MONIKA FLUDERNIK. Narratology in Context. Sonderdrucke aus der Albert-Ludwigs-Universität Freiburg

Transcription

1 Sonderdrucke aus der Albert-Ludwigs-Universität Freiburg MONIKA FLUDERNIK Narratology in Context Originalbeitrag erschienen in: Poetics today 14 (1993), S. [729] - 761

2 Narratology in Context Monika Fludernik English, Vienna Seymour Chatman, Coming to Terms: The Rhetoric of Narrative in Fiction and Film. Ithaca, NY: Cornell University Press, Ansgar NUnning, GrundzOge eines kommunikationstheoretischen ModeIls der erzahlerischen Vermittlung: Die Funktionen der Erzahlinstanz in den Romanen George Eliots. Horizonte: Studien zu Texten und Ideen der europaischen Moderne 2. Trier: Wissenschaftlicher Verlag Trier, Sandy Petrey, Speech Acts and Literary Theory. New York: Routledge, Vaheed K. Ramazani, The Free Indirect Mode: Flaubert and the Poetics of Irony. Charlottesville: University Press of Virginia, Roger Sell, ed., Literary Pragmatics. London: Routledge, In recent years the discipline of narratology has frequently been pronounced "dead," irrelevant, or "out" see, for instance, the hilarious report on the 1990 MLA convention in the New York Times Magazine (Matthews 1991: 57). In particular, narratology's structuralist origins have increasingly come under attack, mostly for a tendency toward formalization and categorization and a lack of ideological awareness, all of which are inherent to structuralism per se (Pavel 1989). As a consequence, narrative studies have been recently departing from the classic formalist concerns of narratology "proper," extending into such bordering areas as psychoanalytic criticism (Brooks 1984), deconstruction (Chambers 1984), feminism (Warhol 1989), and cultural criticism Poetics Today 14:4 (Winter 1993). Copyright 1993 by The Porter Institute for Poetics and Semiotics. CCC /931$2.50.

3 730 Poetics Today 14:4 (Chambers 1991). At the same time, one can now observe a backlash from the "old guard" narratologists, who have started to revise, as well as to reiterate, some of their earlier formalist methodologies, complementing the old paradigms by developing them in new directions. This trend has been apparent since the 1988 Society for the Study of Narrative Literature (SSNL) conference in Columbus, Ohio, where a small but vociferous group of scholars (myself among them) warned of throwing out the baby with the bathwater. We are now confronted with a number of new statements by the grand old wo/men of the discipline: Gerard Genette's Fiction et diction (1991), Seymour Chatman's 1990 work, Coming to Terms (reviewed here), and the series of essays in the three special issues on narratology of this journal: Poetics Today 11:2 (Summer 1990), 11:4 (Winter 1990), and 12:3 (Fall 1991), which include work by Mieke Bal, Gerard Genette, Dorrit Cohn, F. K. Stanzel, and Meir Sternberg in addition to some excellent contributions by Ruth Ronen, Marie-Laure Ryan, Thomas G. Pavel, William Nelles, and Uri Margolin. In contrast to the situation in 1988, narratology today seems to be more alive than ever, even if beset by disputes over its legitimate aims and methods. This new lease on life seems to me to be due in no small measure to the still-growing influence of linguistics on narrative study, specifically of linguistic models that go beyond the narrowly structuralist framework of yore. Although not all of the books discussed in this review essay reflect clearly definable linguistic models, the influence of speech-act theory, discourse analysis, and pragmatics is readily apparent. Whereas earlier narratologists relied heavily on Saussurean and Benvenistean binary opposition (and on markedness) within static systems, the latest trends in linguistics offer paradigms that emphasize textual processes. Linguistic pragmatics, for instance, underscores the reader's role in the creation of a fictional world by identifying active inferential processes and acts of imaginative projection that operate according to criteria of relevance (see, e.g., Sperber and Wilson 1986, as applied in Pilkington 1991), culturally sanctioned frames and scripts (see Schank and Abelson 1977, as applied in, e.g., Polanyi 1985), or strategies of cooperation, as described in discourse analysis (e.g., the "politeness" of the text [Sell, pp ], Gricean cooperative maxims, etc.). The books reviewed below a random collection of recent studies on narratives both specific and abstract, on linguistic approaches to literature, and on individual writers document both the still-increasing tendency to use linguistic paradigms in the analysis of (narrative) texts and the continuing usefulness of "traditional" narratology within certain theoretical and practical limits. Although I will discuss each book separately (they are too different in their approaches and subject mat-

