Literarinesses—A Bag of Three-Sided Coins

Abstract

The theoretical question of what makes texts “literary” has a long tradition in literary studies. At the level of concrete individual encounters/transactions between readers and texts, literariness has been shown to reflect how actual readers pre-categorize, approach, and process texts. Literariness has been approached from three different angles: the study of formal and semantic features of literary language, which dates back to the formalist beginnings of the concept; the study of literary reading modes and the generalized literary categories in which they are grounded; and the study of actual reading experiences. We argue (1) that these three aspects are mutually dependent and, in fact, constitute three sides of the same coin and (2) that different texts and genres instantiate distinct literariness profiles, that is, distinct ‘literarinesses’ in the mind of the reader—what makes a text literary differs between text types. Building on previous work in linguistics, literary studies, psychology, and stylistics, we discuss the cognitive implications of these two central claims for the reader. We also integrate our approach with extant research on genre-specific profiles and develop a set of ideas for future research in this field.

1. Introduction

Literary texts employ a wide variety of phonological, morpho-syntactic, semantic, and pragmatic devices, schemes, and techniques that raise questions of perception, memory, and consciousness, inviting interdisciplinary dialogue with linguistics and cognitive science. Not merely reacting to these linguistic techniques, readers tend to adapt how they read to what they believe to be reading, which in turn influences their reading experiences. In short, the literariness of texts changes how we experience them.

This idea is not new. More than one hundred years ago, Roman Jakobson raised the issue of literariness by claiming that “the subject of literary science is not literature, but literariness, i.e., that which makes a given work a literary work” (Jakobson [1919] 1973, p. 11). Literariness has since been approached from three different angles: the study of (1) formal features of literary language, which dates back to the beginnings of the concept (Jakobson [1919] 1973); (2) literary reading modes and the conventions in which they are grounded (Randall 1985); and (3) actual reading experiences.

At a first glance, discussions of literariness may seem dated in the context of current literary studies, where definitions of “literature” are contested and it has become common to speak of “text” or “discourse” instead, purposefully avoiding Jakobson’s aforementioned question (Klarer 2023). This reflects the well-established understanding that no medium, message, or property of language belongs exclusively to literature. At the same time, the core interest of literary studies, despite recent interest in the new media, in political speeches and in other forms of discourse, remains firmly centered around texts and genres that are not, or not primarily, tied to matters of practical use and that can be described as original in terms of message and/or unusual in terms of form. In addition, literary studies still use poetological approaches, as well as historical approaches, to work with these texts and genres. In that sense, while discussions of what makes a text literary are not at the forefront of literary studies right now (but see more recent empirical takes, e.g., Salgaro and Sopcak 2018) and definitions of literature are viewed as problematic, identifying texts as literary on an operational level appears to present no serious issue. However, this gap between theoretically avoiding a phenomenon while making it the center of a discipline represents an epistemological lacuna in literary studies: what is the nature of the knowledge we are creating? Based on this observation, and also in the light of the radical changes writing and reading face in the age of hypertext fiction, “twitterature”, “insta-poetry” and other digital forms, not to mention AI-created texts, it seems highly relevant today to revisit the term “literariness” and to ask how, why, and under what circumstances we perceive texts as literary (or not) and how we categorize, process, and evaluate them if we do.

In this paper, we contribute to answering these questions with a cognitive approach to literariness that is not only based on the study of individual differences, but also informed by the study of formal text features and genre constraints and by their effects on the reader. In so doing, we posit that the cognitive effects of reading are reactions to a selection and combination of linguistic signs within a medial framework—that is, to a particular text with particular properties. Building on previous work in linguistics, literary studies, psychology, and stylistics, we make two central points: one, that the three different conceptions of literariness outlined above (formal features of language, literary reading modes, and reading experiences) are mutually dependent and, in fact, constitute three sides of the same coin and, two, that different texts and genres instantiate discrete literariness profiles, that is, distinct ‘literarinesses’ that are based on different readerly expectations, behavior, and reading experiences. After introducing the different takes on literariness in more depth, we will discuss these two points to explicate and illustrate our own notion of ‘literarinesses’ in terms of genre-specific text-reader relations during a number of sub-processes of text comprehension. In doing so, we refer to extant research on genre profiles as well as develop hypotheses for future research on genres and text types that have not yet been studied in detail.

2. Three Takes on Literariness

A leading figure in Czech Structuralism, Jakobson pursued a form-based approach, conceiving of literariness as a text property determined by choices in the selection and combination of linguistic signs. However, already the Russian formalists had been well aware that these choices shape and determine the responses of readers and that defamiliarization can be effected by employing unusual formal techniques or devices (Shklovsky [1927] 1965). The form-based approach to literariness was further developed by other Czech structuralists; most notably, Mukařovský described poetic language as being characterized by “the violation of the norm of the standard” language (Mukařovský [1932] 1964, p. 18). Here, too, the effect on the reader was central to the conception of literariness: text elements that deviate from the norm were considered foregrounded and supposed to de-automatize the act of reading. However, Mukařovský ([1932] 1964, p. 21) carefully noted that “every work of poetry is perceived against the background of a certain tradition”, meaning that the degree to which a component of a poetic text is perceived as foregrounded depends not only on the norms of the standard language, but also on the norms of an “automatized poetic canon”. Furthermore, the Czech structuralists acknowledged that the standard language is not homogenous, but functionally and stylistically differentiated, and that its functional dialects come with their own sets of norms and automatizations (Havránek [1932] 1964). The position of the formalists and structuralists has often been reduced to a purely form-based approach and hence been rejected as insufficient, despite the fact that they took into account effects on the reader and advocated a remarkably complex and diversified functional theory of language. In particular, the work of Potebnja (1862), which is influenced by the works of Kant and Wilhelm von Humboldt, reflects the role of subjective creativity during writing and reading. By going through either the process of production or the process of reception, the individual arrives at an aesthetic experience that is fundamentally “literary”.

Moreover, the (not entirely justified) criticism of structuralism and formalism, which emerged with and outlived poststructuralism, disregards that a core understanding of literariness as an inherent feature of literary texts still appears to linger, evidenced, for instance, by studies that continue to describe the artfulness of literary works on the basis of linguistic properties, such as verbal imagery (e.g., Castore 2023; Habibovna and Toshpulat 2025; Lock 2025, to name but very few among the many recent examples).

