Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

FICTION AND DICTION

If I were not so afraid of ridicule, I might have given this study a title that has already been
used too often: "What is literature?": a question to which, as we know, the illustrious text that bears it
as its title does not really answer, which, in short, is proof of great wisdom: stupid questions do not
deserve an answer; therefore, true wisdom would perhaps consist in not asking one. Literature is
undoubtedly several things at once, united (for example) by the rather tenuous link of what
Wittgenstein called a "family resemblance" and whose simultaneous examination is difficult or
perhaps - according to a relationship of uncertainty comparable to those found in physics -
impossible. So, I'm going to stick to just one of these aspects, the one that matters most to me in this
case: the aesthetic aspect. Indeed, there is almost universal consensus, although it is often forgotten,
that literature, among other things, is an art, and no less universal is the evidence that the specific
material of this art is "language," that is, the (since, according to Mallarmé's sober statement, there are
"several") languages, naturally.
The most common formula, which I will therefore adopt as a starting point, is this: literature is
the art of language. A work is literary only if it uses, exclusively or essentially, the linguistic medium.
But this necessary condition is obviously not sufficient: of all the materials that humanity can use for
artistic purposes, among others, language is perhaps the least specific, the least strictly reserved for
that purpose and, therefore, the one whose use is least sufficient to call the activity that uses it artistic.
It is not at all certain that the use of sounds or colours is enough to define music or painting, but there
is no doubt that the use of words and phrases is not enough to define literature, and even less so
literature as art. In the past, Hegel, who saw in literature - and even, to tell the truth, in poetry - a
constitutively indecisive and precarious practice, "in which art begins to dissolve and approaches its
point of transition towards religious representation and the prose of scientific 1 thought" (I am going to
give a free interpretation to that statement by expanding it: towards the prose of ordinary language,
not only religious or scientific, but also utilitarian and pragmatic), noted this negative peculiarity.
And, obviously thinking about that property that language has of overflowing its aesthetic use
everywhere, Roman Jakobson did not assign as the object of poetics literature as a raw or empirical
phenomenon, but rather literariness, defined as "what makes a verbal message a work of art2."
Let us accept, by convention, this definition of literature as an aesthetic aspect of literary
practice and, by methodological choice, the limitation of poetics to the study of this aspect, leaving
aside the question of whether its other aspects - for example, the psychological or the ideological - are
not covered, in fact or in law, by this discipline. I remember, however, that for Jakobson, the question
that constitutes the object of poetics ("what makes a verbal message a work of art") has to do at the

1 Esthétique, «La Poésie», Introduction.

2 Essais de linguistique générale, Paris, Minuit, 1963, p.210.
1
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

same time with two "specific differences": the one that "separates the art of language from the other
arts" and the one that separates it "from other kinds of verbal 3 practices." And again I am going to
leave aside the first of these "specific differences", which concerns what Etienne Souriau called
"comparative aesthetics" and, more precisely, the comparative ontology of the different arts. The
difference that we are going to deal with here, and which, in fact, most scholars of poetics have not
ceased to deal with since Aristotle, is, therefore, that which, by making "a verbal message a work of
art," distinguishes it not from other works of art, but from "other kinds of verbal" or linguistic
practices.
Let us first discard an initial response that presents itself to the naive conscience and that - I
must, moreover, specify - poetics has never, as far as I know, taken into account: the specificity of
literature as an art would be that of the written with respect to the oral, given that literature, in
accordance with etymology, is linked to the written state of language. The existence of innumerable
non-artistic uses of writing and, conversely, the existence of no less innumerable artistic
achievements, improvised or not, in a regime of primary or secondary orality, are enough to dismiss
such a response, whose naivety surely lies in the fact that it forgets a fundamental characteristic of
language as a system and of every verbal statement as a message, namely, its ideality, which allows it
to essentially transcend the particularities of its various materializations: phonetic, graphic or
otherwise. I say "essentially" because this transcendence does not in any way prevent him from
resorting, on the side, to some of these resources, which the passage from one register to another does
not, moreover, entirely annul: for this reason, we do not fail to appreciate with the eye and in silent
reading the sonorities of a poem, in the same way that a trained musician can appreciate those of a
symphony with the simple study of its score. Like painting for Leonardo, and even more so because
of the ideality of its products, literature is a mental thing.
We can thus repeat Jakobson's question in this expanded form, or rather, in this form
protected against any abusive limitation: "What makes a text, oral or written, a work of art?" To this
question, Jakobson's answer is well known—and I will refer to it later—but, as it is only one of the
possible and even existing answers, I would first like to deal with the question itself. It can be
understood, I think, in two quite different ways.
The first consists in taking for granted, in a certain way, definitively and universally
perceptible, the literature of certain texts and asking about the objective reasons, immanent or
inherent to the text itself and that accompany it in all circumstances. So the question of
Jakobson interprets it thus: "What are the texts that are works?" I will call those theories that are
implicitly based on such an interpretation constitutivist or essentialist theories of literariness.
The other interpretation understands the meaning of the question to be more or less this:

