Home > Full text of "Narrative Discourse: An Essay in Method"

Full text of "Narrative Discourse: An Essay in Method"


Narrative Discourse


Translated by Jane E. Lewin
Foreword by Jonathan Culler




"Discours du recit," a portion of Figures HI by Gerard Genette, was
published in French, © Editions due Seuil, 1972.

The publisher gratefully acknowledges the financial assistance of the
French Ministry of Culture in defraying part of the cost of translation.

Copyright © 1980 by Cornell University

All rights reserved. Except for brief quotations in a review, this book,
or parts thereof, must not be reproduced in any form without
permission in writing from the publisher. For information address
Cornell University Press, Sage House, 512 East State Street, Ithaca,
New York 14850.

First published 1980 by Cornell University Press.
First printing, Cornell Paperbacks, 1983.

Quotations from Remembrance of Things Past, by Marcel Proust,
translated by C. K. Scott Moncrieff, copyright 1924, 1925, 1927, 1929,
1930, 1932 and renewed 1952, 1953, 1955, 1957, 1958, 1960 by Random
House, Inc., are reprinted by permission of Random House, Inc., and
Chatto and Windus, Ltd.

Cornell University Press strives to utilize environmentally responsible
suppliers and materials to the fullest extent possible in the publishing of
its books. Such materials include vegetable-based, low-VOC inks and
acid-free papers that are also either recycled, totally chlorine-free, or
partly composed of nonwood fibers.

Library of Congress Cataloging in Publication Data

Genette, Gerard, 1930-
Narrative Discourse.

Translation of Discours du recit, a portion of the 3d vol. of the author's
Figures, essais.

Bibliography: p.

Includes index.

1. Proust, Marcel, 1871-1922. A la recherche du temps perdu.
2. Narration (Rhetoric) 1. Title.

PQ2631.R63A791713 808.3'3 79-13499

ISBN 0-8014-1099-1 (cloth)
ISBN 0-8014-9259-9 (pbk.)

Paperback printing 10 9 8 7


Foreword by Jonathan Culler


Translator's Preface






1 Order


2 Duration


3 Frequency


4 Mood


5 Voice









Anyone who has begun the study of fiction has encountered
terms like point of view, flashback, omniscient narrator, third-person
narrative. One can't describe the techniques of a novel without
such terms, any more than one can describe the workings of a
car without the appropriate technical vocabulary. But while
someone who wanted to learn about cars would have no trouble
finding a manual, there is no comparable work for the student of
literature. These basic concepts have been developed in an ad
hoc, piecemeal fashion and, paradoxically, though they are
supposed to identify all the various elements and possible tech-
niques of the novel, they have not been put together in a sys-
tematic way. Even Wayne Booth's The Rhetoric of Fiction, from
which students of the novel have learned a great deal, is primar-
ily limited to problems of narrative perspective and point of
view. There has been no comprehensive survey.

Gerard Genette's Narrative Discourse is invaluable because it
fills this need for a systematic theory of narrative. As the most
thorough attempt we have to identify, name, and illustrate the
basic constituents and techniques of narrative, it will prove in-
dispensable to students of fiction, who not only will find in it
terms to describe what they have perceived in novels but will
also be alerted to the existence of fictional devices which they
had previously failed to notice and whose implications they had
never been able to consider. Every reader of Genette will find
that he becomes a more acute and perceptive analyst of fiction
than before.


8 Foreword

This is also a major work, however, for those who are in-
terested in narrative theory itself, for it is one of the central
achievements of what was called "structuralism." The struc-
turalist study of literature, associated with the names of Roland
Barthes, Tzvetan Todorov, Genette, and others, sought not to
interpret literature but to investigate its structures and devices.
The project, as defined in Barthes's Critique et verite and To-
dorov's "Poetique" (in Qu'est-ce que le structuralisme?), was to
develop a poetics which would stand to literature as linguistics
stands to language and which therefore would not seek to ex-
plain what individual works mean but would attempt to make
explicit the system of figures and conventions that enable works
to have the forms and meanings they do. ' Structuralists devoted
considerable attention to plot structure, or the "grammar" of
plot, as Todorov called it in his Grammaire du Decameron, and to
the ways in which details of various kinds in a novel are or-
ganized to produce effects of suspense, characters, plot se-
quences, and thematic and symbolic patterns. 2 Though Narra-
tive Discourse does not directly assimilate either of these investi-
gations, it is the centerpiece of the study of narrative, for in
attempting to define the forms and figures of narrative discourse
Genette must deal with all the complex relations between the
narrative and the story it tells. The structures and codes which
Barthes and Todorov studied must be taken up and organized
by a narrative; this activity is Genette' s subject.

But if Narrative Discourse is the culmination of structuralist
work on narrative and shows, in its terminological exuberance, a
Gallic delight in the adventures of thought, it is also wholly
conversant with Anglo-American discussions of narrative,
which it cites, uses, and occasionally refutes. This is no provin-
cial exercise but a broadly based theoretical study.

It is also, however — and this is doubtless more surprising — a

1 For discussion and bibliography see Jonathan Culler, Structuralist Poetics:
Structuralism, Linguistics, and the Study of Literature (Ithaca, N.Y.: Cornell Univer-
sity Press, 1975).

2 See Roland Barthes, S/Z (New York: Hill and Wang, 1974), and Tzvetan
Todorov, The Poetics of Prose (Ithaca, N.Y.: Cornell University Press; London:
Blackwell, 1977).

Foreword 9

remarkable study of Proust's A la recherche du temps perdu. It is as
though Genette had determined to give the lie to the skeptics
who maintained that the structural analysis of narrative was
suited only to the simplest narratives, like folk tales, and, in an
act of bravado, had chosen as his object one of the most com-
plex, subtle, and involuted of narratives. But in fact, this is not
an act of bravura. Genette has long been concerned with Proust,
and the three volumes of his Figures, 3 from which Narrative
Discourse is taken, contain three other essays on Proust's work.
Given the focus on Proust, our ordinary notions of criticism
ask us to choose between two ways of viewing Genette's pro-
ject: either his real goal is the development of a theory of narra-
tive and Proust's great novel is simply being used as a source of
illustrations, or else the theoretical matter is simply a methodo-
logical discussion which is justified insofar as it leads to a better
understanding of A la recherche du temps perdu. In his preface
Genette quite rightly refuses to choose between these alterna-
tives, but this does not mean that his work should be viewed as
something of a compromise, neither one nor the other. On the
contrary, it is an extreme and unusual example of each genre.
On the one hand, the fact that it uses Proust so voraciously gives
it great theoretical power, for it is forced to take account of all the
complexities of Proustian narrative. Not only is this a severe test
of categories, which doubtless leads to the discovery of new
distinctions, but the theory is constantly confronted with
anomalies and must show how they are anomalous. On the
other hand, the fact that Genette is trying to elaborate a theory
of narrative while studying Proust gives him a signal advantage
over other interpreters of the Recherche. He need not hasten to
offer a thematic interpretation of every incident, decide what is
Proust's vision of life, his conception of art. He can dwell on the
strangeness of Proustian discourse, constantly pointing out how

3 Figures (Paris: Seuil, 1966), Figures II (1969), Figures III (1972). In addition to
the three other discussions of Proust (one in each volume) these collections
contain essays dealing with Stendhal, Flaubert, Robbe-Grillet, Barthes, baroque
poets, and various issues in literary and .rhetorical theory. More recently,
Genette has published his immense Mimologiques (Seuil, 1976), a study of writ-
ings through the ages that have denied the arbitrary nature of the linguistic sign.



bizarre a construction this novel is. Compelled by his special
perspective to ask questions about what is usually taken for
granted, he continually tells us things we did not know about
the book and achieves something that most interpreters do not:
he leads us to experience the strangeness of the text.

Since Genette's presentation and Jane Lewin's translation are
admirably clear, there is no need to outline the book's argument,
and one can introduce it simply by indicating several major
areas of interest.

Point of View. One important and original proposal bears on
the traditional notion of point of view. Most theorists, Genette
argues, have failed to distinguish properly between "mood and
voice, that is to say, between the question who is the character
whose point of view orients the narrative perspective? and the very
different question who is the narrator?" Thus, if a story is told
from the point of view of a particular character (or, in Genette's
terms, focalized through that character), the question whether
this character is also the narrator, speaking in the first person, or
whether the narrator is someone else who speaks of him in the
third person, is not a question of the point of view, which is the
same in both cases, but a question of voice. And conversely, in
what is traditionally called a first-person narrative the point of
view can vary, depending on whether events are focalized
through the consciousness of the narrator at the moment of
narration or through his consciousness at a time in the past
when the events took place. Insistence on the difference be-
tween narration and focalization is a major revision of the theory
of point of view.

Focalization. The notion of focalization leads to some interest-
ing problems in its own right. One commentator, Mieke Bal, has
argued persuasively that Genette uses focalization to cover two
cases which are so different that to treat them as variants of the
same phenomenon is to weaken his important new concept. 4 In
what Genette calls internal focalization the narrative is focused
through the consciousness of a character, whereas external focali-

1 Mieke Bal, "Narration et localisation," Poetique, 29 (February 1977), 107-127.



zation is something altogether different: the narrative is focused
on a character, not through him. For example, in Hemingway's
"The Killers" or in the novels of Dashiell Hammett we are told
what the characters do but not what they think or see. To treat
this absence of focalization as another sort of focalization re-
duces the precision of the concept. Bal has proposed emenda-
tions to solve the problems which Genette's theory brings to
light, and Genette seems quite happy to accept modifications.
As he says in his Afterword, the very nature of poetics as a
progressive, cumulative enterprise ensures that his formulations
will one day be relegated to the rubbish heap. If this happens, it
will doubtless be because they have inspired improvements.

The Iterative. Genette's attempt to be comprehensive where
others have proceeded in more piecemeal fashion occasionally
leads to the discovery of topics which have not been much dis-
cussed but which prove, on investigation, to be extremely im-
portant. Studying the possible relationships between the time of
story or plot and the time of the narrative, he determines that
they may be classified in terms of order (events occur in one
order but are narrated in another), pace or duration (the narra-
tive devotes considerable space to a momentary experience and
then leaps over or swiftly summarizes a number of years), and
frequency (the narrative may repeatedly recount an event that
happened only once or may recount once what happened fre-
quently). Now order and pace are well known to students of
narrative: the former involves notions like flashback, foreshadow-
ing, and beginning in medias res, and the latter notions like scene
and summary. But frequency, as it happens, has seldom been
discussed, though it turns out to be a major topic. Repetition, a
common form of frequency, has emerged as the central tech-
nique in certain avant-garde novels, and what Genette calls the
iterative, in which the narrative tells once that something hap-
pened frequently, turns out to have a variety of important
functions. Proust, of course, is much given to the iterative
mode, but he also employs a fascinating figure which Genette
calls the pseudo-iterative: when the story narrates as something
that happened repeatedly an event whose very particularity
makes it seem undeniably singular. Thus, in the long account of



what happened every Sunday at Combray are inserted extended
conversations, unlikely to have been repeated every week. This
mode produces strange narrative effects which have not been
discussed; we owe our growing understanding of them to
Genette's pioneering investigation of the iterative.

Norm and Anomaly. Genette's definition of the figures of fre-
quency has the result of making anomalous (hence the label
"pseudo-iterative") a distinctively Proustian mode. Now one
might expect an account of narrative based on Proustian exam-
ples to work just the other way, making Proust's bizarre tech-
niques the norm; but under each of the major categories — tense,
voice, and mood — something typically Proustian is rendered
anomalous by the system of distinctions. Discussing voice,
Genette concludes that the movement from one level of narra-
tive to another in Proust is often confused and is ruled by trans-
gressions. In the case of mood, not only does Proust prove "in-
assimilable" to the basic distinction between mimesis and
diegesis, but his "polymodality" is "a scandal" for the system of
point of view. At moments when we are looking with Marcel
through a window or keyhole and seeing only those actions he
can see, we will be told the thoughts of the characters we are
supposedly observing. In various ways, as Genette says, "Proust
upsets the whole logic of narrative representation."

This may seem an odd conclusion to reach, since in compari-
son with more recent novelists Proust seems so massively com-
mitted to representing a world and a character's experience of it.
Doubtless, if Proust can always be caught in flagrant violation of
the system, this is because the categories for the description of
narrative discourse are in fact based on what we may for conve-
nience call a model of the real world. According to this model,
events necessarily take place both in a particular order and a
definable number of times. A speaker has certain kinds of in-
formation about events and lacks other kinds. He either experi-
enced them or he did not, and generally he stands in a definable
relationship to the events he recounts. However true this model
may be, there is nothing to prevent narratives from violating
it and producing texts which involve impossible combinations.
A sentence such as "I watched George reach into his briefcase



for something while he thought about whether he might have
lamb for dinner that evening" asserts a combination of knowl-
edge and ignorance that in the world would be most unlikely,
but novels frequently produce such combinations, though sel-
dom within the space of a single sentence. It may well be that
narratives will usually prove anomalous because our models of
narrative procedures are always based on models of reality.

But it might also be the case that Genette's work is testimony
to the power of the marginal, the supplementary, the exception.
It is as though his categories were specifically designed to iden-
tify as anomalous the most salient of Proust's techniques, so that
in a sense these marginal phenomena, these exceptions, in fact
determine the norms; these cases which the system seems to set
aside are in fact crucial to it. In its exemplification of this
paradoxical logic, Genette's work communicates with the most
interesting speculative strain of what is now called "post-
structuralism": Jacques Derrida's investigation of the logic of
marginality or supplementarity that is always at work in our
interpretive schemes. 5 Whether or not one actively pursues
these questions, Genette's Narrative Discourse is a provocative
work, as well as an indispensable tool for students of narrative.

Jonathan Culler
Ithaca, New York

5 See Derrida's Of Grammatology (Baltimore: Johns Hopkins University Press,

Translator's Preface

Marcel Proust's A la recherche du temps perdu, whose narrative
is "the specific subject of this book/' has been translated into
English as Remembrance of Things Past (by C. K. Scott Moncrieff
[vols. 1-6] and Frederick A. Blossom [whose translation of vol-
ume 7 was replaced in 1970 by Andreas Mayor's]; 2 vols., New
York: Random House, 1934; also published in seven separate
volumes by Random House). In this book the French title (which
means literally "in search of time lost") is retained, as are the
French titles of the seven volumes forming the Recherche, listed
here with their standard English translations:

Du cote de chez Swann (Swann's Way)
Part I: Combray (Overture; Combray)
Part II: Un amour de Swann (Swann in Love)
Part III: Noms de pays: le nom (Place-Names: The Name)

A Vombre des jeunes filles en fleurs (Within a Budding Grove)
Part I: Autour de Mme. Swann (Madame Swann at Home)
Part II: Noms de pays: le pays (Place-Names: The Place; Seascape, with
Frieze of Girls)

Le Cote de Guermantes (The Guermantes Way)

Sodome el Gomorrhe (Cities of the Plain)

La Prisonniere (The Captive)

Albertine disparue, later changed to La Fugitive (The Sweet Cheat


Le Temps retrouve (The Past Recaptured)

All quotations from the Recherche are from the Scott Moncrieff
and Mayor translation, except for a very few (indicated in the



Translator's Preface

Translator's Preface


notes) which are my translation, at those places where Genette's
exposition required a strictly literal rendering. In the notes, page
references to the Recherche are to both the two-volume Random
House translation (1934/1970) and the later three-volume
Clarac-Ferre edition (Pleiade, 1954) that is cited by Genette, but
in the body of the text, page numbers — or the number of pages
in a given section — refer only to the Random House edition.

For quotations from French works other than the Recherche, all
translations are mine unless the notes indicate otherwise. (Exist-
ing translations of other works by Proust and of French critical
studies, listed in the Bibliography, have always been used, and
in such cases the notes usually cite only the English edition.) For
quotations from works originally written in English, the original
has been quoted and cited, although Genette sometimes used
French translations, as listed in the Bibliography. And for quota-
tions from works originally written in a language other than
French or English, I have used and cited published English

I have silently modified the French edition of this book by
correcting obvious errors, occasionally supplementing the
documentation, and giving both French and English versions of
quotations from Proust when the French version seemed essen-
tial (mainly in Chapter 3).

The publication history of Proust's novel enters into Genette's
discussion (and explains, as well, the occasional discrepancies
between English and French versions of the Recherche). By 1912,
Proust had written a 1300-page novel in three sections: Du cote
de chez Swann, Le Cote de Guermantes, and he Temps retrouve.
Proust's original first part would have run about 800 pages in
print, but the publisher, Grasset, refused to produce a volume of
that size; his refusal forced Proust to play around with his ma-
terial, shifting it to meet the 500-page limit that Grasset im-
posed for publication in 1913. Then came the war, delaying pub-
lication of the remaining two sections — and giving Proust time
to alter and expand his manuscripts, which he did assiduously.
As a result, when publication was resumed five years later, by
Gallimard, it was with a volume entitled A V ombre desjeunes filles
en fleurs, formerly planned as the opening chapter of the third
volume; and Sodome et Gomorrhe was announced. Le Cote de

Guermantes I was published in 1920, with Guermantes II and
Sodome et Gomorrhe I following in 1921; Sodome II appeared in
May 1922. In November 1922 Proust died. La Prisonniere came
out in 1923, Albertine disparue (changed in 1954 to Proust's origi-
nal title, La Fugitive) in 1925, Le Temps retrouve in 1927. Scott
Moncrieff and Blossom's translation is based on these volumes.

Proust's method of working was such that the published edi-
tions of his novel were rather unreliable — in some cases
thoroughly so, as was learned when his manuscripts became
available in the 1950's. He revised and expanded incessantly,
adding to typescripts and page proofs without mercy. After
1918, in poor health and driving hard to finish his work before
death should come, he put his energies into creation rather than
supervision, with the result that the volumes published in his
lifetime were seen through the press by others, who had a great
deal of difficulty coping with the never-ending flow of revisions.
The volumes published after his death were based either on
manuscripts he had only partially revised or simply on rough
drafts, but considerably rearranged and touched up by the orig-
inal editors, whose first care was to put the drafts in readable
and orderly shape. In 1954, however, Pierre Clarac and Andre
Ferre, having had access to the newly available manuscripts,
published what is now the standard edition of the novel. They
restored the text of the later volumes to the state it had been in
when Proust died. For the earlier volumes, to establish their text
they struggled with Proust's habit of revising and adding, con-
tinually creating his novel, and letting other people — who may
have misunderstood his intentions or his handwriting — see the
work into print.

Because the French text on which the English translation of Le
Temps retrouve was based was the one most changed by the
Clarac-Ferre edition of 1954, in 1970 Andreas Mayor published a
new English translation based on the Clarac-Ferre text. Mayor's
avowed intention, however, was chiefly to please an audience
interested in reading a good narrative; therefore he took the
same kind of liberty with the restored text that the original
French editors had taken with Proust's manuscripts.

Jane E. Lewin
Bethesda, Maryland



The specific subject of this book is the narrative in A la re-
cherche du temps perdu. This statement immediately calls for two
comments, of differing importance. The first bears on the nature
of the Proustian corpus. Everyone today knows that the work
whose canonic text was established in 1954 by the Clarac-Ferre
edition is but the latest form of a work Proust labored at during his
whole life, as it were, a work whose earlier versions are, for the
most part, scattered among Les Plaisirs et les jours (1896), Pas-
tiches et melanges (1919), the various posthumous collections or
previously unpublished works entitled Chroniques (1927), Jean
Santeuil (1952), and Contre Sainte-Beuve (1954), 1 and the eighty-
odd notebooks deposited in the manuscript room of the Bib-
liotheque Nationale beginning in 1962. For this reason, plus the
forced interruption of November 18, 1922, the Recherche, more
than all other works, must not be considered closed; and there-
fore it is always legitimate and sometimes necessary to appeal to
one or another of its variants for comparison with the "defini-
tive" text. The same is true with respect to the handling of the
narrative. We cannot fail to appreciate, for example, how much

1 The dates given here are those of the first publication, but my references are
naturally to the Clarac-Sandre edition in two volumes — jean Santeuil preceded
by Les Plaisirs et les jours; Contre Sainte-Beuve preceded by Pastiches et melanges
and followed by Essais et articles (Pleiade, 1971) — which contain numerous pre-
viously unpublished writings. Even so, while waiting for the critical edition of
the Recherche we must sometimes continue to turn to the Fallois edition of the
Contre Sainte-Beuve for certain pages taken from the Cahiers.




perspective and significance the discovery of the Santeuil text in
the "third person" brings to the narrative system adopted in the
Recherche. Therefore while my study will bear mainly on the
final work, I will occasionally take into account its antecedents,
considering them not for their own sake, which would make
little sense, but for the light they can add.

The second comment concerns the method, or rather the ap-
proach, adopted here. Readers may already have observed that
neither the title nor the subtitle of this book mentions what I
have just designated as its specific subject. The reason is neither
coyness nor deliberate inflation of the subject. The fact is that
quite often, and in a way that may exasperate some readers,
Proustian narrative will seem neglected in favor of more general
considerations; or, as they say nowadays, criticism will seem
pushed aside by "literary theory," and more precisely by the
theory of narrative or narratology. I could justify and clarify this
ambiguous situation in two very different ways. I could
either — as others have done elsewhere — frankly put the specific
subject at the service of the general aim, and critical analysis at
the service of theory: in that case the Recherche would be only a
pretext, a reservoir of examples, and a flow of illustration for a
narrative poetics in which the specific features of the Recherche
would vanish into the transcendence of "laws of the genre." Or,
on the other hand, I could subordinate poetics to criticism and
turn the concepts, classifications, and procedures proposed here
into so many ad hoc instruments exclusively intended to allow a
more precise description of Proustian narrative in its particular-
ity, the "theoretical" detour being imposed each time by the
requirements of methodological clarification.

I confess my reluctance — or my inability — to choose between
these two apparently incompatible systems of defense. It seems
to me impossible to treat the Recherche du temps perdu as a mere
example of what is supposedly narrative in general, or novelistic
narrative, or narrative in autobiographical form, or narrative of
God knows what other class, species, or variety. The specificity
of Proustian narrative taken as a whole is irreducible, and any
extrapolation would be a mistake in method; the Recherche illus-
trates only itself. But, on the other hand, that specificity is not



undecomposable, and each of its analyzable features lends itself to
some connection, comparison, or putting into perspective. Like
every work, like every organism, the Recherche is made up of
elements that are universal, or at least transindividual, which it
assembles into a specific synthesis, into a particular totality. To
analyze it is to go not from the general to the particular, but
indeed from the particular to the general: from that incompa-
rable being that is the Recherche to those extremely ordinary ele-
ments, figures, and techniques of general use and common cur-
rency that I call anachronies, the iterative, focalizations,
paralipses, and so on. What I propose here is essentially a
method of analysis; I must therefore recognize that by seeking
the specific I find the universal, and that by wishing to put
theory at the service of criticism I put criticism, against my will,
at the service of theory. This is the paradox of every poetics, and
doubtless of every other activity of knowledge as well: always
torn between those two unavoidable commonplaces — that there
are no objects except particular ones and no science except of the
general — but always finding comfort and something like attrac-
tion in this other, slightly less widespread truth, that the general
is at the heart of the particular, and therefore (contrary to the
common preconception) the knowable is at the heart of the

But to answer for methodological giddiness, even strabismus,
by invoking science perhaps involves some fraud. I will there-
fore plead the same case differently: perhaps the real relation-
ship between "theoretical" dryness and critical meticulousness
is one of refreshing rotation and mutual entertainment. May the
reader also find in that relationship a sort of periodic diversion,
like the insomniac turning over and over in search of a better
position: amant alterna Camenae. 2

2 [Translator's note.] "Alternate strains are to the Muses dear." Virgil, Ec-
logues, III. 59, trans. James Rhoades, The Poems of Virgil (Chicago: Encyclopaedia
Britannica, 1952).


We currently use the word narrative 1 without paying attention
to, even at times without noticing, its ambiguity, and some of
the difficulties of narratology are perhaps due to this confusion.
It seems to me that if we want to begin to see clearly in this area,
we must plainly distinguish under this term three distinct

A first meaning — the one nowadays most evident and most
central in common usage — has narrative refer to the narrative
statement, the oral or written discourse that undertakes to tell of
an event or a series of events: thus we would term narrative of
Ulysses the speech given by the hero to the Phaeacians in Books
IX -XII of the Odyssey, and also these four books themselves, that
is, the section of Homeric text that purports to be the faithful
transcription of that speech.

A second meaning, less widespread but current today among
analysts and theoreticians of narrative content, has narrative
refer to the succession of events, real or fictitious, that are the
subjects of this discourse, and to their several relations of link-
ing, opposition, repetition, etc. "Analysis of narrative" in this
sense means the study of a totality of actions and situations
taken in themselves, without regard to the medium, linguistic or
other, through which knowledge of that totality comes to us: an

' [Translator's note.] The French word is recit; in Genette's text it functions as
"narrative" does in English, and it has been so translated throughout.



Narrative Discourse

example would be the adventures experienced by Ulysses from
the fall of Troy to his arrival on Calypso's island.

A third meaning, apparently the oldest, has narrative refer
once more to an event: not, however, the event that is re-
counted, but the event that consists of someone recounting
something: the act of narrating taken in itself. We thus say that
Books IX-XII of the Odyssey are devoted to the narrative of Ulys-
ses in the same way that we say Book XXII is devoted to the
slaughter of the suitors: recounting his adventures is just as
much an action as slaughtering his wife's suitors is, and if it goes
without saying that the existence of those adventures in no way
depends on the action of telling (supposing that, like Ulysses,
we look on them as real), it is just as evident that the narrative
discourse ("narrative of Ulysses" in the first meaning of the
term) depends absolutely on that action of telling, since the
narrative discourse is produced by the action of telling in the
same way that any statement is the product of an act of enunciat-
ing. If, on the other hand, we take Ulysses to be a liar and the
adventures he recounts to be fictitious, then the importance of
the act of narrating expands, for on it depend not only the exis-
tence of the discourse but also the fiction of the existence of the
actions that it "relates." The same thing can obviously be said of
the narrating act of Homer himself wherever he undertakes to
tell directly the account of the adventures of Ulysses. Without a
narrating act, therefore, there is no statement, and sometimes
even no narrative content. So it is surprising that until now the
theory of narrative has been so little concerned with the prob-
lems of narrative enunciating, concentrating almost all its atten-
tion on the statement and its contents, as though it were com-
pletely secondary, for example, that the adventures of Ulysses
should be recounted sometimes by Homer and sometimes by
Ulysses himself. Yet we know (and I will return to this later) that
Plato long ago found this subject worth his attention.

As its title indicates, or almost indicates, my study basically
has to do with the most widespread meaning of the term narra-
tive, that is, with narrative discourse, which in literature, and
particularly in the case that interests me, happens to be a narra-
tive text. But, as we will see, analysis of narrative discourse as 1



understand it constantly implies a study of relationships: on the
one hand the relationship between a discourse and the events
that it recounts (narrative in its second meaning), on the other
hand the relationship between the same discourse and the act
that produces it, actually (Homer) or fictively (Ulysses) (narra-
tive in its third meaning). Starting now, therefore, in order to
avoid confusion and semantic difficulties, we must designate
each of these three aspects of narrative reality by univocal terms.
I propose, without insisting on the obvious reasons for my
choice of terms, to use the word story for the signified or narra-
tive content (even if this content turns out, in a given case, to be
low in dramatic intensity or fullness of incident), to use the word
narrative for the signifier, statement, discourse or narrative text
itself, and to use the word narrating for the producing narrative
action and, by extension, the whole of the real or fictional situa-
tion in which that action takes place. 2

My subject here is therefore narrative, in the limited sense that
I will henceforth assign to that term. It is fairly evident, I think,
that of the three levels we have just sorted out, the level of
narrative discourse is the only one directly available to textual
analysis, which is itself the only instrument of examination at
our disposal in the field of literary narrative, and particularly
fictional narrative. If we wanted to study on their own account,
let us say, the events recounted by Michelet in his Histoire de
France, we could have recourse to all sorts of documents external
to that work and concerned with the history of France; or, if we
wanted to study on its own account the writing of that work, we
could use other documents, just as external to Michelet's text,
concerned with his life and his work during the years that he

2 [Translator's note.] "Story" is the French histoire (tell a story — raconter une
histoire); the gerund "narrating" is an English rendering of the French noun
narration, and it is the rendering that will be adhered to throughout. In a note at
this point Genette speaks of the acceptability of his terms with respect to current
French usage, and apropos of histoire ("story"), he refers to Tzvetan Todorov's
by now "fairly well accepted . . . proposal to differentiate 'narrative as discourse'
(first meaning) and 'narrative as story' (second meaning)." He also explains his
use of a term generally unfamiliar in America but used frequently in this book:
"With the same meaning ["story"), I will also use the term diegesis, which comes
to us from the theoreticians of cinematographic narrative."


Narrative Discourse

devoted to that text. Such a resource is not available to someone
interested in either the events recounted by the narrative that
the Recherche du temps perdu constitutes or the narrating act from
which it arises: no document external to the Recherche, and par-
ticularly not a good biography of Marcel Proust, if one existed, 3
could teach us about either those events or that act, since both of
these are fictional and both set on stage, not Marcel Proust, but
the hero and supposed narrator of his novel. I do not mean to
suggest that the narrative content of the Recherche has no con-
nection with the life of its author, but simply that this connec-
tion is not such that the latter can be used for a rigorous analysis
of the former (any more than the reverse). As to the narrating
that produced the narrative, the act of Marcel 4 recounting his
past life, we will be careful from this point on not to confuse it
with the act of Proust writing the Recherche du temps perdu. I will
come back to this subject later; it is enough for the time being to
remember that the 521 pages of Du cote de chez Swann (Grasset
edition) published in November 1913 and written by Proust
some years before that date are supposed (in the present state of
the fiction) to have been written by the narrator well after the
war. It is thus the narrative, and that alone, that informs us here
both of the events that it recounts and of the activity that sup-
posedly gave birth to it. In other words, our knowledge of the
two (the events and the action of writing) must be indirect,
unavoidably mediated by the narrative discourse, inasmuch as
the events are the very subject of that discourse and the activity
of writing leaves in it traces, signs or indices that we can pick up
and interpret — traces such as the presence of a first-person pro-
noun to mark the oneness of character and narrator, or a verb in
the past tense to indicate that a recounted action occurred prior
to the narrating action, not to mention more direct and more

3 The bad ones present no inconvenience here, since their main defect consists
of coolly attributing to Proust what Proust says of Marcel, to Illiers what he says
of Combray, to Cabourg what he says of Balbec, and so on — a technique debat-
able in itself, but not dangerous for us: except for the names, such books never
step outside the Recherche.

4 Here, to refer to both the hero and the narrator of the Recherche, we are
keeping this controversial Christian name. I will explain this in the last chapter.



explicit indications. Story and narrating thus exist for me only
by means of the intermediary of the narrative. But reciprocally
the narrative (the narrated discourse) can only be such to the
extent that it tells a story, without which it would not be narra-
tive (like, let us say, Spinoza's Ethics), and to the extent that it is
uttered by someone, without which (like, for example, a collec-
tion of archeological documents) it would not in itself be a dis-
course. As narrative, it lives by its relationship to the story that it
recounts; as discourse, it lives by its relationship to the narrating
that utters it.

Analysis of narrative discourse will thus be for me, essen-
tially, a study of the relationships between narrative and story,
between narrative and narrating, and (to the extent that they are
inscribed in the narrative discourse) between story and narrat-
ing. This position leads me to propose a new demarcation of the
field of study. My starting point will be the division put forth in
1966 by Tzvetan Todorov. 5 This division classed the problems of
narrative in three categories: that of tense, "in which the rela-
tionship between the time of the story and the time of the dis-
course is expressed"; that of aspect, "or the way in which the
story is perceived by the narrator"; that of mood, in other words,
"the type of discourse used by the narrator." I adopt, without
any amendment, the first category with the definition that I
have just cited, illustrated by Todorov with remarks on "tem-
poral distortions" (that is, infidelities to the chronological order
of events) and on relationships of linking, alternation, or em-
bedding among the different lines of action that make up the
story; but he added considerations about the "time of [narrative]
enunciating" and the time of narrative "perception" (which he
assimilated to the time of the writing and the reading) that seem
to me to exceed the limits of his own definition. I for my part will
hold those considerations in reserve for another order of prob-
lems, obviously connected to the relationships between narra-
tive and narrating. The category of aspect 6 basically covered

5 Tzvetan Todorov, "Les Categories du recit litteraire," Communications, 8

6 Rechristened "vision" in Litterature et signification (1967) and in Qu'est-ce que
le structuralisme? (1968).


Narrative Discourse

questions of narrative "point of view"; and that of mood 7
gathered together the problems of "distance" that American crit-
ics in the Jamesian tradition generally treat in terms of opposi-
tion between showing ("representation" in Todorov's vocabu-
lary) and telling ("narration"), a resurgence of the Platonic
categories of mimesis (perfect imitation) and diegesis (pure narra-
tive), the various ways of representing the speech of characters,
and the modes of explicit or implicit presence in the narrative of
narrator and reader. Just as with the "time of enunciating," here
too I think it is necessary to cut off the last series of problems, in
that it focuses on the act of narrating and its protagonists; on the
other hand, we must gather into a single large category — let us
provisionally call it that of the modalities of representation or the
degrees of mimesis — all the rest of what Todorov split between
aspect and mood. This redistribution thus ends us up with a
division substantially different from the one that inspired it, a
division that I will now formulate on its own account, having
recourse for my terms to a kind of linguistic metaphor that
should certainly not be taken too literally.

Since any narrative, even one as extensive and complex as the
Recherche du temps perdu, 8 is a linguistic production undertaking
to tell of one or several events, it is perhaps legitimate to treat it
as the development — monstrous, if you will — given to a verbal
form, in the grammatical sense of the term: the expansion of a
verb. 1 walk, Pierre has come are for me minimal forms of narra-
tive, and inversely the Odyssey or the Recherche is only, in a
certain way, an amplification (in the rhetorical sense) of state-
ments such as Ulysses comes home to Ithaca or Marcel becomes a
writer. This perhaps authorizes us to organize, or at any rate to
formulate, the problems of analyzing narrative discourse accord-
ing to categories borrowed from the grammar of verbs,

7 Rechristened "register" in 1967 and 1968.

8 Is it necessary to specify that by treating this work as a narrative here we do
not by any means intend to limit it to that aspect? An aspect too often neglected
by critics, but one Proust himself never lost sight of. Thus he speaks of "that
invisible vocation of which these volumes are the history" (RH 1, 1002/P II, 397; my



categories that I will reduce here to three basic classes of deter-
minations: those dealing with temporal relations between narra-
tive and story, which I will arrange under the heading of tense;
those dealing with modalities (forms and degrees) of narrative
"representation," and thus with the mood 9 of the narrative; and
finally, those dealing with the way in which the narrating itself
is implicated in the narrative, narrating in the sense in which I
have defined it, that is, the narrative situation or its instance, 10
and along with that its two protagonists: the narrator and his
audience, real or implied. We might be tempted to set this third
determination under the heading of "person," but, for reasons
that will be clear below, I prefer to adopt a term whose
psychological connotations are a little less pronounced (very lit-
tle less, alas), a term to which I will give a conceptual extension
noticeably larger than "person" — an extension in which the
"person" (referring to the traditional opposition between
"first-person" and "third-person" narratives) will be merely one
facet among others: this term is voice, whose grammatical mean-
ing Vendryes, for example, defined thus: "Mode of action of the
verb in its relations with the subject." 11 Of course, what he is
referring to is the subject of the statement, whereas for us voice,
since it deals with the narrating, will refer to a relation with the
subject (and more generally with the instance) of the enunciat-

9 The term is used here with a sense very close to its linguistic meaning, if we
refer, for example, to this definition in the Littre dictionary: "Name given to the
different forms of the verb that are used to affirm more or less the thing in
question, and to express. . . the different points of view from which the life or
the action is looked at."

10 In the sense in which Benveniste speaks of "instance of discourse" (Problems
in General Linguistics, trans. M. E. Meek [Coral Gables, Fla., 1971], pp. 217-222).
[Translator's note: "instance" with this very particular sense appears through-
out Genette's text. In Benveniste's essay ("The Nature of Pronouns"), the
"instances of discourse" are defined as "the discrete and always unique acts by
which the language is actualized in speech by a speaker" (p. 217); "[each]
instance is unique by definition" (p. 218). The narrating instance, then, refers
to something like the narrating situation, the narrative matrix — the entire set of
conditions (human, temporal, spatial) out of which a narrative statement is

11 Quoted in the Petit Robert dictionary, under Voix.


Narrative Discourse

ing: once more, these terms are merely borrowed, and I make no
pretense of basing them on rigorous homologies. 12

As we have seen, the three classes proposed here, which des-
ignate fields of study and determine the arrangement of the
chapters that follow, 13 do not overlap with but sort out in a more
complex way the three categories defined earlier designating the
levels of definition of narrative: tense and mood both operate at
the level of connections between story and narrative, while voice
designates the connections between both narrating and narrative
and narrating and story. We will be careful, however, not to
hypostatize these terms, not to convert into substance what is
each time merely a matter of relationships.

12 Another — purely Proustological — justification for the use of this term: the
existence of Marcel Muller's valuable book entitled Les Voix narratives dans "A la
recherche du temps perdu" (Geneva, 1965).

13 The first three (Order, Duration, Frequency) deal with time; the fourth, with
mood; the fifth and last, with voice.

1 Order

Narrative Time?

Narrative is a . . . doubly temporal sequence . . . : There is the time
of the thing told and the time of the narrative (the time of the
signified and the time of the signifier). This duality not only ren-
ders possible all the temporal distortions that are commonplace
in narratives (three years of the hero's life summed up in two
sentences of a novel or in a few shots of a "frequentative" montage
in film, etc.). More basically, it invites us to consider that one of
the functions of narrative is to invent one time scheme in terms of
another time scheme. 1

The temporal duality so sharply emphasized here, and re-
ferred to by German theoreticians as the opposition between
erzahlte Zeit (story time) and Erzahlzeit (narrative time), 2 is a
typical characteristic not only of cinematic narrative but also of
oral narrative, at all its levels of aesthetic elaboration, including
the fully "literary" level of epic recitation or dramatic narration
(the narrative of Theramene, 3 for example). It is less relevant

1 Christian Metz, Film Language: A Semiotics of the Cinema, trans. Michael
Taylor (New York, 1974), p. 18. [Translator's note: I have altered this translation
slightly so as to align its terms with the terms used throughout this book.]

2 See Gunther Mtiller, "Erzahlzeit und erzahlte Zeit," Festschrift fiir P . Kluck-
hohn und Hermann Schneider, 1948; rpt. in Morphologische Poetik (Tubingen, 1968).

3 [Translator's note.l A character in Racine's Phedre, proverbial for his narra-
tion of Hippolytus' death.



Narrative Discourse

perhaps in other forms of narrative expression, such as the
roman-photo 4 or the comic strip (or a pictorial strip, like the pre-
della of Urbino, or an embroidered strip, like the "tapestry" of
Queen Matilda), which, while making up sequences of images
and thus requiring a successive or diachronic reading, also lend
themselves to, and even invite, a kind of global and synchronic
look — or at least a look whose direction is no longer determined
by the sequence of images. The status of written literary narra-
tive in this respect is even more difficult to establish. Like the
oral or cinematic narrative, it can only be "consumed," and
therefore actualized, in a time that is obviously reading time,
and even if the sequentiality of its components can be under-
mined by a capricious, repetitive, or selective reading, that un-
dermining nonetheless stops short of perfect analexia: one can
run a film backwards, image by image, but one cannot read a
text backwards, letter by letter, or even word by word, or even
sentence by sentence, without its ceasing to be a text. Books are
a little more constrained than people sometimes say they are by
the celebrated linearity of the linguistic signifier, which is easier
to deny in theory than eliminate in fact. However, there is no
question here of identifying the status of written narrative (liter-
ary or not) with that of oral narrative. The temporality of written
narrative is to some extent conditional or instrumental; produced
in time, like everything else, written narrative exists in space and
as space, and the time needed for "consuming" it is the time
needed for crossing or traversing it, like a road or a field. The
narrative text, like every other text, has no other temporality than
what it borrows, metonymically, from its own reading.

This state of affairs, we will see below, has certain conse-
quences for our discussion, and at times we will have to correct,
or try to correct, the effects of metonymic displacement; but we
must first take that displacement for granted, since it forms part
of the narrative game, and therefore accept literally the quasi-
fiction of Erzahlzeit, this false time standing in for a true time
and to be treated — with the combination of reservation and ac-
quiescence that this involves — as a pseudo-time.

4 [Translator's note.] Magazine with love stories told in photographs.



Having taken these precautions, we will study relations be-
tween the time of the story and the (pseudo-) time of the narra-
tive according to what seem to me to be three essential determi-
nations: connections between the temporal order of succession
of the events in the story and the pseudo-temporal order of their
arrangement in the narrative, which will be the subject of the
first chapter; connections between the variable duration of these
events or story sections and the pseudo-duration (in fact, length
of text) of their telling in the narrative — connections, thus, of
speed — which will be the subject of the second chapter; finally,
connections of frequency, that is (to limit myself to an approxi-
mate formulation), relations between the repetitive capacities of
the story and those of the narrative, relations to which the third
chapter will be devoted.


To study the temporal order of a narrative is to compare the
order in which events or temporal sections are arranged in the
narrative discourse with the order of succession these same
events or temporal segments have in the story, to the extent that
story order is explicitly indicated by the narrative itself or infer-
able from one or another indirect clue. Obviously this recon-
stitution is not always possible, and it becomes useless for cer-
tain extreme cases like the novels of Robbe-Grillet, where tem-
poral reference is deliberately sabotaged. It is just as obvious
that in the classical narrative, on the other hand, reconstitution
is most often not only possible, because in those texts narrative
discourse never inverts the order of events without saying so,
but also necessary, and precisely for the same reason: when a
narrative segment begins with an indication like "Three months
earlier, ..." we must take into account both that this scene
comes after in the narrative, and that it is supposed to have come
before in the story: each of these, or rather the relationship be-
tween them (of contrast or of dissonance), is basic to the narra-
tive text, and suppressing this relationship by eliminating one of
its members is not only not sticking to the text, but is quite
simply killing it.

Pinpointing and measuring these narrative anachronies (as I


Narrative Discourse



will call the various types of discordance between the two order-
ings of story and narrative) implicitly assume the existence of a
kind of zero degree that would be a condition of perfect tem-
poral correspondence between narrative and story. This point of
reference is more hypothetical than real. Folklore narrative
habitually conforms, at least in its major articulations, to
chronological order, but our (Western) literary tradition, in con-
trast, was inaugurated by a characteristic effect of anachrony. In
the eighth line of the Iliad, the narrator, having evoked the
quarrel between Achilles and Agamemnon that he proclaims as
the starting point of his narrative (ex hou de ta prota), goes back
about ten days to reveal the cause of the quarrel in some 140
retrospective lines (affront to Chryses — Apollo's anger —
plague). We know that this beginning in medias res, followed by
an expository return to an earlier period of time, will become
one of the formal topoi of epic, and we also know how faithfully
the style of novelistic narration follows in this respect the style of
its remote ancestor, 5 even in the heart of the "realistic"
nineteenth century. To be convinced of this one need only think
of certain of Balzac's openings, such as those in Cesar Birotteau or
La Duchesse de Langeais. D'Arthez directs Lucien de Rubempre to
follow this principle, 6 and Balzac himself chides Stendhal for not
having begun the Chartreuse with the Waterloo episode, reduc-
ing "everything that precedes it to some narrative by or about
Fabrice while he lies wounded in the Flemish village." 7 We will
thus not be so foolish as to claim that anachrony is either a rarity
or a modern invention. On the contrary, it is one of the tra-
ditional resources of literary narration.

Furthermore, if we look a little more closely at the opening
lines of the Iliad just referred to, we see that their temporal

5 A testimony a con trario is this appraisal Huet gives of Jamblique's Babyloniques:
"The arrangement of his design lacks art. He has roughly followed temporal
order, and did not toss the reader immediately into the middle of the subject as
Homer did" (Traite de I'origine des romans, 1670, p. 157).

6 "Step into the action first. Grab your subject sometimes sideways, some-
times from the rear; finally, vary your plans, so as never to be the same" (Balzac,
Illusions perdues, Gamier ed., p. 230).

7 Balzac, Etudes sur M. Beyle (Geneva, 1943), p. 69.

movement is still more complex. Here they are in the translation
of Andrew Lang, Walter Leaf, and Ernest Myers:

Sing, goddess, the wrath of Achilles Peleus' son, the ruinous
wrath that brought on the Achaians woes innumerable, and
hurled down into Hades many strong souls of heroes, and gave
their bodies to be a prey to dogs and all winged fowls; and so the
counsel of Zeus wrought out its accomplishment from the day
when first strife parted Atreides king of men and noble Achilles.
Who then among the gods set the twain at strife and variance?
Even the son of Leto and of Zeus; for he in anger at the king sent a
sore plague upon the host, that the folk began to perish, because
Atreides had done dishonour to Chryses the priest. 8

Thus, the first narrative subject Homer refers to is the wrath of
Achilles; the second is the miseries of the Greeks, which are in fact
its consequence; but the third is the quarrel between Achilles and
Agamemnon, which is its immediate cause and thus precedes it;
then, continuing to go back explicitly from cause to cause: the
plague, cause of the quarrel, and finally the affront to Chryses,
cause of the plague. The five constituent elements of this open-
ing, which I will name A, B, C, D, and £ according to the order
of their appearance in the narrative, occupy in the story, respec-
tively, the chronological positions 4, 5, 3, 2, and 1: hence this
formula that will synthesize the sequential relationships more or
less well: A4-B5-C3-D2-E1. We are fairly close to an evenly
retrograde movement. 9

We must now go into greater detail in our analysis of anach-
ronies. I take a fairly typical example from Jean Santeuil. The
situation is one that will appear in various forms in the Re-
cherche: the future has become present but does not resemble the
idea of it that one had in the past. Jean, after several years, again

8 Homer, The Iliad, trans. Andrew Lang, Walter Leaf, and Ernest Myers (New
York: Modern Library, n.d.), Book I, 11.1-11. [Translator's note: Genette's refer-
ence in the text is to the French translation by Paul Mazon (Paris, 1962).]

9 And even more so if we take into account the first — nonnarrative — section,
in the present tense of the narrating instance (in Benveniste's sense), which thus
comes at the last possible moment: "Sing, goddess."

!| II


Narrative Discourse



finds the hotel where Marie Kossichef, whom he once loved,
lives, and compares the impressions he has today with those
that he once thought he would be experiencing today:

Sometimes passing in front of the hotel he remembered the rainy
days when he used to bring his nursemaid that far, on a pilgrim-
age. But he remembered them without the melancholy that he
then thought he would surely some day savor on feeling that he
no longer loved her. For this melancholy, projected in anticipation
prior to the indifference that lay ahead, came from his love. And
this love existed no more. 10

The temporal analysis of such a text consists first of number-
ing the sections according to their change of position in story
time. We discover here, in brief, nine sections divided between
two temporal positions that we will designate 2 (now) and 1
(once), setting aside their iterative nature ("sometimes"). Section
A goes in position 2 ("Sometimes passing in front of the hotel he
remembered"), B in position 1 ("the rainy days when he used to
bring his nursemaid that far, on a pilgrimage"), C in 2 ("But he
remembered them without"), D in 1 ("the melancholy that he
then thought"), E in 2 ("he would surely some day savor on
feeling that he no longer loved her"), F in 1 ("For this melan-
choly, projected in anticipation"), G in 2 ("prior to the indif-
ference that lay ahead"), H in 1 ("came from his love"), 7 in 2
("And this love existed no more"). The formula of temporal
positions, then, is as follows:


thus, a perfect zigzag. We will observe in passing that on a first
reading the difficulty of this text comes from the apparently
systematic way in which Proust eliminates the most elementary
temporal indicators (once, now), so that the reader must supply

10 jean Santeuil, Pleiade ed., p. 674. [Translator's note: the rendering given in
the English edition — trans. Gerard Hopkins (New York, 1956), p. 496 — is very
free; for purposes of Genette's analysis, I have used a literal translation of my

them himself in order to know where he is. But simply picking
out the positions does not exhaust temporal analysis, even tem-
poral analysis restricted to questions of sequence, and does not
allow us to determine the status of the anachronies: we have yet
to define the relationships connecting sections to each other.

If we take section A as the narrative starting point, and there-
fore as being in an autonomous position, we can obviously de-
fine section B as retrospective, and this retrospection we may call
subjective in the sense that it is adopted by the character him-
self, with the narrative doing no more than reporting his present
thoughts ("he remembered . . . "); B is thus temporally subordi-
nate to A: it is defined as retrospective in relation to A. C con-
tinues with a simple return to the initial position, without sub-
ordination. D is again retrospective, but this time the retrospec-
tion is adopted directly by the text: apparently it is the narrator
who mentions the absence of melancholy, even if this absence is
noticed by the hero. E brings us back to the present, but in a
totally different way from C, for this time the present is envis-
aged as emerging from the past and "from the point of view" of
that past: it is not a simple return to the present but an anticipa-
tion (subjective, obviously) of the present from within the past;
E is thus subordinated to D as D is to C, whereas C, like A, was
autonomous. F brings us again to position 1 (the past), on a
higher level than anticipation E : simple return again, but return
to 1, that is, to a subordinate position. G is again an anticipa-
tion, but this time an objective one, for the Jean of the earlier
time foresaw the end that was to come to his love precisely as,
not indifference, but melancholy at loss of love. H, like F, is a
simple return to 1. I, finally, is (like C) a simple return to 2, that
is, to the starting point.

This brief fragment thus offers us in miniature a quite var-
iegated sample of the several possible temporal relationships:
subjective and objective retrospections, subjective and objective
anticipations, and simple returns to each of these two positions.
As the distinction between subjective and objective anachronies
is not a matter of temporality but arises from other categories
that we will come to in the chapter on mood, we will neutralize
it for the moment. Moreover, to avoid the psychological conno-


Narrative Discourse

tations of such terms as "anticipation" or "retrospection,"
which automatically evoke subjective phenomena, we will
eliminate these terms most of the time in favor of two others that
are more neutral, designating as prolepsis any narrative maneu-
ver that consists of narrating or evoking in advance an event that
will take place later, designating as analepsis any evocation after
the fact of an event that took place earlier than the point in the
story where we are at any given moment, and reserving the
general term anachrony to designate all forms of discordance
between the two temporal orders of story and narrative (we will
see later that these discordances are not entirely limited to
analepsis and prolepsis). 11

This analysis of syntactic relationships (subordination and
coordination) between sections now allows us to replace our
first formula, which admitted only positions, with a second,
which recognizes connections and interlockings:


Here we clearly see the difference in status between sections
A, C, and I on the one hand, and £ and G on the other, all of
which occupy the same temporal position but not at the same
hierarchical level. We also see that the dynamic relationships
(analepses and prolepses) come at the openings of brackets or
parentheses, with the closings corresponding to simple returns.
Finally, we observe that this fragment is perfectly self-
contained, with the starting positions at each level scrupulously
reinstated: we will see that this is not always the case. Of course,
numerical relationships allow us to recognize analepses and pro-

11 Here begin the problems (and disgraces) of terminology. Prolepsis and
analepsis offer the advantage of being — through their roots — part of a
grammatical-rhetorical family some of whose other members will serve us later;
on the other hand, we will have to play on the opposition between the root
-lepse — which in Greek refers to the fact of taking, whence, in narrative, assum-
ing responsibility for and taking on (prolepsis: to take on something in advance;
analepsis: to take on something after the event) — and the root -lipse (as in ellipsis
or paralipsis) which refers, on the contrary, to the fact of leaving out, passing by
without any mention. But no prefix taken from Greek allows us to subsume the
antithesis prolana. Whence our recourse to anachrony, which is perfectly clear but
lies outside the system, and whose prefix interferes regrettably with analepsis.
Regrettably, but significantly.



lepses, but we can clarify the formula even further, like this, for



This fragment presented the obvious advantage (didactically)
of a temporal structure limited to two positions. That situation is
fairly rare, however, and before leaving the micronarrative level
behind, we will take from Sodome et Gomorrhe a text that is much
more complex (even if we reduce it, as we shall, to its basic
temporal positions, ignoring a few nuances), and that illustrates
well the temporal omnipresence characteristic of Proustian nar-
rative. We are at the soiree given by the Prince de Guermantes,
and Swann has just told Marcel of the Prince's conversion to
Dreyfusism which, with a naive partiality, he sees as proof of
intelligence. This is how Marcel's narrative makes connections (I
put a letter at the beginning of each distinct section):

(A) Swann now found equally intelligent anybody who was of his
opinion, his old friend the Prince de Guermantes and my school-
fellow Bloch, (B) whom previously he had avoided (C) and whom
he now invited to luncheon. (D) Swann interested Bloch greatly
by telling him that the Prince de Guermantes was a Dreyfusard.
"We must ask him to sign our appeal for Picquart; a name like his
would have a tremendous effect." But Swann, blending with his
ardent conviction as an Israelite the diplomatic moderation of a
man of the world, (E) whose habits he had too thoroughly ac-
quired (F) to be able to shed them at this late hour, refused to
allow Bloch to send the Prince a circular to sign, even on his own
initiative. "He cannot do such a thing, we must not expect the
impossible," Swann repeated. "There you have a charming man
who has travelled thousands of miles to come over to our side. He
can be very useful to us. If he were to sign your list, he would
simply be compromising himself with his own people, would be
made to suffer on our account, might even repent of his confi-
dences and not confide in us again." Nor was this all, Swann
refused his own signature. He felt that his name was too Hebraic
not to create a bad effect. Besides, even if he approved of all the
attempts to secure a fresh trial, he did not wish to be mixed up in
any way in the antimilitarist campaign. He wore, (G) a thing he
had never done previously, the decoration (H) he had won as a


Narrative Discourse

young militiaman, in '70, (/) and added a codicil to his will asking
that, (/) contrary to his previous dispositions, (K) he might be
buried with the military honours due to his rank as Chevalier of
the Legion of Honour. A request which assembled round the
church of Combray a whole squadron of (L) those troopers over
whose fate Francoise used to weep in days gone by, when she
envisaged (M) the prospect of a war. (N) In short, Swann refused
to sign Bloch's circular, with the result that, if he passed in the
eyes of many people as a fanatical Dreyfusard, my friend found
him lukewarm, infected with Nationalism, and a militarist.

(O) Swann left me without shaking hands so as not to be forced
into a general leave-taking. 12

We have thus recognized here (once more, extremely crudely
and for purely demonstrative purposes) fifteen narrative sections,
distributed among nine temporal positions. These positions are
the following, in chronological order: (1) the war of 1870; (2) Mar-
cel's childhood in Combray; (3) a time before the Guermantes
soiree; (4) the Guermantes soiree, which we can place in 1898; (5)
the invitation to Bloch (necessarily later than this soiree, from
which Bloch is absent); (6) the Swann-Bloch luncheon; (7) the
addition of the codicil; (8) Swann's funeral; (9) the war whose
prospect Francoise envisaged and which, strictly speaking, oc-
cupies no definite position, since it is purely hypothetical, but
which — in order to place it in time and simplify things — we may
identify with the war of 1914-18. The formula of positions is then
the following:


If we compare the temporal structure of this fragment to that of
the preceding one, we notice, besides the greater number of posi-
tions, a much more complex hierarchical interlocking, since, for
example, M depends onL, which depends onK, which depends
on I, which depends on the large prolepsis D-N. Moreover, cer-
tain anachronies, HkeB andC, are juxtaposed without an explicit
return to the base position: they are thus at the same level of sub-
ordination and are simply coordinate with each other. Finally,

12 RH II, 82-83/P II, 712-713.



the transition from C5 to D6 does not produce a true prolepsis
since we never come back to position 5; it therefore constitutes a
simple ellipsis of the time that passed between 5 (the invitation)
and 6 (the luncheon); the ellipsis, or leap forward without any
return, is obviously not an anachrony but a simple acceleration
of the narrative, which we will study in the chapter on duration:
it certainly has to do with time, but not time approached as order,
which is all that interests us here; we will thus mark the transition
from C to D not with a bracket but simply with a hyphen to indi-
cate sheer succession. This, then, is the complete formula:


We will now leave the micronarrative level behind to examine
the main articulations of the temporal structure of the Recherche.
Needless to say, an analysis at this level cannot consider the
details that belong to another scale, and therefore proceeds by
means of very crude simplification: here we pass from the micro-
structure to the macrostructure.

The first temporal section of the Recherche, which occupies the
first five pages of the book, evokes a moment that is impossible
to date with precision but that takes place fairly late in the hero's
life, 13 at the time when, going to bed early and suffering from
insomnia, he spent a large part of his nights recalling his past.
This first time in the narrative order is therefore far from being
first in the diegetic order. Anticipating the analysis to follow, let
us assign it at once to position 5 in the story. Thus: A5.

The second section (I, 7-33) is the account given by the
narrator — but plainly inspired by the memories of the sleepless
hero (who fulfills here the function of what Marcel Muller calls
the intermediary subject) 14 — of a very limited but very important
episode in his childhood in Combray: the famous scene that he

13 As a matter of fact, one of the rooms evoked is that at Tansonville, where
Marcel slept only during the visit recounted at the end of La Fugitive and the
beginning of the Temps retrouve. The period of the insomnias, necessarily later
than that visit, could coincide with one and/or the other of the cures in a clinic
which follow, and which frame the episode of Paris at war (1916).

14 Muller, Les Voix narratives, Part I, chap. 2, and passim. I will return to
the distinction between hero and narrator in the last chapter.


Narrative Discourse

names "the drama of [his] going to bed," in the course of which
his mother, prevented by Swann's visit from bestowing on him
her ritual goodnight kiss, will finally — decisive "first conces-
sion" — yield to his pleas and spend the night with him: B2.

The third section (I, 33-34) brings us very briefly back to po-
sition 5, that of the insomnias: C5. The fourth probably also
takes place somewhere within that period, since it brings about
a modification in the content of the insomnias: 15 it is the episode
of the madeleine (I, 34-36), in the course of which the hero finds
a whole side of his childhood restored to him, a side of his
childhood ("of Combray, save what was comprised in the
theatre and the drama of my going to bed there") that until then
had remained buried (and preserved) in apparent oblivion: D5'.
Thus a fifth section follows, a second return to Combray but
much vaster than the first in its temporal range since this time it
covers (not without ellipses) the whole of the childhood in
Combray. Combray II (I, 37-142) will thus be for us E2', contem-
poraneous with B2 but largely overflowing it, the way C5 over-
flows and includes D5'.

The sixth section (I, 143) returns to position 5 (insomnias):
thus F5. This position again serves as a springboard for a new
memory-elicited analepsis, whose place is the earliest of all since
it antedates the hero's birth: Un amour de Swann (I, 144-292) is
the seventh section, Gl.

The eighth section is a very brief return (I, 293) to the position
of the insomnias, thus H5. Again this position opens an analep-
sis, one that this time is aborted, although its function as ad-
vance notice or pointer is obvious to the attentive reader: the
evocation in a half-page (still I, 293) of Marcel's room at Balbec is
the ninth section, 14. Immediately coordinated with this, only
now without a perceptible return to the transfer point of the
insomnias, is the narrative (this, too, retrospective with respect
to the starting point) of the dreams of traveling that the hero had
in Paris, several years before his stay in Balbec; the tenth section
will thus be J3: Parisian adolescence, love with Gilberte, partici-

15 After the madeleine, the "total" Combray will be integrated into the insom-
niac's memories.



pation in Mme. Swann's circle, then, after an ellipsis, first stay
at Balbec, return to Paris, entry into the milieu of the Guer-
mantes, etc.: henceforth the movement is established, and the
narrative, in its major articulations, for the most part becomes
regular and conforms to chronological order — so much so that,
at our level of analysis, we may take section ]3 to extend to all
the rest (and the end) of the Recherche.

The formula for this beginning is, then, according to our pre-
vious conventions:

A5[B2]C5[D5'(E2')]F5[G1]H5[I4\[J3 . . .

Thus, the Recherche du temps perdu is launched with a vast
movement of coming-and-going from one key, strategically
dominant position, obviously position 5 (insomnias) and its
variant 5' (madeleine) — positions of the "intermediary subject,"
who is insomniac or beneficiary of the miracle of involuntary
memory. His recollections control the whole of the narrative,
giving point 5-5' the function of a sort of indispensable transfer
point or — if one may say so — of a dispatching narrative: in order
to pass from Combray I to Combray II, from Combray II to Un
amour de Swann, from Un amour de Swann to Balbec, it is always
necessary to come back to that position, which is central even
though excentric (because later). Its control does not loosen until
the transition from Balbec to Paris, even though this latter sec-
tion (J3), inasmuch as it is coordinated with the preceding sec-
tion, is also subordinated to the remembering activity of the
intermediary subject, and so it too is analeptic. The
difference — certainly essential — between this analepsis and all
the preceding ones is that this one remains open, and its extent
merges with almost the whole of the Recherche: which means,
among other things, that this analepsis will rejoin and pass be-
yond its own memory-created starting point without mention-
ing that point and seemingly without noticing it, swallowing it
up in one of its ellipses. We will come back to this particular
characteristic later. At the moment, let us only note this zigzag
movement, this initial — and as it were initiatory, or
propitiatory — stammering: 5-2 -5-5'-2'-5-l -5-4-3. . ., itself al-
ready contained, like all the rest, in the embryonic cell of the first


Narrative Discourse

five pages, which lead us from room to room, from period to
period, from Paris to Combray, from Doncieres to Balbec, from
Venice to Tansonville. Not a motionless shifting back and forth,
however, despite its repeated returns, since, thanks to it, a pin-
pointed Combray 1 is succeeded by a more spacious Combray II,
by an Amour de Swann that is earlier but has an already irrevers-
ible movement, by aNoms de pays: le nom, where finally the narra-
tive definitively sets in motion and adopts its pace.

These complexly structured openings, mimicking, as it were,
the unavoidable difficulty of beginning the better to exorcise it, are
seemingly part of the earliest and most lasting narrative tradi-
tion: we have already noted the sidewise movement at the start
of the Iliad, and must recall here that onto the convention of the
beginning in medias res was added or superimposed, for the
entire classical period, the convention of narrative embeddings
(X tells that Y tells that . . .) — embeddings which are still at work
(and we will return to this later) in jean Santeuil, and which allow
the narrator time to position his voice. The particular charac-
teristic of the exordium of the Recherche is obviously its multi-
plication of memory-created instances, and consequently its
multiplication of beginnings, among which each (except the last)
can seem afterward like an introductory prologue. First begin-
ning (absolute beginning): "For a long time I used to go to bed
early ..." Second beginning (ostensible beginning of the au-
tobiography), five pages later: "At Combray, as every afternoon
ended ..." Third beginning (appearance on stage of involun-
tary memory), twenty-six pages later: "And so it was that, for a
long time afterwards, when I lay awake at night and revived old
memories of Combray ..." Fourth beginning (resumption after
the madeleine, real beginning of the autobiography), four pages
later: "Combray at a distance, from a twenty-mile radius ..."
Fifth beginning, one hundred and seven pages later: ab ovo,
Swann in love (an exemplary novella if there ever was one, ar-
chetype of all the Proustian loves), conjoint (and hidden) births
of Marcel and Gilberte ("We will confess," Stendhal would say
here, "that, following the example of many serious authors, we
have begun the story of our hero a year before his birth." Is not
Swann to Marcel, mutatis mutandis and, I hope, with nothing



untoward in mind, what Lieutenant Robert is to Fabrice del
Dongo?) 16 — fifth beginning, thus: "To admit you to the 'little
nucleus,' the 'little group,' and 'little clan' at the Verdurins' ..."
Sixth beginning, one hundred and forty-nine pages later:
"Among the rooms which used most commonly to take shape
in my mind during my long nights of sleeplessness. . ." im-
mediately followed by a seventh and thus, as it should be, a final
beginning: "And yet nothing could have differed more utterly,
either, from the real Balbec than that other Balbec of which I had
often dreamed ..." This time, the movement is launched: after
this it will never stop.

Reach, Extent

I have said that, in its main articulations, the continuation of
the Recherche was arranged in conformity with chronological
order; but this general course does not exclude the presence of a
great many anachronies in small points: analepses and pro-
lepses, certainly, but also other forms that are more complex or
more subtle, perhaps more specific to Proustian narrative, and
that in any case are remote from both "real" chronology and
classical narrative temporality. Before taking up the analysis of
these anachronies, let us make clear that we are concerned here
only with temporal analysis, and furthermore temporal analysis
limited solely to questions of order: for the time being we are
setting aside questions of speed and frequency and a fortiori
characteristics of mood and voice, which can affect anachronies
as they can affect any other kind of narrative segment. In par-
ticular, we will disregard an essential distinction between, on
the one hand, the anachronies that the narrative takes direct
responsibility for, and that thus stay at the same narrative level
as their surroundings (example, lines 7-12 of the Iliad or the
second chapter of Cesar Birotteau), and, on the other hand, the
anachronies that one of the characters of the first narrative takes

16 But is not Swann's role in the bedtime scene symbolically paternal? After all,
it is he who deprives the child of its mother's presence. The legal father, on the
contrary, appears here with an unpardonable laxity, a bantering and suspect
willingness to oblige: "Go with the boy." What can we conclude from this


Narrative Discourse

on, and that thus appear at a second narrative level (example,
Books IX-XII of the Odyssey [Ulysses' speech], or Raphael de
Valentin's autobiography in the second part of La Peau de cha-
grin). Obviously we will again meet this question (which is not
specific to anachronies although it concerns them in the highest
degree) in the chapter on narrative voice.

An anachrony can reach into the past or the future, either
more or less far from the "present" moment (that is, from the
moment in the story when the narrative was interrupted to
make room for the anachrony): this temporal distance we will
name the anachrony's reach. The anachrony itself can also cover
a duration of story that is more or less long: we will call this its
extent. Thus when Homer, in Book XIX of the Odyssey, evokes
the circumstances long ago in which Ulysses, while an adoles-
cent, received the wound whose scar he still bears when
Euryclea is preparing to wash his feet, this analepsis (filling lines
394^166) has a reach of several decades and an extent of a few
days. So defined, the status of anachronies seems to be merely a
question of more or less, a matter of measurement particular to
each occasion, a timekeeper's work lacking theoretical interest.
It is, however, possible (and, I claim, useful) to categorize —
without too much emphasis — the characteristics of reach and
extent with respect to the ways in which they are connected to
certain "higher" moments in the narrative. This categorization
applies in basically the same way to the two main classes of
anachronies; but for convenience of exposition and to avoid the
risk of becoming too abstract, we will first handle analepses
exclusively, and broaden our procedure afterward.


Every anachrony constitutes, with respect to the narrative into
which it is inserted — onto which it is grafted — a narrative that is
temporally second, subordinate to the first in a sort of narrative
syntax that we met in the analysis we undertook above of a very
short fragment from Jean Santeuil. We will henceforth call the
temporal level of narrative with respect to which anachrony
is defined as such, "first narrative." Of course — and this we



have already verified — the embeddings can be more complex,
and an anachrony can assume the role of first narrative with
respect to another that it carries; and more generally, with re-
spect to an anachrony the totality of the context can be taken
as first narrative.

The narrative of Ulysses' wound deals with an episode that is
quite obviously earlier than the temporal point of departure of
the "first narrative" of the Odyssey, even if, according to this
principle, we allow "first narrative" to include the retrospective
tale Ulysses tells the Phaeacians, which goes back as far as the
fall of Troy. We can thus describe as external this analepsis
whose entire extent remains external to the extent of the first
narrative. We can do the same, for example, with the second
chapter of Cesar Birotteau, whose story, as the title clearly indi-
cates ("Les Antecedents de Cesar Birotteau"), takes place ear-
lier than the drama opened by the nocturnal scene of the first
chapter. Inversely, we will describe as internal analepsis the sixth
chapter of Madame Bovary, dealing with Emma's years in the
convent, which are obviously later than Charles's entrance at
school, which is the novel's starting point; or similarly, the be-
ginning of the Souffrances de I'inventeur, which, after the narra-
tive of the Parisian adventures of Lucien de Rubempre, serves
to acquaint the reader with David Sechard's life in Angouleme
during that period. 17 We can also imagine, and occasionally we
come across, mixed analepses, whose reach goes back to a point
earlier and whose extent arrives at a point later than the begin-
ning of the first narrative: so it is with the story of Des Grieux in
Manon Lescaut, which begins several years before the first meet-
ing with the Man of Quality and continues up to the time of the
second meeting, which is also the time of the narrating.

