Genres, Text Types, or Discourse Modes? Narrative Modalities and Generic Categorization.
I would like to return to the problem of narrative's variegated textual surface structure, picking up where I left this issue of generic classification and text types in Towards a 'Natural' Narratology (1996). In a very brief section of chapter 8 of that book (section 8.4, esp. 356-58), I had proposed a revision and extension of Chatman's triad which I modelled on textlinguistic work found in Longacre's The Grammar of Discourse. I would now like to expand this proposal even further, linking it more comprehensively with the structure of natural narratology. In particular, I wish to discuss some of the theoretical implications of a text-type approach to the definition of narrative. I will start by introducing a few models from text linguistics, especially the model of Virtanen and Warvik with which I was not familiar when writing Towards a 'Natural' Narratology.
1. Text Types
Linguists have realized for some time that textual surface structures display a wide spectrum of forms that vary with the respective type of discourse. Since text linguistics, unlike literary scholarship, does not focus primarily on literary or even on written texts, linguists have had to develop a great number of concepts to account for variety in language use (register e.g.) or for the use of language in specific situations (e.g. telephone conversations; natural narrative; doctor-patient discourse; instruction manuals; cookbooks, etc.). The term "text type" in text linguistics refers to a number of quite distinct phenomena on a variety of different levels. In "Text-Type as a Linguistic Unit," for instance, Esser defines text type as "language variation according to use as opposed to variation according to user" (142).  He distinguishes between extensional definitions (text types as genres); definitions based on external criteria of production; on structurally defined schemata or superstructures (cf. van Dijk); and definitions deriving from "abstracted corpus norms" established by means of statistical analysis (e.g., in the work of Biber). None of these definitions concern the sentence or paragraph level, although the term "text type" is frequently deployed with that reference, too. Whereas literary scholars naturally focus on literary genres (synchronically and diachronically) and subdivide the major genres -- lyric, epic (fiction), and drama -- into ever more specific types (the novel, the detective novel; metaphysical poetry, the sonnet), text linguists have been concerned with finding equivalents to the generic distinctions between the lyric/epic/drama in their mostly non-literary corpora, and have been subdividing their "big" text types into ever more specific "types of texts" (Textsorten).
A good example of such a typology is provided by Kinneavy in A Theory of Discourse (1971). Kinneavy distinguishes between four "aims of discourse" (61), modelled on Jakobson's communicative functions ("Closing Statement"), i.e. the expressive, referential, literary and persuasive text types (as we would say). The category "literary aims of discourse" focusses on the signal (Jakobson's "poetic function") and includes the joke, the movie, the TV show besides drama, ballads, the lyric, the short story, and the like. Kinneavy's expressive category splits into (a) individual and (b) social types, including under (a) conversation, journals, and prayers, and under (b) manifestos, contracts, myths, as well as religious credos. The referential aim of discourse encompasses exploratory texts (dialogues, seminars); scientific texts ("proving a point by arguing from accepted premises" or "by generalizing from particulars"--it therefore best fits what might be called argumentation); and informative texts (news articles, t extbooks). The persuasive category includes religious sermons, editorials, and political or legal oratory.
One can notice immediately how these aims of discourse open up more questions than they answer. For one, written and oral modes are mentioned indiscriminately; very specific situations (the seminar discussion) come to stand beside very general and voluminous textual corpora (legal oratory, scientific discourse). Some prominent text types are missing from the list, e.g., instruction manuals or guidebooks: are these persuasive? Also, why is an editorial a persuasive text type and not, rather, expository or expressive? In other words, Kinneavy's courageous attempt at a typology, which he establishes by aligning a number of textual categories with a common purpose, despite its obvious merits fails to see that most texts of a given "genre" are referential as well as persuasive or expressive. Real texts combine a number of discourse aims with a variety of local realizations that may deviate from these overall aims or straddle and integrate them with each other.
What is important about Kinneavy's model, though, is the basic distinction between a global text type on the one hand (his referential, expressive, persuasive, and literary aims of discourse), and, on the other, its specific realization in an empirical text belonging to a genre (something that linguists also call a text type and that Egon Werlich calls a text form [Text Grammar 46]). Werlich's typology (Typologie der Texte; A Text Grammar of English) distinguishes between five text types--description, narration, exposition, argumentation, and instruction. These five text types are conceived as "an idealized norm of distinctive text structuring which serves as a deep structural matrix of rules and elements for the encoder" (Text Grammar 39). Werlich's model distinguishes between three levels--the ideal type (which is of a global prototypical nature and exists only as an abstract matrix in language users' minds); the text form, i.e., the specific text type (e.g., the self-help manual); and the real manifest te xt as language--sentences, paragraphs, chapters--that includes realizations of the abstract orders of text type and text form. Werlich calls this third level the text idiom.
Werlich's typology makes it easy to analyze texts on a sentence-by-sentence level. It does raise some questions, however, regarding the ideal types. As with Chatman's triad of descriptive, argumentative, and narrative, the category description falls somewhat flat on the level of Werlich's ideal text type since it has few realizations on the text form (i.e., genre) level. Descriptions are a regular constituent of most texts, but texts that are exclusively descriptive in character do not occur with great frequency. In fact, one can argue that descriptive passages serve a crucial expository function within narrative, in procedural (instructional) and in argumentative texts since they define and introduce the phenomena with which the reader then engages narratively, procedurally, or argumentatively. It is no coincidence that the sentence Werlich uses to exemplify exposition (One part of the brain is the cortex or rind) can easily be part of a description of the brain (Text Grammar 40). Indeed, it is hard to see how description could work without conceptual clarification, or the other way round.
