Textual Effects of Metalepsis
Since Gérard Genette coined the term ‘narrative metalepsis’ (hereafter ‘metalepsis’) in Figures III (Genette  1980: 234), the topic has gradually developed from a marginal observation into a central notion in narratological theories. The phenomenon drew critical attention during the heyday of postmodernist literature and contributed significantly to the establishment of postclassical narratology. Since then, a number of theories and typologies on metalepsis have been developed.
The different categorizations and descriptions outlined over the past decades (see Pier 2014 for an overview) illustrate the lack of consensus on the definition of metalepsis. The dichotomy ontological/rhetorical is widely adopted, although each researcher tends to provide these terms with a different meaning. As to whether horizontal transitions, for example, there is no unanimity as to whether they constitute metalepses. Based on the already existent definitions and typologies, I have drawn up my own working model, paying particular attention to the effects caused by metalepses. These effects are highly text-dependent, so that systematizing metaleptic effects cannot easily be achieved. I argue that there are nevertheless tendencies to distinguish these effects depending on the ‘place’ where metalepsis appears. As I will argue in this paper, some metalepses appearing on the extradiegetic level tend to cause a different reading experience than metalepses on the diegetic plane.
In what follows, I shall briefly present my definition and classification and then go on to discuss its most important factors: 1) the story-discourse dichotomy, 2) the agent of metalepsis, 3) the structural paradox created by metalepsis and 4) horizontal metalepses. In the second part of this paper I shall clarify the main distinction of my model: on the one hand are those transgressions that allude to the creation of the text and thus remind the reader of his extratextual reality: extradiegetic metalepses which implicitly or explicitly involve the reader in the plot development. On the other hand, there are more innocent metalepses that do not draw the reader in: diegetic metalepses. The first type, extradiegetic metalepsis, is often dissolved into other subgroups. In contrast, I think this type should be regarded as a separate class because of the particular effect it produces in and on the text. This will be illustrated with examples from highly metaleptic texts as Niebla (Miguel de Unamuno 1914), At Swim-Two-Birds (Flann O’Brien 1939), “Built Up Logically” (Howard Schoenfeld 1950) and Die Rättin (Günter Grass 1986).
2. Working model
The starting point for my definition is Gérard Genette’s original definition of metalepsis as he presented it in Narrative Discourse :
any intrusion by the extradiegetic narrator or narratee into the diegetic universe (or by diegetic characters into a metadiegetic universe, etc.) or the inverse (as in Cortázar), produces an effect of strangeness that is either comical (when, as in Sterne or Diderot, it is presented in a joking tone) or fantastic. (Genette  1980: 236)
I concur with this definition of metalepsis but wish to nuance two factors: 1) I do not agree that a metalepsis can be realized only by the narrator, and 2) I stress the text-internal appearance more than Genette does. This leads me to the following definition: narrative metalepsis is a text-internal transgression of hierarchically ordered diegetic universes which reveals the internal structure of the text. As such, a metalepsis is a paradoxical narrative element because it confronts the reader with the artificial quality of the text. The paradoxical quality of metalepsis will be explored in section 2.3. In my opinion, it is vital to distinguish between diegetic and extradiegetic metalepses. Diegetic metalepses occur between extradiegetic, diegetic and/or hypodiegetic levels; extradiegetic metalepses involve extratextual reality and lend the entire text a self-referential character. A special subclass of extradiegetic metalepses are the ‘unmarked’ forms, to speak with William Nelles (1997: 153). Both the extradiegetic and diegetic variants can be specified as either story or discourse metalepses (cf. 2.1); depending on the agent of the metalepses, they can be labelled narratorial or figural metalepses or metalepses of the narratee (cf. 2.2).
My classification of metalepses is based on the distinction between discourse and story metalepses. Discourse metalepses occur only in language, whereas story metalepses ‘literally’ happen in the text.
A survey of the different typologies developed during the past decades shows that most researchers divide metalepsis into these two basic classes: in the one group we find metalepses which ‘literally’ take place in the text; the other group contains metalepses which do not literally occur but are referred to verbally. Concerning the terminology of those two classes, there is little agreement.
In what has become a widely acknowledged distinction, Marie-Laure Ryan opposed ontological metalepsis to rhetorical metalepsis. The rhetorical variant “opens a small window that allows a quick glance across levels, but the window closes after a few sentences, and the operation ends up reasserting the existence of the boundaries” (Ryan 2005: 441). The ontological form, on the other hand, contains a ‘literal’ crossing of boundaries. This distinction has to a large extent been adopted by other narratologists. I, too, adopt this distinction, but using the terms ‘story’ and ‘discourse’.