3 Ibid.
2
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

"Under what conditions or circumstances can a text, without internal modification, become a work?"
and, therefore, undoubtedly the reverse (but I will return to speak about the modalities of this
reciprocity): "Under what conditions or circumstances can a text, without internal modification, cease
to be a work?" I will call the theory based on this second interpretation the conditionalist theory of
literariness. We could also illustrate it by applying Nelson Goodman's 4 famous formula: replacing the
question What is art? by When is art?, therefore, replace the question “What is literature?” with the
question “When is it literature?”. Since we have admitted with Jakobson that a theory of literariness
is a poetics—this time giving this term no longer the weaker or neutral sense of discipline, but the
strong and committed sense of doctrine or at least of hypothesis—I will qualify the first version as
essentialist poetics and the second as conditionalist poetics. And I will add that the first version is
characteristic of closed poetics and the second of open poetics.
The first type is that of "classical" poetics, in a very broad sense, which sometimes greatly
exceeds the limits of official classicism. Its principle is, therefore, that certain texts are literary by
essence or by nature and for eternity and others are not. But the attitude I describe thus does not yet
define—remember—but an interpretation of the question or, if you prefer, a way of formulating the
question. Thus, it is, in turn, susceptible to variants according to its way of answering its own
question, that is, according to the criterion it proposes to distinguish literary texts from those that are
not: in other words, according to the choice of the criterion of constitutive literariness. The history of
poetics, explicit or implicit, shows that it has been divided between two possible criteria, which I will
very roughly describe as thematic, one, and formal, the other. I hasten to add, although my purpose
here is not historical, that we can describe the history of essentialist poetics as a long and laborious
effort to move from the thematic criterion to the formal criterion or at least to make a place for the
latter alongside the former.
The most vigorous illustration of essentialist poetics in its thematic version is, evidently, that
of Aristotle, who, as everyone knows, thanks to various adaptations has prevailed for more than
twenty centuries in the literary consciousness of the West. I am not the first to observe 5 that in some
respects it seems as if Aristotle had, for his part, noticed the difficulty described much later by Hegel,
that is, the lack of specificity of literary practice, and had decided to resolve it or, at least, to avert it
in the most radical way possible. This solution lies in two words, one of which is, in short, nothing
more than a gloss of the other: poiesis and mimesis.
Poiesis. This term means in Greek —remember— not only "poetry" but also, in a broader
sense, "creation" and the very title of Poetics indicates that the object of this treatise will be the way
in which language can be or become a means of creation, that is, of producing a work. It seems, then,
as if Aristotle had established a division between two functions of language: its ordinary function,

4 «Quan and at-il art?» (1977), in D. Lories, Philosophie analytique et Esthétique, Paris, Méridiens-Klincksieck, 1988.
5 See Käte Hamburger's opinion below.
3
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

which is to speak (legein) to inform, question, persuade, order, promise, etc., and its artistic function,
which is to produce works (poiein). The first corresponds to rhetoric - today we would rather say
pragmatics -; the second, to poetics. But how can language, normally an instrument of
communication and action, become a means of creation? Aristotle's answer is clear: there can only be
creation through language if it becomes a vehicle of mimesis, that is, of representation or, better said,
of simulation of imaginary actions and events, if it serves to invent stories or, at least, to transmit
stories already invented. Language is creative when it is put at the service of fiction, and I am not the
first to propose that mimesis be translated as fiction6. For Aristotle, the creativity of the poet does not
manifest itself at the level of verbal form, but at that of fiction, that is, of the invention and
arrangement of a story. "The poet," he says, "must be more of a creator of stories than of verses, since
through fiction he is a poet and what he pretends are actions." 7 In other words: what the poet does is
not diction, but fiction. This categorical position explains the expulsion - or, rather, the absence - of
all non-fictional poetry, whether lyrical, satirical, didactic or of another kind, from the sphere of
poetics: Empedocles, says Aristotle, is not a poet, he is a naturalist and, if Herodotus had written in
verse, that would in no way modify his status as a historian or attribute to him the slightest
qualification as a poet. Conversely, we can safely infer from this that, if the practice of prose fiction
had existed in his time, Aristotle would have had no objection in principle to admitting it into his
Poetics. That is what Huet would propose twenty centuries later: "Following Aristotle's maxim that
the poet is more of a poet for the fictions he invents than for the verses he composes, we can place the
authors of novels among poets8," and everyone knows the use that Fielding would make of that
authorization for the benefit of what he would call "comic epic in prose." The same can be said,
naturally, of prose theatre, which does not present any more difficulty for a fictionalist type of
poetics.
I will not go any further in describing the system of this poetics: I will only recall 9 that the
sphere of fiction, which is therefore coextensive with that of poetry as creation, is subdivided into
two modes of representation - the narrative and the dramatic - and into two levels of dignity of the
themes represented - the noble and the vulgar - from which are derived those four great genres:
tragedy (noble theme in dramatic mode), epic (noble theme in narrative mode), comedy (vulgar
theme in dramatic mode) and parody (vulgar theme in narrative mode), which the modern novel
replaced quite naturally. It is not the system of genres that interests us here, but the criterion of
literariness that prevails in it and that we can formulate in these terms, in which the Hegelian problem
and the Aristotelian response are combined: the surest way for poetry to escape the danger of
dissolution in the ordinary use of language and become a work of art is narrative or dramatic fiction.

