1 INTRODUCTION

'Begin at the beginning,' the King said,

gravely, 'and go on till you come to
the end: then stop.'
Lewis Carroll,
Alice in Wonderland

When you coin a term, it ought to mark

a real species, and a specific difference;
otherwise you get empty, frivolous
verbiage.
Aristotle,
The Rhetoric

Among the many pressing needs of literary theory-poetics

in the broad sense-is a 'reasoned account of the structure of
narrative, the elements of storytelling, their combination and
articulation. The task is delineated by Aristotle, but delineated
only; the Poetics opens more questions than it answers. There is
a distinguished tradition of Anglo-American studies on narra-
tive: Henry James, Percy Lubbock, Wayne Booth. Less known in
this country but of great importance is a flood of recent work
from Russia and France.
The Russian formalist tradition, especially the work of Vlad-
imir Propp, emphasized simple narratives: folk tales, 1 myths,
romans poliders. But modern narrative fiction entails additional
complexities of structure. The rigid homogeneity of plot and
simplicity of characterization found in the Russian fairy tale are
obviously not typical of many modern narratives. Still, much
can be learned from these investigations, particularly about the
theory of plot and the necessity of separating narrative structure
1. See Morfologia Skazi (Leningrad, 1928), translated by Laurence Scott as
The Morphology of the Folktale, 2d ed. (Austin, 1970). A summary of Propp's
analysis appears in the artide "Les Transfonnations des contes fantastiques,"
Throrie de la litterature, ed. Tzvetan Todorov (paris, 1966), pp. 234-262; and also
Oaude Bremond, "Le message narratif," Communialtions, 4 (1964), 4-10.
16 STORY AND DISCOURSE

from any of its mere manifestations, linguistic or otherwise.

Certain disadvantages must also be considered, particularly
classificatory reductivism. On balance, what constitutes a viable
and modern narrative theory? 2
2. The largest selection of Russian Formalist writings is a recent German
translation entitled Texte der Russischen Formalisten, ed. Juri Striedter (Munich,
1969), in two volumes. A French translation of a smaller selection has been
made by Tzvetan Todorov, under the title Theone de la litterature (paris, 1966),
which contains important articles by Eichenbaum, Shldovsky, Jakobson, Vino-
gradov, Tynianov, Brlk, and Propp, as well as a reminiscent preface by Jakobsen
and introductory essay by Todorov. A still smaller selection is in Russian Formal-
ist Criticism, ed. Lee Lemon and Marion Reis (Uncoln, Neb., 1965). See also
L. Matejka and K. Pormorska, eds., ReJJdings in Russian Poetics (Cambridge,
Mass., 1971), and L. Matejka and 1. Titunik, eds., Semiotics of Art: Prague School
Contn11utions (Cambridge, Mass., 1976).
The relevant works of the French Structuralists are now numerous; the fol-
lowing is only a selection: Oaude Levi-Strauss, "La structure et 1a forme," in
the Italian translation of Propp, Morphologia della fiaba, ed. Gian Bravo (Turin,
1966); Levi-Strauss, Anthropologie structurale (paris, 1958); Claude Bremond, "Le
Message narratif," Communications, 4 (1964), 4-32, and "La Logique des pos-
sibles narratifs," Communications, 8 (1966), 60-76, both included in Bremond,
Logique du redt (Paris, 1973); Roland Barthes, "Introduction a l' analyse structurale
des recits," Communications, 8 (1966), 1-27, English translation by Lionel Duisit
appeared in New Literary History (1975), 237-272; A.-J. Greimas, Semantique
structurale (paris, 1966); Gerard Genette, "Frontieres du recit," Communications,
7 (1966), 152-163, and in particular "Discours du recit," in Figures m (paris,
1972), 67-282 (to be published in an English translation by Cornell University
Press), 'Vhich. has had a great influence on my own work; Tzvetan Todorov,
''Les Categories du recit litteraire," Communications, 8 (1966), 125-151, included
in Utterature et signification (paris, 1967), "Paetique," in Oswald Ducrot, ed.,
Qu'est-ce que Ie structuraIisme? (paris, 1968), pp. 99-166, "Structural AnalysiS of
Narrative," Navel, 3 (Fa111969), 70-76, and Grammaire du Decameron (The Hague,
1970); Christian Me~ "Remarques pour une phenomenologie du narratif," in
his Essais sur la signification au dnima, I (paris, 1968), 25-35, now in English,
Film Language, trans. Michael Taylor (New York, 1974), pp. 16-30; and Roland
Barthes, S/Z, now in English, trans. Richard Miller (New York, 1974).
In Germany and Holland a school of semiologistsapproaches narrative anal-
ysis from the point of view of "text grammar," attempting to extend the proce-
dures of modem linguistics beyond the sentence to larger units of .discourse.
See Teun Van Dijk, Some Aspects of Text Grammars (The Hague, 1972), which
contains an extremely rich bibliography, and his journal, Poetics.
In America, two books have appeared reviewing these developments, one by
Robert Scholes, Structuralism in Literature: An Introduction (New Haven, 1974),
and Jonathan Culler, Structuralist Poetics (Ithaca, N.Y., 1975). Gerald Prince has
attempted a formalization of "stories" along Chomskian lines in A Grammar of
Stories (The Hague, 1973). For my earlier thinking on the subject, see "New
Ways of Narrative Analysis," Language and Style, 2 (1968), 3-36, and "The Struc-
ture of Fiction," University Review, 37 (1971), 199-214.
 INTRODUCTION 17

To begin, let me sketch the general conception of literature

and of art in terms of which the present theory is conceived.