Ryan’s distinction already has a predecessor in William Nelles’ classification. Nelles divides metalepses into, among others, an ontological (or modal) and an epistemological (or verbal) form, a distinction between physically moving to another world and only displaying knowledge of that other world (cf. Nelles 1997: 154). Ryan’s suggestion to break metalepsis down into a rhetorical and an ontological variant is thus derived partly from Nelles’ typology. Nelles further breaks metalepsis down into ‘unmarked’ and ‘distinctly’ marked metalepses. Unmarked border crossings have no independent meaning but only a structural function (153). To illustrate unmarked metalepsis, Nelles refers to an example from Honoré de Balzac’s Illusions perdues: “Pendant que le vénérable ecclésiastique monte les rampes d’Angoulême, il n’est pas inutile d’expliquer…” (“While the venerable churchman climbs the ramps of Angloulême, it is not useless to explain…”; 153). He describes this type of metalepsis not as a pure movement of the narrator or of one of the characters but as a “temporary sharing of a common level” (153). Distinctly marked metalepses, on the other hand, do contain a clear movement from one diegetic level to another. How he distinguishes between distinctly marked metalepses and ontological metalepses is not made clear. Nelles was one of the first scholars to pick up the Genettian narratological term and categorize it. His classification, however, contains so many different subtypes that in the end it is hardly workable.
Monika Fludernik, like Ryan, calls the two subtypes ontological metalepsis and rhetorical or discourse metalepsis (Fludernik 2003: 383). Her classification is based on Genette’s theory, which, according to Fludernik, implicitly distinguishes five subtypes: 1) authorial metalepsis, 2) narratorial metalepsis (or ontological metalepsis type 1), 3) lectorial metalepsis (or ontological metalepsis type 2), 4) rhetorical/discourse metalepsis and 5) pseudo-diegetic or reduced metadiegetic form (which Fludernik, like Genette, does not recognize as being properly metaleptic). These types overlap to some extent, as it is unclear, for example, where the difference lies between authorial and narratorial metalepsis, because both forms refer to the narrator in his capacity as author of the story, thus revealing the fictional nature of the story. Furthermore, her examples of ontological metalepsis type 1 and of discourse metalepsis follow, in my opinion, the same principles. She illustrates the second type of metalepsis (narratorial or ontological metalepsis type 1) with an example from Joseph Andrews which runs as follows: “and indeed Fanny was the only creature whom the daughter would not have pitied in her situation; wherein, tho’ we compassionate her ourselves, we shall leave her for a little while, and pay a short visit to Lady Booby.” (Fielding 1745: 314). As a rhetorical or discourse metalepsis, she cites Balzac’s example from Illusions perdues as well. The “while” formula, according to Fludernik, implies a synchronization of narrating time and narrated time which causes a “projected simultaneity,” giving the illusion that the narrator enters the fictional world he is portraying. Only if he does so, says Fludernik, is he able to talk while the cleric is climbing the stairs (2003: 387). In my view, this metaleptic subform and the narratorial metalepsis quoted from Joseph Andrews overlap to a large extent. Both in the Balzac example and in the one from Joseph Andrews, there is a projection of the narrator – and no literal movement, as the qualification ‘ontological’ implies – into the story, resulting in a synchronization between narrating time and narrated time. Both crossings are, to speak with Nelles, examples of unmarked metalepses. How to distinguish between the ontological type 1 variant and the rhetorical variant thus remains unclear. Fludernik further makes a distinction between literal and metaphorical metalepses. In my opinion, however, this subdivision largely reduplicates her ontological/rhetorical dichotomy, leaving it unclear how the two subdivisions – ontological/rhetorical and literal/metaphorical – can coexist. Fludernik’s classification thus proves not to be as concise as necessary.
Dorrit Cohn distinguishes between metalepsis at the discourse level and metalepsis at the story level, as well. Crossings at the story level result in a violation of the boundary between “the primary story (the reader’s story) and the secondary story (the framed novel)” (Cohn 2012: 106). She describes metalepsis at the discourse level as a kind of figure: the narrator interrupts narration of the related events and gives some side remarks. This leads to a “light-hearted and playful synchronization of the narration with the narrated events” (105). This distinction also fits Ryan’s subdivision into ontological and rhetorical variants, although Cohn specifies that discourse metalepses are realized by the narrator. Her discourse metalepsis thus coincides with Nelles’ unmarked variants.