6 Idem.
7 Poéthique, 1451b.
8 De l'origine des romans, 1670, p. 5.
9 Cf. Introduction à l'architexte, Paris, Seuil, 1979.
4
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

This is exactly what the most brilliant representative of neo-Aristotelian poetics today, Käte
Hamburger, writes:

To the extent that it is satisfying to see the ideas of the "founding fathers" confirmed by facts (even if it is
unproductive to take them dogmatically as a starting point), we can consider it a satisfactory result, a confirmation,
that Hegel's phrase is fully valid where Aristotle drew the line between mimetic art and elegiac art, where he
separated poiein from legein. Hegel's phrase does not have—or not yet—validity for the entire sphere of literature
(for that whole that the German language calls Dichtung), in the cases in which it corresponds to poiein, to mimesis.
In this case, the insurmountable boundary that separates fictional narration from the statement of reality, whatever it
may be, that is, from the enunciative system, prevents literature from falling into the "prose of scientific thought," in
other words, precisely into the system of enunciation. In this case, it is a question of "doing" in the sense of giving
form, producing and reproducing: it is the work of the poietes or the mimetes, who uses language as a material and an
instrument, in the same way that the painter uses colours and the sculptor uses stone. 10

Evidently, this thesis (if not its considerations) is the one that all those—scholars of poetics,
critics, or simple readers—for whom fiction, and more precisely narrative fiction, and therefore, today par
excellence the novel, represents literature itself, adhere to, explicitly or not, consciously or not. Fictionalist
poetics thus turns out to be largely in the majority in public opinion and among the public, possibly the
least educated.
I am not sure that this fervor is due to its theoretical merit, the only one that matters to us here.
This merit lies, for its part, in the solidity of a position that is in some way impregnable or, as Käte
Hamburger suggests, of a secure and well-watertight border: in verse or prose, in narrative or dramatic
mode, fiction has as its typical and manifest trait that of proposing to its public that disinterested pleasure
that bears, as we know better since Kant, the mark of aesthetic judgment. To enter fiction is to leave the
ordinary sphere of the exercise of language, characterized by the concern for truth or persuasion imposed
by the rules of communication and the deontology of discourse. As many philosophers have repeated since
Frege, the fictional statement is neither true nor false (but only - Aristotle would have said - "possible") or
it is both true and false: it is beyond or on this side of the true and the false and the paradoxical contract of
reciprocal irresponsibility that it establishes with its recipient is a perfect emblem of the famous aesthetic
disinterestedness. So, if there is one way, and only one way, for language to become a work of art with
complete certainty, that way is undoubtedly fiction.
The downside of this advantage of impregnability is, evidently, the unbearable limitation of this
position or, if you prefer, the price to be paid is the eviction, which I mentioned above with regard to
Aristotle, of too high a number of texts and even genres whose artistic character, even if it is not so
automatically assured, is still evident. Despite its overall fidelity to the fictionalist principle, classical
poetics could not indefinitely resist the pressure of this evidence, at least as far as non-fictional genres of

10 Logique des genres littéraires (1957), Paris, Seuil, 1986, pp. 207-208.
5
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

poetry were concerned, comfortably federated with the arch-generic term of lyric poetry. I am not going to
go into the details of that story, which, moreover, I have told elsewhere and from another point of view
and which, since the Italian and Spanish Renaissance, has led to the division of the poetic sphere into three
great "types": two - the narrative or "epic" and the dramatic - fictional and one - the lyrical - non-fictional.
This integration of the lyric takes place sometimes in a purely empirical and somewhat surreptitious way
in countless "poetic arts" that propose just as many lists, more or less retouched, of genres, some fictional
and others non-fictional (but this disparity is discreetly overlooked) and other times in a more explicit and
argued way that usually covers with the Aristotelian flag a merchandise that is not at all Aristotelian: for
example, by considering the lyric one of the three fundamental modes of enunciation (that in which the
poet constantly expresses himself in his own name without ever yielding the word to a character), while
for Aristotle, as for Plato, there are only modes of mimetic representation and, therefore, of fiction. Or, as
is perfectly evident in Father Batteux - the last great scholar of classical poetics in the strict sense - who
maintains with a great deal of sophistry that lyrical poetry is also mimetic in the ancient sense, since it can
express "feigned" feelings, and, therefore, also fictional. The day when Batteux's own German translator,
Johann Adolf Schlegel, would contest this somewhat fraudulent annexation in a footnote, observing that
the feelings expressed by the lyric poet could, as Aristotle implied, not be feigned, the monopoly of fiction
over literature would have ended, unless, of course, one returned to the exclusion of lyric poetry, but it was
already too late for such a turn back11.
The new system, illustrated by countless variations on the epic-dramatic-lyrical triad, consists,
therefore, in repudiating the fictional monopoly in favour of a sort of more or less declared duopoly, in
which literariness would henceforth be linked to two major types: on the one hand, fiction (dramatic or
narrative); on the other, lyrical poetry, increasingly designated by the term poetry alone.
The most elaborate and original version of this distribution, despite the faithfully Aristotelian
character (as we have seen) of its initial problematic, is undoubtedly Käte Hamburger's The Logic of
Literary Genres, already cited, which recognizes, in the sphere of Dichtung, only two fundamental
"genres": the fictional or mimetic and the lyrical, both characterized, but each in its own way, by a break
with the ordinary regime of language, which consists of what Hamburger calls "statements of reality,"
authentic speech acts performed on the subject of reality by a real and determined "I-origin." In fiction, we
do not encounter statements of reality, but rather fictional statements whose true "I-origin" is not the
author or the narrator, but rather the fictional characters, whose point of view and spatiotemporal situation
govern the entire enunciation of the story, down to the grammatical detail of its sentences, already a
fortiori of the dramatic text. In lyrical poetry, we undoubtedly encounter statements of reality and,
therefore, authentic acts of language, but whose origin remains undetermined, since, in essence, the
"lyrical I" cannot be identified with certainty either with the poet in person or with any other determined