This distinction is not as useless as it might seem at first sight.
In effect, external analepses and internal analepses (or the inter-
nal part of mixed analepses) function for purposes of narrative
analysis in totally different ways, at least on one point that
seems to me essential. External analepses, by the very fact that
they are external, never at any moment risk interfering with the

'Balzac, Illusions perdues, Gamier ed., pp. 550-643.


Narrative Discourse

first narrative, for their only function is to fill out the first narra-
tive by enlightening the reader on one or another "antecedent."
This is obviously the case with some of the examples already
mentioned, and it is also, and just as typically, the case with
Un amour de Swann in the Recherche du temps perdu. The case is
otherwise with internal analepses: since their temporal field is
contained within the temporal field of the first narrative, they
present an obvious risk of redundancy or collision. We must
therefore examine these problems of interference more closely.
We will set aside at once the internal analepses that I propose
to call heterodiegetic, 18 that is, analepses dealing with a story line
(and thus with a diegetic content) different from the content (or
contents) of the first narrative. Such analepses deal, classically,
either with a character recently introduced whose "antecedents"
the narrator wants to shed light on, like Flaubert for Emma in
the chapter we referred to earlier; or they deal with a character
who has been out of sight for some time and whose recent past
we must catch up with, as is the case for David at the beginning
of the Souffrances de I'inventeur. These are, perhaps, the most
traditional functions of analepsis, and obviously the temporal
coinciding here does not entail real narrative interference. So it
is, for instance, when, at the Prince de Faffenheim's entrance
into the Villeparisis drawing room, a retrospective digression of
several pages informs us of the reasons for this appearance, that
is, the vicissitudes of the Prince's candidacy for the Academy of
Moral Sciences; 19 or when, reencountering Gilberte Swann who
has become Mile, de Forcheville, Marcel has the reasons for this
change in name explained to him. 20 Swann's marriage, the mar-
riages of Saint-Loup and "the Cambremer boy," the death of
Bergotte 21 thus overtake the main line of the story — which is
Marcel's autobiography — after the event, without in any way
disturbing the prerogative of the first narrative.

18 G. Genette, Figures 11 (Paris, 1969), p. 202.

19 RH I, 899-904/P II, 257-263.

20 RH II, 786-792/P III, 574-582.

21 RH I, 358-361/P I, 467-171; RH II, 849^856/P III, 664-673; RH II, 506-510/P III,



Very different is the situation of internal homodiegetic
analepses, that is, internal analepses that deal with the same line
of action as the first narrative. Here the risk of interference is
obvious, and even apparently unavoidable. In fact, now we
must once again differentiate two categories.

The first, which I will call completing analepses, or "returns,"
comprises the retrospective sections that fill in, after the event,
an earlier gap in the narrative (the narrative is thus organized by
temporary omissions and more or less belated reparations, ac-
cording to a narrative logic that is partially independent of the
passing of time). These earlier gaps can be ellipses pure and
simple, that is, breaks in the temporal continuity. Thus, Marcel's
stay in Paris in 1914 is recounted on the occasion of another
Parisian stay, this one in 1916, partially filling in the ellipsis of
several "long years" the hero spent in a clinic; 22 the meeting in
Uncle Adolphe's apartment with the Lady in pink 23 opens, in
the middle of the Combray narrative, a door onto the Parisian
side of Marcel's childhood — a side totally concealed, except for
this, until the third part of Swann. It is obviously in temporal
gaps of this kind that we must hypothetically place certain
events in Marcel's life known to us only by brief retrospective
allusions: a trip to Germany with his grandmother earlier than
the first trip to Balbec, a stay in the Alps earlier than the episode
of Doncieres, a trip to Holland earlier than the Guermantes
dinner, or again — appreciably more difficult to locate, given the
length of military service during that period — the years in the
military parenthetically evoked during the final stroll with Char-
ms. 24

But there is another type of gap, of a less strictly temporal
kind, created not by the elision of a diachronic section but by the

22 RH II, 900-913/P III, 737-755; cf. RH II, 889/P III, 723.

23 RH I, 55-60/P I, 72-80.

24 RH I, 544/P I, 718; RH I, 773/P II, 83; RH I, 1090/P II, 523; RH II, 954/P III, 808.
Supposing, of course, that we take these items of retrospective information
wholly seriously, which is the law of narrative analysis. The critic, however, for
his part can just as well take such allusions to be authorial lapses; or perhaps
Proust's biography is momentarily projected onto Marcel's.


Narrative Discourse

omission of one of the constituent elements of a situation in a
period that the narrative does generally cover. An example: the
fact of recounting his childhood while systematically concealing
the existence of one of the members of his family (which Proust
would be doing vis-a-vis his brother Robert if we took the Re-
cherche for a genuine autobiography). Here the narrative does
not skip over a moment of time, as in an ellipsis, but it sidesteps a
given element. This kind of lateral ellipsis we will call, conform-
ing to etymology and not excessively straining rhetorical usage,
a paralipsis. 25 Like temporal ellipsis, paralipsis obviously lends
itself very nicely to retrospective filling-in. For instance Swann's
death, or more precisely its effect on Marcel (for the death itself
could be considered external to the autobiography of the hero,
and thus heterodiegetic), was not recounted in its place; yet in
principle there is no room for a temporal ellipsis between
Swann's last appearance (at the Guermantes soiree) and the day
of the Charlus-Verdurin concert when the retrospective news of
his death slips in; 26 so we must assume that this very important
event in Marcel's affective life ("The death of Swann had been a
crushing blow to me at the time") was omitted laterally, in
paralipsis. An even more clear-cut example: the end of Marcel's
passion for the Duchesse de Guermantes, thanks to the quasi-
miraculous intervention of his mother, is the subject of a retro-
spective narrative with no specific date ("There had been a day
when"); 27 but since his ailing grandmother is involved in this
scene, we must obviously place it before the second chapter of
Guermantes II (I, 965); but we must also, of course, place it after I,
861-862, where we see that Oriane has not yet "ceased to interest
[him]." Yet there is no identifiable temporal ellipsis; Marcel has
therefore omitted to report to us in its place this nonetheless ex-
tremely important aspect of his inner life. But the most remarkable

25 The rhetoricians' paralipsis is, rather, a false omission, otherwise called
pretention. Here, paralipsis as a narrative trope is contrasted to ellipsis the way
put it aside is contrasted to leave it where it is. We will meet paralipsis again later as
an item of mood.

26 RH II, 518(P III, 199-201. Unless we take as ellipsis the iterative handling of
the first months of joint life with Albertine at the beginning of the Prisonniire.

27 RH I, 983/ P 11, 371.



case — although it is rarely picked up by critics, perhaps because
they refuse to take it seriously — is the mysterious "girl-cousin"
about whom we learn, when Marcel gives Aunt Leonie's sofa to
a go-between, that with her on this same sofa he experienced
"for the first time the sweets of love"; 28 and this happened no-
where else but at Combray, and at a fairly early date, since he
makes clear that the scene of the "initiation" 29 took place "one
hour when my Aunt Leonie had gotten up," and we know in
another connection that in her final years Leonie no longer left
her room. 30 Let us set aside the probable thematic value of this
belated confidence, and let us even admit that the omission of
the event from the narrative of Combray is a purely temporal
ellipsis: the omission of the character from the family tableau
perhaps for that reason comes even closer to being censorship.
This little cousin on the sofa will thus be for us — to each age its
own pleasures — analepsis on paralipsis.

Up to now we have examined the (retroactive) localization of
analepses as if they always involved a unique event to be placed
at one single point in past history. In fact, certain retrospections,
although dealing with individual events, can refer to iterative
ellipses, 31 that is, ellipses dealing not with a single portion of
elapsed time but with several portions taken as if they were alike
and to some extent repetitive. Thus, the meeting with the Lady
in pink can refer us to any day in the winter months when
Marcel and his parents were living in Paris, in any year before
the quarrel with Uncle Adolphe: an individual event, certainly,

28 RH 1, 440/P I, 578.

29 "Girl-cousin (a little one). My initiator: I, 578 [P/RH 1, 440]," imperturbably
and precisely notes the Clarac and Ferre index of the names of the characters.

30 It is true that she has two adjoining rooms, and goes into one while the
other is being aired out (RH I, 37-38/P I, 49). But if that were the situation, the
scene becomes extremely hazardous. On the other hand, the relationship is not
clear between this "sofa" and the bed described on p. 38 (RH I/P I, 50), with its
flowered quilt having a "nondescript, resinous, dull, indigestible, and fruity
smell" where Marcel when very young, "with an unconfessed gluttony," always
returned to "bury" himself. Let us leave this problem to the specialists, and
remember that in the "Confession d'une Jeune Fille" of Les Plaisirs et les jours the
"initiation" involves the fourteen-year-old heroine and a "cousin, a boy of fif-
teen . . . already very depraved" (Pleiade, p. 87; Pleasures and Regrets, trans.
Louise Varese |New York, 1948], p. 34).

31 About the iterative in general we will speak in Chapter 3.


Narrative Discourse

but for us its localization is on the order of the species or the
class (a winter) and not of the individual (a given winter). It is
the same a fortiori when the event recounted by the analepsis is
itself iterative in nature. Thus, in the jeunes Filles en fleurs, the
day of the first appearance of the "little band" ends with a
dinner at Rivebelle that is not the first; for the narrator, this
dinner is the opportunity for a look back at the preceding series,
a look back written mainly in the imperfect tense for repeated
action and telling of all the previous dinners in the account of a
single one: 32 clearly the ellipsis which this retrospection fills in
must itself be iterative. Similarly, the analepsis that ends the
Jeunes Filles, a final glance at Balbec after the return to Paris, 33 in
a synthetic way bears on the whole series of naps that Marcel,
on the doctor's order, had to take every morning until noon
during his entire stay, while his young friends were strolling
along the sunny jetty and the morning concert blared under his
windows; it thus allows this part of the Recherche to end not with
the greyness of a sad return home but with the glorious
pause — the golden stop — of a changeless summer sun.

With the second type of (internal) homodiegetic analepses,
which we will name precisely repeating analepses, or "recalls,"
we no longer escape redundancy, for in these the narrative
openly, sometimes explicitly, retraces its own path. Of course,
these recalling analepses can rarely reach very large textual di-
mensions; rather, they are the narrative's allusions to its own
past, what Lammert calls Ruckgriffe, or "retroceptions." 34 But
their importance in narrative economy, especially with Proust,
amply compensates for their limited narrative scope.

We must obviously set among these recalls the three reminis-
cences owed to involuntary memory during the Guermantes
matinee, all of which (contrary to that of the madeleine) refer to
an earlier time in the narrative: the stay in Venice, the train's
stop in front of a row of trees, the first morning by the sea at

32 RH I, 609-617/P I, 808-823.

33 RH I, 713-714/P I, 953-955.

34 Eberhart Lammert, Bauformen des Erzahlens (Stuttgart, 1955), Part II.



Balbec. 35 These are recalls in the purest form, deliberately cho-
sen or devised because of their casual and commonplace charac-
ter. But at the same time they suggest a comparison between
present and past, a comparison comforting for once, since the
moment of reminiscence is always euphoric, even if it revives a
past that in itself was painful: "I recognized that what seemed to
me now so delightful was that same row of trees which I had
found tedious both to observe and to describe." 36 The compari-
son between two situations that are similar and also different
often motivates as well recalls in which involuntary memory
does not play a role: for instance when the Due de Guermantes's
words about the Princesse de Parme ("she thinks you're charm-
ing") remind the hero — and give the narrator the opportunity to
remind us — of those identical words of Mme. de Villeparisis
about another "highness," the Princesse de Luxembourg. 37
Here the accent is on analogy. The accent is on contrast, on the
other hand, when Saint-Loup introduces his Egeria Rachel to
Marcel, who immediately recognizes her as the little prostitute
from earlier times, she "who, but a few years since, . . . used to
say to the procuress: 'To-morrow evening, then, if you want me
for anyone, you will send round, won't you?' " 38 This sentence
in effect reproduces almost verbatim what "Rachel when from
the Lord" said in the Jeunes Filles en fleurs: "That's settled then; I
shall be free to-morrow, if you have anyone you won't forget to
send for me." 39 The variant in Guermantes is, so to speak, al-
ready foreseen in these terms: "She would simply vary her for-
mula, saying indifferently: 'If you want me' or 'If you want
anybody.'" In this case the recall has an evidently obsessive
precision and puts the two sections in direct communication —
whence the interpolation in the second section of the paragraph

35 RH II, 997-999/P III, 866-869; cf. RH II, 820-840/P III, 623-655 RH II 988/P
III, 855, and RH I, 510-511/P I, 672-674.

36 RH II, 998/P III, 868. Let us remember that the feeling of boredom before the
row of trees had been for Marcel the sign of an abortive literary vocation, and
thus of the failure of his life.

37 RH I, 1022/P II, 425; cf. RH I, 531/P I, 700.

38 RH I, 827/P II, 158.
3 " RH I, 439/P I, 577.


Narrative Discourse

about Rachel's past conduct, which seems as if snatched from
the text of the first section. A striking example of migration, or,
if one wishes, of narrative scattering.

Again a comparison, in La Prisonniere, between the cowardice
Marcel shows toward Albertine and his courage earlier in front
of Gilberte, when he had "still enough strength left to give her
up." 40 This return retroactively confers on the past episode a
meaning that in its own time it did not yet have. Indeed, this is
the most persistent function of recalls in the Recherche, to modify
the meaning of past occurrences after the event, either by mak-
ing significant what was not so originally or by refuting a first
interpretation and replacing it with a new one.

The first modality is signaled very precisely by the narrator
himself when he writes about the incident of the syringas: "At
the actual moment, I saw nothing in all this that was not per-
fectly natural, at the most a little confused, but in any case
unimportant," and again: "incident the cruel significance of which
entirely escaped me and did not enter my mind until long after-
wards." 41 That significance will be delivered up by Andree after
Albertine's death, 42 and this case of deferred interpretation
yields us an almost perfect example of double narrative, first
from the (naive) viewpoint of Marcel, then later from the (en-
lightened) viewpoint of Andree and Albertine, when the clue,
now finally supplied, dissipates every kind of "confusion."
With much greater fullness, the late meeting with Mile, de
Saint-Loup, 43 daughter of Gilberte and Robert, will give Marcel
the opportunity for a general "replay" of the main episodes of
his existence, episodes which until then were lost to insignifi-
cance because of their dispersion and are now suddenly reas-
sembled, now made significant by being bound all together
amongst themselves, because all are bound now to the existence
of this child who was born Swann and Guermantes,

40 RH II, 622/P III, 344.

41 RH II, 415^116/P ID, 54-55: returning home with the syringas, Marcel bumps
into Andree who, making a pretext of some allergy, prevents him from going in
right awav. In fact, that day she had been in a sinful situation with Albertine.

42 RH II, 803-804/P III, 600-601.

43 RH II, 1125-1126/P III, 1029-1030.



granddaughter of the Lady in pink, grandniece of Charlus,
evoker of both the "two ways" of Combray but also of Balbec,
the Champs-Elysees, La Raspeliere, Oriane, Legrandin, Morel,
Jupien . . . : chance, contingency, arbitrariness now suddenly
wiped out, his life's portrait now suddenly "captured" in the
web of a structure and the cohesiveness of a meaning.

This principle of deferred or postponed significance 44 obvi-
ously fits perfectly into the mechanism of enigma, analyzed by
Barthes in SIZ; and that so sophisticated a work as the Recherche
should use this mechanism perhaps surprises those who place
this work at the antipodes from popular novels — which it no
doubt is in its significance and aesthetic value, but not always in
its techniques. There is something of "it was Milady" in the
Recherche, even if only in the humorous form of "it was my old
friend Bloch" in the Jeunes Filles, when the thundering anti-
semite emerges from his tent. 45 The reader will wait more than a
thousand pages before learning, at the same time as the hero (if
he has not already guessed on his own), the identity of the Lady
in pink. 46 After the publication of his article in Le Figaro, Marcel
receives a letter of congratulations signed "Sanilon," written in
a colloquial and charming style: "I was desolate at my inability
to discover who had written to me"; he will know later, and we
will know with him, that it was Theodore, the ex-grocer's assis-
tant and choirman of Combray. 47 Entering the Due de Guer-
mantes's library, Marcel passes a little, provincial, timid, and
shabby bourgeois: it was the Due de Bouillon! 48 A tall woman
makes overtures to him on the street: she will turn out to be
Mme. d'Orvilliers! 49 In the little train of La Raspeliere, a large
common woman with a massive face is reading the Revue des
deux mondes: she will turn out to be Princess Sherbatoff! 50 Some
time after Albertine's death, a blond girl glimpsed in the Bois,

44 See Jean-Yves Tadie, Proust et le roman (Paris, 1971), p. 124.

45 RH I, 558-559/P I, 738.

46 RH I, 907/P II, 267.

47 RH II, 798 and 874/P III, 591 and 701.

48 RH I, 1125 and II, 61/P II, 573 and 681.

49 RH I, 985 and II, 88/P II, 373 and 721.

50 RH II, 184 and 208/P II, 858 and 892.


Narrative Discourse

then on the street, casts his way a glance that inflames him: met
again in the Guermantes drawing room, she will turn out to be
Gilberte! 51 The technique is used so often, it so obviously forms
background and norm, that for contrast or deviation one can
sometimes play on its exceptional absence or zero degree. In the
little train of La Raspeliere there is a glorious girl with dark eyes,
magnolia skin, bold manners, a voice quick, cool, and jocular:
"T should so much like to see her again,' I exclaimed. 'Don't
worry, one always sees people again,' replied Albertine. In this
particular instance, she was wrong; I never saw again, nor did I
ever identify, the pretty girl with the cigarette." 52

But the most typical use of the recall in Proust is without
doubt when an event that at the time of its occurrence has al-
ready been provided with a meaning, has this first interpreta-
tion replaced afterward by another (not necessarily better) one.
This technique is obviously one of the most efficient methods for
circulating meaning in the novel and for achieving the perpetual
"reversal from pro to con" that characterizes the Proustian ap-
prenticeship to truth. Saint-Loup, in Doncieres, meeting Marcel
on the street, apparently does not recognize him and greets him
coldly as if he were a soldier: later we will learn that he had
recognized him but wished not to stop. 53 The grandmother, at
Balbec, insists with an irritating futility that Saint-Loup photo-
graph her with her beautiful hat on: she knew she was dying
and wanted to leave her grandson a memento that would not
show her looking poorly. 54 Mile. Vinteuil's friend, the profaner
of Montjouvain, devoted herself devoutly at the same period to
recreating note by note the indecipherable rough drafts of the
septet, 55 etc. We know the lengthy series of revelations and
confessions by which the retrospective or even posthumous
image of Odette, Gilberte, Albertine, or Saint-Loup dissolves
and reforms: thus, the young man who accompanied Gilberte
one particular evening on the Champs-Elysees "was Lea in male

si RH II, 777 and 786/P III, 563 and 574.

52 RH II, 202/P II, 883.

-™ RH I, 813 and 841/P II, 138 and 176.

54 RH I, 593 and II, 127-128/P I, 786 and II, 776.

55 RH I, 123-127 and II, 562/P I, 160-165 and III, 261.



attire"; 56 on the day of the walk in the suburban village and the
slap on the journalist's cheek, Rachel was for Saint-Loup only a
"screen," and at Balbec he secluded himself with the elevator
boy of the Grand Hotel; 57 the evening of the cattleyas, Odette
was coming from Forcheville's; 58 and there is the whole series of
belated adjustments of Albertine's relationships with Andree,
Morel, and various young girls of Balbec and elsewhere; 59 but
on the other hand, and by an even crueller irony, the sinful
liaison between Albertine and Mile. Vinteuil's friend, the reluc-
tant admission of which crystallized Marcel's passion, was pure
invention: "I stupidly thought that I might make myself seem
interesting to you by inventing the story that I had known the
girls quite well" 60 — the aim is achieved, but by another route
(jealousy, and not artistic snobbishness), and with what out-
come we know.

These revelations of the erotic habits of the male friend or the
loved woman are obviously capital. I would be tempted to find
even more capital ("capitalissime," in Proustian language)—
because it touches the very foundation of the hero's Wel-
tanschauung (the universe of Combray, the opposition of the two
ways, "deepest layer of my mental soil" 61 ) — the series of re-
interpretations for which the late stay at Tansonville will be the
occasion and Gilberte de Saint-Loup the unwitting medium. I
have already tried elsewhere 62 to show the importance, on vari-
ous levels, of the "verification" — which is a refutation — that
Gilberte inflicts on Marcel's system of thought when she reveals
to him not only that the sources of the Vivonne (which he imag-
ined like "something as extraterrestrial as the Gates of Hell")
were only "a sort of rectangular basin in which bubbles rose to
the surface," but also that Guermantes and Meseglise are not so
removed from each other, so "irreconcilable," as he had be-

56 RH I, 474 and II, 868/P I, 623 and III, 695.

57 RH I, 825-844 and II, 859/P I, 155-180 and III, 681.

58 RH I, 177 and 284/P I, 231 and 371.

59 RH II, 744-745/P III, 515; RH II, 751/P III, 525; RH II, 802-803/P III, 599-601.

60 RH II, 370 and 617-618/P II, 1120 and III, 337.

61 RH I, 141/P I, 184.

62 Figures (Paris, 1966), p. 60, and Figures II, p. 242.


Narrative Discourse

lieved, since in a single walk one can "go to Guermantes, taking
the road by Meseglise." The other side of those "new revela-
tions of existence" is this stupefying information that at the time
of the steep path at Tansonville and the flowering hawthorns
Gilberte was in love with him, and the unusual gesture she had
directed at him then was in fact too explicit an advance. 63 Marcel
understands then that he had understood nothing, and —
supreme truth — "that the true Gilberte — the true Albertine —
were perhaps those who had at the first moment yielded them-
selves in their facial expression, one behind the hedge of pink
hawthorn, the other upon the beach," and that he had thus,
through incomprehension — through excess of reflection —
"missed the boat" at that first moment.

With the misunderstood gesture of Gilberte, once again the
whole profound geography of Combray is reshaped. Gilberte, it
turns out, had wanted to take Marcel with her (and other
neighborhood scamps, including Theodore and his sister —
future chambermaid of Baroness Putbus and the very symbol of
erotic fascination) to the ruins of the donjon of Roussainville-
le-Pin: this same phallic donjon, a vertical "confidant," on the
horizon, of Marcel's solitary pleasures in the little room smelling
of orrisroot and of his roaming frenzies in the countryside of
Meseglise. 64 But he did not then suspect that the donjon was
even more: the real place — proffered, accessible and not recog-
nized, "in reality, and so close at hand" 65 — of forbidden plea-
sures. Roussainville, and by metonymy the whole Meseglise
way, 66 are already the Cities of the Plain, "a promised [and]
accursed land." 67 "Roussainville, within whose walls I had

"RH 1, 108 and II, 866/P 1, 141 and III, 694.

64 RH I, 10 and 121/P 1, 12 and 158.

<> 5 RH 11, 868/P III, 697.

f ' 6 That the Meseglise way incarnates sexuality this sentence clearly shows:
"The things for which at that time I so feverishly longed, she had been ready, if
only I had had the sense to understand and to meet her again, to let me taste in
my boyhood. More completely even than 1 had supposed, Gilberte had been in
those days truly part of the 'Meseglise way' " (RH II, 868/P 111, 697).

67 Roussainville under the thunderstorm is obviously (like Paris, later under
the enemy's fire) Sodom and Gomorrah under the thunderbolt from heaven:
"Before our eyes, in the distance, a promised or an accursed land, Roussainville,
within whose walls I had never penetrated, Roussainville was now, when the
rain had ceased for us, still being chastised, like a village in the Old Testament,



never penetrated": what missed opportunity, what regret! Or
what denial? Yes, as Bardeche says, Combray's geography, ap-
parently so innocent, is "a countryside which, like many others,
requires deciphering." 68 But this deciphering, along with oth-
ers, is already at work in the Temps retrouve, and it arises from a
subtle dialectic between the "innocent" narrative and its retro-
spective "verification": such in part are the function and impor-
tance of Proustian analepses.

We have seen how the determination of reach allowed us to
divide analepses into two classes, external and internal, depend-
ing on whether the point to which they reach is located outside
or inside the temporal field of the first narrative. The mixed
class — not, after all, much resorted to — is in fact determined by a
characteristic of extent, since this class consists of external
analepses prolonged to rejoin and pass beyond the starting
point of the first narrative. Extent, once more, controls the dis-
tinction we are now going to talk about as, returning to the two
examples from the Odyssey that we have already met, we now
compare them.

The first is the episode of Ulysses' wound. As we have already
noted, its extent is much less than its reach, much less even than
the distance separating the moment of the wound from the start-
ing point of the Odyssey (the fall of Troy). After having re-
counted the hunt on Parnassus, the battle against the wild boar,
the wound, the healing, the return to Ithaca, the narrative inter-
rupts its retrospective digression point-blank and, skipping over
several decades, comes back to the present scene. 69 The "return

by all the innumerable spears and arrows of the storm, which beat down
obliquely upon the dwellings of its inhabitants, or else had already received the
forgiveness of the Almighty, Who had restored to it the light of His sun, which
fell upon it in rays of uneven length, like the rays of a monstrance upon the
altar" (RH I, 116-117IP I, 152). We will note the presence of the verb "beat down
upon," a muffled redoubling of the link that unites this scene — ahead of time —
to the episode of M. de Charlus during the war, the flagellation functioning both as
"vice" ("sin") and as punishment.

68 Maurice Bardeche, Marcel Proust romancier (Paris, 1971), I, 269.

69 Let us remember that this passage, which some people challenge without
much evidence and despite Plato's testimony (Republic, I, 334 b), has been the
subject of a commentary by Auerbach (Mimesis, trans. Willard Trask [1953; rpt.
Garden City, N.Y., 1957], chap. 1).


Narrative Discourse

to the past" is thus followed by a leap forward (in other words,
an ellipsis), which leaves a whole long portion of the hero's life
in darkness. The analepsis here is as it were pinpointed, re-
counting a moment from the past that remains isolated in its
remoteness, and not seeking to join that moment to the present
by covering an intervening period which is not relevant to the
epic (since the subject of the Odyssey, as Aristotle observed, is
not Ulysses' life but only his return from Troy). I will name this
type of retrospection, which ends on an ellipsis without rejoin-
ing the first narrative, simply partial analepsis.

The second example is Ulysses' narration before the Phaea-
cians. This time, in contrast, having gone back as far as the point
where Fame to some degree lost sight of him — in other words,
to the fall of Troy — Ulysses brings his tale along until it rejoins
the first narrative, covering the entire period extending from the
fall of Troy to his arrival on Calypso's island: complete analepsis,
this time, which joins the first narrative without any gap between
the two sections of the story.

There is no point in dwelling here on the obvious differences
in function between these two types of analepsis. The first
serves solely to bring the reader an isolated piece of information,
necessary for an understanding of a specific moment of the ac-
tion. The second, tied to the practice of beginning in medias res,
aims at retrieving the whole of the narrative's "antecedents." It
generally forms an important part of the narrative, and some-
times, as in La Duchesse de Langeais or The Death of Ivan llych,
even presents the chief part of it, with the first narrative
functioning as the denouement in advance.

Until now we have looked from this point of view only at
external analepses, which we decreed complete inasmuch as
they rejoin the first narrative at its temporal starting point. But a
"mixed" analepsis like Des Grieux's narrative can be said to be
complete in a totally different sense since, as we have already
noted, it rejoins the first narrative not at that one's beginning
but at the very point (the meeting in Calais) when the first was
interrupted to give up its place to the second: in other words,
the extent of the analepsis is rigorously equal to its reach, and
the narrative movement completes a perfect round trip. It is
likewise in this sense that we can speak of complete internal



analepses, as in Les Souffrances de I'inventeur, where the retro-
spective narrative is brought up to the moment when David's
and Lucien's destinies again meet.

By definition, partial analepses pose no problem of joining or
narrative juncture: the analeptic tale plainly interrupts itself on
an ellipsis, and the first narrative picks up right where it had
stopped — picks up either implicitly and as if nothing had sus-
pended it, as in the Odyssey ("Now the old woman took the
scarred limb and passed her hands down it, and knew it by the
touch . . . "), or else explicitly, taking note of the interruption
and, as Balzac likes to do, emphasizing the explanatory function
that was already pointed out at the beginning of the analepsis by
the famous "this is why" or one of its variants. Thus, the big
return to the past in La Duchesse de Langeais, introduced by a
most explicit formulation ("Now here is the adventure that had
brought about the respective situations of the two characters in
that scene"), has an ending not less openly acknowledged: "The
feelings that stirred the two lovers when they met each other
again at the gate of the Carmelites and in the presence of a
Mother Superior should now be understood in all their inten-
sity, and the violence aroused in both of them will no doubt
explain the denouement of this adventure." 70 Proust, who de-
rided the Balzacian "this is why" in Contre Sainte-Beuve, was not
above imitating it at least once in the Recherche. 71 He is equally
capable of resumptions of the same type — like this one that fol-
lows the account of the negotiations about the Academy be-
tween Faffenheim and Norpois: "Thus it was that Prince von
Faffenheim had been led to call upon Mme. de Villeparisis" 72 —
or of resumptions at least explicit enough for the transition to
be perceptible at once: "And now, on my second return to
Paris, ..." or "as I turned over in my mind this recent meeting
with Saint-Loup . . . " 73 But most often his resumption is far
more discreet: the evocation of Swann's marriage, brought on

70 Gamier, pp. 214 and 341.

71 Contre Saint-Beuve, Pleiade ed., p. 271 (Marcel Proust on Art and Literature,
1896-1919, trans. S. T. Warner |New York, 1958], p. 173), and Recherche RH I,
159/P I, 208.

72 RH I, 904/P II, 263.

73 RH II, 913/P III, 755 and RH II, 919/P III, 762.


Narrative Discourse

by one of Norpois's answers in the course of a dinner, is abruptly
cut off by a return to the present conversation ("I began next
to speak of the Comte de Paris. . .")/ as is the evocation later
on of Swann's death, inserted between two of Brichot's sen-
tences with no transition (" 'No,' Brichot went on . . ."). 74 Some-
times the resumption is so elliptical that on a first reading one
experiences some difficulty discovering just where the tem-
poral leap takes place. Thus, when the performance of Vinteuil's
sonata at the Verdurins' reminds Swann of an earlier perfor-
mance, the analepsis, although introduced in the Balzacian way
that we have spoken of ("this is why"), ends, by contrast, with
no other mark of return than a simple paragraph break: "And
so, at last, he ceased to think of it. / But tonight, at Mme. Verdu-
rin's, scarcely had the little pianist begun to play when ..." In
the same way, during the Villeparisis matinee, when Mme.
Swann's arrival reminds Marcel of a recent visit by Morel, the
first narrative connects to the analepsis in a particularly offhand
manner: "I, as I gave him my hand, was thinking of Mme.
Swann and saying to myself with amazement, so far apart, so
different were they in my memory, that I should have hence-
forth to identify her with the 'Lady in pink.' / M. de Charlus was
not long in taking his place by the side of Mme. Swann." 75

As we see, for the attentive reader the elliptical character of
these resumptions, at the end of partial analepses, simply un-
derlines by asyndeton the temporal rupture. With completing
analepses the reverse difficulty obtains, resulting not from the
gap between analeptic narrative and first narrative but, on the
contrary, from their necessary junction. This junction could
hardly be without some degree of overlapping and thus an ap-
pearance of awkwardness, unless the narrator has the skill to
extract from this awkwardness a sort of playful charm. Here, in
Cesar Birotteau, is an instance of overlapping that is not taken
charge of — perhaps not noticed by the novelist himself. The
(analeptic) second chapter ends thus: "A few moments later,
Constance and Cesar were snoring peacefully"; the third chap-

74 RH I, 361/P I, 471 and RH II, 520/P III, 201.

75 RH I, 162/P 1, 211 and RH I, 907/P II, 267.



ter begins in these terms: "As he was going to sleep, Cesar was
afraid that the next day his wife would make some peremptory
objections, and he set himself to get up very early in order to
settle everything": we see that here the resumption is not with-
out a touch of incoherence. The linking up in Les Souff ranees de
I'inventeur is more successful, because here the tapestry worker
has been able to extract a decorative element from the difficulty
itself. Here is the opening of the analepsis: "While the venerable
churchman climbs the ramps of Angouleme, it is not useless to
explain the network of interests into which he was going to set
foot. / After Lucien's departure, David Sechard ..." Here is how
the first narrative resumes, more than one hundred pages fur-
ther on: "At the moment when the old cure of Marsac was
climbing the ramps of Angouleme to go inform Eve of the
condition her brother was in, David had been hidden for eleven
days only two doors from where the worthy priest had just
come out." 76 This play between the time of the story and the
time of the narrating (to tell of David's misfortunes "while" the
cure of Marsac climbs the staircase) will be discussed on its own
account in the chapter on voice; we see how it converts into
humor what was a burden.

The typical behavior of Proustian narrative seems to consist,
quite to the contrary, of eluding the juncture, either by dis-
simulating the end of the analepsis in the sort of temporal dis-
persion that iterative narrative procures (this is the case with
two retrospections concerning Gilberte in La Fugitive, one about
her adoption by Forcheville, the other about her marriage to
Saint-Loup), 77 or else by pretending to be unaware that the
point in the story where the analepsis closes had already been
reached by the narrative. Thus, in Combray, Marcel begins by
mentioning "the interruption which a visit from Swann once
made, and the commentary which he then supplied to the
course of my reading, which had brought me to the work of an
author quite new to me, called Bergotte"; he then goes into the
past to tell how he had discovered that author; six pages further

7h Gamier, pp. 550 and 643.

77 RH II, 792/P III, 582 and RH II, 856/P III, 676.


Narrative Discourse

on, again picking up the thread of his narrative, he makes the
linkage in these terms, as if he had not already named Swann
and called attention to his visit: "One Sunday, however, while I
was reading in the garden, I was interrupted by Swann, who
had come to call upon my parents. — 'What are you reading?
May I look? Why, it's Bergotte!'" 78 Whether trick, oversight, or
offhandedness, the narrative thus avoids acknowledging its own
footprints. But the boldest avoidance (even if the boldness is
pure negligence) consists of forgetting the analeptic character of
a section of narrative and prolonging that section more or less
indefinitely on its own account, paying no attention to the point
where it rejoins the first narrative. That is what happens in the
episode — famous for other reasons — of the grandmother's
death. It opens with an obviously analeptic beginning: "I went
upstairs, and found my grandmother not so well. For some time
past, without knowing exactly what was wrong, she had been
complaining of her health." Then the narrative that has been
opened in the retrospective mood continues uninterruptedly on
up to the death, without ever acknowledging and signaling the
moment (although indeed necessarily come to and passed be-
yond) when Marcel, returning from Mme. de Villeparisis's, had
found his grandmother "not so well." We can never, therefore,
either locate the grandmother's death exactly in relation to the
Villeparisis matinee, or decide where the analepsis ends and the
first narrative resumes. 79 The case is obviously the same, but
on a very much broader scale, with the analepsis opened in Noms
de pays: le pays. We have already seen that this analepsis will
continue to the last line of the Recherche without paying its re-
spects in passing to the moment of the late insomnias, although
these were its source in his memory and almost its narrative
matrix: another retrospection that is more than complete, with
an extent much greater than its reach, and which at an undeter-
mined point in its career is covertly transformed into an anticipa-
tion. In his own way — without proclaiming it and probably even
without perceiving it — Proust here unsettles the most basic

78 RH I, 68 and 74/P I, 90 and 97.

79 RH I, 928-964/P II, 298-345.



norms of narration, and anticipates the most disconcerting pro-
ceedings of the modern novel.


Anticipation, or temporal prolepsis, is clearly much less fre-
quent than the inverse figure, at least in the Western narrative
tradition — although each of the three great early epics, the Iliad,
the Odyssey, and the Aeneid, begins with a sort of anticipatory
summary that to a certain extent justifies the formula Todorov
applied to Homeric narrative: "plot of predestination." 80 The
concern with narrative suspense that is characteristic of the
"classical" conception of the novel ("classical" in the broad
sense, and whose center of gravity is, rather, in the nineteenth
century) does not easily come to terms with such a practice.
Neither, moreover, does the traditional fiction of a narrator who
must appear more or less to discover the story at the same time
that he tells it. Thus we will find very few prolepses in a Balzac, a
Dickens, or a Tolstoy, even if the common practice, as we have
already seen, of beginning in medias res (or yet, I may venture to
say, in ultimas res), sometimes gives the illusion of it. It goes
without saying that a certain load of "predestination" hangs
over the main part of the narrative in Manon Lescaut (where we
know, even before Des Grieux opens his story, that it ends with
a deportation), or a fortiori in The Death of Ivan llych, which
begins with its epilogue.

The "first-person" narrative lends itself better than any other
to anticipation, by the very fact of its avowedly retrospective
character, which authorizes the narrator to allude to the future
and in particular to his present situation, for these to some ex-
tent form part of his role. Robinson Crusoe can tell us almost at
the beginning that the lecture his father gave to turn him aside
from nautical adventures was "truly prophetic," even though at
the time he had no idea of it, and Rousseau, with the episode of
the combs, does not fail to vouch for not only his past innocence
but also the vigor of his retrospective indignation: "In writing

"Tzvetan Todorov, The Poetics of Prose (Ithaca, N.Y., and London, 1977), p.



Narrative Discourse

this I feel my pulse quicken yet." 81 Nonetheless, the Recherche
du temps perdu uses prolepsis to an extent probably unequaled in
the whole history of narrative, even autobiographical narra-
tive, 82 and is thus privileged territory for the study of this type
of narrative anachrony.

Here again, we can easily distinguish internal and external
prolepses. The limit of the temporal field of the first narrative is
clearly marked by the last nonproleptic scene, that is, for the
Recherche (if we draw into the "first narrative" that enormous
anachrony which begins on the Champs-Elysees and never
ends), without any possible doubt, the Guermantes matinee.
Now, it is well known that a certain number of episodes in the
Recherche take place at a point later in the story than this
matinee 83 (most, moreover, are told as digressions during this
same scene): for us these will thus be external prolepses. They
function most often as epilogues, serving to continue one or
another line of action on to its logical conclusion, even if that
conclusion takes place later than the day on which the hero
decides to leave the world and withdraw into his work: quick
allusion to Charlus's death; again an allusion (although more
detailed, with a highly symbolic reach), to the marriage of Mile,
de Saint-Loup ("this daughter, whose name and fortune gave
her mother the right to hope that she would crown the whole
work of social ascent of Swann and his wife by marrying a royal
prince, happening to be entirely without snobbery, chose for her
husband an obscure man of letters. Thus it came about that the
family sank once more, below even the level from which it had
started its ascent"); 84 final appearance of Odette, "showing
signs of senility," nearly three years after the Guermantes

81 Rousseau, Confessions, Pleiade ed., p. 20.

82 The Recherche contains more than twenty proleptic sections of significant
length, not counting simple allusions in the course of a sentence. The analepses
of like definition are not more numerous, but it is true that they take up, by their
extent, the quasi-totality of the text, and that it is atop that first retrospective
layer that analepses and prolepses of the second degree are set.

83 See Tadie, Proust et le roman, p. 376.

84 RH II, 950/P III, 804 and RH II, 1124/P III, 1028.



matinee; 85 Marcel's future experience as a writer, with his an-
guish in the face of death and the encroachments of social life,
the first reactions of readers, the first misapprehensions, etc. 86
The latest of these anticipations is the one, especially improvised
in 1913 for that purpose, which closes the Cote de chez Swann.
That tableau of the Bois de Boulogne "today," in direct contrast
to the one from the years of adolescence, is obviously very close
to the moment of narrating, since that last walk took place,
Marcel tells us, "this year," "one of those mornings early in
November," or so far as we can tell less than two months from
this moment. 87

One further step, therefore, and here we are in the narrator's
present. Prolepses of this type, quite numerous in the Recherche,
almost all correspond to the Rousseauistic model evoked above:
they are testimonies to the intensity of the present memory, and
to some extent authenticate the narrative of the past. For exam-
ple, apropos of Albertine: "Thus it is, calling a halt, her eyes
sparkling beneath her polo-cap, that I see her again to-day,
outlined against the screen which the sea spreads out behind
her"; apropos of the church in Combray: "And so even to-day,
in any large provincial town, or in a quarter of Paris which I do
not know well, if a passer-by who is putting me on the right
road shews me from afar, as a point to aim at, some belfry of a
hospital, or a convent steeple . . . "; apropos of the Baptistery of
Saint Mark's: "A time came for me when, remembering the
Baptistery . . . "; the end of the Guermantes soiree: "I can see all
that departing crowd now, I can see, if I be not mistaken in
placing him upon that staircase, . . . the Prince de Sagan." 88 And
especially, of course, apropos of the scene of his going to bed,

85 RH II, 1063/P III, 951-952.

86 RH II, 1133-1136/P III, 1039-1043.

87 RH I, 321-325/P I, 421-427. 1 will come back later to the difficulties raised by
this passage, written in 1913 but fictively (diegetically) contemporaneous with
the final narrating, and therefore later than the war.

88 RH I, 624/P I, 829; RH I, 50/P I, 67; RH [omitted in the English translation]/P
III, 646; RH II, 87/P II, 720; cf. RH I, 127/P I, 165 (on the village of Combray), RH I,
141-142/P I, 185 (on the Guermantes countryside), RH I, 142/P I, 186 (on the "two
ways"), RH I, 487/P I, 641 (on Mme. Swann), RH II, 202/P II, 883 (on the young
woman from the train of La Raspeliere), RH II, 822/P III, 625 (on Venice), etc.


Narrative Discourse

that poignant testimonial, already commented on in Mimesis,
which we cannot refrain from quoting here in its entirety — a
perfect illustration of what Auerbach calls "the symbolic om-
nitemporality" of the "remembering consciousness," but a per-
fect example also of fusion, quasi-miraculous fusion, between
the event recounted and the narrating instance, which is both
late (final) and "omnitemporal":

Many years have passed since that night. The wall of the staircase,
up which I had watched the light of his candle gradually climb,
was long ago demolished. And in myself, too, many things have
perished which, I imagined, would last for ever, and new struc-
tures have arisen, giving birth to new sorrows and new joys which
in those days I could not have foreseen, just as now the old are
difficult of comprehension. It is a long time, too, since my father
has been able to tell Mamma to "Go with the child." Never again
will such hours be possible for me. But of late I have been increas-
ingly able to catch, if I listen attentively, the sound of the sobs
which I had the strength to control in my father's presence, and
which broke out only when I found myself alone with Mamma.
Actually, their echo has never ceased: it is only because life is now
growing more and more quiet round about me that I hear them
afresh, like those convent bells which are so effectively drowned
during the day by the noises of the streets that one would suppose
them to have been stopped for ever, until they sound out again
through the silent evening air. 89

To the extent that they bring the narrating instance itself di-
rectly into play, these anticipations in the present constitute not
only data of narrative temporality but also data of voice: we will
meet them later under that heading.

89 RH I, 28/P I, 37. Auerbach's commentary, Mimesis, p. 481. Here we cannot
avoid thinking of Rousseau: "Nearly thirty years have passed since I left Bossey,
without my recalling to mind my stay there with any connected and pleasurable
recollections; but, now that I have passed the prime of life and am approaching
old age, I feel these same recollections springing up again while others disap-
pear; they stamp themselves upon my memory with features, the charm and
strength of which increase daily, as if, feeling life already slipping away, I were
endeavoring to grasp it again by its commencement" (Confessions, Pleiade, p. 21-
New York: Modern Library, 1945, p. 20).



Internal prolepses present the same kind of problem that
analepses of the same type do: the problem of interference, of
possible useless duplication between the first narrative and the
narrative taken on by the proleptic section. Here again we will
disregard heterodiegetic prolepses, for which this risk is nil,
whether the anticipation is internal or external; 90 and, among
homodiegetic prolepses, we will again differentiate between
those that fill in ahead of time a later blank (completing pro-
lepses), and those that — still ahead of time — double, however
slightly, a narrative section to come (repeating prolepses).

Examples of completing prolepses are the quick evocation, in
Combray, of Marcel's future years in school; the last scene be-
tween his father and Legrandin; the evocation, apropos of the
scene of the cattleyas, of the sequel of the erotic relations be-
tween Swann and Odette; the anticipatory descriptions of the
changing scene of the sea at Balbec; the advance notice, in the
middle of the first dinner at the Guermantes', of the long series
of like dinners, etc. 91 All these anticipations offset future ellipses
or paralipses. More subtle is the situation of the last scene in
Guermantes (the visit by Swann and Marcel to the Duchess)
which is, we know, 92 reversed with the first in Sodome (the

90 Here is a list of the main ones, in their order of succession in the text: RH II,
24/P II, 630, during the Jupien-Charlus meeting: sequel of the relations between
the two men, advantages Jupien derives from Charlus's goodwill, Francoise's
esteem for the moral qualities of the two inverts; RH II, 101-1 02/P II, 739-741, on
the return from the Guermantes soiree: the Duke's later conversion to
Dreyfusism; RH II, 529-530/P III, 214-216, before the Verdurin concert: Charlus's
later discovery of Morel's connections with Lea; RH II, 604-605/P III, 322-324, at
the end of the concert: Charlus's illness and forgetting his grudge against the
Verdurins; RH II, 932-934/P III, 779-781, during the walk with Charlus: sequel of
his relations with Morel, who loved a woman. We see that all of them have the
function of anticipating a paradoxical evolution, one of those unlooked-for re-
versals that Proustian narrative delights in.

91 RH I, 56/P I, 74; RH I, 99-102/P I, 129-133; RH I, 178-180/P I, 233-234; RH I,
510-511 and 605-608/P I, 673 and 802-806; RH I, 1082-1083/P II, 512-514; cf. RH I,
772/P II, 82-83 (on the room in Doncieres), RH II, 950/P III, 804 (the meeting with
Morel, two years after the walk with Charlus), RH II, 875-876/P III, 703-704 (the
meeting with Saint-Loup in society), etc.

92 "Now this wait on the staircase was to have for me consequences so consid-
erable, and to reveal to me a picture no longer Turneresque but ethical, of so
great importance, that it is preferable to postpone the account of it for a little
while by interposing first that of my visit to the Guermantes when I knew that
they had come home" (RH I, 1125/P II, 573).


Narrative Discourse

Charlus-Jupien "conjunction"), so that we must consider the
first as a prolepsis filling in the ellipsis opened, by this very
anticipation, between Sodome I and Sodome II, and the second as
an analepsis filling in the ellipsis opened in Guermantes by its
own delaying — a rearrangement of interpolations that is obvi-
ously motivated by the narrator's desire to have done with the
properly worldly aspect of the "Guermantes way" before ap-
proaching what he calls the "moral landscape" of Sodom and

Perhaps one will have noticed here the presence of iterative
prolepses, which, like analepses of the same kind, refer us to the
question of narrative frequency. Without discussing that question
here on its own account, I will simply note the characteristic
attitude, which consists, on the occasion of a first time (first kiss
of Swann and Odette, first sight of the sea at Balbec, first eve-
ning at the hotel in Doncieres, first dinner with the Guer-
mantes), of envisaging in advance the whole series of occur-
rences that the first one inaugurates. We will see in the next
chapter that most of the typical big scenes of the Recherche con-
cern an initiation of this kind ("debuts" of Swann at the Verdu-
rins', of Marcel at Mme. de Villeparisis's, at the Duchess's, at the
Princess's); the first meeting is obviously the best opportunity to
describe a scene or a milieu, and moreover it serves as a
paradigm of the others that follow. The generalizing prolepses
more or less make clear this paradigmatic function by opening
out a view onto the later series: "window in which I was, hence-
forward, to plant myself every morning. ..." They are thus, like
any anticipation, a mark of narrative impatience. But they have
also, it seems to me, an inverse weight that is perhaps more
specifically Proustian and that betokens rather a sentiment of
nostalgia for what Vladimir Jankelevitch once called the
"primultimateness" of the first time: that is, the fact that the first
time, to the very extent to which one experiences its inaugural
value intensely, is at the same time always (already) a last
time — if only because it is forever the last to have been the first,
and after it, inevitably, the sway of repetition and habit begins.
Before kissing her for the first time, Swann holds Odette's face
for a moment "at a little distance between his hands": this, the



narrator says, is to give his mind time to catch up and be witness
to the fulfillment of the dream it had cherished for so long. But
there is another reason:

Perhaps, moreover, Swann himself was fixing upon these features
of an Odette not yet possessed, not even kissed by him, on whom
he was looking now for the last time, that comprehensive gaze with
which, on the day of his departure, a traveller strives to bear away
with him in memory the view of a country to which he may never
return. 93

To possess Odette, to kiss Albertine for the first time, is to set
eyes for the last time upon the Odette not yet possessed, the
Albertine not yet kissed: so true is it that in Proust the event —
any event — is only the transition, evanescent and irretrievable (in
the Virgilian sense), from one habit to another.

Repeating prolepses, like analepses of the same type and for
reasons equally obvious, scarcely occur except as brief allusions:
they refer in advance to an event that will be told in full in its
place. As repeating analepses fulfill a function of recall with
respect to the addressee of the narrative, so repeating prolepses
play a role of advance notice, and I will designate them by this
term as well. The canonical formula for them is generally a "we
will see" or "one will see later," and the paradigm or prototype
is this notice apropos of the scene of the sacrilege at
Montjouvain: "We shall see, in due course, that for quite
another reason, the memory of this impression was to play an
important part in my life." An allusion, of course, to the
jealousy that the (false) revelation of relations between Albertine
and Mile. Vinteuil will provoke in Marcel. 94 The role these ad-
vance notices play in the organization and what Barthes calls the

93 RH I, 179/P I, 233. (My emphasis.)

94 RH I, 122/P I, 159 and RH II, 366/P II, 1114. But we must remember that
when he wrote this sentence before 1913 Proust had not yet "invented" the
character of Albertine, who will be worked out between 1914 and 1917. Clearly,
however, he has in mind for the scene of Montjouvain a "fallout" of this order,
which became specified only afterward: an advance notice, thus, doubly pro-



Narrative Discourse



"weaving" of the narrative is fairly obvious, through the expec-
tation that they create in the reader's mind— an expectation that
can be fulfilled immediately, in the case of those advance notices
with a very short reach or a nearby resolution, which, for exam-
ple, at the end of a chapter disclose the subject of the following
chapter by adumbrating it, as happens frequently in Madame
Bovary. 95 The more unbroken construction of the Recherche in
general excludes this kind of effect, but whoever remembers the
end of Part II, chapter 4, in Bovary ("She did not know that
when the gutters of a house are plugged up, the rain forms
pools on the roof; and so she continued to feel secure, when
suddenly she discovered a crack in the wall") will have no trou-
ble recognizing this model of metaphorized presentation in the
opening sentence of the last scene of the Temps retrouve:

But it is sometimes just at the moment when we think that every-
thing is lost that the intimation arrives which may save us; one
knocks at all the doors which lead nowhere, and then one stum-
bles without knowing it on the only door through which one can
enter— which one might have sought in vain for a hundred
years — and it opens of its own accord. 96

But most often the advance notice has a considerably longer
reach. We know that Proust prized the cohesiveness and the
architecture of his work, and that he suffered at seeing so many
effects of distant symmetry and "telescopic" correspondences
misunderstood. The separated publication of the various vol-
umes could not but aggravate the misunderstanding, and it is
certain that long-distance advance notices, as for the scene at
Montjouvain, were supposed to serve to reduce misunderstand-
ing by giving a provisional justification to episodes whose pres-
ence could otherwise seem adventitious and gratuitous. Here
are some further occurrences of this, in the order in which they
come: "As for Professor Cottard, we shall meet him again,

2 !; T C , hap - 3; XI ' Chap ' 4; "' cha P- 5; ��< cha P- 10 - "' ch aP- 13; III, chap 2
»RH II, 997/P III, 866. Cf., this time without metaphor, the anticipated sum-

^ n T%��™ V ^l Urin dinner (RH '' 193/P l 251) or of the Sainte-Euverte soiree
(l<ti i, 247/P 1, 322).

much later in the course of our story, with the 'Mistress,' Mme.
Verdurin, in her country house La Raspeliere"; "we shall see how
this sole social ambition that [Swann] had entertained for his
wife and daughter was precisely that one the realisation of
which proved to be forbidden him by a veto so absolute that
Swann died in the belief that the Duchess would never possibly
come to know them. We shall see also that, on the contrary, the
Duchesse de Guermantes did associate with Odette and Gil-
berte after the death of Swann"; "That I was one day to experi-
ence a grief as profound as that of my mother, we shall find in the
course of this narrative" (this grief is obviously the one that
Albertine's flight and death will provoke); "[Charlus] had recov-
ered [his health], only to fall later into the condition in which we
shall see him on the day of an afternoon party given by the Prin-
cesse de Guermantes." 97

We will not confuse these advance notices, which by defini-
tion are explicit, with what we should instead call mere advance
mentions, 98 simple markers without anticipation, even an allu-
sive anticipation, which will acquire their significance only later
on and which belong to the completely classic art of "prepara-
tion" (for example, having a character appear at the beginning
who will really step in only very much later, like the Marquis de
la Mole in the third chapter of he Rouge et le noir). We can con-
sider as such the first appearance of Charlus and Gilberte at
Tansonville, of Odette as the Lady in pink, or the first mention
of Mme. de Villeparisis on the fifteenth page of Swann, or again,
more obviously functional, the description of the bank at
Montjouvain, "on a level with [the] drawing-room, upstairs,
and only a few feet away from its window," which prepares for
Marcel's situation during the profanation scene; 99 or, more iron-
ically, the idea Marcel rejects of mentioning before M. de Crecy

" RH I, 332 and II, 190 ((.IP I, 433 and II, 866 ff .; RH I, 361 and II, 786 ((.IP I, 471
and III, 575 ff.; RH II, 122 and II, 669 ((.IP II, 768 and III, 415 ff.; RH II, 951 and 11,
992/P III, 805 and III, 859. (My emphasis.)

98 Cf. Raymonde Debray-Genette, "Les Figures du recit dans Un coeur simple,"
Poetique, 3 (1970).

<><> RH I, 108-109/P 1, 141; RH I, 57/P I, 76; RH I, 15/P I, 20; RH I, 86 and 122/P I,
113 and 159.


Narrative Discourse



< I ill


what he believes to be Odette's former "code name," which
prepares for the subsequent revelation (by Charlus) of the au-
thenticity of this name and of the real relationship between the
two characters. 100 The difference between advance notice and
advance mention is clearly discernible in the way in which Proust
prepares, in several stages, for Albertine's entrance. The first
reference, in the course of a conversation at the Swanns': Alber-
tine is named as niece of the Bontemps, and deemed "the quain-
test spectacle" by Gilberte — simple advance mention. The sec-
ond reference — another advance mention — by Mme. Bontemps
herself, who describes her niece as having "impudence," as
being a "little wretch, ... as cunning as a monkey": she has
publicly reminded a minister's wife that the latter's father was a
scullion; this description will be explicitly recalled very much
later, after Albertine's death, and held up as the "insignificant
seed [which] would develop and would one day overshadow
the whole of my life." The third reference — this time a genuine
advance notice:

There was a scene at home because I did not accompany my father
to an official dinner at which the Bontemps were to be present
with their niece Albertine, a young girl still hardly more than a
child. So it is that the different periods of our life overlap one
another. We scornfully decline, because of one whom we love and
who will some day be of so little account, to see another who is of
no account today, with whom we shall be in love tomorrow, with
whom we might, perhaps, had we consented to see her now, have
fallen in love a little earlier and who would thus have put a term to
our present sufferings, bringing others, it is true, in their place. un

Unlike the advance notice, the advance mention is thus in gen-
eral, at its place in the text, only an "insignificant seed," and
even an imperceptible one, whose importance as a seed will not
be recognized until later, and retrospectively. 102 But we must

10 »RH II, 345 and 589/P II, 1085 and III, 301.

"" RH I, 391/P I, 512; RH I, 455/P I, 598, cf. RH II, 1027/P III, 904; RH I 476/P I

102 "The 'soul' of any function is, as it were, its seedlike quality, which enables
the function to inseminate the narrative with an element that will later come to
maturity" (Roland Barthes, "An Introduction to the Structural Analysis of Nar-
rative," NLH, 6 [Winter 1975], 244).

consider the possible (or rather the variable) narrative competence
of the reader, arising from practice, which enables him both to
decipher more and more quickly the narrative code in general or
the code appropriate to a particular genre or a particular work,
and also to identify the "seeds" when they appear. Thus, no
reader of Ivan Ilych (helped, it is true, by the presentation of the
denouement in advance, and by the very title) can fail to identify
Ivan's fall against the French-window fastener as the instrument
of destiny, as the beginning of the death struggle. Moreover,
this very competence is what the author relies on to fool the
reader by sometimes offering him false advance mentions, or
snares 103 — well known to connoisseurs of detective stories. Once
the reader has acquired this second-degree competence of
being able to detect and thus to outmaneuver the snare, the
author is then free to offer him false snares (that are genuine
advance mentions), and so on. Proustian believability, of
course — based, as Jean-Pierre Richard puts it, on the "logic of
inconsistency" 104 — plays on (particularly in what concerns
homosexuality and its subtle variant, heterosexuality) this com-
plex system of frustrated expectations, disappointed suspicions,
surprises looked forward to and finally all the more surprising in
being looked forward to and occurring nonetheless — by virtue
of this principle for all purposes, that "The laborious process of
causation . . . sooner or later will bring about every possible ef-
fect, including (consequently) those which one had believed to
be most nearly impossible": 105 a warning to connoisseurs of
"psychological laws" and realistic motivations.

Still, before leaving narrative prolepses there is a word to say
about their extent, and the distinction possible here too between
partial and complete prolepses — if one is willing to grant com-
pleteness to anticipations prolonged in the time of the story up
to the "denouement" (for internal prolepses) or up to the narrat-
ing moment itself (for external or mixed prolepses). I find hardly
any examples of completeness, and it seems that in fact all pro-
lepses are of the partial type, often interrupted in as abrupt a

"" See Roland Barthes, S/Z (New York, 1974), p. 32.

104 Jean-Pierre Richard, "Proust et l'objet hermeneutique," Poetique, 13 (1973).

'RH I, 361 /P I, 471.


Narrative Discourse

way as they were begun. Marks of prolepsis: "to anticipate for a
moment, since I am still finishing my letter to Gilberte ... ";
"interrupting for a few moments our narrative, which shall be re-
sumed immediately after the closure of this parenthesis ..."; "to
anticipate a little for I am still at Tansonville . . . "; "the next day,
to anticipate . . . "; "I take a leap of many years . . . " 106 Marks of the
end of prolepsis and return to the first narrative: "To return to
this first evening at the Princesse de Guermantes's . . . "; "but it
is time to rejoin the Baron as he advances with Brichot and myself
towards the Verdurins' door. . . "; "to go backwards, to the Ver-
durin soiree. . . "; "But I must return to my narrative. . . "; "But
we have anticipated, and let us now go back three years, to the after-
noon party which is being given by the Princesse de Guerman-
tes . . . " 107 We see that Proust does not always retreat from the
burden of explicitness.

The importance of "anachronic" narrative in the Recherche du
temps perdu is obviously connected to the retrospectively syn-
thetic character of Proustian narrative, which is totally present in
the narrator's mind at every moment. Ever since the day when
the narrator in a trance perceived the unifying significance of his
story, he never ceases to hold all of its threads simultaneously,
to apprehend simultaneously all of its places and all of its
moments, to be capable of establishing a multitude of "tele-
scopic" relationships amongst them: a ubiquity that is spatial
but also temporal, an "omnitemporality" perfectly illustrated by
the passage in the Temps retrouve where the hero, in the pres-
ence of Mile, de Saint-Loup, reconstitutes in a flash the "net-
work of [entangled] memories" that his life has become, and
that will become the fabric of his work. 108

But the very ideas of retrospection or anticipation, which
ground the narrative categories of analepsis and prolepsis in

"> 6 RH II, 101/P II, 739; RH II, 529/P III, 214; RH II, 875/P III, 703; RH II 932/P III
779; RH II, 950/P III, 803. (My emphasis.)

>�� 7 RH II, 85/P II, 716; RH II, 530/P III, 216; RH II, 952/P III, 806; RH II, 1064/P III,
952. (My emphasis.) Of course, these signs of the organization of the narrative
are in themselves marks of the instance of narrating, which we will meet again as
such in the chapter on voice.

,08 RHII, 1126/PIII, 1030.



"psychology," take for granted a perfectly clear temporal con-
sciousness and unambiguous relationships among present,
past, and future. Only because the exposition required it, and at
the cost of excessive schematization, have I until now postulated
this to have always been so. In fact, the very frequency of inter-
polations and their reciprocal entanglement often embroil mat-
ters in such a way as to leave the "simple" reader, and even the
most determined analyst, sometimes with no way out. To con-
clude this chapter we shall examine some of these ambiguous
structures, which bring us to the threshold of achrony pure and

Toward Achrony

Since our first microanalyses we have met examples of com-
plex anachronies: second-degree prolepses in the section taken
from Sodome et Gomorrhe (anticipation of Swann's death on an-
ticipation of his luncheon with Bloch), analepses on prolepses
(retrospection of Francoise at Combray on that same anticipation
of Swann's funeral), and prolepses on analepses (twice in the
excerpt from ]ean Santeuil, recalls of past plans). Such second- or
third -degree effects are likewise frequent in the Recherche at the
level of large or medium-sized narrative structures, even with-
out taking into account that first degree of anachrony which the
quasi-totality of the narrative is.

The typical situation evoked in our fragment of Jean Santeuil
(memories of anticipations) has taken root in the Recherche in the
two characters born by fission from the original hero. The return
to Swann's marriage, in the Jeunes Filles, includes a retrospective
evocation of the plans of worldly ambition for his daughter and
his (future) wife:

But when Swann in his daydreams saw Odette as already his wife
he invariably formed a picture of the moment in which he would
take her — her, and above all her daughter — to call upon the Prin-
cesse des Laumes (who was shortly ... to become Duchesse de
Guermantes). . . . His heart would soften as he invented — uttering
their actual words to himself — all the things that the Duchess
would say of him to Odette, and Odette to the Duchess. . . . He
enacted to himself the scene of this introduction with the same
precision in each of its imaginary details that people shew when


Narrative Discourse




they consider how they would spend, supposing they were to
win it, a lottery prize the amount of which they have arbitrarily
determined. 109

This "waking dream" is proleptic insofar as Swann entertains it
before his marriage, analeptic insofar as Marcel recalls it after
that marriage, and the two movements come together to cancel
each other out, in this way perfectly superimposing the fantasy
on its cruel refutation by the facts, since here is Swann married
for several years to an Odette still unwelcome in the Guer-
mantes salon. It is true that he married Odette when he no
longer loved her, and that "the creature that, in [him], had so
longed to live, had so despaired of living all its life in company
with Odette, . . . that creature was extinct." So here now face to
face in ironic contradiction are the earlier resolutions and the
present realities: resolution to elucidate some day the mysteri-
ous relations between Odette and Forcheville, replaced by a
total lack of curiosity:

Formerly, while his sufferings were still keen, he had vowed that,
as soon as he should have ceased to love Odette, and so to be
afraid either of vexing her or of making her believe that he loved
her more than he did, he would afford himself the satisfaction of
elucidating with her, simply from his love of truth and as a histori-
cal point, whether or not she had had Forcheville in her room that
day when he had rung her bell and rapped on her window with-
out being let in, and she had written to Forcheville that it was an
uncle of hers who had called. But this so interesting problem, of
which he was waiting to attempt the solution only until his
jealousy should have subsided, had precisely lost all interest in
Swann's eyes when he had ceased to be jealous.

Resolution to express some day the indifference that lay ahead,
replaced by the circumspection of real indifference:

But whereas at that other time he had made a vow that if ever he
ceased to love her whom he did not then imagine to be his future
wife, he would implacably exhibit to her an indifference that

109 RH I, 360-361/PI, 470.

would at length be sincere, so as to avenge his pride that had so
long been trampled upon by her — of those reprisals which he
might now enforce without risk. . ., of those reprisals he took no
more thought; with his love had vanished the desire to shew that
he was in love no longer.

The contrast, via the past, between anticipated present and real
present, in Marcel when he is finally "cured" of his passion for
Gilberte: "I had no desire now to see her, not even that desire to
shew her that I did not wish to see her which, every day, when I
was in love with her, I vowed to myself that I would flaunt
before her, when I should be in love with her no longer." Or,
with slightly different psychological significance, when Marcel
again, now Gilberte's "chum" and an intimate in the Swann
dining room, in order to measure the progress he has made,
tries in vain to recover the feeling he had earlier of how inacces-
sible this "inconceivable place" was — not without attributing to
Swann himself analogous thoughts about his life with Odette,
that once "unhoped-for paradise" not to be imagined without
turmoil but now a prosaic and totally charmless reality. 110 What
one had planned does not occur; what one had not dared to
hope for materializes, but only at the moment when one no
longer desires it. In both cases the present superimposes itself
on the previous future whose place it has taken: a retrospective
refutation of a mistaken anticipation.

An inverse movement — a recall that is anticipated, a detour
no longer by the past but by the future — occurs each time the
narrator explains in advance how he will later, after the event,
be informed of a present incident (or of its significance). So, for
example, in telling of a scene between M. and Mme. Verdurin,
he specifies that it will be reported to him by Cottard "a few
years later." The seesawing speeds up in this indication in Com-
bray: "Many years later we discovered that, if we had been fed
on asparagus day after day throughout that whole season, it was
because the smell of the plants gave the poor kitchen-maid, who

110 RH I, 361/P I, 471; RH I, 399-400/P I, 523; RH I, 401/P I, 525; RH II, 83/P II,
713; RH I, 410/P I, 537-538.


Narrative Discourse




had to prepare them, such violent attacks of asthma that she was
finally obliged to leave my aunt's service." 111 It becomes almost
instantaneous in this sentence from La Prisonniere: "I learned
that a death had occurred that day which distressed me greatly,
that of Bergotte" — so elliptical, so discreetly misshaped that the
reader at first thinks he has read: "I learned that day that a death
had occurred." 112 There is the same zigzag round trip when the
narrator introduces a present, or even a past event through the
anticipation of the memory he will have of it later, as we have
already seen for the final pages of the Jeunes Filles en fleurs,
which carry us forward to the first weeks at Balbec through the
future memories of Marcel in Paris; similarly, when Marcel sells
Aunt Leonie's sofa to a go-between, we learn that only "very
much later" will he remember having, very much earlier, used
that sofa with the enigmatic cousin we have already spoken of:
analepsis on paralipsis we called it then, a formula we must
now complete by adding via prolepsis. These narrative contor-
tions would doubtless be enough to bring down upon the hy-
pothetical young lady the suspicious, albeit kindly, glance of
the hermeneut.

Another effect of double structure is that a first anachrony
may invert — necessarily inverts — the relationship between a
second anachrony and the order of arrangement of the events in
the text. Thus, the analeptic status of Un amour de Swann has the
effect that an anticipation (in the time of the story) is able to refer
to an event already covered by the narrative: when the narrator
compares the vesper anguish of Swann deprived of Odette to
the anguish he himself will suffer "some years later" on the
nights when this same Swann will come to dine at Combray,
this diegetic advance notice is at the same time a narrative recall
for the reader, since he has already read the narrative of that
scene some one hundred and ninety pages "earlier"; inversely
and for the same reason, the reference to Swann's earlier an-
guish, in the narrative of Combray, is for the reader an advance

111 RH II, 607/P III, 326; RH I, 95/P I, 124.

1,2 RH II, 506/P III, 182. The Clarac-Ferre resume (P III, 1155) conveys it thusly:
"I learn that day of the death of Bergotte."

notice of the forthcoming narrative of Un amour de Swann. 113 The
specific formula of such double anachronies would thus be
something like this: "It would happen later, as we have already
seen," or: "It had already happened, as we will see later." Retro-
spective advance notices? Anticipatory recalls? When later is
earlier, and earlier later, defining the direction of movement
becomes a delicate task.