As mentioned in the introductory remarks, I also differentiate between an ontological and a rhetorical metaleptic form. But to avoid confusion, I opt to use Cohn’s terms ‘story’ and ‘discourse’: a metalepsis can never be an ontological transgression (only four-dimensional art forms have the possibility to put ontological metalepses on stage). I thus prefer the term ‘story metalepsis’. While ‘rhetorical metalepsis’ could also refer to the trope metalepsis, I speak of ‘discourse metalepsis’ instead. As mentioned above, story metalepsis ‘literally’ takes place in the text, whereas my category of discourse metalepsis contains all linguistically realized metaleptic forms. This contrasts with most other typologies in which discourse/rhetorical metalepses refer to metalepses realized by the narrator. Furthermore, I also distinguish Nelles’ subform of unmarked metalepses from the other variants. This subtype will be clarified and illustrated in the second part of this paper.
Most scholars present a typology in which the rhetorical/discourse variant only refers to a metalepsis established by the narrator or an “authorial voice.” The possibility that characters know about their fictionality and display this knowledge without ‘really’ crossing a diegetic border is overlooked in the majority of the classifications. In my opinion, the agent who realizes the metaleptic movement can give us crucial information considering its textual effect. Metalepses realized by characters often influence the text in a light-hearted way. On the other hand, metalepses produced by the narratee of the story, for example, often offer food for thought about the text-reader relationship. Though it is impossible to systemize the effects an sich, I include the agent of the crossing in my typology, for this can bear directly on the effects of metalepsis. The agent of a metalepsis, whether a discourse or story metalepsis, is either the narrator or a character or the narratee. Metalepses realized by the narrator I refer to as narratorial metalepses. A figural metalepsis is realized by one of the characters. If the narratee is the agent of the metalepsis, I describe the crossing as a metalepsis of the narratee. Thus, if a character in a novel, for example, is the subject of a story metalepsis, such a crossing is a figural story metalepsis. In contrast to most scholars, I do not include the direction of movement into my classification, as I believe that this aspect does not give us any additional relevant information concerning the textual effect of a metalepsis.
Sonja Klimek was one of the first scholars to include the agent of the metalepsis in her overview. She also discerns between descending and ascending metalepses and, like Fludernik, between “literal” metalepses, which are realized ‘literally’, and “metaphorical” metalepses, existing only on the linguistic level (Klimek 2012: 70). In addition, Klimek takes up two other complex metaleptic variants in her research: the ‘möbiusband’ story and the ‘illogical heterarchy’. In a möbiusband story, the intradiegtic level becomes the extradiegetic level of the alleged extradiegesis and vice (69). The term illogical heterarchy refers to a structure in which there is no single highest level. For certain types of narrative, this means that the hierarchy of levels can no longer be determined (cf. McHale 1987: 112–130). I do not integrate Klimek’s two complex metaleptic variants into my model. If texts are metaleptic to the extent that it is no longer possible to identify different subtypes, I plea to call such texts ‘highly metaleptic texts’. Ultimately, it is more important to identify the various tendencies in the use of metalepsis and, above all, to describe the effects produced by metaleptic structures than it is to pin down and label all the different kinds of metalepsis.
2.3. Structural paradox and horizontal transitions
Genette’s discussion in Figures III implicitly says that metalepsis covers only ascending transgressions in which authors mingle in their story. In Métalepse. De la figure à la fiction (2004) he introduces the term antimétalepse for the opposite movement (Genette 2004: 27). This term, however, has not won general acceptance; it is more common to speak of metaleptic movement as either ascending or descending. As pointed out above, I do not agree that this gives us any extra relevant information concerning the effect a metalepsis evokes in the text. The most important information about textual effects is captured by the terms ‘extradiegetic/diegetic’ (commented on in the second part of this paper) and ‘story/discourse’. Discussing the direction of metalepsis brings us now to the contentious question of whether horizontal movements should be included among metalepses.
