The Johns Hopkins University Press Is Collaborating With JSTOR To Digitize, Preserve and Extend Access To New Literary History
The Johns Hopkins University Press Is Collaborating With JSTOR To Digitize, Preserve and Extend Access To New Literary History
The Johns Hopkins University Press Is Collaborating With JSTOR To Digitize, Preserve and Extend Access To New Literary History
The Johns Hopkins University Press is collaborating with JSTOR to digitize, preserve and extend access to
New Literary History.
http://www.jstor.org
Boundaries of Narrative*
Gerard Genette
ITHINTHESPHEREOF LITERATURE,
narrative1 may be defined simply
NEW LITERARYHISTORY
with the following two differences: Socrates denied to the narrative the quality (or, to him, the fault) of imitation, and he took into consideration those
aspects of direct representation (dialogues) that nondramatic poems, like
those of Homer, may include. Thus from classical origins two contradictory
traditions seem to exist whereby narrative would be opposed to imitation as
its antithesis or would constitute one of its modes.
For Plato, the domain of what he calls lexis (or manner of speaking, as
opposed to logos, that which is said) can be theoretically divided into imitation properly speaking (mimesis) and simple narrative (diegesis). By simple
narrative Plato means all that the poet relates "in speaking in his own name,
without trying to make us believe that it is another who speaks."3 Thus in
Book I of the Iliad Homer tells us of Chryses: "He came to the Achaeans' great
boats to buy back his daughter, bringing a tremendous ransom and bearing
the bands of Apollo the archer on the golden staff in his hand. He entreated
all the Achaeans, but especially Atreus' sons, two fine military leaders."4 In
contrast, the next verses consist in imitation, because Homer makes Chryses
himself speak, or rather Homer speaks, pretending to have become Chryses,
and "strives to give us the illusion that it is not Homer speaking, but really
the old man, Apollo's priest." Here is the text of the discourse of Chryses:
"Descendents of Atreus, and you also, well-armed Achaeans, may the gods,
dwellers on Olympus, allow you to destroy Priam's city and then to return
without injury to your homes! But for me, may you also give me back my
daughter! And for that, accept this ransom, out of respect to the son of Zeus,
to Apollo the archer." But Plato adds that Homer could as well have continued his narrative in a purely narrative form by recounting the words of
Chryses instead of quoting them. This would have made the same passage,
in indirect style and in prose: "Having arrived, the priest implored the gods
to allow the Achaeans to take Troy and to keep them from destruction, and
he asked the Greeks to give him back his daughter in exchange for a ransom
and out of respect for the gods."5 Plato's theoretical division, opposing the
two pure and heterogeneous modes of narrative and imitation within poetic
diction, elicits and establishes a practical classification of genres, which includes the two distinct modes (narrative-represented
by the ancient
dithyramb-and mimetic-represented
by the theater) plus a mixed mode,
or more precisely an alternating mode, that of the epic, as we have just seen
in the example from the Iliad.
At first glance, Aristotle's classification is totally different since it consigns
all poetry to imitation, distinguishing only imitative modes: the direct,
which Plato has designated as imitation, and the narrative, which Aristotle,
like Plato, names diegesis. In addition, Aristotle seems not only fully to identify the dramatic genre with the imitative mode (as does Plato) but also fully
to identify the epic genre with the purely narrative mode, without taking
into consideration the mixed character of the epic mode. This reduction may
be related to the fact that Aristotle defines, more strictly than Plato, the
imitative mode by the scenic conditions of the dramatic representation. This
reduction can justify itself equally by the fact that the epic work remains
essentially narrative-no matter what the actual portions of dialogue or di-
BOUNDARIES OF NARRATIVE
rect discourse and even if this portion surpasses that of the narrativebecause its dialogues are of necessity surrounded and brought about by the
narrative parts which properly constitute the content or thread of the epic's
discourse. Moreoever, Aristotle acknowledges Homer's superiority over
other epic poets because Homer personally intervenes as little as possible in
the Iliad and usually puts already defined characters on stage, thus conforming to the poet's role, which is to imitate as much as possible.6 Aristotle
therefore seems implicitly to realize the imitative character of the Homeric
dialogues and thus the mixed character of epic diction which is narrative in
its depth but dramatic in its largest extension.