These proleptic analepses and analeptic prolepses are so
many complex anachronies, and they somewhat disturb our
reassuring ideas about retrospection and anticipation. Let us
again recall the existence of open analepses (analepses whose
conclusion cannot be localized), which therefore necessarily en-
tails the existence of temporally indefinite narrative sections. But
we also find in the Recherche some events not provided with any
temporal reference whatsoever, events that we cannot place at
all in relation to the events surrounding them. To be unplaceable
they need only be attached not to some other event (which
would require the narrative to define them as being earlier or
later) but to the (atemporal) commentarial discourse that accom-
panies them — and we know what place that has in this work. In
the course of the Guermantes dinner, apropos of Mme. de Var-
ambon's obstinacy in relating Marcel by marriage to Admiral
Jurien de la Graviere (and thus, by extension, apropos of the so
frequently made analogous errors in society), the narrator
evokes the error of a friend of the Guermantes' who was rec-
ommending himself to Marcel by making use of the name of a
cousin, Mme. de Chaussegros, a person totally unknown to the
narrator: one can assume that this anecdote, which implies a
certain progress in Marcel's social career, occurs later than the
Guermantes dinner, but nothing permits us to affirm this. After
the scene of the missed introduction to Albertine, in the Jeunes
Filles en fleurs, the narrator offers some reflections on the subjec-
tivity of the feeling of love, then illustrates this theory with the
example of a drawing master who had never known the color of
the hair of a mistress he had passionately loved and who had left

13 RH I, 228 and 23-24/P I, 297 and 30-31.

W 31


Narrative Discourse



him a daughter ("I never saw her except with a hat on"). 114
Here, no inference from the content can help the analyst define
the status of an anachrony deprived of every temporal connec-
tion, which is an event we must ultimately take to be dateless
and ageless: to be an achrony.

Now, it is not only such isolated events that express the narra-
tive's capacity to disengage its arrangement from all depen-
dence, even inverse dependence, on the chronological sequence
of the story it tells. The Recherche presents, at least in two places,
genuine achronic structures. At the end of Sodome, the itinerary of
the "Transatlantic" and its sequence of stops (Doncieres,
Maineville, Grattevast, Hermenonville) gives rise to a short nar-
rative sequence 115 whose order of succession (Morel's misad-
venture at the brothel in Maineville — meeting with M. de Crecy
at Grattevast) owes nothing to the temporal connection between
the two events composing it and everything to the fact (itself,
however, diachronic, but not a diachrony of the events re-
counted) that the little train goes first to Maineville, then to
Grattevast, and that these stations evoke in the narrator's mind,
in that order, anecdotes connected to them. 116 As J. P. Houston
has rightly noted in his study of temporal structures in the Re-
cherche, 117 this "geographic" ordering does no more than repeat
and make clear the ordering — more implicit but more important
in every respect — of the last forty pages of Combray, There the
narrative order is governed by the opposition Meseglise way/
Guermantes way, and by the sites' increasing distance from the
family home in the course of an atemporal and synthetic
walk. 118 First appearance of Gilberte; farewell to the hawthorns;
meeting with Swann and Vinteuil; Leonie's death; profanation

114 RH I, 1072/P II, 498; RH I, 645/P I, 858-859.

115 RH II, 338-346/P II, 1075-1086.

116 "1 confine myself at present, as the train halts and the porter calls out
'Doncieres,' 'Grattevast,' 'Maineville,' etc., to noting down the particular mem-
ory that the watering-place or garrison town recalls to me" (RH II, 339/P II, 1076).

117 ]. P. Houston, "Temporal Patterns in A.L.R.T.P.," French Studies, 16(1962),

118 The greater part of this sequence belongs for this reason in the category of
the iterative. For the moment I am disregarding that aspect in order to examine
only the order of succession of the singular events.

scene at the Vinteuils'; appearance of the Duchess at church;
sight of the steeples of Martinville— this succession has no con-
nection to the temporal order of the events composing it, or only
a partially coincidental connection. The succession depends es-
sentially on the location of the sites (Tansonville— Meseglise
plain— Montjou vain — return to Combray— Guermantes way)
and thus on a very different temporality: on the opposition be-
tween the days of the walk to Meseglise and the days of the walk
toward Guermantes and, within each of these two series, on the
approximate order of the "stations" of the walk. Only by naively
confusing the narrative's syntagmatic order with the story's
temporal order does one imagine, as hurried readers do, that the
meeting with the Duchess or the episode of the steeples comes
later than the scene at Montjouvain. The truth is that the nar-
rator had the clearest of reasons for grouping together, in defiance
of all chronology, events connected by spatial proximity, by cli-
matic identity (the walks to Meseglise always take place in bad
weather, those to Guermantes in good weather), or by thematic
kinship (the Meseglise way represents the erotic-affective side of
the world of childhood, that of Guermantes its aesthetic side); he
thus made clear, more than anyone had done before him and bet-
ter than they had, narrative's capacity for temporal autonomy." 9

But it would be utterly vain to think of drawing definitive con-
clusions merely from an analysis of anachronies, which illustrate
simply one of the constitutive features of narrative temporality.
It is fairly obvious, for example, that distortions of speed con-
tribute to emancipation from narrative temporality quite as
much as transgressions of chronological order do. These are the
subject of our next chapter.

119 Having christened the anachronies by retrospection and anticipation
analepses and prolepses, we could give the name syllepses (the fact of taking
together)— temporal or other— to those anachronic groupings governed by one or
another kinship (spatial, temporal, or other). Geographical syllepsis, for exam-
ple, is the principle of narrative grouping in voyage narratives that are embel-
lished by anecdotes, such as the Memoires d'un touriste or he Rhin Thematic
syllepsis governs in the classical episodic novel with its numerous insertions of
"stones," justified by relations of analogy or contrast. We will meet the notion of
syllepsis again apropos of iterative narrative, which is another variety of it.




2 Duration


At the beginning of the last chapter I recalled what difficulties
the very idea of "time of the narrative" runs up against in writ-
ten literature. It is obviously apropos of duration that these diffi-
culties are so strongly felt, for the data of order, or of frequency,
can be transposed with no problem from the temporal plane of
the story to the spatial plane of the text: to say that episode A
comes "after" episode B in the syntagmatic arrangement of a
narrative text or that episode C is told "twice" is to make state-
ments that have an obvious meaning and that can be clearly
compared with other assertions such as "event A is earlier than
event B in the story's time" or "event C happens only once."
Here, therefore, comparison between the two planes is legiti-
mate and relevant. On the other hand, comparing the "dura-
tion" of a narrative to that of the story it tells is a trickier opera-
tion, for the simple reason that no one can measure the duration
of a narrative. What we spontaneously call such can be nothing
more, as we have already said, than the time needed for read-
ing; but it is too obvious that reading time varies according to
particular circumstances, and that, unlike what happens in
movies, or even in music, nothing here allows us to determine a
"normal" speed of execution.
The reference point, or degree zero, which in matters of order




was the concurrence between diegetic sequence and narrative
sequence, and which here would be rigorous isochrony between
narrative and story, is now therefore absent — even if it be true,
as Jean Ricardou notes, that a scene with dialogue (supposing it
unadulterated by any intervention of the narrator and without
any ellipsis) gives us "a sort of equality between the narrative
section and the fictive section." 1 It is I who emphasize "sort," in
order to insist on the unrigorous, and especially unrigorously
temporal, nature of this equality. All that we can affirm of such a
narrative (or dramatic) section is that it reports everything that
was said, either really or fictively, without adding anything to it;
but it does not restore the speed with which those words were
pronounced or the possible dead spaces in the conversation. In
no way, therefore, can it play the role of temporal indicator; it
would play that role only if its indications could serve to mea-
sure the "narrative duration" of the differently paced sections
surrounding it. Thus a scene with dialogue has only a kind of
conventional equality between narrative time and story time, and
later we will utilize it in this way in a typology of the traditional
forms of narrative duration, but it cannot serve us as reference
point for a rigorous comparison of real durations.

We must thus give up the idea of measuring variations in
duration with respect to an inaccessible, because unverifiable,
equality of duration between narrative and story. But the iso-
chronism of a narrative may also be defined — like that of a pen-
dulum, for example — not relatively, by comparing its duration
to that of the story it tells, but in a way that is more or less
absolute and autonomous, as steadiness in speed. By "speed" we
mean the relationship between a temporal dimension and a
spatial dimension (so many meters per second, so many seconds
per meter): the speed of a narrative will be defined by the rela-
tionship between a duration (that of the story, measured in sec-
onds, minutes, hours, days, months, and years) and a length

1 Jean Ricardou, Problemes du nouveau roman (Paris, 1967), p. 164. Ricardou
contrasts narrating to fiction in the sense in which I contrast narrative (and some-
times narrating) to story (or diegesis): "narrating is the manner of telling, fiction is
what is told" (p. 11).


Narrative Discourse

(that of the text, measured in lines and in pages). 2 The isochro-
nous narrative, our hypothetical reference zero, would thus be
here a narrative with unchanging speed, without accelerations
or slowdowns, where the relationship duration-of-story/
length-of-narrative would remain always steady. It is doubtless
unnecessary to specify that such a narrative does not exist, and
cannot exist except as a laboratory experiment: at any level of
aesthetic elaboration at all, it is hard to imagine the existence of a
narrative that would admit of no variation in speed — and even
this banal observation is somewhat important: a narrative can do
without anachronies, but not without anisochronies, or, if one
prefers (as one probably does), effects of rhythm.

Detailed analysis of these effects would be both wearying and
devoid of all real rigor, since diegetic time is almost never indi-
cated (or inferable) with the precision that would be necessary.
The analysis is relevant, therefore, only at the macroscopic level,
that of large narrative units, granting that the measurement for
each unit covers only a statistical approximation. 3

If we want to draw up a picture of these variations for the
Recherche du temps perdu, we must decide at the very beginning
what to consider as large narrative articulations, and then, to
measure their story time, we must have at our disposal an ap-
proximately clear and coherent internal chronology. If the first
datum is fairly easy to establish, the second is not.

So far as narrative articulations are concerned, we must ob-
serve first that they do not coincide with the work's visible di-
visions into parts and chapters supplied with titles and num-
bers. 4 If for our demarcating criterion, however, we adopt the

2 This procedure is proposed by Gunther Miiller, "Erzahlzeit," and Roland
Barthes, "Le Discours de l'histoire," Information sur les sciences sociales, August

3 Metz (pp. 119 ff.) calls this "the large syntagmatic category" of narrative.

4 We know, besides, that only external constraint is responsible for the exist-
ing break between Swann and the jeunes Filles en fleurs. The relations between
external divisions (parts, chapters, etc.) and internal narrative articulations have
not — up until now, in general and to my knowledge — generated all the attention
they deserve. These relations, however, are what mainly determine the rhythm
of a narrative.



presence of an important temporal and/or spatial break, we can
establish the separation without too much hesitation, as follows
(I give some of these units titles — purely indicative ones — of my
own making):

(1) I, 3-142, leaving out the memory-elicited analepses studied
in the preceding chapter, is the unit devoted to the childhood in
Combray: we, like PrOust himself, will obviously name itCombray.

(2) After a temporal and spatial break, Un amour de Swann, I,

(3) After a temporal break, the unit devoted to the Parisian
adolescence and dominated by love with Gilberte and the dis-
covery of Swann's milieu, occupying the third part of Du cote de
chez Swann ("Noms de pays: le nom") and the first part of the
Jeunes Filles en fleurs ("Autour de Mme. Swann"), I, 293^187: we
will name it Gilberte.

(4) After a break that is both temporal (two years) and spatial
(the movement from Paris to Balbec), the episode of the first stay
at Balbec, corresponding to the second part of the Jeunes Filles
("Noms de pays: le pays"), I, 488-714: Balbec I.

(5) After a spatial break (return to Paris), we will take as one
and the same unit everything coming between the two visits to
Balbec, occurring almost entirely in Paris (with the exception of a
short visit to Doncieres) and in the Guermantes milieu, thus the
complete Cote de Guermantes (I, 719-1141) and the beginning of
Sodome et Gomorrhe (II, 3-109): Guermantes.

(6) The second visit to Balbec, after a new spatial break, in
other words, all the rest of Sodome et Gomorrhe, II, 110-378: we
will christen this unit Balbec II.

(7) After a new change of place (return to Paris), the story of
Albertine's confinement, flight, and death, up to II, 820, in other
words, the entire Prisonniere and most of La Fugitive, up to the
departure for Venice: Albertine.

(8) II, 821-856, the visit to Venice and the trip back: Venice.

(9) II, 856-889, straddling La Fugitive and Le Temps retrouve,
the stay at Tansonville.

(10) After a break that is both temporal (stay in a clinic) and
spatial (return to Paris), II, 890-987: The War.

(11) After a final temporal break (again a stay in a clinic),


Narrative Discourse

comes the final narrative unit, II, 988-1140, s the Matinee
Guermantes. b

With respect to chronology, the task is slightly more delicate,
since in its details the chronology of the Recherche is neither clear
nor coherent. We have no need here to join in an already old
and apparently insoluble debate, whose chief documents are
three articles by Willy Hachez and the books by Hans Robert
Jauss and Georges Daniel, which readers can refer to for a de-
tailed account of the discussion. 7 Let us recall only the two main
difficulties: on the one hand, the impossibility of connecting the
external chronology of Un amour de Swann (references to histori-
cal events requiring the episode to be dated near 1882-1884) to
the general chronology of the Recherche (putting this same epi-
sode about 1877-1878); 8 on the other hand, the disagreement
between the external chronology of the episodes Balbec II and
Albertine (references to historical events that took place between
1906 and 1913) and the general internal chronology (which puts
them back between 1900 and 1902). 9 So we cannot establish an

5 [Translator's note.] The corresponding Pleiade page numbers are: (1) I,
3-186; (2) I, 188-382; (3) I, 383-641; (4) I, 642-955; (5) II, 9-751; (6) II, 751-1131; (7)
III, 9-623; (8) III, 623-675; (9) III, 675-723; (10) III 723-854; (11) III, 854-1048.
Omitted in the English translation are P III, 673-676.

6 We see that the only two times when narrative articulations and external
divisions coincide are the two ends of visit to Balbec (the end of Jeunes Filles and
the end of Sodotne); we can add the times when articulations and subdivisions
coincide: the end of "Combray," the end of "Amour de Swann," and the end of
"Autour de Mme. Swann." All the rest is an overlapping. But of course my
carving up is not sacrosanct, and it lays claim to a value that is no more than

7 Willy Hachez, "La Chronologie etl'age des personnages deA.L.R.T.P., "Bul-
letin de la societe des amis de Marcel Proust, 6 (1956); "Retouches a une
chronologie," BSAMP, 11 (1961); "Fiches biographiques de personnages de
Proust," BSAMP, 15(1965). H. R. Jauss, Zeitund Erinnerung in A. L.R.T.P. (Heidel-
berg, 1955). Georges Daniel, Temps el mystification dans A.L.R.T.P. (Paris, 1963).

8 Added to this chronological disagreement is the one resulting from the ab-
sence in Un amour de Swann of any mention (and of any likelihood) of Gilberte's
birth, which is nonetheless required by the general chronology.

9 We know that these two contradictions result from external circumstances:
the separate writing of Un amour de Swann, integrated after the fact into the
whole, and the late projection onto the character of Albertine of facts linked to
the relations between Proust and Alfred Agostinelli. [Translator's note: Agos-
tinelli was a young man for whom Proust developed an extremely deep affection
in 1913. In 1914 he died in the crash of the plane he was learning to fly, an event
Genette refers to on p. 99.]



approximately coherent chronology except by eliminating these
two external series and adhering to the main series, whose two
fundamental guide marks are, for Guermantes, autumn 1897-
spring 1899 (because of the Dreyfus affair) and, for The War,
naturally 1916. Given these reference points, we establish an
almost homogeneous series, but it still has a few partial
obscurities. These are due, in particular, to: (a) the blurred na-
ture of the chronology of Combray and its poorly defined rela-
tionship to the chronology of Gilberte; (b) the obscurity of the
chronology of Gilberte, not allowing us to ascertain whether one
or two years pass between the two "New Years" mentioned; 10
(c) the indeterminate length of the two stays in a clinic. 11 I will
make short work of these uncertainties by establishing a purely
indicative chronology, since our purpose is only to form an
overall idea of the major rhythms of the Proustian narrative. Our
chronological hypothesis, within the limits of exactitude we have
thus settled on, is therefore as follows:

Un amour de Swann: 1877-1878

(Births of Marcel and Gilberte: 1878)

Combray: 1883-1892

Gilberte: 1892-spring 1895

Balbec I: summer 1897

Guermantes: autumn 1897-summer 1899

Balbec II: summer 1900

Albertine: autumn 1900-beginning 1902

10 RH I, 372 and 462/P I, 486 and 608.

11 The length of the first, between Tansonville and The War (RH II, 890/P III,
723), is not specified by the text ("the long years . . . which I spent far from Paris
receiving treatment in a sanatorium, until there came a time, at the beginning of
1916, when it could no longer get medical staff"), but it is fairly precisely deter-
mined by the context: the terminus ab quo is 1902 or 1903, and the terminus ad quern
is the explicit date of 1916, with the two-month trip to Paris in 1914 (RH II,
900-919/P III, 737-762) being only an interlude within that stay. The length of the
second (between The War and Matinee Guermantes, RH II, 988/P III, 854), which
can begin in 1916, is equally indefinite; but the phrase used ("many years
passed") prevents us from taking it to be very much briefer than the first, and
forces us to put the second return, and therefore the Guermantes matinee (and a
fortiori the moment of the narrating, which comes later by three years at least)
after 1922, the date of Proust's death — which is an inconvenience only if one
claims to identify the hero with the author. That wish is obviously what obliges
Hachez (1965, p. 290) to shorten the second stay to three years at the most, in
defiance of the text.

I' i'l

92 Narrative Discourse

Venice: spring 1902
Tansonville: 1903?
The War: 1914 and 1916
Matinee Guermantes: about 1925

According to this hypothesis, and some other temporal data
of secondary importance, the main variations of speed in the
narrative work out approximately like this:

Combray: 140 pages for about ten years.

Un amour de Swann: 150 pages for some two years.

Gilberts: 200 pages for about two years.

(Here, ellipsis of two years.)

Balbec I: 225 pages for three or four months.

Guermantes: 525 pages for two and one-half years. But we must
specify that this sequence itself contains very wide variations,
since 80 pages tell about the Villeparisis reception, which must
last two or three hours; 110 pages tell about the dinner at the
Duchesse de Guermantes's, lasting almost the same length of
time; and 65 pages tell about the Princess's soiree: in other
words, almost half the sequence is for fewer than ten hours of
fashionable gatherings.

Balbec II: 270 pages for nearly six months, 80 of which are for a
soiree at La Raspeliere.

Albertine: 440 pages for some eighteen months, 215 of which are
devoted to only two days, and 95 of these are for the Charlus-
Verdurin musical soiree alone.

Venice: 35 pages for some weeks.

(Indefinite ellipsis: at least some weeks.)

Tansonville: 30 pages for "some days."

(Ellipsis of about twelve years.)

The War: 100 pages for some weeks, the main part of which is for a
single evening (stroll in Paris and Jupien's male brothel).

(Ellipsis of "many years.")

Matinee Guermantes: 150 pages for two or three hours.

It seems to me, from this very sketchy list, that we can draw at
least two conclusions. First, the range of variations, going from
150 pages for three hours to three lines for twelve years, viz.
(very roughly), from a page for one minute to a page for one
century. Next, the internal evolution of the narrative in propor-
tion as it advances toward its end, an evolution that we can
summarily describe by saying that we observe on the one hand a



gradual slowing down of the narrative, through the growing
importance of very long scenes covering a very short time of
story; and on the other hand, in a sense compensating for this
slowing down, a more and more massive presence of ellipses.
We can easily synthesize these two aspects with the following
phrase: the increasing discontinuity of the narrative. The Proust-
ian narrative tends to become more and more discontinuous,
syncopated, built of enormous scenes separated by immense
gaps, and thus it tends to deviate more and more from the
hypothetical "norm" of narrative isochrony. Let us remember
that we are not by any means dealing here with an evolution
over time that would refer us to a psychological transformation
in the author, since the Recherche was not by any means written
in the order in which it is arranged today. On the other hand, it
is true that Proust, who we well know tended unceasingly to
inflate his text with additions, had more time to increase the
later volumes than the earlier ones; the bulkiness of the later
scenes thus partakes of that well-known imbalance that the pub-
lication delay imposed by the war brought about in the Re-
cherche. But circumstances, if they explain the "stuffing" with
details, cannot account for the overall composition. It certainly
seems that Proust wanted, and wanted from the beginning, this
ever more abrupt rhythm, with a Beethovenian massiveness and
brutality, which contrasts so sharply with the almost impercep-
tible fluidity of the early parts, as if to compare the temporal
texture of the older events with that of the more recent ones — as
if the narrator's memory, while the facts draw nearer, were be-
coming both more selective and more enormously enlarging.

This change in rhythm cannot be accurately defined and in-
terpreted until we connect it to other temporal treatments that
we will study in the next chapter. But from now on we can and
should examine more closely how the more or less infinite di-
versity of narrative speeds is in fact distributed and organized.
Theoretically, indeed, there exists a continuous gradation from
the infinite speed of ellipsis, where a nonexistent section of nar-
rative corresponds to some duration of story, on up to the abso-
lute slowness of descriptive pause, where some section of narra-


Narrative Discourse

five discourse corresponds to a nonexistent diegetic duration. 12
In fact, it turns out that narrative tradition, and in particular the
novel's tradition, has reduced that liberty, or at any rate has
regulated it by effecting a selection from all the possibilities: it
has selected four basic relationships that have become — in the
course of an evolution that the (as yet unborn) history of literature
will some day start to study — the canonical forms of novel
tempo, a little bit the way the classical tradition in music singled
out, from the infinitude of possible speeds of execution, some
canonical movements (andante, allegro, presto, etc.) whose rela-
tionships of succession and alternation governed structures like
those of the sonata, the symphony, or the concerto for some two
centuries. These four basic forms of narrative movement, that
we will hereafter call the four narrative movements, are the two
extremes that I have just mentioned (ellipsis and descriptive
pause) and two intermediaries: scene, most often in dialogue,
which, as we have already observed, realizes conventionally the
equality of time between narrative and story; and what English-
language critics call summary 13 — a form with variable tempo
(whereas the tempo of the other three is fixed, at least in princi-
ple), which with great flexibility of pace covers the entire range
included between scene and ellipsis. We could schematize the
temporal values of these four movements fairly well with the
following formulas, with ST designating story time and NT the
pseudo-time, or conventional time, of the narrative:

12 This formulation can occasion two misunderstandings that I wish to dissi-
pate at once. (1) The fact that a section of discourse corresponds to no duration in
the story does not in itself characterize description: it may also characterize those
commentarial excursuses in the present tense which, ever since Blin and Brom-
bert, we have generally called author's intrusions or interventions, and which we
will meet again in the last chapter. But what is distinctive about these excursuses
is that they are not strictly speaking narrative. Descriptions, on the other hand,
as constituents of the spatio-temporal universe of the story, are diegetic, and thus
when we deal with them we are involved with the narrative discourse. (2) Every
description is not necessarily a pause in the narrative, which we will observe in
Proust himself. So we are not concerned here with description, but with descrip-
tive pause, which is therefore not to be confused either with every pause or with
every description.

13 [Translator's note. ] l have omitted from the text a brief statement on French

Duration 95

pause: NT = n, ST = 0. Thus: NT oo > ST 14
scene: NT = ST
summary: NT < ST

ellipsis: NT = 0, ST = n. Thus: NT < oo ST.

A plain reading of this chart reveals an asymmetry, which is
the absence of a form with variable tempo symmetrical to the
summary and whose formula would be NT > ST. This would
obviously be a sort of scene in slow motion, and we think im-
mediately of the long Proustian scenes, the reading of which
often seems to take longer, much longer, than the diegetic time
that such scenes are supposed to be covering. But, as we shall
see, big scenes in novels, and especially in Proust, are extended
mainly by extranarrative elements or interrupted by descriptive
pauses, but are not exactly slowed down. And needless to say,
pure dialogue cannot be slowed down. So there remains de-
tailed narration of acts or events told about more slowly than
they were performed or undergone. The thing is undoubtedly
feasible as a deliberate experiment, 1S but we are not dealing there
with a canonical form, or even a form really actualized in literary
tradition. The canonical forms are indeed restricted, in fact, to
the four movements I have enumerated.


Now, if we examine from this point of view the narrative
pacing of the Recherche, what we are first compelled to note is
the almost total absence of summary in the form it had during
the whole previous history of the novel: that is, the narration in
a few paragraphs or a few pages of several days, months, or

14 This sign =��>(inhnitely greater), as well as the inverse one <=��(inhnitely
less), are not, 1 am told, mathematically orthodox. I am retaining them, how-
ever, because they seem to me, in this context and for anyone of good will, as
transparent a means as there is to designate an idea that is itself mathematically
suspect, but very clear here.

15 This is somewhat the circumstance with L'Agrandissement by Claude
Mauriac (1963), which devotes some two hundred pages to a period of two
minutes. But there again, the lengthening of the text does not arise from a real
expansion of the time period, but from various insertions (memory-elicited
analepses, etc.).


Narrative Discourse

years of existence, without details of action or speech. Borges
quotes an example of this, taken from Don Quixote, which seems
to me fairly typical:

In the end it seemed to [LotharioJ necessary to take full advantage
of the opportunity which Anselmo's absence gave him, and to
intensify the siege of the fortress. So he assailed her self-love with
praise of her beauty; for there is nothing which reduces and levels
the embattled towers of a beautiful woman's vanity so quickly as
this same vanity posted upon the tongue of flattery. In fact, he
most industriously mined the rock of her integrity with such
charges that Camilla would have fallen even if she had been made
of brass. Lothario wept, beseeched, promised, flattered and
swore, with such ardour and with such signs of real feeling, that
he overcame Camilla's chastity and achieved the triumph which
he least expected and most desired. 16

"Chapters like [this one]," comments Borges, "form the
overwhelming majority of world literature, and not the most
unworthy." He is thinking here, however, less of relations of
speed as such than of the contrast between classical abstraction
(here, despite the metaphors or perhaps because of them) and
"modern" expressivity. If one has one's eye more on the contrast
between scene and summary, 17 one obviously cannot maintain
that texts of this type "form the immense majority of world
literature," for the simple reason that the very brevity of sum-
mary gives it almost everywhere an obvious quantitative in-
feriority to descriptive and dramatic chapters, and that therefore
summary probably occupies a limited place in the whole corpus

"��Cervantes, The Adventures of Don Quixote, Part I, chap. 34, trans. J. M.
Cohen (Harmondsworth: Penguin, 1950), p. 300; quoted in J. L. Borges, Dis-
cussions (Paris, 1966), pp. 51-52. The comparison with a more flippant (but
motivated) summary on an analogous subject, in Fielding, is unavoidable: "Not
to tire the Reader, by leading him thro' every Scene of this Courtship, (which,
tho', in the Opinion of a certain great Author, it is the pleasantest Scene of Life
to the Actor, is, perhaps, as dull and tiresome as any whatever to the Audience)
the Captain made his Advances in Form, the Citadel was defended in Form, and
at length, in proper Form, surrendered at Discretion" (Henry Fielding, Tom
Jones, Book I, chap. 11 [New York: Norton Critical Editions, 19731, p. 52).

17 See Percy Lubbock, The Craft of Fiction (London, 1921).



of narrative, even of classical narrative. On the other hand, it is
obvious that summary remained, up to the end of the nine-
teenth century, the most usual transition between two scenes,
the "background" against which scenes stand out, and thus the
connective tissue par excellence of novelistic narrative, whose
fundamental rhythm is defined by the alternation of summary
and scene. We must add that most retrospective sections, and
particularly in what we have called complete analepses, belong
to this type of narration, of which the second chapter oiBirotteau
gives an example as typical as it is admirable:

A cotter, Jacques Birotteau by name, living near Chinon, took unto
himself a wife, a domestic servant in the house of a lady, who
employed him in her vineyard. Three sons were born to them; his
wife died at the birth of the third, and the poor fellow did not long
survive her. Then the mistress, out of affection for her maid,
adopted the oldest of the cotter's boys; she brought him up with
her own son, and placed him in a seminary. This Francois Birot-
teau took orders, and during the Revolution led the wandering life
of priests who would not take the oath, hiding from those who
hunted them down like wild beasts, lucky to meet with no worse
fate than the guillotine. 18

Nothing of the kind in Proust. With him, narrative cutting is
never accomplished by this sort of acceleration, even in the
anachronies, which in the Recherche are almost always genuine
scenes, earlier or later, and not offhand glances at past or future.
In Proust cutting either arises from a quite different kind of
synthesis, which we will study more closely in the next chapter

18 Gamier, p. 30; Cesar Birotteau, Beatrix, and Other Stories, trans. E. Marriage
and J. Waring (Philadelphia, 1899), p. 22. After Lubbock, the functional relation-
ship between summary and analepsis was clearly indicated by Phyllis Bentley:
"One of the most important and frequent uses of the summary is to convey
rapidly a sketch of past life. The novelist, having excited our interest in his
characters by telling a scene to us, suddenly whizzes his pageant back, then
forward, giving us a rapid summary of their past history, a retrospect" ("Use of
Summary," from Some Observations on the Art of Narrative, 1947; rpt. in Philip
Stevick, ed., The Theory of the Novel [New York, 1967], p. 49).


Narrative Discourse


under the name of iterative narrative, 19 or else it pushes acceler-
ation so far as to cross the limits separating summary from ellip-
sis pure and simple. An example is the way in which the narra-
tive sums up Marcel's years of retirement that precede and fol-
low his return to Paris during the war. 20 The confusion between
acceleration and ellipsis is, moreover, all but obvious in Proust's
famous commentary on a page of the Education sentimentale:

Here there is an implied "silence" of vast duration, 21 and sud-
denly, without the hint of a transition, 22 time ceases to be a matter
of mere successive quarters of an hour, and appears to us in the
guise of years and decades, . . . this extraordinary change of
tempo, for which nothing in the preceding lines has prepared us. 23

Now, Proust has just introduced that passage with these words:
"The finest thing, to my mind, in the whole of Education sen-
timentale, is to be found, not in words at all, but in a passage
where there comes a sudden moment of silence," and he will go
on as follows: "in Balzac, . . . the change of tempo has an active
and documentary character." So we do not know whether his

19 Which the classical novel, by no means ignorant of it, integrated into sum-
mary; example, Birotteau (Gamier, pp. 31-32; Marriage and Waring, pp. 23-24):
"He used to cry sometimes when the day was over and he thought of Touraine,
where the peasant works leisurely and the mason takes his time about laying a
stone, and toil is judiciously tempered by idleness; but he usually fell asleep
before he reached the point of thinking of running away, for his morning's
round of work awaited him, and he did his duty with the instinctive obedience
of a yard dog."

20 RH II, 890/P III, 723: "These ideas, tending on the one hand to diminish,
and on the other to increase, my regret that I had no gift for literature, were
entirely absent from my mind during the long years — in which I had in any case
completely renounced the project of writing — which I spent far from Paris re-
ceiving treatment in a sanatorium, until there came a time, at the beginning of
1916, when it could no longer get medical staff"; and RH II, 988/P III, 854: "The
new sanatorium to which I withdrew was no more successful in curing me than
the first one, and many years passed before I came away."

21 It is the change of chapter between "... and Frederic, gaping, recognized
Senechal" (III, chap. 5) and "He traveled. . . " (III, chap. 6).

22 As if the change of chapter were not, precisely, a transition. But probably
Proust, who is quoting from memory, forgot this detail.

23 Essais el articles, Pleiade, p. 595; "About Flaubert's Style," in Marcel Proust:
A Selection from His Miscellaneous Writings, trans. Gerard Hopkins (London, 1948),
pp. 234-235.



admiration here is for the sudden silence, that is, the ellipsis sep-
arating the two chapters, or for the change of tempo, that is, the
summary in the opening lines of chapter 6. No doubt the truth
is that the distinction matters little to him, so true is it that, ad-
dicted to a kind of narrative "all or nothingness," he himself can
accelerate only (according to his own expression) "wildly," 24
even at the risk of (let us dedicate this metaphor from mechanics
to the spirit of the unfortunate Agostinelli) lifting off. 25


A second negative finding concerns descriptive pauses. Proust
is customarily viewed as a novelist lavish in descriptions, and
no doubt he owes that reputation to an acquaintance with his
work that is apt to be from anthologies, where apparent di-
gressions like the hawthorns at Tansonville, the seascapes of
Elstir, the Princess's fountain, etc., are inevitably isolated. In
fact, the clear descriptive passages are, relative to the scope of
the work, neither very numerous (there are scarcely more than
about thirty) nor very long (most do not exceed four pages): the
proportion is probably lower than in some of Balzac's novels. In
addition, a large number of these descriptions (undoubtedly
more than a third) 26 are the iterative type, that is, they are not
connected to a particular moment in the story but to a series of
analogous moments, and consequently cannot in any way con-
tribute to slowing down the narrative but, indeed, the reverse:
for example, Leonie's room, the church at Combray, the "views

24 "And to make [Time's| flight perceptible novelists are obliged, by wildly
accelerating the beat of the pendulum, to transport the reader in a couple of
minutes over ten, or twenty, or even thirty years" (RH I, 369/P I, 482).

25 The Contre Sainte-Beuve contains this very allusive criticism of the Balzacian
use of summaries: "There are his recapitulations where, without allowing a
moment's breathing-space, he tells us everything we ought to know" (Pleiade,
p. 271; Marcel Proust on Art, p. 173).

26 These figures might seem vague; but it would be absurd to look for precision
apropos of a corpus whose boundaries are themselves very uncertain, since
obviously pure description (purified of any narration) and pure narration
(purified of any description) do not exist, and since the counting of "descriptive
passages" necessarily omits thousands of sentences, portions of sentences, or
descriptive words set among scenes where narrative is dominant. On this mat-
ter, see my Figures II, pp. 56-61.


Narrative Discourse

of the sea" at Balbec, the hotel in Doncieres, the scenery of
Venice, 27 so many pages all synthesizing several occurrences of
the same sight into one single descriptive section. But most im-
portant is this: even when the object described has been met
only once (like the trees at Hudimesnil), 28 or when the descrip-
tion concerns only a single one of its appearances (generally the
first, as with the church at Balbec, the Guermantes fountain, the
sea at La Raspeliere), 29 that description never brings about a
pause in the narrative, a suspension of the story or of (according
to the traditional term) the "action." In effect, Proustian narra-
tive never comes to a standstill at an object or a sight unless that
halt corresponds to a contemplative pause by the hero himself
(Swann in Un amour de Swann, Marcel everywhere else), and
thus the descriptive piece never evades the temporality of the

Of course, such treatment of description is not in itself an
innovation; and, for example, when the narrative in Astree de-
scribes at length the pictures displayed in Celadon's room at the
chateau d'Isoure, we can assume that that description more or
less accompanies Celadon's gaze as he discovers these pictures
on waking up. 30 But we know that the Balzacian novel, on the
contrary, established a typically extratemporal descriptive canon
(furthermore, more in conformity with the model of epic ec-
phrasis), 31 a canon where the narrator, forsaking the course of
the story (or, as in he Pere Goriot or La Recherche de I'absolu,
before arriving there), makes it his business, in his own name
and solely for the information of his reader, to describe a scene
that at this point in the story no one, strictly speaking, is looking
at. For example, as the sentence in the Vieille Fille that opens the
scene at the Cormon townhouse certainly indicates: "Now,
however, it will be necessary to enter the household of that
elderly spinster toward whom so many interests converge, and

27 RH I, 37-38/P I, 49-50; RH I, 45-51/P I, 59-67; RH I, 510-51 1/P I, 672-673; RH
I, 605-608/P I, 802-806; RH I, 784-785/P II, 98-99; RH II, 821-823/P III, 623-625.

28 RH I, 543-545/P 1, 717-719.

29 RH I, 500-502/P I, 658-660; RH II, 43/P II, 656-657; RH II, 212/P II, 897.
'"Honore d'Urfe, Astree, Vaganay ed., I. 40^13.

31 Except for the shield of Achilles (Iliad, Book XVIII), described, as we know,
at the time of its construction by Hephaistus.



within whose walls the actors in this Scene are to meet this very
evening." 32 This "entering" is obviously the doing of the nar-
rator and reader alone, who are going to wander over the house
and the garden while the real "actors in this Scene" continue to
attend to their business elsewhere, or rather wait to go back to
their business until the narrative agrees to return to them and
restore them to life. 33

We know that Stendhal always avoided that canon by pul-
verizing the descriptions, and by almost systematically integrat-
ing what he allowed to remain of them to the level of his charac-
ters' actions — or daydreams. But Stendhal's position, here as
elsewhere, remains marginal and has no direct influence. If we
wish to find in the modern novel a model or a precursor of
Proustian description, we should much rather think of Flaubert.
Not that the Balzacian type is completely foreign to him: see the
scene of Yonville that begins the second part of Bovary. But most
of the time, and even in descriptive passages of a certain extent,
the general movement of the text 34 is governed by the step or
the gaze of one (or several) character(s), and the unfolding of
that movement corresponds exactly to the length of the trip
(Emma's inspection of the house at Tostes, Frederic's and
Rosanette's walk in the forest) 35 or of the motionless contempla-
tion (sight of the garden at Tostes, gallery with colored panes of
glass at la Vaubyessard, view of Rouen). 36

32 Gamier, p. 67; The Old Maid, trans. W. Walton (Washington, D.C., 1898), p.

33 Gautier will use this technique to the point of a flippancy that "bares" it, as
the Formalists would say: "The Marquise inhabited a separate suite, which the
Marquis did not enter unless he was announced. We will commit this impropri-
ety that authors of all times have allowed themselves, and without saying a
word to the buttons who would have forewarned the lodger, we will penetrate
into the bedroom, sure of disturbing no one. The writer of a novel naturally wears
on his finger the ring of Gyges, which makes him invisible" (he Capitaine Fra-
casse, Gamier ed., p. 103). Later we will again meet this trope, the metalepsis,
with which the narrator pretends to enter (with or without his reader) into the
diegetic universe.

34 Setting aside certain descriptive intrusions of the narrator, generally in the
present tense, very brief, and as if unintentional: see my Figures, pp. 223-243.

35 Bovary, Gamier ed. (edited by Gothot-Mersch), pp. 32-34; U Education,
edited by Dumesnil, II, 154-160.

36 Bovary, Pommier-Leleu version, pp. 196-197 and 216; Gamier, pp. 268-269.
The latter is iterative as well.


Narrative Discourse

Proustian narrative seems to have turned this principle of
concurrence into a rule. We know what characteristic habit of
the author himself is reflected in the hero's capacity to come to a
stop for long minutes before an object (hawthorns at Tanson-
ville, pond at Montjouvain, trees at Hudimesnil, apple trees in
bloom, views of the sea, etc.) — an object whose power to fascinate
derives from the presence of a secret not disclosed, a message
still illegible but insistent, a rough sketch and veiled prom-
ise of the ultimate revelation. The duration of these contempla-
tive halts is generally such that it is in no danger of being ex-
ceeded by the duration of the reading (even a very slow reading)
of the text that "tells of" them. So it is, for example, with the
gallery of the Elstir paintings at the Due de Guermantes's, the
evocation of which takes up less than four pages 37 and which
itself — Marcel notices after the event — has delayed him for three
quarters of an hour, during which time the famished Duke leads
some respectful guests, including the Princesse de Parme, in
being patient. In fact, Proustian "description" is less a descrip-
tion of the object contemplated than it is a narrative and analysis
of the perceptual activity of the character contemplating: of his
impressions, progressive discoveries, shifts in distance and
perspective, errors and corrections, enthusiasms or disappoint-
ments, etc. A contemplation highly active in truth, and contain-
ing "a whole story." This story is what Proustian description
recounts. Suppose we reread, for example, the few pages de-
voted to Elstir's seascapes at Balbec. 38 We will see how jammed
they are with terms designating not what the painting of Elstir
is, but the "optical illusions" that it "recreates," and the false
impressions it arouses and dissipates in turn: seem, appear, give
the impression, as if, you felt, you would have said, you thought, you
understood, you saw reappear, they went racing over sunlit fields, etc.
Aesthetic activity here is not repose at all, but this characteristic
is not due only to the sleight-of-hand "metaphors" of the im-
pressionist painter. The same labor of perception, the same
struggle or play with appearances, occurs again in the presence

37 RH I, 1017-1020/P II, 419^22.

38 RH I, 629-632/P I, 836-840.



of the slightest object or landscape. Here is the (very) young
Marcel grappling with Aunt Leonie's handful of dried lime-
flowers: "as though a painter," "the leaves. . . assumed [the ap-
pearance] ... of the most incongruous things imaginable," "A
thousand trifling little details . . . gave me . . . the pleasure of
finding that these were indeed real lime-blossoms," "\ recognized,"
"the rosy. . . glow shewed me that these were petals which,"
etc.: 39 a whole precocious education in the art of seeing, of going
beyond false appearances, of discerning true identities, giving
this description (which, furthermore, is iterative) a story dura-
tion that is packed full. There is the same labor of perception in
front of Hubert Robert's fountain, the description of which I
reprint in its entirety, merely emphasizing the terms that mark
the duration of the scene and the activity of the hero, who is
hidden here by a falsely generalizing impersonal pronoun (a
little like Brichot's "one") that multiplies his presence without
abolishing it:

In a clearing surrounded by fine trees several of which were as old
as itself, set in a place apart, one could see it in the distance, slender,
immobile, stiffened, allowing the breeze to stir only the lighter fall
of its pale and quivering plume. The eighteenth century had re-
fined the elegance of its lines, but, by fixing the style of the jet,
seemed to have arrested its life; at this distance one had the impression
of a work of art rather than the sensation of water. The moist cloud
itself that was perpetually gathering at its crest preserved the
character of the period like those that in the sky assemble round
the palaces of Versailles. But from a closer view one realised that,
while it respected, like the stones of an ancient palace, the design
traced for it beforehand, it was a constantly changing stream of water
that, springing upwards and seeking to obey the architect's tra-
ditional orders, performed them to the letter only by seeming to
infringe them, its thousand separate bursts succeeding only at a
distance in giving the impression of a single flow. This was in reality
as often interrupted as the scattering of the fall, whereas from a
distance it had appeared to me unyielding, solid, unbroken in its
continuity. From a little nearer, one saw that this continuity, appar-
ently complete, was assured, at every point in the ascent of the jet,

'RH I, 39/PI, 51.


Narrative Discourse

wherever it must otherwise have been broken, by the entering into
line, by the lateral incorporation of a parallel jet which mounted
higher than the first and was itself, at an altitude greater but al-
ready a strain upon its endurance, relieved by a third. Seen close at
hand, drops without strength fell back from the column of water
crossing on their way their climbing sisters and, at times, torn,
caught in an eddy of the night air, disturbed by this ceaseless flow,
floated awhile before being drowned in the basin. They teased with
their hesitations, with their passage in the opposite direction, and
blurred with their soft vapour the vertical tension of that stem,
bearing aloft an oblong cloud composed of a thousand tiny drops,
but apparently painted in an unchanging, golden brown which
rose, unbreakable, constant, urgent, swift, to mingle with the
clouds in the sky. Unfortunately, a gust of wind was enough to
scatter it obliquely on the ground; at times indeed a single jet,
disobeying its orders, swerved and, had they not kept a respectful
distance, would have drenched to their skins the incautious crowd
of gazers. 40

We meet this situation again, developed much more exten-
sively, in the course of the Guermantes matinee. Its first
twenty-five pages at least 41 are based on this activity of recog-
nizing and identifying, an activity forced on the hero by the
aging of an entire "society." At first glance these twenty -five
pages are purely descriptive: the sight of the Guermantes salon
after a ten-year absence. In fact, we are definitely dealing in-
stead with a narrative: how the hero, passing from one to
another (or from some to others), must each time make the
e ff or t — sometimes a fruitless one — to recognize, in this little old
man, the Due de Chatellerault; under his beard, M. d'Argen-
court; the Prince d'Agrigente, dignified by age; the young count

of , as an old colonel; Bloch, as poppa Bloch, etc. —

revealing at each encounter "the mental effort that made [him]
hesitate between three or four people," and that other "mental
effort," the even more disturbing one of identification itself:

40 RH II, 43/PII, 656.

41 We are dealing here with the first twenty-five pages of the reception as such
(RH II 1039-1064/P III, 920-952), once Marcel has entered the salon, after the
meditation in the library (RH II, 997-1039/P III, 866-920).



For to "recognize" someone, and, a fortiori, having failed to recog-
nize someone to learn his identity, is to predicate two contradic-
tory things of a single subject, it is to admit that what was here, the
person whom one remembers, no longer exists, and also that what
is now here is a person whom one did not know to exist; and to do
this we have to apprehend a mystery almost as disturbing as that
of death, of which it is, indeed, as it were the preface and the
harbinger. 42

A painful substitution, like the one he must effect at the church
of Balbec, of the real for the imaginary: "my mind . . . was as-
tonished to see the statue which it had carved a thousand times,
reduced now to its own apparent form in stone," a work of art
"transformed, as was the church itself, into a little old woman in
stone whose height I could measure and count her wrinkles." 43
A euphoric superimposition, by contrast: the one setting up a
comparison between the memory of Combray and the scenery
of Venice, "impressions analogous . . . but transposed into a
wholly different and far richer key." 44 Finally a difficult, almost
acrobatic juxtaposition: the pieces of the "countryside at sun-
rise" perceived alternately through the two opposite window
panes of the railroad car between Paris and Balbec, and requir-
ing the hero to be "running from one window to the other to
reassemble, to collect on a single canvas the intermittent, an-
tipodean fragments of [his] fine, scarlet, ever-changing morn-
ing, and to obtain a comprehensive view of it and a continuous
picture." 45

So we see that in Proust contemplation is neither an instan-
taneous flash (like recollection) nor a moment of passive and
restful ecstasy; it is an activity — intense, intellectual, and often
physical — and the telling of it is, after all is said and done, a
narrative just like any other. What we are compelled to con-
clude, therefore, is that description, in Proust, becomes ab-
sorbed into narration, and that the second canonical type of

42 RH II, 1054/P III, 939.

43 RH I, 501-502/P I, 659-660.

44 RH II, 821/P III, 623.

45 RH I, 497/P I, 654-655.


Narrative Discourse

movement — the descriptive pause — does not exist in Proust, for
the obvious reason that with him description is everything ex-
cept a pause in the narrative.


Absence of summary, absence of descriptive pause — on the
roster of Proustian narrative, then, only two of the traditional
movements still exist: scene and ellipsis. Before examining the
temporal pacing and the function of scene in Proust, we will say
a few words about ellipsis. Obviously we are dealing here only
with ellipsis as such, or temporal ellipsis, leaving aside those
lateral omissions for which we have reserved the name paralipsis.

From the temporal point of view, the analysis of ellipses
comes down to considering the story time elided, and here the
first question is to know whether that duration is indicated (defi-
nite ellipses) or not indicated (indefinite ellipses). Thus, between
the end of Gilberte and the beginning of Balbec a two-year ellipsis
occurs that is clearly definite: "I had arrived at a state almost of
complete indifference to Gilberte when, two years later, I went
with my grandmother to Balbec"; 46 on the other hand, we re-
member, the two ellipses relating to the hero's sojourns in a
clinic are (almost) equally indefinite ("long years," "many
years"), and the analyst is reduced to sometimes difficult

From the formal point of view, we will distinguish:

(a) Explicit ellipses, like those I have just quoted. They arise
either from an indication (definite or not) of the lapse of time they
elide, which assimilates them to very quick summaries of the
"some years passed" type (in this case the indication constitutes
the ellipsis as textual section, which is then not totally equal to
zero); or else from elision pure and simple (zero degree of the
elliptical text) plus, when the narrative starts up again, an indi-
cation of the time elapsed, like the "two years later" quoted just
above. This latter form is obviously more rigorously elliptical,
although quite as explicit, and not necessarily shorter; but in this
form the text expresses the perception of narrative void or gap

46 RH I, 488/P I, 642.



more analogically, more "iconically" (in Peirce's or Jakobson's
sense). 47 Both of these forms, in addition, can supplement the
purely temporal indication with a piece of information having
diegetic content, such as "some years of happiness passed," or
"after some years of happiness." These characterizing ellipses are
one of the resources of novelistic narration. In the Chartreuse
Stendhal gives an example that is memorable, and moreover
ingenuously contradictory, after the nocturnal reunion of Fab-
rice and Clelia: "Here, we ask for permission to pass over, with-
out saying a single word about it, a space of three years. . . . After
three years of divine happiness . . . " 48 Let us add that a negative
characterization is a characterization just like any other: an
example is when Fielding, who with some exaggeration flatters
himself on being the first to vary the rhythm of the narrative and
to elide the dead spaces of the action, 49 leaps over twelve years
in the life of Tom Jones, asserting that "nothing worthy of a
Place in this History occurred within that Period." 50 We know
how much Stendhal admired and imitated this flippant manner.
In the Recherche, the two ellipses that frame the episode of the
war are obviously characterizing ellipses, since we learn that
Marcel spent those years in a clinic, being cared for without
being cured, and without writing. But almost equally charac-
terizing, although retrospectively, is the ellipsis opening Balbec
I, for to say "I had arrived at a state almost of complete indif-
ference to Gilberte when, two years later ..." amounts to say-

47 See Roman Jakobson, "Quest for the Essence of Language," Diogenes, 51
(Fall 1965), 21-37.

48 Gamier, p. 474.

49 See Book II, chap. 1, of Tom Jones, where he attacks the dull historians who
"fill up as much Paper with the Detail of Months and Years in which nothing
remarkable happened, as [they employ] upon those notable /Eras when the
greatest Scenes have been transacted on the human Stage," and whose books he
compares "to a Stage-Coach, which performs constantly the same Course,
empty as well as full." In opposition to this somewhat imaginary tradition, he
boasts of inaugurating "a contrary Method," sparing nothing to "open (any
extraordinary Scene] at large to our Reader," while on the contrary ignoring
"whole Years [that] pass without producing any Thing worthy |of] Notice" — like
the "Isagacious] Registers of [the Guild-hall] Lottery" who announce only the
winning numbers (Norton Critical Edition, pp. 58-59).

50 Book III, chap. 1 (Norton, p. 88).


Narrative Discourse

ing, "for two years, I was detaching myself from Gilberte little
by little."

(b) Implicit ellipses, that is, those whose very presence is
not announced in the text and which the reader can infer only
from some chronological lacuna or gap in narrative continuity.
This is the case for the indefinite time elapsing between the
end of the jeunes Filles en fleurs and the beginning of Guer-
mantes: we know Marcel had returned to Paris, to "[his] own
room, the ceiling of which was low"; 51 we meet him next in a
new apartment attached to the Guermantes townhouse, which
presumes the elision of at least a few days, and perhaps consid-
erably more. It is also the case, and in a more puzzling way, for
the few months following the grandmother's death. 52 This ellip-
sis is perfectly mute: we left the grandmother on her deathbed,
most likely at the beginning of the summer; the narrative takes
up again in these terms: "Albeit it was simply a Sunday in
autumn ..." The ellipsis is apparently definite, thanks to this
indication of date, but it is very imprecisely so, and will soon
become rather confused. 53 Above all it is not characterized, and
it will remain not characterized: we will never, even retro-
spectively, know anything of what the hero's life has been dur-
ing these few months. This is perhaps the most opaque silence
in the entire Recherche, and, if we remember that the death of
the grandmother is to a great extent a transposition of the death
of the author's mother, this reticence is undoubtedly not devoid
of significance. 54

s ' RH I, 712/P I, 953.

"Between chapters 1 and 2 of Guermantes 11 (RH I, 964-%5/P II, 345).

53 "First it is an indefinite Sunday in autumn [RH I, 965/P II, 345] and soon it is
the end of autumn [RH I, 994/P II, 3851. However, shortly thereafter |RH I, 999/P
II, 392] Francoise says, 'It's the end of "Sectember" already. . . .' In any case, it is
not a September atmosphere, but a November or even a December one that the
restaurant is deep in where the narrator dines the day before the first invitation
to the Duchesse de Guermantes's. And on leaving her reception, the narrator
asks for his snowboots" (Daniel, Temps et mystification, pp. 92-93).

54 Let us remember that Marcel himself has the habit of interpreting certain
words "in the same way as . . . a sudden silence" (RH II, 439/P III, 88). The
hermeneutics of narrative must also take on these sudden silences, by account-
ing for their "duration," their intensity, and naturally their placement.



(c) Finally, the most implicit form of ellipsis is the purely
hypothetical ellipsis, impossible to localize, even sometimes
impossible to place in any spot at all, and revealed after the
event by an analepsis such as those we already met in the pre-
ceding chapter: 55 trips to Germany, to the Alps, to Holland,
military service. We are obviously there at the limits of the
narrative's coherence, and for that very reason at the limits of
the validity of temporal analysis. But the designation of limits is
not the most trifling task of a method of analysis; and we may
say in passing that perhaps the main justification for studying a
work like the Recherche du temps perdu according to the tra-
ditional criteria of narrative is, on the contrary, to allow one to
establish with precision the points on which such a work, delib-
erately or not, goes beyond such criteria.


If we consider the fact that ellipses, whatever their number
and power of elision may be, represent a practically nonexistent
portion of text, we must surely come to the conclusion that the
whole of Proust's narrative text can be defined as scene, taking
that term in the temporal sense in which we are defining it here
and setting aside for the moment the iterative nature of some of
those scenes. 56 Thus the traditional alternation summary/scene
is at an end. Later we will see it replaced by another alternation,
but now we must note a change in function which in any case
modifies the structural role of the scene.

In novelistic narrative as it functioned before the Recherche,
the contrast of tempo between detailed scene and summary al-
most always reflected a contrast of content between dramatic
and nondramatic, the strong periods of the action coinciding
with the most intense moments of the narrative while the weak
periods were summed up with large strokes and as if from a
great distance, according to the principle that we have seen set
forth by Fielding. The real rhythm of the novelistic canon, still

'P. 51.

'On the dominance of scene, see Tadie, Proust et le roman, pp. 387 ff.


Narrative Discourse

very perceptible in Bovary, is thus the alternation of nondramat-
ic summaries, functioning as waiting room and liaison, with
dramatic scenes whose role in the action is decisive. 57

One can still grant that status to some of the scenes in the
Recherche, like the "drama of bedtime," the profanation at
Montjouvain, the evening of the cattleyas, Charlus's deep anger
at Marcel, the grandmother's death, Charlus's exclusion, and
naturally (although there we are dealing with a completely
internal "action") the ultimate revelation, 58 all of which mark
irreversible stages in the fulfillment of a destiny. But clearly such
is not the function of the longest and most typically Proustian
scenes, those five enormous ones that all by themselves take up
about 450 pages: the Villeparisis matinee, the Guermantes
dinner, the soiree at the Princess's, the soiree at La Raspeliere,
the Guermantes matinee. 59 As we have already observed, each
of these has inaugural importance: each marks the hero's en-
trance into a new place or milieu and stands for the entire series,
which it opens, of similar scenes that will not be reported: other
receptions at Mme. de Villeparisis's and in the Guermantes
milieu, other dinners at Oriane's, other receptions at the Prin-
cess's, other soirees at La Raspeliere. None of these inaugural
social gatherings merits more attention than all the analogous
ones that succeed it and that it represents except by being the
first in each series, and as such arousing a curiosity that habit
will immediately after begin to blunt. 60 So we are not dealing

57 This assertion should obviously be taken with qualifications: for instance, in
the Souffrances de I'inventeur, the most dramatic pages are perhaps those where
Balzac sums up with the spareness of a military historian the procedural battles
waged against David Sechard.

58 RH I, 16-36/P I, 21^8; RH I, 122-127/P I, 159-165; RH I, 173-179/P I, 226-233;
RH I, 1110-1119/P II, 552-565; RH I, 956-964/P II, 335-345; RH II, 537-606/P III,
226-324; RH II, 996-999/P III, 865-869.

59 RH I, 846-920/P II, 183-284; RH I, 1016-1106/P II, 416-547; RH II, 27-89/P II,
633-722; RH II, 190-269/P II, 866-979; RH II, 997-1140/P III, 866-1048.

60 The status of the final scene (the Guermantes matinee) is more complex
because it involves as much (and even more) a farewell to the world as an
initiation. But the theme of discovery is nonetheless present there, in the form, as
we know, of a rediscovery, a recognition made difficult by the mask of aging and
transformation — a reason for curiosity as powerful as, if not more so than, the
reason animating the earlier scenes of entry into society.



here with dramatic scenes, but rather with typical or illustrative
scenes, where action (even in the very broad sense one must
give this term in the Proustian universe) is almost completely
obliterated in favor of psychological and social characterization. 61
This change of function entails a very appreciable modifica-
tion in temporal texture: contrary to the earlier tradition, which
made scene into a place of dramatic concentration almost en-
tirely free of descriptive or discursive impedimenta, and free
even more of anachronic interferences, the Proustian scene — as
]. P. Houston has said 62 — plays in the novel a role of "temporal
hearth" or magnetic pole for all sorts of supplementary informa-
tion and incidents. It is almost always inflated, indeed encum-
bered with digressions of all kinds, retrospections, anticipa-
tions, iterative and descriptive parentheses, didactic interven-
tions by the narrator, etc., all intended to collect in a syllepsis
around the gathering-as-pretext a cluster of events and consid-
erations able to give that gathering a fully paradigmatic impor-
tance. A very approximate breakdown bearing on the large
scenes in question reveals fairly well the relative weight of these
elements that are external to the gathering being told about but
thematically essential to what Proust called his "supernourish-
ment": in the Villeparisis matinee, twenty-five pages out of sev-
enty-five; in the Guermantes dinner, forty-three out of ninety;
in the Guermantes soiree, seventeen out of sixty-two; in the last
Guermantes matinee, finally — the first forty-two pages of which
are taken up with an almost indistinguishable mixture of inter-
nal monologue by the hero and speculative discourse by the
narrator, and the remainder of which is handled (as we will see
later) chiefly in an iterative mode — the proportion is reversed
and it is the strictly narrative moments (barely forty pages out of
one hundred and forty) that seem to emerge from a sort of

61 B. G. Rogers (Proust's Narrative Techniques [Geneva, 1965], pp. 143 ff.) sees
in the unfolding of the Recherche a gradual disappearance of dramatic scenes,
which, according to him, are more numerous in the early parts. His main argu-
ment is that Albertine's death is not cause for a scene. Not a very convincing
proof; the proportion hardly varies in the course of the work, and the relevant
feature is much rather the steady predominance of nondramatic scenes.

62 Houston, pp. 33-34.


Narrative Discourse

descriptive-discursive magma very remote from the usual criteria
of "scenic" temporality and even from all narrative temporality —
like those melodic scraps that one perceives in the opening
measures of "La Valse," through a mist of rhythm and harmony.
But here the haziness is not inceptive, like Ravel's or like that
of the opening pages of Swann, but the contrary: as if in this
final scene the narrative wanted, at the end, to dissolve gradually
and to enact the intentionally indistinct and subtly chaotic
reflection of its own disappearing.

Thus we see that Proustian narrative does not leave any of the
traditional narrative movements intact, and that the whole of
the rhythmic system of novelistic narrative is thereby pro-
foundly affected. But we still have one last modification left to
understand, undoubtedly the most decisive one: its emergence
and diffusion will give the narrative temporality of the Recherche
a completely new cadence — a perfectly unprecedented rhythm.

3 Frequency


What I call narrative frequency, that is, the relations of fre-
quency (or, more simply, of repetition) between the narrative
and the diegesis, up to this time has been very little studied by
critics and theoreticians of the novel. It is nonetheless one of the
main aspects of narrative temporality, and one which, at the
level of common speech, is well known to grammarians under
the category precisely of aspect.

An event is not only capable of happening; it can also happen
again, or be repeated: the sun rises every day. Of course, strictly
speaking the identity of these multiple occurrences is debatable:
"the sun" that "rises" every morning is not exactly the same
from one day to another — any more than the "8:25 p.m.
Geneva-to-Paris" train, dear to Ferdinand de Saussure, is made
up each evening of the same cars hooked to the same locomo-
tive. 1 The "repetition" is in fact a mental construction, which
eliminates from each occurrence everything belonging to it that
is peculiar to itself, in order to preserve only what it shares with
all the others of the same class, which is an abstraction: "the
sun," "the morning," "to rise." This is well known, and I recall
it only to specify once and for all that what we will name here
"identical events" or "recurrence of the same event" is a series
of several similar events considered only in terms of their resemblance.

1 Ferdinand de Saussure, Course in General Linguistics, trans. Wade Baskin
(New York: McGraw-Hill, 1959), p. 108.



Narrative Discourse

Symmetrically, a narrative statement is not only produced, it
can be produced again, can be repeated one or more times in the
same text: nothing prevents me from saying or writing, "Pierre
came yesterday evening, Pierre came yesterday evening, Pierre
came yesterday evening." Here again, the identity and therefore
the repetition are facts of abstraction; materially (phonetically or
graphically) or even ideally (linguistically) none of the occur-
rences is completely identical to the others, solely by virtue of
their co-presence and their succession, which diversify these
three statements into a first, a next, and a last. Here again one
can refer to the famous pages of the Cours de linguistique generate
on the "problem of identities." That is a further abstraction to
take into consideration, and we will do so.

A system of relationships is established between these
capacities for "repetition" on the part of both the narrated
events (of the story) and the narrative statements (of the text) — a
system of relationships that we can a priori reduce to four virtual
types, simply from the multiplication of the two possibilities
given on both sides: the event repeated or not, the statement
repeated or not. Schematically, we can say that a narrative,
whatever it is, may tell once what happened once, n times what
happened n times, n times what happened once, once what
happened n times. Let us linger a bit with these four types of
relations of frequency.

Narrating once what happened once (or, if we want to abbreviate
with a pseudo-mathematical formula: IN/IS). For example, a
statement such as "Yesterday, I went to bed early." This form of
narrative, where the singularness of the narrative statement cor-
responds to the singularness of the narrated event, is obviously
far and away the most common — so common, and apparently
considered so "normal," that it bears no name, at least in our
language. However, to express specifically that we are dealing
with only one possibility among others, I propose to give it a
name. I will hereafter call it singulative narrative — a neologism
that I hope is transparent, and that we will sometimes lighten by
using the adjective "singular" in the same technical sense: a
singulative or singular scene.

Narrating n times what happened n times (nNlnS). For example,



the statement, "Monday, I went to bed early, Tuesday I went to
bed early, Wednesday I went to bed early, etc." From the point
of view we are interested in here, that is, relations of frequency
between narrative and story, this anaphoric type is still in fact
singulative and thus reduces to the previous type, since the
repetitions of the narrative simply correspond — according to a
connection that Jakobson would call iconic — to the repetitions of
the story. The singulative is therefore defined not by the
number of occurrences on both sides but by the equality of this
number. 2

Narrating n times what happened once (nNHS). For example, a
statement like this one: "Yesterday I went to bed early, yester-
day I went to bed early, yesterday I went to bed early, etc." 3
This form might seem purely hypothetical, an ill-formed off-
spring of the combinative mind, irrelevant to literature. Let us
remember, however, that certain modern texts are based on
narrative's capacity for repetition: we may remember, for in-
stance, a recurrent episode like the death of the centipede in La
Jalousie. On the other hand, the same event can be told several
times not only with stylistic variations, as is generally the case in
Robbe-Grillet, but also with variations in "point of view," as in
Rashomon or The Sound and the Fury. 4 The epistolary novel of the
eighteenth century was already familiar with contrasts of this
type, and of course the "repeating" anachronies that we met in
Chapter 1 (advance notices and recalls) belong to this narrative
type, which they bring into existence more or less fleetingly. Let
us also remember (and this is not as foreign to the function of
literature as one might believe) that children love to be told the
same story several times — indeed, several times in a row — or to
reread the same book, and that this predilection is not entirely
the prerogative of childhood: later we will examine in some

2 That is, the formula nN/nS defines equally the first two types, granting that
most often n = 1. To tell the truth, this grid does not take into account a fifth
possible relationship (but one that to my knowledge we have no example of),
where what happened several times would also be recounted several times, but
a different (either greater or lesser) number of times: nNlmS.

3 With or without stylistic variations, such as, "Yesterday I went to bed early,
yesterday I went to bed before it was late, yesterday I put myself to bed early. ..."

4 We will come back to this question in the next chapter.


Narrative Discourse

detail the scene of the "Saturday luncheon at Combray," which
ends on a typical example of ritual narrative. This type of narra-
tive, where the recurrences of the statement do not correspond
to any recurrence of events, I will obviously call repeating narrative.

Finally, narrating one time (or rather: at one time) what happened
n times (lNlnS). Let us go back to our second — singulative
anaphoric — type: "Monday I went to bed early, Tuesday, etc."
Plainly, when such repeating phenomena occur in the story, the
narrative is not by any means condemned to reproduce them in
its discourse as if it were incapable of the slightest effort to
abstract and synthesize: in fact, and except for deliberate stylistic
effect, a narrative — and even the most unpolished one — will in
this case find a sylleptic 5 formulation such as "every day," or
"the whole week," or "every day of the week I went to bed
early." It is well known what variant of this phrase opens the
Recherche du temps perdu. This type of narrative, where a single
narrative utterance takes upon itself several occurrences together 6
of the same event (in other words, once again, several events
considered only in terms of their analogy), we will call iterative
narrative. We are dealing here with a linguistic proceeding that
in its different forms 7 is completely common and probably uni-
versal or quasi-universal — one well known to grammarians,
who have conferred its name upon it. 8 Literature's investment
in it, on the other hand, does not seem to have provoked very
intense interest so far. 9 It is, however, a completely traditional
form: we can find examples of it as early as the Homeric epic,
and throughout the history of the classical and modern novel.

But in the classical narrative and even up to Balzac, iterative
sections are almost always functionally subordinate to singula-

5 In the sense in which we defined narrative syllepsis earlier (p. 85).

6 It is indeed a question of taking on together, synthetically, and not of recount-
ing a single one of them which would stand for all the others, which is a
paradigmatic use of singulative narrative: "I report the conversation at one of
these meals, which may give an idea of the others" (RH II, 289/P II, 1006).

7 For example, the "iterative" or "frequentative" form of the English verb, or
the French imperfect tense for repeated action.

8 In concurrence, then, with "frequentative."

9 Let us mention, however, J. P. Houston's article, already referred to, and
Wolfgang Raible's "Linguistik und Literaturkritik," Linguistik unci Didaktik, 8



tive scenes, for which the iterative sections provide a sort of
informative frame or background, in a mode illustrated fairly
well, for example, in Eugenie Grandet, by the preliminary scene
of daily life in the Grandet family, a scene which serves only to
prepare for the opening of the narrative as such: "In 1819, to-
wards the beginning of the evening, in the middle of November,
big Nanon lit the fire for the first time." 10 The classic function of
iterative narrative is thus fairly close to that of description, with
which, moreover, it maintains very close relations: the "moral
portrait," for example, which is one of the varieties of the de-
scriptive genre, operates most often (see La Bruyere) through
accumulation of iterative traits. Like description, in the tra-
ditional novel the iterative narrative is at the service of the narra-
tive "as such," which is the singulative narrative. The first
novelist who undertook to liberate the iterative from this
functional dependence is clearly Flaubert in Madame Bovary,
where pages like those narrating Emma's life in the convent, her
life at Tostes before and after the ball at La Vaubyessard, or her
Thursdays at Rouen with Leon 11 take on a wholly unusual full-
ness and autonomy. But no novelistic work, apparently, has
ever put the iterative to a use comparable— in textual scope, in
thematic importance, in degree of technical elaboration— to
Proust's use of it in the Recherche du temps perdu.

The first three main sections of the Recherche— that is, Com-
bray, Un amour de Swann, and "Gilberte" (Noms de pays: ' \e nom
and Autour de Madame Swann)— can without exaggeration be
considered essentially iterative. Other than some singulative
scenes (which are, for that matter, dramatically very important,
like Swann's visit, the meeting with the Lady in pink, the Le-
grandin episodes, the profanation at Montjouvain, the appear-
ance of the Duchess at church, and the trip to the steeples at
Martinville), the text of Combray narrates, in the French imper-
fect tense for repeated action, not what happened but what used
to happen at Combray, regularly, ritually, every day, or every

10 Gamier, p. 34.

11 1, chap. 6; I, chap. 7; I, chap. 9; III, chap. 5.