Narrative and Poetics

Formalists and structuralists argue that it is not the literary
text itself that is the subject of poetics but rather-to use Roman
Jakobson's phrase-its "literariness." The question for poetics
(unlike literary criticism) is not "What makes Macbeth great?"
but rather 'What makes it a tragedy?" A statement by Tzvetan
Todorov sums up the position very well:
Literary theory [poetics, poetique] is ... distinct, as is any science, from
the description of literary works. For to describe is to try to obtain, on
the basis of certain theoretical premises, a rationalized representation
of the object of study, while to present a scientific work is to discuss
and transform the theoretical premises themselves, after having experi-
enced the object described. Description is, in literature, a reasoned
resume; it must be done in such a way that the principal traits -of the
object are not omitted and indeed emerge even more evidently. De-
scription is paraphrase that e?ffiibits (rather than conceals) the logical
principle of its own organization. Any work is, in this sense, its own
best possible description: entirely immanent and exhaustive. If we can-
not satisfy ourselves with description it is because our principles differ
too much.
We have seen develop in our own time more and more perfected
techniques for describing the literary work. All the constitutive and
pertinent elements of a poem, for example, will be identified: then their
relative disposition, and finally a new presentation of the same poem,
a presentation that allows us to penetrate more deeply into its meaning.
But description of a work can never lead us to modify our -premises;
it can only illustrate them.
The procedure of the literary theorist ["poetician'1 is quite different.
If he analyzes a poem, it is not to illustrate his premises (or, if he does,
he does so only once, and then for instructional reasons), but to draw
from this analysis conclusions that complete or modify the underlying
premises; in other words, the object of literary theory is not works but
literary discourse, and literary theory will take its place beside the other
sciences of discourse which will have to be established for each of the
kinds ....
Literary theory cannot avoid literature on the way to its own proper
discursive goal; and at the same time it is only in going beyond the
concrete work that it can reach that goal. 3
3. Litterature et signification, p. 7.
18 STORY AND DISCOURSE

On this view literary theory is the study of the nature of

literature. It is not concerned with the evaluation or description
of any particular literary work for its own sake. It is not literary
criticism but the study of the givens of criticism, the nature of
literary objects and their parts. It is, as Rene Wellek and Austin
Warren point out, an "organon of methods." 4
Like modem linguistics, literary theory might well consider a
rationalist and deductive approach rather than the usual empiri-
cist one. It should assume· that definitions are to be made, not
discovered, that the deduction of literary concepts is more testa-
ble and hence more persuasive than their induction. Poetics
should construct "a theory of the structure and functioning of
literary discourse, a theory which presents a set [tableau] of pos-
sible literary objects, such that existing literary works appear as
particular realized cases." 5 Aristotle provides a precedent; the
Poetics is nothing less than a theory of the properties of a certain
type of literary discourse. Northrop Frye is outspokenly deduc-
tive in Anatomy of Criticism. We need not expect actual works to
be pure examples of our categories. The categories plot the ab-
stract network upon which individual works find their place. No
individual work is a perfect specimen of a genre-novel or comic
epic or whatever. All works are more or less mixed in generic
. character;-
To put it another way, genres are constructs or composites of
features. The novel and the drama, for example, require features
like plot and character, which are not essential to the lyric poein;
but all three may utilize the feature of figurative language.
Further, works ordinarily mix features in different dosages: both
Pride and Prej~dice and Mrs. Dalloway contain examples of in-
direct free styie, but the dosage in Mrs. Dalloway is much larger,
making it a qualitatively different kind of novel. We should not
be disconcerted by the fact that texts are inevitably mixed; in
that respect they resemble most organic objects. It is their gen-
eral tendencies that form the subject of rational inquiry.
Narrative theory has no critical axe to grind. Its objective
4. Theory of Literature, 3d ed. (Harmondsworth. England, 1963). p. 19.
5. Todorov, ''Poetique,'' p. 103.
 INTRODUCTION 19

is a grid of possibilities, through the establishment of the mini-

mal narrative constitutive features. It plots individual texts on
the grid and asks whether their accommodation requires adjust-
ments of the grid. It does not assert that authors should or
should not do so-and-so. Rather, it poses a question: What can
we say about the way structures like narrative organize them-
selves? That question raises subsidiary ones: What are the ways
in which we recognize the presence or absence of a narrator?
What is plot? Character? Setting? Point of view?

Elements of a Narrative Theory

Taking poetics as a rationalist discipline, we may ask, as does
the linguist about language: What are the necessary components
-and only those-of a narrative? Structuralist theory argues
that each narrative has two parts: a story (histoire), the content
or chain of events (actions, happenings), plus what may be
called the existents (characters, items of setting); and a discourse
(discours), that is, the expr~ssion, the means by which the con-
tent is communicated. In simple terms, the story is the what in
a narrative that is depicted, discourse the how. The follOwing
diagram suggests itself:

Actions
Events
{ Happenings
Story
{ Characters
Narrative Text { Existents {
Setting
Discourse

This kind of distinction has of course been recognized since the

Poetics. For Aristotle, the imitation of actions in the real world,
praxis, was seen as forming an argument, logos, from which
were selected (and possibly rearranged) the units that formed
the plot, mythos.
The Russian formalists, too, made the distinction, but used
only two terms: the "fable" (tabula), or basic story stuff, the sum
total of events to be related in the narrative, and, conversely,
20 STORY AND DISCOURSE

the 'plot' (sjuzet), the story as actually told by linking the events
together. 6 To fonnalists, fable is "the set of events tied together
which are communicated to us in the course of the work," or
"what has in effect happened"; plot is "how the reader becomes
aware of what happened," that is, basically, the "order of the
appearance (of the events) in the work itself," 7 whether normal
(abc), flashed-back (acb), or begun in medias res (be).
French structuralists also incorporate these distinctions.
Oaude Bremond argues that there exists a
... layer of autonomous significance, endowed with a structure that
can be isolated from the whole of the message: the story [recit]. So any
sort of narrative message (not only folk tales), regardless of the process
of expression which it uses, manifests the same level in the same way.
It is only independent of the techniques that bear it along. It may be
transposed from one to another medium without losing its essential
properties: the subject of a story may serve as' argument for a ballet,
that of a novel can be transposed to stage or screen, one can recount
in words a film to someone who has not seen it. These are words we
read, images we see, gestures we decipher, but through them, it is a
story that we follow; and this can be the same story. That which is
narrated [raconte] has its own proper significant elements, its story-
elements [racontants]: these are neither words, nor images, nor ges-
tures, but the events, situations, and behaviors signified by the words,
images, and gestures. 8
This transposability of the story is the strongest reason for
arguing that narratives are indeed structures independent of
any medium. But what is a structure, and why are we so ready
to classify the narrative as being one? In the best short introduc-
tion to the subject, Jean Pia get shows how disciplines as various
as mathematics, social anthropology, philosophy, linguistics,
and physics have utilized the conception of structure, and how
in each case,' three key notions have been invoked: wholeness,
6. Victor Erlich, Russian Formalism: History, Doctrine, 2d ed. (The Hague,
1965), 240-241.
7. Boris Tomashevsky, Teorija literatury (Poetika) (Leningrad, 1925). The rele-
vant section, "Thematique," appears in Todorov, ed., Theorie de la litterature,
pp. 263-307 and in Lemon and Reis, eds., Russian Formalist Criticism, pp. 61-98.
The quotations here translate the French text in Todorov, ed., p. 268. The dis-
tinction between fabula and sjuiet appears on page 68 of Lemon and Reis.
8. "Le message narratif," p. 4.
 INTRODUCTION 21

