis for detecting unrelia
bility ignores the discrepancy between conflicting points of view within the text as
well as the reader's sense of being in cahoots with the viewpoint that differs from the
narrator's. Texts featuring untrustworthy narrators?like ironical utterances?have a
"target or victim." The reader "feel[s] drawn into a conspiracy with the speaker"
(Sperber and Wilson 313), here the implied author or those textual signals that give
the impression of there being a person behind the narrator of a text. As Booth puts it,
in cases of irony (and unreliability) "the speaker is himself the butt of the ironic
point" (Fiction 304). Attributing unreliability to the narrator, the reader recognizes
the text's implicit joke and sees that the narrator is not what she proposes to be.
Cases of unreliable narration invite the reader to depart from a literal reading,
something?as Booth and Cohn point out?readers tend to resist. As in cases of
irony, the narrated utterance must be turned over and reinterpreted. A gap opens be
tween the literal and the implied; when the reader detects unreliability, she enters the
gap successfully. Textual signals help her decide whether the narrator is fallible or
untrustworthy. The decision allows the reader to predict whether the narrator is likely
to always misreport or is prevented by circumstances from telling the tale straight.
The reader can then assume a strategy by which she can make different types of un
reliable narration comprehensible and render fallible and untrustworthy narrators re
liable in their unreliability.

ENDNOTES
1. Several individuals have kindly guided me through earlier versions of this text. However, I need to
particularly thank Monika Fludernik, in whose colloquium on narratology the idea for this paper was
conceived, and Ansgar Niinning, who has graciously supported my continuing dialogue with his
work.

2. The implied author represents a tool with which the critic can talk about the values, linguistic pecu
liarities, and worldview of a particular text?as they can be documented in a text?without referring
to the personality of the historical person who wrote the book. W. K. Wimsatt and Monroe C. Beard
sley called this type of misreading "the intentional fallacy."

3. Psychological experiments demonstrate that the more difficult it is to achieve membership status in a
particular group, the more group identity is valued by the individual members (Aronson and Mills). It
would be interesting to study whether textual difficulty (i.e., ambivalence, complexity, syntactical
play, multiperspectivity) represents a form of initiation: those readers who are admitted into the
"group" of competent readers (Booth writes in the plural form) are all the more gratified by their
comprehension, depending on how difficult it was to achieve.

4. Fludernik {Fictions 58-65) provides a very helpful discussion of how the adoption of conventional
sender- and speech-oriented linguistic models of communication in literary studies has led to the ne
glect of recipient-oriented ones.

5. "I am trying to be an American writer and claim only the same rights that other American writers
enjoy. On the other hand, my creature Humbert is a foreigner and an anarchist, and there are many
things, besides nymphets, in which I disagree with him" (317).

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
106 Greta Olson

6. For instance, Booth writes that the narrator of James Joyce's Ulysses should not be termed unreliable:
"to be undependable in this sense is not identical with being what I have called unreliable; most un
reliable narrators are dependable in the sense of being consistent" {Fiction 300).

7. Pragmatically, these definitions leave ample room for interpretation, as is documented in the various
translations of the term into German such as "reliability," "believability," "truthfulness," and "trust
worthiness" (Niinning, Unreliable 11).

8. A minor point here is that Booth also names Fleda Vetch, the reflector figure, filter, or focalizer (the
reader may choose her preferred terminology) from Henry James's The Spoils ofPoynton as an ex
ample of an unreliable narrator {Fiction 159).

9. This is my translation of Niinning throughout. English-speaking readers can refer to a lecture that
Niinning delivered on the same subject in 1997, which was published in 1999 ("Reconceptualizing").
Unfortunately, this lecture does not contain the helpful list of textual signals of unreliability or the
suggestions for further research areas concerning this topic that N?nning's introduction to the mono
graph does.