Narrative Discourse

Sunday, or every Saturday, etc. The narrative of Swann's and
Odette's love is also carried on, for the most part, in this mode of
custom and repetition (major exceptions: the two Verdurin
soirees, the scene of the cattleyas, the Sainte-Euverte concert),
just like the love between Marcel and Gilberte (notable singula-
tive scenes: Berma, the dinner with Bergotte). An approximate
count (precision here would not be pertinent) reveals something
like 86 iterative to 52 singulative pages in Combray, 68 to 77 in Un
amour de Swarm, 109 to 85 in Gilberte, or about 265 iterative to 215
singulative pages for the whole of these three sections. Only
with the first visit to Balbec is the predominance of the singula-
tive established (or reestablished, if we think of what the propor-
tion was in the traditional narrative). 12 Yet we note, up to the
end, numerous iterative sections, like the rides at Balbec with
Mme. de Villeparisis in the ]eunes Filles en fleurs; the hero's ma-
neuvers, at the beginning of Guermantes, to meet the Duchess
every morning; the sights of Doncieres; the trips in the little train
of La Raspeliere; life with Albertine in Paris; the outings in Ven-
ice. 13

And we must note the presence of iterative passages within
singulative scenes: for example, at the beginning of the dinner at
the Duchess's, the long parenthesis devoted to the wit of the
Guermantes. 14 In this case, the temporal field covered by the
iterative section obviously extends well beyond the temporal
field of the scene it is inserted into: the iterative to some extent
opens a window onto the external period. So we will describe
parentheses of this type as generalizing iterations, or external itera-
tions. Another, much less classical type of move to the iterative
within a singular scene is partly to treat the duration of the scene
itself in an iterative form, whereupon the scene is then synthe-
sized by a sort of paradigmatic classification of the events com-

12 We would have to have a gigantic set of statistics to establish this proportion
accurately; but probably the iterative's share would not reach anything near ten
per cent.

oRHI 534-548/P I, 704-723; RH I, 755-756/P II, 58-59; RH I, 782-785/P II,
96-100; RH II, 308-364/P II, 1034-1112; RH II, 383-134/P III, 9-81; RH II, 820-S25/P
III, 623-630.

,4 RH I, 1031-1063/P II, 438-183.



posing it. A very clear example of such a treatment, even though
it extends over a necessarily very short period of time, is this
passage about the meeting between Charlus and Jupien, in
which we see the Baron raise his eyes "every now and then"
and dart an attentive look at the tailor: "each time that M. de
Charlus looked at Jupien, he took care that his glance should be
accompanied by a spoken word. . . . Thus, every other minute, the
same question seemed to be being intensely put to Jupien." The
iterative nature of the action is confirmed here by the indication
of frequency, with a wholly hyperbolic precision. 15 We find the
same effect again, on a much vaster scale, in the final scene of
the Temps retrouve, which is treated almost continuously in the
iterative mode. What governs the composition of the text here is
not the diachronic unfolding of the reception at the Princess's, in
the succession of events filling it up, but rather the enumeration
of a certain number of classes of occurrences, each of which
synthesizes several events that are in fact scattered throughout
the "matinee."

In some of the guests I recognized after a while . . . And yet, in
complete contrast with these, I had the surprise of talking to men
and women who had . . . Some men walked with a limp . . . Certain
faces . . . seemed to be muttering a last prayer . . . the white hair of
these women . . . profoundly disquieted me . . . Some of the old
men . . . There were men in the room whom I knew to be re-
lated . . . the [women] who were either too beautiful or too ugly . . .
Others too, both men and women . . . Even in the case of the
men . . . More than one of the men and women . . . Sometimes . . .
But with other people . . . 16

I will call this second type internal or synthesizing iteration, in the
sense that the iterative syllepsis extends not over a wider period
of time but over the period of time of the scene itself.

15 RH II, 6/P II, 605. Without an indication of frequency, but in just as hyper-
bolic a way, cf. RH I, 827/P II, 157: while Saint-Loup went to get Rachel, Marcel
"strolled up and down the road," past the gardens; for these few minutes, "If I
raised my head I could see, now and then, girls sitting in the windows."

16 RH II, 1052-1083/P III, 936-976.


Narrative Discourse

A single scene, furthermore, can contain both types of syllep-
sis. In the course of that same Guermantes matinee, Marcel
evokes in an external iteration the amatory relationship between
the Duke and Odette: "he was always in her house ... he spent
his days and his evenings with Mme. de Forcheville ... he per-
mitted her to receive friends. ... At moments ... the lady in
pink would interrupt him with a sprightly sally. ... It must be

added that Odette was unfaithful to M. de Guermantes " 17

The iterative here obviously synthesizes several months or even
several years of relations between Odette and Basin, and thus
a period of time very much longer than that of the Guermantes
matinee. But it also happens that the two types of iteration
blend to the point that the reader can no longer differentiate
them, or untangle them. For example, in the scene of the Guer-
mantes dinner, near the top of page 1097, we will meet an un-
ambiguous internal iteration: "I cannot, by the way, say how
many times in the course of this evening I heard the word
'cousin' used." But the next sentence, still iterative, can already
bear on a longer period of time: "On the one hand, M. de
Guermantes, almost at every name that was mentioned [in the
course of this dinner, certainly, but perhaps also in a more
habitual way], exclaimed: 'But he's Oriane's cousin!'" The third
sentence perhaps brings us back to the period of the scene: "On
the other hand the word cousin was employed in a wholly dif-
ferent connexion ... by the Turkish Ambassadress, who had
come in after dinner." But the next is an iterative plainly external
to the scene, since it goes on to give a sort of general portrait of
the Ambassadress:

Devoured by social ambition and endowed with a real power of
assimilating knowledge, she would pick up with equal facility the
story of the Retreat of the Ten Thousand or the details of sexual
perversion among birds. . . . She was, incidentally, a dangerous
person to listen to. . . . She was at this period little received in

— so much so that when the narrative returns to the conversa-
tion between the Duke and the Ambassadress, we are not able

17 RH II, 1113-1117/P 111, 1015-1020.



to tell whether we are dealing with this conversation (in the
course of this dinner) or with a wholly other one:

She hoped to give herself a really fashionable air by quoting the
most historic names of the little-known people who were her
friends. At once M. de Guermantes, thinking that she was refer-
ring to people who frequently dined at his table, quivered with joy
at finding himself once more in sight of a landmark and shouted
the rallying-cry: "But he's Oriane's cousin!"

Likewise, one page further on, the iterative treatment that
Proust imposes on the genealogical conversations between the
Duke and M. de Beauserfeuil wipes out all demarcation between
this first dinner at the Guermantes', subject of the present
scene, and the whole of the series it inaugurates.

Thus in Proust the singulative scene itself is not immune to a
sort of contamination by the iterative. The importance of this
mode, or rather of this narrative aspect, is further accentuated by
the very characteristic presence of what I will call the pseudo-
iterative — that is, scenes presented, particularly by their wording
in the imperfect, as iterative, whereas their richness and preci-
sion of detail ensure that no reader can seriously believe they
occur and reoccur in that manner, several times, without any
variation. 18 For example, consider certain long conversations be-
tween Leonie and Francoise (every Sunday at Combray!), be-
tween Swann and Odette, at Balbec with Mme. de Villeparisis,
in Paris at Mme. Swann's, in the pantry between Francoise and
"her" footman, or the scene of Oriane's pun, "Teaser Augus-
tus." 19 In all these cases, and in some others as well, a singular
scene has been converted almost arbitrarily, and without any
modification except in the use of tenses, into an iterative scene.
This is obviously a literary convention (I would readily say nar-
rative license, as we speak of poetic license) that presumes a great
obligingness on the part of the reader or, as Coleridge said, a
"willing suspension of disbelief." This convention is, besides, of

18 Cf. Houston, p. 39.

19 RH I, 77-83/P I, 100-109; RH I, 186-187/P I, 243; RH I, 546-548/P I, 721-723;
RH I, 453^56/P I, 596-599; RH I, 727-732/P II, 22-26; RH I, 1049-1051/P II,


Narrative Discourse

very long standing. I note an example at random in Eugenie
Grandet (dialogue between Mme. Grandet and her husband)
and another in Lucien Leuwen (conversation between Leuwen
and Gauthier), but also in one of Cervantes' exemplary novels:
for example, the monologue of old Carrizales in "The Jealous
Extremaduran," which we are told was spoken "not once but a
hundred times." 20 Every reader naturally interprets this as
hyperbole, not only for its indication of number but also for its
claim of exact identity among several soliloquies almost alike, of
which this one presents a sort of sample. In short, in classical
narrative the pseudo-iterative typically constitutes a figure of
narrative rhetoric which is not required to be taken literally, but
just the reverse. The narrative affirms literally "this happened
every day," to be understood figuratively as "every day some-
thing of this kind happened, of which this is one realization
among others."

It is obviously possible to treat in this way the several exam-
ples of pseudo-iteration noticed in Proust. 21 It seems to me,
however, that their extent, especially when we compare it to the
importance of the iterative in general, prohibits such a limita-
tion. The convention of the pseudo -iterative does not function
in Proust in the intentional and purely figurative mode it takes
in the classical narrative. In Proustian narrative the characteristic
and very marked tendency toward inflating the iterative is in-
tended to be taken in its impossible literalness.

The best (although paradoxical) proof of this is perhaps given
by the three or four times when Proust inadvertently lets a
necessarily singulative passe simple slip into the middle of a
scene presented as iterative — " 'And it will come in the middle
of my luncheon!' she would murmur [ajouta-t-elle a mi-voix] to
herself. ... At the mention of Vigny [Mme. de Villeparisis]
laughed [se mit a rire]. . . . 'The Duchess must be connected with

20 Eugenie Grandet, Gamier ed., pp. 205-206; Lucien Leuwen, Part I, chap. 7;
Cervantes, "The Jealous Extremaduran," in Exemplary Stories, trans. C. A. Jones
(Harmondsworth, England: Penguin, 1972), p. 149.

21 See Pierre Guiraud, Essais de stylistique (Paris, 1971), p. 142.



all that lot,' said [dit\ Francoise" 22 — or links an iterative scene
with a consequence that is singular by definition, as on the page
of the Jeunes Filles en fleurs where we learn from the mouth of
Mme. Cottard that at each of Odette's "Wednesdays" the hero
had "made a complete conquest, first shot, of Mme. Verdurin,"
assuming that action to have a capability for repetition and re-
newal wholly contrary to its nature. 23 We can, no doubt, see in
these apparent blunders the traces of a first draft written in the
singulative, in which Proust supposedly forgot or neglected to
convert certain verbs; but it seems to me sounder to read these
slips as so many signs that the writer himself sometimes "lives"
such scenes with an intensity that makes him forget the distinc-
tion of aspects — and that excludes on his part the purposeful
attitude of the classical novelist using in full awareness a purely
conventional figure. These confusions, it seems to me, instead
reflect in Proust a sort of intoxication with the iterative.

It is tempting to connect this characteristic to what is sup-
posedly one of the dominant features of Proustian psychology,
to wit, a very sharp sense of habit and repetition, a feeling of the
analogy between moments. But the iterative nature of the narra-
tive is not always, as it is in Combray, based on the actually
repetitive and routine aspect of a provincial and petit-bourgeois
life like Aunt Leonie's: this motivation does not apply to the
Parisian milieu or the visits to Balbec and Venice. In fact, and
contrary to what one is often led to believe, the Proustian crea-
ture is as little sensitive to the individuality of moments as he is
spontaneously sensitive to the individuality of places. Moments
in Proust have a strong tendency to resemble and blend with

22 RH I, 43/P I, 57; RH I, 547/P I, 722; RH I, 729/P II, 22. Another out-of-place
passe simple ('"I am sure . . . / said my aunt in a resigned tone [dit mollement
ma tante]") is present in the Clarac-Ferre edition (RH I, 79/P I, 104), as in the
NRF [Gallimard] edition of 1917, but the original (Grasset, 1913, p. 128) gave the
"correct" form: "would say" \"disait"\. This variant seems to have escaped Clarac
and Ferre, who do not call attention to it. The 1917 correction is hard to explain,
but the principle of lectio difficilior gives it precedence by the very reason of its

21 RH 1, 462/PI, 608.


Narrative Discourse

each other, and this capacity is obviously the very condition for
experiencing "involuntary memory." This contrast between the
"singularism" of his spatial sensitivity and the "iteratism" of his
temporal sensitivity is well illustrated, for example, in the sen-
tence from Swann where he speaks of the Guermantes land-
scape, a landscape "whose individuality sometimes, at night, in
my dreams, binds me with a power that is almost fantastic": 24
individuality of place, indefinite, quasi-erratic ("sometimes")
recursiveness of the moment. We find the same contrast again in
this passage from La Prisonniere, where the singularity of a real
morning is blotted out in favor of the "ideal morning" that it
evokes and represents:

Because I had refused to savour with my senses this particular
morning, I enjoyed in imagination all the similar mornings, past or
possible, or more precisely a certain type of morning of which all
those of the same kind were but the intermittent apparition which
I had at once recognised; for the keen air blew the book open of its
own accord at the right page, and I found clearly set out before my
eyes, so that I might follow it from my bed, the Gospel for the day.
This' ideal morning filled my mind full of a permanent reality,
identical with all similar mornings, and infected me with . . . cheer-
fulness. 25

But the mere fact of recurrence is not what defines the most
rigorous form of iteration, the form that is apparently most satis-
fying to the spirit— or most soothing to Proustian sensibility.
The repetition also has to be regular, has to obey a law of fre-
quency, and this law has to be discernible and formulable, and
therefore predictable in its effects. At the time of the first visit to
Balbec, when he has not yet become the intimate of the "little
band," Marcel contrasts these young girls, whose habits are still

24 RH I, 142/P 1, 185. (My emphasis.)

25 RH II, 395/P HI, 26. That these "identities" are a mental construction obvi-
ously does not escape Proust, who writes further on (RH II, 434/P III, 82): "But
each day was for me a different country"; and apropos of the sea at Balbec: "For
none of those Seas ever stayed with us longer than a day. On the morrow there
would be another, which sometimes resembled its predecessor. But I never saw
the same one twice" (RH I, 534/P I, 705. But "twice" perhaps means here "twice
in a row").



unknown to him, with the little traffickers on the beach, with
whom he is already familiar enough to know "where and at what
time it will be possible to see them again." The young girls, on
the contrary, are absent "certain days" that are apparently in-

not knowing the cause of their absence I sought to discover
whether it was something fixed and regular, if they were to be seen
only every other day, or in certain states of the weather, or if there
were days on which no one ever saw them. I imagined myself already
friends with them, and saying: "But you weren't there the other
day?" "Weren't we? Oh, no, of course not; that was because it was a
Saturday. On Saturdays we don't ever come, because. . . "If it
were only as simple as that, to know that on black Saturday it was
useless to torment oneself, that one might range the beach from
end to end, sit down outside the pastry-cook's and pretend to be
nibbling an eclair, poke into the curiosity shop, wait for bathing
time, the concert, high tide, sunset, night, all without seeing the
longed-for little band. But the fatal day did not, perhaps, come once
a week. It did not, perhaps, of necessity fall on Saturdays. Perhaps
certain atmospheric conditions influenced it or were entirely uncon-
nected with it. How many observations, patient but not at all
serene, must one accumulate of the movements, to all appearance
irregular, of those unknown worlds before being able to be sure
that one has not allowed oneself to be led astray by mere coinci-
dence, that one's forecasts will not be proved wrong, before one
elucidates the certain laws, acquired at the cost of so much painful
experience, of that passionate astronomy. 26

I have italicized here the most obvious marks of this anxious
search for a law of recurrence. Certain ones — once a week, every
other day, in certain states of the weather — we will recall a little
later. For the moment, let us notice the most emphatic one, and
perhaps superficially the most arbitrary one: Saturdays. It sends
us back without any possible hesitation to a passage in Swann
where the specific nature of Saturdays has already been ex-
pressed. 27 At Combray that is the day when, in order to leave
Francoise time in the afternoon to go to the market at Roussain-

26 RH I, 625-626/PI, 831.

27 RH I, 84-85/P1, 110-111.



Narrative Discourse

ville, lunch is put forward an hour. A "weekly exception" to
custom, itself obviously a custom in the second degree, one of
those variations that, "repeated at regular intervals and in iden-
tical form, did no more, really, than print a sort of uniform
pattern upon the greater uniformity of LLeonie's] life," which
she, and all her household with her, "clung to ... as much as to
the rest"— and all the more so since the regular "asymmetry" of
Saturdays, unlike that of Sundays, is specific and original, pecu-
liar to the hero's family and almost incomprehensible to others.
Whence the "civic," "national," "patriotic," "chauvinist" na-
ture of the event, and the atmosphere of ritual surrounding it.
But most characteristic in this text, perhaps, is the idea (ex-
pressed by the narrator) that this custom, becoming "the favour-
ite theme for conversations, for pleasantries, for anecdotes
which can be embroidered . . . would have provided a nucleus,
ready-made, for a legendary cycle, if any of us had had the epic
mind"— the classic passage from ritual to explanatory or illus-
trative myth. The reader of the Recherche is well aware who, in
that family, has "the epic mind" and will one day write the
family's "legendary cycle," but the main point here is the spon-
taneously established link between narrative inspiration and re-
petitive event, that is, in one sense, the absence of event. We are
present to some extent at the birth of a vocation, which is in fact
that of iterative narrative. But that is not all: the ritual was once
(or perhaps several times, but to a certainty only occasionally
and not every Saturday) slightly violated (and thus confirmed)
by the visit of a "barbarian" who, nonplussed to find the family
at lunch so early, heard from the paterfamilias, guardian of tra-
dition, the response: "You see, it's Saturday!" This irregular,
perhaps singular, event is immediately integrated into the cus-
tom in the form of a tale by Francoise, a tale which will be
repeated dutifully from then on, no doubt every Saturday, to
the general satisfaction:

and then, to add to her own enjoyment, [she] would prolong the
dialogue, inventing a further reply for the visitor to whom the
word "Saturday" had conveyed nothing. And so far from our
objecting to these interpolations, we would feel that the story was



not yet long enough, and would rally her with: "Oh, but surely he
said something else as well. There was more than that, the first
time you told it."

My great-aunt herself would lay aside her work, and raise her
head and look on at us over her glasses.

Such in fact is the first manifestation of the "epic" spirit. All that
is left is for the narrator to treat that element of the Sabbath
ritual like the others, that is, in the iterative mode, in order to
"iteratize," as it were, the deviant event in its turn, in accord
with this irresistible process: singular event — repetitive
narrating — iterative narrative (of that narrating). Marcel tells (at)
one time how Francoise told often what happened undoubtedly
only once: or how to turn a unique event into the subject of an
iterative narrative. 28

Determination, Specification, Extension

Every iterative narrative is a synthetic narrating of the events
that occur and reoccur in the course of an iterative series that is
composed of a certain number of singular units. Take the series:
Sundays in the summer of 1890. It is composed of a dozen real
units. The series is defined, first, by its diachronic limits (be-
tween the end of June and the end of September in the year
1890), and then by the rhythm of recurrence of its constituent
units: one day out of seven. We will term the first distinguishing
characteristic determination, and the second, specification. Finally,
we will term the diachronic extent of each of the constituent
units, and consequently of the constituted synthetic unit, exten-
sion: for instance, the account of a Sunday in summer covers a
synthetic duration that could be twenty-four hours but can just
as easily (as is the case in Combray) be limited to about ten hours,
from getting up to going to bed.

28 In an earlier version (Contre Sainte-Beuve, ed. Bernard de Fallois, pp. 106-
107) — a version which, let us note in passing, is set in Paris, and in which the
cause of the Sabbath asymmetry is therefore not the Roussainville market but a
class given at the beginning of the afternoon by the hero's father — the commem-
oration of the incident is not narrative only; it is a mimetic ritual which consists
of "bringing about the scene" (that is, its repetition) by "inviting on purpose"
some barbarians.


Narrative Discourse

Determination. The diachronic limits of a series can be left im-
plicit, especially when we are dealing with a recurrence that in
practice can be considered unlimited: if I say, "the sun rises
every morning," it simply would be ridiculous to want to state
precisely since when and until when. The events that
novelistic-type narration is concerned with are obviously less
permanent, so those series are generally defined by the indica-
tion of their beginning and their end. But this determination can
very easily remain indefinite, as when Proust writes: "After a
certain year we never saw [Mile. Vinteuil] alone." 29 Sometimes it
is definite, made so either by an absolute date ("When spring
drew round, ... I used often to see [Mme. Swann] entertaining
her guests in her furs") 30 or (more often) by reference to a singu-
lar event. For example, the breach between Swann and the Ver-
durins puts an end to one series (meetings between Swann and
Odette at the Verdurins') and at the same time inaugurates
another (obstructions placed by the Verdurins in the way of
Swann and Odette's affair): "And so that drawing-room which
had brought Swann and Odette together became an obstacle in
the way of their meeting. She no longer said to him, as she had
said in the early days of their love . . . " 31

Specification. It too can be indefinite, that is, indicated by an
adverb of the type sometimes, certain days, often, etc. It can on the
other hand be definite, either in an absolute way (this is fre-
quency as such: every day, every Sunday, etc.) or else in a more
relative and more irregular way, nonetheless expressing a very
strict law of concomitance, like that presiding over the choice of
walks at Combray: the Meseglise way on days of unsettled weather,
the Guermantes way on days of clear weather. 31 Definite or not,
these are simple specifications, or rather I have presented them
as such. There also exist complex specifications, where two (or
several) laws of recurrence are superimposed, which is always

29 RH I, 113/P I, 147.
3 <> RH I, 482/P I, 634.

31 RH I, 221/P I, 289.

32 RH I, 115/P I, 150 and RH I, 127/P I, 165.



possible when iterative units can be embedded in each other: for
example, the simple specification every month of May and the
simple specification every Saturday, which combine in the com-
plex specification every Saturday in the month of May. 33 And we
know that all the iterative specifications of Combray (every day,
every Saturday, every Sunday, every day of good or bad
weather) are themselves governed by the overarching specifica-
tion every year between Easter and October — and also by the determi-
nation during my years of childhood. We can obviously produce
much more complex definitions, such as, for example, "every
hour on Sunday afternoons in the summer when it didn't rain,
between my fifth and fifteenth years": this is approximately the
law of recurrence governing the piece about the passing of the
hours during the hero's reading in the garden. 34

Extension. An iterative unit can have so slight a duration that it
gives no hold for narrative expansion: for example, a statement
such as "every evening I go to bed early" or "every morning my
alarm goes off at seven o'clock." Such iterations are to some
extent pinpointed. On the other hand, an iterative unit such as
sleepless night or Sunday at Combray has enough extent to become
the subject of an expanded narrative (five and forty-five pages,
respectively, in the text of the Recherche). So it is here that the
specific problems of iterative narrative appear. In effect, if in
such a narrative one wanted to retain only the invariant features
common to all the units of the series, one would be doomed to
the diagrammatic barrenness of a fixed timetable, like "to bed at
nine o'clock, an hour of reading, several hours of sleeplessness,
sleep in the early morning," or "getting up at 9 o'clock, break-
fast at 9:30, mass at 11, lunch at 1, reading from 2 to 5, etc." — an
abstraction which is obviously due to the synthetic nature of the
iterative but which is unable to satisfy either the narrator or the
reader. That is when, to "concretize" the narrative, internal de-
terminations and specifications of the iterative series step in, offer-
ing a means of diversification (a means, therefore, of variation).

"RH I, 85/P I, 112.
34 RH I, 66/P I, 87-88.


Narrative Discourse

In fact, as we have already glimpsed, determination does not
mark only the outer limits of an iterative series; it can just as
easily punctuate its stages, and divide it into subseries. For in-
stance, I said that the breach between Swann and the Verdurins
brought one series to an end and inaugurated another; but we
could just as well say, moving to the higher unit, that this singu-
lar event determines, in the series "meetings between Swann
and Odette," two subseries (before the breach/after the breach),
each of which functions as a variant of the synthetic unit: meet-
ings at the Verdurins'/meetings away from the Verdurins'. More
plainly still, we can take as internal determination the interposi-
tion, into the series of Sunday afternoons at Combray, of the
meeting with the Lady in pink at Uncle Adolphe's 35 — a meeting
whose consequence will be the falling-out between Marcel's
uncle and parents, and the shutting up of his "little sitting-
room." Hence this simple variation: before the Lady in pink,
Marcel's routine includes a stop in his uncle's little room; after
the Lady in pink, this ritual disappears and the boy goes directly
up to his bedroom. 36 Similarly, a visit from Swann 37 will deter-
mine a shift in the subject (or at least in the setting) of Marcel's
amorous daydreams: before this visit, and under the influence
of his earlier reading, they take place against a background of a
wall gay with purple flowers, shaped like bunches of grapes,
hanging over water; after this visit and Swann's disclosure of
the amicable relations between Gilberte and Bergotte, these
daydreams will stand out from "a wholly different background,
the porch of a gothic cathedral" (like those Gilberte and Bergotte
visit together). But previously these fantasies had been changed
by a piece of information (from Dr. Percepied) about the flowers
and spring waters of the Guermantes park: 38 the watery erotic
region had been identified with Guermantes, and its heroine
had taken on the features of the Duchess. So we have here an
iterative series, amorous daydreams, subdivided by three singular
events (reading, Percepied information, Swann information)

35 RH I, 55-60/P I, 72-80.

36 RH I, 61/P I, 80.

" RH I, 68-76/P 1, 90-100.
38 RH 1, 132/P I, 172.



into four "determined" sections (before reading, between read-
ing and Percepied, between Percepied and Swann, after Swann)
that constitute the same number of variants (daydreams without
a distinct setting/in a river setting/in the same setting identified
with Guermantes and the Duchess/in a Gothic setting with Gil-
berte and Bergotte). But this series is dislocated, in the text of
Combray, by the system of anachronies: the third section, whose
chronological position is obvious, will not be mentioned until
some sixty-four pages later, on the occasion of the walks by the
Guermantes way. Analysis must therefore reconstitute it here,
despite the actual order of the text, as an underlying and hidden
structure. 39

We should not, however, infer too quickly from this notion of
internal determination that the interposition of a singular event
always has the effect of "determining" the iterative series. As
we will see later, the event can be simply an illustration, or on
the contrary an exception without follow-up, producing no
change. An example is the episode of the steeples of Martin ville,
after which the hero will resume as if nothing had happened ("I
never thought again of this page") 40 his previous custom of
carefree and (apparently) spiritually profitless walks. So we
must differentiate, among singulative episodes interposed into
an iterative section, between those which have a determinative
function and those which do not.

Beside these definite internal determinations, we find indefi-
nite ones, of a type we have already met: "starting from a certain
year." The walks by the Guermantes way give an example that
is remarkable in its conciseness and the apparent confusion of its
writing: "And then it happened that [Puis il arriva que], going the
'Guermantes way,' I passed occasionally \je passai parfois] by a row
of well-watered little gardens, over whose hedges rose clusters
of dark blossoms. I would stop before them, hoping to gain

-'Another series, only a few pages away— that of the daydreams of literary
ambition— undergoes a modification of the same order after the Duchess's ap-
pearance in church: "How often, after that day, in the course of my walks along
the 'Guermantes way,' and with what an intensified melancholy did I reflect on
my lack of qualification for a literary career" (RH I 137/P I 178)

40 RH I, 140/P I, 182.


Narrative Discourse

some precious addition to my experience, for I seemed to . . . " 41
We are indeed dealing with an internal determination: starting
from a certain date, the walks along the Vivonne include an
element which was lacking until then. The difficulty of the text is
due in part to the paradoxical presence of an iterative in the
passe simple ("I passed occasionally [je passai parfois]") —
paradoxical but perfectly grammatical, just like the iterative
passe compose of the opening sentence of the Recherche, which
could also just as well be written in the passe simple ("*For a
long time I went to bed early l*Longtemps je me couchai de
bonne heure]"), but not in the imperfect, which does not have
enough syntactic autonomy to begin an iteration. The same pat-
tern occurs elsewhere after a definite determination: "Once we
had got to know [Unefois que nous connumes] this road, for a change
we would return [revinmes] — that is, if we had not taken it on the
outward journey — by another which ran through the woods of
Chantereine and Canteloup." 42

The variants obtained by internal determination are still, I
emphasize, iterative in kind: there are several reveries in a
Gothic setting, as there are several reveries in a river setting; but
the relationship they maintain is diachronic in kind, and there-
fore singulative, like the unique event that separates them: one
subseries comes after the other. Internal determination therefore
arises from singulative segments in an iterative series. Internal
specification, on the other hand, is a technique of purely iterative
diversification, since it consists simply of subdividing the recur-
rence to get two variants in a (necessarily iterative) relationship
of alternation. For example, the specification every day can be
divided into two halves that are not successive (as they are in
every day before/after such an event), but alternating, in the sub-
specification one day out of two. We have already met one form —
less rigorous, actually — of this principle in the opposition good
weatherlbad weather that articulates the law of recurrence of the
walks at Combray (apparently every afternoon except Sunday). We
know that a considerable part of the text of Combray is composed

41 RHI, 132/P 1, 172.
42 RH I, 545/P I, 720.



in accord with that internal specification, which governs the
alternation walks toward Mesegliselwalks toward Guermantes: "the
habit we had of never going both ways on the same day, or in
the course of the same walk, but the 'Meseglise way' one time
and the 'Guermantes way' another" 43 — an alternation in the
temporality of the story, which the arrangement of the narra-
tive, as we have already seen, 44 is careful not to respect, devot-
ing one segment (pages 103-127) to the Meseglise way, then
another (pages 127-141) to the Guermantes way. 45 So much so
that the totality of Combray 11 (after the detour via the madeleine)
is arranged approximately according to these iterative specifi-
cations: (1) every Sunday, pages 37-103 (with a parenthetical
every Saturday, pages 83-88); (2) every (week) day of unsettled
weather, pages 103-127; (3) every day of fine weather, paees 127-
141. 4 «

There we were dealing with a definite specification. Other
occurrences of this technique appear in the Recherche, but they
are never exploited in so systematic a way. 47 Most often, in fact,
the iterative narrative is articulated through indefinite specifi-
cations of the type sometimes/sometimes, which permits a very flex-
ible system of variations and very elaborate diversification with-
out ever leaving the iterative mode. For example, the literary
anxieties of the hero during his walks to Guermantes are divided
into two classes {sometimes . . . but other times) according to wheth-

RH I, 104/P I, 135. The term alternation, and Proust's own expression (one
time the Meseglise way and another time the Guermantes way), should not lead
us to believe in so regular a succession, which would assume that the weather in
Combray is fine strictly one day out of two; in fact, it seems that the walks of the
Guermantes way are much rarer (see RH I, 102'P I 133)

44 Pp. 84-85.

45 We are in fact dealing with a three-term specification (days of fine weather/
of unsettled weather/of bad weather), the third of which involves no narrative
expansion: "If the weather was bad all morning, my family would abandon the
idea of a walk, and 1 would remain at home" (RH I, 117/P I, 153)

46 The composition of Combray I, if we set aside the memory-grounded open-
ing of pp. 3-7 |RH I/P I, 3-91 and the transition (the madeleine) of pp. 33-36 [RH
J/P I, 43-48], is governed by the succession of an iterative section (every evening,
RH I, 7-17/P I, 9-21) and a singulative section (the evening of Swann's visit, RH I,
17-33/PI, 21^13). *

47 For example, Eulalie's dominical visits, sometimes with, sometimes without
the Cure of Combray (RH I, 82/P I, 108).


Narrative Discourse

er he reassures himself about his future by counting on the
miraculous intervention of his father or whether he sees him-
self desperately alone facing the "nullity of [his] intellect." 48 The
variations of the walks to Meseglise according to the degrees of
"bad weather" fill, or rather engender, a text of three pages 49
composed according to this system: often (threatening weather)/
at other times (shower during the walk, shelter in the Roussainville
wood)/fl/so often (shelter under the portal of Saint-Andre-des
Champs)lsometimes (weather so worsened that there is a return
home). The system, moreover, is a little more complex than this
enumeration paralleling the text indicates, for variants two and
three are in fact subclasses of one same class: sudden shower.
The real structure is thus:

1. Weather threatening but without sudden shower.

2. Sudden shower:

a. shelter in the woods,

b. shelter under the portal.

3. Weather definitely worsened. 50

But the most characteristic example of constructing a text on
the resources of internal specification alone is undoubtedly the
portrait of Albertine that comes near the end of the Jeunes Filles
en fleurs. Its theme is, as we know, the variety of Albertine's
countenance, which symbolizes the mobile and elusive nature of
the young girl, a "creature in flight" par excellence. But however
changeful she may be, and even though Proust uses the expres-
sion "each of these Albertines," the description treats "each" of
these variants not as an individual, but as a type, a class of

48 RH I, 132-133/P I, 173-174.

«RHI, 115-11 7/P I, 150-153.

50 Another complex system of internal specifications is the meetings (and
nonmeetings) with Gilberte on the Champs-Elvsees, a system articulated like
this (RH I, 302/P I, 395):

(1) days of Gilberte's presence

(2) days of her absence

(a) given notice of

— for lessons
— for a party

(b) impromptu

(c) impromptu but foreseeable (bad weather).



occurrences: certain days I other days I other times I sometimes I often I
most often I it might happen I at times, even . . . : as much as a
collection of faces, this portrait is a catalogue of frequentative

So it was with Albertine as with her friends. On certain days, slim,
with grey cheeks, a sullen air, a violet transparency falling ob-
liquely from her such as we notice sometimes on the sea, she
seemed to be feeling the sorrows of exile. On other days her face,
more sleek, caught and glued my desires to its varnished surface
and prevented them from going any farther; unless 1 caught a
sudden glimpse of her from the side, for her dull cheeks, like
white wax on the surface, were visibly pink beneath, which made
me anxious to kiss them, to reach that different tint which thus
avoided my touch. At other times happiness bathed her cheeks
with a clarity so mobile that the skin, grown fluid and vague, gave
passage to a sort of stealthy and subcutaneous gaze, which made it
appear to be of another colour but not of another substance than
her eyes; sometimes, instinctively, when one looked at her face
punctuated with tiny brown marks among which floated what
were simply two larger, bluer stains, it was like looking at the egg
of a goldfinch — or often like an opalescent agate cut and polished
in two places only, where, from the heart of the brown stone,
shone like the transparent wings of a sky-blue butterfly her eyes,
those features in which the flesh becomes a mirror and gives us the
illusion that it allows us, more than through the other parts of the
body, to approach the soul. But most often of all she shewed more
colour, and was then more animated; sometimes the only pink
thing in her white face was the tip of her nose, as finely pointed as
that of a mischievous kitten with which one would have liked to
stop and play; sometimes her cheeks were so glossy that one's
glance slipped, as over the surface of a miniature, over their pink
enamel, which was made to appear still more delicate, more pri-
vate, by the enclosing though half-opened case of her black hair;
or if might happen that the tint of her cheeks had deepened to the
violet shade of the red cyclamen, and, at times, even, when she was
flushed or feverish, with a suggestion of unhealthiness which
lowered my desire to something more sensual and made her
glance expressive of something more perverse and unwholesome,
to the deep purple of certain roses, a red that was almost black;
and each of these Albertines was different, as in every fresh appear-
ance of the dancer whose colours, form, character, are transmuted


Narrative Discourse

according to the innumerably varied play of a projected
limelight. 51

Naturally, the two techniques (internal determination and
specification) can come into play together in the same section.
That is what happens in a very clear and felicitous way in the
paragraph that begins the section of Combray devoted to the
"two ways" — begins it by evoking through anticipation the re-
turns home from the walk:

We used always to return from our walks in good time to pay aunt
Leonie a visit before dinner. In the first weeks of our Combray holi-
days, when the days ended early, we would still be able to see, as
we turned into the Rue du Saint-Esprit, a reflection of the western
sky from the windows of the house and a band of purple at the
foot of the Calvary, which was mirrored further on in the pond; a
fiery glow which, accompanied often by a cold that burned and
stung, would associate itself in my mind with the glow of the fire
over which, at that very moment, was roasting the chicken that
was to furnish me, in place of the poetic pleasure I had found in
my walk, with the sensual pleasures of good feeding, warmth and
rest. But in summer, when we came back to the house, the sun
would not have set; and while we were upstairs paying our visit to
aunt Leonie its rays, sinking until they touched and lay along her
window-sill, would there be caught and held by the large inner
curtains and the bands which tied them back to the wall, and split
and scattered and filtered; and then, at last, would fall upon and
inlay with tiny flakes of gold the lemonwood of her chest-of-
drawers, illuminating the room in their passage with the same
delicate, slanting, shadowed beams that fall among the boles of
forest trees. But on some days, though very rarely, the chest-of-
drawers would long since have shed its momentary adornments,
there would no longer, as we turned into the Rue du Saint-Esprit,
be any reflection from the western sky burning along the line of
window-panes; the pond beneath the Calvary would have lost its
fiery glow, sometimes indeed had changed already to an opalescent
pallor, while a long ribbon of moonlight, bent and broken and
broadened by every ripple upon the water's surface, would be
lying across it, from end to end. 52

51 RH I, 708/P I, 946-947. (My emphasis.)
"RH I, 102/PI, 133.



The first sentence here lays down an absolute iterative princi-
ple, "We used always to return from our walks in good time";
opening within it is a diversification by internal determination,
spring/summer, 53 which governs the following two sentences; fi-
nally, an internal specification, which seems to bear on both of
the two preceding sections, introduces a third exceptional (but
not singulative) variant, "on some days, though very rarely" (these
are apparently the days of walking toward Guermantes). The
complete iterative system, then, is articulated according to the
diagram, which reveals, under the apparently even continuity of
the text, a more complex and more entangled hierarchical struc-

always early

fairly early

{spring: twilight
{summer: sun

[often: cold

rarely f(zero)

later: already dark \ sometimes: opalescent

(One may perhaps find, and quite rightly, that such a
schematization does not account for the "beauty" of this page;
but such is not its purpose. The analysis here is not placed at the
level of what in Chomskian terms would be called "surface
structures," or in Hjelmslevo-Greimassian terms stylistic "man-
ifestation," but at the level of "immanent" temporal structures
that give the text its skeleton and its foundation — and without
which it would not exist [since in this case, without the system
of determinations and specifications we have reconstituted, the
text would necessarily, and flatly, be limited to its first sentence
alone]. And, as usual, the analysis of foundations discloses,
beneath the smooth horizontality of successive syntagms, the
uneven system of paradigmatic selections and relationships. If
the object of analysis is indeed to illuminate the conditions of

53 A determination that is itself iterative, since it is repeated every year. The
opposition spring/summer, which at the level of a single year is pure determina-
tion, thus becomes, if one encompasses the totality of the Combray period, a
combination of determination and specification.


Narrative Discourse

existence — of production — of the text, it is not done, as people
often say, by reducing the complex to the simple, but on the
contrary by revealing the hidden complexities that are the secret
of the simplicity.)

This "impressionist" theme of the variations, according to
time and season, in the illumination and thus in the very image
of the site 54 — the theme of what Proust calls the "varied land-
scape of the hours" — again governs the iterative descriptions of
the sea at Balbec, and particularly the one on pages 605-608 of
the Jeunes Filles en fleurs:

Regularly, as the season advanced, the picture that I found there
in my window changed. At first it was broad daylight. . . .

Presently the days grew shorter. ... A few weeks later, when I
went upstairs, the sun had already set. Like the one that I used to
see at Combray, behind the Calvary, when I was coming home
from a walk and looking forward to going down to the kitchen
before dinner, a band of red sky over the sea . . .

This first series, variations by determination, is followed by
another, variations by specification:

1 was on all sides surrounded by pictures of the sea.

But as often as not they were, indeed, only pictures. . . . At one
time it was an exhibition of Japanese colour-prints. ... I had more
pleasure on evenings when a ship . . . Sometimes the ocean . . .
Another day the sea . . . And sometimes . . .

The same pattern occurs two pages later, apropos of the arrivals
at Rivebelle and even closer to the Combrayan version, although
that one is not recalled this time: "At first, when we arrived
there, the sun used just to have set, but it was light still. . . .
Presently night had always fallen when we left the carriage." In
Paris, in La Prisonniere, the mode of variation will be rather of an
auditory kind: it is the morning nuances of the sound of the bells
or the noises of the street that inform Marcel, still buried be-

54 "Difference of lighting modifies no less the orientation of a place . . . than
would a distance in space actually traversed in the course of a long journey" (RH
I, 511/P I, 673).



neath his bedclothes, what the weather is. 55 Remaining constant
is the extraordinary sensitivity to variations of climate, the al-
most maniacal attention (which Marcel figuratively inherits from
his father) to the movements of the barometer within, and, with
respect to what concerns us here, the characteristic fertile bond
between the temporal and the meteorological, developing to its
furthest consequence the ambiguity of the temps franqais — 1
mean the French word temps, expressing both time and weather:
an ambiguity already exploited by the magnificently pre-
monitory title of one of the sections of Les Plaisirs et les jours:
"Reveries couleur du Temps." The return of the hours, the days,
the seasons, the circularity of the cosmic movement, remains
both the most constant motif and the most exact symbol of what
I will readily call Proustian iteratism.

Such are the resources of strictly iterative diversification
(internal determination and internal specification). When these
are exhausted, two recourses still remain, whose common fea-
ture is that they put the singulative at the service of the iterative.
The first we already know: the convention of the pseudo-
iterative. The second is not a figure; it consists of invoking — in a
completely literal and avowed way — a singular event, either as
illustration and confirmation of an iterative series (it is thus
that . . . ), or as exception to the rule that has just been estab-
lished (once, however . . .). An example of the first function is this
passage from the jeunes Filles en fleurs: "Now and then [this is the
iterative law] a pretty attention from one or another of them
would stir in me vibrations which dissipated for a time my de-
sire for the rest. Thus one day Albertine [this is the singular
illustration] . . . " 56 An example of the second is the episode of
the steeples of Martinville, plainly presented as a deviation from
habit: ordinarily, once he was back from his outing, Marcel for-

55 RH II, 383/P III, 9; RH II, 434/P III, 82; RH II, 459/P III, 116.

56 RH 1, 682/P I, 911. 1 would hesitate, on the other hand, to say the same of the
three episodes that illustrate Marcel's "progress" with Gilberte ("One day," gift
of the agate marble; "Another time," gift of the pamphlet by Bergotte; "And . . .
another day": " 'you may call me "Gilberte," ' " RH I, 307-308/P I, 402-403), be-
cause these three "examples" perhaps exhaust the series, like the "three halting-
points" in the progress of forgetting after Albertine's death (RH II, 774-820/P III,
559-623). Which amounts to an anaphoric singulative.


Narrative Discourse

doh^thT" 8510 ^ ^ h3d eX P erience d and did not try to de-
cipher their significance; "Once, however "" he eoes fl
and wntes down immediately the descriptive ptcf that 7s h I
ftrst work and the sign of his vocation. Even more eX pUc t m i

everything was quite normal when I returnpH f™™ maaem >
the Duchess.''-^ Thus thrmJv 1 r « urne £ d f ™ my visit to

vely either by respecting its code or by transgressing fi�� whkh
is another way of manifesting it. �� ' *

Internal and External Diachrony

^^z:Zz'; g :tt "r ilera,,ve unit as ™ r "^

only two variants of the same theme. And in flct anT I™

.o *, e„ d „ f lhe senes , mrying with0 :;ri ; m T t s sTZL 8

what happens in thp firct ,-..,„„ r ^ s actually

. rr* llb in tne «rst pages of Swann, where the onlv t«™
poral indications are either of tho f .• , y tem "

57 RH I, 138/PI, 179

58 RH II, 415-116/P III, 54-55.



now �� �� �� now) or else are devoted to the internal duration of the
synthetic night, whose unfolding governs the progression of the
text (when I had put out my candle . . . half an hour later . . . then . . . at
the same time . . . gradually . . . then . . .), without anything to indi-
cate that the passage of years alters this unfolding in any way

But with the play of internal determinations, the iterative nar-
rative can, as well, take real diachrony into account and inte-
grate it into the iterative's own temporal progression — can re-
count, for example, the unit Sunday in Combray, or walks about
Combray, noting the changes that the time elapsed (about ten
years) in the course of the real series of weeks spent at Combray
has brought to the unfolding of the unit. These changes can be
looked at not as interchangeable variations, but as irreversible
transformations: deaths (Leonie, Vinteuil), fallings-out
(Adolphe), the hero's maturation and aging (new interests: Ber-
gotte; new acquaintances: Bloch, Gilberte, the Duchesse de
Guermantes; decisive experiences: discovery of sexuality;
traumatizing scenes: "first concession," profanation at
Montjouvain). Inevitably the question then arises of the rela-
tions between internal diachrony (that of the synthetic unit) and
external diachrony (that of the real series), and of their possible
interferences. That is what in fact takes place in Combray II, and
J. P. Houston was able to maintain that there the narrative
moved forward simultaneously along the three time periods of
the day, the season, and the years. 59 Matters are not entirely so
clear and systematic, but it is true that in the section devoted to
Sundays, the morning comes at Easter and the afternoon and
evening on Ascension Day, and that Marcel's pursuits in the
morning seem to be those of a child and in the afternoon, those
of an adolescent. Even more clearly, the two walks, and in par-
ticular the walk toward Meseglise, take account, in the succes-
sion of their singular or customary episodes, of the flow of the
months of the year (lilacs and hawthorns in bloom at Tanson-
ville, autumn rains at Roussainville) and the flow in the years of
the hero's life (a very young child at Tansonville, an adolescent

"Houston, p. 38.



Narrative Discourse

tormented by desire at Meseglise, with the final scene explicitly
later still). 60 And we have already noted the diachronic break
that the Duchess's appearance in church introduces into the
walks to Guermantes. In all these cases, therefore, Proust suc-
ceeds in treating the internal and external diachronies in an
approximately parallel way — thanks to a skillful arrangement of
episodes — without overtly departing from the frequentative
tense that he took as the base for his narrative. Likewise, the
love between Swann and Odette, between Marcel and Gilberte,
will develop to some extent by iterative plateaux, marked by a
very characteristic use of those thenceforth' s, since's, mow's, 61
which treat every story not as a train of events bound by a
causality but as a succession of states ceaselessly substituted for
each other, with no communication possible. More than usual,
the iterative here is the temporal mode (the aspect) of that sort of
perpetual forgetting, of innate incapacity on the part of the
Proustian hero (Swann always, Marcel before the revelation) to
perceive the continuity of his life, and thus the relation of one
"time" to another. When Gilberte, whose inseparable compan-
ion and "great favorite" he now is, points out to him the prog-
ress of their friendship since the time of the games of prisoner's
base on the Champs-Elysees, Marcel, for want of being able to
reconstitute in himself a situation now past and therefore de-
stroyed, is as incapable of measuring that distance as he will be
later of conceiving how he could once have loved Gilberte, and
how he could have imagined the time when he would no longer
love her as so different from what in fact it becomes:

60 "Some years later" (RH I, 122/P I, 159).

61 "Now [translator: my translation!, every evening. . . " (RH I, 180/P I, 234);
"there was one thing that was, now, invariable" (RH I, 180/P I, 235); "Now |his
jealousy] had food in store, and Swann could begin to grow uneasy afresh every
evening" (RH I, 217/P I, 283); "Gilberte's parents, who for so long had prevented
me from seeing her, now ..." (RH I, 385/P I, 503); "Now, whenever I had to write
to Gilberte. . ." (RH I, 481/P I, 633). Let us leave to the computer the trouble
of making this list complete for the whole of the Recherche; here are three more
occurrences of it, very close together: "It was already nightnou? when I exchanged
the warmth of the hotel ... for the railway carriage into which I climbed with
Albertine" (RH II, 310/P II, 1036); "Included in the number of Mme. Verdurin's
regular frequenters . . . had been, for some months now, M. de Charlus" (RH II,
310/P II, 1037); "Now it was, quite unconsciously, because of that vice that they
found him more intelligent than the rest" (RH "ll, 313/P II, 1041).



She spoke of a change the occurrence of which I could verify onlv
by observing it from without, finding no trace of it within myself,
for it was composed of two separate states on both of which 1 could
not, without their ceasing to be distinct from one another, succeed
in keeping my thoughts fixed at one and the same time. 62

To keep his thoughts fixed on two moments at the same time is
almost always, for the Proustian creature, to consider them
identical and to merge them: this strange equation is itself the
law of the iterative.

Alternation, Transitions

It is as though Proustian narrative substituted for summary,
which is the synthetic form of narration in the classical novel
and which, as we saw, is absent from the Recherche, a different
synthetic form, the iterative: a synthesis not by acceleration, but
by assimilation and abstraction. Thus the rhythm of the narra-
tive in the Recherche is essentially based not, like that of the
classical novel, on the alternation of summary and scene, but on
another alternation, that of iterative and singulative.

Generally that alternation overlays a system of functional
subordinations that analysis can and should elucidate. We have
already encountered the system's two basic types of relation-
ship: the iterative section with a descriptive or explanatory func-
tion subordinated to (and generally inserted within) a singula-
tive scene (example, the wit of the Guermantes in the dinner at
Oriane's), and the singulative scene with an illustrative function
subordinated to an iterative development (example, the steeples
of Martinville, in the series of walks to Guermantes). But more
complex structures exist: when, for example, a singular anec-
dote illustrates an iterative development that is itself subordi-
nated to a singulative scene (for instance, Princesse Mathilde's
reception, 63 illustrating the wit of the Guermantes), or inversely,
when a singulative scene subordinated to an iterative section
calls up in its turn an iterative parenthesis, which is what hap-
pens when the episode of the meeting with the Lady in pink —

"RHI, 410-411/PI, 538.

63 RH 1, 1052-1053/P II, 468-169.


Narrative Discourse

told, as we have already seen, for its indirect effects on the
hero's Sundays in Combray — opens with a development de-
voted to Marcel's youthful passion for the theatre and actresses,
a development necessary to explain his unexpected visit to his
Uncle Adolphe. 64

But it sometimes happens that the relationship eludes all
analysis, and even all definition, the narrative passing from one
aspect to the other without worrying about their reciprocal
functions, and even apparently without noticing them. Robert
Vigneron came across such effects in the third part of Swann,
and believed it possible to attribute what appeared to him "inex-
tricable confusion" to the last-minute reshuffling imposed by
the split edition of the first volume of the Grasset edition: in
order to put the brilliant piece on the Bois de Boulogne "today"
at the end of that volume (and thus at the end of Du cote de chez
Swann) and connect it somehow or other to what precedes it,
Proust supposedly had to change quite decidedly the order of
the various episodes placed on pages 482-511 of the Grasset
edition. 65 But these interpolations would have entailed various
chronological difficulties that Proust would not have been able
to mask except at the cost of a temporal "camouflage" whose
crude and clumsy medium would be the (iterative) imperfect:

To dissimulate this chronological and psychological confusion, the
author tries to disguise single actions as repeated actions and slyly
daubs his verbs with a whitewash of imperfects. Unfortunately,
not only does the singularity of some actions make their habitual
repetition unlikely, but, even worse, in places obstinate passes
definis elude the whitewash and reveal the trick. 66

Relying on this explanation, Vigneron went so far as to reconsti-
tute by way of hypothesis the "original order" of the text that
was so unseasonably disarranged. A most risky reconstitution, a

M RH I, 55-57/P I, 72-75.

65 [Translator's note.] The pages Genette refers to from the Grasset edition
appear in a slightly different version on RH I, 301-318/P I, 394^17.

66 Robert Vigneron, "Structure de Swann: pretentions et defaillances," MP, 44
(November 1946), 127.



most fragile explanation: we have already met several examples
of pseudo-iterative (for that is indeed what we are dealing with
here) and of aberrant passes simples in the parts of the Recherche
that did not suffer in any way from the forced truncation of 1913
and those that we can note at the end of Swann are not the most
surprising ones.

Let us look a little more closely at one of the passages
Vigneron incriminates: it is pages 486-189 of the Grasset edi-
tion. 67 Their subject is those winter days when the Champs-
Elysees are covered with snow, but when a ray of unexpected
sunlight in the afternoon sends Marcel and Francoise forth on an
impromptu walk, with no hope of meeting Gilberte. As
Vigneron says in different words, the first paragraph ("And on
those days when") is iterative: its verbs are in the imperfect
tense for repeated action. "In the next paragraph," writes
Vigneron ("Francoise found it too cold"), "the imperfects and
the passes simples follow each other with no apparent reason,
as if the author, incapable of definitively adopting one point of
view rather than the other, had left his temporal transpositions
incomplete." To let the reader decide, I will quote that para-
graph here as it appears in the edition of 1913:

Frangoise found it too coid to stand about, so we walked to the Pont
de la Concorde to see the Seine frozen over, on to which everyone,
even children, walked fearlessly, as though upon an enormous
whale, stranded, defenceless, and about to be cut up. We returned
to the Champs-Elysees; I was growing sick with misery between the
motionless wooden horses and the white lawn, caught in a net of
black paths from which the snow had been cleared, while the statue
that surmounted it held in its hand a long pendent icicle which
seemed to explain its gesture. The old lady herself, having folded
up her Debats, asked a passing nursemaid the time, thanking her
with "How very good of you!" then begged the roadsweeper to
tell her grandchildren to come, as she felt cold, adding "A
thousand thanks. I am sorry to give you so much trouble!" Sud-
denly the sky was rent in two: between the punch-and-judy and
the horses, against the opening horizon, I had just seen, like a

67 RH I, 303-305/P I, 397-399.


Narrative Discourse

miraculous sign, Mademoiselle's blue feather. And now Gilberte
was running at full speed towards me, sparkling and rosy beneath a
cap trimmed with fur, enlivened by the cold, by being late, by her
anxiety for a game; shortly before she reached me, she slipped on a
piece of ice and, either to regain her balance, or because it appeared
to her graceful, or else pretending that she was on skates, it was
with outstretched arms that she smilingly advanced, as though to
embrace me. "Bravo! bravo! that's splendid; 'topping,' 1 should
say, like you — 'sporting,' I suppose 1 ought to say, only I'm a
hundred-and-one, a woman of the old school," exclaimed the lady,
uttering, on behalf of the voiceless Champs-Elysees, their thanks
to Gilberte for having come, without letting herself be frightened
away by the weather. "You are like me, faithful at all costs to our
old Champs-Elysees; we are two brave souls! You wouldn't be-
lieve me, I dare say, if I told you that 1 love them, even like this.
This snow (I know, you'll laugh at me), it makes me think of
ermine!" And the old lady began to laugh herself.

Francoise avail trop froid pour rester immobile, nous allames jus-
qu'au pont de la Concorde voir la Seine prise, dont chacun, et
meme les enfants s'approchaient sans peur comme d'une immense
baleine echouee, sans defense, et qu'on allait depecer. Nous reven-
ions aux Champs-Elysees; je languissais de douleur entre les
chevaux de bois immobiles et la pelouse blanche prise dans le
reseau noir des allees dont on avait enleve la neige et sur laquelle la
statue avait a la main un jet de glace ajoute qui semblait l'explica-
tion de son geste. La vieille dame elle-meme ayant plie ses Debats
demanda l'heure a une bonne d'enfants qui passait et qu'elle remer-
cia en lui disant: "Comme vous etes aimable!" puis priant le can-
tonnier de dire a ses petits enfants de revenir, qu'elle avait froid,
ajouta: "Vous serez mille fois bon. Vous savez que je suis confuse!"
Tout a coup l'air se dechirait: entre le guignol et le cirque, a l'hori-
zon embelli, sur le del entrouvert, je venais d'apercevoir, comme
un signe fabuleux, le plumet bleu de Mademoiselle. Et deja Gil-
berte courait a toute vitesse dans ma direction, etincelante et rouge
sous un bonnet carre de fourrure, animee par le froid, le retard et
le desir du jeu; un peu avant d'arriver a moi, elle se laissa glisser
sur la glace et, soit pour mieux garder son equilibre, soit parce
qu'elle trouvait cela plus gracieux, ou par affectation du maintien
d'une patineuse, c'est les bras grands ouverts qu'elle avancait en
souriant, comme si elle avait voulu m'y recevoir. "Brava! Brava! ca
c'est tres bien, je dirais comme vous que c'est chic, que c'est crane,



si je n'etais pas d'un autre temps, du temps de l'ancien regime,
s'ecria la vieille dame prenant la parole au nom des Champs-
Elysees silencieux pour remercier Gilberte d'etre venue sans se
laisser intimider par le temps. Vous etes comme moi, fidele quand
meme a nos vieux Champs-Elysees; nous sommes deux in-
trepides. Si je vous disais que je les aime meme ainsi. Cette neige,
vous allez rire de moi, fa me fait penser a de l'hermine!" Et la
vieille dame se mit a rire.

Let us agree that in this "state" the text corresponds fairly well
to Vigneron's harsh description of it: iterative and singulative
forms are entangled in a way that leaves the verbal aspect in
utter irresolution. But this ambiguity does not thereby justify the
explanatory hypothesis of an "incomplete temporal transposi-
tion." I believe that I even glimpse a presumption, at least, to
the contrary.

Indeed, if we examine more carefully the verbal forms
italicized here, we notice that all the imperfects except one can
be interpreted as imperfects of concomitance, which means that
the whole of the piece may be defined as singulative, with all the
verbs that strictly describe events, except one, being in the passe
defini: we walked [allames], the old lady asked [demanda; remercia,
ajouta], Gilberte slipped [se laissa glisser], the old lady exclaimed,
began to laugh [s'ecria, se mit a rire]. "Except one/' I said, which
is obviously, "Suddenly, the sky was rent in two [Tout a coup le
ciel se dechirait]". The very presence of the adverb suddenly pre-
vents this imperfect from being read as durative and requires it
to be interpreted as iterative. It alone jars in an irreducible way
in a context interpreted as singulative, and thus it alone intro-
duces into the text that "inextricable confusion" Vigneron
speaks of. 68 Now it happens that that form is corrected in the
1917 edition, which gives the expected form: "l'air se dechira."
That correction, it seems to me, is enough to pull this paragraph

6 We can also, to tell the truth, hesitate at "We returned to the Champs-
Elysees [nous revenions aux Champs-Elysees]," which does not easily reduce to
an imperfect of concomitance, since the events that it would go with are a little
subsequent to it ("The old lady . . . asked ... the time [la vieille dame ... de-
manda l'heurel"). But contamination by the context can sufficiently explain its


Narrative Discourse

from "confusion" and to push it entirely under the temporal
aspect of the singulative. Vigneron's description does not apply,
therefore, to the definitive text of Swann, the last to appear in
the author's lifetime. And as to explaining the problem by call-
ing it an "incomplete transposition" of the singulative into the
iterative, we see that that one correction goes exactly in the
opposite direction: far from "completing" in 1917 the
"whitewashing with imperfects" of a text in which in 1913 he
would carelessly have left too many passes simples, Proust 59 on
the contrary brings over to the singulative the only undeniably
iterative form on that page. Vigneron's interpretation, already
fragile, thereupon becomes untenable.

We are talking only, I hasten to make clear, about the cir-
cumstantial explanation Vigneron quite uselessly sought for the
confusions at the end of Swann, as if all the rest of the Proustian
narrative were a model of coherence and clarity. The same critic,
however, has rightly noticed elsewhere 70 the wholly retrospect-
ive unity imposed by Proust on "heteroclite" materials, and has
described the entire Recherche as a "Harlequin's cloak whose
multiple pieces, however rich the fabric may be, however indus-
triously they may have been brought together, recut, adjusted
and stitched, still betray, by differences in texture and color,
their diverse origins." 71 That is undeniable, and the subse-
quent publication of the various "first versions" has done, and
very probably will do, nothing but confirm that intuition. There
is some "collage," or rather some "patchwork," in the Re-
cherche, and its unity as narrative is indeed — like, according to
Proust, the unity of the Comedie humaine or of the Ring of the

" 9 Or perhaps somebody else: relying on a 1919 letter, Clarac and Ferre write,
"It seems therefore that Proust may not have supervised the new edition of
Swann which came out in 1917" (P I, xxi). But this uncertainty does not remove
all authority from the correction, which, moreover, Clarac and Ferre themselves
adopt. Besides, Proust cannot be totally unconnected with the variants of 1917: it
certainly must have been he who requested the corrections shifting Combray,
for the reasons we know, from Beauce to Champagne.

70 Vigneron, "Structure de Swann: Combray ou le cercle parfait," MP, 45
(February 1948), 185-207.

71 Vigneron, "Structure de Swann: Balzac, Wagner et Proust," French Review,
19 (May 1946), 384.



Nibelung—a unity after the event, claimed all the more keenly
because it is constructed later and more laboriously with mate-
rials from every source and from every period. We know that
Proust, far from considering this type of unity as "illusory"
(Vigneron), judged it "not fictitious, perhaps indeed all the
more real for being ulterior, for being born of a moment of
enthusiasm when it is discovered to exist among fragments
which need only to be joined together. A unity that has been
unaware of itself, therefore vital and not logical, that has not
banned variety, chilled execution." 72 We can only, it seems to
me, grant his basic point, perhaps adding, however, that he
underestimates here the fragments' resistance to being "joined
together." It is no doubt this resistance that the chaotic (accord-
ing to the norms of classical narration) episode of the Champs-
Elysees (among others) bears the trace of, more than of a rushed
publication. We can be convinced of this by comparing the pas-
sage in question here with two of its earlier versions: that in Jean
Santeuil, which is purely singulative, and that in Contre Sainte-
Beuve, which is completely iterative. 73 Proust, at the moment of
connecting the pieces together for the last version, could have
hesitated to choose, and could have finally decided, consciously
or not, on the absence of choice.

Whatever the cause, the most relevant hypothesis as to how
this should be read continues to be that this passage is com-
posed of an iterative beginning (the first paragraph) and a sin-
gulative continuation (the second, which we have just exam-
ined, and the third, whose temporal aspect has no ambiguity):
this would be banal if the temporal standing of this singulative
in relation to the preceding iterative were indicated, if only by a

72 RH II, 491/P III, 161. Cf. Marcel Proust on Art: "Some portions of [Balzac's]
great sequences were not linked up . . . till afterwards. What does that matter?
Wagner had composed The Good Friday Music before he thought of writing
Parsifal, and put it into the opera later on. But the additions Balzac made, these
lovely things that are brought in, the new relationships suddenly perceived by
his genius between separate parts of his work, which rejoin each other, come to
life, are henceforth inseparable, are they not his finest creative intuitions 7 " (p
182). F

73 /ran Santeuil, trans. Hopkins, pp. 49-52; Contre Sainte-Beuve, ed. Fallois, p.
Ill (Marcel Proust on Art, pp. 75-76).


Narrative Discourse

"once" that would isolate it in the series to which it belongs. 74
But nothing of the kind: the narrative passes without warning
from a habit to a singular event as if, instead of the event being
placed somewhere within or in connection to the habit, the habit
could become, indeed could be at the same time, a singular
event — which is, strictly speaking, inconceivable, and is thus, in
the Proustian text as it is, an occasion of irreducible unrealism.
There are others, of the same kind. For example, at the end of
Sodome et Gomorrhe, the account of M. de Charlus's trips in the
little train of La Raspeliere and of his relations with the other
followers begins in a very precisely specified iterative ("Regu-
larly, thrice weekly, . . . "), is then restricted by internal deter-
mination ("the very first times . . . "), and goes on for two pages
in an indeterminate singulative ("[Cottard] said [dit], either from
malice. . . "). 7S We see that here simply amending the iterative
plural ("the very first times") to a singular ("the very first time")
would be enough for everything to become orderly again. But
anyone who ventured to plunge into that course would have a
little more trouble with "Teaser Augustus," which is iterative on
pages 1049-1051 [RH I/P II, 464^66] but abruptly becomes sin-
gulative in the middle of page 1051 and continues so to the end
of the episode. And would have more trouble still, with the
narrative of the dinner at Rivebelle, in the Jeunes Filles en fleurs,
which is inextricably both a synthetic dinner, told in the imper-
fect ("At first, when we arrived there [Les premiers temps,
quand nous y arrivions]"), and a singular dinner, told in the
passe defini ("I noticed one of these servants. ... A young, fair
[woman] . . . gazed at me [je remarquai un de ces servants. . . .
Une jeune fille blonde me regarda]"); and since it deals with the
evening of the first appearance of the young girls we can date it
accurately, but no temporal indication places it in relation to the

74 The third paragraph does carry such an indication: "The first of these days
..." (called by Vigneron a "laboured connection" but habitual in Proust; for
example, at the inn in Doncieres, RH 1, 784/P II, 98, where "the first day"
associates a singulative illustration with a beginning of an iterative scene). But
this indication is not good retroactively for the second paragraph, whose inde-
terminateness it simply heightens through contrast.

75 RH II, 310-312/P II, 1037-1040. [Translator's note: partly my translation.]



series it belongs to and in which it gives the impression— a
rather disconcerting one — of floating. 76

Most often, actually, these points of contact between iterative
and singulative, with no assignable temporal relationship, are,
deliberately or not, masked by the interposition of neutral sec-
tions whose aspect is indeterminate, whose function, as Hous-
ton observes, seems to be to prevent the reader from noticing
the change of aspect. 77 These neutral sections can be of three
kinds: they may be discursive excursuses in the present tense,
such as a fairly long one in the transition between the iterative
beginning and the singulative continuation of La Prisonniere; 78
but this kind obviously has extranarrative status. It is otherwise
for the second type, accurately noted by Houston, which is dia-
logue (possibly restricted to a single rejoinder) without a declara-
tive verb; 79 the example Houston cites is the conversation be-
tween Marcel and the Duchess about the dress she wore to the
Sainte-Euverte dinner. 80 By definition, abruptive dialogue has
no determination of aspect, since it is deprived of verbs. The
third type is more subtle, for here the neutral section is in fact a
mixed or, more exactly, an ambiguous section: it consists of
interposing between iterative and singulative some imperfects
whose aspectual value remains indeterminate. Here is an exam-
ple taken from Un amour de Swann: we are first in the singulative;
Odette one day asks Swann for money to go without him to
Bayreuth with the Verdurins;

Of him she said not a word; it was to be taken for granted that their
presence at Bayreuth would be a bar to his [singulative descriptive

76 RH 1, 609-61 9/P 1, 808-822.

77 Houston, pp. 35-36.

78 RH 11, 434-^135/P 111, 82-83.

79 It is what Fontanier calls abruption: "The figure by which one removes the
customary transitions between the parts of a dialogue, or before direct speech, in
order to make its presentation more animated and more interesting" (Les Figures
du discours [1821-1827; Paris: Flammarion, 1968], pp. 342-343).

80 RH II, 403/P III, 37. The singulative section introduced here comes to an end
further on (RH II, 408/P III, 43) with a new abruptive dialogue.


Narrative Discourse

Then that annihilating answer, every word of which he had
carefully rehearsed overnight, without venturing to hope that it
could ever be used [ambiguous pluperfect], he had the satisfaction
of having it conveyed to her [iterative imperfect J. 81

De lui, elle ne disait pas un mot, il etait sous-entendu que leur
presence excluait la sienne [singulative descriptive imperfects].

Alors cette terrible reponse dont il avait arrete chaque mot la
veille sans oser esperer qu'elle pourrait servir jamais [ambiguous
pluperfect], il avait la joie de la lui faire porter [iterative imperfect].

A transformation even more effective in its abruptness is the
return to the iterative that closes the singulative episode of the
trees at Hudimesnil, in the ]eunes Filles en fleurs:

And when, the road having forked and the carriage with it, I turned
my back on them and ceased to see them, while Mme. de
Villeparisis asked me what I was dreaming about, I was as wretched
as though I had just lost a friend, had died myself, had broken faith
with the dead, or had denied my God [singulative imperfects].

It was time to be thinking of home [ambiguous imperfectl. Mme.
de Villeparisis . . . told her coachman to take us back by the old
Balbec road [iterative imperfect]. 82

Quand, la voiture ayant bifurque, je leur tournai le dos et cessai de
les voir, tandis que Mme. de Villeparisis me demandait pourquoi
j'avais l'air reveur, j'eta's triste comme si je venais de perdre un
ami, de mourir a moi-meme, de renier un mort ou de meconnaitre
un dieu [singulative imperfects].

II fallait songer au retour [ambiguous imperfect]. Mme. de
Villeparisis . . . disait au cocher de prendre la vieille route de Balbec
[iterative imperfect |.