A few researchers, among them Alexander Bareis, have argued that it is not always easy to distinguish between metalepsis and other, similar phenomena. There are, for example, intertextual references or illusion-breaking elements (cf. Bareis 2008: 210–211). To counter this argument, I stress that metalepsis is a text-internal border transition that creates a structural paradox in the text. By ‘structural paradox’, I refer to the fact that the mimetic quality of a story is disrupted by the metalepsis. It is normally assumed that a novel seeks to present a factual representation of reality. The reader agrees with this assumption and considers the related facts as ‘truthful’. The nineteenth-century English poet Samuel Taylor Coleridge described this agreement between reader and text as the willing suspension of disbelief. Metalepsis causes a conflict in the mimetic assumption, because it creates an illogical break in the construction of the text. The violation of the border between representation and the world in which this representation is created is not compatible with the mimetic quantity of the novel because it conflicts with the logical laws of extratextual reality. No matter how hard the reader tries to accept the content of the novel as a mimetic factual representation of reality, the structure of the text makes this impossible. The intrusion of a metalepsis thus influences the structure of a text to the extent that the report cannot be a factual representation of reality. Metalepsis disrupts the mimetic character of a text and can thus be characterised as para-dox, beyond the doxa of the mimesis. In this paper, the structural paradox created by a metalepsis refers to the contrast between the assumption of the mimetic nature of narrative and the structural paradox caused by metalepsis. It is important to stress that this structural paradox concerns only the form of metalepsis and not its effect. In its form, a metalepsis is paradoxical, but this does not mean that it always has an illusion-breaking effect.
Stressing the text-internal appearance of metalepsis has repercussions on the discussion of whether or not to include horizontal transgressions among metalepses or not. Frank Wagner (2002) provides an outline of all possible metaleptic crossings. He distinguishes between ascending (from a higher to a lower diegetic level) and descending metalepses (from a lower to a higher diegetic level) (Wagner 2002: 235–253). Furthermore, he includes violations of boundaries between stories at the same diegetic level, calling them ‘auto-intertextual’ violations (245 ff.). Such devices were not regarded as being metaleptic by Genette in Figure III. In Métalepse. De la figure à la fiction, however, he includes transgressions between diegetic worlds at the same level as metalepses. There are plenty of researchers who have followed Wagner’s proposal and also consider horizontal crossings as metalepses. Gerald Prince, for example, suggests the term perilepsis for such border crossings (2005: 628), and Grabe et al. (2006) have described them as horizontal metalepses (Lang 2006: 34–44). Klaus Meyer-Minnemann (2002: 146–149) and Sabine Schlickers (2005) also acknowledge horizontal crossings as metalepses. Sophie Rabau refers to a horizontal metalepsis of the enunciation (2005: 59–72) and to discourse metalepsis as ‘heterometalepsis’. Karin Kukkonen notes that the hierarchical relation between the fictional and the real world or, better, between the represented world and the world which represents, is of great importance for a proper understanding of the notion of metalepsis, implicitly saying that horizontal movements along a given hierarchical level cannot be defined as metalepses. Nevertheless, she argues in favour of the existence of horizontal metalepses and in effect ignores this hierarchical relation (Kukkonen 2011: 8).
As outlined by Bareis (2008), the problem with horizontal movements is that they overlap to a great extent with intertextual references. If, for example, Oskar Matzerath, Günter Grass’ protagonist in Die Blechtrommel (1959), reappears in Die Rättin (1986) and this movement is called a horizontal metalepsis, I would be willing to follow this way of thinking. I find it already more difficult to interpret the emergence of the character Hoftaller in Grass’ Ein weites Feld (1995) as a horizontal metalepsis, because Hoftaller is – or only refers to? – the spy Tallhover, who originally appeared in Joachim Schädlichs novel of the same name (1986). To understand Oskar Shell in Jonathan Safran Foers Extremely Loud and Incredibly Close (2005), who clearly incorporates characteristic features of his older “brother” Oskar Matzerath as getting there because of a horizontal metaleptic movement out of Günter Grass’ Die Blechtrommel, is too big a step for me because the distinction from intertextuality completely disappears. Marie-Laure Ryan suggests dubbing the introduction of already existing literary material into new texts ‘transfictionality’, and thus not a metaleptic form, defining transfictionality as “the migration of elements such as characters, plot structures or settings from one fictional text to another” (Ryan 2013: par. 23). She considers already existing literary figures who get mixed up in other diegetic worlds as counterparts of the original figures. I agree with Ryan on these points, because defining these movements as transfictional and defining Hoftaller, for example, as a counterpart of Tallhover is much more concise than heaping together all the terms. All in all, it is more satisfactory to distinguish metaleptic movements from other similar textual phenomena. Moreover, the horizontal crossing of, for example, Oskar Matzerath from Die Blechtrommel into Die Rättin does not meet the condition of a structural paradox because there is no text-internal crossing. His reappearance in no way affects the structure of the novel Die Rättin and therefore does not lay bare the fictional quality in a paradoxical way. His ‘resurrection’ should thus be considered a ‘regular’ fictional story element instead of a paradoxical narrative element.
