The notion of text type is an abstract category designed to characterize the main structure of a particular text or one of its parts according to its dominant properties. It is intended to integrate common features of historically varying genres (novella, novel, short story, etc.) and thus to reduce the complexity of the many overlapping kinds of texts to distinct textual phenomena. In virtue of narratology’s traditional focus on time, these phenomena are semantic properties that constitute the temporal character of the text (passage). Thus, the text type ‘narrative’ is defined by the property ‘change of state’ of concrete objects and the text type ‘description’, accordingly, by the property ‘is about states’ of concrete objects. The text type ‘argument’ is defined by logical-semantic relations between abstract objects instead of temporal-semantic properties. There are many other typologies of text types, often including more types. But for the sake of consistency, the following account will be restricted to these three. One requirement of the notion is that the various text types be mutually exclusive.
The term ‘modes (or types) of discourse’, sometimes used synonymously with ‘text type’, could be restricted to the characterization of texts according to pragmatic properties (e.g. the speaker’s purpose). Thus any text may be used to persuade somebody. Its mode of discourse is then persuasive, even though the text type being used may vary (Virtanen 1992). The most appropriate text type in this case (or the text type most often used in connection with the purpose to persuade) may be the text type ‘argument’. But it need not be. The persuasive mode of discourse can be instantiated by any text type, depending on pragmatic concerns. The notion ‘mode of discourse’ is thus context-sensitive; that of ‘text type’ is not.
Another category that is closely related to the notion of text type is ‘genre’. However, text type and genre should be kept strictly apart from each other as well. Unlike the numerous historically generated subclasses of genre (such as novel, sonnet, recipe, homepage) that have evolved by chance, typologies of text type include a limited number of different items and aim at a complete set of all possible types that can make up any text. Moreover, in contrast to genre, whose members are, by definition, entire texts, single text types mainly refer to parts of texts depending on whether the passage exhibits the semantic profile in question or not. As a rule, the definition of text types is based on text-internal data whereas definitions of (non-literary) genres follow various text-external and text-internal criteria alike (consider the letter and its many subclasses).
One important consequence that follows from this definition is that narrative as a genre is distinguished from the text type ‘narrative’. The text type ‘narrative’ derives from the prevailing quality of texts considered to be prototypical for the genre narrative or fiction, members of which are often not pure narratives in the sense of text type. While any text that is called, say, a novel belongs to the genre narrative, probably no novel is contains only the text type ‘narrative’. Usually, novels exhibit all text types. However, any experimental literary text that is called a novel belongs to the genre narrative, even if it is mainly characterized by the text type ‘description’. The problem of equivocation (one term denoting different notions) occurs in every case. This can be avoided when another term is available: thus the term ‘ekphrasis’ denotes a descriptive genre whereas ‘description’ denotes the text type usually dominating ekphrasis. Yet ‘description’ is by no means restricted to this latter use, and the term ‘ekphrasis’ mainly refers to literary descriptions depicting pieces of visual art (Henkel 1997; Klarer 2005).
There are many varying classifications and typologies, each including different types (Georgakopoulou 2005). The text types ‘description’ and ‘narrative’, though, seem to be part of almost all typologies (except for Fludernik 2000; see below, § 3.2). For example, in addition to those mentioned in the definition above, exposition and instruction are discussed as text types by Werlich (1976), van Dijk (1980) adds scientific inquiries to the list of text types, and de Beaugrande and Dressler (1981) include didactic texts. Sometimes, the notion of text type is meant to characterize entire texts, sometimes not; some authors focus on semantic features, others on pragmatic features. Heinemann’s (2000b) survey of the notion of text type in linguistics shows that the linguistic typologies of texts follow the application of different criteria: grammatical properties of texts, semantic properties of texts, situational context, function, etc. This practice has brought about a huge variety of heterogeneous concepts. There is no agreement on which notion should satisfy which criteria. And, what is more, even the use of particular terms is not regular. Thus linguists often use our term in the sense of what above was called a genre. For an extremely fine-grained classification with hundreds of genres, termed “text types,” see Görlach (2004: 23–88).
Linguistic research often aims at a classificatory system of its categories. An exception is Virtanen’s (1992, 2010) two-fold model of discourse types and text types, conceiving of types as functional categories. The definition above indicates that text types in the sense suggested here are not classifying but functional or comparative concepts, too (for this distinction, see Carnap 1950: 8; Strube 1993: 59–65). Classifying concepts are applied according to “either–or” decisions (a proposition is either true or not true); the application of a comparative concept, by contrast, depends on to what extent it is appropriate. Most typologies disregard the difference between classifying and comparative concepts. This leads to the construction of hierarchies of types and subtypes that correspond to each other in the manner of species and subspecies.