11 On this subject, see Introduction à l'architexte, pp. 41-42.

6
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

subject. The putative statement of a literary text is, therefore, never a real person, but either a fictitious
character (in fiction) or an indeterminate self (in lyric poetry), which constitutes in a certain way an
attenuated form of fictitiousness12: perhaps we are not so far from Batteux's stratagems for bringing
lyricism into fiction.
But, as we have been able to observe in passing, this bipartition (and some others) does not oppose
the essentially thematic character of the fictional criterion (representation of imaginary events) to a
symmetrically formal character of the poetic criterion: like the defenders of the classical-romantic triad,
Käte Hamburger defines the lyrical by an attitude of enunciation rather than by a state of language. The
formal criterion, which I previously stated as a symmetrical correspondence to the thematic criterion of the
Aristotelian tradition, can be found in another tradition, which goes back to German Romanticism and
which was illustrated above all, beginning with Mallarmé and continuing with Russian Formalism, in the
idea of a “poetic language” distinct from prosaic or ordinary language by formal characteristics
superficially linked to the use of the verb, but more fundamentally by a change in the use of language, no
longer treated as a transparent means of communication, but as a sensitive, autonomous and non-
exchangeable material, in which a mysterious formal alchemy, by remaking “with several words a whole,
new word, foreign to language and like a sorceress,” “compensates for the defect of languages” and brings
about the “inseparable union of sound and meaning.” I have just spliced into the same sentence some
fragments of formulas by Mallarmé and Valéry, which are indeed very similar in this respect. But it is
surely to the second that we owe - although taken, in the distant past, from Malherbe - the most expressive
image of this theory of poetic language: poetry is to prose, or ordinary language, what dance is to
marching, that is, a use of the same resources, but "coordinated and excited in a different way," in a system
of "acts that [from now on] have their end in themselves." Thanks to which, and contrary to the ordinary
message, whose function is to annul itself in its understanding and in its result, the poetic text is not
annulled in anything but itself: its meaning does not erase, does not make one forget, its form, it is
inseparable from it, since no knowledge results from it that can be used for any act that forgets its cause.
"The poem [indestructible because it is irreplaceable] does not die because it has lived; it is expressly made
to be reborn from its ashes and to return indefinitely to what it has just been. Poetry is recognized in that
property of tending to be reproduced in its form: it incites us to reconstruct it identically. 13
The theoretical consequence of this tradition is, evidently, Jakobson's concept of poetic function,
defined as the insistence of the text on its verbal form, a form that is thereby made more perceptible and in
a certain way intransitive. In poetry, Jakobson wrote as early as 1919, "the communicative function, which
is both characteristic of common language and of emotional language, is reduced to a minimum 14" in
favour of a function that cannot, therefore, be described as anything other than aesthetic and through which

12 See Jean-Marie Schaeffer, “Fiction, Feint et Narration,” Critique, June 1987.

13 Valéry, Oeuvres, Paris, Gallimard, «Bibl. of the Pleiades», I, pp. 1324, 1331.
14 "La nouvelle poésie russe", in Questions de poétique, Paris, Seuil, p. 15.
7
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

language is immobilised in the self-sufficient existence of the work of art. To the question we have chosen
as our starting point, "What makes certain texts works of art?", Jakobson's answer, as, in other words, that
of Mallarmé or Valéry, is very clearly: the poetic function. The most dense formulation of this new
criterion also appears in that text from 1919 that Jakobson has not ceased to clarify and justify since then:
"Poetry is language in its aesthetic function." If we remember that, in the classical tradition, the formula
was, equally abruptly and exclusively, something like: "The aesthetic function of language is fiction," we
gauge the distance and understand why, a few years ago, Tzvetan Todorov wrote, more or less, that poetics
(but I will clarify, for my part: essentialist poetics) had two rival definitions of literariness: one through
fiction, the other through poetry.15
Each of them, in its own way, can legitimately claim to respond to Hegel's concern about the
guarantee of specificity of literary art. However, it is quite clear that neither of them can legitimately claim
to cover the entirety of this sphere. I am not going to go back to the specious nature of Batteux's arguments
in the service of a hegemony of fictionalist poetics over lyrical genres, and I must remember that
"poeticist" poetics has never seriously attempted to annex itself to the sphere of fiction as such: at most, it
appears to ignore or disdain this form of literature, by throwing it into the amorphous limbo of vulgar
prose without formal impositions (see what Valéry says about the novel), just as Aristotle threw all non-
fictional poetry into that of a more or less didactic discourse. The most prudent thing to do, therefore,
seems to be provisionally to attribute to each its share of truth, that is, a portion of the literary sphere: to
the thematic definition, the empire of prose fiction; to the formal definition, the empire of the poetic in the
strong sense, since, evidently, both apply together to that vast empire of the medium that is poetic fiction
of the epic type, classical tragedy and comedy, romantic drama or verse novel in the style of Jocelyn or
Eugene Onegin. Let us note in passing that Aristotle's entire sphere passes under condominium, but it is
not my fault that the Iliad is written in verse.
The most serious thing, however, is not this partial and perhaps opportune competition or bi-
belonging; as two precautions are better than one, it is surely not inappropriate for a text to satisfy two
criteria of literariness at the same time: for its fictional content and for its poetic form. The most serious
thing is the inability of our two essentialist poetics, even when united - albeit by force - to encompass by
themselves the entire literary sphere, since their double apprehension escapes the very considerable
domain of what I am provisionally going to call non-fictional prose literature: history, eloquence, essay,
autobiography, for example, not to mention texts whose extreme singularity prevents them from adhering
to any genre. Perhaps it is now better understood why I said above that essentialist poetics are closed
poetics: for them, they do not belong to literature but rather to texts a priori marked by the generic or,
better said, archigeneric seal of fictionality and/or poeticity. In doing so, they prove incapable of accepting
texts that, by not belonging to this canonical list, could enter and leave the literary sphere at the whim of