10. The vehemence with which Niinning attacks the inconsistencies of the implied author model of unre
liability might lead one to wonder if he does not in fact wish to accuse Booth of being something of
an unreliable narrator himself. For the pros and cons of the implied author, see, for instance, Chatman
(74-89), who is pro, and Niinning ("Renaissance"), who is con. Lanser ("[Im]plying") offers a bal
anced view and Phelan responds specifically to N?nning's critique of the term ("Can Readers").
Whether or not one adopts the model of an implied author as a reading strategy, one has to infer a
point of view other than the narrator's in order to attribute unreliability.

11. This list is based in part on the contributions of Gaby Allrath and Dagmar Busch to N?nning's mono
graph. The similarity between it and Booth's list of signals for textual irony is evident.

12. N?nning cites Culler on this process: "At the moment when we propose that a text means something
other than what it appears to say we introduce, as hermeneutic devices which are supposed to lead us
to the truth of the text, models which are based on our expectations about the text and the world"
(157; qtd. in Unreliable 24).

13. Fludernik points out that personified narrators are textual illusions: "the narrative instance Erz?hler is
a discourse-theoretical construct and not an empirical reality" {Fictions 453).

14. Manfred Jahn, Dornt Cohn, and Tamar Yacobi explore the possibility of heterodiegetic narrator unre
liability in recent essays, as does Cohn in The Distinction of Fiction, chapter 8. I would agree with
Cohn and Yacobi that the same inferences about character, trustworthiness, and personality can be ap
plied to narrators who do not take part in their stories. The degree to which these disembodied voices
appear to be part of full-fleshed characters determines the reader's perception of whether they can be
thought of as fallible and trustworthy. The less personalized the narrative voice is, I would argue, the
more inappropriate it is to infer unreliability. For instance, in many modern and postmodern texts in
which there is no apparent narrator viewpoint, such as in Samuel Beckett's short stories in Six
Residua, questions of reliability become mute.

15. Phelan makes explicit the fluid movement between the reader's analysis of character (what I call per
sonality) and the narrator's discourse: "Thus, interpreters will examine the homodiegetic narrator's
character?including such aspects of character as motives, values, beliefs, interests, psychology, race,
class, and gender (to the extent these matters can be inferred from events and descriptions)?for clues
to the narration and the character's narration for clues to the character" {Narrative 111).

16. In his earlier lecture on unreliability, N?nning does describe a separation between fallible and unreli
able narrators and considers adopting this differentiation ("Reconceptualizing" 68). Yet in the intro
duction to the monograph, he argues that this distinction fails to clear up the problems of imprecision
inherent in Booth's definition and its usage {Unreliable 12).

17. It goes without saying that as textual illusions, these narrators appear to inhabit their fictional worlds.

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 107

18. The fundamental attribution error refers to the general tendency to attribute the causes of behaviors
to personal dispositions rather than external, situational circumstances. The actor-observer bias de
scribes how actors tend to look for the causes of their own behaviors in external circumstances rather
than in their own personal characteristics. A similar bias occurs in causal perceptions about the be
haviors of members and non-members of groups. H. H. Kelley's notion of the covariation principle
proposes a model for making attributions about the causes of given behaviors on the basis of different
classes of information (distinctiveness, consensus, and consistency).

19. Similarly, Dorrit Cohn differentiates between "a factual kind of unreliability" and "an ideological
kind that is attributed to the narrator who is biased or confused" ("Discordant" 307). She concentrates
on the latter in her essay, calling this type of narration "discordant" to denote the lack of harmony be
tween the narrator and the story she tells.

20. I owe this insight to James Phelan and his critical reading of this text.

WORKS CITED
Allrath, Gaby. " 'But why will you say that I am mad?' Textuelle Signale f?r die Ermittlung von unreli
able narration'.' In Unreliable Narration: Studien zur Theorie und Praxis unglaubw?rdigen Erz?h
lens in der englischsprachigen Erz?hlliteratur, edited by Ansgar N?nning, 59-80. Trier: WVT, 1998.

