1My work on Mallarmé revolves around the question of what he means by the word "fiction", when he respects and praises it so much from the 1870s until his death. It is all the more interesting that this term, often confused with that of novel, implies first of all the narration or description of an imaginary world, of which the poet only suggests some elements, and that philosophy of fiction has, for its part, developed with its debates on narrative genres, using the results of narratology. There is more. It often happens that Mallarmé is satisfied, in his poems, to designate certain gestures and to superimpose metaphors on them, without even constituting almost any fictional world. This is why I was obliged to reflect on fictionality without a world, by referring to the authors of the philosophy of fiction.

2Interdisciplinary research on fiction and fictionality still debates the matter of fictional worlds and narrative genres. Here, I ask a specific question : how can we deal with non-narrative works that do not feature a fictional world ? Whether this "worldlessness" is fictional possibility or impossibility counts among questions as important as that of the boundary between fact and fiction. I will explore the treatment of that question with respect to Genette, Schaeffer, and Walton. Of course, one could treat this within the framework of lyricology. However, “today the theory of lyric is fragmented along linguistic, national and disciplinary lines. This is apparent even in different traditions of naming and delineating the field.” (Hillebrandt et al., 2017, p. 2) Besides, it is not clear whether Mallarmé's works would fit neatly into the category of lyric poetry. I would rather like to focus on the figurative and metaphorical expressions that poetry often makes use of. Even in fiction theory, the treatment of metaphor is delicate and often lies in the gray area between what is fictional and what is not fictional. Insofar as fiction theory is constructed as feint, disguise, pretense, or make-believe, metaphor is deeply involved in it. Therefore, reconsidering the relationship between novels and poetry or what is fictional and what is not fictional, from the perspective of the treatment of metaphor, should serve as a reconsideration of the fictional impossibility. The concern of this paper is to examine this relationship.

1. Schaeffer’s notion of fiction

3Schaeffer's theory of fiction is characterized by three features : feint, immersion, and modeling. But, in my opinion, this theory has certain limits. One of these limits is that this theory places immersion at the heart of the fictional experience. Although "immersion" is Schaeffer’s technical term and it includes the blocking by consciousness of the deception that was accepted by perception, the fact remains that, as far as "immersion" is concerned a certain passivity, being strongly attached to the object as well as to some place where someone is immersed. There are two problems with this theory. First, not all fiction adopts the mode of immersion. Second, his emphasis on immersion presupposes a fictional world (such as the story in a novel or the space represented in a landscape painting, or some other type of “work world”).

4Schaeffer's fictional world is the object in which one is immersed. This world or this “universe” is considered sometimes as a "story", sometimes as a "group of actions, of events, of feelings and so on" (Schaeffer, [1999] 2010, p. 172). Thus, from now on, in this paper, I use the “work world” to refer to the “fictional world presented by the work”.

5This is not the case with Kendall Walton, as will be discussed later. His book Mimesis as make-believe, contains a section entitled “prop without world,” (Walton, 1993, p. 61), in which the author argues that when we play with dolls, there is make-believe, but there is no fictional world. In the same book, further on, music is described as an artistic genre whose works have no fictional world of their own. In any case, I believe that there are two kinds of fiction, two kinds of make-believe in Walton. First, they are forms of play, constituting a “game world”. Secondly, they are fictional works, constituting not only a game world, but also a fictional world.

6The merit of the theories of fiction, which are based on feint, pretense, or make-believe, lies in that they can deal with fictions without appealing to the concept of a fictional world. However, Schaeffer’s theory does not make full use of this advantage. Later, I will clear up this point by overlapping these theories with each other (for example, that of Schaeffer with that of Walton) in this paper. While starting with feints, his theory of fiction is in fact a theory of fictional worlds. In this view, fiction is the business of fictional worlds.

7I find the same kind of vein of ideas in Françoise Lavocat’s book, Fait et Fiction, when she writes : “We argue indeed that fictions are possible worlds of a quite particular nature” (Lavocat, 2016, p. 532). Like Schaeffer, Lavocat does not seem to consider fictions without fictional worlds either. This lack is all the more serious that her book is a monumental interdisciplinary synthesis. Incidentally, it is interesting that Olivier Caïra widens the notion of fiction towards the non-mimetic space, by trying to define the fiction without mimesis (Caïra, 2011, p. 87). However, the fiction without world does not mean the fiction without mimesis. Before moving on to fiction without mimesis, it is necessary to consider in advance different modes of mimesis. I think that, in principle, fiction can be a kind of mimesis without constituting a fictional world.