The difference between the classifications of Plato and Aristotle can thus
be reduced to a simple variation in terms. These two classifications agree on
the essential opposition between the dramatic and the narrative. The dramatic is considered by the two philosophers as more fully imitative than the
narrative. There is an agreement on this fact, in some ways underlined by a
disagreement in value. Plato condemns poets, from the playwrights to
Homer, as imitators, too mimetic for narrative poets. Plato admits into his
City only one ideal poet whose austere diction would be as unmimetic as
possible. In direct contrast with Plato, Aristotle places tragedy above the epic
and praises Homer for all that brings his writing closer to dramatic diction.
The two systems are thus identical except for a reversal of value. For Plato, as
for Aristotle, narrative is a weakened, attenuated mode of literary representation. And at this point, it would be difficult to conceive of other interpretations of narrative.
However, it is now necessary to introduce an observation with which
neither Plato nor Aristotle seems to have been concerned, and which will
restore to narrative its value and its importance. Direct imitation as it functions on stage consists of gestures and acts of speech. Insofar as it consists of
gestures, it can obviously represent actions, but here it escapes from the
linguistic plane where the specific activity of the poet exercises itself. To the
extent that it consists of acts of speech, of discourses delivered by characters
(and evidently in a narrative work the portion of direct imitation is reduced
to that), the literary work is not, properly speaking, representational. The
work binds itself to reproduce, as is, a real or fictitious discourse. One can
say that the above-cited verses of the Iliad (12-16) give us a verbal representation of the acts of Chryses, but one can't say as much of the five following
verses. They do not re-present the discourse of Chryses: if they concern a
discourse that was really uttered, the verses repeat it literally, and if they
concern a fictitious discourse, they constitute it literally. In both cases, the
work of re-presentation is nonexistent. In both cases, the five lines of Homer
are completely identical with Chryses' discourse. Evidently it is not the same
for the five narrative verses which come before and which are not in any way
lost in the acts of Chryses. As William James says, "The word 'dog' doesn't
bite." If poetic imitation is considered to be the verbal representation of a
nonverbal reality, and in certain cases of a verbal reality (as pictorial imitation
is considered the representation by a picture of a nonpictorial reality and
sometimes of a pictorial reality), then imitation must be found in the five
NEW LITERARYHISTORY
narrative verses but not in the five dramatic verses. These dramatic verses
consist simply of interpolating in the middle of a text representing events
another text drawn directly from these events. It is as if, anticipating certain
moder procedures, a Dutch painter of the seventeenth century placed in the
center of a still life a real oyster shell rather than the painted image of a shell.
This rather obvious comparison is included in order'to emphasize the profoundly heterogeneous character of a mode of expression to which we are so
accustomed that we do not perceive the most abrupt changes of register.
According to Plato, the mixed narrative (the most popular and universal
mode of telling a story) imitates alteratingly in the same tone and, as
Michaux would say, "without even noticing the difference," a nonverbal
matter which the narrative must represent as well as it can and a verbal
matter which presents itself as itself and which the narrative need simply
quote. If the narrative is rigorously faithful to historical events, the
historian-narrator must be very sensitive to the changing of orders when he
goes from the narrative work of telling the completed acts to the mechanical
transcription of the spoken words. But if the narrative is partially or totally
fictitious, the work of fiction, weighing equally on the verbal and the nonverbal contexts, doubtless has as an effect the masking of the difference
which separates these two types of imitation. One type of imitation is in
direct relation while the other type calls for the intervention of a more complex system of mediation. While we can admit with some difficulty that the
operations of conceiving of acts and conceiving of words do proceed from a
similar mental operation, "to say" these acts and "to say" these words constitute two very different verbal operations. More precisely, only the first act
constitutes a real operation, that is, an act of diction in the Platonic sense,
which requires a series of transpositions and equivalencies as well as a set of
unavoidable choices between those elements of the story to be retained and
those to be omitted, choices between the different possible points of view,
etc. All of these operations are obviously absent when the poet or historian
limits himself to transcribing a discourse. One can, and in fact must, contest
that distinction between the act of mental representation and the act of
verbal representation-between
logos and lexis. However, this leads to contesting the very notion of imitation that conceives of poetic fiction as a
simulacrum of reality, which is just as transcendent to the discourse which
transmits it as the historical event is exterior to the discourse of the historian,
or as a represented landscape is exterior to the painting which represents it. It
is a theory which makes no distinction between fiction and representation, for
the object of the fiction is reducible to a sham reality awaiting its representation. But it appears that in this perspective the very notion of imitation on
the level of lexis is a pure mirage which fades away as one approaches it.