Aronson, E., and, J. Mills. "The effect of severity of initiation on liking for a group." Journal of Abnor
mal and Social Psychology 59 (1959): 177-81.

Beardsley, Monroe, and W. K. Wimsatt. "The Intentional Fallacy." 1946. In 20th Century Literary Crit
icism, edited by David Lodge, 332-45. London: Longman, 1972.

Booth, Wayne. The Rhetoric of Fiction. Chicago: Chicago Univ. Press, 1961.

-. A Rhetoric of Irony. Chicago: Chicago Univ. Press, 1974.

Busch, Dagmar. "Unreliable Narration aus narratologischer Sicht: Bausteine f?r ein erz?hltheoretisches
Analyseraster." In Unreliable Narration: Studien zur Theorie und Praxis unglaubw?rdigen Erz?h
lens in der englischsprachigen Erz?hlliteratur, edited by Ansgar N?nning, 41-58. Trier: WVT,
1998.

Chandler, Raymond. The Big Sleep. London: Penguin, 1939.

Chatman, Seymour. Coming to Terms: The Rhetoric of Narrative in Fiction and Film. Ithaca: Cornell
Univ. Press, 1990.

Clemens, Samuel Langhorne. The Adventures of Huckleberry Finn. 1876. London: Penguin, 1994.

Cohn, Dorrit. "Discordant Narration." Style 34.2 (2000): 307-16.

-. The Distinction of Fiction. Baltimore: Johns Hopkins Univ. Press, 1999.

Collins, Wilkie. The Woman in White. 1868. London: Penguin, 1994.

Conrad, Joseph. Lord Jim. 1900. London: Penguin, 1986.

Culler, Jonathan. Structuralist Poetics: Structuralism, Linguistics and the Study of Literature. London:
Routledge, 1975.

Defoe, Daniel. The Fortunes and Misfortunes of the Famous Moll Flanders. 1722. Edited by Edward
Kelly. Norton: New York, 1973.

Dostoevsky, Fyodor. Notes from the Underground. 1864. Translated and edited by M. R. Katz. New York:
Norton, 1989.

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 108 Greta Olson

Fludernik, Monika. "Defining (In)sanity: The Narrator of The Yellow Wallpaper and the Question of Un
reliability." In Grenz?berschreitungen: Narratologie im Kontext / Transcending Boundaries: Narra
tology in Context, edited by Walter Gr?nzweig and Andreas Solbach, 75-95. T?bingen: Gunter Narr
Verlag, 1999.

-. The Fictions of Language and the Languages of Fiction: The Linguistic Representation of Speech
and Consciousness. London: Routledge, 1993.

-. Towards a 'Natural'Narratology. London: Routledge, 1996.

Goetsch, Paul. "Defoes 'Moll Flanders' und der Leser." Germanisch-Romanische Monatsschrift 30
(1980): 271-88.
Grice, H. Paul. "Further Notes on Logic and Conversation." 1975. In Syntax and Semantics 9: Pragmat
ics, edited by Peter Cole and Jerry L. Morgan, 113-27. New York: Academic Press, 1978.

-. "Logic and Conversation." 1967. In Syntax and Semantics 3: Speech Acts, edited by Peter Cole
and Jerry L. Morgan, 41-58. New York: Academic Press, 1975.

Hof, Renate. Das Spiel des 'unreliable narrator': Aspekte unglaubw?rdigen Erz?hlens im Werk von
Vladimir Nabakov. Munich: Fink, 1984.

Jahn, Manfred. "Das Ph?nomen der Unzuverl??lichkeit in der Erz?hlsituation." In Unreliable Narra
tion: Studien zur Theorie und Praxis unglaubw?rdigen Erz?hlens in der englischsprachigen Erz?h
lliteratur, edited by Ansgar N?nning, 81-106. Trier: WVT, 1998.