2. Schaeffer’s discussion of “diction”

8Many of Mallarmé’s works are almost worldless fictions. They are apparently difficult and refuse immersion while they are full of suggestions, which are graphically or sonically imitative. The worldless character is not unique to his works. Lyric and other rhyming non-narrative poems do not necessarily depict a fictional world though there are elaborated tropes, proverbs and emotional expressions. Nevertheless, Mallarmé, heir to this tradition, clearly celebrates “fiction” when he speaks of his own aesthetics. As a researcher, I would like to give some explanations to this gesture.

9So, do Mallarmé’s works fit in Schaeffer’s theory ? Hardly—because Schaeffer’s notion of fiction is restricted to the matter of fictional worlds. As Marielle Macé remarks, (2007) Schaeffer implicitly makes a distinction between fiction and figure, leaving the latter aside. Certainly, it is prudent to attribute the distinction to John Searle’s theory, which he refers to. But I believe it is not unrelated to Genette’s discussion of fiction and diction. Before entering this point, we should clarify that Schaeffer’s conception of fiction is much closer to that of Searle than that of Genette : the latter considers fiction as an indirect speech act whereas Searle and Schaeffer rather think fiction as a global pragmatic combination of discourse, not any speech act. In Métalepse, Genette considers figurative discourse (among which metaphoric expressions) as small fictions, overlapping fictional discourse and figurative discourse, while Searle and Schaeffer make a rigid distinction between these two discourses, Searle attributing the former to the pragmatic level, the latter to the semantic level, Schaeffer rather proposing to situate the two discourses separately on a common ground of mental simulation, not a direct link of the two.

10Let us return to the discussion of fiction and diction. According to Genette’s book Fiction et diction, ([1991] 2004) fiction is thematic, or it is about structures concerning narrative contents, whereas diction is “rhematic”, or it is about structures concerning ways of speaking. Next, Genette divides fictions into constitutive and conditional ones, and he divides dictions into constitutive and conditional ones. He presents novels as representative examples of fiction, and poems as representative examples of diction, making mention of Mallarmé. Genette exploits speech acts developing Searle’s theory. His argument is all the more remarkable as a gesture separating what is fictional from what is not fictional, that Schaeffer’s theory seemingly takes up this argument in order to construct his theory of fiction.

11Although Genette does not specify it, the relationship between fiction and diction is not exclusive. It is possible to superimpose fiction on diction. This is how many traditional rhymed works are structured. And while diction does not necessarily require fiction, some dictions can be partially fictional. Moreover, when we pay attention to the fictional aspect of a literary text, we can set aside the aspect of diction as conditional. In the same way, paying attention to the aspect of diction, we can set aside the fictional aspect as conditional. Genette’s text is too reticent about this delicate relationship between the two.

12Nonetheless, Schaeffer and Genette have something in common. They represent the distinction between fictional discourse (fiction) and figurative discourse (diction) by the literary genres of novels and poetry (with frequent references to Mallarmé : see Genette, 2004 ; Schaeffer, 2011, p. 141 sq.), also setting common ground in a framework other than pragmatics. It seems that Schaeffer’s theory of diction or figurative discourse, inspired by Reuven Tsur’s, takes the form of cognitive stylistics, (2009, 2010b and 2011) where he often discusses poetry, even if he doesn’t necessarily refer to metaphors. I can relate his theory of diction to his theory of fiction to further explore the relationship between the two. To do this, it is necessary to see how fiction is found in diction and how diction is found in fiction.

13Schaeffer’s theory of diction is based on cognitive stylistics. For example, in one of his articles, (2011) he analyzes the modalities of cognitive attention that characterize aesthetic experience and distinguish it from other attentional strategies. Let me summarize and discuss this paper. The hypothesis he defends is that when an object is invested by the aesthetic relation, the properties of this object are weighted differently than when it is approached in a non-aesthetic attitude. Thus, in the case of a text, certain characteristics recede into the background—for example, that of referential force—while others—for example, stylistic characteristics, i.e., to put it briefly, the exemplified properties of the work—are brought to the fore (incidentally, this sort of bipolar oscillation of attention also appears in Schaeffer's theory of fiction because the degree of fictional immersion depends on how much attention is paid to the outside world and to the work of fiction).