More drawn out, by contrast, but extraordinarily skillful in
maintaining its irresoluteness for twenty or so lines, is this tran-
sition in Un amour de Swann:

81 RH I, 231/PI, 301.

82 RH I, 545/PI, 719.



But she saw that his eyes remained fixed upon the things that he
did not know, and on that past era of their love, monotonous and
soothing in his memory because it was vague, and now rent, as
with a sword-wound, by the news of that minute on the Island in
the Bois, by moonlight, while he was dining with the Princesse des
Laumes. But he had so far acquired the habit of finding life inter-
esting — of marvelling at the strange discoveries that there were to
be made in it — that even while he was suffering so acutely that he
did not believe it possible to endure such agony for any length of
time, he was saying to himself: "Life is indeed astonishing, and
holds some fine surprises; it appears that vice is far more common
than one has been led to believe. Here is a woman in whom I had
absolute confidence, who looks so simple, so honest, who, in any
case, even allowing that her morals are not strict, seemed quite
normal and healthy in her tastes and inclinations. I receive a most
improbable accusation, I question her, and the little that she admits
reveals far more than I could ever have suspected." But he could not
confine himself to these detached observations. He sought to form
an exact estimate of the importance of what she had just told him,
so as to know whether he might conclude that she had done these
things often, and was likely to do them again. He repeated her words
to himself: "I knew quite well what she was after." "Two or three
times." "I've heard that tale before." But they did not reappear in his
memory unarmed; each of them held a knife with which it stabbed
him afresh. For a long time, like a sick man who cannot restrain him-
self from attempting, every minute, to make the movement that, he
knows, will hurt him, he kept on murmuring to himself. 83

Mais elle vit que ses yeux rcstaient fixes sur les choses qu'il ne
savait pas et sur ce passe de leur amour, monotone et doux dans sa
memoire parce qu'il etait vague, et que dechirait maintenant
comme une blessure cette minute dans l'ile du Bois, au clair de
lune, apres le diner chez la princesse des Laumes. Mais il avait
tellement pris l'habitude de trouver la vie interessante — d'admirer
les curieuses decouvertes qu'on peut y faire — que tout en souffrant
au point de croire qu'il ne pourrait pas supporter longtemps une
pareille douleur, il se disait: "La vie est vraiment etonnante et
reserve de belles surprises; en somme le vice est quelque chose de

1 RH I, 281/P I, 366-367.


Narrative Discourse

plus repandu qu'on ne le croit. Voila une femme en qui j'avais
confiance, qui a l'air si simple, si honnete, en tout cas, si meme elle
etait legere, qui semblait bien normale et saine dans ses gouts: sur
une denonciation invraisemblable, je l'interroge, et le peu qu'elle
m'avoue revele bien plus que ce qu'on eut pu soupgonner." Mais il
ne pouvait pas se borner a ces remarques desinteressees. II cher-
chait a apprecier exactement la valeur de ce qu'elle lui avait raconte,
afin de savoir s'il devait conclure que ces choses, elle les avait jaites
souvent, qu'elles se renouvelleraient . 11 se repi'tait ces mots qu'elle
avait dits: "Je voyais bien ou elle voulait en venir," "Deux ou trois
fois," "Cette blague!," mais ils ne reparaissaien t pas desarmes dans
la memoire de Swann, chacun d'eux tenait son couteau et lui en
portait un nouveau coup. Pendant bien longtemps, comme un
malade ne peut s'empecher d'essayer a toute minute de faire le
mouvement qui lui est douloureux, il se redisait ces mots.

We see that the transformation is truly reached, unequivo-
cally, only starting with "For a long time [Pendant bien
longtemps]," which assigns a clearly iterative value to the im-
perfect "he kept on murmuring to himself [il se redisait ces
mots]" — an iterative value that the whole of the following pas-
sage will keep. Apropos of a transition of this kind (but more
elaborated — more than five pages — and, to tell the truth, less
pure, since it also includes several paragraphs of reflections in
the narrator's present and a brief interior monologue of the
hero) — the transition in La Prisonniere separating and joining the
narrative of an "ideal" Parisian day and the account of a certain
real day in February 84 — J. P. Houston rightly evokes "those
Wagnerian scores where the tonality shifts continuously with-
out any change in the key signature." 85 Proust knew, indeed,
how to exploit with great harmonic subtlety the capacities for
modulation which the ambiguity of the French imperfect tense
admits, as if he had wished, before mentioning it explicitly ap-
ropos of Vinteuil, to fashion almost a poetic equivalent of the
chromaticism of Tristan.

All of that, we imagine, cannot be simply the result of material
contingencies. Even if we must make (considerable) allowances

84 RH II, 434-438/P III, 81-

85 Houston, p. 37.



for external circumstances, there undoubtedly remains in Proust,
at work in such pages just as we have already met it else-
where, a sort of undertow of will — perhaps scarcely conscious

to liberate the forms of narrative temporality from their dramatic
function, to let them play for their own sake, and (as he says
apropos of Flaubert) to treat them in terms of music. 86

The Game with Time

We still have a word to say on the category of narrative time as
a whole, with respect to the general structure of the Recherche
and with respect to the place that work has in the evolution of
novelistic forms. More than once we were able to observe, in-
deed, the tight actual solidarity of various phenomena that we
had to separate for purposes of exposition. For example, in tra-
ditional narrative, analepsis (an aspect of sequence) most often
takes the form of summary (an aspect of duration, or of speed);
summary frequently has recourse to the services of the iterative
(an aspect of frequency); description is almost always at the same
time pinpointed, durative, and iterative, without ever forbid-
ding itself the beginnings of diachronic movement — and we
have seen how in Proust this tendency goes so far as to reabsorb
description into narrative; there exist frequentative forms of el-
lipsis (for example, all Marcel's Parisian winters during the
period of Combray); the iterative syllepsis is not only an aspect
of frequency: it also affects sequence (since by synthesizing
"similar" events it abolishes their succession) and duration
(since at the same time it eliminates their time intervals); and we
could extend this list further. So we can characterize the tem-
poral stance of a narrative only by considering at the same time
all the relationships it establishes between its own temporality
and that of the story it tells.

We observed in the chapter on sequence that the main anach-
ronies of the Recherche all come at the beginning of the work,

86 "With Balzac, the change of tempo has an active and documentary charac-
ter. Flaubert was the first novelist to free this change from all parasitic growths of
historical scavenging. He treated it in terms of music. Nobody before him had
ever done that" (Essais et articles, Pleiade, p. 595; "About Flaubert's Style," in
Proust: A Selection, p. 235).


Narrative Discourse

chiefly in Du cote de chez Swann, where we saw the narrative
begin as if with difficulty, hesitatingly, interrupted by incessant
backs-and-forths between the remembering position of the "in-
termediary subject" and various diegetic positions, which were
sometimes redoubled (Combray I and Combray 11), until, at Bal-
bcc, the narrative concluded a sort of general agreement with
chronological succession. We cannot miss connecting that as-
pect of sequence with an aspect of frequency just as unmistak-
able: the dominance of the iterative in this same section of text.
The initial narrative sections are mainly iterative plateaux
(childhood in Combray, Swann's love, Gilberte) that occur to
the mind of the intermediary subject — and, through him, to the
narrator — like so many almost motionless moments when the
passage of time is masked behind repetition. The anachronism
of the memories ("voluntary" or not) and their static nature are
obviously in league with each other in that they both arise from
the work of memory, which reduces (diachronic) periods to
(synchronic) epochs and events to pictures — epochs and pic-
tures that memory arranges in an order not theirs, but its own.
The remembering activity of the intermediary subject is thus a
factor in (I should rather say a means of) the emancipation of the
narrative with respect to diegetic temporality on the two con-
nected planes of simple anachronism and iteration, which is a
more complex anachronism. But starting with Balbec, and espe-
cially with Guermantes, the simultaneous restoration of
chronological order and the dominance of the singulative —
plainly associated with the gradual disappearance of the re-
membering instance and thus with the emancipation, this time,
of the story, which regains its hold over the narrative 87 — brings
us back to apparently more traditional paths, and one might
prefer the subtle temporal "confusion" of Swann to the sobered
arrangement of the Balbec- Guermantes- Sodome series. But with
that series the distortions of duration take over, subjecting a
temporality whose rights and norms have apparently been re-

87 It is in fact as if the narrative, caught between what it tells (the story) and
what tells it (the narrating, led here by memory), had no choice except domina-
tion by the former (classical narrative) or domination by the latter (modern
narrative, inaugurated with Proust); but we will bring this point up again in the
chapter on voice.



established to a distorting operation (enormous ellipses, mon-
strous scenes) that no longer comes from the intermediary subject
but directly from the narrator — who in his impatience and grow-
ing anguish is desirous both of loading his final scenes, like Noah
his ark, to the bursting point, and of jumping to the denoue-
ment (for it is one) that will finally give him being and legitimate
his discourse. This is to say that there we touch on another
temporality, no longer the temporality of the narrative but in the
final instance governing it: the temporality of the narrating it-
self. We will meet it again below. 88

Interpolations, distortions, temporal condensations 89 —
Proust, at least when he is aware of them (he seems never, for
example, to have perceived the importance for him of the itera-
tive narrative), justifies them constantly (according to an already
old tradition that will not die with him) by a realistic motivation:
he invokes in turn the concern to tell things as they were "lived"
at the time and the concern to tell them as they were recalled
after the event. Thus, the anachronism of the narrative is now
that of existence itself, 90 now that of memory, which obeys other
laws than those of time. 91 The variations in tempo, likewise, are

"Chapter 5. We may deplore this quartering of the problems of narrative
temporality, but any other distribution would have the effect of underestimating
the importance and the specificity of the narrating instance. In matters of "writ-
ing" our only choice is between drawbacks.

89 Here these three terms obviously designate the three main kinds of tem-
poral "distortion," according to whether they affect order, duration, or fre-
quency. The iterative syllepsis condenses several events into a single narrative;
the alternation scenes/ellipses distorts duration; finally, let us recall that Proust
himself named "interpolations" the anachronies he admired in Balzac: "Make
plain Balzac's . . . interpolation of passages of time, like geological formations
where lava from different epochs lies intermingled (La Duchesse de Langeais,
Sarrazine)" (Marcel Proust on Art, p. 180).

90 "For often we find a day, in [a season], that has strayed from another
season, and makes us live in that other ... by inserting, out of its turn, too early
or too late, this leaf, torn from another chapter, in the interpolated calendar of
Happiness" (RH I, 295/P I, 386-387); "So it is that the different periods of our life
overlap one another" (RH I, 476/P I, 626); "... our life being so careless of
chronology, interpolating so many anachronisms in the sequence of our davs"
(RH I, 488/P I, 642).

1)1 "As our memory presents things to us, as a rule, not in their chronological
sequence but as it were by a reflexion in which the order of the parts is reversed"
(RH I, 440/P I, 578).


Narrative Discourse

now the doing of "life," 92 now the work of memory, or rather of
forgetfulness. 93

These contradictions and compliancies would dissuade us, if
there were any need to, from granting too much credit to those
retrospective rationalizations that great artists are never nig-
gardly with, and this in direct proportion to their genius, in other
words, to the lead their practice has over any theory — in-
cluding their own. The role of the analyst is not to be satisfied
with the rationalizations, nor to be ignorant of them, but rather,
having "laid bare" the technique, to see how the motivation that
has been invoked functions in the work as aesthetic medium. We
would thus readily say, in the manner of the early Shklovsky,
that in Proust, for example, "reminiscence" is at the service of
metaphor and not the reverse; that the intermediary subject's
selective amnesia is there so that the narrative of childhood may
open with the "drama of going to bed"; that the "jog-trot" of
Combray serves to trigger the horizontal escalator of iterative

92 "In our life the days are not all equal. To reach the end of a day, natures that
are slightly nervous, as mine was, make use, like motor-cars, of different
'speeds.' There are mountainous, uncomfortable days, up which one takes an
infinite time to pass, and days downward sloping, through which one can go at
full tilt, singing as one goes" (RH 1, 298/P 1, 390-391); "The time which we have
at our disposal every day is elastic; the passions that we feel expand it, those that
we inspire contract it; and habit fills up what remains" (RH 1, 465/P 1, 612).

93 "Oblivion does not fail to alter profoundly our notion of time. There are
optical errors in time as there are in space. . . . This oblivion of so many things . . .
by its fragmentary, irregular interpolation in my memory... confused, de-
stroyed my sense of distances in time, contracted in one place, extended in
another, and made me suppose myself now farther away from things, now far
closer to them than I really was" (RH II, 799-800/P III, 593-594). Throughout this
we are dealing with time as it is lived or remembered "subjectively," with the
"optical illusions of which our first sight ... is composed," and of which Proust,
like Elstir, wants to be the faithful interpreter. But we see him for that matter
justify his ellipses, for example, by the concern to make perceptible to the reader
a flight of time which "life," ordinarily, screens from us, and of which we have
only a knowledge acquired from books: "In theory one is aware that the earth
revolves, but in practice one does not perceive it, the ground upon which one
treads seems not to move, and one can live undisturbed. So it is with Time in
one's life. And to make its flight perceptible novelists are obliged, by wildly
accelerating the beat of the pendulum, to transport the reader in a couple of
minutes over ten, or twenty, or even thirty years" (RH I, 369/P I, 482). We see
that realistic motivation adapts itself equally to subjectivism and scientific objec-
tivity: sometimes I distort to show things as they are illusively experienced,
sometimes I distort to show things as they really are, which experience conceals
from us.



imperfects; that the hero makes two stays in a clinic to provide
the narrator with two fine ellipses; that the little madeleine has
broad shoulders. And Proust himself said it clearly at least once:

Leaving aside, for the moment, all question of the value I attach to
such unconscious memories, on which, in the final volume ... I
base my whole theory of art, let me concentrate attention on the
purely compositional aspect of the matter, and point out that, in
order to pass from one plane to another, I make use, not of "fact,"
but of something in which I find a greater degree of purity and
significance, as a link — namely, a phenomenon of memory. Now,
open the Memoires d'Outre-Tombe, or Gerard de Nerval's Filles du
Feu, and you will find that two great writers, whom it is the
fashion to impoverish and devitalise by applying to them an over-
formal interpretation, were perfectly familiar with this method of
sudden transition. 94

Involuntary memory, ecstasy of the intemporal, contemplation
of eternity? Perhaps. But also, when we concentrate on the
"purely compositional aspect of the matter," significant link and
method of transition. And let us relish in passing, in this
craftsman's 95 confession, that strange repentance about the
writers "whom it is the fashion to impoverish and devitalise by
applying to them an over-formal interpretation." That is one
stone that falls back into its own garden, but it has not yet been
shown how "over-formal" interpretation impoverishes and de-
vitalizes. Or rather, Proust himself proved the contrary by point-
ing out, for example about Flaubert, how a particular use "of the
past definite, the past indefinite, the present participle, and of
certain pronouns and prepositions, has renewed our vision of
things almost to the same extent as Kant, with his Categories,
renewed our theories of knowledge and of the reality of the
external world." 96 To put it another way, and to parody Proust's
own formula, vision can also be a matter of style and of technique.

94 Essais et articles, Pleiade, p. 599; "About Flaubert's Style," in Proust: A
Selection, p. 239.

95 It is apropos of Wagner that Proust speaks of the "delight of the craftsman"
(RH II, 491/P III, 161).

96 Essais et articles, Pleiade, p. 586; "About Flaubert's Style," in Proust: A
Selection, p. 224.


Narrative Discourse

We know with what ambiguity— to all appearances
unbearable— the Proustian hero devotes himself to the search
for and the "adoration" of both the "extra-temporal" and "time
in its pure state"; how he wants himself, and with him his future
work, to be both together "outside time" and "in Time." What-
ever the key to this ontological mystery may be, perhaps we see
better now how this contradictory aim functions in and takes
possession of Proust's work: interpolations, distortions,
condensations— the Proustian novel is undoubtedly, as it pro-
claims, a novel of Time lost and found again, but it is also, more
secretly perhaps, a novel of Time ruled, captured, bewitched,
surreptitiously subverted, or better: perverted. Apropos of this
novel, how could we not speak— as its author does about dream-
ing (and perhaps not without some ulterior motive of
connection)— of the "formidable game it creates with Time"? 97

"P III 912 ITranslator's note: my translation; the RH translation is on II,
1032 I In'passing let us emphasize the verb used here: "create (and not: play) a
game with Time" is not only to play with Time, it is also to make a game of it. But a
"formidable" game. In other words, also a dangerous one.

4 Mood

Narrative Moods?

If the grammatical category of tense clearly applies to the
stance of narrative discourse, that of mood might seem a priori
to be irrelevant here. Since the function of narrative is not to
give an order, express a wish, state a condition, etc., but simply
to tell a story and therefore to "report" facts (real or fictive), its
one mood, or at least its characteristic mood, strictly speaking
can be only the indicative — and at that point we have said every-
thing there is to say on this subject, unless we stretch the lin-
guistic metaphor a little more than is fitting.

Without denying the metaphoric extension (and therefore the
distortion), we can meet the objection by saying that there are
not only differences between affirming, commanding, wishing,
etc., but there are also differences between degrees of affirma-
tion; and that these differences are ordinarily expressed by
modal variations, be they the infinitive and subjunctive of indi-
rect discourse in Latin, or, in French, the conditional that indi-
cates information not confirmed. This obviously is the function
the Littre dictionary is referring to when it defines the gram-
matical meaning of mood: "name given to the different forms
of the verb that are used to affirm more or less the thing in
question, and to express . . . the different points of view from
which the life or the action is looked at," and this definition on
good authority is very valuable to us here. Indeed, one can tell
more or tell less what one tells, and can tell it according to one point



Narrative Discourse

of view or another; and this capacity, and the modalities of its use,
are precisely what our category of narrative mood aims at. Narra-
tive "representation," or, more exactly, narrative information,
has its degrees: the narrative can furnish the reader with more or
fewer details, and in a more or less direct way, and can thus
seem (to adopt a common and convenient spatial metaphor,
which is not to be taken literally) to keep at a greater or lesser
distance from what it tells. The narrative can also choose to regu-
late the information it delivers, not with a sort of even screening,
but according to the capacities of knowledge of one or another
participant in the story (a character or group of characters), with
the narrative adopting or seeming to adopt what we ordinarily
call the participant's "vision" or "point of view"; the narrative
seems in that case (continuing the spatial metaphor) to take on,
with regard to the story, one or another perspective. "Distance"
and "perspective," thus provisionally designated and defined,
are the two chief modalities of that regulation of narrative informa-
tion that is mood— as the view I have of a picture depends for
precision on the distance separating me from it, and for breadth
on my position with respect to whatever partial obstruction is
more or less blocking it.


This problem was addressed for the first time, it seems, by
Plato, in Book III of The Republic. 1 As we know, Plato contrasts
two narrative modes, according to whether the poet "himself is
the speaker and does not even attempt to suggest to us that
anyone but himself is speaking" (this is what Plato calls pure
narrative), 2 or whether, on the other hand, the poet "delivers a
speech as if he were someone else" (as if he were such-and-such
a character), if we are dealing with spoken words (this is what
Plato properly calls imitation, or mimesis). And to really exhibit

1 Plato, Republic, 392 c to 395. [Translator's note: all quotations in this chapter
from the Republic are from the translation of Paul Shorey, Cambridge, Mass.:
Loeb Classical Library, 1937.] Cf. my Figures //, pp. 50-56.

2 The common translation of haple diegesis as "simple narrative" seems to me a
little off the mark. Haple diegesis is the narrative not mixed (in 397 b, Plato says:
akraton) with mimetic elements: therefore, pure.



the difference, Plato goes so far as to rewrite as diegesis the end
of the scene between Chryses and the Achaeans, a scene which
Homer had treated as mimesis, that is, as direct speech in the
manner of drama. The scene in direct dialogue then becomes a
narrative mediated by the narrator, where the "replies" of the
characters are dissolved and condensed into indirect discourse.
Indirection and condensation — later we will again meet these
two distinctive features of "pure narrative," in contrast to
"mimetic" representation borrowed from the theatre. In these
terms, adopted provisionally, "pure narrative" will be taken to
be more distant than "imitation": it says less, and in a more
mediated way.

We know how this contrast — somewhat neutralized by Aris-
totle (who makes pure narrative and direct representation two
varieties of mimesis), 3 and (for that very reason?) neglected by
the classical tradition (which in any case paid little attention to
the problems of narrative discourse) — abruptly surged forth
again in novel theory in the United States and England at the
end of the nineteenth century and the beginning of the twen-
tieth, with Henry James and his disciples, in the barely trans-
posed terms of showing vs. telling, which speedily became the
Ormazd and the Ahriman 4 of novelistic aesthetics in the
Anglo-American normative vulgate. 5 From this normative point
of view, Wayne Booth, throughout his Rhetoric of Fiction, deci-
sively criticized that neo- Aristotelian valuing of the mimetic. 6
From our own strictly analytic point of view it must be added (as
Booth's discussion, moreover, reveals in passing) that the very
idea of showing, like that of imitation or narrative representation
(and even more so, because of its naively visual character), is

3 Aristotle, Poetics, 1448 a.

4 [Translator's note.] The Zoroastrian good and evil principles, respectively;
the first created and governs the world, while the second seeks to destroy the
other's beneficent work.

5 See in particular Percy Lubbock, The Craft of Fiction. For Lubbock, "The art of
fiction does not begin until the novelist thinks of his story as a matter to be
shown, to be so exhibited that it will tell itself" (p. 62).

h Wayne C. Booth, The Rhetoric of Fiction (Chicago, 1961). Let us note that,
paradoxically, Booth belongs to the neo-Aristotelian school of the "Chicago


Narrative Discourse

completely illusory: in contrast to dramatic representation, no
narrative can "show" or "imitate" the story it tells. All it can do
is tell it in a manner which is detailed, precise, "alive," and in
that way give more or less the illusion of mimesis— which is the
only narrative mimesis, for this single and sufficient reason: that
narration, oral or written, is a fact of language, and language
signifies without imitating.

Unless, of course, the object signified (narrated) be itself lan-
guage We observed just above, when we recalled the Platonic
definition of mimesis (the poet can deliver a speech as someone
else) that then we are dealing with spoken words. But what
happens when we are dealing with something else: not words,
but silent events and actions? How then does mimesis function,
and how will the narrator "suggest to us that ... he lis] someone
else"? (I do not say the poet, or the author: whether the narra-
tive be taken charge of by Homer or by Ulysses is simply to
transfer the problem.) How can one handle the narrative object
so that it literally "tells itself" (as Lubbock insists) without any-
one having to speak for it? Plato knows better than to answer
this question, and even than to ask it, as if his exercise in rewrit-
ing bore only on speech, and-for the opposition between
diegesis and mimesis-contrasted only two kinds of dialogue,
dialogue in indirect style and dialogue in direct style. The truth
is that mimesis in words can only be mimesis of words. Other
than that, all we have and can have is degrees of diegesis. So we
must distinguish here between narrative of events and "narra-
tive of words."

Narrative of Events

The Homeric "imitation" of which Plato offers a translation
into "pure narrative" includes only a brief section that is not in
dialogue. Here it is in its original version: "So said he, and the
old man was afraid and obeyed his word, and fared silently
along the shore of the loud-sounding sea. Then went that aged
man apart and prayed aloud to king Apollo, whom Leto of the
fair locks bare." 7 Here it is in its Platonic rewriting: "And the old

'Iliad, I, 11.33-36, trans. Lang, Leaf, and Myers.



man on hearing this was frightened and departed in silence, and
having gone apart from the camp he prayed at length to

The most evident difference is obviously in length (eighteen
words to thirty in the Greek texts, twenty-six to forty-three in the
English translations). Plato achieves this condensation by
eliminating redundant information ("so said he," "obeyed,"
"whom Leto bare"), and also by eliminating circumstantial and
"picturesque" indicators: "of the fair locks," and especially
"along the shore of the loud-sounding sea." This shore of the
loud-sounding sea, a detail functionally useless in the story, is —
despite the stereotyped nature of the formula (which recurs sev-
eral times in the Iliad and the Odyssey), and beyond the enor-
mous differences in style between the Homeric epic and the
realistic novel — fairly typical of what Barthes calls a realistic ef-
fect. 8 The loud-sounding shore serves no purpose other than to
let us understand that the narrative mentions it only because it
is there, and because the narrator, abdicating his function of
choosing and directing the narrative, allows himself to be gov-
erned by "reality," by the presence of what is there and what
demands to be "shown." A useless and contingent detail, it is
the medium par excellence of the referential illusion, and there-
fore of the mimetic effect: it is a connotator of mimesis. So Plato,
with a sure hand, suppresses it in his translation as a feature
incompatible with pure narrative.

The narrative of events, however, whatever its mode, is al-
ways narrative, that is, a transcription of the (supposed) non-
verbal into the verbal. Its mimesis will thus never be anything
more than an illusion of mimesis, depending like every illusion
on a highly variable relationship between the sender and the
receiver. It goes without saying, for example, that the same text
can be received by one reader as intensely mimetic and by
another as an only slightly "expressive" account. Historical
evolution plays a critical role here, and it is likely that the audi-
ence for the classics, which was so sensitive to Racinean "figura-
tion," found more mimesis than we do in the narrative style of a

"Barthes, "L'Effet de reel," Communications, 11 (1968), 84-89.


Narrative Discourse

d'Urfe or a Fenelon, but would undoubtedly have found the so
richly and minutely described accounts in the naturalistic novel
to be only chaotic proliferation and "murky mess," and would
thus have missed the mimetic function of those accounts. We
have to make allowance for this relationship, which varies ac-
cording to individuals, groups, and periods, and does not,
therefore, depend exclusively on the narrative text.

The strictly textual mimetic factors, it seems to me, come
down to those two data already implicitly present in Plato's
comments: the quantity of narrative information (a more de-
veloped or more detailed narrative) and the absence (or minimal
presence) of the informer— in other words, of the narrator.
"Showing" can be only a way of telling, and this way consists of
both saying about it as much as one can, and saying this "much"
as little as possible [en dire le plus possible, et ce plus, le dire le
moins possible]: speaking, Plato says, "as if the poet were
someone else"— in other words, making one forget that it is the
narrator telling. Whence these two cardinal precepts of showing:
the Jamesian dominance of scene (detailed narrative) and the
(pseudo-)Flaubertian transparency of the narrator (canonic
example: Hemingway's "The Killers," or "Hills Like White
Elephants"). Cardinal precepts and, above all, interrelated pre-
cepts: pretending to show is pretending to be silent. Finally,
therefore, we will have to mark the contrast between mimetic
and diegetic by a formula such as: information + informer = C,
which implies that the quantity of information and the presence
of the informer are in inverse ratio, mimesis being defined by a
maximum of information and a minimum of the informer,
diegesis by the opposite relationship.

As we see immediately, this definition, on the one hand,
sends us back to a temporal determination— narrative speed-
since it goes without saying that the quantity of information is
solidly in inverse ratio to the speed of the narrative; and on the
other hand it sends us to a datum of voice— the degree to which
the narrating instance is present. Mood here is simply a product
of features that do not belong to it in its own right, and so we
have no reason to linger over it— except to note this: that the
Recherche du temps perdu in itself constitutes a paradox — or a



contradiction — completely unassimilable by the mimetic
"norm" whose implicit formula we have just elucidated. In-
deed, as we saw in Chapter 2, Proustian narrative consists on
the one hand almost exclusively of "scenes" (singulative or
iterative), in other words, of a narrative form that is most rich in
information, and thus most "mimetic"; but on the other hand,
as we shall see more closely in the next chapter (and as the most
unsophisticated reading readily testifies), the narrator's pres-
ence is constant, and so intense as to be completely contrary to
the "Flaubertian" rule. The narrator is present as source,
guarantor, and organizer of the narrative, as analyst and com-
mentator, as stylist (as "writer," in Marcel Muller's vocabulary)
and particularly — as we well know — as producer of "meta-
phors." Proust then would be — like Balzac, like Dickens, like
Dostoevski, but in an even more pronounced and thus more
paradoxical way — simultaneously at the extreme of showing and
at the extreme of telling (and even a little further than that, in
this discourse sometimes so liberated from any concern with a
story to tell that it could perhaps more fittingly be called simply
talking). All this is both well known and impossible to demon-
strate without an exhaustive analysis of the text. As illustration,
I will content myself here with invoking once again the scene of
the bedtime in Combray, already quoted in Chapter l. 9 Nothing
is more intense than this vision of the father, "an immense
figure in his white nightshirt, crowned with the pink and violet
scarf of Indian cashmere in which ... he used to tie up his
head," candle in hand, with his fantastic reflection on the wall of
the staircase, and the child's sobs, so long suppressed, bursting
out when he is alone once more with his mother. But at the same
time nothing is more explicitly mediated, avouched as memory,
and as memory both very old and very recent, perceptible anew
after years of oblivion, now that "life is more quiet" around a
narrator on the threshold of death. It cannot be said that this
narrator here lets the story tell itself, and it would be too little to
say that he tells it without any care to efface himself before it:
what we are dealing with is not the story, but the story's "im-

'P. 70.


Narrative Discourse

age," its trace in a memory. But this trace, so delayed, so remote,
so indirect, is also the presence itself. In this mediated intensity is
a paradox which, quite obviously, is such only according to the
norms of mimetic theory: a decisive transgression, a rejection
pure and simple — as we watch — of the millennial opposition
between diegesis and mimesis.

We know that for post-Jamesian partisans of the mimetic
novel (and for James himself), the best narrative form is what
Norman Friedman calls "the story told as if by a character in the
story, but told in the third person" (a clumsy formula that evi-
dently refers to the focalized narrative, told by a narrator who is
not one of the characters but who adopts the point of view of
one). Thus, continues Friedman summarizing Lubbock, "the
reader perceives the action as it filters through the conscious-
ness of one of the characters involved, yet perceives it directly as
it impinges upon that consciousness, thus avoiding that removal
to a distance necessitated by retrospective first-person narra-
tion." 10 The Recherche du temps perdu, a narration doubly, some-
times triply, retrospective, does not, as we know, avoid that
distance; very much to the contrary, it maintains and cultivates
it. But the marvel of Proustian narrative (like that of Rousseau's
Confessions, which here again we must put side by side with it) is
that this temporal distance between the story and the narrating
instance involves no modal distance between the story and the
narrative: no loss, no weakening of the mimetic illusion. Ex-

10 Norman Friedman, "Point of View in Fiction," PMLA, 70 (1955); rpt. in
Philip Stevick, ed., The Theory of the Novel (New York, 1967), p. 113. This alleged
disability of the autobiographical novel is described more precisely by A. A.
Mendilow: "Contrary to what might be expected, a novel in the first person
rarely succeeds in conveying the illusion of presentness and immediacy. Far
from facilitating the hero-reader identification, it tends to appear remote in time.
The essence of such a novel is that it is retrospective, and that there is an avowed
temporal distance between the fictional time — that of the events as they
happened — and the narrator's actual time — his time of recording those events.
There is a vital difference between writing a story forward from the past, as in the
third person novel, and writing one backward from the present, as in the
first person novel. Though both are equally written in the past, in the former
the illusion is created that the action is taking place; in the latter, the action is felt
as having taken place" (Time and the Novel |New York, 1952], pp. 106-107).



treme mediation, and at the same time utmost immediacy. That
too is perhaps symbolized by the rapture of reminiscence.

Narrative of Words

If the verbal "imitation" of nonverbal events is simply a Utopia
of illusion, the "narrative of words" can, by contrast, seem con-
demned a priori to that absolute imitation which, as Socrates
demonstrated to Cratylus, would, if it truly presided over the
creation of words, make of language a reduplication of the
world: "Everything would be duplicated, and no one could tell
in any case which was the real thing and which the name." 11
When Marcel, on the last page of Sodome et Gomorrhe, declares to
his mother, "It is absolutely necessary that I marry Albertine,"
there is no difference between the statement present in the text
and the sentence purportedly spoken by the hero other than
what derives from the transition from oral language to written.
The narrator does not narrate the hero's sentence; one can
scarcely say he imitates it: he recopies it, and in this sense one
cannot speak here of narrative.

Yet that is indeed what Plato does when he imagines what the
dialogue between Chryses and Agamemnon would become if
Homer reported it "not as if made or being Chryses [and
Agamemnon], but still as Homer," since he adds right here: "It
would not be imitation but narration, pure and simple." It is
worth the trouble to return again to that strange rewriting, even
if the translation lets some nuances escape. Let us be satisfied
with a single portion, composed of Agamemnon's answer to
Chryses' supplications. Here is what this discourse was in the

"Let me not find thee, old man, amid the hollow ships, whether
tarrying now or returning again hereafter, lest the staff and fillet of
the god avail thee naught. And her will I not set free; nay, ere that
shall old age come on her in our house, in Argos, far from her
native land, where she shall ply the loom and serve my couch. But

11 Plato, Cratylus, 432 d, trans. H. N. Fowler (Cambridge, Mass.: Loeb Classi-
cal Library, 1926), p. 165.


Narrative Discourse

depart, provoke me not, that thou mayest the rather go in
peace." 12

Here is what it becomes in Plato:

Agamemnon was angry and bade him depart and not come again
lest the scepter and the fillets of the god should not avail him. And
ere his daughter should be released, he said, she would grow old
in Argos with himself, and he ordered him to be off and not vex
him if he wished to get home safe.

Here we have side by side two possible states of the discourse
of characters, which we shall provisionally describe in a rather
rough way: in Homer, an "imitated" discourse — that is, dis-
course fictively reported as it supposedly was uttered by the
character; in Plato, a "narratized" discourse— that is, discourse
treated like one event among others and taken on as such by the
narrator himself. In Plato's narratized discourse Agamemnon's
speech becomes an action, and nothing external distinguishes
between what comes from the answer Homer gives his hero (he
"bade him depart") and what is taken from the narrative lines
that precede (he "was angry") — in other words, nothing exter-
nal distinguishes between what was words in the original and
what was gesture, posture, state of mind. Without any doubt we
could push further the reduction of speech to event, for example
by writing once and for all: "Agamemnon refused and dis-
missed Chryses." There we would have the pure form of nar-
ratized speech. In Plato's text, the care to retain a few more
details has disturbed that purity by introducing into it elements
of a sort of intermediary degree, written in an indirect style,
more or less closely subordinated ("his daughter should [not] be
released, he said"; "lest the scepter. . . should not avail him").
For this intermediary degree we will reserve the name of trans-
posed speech. This tripartite division applies to "inner speech" as
well as to words actually uttered, a distinction, moreover, that is
not always relevant when we are dealing with a soliloquy. See,

2 lHad, I, 11.26-32, trans. Lang, Leaf, and Myers.



for example, that monologue — internal or external? — of Julien
Sorel receiving Mathilde's declaration of love, punctuated by
"Julien said to himself," "he cried," "he added": it would be
quite useless to wonder whether or not those expressions
should be taken literally. 13 The novelistic convention, perhaps
truthful in this case, is that thoughts and feelings are no dif-
ferent from speech, except when the narrator undertakes to
condense them into events and to relate them as such.

So we will distinguish these three states of characters' speech
(uttered or "inner"), connecting them to our present subject,
which is narrative "distance."

1. Narratized, or narrated, speech is obviously the most distant
and generally, as we have just seen, the most reduced. Let us
suppose that the hero of the Recherche, instead of reproducing
his dialogue with his mother, should simply write at the end of
Sodome: "I informed my mother of my decision to marry Alber-
tine." If we were dealing not with his words but his "thoughts,"
the statement could be even briefer and closer to pure event: "I
decided to marry Albertine." On the other hand, the narrative of
the inner debate leading to that decision, conducted by the nar-
rator in his own name, can be developed at greater length within
the form traditionally referred to by the term analysis, a form we
can consider to be like a narrative of thoughts, or narratized
inner speech.

2. Transposed speech, in indirect style: "I told my mother that I
absolutely had to marry Albertine" (uttered speech); "I thought
that I absolutely had to marry Albertine" (inner speech). Al-
though a little more mimetic than narrated speech, and in prin-
ciple capable of exhaustiveness, this form never gives the reader
any guarantee — or above all any feeling — of literal fidelity to the
words "really" uttered: the narrator's presence is still too per-
ceptible in the very syntax of the sentence for the speech to
impose itself with the documentary autonomy of a quotation. It

13 Stendhal, Le Rouge et le noir, Book II, chap. 13 (Gamier, p. 301). Similarly,
Mathilde, busy sketching in her album, "cried with rapture" (II, chap. 19; Gamier,
p. 355). Julien goes so far as to "reflect" with a Gascon accent: " 'It's a question of
honnur,' he said to himself" (II, chap. 15; Gamier, p. 333).


Narrative Discourse

is, so to speak, acknowledged in advance that the narrator is not
satisfied with transposing the words into subordinate clauses,
but that he condenses them, integrates them into his own
speech, and thus expresses them in his own style, like Francoise
representing the civilities of Mme. de Villeparisis. 14

It is not entirely the same with the variant known by the name
of "free indirect style," where economizing on subordination
allows a greater extension of the speech, and thus a beginning of
emancipation, despite the temporal transpositions. But the main
difference is the absence of a declarative verb, an absence which
can (unless the context provides indicators) involve a double
confusion. First, between uttered speech and inner speech. In a
statement such as, "I went to find my mother: it was absolutely
necessary that I marry Albertine," the second clause can express
equally well the thoughts Marcel has while seeking out his
mother or the words he addresses to her. Next and especially,
confusion between the speech (uttered or inner) of the character
and that of the narrator. Marguerite Lips quotes some striking
examples of this, 15 and we know the remarkable advantage
Flaubert derived from this ambiguity, which permits him to
make his own language speak this both loathsome and fascinat-
ing idiom of the "other" without being wholly compromised or
wholly innocent.

3. The most "mimetic" form is obviously that rejected by
Plato, where the narrator pretends literally to give the floor to
his character: "I said to my mother (or: I thought): it is absolutely
necessary that I marry Albertine." This reported speech, dra-
matic in type, has been adopted since Homer as the basic form
of dialogue (and of monologue) in the "mixed" 16 narrative first
of the epic and then of the novel; and Plato's appeal for the

14 " 'She said to me, "You'll be sure and bid them good day" ' " (RH I, 529/P I,
697). The paradox here is that the representation professes to be a literal quota-
tion, emphasized by a vocal imitation. But if Franchise had been satisfied with a
"She told me to bid you good day," she would be following the norm of indirect

15 Marguerite Lips, he Style indirect libre (Paris, 1926), pp. 57 ff.
"��Mixing diegesis and mimesis in the Platonic sense.



purely narrative was all the less effective since Aristotle lost no
time upholding, with the authority and success we know of, the
superiority of the purely mimetic. We should not fail to ap-
preciate the influence that this prerogative, massively granted to
dramatic style, exerted for centuries on the evolution of narra-
tive genres. It is expressed not only by the canonization of
tragedy as the supreme genre in the entire classical tradition, but
also, more subtly and well beyond classicism, in that sort of
tutelage exercised over narrative by the dramatic model, ex-
pressed so well by the use of the word "scene" to designate the
basic form of novelistic narration. Up to the end of the
nineteenth century, the novelistic scene is conceived, fairly pit-
eously, as a pale copy of the dramatic scene: mimesis at two
degrees, imitation of imitation.

Curiously, one of the main paths of emancipation of the
modern novel has consisted of pushing this mimesis of speech
to its extreme, or rather to its limit, obliterating the last traces of
the narrating instance and giving the floor to the character right
away. Let us imagine a narrative beginning (but without quota-
tion marks) with this sentence: "It is absolutely necessary that I
marry Albertine," and continuing thus up to the last page, ac-
cording to the order of the hero's thoughts, perceptions, and
actions performed or undergone. "The reader [would be] in-
stalled in the thought of the main character from the first lines
on, and it is the uninterrupted unfolding of that thought which,
substituting completely for the customary form of narrative,
[would] apprise us of what the character does and what hap-
pens to him." The reader has perhaps recognized this as Joyce's
description of Les Lauriers sont coupes by Edouard Dujardin 17 —
as, in other words, the most exact definition of what has been
quite unfortunately christened "interior monologue," and
which it would be better to call immediate speech: for the main
point, which did not escape Joyce, is not that the speech should

17 Reported by Valery Larbaud in his preface to the 10/18 edition of Dujardin's
Les Lauriers sont coupes, p. 8. This conversation took place in 1920 or shortly
thereafter. Let us recall that the novel dates from 1887.


Narrative Discourse

be internal, but that it should be emancipated right away ("from
the first lines") from all narrative patronage, that it should from
the word go take the front of the "stage." 18

We know, from Joyce to Beckett, to Nathalie Sarraute, to
Roger Laporte, what that strange little book's posterity has been
and what revolution in the history of the novel that new form
effected in the twentieth century. 19 It is not part of our purpose
to dwell on that here, but only to note the generally misun-
derstood relationship between immediate speech and "reported
speech," which are formally distinguished from one another
only by the presence or absence of a declarative introduction. As
the example of Molly Bloom's monologue in Ulysses shows, or
the first three sections of The Sound and the Fury (successive
monologues of Benjy, Quentin, and Jason), the monologue does
not have to be coextensive with the complete work to be ac-
cepted as "immediate"; it is sufficient, whatever the
monologue's extent may be, for it to happen on its own, without
the intermediary of a narrating instance which is reduced to
silence and whose function the monologue takes on. We see
here the essential difference between immediate monologue and
free indirect style, which are sometimes erroneously confused or
improperly put together: in free indirect speech, the narrator
takes on the speech of the character, or, if one prefers, the
character speaks through the voice of the narrator, and the two
instances are then merged; in immediate speech, the narrator is
obliterated and the character substitutes for him. In the case of an

18 Dujardin himself insists more on a stylistic criterion, which is the necessarily
formless — according to him — nature of interior monologue: "a discourse with-
out an auditor and unspoken, by which a character expresses his most intimate
thoughts, those closest to the unconscious, prior to all logical organization, or,
simply, thought in its dawning state — expresses it by means of direct phrases
reduced to their syntactical minimum, in such a way as to give the impression of
a hodgepodge" (he Monologue interieur [Paris, 1931 1, p. 59). The bond here be-
tween intimacy of thought and the nonlogical and nonarticulated nature of it is,
clearly, a prejudice of the age. Molly Bloom's monologue corresponds fairly well
to that description, but those of Beckett's characters are, on the contrary, rather
hyperlogical and ratiocinating.

19 On this subject see L. E. Bowling, "What Is the Stream of Consciousness
Technique?" PMLA, 65 (1950), 333-345; Robert Humphrey, Stream of Consciousness
in the Modern Novel (Berkeley, 1954); Melvin Friedman, Stream of Consciousness: A
Study in Literary Method (New Haven, 1955).



isolated monologue, one not taking up the whole of the narra-
tive, as in Joyce or Faulkner, the narrating instance is main-
tained (but in the background) by the context: all the chapters
before the last one in Ulysses, the fourth section in The Sound and
the Fury. When the monologue blends with the whole of the
narrative, as in the Lauriers, or Martereau, or Fugue, the higher
(i.e., narrating) instance is annulled, and we are again in the
presence of a narrative in the present tense and "in the first
person." Here we verge on problems of voice. For the moment
let us not go further, and let us return to Proust.

Needless to say, unless one is deliberately trying to prove a
point (like the rejection, in Plato's rewriting of Homer, of all
reported speech), the different forms we have just distinguished
in theory will not be so clearly separated in the practice of texts.
Thus, we have already been able to note in the text Plato pro-
posed (or at any rate in its English translation) an almost imper-
ceptible sliding from narrated speech to transposed speech, and
from indirect style to free indirect style. The same chain is
found, for example, in this passage of Un amour de Swann, where
the narrator first describes in his own name the feelings of
Swann admitted to Odette's and confronting the agonies cus-
tomary to him in his present situation: "And then ... all the
terrible and disturbing ideas which he had formed of Odette
melted away and vanished in the charming creature who stood
there before his eyes"; then, introduced by the phrase "He had
the sudden suspicion," here is a whole series of the character's
thoughts reported in indirect style:

that this hour spent in Odette's house, in the lamp-light was,
perhaps, after all, not an artificial hour . . . ; that, if he himself had
not been there, she would have pulled forward the same armchair
for Forcheville . . . ; that the world inhabited by Odette was not
that other world, fearful and supernatural, in which he spent his
time in placing her — and which existed, perhaps, only in his im-
agination, but the real universe . . . ;

then Marcel lends his voice, in free indirect style (and with the
grammatical transpositions that implies) to Swann's own inner


Narrative Discourse

Ah! had fate but allowed him to share a single dwelling with Odette,
so that in her house he should be in his own; if, when asking his
servant what there would be for luncheon, it had been Odette's bill
of fare that he had learned from the reply; if, when Odette wished to
go for a walk, in the morning, along the Avenue du Bois-de-
Boulogne, his duty as a good husband had obliged him, though he
had no desire to go out, to accompany her . . . ; then how com-
pletely would all the trivial details of Swann's life, which seemed to
him now so gloomy, simply because they would, at the same time,
have formed part of the life of Odette, have taken on ... a sort of
superabundant sweetness and a mysterious solidity;

then, after that sort of mimetic atmosphere, the text returns to
subordinated indirect style:

And yet he was inclined to suspect that the state for which he so
much longed was a calm, a peace, which would not have created
an atmosphere favourable to his love. . . . He told himself that, when
he was cured of it, what Odette might or might not do would be
indifferent to him,

to return finally to the opening mode of narratized speech ("he
feared death itself no more than such a recovery"), a mode that
allows the text imperceptibly to move on to the narrative of
events: "After these quiet evenings, Swann's suspicions would
be temporarily lulled; he would bless the name of Odette, and
next day, in the morning, would order the most attractive jewels
to be sent to her." 20

These gradations or subtle blends of indirect style and nar-
rated speech ought not to blind us to the Proustian narrative's
characteristic use of reported inner speech. Whether Marcel is
involved or Swann, the Proustian hero, especially in his
moments of ardent emotion, readily articulates his thoughts as a
genuine monologue, enlivened by a fully theatrical rhetoric.
Here is Swann when angry:

"But I've been a fool, too," he would argue. "I'm paying for other
men's pleasures with my money. All the same, she'd better take

>RH I, 229-230/P I, 298-300. (My emphasis.)



care, and not pull the string too often, for I might very well stop
giving her anything at all. At any rate, we'd better knock off
supplementary favours for the time being. To think that, only
yesterday, when she said she would like to go to Bayreuth for the
season, I was such an ass as to offer to take one of those jolly little
places the King of Bavaria has there, for the two of us. However
she didn't seem particularly keen; she hasn't said yes or no yet.
Let's hope that she'll refuse. Good God! Think of listening to
Wagner for a fortnight on end with her, who takes about as much
interest in music as a fish does in little apples; it will be fun!" 21

Or Marcel trying to reassure himself after Albertine's departure:

"All this means nothing," I told myself, "It is even better than I
thought, for as she doesn't mean a word of what she says she has
obviously written her letter only to give me a severe shock, so that
I shall take fright, and not be horrid to her again. I must make
some arrangement at once: Albertine must be brought back this
evening. It is sad to think that the Bontemps are no better than
blackmailers who make use of their niece to extort money from
me. But what does that matter?" 22

Moreover, sometimes it happens that Swann, at least, speaks
"to himself, aloud," and, what is more, on the street, when he is
returning home furious after having gotten himself excluded
from the party at Chatou:

"What a fetid form of humour!" he exclaimed, twisting his mouth
into an expression of disgust so violent that he could feel the muscles
of his throat stiffen against his collar. ... "I dwell so many miles
above the puddles in which these filthy little vermin sprawl and
crawl and bawl their cheap obscenities, that I cannot possibly be
spattered by the witticisms of a Verdurin!" he cried, tossing up his
head and arrogantly straightening his body. . . .

He had long since emerged from the paths and avenues of the
Bois, he had almost reached his own house, and still, for he had
not yet thrown off the intoxication of grief, or his whim of insincer-
ity, but was ever more and more exhilarated by the false intonation,

21 RH I, 231/P I, 300-301. This monologue is pseudo-iterative as well

22 RH II, 676-678/P III, 421-122.


Narrative Discourse

the artificial sonority of his own voice, he continued to perorate aloud
in the silence of the night. 23

We see that here the sound of the voice and the factitious intona-
tion form part of the thought, or rather reveal it beyond the
emphatic disclaimers of bad faith:

Doubtless Swann's voice shewed a finer perspicacity than his own
when it refused to utter those words full of disgust at the Verdu-
rins and their circle, and of joy at his having shaken himself free of
it, save in an artificial and rhetorical tone, and as though his words
had been chosen rather to appease his anger than to express his
thoughts. The latter, in fact, while he abandoned himself to invec-
tive, were probably, though he did not know it, occupied with a
wholly different matter.

This "matter," which is more than different from— which is di-
ametrically opposed to— the scornful speech Swann addresses
to himself, is obviously to reingratiate himself at any cost with
the Verdurins and get himself invited to the dinner at Chatou.
Such is often the duplicity of inner speech, and nothing can
reveal it better than these insincere monologues uttered aloud,
like a scene, a "comedy" that one is acting in for oneself.
"Thought" is indeed speech, but at the same time this speech,
"oblique" and deceitful like all the others, is generally unfaithful
to the "felt truth"— the felt truth which no inner monologue can
reveal and which the novelist must ultimately show glimpses of
through the concealments of bad faith, which are "conscious-
ness" itself. That is expressed fairly well in the passage of the
Temps retrouve which follows the famous assertion that "The
function and task of a writer are those of a translator":

And if in some cases— where we are dealing, for instance, with the
inaccurate language of our own vanity— the rectification of an ob-
lique interior discourse which deviates gradually more and more
widely from the first and central impression, so that it is brought
back into line and made to merge with the authentic words which
the impression ought to have generated, is a laborious undertak-

23 RH I, 219-222/P I, 286-289. (My emphasis.)



ing which our idleness would prefer to shirk, there are other
circumstances — for example, where love is involved — in which
this same process is actually painful. Here all our feigned indif-
ferences, all our indignation at the lies of whomever it is we love
(lies which are so natural and so like those that we perpetrate
ourselves), in a word all that we have not ceased, whenever we are
unhappy or betrayed, not only to say to the loved one but, while
we are waiting for a meeting with her, to repeat endlessly to our-
selves, sometimes aloud in the silence of our room, which we disturb
with remarks like: "No, really, this sort of behavior is intolerable,"
and: "I have consented to see you once more, for the last time, and
I don't deny that it hurts me," all this can only be brought back
into conformity with the felt truth from which it has so widely
diverged by the abolition of all that we have set most store by, all
that in our solitude, in our feverish projects of letters and schemes,
has been the substance of our passionate dialogue with ourselves. 24

Although we would perhaps expect from Proust —
chronologically situated as he is between Dujardin and Joyce —
some movement in the direction of the "interior monologue"
after the style of the Lauriers or Ulysses, 25 yet we know he pre-
sents almost nothing in his work which we can liken to that. It
would be totally mistaken to describe as such the passage in the
present tense ("I drink a second mouthful, in which I find noth-
ing more than in the first ... ") which is inserted in the episode
of the madeleine 26 and whose stance much more recalls the
narrative present of philosophical experience, as we find it for
example in Descartes or Bergson; the hero's supposed soliloquy
here is very firmly taken charge of by the narrator for obvious
purposes of demonstration, and nothing is more remote than
this from the spirit of the modern interior monologue, which
encloses the character in the subjectivity of a "real experience"

24 RH II, 1016/P III, 890-891. (My emphasis.)

25 On this subject see Michel Raimond, La Crise du reman (Paris, 1966), pp.
277-282, who examines Robert Kemp's view, expressed in 1925, of a Proust
employing the interior monologue, and decides, like Dujardin, in the negative:
"These vistas seem to lead him sometimes to the frontiers of the interior
monologue, but he never crosses them, and most of the time he stays clear of

"RHI, 34-35/PI, 45^16.


Narrative Discourse

without transcendence or communication. The only case in
which the form and spirit of immediate monologue appear in
the Recherche is the one J. P. Houston notes — while describing it
precisely as "quite a rarity in Proust" — on page 436 of La Pris-
onniere. 27 But Houston quotes only the opening lines of this
passage, which despite all their animation perhaps come under
free indirect style; and it is the subsequent lines which, aban-
doning all temporal transposition, constitute the genuine Joycean
hapax of the Recherche. Here is the whole of this passage,
in which I emphasize the several phrases where immediate
monologue is incontestable:

Those morning concerts at Balbec were not remote in time. And
yet, at that comparatively recent moment, 1 had given but little
thought to Albertine. Indeed, on the very first mornings after my
arrival, I had not known that she was at Balbec. From whom then
had I learned it? Oh, yes, from Aime. It was a fine sunny day like
this. He was glad to see me again. But he does not like Albertine. Not
everybody can be in love with her. Yes, it was he who told me that she was
at Balbec. But how did he know? Ah! he had met her, had thought
that she had a bad style. 28

When all is said and done, then, Proustian handling of inner
speech is extremely traditional, although not completely for tra-
ditional reasons, showing a very marked — and to some people,
paradoxical — aversion to what Dujardin calls the mental
"hodgepodge," "thought in a dawning state," represented by
an infraverbal flux reduced to the "syntactical minimum." Noth-
ing is more foreign to Proustian psychology than the Utopia of
an authentic interior monologue whose inchoateness sup-
posedly guarantees transparency and faithfulness to the deepest
eddies of the "stream of consciousness" — or of unconscious-

The single apparent exception is the last sentence in Marcel's
dream at Balbec: 29 "You know quite well I shall always stay

27 Houston, p. 37.
2B RH II, 436/P III, 84.
M RH II, 117-118/PII, 762.



beside her, dear, deer, deer, Francis Jammes, fork" — which con-
trasts with the perfectly articulate character of the words ex-
changed in this dream until then. 30 But if we look at it a little
more closely, this contrast itself carries a very precise meaning:
immediately after that sentence with the conspicuous incoher-
ence, the narrator adds:

But already I had retraced the dark meanderings of the stream,
had ascended to the surface where the world of living people
opens, so that if I still repeated: "Francis Jammes, deer, deer," the
sequence of these words no longer offered me the limpid meaning
and logic which they had expressed to me so naturally an instant
earlier and which I could not now recall. I could not even under-
stand why the word 'Aias' which my father had just said to me,
had immediately signified: "Take care you don't catch cold,"
without any possible doubt.

This means that the infralinguistic sequence deer, Francis Jammes,
fork is by no means given as an example of dream language, but
as evidence of rupture and incomprehension, at waking, be-
tween that language and the alert consciousness. In the space of
the dream, everything is clear and natural, expressed by
speeches with perfect linguistic coherence. It is at waking — in
other words, at the moment when this coherent universe gives
up its place to another (whose logic is different) — that what was
"limpid" and "logical" loses its transparency. Similarly, when
the sleeper of the opening pages of Swann is emerging from his
first sleep, the subject of his dream (his being a church, a quar-
tet, the rivalry of Francois I and Charles V) "would begin to seem
unintelligible, as the thoughts of a former existence must be to a
reincarnate spirit." 31 The infralinguistic "hodgepodge" is thus
in Proust never the speech of a supposedly alogical depth, even
the depth of dream, but is only the means of representing, by a
sort of transitory and borderline misunderstanding, the gulf be-
tween two logics, each as distinct as the other.

30 As in Swann's, RH I, 290-292/P I, 378-381.

31 RH I, 3/P I, 3. (My emphasis.)


Narrative Discourse

As to "outer" speech— that is, the stance of what we tra-
ditionally call "dialogue," even if it involves more than two
characters— we know that Proust here completely parts com-
pany with the Flaubertian practice of free indirect style. Mar-
guerite Lips has noted two or three examples of it, 32 but they stand
as exceptions. That ambiguous transfusion of speeches, that
confusion of voices is deeply foreign to his style, which here is
linked much more to the Balzacian model, marked by the pre-
dominance of reported speech and of what Proust himself calls
"objectivized language"— that is, linguistic autonomy granted
to the characters, or at any rate to some of thern:

Because in some respects Balzac is a slapdash writer, one might
suppose that he did not trouble to make his characters talk like
themselves, or that if he had tried to, he would not have been able
to resist drawing attention to it at every turn. However, it is quite
the opposite: the man who artlessly reels out his views on history,
art, and so forth, keeps the most deep-laid schemes under cover,
and leaves the truth of the dialogue to speak for itself, without
attempting to underline what he does so artfully that it might go
unnoticed. When he makes the lovely Mme. de Roguin, that born
Parisian whom Tours knew as the country prefect's wife, talk

"The example of Francoise's menus, RH I, 54/P I, 71: "a brill, because the
fish-woman had guaranteed its freshness; a turkey, because she had seen a
beauty in the market at Roussainville-le-Pin, ..." where the citational nature is
not very marked, except in "a roast leg of mutton, because the fresh air made
one hungry and there would be plenty of time for it to 'settle down' in the seven
hours before dinner" (Lips, p. 46); and this other one, more obvious because of
the interjection: "We would fly upstairs to my aunt Leonie's room to reassure
her, to prove to her by our bodily presence that all her gloomy imaginings were
false, that, on the contrary, nothing had happened to us, but that we had gone
the 'Guermantes way,' and, good lord, when one took that walk, my aunt knew
well enough that one could never say at what time one would be home" (RH 1,
102-103/P I, 133-134; Lips, p. 99). Here is another, where the source of the
discourse (again Francoise) increasingly stands out: "She was quite overcome
because there had just been a terrible scene between the lovesick footman and
the tale-bearing porter. It had required the Duchess herself, in her unfailing
benevolence, to intervene, restore an apparent calm to the household and for-
give the footman. For she was a good mistress, and that would have been the
ideal 'place' if only she didn't listen to 'stories' " (RH I, 935/P II, 307). We see that
Proust does not dare to take on the servant's lexicon without quotation marks: a
sign of great timidity in the use of free indirect style.



about how the Rogrons have furnished their house, how infallibly
all those sallies are hers, not Balzac's! 33

This autonomy is sometimes disputed, and Malraux, for exam-
ple, deems it "altogether relative." 34 No doubt it is excessive to
say, like Gaetan Picon (whom Malraux is answering here), that
Balzac "seeks to give each character a personal voice," if personal
voice means own individual style. "Verbal features of character"
emerge in Balzac (as in Moliere) through meaning rather than
through style, and the most conspicuous pronunciations
(Nucingen's or Schmucke's German accent or the concierge par-
lance of La Cibot) are group languages rather than personal
styles. Nonetheless, the attempt at characterization is obvious
and, whether idiolect or sociolect, the characters' parlance is
indeed "objectivized," with a marked differentiation between
the narrator's speech and the characters' speech — and the
mimetic effect is thus probably more intense than in the work of
any previous novelist.

Proust, for his part, will push the effect much further, and the
mere fact that he should have noted it and somewhat exagger-
ated its presence in Balzac shows well, as do all critical distor-
tions of this type, what his own course was. Clearly no one else,
either before him or after, and to my knowledge not in any
language, has so nailed down the "objectivization" — and this
time the individuation — of the characters' style. I have touched
on this subject elsewhere; 35 an exhaustive study of it would
require a comparative stylistic analysis of the speeches of Char-
lus, Norpois, Francoise, etc., not without unavoidable refer-
ences to the "psychology" of these characters — and would re-
quire also a comparison between the technique of these imagi-
nary (or partially imaginary) pastiches and the technique of the
real pastiches of the Affaire Lemoine and elsewhere. To do that is
not our purpose here. It is enough to recall the importance of the

33 Marcel Proust on Art, p. 179.

14 Gaetan Picon, Malraux par lui-meme (Paris, 1953), p. 40.

35 Figures 11, pp. 223-294. Cf. Tadie, Proust et k roman, chap. 6.


Narrative Discourse

fact; but we must also mention the unevenness of its dispersion.
Indeed, it would be excessive and hasty to say that all Proust's
characters have an idiolect, and all with the same continuous-
ness and intensity. The truth is that nearly all of them do pre-
sent, at least at some time, some eccentric characteristic of lan-
guage, an incorrect or dialectical or socially imprinted turn of
phrase, a typical acquisition or borrowing, a blunder, howler, or
tell-tale slip, etc.; we can say that none of them escapes that
minimal state of connotative relationship with language, except
perhaps the hero himself, who as such speaks very little, for that
matter, and whose role here is rather as observer, apprentice,
and decoder. At a second level are the characters marked by a
recurrent linguistic characteristic, which belongs to them like a
tic or a personal and/or class marker: Odette's Anglicisms, Ba-
sin's improprieties, Bloch's schoolboy pseudo-Homerisms,
Saniette's archaisms, the blunders of Franchise or of the director
at Balbec, Oriane's puns and provincialisms, Saint-Loup's
social-club jargon, the Sevigne style of the hero's mother and
grandmother, errors in pronunciation by the Princess Sher-
batoff, Breaute, Faffenheim, etc. This is where Proust is closest
to the Balzacian model, and this is the practice which has been
most often imitated since. 36 The highest level is that of personal
style as such, 37 both specific and continuous, as we find it with
Brichot (demagogic professor's pedantism and familiarism),
with Norpois (officious truisms and diplomatic periphrases),
with Jupien (classical purity), with Legrandin (decadent style),
and especially with Charlus (furious rhetoric). "Stylized"
speech is the extreme form of the mimesis of speech, where the
author "imitates" his character not only in the tenor of his re-
marks but in the hyperbolic literalness of pastiche, which is
always a little more idiolectical than the authentic text, as
"imitation" is always a caricature through accumulation and ac-
centuation of specific characteristics. And so Legrandin or Char-

36 £ ven by Malraux, who did not fail to give tics of language to some of his
heroes (Katow's elisions, Clappique's "my good man," Tchen's "Nong,"
Pradas's "concretely," Garcia's obsession with definitions, etc.).

37 Which does not mean that here the idiolect is devoid of all representative
value: Brichot speaks as a Sorbonne man, Norpois as a diplomat.



lus always gives the impression of imitating himself, and finally
of caricaturing himself. Here the mimetic effect is thus at its
height, or more exactly at its limit: at the point where the ex-
treme of "realism" borders on pure unreality. The narrator's
unerring grandmother says rightly that Legrandin talks "a little
too much like a book." 38 In a larger sense, this risk lies heavy
over any too-perfect mimesis of language, which finally annuls
itself in the circularity— already noted by Plato — of the link to its
shadow: Legrandin talks like Legrandin (in other words, like
Proust imitating Legrandin), and speech, finally, sends one back
to the text that "quotes" it (in other words, to the text that in fact
constitutes it).

This circularity perhaps explains why a technique of "charac-
terization" as effective as stylistic autonomy does not, in Proust,
result in the composition of substantial and well-defined charac-
ters in the realistic sense of the term. We know that Proustian
"characters" remain, or rather become, down through the pages
more and more indefinable, ungraspable, "creatures in flight,"
and the incoherence of their behavior is obviously the main
reason for this, and the reason most carefully arranged for by
the author. But the hyperbolic coherence of their language, far
from compensating for that psychological evanescence, quite
often simply accentuates it and aggravates it. A Legrandin, a
Norpois, even a Charlus does not completely escape the
exemplary fate of lesser characters like the director at Balbec,
Celeste Albaret, or the footman Perigot Joseph: blending with,
to the point of amounting to, his language. Here the strongest
verbal existence is the sign and the beginning of a disappear-
ance. At the limit of stylistic "objectivization," the Proustian
character finds this highly symbolic form of death: doing away
with himself in his own speech.


What we are calling, for the moment and through metaphor,
narrative perspective — in other words, the second mode of regu-
lating information, arising from the choice (or not) of a restric-

38 RH I, 51/P I, 67-68.


Narrative Discourse

tive "point of view" — is, of all the questions having to do with
narrative technique, the one that has been most frequently stud-
ied since the end of the nineteenth century, with indisputable
critical results, like Percy Lubbock's chapters on Balzac,
Flaubert, Tolstoy, or James, or Georges Blin's chapter on "re-
strictions of field" in Stendhal. 39 However, to my mind most of
the theoretical works on this subject (which are mainly
classifications) suffer from a regrettable confusion between
what I call here mood and voice, a confusion between the ques-
tion who is the character whose point of view orients the narra-
tive perspective? and the very different question who is the
narrator? — or, more simply, the question who sees? and the ques-
tion who speaks? We will return later to this apparently obvious
but almost universally disregarded distinction. Thus Cleanth
Brooks and Robert Penn Warren proposed in 1943, under the
term focus of narration — which they explicitly (and very happily)
proposed as an equivalent to "point of view" — a four-term
typology, summed up in the table below. 40

Internal analysis of events

Outside observation of events

Narrator as a
character in
the story

Narrator not a
character in
the story

1. Main character tells
his story

4. Analytic or omniscient
author tells storv

2. Minor character tells
main character's story

3. Author tells story
as observer

Now, it is obvious that only the vertical demarcation relates to
"point of view" (inner or outer), while the horizontal bears on
voice (the identity of the narrator), with no real difference in
point of view between 1 and 4 (let us say Adolphe and Armance)

"Georges Blin, Stendhal et les vroblemes du roman (Paris, 1954), Part II. For a
"theoretical" bibliography on this subject, see Francoise van Rossum-Guyon,
"Point de vue ou perspective narrative," Poetique, 4 (1970). From the historical
angle, see Richard Stang, The Theory of the Novel in England, 1850-1870 (New
York, 1959), chap. 3; and Raimond, La Crise du roman, Part IV.

""Cleanth Brooks and Robert Penn Warren, Understanding Fiction (New York,
1943), p. 589.



and between 2 and 3 (Watson narrating Sherlock Holmes, and
Agatha Christie narrating Hercule Poirot). In 1955, F. K. Stanzel
distinguished three types of novelistic "narrative situations":
the auktoriale Erzahlsituation, which is that of the "omniscient"
author (type: Tom Jones); the Ich Erzahlsituation, where the nar-
rator is one of the characters (type: Moby Dick); and the personate
Erzahlsituation, a narrative conducted "in the third person" ac-
cording to the point of view of a character (type: The Ambas-
sadors). 41 Here again, the difference between the second and
third situations is not in "point of view" (whereas the first is
defined according to that criterion), since Ishmael and Strether
in fact occupy the same focal position in the two narratives: they
differ only in that in one the focal character himself is the nar-
rator, and in the other the narrator is an "author" absent from
the story. In the same year Norman Friedman, on his part, pre-
sented a much more complex classification with eight terms: two
types of "omniscient" narrating with or without "authorial in-
trusions" (Fielding or Thomas Hardy); two types of "first-
person" narrating, I-witness (Conrad) or I-protagonist (Dickens,
Great Expectations); two types of "selective-omniscient" narrat-
ing, that is, with restricted point of view, either "multiple"
(Virginia Woolf, To the Lighthouse), or single (Joyce, Portrait of
the Artist); finally, two types of purely objective narrating, the
second of which is hypothetical and, moreover, not easily dis-
tinguishable from the first: the "dramatic mode" (Hemingway,
"Hills Like White Elephants") and "the camera," a recording
pure and simple, without selection or organization. 42 Clearly,
the third and fourth types (Conrad and Dickens) are distin-
guished from the others only in being "first-person" narratives,
and the difference between the first two (intrusions of the au-
thor or not: Fielding or Hardy) is likewise a fact of voice, relating
to the narrator and not to the point of view. Let us recall that
Friedman describes his sixth type (Portrait of the Artist) as "a
story told as if by a character in the story, but told in the third

41 F. K. Stanzel, Narrative Situations in the Novel, trans. J. P. Pusack
(Bloomington, lnd., 1971).

42 N. Friedman, "Point of View in Fiction."


Narrative Discourse

person," a formulation that attests to obvious confusion be-
tween the focal character (what James called the "reflector") and
the narrator. The same assimilation, obviously intentional, oc-
curs with Wayne Booth, who in 1961 gave the title "Distance
and Point of View" to an essay devoted in fact to problems of
voice (the distinction between implied author and narrator — a
narrator who is dramatized or undramatized and reliable or unreli-
able) as, for that matter, he explicitly stated in proposing "a
richer tabulation of the forms the author's voice can take." 43
"Strether," continued Booth, "in large part 'narrates' his own
story, even though he is always referred to in the third person":
is his status, then, identical to Caesar's in the Gallic War? We see
what difficulties the confusion between mood and voice leads
to. In 1962, finally, Bertil Romberg took up Stanzel's typology
again, and completed it by adding a fourth type: objective narra-
tive in the behaviorist style (Friedman's seventh type); 44 whence
this quadripartition: (1) narrative with omniscient author, (2)
narrative with point of view, (3) objective narrative, (4) narrative
in the first person — where the fourth type is clearly discordant
with respect to the principle of classification of the first three.
Here Borges would no doubt introduce a fifth class, typically
Chinese: that of narratives written with a very fine brush.