Underlining the restriction of structural paradox in the definition of metalepsis sheds new light on the discussion about horizontal transitions: assuming that a metalepsis creates a structural paradox in the text in which it appears also means not acknowledging horizontal border crossings as being metaleptical. Instead, I concur with Ryan who distinguishes between metalepsis and transfictionality. Furthermore, this added restriction also makes it possible to discern between regular fictional story elements – which do not create a structural paradox in the text – and metaleptic crossings.
3. Metaleptic effects
3.1. Self-referential texts?
In Genette’s basic definition of metalepsis it is mentioned that metalepsis “produces an effect of strangeness that is either comical […] or fantastic” ( 1980: 234). Genette takes an example from Jorge Luis Borges to describe what he regards as “the most troubling thing about metalepsis” (236). In an essay on Don Quixote, Borges quotes several passages from texts displaying characters who become the readers or spectators of their own story. Concerning such reflections, Borges raises the possibility that if characters in a fictional story can turn into readers, then we, the readers, could be fictitious just as well (Borges  1992: 59). Borges here originally referred to a mise en abyme structure. Genette, however, projects this possible outcome onto metaleptic texts and concludes that
the most troubling thing about metalepsis indeed lies in this unacceptable and insistent hypothesis, that the extradiegetic is perhaps always diegetic, and that the narrator and his narratees – you and I – perhaps belong to some narrative. (Genette  1980: 236)
This hypothesis of course lets our imagination run wild and it explains, in part, why metalepsis in the (post)modern era is such a cherished intervention. But this does not mean that every metalepsis has a such outcome. The effect depends largely on the texts in which the metalepsis appear. It is thus very hard to establish a systematization of the textual effects of metalepses. Indeed, most typologies do not take these effects into consideration. That the range of possible outcomes goes beyond ‘strange’, ‘comical’, ‘fantastic’ or ‘confusing’ is, however, obvious.
To establish an unambiguous effect for each text and for every reader is thus, as indicated above, impossible. But as John Pier notes, there are some trends in the effect depending on the particular form of metalepsis. According to Pier, story metalepses, both narratorial and figural, are more likely to confirm the immersion of the reader in the story or even to strengthen it. Discourse metalepses, and especially the figural variants, confuse this immersion for a moment, but do not make it impossible (Pier 2005: 253). In my opinion, however, this difference in effect can be traced back to a more basic distinction: the difference between metalepses which are situated on the highest discourse level and those that are situated on a lower level. Because the place where a metalepsis appears in the text may influence the effect produced by the metalepsis, it seems advisable to me to systematize this dichotomy. The first step is to break metalepsis down into diegetic and extradiegetic forms. Diegetic metalepses occur within the text, i.e. between the extradiegetic, diegetic and/or hypodiegetic levels. As explained above, diegetic metalepses can be divided into story and discourse metalepses. For each metalepsis, it must be determined whether it is a narratorial metalepsis, a figural metalepsis or a metalepsis which involves the narratee of the story. An extradiegetic metalepsis is by definition a discourse metalepsis and is located primarily at the highest discourse level of the text. It refers (mostly) – implicitly or explicitly – to the creation of the text the reader is holding in his hand. As such, the reader is reminded of the reality outside the text and finds himself confronted with its artificial quality. Extradiegetic metalepses are realized by the narrator, although there are a few exceptions. Whereas diegetic metalepses only lay bare the structure of the told story and as such stress its fictionality, extradiegetic metalepses not only reveal the told story as fictitious but they make sure the entire text refers to itself as a text, thus questioning its status as a medium to display (a) reality. Consequently, the reader is obliged to take up a position towards the text: is he willing to accept the paradox and still suspend his disbelief? Or will he approach the text now as the artificial product it is and reflect on the medium? The reader is called on to decide on how the events will develop. I believe it can be said that diegetic metalepses tend to enhance reader immersion whereas extradiegetic metalepses mostly break the reader’s illusion.