The notion of genre is a classifying concept: either the text t belongs to the genre G or not. In this vein, the notion of text type classifies short stretches of text according to the temporal meaning of the predicate used in the passage. However, if applied to a text passage exceeding the scope of a simple clause, the notion of text type is a comparative concept because it is meant to characterize the temporal-semantic profile of a text passage according to its dominant temporal-semantic properties. The attribute ‘dominant’ means that in every text there are other temporal-semantic properties of lesser intensity than the dominant property. Thus a text (or passage) is not either descriptive or not but more or less descriptive, depending on the extent to which it exhibits descriptive markers (e.g. verbs that refer to states). A text is of a particular text type to the extent it displays those properties that determine this text type.
The underlying reason for this conceptual ambiguity is that text types are defined with regard to simple clauses that prototypically exhibit the respective properties supposed to dominate the entire text. In the reality of texts, however, the various properties that determine the different text types can be instantiated even by a mere sentence. The difficulty arises because the notion of text type, meant to characterize aspects on the level of texts, is defined in terms that are derived from the level of sentences.
As mentioned above, the comparative concept of text type must be distinguished from the classifying notion of genre. While genres single out entire texts according to heterogeneous features (e.g. formal features in the case of sonnets and paratextual information in the case of homepages), text types try to capture semantic relations between textual surface structures (of items of discours) and content structures (of items of histoire). In this sense, the division of text types follows a semantic criterion. Therefore, single text types are to be distinguished from each other by identifying not only what kind of linguistic device is used—event verbs or static verbs (for a more fine-grained verb classification, see Vendler 1957)—but also by what a text is about: time-bound events, states, or timeless universals (abstract objects, relations, etc.). The reason is that words are not always used according to their primary meaning. For example, a particular text passage may predominantly consist of static verbs that refer to states only prima facie but indirectly or figuratively express a change of state. As a consequence, the notion of text type has another component that has not been mentioned so far. It depends not only on the extent to which a text is characterized by a particular text type according to the number of words determining this text type, but also on the degree to which these words express their direct meaning. For example, a text is more descriptive the more its meaning corresponds to the direct meaning of its static verbs; conversely, the extent to which a text is descriptive curtails the function of static verbs to indicate a change of state. This holds for elliptic contexts, for example, where a description using static verbs refers to a change of state without directly naming it. Take the static verb ‘to stand’: this verb refers to a state, but if it is used in the context of, say, a murder (‘after stabbing her victim, the murderer raised her head. Her daughter stood in the door’), the state of standing in the door functions as (part of) an event. (Even the murder could be expressed by static verbs.) Another example is the iterative use of event verbs (discours component) that makes a passage descriptive from the point of histoire.
Text types are thus determined quantitatively as well as qualitatively. First, all texts, not only narratives, usually do not deal with only one kind of object. Even a single sentence may contain lexical items with narrative as well as descriptive meanings. Second, these linguistic devices may not always directly correspond to the object they actually refer to in a given context, but may have an additional meaning that is sometimes even opposed to its direct meaning with regard to its temporal-semantic features. For this reason, text types supervene on texts, and this is why the ascription of text types to texts is seldom unambiguous. Text types are not only bottom-up abstractions of texts but also top-down structures that have an impact on the meaning of the text, mainly with regard to the temporal characteristics of its contents/meaning in relation to the lexical devices. Thus narrative is a type of text that predominantly represents events featuring event verbs; description is a type of text that predominantly represents states (of objects) featuring static verbs; and argument is a type of text that predominantly represents omnitemporal and logical relations (primarily between abstract objects such as concepts), i.e. universal propositions concerning objects outside time. These three text types do not cover all forms of texts, however since, for a starter, they may be ascribed to assertive sentences. Features of other text levels (e.g. grammatical mode or dialogue) are not encompassed by this typology. Among others, explanation is one more candidate for text type (Herman 2008) as it focuses on the causal characteristics of the represented events. No system of text types has been generally accepted so far.
Text types makes it possible to link textual cues to interpretive ascriptions. Thus they may help to answer questions as to how events are represented in a particular passage: by naming the event directly or by naming it indirectly through the depiction of related states that only hint at the event. Interpretation sometimes benefits from the disadvantage of ambiguity. Thus the two components that determine a text type—the level of conventionalized meaning according to the dictionary and the level of meaning actualized in a given text—can correspond just as they may be inconsistent with each other. Other questions with a wider interpretive focus might bear on why certain passages of a text are composed narratively but others descriptively and what this means for different interpretive purposes such as the work’s aesthetic form or the reception of the text.