transformation, and self-regulation. Any group of objects with-

out these characteristic properties is merely an aggregate, not a
structure. Let us examine narratives in terms of the three prop-
erties to see whether they are in fact structures.
Clearly a narrative is a whole because it is constituted of ele-
ments--events and existents-that differ from what they consti-
tute. Events and existents are single and discrete, but the nar-
rative is a sequential composite. Further, events in the narra-
tive (as opposed to the chance compilation) tend to be related
or mutually entailing. If we were to extract randomly from cock-
tail chatter a set of events that happened at different times and
different places to different persons, we would clearly not have
a narrative (unless we insisted upon inferring one--a possibility
- I will discuss below). The events in a true narrative, on the
other hand, "come on the scene as already ordered," in Piaget's
phrase. Unlike a random agglomerate of events, they manifest
a discernible organization.
Second, narratives entail both transformation and self-regula-
tion. Self-regulation means that the structure maintains and
closes itself, in Piaget's words, that "transformations inherent
in a structure never lead beyond the system but always en-
gender elements that belong to it and preserve its laws. . . . In
adding or subtracting any two whole numbers, another whole
number is obtained, and one which satisfies the laws of the
'additive group' of whole numbers. It is in this sense that a
structure is 'cIosed."'9 The process by which a narrative event
is expressed is its "transformation" (as in linguistics an element
in the "deep structure" must be "transformed" in order to occur
in the surface representation). However this transformation
takes place--whether, for example, the author elects to order
the reporting of events according to their causal sequence or
to reverse them in a flashback effect-only certain possibilities
can occur. Further, the narrative will not admit events or other
kinds of phenomena that do not "belong to it and preserve its
laws." Of course certain events or existents that are not immedi-
9. Jean Piaget, Structuralism, trans. Chaninah Maschler (New York, 1970),
p.14.
22 STORY AND DISCOURSE

ately relevant may be brought in. But at some point their rele-
vance must emerge, otherwise we object that the narrative is
"ill-formed. "
So the evidence for calling narratives "structures" seems
strong enough, even in the rigorous sense of the structuralists.
So far we have spoken only of the story component of nar-
ratives. Narrative discourse, the "how," in tum divides into two
subcomponents, the narrative form itself-the structure of nar-
rative transmission-and its manifestation-its appearance in a
specific materializing medium, verbal, cinematic, balletic, musi-
cal, pantomimic, or whatever. Narrative transmission concerns
the relation of time of story to time of the recounting of story,
the source or authority for the story: narrative voice, "point of
view," and the like. Naturally, the medium influences the
transmission, but it is important for theory to distinguish the
two.

Is Narrative a Semiotic Structure?

Narrative is a structure: we may go on to ask if it is indepen-
dently meaningful, that is, conveys a meaning in and of itself,
separately from the story it tells. Linguistics and semiotics, the
general science of signs, teach us that a simple distinction be-
tween expression and content is insufficient to capture all ele-
ments of the communicative situation. Crosscutting this distinc-
tion, there is that between substance and form. The following
diagram is familiar to everyone who. has read Ferdinand de
5aussure and Louis Hjelmslev:

Expression Content
Substance
Form

Units of the expression plane convey meanings, that is, units

of the content plane. In languages, the substance of expression
is the material nature of the linguistic elements, for example,
the actual sounds made by voices, or marks on paper. The sub-
,
INTRODUCTION 23

stance of content (or "meaning") is, on the other hand, lithe

whole mass of thoughts and emotions common to mankind in-
dependently of the language they speak." 10 Now each language
(reflecting its culture) divides up these mental experiences in
different ways. Hence the form of the content is "the abstract
structure of relationships which a particular language imposes
... on the same underlying substance." 11 The vocal apparatus
is capable of an immense variety of sounds, but each language
selects a relatively small number through which to express its
meanings. English,. for example, makes a three-way distinction
between high-front vowel sounds, as in beat, bit, and bait,
whereas most other European languages have only two units
. within the same phonic range; in French, for instance, there are
the vowels in qui and quai, but nothing between. So linguists
distinguish the substance of (phonic) expression, the myriad
audible sounds utilized by a given language, from the form of
expression, the small set of discrete phonemes or range of
phonic oppositions characteristic of it.
If narrative structure is indeed semiotic-that is, communi-
cates meaning in its own right, over and.above the paraphrase-
able contents of its story-it should be explicable in terms of
the quadripartite array above. It should contain (1) a form and
substance of expression, and (2) a form and substance of con-
tent.
What in na:t:rative is the province of expression? Precisely
the narrative discourse. Story is the content of the narrative
expression, while discourse is the form of that expression. We
must distinguish between the discourse and its material mani-