Let me now clarify the distinction diegetic-extradiegetic with concrete samples. When, for instance, we read how Dermot Trellis, an invented narrator-character by the unnamed I-narrator of Flann O’Brien’s metaleptic novel At Swim-Two-Birds, impregnates Sheila Lamont, one of his own characters, we can easily call this metalepsis a story metalepsis. The agent of the movement differs depending on how we interpret Dermot Trellis: do we consider him as one of the characters in the story the unnamed extradiegetic I-narrator is telling us (figural), or do we approach him as the diegetic author (narratorial) of the events he relates? In Jonathan Carroll’s The Land of Laughs (1982) we encounter a figural story metalepsis when it becomes clear that the little village in which France Gallen, a famous writer, lives is actually inhabited by creatures he brought into being in his books. In Miguel de Unamuno’s Mist (Niebla), the leading character, Augusto Pérez, travels towards an author whose essays he has read. During their conversation, Pérez realises he is nothing but a creature of this author’s imagination. The meeting between them can be described as a figural story metalepsis. In John Fowles’ The French Lieutenant’s Woman (1969) the extradiegetic-heterodiegetic narrator (whose name is John Fowles) suddenly finds himself in a railroad carriage sitting next to his character Charles. The narrator thus moves into his story, a transgression which I describe as a narratorial story metalepsis. All these metalepses involve a transgression which actually takes place, a story metalepsis.
However, there are plenty of transgressions which do not contain a ‘literal’ crossing of internal borders. Thus when the leading characters in Günter Grass’ The Box. Tales from the Darkroom (Die Box. Dunkelkammergeschichten, 2008) observe “It’s possible even we, sitting here and talking are just figments of his imagination” (Grass  2010: 107), they only refer to their fictional status, a transgression I call figural discourse metalepsis. In Jostein Gaarder’s Sophie’s World (Sofies verden, 1991) Sophie and Albert become aware of the fact that they only live their life in a novel. This awareness again points to a figural discourse metalepsis which is followed by a figural story metalepsis when Sophie and Alberto flee their world and end up in the world of their author. The aforementioned examples from The Box and Sophie’s World stage characters who are aware of their fictional status. It is also possible that a heterodiegetic narrator himself refers to the fictional status of his characters, without them experiencing this. Dermot Trellis in At Swim-Two-Birds, for example, refers to his characters as fictional creatures and thus evokes a narratorial discourse metalepsis. In The Box Grass applies a similar technique. The novel starts like a fairy tale with a heterodiegetic narrator reporting: “Once upon a time, there was a father, who, having grown old in years, called together his sons and daughters – four, five, six, eight in all. For a long time they resisted, but in the end they granted his wish” (Grass  2010: 1). Though The Box pretends to be told from the perspective of Grass’ children, the reader, from the first page on, is warned about the status of this account:
Now they are seated around a table and all begin to talk at once, all products of their father’s whimsy, using words he has put in their mouths, yet obstinate, too, determined not to spare his feelings despite their love for him. (1)
We encounter the same principle when the narrator of a storyline in Grass’ doomsday novel The Rat (1986), who presents himself, at the beginning of his story, as a heterodiegetic, utters: “Slowly, because that’s how I want it, they get into the habit of calling one another by their functions” (Grass  1987: 23). In both examples the narrator lays bare the fictional account of the story he is telling by a narratorial discourse metalepsis without informing his characters about this (at this point of the story).
At first sight, the following metalepsis from Mist is built up according to the same rules: the narrator interrupts, out of the blue, a conversation between Augusto Pérez and Victor Gotí who, ironically, are talking about the Cartesian cogito ergo sum. The narrator addresses himself to his reader with the following words:
While Augusto and Victor were carrying on this “nivolistic” conversation, I, the author of this nivola, which you, my dear reader, are holding in your hand and reading –, I was smiling enigmatically at the sight of my “nivolistic” characters advocating my case and justifying my methods of procedure and I said to myself, “Think how far these poor fellows are from suspecting that they are only trying to justify what I am doing with them! In the same fashion, whenever a man is seeking for reasons wherewith to justify himself, he is, strictly speaking, only seeking to justify God. And I am the God of these two Poor ‘nivolistic’ devils. (de Unamuno  2000: 252)
In this intrusion, the extradiegetic heterodiegetic narrator reveals himself as the (fictionalized!) author of the nivola the reader is holding in his hands. I describe this address to the reader as an extradiegetic narratorial discourse metalepsis. Also,The Life and Opinions of Tristram Shandy, Gentleman (Sterne 1759–1767) is burgeoning with similar reader-apostrophes. In fact, Tristram Shandy even goes one step further: the narrator not only addresses his narratee, he also tries to involve him in the plot by, for instance, asking to bring his father to bed. Terry Pratchett in Mort. A Novel of Discworld (1987) uses another way to mingle the telling and the told: “‘You shouldn’t - - - - them, then’, muttered one of his henchmen, effortlessly pronouncing a row of dashes” (Pratchett 2013: 142). The level of presentation and the way in which this happens are literally projected into the told story.