The present account primarily relies on research in literary studies and narratology. As a linguistic notion, text types are rooted in the development of text linguistics starting in the 1960s. The preferred objects of text linguistics are functional texts, to the effect that literary texts are often excluded from linguistic research. (For the wide range of competing concepts in linguistics, see Heinemann 2000a, 2000b, and Schlüter 2001: 67–144; for an overview focusing on literary studies, see Dammann 2000). While the term “text type” came in the wake of this linguistic tradition, the notion is much older. It existed under the disguise of terms found in various languages and was not terminologically differentiated.
The first differentiation of notions that share features with the modern concepts of genre and text type can be found in Plato and Aristotle. Plato divides what he calls diegesis into three kinds (Republic, III, 392c ff.). A text of the first kind is directly ascribed to the author, a text of the second kind to a character, and a text of the third kind is mixed (Halliwell → Diegesis – Mimesis). The underlying criterion of the typology is the answer to the question “Who speaks?” The notions are obviously classifiers and can be linked to genres such as drama. However, the third kind—“mixed”—is different from the other two in that it is a hybrid, containing the features of both. Were this conception pursued further, the first two kinds would function within the third kind as text types in our sense.
Plato’s (and Aristotle’s) legacy is a long and complicated one (see Trappen 2001). An influential reformulation of the division of literary genres is suggested by Goethe ( 1994: 206): “There are only three natural forms of poetry: The clearly telling, the enthusiastically excited, and the personally acting: Epic, Lyric, and Drama. These three modes can work together or separately.” His understanding of the “three natural forms of poetry” is similar to that of text types in several aspects.
Referring to a number of literary genres (termed “Dichtarten”) such as drama, elegy, novel, parody, and satire, Goethe assumes that all genres can be reduced to those three forms which he conceives of as a kind of deep structures (“wesentliche Formen”) that may be observed in every literary work, independently of its genre. Every single work is characterized by a particular compound of the three forms. In the present context it is important to note that Goethe does not attach the three forms to groups of texts, thereby dismissing the idea of classifying texts. Instead, he suggests that the forms usually occur together (although he refers to Homer’s epics as examples of pure Epic).
As in the case of text types, Goethe’s notions are not meant to classify texts but to characterize the special shape of individual texts. Like text types, Epic, Lyric, and Drama are considered to be historically stable notions that refer to gradual properties. Furthermore, these terms are apparently derived from the surface structures of texts, i.e. from historically varying genres that Goethe considered paradigmatic (such as Homer’s epics). At the same time, they are supposed to refer to deep structures that can be observed in texts of all genres. However, while Goethe considers only works of literary art, text types are neutral to the question of literariness.
Goethe’s conception remained influential up to the middle of the 20th century. Staiger (1946) develops a threefold typology of literary modes on the basis of Goethe’s original conception. The genre triad is replaced by the comparative/functional categories in the sense of ‘natural forms’ even though Staiger implicitly retains the classifying genre notions. He refers to them using nouns and to the comparative categories using adjectives. On this basis, Staiger ( 1957: 112) concludes that Kleist’s narratives, being novellas, are epics in the generic/classifying sense, and dramatic in the comparative/functional sense, thus blurring the conceptual shift between the two categories. While the classifying notions, according to Staiger (1946: 8), refer to patterns that underlie constant change and innovation, the comparative notions refer to the “tonality” or mode of texts and are supposedly invariant. Staiger ties the comparative categories to general anthropological attitudes such that lyric texts allegedly reveal the lyric dimension of human nature.
Although the terms ‘natural forms’ and ‘text types’ as well as their scopes are different, and although the criteria by which the notions are determined diverge widely, Goethe’s idea of modifying the traditional generic notions in a comparative/functional sense continues to be felt in the modern concept of text type.
A related concept in recent/modern literary studies is Hempfer’s (1973) ways of writing (“Schreibweisen”). His approach is worth mentioning because he distinguishes what Staiger and his followers amalgamate. Ways of writing are meant to be historically stable notions capturing deep structures while genres (“Gattungen”) are historically varying conventions. Hempfer’s examples of ways of writing are the narrative (“das Narrative”), the dramatic (“das Dramatische”), the satirical (“das Satirische”), etc. The transhistorical deep structures of the narrative and the dramatic are determined by the communicative situation. The deep structures of satire, by contrast, are determined by different properties. While Hempfer’s distinction of the narrative and the dramatic goes back to Plato, his notions are different from the point of view of concept structure. Again, text types and ways of writing have in common that they do not classify texts but characterize them. However, in contrast to text types, Hempfer’s ways of writing, like Goethe’s and Staiger’s, are obviously derived from generic notions and lack a unified criterion. Hempfer’s structural approach is refined and naturalized by Zymner (2003) who conceives of ways of writing as dispositions that can have certain effects.