14According to Schaeffer, the significance of stylistic properties results from the conjunction of their perceptibility (hence the author’s poietic strategy) and the adoption of a profile, an attentional style on the part of the reader. In the end, it is the reader’s “attentional style” does or does not make the stylistic properties of texts operative (Schaeffer, 2011, p. 139). Here, it is prudent to review the following points : his notion of attentional style does not take sides in the debate about the status of stylistic properties that are activated by the aesthetic attitude. This notion is particularly neutral regarding the question of the extent to which stylistic properties are intentional, textual, attentional, or all three. Whatever the native status of stylistic properties, they are operative only insofar as they are “perceptible” to a reader. As in the case of meaning, this does not eliminate any question of “accuracy” or “fallacy” (ibid, p. 141). If we are willing to admit that stylistic properties are properties exemplified by texts as in Goodman and Genette, then, insofar as the properties exemplified by a text are a subclass of the properties actually possessed by a text, the minimal constraint on the identification of stylistic facts is that they must be part of the traits actually possessed by the work. According to this hypothesis, the variability of stylistic identification can be explained by two facts : firstly, not all properties possessed by a text are perceived as exemplified by it on this or that reactivation ; secondly, the properties intentionally exemplified by the text do not necessarily coincide with the attentional exemplifications on the part of the reader.

15There are many classifications of cognitive styles, but, he says, they all have the same structure. This structure is bipolar : gestalt style vs. analytic style ; contextual field maximizing style vs. contextual field minimizing style ; holistic style vs. serial style ; adaptive style vs. innovative style ; convergent style vs. divergent style. Of all these pairs, the model that distinguishes between convergent and divergent styles seems to be the most useful for grasping the specificity of the cognitive strategy that presides over the aesthetic relation, since it emphasizes what is probably its main feature : delayed categorization (see Schaeffer, 2010b).

16To understand the relationship between divergent style and creativity, it is useful to look at how it differs from convergent style. The convergent style tends to minimize the attentional cost invested in extracting relevant information : it favors speed, high selectivity, use of immediate context, global coherence and hierarchization of several processes. The divergent style, in contrast, favors segmentation, low selectivity, delay of integration and categorical coherence. It gives preference to attention-guided top-down processing over input-guided processing, and overall tends towards a de-hierarchization of several processing : thus, when we approach a text from an aesthetic perspective, we often pay conscious attention to syntactic rhythm and sound, whereas when we read the same text from a practical perspective, these two textual features (which are essential for the construction of meaning) are processed automatically, without becoming the object of an attentive attitude. These divergent cognitive style characteristics—the importance of top-down processing, de-hierarchization, delayed categorization—all contribute to making cognitive processing uneconomical, sometimes leading to attentional overload. In his opinion, this hypothesis joins multiple current considerations in the anthropology of art, in particular those formulated within the framework of the theory of costly signaling.

17Incidentally, as Schaeffer confirms, while all authors agree on the relevance of purely linguistic facts (phonological, syntactic, and semantic) at the infraphrastic and phrastic levels, the same cannot be said when it comes to setting a possible limit upwards (towards macrodiscursive formal features). He writes : “For some, stylistic facts encompass speech acts, others include narrative structures, still others the metrical organization of poetic forms, and so on” (Schaeffer, 1997, p. 14). That is why, when he speaks of stylistics, he is considering different kinds of units, from poetic forms to figures like metaphors, and even narratives. Within this framework, Schaeffer states :

Thus in a suspense-based narrative, categorization delay is maximized at the situational and sequential unit levels : it can be maintained for a very long time insofar as the units it “affects” exploit not working memory but our recall capacities, which are much more powerful than working memory. On the other hand, in a poem, the delay in categorization very often plays out at the level of the propositional unit, notably through the use of non-standard syntactic constructions : it goes without saying that at this level the delay in categorization cannot be pushed beyond a limit that is quickly reached, because phrasal syntactic integration exploits working memory : too great a delay in categorization risks exceeding the capacities of working memory and thus disintegrating the text. Mallarmé in France, Hölderlin in Germany, were virtuosos in exploiting categorization delay at the phrasal level. (2011, p. 148)