Although these conceptual drawbacks persist in all models (including my own), what is extremely important in Werlich's proposals is the alignment of text idiom with textual function. Werlich does not spell this out, but it is implicit in his remarks. Whereas text types serve a communicative function (one deploys a narrative text type in order to present phenomena in time), empirical linguistic surface-structure sentences serve a function within this specific discourse. One can thus begin to comprehend how a descriptive passage in a narrative text sets the scene for imminent events, or how argumentation by the narrator evaluates the action for the benefit of the narratee.
I now turn to the excellent paper by Tuija Virtanen and Brita Warvik, "Observations sur les types de texte." Their typology integrates the models of Kinneavy, Werlich, Longacre (The Grammar of Discourse) and Jean-Michel Adam ("Quels types de textes?") and presents a very complex multi-level structure. Briefly, Virtanen and Warvik adopt from Longacre the emphasis on oral discourse types, and from Adam the idea of including criteria from speech act theory.
Longacre had distinguished four text types (based on the combinations of two binary oppositions: texts with or without temporal succession and texts with or without an "agent orientation"). His four categories are narration, procedural discourse, behavioral discourse, and expository discourse. Longacre thus eliminates description as a global text type, and he introduces a category of the behavioral that relates to speech act-oriented discourses (the prayer, political speeches). Thus, in Longacre's schema, whereas argument is not a text type but subtends expository (scientific) and behavioral (persuasive) discourses, the concept of the procedural can subsume description and exposition in the instructional function that underlies this type. The major difference between Longacre and Werlich lies in Longacre's concentration on oral modes. Moreover, Longacre is more interested in the ideal text type than in the sentence-by-sentence categorization of textual surface structure.
Adam expands Werlich's typology to propose eight text types. He is not interested in surface-structure phenomena but in global or deep-structure qualities of text. Adam's eight text types are:
- le type textuel narratif -- speech act: assertion
- le type textuel descriptif -- speech act: assertion
- le type textuel explicatif -- speech act: assertion
- le type textuel argumentatif -- speech act: convince
- le type textuel injonctif -- speech act: directive
- le type textuel predictif -- speech act: prophesy
- le type textuel conversationnel -- speech acts: question; excuse; promise; genres: interview; dialogue
- le type textuel rhetorique -- genres: poetry; songs; proverbs; graffiti
As we will see, this schema anticipates my argument for a conversational text type and a reflective text type, Adam's "rhetorical" text type. Adam is quite original in aligning text types with speech acts, but as his schema demonstrates, such an alignment does not yield a well-structured symmetry, for while three text types correlate with assertive speech acts alone, other text types conjoin several different speech acts. The positing of prediction as a separate text type also seems somewhat de trop: after all, this text type cannot claim a large number of texts and genres as its province. This criticism, however, is not meant to derogate the model; what it attempts to point out is that a really convincing list of text types is extremely difficult to come by and that such listing is fraught with a theoretical minefield (speech acts versus discourse functions versus cognitive parameters as constitutive categories of the typology).
Virtanen and Warvik's model (cf. figure 1) extends the three levels of Werlich's schema. Moreover, the authors proffer important theoretical discussions of the (non-)adequation between text types and surface-structure phenomena. Virtanen and Warvik first of all propose the existence of an overall level of cognitive processes, thereby extracting from Werlich's text types those features that, from a cognitive point of view, allow us to establish categorically what our discourses are about from a conceptual point of view.  Virtanen and Warvik then propose a second level relating to the functions of discourse (Kinneavy's criterion) that enables encoders then to pick a discourse type ("types de discours"), i.e., what Werlich and others call a text type. Virtanen and Warvik do not list any of these discourse types in a comprehensive manner, but in their discussion refer to an argumentative discourse type and to a narrative discourse type. In Virtanen's more explicit "Issues of Text Typology" (1992), the list in cludes narrative, description, instruction, exposition, and argument.
The next level is taken up by text types ("types de textes"). These are subdivided in accordance with multiple parameters (spoken-oral; formal-informal; fictional-nonfictional; monologic-dialogic). These correspond to the genres in my model or to Werlich's text forms. On yet another level, Virtanen and Warvik introduce textual strategies that allow one to organize the text: "Sur un plan plus concret, on choisit une strategie textuelle pour organiser le texte" (107). ["On a more specific level, one chooses a textual strategy with the purpose of organizing the text."] These levels are further extended top-down to include grammatical choice and, finally, the actual text itself, the "texte actualisee."
It is not quite clear from Virtanen and Warvik's essay whether their "strategies textuelles" concern the choice of text idioms la Werlich, i.e., sentences of narration, description, and the like. In "Observations sur les types de texte," the example provided to note the fact that most texts are not "unitypal" suggests that this is indeed the case:
Pour donner un example, nous pouvons partir du processus cognitif du jugement et notre but sera alors de convaincre; donc nous choisissons le type de discours argumentatif. Mais ce type n'est pas necessairement actualise par l'argumentation; nous pouvons tres bien choisir, par exemple, la narration et y ajouter un peu de description et un morceau d'evaluation. (107-08) ["To give an example, we can take the cognitive process of judgment as a starting point, and our goal will then be to convince someone. Hence, we choose the argumentative discourse type. However, this discourse type is not necessarily actualized by means of argumentation; we could just as easily choose narration and add a little bit of description and a bit of evaluative commentary."]
But in Virtanen's "Issues of Text Typology" these levels are reduced to two levels, discourse types and text types, which are constituted by discourse function and text strategy, respectively. In this model, the generic types again disappear, and what is referred to as text types are argumentative or descriptive sentences.