In the middle of the 20th century, two important developments took place that transformed the prevailing conception of language and inspired new approaches to literature and literariness. For one, pragmatics emerged as a scientific theory of verbal behavior and as an academic discipline (e.g., Austin 1962; Searle 1969), continuing and further developing the functional view of language that had characterized Czech Structuralism. Furthermore, the focus of linguistic inquiry shifted away from the Structuralist interest in language as a semiotic system towards the development of more concrete, psychologically oriented theories that focused on the representation of the semiotic system in the mind of the speaker/hearer (e.g., Chomsky 1957, 1965), on the way it is used in real-time comprehension (e.g., Bever 1970), and on how it interacts with other aspects of the human mind (e.g., Kintsch and Keenan 1973; Miller and Isard 1964). Around the same time, researchers in psychology, linguistics, and artificial intelligence began to take an increased interest in reading (e.g., Kintsch and van Dijk 1978; Mandler and Johnson 1977; Schank and Abelson 1975).

In the field of literary studies, two major developments transformed the understanding of literariness: First, the infamous “death of the author” (Barthes 1967) led to an growing focus on the reader, resulting in an unprecedented rise of reception theories that became more and more informed by psychology and cognitive science. While Barthes was of course not the first to point to the reader (cf., e.g., Potebnja 1862), his work caused a seismic shift in literary studies. Second, in the context of this budding field of reception research, Jauß (1967) and Iser (1976) extended the focus the structuralists and formalists had placed on formal features of language to interactions between semantic content and linguistic realization. They proposed that the function and decisive characteristic of literature was its ability to reframe real-world or real-world-compatible information in unfamiliar contexts, thereby allowing readers “to redefine, modify or suspend [their] usual schemata” (Salgaro 2015). This approach to schema theory has remained popular in literary studies, mostly through the works of Eco (1979) and Stockwell (2002), and has, for instance, been applied to the study of narratology (e.g., Fludernik 1996; Herman 2009 cf. Emmott and Alexander 2014, for a review) or the study of poetry (e.g., Hühn and Kiefer 2005). In a similar vein, Cook (1994, p. 182) defined literariness as a type of “discourse deviation”, which contradicts the reader’s assumptions, but also offers new semantic impulses and brings about new links between information not previously linked on the mind of the reader.

In the wake of the so-called “cognitive turn” in the middle of the 20th century, a second family of approaches to literature and literariness emerged that shifted the focus away from the text and towards literary modes of reading. This shift had already announced itself in some early 20th century theories (Richards 1924; Rosenblatt 1938), and it reflected a growing interest in the reader’s knowledge (e.g., Bierwisch 1965; Ryan 1979) and their activity (e.g., Fish 1970; Iser 1976; Rosenblatt 1978; Smith 1968). According to this view, a text becomes literary if and when the reader perceives it as literary, as explicated by Stanley Fish, one of the most eminent proponents of this position: “Literature is […] an open category, not definable by fictionality, or by disregard of propositional truth, or by a statistical predominance of tropes and figures, but simply by what we decide to put into it.“ (Fish 1973, p. 52). Hence, the reader and not the text becomes the locus of a dichotomy between literary and non-literary texts; that is, readers are the ones who classify what they read as literary or non-literary and adapt their reading behavior accordingly.

To Fish, “[w]hat characterizes literature […] is not formal properties, but an attitude […] towards properties that belong by constitutive right to language.” Parts of literary studies have judged Fish harshly for this position, accusing him of completely disregarding medium, message, and formal properties of the text. However, Fish does acknowledge that textual features impact reading behavior (Fish 1980). What is more, the idea that carefully crafted texts control the temporal flow of the reading experience is central to his method of functional and incremental close reading: “Whatever the size of the unit, the focus of the method remains the reader’s experience of it, and the mechanism of the method is the magic question, what does this------do?“ (Fish 1970, p. 138). Attributions of literariness, he insists, are constituted by what we have learned about literature through reading or being taught, and they can be understood differently according to perceived context (Fish 1980). As such, to him, they stem from the reader, from conventions established and passed on between readers (e.g., in class), and are influenced by assumptions made on the basis of extra-textual context.

This attitude involves literature- and genre-specific strategies pertaining to both reading and interpretation (de Beaugrande 1978; Culler 1975; Rosenblatt 1978, 1988); it effectively makes literary reading a mode of context-dependent language use and, as such, ultimately a pragmatic phenomenon (e.g., Leech 1969; Pratt 1977). Psychological models of discourse comprehension apply this idea to text processing in general and posit that text-type specific strategies extend beyond interpretive ones and affect almost the entire reading process. Strategic processing is assumed to involve the way readers attend to, gather, process, represent, and store the information contained in a text (e.g., Kintsch 1998; van Dijk and Kintsch 1983; Zwaan 1993). Revealing behavioral and neural correlates of text-type specific reading strategies, empirical studies of literature provided compelling evidence for literary and genre-specific modes of reading, evaluating, and interpreting texts (e.g., Blohm et al. 2017, 2021, 2022; Fischer et al. 2003; Gavaler and Johnson 2019; Gibbs et al. 1991; Hanauer 1998; Zwaan 1993; Knoop et al. 2024; Wagner and Betz 2023). This body of research reports a higher acceptance for lexical, grammatical, or semantic properties that are at odds with the norms of the written standard but are apparently licensed when seen against the background of a certain tradition.

A third approach to literariness centers on actual subjective reading experiences. Already alluded to in both Richards’ notion of the reader’s “stream of reactions” to a text (Richards 1924) and in Rosenblatt’s transactional theory (Rosenblatt 1938, 1978, 1988), this position is closely tied to the idea of literary reading modes. However, it does not reduce literariness to a reading stance involving selective attention, interpretive strategies, and specific behavioral correlates but rather highlights the individual experience during the act of reading. Most notably, Miall and Kuiken (1994, 1998, 1999) developed a three-component model of literariness involving “foregrounded stylistic or narrative features, readers’ defamiliarizing responses to them and the consequent modification of personal meanings” (Miall and Kuiken 1999, p. 121). In this view, text properties and reader characteristics are still important, but they are considered necessary yet insufficient conditions for literariness. More recently, Appel and colleagues (Appel et al. 2021) proposed an understanding of literariness that “is introduced as (a) a function of specific textual features that create linguistic foregrounding and (b) the positioning of a text as literary through para-textual signifiers (such as non-fiction and fiction labelling)” (p. 177). This approach differs fundamentally from earlier elaborations on literariness in that its focus on foregrounding shifts the experience of literariness onto the individual reading experience: different aspects of the text may appear foregrounded to different readers and during different readings. The weak point of these approaches lies in the fact that elements which may seem quite conventional and are simply too subtle to be perceived as foregrounded, such as poetic meter, can have a considerable impact on the aesthetic experience and on the perception of literariness.