18This remark is not unconnected with his theory of fiction. From the perspective of his cognitive stylistics, the difference between novels and poetry, fiction and diction, or fiction discourse and figure discourse, is explained by the difference between the units of the object of attention as well as between the human capacities used to pay attention to it. When a situation or sequence that constitutes a fictional world becomes an object of attention through recall capacities, we can recognize fiction in it. When a lesser unit, a propositional unit, is the object of attention by working memory, we can only recognize diction in it. This explains, in part, why metaphor does not belong to fiction in his theory of diction.

19Considering that his theory of fiction also describes the degree of immersion through attention, his philosophy shows room for integrating theory of fiction into cognitive stylistics from the perspective of attentional attitudes that take pragmatic frameworks into consideration. In contrast, from the side of his theory of fiction, the classification of imitation gives different positions to fiction and metaphor in the framework of mental simulation. From the standpoint of mental simulation, it can be said to indicate the room where fiction and diction can be integrated. Marielle Macé aptly points out : “Faced with this complex geography of possible relationships between figure and fiction, Jean-Marie Schaeffer proposes to consider, rather than a link, a common ground ; this common ground enables us to understand why we can ask the question of this link, but also why there is no solution to this questioning. This common ground has been opened up by the thoughts of mental simulation. Here, we observe that the mental processing of fiction and the mental processing of the figure both involve simulation processes” (2007).

20Thus, Schaeffer proposes a common ground for the different things, both from the side of his theory of fiction and from the side of his cognitive stylistics. In both cases, however, fiction and metaphor do not overlap. There seems to be no such room in his thinking for metaphor to include some degree of fiction.

3. Metaphors : an auxiliary line

21We have seen Genette’s distinction between fiction and diction and Schaeffer’s distinction between fictional discourse and figurative discourse. Such a distinction is not found only in France, but also in Kendall Walton's discussion in Anglophone analytic aesthetics.

22The American philosopher Kendall Walton deals with these relationships between metaphor and fiction. After the publication of his first major book Mimesis as Make-believe in 1990, he analyzed metaphors from the perspective of his theory of fiction, and these results are included in his latest book In Other Shoes in 2014. In what follows, let me consider one of these articles on metaphors, entitled “Metaphor and prop-oriented make-believe.” (Walton, 2015)

23According to Walton, fiction in the broad sense of the term takes the form of “make-believe.” In this make-believe, props are used to imagine. For example, when we play with a tree stump that we see by chance in the forest, considering that this stump is an animal, a bear, but that the fact of imagining there is a bear is a fiction. The stump is then the prop to constitute a fiction. In the case of a work of fiction, the work is a prop. But there is a difference between the stump and the work. The stump is not there to be used as a prop for fiction, but the work is exactly invented for the purpose of imagining the fiction.

24In the case of literature, Walton’s focus is on works that have a narrative, such as novels and stories, i.e., works that have a fictional world. In contrast, he treats poetry quite differently from narrative works. For him, poetry is “thoughtwriting.” Speeches written by speechwriters are not serious statements of the writer but are prepared for use by others. Similarly, poetry is written by a poet, a thoughtwriter, for the use of others. Thoughts here include not only intellectual ideas, but also “any feelings, emotions, sentiments, attitudes, etc.” (Walton, 2011, p. 462). We learn to express ourselves from the words and thoughts of others. The poet is the best example of such others. Therefore, the poet, like a speechwriter, does not express thoughts “seriously”, but only mentions them. In his opinion, poetry does not necessarily have a narrator like novel’s one, and there is no need to assume any kind of persona in poetry. Also, since thoughtwriting is not a use but a mention, a poem is not a work of fiction in the sense that it does not present its own fictional world and does not prescribe us to imagine it, as do novels and stories. In the absence of persona and work world, poetry is more like music. Walton says : “There need be no imagining at all, and no prescriptions to imagine. Nothing is true in the world of the poem : there is no fictional world. The poem isn’t a work of fiction, any more than a speech written by a speechwriter is” (p. 463). Then, Walton asks himself about the nature of poetry : “Is it nonfiction ? Yes, if that just means that it is not fiction” (p. 463). Nonfiction in this paper should be understood in this sense.