In "Issues of Text Typology," Virtanen additionally discusses the combinability between text types and discourse types for a number of texts. Thus, the narrative text type, she argues, can be used in any discourse type, whereas argumentation mostly serves to constitute the argumentative discourse type. The model also stipulates that narrative is least able to occur without the actualization of the narrative text type, whereas argument can take any shape on the text-type level (see figure 3 in "Issues of Text Typology"). This schema therefore posits a symmetry between discourse types and text types: there is for each discourse type one prototypical text type, and the other way round, even if in the realization of ideal types in empirical texts no neat correlation can be observed.
So far, the linguistic models have attempted to place narrative within a broad spectrum of other text types. The most important insights from these contributions can be summarized as follows: (1) One has to distinguish between global text types that are defined as ideal text types, on the one hand, and realizations of these text types on the linguistic surface structure, on the other. (2) On the linguistic surface structure a combination of discourse types is found to interact. Their choice depends on the discourse strategies (Virtanen and Warvik) that speakers or writers deploy. (3) Generic expectations have a great influence on the constitution of text type and on the choice of discourse strategies. (4) Form (i.e., type of text type or discourse type) and function (the specific discourse purpose to be achieved) must be distinguished. A one-to-one relationship between form and function can not necessarily be assumed to exist.
Bearing these insights in mind, let me now turn to my own proposals as discussed in Towards a Natural' Narratology.
2. Macrogeneric Text Type, Genre and Discourse Mode: A Narratological Model
In the wake of Helmut Bonheim, both Seymour Chatman, in Coming to Terms, and Jon Adams, in Narrative Explanation, have distinguished three narratologically relevant text or discourse types: narrative, description, and argument (Chatman); narrative, description, and exposition (Adams). These three categories quite obviously reflect on elements observable in narrative discourse: the narrator's diegetic narration (exclusive of the representation of characters' discourse) clearly conjoins sentences that refer to plot, sentences that descriptively set the scene, and sentences that are argumentative (evaluative, reader-addressed, presenting an argument about the fictional world or a commentary by the narrator on it). Delimiting these three types of sentences may be problematic, however. Thus, exposition may be evaluative and descriptive and even refer to events. Even more problematic is the exclusion of dialogue (and thought representation) from the list since dialogue often takes up a large proportion of the surfa ce text, and these models pretend to work on the basis of a bottom-up methodology. Perhaps most suspect of all is that both models posit the existence of 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.
In Towards a 'Natural' Narratology, I therefore proposed (a) to distinguish between three levels of text types and (b) to start out from a functional approach that takes the oral language as its primary model. My presupposition is that the category of genre also pertains in nonliterary discourse (cf. Weinrich 1972).  My model has many similarities to Virtanen and Warvik's schema. The three levels I proposed were:
(1) The level of the macro-genre, Virtanen and Warvik's discourse type. This level is constituted by the functions of communication.
(2) The level of genre. Traditional genre expectations are operative here.
(3) The level of discourse mode on the surface level of texts. On this level, the function, for instance of an argumentative or descriptive passage, within the schema of the specific genre is at issue.
If one starts out from the oral language, a number of additional macro-genres suggest themselves for consideration. Extending what I said in Towards a 'Natural' Narratology, one could posit the macro-genre conversation besides narrative, argument, and instruction, and one might even want to add a metalinguistic category (the reflective text type). Before arguing for this particular set of macro-genres, let me first present the rest of the schema. On the specific level of empirical texts, these macro-genres are realized as textual genres (written language) and text types (Textsorten; oral language). The diversity of these genres arises from the multiplicity of their contexts of use; the categorization into genres within a macro-genre is effected on the basis of the discourse functions constitutive of the macro-genres.
On the third level, which I see as entirely distinct from the two generic levels, the surface structure of texts and the specific functional correlates within specific genres or text types determine the discourse modes that I have enumerated in figure 2. For instance, description within narrative serves an orientational function (exposition in the sense of providing background information for the better reliefing of narrative foreground), but within scientific prose, for example, description may be part of an exposition or part of a directive sequence in a guide book. Likewise, what is usually considered to be narrative report (the presentation of event sequences) corresponds to the argumentative backbone in historiography and to the procedural core of instructional discourse. Some discourse modes, like the oral mode of phatic expressions (mmh, right, good, whatever), do not have equivalents in other text types because they are specific to conversational exchange. The list of discourse modes is therefore ope n, and individual discourse modes perform entirely different functions in their various generic contexts.
Before going into the model in more detail, let me start by emphasizing that the proposal I am presenting for consideration is not meant to be the last word on this issue of text types and related concepts. As a literary scholar with a great deal of interest in linguistics, what I am doing here is offering a few suggestions from a narratological perspective. It is up to text linguistics to add further categories, or to reconceptualize the schema in any way that linguists find corresponds with their material. The model in figure 2 is not meant to be a solution to what are clearly quite intractable problems for literary scholars and linguists alike, but tentatively to redress what I have seen as some shortcomings in previous models, particularly from a narratological perspective.