Language can but perfectly imitate language; more exactly, a discourse can
but perfectly imitate a perfectly identical discourse. In short, a discourse can
but imitate itself. As far as lexis is concerned, direct imitation is a tautology.
We are thus led to the unexpected conclusion that, as far as representation
is concerned, the only mode that literature knows is narrative, the verbal
equivalent of nonverbal events and (as illustrated in Plato's example) of
BOUNDARIESOF NARRATIVE
NEWLITERARY
HISTORY
sequently, no verb is totally free of descriptive implication. One can thus say
that description is more indispensable than narration since it is easier to
describe without narrating. Perhaps this is so because objects can exist without movement but not movement without objects. But this basic situation
already indicates the nature of the relationship which unites the two functions in the overwhelming majority of literary texts: description could be
conceivable independent of narration, but one never actually finds it in a free
state. Narration cannot exist without description, but this dependence does
not stop it from always playing the first role. The description is naturally
ancilla narratienis [the handmaiden of the narration], the slave always necessary, always submissive, never emancipated. Narrative genres like the epic,
the short story, the novella, the novel, do exist where description can occupy
a very large place, indeed the largest, without ceasing to be, by vocation, a
simple auxiliary of narrative. However, purely descriptive genres never
exist, and we can hardly imagine a work where the narrative acts as auxiliary
to the description, outside of the didactic domain or semididactic fiction like
that of Jules Verne.
Thus the study of the relations between narration and description amounts
essentially to a consideration of the diegetic functions of the description, i.e.,
the role played by the descriptive passages or aspects in the general system
of the narrative. Without attempting here to enter into the details of such a
study, one can recall two relatively distinct functions of description in the
"classical" literary tradition (from Homer to the end of the nineteenth century). The first is of a decorative order. One realizes that traditional rhetoric
groups description and the other stylistic figures among the ornaments of the
discourse. The extended and detailed description appears as a pause and an
amusement during the narrative, in a purely aesthetic role similar to that of a
piece of sculpture in a classical building. The most famous example of this is
possibly the description of Achilles' shield in Book XVIIIof the Iliad.8 Boileau
is doubtless thinking of this decorative role when he praises richness and
pomp in this type of excerpt. The baroque period distinguishes itself by a
proliferation of the descriptive excursus (very noticeable in works such as the
Moyse Sauve of Saint-Amant) and in its decline resulting in the destruction of
the equilibrium of the narrative poem.
The second major function of the description, and the most evident now
because it has imposed itself on the tradition of the novel with Balzac, is at
once explicative and symbolic. In Balzac and his realist successors, physical
portraits and descriptions of clothing and furnishings tend to reveal and at
the same time to justify the psychology of the characters, of which they are at
once sign, cause, and effect. The description becomes a major element of the
exposition as it was not during the classical period. One need only consider
the houses of Mlle. Cormon in La Vieille Fille or of Balthazar Claes in La
Recherche de l'Absolu. All of this is too well established to need further emphasis. Let us simply note that the evolution of narrative forms, by substituting the signifying description for the ornamental description, has tended to
reinforce the narrative's domination (at least until the beginning of the twentieth century). Description has doubtless lost in autonomy what it has gained
BOUNDARIESOF NARRATIVE
NEWLITERARY
HISTORY
OFNARRATIVE
BOUNDARIES
distinguished by the exclusive use of the third person and of such forms as
the preterit and the pluperfect. No matter what the details and the variations
from one idiom to another, all of these differences dovetail sharply into an
opposition between the objectivity of narrative and the subjectivity of discourse. But we must specify that this is a question of an objectivity and a
subjectivity defined by a linguistic order of criteria. The discourse which
explicitly or not contains the presence of (or reference to) "I" is subjective,