10. John Lyons, Introduction to Theoretical Linguistics (Cambridge, 1969), p. 56.

11. Ibid., p. 55. John Lyons' example: the English word "brother-in-law" can
be translated into Russian as zjatj, shurin, svojak, or deverj; "and ... zjatj must
sometimE!S be translated as son-in-law. From this it should not be concluded,
however, that the word zjatj has two meanings. and that in one of its meanings
it is equivalent to the other three. All four words in Russian have a different
meaning. It so happens that Russian brings together (under zjatj) both sister's
husband and daughter's husband, but distinguishes wife's brother (shurin),
wife's sister's husband (svojak) and husband's brother (deverj). So there is really
no word which means 'brother-in-law' in Russian, just as there is no word
which means 'zjatj' in English."
24 STORY AND DISCOURSE

festation-in words, drawings, or whatever. The latter is clearly

the substance of narrative expression, even where the manifesta-
tion is independently a semiotic code. But commonly codes
serve other codes as substance; for instance, Barthes has shown
that in the world of fashion, the codes of clothing "enjoy the
status of systems only in so far as they pass through the relay
of language, which extracts their signifieds (in the forms of
usages or reasons}." He concludes that "it is ... difficult to
conceive a system of images and objects whose signifieds can
exist independently of language." 12 In precisely the same way,
narratives are langues conveyed through the paroles of concrete
verbal or other means of communication.
As for narrative content, it too has a substance and a form.
The substance of events and existents is the whole universe,
or, better, the set of possible objects, events, abstractions, and
so on that can be "imitated" by an author (film director, etc.).
Thus:
Expression Content
Media insofar as they Representations of ob-
can communicate sto- jects & actions in real
ries. (Some media are & imagined worlds
Substance semiotic systems in that can be imitated in
their own right.) a narrative medium, as
filtered through the
codes of the author's
society.
Narrative discourse Narrative story com-
(the structure of narra- ponents: events, exis-
tive transmission) con- tents, and their con-
Form sisting of elements nections.
shared by narratives
in any medium what-
soever.

But what does it mean practically to say that narrative is a

meaningful structure in its own right? The question is not "What
12. Roland Barthes, Elements of Semiology, trans. Annette Lavers and Colin
Smith (Boston, 1967), p. 10.
 INTRODUCTION 25

does any given story mean?" but rather "What does narrative
itself (or narrativizing a text) mean?" The signifies or signifieds
are exactly three-event, character, and detail of setting; the
signifiants or signifiers are those elements in the narrative state- .
ment (whatever the medium) that can stand for one of these
three, thus any kind of physical or mental action for the first,
any person (or, indeed, any entity that can be personalized) for
the second, and any evocation of place for the third. We are
justified, I believe, in arguing that narrative structure imparts
meanings, of the three kinds listed above, precisely because it
can endow an otherwise meaningless ur-text with eventhood,
characterhood, and settinghood, in a normal one-to-one stand-
ing-for relationship. There are animated cartoons in which a
completely contentless object is endowed with characterhood,
that is, takes on the meaning "character" because it engages in
a suitably anthropomorphic action (that is, a movement on the
screen that is conceived as an instance of human movement).
An example is the film by !=huck Jones called The Dot and the
Line, whose plot runs roughly as follows: a line courts a dot,
but the dot is going around with a squiggle, a sort of hip joke-
ster. Whatever we think of the dot and the line as geometric
familiars, the squiggle is surely without meaning until it moves.
That is, as a drawn object projected on the screen, no one would
identify it as anything but a random assemblage of swirling
lines. In context, however, in its visible movement-relations
with the dot and the line, it becomes a character. (It is true that
a narrator tells the story through voice-over, but the story would
be comprehensible even if seen without the sound track.)
This book is essentially aboufthe form of narrative rather than
its substance, but substance will be discussed where it seems to'
facilitate an understanding of narrative form. For instance it is
clear that verbal narratives express narrative contents of time
summary more easily than do cinematic narratives, while the
latter more easily show spatial relations. A purely gratuitous
visual link may tie together two shots (the line of the roof-
support in Charles Foster Kane's childhood home in Citizen Kane
"turns into" a string wrapping a Christmas package given him
by his coldhearted guardian; the sweep of the curve of a Holly-
26 STORY AND DISCOURSE

wood starlet's body on top of St. Peter's in La Dolce Vita "turns

into" the sweep of a saxophone braying in an outdoor night-
club).
The above considerations prompt a redrawing of our first
diagram:

Events {ActiOnS
Happenings Form
of
Characters Content
Existents
{ Settings

People, things, etc., as pre- } Substance

processed by the author's = of
cultural codes . Content
Narrative
Form
Structure of
narrative transmission }= of
Expression
Discourse
(Expression) Verbal }.
Cinematic Substance
Manifestation Balletic = of
{ Pantomimic Expression
etc.

Manifestation and Physical Object

Story, discourse, and manifestation must further be distin-
guished from the mere physical disposition of narratives-the
actual print in books, movements of actors or dancers or mario-
nettes, lines on paper or canvas, or whatever.
This issue is resolved by phenomenological aesthetics, partic-
ularly by Roman Ingarden, who has established the funda-
mental difference between the "real object" presented to us in
museums, libraries, the theater, and so on, and the "aesthetic
object." 13 The real object is the thing in the outside world-the
13. Roman lngarden, "Aesthetic Experience and Aesthetic Object," in
Nathaniel Lawrence and Daniel O'Conner, eds., Readings in Existential Phenom-
ellology (Englewood Cliffs, N.J., 1967),304.
 INTRODUCTION 27

piece of marble, the canvas with pigment dried on it, the air-
waves vibrating at certain frequencies, the pile of printed pages
sewn together in a binding. The aesthetic object, on the other
hand, is that which comes into existence when the observer
experiences the real object aesthetically. Thus it is a construc-
tion (or reconstruction) in the observer's mind. Aesthetic ob-
jects may exist in the absence of a real object. One can have an
aesthetic experience through purely fictitious objects; for ex-
ample, we may "only imagine the 1etters' or the corresponding
sound, e.g., when we are repeating a poem from memory."
Thus the material book (or whatever) is not"a literary work, but
only a means to 'fix' the work, or rather to make it accessible
to the reader." To a certain point, the physical condition of a
book (or other artifact) does not affect the nature of the aesthetic
object fixed by it: David Copperfield remains David Copperfield
whether it is read in an elegant library edition or a dirty, water-
stained paperback version. Further, mere reading is not an aes-
thetic experience, just as merely looking at a statue is not one.
They are simply preliminary to the aesthetic experience. The
perceiver must at some point mentally construct the "field" or
"world" of the aesthetic object.
The aesthetic object of a narrative is the story as articulated by
the discourse, what Susanne Langer would call the "virtual"
object of the narrative. 14 A medium-language, music, stone,
paint and canvas, or whatever-actualizes the narrative, makes
it into a real object, a book, a musical composition (vibrating
sound waves in an auditorium or on a disc), a statue, a painting:
but the reader must unearth the virtual narrative by penetrating
its medial surface. (See the discussion below of "reading"
versus "reading out.") .