25Eileen John offers some comments, supplementing and correcting Walton’s ideas. First, don’t we really need personas in poetry ? She responds : to distinguish poetry from things like phone messages, prayers, songs, bumper stickers, and greeting cards, if poetry does have some kind of expression of thought, it still cannot help but assume a narrator as some kind of persona, and there is minimal fiction intervening. Novels and poems are not likely to be clearly separated into fiction and “nonfiction” (p. 463). John argues :

Emphasizing that there is a substantial contrast between poetry and novels in this regard seems right, as the importance of a prescribed fictional world varies radically between the two forms, but the contrast need not be absolute. My sense is that even the thoughtwriting alternative, with the reader serving as the genuine expresser, needs at least a hint of fiction, in a perhaps very attenuated positing of an expressive subject. As Walton shows in much of his work, there are hints of fiction in many things we do, and expression in poetry seems to be one of the fleeting, flexible sites for this, where encountering what seems to be another’s expression can easily flow into our own expression. (John, 2013, p. 459)

26Second, although it is said that we borrow thoughts from poetry, poetry is often a dense linguistic construction so that it often offers too much thought, and it prompts experiences of thought in which we do not fully claim thought as our own. Here John touches on issues such as the categorization delay by costly signaling as Schaeffer argues. Third, thoughts are difficult to appropriate without feeling that we can be responsible for those thoughts. From such a multifaceted perspective, she points out the roughness of Walton’s argument.

My inability to claim these thoughts may be because I am uncertain about how to think these things and why they should be thought. The “how” and “why” aspects of thought include such things as epistemic and rhetorical, as well as affective “moods” of thought […]. This brings into play the role of past experience and memory, evidence and standards of justification, sense of context, depth and shallowness of understanding, and interests and desires of the thinker. (p. 463)

27However, John also acknowledges that metaphor is a relatively easy thoughtwriting to incorporate. Referring to Richard Moran, she argues : “Whether through metaphor or other means, a poem can signal possibilities for thought about the world that are not fully mapped out by the conceptualizing vehicles we regularly use” (p. 464).

28With the above in mind, I will examine how metaphors are positioned in Walton’s theory of fiction. From Walton’s point of view, what is metaphor ? It is appropriate here to ask what the relationship is between fiction and metaphor. For metaphors have an aspect of make-believe, but there are also so-called dead metaphors. In principle, metaphors are understood without being aware of the comparison, without being accompanied by the immersion either.

1) Metaphor is a prop-oriented make-believe.

29In general, a prop is a means of make-believe, but this make-believe sometimes becomes a means of understanding a prop itself. A prop then becomes an object of attention, its make-believe providing a convenient or instructive way to understand the prop. In this case, one does not need to pay much attention to the content of the make-believe. This is called “prop-oriented make-believe”.

30For example, when someone asks where the town of Crotone in Italy is located, sometimes the answer is, “It’s on the arch of the Italian boot.” The purpose of this answer is to understand where the city of Crotone in Italy is located rather than to enjoy the make-believe of comparing Italy to a boot. In this case, the boot is the prop for make-believe, but the object of attention is the prop itself rather than the content of its make-believe. This is what Walton calls prop-oriented make-believe. It contains not only metaphors, but also different metaphorical expressions, textual or visual. To enjoy this make-believe, we do not need to immerse ourselves in an imaginary boot. “Our participation is minimal at best” (Walton, p. 177) and it remains an “implied game” (p. 180). Then, for him, some seemingly dead metaphors (like “saddle” of a mountain) are also included in make-believe, which are very familiar ones, but which are not so dead as to be denied being a metaphor (Walton, 2015, p. 185).

2) There are both prop-oriented and content-oriented make-believe.

31Metaphors are not only prop-oriented. When one says “this melody is melancholic,” it is a prop-oriented make-believe in the sense that one uses the metaphor to name or classify the melody. At the same time, however, this sense of melancholy cannot be recognized by the formal attributes of the sound alone, so a playful way of imagining that something is melancholy while listening to the music is needed. In this case, on the one hand, the attention is focused on the melody as a prop. But on the other hand, the expressive quality of the melancholy—that is, the content of the make-believe—is essential for the formal structure of the melody. I think some of abstract examples of catachresis function in this way, as both prop- and content-oriented make-believe.