After these caveats, allow me to proceed into more detail. I have proposed two new macro-genres, conversation and reflective discourse, a move that requires some justification. Much of the oral language cannot be aligned exclusively with an argumentative--persuasive or didactic or instructional--purpose, nor does it serve to express the speaker's experience (narrative). Much conversational exchange merely serves the purpose of what Jakobson called the phatic function-staying in touch, establishing and preserving contact, furthering social interaction. Most conversation is in fact seemingly purposeless, does not have predetermined aims or clear procedural routines, or at best moves from one short-term goal to the other. Indeed, much of conversational narrative can be identified as an exchange of news, have-you-heard-the-latest? type of discourse, rather than an instance of full-blown narrative experientiality-and to this extent much conversational narrative has a low degree of narrativity in terms of natural narratology. Conversation has been proposed as a macrogenre to account for the prominence of interactivity in the oral language. In written texts, there are not many such interactive genres. On the level of discourse mode, the textual manifestation that most closely corresponds to interaction is dialogue, a prominent textual surface element in narrative. The reader will have noted that I have placed drama among the narrative genres. I did this to mark the experiential, narrative structure of dramatic literature. In literary drama the interaction is between the characters on stage; their interaction belongs with the conversational macro-genre. The play as a whole, however, is a narrative, and that narrative is represented by an enactment of it. This interaction between characters should not be identified with interaction between communicative partners on the discourse level (in their dialogue); indeed, drama is one of the literary forms that has least direct persuasive force since its mode of rhetoric is for t he most part an indirect one (to be decoded by the audience). The conversational form of drama therefore serves a representational (mimetic) and medium-related function.
My second new category requiring some explanation is the metalinguistic genre, which I have called "reflective." In a sense such a category is the descendent of Jakobson's poetic and metalinguistic functions or of Kinneavy's literary aim of discourse. Moreover, in its reflective garb, this macro-genre has much to recommend itself as a quite general discourse function. All discourse, written and oral, narrative or expository, is usually framed by two additional discourse levels: an orientational, background level as well as a commentatory level. In other words, whatever type of text we are writing, we always add comments to it or fill in additional information. Thus, in a historical text, comments by the historiographer on the purposes of his endeavor or evaluative asides occur as frequently as explanatory comments on the antecedents of certain developments. Likewise, in a cookbook some general comments may be found, for instance advice on how important it is to have an additional ten eggs on hand, and specifi c instructions may be expanded by explanatory asides on how precisely to fold in the eggs or whip the cream. Furthermore, discourses in a variety of ways reflect on themselves or step away from communicative purposes to indulge in abstraction, speculation, and play. Such uses of discourse need not be literary as such, but they can be aligned with a number of historical genres or text types, some of which -- such as jokes - are frequently discussed as potentially literary texts. My main intention in adding the category of reflective discourse, however, was to find a slot for the genre of poetry (in the sense of lyric poetry). A poem represents perhaps the most perfect example of a literary text without a clearly determinable communicative purpose. Even where the mood of the speaker is at issue, a poem seems to consist in an expression of feeling for its own sake. It is for this reason that the reflective category has been added to the list of macro-genres. I will come back to the genre of poetry and its narrat ological relevance in section 3 below.
The second most crucial reorientation apparent from figure 2 concerns the introduction and naming of a level of discourse modes. As I already noted, these do not directly correlate with genres or macro-genres but, instead, constitute text-linguistic units within genre-specific schemata. Adam and Esser already gestured in this direction when they saw the text type narrative as structured by a split into story and evaluation, with story subdividing into orientation and story, and story into narrative episodes that consist of recursive event-resolution segments. In the wake of my proposals in Towards a 'Natural' Narratology, one can specify the structure of narrative even further. Thus, for conversational storytelling, it can be demonstrated that the discourse mode report occurs in narrative clauses inside episodes as well as (less consistently) in some flashback (delayed orientation) sections; description occurs prototypically in orientation and delayed orientation sections; argumentation in narratorial comment and in the final evaluation (the coda); and the abstract could be argued to belong to the expositional discourse mode, with interactional discourse (reader address) also confined to the framing narratorial segments (abstract, coda, narratorial comment). Dialogue, a central feature of narrative episodes, is another prominent discourse mode.
In the case of other genres, the structural patterns and functionally relevant discourse modes vary greatly. Rhetorical genres, for instance, could be analyzed in terms of classical rhetoric categories such as dispositio (cf. Adam's treatment of a commercial for a dishwashing liquid--42); conversation might require a quite different analysis that would have to recur to structural patterns such as turns, feedback, or topic-changing and attention-preserving mechanisms; and argumentative prose could be structured in terms of logical procedural progression with mechanisms for completion, thematic expansion and sequence formation (cf. Werlich, Text Grammar 24-38). Mulcahy and Samuels quote a typology by B.J.F. Meyer (1985) in which different types of expository writing are outlined:
Meyer classifies text according to five types: collection (list-like), causation, response (problem-solution), comparison, and description. Each of these classifications prescribes a specific type of logical relationship. For instance, causation specifies a causal relationship in which one idea is the cause and the related idea is the consequence. [...] Meyer & Freedle (1979) noted the effects of four types of top-level discourse structures on recall. They noted that subjects who listened to adversative (contrastive) and covariance (cause-effect) structures remembered more than subjects who listened to attribution (list-like) and response (problem-solution) structures. (251)
Such global structures for expository writing also allow for functional subdivision in parallel to that of episode elements in narrative. The point I want to make here concerns the variability of discourse modes (there are a great number of discourse modes, quite a few of which are peculiar to only one or two genres) and their functional multivalence: Report or description, for instance, have an entirely different status and function in narratives than they do in expository writing. Again, as a literary scholar I am here turning the job over to linguists to do more with such a model. My concern, basically, is a narratologically fruitful approach to the issue of text types, i.e., macro-genres.