Narrative Inference, Selection, and Coherence

If discourse is the class of all expressions of story, in what-
ever medium possible to it (natural language, ballet, "program"
music, comic strips, mime, and so on), it must be an abstract
class, containing only those features that are common to all
14. Susanne Langer, Feeling and Form (New York, 1953), p. 48, and passim.
The section on narrative is on pages 260-265.
28 STORY AND DISCOURSE

actually manifested narratives. The principal features are order

and selection. The first I have already spoken of; the second
is the capacity of any discourse to choose which events and
objects actually to state and which only to imply. For example,
in the "complete" account, never given in all its detail, the
"ultimate argument," or logos, each character obviously must
first be born. But the discourse need not mention his birth, may
elect to take up his history at the age of ten or twenty-five or fifty
or whenever suits its purpose. Thus story in one sense is the
continuum of events presupposing the total set of all conceiv-
able details, that is, those that can be projected by the normal
laws of the physical universe. In practice, of course, it is only
that continuum and that set actually inferred by a reader, and
there is room for difference in interpret~tion.
A narrative is a communica tion; hence, it presupposes two
parties, a sender and a receiver. Each party entails three differ-
ent personages. On the sending end are .the real author, the
implied author, and the narrator (if any); on the receiving end,
the real audience (listener, reader, viewer), the implied audi-
ence, and the narratee. (These distinctions will ~ amplified in
Chapter 4.) .
The sense modality in which narrative operates may be either
visual or auditory or both. In the visual category are nonverbal
narratives (painting, sculpture, ballet, pure or "unbubbled"
comic strips, mime, etc.), plus written texts. In the auditory
category are bardic chants, musical narratives, radio plays, and
other oral performances. But this distinction conceals an impor-
tant commonality between written and oral texts. All written
texts are realizable. orally; they are not being performed but
could be at any moment. That is, they are innately susceptible
of performance.
Whether the narrative is experienced through a performance
or through a text, the members of the audience must respond
with an interpretation: they cannot avoid participating in the
transaction. They must fill in gaps with essential or likely
events, traits and objects which for various reasons have gone
unmentioned. If in one sentence we are told that John got
 INTRODUCTION 29

dressed and in the next that he rushed to an airport ticket

counter, we surmise that in the interval occurred a number of
artistically inessential yet logically necessary events: grabbing
his suitcase, walking from the bedroom to the living room and
out the front door, then to his car or to the bus or to a taxi,
opening the door of the car, getting in, and so on. The audi-
ence's capacity to supply plausible details is virtually limitless,
as is a geometer's to conceive of an infinity of fractional spaces
between two points. Not, of course, that we do so in normal
reading. We are speaking only of a logical property of narra-
tives: that they evoke a world of potential plot details, many of
which go unmentioned but can be supplied. The same is true
of character. We may project any number of additional details
about characters on the basis of what is expressly said. If a
girl is portrayed as "blue-eyed," "blonde," and "graceful," we
assume further that her skin is fair and unbleInished, that she
speaks with a gentle voice, that her feet are relatively small, and
so on. (The facts may be other, but we have to be told so, and
our inferential capacity remains undaunted. Indeed, we go on to
infer a variety of details to account for the "discrepancy. ")
Thus there is a special sense in which narratives may be said
to select. In nonnarrative paintings, selection means the separa-
tion of one portion from the rest of the universe. A painter or
photographer will frame this much imitated nature, and the rest
is left beyond the frame. Within that frame the number of details
explicitly presented is a stylistic, rather than a general structural
question. A Dutch still life painter may include an exact repro-
duction of minutiae in the set-up before him, down to the small-
est dewdrop on a peach, while an Impressionist may dash off
a distant pedestrian with a single brush stroke. But a narrative,
as the product of a fixed number of statements, can never be
totally "complete," in the way that a photographic reproduction
is, since the number of plausible intermediate actions or prop-
erties is virtually infinite. In a highly realistic painting, what is
shown is determined by what was visible to the painter, and
that is a function of his distance from the depicted scene. Scale,
then, controls the number of details. But narratives are not re-
30 STORY AND DISCOURSE

stricted by spatial scale and undergo no such control: a visual

narrative, a comic strip or movie, can move from close to long
shot and return with no effort. And there is a virtually infinite
continuum of imaginable details between the incidents, which
will not ordinarily be expressed, but which could be. The author
selects those events he feels are sufficient to elicit the necessary
sense of continuum. Normally, the audience is content to accept
the main lines and to fill in the interstices with knowledge it
has acquired through ordinary living and art experience.
So far we have considered gaps common to all narratives
regardless of medium. But there is also a class of indetermina-
cies-phenomenologists call them Unbestimmtheiten-that arise
from the peculiar nature of the medium. The medium may spe-
cialize in certain narrative effects and not others. For instance,
the cinema can easily-and does routinely-present characters
without expressing the contents of their minds. It is usually
necessary to infer their thinking from what they overtly say and
do. Verbal narrative, on the other hand, finds such a restriction
difficult-even Ernest Hemingway, at such pains to avoid di-
rectly stating his characters' thoughts and perceptions, some-
times "slips." Conversely, verbal narrative may elect not to
present some visual aspect, say, a character's clothes. It remains
totally imbestimmt about them, or describes them in a general
way: "He was dressed in street clothes." The cinema, however,
cannot avoid a rather precise representation of visual detail.
It cannot "say," simply, "A man came into the room." He must
be dressed in a certain way. In other words clothing, unbestimmt
in verbal narrative, must be bestimmt in a film.
Another restriction on selection an~ inference is coherence.
Narrative existents must remain the same from one event to the
next. If they do not, some explanation (covert or overt) must
occur. If we have a story like "Peter fell ill. Peter died. Peter
was buried," we assume that it is the same Peter in each case.
In E. M. Forster's example, "The king died, and then the queen
died of grief," we assume that the queen was in fact the wife of
that king. If not, there would have to be some explanation of
the queen's death, for example, "Though she did not know him,
 INTRODUCTION 31

she died of the grief she felt for the decay of royal houses."
Some principle of coherence must operate, some sense that the
identity of existents is fixed and continuing. Whether or not the
events must also be causally linked is not so clear.
The drawing of narrative inferences by the reader is a low-
level kind of interpretation. Perhaps it doesn't even deserve the
name, since "interpretation" is so well established as a synonym.
for "exegesis" in literary criticism. This narrative filling-in is all
too easily forgotten or assumed to be of no interest, a mere reflex
action of the reading mind. But to neglect it is a critical mistake,
for this kind of inference-drawing differs radically from that re-
quired by lyric, expository, and other genres.