but this "I" is defined only as the person who maintains the discourse. So,
too, the present (the time par excellence of the discursive mode) is only
defined as the instant in which the discourse is held, its usage marking "the
coincidence of the described event with the segment of the discourse describing it."13Inversely, the objectivity of narrative is defined by the absence
of all reference to the narrator. "Truly there is no longer a 'narrator.' The
events are chronologically recorded as they appear on the horizon of the
story. Here no one speaks. The events seem to tell themselves."14
This is doubtless the perfect description of pure narrative in its essence
and in its radical opposition to all personal forms of expression by the
speaker. It is narrative as one might ideally conceive of it, and as one might
find it in several good examples like those chosen by Benveniste from the
historian Glotz and from Balzac. Consider the following excerpt from Gambara:
Aftera tour of the gallery, the young man looked first at the sky and then at his watch,
made a gesture of impatience, entered a tobacconist's,and there lit up a cigar, placed
himself in frontof a mirrorand glancedat his outfit which was a bit more opulent than
the norms of taste in Francepermit. He re-adjustedhis collarand his blackvelvet vest
over which crossedand re-crossedone of those heavy gold chainsmade in Genoa.Then,
in one gesture, casting his velvet-lined coat over his left shoulder and draping it with
elegance, he resumed his walk without permitting himself to be distracted by the
bourgeois ogling to which he was subject. When the small shops began to turn on
their lights and the night appeareddark enough to him, he headed toward the Place
du PalaisRoyallike a man afraidof being recognized, for he skirtedthe place up until
the fountain in order to gain in the shelter of the coaches the entry to the Rue Froidmanteau....
At this level of purity, the diction of the narrative is in a sense the absolute
transitivity of the text, the total absence (except for several twists to which
we shall return later) not only of the narrator but also of the act of narration
itself by the rigorous expunging of all reference to the very instance of the
discourse which constitutes the narration. The text is there before our eyes
without being uttered by anyone, and none (or almost none) of the information it contains demands to be referred to its source, evaluated in terms of its
distance or its relation to the speaker and to the act of speaking. If we
compare such an utterance with a sentence like "I was waiting to write to you
that I had a fixed abode. Finally I have decided. I will spend the winter here,"15
we realize to what extent the autonomy of the narrative is contrary to the
dependency of the discourse whose essential determinants (who is "I"? who
is "you"? what place is designated by "here"?) can only be decoded by
10
NEW LITERARYHISTORY
BOUNDARIESOF NARRATIVE
11
into narrative remains discourse and forms a sort of cyst, easily recognized
and localized. One might say that the purity of narrative is more obvious
than that of discourse.
The reason for this lack of symmetry is quite simple, but it highlights for
us a definitive trait of narrative. Actually, discourse has no purity to preserve
since it is the natural mode of language, the broadest and most universal
mode, by definition open to all forms. On the contrary, narrative is a particular mode, marked and defined by a certain number of exclusions and restrictive conditions (no present tense, no first person, etc.) Discourse can "narrate" without ceasing to be discourse. Narrative can't "discourse" without
betraying itself. On the other hand, narrative can't abstain from discourse
without succumbing to dryness and sterility, which is why narrative exists
nowhere in its pure state. The slightest general observation, the least adjective a little more than descriptive, the most discreet comparison, the most
humble "perhaps," the most inoffensive of logical articulations introduces
into the fabric of narrative a type of utterance which is foreign and somehow
resistant to it. Numerous and minute textual analyses are necessary to study
the particulars of these sometimes microscopic intrusions. One of the aims of
such a study would be to enumerate and classify the means by which narrative literature (and especially the novel) has tried to organize in an acceptable
way within the interior of its own lexis the delicate relationships which the
demands of narrative and the necessities of discourse bring together.