3) There are examples of combining both orientations.

32When one says, “there was anger in the ray of the sun,” this metaphorical expression is primarily a content-oriented make-believe in that it invites us to imagine that there is anger in the daylight, but it is also prop-oriented because it teaches us about the daylight before us. However, unlike the previous case, this time the make-believe is not used to classify anything. It leads us to suggest the idea that the light of this day is full of anger. Indeed, Walton writes : “The metaphor shows us a way of regarding sunlight as making it fictional that there is anger” (Walton, p. 193). In a way, this expression strives to show what can only be presented by combining daylight and anger. After analyzing this type of metaphor, stating that it is also found in figurative art, Walton mentions a Japanese brush painting :

A Japanese brush painting of a flower may be interesting not (or not merely) because of what it makes fictional, but because of how it makes it fictional, because of the manner in which the brush strokes work to generate the fictional truths. (p. 195)

33Again, the painted flower functions as a make-believe. Furthermore, the interest of this painting lies in “what it makes fictional” i.e., the prop that makes the painting look like a flower, as well as in “how it makes it fictional.” This painting is not only a prop, but also the one designed for the unique combination of the prop and the content, that is, the combination of the brushstrokes on the one hand and the painted flower on the other. Furthermore, these paintings can also be decorative. In his first major book, in “Appreciation without Participation” (1993, p. 274-289), Walton examines decoration and metafiction and finally explores the significance in human life, presented by fictions that interrupt participation.

34Thus, what is interesting about Walton’s article is that he analyzes different modes of fiction and different uses of fiction, taking into account appreciation without participation or immersion, unlike Schaeffer, who poses a theory that almost insists that there is no fiction without immersion. Incidentally, in his first major work, Walton appears to regard fiction as an alien thing that is not even a symbolic system (Chapter 3, Section 7), whereas, in his second major work, he describes metaphors as “make-believe.” Do these metaphors, which are often found in symbol systems, not contaminate the symbol system with fictions ? I will consider this point at length in the future.

4. Applying theories to the analysis of a poem

35Now, I have confirmed that according to Schaeffer, fiction is a mechanism inviting “immersion”, whereas a poem like formal verse is a mechanism inviting “cognitive divergence”. Hence Schaeffer’s following problem : his fiction theory does not deal well with a metaphorical expression that does not lead readers to immerse themselves in a fictional world, as I have drawn an auxiliary line by referring to Walton’s argument. While Walton, for his part, argues that poetry isn’t a work of fiction but a thoughtwriting, he also shows us in what sense metaphors can be a make-believe where we could find a kind of minimal fiction, as Eileen John derives from his theory. Now I’m ready to approach Mallarmé’s poem by considering, on the one hand, the stratification and the relationships between layers in favor of cognitive divergence, and, on the other hand, various forms of make-believe that do not necessarily require immersion.

Original (1998, p. 59-60) || Translation (Blackmore and Blackmore, 2006, p. 211)
Toute l’âme résumée All the soul that we evoke (summarize)
Quand lente nous l’expirons when we shed it lingering
Dans plusieurs ronds de fumée into various rings of smoke
Abolis en autres ronds each effaced by a new ring

Atteste quelque cigare testifies to some cigar
Brûlant savamment pour peu burning with much artifice
Que la cendre se sépare as the ash falls away far
De son clair baiser de feu from its lucid fiery kiss

Ainsi le chœur des romances should the choir of lyric art
À la lèvre vole-t-il fly toward your own lips thus
Exclus-en si tu commences exclude from it if you start
Le réel parce que vil the real which is villainous

Le sens trop précis rature sense too definite cancels your
Ta vague littérature. indistinct literature.

36In this poem, I’m not sure if there is a fictional world in which I can immerse myself. Is it a work of diction ? Yes, I find the style divergent according to which the delay in categorization plays at the phrasal level, since this poem consists of only one sentence, what Schaeffer calls a “propositional unit.” But it is not to say that there is nothing fictional here. Rather, this work is populated with minimal fictions like metaphors, as Walton names it prop-oriented make-believe(s). Incidentally, what Walton calls metaphor does not only mean so-called metaphors, but also various linguistic expressions that compare one thing to another, as well as various visual expressions, from road signs to decorative patterns, as in the case of Japanese brush painting of a flower. Let me analyze this poem.