What is new about the model I have proposed relates primarily to the functional alignment of discourse modes within genres. As in Virtanen and Warvik and Kinneavy, the macro-genres are ideal types, and so are genres and text types. My main concern in proposing the model has been to explicate the macro-genre narrative as an ideal type and to relate its multiform textual surface structure on the sentence level (i.e., the discourse modes) to the abstract text-type category (here split into macro-genre and specific generic moulds). I therefore now return to my narratological origins in spelling out what this model might do in practical narratological terms.
3. Consequences for Natural Narratology
In this final section of the paper I want to outline a few consequences of my conception of macro-genres for narratology.
The provisional model of macro-genres presented here is, first of all, one that takes naturally occurring discourses as its prototypes. Therefore, all macro-genres correspond to discourse functions observed in oral language. Nevertheless, some discourse functions have--in terms of actualized genres--developed more strongly in the written language, whereas others (the conversational macro-genre) continue to be constitutive of orality even though they are adaptable to written discourse. As a consequence of the historical move from orality to literacy, the surface structure of texts begins to evolve in different ways. As far as the narrative macro-genre is concerned, for instance, the discourse modes observable in natural narrative increasingly lose their original functions within the structure of the (oral) narrative episode and entail changes in narrative structure that require concomitant functional realignments and modifications. Thus, the originary episodic pattern of oral narrative develops towards a repo rt-plus-scene structure constitutive of the novel as a genre. A similar type of development can be observed for such macro-genres as argumentative or instructional discourse because each written text type tends to evolve a tradition of its own in which quite specific structures and techniques establish themselves as core elements of the genre. One need only think of the topic sentence in essay writing to realize that such features are both structurally central and historically relative--topic sentences are not the norm in French or German argumentative prose, and these days they seem to have gone out of fashion to some extent even in English language composition. My model is therefore open to a diachronic perspective.
A second important consequence of integrating narrative with other macro-genres is the reposing of the definitional question of narrative. If narrative is definable in relation to a spectrum of five or seven text types (macro-genres), then one is tempted to set narrative in contrast to instruction or reflection in a quite essentialist manner, or, relationally, to posit systematically complementary discourse functions for each macro-genre. Here it is extremely important to remember that macro-genres are abstractions and do not exist; in fact, they "exist" much less concretely than do genres manifested in a huge number of actual texts and do seem to influence composition by the weight of tradition (the bulk of former such texts) and the hypostacized "genre" concept emanating from that tradition or practice. Moreover, by contrasting macro-generic features, one is actually contrasting idealized concretizations of discourse functions.
It might, in fact, even be possible to eliminate the level of macro-genres by a swift slash with Occam's razor. After all, the purpose that macro-genres serve is that of ordering genres into manageable groups or categories. Despite some basic distinctions that apparently form part of speakers' linguistic competence (Faigley and Meyer), the traditional categorizations repose on very arbitrary groupings, as the comparison of different models of text types (including my own) has no doubt illustrated. It is because we want to find a common denominator among texts that we regard as narrative, expository, or instructional that we begin to search for a discourse function that might unite the very disparate genres within these categories. Thus, to posit that readers first choose a discourse function before they decide on a text type or genre seems to me to invert the order in which, cognitively speaking, textualizing processes actually evolve. Speakers, I would argue, start out from a universe of discourses, and wit hin that universe pick out the genre that seems most appropriate to their current concerns. I would presume that they also tactically choose one genre over another because of the textual strategies available in a specific genre.
Having deconstructed the idea of the macro-genre and revealed it to be a mere red herring in text linguistics and literary theory, where do we go from here?
Whereas text linguistics can perhaps afford to shrug its shoulders and simply turn to the analysis of individual genres (cookbooks, instruction manuals, TV commercials), postponing theorization to the stage when it can safely be performed in an inductive manner, literary scholars find themselves in the unenviable position of having to search through their own rubbish bin: even if one agrees that "narrative" as a macro-genre may be an illusionary concept (and, even more so, an elusive one), the fact remains that there exist huge numbers of novels, poems, plays, histories, and biographies, and that one has to make categorical distinctions among them. Since literary critics do not regularly concern themselves with scientific prose or instruction manuals, they focus on a very narrow set of texts, so called literary texts, that is. As a consequence, distinctions among them loom large, and the constraint to set apart what readers intuit to be different groups of genres results in theoretical disciplines such as po etics, genre theory, narratology, drama theory, or the theory of poetry. Why is it so much more important to literary scholars that they can "prove" a text to be narrative rather than, say, lyric? Perhaps this is the case because there are so many poems or prose texts that defy categorization, that force readers to decide whether to read them as narrative rather than as poetry? Because we read literary texts quite differently when we read them as narratives rather than as poems or essays. When we encounter a commercial, we read this as a commercial, and the problem of whether a commercial is an instructional text (go and buy this) or an argumentative text (this article is good because ...) arises for the text linguist but presumably not for the consumer. By contrast, reading a passage from Gertrude Stein's later work as narrative will yield a quite different reading experience from its being read as poetry.
Which brings us back to the much-vexed question of the distinctions between the narrative and poetic genres, an issue recently taken up again by Werner Wolf in "Aesthetic Illusion in Poetry." (The distinctness of drama in any discussion makes it a moot point in this context; even if narratologists lay claim to drama as a narrative genre, drama will always be clearly defined on account of its dialogue form and performative and visual nature.)