15 "La notion de littérature", in Les Genres du discours, Paris, Seuil, 1978.

8
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

circumstances and according to certain conditions - we could say - of heat and pressure. It seems that this
is where it is necessary to resort to another poetics, which I describe as conditionalist.

Unlike the other, this poetics has hardly been expressed in doctrinal or demonstrative texts, for the
simple reason that it is more instinctive and essayistic than theoretical, since it relies on the judgment of
taste, which, as everyone knows, is subjective and unmotivated, the criterion of all literary works. Its
principle is, approximately, this: "I consider literary any text that provokes aesthetic satisfaction in me." Its
only relationship with universality belongs, as Kant demonstrated, to the sphere of desire or pretension: I
want everyone to judge beautiful what seems beautiful to me and I find it difficult to understand why they
do not do so. But since we have made great progress (which some deplore) towards cultural relativism
over the last two centuries, it often happens, and increasingly, that this claim to universality is left in the
garb of "classical" humanism in favour of a more blatantly egocentric appreciation: "It is literature that I
decree such, I say so and that's it, or, if anything, my friends and I, my "modernity" of choice and I." To
illustrate this declared subjectivism, I refer, for example, to Roland Barthes' The Pleasure of the Text, but it
is clear that many of our literary attitudes are unconsciously inspired by this poetics. This new vulgate,
elitist in its very beginning, is surely the work of a smaller and more enlightened cultural stratum than that
which finds in fiction an automatic and comfortable criterion of literariness. But sometimes it coexists with
it, even if incoherently, and at least in a form in which the descriptive gives way to the evaluative, in
judgments in which the diagnosis of literariness is equivalent to a quality label: as when a supporter of the
functional criterion refuses, nevertheless, to grant it to a newsstand novel, considering it too "poorly
written" to "be literature," which is equivalent, in short, to considering fictionality a necessary but not
sufficient condition of literariness. My belief is exactly the opposite and I will return to this aspect later.
Essentially, it seems to me that this conditionalist poetics proceeds, in fact, if not in principle, from
a subjectivizing interpretation, extended to prose, of the criterion of Valéry-Jakobson: a text is literary (and
not just poetic) for those who are more interested in its form than in its content, for those who, for
example, appreciate its writing and at the same time reject or overlook its meaning. Moreover, I must
remember that this extension of the criterion of intransitivity to prose was recognized in advance by
Mallarmé in the name of the omnipresence of Verse far beyond what he called "official verse": "Verse is
everywhere in the language in which there is rhythm [...]. Whenever there is an effort to achieve a style,
there is versification.»II16 For us, the term style, with or without effort, is, evidently, the key to that poetic
or literary capacity of all kinds of texts, of that transcendence of the «poetic function» with respect to the
canonical limits, today very blurred -or displaced-, moreover, of the metric form.
What is at stake here, then, is the capacity of any text whose original or originally dominant
function was not aesthetic, but, for example, didactic or polemical, to survive that function or to submerge
it thanks to a judgment of individual or collective taste that brings its aesthetic qualities to the fore. Thus, a
16 Oeuvres complètes, Paris, Gallimard, «Bibl. of the Pleiades», p. 867.
9
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