371) These verses consist of seven syllables, and by its brevity, the 14 lines of verse take on a long vertical form, compared to the smoke of a cigar. I find here a visual metaphorical relationship between the poem’s form and the smoke of a cigar. The former constitutes the prop to imagine the latter as a make-believe. In this case, the make-believe is oriented toward its prop rather than its content, since the reader’s attention is likely to be directed toward the form of expression that is nicely arranged in the form of cigar smoke rather than the cigar smoke itself.

382) The frequent use of the nasal vowel “o” in the first quatrain alludes to an exhalatory motion, while the frequent use of p/b and f/v in the second quatrain hints at a lip-smacking motion, both of which are related to smoking. I find here a sonic metaphorical relationship between the frequent use of specific sounds and a series of cigar smoking motions. Again, the former constitutes the prop to imagine the latter as a make-believe. In this case also, the make-believe is oriented toward its prop rather than its content, since the reader’s attention is likely to be directed toward the sounds that mimic cleverly a series of cigar smoking motions rather than smoking motions themselves.

393) This set of smoking themes are compared to another set of psychic themes by the word “soul”. The scene of exhaling cigar smoking resembles that of a soul leaving the body and expiring. Furthermore, these psychic themes lead in turn to the metaphor of the author’s expression, i.e., the literary work that he has breathed his soul into. The three sets of themes (smoking-dying-writing), two by two, constitute each make-believe. And since the reader’s attention is likely to be directed toward the circulation through the three sets of themes, every make-believe is both prop-oriented and content-oriented.

404) The poem forms a Shakespearean sonnet, the last two lines of which rhyme together. But, in terms of content, all the 14 lines are divided into eight (the first two quatrains), four (the third quatrain) and two (the last distich), each group repeating more or less the previous content (the idealistic aesthetics that aims to abstract and gasify the reality in a summary way, shaving off the materiality and coarseness of reality), while reducing the number of lines by half. In other words, this structure itself is the metaphor of a summary of “All the soul” that is also the literary work. Even here, I find a visual metaphorical relationship between the poem’s structure and the content its first line suggests. At first glance, this metaphor resembles the first visual one. Therefore, the structure constitutes the prop to imagine the content as make-believe. This view is partially correct as it is naturally a relationship between form and content. But, in the present case, because the poem’s structure is only discovered and meaningful considering the summary of all the soul that the first line suggests, also because the object of attention is rather the prop to imagine a fiction than the fiction itself, this make-believe is still both content-oriented and prop-oriented.

41In terms of Schaffer’s stylistics, Mallarmé’s works show a typical “cognitive divergence”, its multiple layers tracing partially the same or similar themes. In terms of Walton’s theory of prop-oriented make-believe, rather than conveying a fictional world that the work possesses itself, his works exploit multiple layers of metaphors, from typography to versification through sound symbolism and so-called metaphors. Readers can enjoy them on multiple levels. But because of this, the “delay in categorization” is inevitable. Thus, this poem is layered with several levels of “minimal fictions”, finding their place at the limits of immersive fiction and divergent style, between work world and thoughtwriting, if we clearly admit another mode of access to fiction than immersion as well as another kind of fiction than “worldful” make-believe. In this sense, it is suggestive that, while discussing “the unholy dismantling of fiction and consequently of the literary mechanism”, Mallarmé states that fiction or literary mechanism serves “a play (un jeu)” (Mallarmé, 2003, p. 67).

42I think this view can explain Mallarmé’s notion of fiction, if not totally, at least partially, because he himself utilized, in his poetry, various kinds of mimetic devices, visual ones, sonic ones or synesthetic ones, within the framework of his poetics of allusion.

43Fiction should be considered together with “nonfiction.” It is therefore as important to question the boundary between fiction and nonfiction in verbal expression as it is to question the boundary between fact and fiction.