All three approaches identify crucial and complementary rather than mutually exclusive aspects of literariness. Indeed, none of the proponents claimed that the different aspects can be neatly distinguished (cf. Winko 2009). Both Miall and Kuiken and Appel and colleagues emphasize this point by integrating all three components into their definitions of literariness. Nevertheless, the implications of their focus are not to be underestimated: by treating formal features of language as response triggers, they shift the issue of literariness entirely from text to discourse.

We offer an alternative proposal that is fundamentally integrative as well; however, our approach does not primarily hinge on the unique and the subjective, including foregrounded textual elements. Instead, it highlights the importance of the prototypical and the conventional in the formation of generalized semantic categories while simultaneously identifying sources of inter-individual variation and offering sufficient flexibility to account for subjective experience. In so doing, we suggest that a concurrent focus on both text and discourse is not only possible, but in fact mutually beneficial: the study of textual features typically forms the basis of testable hypotheses regarding effects on the reader. In turn, measured effects on the reader allow assumptions about the workings of different textual elements and of their contribution to perceived literariness.

This position, like others before it (cf. Appel et al. 2021; Miall and Kuiken 1994; Winko 2009), integrates and relates both text- and reader-specific properties and situates literariness in the dynamics of the reading. It builds on three central points: First, we assume that every single textual element, whether foregrounded or highly automatized, affects the reader’s mental representation of the text and thus contributes to the subjective reading experience. In that sense, unlike Appel et al. (2021) or Miall and Kuiken (1994), we do not attribute any exclusive effects to foregrounded textual elements. Recent research (e.g., Menninghaus et al. 2017) has shown that even subtle features or highly generalized literary categories can have strong emotional and cognitive effects, thereby altering the aesthetic experience. Second, and related to that, while our approach is obviously a cognitive one, we attribute paramount importance to the nature of all textual elements (foregrounded or not), assuming that they all serve to trigger, feed, direct and, generally, affect distinct cognitive processes; therefore, small-scale, in-depth text analyses that explore the text’s full formal and semantic potential should be linked to empirical investigations of their cognitive effects. Thereby, the mastery of the writer, in its historical contexts, can be analyzed alongside effects on the reader, and viable hypotheses regarding the effects of comparable works or passages can be derived. This can include aspects as seemingly trivial as arranging semantic information in a manner that makes it appear suspenseful in a crime novel, whereas the same information could be arranged to be simply informative in a newspaper article. Moreover, our text-analysis-based approach allows for a better understanding of genre- and text-specific distributions and relations of the cognitive effects of generalized literary categories vs. novel textual features.

Third, we suggest that the reader’s text categorization is one of the main determinants of text processing and evaluation and acts as an additional constraint on the construction of meaning. For instance, if a reader categorizes a text as a fairy tale, this pre-categorization impacts the semantic integration of counterintuitive information within a time window as early as 400 ms (Knoop et al. 2024). In this respect, our position is closely linked to the notion of literary reading modes as advocated, for instance, by Stanley Fish and Louise Rosenblatt.

3. Two Claims on the Nature of Literariness

Synthesizing the three predominant approaches to literariness, we formulate our two initial claims more precisely:

(1)

The three different conceptions of literariness outlined in the brief historical sketch above are mutually dependent and, in fact, constitute three sides of the same coin. Literariness is constituted of readers’ text-feature configurations that are associated with literary genres, resulting reading modes, and actual reading experiences. As such, we understand literariness as a quality of the reading experience. What may seem like a bold claim to some and trivial to others (“no reader, no literariness”) fundamentally hinges on the observation that readers are the ones to furnish texts with the label “literary”. As a result, literariness has always been a fluid category: while some texts have become firm parts of literary canons, others have been moved in and out; and even the category itself has broader or narrower boundaries depending on who defines it, ranging from all writings to texts read for pleasure, to works of art that are meaningful in both content and form, to works with cultural significance or societal value, to texts drawn from authorial imagination rather than practical purpose, to texts displaying formal or semantic creativity (cf., e.g., Culler 2007; Eagleton 1983; Farber 2005). Many definitions are based on value judgments, but a few are based on prototypes (e.g., Meyer 1997). Either way, literariness remains a dynamic idea, attributed to texts by readers based on different, pre-determined categories. Literary theory traditionally only acknowledged attributions of literariness by ‘professional readers’, i.e., literary scholars, whereas our approach pertains to any act of reading, regardless of the reader’s professional status and experience.

(2)

Genre profiles/schemata constitute distinct ‘literarinesses’: each literary text instantiates one or more of these profiles to varying degrees.

We propose a framework that incorporates categorical-text-type classifications (e.g., literary/non-literary, prose/poetry, ballad/sonnet, …) but at the same time assumes degrees of literariness as partial conformity with genre prototypes. For instance, if readers notice that the protagonist of a novel resembles the hero of a tragedy, they will probably not start doubting that they are reading a novel, but the resonance with the tragedy schema is likely to influence their further reading experience. In other words, we claim that genre-specific text properties resonate with the reader the same way as Miall and Kuiken (1999) assume for stylistic variation, i.e., prompting the reader to pursue particular themes and reading strategies.

4. Three Components of the Act of Reading

We assume three basic components of the act of reading: the reading situation, the reader, and the medium that contains an ordered set of linguistic signs (text-as-signs). Thus, at the most general level, the act of reading can be seen as a real-world event “involving a particular reader and a particular configuration of marks on a page, and occurring at a particular time in a particular context” (Rosenblatt 1988, p. 4).

The reading situation includes the immediate environment of the reader but also serves as an interface with the historical and cultural context of the act of reading; only those aspects of the reading situation are considered relevant that potentially influence how the reader categorizes, processes, and evaluates the text.