Wolf does not create an inviolable barrier between poetry and what he calls aesthetic illusion (which correlates with mimesis as prototypically constitutive of narrative, drama, painting, film). According to Wolf, Keats's "Grecian Urn" can be read as an illusionistic text in which the "perception of the urn," rather than being a merely static description, turns into "a dynamic imaginative activity" (282-83) and therefore fulfills a requirement of experientiality (Fludernik) that for Wolf constitutes aesthetic illusionism.  In other words, by positing the viewer as engaged in a specific act of perception, one specific in space and time, Wolf is able to move from the displacement, abstraction, and generality of poetry to aesthetic illusion. But Wolf does not align this poem with narrativity.  Wolf's proposals, which are a very interesting reworking of his earlier positions in Asthetische Illusion (1993), also introducing the possible-worlds aspects of aesthetic illusion, are located on the level of inter literary analysis. They split the literary field into illusionistic, antiillusionistic, and nonillusionistic texts. For Wolf, the genre narrative is distinctly illusionistic, even if combinable with some antiillusionistic elements. Poetry, by contrast, is prototypically conceived of as nonillusionistic but may assume illusionistic features if read narratively (the narrative poem) or experientially, as in the case of Keats's "Grecian Urn."
Problems arise, however, if one attempts to integrate Wolf's model with text-type theory. The main one is that because Wolf very laudatorily attempts to define aesthetic illusion as well as poetry on the basis of a cluster of features partly in binary opposition, he deploys features that do not easily compare with instructional prose or with scientific texts. Thus, Wolf s list of the characteristics of poetry ("Aesthetic Illusion" 261-63) combines (1) brevity and reduction (versus the novel?--but this is a feature also typically adduced for the short story); (2) the presence of a lyric persona (fiction, too, usually is supposed to have a speaker); (3) the "emphasis on the utterance as the product of consciousness rather than on the content of the utterance (as in narrative fiction)"; (4) dereferentialization (i.e., lack of specificity, a feature that is prominent in narrative); (5) foregrounded self-referentiality (i.e., Jakobson's poetic function; this is a feature shared with anti-illusionistic narrative); (6) musicality (i.e., the rhythmic and acoustic features of poetry); and (7) artificiality (i.e., the foregrounding of formal features of poetry such as versification). It is interesting to note that a narrative that is particularly anti-illusionistic, having qualities (3), (4), (5) and (7), would--according to this model--appear to be more "lyrical" than prototypically narrative.
Wolf's essay is a particularly noteworthy attempt to set up a definition of poetry in contradistinction to narrative, thereby redefining narrative in opposition to poetry. This move constitutes a departure from the traditional constitutive opposition of narrative versus drama that is the basis of F.K. Stanzel's narrative theory and implicit in Genette's formulations. Wolf is astute in not setting up clear binary oppositions between narrative and poetry on each and every count. What immediately emerges as a crucial consequence of this model, however, is its lack of common ground with such text-type oppositions as Chatman's narrative, argument, and description, and expository and instructional writing in some other typologies. Nor does Wolf address the question of fictionality, although his alignment of aesthetic illusion qua experientiality with possible world structure suggests that lyric poetry is only partly a fictional genre. But the question of fictionality as a qualifying category does not really arise because his emphasis is exclusively on the literary realm. In Wolf's very interesting model, the literary realm is entirely cut off from that of other nonliterary texts. The question whether, or to what extent, his categories might apply to, or compare with, historical writing, instructional prose or expository discourse simply does not figure. The very strength of the model lies precisely in its concentration on the literary field, and within that it supplies excellent distinctions.
Natural narratology, by contrast, has set itself the task to encompass the entire range of literary and nonliterary texts, and--like Chatman--to integrate narrative with other macro-genres. I have now modified my earlier proposals to add the reflective macro-genre to the more traditional list, and it might therefore be of use to show more precisely how natural narratology combines the linguistic framework of text types with the requirements of narratological and poetic analysis. Comparing narrative with other macro-genres (including the reflective) puts the emphasis on the discourse function(s) of narrative that--in natural narrative--consist in the discursive mastering of experience. Experientiality was defined as the conjunction of tellability and point (Towards a 'Natural' Narratology 29), that is, in relation to the narrative dynamics of embodied experience and human consciousness. Reflectivity, by contrast, although it is also an expressive discourse function, seems to operate like a kind of equivalent of "art for art's sake" in the realm of discourse. As Wolf points out, whereas the self-referential strain of poetry can be treated as an indicator of dereferentialization, narrative (and all other major macro-genres) are inherently referential (whether they refer to a possible world or the addressee's and speaker's shared world).
What is the status of aesthetic illusion and of consciousness in these models? Aesthetic illusion, in the framework of natural narratology, corresponds to a naturalization of a text as mimetic. For those unfamiliar with Towards a 'Natural' Narratology, naturalization is the process by means of which readers decode a text as narrative. One of the major consequences of the model of natural narratology has been the integration of twentieth-century texts that--according to traditional narratological accounts--were only marginally narrative into the realm of narrativity. Thus, such plotless Modernist texts as Virginia Woolf's "The Mark on the Wall" display a high degree of experientiality and are therefore prototypically narrative. Not only is the speaker in "The Mark on the Wall" represented as an experiencer, a consciousness, but the text is also an example of extensive reflectivity and therefore even more closely bound up with human consciousness. Such texts are narrativized accordingly as instances of represe nted consciousness. Other, more metafictional exercises, such as the texts of B.S. Johnson, likewise can be narrativized as exercises in narratorial reflectivity: the speaker's reflective and critical consciousness constitutes the experientiality of the text.