A Sketch of Narrative Structure

Narrative discourse consists of a connected sequence of nar-
rative statements, where "statement" is quite independent of the
particular expressive medium. It includes dance statement, lin-
guisticstatement, graphic,statement, and so on. (The nature of
the connection will be taken up in detail in Chapter 2.) "Narra-
tive statement" and "to state narratively" are used here as tech-
nical terms for any expression of a narrative element viewed
independently of its manifesting substance. The term has a
broad discoursive sense, not a grammatical one. For example, a
narrative statement may be manifested by questions or com-
mands as well as by declarative constructions in natural lan-
guage.
Narratives are communications, thus easily envisaged as the
movement of arrows from left to right, from author to audience.
But we must distinguish between real and implied authors and
audiences: only implied authors and audiences are immanent
to the work, constructs of the narrative-transaction-as-text. The
real author and audience of course communicate, but only
through their implied counterparts. What is communicated is
story, the formal content element of narrative; and it is com-
municated by discourse, the formal expression element. The dis-
course is said to "state" the story, and these statements are of
two kinds-process and stasis-according to whether someone
 32 STORY AND DISCOURSE

did something or something happened; or whether something

simply existed in the story. Process statements are in the mode
of DO or HAPPEN, not as actual words in English or any natural
language (these form the substance of the expression), but as
more abstract expressional categories. Both the English sentence
"He stabbed himself" and a mime's plunging an imaginary
dagger into his heart manifest the same narrative process state-
ment. Stasis statements are in the mode of IS. A text that con-
sisted entirely of stasis statements, that is, stated only the
existence of a set of things, could only imply a narrative. Events
are either logically essential, or not ("kernels" versus "satel-
lites"). Further, they are either acts or actions, in which an exis-
tent is the agent of the event, or happenings, where the existent
is the patient. An existent, in turn, is either a character or an
element of setting, a distinction based on whether or not it per-
forms a plot-significant action. A stasis statement may com-
municate either or both of two aspects: the identity of an exis-
tent or one of its qualities, for example, traits (see Chapter 3).
A process statement may be said either to rerount or to enad
an event according to whether or not it is explicitly presented,
that is, uttered as such by a narrator. These distinctions were
already noted by the ancients. The difference between narration
proper; the recounting of an event (the subject of Chapter 5),
and enactment, its unmediated presentation (the subject of
Chapter 4), corresponds to the classical distinction between
. diegesis and mimesis (in Plato's sense of the word), or, in modern
terms, between telling and showing. Dialogue, of course, is the
preeminent enactment. The contrast between narration proper
and enactment is demonstrated in the two basic forms for de-
picting a character's speech-indirect versus direct: "John said
that he was tired" versus "'I'm tired' [said John]." The first· -
necessarily entails a person telling what John said, while the
second simply has John saying something-in the audience's
presence, so to speak.
Correspondingly, a stasis statement is either unmediated,
that is, it exposes, or mediated, that is, it presents. This is the dif-
ference between "John was angry" and "Unfortunately, John
was angry." Crosscutting this distinction is that of aspect: the
 INTRODUCTION 33

stasis statement may either identify ("John was a clerk") or

qualify ("John was angry").
Further, events may imply or index existents; and, vice versa,
existents may project events. For example, "John seduced Mary"
indexes "there is a character named John"· and "John is a
seducer"; while "John is a loser" projects "John has lost many
times and will continue to do so." Finally, one event may imply
another, one existent another: "John murdered Mary" implies
either "He was later caught" or "He escaped justice"; "John is a
murderer" implies ''He is not a very pleasant fellow."
"John left" or "John was tall" are as close as narrative can
come to stage imitation, an actor walking into the wings or the
choice of a tall rather than short actor. So it seems reasonable
to call the narrative statements of such actions and presentations
"unnarrated." But "John left, unfortunately" or '10hn was tall,
unfortunately" necessarily presuppose a speaker who has taken
it upon himself to judge what is and what is not unfortunate.
They are clearly interpretive statements, and interpretation im-
plies a narrator.
In the strict sense, of course, all statements are "mediated,"
since they are composed by someone. Even dialogue has to be
invented by an author. But it is quite clear (well established in
theory. and criticism) that we must distinguish betwee"n the nar-
rator, or speaker, the one currently "telling" the story, and the
author, the.. ultimate designer of the fable, who also decides,
for example, whether to have a narrator, and if so, how promi-
nent he should be. It is a fundamental convention to ignore the
--
"

author, but not the narrator. The narrator may be overt-a real
character (Conrad's Marlow) or an intrusive outside party (the
narrator of Tom Jones). Or he may be "absent," as in some of
Hemingway's or Dorothy Parker's stories containing only dia-
logue and uncommented-upon action. The "narrator," when he
appears, is a demonstrable, recognizable entity immanent to the
narrative itself. Every narrative, even one wholly "shown" or
unmediated, finally has an author, the one who devised it. But
"narrator" should not be used in that sense. Rather it should
mean only the someone-person or presence-actually telling
the story to an audience, no matter how minimally evoked his
34 STORY AND DISCOURSE