4.1. The Knowledge of the Reader

In a typical act of reading, the reader is the only agent, initiating, interrupting, and resuming the reading process at will. Readers incrementally convert the linguistic signs into mental representations which, for them, constitute the text (text-as-representations). At first glance, the fundamental distinction between text-as-signs and text-as-representations (cf. Rosenblatt 1978; Schmidt 1982) may seem trivial; its far-reaching consequences, however, are too often disregarded. The text-as-representations and the experience of constructing it are all that is accessible to readers and their judgment, and all that they react to. But no matter how carefully we analyze the written text (text-as-signs), this analysis neither gives access to the text representations readers form, nor does it allow us to infer how they respond to them; all we can (currently) hope for is making informed guesses based on empirically validated theories.

Whereas the text-as-signs holds all the cues, the reader provides the knowledge required for turning them into mental representations (Frye 1947, pp. 427–28; Kintsch 1998). Four types of knowledge are essential to literary comprehension: world knowledge, linguistic competence and reading skills, pragmatic competence, and literary competence.

First and foremost, the reader brings their world knowledge to the act of reading: lexical knowledge of concepts concrete and abstract; propositional knowledge that specifies relations between concepts (e.g., “Mark Twain wrote Huckleberry Finn”, “The sky is blue”); and so-called script knowledge (Schank and Abelson 1977), i.e., partially ordered sequences of propositions that denote states and events in real-world contexts (e.g., the prototypical sequence of actions involved in going to a restaurant) and that emerge from the reader’s event memory.

Secondly, the act of reading depends on the reader’s linguistic competence and reading skills: To varying degrees, readers have internalized the semiotic system that links the form and arrangement of visual and acoustic signs to conceptual nodes (the lexicon) and to the relations that hold between them (the grammar). Linguistic competence develops from early childhood into adolescence and shows substantial differences between children and adults and between early and late language learners. While there is a considerable degree of inter-individual variation, the functionality of language attests to the large amount of overlap between members of a speech community.

Usually, adult readers will have accumulated considerable knowledge about how the semiotic system can and should be used in different contexts. In daily life, their pragmatic competence allows them to manipulate verbal signs to attain desired goals but also to understand more than their interlocutors actually express, i.e., to infer their intended meaning or their desired goals. Pragmatic competence emerges in fairly early stages of first language acquisition, but its development extends beyond the maturation of the phonological and grammatical subsystems.

Native-speaking adult readers with years of reading experience (and likely explicit instruction) will be familiar with a range of literary and non-literary texts, and the peculiarities of the written modality will become an integral part of their pragmatic competence. As far as these texts are typically considered literary, this part of the readers’ pragmatic knowledge can be said to be literature-specific, and thus constitutes the basis of their literary competence. Over their lifetime, readers acquire and extend an individual literary corpus, leading to the emergence of prototypical categories based on multiple repetitions of similarities and contrasts. Based on the fundamental idea that multiple repetitions of occurring patterns lead to the emergence of schematic representations (Bartlett 1932; Ghosh and Gilboa 2014), we conceptualize the reader’s literary competence as the generalizations across their individual literary corpus, i.e., as the categories developed on the basis of prior reading experiences. For the present purpose, we are mainly concerned with genre categories and the emergence of genre schemata that presumably show considerable overlap with the schemata of other readers with similar experience (Ghosh and Gilboa 2014; Viehoff 1995). As will become apparent in our discussion of genre-specific literary reading, these genre schemata play a central role in our conception of literariness.

But what exactly is a genre? In the broadest sense of the term, text genres are classes of written works that are grouped together because they share a particular profile comprised of formal and semantic properties (Norton and Rushton 1941). These shared properties are then tied to that particular genre. Of course, genre boundaries tend to be fuzzy and marked by works that do not meet all possible properties of the genre, and new developments may change the scope of genre schemata. In addition, genre emerges as one of several generalized semantic categories from each reader’s individual reading experiences and therefore remains dynamic (cf. Schneider and Hartner 2014). Whereas some genres, such as novels or tragedies, are considered exclusively literary, some genres may occur in literary and non-literary contexts alike, such as letters, diaries, essays, or political writings. We assume that literary genres are broadly defined by the relative irrelevance of any functional purpose beyond the (extended) act of reading. For instance, while letters have undeniable practical functions, such as informing a relation about certain events or sharing one’s feelings with them, the letter as a representative of a literary genre, such as Kafka’s letter’s to Felice, have moved beyond that purpose and are now being read because they are formally and semantically interesting, both as testaments to the context of their time and of the author’s life, and as reading experiences in their own right. Representatives of other genres, like the fantasy novel, are usually never read in the context of practical purpose beyond the act of reading, not even at the time of their emergence.

Moreover, literary genres are part of readers’ generalized text-type inventories, i.e., they are not fundamentally distinct from other written communication but differ in their specifics and affordances; they form a hierarchy of genres and subgenres, where each subgenre exhibits properties of the super-ordinate genre and additional properties that distinguish it from other subgenres (Steen 1999). Crucially, we situate this hierarchy in readers’ literary competence and conceive the distinctive properties as genre-specific pragmatic defaults and prototypical properties of the text-as-representations. There are instances where a genre is characterized by text-internal aspects that impose very general restrictions on the properties of the text-as-signs, such as the graphic layout of a poem or the mere text length characteristic of the novel; in such cases, the distinctive properties belong to both text-as-signs and text-as-representations and may serve as strong genre cues in text categorization. But how can the less obvious distinctive properties of text structure and content be reconciled with the representational view of the literary text? Assuming a multi-layered text representation (e.g., Ingarden 1931; van Dijk and Kintsch 1983; Zwaan 1993), we propose that genres correspond to sets of constraints on different levels of the text-as-representation, such as restrictions on phonological or lexical structure in poetry (Jakobson 1960) or on character and event configurations in classical tragedies and comedies (Schauber and Spolsky 1986).

4.2. The Text as Representation

Having identified the basic components of an act of reading and the domains of the reader’s knowledge that are indispensable for successful (literary) comprehension, we now turn to a more fine-grained description of the text-as-representations and the processes involved in constructing it.

Across perceptual modalities, comprehension involves the construction of mental representations based on sensory input and prior knowledge (e.g., Bartlett 1932; Kintsch 1998). Language comprehension corresponds to the incremental mapping of acoustic or visual cues (sensory input) onto conceptual representations (prior knowledge) and their relations. The comprehension system relies on the stored code (linguistic competence) and a processing system that recognizes the linguistic signs and their arrangement and transforms them into intermediate lexical, phonological, grammatical, and semantic representations of words, phrases, and sentences.