In these three examples (Mist, Tristram Shandy and Mort. A Novel of Discworld) a similar procedure lies at the basis of the metalepses: the narrator makes his readers knowledgeable about the fictional quality of the account he relates. At first sight, the last three examples tie in with the other diegetic narratorial discourse metalepses from At Swim-Two-Birds, The Box and The Rat. Yet, if we analyse these examples closely, a distinction comes to the surface. The first group of examples all have a paradoxical quality, as explained in the beginning of this paper: they show that what is being told is constructed. Because the metalepses occur inside the diegesis or involve this level, only the internal structure of the story is uncovered so that the status of the work in its entirety is not influenced by these paradoxical transgressions. The extradiegetic narratorial discourse examples in the second group (Mist, Tristram Shandy and Mort. A Novel of Discworld) all appeal to some extent to the creation of the text and as such remind the reader of the extratextual reality. They lay bare the construction of the entire text, both story and discourse, and not only of the story told in the text. As a result, these texts become self-referential, referring in the text to the text, whereas the metalepses in the first group only cause the construction of the told story to be shown, lending the texts a paradoxical character but not making them self-referential. The distinctive quality of an extradiegetic metalepsis thus lies in its self-referentiality. As argued above, extradiegetic metalepses tend to evoke a different reading experience. In particular, the novels Mist and Tristram Shandy drive the reader to rethink the text-author relation and the text-reader relation. Mist even goes so far as to question the very ontological status of the reader himself. It is up to the reader how he will approach the text. His attitude towards the text obviously influences how the text will develop: does he recognize the questions and themes brought up, or will he consider them as Spielerei and willingly sustain in his ‘belief’.
An extradiegetic metalepsis is very often conceived as a digression of the narrator (thus being a narratorial discourse metalepsis), which puts the story on hold and lets the discourse go on for a while on its own, as the reader apostrophe in Mist illustrates. However, this is no condition for the appearance of extradiegetic metalepses: in Schoenfeld’s “Built Up Logically”, there is a simultaneity between story and discourse:
“Is there a typewriter here?” I asked. “On the desk,” Sally said. […] I nodded, inserted a sheet of paper in the typewriter, and went on with the story: […] By a coincidence arranged by me as the legitimate author of the story, the pistol exploded on landing, sending a bullet into the brain of Frank who was still asleep across the street on the front stoop of a brownstone house. Frank slumped forward and rolled into the gutter, dead, a grim monument and warning to all characters with rebellious spirits. I grinned and added the last two words to the story: THE END. (Schoenfeld 1950: 40)
In this example, the metalepsis does not interrupt the story. Story and discourse coincide, so that the extradiegetic narratorial discourse metalepis becomes a part of the story as it goes on.
Furthermore, the kind of narrator also plays a decisive role. If an extradiegetic homodiegetic narrator refers for example to his power over the diegetic characters, this reference automatically becomes an extradiegetic metalepsis, whereas an extradiegetic heterodiegetic narrator, alluding to the will the characters are submitted to, does not automatically create an extradiegetic metalepsis. In novels with a homodiegetic narrator, extradiegetic metalepses are more common than in works in which a heterodiegetic narrator relates the story. Although extradiegetic metalepses are mostly narratorial transgressions, we occasionally encounter a figural variant, as in Henry N. Beard and Douglas Kennedy’s Bored of the Rings: when Bromosel hears “you cash in your chips around page eighty-eight,” he “looked up to the top of the page and winced” (1993: 57). Further on in the novel, the reader is again reminded of the fictional quality by similar thoughts of Bromosel: “‘No’, agreed Bromosel, looking across the grey surface of the page to the thick half of the book still in the reader’s right hand” (69). Bromosel, a character in the novel, is the one who provides this parody of Lord of the Rings with a self-referential character by mentioning the material conditions of the text.
