The Problem of Fiction in Literary Theory
Classical approaches to defining fiction
Following Jean-Marie Schaeffer (2014 ), we subdivide the debate on how to differentiate fiction from non-fiction into three groups.
- 1) The semantic definition is inspired by philosophical theories of reference and rests on the idea of an ontological divide between fiction and non-fiction: either a narrative has a reference to the “real” world (in the case of factual narratives) or not (in the case of fictional narratives). A challenge for this approach arises from so-called immigrant objects (Zipfel 2001, 92), elements imported from the real world into a fictional universe (e. g., the notorious example of Napoleon’s appearance in a novel). Lubomir Dolezel offers a solution by stating that real-world elements, when appearing in a fictive world, undergo “a substantial transformation at the world boundary into non-actual possibles, with all the ontological, logical and semantical consequences” (1988, 485). However, Dolezel does not explain what this mystic conversion (e. g., of real-world persons) means in concrete terms; we agree with Frank Zipfel (2001, 94) that it is extremely counter-intuitive to label real persons or places in fiction as fictive objects out of the desire to achieve theoretical rigour. Wolf Schmid (2010 , 32), on the other hand, highlights the consequent danger of gradating the fictive.
- 2) A syntactic definition seeks to distinguish fictional and factual narratives by means of distinctive formal criteria that are inherent to the text. Probably the best-known representative is Kate Hamburger, who describes assumed fiction-signals in third-person narratives. This occurs with the use of the epic preterit that ceases to indicate the past in fictional contexts, the use of verbs of inner action to portray a character’s inner thoughts and the use of free indirect speech (1973 [1957/1968], 59-133). Because of her logico-linguistic approach, Hamburger was led to exclude first-person narration from the realm of fiction which, as Schaeffer (2013, 189) rightly points out, is a highly counter-intuitive conclusion. Moreover, as highlighted by Walsh (2007, 44-45), the supposed essential formal characteristics of fictional narratives undergo changes and are thus not universally valid. Henrik Skov Nielsen (2011, 114) also objects to a syntactic definition, pointing out that alleged fictional techniques can be borrowed by non-fictional texts. It follows then that inherent signals of fiction are neither necessary nor sufficient to identify a work of fiction as such.1
- 3) A pragmatic definition is proposed by John Searle. By way of contrast to Hamburger, he dismisses the existence of a “textual property, syntactical or semantic” (Searle 1975, 325) that allows a clear distinction between fictional and factual narratives. Instead, he identifies as the decisive criterion the “illocutionary intentions of the author” (325) and defines fiction as a pretended speech act. Searle’s approach has been contested, among others, by Kendall Walton (1990), who rejects the idea that intentionality can suffice as a valid criterion for the definition of fiction. Nevertheless, Schaeffer points out the importance of the interplay between a “fictional intention” and the perception of a given work as fiction: “[I]f it is true that fictional intention cannot define fiction as a pragmatic stance, it is nevertheless the existence of a shared intention which explains the fact that the emergence of fictional devices has the cultural and technical history it has” (2014 , 190). In short: fictionality, as defined in approaches which emphasize the pragmatic dimension, is not based on intrinsic features of a given work but rather on conventions and on a kind of contract between author and receiver. 
Nielsen (2011, 113-114) terms all three approaches that seek to define cate- gorial differences between fiction and non-fiction as separatist or exceptionalist. Furthermore, he identifies two positions that do not consider fiction and non-fiction to be opposites:
- 4) A panfictionalist approach claims that all narratives are artificial and therefore fictionalizing because they are the result of a transformation of facts into language. Hayden White, one of the best known representatives of this position, states that historians, for instance, charge real events with the symbolic significance of a plot structure, a feature that is also typical of fiction (1978, 92). Nielsen is a resolute opponent of panfictionalist approaches: even if we agreed that all fiction is artificial and that all narrative is artificial, the logical consequence would not be that all narrative is fictional (2011, 114).
- 5) A similarist or non-fictionalist approach, inspired by cognitive theory and furthered by Monika Fludernik’s Towards a ‘Natural’ Narratology (1996), is sup?ported by David Herman, for example. According to Herman, conversational narratives are prototypical, and all other narrative forms derive from these. Moreover, he argues that we always activate the same cognitive repertoire and thus the same translation apparatus to understand all narratives (2009, 6). Diverging from Herman’s very popular approach, Nielsen and his co-authors argue that we activate a particular translation machine when dealing with fictionality (Jacobsen et al. 2013, 19-20).
To summarize the debate, most recent contributions agree that none of the elements thus far examined deliver clear-cut criteria for identifying fictional narratives. On the other hand, approaches that eliminate the divide between fiction and non-fiction are also widely rejected. It seems that everybody recognizes fiction when they are confronted with it, but why this happens is still a mystery.
-  Two years later, Nielsen and his co-authors in Fiktionalitet were to outline ten theses aboutfictionality, where this position is summarized as thesis 5 (Jacobsen et al. 2013, 44). Again twoyears later, Nielsen et al. reworded this conclusion in the article “Ten theses about fictionality”as thesis 6: “No formal technique or other textual feature is in itself a necessary and sufficientground for identifying fictive discourse” (Nielsen, Phelan, Walsh 2015, 66-67).
-  For a more detailed critique of Searle’s approach, see Zipfel (2001, 185-195).