While pragmatic competence generally allows for context-appropriate interpretation of sentence meanings, the interpretation of verbal utterances in literary contexts is co-determined by additional conventions specified in the stored genre schemata (e.g., Bierwisch 1965; Culler 1975; Pratt 1977; Pilkington 2000; Schmidt 1982; Weninger 1994). Sentence by sentence, the reader constructs a representation of the entire discourse.

Similar to Ingarden’s (1931, 1968) proposed stratification of the literary work of art, we suggest a three-level system of text representation that is adopted from theories of discourse comprehension and that includes a surface structure level, a propositional level referred to as text base, and a situational level (e.g., Graesser 1981; Kintsch 1998; van Dijk and Kintsch 1983; Zwaan 1993).

By focusing their attention on the text-as-signs, the reader initiates and continues the act of reading. They move along the signs of the text but often proceed non-linearly and create unique rearrangements of the signs (cf. Barthes [1973] 1975). Taking in one perceptual chunk after the other, they decode the contained information and construct surface structure representations of words, phrases, and sentences into which they integrate incoming material. At this level, we distinguish lexical representations (words) that link visual forms to conceptual and phonological representations (meanings and sounds) and grammatical representations (morpho-syntax) activated and constructed during incremental comprehension.

All of these surface structure representations are grounded in the semiotic system shared by the speech community, and they are tightly linked to the stimulus (text-as-signs). In this sense, they supposedly show the least amount of variation between readers, even if only imperfectly represented in the orthographic system (e.g., Franz et al. 2022; Šturm and Volín 2023).

Due to severe memory restrictions and the limitations of the available sign repertoires (e.g., grapheme/phoneme inventories), surface structure representations are typically short-lived, i.e., their activation level decays rapidly as they are supplanted by incoming words. This means that readers scans along the text-as-signs, but at each point during reading they have only a tiny portion of its form accessible in active memory and forget much of it as they move along.

Of course, the processed surface information is not lost. Instead, the reader re-codes it as propositions, a representational format that corresponds to the literal meaning of an expression or utterance and that is typically formalized as predicate-argument structures (x is y; x does y to z). Propositions can be related by argument overlap, e.g., when multiple actions are performed by the same character or when several scenes are set in the same fictional environment, or by inherent causal or temporal connections (x happened because y occurred) that may be linguistically flagged or merely inferred and supplied by the reader. Many texts express thousands of (usually) related propositions that form large networks: the text base. The reader incorporates incoming propositions into the text base while simultaneously reducing and organizing it (Kintsch and van Dijk 1978). Supplying information that remains unexpressed, and substituting related propositions with higher-order propositions (walking to the station, taking the bus, entering the building = going to work), they attempt to combine semantic building blocks into complex hierarchical networks that, to them, constitute the meaning of the text (e.g., Graesser 1981; Iser 1976; Kintsch 1998; van Dijk and Kintsch 1983; cf. Culpeper 2001, on the creation of characters as persons on the reader’s mind). Contrary to some models of text comprehension (e.g., Kintsch 1998), we conceptualize the text base as the entire propositional network available to the reader, including not only the propositions contributed by the text-as-signs but also those inferred and supplied by the reader themselves; what we thus highlight is the shared representational format of abstract meaning and not the degree of stimulus-control. However, as the text base incorporates information that is not supplied by the symbols of the text, and as it is transformed by operations that are reader- and not stimulus-controlled, there is considerable inter-individual variation already at this stage of literary comprehension.

Situation models are mental representations of described events and states of affairs (e.g., Johnson-Laird 1983; van Dijk and Kintsch 1983; Zwaan and Radvansky 1998; Zwaan 1991, 1993). In constructing these models, the reader merges individual event memory and world knowledge with the propositional information derived from the text-as-signs. In other words, situation models fuse individual experience and textual information into personal conceptions of characters and events described (van Dijk and Kintsch 1983). Of all assumed levels of text representation, this is the most subjective one, i.e., the one that supposedly varies most between individual readers.

Based on this representational system, we can now refine the notion of literary genres as sets of constraints on representational levels. Following Steen’s (1999) hierarchical approach to literature and literary genres, we assume that basic genres (poetry, prose, and drama) provide the default weightings for the relative informativity and importance of the different levels of the to-be-constructed text representation (Hanauer 1998; Jakobson 1935; Zwaan 1993) and that they specify the defaults and permissible ranges of values for text complexity and discourse types within the text (dialogue, description, narration, etc.). Subgenres are distinguished by more and more specific constraints on individual levels of text representation, with categorical differentiation being greatest with respect to properties of the dominant level. For instance, many poetic subgenres are demarcated by poetic forms (phonological surface structure) whereas subgenres of fictional prose are often defined in terms of text world attributes or plot development (text base/situational level).

Regarding the major literary genres, at least, we have some evidence that each is also associated with genre-specific reading and evaluation strategies in student populations (e.g., Blohm et al. 2017, 2021, 2022; Fischer et al. 2003; Gibbs et al. 1991; Hanauer 1998; Koops van‘t Jagt et al. 2014; Wallot and Menninghaus 2018; Zwaan 1993). Whether or not subgenres lead to observable differences in reading behavior depends to a large degree on the reader’s literary competence. The greater the size of the individual corpus and the degree of categorical differentiation, the more stable and nuanced the reader’s expectations and reading strategies will become.

4.3. Text Categorization, Text Processing, and Text Evaluation

We look at three sub-processes of the act of reading: text categorization, text processing, and text evaluation. This distinction identifies three functional domains of the reader‘s mental activity; it does not imply that the identified sub-processes cannot be further broken down nor does it exclude that the reader engages in other activities that are not directly linked to the reading process per se, most notably conscious reflection on and interpretation of the constructed text-as-representations.

In many instances, the starting point is a text categorization judgment (e.g., Blohm et al. 2017; Hanauer 1995, 1997, 2001; Zwaan 1993) which activates the appropriate genre schema and thus makes the associated knowledge available to the processing system. The categorization judgment may be based on text-external or -internal factors. External factors comprise properties of the medium (e.g., cover, peritext, and text layout) or aspects of the reading situation (e.g., genre labels in bookstores or a class on romantic poetry) but also other people’s judgments (e.g., teachers). If no external genre cues are available, or if available external cues are insufficient for singling out a particular genre, then categorization will be deferred to the actual reading phase and rely solely on properties of the text-as-representations (text-driven schema activation). Differences in expertise appear to have a considerable bearing on the process of text categorization: Hanauer (1995, 1996) has been able to show that the nature and weighting of the cues used for text categorization changes with increasing expertise. Kao and Jurafsky (2012) found similar effects in poetry production by published vs. lay poets. In a study on poetry, Fokin et al. (2022) provide evidence that expertise affects genre-specific reading modes (e.g., reading speed).