The question here is to what extent mimesis interrelates with consciousness. Mimesis certainly comprises the representation of consciousness as part of our everyday world experience. It does not necessarily entail considerations of temporal and spatial anchoring or of temporality, both of which are central to the notion of narrative experientiality and therefore constitutive of narrativity. Wolf's aesthetic illusion, which I take to be a cognitively updated version of mimesis (and, possibly, realism), correlates with the illusion of life-likeness. Since narrative experientiality is part of the general human experience, it can become a signified of aesthetic illusionism. Narrativity, which is established by the projection of experiential consciousness (as agent, patient, viewer, or simply as reflective mind) within specific spatio-temporal parameters, therefore participates in illusionism, but illusionism is not necessarily constitutive of narrative. Keats's "Ode to a Grecian Urn" can be discussed as an insta nce of aesthetic illusion, of ekphrasis. It is not an instance of narrative, however, since the speaker remains too abstract to figure as a narrative persona. On the other hand, a poem like Marlowe's "The Passionate Shepherd to his Love" ("Come live with me, and be my love") is arguably more narrative since it invokes an entire situation of wooing that might be specific in place and time. Coleridge's "Kubla Khan" is perhaps even more easily recuperable as narrative, either as a marginally narrative account of Kubla Khan's specific experience of omens ("And 'mid this tumult Kubla heard from far / Ancestral voices prophesying war!"), or--much more convincingly--as the account of the speaker's dream vision, i.e., his consciousness. Narrative poems, such as Thomas Gray's "Ode on the Death of a Favourite Cat" ("'Twas on a lofty vase's side [...] The pensive Selima reclined [...] She tumbled headlong in.") or Coleridge's "Ancient Mariner," need no additional effort at narrativization; their narrativity is easily as sured on the basis of a proper plot and correlative experientiality.
Poetry can, therefore, both in Wolf's model and in my own, share illusionistic and even narrative features. But prototypical poetry tends to be both nonnarrative and nonillusionistic. Within a text-type approach to literary genres, however, aesthetic illusion (whether in conjunction with, or as a replacement of, narrativity) has little usefulness as a defining feature of narrative in contrast to instruction or conversation or other macro-genres. In fictional narratives, precisely on account of their fictionality, aesthetic illusionism can lay claim to an important status in the critical vocabulary. When one extends the study of narrative to the realm of all discourses, different distinctions become operative.
But, as in the discussion about macro-genres, this analysis is actually de trop. What readers really do when they read a text largely depends to start with on the observable generic alignment of that text. In so far as a text is read as narrative (and poems will less easily give rise to such a reading), it is read as a specific instantiation of experientiality in time, and this reading will imply a good measure of mimetic illusionism. Narrativization, moreover, manages to explain how experimental texts that are plainly antiillusionistic nevertheless allow themselves to be read in accordance with narrative frames. Poems, which are prototypically conceived of as nonillusionistic, can also be read as narratives, but only to the extent that they engage with specificity  and temporality, and with the discursive mastery of experience.
At this point we have, however, left behind our linguistic model of text types based on inductive abstractions and have moved instead to a reception-oriented model of the dynamic decoding process, a model that opens itself to diachronic analysis. Within the confines of the mimetic language game, the purely literary conceptions of Wolf's model pertain; when expanded to include the oral language, nonfiction and a diachronic perspective, narrativity--if we are to create abstract entities at all--needs to be conceived in a broader and more durable shape. It must operate as a text type constituent as well as a category of literary production, and it must be flexible enough to allow for readers' or speakers' appropriations, modifications, or revaluations. It is from this broad, inclusive perspective that natural narratology can fruitfully engage with text linguistics and become a partner in a lively crossdisciplinary exchange. After all, genre is pot a concept confined to literary criticism, but a term relating to categorial analysis. And categories of the mind are shared among us all.
Monika Fludernik (email@example.com) is professor of English literature at the University of Freiburg/Germany. She is the author of The Fictions of Language and the Languages of Fiction (Routledge, 1993) and Towards a 'Natural' Narratology (Routledge, 1996), which was awarded the Perkins Prize by the Society for the Study of Narrative Literature (SSNL). She has edited a special issue on second-person fiction (Style 28.3, 1994), a special issue of The European Journal of English Studies on "Language and Literature" (EJES 2.2, 1998), and (coedited with Donald and Margaret Freeman) a special issue of Poetics Today 20.3 (1999) on "Metaphor and Beyond: New Cognitive Developments." Further publications include two collections of essays (Hybridity and Postcolonialism: Twentieth-Century Indian Literature, Tubingen: Stauffenburg, 1998; and Das 18. Jahrhundert, Trier: WYT, 1998). Forthcoming are two further collections, Romantik (Trier: WVT, 2000) and Fin de siecle (Trier: WVT, 2001), and a monograph on Gabr iel Josipovici (New York: Lang, 2000).
(1.) See de Beaugrande (197-99) for a comprehensive user-oriented typology.
(2.) According to Faigley and Meyer's "Rhetorical Theory," readers are able to recognize text types as part of their linguistic competence.
(3.) "Wenn die Literaturwissenschaft, auBer den Texten der schonen Literatur oder Dichtung, auch nichtpoetische ('expositorische') Texte untersuchen will, ist der historisch-literarische Gattungsbegriff ohne besondere Schwierigkeiten auf diesen Untersuchungsbereich ubertragbar" (Weinreich 161). 'If literary criticism analyzes non-poetic ('expository') texts besides belles lettres or poetry, it can do so without any difficulty and can employ the concept of genre (a term from literary history) also in reference to such non-poetic texts."
(4.) Wolf's original definition of aesthetic illusion in Asthetische Illusion und Illusionsdurchbrechung in der Erzahlkunst did not yet repose on a criterion of experientiality. I introduced the term in Towards a 'Natural' Narratology to define narrativity and to link narrative with a consciousness factor whether of mediation or participation.