What all these conceptual variants, including text types, have in common is that they are categories aimed at characterizing aspects of style. They differ from one another in that they focus on different aspects. While text types, as defined here, refer to temporal structures, Goethe’s natural forms and Staiger’s styles evoke traits of human nature; as for Hempfer’s ways of writing, they are rooted in, among other things, communicative situations.
The first conception of text type in the 20th century appears to have originated from Russian formalism. In an article on Pushkin’s “The Shot,” Petrovskij ( 2009) analyzed the story’s narrative dynamics and characterized various sections of the text according to their narrative (temporal) and descriptive (static) structures. Similarly, Trubeckoj ( 1980) used this distinction to characterize an Old Russian travelogue by Afanasij Nikitin. However, these early examples have not had any influence on recent developments.
Starting with Genette, the narratological discussion of text types has long been devoted to the relation between the text types ‘narrative’ and ‘description’. Genette’s assertion that “description is naturally […] the handmaiden of the narration” ( 1976: 6) caused much ado in narratology (e.g. Klaus 1982; Chatman 1990: 6–37; Ronen 1997). One of the main reasons was the lack of distinction between narrative as a generic or classifying concept covering kinds of texts and narrative as a typological, comparative concept designed to capture deep structures, functions, etc. Although he does not mention it explicitly and sometimes blurs the difference himself, Genette ( 1976) seems to think of narrative as a generic concept, and his conclusion as to what description is in relation to narrative thus entails that description is not on the same level as narrative in the generic sense. Thus description for Genette is something like a text type while narrative is both a sort of literary genre (in fact, “the only mode that literature knows” ) and a text type (“the narration properly speaking”): “Every narrative includes two types of representation, although they are blended together and always in varying proportions: representations of actions and events, which constitute the narration properly speaking, and representations of objects or people, which make up the act of what we today call ‘description’.” (5) Here, “narration” and “description” evidently denote text types. However, Genette does not aim at a theory of these typological notions but looks at whether the notion of description is suitable to limiting the generic scope of narrative (in his opinion, not at all).
In the wake of Genette’s article narratologists and other literary scholars began to investigate description and its manifestations and functions in literary discourse (e.g. Hamon  1982, 1993). For Hamon, description is a way to insert knowledge in narratives. He is less interested in a theory of text types than in literary devices that implement knowledge in narratives. Hamon’s model for descriptions is an encyclopedia entry which literary descriptions expand in many ways. Adam (1992) develops Hamon’s approach further and combines it with text linguistics. Much of what is published on the subject is devoted to differentiating description and its functions (e.g. Lodge 1977; Gelley 1979; Beaujour 1981; Kittay 1981; Mosher 1991). While Hamon investigated the role of description in 19th-century prose, Scherpe (1996) and Pflugmacher (2007) focus on the practice of description in modern writing. Ibsch (1982) contrasts examples of the two periods. A recent study that uses other text types to explain a peculiarity of a literary text is Abbott (2011).
One of the major contributions to the theory of text types is made by Chatman (1990). He considers not only fiction but also film in terms of a theory of text types. His first step is to modify the meaning of the term ‘text’ according to the premise that narrative is a structure existing independently of the medium. Chatman’s notion of text is not restricted to pieces of spoken or written discourse but comprises “any communication that temporally controls its reception by the audience.” (1990: 7, original emphasis) For Chatman, films and literary works are both texts. His criterion, which has its roots in Lessing ( 1984), is that works of literature and films, contrary to pictures, are deployed in time.
Chatman defines three text types: narrative, description, and argument. However, he does not distinguish these types from each other by using one and the same criterion. The text type ‘description’ is said to “render the properties of things” with the subject as criterion while the text type ‘argument’ is defined on the basis of the communicative goal that a producer of a text pursues, namely, “to persuade an audience” (9). Furthermore, “Argument is the text-type that relies on ‘logic,’ at least in the informal sense.” (10) Finally, Chatman explicitly conceives of text types as macro genres to which the known literary genres are subordinated. “[…] Westerns are generic subclasses of the Narrative text-type. A Theophrastian character is a subclass of Description. A sermon is a subclass of Argument” (ibid.). He clearly considers text types to be generic categories. Implicitly, however, his conception of text types aims at a functional, not a generic account.