Once the network of generic knowledge is accessible to the processing system, it is used to optimize text processing in accordance with the reader’s goals (e.g., Schmalhofer and Glavanov 1986; Vipond and Hunt 1984), resulting in observable modulations of reading behavior and text memory (Hanauer 1998; Schumacher and Avrutin 2011; Zwaan 1993). The processing phase encompasses all of the mechanisms and sub-processes that are necessary to construct surface structure representations from linguistic signs (e.g., word recognition, morphological decomposition, and syntactic parsing) and higher-order representations from the surface form (e.g., sentence comprehension and discourse model building and updating). This also includes mental simulation (Mak and Willems 2021), the complex activation and blending of textual information from literary and non-literary sources known to the reader (Schneider and Hartner 2014), and interpretive inferences (McCarthy et al. 2021).

It is sometimes claimed that literature- and genre-specific language processing involves selective attention to certain aspects of a text (e.g., Hanauer 1998; Rosenblatt 1978, 1988). Although intuitively plausible, this hypothesis is, to the best of our knowledge, not supported by direct evidence; however, observed genre-dependent differences in reading behavior and text memory (e.g., Hanauer 1998; Tillmann and Dowling 2007; Zwaan 1993) seem to suggest that readers allocate attention differently when reading different genres.

Based on these and similar findings (Blohm et al. 2017, 2022; Fischer et al. 2003; Schumacher and Avrutin 2011; van Dijk and Kintsch 1983), we assume that each genre is associated with specific expectations that co-determine what a reader attends to and how they gather, process, and store the information accessible in a text.

In addition to the processes that are necessary to construct representations from sensory input and lower-level representations and to integrate incoming information, we assume evaluation processes for each level of representation. These comprise the assessment of well-formedness (e.g., conformity with the phonotactic or syntactic rules of the language), coherence, plausibility, or novelty, i.e., consistence with prior experience. They may trigger controlled repair processes that operate within the respective representational level, e.g., bridging inferences if propositions cannot be related on the basis of the expressed information. In general, there is a gradient increase from more automatic evaluation at the lower representational levels to more controlled and self-related evaluation (significance, valence) at the higher levels that mirrors the increase in diversification and subjectivity as the textual information is transformed into less and less stimulus-bound representations. We assume that, analogous to the genre-specific defaults for text processing, each genre comes with its own defaults for text evaluation (Blohm et al. 2017; Culler 1975; Fish 1973; Gibbs et al. 1991; Schmidt 1982; Schumacher and Avrutin 2011) that emerge from prior reading experiences (Nenadić et al. 2019).

5. Literarinesses—A Bag of Three-Sided Coins

5.1. Literariness—The Three-Sided Coin

We have established how we conceive literariness as a property of the act of reading (Fish 1973; Hoffstaedter 1987) that comprises the reader’s mental representation of formal and semantic text features, genre-specific reading modes including text processing and evaluation, and actual reading experiences. In the following, we will show how genre (and author) schemata accommodate and relate these three different aspects of literariness, and what role they play in the act of reading.

5.1.1. Literariness I: Text-Feature Configurations Associated with Literary Genres

The first component of literariness lies in the formal and semantic text features as represented by the reader. More precisely, it lies in the configurations of surface structure and text base elements that become part of the respective literary genre schemata with repeated exposure. Instances of formal literariness features include formulaic expressions (like “Once upon a time” in fairy tales) or lineation and parallelistic sound patterning (in poetry), whereas examples of semantic literariness features include genre-specific stock characters (in comedy), prototypical settings (like in the Gothic novel), or plot configurations (in classical tragedy). Of course, these features are not restricted to literary texts only, but can also occur in non-literary texts (e.g., de Beaugrande 1978; Culler 1975; Fish 1973), or even be associated with other non-literary genre schemata. For instance, rhyme is a typical feature of advertising. The properties of the pragmatic model constructed by the reader (who is telling what to whom?) might also be included among the key determinants of Literariness I (van Dijk 1997).

What we claim, then, is not that sets of features offer definite genre demarcations, but that genres are characterized by prototypical and permissible text-representational profiles and that the profiles of literary genres are distinguishable from those of the non-literary ones. We further propose that distinctive features resonate with the genre schema they are associated with: If a given text results in representations that provide sufficient match with a stored genre schema, then the activation level of that schema will increase to such a degree that it affects text processing and evaluation, and the sufficiently competent reader will perceive distinctly literary (poetic, tragic, etc.) qualities in the text representation. Hence, Literariness I hinges on the ability of the reader to recognize formal, semantic, and discourse properties and thus depends on their linguistic, pragmatic, and literary competence.

An interesting and poorly understood tension arises from this text-schema link if the recognized features contradict the reader’s text categorization judgment. In such cases, readers might not alter the actual categorization but still read under the influence of two (or more) genre schemata, and thus accept, for instance, absurd occurrences in the context of what they consider to be a fairy tale (Knoop et al. 2024). However, some recent evidence suggests that actual reading experiences can override schematic genre expectations (Gavaler and Johnson 2019). These discrepancies in available evidence might be due to the fact that prototypical expectations differ greatly depending on genre.

Not all literary texts can be expected to exhibit large amounts of distinctive formal and semantic features. For one, genre restrictions on representational levels emerge from configurational patterns that reoccur across texts, but representational levels remain unconstrained if inter-textual diversity is too great to allow for generalizations. Besides, some authors are known for the utter sparsity of their style, or even for presenting stylistically entirely unadorned everyday texts as literature. A famous, and possibly the most extreme example for the latter are so-called ready-mades or “found poems”, which use preexisting texts from non-literary sources and contexts and present them as poetry, sometimes, but not always, by adding genre-specific formal features (Literariness I) such as verse-like line breaks (e.g., Handke 1969; Seely 2003). If readers assume that they are indeed literature—for instance, because they appear in a literary anthology or because the author has made their intentions explicit in a paratext—they will process them as such.