voice or the audience's listening ear. A narrative that does not

give the sense of this presence, one that has gone to noticeable
lengths to efface it, may reasonably be called "nonnarrated" or
"unnarrated." (The seeming paradox is only terminological. It
is merely short for "a narrative that is not explicitly told" or
"that avoids the appearance of being told.") Thus there is no
reason for positing some third category of narrative (like "dra-
matic" or "objective" or the like) since that is essentially" 'non-
narrated' narrative."
It might seem that a discussion of existents is superfluous
or at least secondary to a minimal narrative theory. But one
cannot account for events without recognizing the existence of
things causing or being affected by those events. At the level of
discourse, no statement of an event can be made, in any me-
dium, that does not include a subject. It is true, of course, that
the narrative may have very little or even no overt description;
but a narrative without an agent performing actions is impos-
sible. A minimal kind of description is thereby entailed; for ex-
ample, if we are told (or shown) absolutely nothing more about
someone than that he loves a woman, we have at least the im-
plicit description "He is a lover" (the character has been "in-
dexed" by one process statement).
By way of example, let us consider a narrative in pictures,
rather than words, partly to underline the generality of narrative
components (they can occur in media other than natural lan-
guage), and partly because the rest of this book will cite only
verbal or cinematic examples. Picture narratives have, of course,
been common for centuries, as the Bayeux Tapestry and paint-
ings like Benozzo Gozzoli's Dance of Salome and the Beheading of
John the Baptist attest. In its simplest form, the picture narrative
represents the events in a clear sequence, say left to right, on
the analogy of western alphabets. But the order might be dif-
ferent: in Gozzoli's painting, for example, Salome dances for
Herod in the rightmost section of the painting, and a later event
-a soldier holding the sword over John's head-occurs in the
leftmost portion. It is in the middle that the final event occurs:
Salome gives the head to her mother.
The Dance of Salome and the Beheading of John the Baptist, by Benozzo Gozzoli, National Gallery of Art,
Washington, D.C., Samuel H. Kress Collection.
Short Ribs, by Frank O'Neal. Reprinted by permission of Newspaper Enterprise Association.
 INTRODUCTION 37

A Comic Strip Example

Picture narratives can be divided up into frames, the tech-
nique of the modern comic strip. Comic strips without dialogue,
captions, or balloons are relatively pure (if banal) examples of
narrative in picture form and as such conveniently illustrate my
diagram of the narrative situation.
The comic strip I have chosen appeared in 1970 in the Sunday
supplement of the San Francisco Chronicle. For convenience of
discussion I have labeled the ten frames 0 through IX (0 con-
taining introductory or "front" matter). With the exception of
the bubble in I and the signs in IV, VI, and VII, there are no
words in this narrative. Even those words could have been re-
placed by visual indications to distinguish the casino from the
pawn shop. The traditional three balls could have been used for
the one and perhaps a pair of dice for the other. But as it stands,
the Inedium is mixed: we must distinguish (1) drawing, either
representational or stylized-conventional, from two uses of
words: (2) dialogue (in the comic-strip convention of the ''bub-
ble") and (3) legend (the signs identifying the two buildings).
The story might be verbalized as follows: There once was a king.
Standing on the tower of his castle, he saw something "that looked
like fun" through his binoculars. He rushed downstairs and out of the
castle and soon arrived at the Royal Casino. He played dice and lost.
Leaving dejectedly, he happened upon the Royal Loan Company. A
crafty thought came to him. He pawned his crown for a bundle of
money so that he could go back to the Royal Casino to gamble some
more.
These are abstract narrative statements, hence I have italicized
them. This English-language version is not at all the story per se;
it is but one more (and poorer) manifestational representation
of it. Story, in my technical sense of the word, exists only at
an abstract level; any manifestation already entails the selection
and arrangement performed by the discourse as actualized by a
given medium. There is no privileged manifestation.
Further, though the above is, I think, a reasonably complete
depiction of "what happens" in the story, it cites only some
among an infinity of possible events. For example, the very exis-
tence of the king presupposes the event of his birth, his royalty
38 STORY AND DISCOURSE

presupposes the existence of a father (or some anc~stor) who

was a king, the event of his coronation, and so on. These are
essential (if trivial) to an understanding of the actual story. Even
in this simple narrative, important events are left to the reader's
inference. The crucial event-the loss at the dice table-occurs
in the space between frame V and frame VI (just as the murder
occurs in a "hole" in the plot of Robbe-Grillet's Le Voyeur). The
cartoonist could have shown a croupier taking the money away
from the king, but he elected not to, leaving the burden of
inference to the reader. We infer the central event from exis-
tents, the dejected appearance of the king in frame VI, at odds
with his animation in II and IV, his arms now hanging slackly
down his body, and his mouth drawn downward. In other
words, in terms of the diagram, the stasis statement The king
is dejected has projected the event of his gambling loss.
Let us now consider a few of the abstract narrative statements
in this story.
From frame 0, we conclude that There is a king (an unmedi-
, ated stasis statement exposes (identifies) a character, manifested
by a simple representational drawing, especiaUy the pronged
crown and the ermine collar and cuffs); The king is excited (stasis
statement exposes a trait of the character, manifested by a con-
ventional schema, the curved lines over the eyes to suggest
movement); and The king is looking through binoculars (process
statement enacts an action, manifested by representational
drawing). These events are in the story. The discourse is primar-
ily unmediated, that is, there is no audible or visible narrator.
However, there is an important discoursive feature communi-
cated by the fact that what the king sees through binoculars !
has been intentionally deleted from this frame. Hence we J
are also unable to see it. It is clear that we are being invited
to share the king's point of view or perspective, in what film-
makers would call an offscreen "eye-line match." (For "point of
view," see Chapter 4.) Later, especially beginning with frame
IV, we are looking objectively at the king, not with him; the
point of view shifts. .
From frame I we read out such narrative statements as It is
a sunny day; The king is on his tower; He thinks to himself . .. (mani-
 INTRODUCTION 39