It is certainly legitimate to envisage a typology of "narrative
situations" that would take into account the data of both mood
and voice; what is not legitimate is to present such a classifica-
tion under the single category of "point of view," or to draw up
a list where the two determinations compete with each other on
the basis of an obvious confusion. And so it is convenient here
to consider only the purely modal determinations, those that
concern what we ordinarily call "point of view" or, with Jean
Pouillon and Tzvetan Todorov, "vision" or "aspect." 45 Granting
this restriction, the consensus settles with no great difficulty on
a three-term typology. The first term corresponds to what

43 Booth, "Distance and Point of View," Essays in Criticism, 11 (1961), 60-79.

44 Bertil Romberg, Studies in the Narrative Technique of the First-Person Novel,
trans. Michael Taylor and Harold H. Borland (Stockholm, 1962).

45 Jean Pouillon, Temps et roman (Paris, 1946); Todorov, "Les Categories du
recit litteraire."



English-language criticism calls the narrative with omniscient
narrator and Pouillon calls "vision from behind," and which
Todorov symbolizes by the formula Narrator > Character (where
the narrator knows more than the character, or more exactly says
more than any of the characters knows). In the second term,
Narrator = Character (the narrator says only what a given charac-
ter knows); this is the narrative with "point of view" after Lub-
bock, or with "restricted field" after Blin; Pouillon calls it "vision
with." In the third term, Narrator < Character (the narrator says
less than the character knows); this is the "objective" or "be-
haviorist" narrative, what Pouillon calls "vision from without."
To avoid the too specifically visual connotations of the terms
vision, field, and point of view, I will take up here the slightly more
abstract term focalization 46 which corresponds, besides, to
Brooks and Warren's expression, "focus of narration." 47


So we will rechristen the first type (in general represented by
the classical narrative) as nonfocalized narrative, or narrative with
zero focalization. The second type will be narrative with internal
Realization, whether that be (a) fixed — canonical example: The
Ambassadors, where everything passes through Strether; or,
even better, What Maisie Knew, where we almost never leave the
point of view of the little girl, whose "restriction of field" is
particularly dramatic in this story of adults, a story whose sig-
nificance escapes her; (b) variable — as in Madame Bovary, where
the focal character is first Charles, then Emma, then again
Charles; 48 or, in a much more rapid and elusive way, as with

46 Already used in my Figures 11, p. 191, apropos of Stendhalian narrative.

47 YJ e can d raw a parallel between this tripartition and the four-term classifica-
tion proposed by Boris Uspenski (A Poetics of Composition, trans. Valentina Zava-
rin and Susan Wittig IBerkeley, 1973 1) for the "plane of psychology" of his
general theory of point of view (see the "clarification" and documents presented
by Todorov in Poetique, 9(February 1972)). Uspenski distinguishes two types in
the point-of-view narrative, according to whether the point of view is constant
(fixed on a single character) or not: this is what I propose to call fixed or variable
internal focalization, but for me these are only subclasses.

48 On this subject see Lubbock, The Craft of Fiction, chap. 6, and Jean Rousset,
"Madame Bovary ou le Livre sur rien," Forme el signification (Paris, 1962).


Narrative Discourse

Stendhal; or (c) multiple — as in epistolary novels, where the
same event may be evoked several times according to the point
of view of several letter-writing characters; 49 we know that
Robert Browning's narrative poem The Ring and the Book (which
relates a criminal case as perceived successively by the mur-
derer, the victims, the defense, the prosecution, etc.) was for
several years the canonical example of this type of narrative, 50
before being supplanted for us by the film Rashomon. Our third
type will be the narrative with external localization, popularized
between the two world wars by Dashiell Hammett's novels, in
which the hero performs in front of us without our ever being
allowed to know his thoughts or feelings, and also by some of
Hemingway's novellas, like "The Killers" or, even more, "Hills
Like White Elephants," which carries circumspection so far as
to become a riddle. But we should not limit this narrative type to a
role only in works at the highest literary level. Michel Raimond
remarks rightly that in the novel of intrigue or adventure,
"where interest arises from the fact that there is a mystery," the
author "does not tell us immediately all that he knows"; 51 and in
fact a large number of adventure novels, from Walter Scott to
Jules Verne via Alexandre Dumas, handle their opening pages
in external focalization. See how Phileas Fogg is looked at first
from the outside, through the puzzled gaze of his contem-
poraries, and how his inhuman mysteriousness will be main-
tained until the episode that will reveal his generosity. 52 But
many "serious" novels of the nineteenth century practice this
type of enigmatic introit: examples, in Balzac, are La Peau de
chagrin or L'Envers de I'histoire contemporaine, and even Le Cousin
Pons, where the hero is described and followed for a long time as

49 See Rousset, "Le Roman par lettres," Forme et signification, p. 86.

"> See Raimond, pp. 313-314. Proust was interested in that book: see Tadie, p.


51 La Crise du roman, p. 300.

52 It is the rescue of Aouda, in chapter 12. Nothing prevents a writer from
indefinitely prolonging this external point of view with respect to a character
who will remain mysterious up to the end: that is what Melville does in The
Confidence-Man, or Conrad in The Nigger of the "Narcissus."



an unknown person whose identity is problematic. 53 And other
motives can justify recourse to this narrative behavior, like the
reason of propriety (or the roguish play with impropriety) for
the scene of the carriage in Bovary, which is narrated entirely
from the point of view of an external, innocent witness. 54

As this last example certainly shows, the commitment as to
focalization is not necessarily steady over the whole length of a
narrative, and variable internal focalization, a formula already
very flexible, does not apply to the whole of Bovary: not only is
the scene of the carriage in external focalization, but we have
already had occasion to say that the view of Yonville that begins
the second part is not any more focalized than most Balzacian
descriptions. ss Any single formula of focalization does not,
therefore, always bear on an entire work, but rather on a definite
narrative section, which can be very short. 56 Furthermore, the
distinction between different points of view is not always as
clear as the consideration of pure types alone could lead one to
believe. External focalization with respect to one character could
sometimes just as well be defined as internal focalization
through another: external focalization on Phileas Fogg is just as
well internal focalization through Passepartout dumbfounded
by his new master, and the only reason for being satisfied with

51 This initial "ignorance" has become a topos of novelistic beginning, even
when the mystery is to be immediately dispelled. For example, in the fourth
paragraph of the Education sentimentale: "A young man eighteen years old with
long hair and holding an album under his arm ..." It is as if, to introduce him,
the author had to pretend not to know him; once this ritual is over, he can go on
without further affectations of mystery: "M. Frederic Moreau, newlv
graduated ..." The two periods may be very close together, but they must be
distinct. This rule operates still, for example, in Germinal, where first the hero is
"a man," until he introduces himself: "My name is Etienne Lantier," after which
Zola will call him Etienne. On the other hand, the rule no longer operates in
James, who from the very beginning establishes a familiar relationship with his
heroes: "Strerher's first question, when he reached the hotel. . . " (The Ambas-
sadors); "She waited, Kate Croy, for her father to come in ... " (The Wings of the
Dove); "The Prince had always liked his London ..." (The Golden Bowl). These
variations would be worth an overall historical study.

54 III, chap. 1. Cf. Sartre, L'Uiot de la famille (Paris, 1971), pp. 1277-1282

55 P. 101.

5f, See Raymonde Debray-Genette, "Du mode narratif dans les Trois Contes,"
Litterature, 2 (May 1971).


Narrative Discourse

the first term is Phileas's status as hero, which restricts
Passepartout to the role of witness. And this ambivalence (or
reversibility) is equally noticeable when the witness is not per-
sonified but remains an impersonal, floating observer, as at the
beginning of La Peau de chagrin. Similarly, the division between
variable focalization and nonfocalization is sometimes very dif-
ficult to establish, for the nonfocalized narrative can most often
be analyzed as a narrative that is multifocalized ad libitum, in
accordance with the principle "he who can do most can do
least" (let us not forget that focalization is essentially, in Blin's
word, a restriction); and yet, on this point no one could confuse
Fielding's manner with Stendhal's or Flaubert's. 57

We must also note that what we call internal focalization is
rarely applied in a totally rigorous way. Indeed, the very princi-
ple of this narrative mode implies in all strictness that the focal
character never be described or even referred to from the out-
side, and that his thoughts or perceptions never be analyzed
objectively by the narrator. We do not, therefore, have internal
focalization in the strict sense in a statement like this one, where
Stendhal tells us what Fabrice del Dongo does and thinks:

Without hesitation, although ready to yield up his soul with dis-
gust, Fabrizio flung himself from his horse and took the hand of the
corpse which he shook vigorously; then he stood still as though
paralysed. He felt that he had not the strength to mount again.
What horrified him more than anything was that open eye.

On the other hand, the focalization is perfect in the following
statement, which is content to describe what its hero sees: "A
bullet, entering on one side of the nose, had gone out at the

57 Balzac's position is more complex. One is often tempted to see Balzacian
narrative as the very type of narrative with an omniscient narrator, but to do that
is to neglect the part played by external focalization, which I have just referred to
as a technique of opening; and neglects also the part played by more subtle
situations, as in the first pages of Line double famille, where the narrative focalizes
sometimes on Camille and her mother, sometimes on M. de Granville — each of
these internal focalizations serving to isolate the other character (or group) in its
mysterious externality: a rearrangement of curiosities that can only quicken the
reader's own.



opposite temple, and disfigured the corpse in a hideous fashion.
It lay with one eye still open." 58 Jean Pouillon very accurately
notes the paradox when he writes that, in "vision with," the
character is seen

not in his innerness, for then we would have to emerge from the
innerness whereas instead we are absorbed into it, but is seen in
the image he develops of others, and to some extent through that
image. In sum, we apprehend him as we apprehend ourselves in
our immediate awareness of things, our attitudes with respect to
what surrounds us — what surrounds us and is not within us. Con-
sequently we can say in conclusion: vision as an image of others is
not a result of vision "with" the main character, it is itself that
vision "with." 59

Internal focalization is fully realized only in the narrative of
"interior monologue," or in that borderline work, Robbe-
Grillet's la ]alousie, 60 where the central character is limited abso-
lutely to— and strictly inferred from— his focal position alone. So
we will take the term "internal focalization" in a necessarily less
strict sense — that term whose minimal criterion has been
pointed out by Roland Barthes in his definition of what he calls
the personal mode of narrative. 61 According to Barthes, this crite-
rion is the possibility of rewriting the narrative section under
consideration into the first person (if it is not in that person
already) without the need for "any alteration of the discourse
other than the change of grammatical pronouns." Thus, a sen-
tence such as "[James Bond] saw a man in his fifties, still
young-looking. . . " can be translated into the first person ("I
saw . . . ")— and so for us it belongs to internal focalization. On
the other hand, Barthes continues, a sentence like "the tinkling
of the ice cubes against the glass seemed to awaken in Bond a

58 Charterhouse of Parma, chap. 3, trans. C. K. Scott Moncrieff (New York-
Liveright, 1925), p. 48.

59 Temps et roman, p. 79.

60 Or, in the movies, Robert Montgomery's The Lady in the Lake, where the
protagonist's part is played by the camera.

61 Barthes, "An Introduction to the Structural Analysis of Narrative," p. 262.


Narrative Discourse

sudden inspiration" cannot be translated into the first person
without obvious semantic incongruity. 62 Here we are typically
in external localization, because of the narrator's marked igno-
rance with respect to the hero's real thoughts. But the conve-
nience of this purely practical criterion should not tempt us to
confuse the two instances of the focalizing and the narrating,
which remain distinct even in "first-person" narrative, that is,
even when the two instances are taken up by the same person
(except when the first-person narrative is a present-tense inte-
rior monologue). When Marcel writes, "I saw a man of about
forty, very tall and rather stout, with a very dark moustache,
who, nervously slapping the leg of his trousers with a switch,
kept fastened upon me a pair of eyes dilated with observa-
tion," 63 the identity of "person" between the adolescent of Bal-
bec (the hero) who notices a stranger and the mature man (the
narrator) who tells this story several decades later and knows
very well that that stranger was Charlus (and knows all that the
stranger's behavior means) must not conceal the difference in
function and, particularly, the difference in information. The
narrator almost always "knows" more than the hero, even if he
himself is the hero, and therefore for the narrator focalization
through the hero is a restriction of field just as artificial in the
first person as in the third. We will soon come again to this
crucial question apropos of narrative perspective in Proust, but
we must still define two ideas indispensable to that study.


The variations in "point of view" that occur in the course of a
narrative can be analyzed as changes in focalization, like those
we have already met in Madame Bovary: in such a case we can
speak of variable focalization, of omniscience with partial restric-
tions of field, etc. This is a perfectly defensible narrative course,
and the norm of coherence raised to a point of honor by post-

1,2 Proust notices in Le Lys dans la vallee this sentence that he rightly says
manages however it can: " "... I walked down to the meadows to see once again
the Indre and its islets, the valley and its hillsides, of which 1 appeared a passionate
admirer'" (Marcel Proust on Art, p. 172).

63 RH I, 568/PI, 751.



Jamesian criticism is obviously arbitrary. Lubbock requires the
novelist to be "consistent on some plan, to follow the principle
he has adopted," 64 but why could this course not be absolute
freedom and inconsistency? Forster 65 and Booth have well
pointed out the futility of pseudo-Jamesian rules, and who
today would take seriously Sartre's remonstrances against
Mauriac? 66

But a change in focalization, especially if it is isolated within a
coherent context, can also be analyzed as a momentary infrac-
tion of the code which governs that context without thereby
calling into question the existence of the code— the same way
that in a classical musical composition a momentary change in
tonality, or even a recurrent dissonance, may be defined as a
modulation or alteration without contesting the tonality of the
whole. Playing on the double meaning of the word mode, which
refers us to both grammar and music, 67 I will thus give the
general name alterations to these isolated infractions, when the
coherence of the whole still remains strong enough for the no-
tion of dominant mode/mood to continue relevant. The two con-
ceivable types of alteration consist either of giving less informa-
tion than is necessary in principle, or of giving more than is
authorized in principle in the code of focalization governing the
whole. The first type bears a name in rhetoric, and we have
already met it apropos of completing anachronies: 68 we are deal-
ing with lateral omission or paralipsis. The second does not yet
bear a name; we will christen it paralepsis, since here we are no
longer dealing with leaving aside (-lipsis, from leipo) informa-
tion that should be taken up (and given), but on the contrary
with taking up (-lepsis, from lambano) and giving information
that should be left aside.
The classical type of paralipsis, we remember, in the code of

64 The Craft of Fiction, p. 72.

65 E. M. Forster, Aspects of the Novel (London, 1927).

""J. P. Sartre, "Francois Mauriac and Freedom," in Literary and Philosophical
Essays, trans. Annette Michelson (New York, 1955), pp. 7-23.

" [Translator's note.] In French the word mode includes two meanings that in
English require separate words: (grammatical) "mood" and (musical) "mode "

68 P. 52.


Narrative Discourse

internal focalization, is the omission of some important action or
thought of the focal hero, which neither the hero nor the nar-
rator can be ignorant of but which the narrator chooses to con-
ceal from the reader. We know what use Stendhal made of this
figure, 69 and Jean Pouillon evokes precisely this fact apropos of
his "vision with," whose main disadvantage seems to him to be
that the character is too well known in advance and holds no
surprise in store — whence this defense, which Pouillon deems
clumsy: deliberate omission. A solid example: Stendhal's dis-
simulation, in Armance, through so many of the hero's
pseudo-monologues, of that hero's central thought, which ob-
viously cannot leave him for a minute: his sexual impotence.
That affectation of mystery, says Pouillon, would be normal if
Octave were seen from without,

but Stendhal does not remain outside; he makes psychological
analyses, and in that case it becomes absurd to hide from us what
Octave himself must certainly know; if he is sad, he knows the
cause, and cannot experience that sadness without thinking of it:
Stendhal therefore ought to inform us of it — which, unfortunately,
he does not do; he obtains an effect of surprise when the reader
has understood, but it is not the main purpose of a character in a
novel to be an enigma. 70

This analysis, we see, assumes the resolution of a question that
has not been totally resolved, since Octave's impotence is not
exactly a datum in the text, but never mind that here: let us take
the example with its hypothesis. This analysis also includes
some opinions that I will avoid adopting as my own. But it has
the merit of describing well the phenomenon — which, of course,
is not exclusive to Stendhal. Apropos of what he calls the "in-
termingling of the two systems," Barthes rightly mentions the
"cheating" that, in Agatha Christie, consists of focalizing a nar-
rative like The Sittaford Mystery or The Murder of Roger Ackroyd
through the murderer while omitting from his "thoughts" sim-
ply the memory of the murder; 71 and we know that the most

69 See my Figures 11, pp. 183-185.

70 Temps et roman, p. 90.

71 Barthes, "An Introduction to the Structural Analysis of Narrative," p. 263.



classical detective story, although generally focalized through
the investigating detective, most often hides from us a part
his discoveries and inductions until the final revelation % ��

The inverse alteration, the excess of information or paralepsis
can consist of an inroad into the consciousness of a character in
the course of a narrative generally conducted in external focali-
zation. We can take to be such, at the beginning of the Peau de
chagrin, statements like "the young man did not understand his
ruin" or "he feigned the manner of an Englishman, "" which
contrast with the very distinct course of external vision adopted
until then, and which begin a gradual transition to internal
focalization. Paralepsis can likewise consist, in internal focaliza-
tion, of incidental information about the thoughts of a character
other than the focal character, or about a scene that the latter is
not able to see. We will describe as such the passage in Maisie
devoted to Mrs. Farange's thoughts, which Maisie cannot know:
"The day was at hand, and she saw it, when she should feel
more delight in hurling Maisie at [her father] than in snatching
her away." 74

A final general comment before returning to Proustian narra-
tive: we should not confuse the information given by a focalized
narrative with the interpretation the reader is called on to give of
it (or that he gives without being invited to). It has often been
noted that Maisie sees or hears things that she does not under-
stand but that the reader will decipher with no trouble. The eyes
"opened wide with attention" of Charlus looking at Marcel at
Balbec can, for the informed reader, be a sign, which completely
escapes the hero, like the whole of the Baron's behavior with
respect to him up to Sodome I. Bertil Romberg analyzes the case
of a novel by J. P. Marquand, H. M. Pulham, Esquire, where the
narrator, a trusting husband, is present at scenes between his
wife and a male friend that he recounts without thinking any-
thing amiss but whose meaning cannot escape the least subtle

. Another unmistakable paralipsis, in Michel Stwgoff: starting with Part 11

chapter 6, Jules Verne conceals from us what the hero knows very well viz that
ne was not blinded by Ogareff's incandescent sword.

73 Gamier, p. 10.

74 Henry James, What Maisie Knew (New York: Scribner's, 1908), p. 19.


Narrative Discourse

reader. 75 This excess of implicit information over explicit infor-
mation is the basis of the whole play of what Barthes calls indi-
ces, 76 which functions just as well in external focalization: in
"Hills Like White Elephants," Hemingway reports the conversa-
tion between his two characters while fully abstaining from in-
terpreting it; so here it is as if the narrator, like Marquand's hero,
did not understand what he relates; this in no way prevents the
reader from interpreting it in conformity with the author's inten-
tions, as each time a novelist writes "he felt a cold sweat run
down his back" we unhesitatingly construe "he was afraid."
Narrative always says less than it knows, but it often makes
known more than it says.


Let us repeat it again: use of the "first person," or better yet,
oneness of person of the narrator and the hero, 77 does not at all
imply that the narrative is focalized through the hero. Very
much to the contrary, the "autobiographical" type of narrator,
whether we are dealing with a real or a fictive autobiography,
is — by the very fact of his oneness with the hero — more "natur-
ally" authorized to speak in his own name than is the narrator of
a "third-person" narrative. There is less indiscretion from Tris-
tram Shandy in mixing the account of his present "opinions"
(and thus of his knowledge) with the narrative of his past "life"
than there is on Fielding's part in mixing the account of his with
the narrative of the life of Tom Jones. The impersonal narrative
therefore tends toward internal focalization by the simple trend
(if it is one) toward discretion and respect for what Sartre would
call the "freedom" — in other words, the ignorance — of its
characters. The autobiographical narrator, having no obligation
of discretion with respect to himself, does not have this kind of
reason to impose silence on himself. The only focalization that
he has to respect is defined in connection with his present in-

75 Romberg, p. 119.

76 Barthes, "An Introduction to the Structural Analysis of Narrative," p. 247.

77 Or (as we will see in the following chapter) of the narrator and an observer
of the Watson type.



formation as narrator and not in connection with his past infor-
mation as hero. 78 He can, if he wants, choose this second form
of focalization (focalization through the hero), but he is not at all
required to; and we could just as well consider this choice, when
it is made, as a paralipsis, since the narrator, in order to limit
himself to the information held by the hero at the moment of the
action, has to suppress all the information he acquired later,
information which very often is vital.

It is obvious (and we have already seen one example) that
Proust to a great extent imposed that hyperbolic restriction on
himself, and that the narrative mood of the Recherche is very
often internal focalization through the hero. 79 In general it is the
"hero's point of view" that governs the narrative, with his re-
strictions of field, his momentary ignorances, and even what the
narrator inwardly looks on as his youthful errors, naivetes, "il-
lusions to lose." In a famous letter to Jacques Riviere, Proust
insisted on his carefulness in dissimulating what was at the back
of his mind (identified here with the mind of Marcel-narrator)
up until the moment of the final revelation. The apparent mean-
ing of the final pages of Swarm (which, we must remember, tell
an experience in principle very recent) is, he says forcefully,

the opposite of my conclusion. It is a stage, seemingly subjective
and amateurish, on the way to the most objective and non-foolish
conclusion. If one inferred from it that my meaning is a disil-
lusioned skepticism, that would absolutely be as if a viewer, at the
end of the first act of Parsifal, after seeing a character understand
nothing of the ceremony and be chased off by Gurnemanz, as-
sumed that Wagner meant that simplicity of heart leads to noth-

78 Of course, this distinction is relevant only for the classical form of autobio-
graphical narrative, where the narrating is enough subsequent to the events for
the narrator's information to differ appreciably from the hero's. When the narrat-
ing is contemporaneous with the story (interior monologue, journal, corre-
spondence), internal focalization on the narrator amounts to focalization on the
hero. J. Rousset shows this well for the epistolary novel (Forme et signification, p.
70). We will come back to this point in the following chapter.

79 We know that he was interested in the Jamesian technique of point of view,
and especially the technique in Maisie (Walter Berry, N.R.F., hommave a Marcel
Proust [Paris: Gallimard, 1927], p. 73).


Narrative Discourse

Similarly, the experience of the madeleine (it too, however, is
recent) is reported in Sivann, but not explained, since the pro-
found reason for the pleasure of the reminiscence is not dis-
closed: "I will not explain it until the end of the third volume."
For the moment, one must respect the hero's ignorance, and
deal carefully with the evolution of his thought and the slow
work of vocation.

But this evolution of a thought, 1 did not want to analyze it
abstractly but to recreate it, make it live. So I am forced to paint the
mistakes, without thinking I have to say that I take them for mis-
takes; too bad for me if the reader thinks I take them for the truth.
The second volume will accentuate this misunderstanding. I hope
that the last will dissipate it. 80

We know that the last did not dissipate all of it. This is the
obvious risk of focalization, a risk that Stendhal pretended to
insure himself against by means of notes on the bottom of the
page: "It is the opinion of the hero, who is mad and will re-

It is obviously with respect to the main point — that is, with
respect to the experience of involuntary memory, and the liter-
ary vocation connected to it — that Proust was most careful in
handling the focalization, forbidding himself to give any prema-
ture sign, any indiscreet encouragement. The "proofs" of Mar-
cel's inability to write, of his incurable dilettantism, of his grow-
ing distaste for literature, do not stop accumulating until the
dramatic peripeteia in the courtyard of the Guermantes
townhouse — all the more dramatic since the suspense has been
built up for a long time by a focalization that on this point was
very rigorous. But the principle of nonintervention bears on
many other subjects — like homosexuality, for example, which,
despite the premonitory scene of Montjouvain, will remain for
the reader as for the hero, until the opening pages of Sodome, a
continent one-hundred times met but never recognized.

The most massive investment in this narrative course (that is,

80 Choix de lettres, ed. Philip Kolb (Paris, 1965), 7 February 1914, pp. 197-199.



focalization through the hero) is undoubtedly the handling of
the amorous relationships of the hero, and also of that second-
degree hero, Swann, in Un amour de Sivann. Here internal focali-
zation recovers the psychological function that the Abbe Prevost
had given it in Manon Lescaut: systematically adopting the
"point of view" of one of the protagonists permits an author to
leave the feelings of the other one almost completely in shadow,
and thus to construct for that other, at little cost, a mysterious
and ambiguous personality, the very one for which Proust will
coin the name "creature in flight" (fugitive). We do not know, at
each state of their passion, any more than Swann or Marcel
knows about the inner "truth" of an Odette, a Gilberte, an Al-
bertine, and nothing could more effectively illustrate the essen-
tial subjectivity of love according to Proust than that constant
evanescence of its object: the creature in flight is by definition
the creature loved. 81 Let us not take up again here the list (al-
ready evoked apropos of analepses with a corrective function) of
episodes (first meeting with Gilberte, false confession of Alber-
tine, incident of the syringas, etc.) whose real significance will
not be discovered by the hero — and with him by the reader—
until long after.

To these temporary ignorances or misunderstandings we
must add some points of definitive opaqueness, where the
perspectives of hero and narrator coincide; for instance, we will
never know what Odette's "true" feelings for Swann were, or
Albertine's for Marcel. A passage in the Jeunes Filles en fleurs
illustrates well this somewhat interrogative attitude of the narra-
tive in the face of those impenetrable creatures, when Marcel,
dismissed by Albertine, wonders for what reason the girl could
possibly have refused him a kiss after a series of such clear

of her attitude during that scene I could not arrive at any satisfac-
tory explanation. Taking first of all the supposition that she was
absolutely chaste (a supposition with which I had originally ac-
counted for the violence with which Albertine had refused to let

' On Marcel's ignorance with respect to Albertine, see Tadie, pp. 40^12.


Narrative Discourse

herself be taken in my arms and kissed, though it was by no means
essential to my conception of the goodness, the fundamentally
honourable character of my friend), I could not accept it without a
copious revision of its terms. It ran so entirely counter to the
hypothesis which 1 had constructed that day when I saw Albertine
for the first time. Then ever so many different acts, all acts of
kindness towards myself (a kindness that was caressing, at times
uneasy, alarmed, jealous of my predilection for Andree) came up
on all sides to challenge the brutal gesture with which, to escape
from me, she had pulled the bell. Why then had she invited me to
come and spend the evening by her bedside? Why had she spoken
all the time in the language of affection? What object is there in
your desire to see a friend, in your fear that he is fonder of another
of your friends than of you; why seek to give him pleasure, why
tell him, so romantically, that the others will never know that he
has spent the evening in your room, if you refuse him so simple a
pleasure and if to you it is no pleasure at all? I could not believe, all
the same, that Albertine's chastity was carried to such a pitch as
that, and I had begun to ask myself whether her violence might
not have been due to some reason of coquetry, a disagreeable
odour, for instance, which she suspected of lingering about her
person, and by which she was afraid that I might be disgusted, or
else of cowardice, if for instance she imagined, in her ignorance of
the facts of love, that my state of nervous exhaustion was due to
something contagious, communicable to her in a kiss. 82

Again, we must interpret as indices of focalization those open-
ings onto the psychology of characters other than the hero
which the narrative takes care to make in a more or less
hypothetical form, as when Marcel guesses or conjectures the
thought of his interlocutor according to the expression on that
person's face:

I could see in Cottard's eyes, as uneasy as though he were afraid of
missing a train, that he was asking himself whether he had not
allowed his natural good-humour to appear. He was trying to
think whether he had remembered to put on his mask of coldness,
as one looks for a mirror to see whether one has not forgotten to tie

82 RH I, 703-704/P I, 940-941.



one's tie. In his uncertainty, and, so as, whatever he had done, to
put things right, he replied brutally. 83

Since Spitzer, 84 critics have often noted the frequency of those
modalizing locutions (perhaps, undoubtedly, as if, seem, appear)
that allow the narrator to say hypothetically what he could not
assert without stepping outside internal focalization; and thus
Marcel Muller is not wrong in looking on them as "the alibis of
the novelist" 85 imposing his truth under a somewhat hypocriti-
cal cover, beyond all the uncertainties of the hero and perhaps
also of the narrator. For here again the narrator to some extent
shares the hero's ignorance; or, more exactly, the ambiguity of
the text does not allow us to decide whether the perhaps is an
effect of indirect style — and, thus, whether the hesitation it de-
notes is the hero's alone. Further, we must note that the often
multiple nature of these hypotheses much weakens their func-
tion as unavowed paralepsis, while at the same time it accen-
tuates their role as indicators of focalization. When the narrative
offers us, introduced by three perhaps's, three explanations to
choose from for the brutality with which Charlus answers Mme.
de Gallardon, 86 or when the silence of the elevator operator at
Balbec is ascribed with no preference to eight possible causes, 87
we are not in fact any more "informed" than when Marcel ques-
tions himself before us on the reasons for Albertine's refusal.
And here we can hardly go along with Muller, who reproaches
Proust for replacing "the secret of each creature with a series of
little secrets": Proust, in giving the idea that the real motive is
necessarily found among those he enumerates, thus suggests,

83 RH I, 381/P I, 498. Cf. an analogous scene with Norpois, RH I, 367/P I,

84 Leo Spitzer, "Zum Stil Marcel Prousts," in Stilstudien (Munich, 1928); trans,
in Etudes de style (Paris, 1970), pp. 453^155.

85 Voix narratives, p. 129.

86 RH II, 41/P II, 653.

87 "He vouchsafed no answer, whether from astonishment at my words,
preoccupation with what he was doing, regard for convention, hardness of
hearing, respect for holy ground, fear of danger, slowness of understanding, or
by the manager's orders" (RH I, 505/P I, 665).


Narrative Discourse

according to Muller, that "the behavior of a character is always
amenable to a rational explanation." 88 The multiplicity of con-
tradictory hypotheses suggests much more the insolubility of
the problem, and at the very least the incapacity of the narrator
to resolve it.

We have already noted the highly subjective nature of Proustian
descriptions, always bound to a perceptual activity of the
hero's. 89 Proustian descriptions are rigorously focalized: not only
does their "duration" never exceed that of real contemplation,
but their content never exceeds what is actually perceived by the
contemplator. Let us not come back to this subject, which is well
understood; 90 let us simply recall the symbolic importance in the
Recherche of scenes to which the hero, through an often mirac-
ulous chance, comes unexpectedly, and of which he perceives
only one part, and whose visual or auditory restriction the nar-
rative scrupulously respects: Swann in front of the window
which he takes for Odette's, able to see nothing between the
"slanting bars of the shutters," but only to hear, "in the silence
of the night, the murmur of conversation"; 91 Marcel at Mont-
jouvain, witness through the window of the scene between the
two young women but unable to make out Mile. Vinteuil's look
or hear what her friend murmurs in her ear, and for whom the
scene will stop when she comes, seeming "weary, awkward,
preoccupied, sincere, and rather sad," to close the shutters and
the window; 92 Marcel again, spying from the top of the stair-
case, then from the neighboring shop, on the "conjunction" of
Charlus and Jupien, the second part of which will be reduced
for him to a purely auditory perception; 93 Marcel still, coming
unexpectedly on Charlus's flagellation in Jupien's male bordello
via a "small oval window opening onto the corridor." 94 Critics


'' Voix narratives, p. 128.

1 Pp. 99-106.

' On the "perspectivism" of Proustian description, see Raimond, pp. 338-

1 RH 1, 209-21 1/P 1, 272-275.
'- RH 1, 122-125/P 1, 159-163.
' RH II, 8-9/P II, 609-610.
1 RH II, 959/PIII, 815.



generally insist, and rightly so, on the unlikelihood of these sit-
uations, 95 and on the hidden strain they inflict on the principle
of point of view; but we should first recognize that here, as in
any fraud, there is an implicit recognition and confirmation of
the code: these acrobatic indiscretions, with their so marked re-
strictions of field, attest to the difficulty the hero experiences in
satisfying his curiosity and in penetrating into the existence of
another. Thus they are to be set down to internal focalization.
As we have already had occasion to note, the observance of
this code goes sometimes so far as to become that form of hyper-
restriction of field that we call paralipsis: the end of Marcel's
passion for the Duchess, Swann's death, the episode of the little
girl-cousin at Combray have provided us with some examples. It
is true that the existence of these paralipses is known to us only
by the disclosure made later by the narrator — is made known,
thus, by an intervention that, for its part, would be due to par-
alepsis if we considered focalization through the hero to be what
the autobiographical form requires. But we have already seen
that this is not so, and that that very widespread idea follows
simply from an equally widespread confusion between the two.
The only focalization logically implied by the "first-person" nar-
rative is focalization through the narrator, and we shall see that
in the Recherche this second narrative mood coexists with the first.

One obvious manifestation of this new perspective is the ad-
vance notices we met in the chapter on order. When it is said,
apropos of the scene at Montjouvain, that later this scene will
exert a decisive influence on the hero's life, such notification
cannot be the hero's doing, but must of course be the narrator's —
like, more generally, all forms of prolepsis, which (except for an
intervention of the supernatural, as in prophetic dreams) always
exceed a hero's capacities for knowledge. Likewise, complemen-

95 Beginning with Proust himself, clearly anxious to forestall criticism (and to
divert suspicion): "Certainly, the affairs of this sort of which I have been a
spectator have always been presented in a setting of the most imprudent and
least probable character, as if such revelations were to be the reward of an action
full of risk, though in part clandestine" (RH II, 8/P II, 608).


Narrative Discourse

tary information introduced by locutions of the type "I have
learned since . . ." 96 — which belongs to the subsequent experience
of the hero, in other words, to the experience of the narrator —
arises from anticipation. It is not correct to set such interventions
down to the "omniscient narrator": 97 they represent simply the
autobiographical narrator's share in the report of facts still un-
known to the hero, but the narrator does not think himself
obliged on that account to put off mentioning them until the hero
should have acquired knowledge of them. Between the informa-
tion of the hero and the omniscience of the novelist is the infor-
mation of the narrator, who disposes of it according to his own
lights and holds it back only when he sees a precise reason for
doing so. The critic can contest the opportuneness of these
complements of information, but not their legitimacy or their
credibility in a narrative whose form is autobiographical.

Further, we must certainly recognize that this holds true not
only for prolepses giving explicit and avowed information. Even
Marcel Muller notes that a formula like "I was ignorant
that . . ," 98 — a real defiance of focalization through the hero — "can
mean J have learned since, and with these two 7's we would un-
questionably be kept on the Protagonist's plane. The ambiguity
is frequent," he adds, "and the choice between Novelist and Nar-
rator for the attribution of a given item of information is often
arbitrary." 99 It seems to me that methodological soundness here
forces us, at least for a preliminary period, to attribute to the
(omniscient) "Novelist" only what we really cannot attribute to
the narrator. We see in this case that a certain amount of infor-
mation which Muller attributes to the "novelist who can walk
through walls" 100 can be ascribed without prejudice to the later

•>" RH 1, 148/P I, 193; RH I, 1057/P II, 475; RH I, 1129/P II, 579; RH II, 290/P II,
1009; RH II, 506/P III, 182; RH II, 607/P III, 326; RH II, 995/P III, 864, etc. It is
different for information of the type I had been told that ... (as for Un amour de
Swann), which is one of the hero's modes of knowledge (by hearsay).

97 As Muller has correctly observed: "We are of course leaving aside the
cases — fairly numerous — where the Narrator anticipates what is still the hero's
future by drawing from what his own (the Narrator's) past is. In such cases there
is no question of the Novelist's omniscience" (Voix narratives, p. 110).

98 RH I, 1111/P II, 554; RH II, 288/P II, 1006.

99 Voix narratives, pp. 140-141.

100 Voix narratives, p. 110.



knowledge of the Protagonist: for instance, Charlus's visits to
Brichot's class, or the scene that unfolds at Berma's while Marcel
attends the Guermantes matinee, or even the dialogue between
the relatives on the evening of Swann's visit, if indeed the hero
really could not hear it at the time. 101 Similarly, many details
about the relations between Charlus and Morel can in one way
or another have come to the narrator's knowledge. 102 The same
hypothesis holds for Basin's infidelities, his conversion to
Dreyfusism, his late liaison with Odette, for M. Nissim Ber-
nard's unhappy love affairs, etc. 103 — so many indiscretions and
so much gossip, whether true or false, are not at all improbable
in the Proustian universe. Let us remember finally that it is to a
tale of this kind that the hero's knowledge of the past love be-
tween Swann and Odette is attributed, a knowledge so precise
that the narrator thinks he has to make excuses for it in a way
that may seem rather clumsy, 104 and that furthermore does not
spare the only hypothesis capable of accounting for the focaliza-
tion through Swann in this narrative within the narrative:
namely, that whatever the eventual way stations, the first
source can only have been Swann himself.

The real difficulty arises when the narrative reports to us, on
the spot and with no perceptible detour, the thoughts of another
character in the course of a scene where the hero himself is
present: Mme. de Cambremer at the Opera, the usher at the
Guermantes soiree, the historian of the Fronde or the librarian at
the Villeparisis matinee, Basin or Breaute in the course of the
dinner at Oriane's. 105 In the same way we have access, without
any apparent way station, to Swann's feelings about his wife or
to Saint-Loup's about Rachel, 106 and even to the last thoughts of

1111 RH II, 583/P III, 291-292; RH II, 1098-1101/P III, 995-999; RH I, 26-27/P I, 35.

102 Including the risque scene of the Maineville bordello, the account of which
is vouched for (RH II, 343/P II, 1082).

101 RH II, 101/P II, 739; RH II, 111 3-11 15/P III, 1015-1018; RH II, 182/P II,

104 RH I, 143/P I, 186.

105 RH I, 753-754/P II, 56-57; RH II, 29/P II, 636; RH I, 869/P II, 215; RH I, 893/P
II, 248; RH I, 1090/P II, 524; RH I, 1 025-1 026/P II, 429-430.

106 RH I, 398-401/P I, 522-525; RH I, 801/P II, 122; RH I, 826/P II, 156; RH I,
830-831/P II, 162-163.


Narrative Discourse

Bergotte on his deathbed, 107 which, as has often been noted,
cannot in point of fact have been reported to Marcel since no
one — for a very good reason — could have knowledge of them.
That is one paralepsis to end all paralepses; it is irreducible by any
hypothesis to the narrator's information, and one we must in-
deed attribute to the "omniscient" novelist — and one that would
be enough to prove Proust capable of transgressing the limits of
his own narrative "system."

But evidently we cannot restrict the part played by paralepsis
to this scene alone, on the pretext that this is the only one to
present a physical impossibility. The decisive criterion is not so
much material possibility or even psychological plausibility as it
is textual coherence and narrative tonality. Thus, Michel
Raimond attributes to the omniscient novelist the scene during
which Charlus takes Cottard into a nearby room and talks to him
without witnesses. 108 In principle nothing prohibits us from as-
suming that this dialogue, like others, 109 was reported to Marcel
by Cottard himself, but nonetheless the reading of this passage
gives the idea of an immediate narrating without way stations,
and the same is true for all those that I mentioned in the preced-
ing paragraph, and for some others as well. In all these Proust
manifestly forgets or neglects the fiction of the autobiographical
narrator and the focalization which that implies — and a fortiori
the focalization through the hero that is its hyperbolic form — in
order to handle his narrative in a third mood, which is obviously
zero-focalization, in other words, the omniscience of the classi-
cal novelist. Which, let us note in passing, would be impossible if
the Recherche were — as some people still want to see it — a true
autobiography. Whence these scenes — scandalous, I would
imagine, for the purists of "point of view" — where J and others
are handled on the same footing, as if the narrator had exactly
the same relationship to a Cambremer, a Basin, a Breaute, and
his own past "me": "Mme. de Cambremer remembered having
heard Swann say . . . / For myself, the thought of the two

1117 RH II, 509/P III, 187.

'<> 8 RH II, 335-336/P II, 1071-1072. Raimond, p.


lew p or example, the conversation between the Verdurins about Saniette, RH
II, 607/P III, 326.



cousins . . . / Mme. de Cambremer was trying to make out
exactly how. . . / For my own part, I never doubted. "no
Plainly such a text is constructed on the antithesis between
Mme. de Cambremer's thoughts and Marcel's, as if somewhere
there existed a point from which my thought and someone else's
thought would seem symmetrical to me — the height of deper-
sonalization, which unsettles a little the image of the famous
Proustian subjectivism. Whence further that scene at
Montjouvain, in which we have already noted the very rigorous
focalization (through Marcel) with respect to visible and audible
actions, but which for thoughts and feelings, on the other hand,
is entirely focalized through Mile. Vinteuil: 111 "she felt. . . she
thought . . . she felt that she had been indiscreet, her sensitive
heart took fright . . . she pretended . . . she guessed . . . she
realised . . . " — as if the witness could neither see all nor hear all,
and nevertheless divined all the thoughts. But the truth quite
obviously is that two concurrent codes are functioning here on
two planes of reality which oppose each other without colliding.
This double focalization 112 certainly corresponds to the antithe-
sis organizing the entire passage (like the entire character of Mile.
Vinteuil, "shy maiden" and "battered old campaigner"), an
antithesis between the brutal immorality of the actions (per-
ceived by the hero-witness) and the extreme delicacy of the
feelings, which only an omniscient narrator, capable like God
himself of seeing beyond actions and of sounding body and
soul, can reveal. 113 But this scarcely conceivable coexistence can
serve as an emblem of the whole of Proust's narrative practice,

110 RH I, 754/PII, 57.

111 With the exception of one sentence (RH 1, 125/P 1, 163) focalized through her
friend, a "probably" (RH I, 123/P I, 161), and a "may well have" (RH I, 125/P I,
162). [Translator's note: partly my translation.]

112 B. G. Rogers, Proust's Narrative Techniques (Geneva, 1965), p. 108, speaks of
"double vision" apropos of the concurrence between the "subjective" hero and
the "objective" narrator.

113 On the technical and psychological aspects of this scene, see Muller's
excellent commentary (pp. 148-153), which, in particular, points out well how the
hero's mother and grandmother are indirectly but closely implicated in this act of
filial "sadism," whose personal resonances in Proust are immense and which
obviously recalls the "Confession d'une jeune fille" of Les Plaisirs et les jours, and
the "Sentiments filiaux d'un parricide."

F ;


Narrative Discourse



which plays without a qualm, and as if without being aware of
it, in three modes of focalization at once, passing at will from the
consciousness of his hero to that of his narrator, and inhabiting
by turns that of his most diverse characters. This triple narrative
position is not at all comparable to the simple omniscience of the
classical novel, for it not only defies, as Sartre reproached
Mauriac for defying, the conditions of the realistic illusion: it
also transgresses a "law of the spirit" requiring that one cannot
be inside and outside at the same time. To resume the musical
metaphor used above, we could say that between a tonal (or
modal) system with respect to which all infractions (paralipses
and paralepses) can be defined as alterations, and an atonal
(amodal?) system where no code prevails anymore and where
the very notion of infraction becomes outworn, the Recherche
illustrates quite well an intermediary state: a plural state, com-
parable to the polytonal (polymodal) system ushered in for a
time, and in the very same year, 1913, by the Rite of Spring. One
should not take this comparison too literally; 114 let it at least
serve us to throw light on this typical and very troubling feature
of Proustian narrative, which we would like to call its polymodal-

(the height of dialogic mimesis) but finally absorbing the charac-
ters in an immense verbal game (the height of literary
gratuitousness, the antithesis of realism); and, finally, the con-
currence of theoretically incompatible focalizations, which
shakes the whole logic of narrative representation. Again and
again we have seen this subversion of mood tied to the activity,
or rather the presence, of the narrator himself, the disturbing
intervention of the narrative source — of the narrating in the nar-
rative. It is this last instance — that of voice — which we must now
look at for its own sake, after having met it so often without
wanting to.

Let us recall to finish this chapter that this ambiguous — or
rather, complex — and deliberately nonorganized position
characterizes not only the system of focalization but the entire
modal practice of the Recherche: at the level of the narrative of
actions, the paradoxical coexistence of the greatest mimetic in-
tensity and the presence of a narrator, which is in principle
contrary to novelistic mimesis; the dominance of direct dis-
course, intensified by the stylistic autonomy of the characters

n4 yye know (George Painter, Proust: The Later Years [New York: Atlantic-Little,
Brown, 1965], pp. 340-342) what a fiasco the meeting arranged in May 1922 be-
tween Proust and Stravinsky (and Joyce) was. We could just as well draw a parallel
between Proustian narrative practice and those multiple and superimposed
visions so well expressed, still in the same period, by Cubism. Is it that kind of
portrait that these lines from the preface toPropos de peintre refer to: "the admira-
ble Picasso, who, in fact, has concentrated all Cocteau's features in a portrait of
such noble rigidity ..." {Essaiset articles, Pleiade, p. 580; "Preface to Jacques Emile
Blanche's Propos de Peintre: De David a Degas," in Proust: A Selection, p. 253)?

5 Voice

The Narrating Instance

"For a long time I used to go to bed early": obviously, such a
statement— unlike, let us say, "Water boils at one-hundred de-
grees Celsius" or "The sum of the angles of a triangle is equal to
two right angles"— can be interpreted only with respect to the
person who utters it and the situation in which he utters it. I is
identifiable only with reference to that person, and the com-
pleted past of the "action" told is completed only in relation to
the moment of utterance. To use Benveniste's well-known terms
again, the story here is not without a share of discourse, and it is
not too difficult to show that this is practically always the case. 1
Even historical narrative of the type "Napoleon died at Saint
Helena" implies in its preterite that the story precedes the nar-
rating, and I am not certain that the present tense in "Water
boils at one-hundred degrees" (iterative narrative) is as atem-
poral as it seems. Nevertheless, the importance or the relevance
of these implications is essentially variable, and this variability
can justify or impose distinctions and contrasts that have at least
an operative value. When I read Gambara or he Chef-d'oeuvre
inconnu, I am interested in a story, and care little to know who
tells it, where, and when; if I read Facino Cane, at no time can I
overlook the presence of the narrator in the story he tells; if it is

1 On this subject see my Figures 11, pp. 61-69.



La Maison Nucingen, the author makes it his business to draw my
attention to the person of the talker Bixiou and the group of
listeners he addresses; if it is UAuberge rouge, I will undoubtedly
give less attention to the foreseeable unfolding of the story
Hermann tells than to the reactions of a listener named Taillefer,
for the narrative is on two levels, and the second — where someone
narrates — is where most of the drama's excitement is.

This kind of effect is what we are going to look at under the
category of voice: "the mode of action," says Vendryes, "of the
verb considered for its relations to the subject" — the subject here
being not only the person who carries out or submits to the
action, but also the person (the same one or another) who re-
ports it, and, if need be, all those people who participate, even
though passively, in this narrating activity. We know that lin-
guistics has taken its time in addressing the task of accounting
for what Benveniste has called subjectivity in language, 2 that is,
in passing from analysis of statements to analysis of relations
between these statements and their generating instance — what
today we call their enunciating. It seems that poetics is experienc-
ing a comparable difficulty in approaching the generating in-
stance of narrative discourse, an instance for which we have
reserved the parallel term narrating. This difficulty is shown
especially by a sort of hesitation, no doubt an unconscious
one, to recognize and respect the autonomy of that instance,
or even simply its specificity. On the one hand, as we have
already noted, critics restrict questions of narrative enunciating
to questions of "point of view"; on the other hand they
identify the narrating instance with the instance of "writing,"
the narrator with the author, and the recipient of the narrative
with the reader of the work: 3 a confusion that is perhaps legiti-
mate in the case of a historical narrative or a real autobiography,
but not when we are dealing with a narrative of fiction, where
the role of narrator is itself fictive, even if assumed directly by
the author, and where the supposed narrating situation can be
very different from the act of writing (or of dictating) which

2 Benveniste, "Subjectivity in Language," Problems, pp. 223-230.

3 For example Todorov, "Les Categories du recit litteraire," pp. 146-147.


Narrative Discourse

refers to it. It is not the Abbe Prevost who tells the love of
Manon and Des Grieux, it is not even the Marquis de Renon-
court, supposed author of the Memoires d'un homme de qualite; it
is Des Grieux himself, in an oral narrative where "I" can desig-
nate only him, and where "here" and "now" refer to the spatio-
temporal circumstances of that narrating and in no way to the
circumstances of the writing of Manon Lescaut by its real author.
And even the references in Tristram Shandy to the situation of
writing speak to the (fictive) act of Tristram and not the (real)
one of Sterne; but in a more subtle and also more radical way,
the narrator of Pere Goriot "is" not Balzac, even if here and there
he expresses Balzac's opinions, for this author-narrator is some-
one who "knows" the Vauquer boardinghouse, its landlady and
its lodgers, whereas all Balzac himself does is imagine them; and
in this sense, of course, the narrating situation of a fictional
account is never reduced to its situation of writing.

So it is this narrating instance that we have still to look at,
according to the traces it has left— the traces it is considered to
have left — in the narrative discourse it is considered to have
produced. But it goes without saying that the instance does not
necessarily remain identical and invariable in the course of a
single narrative work. Most of Manon Lescaut is told by Des
Grieux, but some pages revert to M. de Renoncourt; inversely,
most of the Odyssey is told by "Homer," but Books IX-XII revert
to Ulysses; and the baroque novel, The Thousand and One Nights,
and Lord Jim have accustomed us to much more complex situa-
tions. 4 Narrative analysis must obviously take charge of the

4 On the Thousand and One Nights, see Todorov, "Narrative-Men," in Poetics of
Prose: "The record [for embedding] seems to be held by the narrative which
offers us the story of the bloody chest. Here
Scheherazade tells that
Jaafer tells that
the tailor tells that
the barber tells that
his brother (and he has six brothers) tells that . . .
The last story is a story to the fifth degree" (p. 71). But the term "embedding"
does not do justice to the fact precisely that each of these stories is at a
higher "degree" than the preceding one, since its narrator is a character in
the preceding one; for stories can also be "embedded" at the same level, simply
by digression, without any shift in the narrating instance: see Jacques's pa-
rentheses in the Fataliste.



study of these modifications — or of these permanences: for if it
is remarkable that Ulysses' adventures are told by two different
narrators, it is proper to find it just as noteworthy that the loves
of Swann and of Marcel are told by the same narrator.

A narrating situation is, like any other, a complex whole
within which analysis, or simply description, cannot differentiate
except by ripping apart a tight web of connections among the
narrating act, its protagonists, its spatio-temporal determina-
tions, its relationship to the other narrating situations involved
in the same narrative, etc. The demands of exposition constrain
us to this unavoidable violence simply by the fact that critical
discourse, like any other discourse, cannot say everything at
once. Here again, therefore, we will look successively at ele-
ments of definition whose actual functioning is simultaneous:
we will attach these elements, for the most part, to the
categories of time of the narrating, narrative level, and "person"
(that is, relations between the narrator — plus, should the occa-
sion arise, his or their narratee[s] s — and the story he tells).

Time of the Narrating

By a dissymmetry whose underlying reasons escape us but
which is inscribed in the very structures of language (or at the
very least of the main "languages of civilization" of Western
culture), I can very well tell a story without specifying the place
where it happens, and whether this place is more or less distant
from the place where I am telling it; nevertheless, it is almost
impossible for me not to locate the story in time with respect to
my narrating act, since I must necessarily tell the story in a
present, past, or future tense. 6 This is perhaps why the temporal
determinations of the narrating instance are manifestly more
important than its spatial determinations. With the exception of

5 This is what I will call the receiver of the narrative, patterned after the
contrast between sender and receiver proposed by A. J. Greimas (Sernantique
structural [Paris, 19661, p- 177).

6 Certain uses of the present tense do indeed connote temporal indefiniteness
(and not simultaneousness between story and narrating), but curiously they
seem reserved for very particular forms of narrative (joke, riddle, scientific prob-
lem or experiment, plot summary) and literature does not have much invest-
ment in them. The case of the "narrative present" with preterite value is also


Narrative Discourse

second-degree narratings, whose setting is generally indicated
by the diegetic context (Ulysses with the Phaeacians, the land-
lady of Jacques le fataliste in her inn), the narrating place is very
rarely specified, and is almost never relevant: 7 we know more or
less where Proust wrote the Recherche du temps perdu, but we are
ignorant of where Marcel is considered to have produced the
narrative of his life, and we scarcely think of worrying about it.
On the other hand, it is very important to us to know, for exam-
ple, how much time elapses between the first scene of the Re-
cherche (the "drama of going to bed") and the moment when it is
evoked in these terms: "Many years have passed since that
night. The wall of the staircase, up which I had watched the
light of his candle gradually climb, was long ago demolished";
for this temporal interval, and what fills it up and gives it life, is
an essential element in the narrative's significance.

The chief temporal determination of the narrating instance is
obviously its position relative to the story. It seems evident that
the narrating can only be subsequent to what it tells, but this
obviousness has been belied for many centuries by the existence
of "predictive" narrative 8 in its various forms (prophetic,
apocalyptic, oracular, astrological, chiromantic, cartomantic,
oneiromantic, etc.), whose origin is lost in the darkness of
time — and has been belied also, at least since Les Lauriers sont
coupes, by the use of narrative in the present tense. We must
consider, further, that a past-tense narrating can to some extent
be split up and inserted between the various moments of the
story, much like a "live" running commentary 9 — a common

7 It could be, but for reasons which are not exactly spatial in kind: for a
"first-person" narrative to be produced in prison, on a hospital bed, in a
psychiatric institution, can constitute a decisive element of advance notice about
the denouement.

8 1 borrow the term "predictive" from Todorov, Grammaire du Decameron (The
Hague, 1969), p. 48, to designate any kind of narrative where the narrating
precedes the story.

9 Radio or television reporting is obviously the most perfectly live form of this
kind of narrative, where the narrating follows so closely on the action that it can
be considered practically simultaneous, whence the use of the present tense. We
find a curious literary use of simultaneous narrative in chapter 29 of Ivanhoe,
where Rebecca is telling the wounded Ivanhoe all about the battle taking place at
the foot of the castle, a battle she is following from the window.



practice with correspondence and private diary, and therefore
with the "novel by letters" or the narrative in the form of a
journal (Wuthering Heights, Journal d'un cure de campagne). It is
therefore necessary, merely from the point of view of temporal
position, to differentiate four types of narrating: subsequent (the
classical position of the past-tense narrative, undoubtedly far
and away the most frequent); prior (predictive narrative, gener-
ally in the future tense, but not prohibited from being conju-
gated in the present, like Jocabel's dream in Moyse sauve); simul-
taneous (narrative in the present contemporaneous with the ac-
tion); and interpolated (between the moments of the action).

The last type is a priori the most complex, since it involves a
narrating with several instances, and since the story and the
narrating can become entangled in such a way that the latter has
an effect on the former. This is what happens particularly in the
epistolary novel with several correspondents, 10 where, as we
know, the letter is at the same time both a medium of the narra-
tive and an element in the plot. 11 This type of narrating can
also be the most delicate, indeed, the one most refractory to
analysis, as for example when the journal form loosens up to
result in a sort of monologue after the event, with an indefi-
nite, even incoherent, temporal position: attentive readers of
L'Etranger have not missed these uncertainties, which are one of
the audacities — perhaps unintentional — of that narrative. 12 Fi-
nally, the extreme closeness of story to narrating produces here,
most often, 13 a very subtle effect of friction (if I may call it that) be-
tween the slight temporal displacement of the narrative of
events ("Here is what happened to me today") and the complete
simultaneousness in the report of thoughts and feelings ("Here

10 On the typology of epistolary novels according to the number of corre-
spondents, see Rousset, "Une forme litteraire: le roman par lettres," Forme et
signification, and Romberg, Studies, pp. 51 ff.

11 An example is when, in Les Liaisons dangereuses, Mme. de Volanges discov-
ers Danceny's letters in her daughter's writing desk — a discovery whose conse-
quences Danceny is notified of in letter 62, typically "performative." Cf. To-
dorov, Litterature et signification (Paris, 1967), pp. 44-46.

12 See B. T. Fitch, Narrateur et narration dans "1'Etranger" a" Albert Camus, 2d rev.
ed. (Paris, 1968) pp. 12-26.

13 But there also exist delayed forms of journal narrating: for example, the "first
notebook" of the Symphonie pastorale, or the complex counterpoint of L'Emploi du


Narrative Discourse

is what I think about it this evening")- The journal and the
epistolary confidence constantly combine what in broadcasting
language is called the live and the prerecorded account, the
quasi-interior monologue and the account after the event. Here,
the narrator is at one and the same time still the hero and al-
ready someone else: the events of the day are already in the
past, and the "point of view" may have been modified since
then; the feelings of the evening or the next day are fully of the
present, and here focalization through the narrator is at the
same time focalization through the hero. Cecile Volanges writes
to Mme. de Merteuil to tell her how she was seduced, last night,
by Valmont, and to confide to her her remorse; the seduction
scene is past, and with it the confusion that Cecile no longer
feels, and can no longer even imagine; what remains is the
shame, and a sort of stupor which is both incomprehension and
discovery of oneself: "What I reproach myself for most, and
what, however, I must talk to you about, is that I am afraid I
didn't defend myself as much as I could have. I don't know how
that happened: surely I don't love M. de Valmont, very much
the opposite; and there were moments when I acted as if I did
love him. . . ," 14 The Cecile of yesterday, very near and already
far off, is seen and spoken of by the Cecile of today. We have here
two successive heroines, (only) the second of whom is (also) the
narrator and gives her point of view, the point of view — dis-
placed just enough to create dissonance — of the immediate post-
event future. 15 We know how the eighteenth-century novel, from
Pamela to Obermann, exploited that narrative situation propitious
to the most subtle and the most "irritating" counterpoints: the
situation of the tiniest temporal interval.

The third type (simultaneous narrating), by contrast, is in
principle the simplest, since the rigorous simultaneousness of
story and narrating eliminates any sort of interference or tem-
poral game. We must observe, however, that the blending of the
instances can function here in two opposite directions, accord-
ing to whether the emphasis is put on the story or on the narra-

14 Letter 97.

15 Compare letter 48, from Valmont to Tourvel, written in Emilie's bed, "live"
and, if I may say so, at the event.



tive discourse. A present-tense narrative which is "behaviorist"
in type and strictly of the moment can seem like the height of
objectivity, since the last trace of enunciating that still subsisted
in the Hemingway-style narrative (the mark of temporal interval
between story and narrating, which the use of the preterite un-
avoidably comprises) now disappears in a total transparency of
the narrative, which finally fades away in favor of the story.
That is how the works that come under the heading of the
French "new novel," and especially Robbe-Grillet's early
novels, 16 have generally been received: "objective literature,"
"school of the look" — these designations express well the sense
of the narrating's absolute transitivity which a generalized use of
the present tense promotes. But inversely, if the emphasis rests
on the narrating itself, as in narratives of "interior monologue,"
the simultaneousness operates in favor of the discourse; and
then it is the action that seems reduced to the condition of sim-
ple pretext, and ultimately abolished. This effect was already
noticeable in Dujardin, and became more marked in a Beckett, a
Claude Simon, a Roger Laporte. So it is as if use of the present
tense, bringing the instances together, had the effect of un-
balancing their equilibrium and allowing the whole of the narra-
tive to tip, according to the slightest shifting of emphasis, either
onto the side of the story or onto the side of the narrating, that
is, the discourse. And the facility with which the French novel in
recent years has passed from one extreme to the other perhaps
illustrates this ambivalence and reversibility. 17

The second type (prior narrating) has until now enjoyed a
much smaller literary investment than the others, and certainly
even novels of anticipation, from Wells to Bradbury — which
nevertheless belong fully to the prophetic genre — almost always
postdate their narrating instances, making them implicitly sub-
sequent to their stories (which indeed illustrates the autonomy
of this fictive instance with respect to the moment of actual

16 All written in the present tense except Le Voyeur, whose temporal system, as
we know, is more complex.

17 An even more striking illustration is La Jalousie, which can be read ad libitum
in the objectivist mode with no jealous person in the narrating, or purely as the
interior monologue of a husband spying on his wife and imagining her adven-
tures. Indeed, when this work was published in 1959 it played a pivotal role.


Narrative Discourse

writing). Predictive narrative hardly appears at all in the literary
corpus except on the second level: examples, in Saint- Amant's
Moyse sauve, are Aaron's prophetic narrative (sixth part) and
Jocabel's long premonitory dream (fourth, fifth, and sixth parts),
both of which are connected with Moses' future. 18 The common
characteristic of these second narratives is obviously that they
are predictive in relation to the immediate narrating instance
(Aaron, Jocabel's dream) but not in relation to the final instance
(the implied author of Moyse sauve, who explicitly identifies him-
self with Saint-Amant): clear examples of prediction after the

Subsequent narrating (the first type) is what presides over the
immense majority of the narratives produced to this day. The
use of a past tense is enough to make a narrative subsequent,
although without indicating the temporal interval which sep-
arates the moment of the narrating from the moment of the
story. 19 In classical "third-person" narrative, this interval ap-
pears generally indeterminate, and the question irrelevant, the
preterite marking a sort of ageless past: 20 the story can be dated,
as it often is in Balzac, without the narrating being so. 21 It some-
times happens, however, that a relative contemporaneity of
story time and narrating time is disclosed by the use of the
present tense, either at the beginning, as in Tom Jones 22 or Le

18 See my Figures 11, pp. 210-211.

19 With the exception of the passe compose, which in French connotes relative
closeness: "The perfect creates a living connection between the past event and
the present in which its evocation takes place. It is the tense for the one who
relates the facts as a witness, as a participant; it is thus also the tense that will be
chosen by whoever wishes to make the reported event ring vividly in our ears
and to link it to the present" (Benveniste, "The Correlations of Tense in the
French Verb," Problems, p. 210). L'Etranger, of course, owes a great deal to the
use of this tense.

20 Kate Hamburger (The Logic of Literature, trans. Marilynn J. Rose, 2d ed.
[Bloomington, Ind., 1973]) has gone so far as to deny any temporal value to the
"epic preterite." In this extreme and strongly contested position there is a certain
hyperbolic truth.

21 On the other hand, Stendhal does like to date, and more precisely to ante-
date, for reasons of political prudence, the narrating instance of his novels: Le
Rouge (written in 1829-1830) at 1827, La Chartreuse (written in 1839) at 1830.

22 "In that Part of the western Division of this Kingdom, which is commonly
called Somersetshire, there lately lived (and perhaps foes sfi/0 a Gentleman whose
Name was Allworthy" (Tom Jones, Book I, chap. 2 [Norton, p. 27]).



Pere Goriot, 23 or at the end, as in Eugenie Grandet 24 or Madame
Bovary. 2S These effects of final convergence (the more striking of
the two types) play on the fact that the very length of the story
gradually lessens the interval separating it from the moment of
the narrating. But the power of these final convergences results
from their unexpected disclosure of a temporal isotopy (which,
being temporal, is also to a certain extent diegetic) between the
story and its narrator, an isotopy which until then was hidden
(or, in the case of Bovary, long forgotten). In "first-person" nar-
rative, on the other hand, this isotopy is evident from the begin-
ning, where the narrator is presented right away as a character
in the story, and where the final convergence is the rule, 26 in
accordance with a mode that the last paragraph of Robinson
Crusoe can furnish us with a paradigm of: "And here, resolving
to harrass my self no more, I am preparing for a longer Journey
than all these, having liv'd 72 Years, a Life of infinite Variety,
and learn'd sufficiently to know the Value of Retirement, and
the Blessing of ending our Days in Peace." 27 No dramatic effect
here, unless the final situation should itself be a violent de-
nouement, as in Double Indemnity, in which the hero writes the
last line of his confession-narrative before slipping with his

23 "Madame Vauquer, whose maiden name was De Conflans, is an elderly
woman who for forty years has kept, in Paris, a family boardinghouse" (Pere
Goriot, trans. J. M. Sedgwick [New York: Rinehart, 1950J, p. 1).

24 "Her face is very pale and quiet now, and there is a tinge of sadness in the
low tones of her voice. She has simple manners" (Eugenie Grandet, trans. E.
Marriage [Philadelphia: Gebbie, 1899], p. 223).

25 "The devil himself doesn't have a greater following than [M. Homais]: the
authorities treat him considerately, and public opinion is on his side. He has just
been awarded the cross of the Legion of Honor" (Madame Bovary, trans. F. Steeg-
muller [New York: Random House, 1957], p. 396). Let us remember that the
opening pages ("We were in study-hall ..." [Steegmuller, p. 3]) already indicate
that the narrator is contemporary with the hero, and is even one of his fellow

26 The Spanish picaresque seems to form a notable exception to this "rule," at
any rate Lazarillo, which ends in suspense ("It was the time of my prosperity,
and I was at the height of all good fortune"). Guzman and Buscon also, but while
promising a continuation and end, which will not come.

27 Robinson Crusoe (Oxford: Blackwell, 1928), III, 220. Or, in a more ironic
mode, Gil Bias: "It is three years since then, my friend the reader, that I have
been leading a delightful life with such dear people. As a crowning satisfaction,
heaven was pleased to bestow on me two children, whose upbringing will
become the pastime of my old age, and whose father I dutifully think I am."


Narrative Discourse

accomplice into the ocean where a shark awaits them: "I didn't
hear the stateroom door open, but she's beside me now while
I'm writing. I can feel her. / The moon." 28

In order for the story to overtake the narrating in this way, the
duration of the latter must of course not exceed the duration of
the former. Take Tristram's comic aporia: in one year of writing
having succeeded in telling only the first day of his life, he
observes that he has gotten 364 days behind, that he has there-
fore moved backward rather than forward, and that, living 364
times faster than he writes, it follows that the more he writes the
more there remains for him to write; that, in short, his undertak-
ing is hopeless. 29 Faultless reasoning, whose premises are not at
all absurd. Telling takes time (Scheherazade's life hangs by that
one thread), and when a novelist puts on his stage an oral nar-
rating in the second degree, he rarely fails to take that into
account: many things happen at the inn while the landlady of
Jacques tells the story of the Marquis des Arcis, and the first part
of Manon Lescaut ends with the remark that since the Chevalier
spent more than an hour on his tale, he certainly needs supper
in order to "get a little rest." We have a few reasons to think that
Prevost, for his part, spent much more than an hour writing
those some one-hundred pages, and we know, for example,
that Flaubert needed almost five years to write Madame Bovary.
Nevertheless — and this is finally very odd — the fictive narrating
of that narrative, as with almost all the novels in the world
except Tristram Shandy, is considered to have no duration; or,
more exactly, everything takes place as if the question of its
duration had no relevance. One of the fictions of literary
narrating — perhaps the most powerful one, because it passes
unnoticed, so to speak — is that the narrating involves an instan-
taneous action, without a temporal dimension. Sometimes it is
dated, but it is never measured: we know that M. Homais has
just received the cross of the Legion of Honor at the moment
when the narrator writes that last sentence, but we do not know


'James M. Cain, Double Indemnity, in Cain X3 (New York: Knopf, 1969), p.
' Sterne, Tristram Shandy, Book IV, chap. 13.



what was happening while the narrator was writing his fir
one. Indeed, we even know that this question is absurd: nothi
is held to separate those two moments of the narrating instance
except the atemporal space of the narrative as text. Contrary to
simultaneous or interpolated narrating, which exist through
their duration and the relations between that duration and the
story's, subsequent narrating exists through this paradox: it
possesses at the same time a temporal situation (with respect to
the past story) and an atemporal essence (since it has no dura-
tion proper). 30 Like Proustian reminiscence, it is rapture, "a mo-
ment brief as a flash of lightning," a miraculous syncope, "a
minute freed from the order of [TJime." 31

The narrating instance of the Recherche obviously corresponds
to this last type. We know that Proust spent more than ten years
writing his novel, but Marcel's act of narrating bears no mark of
duration, or of division: it is instantaneous. The narrator's pres-
ent, which on almost every page we find mingled with the
hero's various pasts, is a single moment without progression.
Marcel Muller thought he found in Germaine Bree the hypoth-
esis of a double narrating instance — before and after the final
revelation — but this hypothesis has no basis, and in fact all I see
in Germaine Bree is an improper (although common) use of
"narrator" for hero, which perhaps led Muller into error on that
point. 32 As for the feelings expressed on the final pages oiSwann,
which we know do not correspond to the narrator's final con-
viction, Muller himself shows very well that they do not at all
prove the existence of a narrating instance prior to the revela-
tion; 33 on the contrary, the letter to Jacques Riviere quoted
above 34 shows that Proust was anxious to tune the narrator's

30 Temporal indications of the kind "we have already said" and "we will see
later," etc., do not in fact refer to the temporality of the narrating, but to the
space of the text ( = we have said above, we will see further on . . . ) and to the
temporality of reading.