In summary, the first side of literariness is the mental representation of text features that are linked to literary genre schemata in the mind of the reader.

5.1.2. Literariness II: Reading Modes Associated with Literary Genres

In the historical sketch of previous approaches to literariness we have highlighted a school of thought that argued that a text becomes literary if and when the reader perceives it as literary (e.g., Fish 1973). In cognitive terms, this means that readers categorize texts as literary ones and adopt appropriate processing routines, interpretive strategies, and standards of evaluation. In short, they pursue literary reading modes. These reading modes constitute the second component of our concept of literariness. It becomes evident during text processing, as readers adapt their reading strategies according to the representational profile they expect a certain text to result in (Blohm et al. 2017, 2022; Culler 1975; Fish 1973; Gibbs et al. 1991; Knoop et al. 2024). For instance, if faced with a ready-made in a literary anthology, a reader is likely to base their categorization judgment on this contextual cue and to adopt a “poetic” reading mode (Literariness II) with observable changes in reading behavior, text memory, and text evaluation (e.g., Hanauer 1998; Zwaan 1993); line structure will be perceived and represented as poetic verse (Literariness I), whereas that same line structure might be seen as a mere result of column width in the context of a newspaper. Reading modes are often tied to genre expectations, but depending on literary competence and individual interests, they may also relate to authorial style or to expectations specific to a particular work of art if the reader has prior information about it. Similarly, a differentiation between fiction and non-fiction has been shown to affect reading modes (e.g., Altmann et al. 2014; Green and Brock 2000; Hanauer 2018; Schmitz et al. 2017).

In short, readers adopt certain assumptions about the texts they are about to read, and about their literariness. These assumptions shape the actual reading experience. This side of literariness is dependent on Literariness I in that the assumptions of readers hinge on prior reading experiences and hence on experiences with the representation of certain text features and how they fit into genre schemata. As the individual corpus of feature representations is broadened, the creation of schemata is likely to become more nuanced.

5.1.3. Literariness III: The Reading Experiences Literature Offers

Literary texts offer reading experiences that other texts do not (and vice versa). Professional emails, for instance, are rarely as musical as poetry or as suspenseful as good crime fiction nor do most of them deal with interesting things like murder, sex, travels in time and space, or alien invasions. These reading experiences afforded by literary texts (and rarely by others) are the third aspect of the act of reading that may be considered distinctly literary, in addition to characteristic representational profiles (Literariness I) and the appropriate reading modes (Literariness II). In narratology, the concept of “experientiality”, first introduced by Fludernik (1996; cf. Herman 2009; Schneider 2000), has become particularly prominent in describing these experiences. Originally, “experientiality” referred to the “implication or activation of a number of cognitive parameters, i.e., basic structures of human engagement with the world that straddle the divide between real-life experience and semiotic representations of experience” (Caracciolo 2014, p. 149). The notion has since been broadened to include a mixture of reading modes and subjective reading experiences with the aim of highlighting “the continuum between the textual representation of fictional (i.e., characters’) experiences and the creation of ‘story-driven’ experiences in narrative audiences” (Caracciolo 2014, p. 151). In addition, the concept has been applied not just to narrative texts, but also to genres traditionally featuring low levels of narrativity, like lyric poetry (cf., e.g., Alber 2002).

The present paper focuses on but one aspect in this vast field of cognitive engagement with literature before the backdrop of the reader’s past experiences (literary and otherwise): We claim that reading is experienced as literary if it results in characteristic representations and/or if it is carried out in a strategic and genre-specific manner (contributions of Literariness I + II), or if it offers possibilities not afforded by non-literary texts. Hence, Literariness III lies in how it feels to construct and ponder the text-as-representations.

As far as they are consciously experienced, representational profiles and reading modes contribute to Literariness III. As pointed out earlier, there are representations and processes that are more accessible and open to introspection than others. For instance, propositional networks are more accessible than grapheme/phoneme representations, and evaluation is more consciously performed than many processing steps. We assume, therefore, that the (subjective) meaning and the evaluation of a text are key determinants of the reader’s actual experience.

Moreover, we may encounter literary texts that are, for instance, so skillfully narrated, so fantastic, or so moving that they result in reading experiences that are hardly conceivable with non-literary genres. Additionally, since reading literary texts is not primarily information-driven (Vipond and Hunt 1984), readers are free of to engage in activities such as pursuing subjective interpretations or consciously evaluating narrated events for self-relevance. In this view, reading experiences can be said to be literary as far as they are characterized by affordances for kinds of experiences that are not offered by non-literary texts and genres.

Hence, Literariness III, which emerges in the actual reading experience, is a product of individual text understanding which is likely to be influenced by Literariness I and II. Reversely, if the actual reading experience contains elements novel to the reader, it may change their assumptions about literariness in general and about genre schemata in particular. We can therefore assume an interdependency between Literariness I and II and Literariness III. Elements that conform with genre schemata are likely to require lower processing efforts but are not necessarily more or less memorable than novel and surprising elements (e.g., Knoop et al. 2024).

As has become apparent in our discussion of the different aspects of literariness, we assume that the defaults of literary reading are determined by categorical decisions on the part of the reader: expectations of probable text features, appropriate processing routines, and expectable affordances. In this sense, it is the reader who “makes a given work a literary work” (Jakobson [1919] 1973, p. 11), in every individual act of reading; but their decision is based on their understanding of generalized text categories (e.g., Eagleton 1983, p. 10; Fish 1973). Simultaneously, multi-dimensionality and gradience in literariness arise from our distinction of Literariness I (text features), II (reading modes), and III (reading experiences), the proposed genre differentiation, as well as the inherent complexities of the assumed levels of representation.

5.2. Literarinesses as Genre Profiles

In the following, we will provide a tentative sketch of genre profiles based on existing empirical evidence.

Table 1. Representational profiles and genre-specific reading modes for news reports, novels, and poetry.