We know that the novel has never succeeded in resolving the problem
posed by these relationships in a definitive manner. Sometimes, as was the
case in the classical period with a Cervantes, a Scarron, a Fielding, the
author-narrator, complacently assuming his own discourse, intervenes in
the narrative with an ironically insistent indiscretion, addressing his reader
in a tone of familiar conversation. On the other hand, we see in the same
period that he will sometimes transfer all the responsibilities of the discourse
to one major character who will speak, i.e., will at the same time narrate and
comment on the events in the first person. This is the case of the picaresque
novels, from Lazarillo de Tormes to Gil Bias, and of other fictitiously autobiographical novels like Manon Lescaut or La Vie de Marianne. Or at times unable
to be reduced to speaking in his own name or to conferring the task on one
character, the author-narrator will divide the discourse among several
characters, perhaps in the form of letters as the eighteenth-century novel did
(La Nouvelle Heloise, Les Liaisons Dangereuses) or possibly in making the
interior discourses of his major characters assume the narrative in succession
in the smoother and more subtle manner of a Joyce or a Faulkner. The only
time when the equilibrium between narrative and discourse seems to have
been assumed with a perfectly clear awareness, without scruple or ostentation, is evidently the nineteenth century, the classic age of objective narrative, of Balzac and Tolstoy. In contrast, one can see to what extent the modem period has accentuated a consciousness of this difficulty, up to the point
of making certain manners of expression almost physically impossible for
certain aware and exacting writers.
For example, we can clearly see how the effort to bring the narrative to its
NEWLITERARY
HISTORY
12
highest degree of purity has led certain American writers like Hammett and
Hemingway to exclude the exposure of any psychological motivation (always
difficult to employ without recourse to general considerations of a discursive
manner), qualifying statements implying a personal valuation by the narrator, logical liaisons, etc., up to the reduction of novel style to the abrupt
succession of short sentences without conjunctions that Sartre posited of
Camus' L'Etranger in 1943 and that we rediscovered ten years later in
Robbe-Grillet. What has often been interpreted as an application of behaviorist theories to literature was perhaps only the result of a particularly
acute sensitivity to certain incompatibilities of language. All of the fluctuations in the writing of modern novels are doubtless worth being analyzed
from this point of view, and particularly the contemporary tendency, possibly the reverse of that of L'Etranger and Robbe-Grillet, which is totally revealed in a Sollers or a Thibaudeau, to absorb the narrative into the actual
discourse of the writer in the act of writing, into what Michel Foucault calls
"the discourse linked to the act of writing, contemporary with its unfolding
and enclosed within it."17 It seems here as if literature has exhausted or
overflowed the resources of its representational mode and wants to fall back
on the vague murmur of its own discourse. Perhaps the novel, after poetry,
will definitively leave the age of representation. Perhaps narrative, in the
negative singularity that we have attributed to it, is, like art for Hegel,
already for us a thing of the past which we must hasten to consider as it
passes away, before it has completely deserted our horizon.
ECOLEPRATIQUEDES HAUTESETUDES,
PARIS
NOTES
1 [The term rdcit is translated throughout this essay as "narrative,"meaning the
story, that which is narrated.The termnarrationis translatedas "narration,"meaning
the act or process of narrating.Because "narrative"and "narration"are used interchangeably in English, the distinction in the Frenchshould be kept in mind. Tr.]
2 Poetics, 1448a.
3 Republic,Les Belles Lettres(Paris, 1967), III, 393a.
4 TheIliad, 1.12-16.
5 Republic,III, 393e.
6 Poetics, 1460a.
7 - One does find the conceptin Boileau, in relationto the epic: "Belively and quick in
your narrations,/Berich and decorous in your descriptions" (L'artpo6tique[Paris:
Gallimard,Bibliotheque de la Pleiade, 1966], III.257-58,p. 175).
8 At least as the classical tradition interpretedand imitated it. One must note that
description tends there to animate itself and thus to "narrativize"itself.
9 Poetics, 1459b.
10 Ibid., 1447b.
11 Since it's the diction which counts here, one may exclude from this listing, as
BOUNDARIESOF NARRATIVE
13
Aristotle does, the Socratic dialogues of Plato and all the accounts in the dramatic form
which arise from imitation in prose.
12 "Les relations de temps dans le verbe francais," Problemes de linguistique gnd?rale
(Paris, 1966), pp. 237-50.
13 "De la subjectivite dans le langage," ibid., p. 262.
14 "Les relations de temps dans le verbe franqais," ibid., p. 241.
15 Senancour, Oberman (Paris and Grenoble: Arthaud, 1967), I, 35.
16 (Paris: Le livre de Poche, 1964), I, 49.
17 "L'arriere-fable," L'Arc (Aix-en-Provence, 1965), p. 6. This is a special number
devoted to the work of Jules Verne.