page of history or memoirs can outlive its scientific value or its documentary interest; thus, a letter or a
speech can find admirers beyond its original destination and its practical motive; thus, a proverb, a maxim,
an aphorism can move or seduce readers who do not at all recognize their value as truth. In addition, an
Italian proverb gives us the formula for this type of attitude: Se non é vero, é ben trovato; freely translated:
"I do not agree, but it is well said." And it would be tempting to establish a relationship of incompatibility
between the aesthetic attitude and theoretical or pragmatic adherence, since the former is in a certain way
liberated by the weakening or disappearance of the latter, as if intelligence could not be totally convinced
and totally seduced at the same time. But we must certainly resist this temptation: as Mikel Dufrenne
rightly says, “a church can be beautiful even when secularized 17.” The fact is that over the centuries the
sphere of conditional literariness has been seen to expand incessantly as a result of an apparently constant,
or perhaps increasing, tendency towards aesthetic recovery, which acts in this as in other spheres and
credits art with a large part of what the action of time subtracts from the sphere of truth or utility: that is
why it is easier for a text to enter the literary sphere than to leave it.
But, although conditionalist poetics has by definition the power to explain conditional literarities in
the name of an aesthetic judgment, this power cannot be extended, despite what its supporters
spontaneously believe, to the sphere of constitutive literarities. If an epic, a tragedy, a sonnet or a novel are
literary works, it is not by virtue of an aesthetic evaluation, even if it is universal, but by an inherent trait,
such as fictionality or poetic form. If Britannicus is a literary work, it is not because I like it or even
because everyone likes it (which I doubt), but because it is a play, just as if Opus 106 or the View of Delft
is a musical or pictorial work, it is not because that sonata or that painting seduces one, ten or a hundred
million fans, but because they are a sonata and a painting. The worst painting, the worst sonata, the worst
sonnet are still painting, music or poetry, for the simple reason that they cannot be anything else except in
addition. And what is sometimes called a "dead genre" - say, arbitrarily, the epic or the sonnet - is simply a
form which has become, finally or temporarily, sterile and unproductive, but whose past productions retain
their label of literariness, however academic or dusty: even if no one wrote sonnets and even if no one
read sonnets, it would still be true that the sonnet is a literary genre and therefore that any sonnet, good or
bad, is a literary work. The literariness that constitutes the genres of fiction or poetry - like the "artistic"
nature that constitutes most other arts - is in a certain sense, within the limits of the cultural history of
humanity, imprescriptible and independent of any evaluation. The judgments and attitudes of
conditionalist poetics regarding them are sometimes impertinent, because they are superfluous, when they
are positive ("This tragedy is literature because I like it"), or ineffective, when they are negative ("This
tragedy is not literature because I do not like it"). Any possible claim of conditionalist poetics to govern
the entire sphere would therefore be abusive and literally illegitimate, excessive in relation to its right.
Now, we have seen that, on the other hand, only she could explain conditional literarities, those that are

17 Esthétique et Philosophie, Paris, Klincksieck, 1980, I, p. 29.

1
0
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

not due to either fictional content or poetic form. The consequence is therefore evident: we should not
replace essentialist poetics with conditionalist poetics, but rather make room for the latter alongside them,
since each of them is governed exclusively by its own spring of legitimacy, that is, its relevance. The error
of all poetics since Aristotle would undoubtedly have been that they all hypostatized in "literature par
excellence", or even in the only literature "worthy of the name", the sector of literary art to which their
criteria were applicable and for which they had been conceived. None of these poetics, understood literally
in their claim to universality, is valid, but all of them are valid in their sphere and retain, in any case, the
merit of having revealed and given value to one of the multiple criteria of literariness. As a plural
phenomenon, literariness demands a pluralist theory that takes into account the various ways in which
language escapes and survives its practical function and produces texts that are susceptible to recognition
and appreciation as aesthetic objects.
From this need results a distribution that I will outline in the following way. Human language
knows two regimes of literariness: the constitutive and the conditional. According to traditional categories,
the constitutive governs two large types or sets of literary practices: fiction (narrative or dramatic) and
poetry, not to mention their possible combination in fiction in poetic form. Since, to my knowledge, we do
not have, in any language, a convenient and positive term (that is, apart from the very crude non-fiction) to
designate this third type, and this terminological gap continues to put us in a difficult position, I propose to
call it diction, which at least has the appeal - if it has one - of symmetry. Fictional literature is that which
imposes itself essentially by the imaginary character of its objects, diction literature is that which imposes
itself essentially by its formal characteristics: not to mention, once again, amalgams or mixtures, but it
seems useful to me to maintain the distinction at the level of essence and the theoretical possibility of pure
states: for example, that of a story that excites us whatever the mode of representation (that story. As is
well known, it was for Aristotle, and still is for some, that of Oedipus, or the symmetrical of a formula that
fascinates us, independently of any perceptible meaning: that was, according to Valéry, the case of many
beautiful verses, which "affect us without communicating much to us" and which "perhaps communicate
to us that they have nothing to communicate to us."18
The reader will no doubt have noticed that I have added poetry to my new category of diction,
which is therefore no longer third but second. Indeed, as Mallarmé well knew, poetry is nothing but a
particularly marked and codified form - and, therefore, in its traditional states (I will return to this),
properly constitutive - of literature by diction. Thus, there are dictions of constitutive literariness and
dictions of conditional literariness, while fiction, for its part, is always constitutively literary 19. So, I am
going to represent this asymmetrical situation using the following diagram:

18 Op. cit., p. 1333.

19 Verbal fiction, of course. The other forms (plastic, cinematographic or otherwise) of fiction correspond to other arts, although the
reasons given by Käte Hamburger for comparing cinema with narrative fiction are not without weight.
1
1
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