Text Type	Non-Literary	Literary
Genre	News Report	Novel	Poetry
representational profile
text-base complexity	low to medium	high	low to medium
weighting	situation model > text base > surface structure 1	situation models ≥ text base > surface structure 1	surface structure > situation models > text base 2,3
constraints
surface structure	conciseness		graphic layout; brevity; phonological patterning
text base	coherence; real-world compatibility	coherence
situation model	concreteness
reading modes
reading speed	> novels 1,4	<news reports 1,4 >poetry 5,6	<literary prose 5,6 <encyclopedia entries 2
	newspaper headlines are processed differently from ‘mere’ sentences 7 license otherwise unacceptable article omission 7,8

¹ e.g., Zwaan (1993). ² e.g., Hanauer (1998). ³ e.g., Tillmann and Dowling (2007). ⁴ e.g., Brysbaert (2019). ⁵ e.g., Fischer et al. 2003. ⁶ e.g., Blohm et al. (2022). ⁷ e.g., Schumacher and Avrutin (2011). ⁸ e.g., De Lange (2008).

specifies properties of the less subjective aspects of literariness, i.e., the representational profiles (Literariness I) and the genre-specific reading modes (Literariness II). We compare two prose genres, one literary (the novel) and one non-literary (the news report), to a lyrical genre (poetry).

The two prose genres are similar in terms of the relative weighting of representational levels and share coherence constraints on the level of the text base but are differentiated by a real-world-compatibility constraint on the level of the text base that is operative in the news report reading mode but absent in fictional prose (Hartung et al. 2017). In this view, fictionality is not a defining but a permissible feature of literary prose. The two literary genres differ in their representational profiles but have overlapping associated processing routines, which are characterized by reduced processing speed relative to non-literary genres and by better surface structure memory after reading (Hanauer 1998; Zwaan 1993). Some cells of the table remain empty; this reflects that empirical evidence is not (yet) available. As argued above, genres may be underspecified with respect to particular representational levels due to the diversity of the genre and the absence of clear-cut conventions.

This profile-based approach and the previous research that supports it represent but the starting point to more elaborate explorations of literary genres. Specifically, we believe that our understanding of literariness can be greatly advanced by a more systematic study of (1) the genre conceptions of readers (Knoop et al. 2016; Blohm and Knoop 2022); (2) literary reading modes (e.g., Blohm et al. 2022; Fokin et al. 2022; Zwaan 1994); and (3) contrastive studies of reading experiences associated with literary genres (e.g., Nenadić et al. 2019). Methodologically, experimental investigations of genre-specific text processing are of course particularly well-suited to inform an approach that centers on mental representations. Computational studies as practiced in the framework of the Digital Humanities are extremely useful to complement such an approach (cf. Jacobs et al. 2017), though in isolation they cannot inform us about the mental representations of actual readers. For instance, they allow the study of characteristic configurations of surface structure and text base elements, and thus to approximate the generalizations subsumed in literary genre profiles. Note that most experimental studies, too, disregard the truly subjective and idiosyncratic and instead seek to unveil general principles and effects that are stable across readers.

One theoretical aspect that requires further attention and elaboration is genre ambiguity, which can manifest in different ways: either the reader finds themselves unable to classify a genre, or the text belongs to a deliberately ambiguous (and lesser known) genre, such as the prose poem, or the text does not fully conform to any known genre, like Christoph Ransmayr’s free verse novel Der fliegende Berg (The Flying Mountain). We hypothesize that, in these cases, the representational profiles would be mixed versions of the above-mentioned genre profiles and reading modes. In the case of Ransmayr’s Der fliegende Berg, we would thus expect higher text-base complexity than in poetry, but an initial focus on surface structure that, over the course of the novel, might be replaced by an increased focus on situation models. The surface structure would be marked by the graphic layout of a poem, but the text base would be more coherent than expected of a poem. In terms of reading modes, we would expect the reading speed to decrease due to the verse structure. Non-literary texts that create literary effects might similarly elicit mixed representational profiles and reading modes. Depending on the extent to which they do, these mixed effects might occur locally; if they occur globally, however, this would likely change the reader’s categorization from “not literary” to “literary”.

Another area where the issue of literariness might be worth a second glance are digital-born texts. Since digital literature is a relatively young phenomenon, the likelihood that these texts will be perceived as “literary” may be a little lower among more conservative readerships; however, much digital literature (such as “insta-poetry” or hypertext fiction), builds on poetological precedents and frameworks that render it likely to be recognized as representatives of known literary forms. In this case, we would assume similar representational profiles and reading modes as for more traditional forms, though depending on genre-fluidity they, too, may be mixed, especially if active reader interaction is part of the layout, like in hypertext fiction. Post-digital forms, in which software, algorithms and coding play a role are, as yet, unchartered territory. We would assume, however, that representational profiles and reading modes would be largely influenced by the depth of engagement the reader is capable of: A technically less informed reader might view these types of information as purely illustrative in nature, similar to images in earlier texts. By contrast, for a reader who is able to read this information, we would assume that it would fundamentally affect aspects like text-base complexity, weighting, surface structure, text base, situation model, and reading speed. Yet for both types of readers, the categorization of a text as literary (or not) may be affected by their assumptions about whether literariness is restricted to verbal text proper or can be extended to other semiotic systems and their medial basis.

6. Conclusions

This paper builds and elaborates on earlier proposals arguing that literariness reflects aspects of the act of reading. We have argued for a crucial role of the genre schemata of readers, which emerge from multiple reading experiences and influence readerly expectations, their processing modes, and their standards of evaluating a text. We have detailed how traditional approaches to literariness can be reconciled by integrating Rosenblatt’s transactional view of the literary work and its idea of literary reading modes with the notion of stratified text representations. We aimed to show how literariness emerges during distinct stages of the act of reading and how its aspects are functionally related. Central to our argument was the idea that three aspects are central to the emergence of literariness: (1) the representation of text features on the reader’s mind, which are typically related to certain genres and which influence the reader’s formation of genre schemata and their classification of a text as literary or non-literary; (2) the reading modes this classification triggers, and which are evolved and adapted in interaction with the text-as-signs; and (3) the reader’s individual experiences, which further shape both the link between text features and genre schemata and the subsequently emerging reading modes. Contrary to previous approaches, we have argued that even seemingly trivial affordances of literary texts (e.g., the joy that lies in solving fictional mysteries, riddles, and crimes) may constitute literariness. All three aspects result in distinctly literary qualities in the act of reading. Finally, we have pointed out that the readers‘ conceptions of literature comprise distinct literary categories (e.g., genres) that constitute distinct “literarinesses“ whose default expectations we suggested to formalize as (genre) profiles. Based on the conception of literariness as a bag of three-sided coins that follows from the combination of distinct literary categories and the three aspects of literary reading, we have sketched ideas for future research.