fested by conventional schema, namely, dialogue bubble with

visual "thought" attachment, that is, disconnected bubbles of
decreasing size); ... "that looks like fun" (manifested by the
king's thought, in printed words); and There is a building below
and to the right.
Certain additional inferences can be made. We assume that it
is a sunny day from the appearance of the elliptical (red) object.
That it is not more perfectly round is a bit confusing until,
putting it together with the anachronistic binoculars (and, later,
the crazy tilt of the tower), we make some such inference as
"These events are taking place in a zany world." That it is the
king's tower is self-evident from common knowledge about
kings, through what Barthes calls the Referential or Gnomic or
Cultural Code. 1s The verbal statement "That looks like fun"
explains the excited interest of the raised eyebrows of frame 0,
and both are confirmed by the raised lines of the mouth in frame
I to show a smile of eager interest. We are still associated with
the king's point of view, but it is not yet clear what the object
of his attention is. Since he is looking in the direction of the
building, either the "fun" is there or it is still somewhere be-
yond the frame.
In frame II, we have There is a bird sitting in a window of the
tower; The king runs . .. down the stairs (expressed as a moment
when both feet are off the ground); . . . rapidly ... (manifested
by the conventional notation of a cloud of dust behind him and
the curved "motion-lines" above his head).
Frame ill has The king runs towards the building that "looks like
fun" (using the conventional schema of puffs of dust again to
mean "speed"). That it is the building associated with "fun"
is inferred from the fact that no other building is visible and that
it resembles the one in frame I. In frame IV, The king is about to
enter the Royal Casino (natural-language used as legend on sign).
We infer that the Royal Casino is the place that "looked like
fun" since it is the goal (directionally) toward which the king has
aspired since the first frame. The king is eager to have fun (from
the smile on his face).
15. 5/Z, trans. Richard Miller, p. 18: "made in a collective and anonymous
voice originating in traditional human experience."
40 STORY AND DISCOURSE

In frame V, The king shoots crap (conventional devices for

showing two kinds of motion imparted to the dice). We infer
that the king did in fact enter the Royal Casino between frames
N and V from cultural knowledge that throwing dice is cus-
tomarily done in casinos. The act is shown in a visual synec-
doche ("close-up" in cinematic language); only the royal hand
and cuff are visible as they throw the dice.
Frame VI has The king leaving slowly (both feet are on the
ground); The king is dejected (the downward cast of the mouth
and the arms straight at the side). We infer, again "gnomically,"
that the king has lost all the money he had with him. A prior
event is inferred from the stasis statement of an existent. In
frame Vll, The king stumbles upon the Royal Loan Company (curved
lines to show the "double take"; legend in natural language to
identify the loan company).
In frame VIll, The king has a thought, though strictly speaking,
this is an inference drawn from a stasis statement like The king
looks pensive (his pensiveness is read out of the hand placed
over the mouth); The king looks crafty (conveyed by a gesture, the
"devilish" angle of the eyebrows from which we infer, by
metonymy, 'The king's thought is wicked'; thus, one stasis
statement is inferred from another stasis statement). Frame IX
shows~The king leaving the Royal Loan Company with a bag of mon-
ey (symbolic device as legend); ... without his crown . .. (the
absence of a previously represented prop), from which we in-
fer that the king has pawned his crown; and The king is on his
way back to the casino, from which we infer . . . to gamble some
more.
Having read out the story, we are.disposed to interpret the
king's character, that he is silly or the like, at least that he holds
his royalty in light regard. Thus we infer a character trait from -
the whole action, that is, the set of events has indexed the king's
character. We need not end our interpretation there, of course.
We might conclude that the whole of the king's performance is
an exercise in futility, since, if "Royal" means what it says, he
owns both the casino and the loan company. He is the only
human figure in the entire narrative; the whole kingdom is his
alone, and he seems its only inhabitant. He loses money to him-
 INTRODUCTION 41

self and then pawns his crown for more of his own money so
that he can lose again. To himself. But then we were warned
that his is a zany world. Those who' feel prompted to search
deeper interpretive mines-say Freudian or Marxist-are wel-
come to do SO.16

"Reading" and "Reading Out"

Though this chapter has treated story as an object, I do not
mean to suggest that it is a hypostatized object, separate from
the process by which it emerges in the consciousness of a
"reader" (using that term to include not only readers in their
armchairs, but also audiences at movie houses, ballets, puppet
shows, and so on). I have attempted to demonstrate the process
by which one reads the relevant narrative features out of or
through one sort of nonverbal manifestation, namely the comic
strip. This kind of "reading out" is qualitatively different from
ordinary reading, though so familiar as to seem totally "nat-
ural." But the conventions are there and are crucial, even if
patently self-evident and self-instructional-the arbitrary fig-
ures, like the frame, the puffs of smoke to indicate speed, and
the bubbles for dialogue or thinking are effortlessly learned by
very small children. But that they are conventions is clear
enough. From the surface or manifestation level of reading, one
works through to the deeper narrative level. That is the process
I call, technically, reading out. Reading out is thus an "interlevel"
term, while mere "reading" is "intralevel." I am trying to avoid
technical vocabulary wherever possible, but this seems a neces-
16. A learned reader comes to quite another interpretation of this comic strip.
Noting that it makes "a big difference" that the Royal Casino is simply marked
"Casino" in the sign in the final frame, he feels that despite the "Royal," the
king does not own the casino or loan company at all. If he did, "a) he wouldn't
need the loan company and b) they wouldn't require his crown as collateral.
The world of Short Ribs is set initially by the king looking excitedly through his
binoculars. It's a modern world, then, with advanced technology in which the
king is isolated ... in his tower. He hopes for an improvement in his state
of affairs and descends to a world that seems to be his. But it isn't, we learn.
The kernel of truth here applies to all of us and doesn't require (though it would
support) a Marxian or other socioeconomic interpretation." The strip has elicited
other profound interpretations from students and colleagues. Oearly herme-
neutics has found a glorious new medium to munch on, along with Sunday
pancakes.
42 STORY AND DISCOURSE

sary distinction, and reading out a relatively transparent term

for "decoding from surface to deep narrative structures." Narra-
tive translation from one medium to another is possible because
roughly the same set of events and existents can be read out.
Obviously this book is more concerned with reading narra-
tives out than with simply reading their surfaces. I do not mini-
mize the problems entailed in surface reading, itself a pro-
foundly cultural and by no means "natural" process. Witness
the reports of anthropologists that aborigines have difficulty in
even seeing what are, to us, "self-evident" video and cinematic
images. But it is at the "reading-out" level that occur the prob-
lems of the elementary literature class, where students under-
stand the meaning of every sentence in isolation, but cannot
make any sense (or any satisfying sense) out of the whole nar-
rative text.

Let us now turn to a more detailed examination of story, con-

sidering first the event-dimension or plot. .