This painting, deliberately lame, requires several observations. The first is of a terminological
nature: I have substituted, without warning, formal, which everyone can (or believes) understand, with the
adjective rhematic, which requires some clarification. As I have done elsewhere, 20 I very freely borrow the
term rheme from linguistics to designate, as opposed to the subject of a discourse, the discourse considered
in itself (a title like Petits Poèmes en prose is rhematic because it specifies not the subject of that
collection, like Le Spleen de Paris, but in a certain sense the collection itself: not what it says, but what it
is). Now, for reasons that will become clearer in the last chapter, it seems to me that we can define diction,
whatever its regime, through the being of a text, as distinct, although inseparable, from its saying: in
Goodmanian terms (as we will see), by its exemplifying capacities, as opposed to its denotative function. In
my sense, rhematic is broader than formal, because the "form" (whether a vowel is open or closed, whether
a phrase is short or long, whether a poem is in octosyllables or alexandrines) is nothing more than an
aspect of the being of a text or one of its elements. The French word nuit denotes (among other things) the
night and exemplifies, or can exemplify, all the "formal" properties, that is to say, undoubtedly material
and sensible, of its signifier, but also others: for example, the fact of being a feminine word, which is not a
formal property, since its homonym nuit, from the verb nuir ("to harm"), has no gender and therefore no
sexual connotations. The exemplary capacities of a word, a phrase, a text therefore exceed their purely
formal properties. And, if diction is the way in which these capacities are manifested and affect the reader,
the designation of rhematic for its criterion of literariness will be more correct, because it is more
complete, than that of formal; I do not take into account the advantage, for once formal - and, again, in
case it is - of symmetry.
The second observation concerns the division between two regimes of literariness by diction,
which are not separated by any watertight border. In fact, for a century now it has become increasingly
evident that the distinction between prose and poetry can be based on other, less categorical, criteria than
that of versification and that these criteria, which are otherwise heterogeneous and more or less cumulative
(for example: preferred themes, density of "images", graphic 21 arrangement) give way, under the name of
"prose poem", "poetic prose" or any other, to intermediate states that give this opposition a character that is
not absolute, but gradual and polar.
Third observation: saying that (verbal) fiction is always constitutively literary does not mean that a

20 See Seuils, Paris, Seuil, 1987, p. 75.

21 See CL Stevenson, "What's a poem?" (1957), Poétique, 83, September 1990.
1
2
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

fictional text is always constitutively functional. Just as a sentence whose meaning we do not understand,
which disgusts us or leaves us indifferent can seduce us by its form, so too a story that others consider true
may perhaps leave us totally incredulous, but seduce us as a kind of fiction: in that case there will
undoubtedly be a kind of conditional fictionality, a true story for some and fiction for others. This is more
or less the case with what is commonly called "myth," a type of narrative that is clearly situated on an
indecisive and shifting border of fiction.22 But this should not lead us to write the word myth in the box that
has been left empty, since this box is not intended for conditionally fictional texts, but for conditionally
literary fictions, a concept that seems to me to be a bit contradictory. To accept a religious story as a myth
is to accept it, at the same time and approximately, as a literary text, as abundantly demonstrated by the
use that our culture makes of "Greek 23 mythology." So the box will remain empty unless we grant that a
conditionally fictional text is, for that reason and in that (derived) sense, conditionally literary.
The fourth observation is a question. Although their criteria are different (one thematic, the other
rhematic), is there nothing in common between these two modes of literariness that are fiction and diction?
In other words, are the ways in which these two modes determine a judgment of literariness radically
heterogeneous in principle? If so, the very concept of literariness could very well be, in turn,
heterogeneous and encompass two aesthetic functions that are absolutely irreducible to one another. But I
don't think that's the case. The common feature seems to me to consist in that character of intransitivity
that formalist poetics reserved for poetic discourse (and possibly for stylistic purposes), intransitive
because its meaning is inseparable from its verbal form: untranslatable in other terms and, therefore,
destined to be "reproduced [incessantly] in its form."24
The fictional text is also intransitive, and in a way that is not due to the unchangeable nature of its
form, but to the fictional nature of its object, which determines a paradoxical function of pseudoreference
or denotation without the denoted. Nelson Goodman 25 characterizes this function - which speech act theory
describes in terms of feigned assertions, narratology as a dissociation between author (real utterer) and
narrator (fictional26 utterer), and others, such as Käte Hamburger, by a substitution of the author's I-origin
by the fictitious I-origin of the characters - in logical terms, as consisting of "monadic" or "single-place"
predicates: a Pickwickian description is nothing other than a Pickwickian-description, indivisible, in the
sense that it refers to nothing outside itself.27 If Napoleon designates an effective member of the human

22 See P. Veyne, Les Grecs ont-ils cru à leurs mythes?, Paris, Seuil, 1983, and T. Pavel, Univers de la fiction (1986), Paris, Seuil,
1988.
23 This sufficient condition is obviously not a necessary condition: one can accept a religious story as both true and literary: a
literature that then no longer owes anything to fictionality. It can certainly be welcomed, too, as going beyond those overly simple
categories, both as myth and as truth: see Northrop Frye and the Bible.
24 These formulas (rituals) may seem more metaphorical than rigorous. They are above all because they describe the phenomenon
through its psychological effects. To define it in more literally semiotic terms, it is surely necessary to resort, as I will do in the last
section on style, to the Goodmanian concept of exemplification. A text is rhematically "intransitive" when (or rather: to the extent that)
its exemplary properties exceed its denotative function.
25 I return to these two relatively interchangeable descriptions in the two chapters that follow.
26 Langages de l'art (1968), Paris, Jacqueline Chambon, 1990, chaps. IV, «The fictions».
27 This is obviously applicable to Dickens's description of Pickwick, which in fact serves to constitute it by pretending to "describe" it.
Later descriptions (or dépictions) made by commentators or illustrators are, for their part, transitive and verifiable as paraphrases of
1
3
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