31 RH II, 1001 and 1002/P III, 872 and 873.

"Muller, p. 45; Germaine Bree, Marcel Proust and Deliverance from Time, trans.
C. J. Richards and A. D. Truitt, 2d ed. (New Brunswick, N 1 1969), pp 19-20
"Muller, p. 46.
34 Pp. 199-200.


Narrative Discourse



discourse to the hero's "errors," and thus to impute to the nar-
rator a belief not his own, in order to avoid disclosing his own
mind too early. Even the narrative Marcel produces after the
Guermantes soiree, the narrative of his beginnings as a writer
(seclusion, rough drafts, first reactions of readers), which nec-
essarily takes into account the length of writing ("like him
too, ... I had something to write. But my task was longer than
his, my words had to reach more than a single person. My task
was long. By day, the most I could hope for was to try to sleep.
If 1 worked, it would be only at night. But I should need many
nights, a hundred perhaps, or even a thousand") 35 and the in-
terrupting fear of death — even this narrative does not gainsay
the fictive instantaneousness of its narrating: for the book Marcel
then begins to write in the story cannot legitimately be identified
with the one Marcel has then almost finished writing as nar-
rative — and which is the Recherche itself. Writing the fictive book,
which is the subject of the narrative, is, like writing every book, a
"task [that] was long." But the actual book, the narrative-book,
does not have knowledge of its own "length": it does away with
its own duration.

The present of Proustian narrating — from 1909 to 1922 —
corresponds to many of the "presents" of the writing, and we
know that almost a third of the book — including, as it happens,
the final pages — was written by 1913. The fictive moment of
narrating has thus in fact shifted in the course of the real writing;
today it is no longer what it was in 1913, at the moment when
Proust thought his work concluded for the Grasset edition.
Therefore, the temporal intervals he had in mind — and wanted
to signify — when he wrote, for example apropos of the bedtime
scene, "Many years have passed since that night," or apropos of
the resurrection of Combray by the madeleine, "I can measure
the resistance, I can hear the echo of great spaces traversed" —
these spaces have increased by more than ten years simply be-
cause the story's time has lengthened: the signified of these
sentences is no longer the same. Whence certain irreducible con-
tradictions like this one: the narrator's today is obviously, for us,

35 RH II, 1136/P1II, 1043.

later than the war, but the "Paris today" of the last pages of
Swann remains in its historical determinations (its referential
content) a prewar Paris, as it was seen and described in its better
days. The novelistic signified (the moment of the narrating) has
become something like 1925, but the historical referent, which
corresponds to the moment of the writing, did not keep pace
and continues to say: 1913. Narrative analysis must register
these shifts— and the resulting discordances — as effects of the
actual genesis of the work; but in the end analysis can look at the
narrating instance only as it is given in the final state of the text,
as a single moment without duration, necessarily placed several
years after the last "scene," therefore after the war, and even, as
we have seen, 36 after the death of Marcel Proust. This paradox,
let us remember, is not one: Marcel is not Proust, and nothing
requires him to die with Proust. What is required, on the other
hand, is that Marcel spend "many years" after 1916 in a clinic,
which necessarily puts his return to Paris and the Guermantes
matinee in 1921 at the earliest, and the meeting with an Odette
"showing signs of senility" in 1923. 37 That consequence is a

Between this single narrating instant and the different
moments of the story, the interval is necessarily variable. If
"many years" have elapsed since the bedtime scene in Com-
bray, it is only "of late" that the narrator has again begun to hear
his childhood sobs, and the interval separating the narrating
instant from the Guermantes matinee is obviously smaller than
the interval separating narrating instant and the hero's first
arrival in Balbec. The system of language, the uniform use of the
past tense, does not allow this gradual shrinking to be imprinted
in the very texture of the narrative discourse, but we have seen
that to a certain extent Proust had succeeded in making it felt, by
modifications in the temporal pacing of the narrative: gradual
disappearance of the iterative, lengthening of the singulative
scenes, increasing discontinuity, accentuation of the rhythm—

36 P. 91.

37 This episode takes place (RH II, 1063/P III, 951) "Less than three years"
thus more than two years— after the Guermantes matinee.

11 ' ill


Narrative Discourse

:«!, :

as if the story time were tending to dilate and make itself more
and more conspicuous while drawing near its end, which is also
its origin.

According to what we have already seen to be the common
practice of "autobiographical" narrating, we could expect to see
the narrative bring its hero to the point where the narrator
awaits him, in order that these two hypostases might meet and
finally merge. People have sometimes, a little quickly, claimed
that this is what happens. 38 In fact, as Marcel Muller well notes,
"between the day of the reception at the Princess's and the day
when the Narrator recounts that reception there extends a
whole era which maintains a gap between the Hero and the
Narrator, a gap that cannot be bridged: the verbal forms in the
conclusion of the Temps retrouve are all in the past tense." 39 The
narrator brings his hero's story— his own story — precisely to the
point when, as Jean Rousset says, "the hero is about to become
the narrator"; 40 I would say rather, is beginning to become the
narrator, since he actually starts in on his writing. Muller writes
that "if the Hero overtakes the Narrator, it is like an asymptote:
the interval separating them approaches zero, but will never
reach it," but his image connotes a Stemeian play on the two
durations that does not in fact exist in Proust. There is simply
the narrative's halt at the point when the hero has discovered
the truth and the meaning of his life: at the point, therefore,
when this "story of a vocation" — which, let us remember, is the
avowed subject of Proustian narrative — comes to an end. The
rest, whose outcome is already known to us by the very novel
that concludes here, no longer belongs to the "vocation" but to
the effort that follows it up, and must therefore be only sketched

38 In particular Louis Martin-Chauffier: "As in memoirs, the man who writes
and the man whose life we see are distinct in time, but tend to catch up with
each other in the long run; they are moving towards the day when the progress
of the hero through his life stops at the table, where the narrator, no longer
separated from him in time nor tied to him by memory, invites him to sit down
beside him so that both together may write: the End" ("Proust and the Double
I," Partisan Review, 16 [October 1949], 1012).

39 Muller, pp. 49-50. Let us remember, however, that certain anticipations
(like the last meeting with Odette) cover a part of that "era."

40 Rousset, p. 144.



in. The subject of the Recherche is indeed "Marcel becom
writer," not "Marcel the writer": the Recherche remains a noil
of development, and to see it as a "novel about the novelist
like the Faux Monnayeurs [The Counterfeiters], would be to distort
its intentions and above all to violate its meaning; it is a novel
about the future novelist. "The continuation," Hegel said pre
cisely apropos of the Bildungsroman, "no longer has any thine
novehshc about it." Proust probably would have been elad to
apply that formulation to his own narrative: what is novelistic is
the quest, the search [recherche], which ends at the discovery (the
revelation), not at the use to which that discovery will after
ward be put. The final discovery of the truth, the late encounter
with the vocation, like the happiness of lovers reunited can
be only a denouement, not an interim stopping place- and in
this sense, the subject of the Recherche is indeed a traditional
subject. So it is necessary that the narrative be interrupted be-
fore the hero overtakes the narrator; it is inconceivable for them
both together to write: The End. The narrator's last sentence is
when— is that—the hero finally reaches his first. The interval
between the end of the story and the moment of the narrating is
therefore the time it takes the hero to write this book, which is
and is not the book the narrator, in his turn, reveals to us in a
moment brief as a flash of lightning.

Narrative Levels

When Des Grieux, having reached the end of his narrative
states that he has just sailed from New Orleans to Havre-de-
Grace, then from Havre to Calais to meet his brother who is
waiting for him several miles away, the temporal (and spatial)
interval that until then separated the reported action from the
narrating act becomes gradually smaller until it is finally reduced
to zero: the narrative has reached the here and the now, the story
has overtaken the narrating. Yet a distance still exists between
the final episodes of the Chevalier's loves and the room in the
Lion d'or" with its occupants, including the Chevalier himself
and his host, where after supper he recounts these episodes to
the Marquis de Renoncourt: the distance between episodes and
inn lies neither in time nor in space, but in the difference be-


Narrative Discourse

tween the relations which both the episodes and the inn main-
tain at that point with Des Grieux's narrative. We will distin-
guish those relations in a rough and necessarily inadequate way
by saying that the episodes of the Chevalier's loves are inside
(meaning inside the narrative) and the inn with its occupants is
outside. What separates them is less a distance than a sort of
threshold represented by the narrating itself, a difference of
level. The "Lion d'or," the Marquis, the Chevalier in his function
as narrator are for us inside a particular narrative, not Des
Grieux's but the Marquis's, the Memoires d'un homme de aualite;
the return from Louisiana, the trip from Havre to Calais, the
Chevalier in his function as hero are inside another narrative,
this one Des Grieux's, which is contained within the first one, not
only in the sense that the first frames it with a preamble and a
conclusion (although the latter is missing here), but also in the
sense that the narrator of the second narrative is already a
character in the first one, and that the act of narrating which
produces the second narrative is an event recounted in the first

We will define this difference in level by saying that any event
a narrative recounts is at a diegetic level immediately higher than the
level at which the narrating act producing this narrative is placed. M.
de Renoncourt's writing of his fictive Memoires is a (literary) act
carried out at a first level, which we will call extradiegetic; the
events told in those Memoires (including Des Grieux's narrating
act) are inside this first narrative, so we will describe them as
diegetic, or intradiegetic; the events told in Des Grieux's narrative,
a narrative in the second degree, we will call metadiegetic. 41 In

41 These terms have already been put forth in my Figures 11, p. 202. The prefix
meta- obviously connotes here, as in "metalanguage," the transition to the sec-
ond degree: the mctanarrative is a narrative within the narrative, the metadiegesis
is the universe of this second narrative, as the diegesis (according to a now
widespread usage) designates the universe of the first narrative. We must admit,
however, that this term functions in a way opposite to that of its model in logic
and linguistics: metalanguage is a language in which one speaks of another
language, so metanarrative should be the first narrative, within which one
would tell a second narrative. But it seemed to me that it was better to keep the
simplest and most common designation for the first degree, and thus to reverse
the direction of interlocking. Naturally, the eventual third degree will be a
meta-metanarrative, with its meta-metadiegesis, etc.



the same way, M. de Renoncourt as "author" of the Memoires is
extradiegetic: although fictive, he addresses the actual public,
just like Rousseau or Michelet; the same Marquis as hero of the
same Memoires is diegetic, or intradiegetic, and so also is Des
Grieux the narrator at the "Lion d'or," as well as the Manon
noticed by the Marquis at the first meeting in Pacy; but Des
Grieux the hero of his own narrative, and Manon the heroine,
and his brother, and the minor characters, are metadiegetic.
These terms (metadiegetic, etc.) designate, not individuals, but
relative situations and functions. 42

The narrating instance of a first narrative is therefore ex-
tradiegetic by definition, as the narrating instance of a second
(metadiegetic) narrative is diegetic by definition, etc. Let us em-
phasize the fact that the possibly fictive nature of the first in-
stance does not modify this state of affairs any more than the
possibly "real" nature of the subsequent instances does: M. de
Renoncourt is not a "character" in a narrative taken charge of by
the Abbe Prevost; he is the fictive author of Memoires, whose real
author, of course, is Prevost, just as Robinson Crusoe is the
fictive author of the novel by Defoe that bears his name; sub-
sequently, each of them (the Marquis and Crusoe) becomes a
character in his own narrative. Neither Prevost nor Defoe enters
the space of our inquiry, which, let us recall, bears on the narrat-
ing instance, not on the literary instance. M. de Renoncourt and
Crusoe are author-narrators, and as such they are at the same
narrative level as their public — that is, as you and me. This is not
the case with Des Grieux, who never addresses himself to us,
but only to the patient Marquis; and inversely, even if this fictive
Marquis had met a real person at Calais (say, Sterne on a jour-
ney), this person would nonetheless be diegetic, even though
real — just like Richelieu in Dumas, Napoleon in Balzac, or the

42 The same character can, moreover, assume two identical (parallel) narrative
functions at different levels: for example, in Sarrasine, the extradiegetic narrator
himself becomes intradiegetic narrator when he tells his companion the story of
Zambinella. Thus he tells us that he tells this story — a story of which he is not
the hero: this situation is the exact opposite of the (much more common) one of
Manon, where the first narrator becomes on the second level the listener of
another character who tells his own story. The situation of a double narrator
occurs only, to my knowledge, in Sarrasine.


Narrative Discourse

Princesse Mathilde in Proust. In short, we shall not confound
extradiegetic with real historical existence, nor diegetic (or even
metadiegetic) status with fiction: Paris and Balbec are at the
same level, although one is real and the other fictive, and every
day we are subjects of a narrative, if not heroes of a novel.

But not every extradiegetic narrating is necessarily taken up as
a literary work with its protagonist an author-narrator in a posi-
tion to address himself, like the Marquis de Renoncourt, to a
public termed such. 43 A novel in the form of a diary (like the
Journal d'un cure de campagne or the Symphonie pastorale) does not
in principle aim at any public or any reader, and it is the same
with an epistolary novel, whether it include a single letter writer
(like Pamela, Werther, or Obermann, often described as journals
disguised as correspondence) 44 or several (like La Nouvelle
Heloi'se or Les Liaisons dangereuses). Bernanos, Gide, Richardson,
Goethe, Senancour, Rousseau, and Laclos present themselves
here simply as "editors," but the fictive authors of these diaries
or "letters collected and published by . . . " — as distinct from
Renoncourt, or Crusoe, or Gil Bias — obviously did not look on
themselves as "authors." What is more, extradiegetic narrating
is not even necessarily handled as written narrating: nothing
claims that Meursault or The Unnamable wrote the texts we read
as their interior monologues, and it goes without saying that the
text of the Lauriers sont coupes cannot be anything but a "stream
of consciousness" — not written, or even spoken — mysteriously
caught and transcribed by Dujardin. It is the nature of im-
mediate speech to preclude any formal determination of the
narrating instance which it constitutes.

Inversely, every intradiegetic narrating does not necessarily
produce, like Des Grieux's, an oral narrative. It can consist of a
written text, like the memoir with no recipient written by
Adolphe, or even a fictive literary text, a work within a work,
like the "story" of the Curious Impertinent discovered in a cloak
bag by the curate in Don Quixote, or the novella "L'Ambitieux

43 See the "Notes by the Author" published at the head of Marion Lescaut.

44 There remains, however, an appreciable difference between these "epistol-
ary monodies," as Rousset calls them, and a diary: the difference is the existence
of a receiver (even a mute one), and his traces in the text.



par amour" published in a fictive magazine by the hero of Albert
Savarus, the intradiegetic author of a metadiegetic work. But the
second narrative can also be neither oral nor written, and can
present itself, openly or not, as an inward narrative (for in-
stance, Jocabel's dream in Moyse same) or (more frequently and
less supernaturally) as any kind of recollection that a character
has (in a dream or not). Thus (and this detail made a strong
impression on Proust) the second chapter of Sylvie is interrupted
by the episode ("memory half dreamed") of Adrienne's song: "I
went back to bed and could find no rest there. As I lay between
sleeping and waking, my whole youth passed through my

memory I visualized a chateau from the time of Henry IV. " 4 s

Finally, the second narrative can be handled as a nonverbal
representation (most often visual), a sort of iconographic docu-
ment, which the narrator converts into a narrative by describing
it himself (the print representing the desertion of Ariadne, in The
Nuptial Song of Peleus and Thetis, or the tapestry of the flood in
Moyse same), or, more rarely, by having another character
describe it (like the tableaux of Joseph's life commented on by
Amram, also in Moyse sauve).

Metadiegetic Narrative

Second-degree narrative is a form that goes back to the very
origins of epic narrating, since Books IX-XII of the Odyssey, as
we know, are devoted to the narrative Ulysses makes to the
assembled Phaeacians. Via Virgil, Ariosto, and Tasso, this tech-
nique (which the Thousand and One Nights has an enormous in-
vestment in, as we know in another connection) enters the
novelistic tradition in the baroque period, and a work like As-
tree, for example, is in large part composed of narratives ob-
tained by one or another character. The practice continues in the

45 So we have there an analepsis which is metadiegetic— obviously not the
case of every analepsis. For example, in that same Sylvie, the retrospection of
chapters 4, 5, and 6 is taken on by the narrator himself and not obtained through
the hero's memory: "While the carriage is climbing the slopes, let us recollect the
time when I came here so often." Here the analepsis is purely diegetic— or, if we
wish to mark more clearly the equality of narrative level, it is isodiegetic. (Proust's
comments are in Marcel Proust on Art, p. 147, and the Recherche, RH II, 1038/P III



Narrative Discourse

eighteenth century, despite the competition of new forms like
the epistolary novel; we certainly see it in Manon Lescaut, or
Tristram Shandy, or Jacques le fataliste. And even the advent of
realism does not prevent it from surviving in Balzac (La Maison
Nucingen, Autre etude de femme, L'Auberge rouge, Sarrasine, LaPeau
de chagrin) and Fromentin (Dominique); we can even observe a
certain exacerbation of the topos with Barbey, or in Wuthering
Heights (Isabella's narrative to Nelly, reported by Nelly to
Lockwood, noted by Lockwood in his journal), and especially in
Lord ]im, where the entanglement reaches the bounds of general
intelligibility. The formal and historical study of this technique
would go well beyond our intention, but for the sake of what
follows it is necessary here at least to differentiate the main
types of relationships that can connect the metadiegetic narra-
tive to the first narrative, into which it is inserted.

The first type of relationship is direct causality between the
events of the metadiegesis and those of the diegesis, conferring
on the second narrative an explanatory function. It is the Balzac-
ian "this is why," but taken on here by a character, whether the
story he tells is someone else's (Sarrasine) or, more often, his
own (Ulysses, Des Grieux, Dominique). All these narratives an-
swer, explicitly or not, a question of the type "What events have
led to the present situation?" Most often, the curiosity of the
intradiegetic listener is only a pretext for replying to the curiosity
of the reader (as in the expository scenes of classical drama), and
the metadiegetic narrative only a variant of the explanatory
analepsis. Whence certain discordances between the alleged
function and the real function — generally resolved in favor of
the latter. For instance, in Book XII of the Odyssey, Ulysses inter-
rupts his narrative at the arrival on Calypso's island, although
most of his audience does not know what follows; the pretext is
that he told it briefly the day before to Alcinous and Arete (Book
VII); the real reason is obviously that the reader knows it in
detail by the direct narrative in Book V. "It liketh me not twice,"
says Ulysses, "to tell a plain-told tale": 46 this reluctance is, to
begin with, the poet's own.



The second type consists of a purely thematic relationship,
therefore implying no spatio-temporal continuity between
metadiegesis and diegesis: a relationship of contrast (the de-
serted Ariadne's unhappiness, in the midst of Thetis' joyous
wedding) or of analogy (as when Jocabel, in Moyse sauve, hesi-
tates to execute the divine command and Amram tells her the
story of Abraham's sacrifice). The famous structure en abyme, not
long ago so prized by the "new novel" of the 1960's, is obviously
an extreme form of this relationship of analogy, pushed to the
limits of identity. Thematic relationship can, moreover, when it
is perceived by the audience, exert an influence on the diegetic
situation: Amram's narrative has as its immediate effect (and,
moreover, as its aim) to convince Jocabel; it is an exemplum with
a function of persuading. We know that regular genres, like the
parable or the apologue (the fable), are based on that monitory
effect of analogy: before the rebelling populace, Menenius
Agrippa tells the story of the Members [of the body] and the Belly;
then, adds Titus Livius, "Drawing a parallel from this to show
how like was the internal dissension of the bodily members to
the anger of the plebs against the Fathers, he prevailed upon the
minds of his hearers." 47 In Proust we will find a less curative
illustration of this force of example.

The third type involves no explicit relationship between the
two story levels: it is the act of narrating itself that fulfills a
function in the diegesis, independently of the metadiegetic
content — a function of distraction, for example, and/or of
obstruction. Surely the most illustrious example is found in the
Thousand and One Nights, where Scheherazade holds off death
with renewed narratives, whatever they might be (provided
they interest the sultan). We notice that, from the first type to
the third, the importance of the narrating instance only grows.
In the first type, the relationship (of linking) is direct; it is not via
the narrative, which could very well be dispensed with: whether
Ulysses tells about it or not, the storm is what cast him up on the
shore of Phaeacia, and the only transformation his narrative
introduces is of a purely cognitive order. In the second type, the

"''Odyssey, Book XII, 11.452-453, trans. S. H. Butcher and A. Lang (New York:
Modern Library, 1950), p. 194.

47 Livy, From the Founding of the City, Book II, chap. 32, trans. B. O. Foster
(London: Loeb Classical Library, 1925), p. 325.


Narrative Discourse

relationship is indirect, rigorously mediated by the narrative,
which is indispensable to the linking: the adventure of the
members and the belly calms the populace on condition that
Menenius tell it to the plebs. In the third type, the relationship is
only between the narrating act and the present situation, with
the metadiegetic content (almost) not mattering any more than a
Biblical message does during a filibuster at the rostrum of the
United States Senate. This relationship indeed confirms, if there
were a need to, that narrating is an act like any other.


The transition from one narrative level to another can in prin-
ciple be achieved only by the narrating, the act that consists
precisely of introducing into one situation, by means of a dis-
course, the knowledge of another situation. Any other form of
transit is, if not always impossible, at any rate always transgres-
sive. Cortazar tells the story of a man assassinated by one of the
characters in the novel he is reading; 48 this is an inverse (and
extreme) form of the narrative figure the classics called author's
metalepsis, which consists of pretending that the poet "himself
brings about the effects he celebrates," 49 as when we say that
Virgil "has Dido die" in Book IV of the Aeneid, or when Diderot,
more equivocally, writes in Jacques le fataliste: "What would pre-
vent me from getting the Master married and making him a cuck-
old?" or even, addressing the reader, "If it gives you pleasure,
let us set the peasant girl back in the saddle behind her escort, let
us let them go and let us come back to our two .travelers." 50 Sterne
pushed the thing so far as to entreat the intervention of the
reader, whom he beseeched to close the door or help Mr.
Shandy get back to his bed, but the principle is the same: any
intrusion by the extradiegetic narrator or narratee into the dieget-

48 Cortazar, "Continuidad de los Parques," in Final del juego.

49 Pierre Fontanier, Commentaire raisonne sur "Les Tropes" de Dumarsais, vol. 2 of
Dumarsais' Les Tropes (1818; repr. Geneva: Slatkine Reprints, 1967), p. 116.
Moyse sauve inspires Boileau (Art poetique, I, 25-26) with this unsparing metalep-
sis: "And [Saint Amant], following Moses o'er the sandy plain, / Perished with
Pharaoh in the Arabian main" (The Art of Poetry: The Poetical Treatises of Horace,
Vida, and Boileau, trans. Soame, ed. Albert S. Cook |Boston: Ginn and Co., 1892|,
p. 160).

50 Gamier, pp. 495 and 497.



ic universe (or by diegetic characters into a metadiegetic uni-
verse, etc.), or the inverse (as in Cortazar), produces an effect of
strangeness that is either comical (when, as in Sterne or Diderot,
it is presented in a joking tone) or fantastic.

We will extend the term narrative metalepsis 51 to all these
transgressions. Some of them, as ordinary and innocent as those
of classical rhetoric, play on the double temporality of the story
and the narrating. Here, for example, is Balzac, in a passage
already quoted from Illusions perdues: "While the venerable
churchman climbs the ramps of Angouleme, it is not useless to
explain ... ," as if the narrating were contemporaneous with the
story and had to fill up the latter's dead spaces. This is the very
prevalent model Proust follows when he writes, for example,
"but I have no time left now, before my departure for Balbec . . . , to
start upon a series of pictures of society," or "I confine myself at
present, as the train halts and the porter calls out 'Doncieres,'
'Grattevast ,' 'Maineville,' etc., to noting down the particular
memory that the watering-place or garrison town recalls to me,"
or again: "But it is time to rejoin the Baron as he ad-
vances..." 52 Sterne's temporal games, of course, are a bit
bolder, a bit more literal, in other words, as when the digres-
sions of Tristram the (extradiegetic) narrator require his father
(in the diegesis) to prolong his nap by more than an hour, 53 but
here, too, the principle is the same. 54 In a certain way, the
Pirandello manner of Six Characters in Search of an Author or To-
night We Improvise, where the same actors are in turn characters
and players, is nothing but a vast expansion of metalepsis; so is
everything deriving from that manner in the plays of Genet, for
example, and so are the changes of level in the Robbe-Grillet
type of narrative (characters escaped from a painting, a book,
a press clipping, a photograph, a dream, a memory, a fantasy,

51 Metalepsis here forms a system with prolepsis, analepsis, syllepsis, and paralep-
sis, with this specific sense: "taking hold of (telling) by changing level."

52 RH II, 102-103/P II, 742; RH II, 339/P II, 1076; RH II, 530/P III, 216. Or again,
RH II, 292/P II, 1011: "Let us for the moment say simply this, while Albertine waits
for me . . .

53 Sterne, Tristram Shandy, 111, chap. 38, and IV, chap. 2.

54 I owe the distant revelation of the metaleptic game to this lapse, perhaps a
deliberate one, by a history teacher: "We are going to study the Second Empire
now from the coup d'etat to the Easter vacation."


Narrative Discourse

etc.)- All these games, by the intensity of their effects, demon-
strate the importance of the boundary they tax their ingenuity to
overstep, in defiance of verisimilitude — a boundary that is pre-
cisely the narrating (or the performance) itself: a shifting but sacred
frontier between two worlds, the world in which one tells, the
world of which one tells. Whence the uneasiness Borges so well
put his finger on: "Such inversions suggest that if the characters
in a story can be readers or spectators, then we, their readers or
spectators, can be fictitious." 55 The most troubling thing about
metalepsis indeed lies in this unacceptable and insistent
hypothesis, that the extradiegetic is perhaps always diegetic,
and that the narrator and his narratees — you and I — perhaps
belong to some narrative.

A less audacious figure, but one we can connect to metalepsis,
consists of telling as if it were diegetic (as if it were at the same
narrative level as its context) something that has nevertheless
been presented as (or can easily be guessed to be) metadiegetic
in its principle or, if one prefers, in its origin: as if the Marquis de
Renoncourt, after having acknowledged that he has gotten the
story of Des Grieux's loves from Des Grieux himself (or even
after having let Des Grieux speak for several pages), sub-
sequently took back the floor to tell that story himself, no longer
"speaking," Plato would say, "as if he had become Des Grieux."
The prototype of this technique is undoubtedly the Theaetetus,
which, as we know, consists of a conversation among Socrates,
Theodorus, and Theaetetus, which Socrates himself told to Eu-
cleides, who tells it to Terpsion. But, says Eucleides, "to avoid in
the written account the tiresome effect of bits of narrative inter-
rupting the dialogue, such as 'and I said' or 'and I remarked'
wherever Socrates was speaking of himself, and 'he asserted' or
'he did not agree,' where he reported the answer," the conver-
sation has been reworded into the form of "a direct conversation
between the actual speakers." 56 These forms of narrating where

55 Borges, Other Inquisitions, 1937-1952, trans. R. Simms (Austin, 1964), p. 46.

56 Plato, Theaetetus, 143 c, in Plato's Theory of Knowledge: The "Theaetetus" and
the "Sophist" of Plato, trans. Francis M. Cornford (London: Routledge and Kegan
Paul, 1935), p. 17.



the metadiegetic way station, mentioned or not, is immediately
ousted in favor of the first narrator, which to some extent

economizes on one (or sometimes several) narrative level(s)

these forms we will call reduced metadiegetic (implying: reduced
to the diegetic), or pseudo- diegetic.

In fact, the reduction is not always obvious; more precisely,
the difference between metadiegetic and pseudo-diegetic is not
always perceptible in the literary narrative text, which (unlike
the cinematographic text) does not have at hand features ca-
pable of indicating the metadiegetic nature of a section, 57 except
by a shift in person: if M. de Renoncourt took Des Grieux's place
to tell the latter's adventures, the substitution would be indicated
immediately in the transition from I to he; but when the hero of
Sylvie relives in a dream a moment from his youth, nothing
allows us to decide whether the narrative is then a narrative of
that dream or a direct narrative, beyond the dream instance, of
the earlier moment.

From Jean Santeuil to the Recherche, or
The Triumph of the Pseudo-diegetic

After that additional detour, it will be easier for us to charac-
terize the narrative choice Proust made, deliberately or not, in
the Recherche du temps perdu. But before we can do that we must
remember what the choice was in his first large narrative work,
or, more precisely, in the first version of the Recherche, that is, in
Jean Santeuil. In that book the narrating instance is split in two:
the extradiegetic narrator, who does not have a name (but he is a
first hypostasis of the hero, and we see him in situations later
assigned to Marcel), is on vacation with a friend at the Bay of
Concarneau; the two young men strike up a friendship with a
writer named C. (the second hypostasis of the hero), who at
their request undertakes each evening to read them the pages he
wrote, during the day, of a novel in progress. These fragmen-
tary readings are not transcribed, but some years later, after C.'s

57 Such as the blur, slow motion, voice-off, transition from color to black and
white or the reverse, etc. Conventions of this kind, moreover, could have been
established in literature (italics, bold-faced type, etc.).


Narrative Discourse

death, the narrator, who has somehow gotten hold of a copy of
the novel, decides to publish it: it is Jean Santeuil, whose hero is
obviously a third outline of Marcel. This domino structure is
fairly archaic, differing only in two minor ways from the tradi-
tion represented by Marion Lescaut: the intradiegetic narrator
here does not tell his own story, and his narrative is not oral but
written, and even literary, since it involves a novel. We will
return later to the first difference, which touches on the problem
of "person," but here we must emphasize the second, which, at
a period when those techniques were no longer much used,
attests to a certain timidity at novelistic writing and an obvious
need for "distancing" with respect to this biography of Jean —
much closer to autobiography than the Recherche is. The narra-
tive splitting in two is further heightened by the literary — and,
what is more, fictive (because novelistic) — nature of the
metadiegetic narrative.

From this first attempt we should retain the fact that Proust
was familiar with the use of "Chinese-box" narrative, and that
he had submitted to its temptation. Moreover, he alludes to this
technique at one point in La Fugitive:

Novelists sometimes pretend in an introduction that while travel-
ling in a foreign country they have met somebody who has told
them the story of a person's life. They then withdraw in favour of
this casual acquaintance, and the story that he tells them is noth-
ing more or less than their novel. Thus the life of Fabrice del
Dongo was related to Stendhal by a Canon of Padua. How gladly
would we, when we are in love, that is to say when another
person's existence seems to us mysterious, find some such well-
informed narrator! And undoubtedly he exists. Do we not our-
selves frequently relate, without any trace of passion, the story of
some woman or other, to one of our friends, or to a stranger, who
has known nothing of her love-affairs and listens to us with keen
interest? 58

We see that the comment does not concern only literary crea-
tion, but extends to the most common narrative activity, such as

>RH II, 768/P III, 551.



can be pursued, for example, in Marcel's existence: these narra-
tives told by X to Y apropos of Z are the very fabric of our
"experience," a large part of which is narrative in kind.

Those antecedents and that allusion only throw into greater
relief the dominant feature of the narrating in the Recherche,
which is the almost systematic elimination of metadiegetic narrative.
In the first place, the fiction of the discovered manuscript disap-
pears in favor of a direct narrating in which the narrator-hero
openly presents his narrative as a literary work, and thus takes
up the role of (fictive) author, like Gil Bias or Crusoe, in im-
mediate contact with the public. Whence the use of the phrase
"these volumes" or "this work" 59 to refer to his narrative;
whence the editorial "we," 60 those addresses to the reader, 61
and even this humorous pseudo-dialogue in Sterne's or Di-
derot's manner: " 'All this,' the reader will remark, 'tells us noth-
ing as to . . . ' It is indeed a pity, gentle reader. And sadder than
you think. ... 'In a word, did Mme. d'Arpajon introduce you to
the Prince?' No, but be quiet and let me go on with my story." 62
The fictive novelist of Jean Santeuil did not permit himself that
much, and this difference measures the progress achieved in
emancipating the narrator. Second, metadiegetic insertions are
almost completely missing from the Recherche: under this head-
ing we can hardly point to anything except Swann's narrative to
Marcel about his conversation with the Prince de Guermantes
who has converted to Dreyfusism, 63 Aime's reports about Alber-
tine's past behavior, 64 and above all the narrative assigned to the
Goncourts about a dinner at the Verdurins'. 65 We will notice,
moreover, that in these three cases the narrating instance is

59 "That invisible vocation of which these volumes are the history" (RH I, 1002/P
II, 397); "The proportions of this work. . . " (RH II, 33/P II, 642); "this book in
which there is not a single incident which is not fictitious ..." (RH II, 981/P III

60 "We suppose M. de Charlus . . . " (RH II, 291/P II, 1010).

61 "We must warn the reader ..." (RH II, 406/P III, 40); "Before we come back
to Jupien's shop, the author would like to say how deeply he would regret it
should any reader be offended ..." (RH II, 410/P III, 46)

« RH II, 39-40/P II, 651-652.

63 RH II, 77-82/P II, 705-712.

64 RH II, 744-745/P III, 515-516; RH II, 750-751/P III, 524-525.

65 RH II, 880-885/P III, 709-717.


Narrative Discourse

highlighted and competes in importance with the event being
related: Swann's naive partiality interests Marcel much more
than the Prince's conversion does; Aime's writing style, with its
parentheses and quotation marks, is an imaginary pastiche; and
the pseudo-Goncourt, a real pastiche, serves here as a page
from literature and a testimony to the vanity of Letters much
more than as evidence about the Verdurin salon. For these vari-
ous reasons it was not possible to reduce those metadiegetic
narratives, that is, to have the narrator take control of them.

Everywhere else, on the other hand, the narrative in the Re-
cherche constantly practices what we have christened the
pseudo-diegetic: that is, a narrative second in its origin is im-
mediately brought to the first level and taken charge of, what-
ever its source might be, by the narrator-hero. Most of the
analepses noted in Chapter 1 originate either in memories the
hero recalls (and thus in a sort of inward narrative in the manner
of Nerval) or else in reports made to the hero by a third person.
Coming under the first type, for example, are the last pages of
the jeunes Filles en fleurs, evoking the sun-bathed mornings of
Balbec — but doing so through the memory of them that the
hero, back in Paris, has preserved: "What my mind's eye did
almost invariably see when I thought of Balbec were the hours
which every morning during the fine weather . . . "; 66 after this
the evocation forgets its memory-elicited pretext and to the last
line unfolds on its own account as direct narrative, so that many
readers do not notice the spatio-temporal detour that gave rise
to it and think it a simple isodiegetic "return backward" without
a change in narrative level. Also coming under the first type is
the return to 1914, during the stay in Paris in 1916, introduced
with this sentence: "I reflected that it was a long time since I had
seen any of the personages who have been mentioned in this
work. In 1914, it was true . . . "; 67 then comes a direct narrative
about that first return, as if it were not a memory evoked during
the second return, or as if the memory were in this case only a
narrative pretext, what Proust precisely calls a "method of tran-

66 RH I, 712-713/P I, 953.
" RH II, 900/P III, 737.



sition." Another example of the first type comes some pages
later, where the passage devoted to Saint-Loup's visit, 68 which
begins as an isodiegetic analepsis, ends with a sentence which
reveals, after the event, its source as memory: "As I turned over
in my mind this recent meeting with Saint-Loup. ..." But above
all we should remember that Combray I is an insomniac's reverie,
that Combray II is an "involuntary memory" called forth by the
taste of the madeleine, and that everything after that, starting
with Un amour de Swann, is again an evocation of the insomniac.
The whole Recherche is in fact a huge pseudo-diegetic analepsis
in the name of the memories of the "intermediary subject" —
memories which the final narrator immediately claims and takes
control of.

To the second type (reports made to the hero by a third per-
son) belong all those episodes, evoked in the preceding chapter
apropos of the problems of focalization, that took place out of
the hero's presence and that the narrator could therefore not be
informed of except by an intermediary narrative. Examples are the
circumstances of Swann's marriage, the negotiations between
Norpois and Faffenheim, Bergotte's death, Gilberte's conduct
after Swann's death, the missed reception at Berma's. 69 As we
have seen, the source of all this information is sometimes stated,
sometimes implicit, but in every case Marcel jealously incorpo-
rates into his own narrative what he has gotten from Cottard,
from Norpois, from the Duchess, or from God knows whom, as
if he could not bear to give up to anyone else the slightest part of
his narrative privilege.

The most typical and naturally the most important case is Un
amour de Swann. With respect to its source this episode is doubly
metadiegetic, first since the details were reported to Marcel by
an undetermined narrator at an undetermined time, and then
because Marcel is remembering these details in the course of
certain sleepless nights. These are memories of earlier narra-
tives, therefore, from which the extradiegetic narrator once

68 RH II, 914-919/P III, 756-762.

69 RH I, 358-361/P I, 467-171; RH I, 899-904/P II, 257-263; RH II, 506-510/P III,
182-188; RH II, 786-792/P III, 574-582; RH II, 1098-1101/P III, 995-998.


Narrative Discourse

again gathers up the whole kitty and in his own name tells this
whole story that took place before he was born — not without
introducing into it subtle marks of his subsequent existence, 70
which are there like a signature and prevent the reader from
forgetting him for too long: a fine example of narrative egocen-
trism. In ]ean Santeuil Proust had savored the antiquated plea-
sures of the metadiegetic, and it is as if he had vowed not to
come back to them any more, and to reserve for himself (or for
his spokesman) the whole of the narrating function. An Amour
de Swann told by Swann himself would have compromised this
unity of instance and this monopoly by the hero. In the defini-
tive economy of the Recherche, Swann, the ex-hypostasis of
Marcel, 71 must be no more than an unhappy and imperfect pre-
cursor. He therefore has no right to the "floor," that is, to the
narrative — and even less (we will come back to this) to the dis-
course that transmits it, accompanies it, and gives it its meaning.
This is why it is Marcel, and only Marcel, who in the final in-
stance, and scorning all the others, must recount that love affair
which is not his own.

But which prefigures his (as everyone knows) and to a certain
extent brings it to pass. Here again we meet the indirect influ-
ence, analyzed above, of certain metadiegetic narratives:
Swann's love for Odette in principle has no direct impact on
Marcel's fate, 72 and on that ground the classical norm would
undoubtedly deem Swann's love purely episodic; but on the
other hand its indirect impact — that is, the influence of the
knowledge Marcel has of that love, gained through a
narrative — is considerable, as he himself testifies in this passage
from Sodome:

70 "l used often to recall to myself when, many years later, i began to take an
interest in his character because of the similarities which, in wholly different
respects, it offered to my own . . ." (RH I, 148/P 1, 193); "And he did not have (asi
had, afterwards, at Combray in my childhood). . ." (RH I, 227/P I, 295); "as /
myself was to go . . ." (RH I, 228/P I, 297); "my grandfather" (RH I, 149, 238/P 1,
194, 310); "my uncle" (RH I, 239-240/P I, 311-312), etc.

71 In Jean Santeuil, the two characters appear merged; and again in certain
sketches of the Cahiers. See for example Andre Maurois, Proust: Portrait of a
Genius, trans. Gerard Hopkins (New York, 1950), p. 152.

72 Unless we count as such the very existence of Gilberte, the "fruit" of that



I thought then of all that I had been told about Swann's love for
Odette, of the way in which Swann had been tricked all his life.
Indeed, when I come to think of it, the hypothesis that made me
gradually build up the whole of Albertine's character and give a
painful interpretation to every moment of a life that I could not
control in its entirety, was the memory, the rooted idea of Mme.
Swann's character, as it had been described to me. These accounts
helped my imagination, in after years, to take the line of suppos-
ing that Albertine might, instead of being a good girl, have had the
same immorality, the same faculty of deception as a reformed
prostitute, and I thought of all the sufferings that would in that
case have been in store for me had I ever really been her lover. 73

"These accounts helped. . . ": it is because of the narrative of
Swann in love that Marcel will one day be able actually to imag-
ine an Albertine like Odette — unfaithful, given to vice,
unattainable — and consequently to fall in love with her. We know
what happens then. The power of narrative . . .

Let us not forget, after all, that if Oedipus can do what every
man, so they say, goes only so far as wishing to do, it is because
an oracle told in advance that one day he would kill his father
and marry his mother: without the oracle, no exile, thus no
incognito, thus no parricide and no incest. The oracle in Oedipus
the King is a metadiegetic narrative in the future tense, the mere
uttering of which will throw into gear the "infernal machine"
capable of carrying it out. This is not a prophecy that comes true;
it is a trap in the form of a narrative, a trap that "takes." Yes,
the power (and cunning) of narrative. Some give life
(Scheherazade), some take life. And we do not properly under-
stand Un amour de Swann unless we realize that this love told is
an instrument of Destiny.


Readers may have noticed that until now we have used the
terms "first-person — or third-person — narrative" only when
paired with quotation marks of protest. Indeed, these common
locutions seem to me inadequate, in that they stress variation in
the element of the narrative situation that is in fact invariant — to

' RH II, 147/P II, 804.


Narrative Discourse

wit, the presence (explicit or implicit) of the "person" of the
narrator. This presence is invariant because the narrator can be
in his narrative (like every subject of an enunciating in his enun-
ciated statement) only in the "first person" — except for an enal-
lage of convention as in Caesar's Commentaries; and stressing
"person" leads one to think that the choice the narrator has to
make — a purely grammatical and rhetorical choice — is always of
the same order as Caesar's in deciding to write his Memoirs "in"
one or another person. In fact, of course, this is not the issue.
The novelist's choice, unlike the narrator's, is not between two
grammatical forms, but between two narrative postures (whose
grammatical forms are simply an automatic consequence): to
have the story told by one of its "characters," 74 or to have it told
by a narrator outside of the story. The presence of first-person
verbs in a narrative text can therefore refer to two very different
situations which grammar renders identical but which narrative
analysis must distinguish: the narrator's own designation of
himself as such, as when Virgil writes "I sing of arms and the
man. . . ," or else the identity of person between the narrator
and one of the characters in the story, as when Crusoe writes "I
was born in the year 1632, in the city of York. ..." The term
"first-person narrative" refers, quite obviously, only to the sec-
ond of these situations, and this dissymmetry confirms its unfit-
ness. Insofar as the narrator can at any instant intervene as such
in the narrative, every narrating is, by definition, to all intents
and purposes presented in the first person (even if in the edito-
rial plural, as when Stendhal writes, "We will confess that . . . we
have begun the story of our hero . . . "). The real question is
whether or not the narrator can use the first person to designate
one of his characters. We will therefore distinguish here two types
of narrative: one with the narrator absent from the story he tells
(example: Homer in the Iliad, or Flaubert in L' Education sentimen-

74 This term [personnages lis used here for lack of a more neutral or more exten-
sive term which would not unduly connote, as this one does, the "humanness"
of the narrative agent, even though in fiction nothing prevents us from entrust-
ing that role to an animal (Menwires d'un am [Memoirs of a Donkey]) or indeed to
an "inanimate" object (I don't know whether we should put into this category
the successive narrators of the Bijoux indiscrets [Indiscreet Jewels]).



tale), the other with the narrator present as a character in the
story he tells (example: Gil Bias, or Wuthering Heights). I call the
first type, for obvious reasons, heterodiegetic, and the second
type homodiegetic.

But from the examples selected no doubt a dissymmetry in the
status of these two types already emerges. Homer and Flaubert
are both totally, and therefore equally, absent from the two nar-
ratives in question; on the other hand, we cannot say that Gil
Bias and Lockwood are equally present in their respective narra-
tives: Gil Bias is incontestably the hero of the story he tells,
Lockwood is incontestably not (and we could easily find exam-
ples of even weaker "presence"; I will come back to this
momentarily). Absence is absolute, but presence has degrees.
So will have to differentiate within the homodiegetic type at
least two varieties: one where the narrator is the hero of his
narrative (Gil Bias) and one where he plays only a secondary
role, which almost always turns out to be a role as observer and
witness: Lockwood, the anonymous narrator of Louis Lambert,
Ishmael in Moby Dick, Marlow in Lord Jim, Carraway in The
Great Gatsby, Zeitblom in Doctor Faustus — not to mention the
most illustrious and most representative one of all, the transpar-
ent (but inquisitive) Dr. Watson of Conan Doyle. 75 It is as if the
narrator cannot be an ordinary walk-on in his narrative: he can
be only the star, or else a mere bystander. For the first variety
(which to some extent represents the strong degree of the
homodiegetic) we will reserve the unavoidable term autodiegetic.

Defined this way, the narrator's relationship to the story is in
principle invariable: even when Gil Bias and Watson
momentarily disappear as characters, we know that they belong
to the diegetic universe of their narrative and that they will
reappear sooner or later. So the reader unfailingly takes the
transition from one status to the other — when he perceives
it — as an infraction of an implicit norm: for instance the (dis-
creet) disappearance of the initial witness-narrator of the Rouge

75 A variant of this type is the narrative with a collective witness as narrator:
the crew of The Nigger of the "Narcissus, " the inhabitants of the small town in "A
Rose for Emily." We remember that the opening pages of Bovary are written in
this mode.


Narrative Discourse

or Bovary, or the (noisier) one of the narrator of Lamiel, who
openly leaves the diegesis "in order to become a man of letters.
Thus, O benevolent reader, farewell; you will hear nothing more
of me." 76 An even more glaring violation is the shift in
grammatical person to designate the same character: for in-
stance, in Autre etude de femme, Bianchon moves all of a sudden
from "I" to "he," 77 as if he were unexpectedly abandoning the
role of narrator; for instance, in jean Santeuil, the hero moves
inversely from "he" to "I." 78 In the field of the classical novel,
and still in Proust, such effects obviously result from a sort of
narrative pathology, explicable by last-minute reshufflings and
states of textual incompleteness. But we know that the contem-
porary novel has passed that limit, as it has so many others, and
does not hesitate to establish between narrator and character(s)
a variable or floating relationship, a pronominal vertigo in tune
with a freer logic and a more complex conception of "personal-
ity." The most advanced forms of this emancipation 79 are
perhaps not the most perceptible ones, because the classical
attributes of "character"— proper name, physical and moral
"nature"— have disappeared and along with them the signs that
direct grammatical (pronominal) traffic. It is undoubtedly Borges
who offers us the most spectacular example of this violation-
spectacular precisely because it is put down in a completely
traditional narrative system, which accentuates the contrast — in
the story entitled "The Form of the Sword": 80 the hero begins to
tell his vile adventure while identifying himself with his victim,
before confessing that he is in fact the other, the dastardly in-
former who until then was dealt with, with all due contempt, in
the "third person." Moon himself supplies the "ideological"
comment on this narrative technique: "What one man does is

76 Stendhal, Lamiel (Paris: Divan, 1948), p. 43. The inverse case, the sudden
appearance of an autodiegetic "I" in a heterodiegetic narrative, seems more rare.
The Stendhalian "I believe" (Leuwen, p. 117, Chartreuse, p. 76) can belong to the
narrator as such.

77 Balzac, Autre etude de femme (Geneva: Skira), pp. 75-77.
is ]ean Santeuil, Pleiade, p. 319; trans. Hopkins, pp. 118-119.
79 See for example J. L. Baudry, Personnes (Paris: Seuil, 1967).

80 In Ficciones, ed. Anthony Kerrigan (New York: Grove Press, 1962), pp.



something done, in some measure, by all men. ... I am all oth-
ers, any man is all men." The Borgesian fantastic, in this respect
emblematic of a whole modern literature, does not accept person.

I do not intend to stretch Proustian narrating in this direction,
although in Proust the process of the disintegration of "charac-
ter" is amply (and notoriously) begun. The Recherche is funda-
mentally an autodiegetic narrative, where, as we have seen, the
narrator-hero never, as it were, yields the privilege of the narra-
tive function to anyone. Here what is most important is not the
presence of this completely traditional form, but first the con-
version it results from, and next the difficulties it encounters in a
novel like this one.

Since it is a "disguised autobiography," it seems on the whole
quite natural and a matter of course that the Recherche should be
a narrative in autobiographical form written "in the first per-
son." This naturalness is obviously deceptive, for Proust's initial
plan, as Germaine Bree suspected in 1948 and as the publication
of Jean Santeuil has since confirmed, made no place (except a
preliminary one) for that narrative course. Jean Santeuil, let us
remember, is deliberately heterodiegetic in form. Such a detour
prohibits us, then, from looking on the narrative form of the
Recherche as the direct extension of an authentically personal
discourse, whose discordances with respect to the real life of
Marcel Proust would constitute only secondary deviations. "His
use of the first person then," Germaine Bree accurately ob-
serves, "was the result of a conscious esthetic choice and not
proof that he considered his work as a confession or an au-
tobiography." 81 To have "Marcel's" life be told by "Marcel"
himself, after having had "Jean's" be told by the writer "C,"
arises indeed from a narrative choice as distinct, and thus as
significant, as Defoe's choice for Robinson Crusoe or Lesage's for
Gil Bias — and even more significant, because of the detour. But
we cannot fail to notice also that that conversion from the hetero-
diegetic to the autodiegetic accompanies and completes the other
conversion, already mentioned, of the metadiegetic to the diege-

1 Bree, p. 8.


Narrative Discourse

tic (or pseudo-diegetic). From Santeuil to the Recherche, the hero
could move from "he" to "I" without the stratification of the
narrating instances necessarily disappearing: it would be enough
for C.'s "novel" to be autobiographical, or even simply auto-
diegetic in form. Inversely, the double instance could be reduced
without modifying the relationship between hero and narrator:
it would be enough to suppress the preamble and begin with
something like, "For a long time Marcel had gone to bed
early. ..." We must therefore look at the full significance of the
dual conversion enacted by the transition from the narrative
system of Jean Santeuil to the narrative system of the Recherche.
If in every narrative we define the narrator's status both by its
narrative level (extra- or intradiegetic) and by its relationship to
the story (hetero- or homodiegetic), we can represent the four
basic types of narrator's status as follows: (1) extradiegetic-
heterodiegetic — paradigm: Homer, a narrator in the first degree
who tells a story he is absent from; (2) extradiegetic-
homodiegetic — paradigm: Gil Bias, a narrator in the first degree
who tells his own story; (3) intradiegetic-heterodiegetic —
paradigm: Scheherazade, a narrator in the second degree who
tells stories she is on the whole absent from; (4) intradiegetic-
homodiegetic — paradigm: Ulysses in Books IX -XII, a narrator in
the second degree who tells his own story. In this system the
(second) narrator of the quasi-totality of the narrative in San-
teuil, the fictive novelist C, falls into the same category that
Scheherazade does as intra-heterodiegetic, and the (single) nar-
rator of the Recherche into the diametrically (diagonally) opposite
category (whatever arrangement the entries are given) that Gil
Bias does, as extra-homodiegetic:







Gil Bias






We are dealing here with an absolute reversal, since we move
from a situation characterized by the complete dissociation of
the instances (first and extradiegetic author-narrator: "I"; sec-
ond narrator, intradiegetic novelist: "C"; metadiegetic hero:
"Jean") to the inverse situation, characterized by the merging
of all three instances in one single "person": the author-
narrator-hero Marcel. Most obviously significant in this turn-
around is the late, and deliberate, assumption of the form of
direct autobiography, which we must immediately connect to
the apparently contradictory fact that the narrative content of
the Recherche is less directly autobiographical than the narrative
content of Santeuil 82 — as if Proust first had had to conquer a
certain adhesion to himself, had to detach himself from himself,
in order to win the right to say "I," or more precisely the right to
have this hero who is neither completely himself nor completely
someone else say "I." So the conquest of the I here is not a
return to and attendance on himself, not a settling into the com-
fort of "subjectivity," 83 but perhaps exactly the opposite: the
difficult experience of relating to oneself with (slight) distance
and off-centering — a relationship wonderfully symbolized by
that barely suggested, seemingly accidental semihomonymy
of the narrator-hero and the signatory. 84

But this explanation clearly pays particular attention to the

82 See Tadie, pp. 20-23.

83 The famous Proustian "subjectivism" is nothing less than a proof of subjec-
tivity. And Proust himself did not fail to get angry at the too-facile conclusions
people drew from his narrative choice: "As I had the misfortune to begin my
book with I and could not change it anymore, I am 'subjective' in aeternum. If I
had begun instead, 'Roger Mauclair was occupying a summer house,' I would be
classified 'objective'" (to }. Boulanger, 30 November 1921, Correspondance
generale [Paris, 1932], III, 278).

84 On this controversial question, see M. Suzuki, "Le 'je' proustien," Bulletin
de la Societe des amis de Marcel Proust, 9 (1959); Harold Waters, "The Narrator, not
Marcel," French Review, 33 (February 1960), 389-392; and Muller, pp. 12 and
164-165. We know that the only two occurrences of this first name in the Re-
cherche are late (RH II, 429 and 488/P III, 75 and 157), and that the first is not
without a reservation. But it seems to me that this is not enough for us to reject
it. If we were to contest everything that is said only once ... On the other hand,
naming the hero Marcel is obviously not identifying him with Proust; but this
partial and fragile coincidence is highly symbolic.


Narrative Discourse

transition from heterodiegetic to autodiegetic and leaves some-
what in the background the suppression of the metadiegetic
level. The ruthless condensing of instances was perhaps already
underway in those pages of Jean Santeuil where the "I" of the
narrator (but which one?) supplanted as if inadvertently the
"he" of the hero: a result of impatience, undoubtedly, but not
necessarily impatience to "express himself" or to "narrate him-
self" by removing the mask of the novelistic fiction; irritation,
rather, at the obstructions or hindrances that the dissociation of
instances puts in the way of the stance of the discourse — which,
even in Santeuil, is not just a narrative discourse. Undoubtedly,
to a narrator so eager to accompany his "story" with that sort of
running commentary that is its underlying justification, nothing
is more annoying than to have to shift "voice" incessantly, nar-
rating the hero's experiences "in the third person" and then
commenting on them in his own name, with a continually re-
peated and always discordant intrusion. Whence the temptation
to leap over the obstruction, and lay claim to and finally annex
the experience itself, as in the passage where the narrator, after
having told the "feelings recaptured" by Jean when the coun-
tryside of Lake Geneva reminds him of the sea at Beg Meil,
continues with his own reminiscences, and his resolution to
write "only of what the past brings suddenly to life in a smell, in
a sight, in what has, as it were, exploded within me and set the
imagination quivering, so that the accompanying joy stirs me to
inspiration." 85 We see that here we are no longer dealing with
inadvertence: it is the narrative course as a whole chosen for
Santeuil which is revealed as inadequate, and which finally gives
way before the deepest needs and instances of the discourse. Such
"accidents" prefigure both the failure (or rather the approaching
abandonment) of Santeuil and its later resumption in the right
voice of the Recherche, the voice of direct autodiegetic narrating.
But, as we saw in the chapter on mood, this new course itself
is not without problems, since now into a narrative in autobio-
graphical form there has to be integrated a whole social chronicle
that often goes beyond the field of the hero's direct knowledge

85 Jean Santeuil, Pleiade, p. 401; trans. Hopkins, p. 410.



and sometimes, as is the case with Un amour de Swann, does not
easily enter even the narrator's knowledge. In fact, as B. G.
Rogers has shown, the Proustian novel manages only with much
difficulty to reconcile two contradictory courses. 86 The first is
that of an omnipresent speculative discourse, which barely ac-
commodates itself to classical "objective" narrating and which
requires the experience of the hero to merge with the past of the
narrator, who will thus be able to comment on it without seeming
to intrude (whence the ultimate adoption of a direct autodiegetic
narrating where the voices — of hero, narrator, and an author
turned toward a public to instruct and persuade — may mingle
and blend). The second is that of a comprehensive narrative con-
tent that widely overflows the hero's inner experience and at
times requires a quasi-"omniscient" narrator (whence the em-
barrassments and pluralities of focalization we have already met).
The narrative course in Jean Santeuil was doubtless untenable,
and its abandonment seems to us retrospectively "justified"; the
course in the Recherche is better suited to the needs of Proustian
discourse, but it is not by any means perfectly coherent. In fact,
the Proustian plan could be fully satisfied by neither the one nor
the other: neither the too-remote "objectivity" of heterodiegetic
narrative, which kept the narrator's discourse set apart from the
"action" (and thus from the hero's experience), nor the "subjec-
tivity" of autodiegetic narrative, too personal and seemingly too
confined to encompass without improbability a narrative con-
tent widely overflowing that experience. We are dealing here,
let us make clear, with the fictive experience of the hero, which
Proust, for well-known reasons, wished more limited than his
own personal experience. In a sense, nothing in the Recherche
exceeds Proust's experience, but everything he thought it neces-
sary to assign to Swann, Saint-Loup, Bergotte, Charlus, Mile.
Vinteuil, Legrandin, and many others obviously exceeds Mar-
cel's experience: a deliberate dispersion of the autobiographical
"material," which is responsible for certain narrative problems.
So — to cite only the two most flagrant paralepses — we can find it
strange that Marcel should have had access to Bergotte' s final

'' Rogers, pp. 120-141.


Narrative Discourse

thoughts, but not that Proust should have, since he had "lived"
them himself at the Jeu de Paume on a certain day in May 1921;
similarly, we can wonder that Marcel should so well read Mile.
Vinteuil's ambiguous feelings at Montjouvain, but much less, I
think, that Proust should have been able to ascribe them to her.
All this, and a lot more, comes from Proust, and we will not go
so far in disdaining the "referent" as to pretend to be unaware of
it; but we also know that he wanted to get it off his hands by
getting it off his hero's hands. So he needs both an "omniscient"
narrator capable of dominating a moral experience which is now
objectivized and an autodiegetic narrator capable of personally
taking up, authenticating, and illuminating by his own com-
mentary the spiritual experience which gives all the rest its ulti-
mate meaning and which, for its part, remains the hero's
privilege. Whence that paradoxical — and to some people
shameful — situation of a "first-person" narrating that is never-
theless occasionally omniscient. Here again — without wanting
to, perhaps unknowingly, and for reasons that result from the
profound (and profoundly contradictory) nature of its
purpose — the Recherche attacks the best-established convention
of novelistic narrating by cracking not only its traditional
"forms," but also — a more hidden and thus more decisive
loosening — the very logic of its discourse.


As in any narrative in autobiographical form, 87 the two actants
that Spitzer called erzahlendes Ich (the narrating I) and erzahltes
Ich (the narrated I) are separated in the Recherche by a difference
in age and experience that authorizes the former to treat the
latter with a sort of condescending or ironic superiority, very
noticeable for example in the scene of Marcel's missed introduc-
tion to Albertine, or that of the kiss denied. 88 But peculiar to the
Recherche, distinguishing it from almost all other autobiog-
raphies real or fictive, is that added to this essentially variable

87 In question here is classical autobiography, with subsequent narrating, and
not interior monologue in the present tense.

88 RH I, 642-643/P I, 855-856 and RH I, 698-699/P I, 933-934.



difference, inevitably decreasing in proportion as the hero pro-
gresses in "apprenticeship" to life, is a more radical and
seemingly absolute difference that is not reducible simply to
"development": the difference caused by the final revelation,
the decisive experience of involuntary memory and aesthetic
vocation. Here the Recherche parts company with the Bil-
dungsroman tradition and approaches certain forms of religious
literature, like Saint Augustine's Confessions: the narrator does
not simply know more, empirically, than the hero; he knows in
the absolute sense, he understands the Truth — a truth which the
hero does not approach with a gradual and continuous move-
ment, but which, quite to the contrary, despite the omens and
notices that have here and there preceded it, rushes in on him at
the very moment when in a certain way he feels himself more
distant than ever from it: "one knocks at all the doors which lead
nowhere, and then one stumbles without knowing it on the only
door through which one can enter — which one might have
sought in vain for a hundred years — and it opens of its own
accord." 89

This particular characteristic of the Recherche involves a crucial
consequence with respect to relations between the hero's dis-
course and the narrator's. Up until that moment, indeed, these
two discourses had been juxtaposed, interwoven, but, except
for two or three exceptions, 90 never completely merged: the
voice of error and tribulation could not be identified with the
voice of understanding and wisdom — Parsifal's voice with Gur-
nemanz's. On the contrary, starting with the final revelation (to
turn inside out the term Proust applied to Sodome I), the two
voices can blend and merge, or spell each other in a single

89 RH II, 997/P III, 866.

90 Usually during moments of aesthetic meditation, apropos of Elstir (RH I,
1017-1020/P II, 419-422), Wagner (RH II, 489-492/P III, 158-162), or Vinteuil (RH
II, 555-559/P III, 252-258), when the hero has a presentiment which will be
confirmed by the final revelation. Sodome I, which in one sense is a first revela-
tion scene, also presents features of coincidence between the two discourses, but
there the narrator takes care, at least once, to correct an error of the hero's (RH II,
24-25/P II, 630-631). An inverse exception is the final group of pages in Swann,
where it is the narrator who makes a pretense of sharing the point of view of
the character.


Narrative Discourse

speech, since henceforth the hero's I thought can be written "I
understood," "I observed," "I began to divine," "I was aware,"
"I knew," "I saw clearly," "the thought came to me," "I had
arrived then at the conclusion," "I understood," etc. 91 — that is,
can coincide with the narrator's J know. Whence that sudden
proliferation of indirect discourse, and its alternation with the
narrator's present discourse, without opposition or contrast. As
we have already noticed, the hero of the matinee is not yet
identified with the final narrator in act, since the work written by
the latter is yet to come for the former; but the two instances
have already met in "thought," that is, in speech, since they
share the same truth, which now can slip without clashing and
the need for correction from one discourse to the other, from
one tense (the hero's imperfect) to the other (the narrator's
present) — as is made very clear by this final sentence, so supple,
so free (so omnitemporal, Auerbach would say), a perfect illustra-
tion of its own subject:

But at least, if strength were granted me for long enough to accom-
plish my work, I should not fail, even if the result were to make
them resemble monsters, to describe men first and foremost as
occupying a place, a very considerable place compared with the
restricted one which is allotted to them in space, a place on the
contrary immoderately prolonged — for simultaneously, like giants
plunged into the years, they touch epochs that are immensely far
apart, separated by the slow accretion of many, many days — in the
dimension of Time.

Du moins, si elle m'etait laissee assez longtemps pour accomplir
mon oeuvre, ne manquerais-)e pas d'abord d'y decrire les hommes
(cela dut-i\ les faire ressembler a des etres monstrueux) comme
occupant une place si considerable, a cote de celle si restreinte qui
leur est reservee dans l'espace, une place au contraire prolongee
sans mesure — puisqu'ils touchent simultanement, comme des
geants plonges dans les annees, a des epoques si distantes, entre
lesquelles tant de jours sont venus se placer — dans le Temps.

91 RH II, 999-1 023/P III, 869-899.



Functions of the Narrator

That final modification, therefore, involves in a very percep-
tible way one of the main functions of the Proustian narrator. It
can seem strange, at first sight, to attribute to any narrator a role
other than the actual narrating, the act of telling the story, but in
fact we know well that the narrator's discourse, novelistic or
not, can take on other functions. Perhaps it is worth the trouble
to make a quick survey of them in order to appreciate better the
distinctiveness, in this respect, of Proustian narrating. It seems
to me that we can distribute these functions (rather as Jakobson
distributes the functions of language) 92 in accordance with the
several aspects of narrative (in the broad sense) to which they
are connected.

The first of these aspects is obviously the story, and the func-
tion connected to it is the properly narrative function, which no
narrator can turn away from without at the same time losing his
status as narrator, and to which he can quite well try — as some
American novelists have — to reduce his role. The second aspect
is the narrative text, which the narrator can refer to in a dis-
course that is to some extent metalinguistic (metanarrative, in
this case) to mark its articulations, connections, interrelation-
ships, in short, its internal organization: these "stage directions"
of the discourse, 93 which Georges Blin called "directing indi-
cations," 94 belong to a second function that we can call direct-
ing function.

The third aspect is the narrating situation itself, whose two
protagonists are the narratee — present, absent, or implied — and
the narrator. The function that concerns the narrator's orienta-
tion toward the narratee — his care in establishing or maintaining
with the narratee a contact, indeed, a dialogue (actual, as in La
Maison Nucingen, or fictive, as in Tristram Shandy) — recalls both

92 Roman Jakobson, "Closing Statement: Linguistics and Poetics," in Thomas
A. Sebeok, ed., Style in Language (Cambridge, Mass.: M.I.T. Press, 1960), pp.

93 Barthes, "Le Discours de l'histoire," p. 66.

94 Regiebemerkungen (Stendhal et les problems du roman, p. 222).


Narrative Discourse

Jakobson's "phatic" (verifying the contact) and his "conative"
(acting on the receiver) functions. Rogers calls narrators of the
Shandian type, always turned toward their public and often
more interested in the relationship they maintain with that pub-
lic than in their narrative itself, "raconteurs." 95 At one time they
would have been called "talkers," and perhaps we should name
the function they tend to privilege the function of communication.
We know what importance it acquires in the epistolary novel,
and perhaps particularly in those forms that Jean Rousset calls
"epistolary monodies," such as, obviously, the Lettres por-
tuguaises, where the absent presence of the receiver becomes the
dominant (obsessive) element of the discourse.

The narrator's orientation toward himself, finally, brings
about a function very homologous with the one Jakobson
names, a little unfortunately, the "emotive" function: this is the
one accounting for the part the narrator as such takes in the
story he tells, the relationship he maintains with it — an affective
relationship, of course, but equally a moral or intellectual one. It
may take the form simply of an attestation, as when the narrator
indicates the source of his information, or the degree of preci-
sion of his own memories, or the feelings which one or another
episode awakens in him. 96 We have here something which
could be called testimonial function, or function of attestation. But
the narrator's interventions, direct or indirect, with regard to the
story can also take the more didactic form of an authorized
commentary on the action. This is an assertion of what could be
called the narrator's ideological function; 97 and we know that Bal-
zac, for example, greatly developed this form of explanatory and

95 Rogers, p. 55.

96 "In writing this I feel my pulse quicken yet; those moments will always be
with me, were I to live a hundred thousand years" (Rousseau, Confessions,
already quoted on pp. 67-68). But the narrator's attestation may also bear on
events contemporary with the act of narrating and unconnected to the story he is
telling: for example, the pages in Doctor Faustus on the war that rages while
Zeitblom is writing his memories of Leverkuhn.

97 Which is not necessarily the author's: the judgments of Des Grieux do not
a priori commit the Abbe Prevost, and those of the fictive author-narrator of
Leuwen or the Chartreuse by no means commit Henry Beyle.



justificatory discourse — for him, as for so many others, a vehicle
of realistic motivation.

These five functions are certainly not to be put into watertight
compartments; none of the categories is completely unadulter-
ated and free of complicity with others, none except the first is
completely indispensable, and at the same time none, however
carefully an author tries, can be completely avoided. It is rather a
question of emphasis and relative weight: everyone knows that
Balzac "intervenes" in his narrative more than Flaubert, that
Fielding addresses the reader more often than Mme. de La
Fayette does, that the "directing indications" are more indis-
creet in James Fenimore Cooper 98 or Thomas Mann 99 than in
Hemingway, etc., but we will not claim to derive some cumber-
some typology from that.

Nor will we go back to the various manifestations, already
encountered elsewhere, of the Proustian narrator's extranarra-
tive functions: addresses to the reader, organization of the nar-
rative by means of advance notices and recalls, indications of
source, memory-elicited attestations. What remains for us to
emphasize, here, is the situation of the narrator's quasi-
monopoly with regard to what we have christened the ideologi-
cal function, and the deliberate (nonobligatory) nature of this
monopoly. In fact, of all the extranarrative functions, this is the
only one that does not of necessity revert to the narrator. We
know how careful great ideological novelists like Dostoevski,

98 "It is necessary, in order that the thread of the narrative should not be spun
to a length which might fatigue the reader, that he should imagine a week to
have intervened between the scene with which the preceding chapter closed and
the events with which it is our intention to resume its relation in this"; "It is
proper that the course of the narrative should be stayed, while we revert to those
causes which have brought in their train of consequences, the singular contest
just related. The interruption must necessarily ..." etc. (James Fenimore Cooper,
The Prairie, chaps. 8, 15 [New York: Holt, Rinehart and Winston, 1950], pp.
92, 178).