Kelley, H. H. Attribution Theory in Social Psychology. In Nebraska Symposium on Motivation, Vol. 15,
edited by David Levine, 192-238. Lincoln: Nebraska Univ. Press, 1967.

Lanser, Susan S. "(Im)plying the Author." Narrative 9 (2001): 153-60.

-. The Narrative Act: Point of View in Prose Fiction. Princeton: Princeton Univ. Press, 1981.

Nabokov, Vladimir. "On a Book Entitled Lolita." In Lolita, 313-19. New York: G. P. Putnam's Sons,
1955.

Niinning, Ansgar. "Reconceptualizing the Theory and Generic Scope of Unreliable Narration." In Recon
ceptualizing Trends in Narratological Research, edited by John Pier, 63-84. Tours: Tours Univ.
Press, 1999.

-. "Renaissance eines anthropomorphisierten Passepartouts oder Nachruf auf ein literaturkritisches

Phantom? ?berlegungen und Alternativen zum Konzept des 'implied author.' " Deutsche Viertel
jahresschrift f?r Literaturwissenschaft und Geistesgeschichte 67 (1993): 1-25.

-. "Unreliable Narration zur Einf?hrung: Grundz?ge einer kognitiv-narratologischen Theorie und

Analyse unglaubw?rdigen Erz?hlens." In Unreliable Narration: Studien zur Theorie und Praxis
unglaubw?rdigen Erz?hlens in der englischsprachigen Erz?hlliteratur, edited by Ansgar N?nning,
3-40. Trier: WVT, 1998.

N?nning, Vera. "Die Historische Variabilit?t von Werten und Normen: The Vicar ofWakefield als Testfall
f?r eine kulturgeschichtliche Erz?hlforschung." Unreliable Narration: Studien zur Theorie und
Praxis unglaubw?rdigen Erz?hlens in der englischsprachigen Erz?hlliteratur, edited by Ansgar
N?nning, 257-85. Trier: WVT, 1998.

Phelan, James. "Can Readers Infer What Authors Imply?" Paper presented at the annual convention of
the Modern Language Association, New Orleans, La., December 2001.

-. Narrative as Rhetoric: Technique, Audiences, Ethics, Ideology. Columbus: Ohio State Univ.
Press, 1996.

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms
 Fallible and Untrustworthy Narrators 109

Phelan, James, and Mary Patricia Martin. "The Lessons of 'Weymouth' : Homodiegesis, Unreliability,
Ethics, and The Remains of the Day." In Narratologies: New Perspectives on Narrative Analysis,
edited by David Herman, 88-109. Columbus: Ohio State Univ. Press, 1999.

Poe, Edgar Allan. "The Tell-Tale Heart." 1843. In Collected Works of Edgar Allan Poe. Vol. Ill Tales and
Sketches 1843-1849, edited by Thomas Mabbott, 792-99. Cambridge: Harvard Univ. Press, 1978.

Riggan, William. Picaros, Madmen, Naifs, and Clowns: The Unreliable First-Person Narrator. Norman:
Oklahoma Univ. Press, 1981.

Sperber, Dan, and Deirdre Wilson. "Irony and the Use-Mention Distinction." In Radical Pragmatics,
edited by Peter Cole, 257-317. New York: Academic Press, 1981.

Stanzel, Franz. A Theory of Narrative. Translated by Charlotte Goedsche. Cambridge: Cambridge Univ.
Press, 1984.

Wall, Kathleen. "The Remains of the Day and Its Challenges to Theories of Unreliable Narration." Jour
nal of Narrative Technique 24 (1994): 18-24.

Yacobi, Tamar. "Package Deals in Fictional Narrative: The Case of the Narrator's Unreliability." Narra
tive 9 (2001): 223-29.

This content downloaded from 132.194.27.171 on Tue, 17 Jan 2017 18:41:33 UTC
All use subject to http://about.jstor.org/terms