The problems of definition aside, Chatman’s idea is that “text-types routinely operate at each other’s service.” (ibid., original emphasis) The principle underlying this idea is that many texts display another structure on the textual surface than they do on a deeper level. Thus a fable is a narrative on the surface level. However, it is not only a narrative but essentially something more because it includes a moral and as such displays the underlying text type ‘argument’ at the service of which the narrative structure at the surface operates.
It is clear, then, that Chatman investigates the functional relations between generic ascriptions and ascriptions of underlying meaning structures conceived of as text types. His is a fruitful approach in that it helps to map interpretive ascriptions of meaning onto a system of interrelated notions that can be used to lay bare semantic properties of texts. More implicitly than explicitly, Chatman shows that the notion of text type in the narratological sense is a Janus-faced thing. On the one hand, text types capture semantic properties of texts according to lexical distribution (level of single words); on the other hand, they are meant to link the results of this procedure to the overall meaning of the text. This is what Chatman seems to have in mind when he models text types on two levels. Sometimes the two levels correspond to each other (a descriptive wording resulting in a description without any other function), and sometimes they do not (descriptive wording resulting in a narrative: in Chatman’s terms, the text type description “operating at the service of” narrative). Similarly, Genette ( 1976: 7) considers Robbe-Grillet’s nouveau roman “an effort to constitute a narrative (a story) by the almost exclusive means of description.”
In an alternative approach, Fludernik (2000) suggests a threefold system intended to cover not only literary but also conversational/oral discourse. She distinguishes three levels: macro genre, genre, and discourse mode. Her macro genres prima facie correspond to the notion of text type in three ways. They are intended to systematically cover all texts; they are mutually exclusive; and they are derived from linguistic text typology. Fludernik adds three types and excludes description from her list: in addition to narrative and argumentative, she enumerates instructive, conversational, and reflective macro genres. She criticizes the assumption of the text type description “as a general text type, since description is very rarely a unitype text type, i.e., there are extremely few purely descriptive texts around” (2000: 280).
Aside from the question as to whether there are any texts at all that display pure macro genres, Fludernik obviously conceives of macro genres as classifying concepts. Texts either belong to one of the macro genres or not. Hence she assigns particular genres of the second level to the macro genres: novel, drama, and film are subordinated to narrative, scientific text and historiography to argumentative text, and philosophy to the reflective macro genre, to name but three of them. She also refers to this second level as “text types,” a term not to be confused with the notion of text type as explicated above. Neither this nor the first level notion has anything to do with text types in our sense.
On the third level, “discourse modes” are determined by “the surface structure of texts and the specific functional correlates within specific genres” (281). Thus, for instance, exhortations (a discourse mode) are subordinated to the genre of sermons which, in turn, are subordinated to the macro genre of instructive texts. It is the notion of discourse mode that corresponds to the notion of text type. This third level comes in response to the need for a non-generic category that is more flexible in order to capture the manifold typological heterogeneities within one and the same text, let alone within one and the same macro genre. Conversely, this notion explains the fact that different macro genres may display one and the same discourse mode; for instance, a narrative, understood as macro genre, may contain evaluative clauses typical for the argumentative macro genre, while argumentative texts may contain event phrases typical for the narrative macro genre as an illustration of an argument. The similarity between Fludernik’s discourse modes and text types is that both are comparative notions. Also, discourse modes refer to passages of texts instead of entire texts and exhibit a functional relation. However, they are derived from heterogeneous textual phenomena and are not mutually exclusive. Fludernik’s enumeration of discourse modes is an open list (of thirteen items) that lacks a unified criterion. Her main interest concerns not discourse modes but macro genres.
The main problems with the notion of text type are that there are so many competing approaches and terms with similar but not identical meanings and that text types, as presented here, are not expressed by a stable term. Basically, the theory of text types suffers from general disagreement as to what text types are. Arguments are drawn to advancing ever more suggestions for putative text types that have yet to be considered. As a result, no explication of text types has been achieved to cover all aspects of the notion. What is required instead is that text types should be explicated with regard to the purposes they are supposed to serve. Are text types meant to describe the particular dynamics of a text, its profile according to the represented temporal structure of the histoire, of the discours, or of both?
Another problem is the discrepancy between the definition/explication of the concept and its application. Text types as explicated here are derived from the properties of sentences. In what kind of relation to textual properties do sentence properties stand? Even if text grammars have studied this question in quite some detail, agreement has not been achieved. Are concepts capturing those sentence properties appropriate means to capture properties of texts? In other words: what exactly does it mean when properties supervene on texts?