44In considering the relationship between fiction and nonfiction, novels and poetry are often contrasted. However, these two types of relationships are not parallel. To see this, we need only consider the treatment of metaphor, which involves minimal pretense or make-believe. This treatment by Genette is typical. He counterposes diction to fiction and assigns poetry and novels to each category Mod. While neither the relationship between fiction and diction and their overlapping parts nor the relationship between novels and poetry and their overlapping parts are fully discussed, in Fiction and Diction, both metaphor and fiction are seen as a type of speech act, whereas in Metalepsis, as the subtitle “from Figure to Fiction” suggests, metaphor is discussed as a minimal fiction. The case of Schaeffer is different. Although his theory of fiction tends to attract attention, his theory of fiction is clearer in its entirety when understood together with his theory of what is not fiction, that is, his cognitive stylistics. In fact, he himself deals with novels in his fiction theory and poetry in his stylistics. And in the latter, he explicitly refers to Mallarmé, whose works are filled with metaphors. Conversely, as Marielle Macé reports, it appears that, from his fiction theory, he tries to place metaphor in a different position within mental simulation in general. Taking his cue from cognitive science, he seemingly positions metaphor as a different kind of mental simulation from fiction in the extension of his fiction theory, while his stylistics places fiction under a gradational continuum as a different level of attentional object units. In addition to the mismatch between the positioning of metaphor in his fiction theory and his stylistics, his concept of fiction seems therefore to be too limited by immersion and fictional worlds. What about Walton’s case ? He is, in a sense, the most exemplary fiction theorist to begin with make-believe without a fictional world, exquisitely analyzing metaphors as a kind of make-believe, a prop-oriented one. Nevertheless, he still contrasts fiction and nonfiction by means of novels and poetry, and poetry is completely ejected to the side of nonfiction. Against this too sharp distinction, I’m obliged to remind myself of the precision of Eileen John’s additions and corrections, which, in keeping with Walton’s intention, recognize a hint of fiction or a minimal fiction in poetry.

45Mallarmé was the first of the above three to embody the gap between the relationship fiction-nonfiction and the relationship novel-poetry. While praising fiction, criticizing the novel genre itself, and placing himself in the genre of poetry, he wrote works, utilizing hints and allusions, and in an abbreviated manner that did not necessarily reveal a fictional world. Therefore, I analyzed the poet’s work both with Schaeffer’s stylistics based on costly signaling, which allows for some continuity between fiction and metaphor, and with Walton’s fiction theory, understood in favor of Eileen John’s remarks, which accounts for minimal fictions of worldless make-believe. Of course, this does not explain all of the discrepancies between Mallarmé’s conception of fiction and his actual works. Some of his works do not explicitly depict a fictional world, but they do hint at its existence. It is also undeniable that, due to the nature of rhymed poetry, his other works are made to be read in a “half-real” way, without the need to immerse oneself in or even imagine a fictional world, while alluding to it. Jesper Juul compares video games to various artistic genres, noting that : “Of all cultural forms that project fictional worlds, the video game is a special form in which players can meaningfully engage with the game even while refusing to imagine the world that the game projects—the rules of a game are often sufficient to keep the player’s interest. Perhaps this places games on par with songs, opera, and ballet—cultural forms that can project fiction but can also be enjoyed even when one does not imagine the worlds that they project” (Juul, 2011, p. 200). Of course, whether Mallarmé is really projecting a (worldful) fiction is the question of this paper, but still, this ludologist comment is interesting.

46In previous decades, whenever Mallarmé was used as experimental material for radical theories, he seemed to be an anti-mimetic poet, whose works were questioned for how they contained something other than ordinary descriptions. For instance : “while Mallarmé was pretending to describe ‘something,’ he was in addition describing the operation of writing” (Derrida, [1972] 1981, p. 253). Conversely, in later historical studies, Mallarmé was rather a poet of mimesis, whose works were examined for how they depicted a fictional world and how they reflected the society of the time. Thus, when we think of fiction, we tend intuitively toward one or the other. What is important today, however, is a subtle conception of fiction that does not fall into either extreme.

47Thus, while Lavocat argued for a border, I rather argue against another border : the existing border between fiction and nonfiction. If a theoretical development is put in place that can adequately deal with fiction up to the point of worldlessness, rather than considering fiction in terms of the presence or absence of a fictional world, the position of fictional impossibility will shift considerably. The question then will not concern impossible possible worldfulness, but impossible possible worldlessness. Then, it will be confirmed that our fictional competence allows us to superimpose a metaphorical layer on prosodic layers or typographic layers, but also suggested that our cognitive disposition to enjoy an impossible fictional world may be based on the disposition to enjoy a worldless fiction.