It goes without saying that texts very often present both diegetic and extradiegetic metalepses. In the above-mentioned novel Mist, for instance, the reader is first addressed by the (extradiegetic) narrator, preparing him for what is to come, and only afterwards do we learn how Augusto decides he wants to meet the author of the interesting essays he is reading and how, during this conversation, he realizes that he himself is also a product of this author’s imagination (diegetic). In the very illogical short story “Built Up Logically,” diegetic and extradiegetic metalepses are also mixed up. Nevertheless, there is an opposing tendency in the use of each kind of metalepsis: whereas diegetic metalepses appear mostly in fantasy novels, stressing the imaginative quality of the stories and, paradoxically, increasing reader immersion in the text, extradiegetic metalepses tend to occur in ‘realistic’ texts, thematizing (though sometimes only implicitly) the relation between text and reader and between text and reality, and questioning the medium’s capacity of a text to capture reality. Extradiegetic metalepses sometimes even cast doubt upon the ontological status of the reader.
3.2. Unmarked metalepses
Within the group of extradiegetic metalepses there is in my opinion a special form which can be illustrated with an example from Jean Pauls Blumen-, Frucht- und Dornenstücke oder Ehestand, Tod und Hochzeit des Armenadvokaten F. St. Siebenkäs (1796–1797). When the narrator describes how Siebenkäs and his wife go to bed, he comments on this event as follows: “Now I wish the entire royal family a good night and I hope they will awake safe and sound in the eight chapter” (Jean Paul 1796–1797: 43, translation mine). This example does not substantively influence the development of the story but only comments on the events in the story. In my opinion, the above-mentioned examples from Illusions perdues and Joseph Andrews also fit this category. Bernd Häsner considers only such transgressions as being properly metaleptic, because he reserves the term metalepsis exclusively for illogical or impossible relations which manifest themselves as an illusionistic simultaneity and contiguity of discourse and story (cf. Häsner 2001: 30). Although I do not agree with Häsner in considering only such relations to be metaleptic, I do think this group of metalepses should form a separate subclass. While some scholars recognize the existence of this particular group of transgressions, they do not take into consideration their effects. Nelles, for example, describes such transgressions as unmarked metalepses, Pier characterizes them as minimal metalepses and Fludernik, as mentioned above, shifts between defining them as discourse metalepses and type 1 ontological metalepses. According to Genette, such transgressions are not metalepses but narrative syllepses. Strange enough, however, he defines the widely cited example from Balzac’s Illusion perdues not as a syllepsis but as metalepsis. Also Grabe et al. (2006) and Liviu Lutas (2011) advocate defining this phenomenon as a syllepsis, based, however, on a different argument. Lutas claims that the temporal dimension is crucial for a syllepsis (2011: 55), referring to Genette’s remark that syllepsis affects the succession and duration of related events ( 1980: 155). Grabe et al. (2006) describe syllepsis as the simultaneity of non-simultaneous events. According to their investigation, syllepsis does not cross any borders but fades them out. I, in contrast, agree with Fludernik, who argues that the transitions in the examples mentioned above are realized because in such digressions the extradiegetic narrator projects himself into his story. It is precisely this projection that incorporates the border transition. By entering his story, the narrator emphasizes his controlling influence and acknowledges that he is not reporting a story but inventing it. Such digressions of the narrator do not change the course of the story but simply mark simultaneity between narrative time and narrated time: the narrator takes his reader by the hand and accompanies him as the story develops. As Fludernik observed, such techniques originate from the oral storytelling tradition and were used until the fifteenth century to structure longer texts or to facilitate the transition between different scenes (2003: 389). Unmarked metalepses thus have a long history but remain rather ‘innocent’: they create a pause in the story and let the discourse continue on its own for a while, but in no way do they influence the development of the events.
In this paper I have presented an alternative classification for metalepsis. I consider metalepsis to be a text-internal transgression between hierarchically arranged universes. Because metalepsis is a text-internal transition, it creates a structural paradox in the text. Whereas most typologies only distinguish between discourse and story variants, I think it is important to discern primarily another, broader difference: a distinction between metalepses on the diegetic plane and those on the extradiegetic plane, that is, a distinction between metalepses that only influence the told story and metalepses that affect the entire text, i.e. story and the highest discourse level. The different effects these metalepses produce are often attributed to the story-discourse dichotomy. However, I believe that they should be explained from this broader perspective. Diegetic metalepses appear more frequently in fantasy novels and tend to immerse the reader in the fantastic story world. Texts displaying numerous extradiegetic transitions, on the other hand, become self-referential and are often food for thought about the status of the text in relation to its author and its reader audience and about the medium ‘text’. Moreover, because they appeal in some way to the extratextual reality of the reader, they compel him to think about his own position with regard to the text and thus make sure he becomes a parameter influencing the further development of the text.