species, Sherlock Holmes or Gilberte Swann do not designate anyone outside the text of Doyle or Proust; it
is a designation that turns on itself and does not leave its own sphere. The fictional text does not lead to
any extratextual reality; everything that it (constantly) takes from reality ("Sherlock Holmes lived at 221 B
Baker Street," "Gilbert Swann had black eyes," etc.) becomes an element of fiction, like Napoleon in War
and Peace or Rouen in Madame Bovary. Thus, it is, in its own way, intransitive, not because its statements
are perceived as intangible (they may be, but these are cases of combining fiction and diction), but because
the beings to which they apply have no existence outside of them and they refer to us in an infinite
circularity. In both cases, this intransitivity, due to thematic emptiness or rhematic opacity, makes the text
an autonomous object and its relationship with the reader an aesthetic relationship, in which the meaning is
perceived as inseparable from the form.
The fifth observation is an objection. Nothing guarantees a priori that conditional literarities, even
when we exclude fiction from them, are inevitably of rhematic criteria. A non-fictional prose text may well
provoke an aesthetic reaction that is due not to its form but to its content: for example, we can recognize
and appreciate as aesthetic objects, independently of the way in which they are narrated, a real action or
event recounted by a historian or an autobiographer (let us say, at random, the torment of the Princess of
Lamballe in Michelet or the episode of the cherries in the Confessions, but this would obviously be the
case in the story of Oedipus, if we considered it authentic), like any other element of reality. But, apart
from the fact that an aesthetic object is not the same as a work of art (I will return to this later), it seems to
me that in this kind of case, if the authenticity of the fact is conclusively demonstrated and clearly
perceived - and, moreover, even if it is illusory - the possible aesthetic judgment will not refer to the text,
but to a fact external to it, or supposed to be such, and whose aesthetic merit, to put it naively, does not
correspond to its author, just as the beauty of its model does not depend on the talent of a painter. Such an
analysis obviously presupposes a separation between history and narrative, and between authentic and
fictional, which is purely theoretical: every narrative introduces into its history a "creation of intrigue"
which is already a creation of fiction and/or diction. But that is precisely what I mean: the aesthetic value
of an event, outside of any narrative or dramatic representation, is not assignable to any text, and that of a
story, or a drama, always corresponds to fiction, to diction or (most frequently) to some cooperation of the
two, whose roles as a whole and as a cast are hardly measurable.
The sixth and final observation is more fundamental and concerns the very concept of conditional
literariness and its relation to our initial question, inherited from Jakobson (or Hegel): "What makes a text
a work?" We have seen that Jakobson's answer was: the poetic function, determined, if not exclusively by
metric forms, at least by formal features clearly determined by the famous "principle of equivalence"; the
fictionalist answer is equally clear and categorical and these two answers, once again, delimit in their
entirety the sphere of constitutive literarities. We can without hesitation consider as works, that is,

Dickens's description. On these questions, which are widely debated by modern philosophy, see Pavel, chap. I, “The Essences of
Fiction,” and the texts to which it refers.
1
4
 Gérard Genette: Fiction and Diction, Lumen, Barcelona, 1991 10

productions with an intentional aesthetic character, the texts that satisfy one or the other of these two
criteria (or both): they correspond, therefore, not only to the aesthetic category, but also (more strictly) to
the artistic category. But texts of conditional literariness do not indisputably correspond to this last
category, since their intentionally aesthetic character is not guaranteed: a page by Michelet or
Demosthenes is not distinguished from a page by another historian or orator of his category except by an
aesthetic "quality" (essentially: stylistic) that the reader must judge freely and of which we cannot say that
it was desired or even noticed by its author. For some readers, it is an indisputable aesthetic object, but the
term work of art, whose definition also implies an aesthetic intention, is not applicable to it literally, but in
a broad and somewhat metaphorical28 sense, as when it is said of a tribulum or an anvil that it is an
"authentic work of art." Conditional literarities do not, therefore, literally answer Jakobson's question,
since they do not determine intentional works, but only aesthetic (verbal) objects. But perhaps the question
was wrongly phrased in one sense. In what sense? In the sense that the intentional (and therefore artistic,
stricto sensu) character of a text matters less than its aesthetic character.
This question refers to a secular opposition between defenders, like Hegel, of a constitutive
aestheticity (that of art), for whom nothing is beautiful if it has not been desired as such and produced by
the spirit29, and those, like Kant, for whom the aesthetic object par excellence is a natural object or one that
appears to be one, when art hides art. This is not the place to debate it, since the sphere of literature is
surely too limited to deal validly with the relationships between the aesthetic and the artistic. Let us just
say that Jakobson's question (which is aimed - remember - at defining the object of poetics) can be
advantageously expanded in these terms: "What makes a text an aesthetic object?", and that "being a work
of art" is perhaps only one answer among several.

28 The expression "becoming (or ceasing to be) a work of art," used above, must therefore be understood in this broad sense. Strictly
speaking, a text can only become or cease to be an aesthetic object.
29 For example, when Monroe Beardsley writes: “Because of their specialized function, works of art are richer sources of aesthetic
value and provide it in a higher degree” (Aesthetics, 1958, 2nd ed., Indianapolis, Hackett, 1981, p.XX).
1
5