(5.) Wolf continues to link narrativity with plot density ("Ereignishaftigkeit -- personal communication).
(6.) Compare also Pfister 12-14.
(7.) This is figure 6 ("Les niveaux d'analyse typologique des textes") in Virtanen and Warvik 106.
Adam, Jean-Michel. "Quels types de textes?" Le francais dans le Monde 192 (1985): 39-43.
Adams, Jon-K. Narrative Explanation: A Pragmatic Theory of Discourse. Aachen British and American Studies 7. Frankfurt and Berlin: Lang, 1996.
de Beaugrande, Robert. Text, Discourse and Process: Toward a Multidisciplinary Science of Texts. Advances in Discourse Processes 4. Norwood, NJ: Ablex, 1980.
Biber, Douglas. Variation across Speech and Writing. Cambridge: Cambridge UP, 1995.
Bonheim, Helmut. The Narrative Modes: Techniques of the Short Story. Cambridge: Brewer, 1982.
Chatman, Seymour. Coming to Terms: The Rhetoric of Narrative in Fiction and Film. Ithaca, NY: Cornell UP, 1990.
Clark, Robert, and Thomas Healy, eds. The Arnold Anthology of British and Irish Literature in English. London: Arnold, 1997.
Coleridge, Samuel Taylor. "Kubla Khan." Clark and Healy 717-18.
___. "The Rime of the Ancient Mariner." Clark and Healy 699-717.
van Dijk, Teun A. Textwissenschaft: Eine interdisziplinare Einfuhrung. Munich: Deutscher Taschenbuchverlag (dtv), 1980.
Esser, Jurgen. "Text-Type as a Linguistic Unit." Anglistentag 1990 Marburg. Proceedings. Ed. Claus Uhlig and Rudiger Zimmermann. Tubingen: Niemeyer, 1991. 142-53.
Faigley, Lester, and Paul Meyer. "Rhetorical Theory and Readers' Classifications of Text Types." Text 3.4 (1983): 305-25.
Fludernik, Monika. Towards a 'Natural' Narratology. London: Routledge, 1996.
Gray, Thomas. "Ode on the Death of a Favourite Cat." Clark and Healy 625-26.
Jakobson, Roman. "Closing Statement. Linguistics and Poetics." 1958. Roman Jakobson: Language in Literature. Ed. Krystyna Pomorska and Stephen Rudy. Cambridge, MA: The Belknap Press of Harvard UP, 1987. 62-94.
Keats, John. "Ode on a Grecian Urn." Clark and Healy 764-66.
Kinneavy, James L. A Theory of Discourse. The Aims of Discourse. Englewood Cliffs, NJ: Prentice-Hall, 1971.
Longacre, Robert E. The Grammar of Discourse. New York: Plenum Press, 1983.
Marlowe, Christopher. "The Passionate Shepherd to his Love." Clark and Healy l68.
Meyer, B. J. F. "Prose Analysis: Purposes, Procedures, and Problems." Understanding Expository Text. Ed. B. Britton and J. Black. Hillsdale, NJ: Lawrence Eribaum Associates, 1985.
___, and R. Freedle. The Effects of Dfferent Discourse Types on Recall. Princeton, NJ: Educational Testing Service, 1979.
Mulcahy, Patricia I., and S. Jay Samuels. "Problem-Solving Schemata for Text Types: A Comparison of Narrative and Expository Text Structures." Reading Psychology: An International Quarterly 8 (1987): 247-56.
Pfister, Manfred. "The Philosophical and the Dramatic Dialogue." Comparative Criticism 20 (1998): 3-15.
Stanzel, Franz Karl. A Theory of Narrative. Cambridge: Cambridge UP. 1984.
Virtanen, Tuija. "Issues of Text Typology: Narrative--A 'Basic' Type of Text?" Text 12.2 (1992): 293-310.
___, and Brita Warvik. "Observations sur les types de texte." Communications. 8 Recontre des professeurs de francais de I'enseignement superieur. Ed. Juhani Harma and Iris Makinen-Schwanck. Helsinki: Publications du Departement des Langues Romanes 6, 1987. 91-114.
Weinrich, Harald. "Thesen zur Textsorten-Linguistik." Textsorten: Differenzierungskriterien aus Iinguistischer Sicht. Ed. Elisabeth Gulich and Wolfgang Raible. Frankfurt: Athenaum, 1972. 161-69.
Werlich, Egon. Typologie der Texte: Entwurf eines textlinguistischen Modells zur Grundlegung einer Textgrammatik. Heidelberg: Quelle & Meyer, 1975.
___. A Text Grammar of English. 2nd ed. Heidelberg: Quelle & Meyer. 1983.
Wolf, Werner. Asthetische Illusion und Illusionsdurchbrechung in der Erzahlkunst: Theorie und Geschichte mit Schwerpunkt auf englischem illusionsstorenden Erzahlen. Tubingen: Niemeyer, 1993.
___. "Aesthetic Illusion in Lyric Poetry?" Poetica 30.3-4 (1998): 251-89.
Woolf, Virginia. "The Mark on the Wall." A Haunted House, and Other Short Stories. New York: Harcourt Brace Jovanovich, 1972. 37-46.
|Printer friendly Cite/link Email Feedback|
|Date:||Jun 22, 2000|
|Previous Article:||What Do We Mean When We Say "Narrative Literature?" Looking for Answers Across Disciplinary Borders.|
|Next Article:||Lateral Reflexivity: Levels, Versions, and the Logic of Paraphrase .|