99 "Since the foregoing section has swollen out of all conscience, I shall do well
to begin a new one." "The chapter just finished is also, for my taste, much too
extended." "I will not look back, I will take care not to count the pages I have
covered between the last Roman numeral and this one I have just written down"
(Thomas Mann, Doctor Faustus, chaps. 4, 5, 9, trans. H. T. Lowe-Porter [New
York: Knopf, 1948], pp. 21, 30, 70).


Narrative Discourse



Tolstoy, Mann, Broch, Malraux were to transfer onto some of
their characters the task of commentary and didactic
discourse — going so far as to transform such scenes from The
Possessed, The Magic Mountain, or L'Espoir [Man's Hope] into ver-
itable colloquia of speculation. Nothing of the sort takes place in
Proust, who, other than Marcel, has given himself no "spokes-
man." A Swann, a Saint-Loup, a Charlus, despite all their intel-
ligence, are objects of observation, not organs of truth or even
genuine interlocutors (we know, moreover, what Marcel thinks
of the intellectual qualities of conversation and friendship): their
errors, their absurdities, their failures and fallings-off are more
instructive than their opinions. Even such figures of artistic crea-
tion as Bergotte, Vinteuil, or Elstir do not intervene, so to speak,
as custodians of an authorized speculative discourse: Vinteuil is
mute and Bergotte is reticent or trivial, and the meditation on
their work reverts to Marcel; 100 Elstir begins, symbolically, with
M. Biche's art-student antics, and the statements he makes at
Balbec matter less than the silent teaching of his canvases. In-
tellectual conversation is a genre plainly contrary to Proustian
taste. We know the disdain inspired in him by everything that
"thinks" — like, according to him, the Hugo of the early poems,
"instead of contenting himself, like Nature, with supplying food
for thought." 101 All humanity, from Bergotte to Franchise and
from Charlus to Mme. Sazerat, is before him like "Nature,"
entrusted with provoking thought, not expressing it. An ex-
treme case of intellectual solipsism. Ultimately, and in his own
way, Marcel is an autodidact.

The consequence is that no one — except the hero under cer-
tain conditions — is able or allowed to contest with the narrator
his privilege of ideological commentary: whence the well-
known proliferation of this "auctorial" discourse, to borrow
from German critics a term which indicates both the presence of
the author (actual or fictive) and the sovereign authority of that

100 " 'Was this perhaps that happiness which the little phrase of the sonata
promised to Swann and which he, because he was unable to find it in artistic
creation, mistakenly assimilated to the pleasures of love. . . '"(RH II, 1006/P III,

1(11 RH I, 1107/P II, 549.

presence in his work. The quantitative and qualitative impor-
tance of this psychological, historical, aesthetic, metaphysical
discourse is such, despite the denials, 102 that we can undoubt-
edly attribute to it the responsibility— and in one sense the
credit— for the strongest shock given in this work, and by this
work, to the traditional equilibrium of novelistic form. If the
Recherche du temps perdu is experienced by everyone as being
"not completely a novel any more" and as the work which, at its
level, concludes the history of the genre (of the genres) and,
along with some others, inaugurates the limitless and indefinite
space of modern literature, the cause is obviously — and this time
too despite the "author's intentions" and through the effect of a
movement all the more irresistible because involuntary — this in-
vasion of the story by the commentary, of the novel by the
essay, of the narrative by its own discourse.

The Narratee

Such speculative imperialism, such certainty of truth, could
lead one to think that the receiver's role here is purely passive,
that he is limited to receiving a message he must take or leave
and to "consuming" after the event a work that was completed
far from him and without him. Nothing would be more contrary
to Proust's convictions, to his own experience of reading, and to
the most powerful demands of his work.

Before considering this final dimension of the Proustian nar-
rating instance we must say a more general word about this
personage that we have called the narratee, and whose function
in the narrative seems so variable. Like the narrator, the narratee
is one of the elements in the narrating situation, and he is neces-
sarily located at the same diegetic level; that is, he does not
merge a priori with the reader (even an implied reader) any
more than the narrator necessarily merges with the author.

To an intradiegetic narrator corresponds an intradiegetic nar-

102 "Hence the temptation for the writer to write intellectual works, which is,
however, a gross mistake. A work in which there are theories is like an object
which still has the ticket that shows its price" (RH II, 1009/P III, 882). Doesn't the
reader of the Recherche know what it costs?


Narrative Discourse

ratee; and the narrative of Des Grieux or Bixiou is not addressed
to the reader of Manon Lescaut or of La Maison Nucingen, but
indeed only to M. de Renoncourt, only to Pinot, Couture, and
Blondet; they alone are designated by the "second-person"
marks present on occasion in the text, just as the "second-
person" marks we find in an epistolary novel can designate only
the epistolary correspondent. We, the readers, cannot identify
ourselves with those fictive narratees anymore than those in-
tradiegetic narrators can address themselves to us, or even as-
sume our existence. 103 For we can neither interrupt Bixiou nor
write to Mme. de Tourvel.

The extradiegetic narrator, on the other hand, can aim only at
an extradiegetic narratee, who merges with the implied reader
and with whom each real reader can identify. This implied
reader is in principle undefined, although Balzac does turn
particularly sometimes toward a reader from the provinces,
sometimes toward a Parisian reader, and Sterne sometimes calls
him Madam or Sir Critick. The extradiegetic narrator can also
pretend, like Meursault, to address no one, but this posture —
fairly widespread in the contemporary novel — obviously cannot
change the fact that a narrative, like every discourse, is necessar-
ily addressed to someone and always contains below the surface
an appeal to the receiver. And if the existence of an intradiegetic
narratee has the effect of keeping us at a distance, since he is
always interposed between the narrator and us — as Finot, Cou-
ture, and Blondet are interposed between Bixiou and the nosy
listener behind the partition, for whom that narrative was not
intended (but, Bixiou says, "there is always someone off to the
side") — it is also true that the more transparent the receiving
instance and the more silent its evocation in the narrative, so
undoubtedly the easier, or rather the more irresistible, each real
reader's identification with or substitution for that implied in-
stance will be.

It is indeed this relationship — despite some rare and fully

103 A special case is the metadiegetic literary work, of the Curious Impertinent or
jean Santeuil kind, which can possibly aim at a reader, but a reader who in prin-
ciple is himself fictive.



needless challenges we have already called attention to that

the Recherche maintains with its readers. Every one of them
knows himself to be the implied— and anxiously awaited—
narratee of this swirling narrative that, in order to exist in its
own truth, undoubtedly needs, more than any other narrative
does, to escape the closure of "final message" and narrative
completion, to resume endlessly the circular movement from the
work to the vocation it "tells" and from the vocation back to the
work it gives rise to, and so on unceasingly.

As the very terms of the famous letter to Riviere make clear, 104
the "dogmatism" and "structure" of the Proustian work do not
dispense with a continual resort to the reader, who is entrusted
not only with "guessing" them before they are expressed, but
also, once they have been revealed, with interpreting them and
placing them back into the movement which both generates
them and carries them off. Proust could not exempt himself
from the rule he enunciates in the Temps retrouve, a rule granting
the reader the right to translate the universe of the work into his
own terms in order then to "give to what he is reading its full
general import": whatever apparent infidelity they commit, "in
order to read with understanding many readers require to read
in their own particular fashion, and the author must not be
indignant at this; on the contrary, he must leave the reader all
possible liberty," because the work is ultimately, according to
Proust himself, only an optical instrument the author offers the
reader to help him read within himself. "For it is only out of
habit, a habit contracted from the insincere language of prefaces
and dedications, that the writer speaks of 'my reader.' In reality
every reader is, while he is reading, the reader of his own
self." 105

Such is the vertiginous status of the Proustian narratee: in-
vited not, like Nathanael, to "throw this book away," 106 but to
rewrite it, being totally unfaithful and wonderfully exact, like

104 "At last I find a reader who guesses that my book is a dogmatic work and a
structure!" (Choix de lettres, p. 197).

105 RH II, 1031-1032/P III, 910-911.

106 (Translator's note.] Nathanael is the character addressed by the "first-
person" narrator of Gide's Les Nourritures terrestres (The Fruits of the Earth).


Narrative Discourse

Pierre Menard inventing Quixote word for word. 107 Everyone
understands what is expressed by that fable, circulated from
Proust to Borges and from Borges to Proust, and illustrated per-
fectly in the small adjoining drawing rooms of La Maison
Nucingen: the real author of the narrative is not only he who tells
it, but also, and at times even more, he who hears it. And who is
not necessarily the one it is addressed to: there are always
people off to the side.


7 (Translator's note.] In Borges' s story "Pierre Menard, Author of Don Qui.


To conclude without useless recapitulations, here are some
words of self-criticism, or, if one likes, of excuse. The categories
and procedures put forward here are certainly not faultless in
my eyes: it has been a question, as it often is, of choosing be-
tween drawbacks. In an area we regularly grant to intuition and
empiricism, the proliferation of concepts and terms will doubt-
less have annoyed more than one reader, and I do not expect
"posterity" to retain too large a part of these propositions. This
arsenal, like any other, will inevitably be out of date before
many years have passed, and all the more quickly the more
seriously it is taken, that is, debated, tested, and revised with
time. One of the characteristics of what we can call scientific effort
is that it knows itself to be essentially decaying and doomed to
die out: a wholly negative trait, certainly, and one rather melan-
choly to reflect on for the "literary" mind, always inclined to
count on some posthumous glory; but if the critic can dream of
an achievement in the second degree, the poetician for his part
knows that he labors in — let us say rather at — the ephemeral, a
worker aware of becoming un-worked.

Therefore I think, and hope, that all this technology —
prolepses, analepses, the iterative, focalizations, paralipses, the
metadiegetic, etc. — surely barbaric to the lovers of belles lettres,
tomorrow will seem positively rustic, and will go to join other
packaging, the detritus of Poetics; let us only hope that it will
not be abandoned without having had some transitory useful-



Narrative Discourse



ness. Occam, already uneasy about the progress of intellectual
pollution, forbade us ever needlessly to invent creatures of
reason — today we would say theoretical objects. I would be an-
noyed with myself if I fell short of this rule, but it seems to me
that at least some of the literary forms designated and defined
here call for further investigations, which for obvious reasons
were not more than touched on in this work. So I hope to have
furnished the theory of literature and the history of literature
with some objects of study that are no doubt minor, but a little
trimmer than the traditional entities, such as "the novel" or

The specific application of these categories and procedures to
the Recherche du temps perdu was perhaps even more offensive,
and I cannot deny that the purpose of my work is defined almost
exactly by the opposite view to what is expressed in the prelimi-
nary statement of a recent, excellent study on the art of the novel
in Proust, a statement which no doubt meets immediately with
the unanimous acceptance of well-thinking people:

We did not want to impose on Proust's work categories external to
it, or a general idea of the novel or of the way in which one should
study a novel; we did not want a treatise on the novel, with illustra-
tions taken from the Recherche, but concepts arising from the work,
and allowing us to read Proust as he read Balzac and Flaubert. The
onlv theory of literature is in criticism of the particular. 1

We can certainly not maintain that here we are using concepts
exclusively "arising from the work," and the description here
of Proustian narrative can hardly be considered to conform to
Proust's own idea of it. Such a gap between indigenous theory
and critical method might seem inappropriate, like all anachro-
nisms. It seems to me, however, that one should not rely blindly
on the explicit aesthetics of a writer, even if he is a critic as in-
spired as the author of the Contre Sainte-Beuve. The aesthetic con-
sciousness of an artist, when he is major, is so to speak never at
the level of his practice, and this is only one manifestation of what

1 Tadie, Proust et le roman,


Hegel symbolized by the late flight of Minerva's owl. We do not
have at our disposal one hundredth of Proust's genius, but we
do have the advantage over him (which is a little like the live don-
key's advantage over the dead lion) of reading him precisely from
the vantage point of what he contributed to fathering (fathering
that modern literature which owes him so much) and thus the
advantage of perceiving clearly in his work what was there only
in its nascent state — all the more nascent because with him the
transgression of norms, the aesthetic invention, are most often,
as we have seen, involuntary and sometimes unconscious. His
goal was otherwise, and this scorner of the avant-garde is almost
always a revolutionary despite himself (I would certainly say that
that is the best way to be one if I didn't have the faint suspicion
that it is the only way). To repeat it once more and following so
many others, we read the past by the light of the present, and is
not that how Proust himself read Balzac and Flaubert, and does
one really believe that his were critical concepts "arising from"
the Comedie humaine or the Education sentimentale?

In the same way, perhaps, the sort of scanning "imposed" here
on the Recherche has allowed us, I hope, to reveal in that novel,
under this new lighting, some aspects that Proust himself, and
until now Proustian criticism, often overlooked (the importance
of iterative narrative, for example, or of the pseudo-diegetic), or
has allowed us to characterize more precisely features already
spotted, such as anachronies or multiple focalizations. The "grid"
which is so disparaged is not an instrument of incarceration, of
bringing to heel, or of pruning that in fact castrates: it is a proce-
dure of discovery, and a way of describing.

That does not mean — as readers may already have noticed —
that its user forbids himself all preference and all aesthetic
evaluation, or even all bias. It has no doubt become evident, in
this comparison of Proustian narrative with the general system
of narrative possibilities, that the analyst's curiosity and predi-
lection went regularly to the most deviant aspects of Proustian
narrative, the specific transgressions or beginnings of a future
development. This systematic valuing of originality and inno-
vation is perhaps somewhat unsophisticated and altogether


Narrative Discourse

romantic as well, but today no one can entirely escape it. Roland
Barthes in S/Z gives a highly convincing justification of it: "Why
is the writerly [what can be written today] our value? Because
the goal of literary work (of literature as work) is to make the
reader no longer a consumer, but a producer of the text." 2 The
preference for what, in Proust's text, is not only "readerly"
(classical) but "writerly" (let us roughly interpret: modern)
perhaps expresses the critic's desire, or even the poetician's,
when in contact with the aesthetically "subversive" points of the
text, to play a role vaguely more active than simply that of ob-
server and analyst. The reader, here, believes he is participating
in and to a minute extent (minute, but decisive) contributing to
creation; and perhaps, by recognition alone — or rather by bring-
ing to light features which the work invented, often without its
author's knowledge — in reality he is. This contribution, or even
this intervention, was, again let us remember, a little more than
legitimate in Proust's eyes. The poetician for his part is also the
"reader of his own self," and to discover is always (as modern
science also tells us) somewhat to invent.

Another choice made, in this case a choice refused, will
perhaps explain why this "conclusion" is not one — I mean: why
readers will not find here a final "synthesis" in which all the
characteristic features of Proustian narrative noted in the course
of this study will meet and justify themselves to each other.
When such convergences or correlations appear unchallenge-
able (for example, between the disappearance of summary and
the emergence of the iterative, or between polymodality and the
elimination of the metadiegetic), we have not failed to acknowl-
edge them and to elucidate them. But it would be unfortunate, it
seems to me, to seek "unity" at any price, and in that way to
force the work's coherence — which is, of course, one of criti-
cism's strongest temptations, one of its most ordinary (not to say
most common) ones, and also one most easy to satisfy, since all
it requires is a little interpretative rhetoric.

Now, if we cannot deny in Proust the will for coherence and
the striving for design, just as undeniable in his work is the

2 Barthes, S/Z, p. 4.



resistance of its matter and the part played by what is
uncontrolled — perhaps uncontrollable. We have already noted
the retroactive nature (here as in Balzac or Wagner) of a unity
belatedly won over material that was heterogeneous and not
originally in harmony. Just as obvious is the part played by the
incompletion due to the somewhat supplementary labor which
the accidental stay of 1914 brought to the work. The Recherche du
temps perdu was, without doubt, at least in Proust's mind, a
"finished" work: that was in 1913, and the perfect ternary com-
position of that period {Cote de chez Swann, Cote de Guermantes,
Temps retrouve) bears witness to it in its own way. But we know
what happened to it, and no one can claim that the present
structure of the Recherche is the result of anything other than
circumstances: one active cause, the war, and one negative
cause, death. Nothing, certainly, is easier than to justify the
action of chance and "demonstrate" that the Recherche finally,
on November 18, 1922, found the perfect balance and the exact
proportion which had been missing until then, but it is just this
easy way out that we are rejecting here. If the Recherche was
complete once, it is not so anymore, and the way in which it
admitted the extraordinary later expansion perhaps proves that
that temporary completion was, like all completion, only a retro-
spective illusion. We must restore this work to its sense of un-
fulfillment, to the shiver of the indefinite, to the breath of the
imperfect. The Recherche is not a closed object: it is not an object.
Here again, no doubt Proust's (involuntary) practice goes be-
yond his theory and his plan — let us say at least that it corre-
sponds better to our desire. The harmonious triptych of 1913 has
doubled its area, but on one side only, the first panel necessarily
remaining consistent with the original blueprint. This imbal-
ance, or decentering, pleases us as it is and in its unpremeditated-
ness; and we will be very careful not to motivate it by "explain-
ing" a nonexistent closure and an illusory design, and not to
reduce improperly what Proust, apropos of something else,
called the "contingency of the narrative." 3 Laws of Proustian
narrative are, like that narrative itself, partial, defective, perhaps

3 ]ean Santeuil, Pleiade, p. 314. [Translator's note: my translation; the Hopkins
translation, which is very free, is on p. 115.]


Narrative Discourse

foolhardy: quite empirical and common laws which we should
not hypostatize into a Canon. Here the code, like the message,
has its gaps and its surprises.

But undoubtedly this rejection of motivation is in its own way
a motivation. We do not escape the pressure of the signified: the
semiotic universe abhors a vacuum, and to name contingency is
already to assign it a function, to give it a meaning. Even — or
especially? — when he is silent, the critic says too much. Perhaps
the best thing would be, as with Proustian narrative itself, never
to "finish," which is, in one sense, never to start.



Works by Proust

A la recherche du temps perdu. Ed. Pierre Clarac and Andre Ferre. 3 vols.
Paris: La Pleiade, Gallimard, 1954. (Tr. Remembrance of Things Past.
Trans. C. K. Scott Moncrieff and Andreas Mayor. 2 vols. New York:
Random House, 1934, 1970.)

Jean Santeuil, preceded by Les Plaisirs et les jours. Ed. Pierre Clarac and
Yves Sandre. Paris: La Pleiade, Gallimard, 1971. (Tr. jean Santeuil.
Trans. Gerard Hopkins. New York: Simon & Schuster, 1956. Tr.
Pleasures and Regrets. Trans. Louise Varese. New York: Crown, 1948.)

Contre Sainte-Beuve, preceded by Pastiches et melanges and followed by
Essai's et articles. Ed. Pierre Clarac and Yves Sandre. Paris: La Pleiade,
Gallimard, 1971. (Tr. [Contre Sainte-Beuve] Marcel Proust on Art and
Literature, 1896-1919. Trans. Sylvia Townsend Warner. New York:
Meridian, 1958. Tr. [selections from Pastiches and Essai's] Marcel
Proust: A Selection from His Miscellaneous Writings. Trans. Gerard Hop-
kins. London: Allan Wingate, 1948.)

Correspondance generate. Paris: Plon, 1930-1936.

Choix de lettres. Ed. Philip Kolb. Paris: Plon, 1965.



For various rough drafts or variants of the Recherche, see the follow-

Du cote de chez Swarm. Paris: Grasset, 1913.

Chroniques. Paris: Gallimard, 1927.

Contre Sainte-Beuve, followed by Nouveaux Melanges. Ed. Bernard de
Fallois. Paris: Gallimard, 1954.

Textes retrouves. Ed. Philip Kolb and L. B. Price. Urbana: University ot
Illinois Press, 1968. And Cahiers Marcel Proust. Paris: Gallimard, 1971.







1 1 i

Andre Maurois. A la recherche de Marcel Proust. Paris: Hachette, 1949.

(Tr. Proust: Portrait of a Genius. Trans. Gerard Hopkins. New York:

Harper, 1950.)
Maurice Bardeche. Marcel Proust romancier. Vol. I. Paris: Les Sept

Couleurs, 1971.

Critical and Theoretical Studies

Aristotle. Poetics.

Auerbach, Erich. Mimesis: The Representation of Reality in Western Litera-
ture. Trans. Willard Trask. 1953; rpt. Garden City, N.Y.: Anchor-
Doubleday, 1957.

Balzac, Honore de. Etudes sur M. Beyle. 1840; rpt. Geneva: Skira, 1943.

Bardeche, Maurice. Marcel Proust romancier. Vol. I. Paris: Les Sept
Couleurs, 1971.

Barthes, Roland. "Introduction a 1'analyse structurale des recits."
Communications, 8. (Tr. "An Introduction to the Structural Analysis of
Narrative." NLH, 6 [Winter 19751, 237-272.)

. "Le Discours de l'histoire." Information sur les sciences sociales,

August 1967.

. "L'Effet de reel." Communications, 11 (1968).

. S/Z. Paris: Seuil, 1970. (Tr. SIZ. Trans. Richard Miller. New York:

Hill and Wang, 1974.)

Bentley, Phyllis. "Use of Summary." In Some Observations on the Art of
Narrative. 1947. Rpt. in The Theory of the Novel. Ed. Philip Stevick.
New York: Free Press, 1967, pp. 47-52.

Benveniste, Emile. Problemes de linguistique generate. Paris: Gallimard,
1966. (Tr. Problems in General Linguistics. Trans. Mary Elizabeth Meek.
Coral Gables, Fla.: University of Miami Press, 1971.)

Blin, Georges. Stendhal et les problemes du roman. Paris: Corti, 1954.

Booth, Wayne. "Distance and Point of View." Essays in Criticism, 11
(1961), 60-79. (Tr. "Distance et point de vue." Poetique, 4 [1970].)

. The Rhetoric of Fiction. Chicago: University of Chicago Press, 1961.

Borges, Jorge Luis. Discussions. Paris: Gallimard, 1966.

. Other Inquisitions, 1937-1952. Trans. Ruth L. C. Simms. Austin:

University of Texas Press, 1964.

Bowling, Lawrence E. "What Is the Stream of Consciousness Tech-
nique?" PMLA, 65 (1950), 333-345.

Bree, Germaine. Du temps perdu au temps retrouve. 1950; rpt. Les Belles
Lettres, 1969. (Tr. Marcel Proust and Deliverance from Time. Trans. C. J.
Richards and A. D. Truitt. 2d ed. New Brunswick, N.J.: Rutgers Uni-
versity Press, 1969.)

Brooks, Cleanth, and Robert Penn Warren. Understanding Fiction. New

York: Crofts, 1943.
Daniel, Georges. Temps et mystification dans A.L.R.T.P. Paris- Nizet

Debray-Genette, Raymonde. "Les Figures du recit dans Un coeur sim-
ple." Poetique, 3 (1970).
. "Du mode narratif dans les Trois Contes." Litterature, 2 (May

Dujardin, Edouard. Le Monologue interieur. Paris: Messein, 1931.
Feuillerat, Albert. Comment Proust a compose son roman. New Haven:

Yale University Press, 1934.
Fitch, Brian T. Narrateur et narration dans "I'Etranger" d' Albert Camus.

Paris: Archives des Lettres modernes, No. 34, 1961. 2d rev. ed., 1968.
Forster, E. M. Aspects of the Novel. London: Edward Arnold, 1927.
Friedman, Melvin. Stream of Consciousness: A Study in Literary Method.

New Haven: Yale University Press, 1955.
Friedman, Norman. "Point of View in Fiction." PMLA, 70 (1955). Rpt.

in The Theory of the Novel. Ed. Philip Stevick. New York: Free Press,

1967, pp. 108-137.
Genette, Gerard. Figures. Paris: Seuil, 1966.

. Figures II. Paris: Seuil, 1969.

Greimas, A. J. Semantique structurale. Paris: Larousse, 1966.
Guiraud, Pierre. Essais de stylistique. Paris: Klincksieck, 1971.
Hachez, Willy. "La Chronologie et l'age des personnages de

A.L.R.T.P." Bulletin de la Societe des amis de Marcel Proust, 6


. "Retouches a une chronologie." BSAMP, 11 (1961).

. "Fiches biographiques de personnages de Proust." BSAMP, 15

Hamburger, Kate. Die Logik der Dichtung. Stuttgart: Ernst Klett Verlag,

1957. (Tr. The Logic of Literature. Trans. MarilynnJ. Rose. 2d. rev. ed.

Bloomington: Indiana University Press, 1973.)
Houston, J. P. "Temporal Patterns in A.L.R.T.P." French Studies, 16

(January 1962), 33^4.
Huet, J. B. Traite de I'origine des romans. 1670.

Humphrey, Robert. Stream of Consciousness in the Modern Novel. Ber-
keley: University of California Press, 1954.
Jakobson, Roman. "Quest for the Essence of Language." Diogenes 51

(Fall 1965), 21-37.
Jauss, H. R. Zeit und Erinnerung in Marcel Prousts A.L.R.T.P. 1965; rpt.

Heidelberg: Carl Winter, 1970.
Larnmert, Eberhart. Bauformen des Erzahlens. Stuttgart: J. B. Metzlersche

Verlag, 1955.








Lefebve, Maurice-Jean. Structure du discours de la poesie et du recit.

Neuchatel: La Baconniere, 1971.
Lips, Marguerite. Le Style indirect libre. Paris: Payot, 1926.
Lubbock, Percy. The Craft of Fiction. London, 1921; rpt. New York:

Viking, 1957.
Martin-Chauffier, Louis. "Proust et le double 'Je' de quatre personnes."

In Problemes du roman (Confluences). 1943. Partially rpt. in Les Critiques

de notre temps et Proust. Ed. Jacques Bersani. Paris: Gamier, 1971, pp.

54-66. (Tr. "Proust and the Double I." Partisan Review, 16 [October

1949J, 1011-1026.)
Maurois, Andre. A la recherche de Marcel Proust. Paris: Hachette, 1949.

(Tr. Proust: Portrait of a Genius. Trans. Gerard Hopkins. New York:

Harper, 1950.)
Mendilow, A. A. Time and the Novel. New York: Humanities Press,

Metz, Christian. Essais sur la signification au cinema. Vol. I. Paris:

Klincksieck, 1968. (Tr. Film Language: A Semiotics of the Cinema. Trans.

Michael Taylor. New York: Oxford University Press, 1974.)
Miiller, Gunther. "Erzahlzeit und erzahlte Zeit," In Festschrift fur P.

Kluckhohn und Hermann Schneider. 1948. Rpt. in Morphologische Poetik.

Tubingen, 1968.
Muller, Marcel. Les Voix narratives dans A.L.R.T.P. Geneva: Droz, 1965.
Painter, George D. Proust: The Early Years and Proust: The Later Years.

New York: Atlantic -Little, Brown, 1959 and 1965. (Tr. Paris: Mercure,

Picon, Gaetan. Malraux par lui-meme. Paris: Seuil, 1953.
Plato. Republic.

Pouillon, Jean. Temps et roman. Paris: Gallimard, 1946.
Raible, Wolfgang. "Linguistik und Literaturkritik." Linguistik und Di-

daktik, 8 (1971).
Raimond, Michel. La Crise du roman, des lendemains du naturalisme aux

annees 20. Paris: Corti, 1966.
Ricardou, Jean. Problemes du nouveau roman. Paris: Seuil, 1967.
Richard, Jean-Pierre. "Proust et l'objet hermeneutique." Poetique, 13

Rogers, Brian G. Proust's Narrative Techniques. Geneva: Droz, 1965.
Romberg, Bertil. Studies in the Narrative Technique of the First-Person

Novel. Trans. Michael Taylor and Harold H. Borland. Stockholm:

Almqvist and Wiksell, 1962.
Rossum-Guyon, Francoise van. "Point de vue ou perspective narra-
tive." Poetique, 4 (1970).

. Critique du roman. Paris: Gallimard, 1970.

Rousset, Jean. Forme et signification. Paris: Corti, 1962.

Sartre, Jean-Paul. "M. Francois Mauriac et la liberie. " 1939. In Situa-
tions I. Paris: Gallimard, 1947. (Tr. "Francois Mauriac and Freedom."
In Literary and Philosophical Essays. Trans. Annette Michelson. New
York: Criterion Books, 1955, pp. 7-23.)

�� . L'ldiot de la famille. Paris: Gallimard, 1971.

Spitzer, Leo. "Zum Stil Marcel Prousts." In Stilstudien. Munich:
Hueber, 1928. (Tr. "Le Style de Marcel Proust." In Etudes de style.
Paris: Gallimard, 1970.)

Stang, Richard. The Theory of the Novel in England, 1850-1870. New
York: Columbia University Press, 1959.

Stanzel, F. K. Die typischen Erzahlsituationen in Roman. Vienna: Wilhelm
Braumiiller, 1955. (Tr. Narrative Situations in the Novel: Tom fones,
Moby-Dick, The Ambassadors, Ulysses. Trans. James P. Pusack.
Bloomington: Indiana University Press, 1971.)

Tadie, Jean-Yves. Proust et le roman. Paris: Gallimard, 1971.

Todorov, Tzvetan. "Les Categories du recit litteraire." Communications,
8 (1966).

. Litterature et signification. Paris: Larousse, 1967.

. "Poetique." In Qu'est-ce que le structuralisme? Paris: Seuil, 1968.

. Grammaire du Decameron. The Hague: Mouton, 1969.

. Poetique de la prose. Paris: Seuil, 1971. (Tr. The Poetics of Prose.

Trans. Richard Howard. Ithaca, N.Y.: Cornell University Press; Lon-
don: Blackwell, 1977.)

. "La Poetique en U.R.S.S." Poetique 9 (1972).

Uspenski, Boris. Poetika Kompozicii. Moscow, 1970. (Tr. A Poetics of
Composition: The Structure of the Artistic Text and Typology of a Compo-
sitional Form. Trans. Valentina Zavarin and Susan Wittig. Berkeley:
University of California Press, 1973.) See also "Poetique de la compo-
sition." Poetique, 9 (1972).

Vigneron, Robert. "Genese de Swann." Revue d'histoire de la philosophie,
January 1937, pp. 67-115.

. "Structure de Swann: Balzac, Wagner et Proust." French Review,

19 (May 1946), 370-384.

. "Structure de Swann: pretentions et defaillances." MP, 44

(November 1946), 102-128.

. "Structure de Swann: Combray ou le cercle parfait." MP, 45 (Feb-
ruary 1948), 185-207.

Waters, Harold. "The Narrator, Not Marcel." French Review, 33 (Feb-
ruary 1960), 389-392.

Wellek, Rene, and Austin Warren. Theory of Literature. New York: Har-
court Brace, 1949. (Tr. La Theorie litteraire. Paris: Seuil, 1971.)

Zeraffa, Michel. Personne et personnage, le romanesque des annees 1920 aux
annees 1950. Paris: Klincksieck, 1969.





Abruption, 151n

Achrony. See Order

Adolphe (Constant), 186, 230

Advance mention. See Order: prolep-

Advance notice. See Order: prolepsis

Adventure novels, 190

Agostinelli, Alfred, 90n, 99

L'Agrandissement (Claude Mauriac),

Albert Savarus (Balzac), 230-231

Alterations. See Mood, perspective

Alternation. See Frequency: singula-

Anachrony. See Order

Analepsis. See Order

Anisochrony, 86-88

Anticipation. See Order: prolepsis

Apologue, 233

Ariosto, Ludovico, 231

Aristotle, 163, 173

Armance (Stendhal), 186, 196

Aspect, 29-30, 113

L'Astree (Urfe, Honore d'), 100, 231

L'Auberge rouge (Balzac), 213, 232

Auctorial discourse, 258-259

Auerbach, Erich, 61n, 70, 254

Author: lapses of, 51n; intrusions of,
94n; distinct from narrator, 213, 259

Autobiographical narrating: and time
of the narrating, 226

Autobiographical narrator: and dis-
tance, 168n; and focalization, 198-
199, 205, 206; as author, 213

Autodiegetic. See Voice: person

Autre etude de fernme (Balzac), 232, 246

Balzac: openings, 36; narrative
juncture, 63; prolepsis, 67; Proust
on anachrony in, 157n; Proust on
change of tempo in, 98, 155n;
pauses, 99; description, 100-101,
191; iterative narrative, 116-117;
Proust on unity in, 149n; relation to
mimetic norm, 167; dialogue, 182-
183, 184; focalization, 190-191, 192n;
time of the narrating, 220; narrative
level, 229, 232; functions of the nar-
rator, 256-257; his implied reader,
260; Albert Savarus, 230-231; L'Au-
berge rouge, 213, 232; Autre etude de
femme, 232, 246; Cesar Birotteau, 36,
47, 49, 64-65, 97; he Chef-d'oeuvre in-
connu, 212; La Comedie humaine,
148-149; Le Cousin Pons, 190-191; La
Duchesse de Langeais , 36, 62, 63;
L'Envers de I'histoire contemporaine ,
190; Eugenie Grandet, 117, 122, 221;
Facino Cane, 212; Gambara, 212; Les
Illusions perdues, 49, 235; Louis Lam-
bert, 245; Le Lys dans la vallee, 194n;
La Maison Nucingen, 213, 232, 255,
260, 262; La Peau de chagrin, 48, 190,
192, 197, 232; Le Pere Goriot, 100,
214, 220-221; La Recherche del' absolu,
100; Sarrasine, 229n, 232; Les Souf-
frances de I'inventeur, 49, 50, 62-63, 65,
110; line double famille, 192n; La
Vieille Fille, 100-101

Barbey d'Aurevilly, Jules Amedee,

Bardeche, Maurice, 61

Baroque novel, 214, 231




Barthes, Roland: enigma, 57; advance
notices, 73-74; seed, 76n; snares, 77;
measuring speed, 88n; realistic ef-
fect, 165; internal focalization, 193-
194; paralipsis, 196; indices, 198;
functions of the narrator, 255n
Baudry, J. L., 246n
Beckett, Samuel, 174, 219, 230
Beginnings. Set' Openings; In medias

Behaviorist narrative, 219
Behaviorist narrator, 189
Bentley, Phyllis, 97n
Benveniste, Emile, 31n, 212, 213, 220n
Bernanos, Georges; journal d'un cure

de campagne, 217, 230
Berry, Walter, 199n
Les Bijoux indiscrets (Diderot), 244n
Bildungsroman, 227, 253
Blin, Georges, 94n, 186, 189, 192, 255
Boileau-Despreaux, Nicolas, 234n
Bond, James, 193-194
Booth, Wayne, 163, 188, 195
Borges, J. L., 96, 188, 236; "The Form
of the Sword," 246-247; "Pierre
Menard, Author of Don Quixote,"
Boulanger, ]., 249n
Bowling, L. E., 174n
Bradbury, Ray, 219
Bree, Germaine, 223, 247
Broch, Hermann, 258
Brombert, Victor H., 94n
Bronte, Emily: Wuthering Heights, 217,

232, 245
Brooks, Cleanth, 186, 189
Browning, Robert: The Ring and the

Book, 190
Buscon (Quevedo), 221n
Butor, Michel: L'Emploidu temps, 217n

Caesar, Julius: Commentaries on the
Gallic and Civil Wars, 188, 244

Cain, James: Double Indemnity, Ill-

Camus, Albert: L'Etranger, 217, 220n,
230, 260

he Capitaine Fracasse (Gautier), lOln

Catullus, The Nuptial Song ofPeleus and
Thetis, 231, 233

Cervantes: Don Quixote, 96, 230, 260n,
262; Exemplary Stories, 111

Cesar Birotteau. See Balzac
Character: and "person," 246
Characterization, 182-185
La Chartreuse de Parme. See Stendhal
Chateaubriand: Memoires d'Outre-

Tombe, 159
Cheating: and focalization, 196-197
he Chef-d'oeuvre inconnu (Balzac), 212
Children: and repeating narrative, 115
Christie, Agatha, 187; The Murder of
Roger Ackroyd, 196; The Sittaford
Mystery, 196
Classical narrative: Duration —

summary, 95-97; description, 100-
101; ellipsis, 107; scene, 109-110,
111; Frequency — the iterative, 98n,
116-117; pseudo-iterative, 121-122,
123; norms of orderliness, 149;
Mood — contrast diegesis/mimesis,
163; basic form of dialogue, 172-173;
focalization, 190-191, 208, 210;
Order— 35, 36, 67; Tense— rhythm
of, 97, 109-110, 143; as a whole, 155;
Voice — domination by story, 156n;
time of the narrating, 217, 220-223;
narrative level, 231-232; person,
246; speculative discourse, 251. See
also Epic
Code, narrative: deciphering it, 77
Code of focalization, 195, 205, 209, 210
La Comedie humaine (Balzac), 148-149
Compositional aspects, 158-159
Conrad, Joseph, 187; Lord jim, 214,
232, 245; The Nigger of the "Narcis-
sus," 190n, 245n
Constant, Benjamin: Adolphe, 186, 230
Cooper, James Fenimore, 257
Correspondence: and time of the nar-
rating, 199n, 217
Cortazar, Julio: Final del juego, 234, 235
Le Cousin Pons (Balzac), 190-191
Cubism, 210n

The Curious Impertinent (Don Quixote),
230, 260n

Daniel, Georges, 90, 108n
Debray-Genette, Raymonde, 75, 191n
Defoe, Daniel: Robinson Crusoe, 67,

221, 229, 230, 239, 244, 247
Description. See Duration



Detective stories, 77, 196-197

Determination. See Frequency: itera-

Diachrony. See Frequency: iterative

Dialogue: and duration, 87; abruptive,
151n; and aspect (frequency), 151.
See also Mood

Dickens, Charles, 67, 167; Greaf Expec-
tations, 187

Diderot, Denis: Les Bijoux indiscrets,
244n; Jacques le fataliste, 216, 222,
232, 234, 235, 239

Diegesis: and definition of narrative,
27n. See also Mood, distance

Diegetic (level). See Voice: narrative

Discourse, narrative. See Narrative:

Distance. See Mood, distance

Dominique (Fromentin), 232

Dostoevski, Fyodor, 167, 257; The Pos-
sessed, 258

Double focalization, 209

Double narrative, 56

Double structure, 79-83

Double vision, 209n

Doyle, Arthur Conan, 187, 198n, 245

Drama: narration in, and temporal du-
ality, 33; style of, its influence on
narrative, 173; expository scenes in,

La Duchesse de Langeais. See Balzac

Dujardin, Edouard: definition of im-
mediate monologue, 174n, 180; Les
Lauriers sont coupes, 173, 174, 175,
179, 216, 219, 230; Le Monologue
interieur, 174n

Dumas, Alexandre, 190, 229

Duration: defined, 86-88; in Proust,
88-93, 143, 156-157, 225-226; narra-
tive movements, 93-95

— descriptive pause and description:
defined, 93-94; boundaries of, 99n;
in Proust, 99-100, 102-106; in classi-
cal novel, 100-101; intrusions of nar-
rator, lOln; iterative, 103, 117;
focalization, 204

—ellipsis: defined, 43, 51, 52, 93; itera-
tive, 53; in Proust: 106-109, and
completing analepsis, 51, iterative,
and recalls, 53-54, iterative, and
rhythm, 93, confusion with acceler-

ations, 98-99, frequentative forms
of, 155

— scene: and definition of duration,
87; defined, 94; and slow motion,
95; and rhythm of classical novel,
97, 143; in classical novel, 109-110,
111; dominant in "showing," 166;
effect of dramatic model on, 173; in
Proust: 109-112, and iterative pro-
lepsis, 72, and narrative
impatience/nostalgia, 72-73

— summary: defined, 94; in classical
novel, 95-97; and iterative narrative,
98n, 155; and rhythm of classical
novel, 109-110; and analepsis, 155;
in Proust: 95, 97-99, replaced by the
iterative, 143

L'Education sentimentale . See Flaubert

Ellipsis. See Duration

Embeddings, 46. See also Voice: narra-
tive level

L'Emploi du temps (Butor), 217n

Enigma, 57-58

Enunciating, the: defined, 213

L'Envers de I'histoire contemporaine (Bal-
zac), 190

Epic: and temporal duality, 33; open-
ing, 36; prolepsis, 67; description,
100; iterative, 116; narrative level,
231. See also Virgil; Iliad; Odyssey

Episodic novel, 85n, 238, 242 '

Epistolary monodies, 230n

Epistolary novel: and repeating narra-
tive, 115; focalization, 190, 199n;
voice, 217, 218, 230, 232, 256, 260

L'Espoir (Malraux), 258

L'Etranger (Camus), 217, 220n, 230,

Eugenie Grandet. See Balzac

Events, narrative of. See Mood, dis-

Extension. See Frequency: iterative

Extent. See Order: analepsis

Extradiegetic. See Voice: narrative

Fable, 233

Facino Cane (Balzac), 212

Faulkner, William: "A Rose for

Emily," 245n; The Sound and the

Fury, 115, 174, 175


Les Faux Monnayeurs (Gide), 227
Fenelon, Francois de Salignac de la

Mothe, 165-166
Fielding, Henry: influence on Stend-
hal, 107; focalization, 192; addresses
to the reader, 257; Tom Jones, and
summary, 96n, and ellipsis, 107,
109, and focalization, 187, 198, and
time of the narrating, 220
Filles du Feu (Nerval), 159
Film: and temporal duality, 33-34;
internal focalization, 193n; indi-
cations of metadiegesis, 237
Final del Juego (Cortazar), 234
First-person narrative: and prolepsis,
67-68. See also Mood, perspective;
Voice: person
Fitch, B. T., 217n
Fitzgerald, F. Scott: The Great Gatsby,

Flaubert: Proust on musical treatment
in, 155; Proust on style as vision in,
159; transparent narrator, 166; dia-
logue, 182; focalization, 192;
functions of the narrator, 257;
L'Education sentimentale, 98, 101,
191n, 244, 245; Madame Bovary, and
Order: internal analepsis, 49,
heterodiegetic analepsis, 50, ad-
vance notice, 74; and Duration: de-
scription, 101, rhythm, 110; and
Frequency: iterative narrative, 117;
and Mood: focalization, 189, 191,
194; and Voice: time of the narrat-
ing, 221, 222-223, person, 245n, 246
Focalization. See Mood, perspective
Folklore narrative, 36
Fontanier, Pierre, 151n, 234n
"The Form of the Sword" (Borges),

Forster, E. M., 195

Frame stories: and narrative level, 228
Free indirect style. See Mood, dis-
tance: narrative of words
Frequency: defined, 113-114
— iterative: defined, 116; and classical
novel, 98n, 116-117; syllepsis, 85n,
116, 118-119; internal, or synthesiz-
ing, and external, or generalizing,
118, 120-121; determination, 127,
128; specification, 127, 128-129; ex-
tension, 127, 129; diversification in,


129-140; internal determination,
130-132, 136-137, 138-139; internal
specification, 132-135, 136-137,
138-139; diachrony, internal and ex-
ternal, 140-143; law of the iterative,
143; and Order: temporal disper-
sion, 65, achronic structures, 84n, as
complex anachronism, 156; and Du-
ration: description, 103, 117, sum-
mary, 155. See also singulative, below
— iterative in Proust: 117-155; ellipses
and recalls, 53-54; generalizing pro-
lepses, 72-73; and pauses, 99-100;
and scene, 111; intoxication with,
123; condition for involuntary
memory, 124; and law of frequency,
124-125; birth of, 125-127; diversifi-
cation in, 129-140; frequentative
locutions, 135; iteratism, 139; inter-
nal and external diachrony, 140-143;
causality, 142; rhythm of, 143; sub-
stitutes for summary, 143; and
anachrony, 156. See also singulative,
— pseudo-iterative: defined, 121; and
classical narrative, 121-122, 123; in
Proust, 121, 122-123; as technique of
diversification, 139; and Vigneron's
hypothesis, 145; mentioned, 177n
— repeating narrative, 115-116
— singulative: defined, 114-115; in
Proust: 117, 118, contaminated by
iterative, 121, within iterative sec-
tions, 131-132, at service of itera-
tive, 139-140, alternates with itera-
tive, 143-144, transitions to and
from iterative, 144-154
Friedman, Melvin, 174n
Friedman, Norman, 168, 187-188
Fromentin, Eugene: Dominique, 232
Fugue (Laporte), 175

Gambara (Balzac), 212

Gautier, Theophile: Le Capitaine

Fracasse, lOln
Genet, Jean, 235
Germinal (Zola), 191n
Gide, Andre: Les Faux Monnayeurs,

227; Les Nourritures terrestres, 261n;

La Symphonie pastorale, 217n, 230
Gil Bias. See Lesage



Goethe, Tohann Wolfgang von: The

Sorrows of Young Werther, 230
Goncourt, Edmund and Jules de, 240
Greimas, A. J., 137, 215n
Guiraud, Pierre, 122n
Guzman (Aleman), 221n

Hachez, Willy, 90, 91n

Hamburger, Kate, 220n

Hammett, Dashiell, 190

Hardy, Thomas, 187

Hegel, Georg Wilhelm Friedrich, 227

Hemingway, Ernest, 219, 257; "Hills
Like White Elephants," 166, 187,
190, 198; "The Killers," 166, 190

Hero: information of, 198-199, 206;
distinct from narrator, 223. See also
Mood, perspective

Heterodiegetic. See Voice: person

Histoire de France (Michelet), 27-28, 229

Historical narrative, 213

History of literature, 94

Homer: as narrator, 248. See also Clas-
sical narrative; Epic; Iliad; Odyssey

Homodiegetic. See Voice: person

Houston, J. P., 84, 111, 116n, 121n,
141, 151, 154, 180

Huet, Pierre Daniel: Traite de I'origine
des rontons , 36n

Hugo, Victor, 258; Le Rhin, 85n

Humphrey, Robert, 174n

Identity: and definition of frequency,

Iliad: anachrony, 36, 37, 46, 47, 67; de-
scription, lOOn; Platonic distinction
diegesis/mimesis, 163; rewritten by
Plato, 164-166, 169-170; voice, 244,

Les Illusbns perdues (Balzac), 49, 235

Imitation. See Mood, distance

Immediate speech. See Mood, dis-
tance: narrative of words

Implied reader, 260

Indices of focalization, 198, 202-203

Information: and definition mimesis/
diegesis, 166; and focalization, 194,
197-199, 206. See also Mood

Informer: and definition mimesis/
diegesis, 166

In medias res: as epic topos, 36; and
narrative embeddings, 46; and

complete analepsis, 62; and prolep-
sis, 67
Instance, narrating. See Voice
Interior monologue: defined, 173-174.
See also Mood, distance: narrative of
Interpolated narrating. See Voice: time

of the narrating
Intradiegetic. See Voice: narrative level
Isochrony: and definition of duration,

Iterative. See Frequency

Jacques le fataliste. See Diderot
Jakobson, Roman, 107, 115, 255-256
La Jalousie. See Robbe-Grillet
James, Henry: 163, 166, 188, 199n; The
Ambassadors, 187, 189, 191n; The
Golden Bowl, 191n; What Maisie
Knew, 189, 197, 199n; The Wings of
the Dove, 191n. See also Jamesians;
Lubbock, Percy
Jamesians, 163, 168, 194-195
Jankelevitch, Vladimir, 72
Jauss, Hans Robert, 90
Journal d'un cure de campagne (Ber-

nanos), 217, 230
Joyce, James: defines immediate
speech, 173; influenced by Les
Lauriers, 174; meets Proust, 210n;
Ulysses, 174, 175, 179; A Portrait of
the Artist as a Young Man, 187

Kemp, Robert, 179n

La Bruyere, Jean de, 117
Laclos, Choderlos de: Les Liaisons

danger euses, 217n, 218, 230
Lammert, Eberhart, 54
La Fayette, Madame de, 257
Lamiel (Stendhal), 246
Laporte, Roger, 174, 219; Fugue, 175
Les Lauriers sont coupes. See Dujardin
Lazarillo, 221n
-lepse, 40n, 195
Lesage, Alain Rene: Gil Bias, 221n,

230, 239, 245, 247, 248
Les Lettres portuguaises (Marianna Al-

coforado), 256
Level, narrative. See Voice
Les Liaisons dangereuses. See Laclos



Limits, 109. See also Transgression

Linguistics, 213

Lips, Marguerite, 172, 182

-lipse, 40n, 195

Literature, history of, 94

Livy: From the Founding of the City, 233,

Louis Lambert (Balzac), 245
Lubbock, Percy, 96, 163n, 164, 168,

186, 189, 194-195. See also James,

Henry; Jamesians
Lucien Leuwen (Stendhal), 122, 246n
he Lys dans la vallee (Balzac), 194n

Madame Bovary. See Flaubert

La Maison Nucingen. See Balzac

Malraux, Andre, 183, 184n, 258; L'Es-
poir, 258

Mann, Thomas, 257; Doctor Faustus,
245, 256n; The Magic Mountain, 258

Manon Lescaut. See Prevost

Marquand, J. P.: H. M. Pulham, Es-
quire, 197-198

Martereau (Sarraute), 175

Martin-Chauffier, Louis, 226n

Mauriac, Claude: L'Agrandissement,

Mauriac, Francois, 195, 210

Maurois, Andre, 242n

Melville, Herman: The Confidence-
Man, 190n; Moby Dick, 187, 245

Memoires d'Outre-Tombe (Chateau-
briand), 159

Memoires d'un ane (Segur), 244n

Memoires d'un louriste (Stendhal), 85n

Mendilow, A. A., 168n

Meta-: defined, 228n

Metadiegetic. See Voice: narrative

Metalepsis. See Voice: narrative level

Metz, Christian, 33, 88n

Meursault (L'Etranger), 230, 260

Michelet, Jules: Histoire de France,
27-28, 229

Michel Strogoff (Verne), 197n

Mimesis. See Mood, distance

Modalizing locutions, 203

Moliere, 183

Montgomery, Robert: The Lady in the
Lake, 193n

Mood: defined, 30, 31, 32, 161-162; in
Proust, summarized, 210-211

Mood, distance, 162-185; diegesis and
mimesis, 162-170, in Proust, 167-
169; narrative of events, 164-169;
showing (vs. telling), 30, 163, 164,
166, 167

— narrative of words, 164, 169-175, in
Proust, 175-185; inner speech, 170-
171, in Proust, 176-178; outer
speech (dialogue), 182-185; nar-
ratized speech, 170, 171; transposed
speech, 170, 171-172; free indirect
style, 172, 174, 203; reported
speech, 170, 172-173, 174, 182-185;
immediate speech, 173-174, and
voice, 175, in modern novel, 179-
180, in Proust, 179-181, and internal
focalization, 193, and time of the
narrating, 219, and narrative level,

Mood, perspective, 185-210

— focalization, 189-194; alterations,
194-197, 210; polymodality, 198-
210; and time of the narrating, 199n;
risk of, 200; and repeating
analepses, 201; indices of, 202-203;
modalizing locutions, 203; and free
indirect style, 203; and unavowed
paralepsis, 203; and description,
204; code of, 205, 210; and paralep-
sis, 205, 210; and paralipsis, 205,
210; and autobiography, 205, 206;
and prolepsis, 205, 206-207; and
supernatural intervention, 205; and
information of hero and narrator,
206; concurrent codes of, 209; de-
personalization, 209; double, 209;
and planes of reality, 209; tonality in
music, 210; and interpolated narrat-
ing, 218; and the pseudo-diegetic,
241-242; and person, 251-252

— paralepsis: defined, 195; discussed,
197; unavowed, 203; decisive crite-
rion for, 208; in Proust, 207-209,

— paralipsis, 195-197; in Proust,
52-53, 199-201, 205

— point of view, 186-191; and repeat-
ing narrative, 115; Jamesians on,
and distance, 168; and information
of hero and narrator, 198-199; critics
restrict enunciating to, 213

Moral portrait, 117



Movements. See Duration: narrative

Moyse sauve. See Saint Amant
Muller, Marcel, 32n, 43, 167, 203-204,

206, 209n, 223, 226, 249n
Muller, Gunther, 33, 88n
Music, 94, 154-155, 195, 210
Myth: from ritual, 125-127

Narratee. See Reader; Receiver; Voice:

Narrating, the: balanced with story,
156n, 218-219. See Narrative; Voice

Narrative: defined, 25-29; function of,
161; pure, see Mood, distance

Narrative competence (reader's), 77

Narrative information: as mimetic fac-
tor, 166

Narrative license, 121

Narrative organization. See Order:

Narrative scattering, 55-56

Narrative temporality, 33-35, 86-88;
play on, 65, 235; autonomy of, 85;
musical treatment of, 155; as a
whole, 155; emancipation of, 156

Narratized speech. See Mood, dis-
tance: narrative of words

Narrator: descriptive intrusions of,
lOln; as mimetic factor, 166; trans-
parent in "showing," 166; informa-
tion of, 198-199, 206; typology of,
189. See also Mood, distance; Mood,
perspective; Voice: person

Nathanael (Les Nourritures terrestres),

Naturalistic novel, 166

Nerval, Gerard de, 240; Filles du Feu ,
159; Sylvie, 231, 237

"New novel," 219, 233

Les Nourritures terrestres (Gide), 261n

La Nouvelle Helo'ise (Rousseau), 230

The Nuptial Song of Peleus and Thetis
(Catullus), 231, 233

Obermann (Senancour), 218, 230
Objective literature, 219
Objective narrator, 189
Objectivization: and mimesis of

speech, 182-185
Odyssey: and definition of narrative,

25-27; reach and extent, 48; external

analepsis, 49; narrative level, 49;
partial and complete analepsis,
61-62; narrative juncture, 63; pro-
lepsis, 67; mentioned, 30, 48, 164,
214-215, 216, 231, 232, 233, 248

Omniscient narrator: defined, 189

Omnitemporality, 70, 254

Openings, 46, 191n

Oral narrative: and temporal duality,


— achrony: defined, 40, 84; in Proust,

— anachrony: defined, 35-36; in Iliad,
36, 37; in classical novel, 36; in Jean
Santeuil, analyzed, 37-41;
subjective-objective, 39; and narra-
tive level, 47-48, 48-49; and psy-
chology, 78-79; repeating (advance
notice/recall) and repeating narra-
tive (frequency), 115; dislocates
iterative series, 131; as iteration, 156;
and realistic motivation, 157; in
Proust: micronarrative analysis,
41-43, temporal structure of Re-
cherche, 43-47, quantity of, 68n, and
synthetic narrative, 78, complex,
79-83, as scene, 97, reviewed, 155-

— analepsis: defined, 40; open, 45, 66,
83; reach, defined, 48; extent, de-
fined, 48; external, internal, and
mixed, 49, 61; heterodiegetic, 50;
traditional functions of, 50;
homodiegetic, 51; completing (re-
turns), 51-54; paralipsis, 52-53; re-
peating (recalls), 54-61, and focali-
zation, 201; and double narrative,
56; and enigma, 57-58; partial and
complete, 62-67; and summary, 97,
155; reveals ellipsis, 109; and narra-
tive level, 231n; and metadiegetic
narrative, 232; and the pseudo-
diegetic, 240-241

— prolepsis: defined, 40; and organiza-
tion of narrative, 51, 73-74, 78n; and
Western narrative tradition, 67; in
Proust, 68-78; external, 68-70; and
narrating instance, 70; internal, and
narrative interference, 71; hetero-
diegetic and homodiegetic, 71;
completing and repeating, 71-
73; iterative, 72-73; and narrative






Order: prolepsis (cont.)
impatience/nostalgia, 72-73; repeat-
ing (advance notice), 73-77, contrast
to advance mention, 75-77; advance
mention and snares, 77; partial and
complete, 77-78; and focalization,
205-207; and narrating place, 216n

Painter, George, 210n

Parable, 233

Paralepsis. See Mood, perspective

Paralipsis: and completing analepses,
defined, 52. See also Mood, per-

Pastiche, 183, 184

Pause. See Duration

La Peau de chagrin. See Balzac

he Pere Goriot. See Balzac

Person. See Voice

Personality: in modern literature, 246

Perspective. See Mood, perspective

Picaresque, 221n

Picon, Gaetan, 183

"Pierre Menard, Author of Don
Quixte" (Borges), 262

Pirandello, Luigi: Six Characters in
Search of an Author, 235; Tonight We
Improvise, 235

Plato: on mimesis and diegesis, 162-
166; on creation of language
(Cratylus), 169; his rewriting of
Homer, 164-166, 169-170;
Theaetetus, 236; mentioned, 26, 30,
61n, 172-173, 175, 185

Point of view. See Mood, perspective

Poirot, Hercule, 187. See also Christie,

Polymodality. See Mood, perspective

Pouillon, Jean: point-of-view typol-
ogy, 188-189; on internal focaliza-
tion, 193; on paralipsis, 196

Predictive narrative. See Voice: time of
the narrating

Prevost, Abbe: Manon Lescaut and or-
der, 49, 62; and focalization, 201;
and voice, 214, 222, 227-230, 232,
238, 256n, 260

Prior narrating. See Voice: time of the

Prolepsis. See Order

Prophetic: novels of anticipation, 219.
See also Voice: time of the narrating

Proust, Marcel: biographies of, 28. A

la recherche du temps perdu: Marcel as
hero, narrator, and author, 28, not
Proust, 225, 249n; dating Swann, 28,
69, 88n, 90n, 91n, 93, 144-145, 224;
hero not narrator, 43n; intermediary
subject, 43, 45, 156, 157 (see Muller,
Marcel); opening, 46^17, 156; and
Proust's biography, 51n, 52, 108,
208, 209n, 238, 249, 251-252; girl-
cousin, 53, 82, 205; enigma, 57-58;
reinterpretations, 58-61; omnitem-
porality, 70, 78; creation of Alber-
tine, 73n, 90n; structure, 74, 148-
149; believability, 77; hero(es) born
from Santeuil, 79, 237-238; discourse
(commentarial, extranarrative,
speculative), 83, 94n, 95, 111, 151,
242, 250-251; characterization re-
placing action, 111; "supernourish-
ment," 111; Proustian psychology,
123-124; spatial sensitivity, 123-124;
temporal sensitivity, 123-124; cos-
mic movement, 139; causality, 142;
unrealism, 150-151; inner truth and
the novelist, 178-179; relation to
Dujardin and Joyce, 179; deper-
sonalization, 209; subjectivism, 209,
249n; novel of development, 227;
"force of example," 233. Jean San-
teuil: anachrony, analysis of pas-
sage, 37^11; complex anachronies
in, 79; hero and hero(es) of Re-
cherche, 79, 237-238; and transitions
between iterative and singulative,
149; as autobiography, 238, 249; nar-
rative level, 237-238, 239, 242; per-
son, 246, 247-250, 251; narratee,

Pseudo-diegetic. See Voice: narrative

Pseudo-iterative. See Frequency

Pseudo-time, 33-35

Raible, Wolfgang, 116n

Raimond, Michel, 179n, 186n, 190,
204n, 208

Rashomon, 115, 190

Ravel, Maurice: La Valse, 112

Reach. See Order: analepsis

Reader: narrative competence of, 77;
informed by Balzacian description,
100-101; his interpretation distinct
from information, 197-198 (see also

Reader (cont.)

Mood, perspective); distinct from
recipient of narrative, 213, 259; nar-
rative level of, 229; and function of
metadiegetic narrative, 232; trou-
bled by metalepsis, 236; identifies
with implied reader, 260; in Proust,
261-262. See also Receiver; Voice:

Realism: characterization, and
mimesis of speech, 185. See also Un-

Realistic effect: connotes mimesis,

Realistic motivation, 77, 157-158

Realistic novel: and metadiegetic nar-
rative, 232

Reality: two planes of, and codes of
focalization, 209

Recalls. See Order: analepsis

Receiver, 165-166, 213, 215n. See also
Reader; Voice: person

La Recherche de I'absolu (Balzac), 100

Reduced metadiegetic. See Voice: nar-
rative level

Referential illusion, 165

Repeating narrative, 115-116

Repetition: and definition of fre-
quency, 113-114

Reported speech. See Mood, distance:
narrative of words

Representation. See Mood: definition
of; Mood, distance

Retrospection. See Order: analepsis

Returns. See Order: analepsis

Le Rhin (Hugo), 85n

Rhythm: and definition of duration,
88; of classical narrative, 97, 109-
110, 143; and realistic motivation,
157-158. See also Duration

Ricardou, Jean, 87

Richard, Jean-Pierre, 77

Richardson, Samuel: Pamela, 218,

Ritual: passage to myth, 125-127

Ritual narrative, 116

Riviere, Jacques, 199, 223-224, 261

Robbe-Grillet, Alain, 35, 219, 235; La
Jalousie, 115, 193, 219n; Le Voyeur,
219n. See also "New novel"

Rogers, B. G., llln, 209n, 251,

Romberg, Bertil, 188, 197, 217n

Le Rouge et le noir. See Stendhal
Rousseau, Jean-Jacques: Confessions,

67-68, 69, 70n, 168, 229, 256n; La

Nouvelle Heloise, 230
Rousset, Jean, 189n, 190n, 199n, 217n,

226, 230n, 256

Saint Amant, Marc Antoine Gerard:

Moysesauve, 217, 220, 231, 233, 234n
St. Augustine: Confessions, 253
Sarrasine. See Balzac
Sarraute, Nathalie, 174; Martereau, 175
Sartre, Jean-Paul, 191n, 195, 198, 210
Saussure, Ferdinand de, 113, 114
Scene. See Duration
Scheherazade: as narrator, 248. See
also The Thousand and One Nights
Scott, Walter: Ivanhoe, 216n
Second-degree narrative. See Voice:

narrative level
Seed, 76-77
Segur, Sophie: Memoires d'un ane,

Senancour, Etienne Pivert de: Ober-

mann, 218, 230
Shklovsky, Victor, 158
Showing. See Mood, distance
Simon, Claude, 219
Simultaneous narrating. See Voice:

time of the narrating
Singular/singulative. See Frequency
Snares, 77
Soliloquy, 170-171
Sophocles: Oedipus the King, 243
Les Souffrances de I'invenieur. See Balzac
Specification. See Frequency: itera-
Speech. See Mood, distance: narrative

of words
Speed: and definition of duration,
87-88; and definition mimesis/
diegesis, 166. See also Duration: nar-
rative movements
Spitzer, Leo, 203, 252
Stang, Richard, 186n
Stanzel, F. K., 187, 188
Stendhal: and description, 101; influ-
ence of, 101; influenced by Fielding,
107; and focalization, 190, 192; use
of paralipsis, 196; and risk of focali-
zation, 200; and time of the narrat-
ing, 220n; and person, 244; and
ideological function of narrator,



Stendhal (cont.)
256n; Armance, 186, 196; La Char-
treuse de Parme — its opening, 36, and
ellipsis, 107, and focalization, 192-
193, mentioned, 46^7, 220n, 238,
246n; Lamiel, 246; Lucien Leuwen,
122, 246n; Memoires d'un touriste,
85n; Le Rouge et le noir — and ad-
vance mention, 75, and inner
speech, 171, and person, 245; men-
tioned, 220n

Sterne, Laurence: Tristram Shandy and
focalization, 198; and narrating in-
stance, 214; and time of the narrat-
ing, 222, 226; and narrative levels,
232; and metalepsis, 234, 235, 239;
and functions of the narrator, 255-
256; implied reader of, 260

Story: balanced with the narrating,
156n, 218-219. See Narrative, defini-
tion of

Stravinsky, Igor: The Rite of Spring , 210

Stream of consciousness, 230. See also
Mood, distance: narrative of words

Structure en abyme, 233

Subsequent narrating. See Voice: time
of the narrating

Summary. See Duration

Supernatural: and focalization, 205

Suspense in narrative, 67

Suzuki, M., 249n

Syllepsis: denned, 85n; within scene,
111. See also Frequency: iterative

Sylvie (Nerval), 231, 237

La Symphonie pastorale (Gide), 217n,

Tadie, Jean- Yves, 57, 68, 109n, 183n,

190n, 201n, 249n
Tasso, Bernardo, 231
Technique: as vision, 159
Telling. See Mood, distance
Tempo. See Duration: narrative

Temporal distortions: and realistic

motivation, 157-158
Temporal duality, 33-34
Temporality. See Narrative temporal-

Tense: defined, 29, 31-32; as a whole,

155; in Proust, 155-160
The Thousand and One Nights, 214, 222,

231, 233, 243, 248


Todorov, Tzvetan, 27n, 29-30, 67,
188-189, 213n, 214n, 216n, 217n

Tolstoy, Leo, 67, 258; The Death of Ivan
Ilych, 62, 67, 77

Tragedy: in classical tradition, 173

Traite de iorigine des romans (Huet), 36n

Transgression, 168, 208, 252, 259. See
also Limits

Transitions. See Frequency: singula-

Transposed speech. See Mood, dis-
tance: narrative of words

Line double famille (Balzac), 192n
The Unnamable (Beckett), 230
Unrealism, 150
Urfe, Honore d', 165-166; L'Astree,

100, 231
Uspenski, Boris, 189n

La Valse (Ravel), 112
Van Rossum-Guyon, Francoise, 186n
Vendryes: definition of voice, 31, 213
Verne, Jules: Around the World in
Eighty Days, 190, 191-192; Michel
Strogoff, 197n
La Vieille Fille (Balzac), 100-101
Vigneron, Robert, 144-148, 149, 150n
Virgil: Aeneid, 67, 231, 234, 244
Voice: denned, 31, 32, 212-213; and
definition of mimesis/diegesis, 166;
distinguished from point of view,
— narrating instance: defined, 31,
212-215; and prolepsis, 70; obliter-
ated in modern novel, 173-174; and
immediate speech, 174-175; distin-
guished from focalizing, 194; dis-
tinct from the writing, 213, 219-220,
229; not invariant, 214-215; place of,
215-216; in Proust: and anachrony,
78n, and tense, 157, time and place
of, 216
—narrative level, 227-237, in Proust,
230, 237-243; denned, 227-228; ex-
tradiegetic, 228-231; diegetic, or in-
rradiegetic, 228-231; metadiegetic,
or second-degree narrative — 228-
234, and place of narrating, 216, and
predictive narrative, 220, and time
of narrating, 222, and analepsis,
232, and influence of, 242-243; re-
duced metadiegetic, or pseudo-



Voice: narrative level (cont.)

diegetic— 236-237, in Proust, 237-
243, and metalepsis, 236-237, de-
fined, 237, 240, and analepsis, 240-
241, and person, 237, 247-248;
metalepsis, lOln, 234-237; and im-
mediate speech, 230; and analepsis,
231n; and status of narrator, 248;
and narratee, 259-260

—person, 238, 243-247, in Proust,
247-254; defined, 215; and narrative
level, 237, 247-248; shifts in, 246;
narrator — and duration, 157, and
mood, 211, and author, 212-213,
259, first-person, and final con-
vergence, 221, not equivalent to
hero, 223, homodiegetic, 245,
heterodiegetic, 245, autodiegetic,
245, status denned and discussed,
248, and focalization, 251-252,
functions of, 255-259; narratee —
215n, 259-262, and repeating anach-
ronies, 73. See also Reader; Re-

— time of the narrating, 157, 215-223,
in Proust, 223-227; and focalization,

199n; interpolated narrating, 216n,
217-218, 223; prior narrating (pre-
dictive narrative), 216, 217, 219-220;
simultaneous narrating, 156n, 215n,
216, 217, 218-219, 223; subsequent
narrating, 217, 220-223
Voyage narratives, 85n
Le Voyeur (Robbe-Grillet), 219n

Wagner, Richard, 149n, 154, 159n;
Parsifal, 199, 253; The Ring of the
Nibelung, 148-149

Warren, Robert Penn, 186, 189

Waters, Harold, 249n

Watson. See Doyle

Wells, H. G., 219

Woolf, Virginia: To the Lighthouse, 187

Words, narrative of. See Mood, dis-

Writing, the: distinct from the narrat-
ing, 213, 214, 219-220, 229

Written narrative: and temporal dual-
ity, 34

Zola, Emile: Germinal, 191n

Set Home | Add to Favorites

All Rights Reserved Powered by Free Document Search and Download

Copyright © 2011
This site does not host pdf,doc,ppt,xls,rtf,txt files all document are the property of their respective owners. complaint#nuokui.com