4 Fludernik Narratology 731 ter to allow facile comparisons), this essay will concentrate on a number of crucial issues that are raised in these texts. I will mainly consider two closely related aspects of narrative fiction: the communicative situation in narrative, and the pragmatics (in a linguistic sense) of reading. The issue of a communicative situation in narrative is overtly addressed in both Chatman's Coming to Terms and Sandy Petrey's Speech Acts and Literary Theory, but from two very different perspectives. Chatman's renewed plea for the existence of the implied author is embedded in a very traditional narratological framework, which he modifies and reanalyzes, and can be compared more profitably with the theoretical first part of Ansgar Niinning's book on the narrator function in the novels of George Eliot. Petrey's more theoretical deliberations on the consequences of speech-act theory for the analysis of literature (with special reference to the debate between deconstructionists and speech-act theorists) have little to say about narrative fiction and can be compared more fruitfully with the pragmatic concerns of both Vaheed Ramazani's Flaubert study and Roger Sell's edited collection on literary pragmatics. Where communication and pragmatics do connect is precisely in the effects of communication. While Chatman and Niinning are primarily concerned with the agents of narrative communication, Petrey concentrates on literature as an illocutionary speech act that therefore has inevitable (pragmatic, if you will) effects on the reader, so he would agree with Sell in his concern for relevant context. One contextual phenomenon that recurs as an issue in Sell's collection is irony, which brings Ramazani into the picture as a point of comparison. Niinning's interesting study of the characteristics of the narrator function in George Eliot's fiction is divided into two parts. Part 1 proposes a communicative model of narrative transmission, while part 2 (chapter 3) traces the development of the narrator function from Scenes of Clerical Life through Daniel Deronda. For obvious reasons of space and relevance, I will concentrate here on part 1 (which will also lead into Chatman's presentation of a roughly equivalent set of issues). Niinning's book is based on his 1989 dissertation (University of Cologne) and is noteworthy both for its theoretical depth and for the importance of its insights into George Eliot's narrative art. Nevertheless, like most two-part studies of its kind, it suffers from a certain dullness (owing to a pronounced lack of examples) in the theoretical first half and a corresponding repetitiveness and overkill in the subsequent work-by-work "application" of the theory. This is not entirely Niinning's fault, but the consequence of a worldwide conspiracy among those who set dissertation formats, with the resuit that dissertations are required to be as "unreaderly" as possible. Niinning's patient analysis of Eliot's novels supersedes recent accounts

5 732 Poetics Today 14:4 of her work, such as those by Gillian Beer (1986), Rosemary Clark-Beattie (1985), Suzanne Graver (1984), Barbara Hardy (1982), Karen B. Mann (1983), or K. M. Newton (1981). His study amply documents the workings of an active narrator in all of Eliot's fiction and traces the gradual depersonalization of that narrative instance. The narrators in Eliot's later novels also cease projecting a male persona, providing instead contradictory or ambivalent clues to the gender of the narrative voice. Nil nning concentrates on the development of narratorial evaluation, addresses to the narratee, and the balance between critical and empathetic presentations of the characters in Eliot's fiction. A comprehensive narratological analysis of this kind has been long overdue, and it should certainly be the starting point for any further refinements. Nanning's theoretical part, on which I will concentrate here, proposes a five-level model of narrative communication: N1 (N for Niveau, i.e., "level") is the level of communication between characters within the fiction, which is embedded in N2 (narrator-narratee communication). The N3 level is that of the hypothetical implied author, which. (deliberately not whom) Nanning prefers to call "Subjekt des Werkganzen" (the subject of the fiction as a whole); N3 in particular includes all of the textual paraphernalia (titles, marginalia, epigraphs, etc.) not ascribable to a narrator, in addition to the full linguistic text, and it also covers the structural processes of temporal reordering, juxtaposition of characters, and those effects of irony that depend on incompatibilities between levels Ni (the characters) and N2 (the narrator). By contrast, N4 is the locus of unreliability, the presence of which depends on a finding of incompatibility between textual norms on levels N2 (the narrator's presentation) and N3 (the differing implications of the full text). Nanning defines N4 as the communication level between the "empiric author in his role as literary producer" and the "empiric reader in his role as recipient of literature" (p. 26). Finally, N5 signifies the level of the real author and reader in their respective social roles. As one can already see from this brief summary, Nanning reshuffles the well-established system of narrative instances made familiar to us by Gerald Prince, Chatman, and Bal. Chatman continues to adhere to these traditional models in Coming to Terms by postulating distinctions among the story plane, that is, characters' communication (L1), the discourse plane (L2), or the level of narrative transmission, the implied author (L3), the "career author," that is, the image of "Dickens" or "Rushdie" that one constructs from all fictional works by the author. (L4), and the real, historical author (L5). Chatman's innovations pertain less to the qualities of the implied author as with Wayne Booth, these mainly amount to a locus for resolving unreliability and for situating the "intentions" of the fiction as a whole; Chatrnan's particular

6 Fludernik Narratolo9y 733 contribution concerns the explicit postulation of a cinematic narrator (who shows rather than narrates) as a vehicle of narrative transmission. (I will return to Chatman's crucial L 1 /L2 distinction and its significance below.) Nanning's schema raises some very interesting questions and problems. Levels N1 and N2, which are relatively uncontroversial, are defined by their relation of embedding: the characters' communication is embedded in the narrator's linguistic account of the fiction. Nanning here relies on Bal's (1981) definition of embedding, which stipulates a homogeneity between the embedded and embedding levels (Ni and N2). Nanning thus explicitly rejects Stanzel's (and, implicitly, Chatman's [1978] and Bal's [1985]) stipulation of ontological incompatibility between the narrator and character levels. The characteristic of homogeneity for Ni and N2, however, seems to rely exclusively on their being "fictional utterances" N3 is declared to be nonhomogeneous. Here, Nanning seems to have been led astray by a non sequitur: after all, titles and epigraphs are at least as fictional as a narrator's metanarrative commentary. The only homogeneity between Ni and N2 that I can discover lies, on the contrary, in the fact of utterance both narrator and characters are (or can be) explicitly personalized to the extent of acquiring the capacity for utterance and hence (potentially) for communication. However, utterance does not equal narration, and Bal (1985) quite correctly distinguishes between the narrator as communicator and/or metanarrative commentator, on the one hand, and the narrator as the instance of narrative transmission (of the histoire), on the other. Since the characters' speech acts unless these are narratives in their own right are simply part of the plot, Genette is quite correct in analyzing Ni as the signified of the enunciation on the narrative level (N2). It is therefore even odder for Nanning to state that direct and indirect (I) speech presentation be long to Ni, but that free indirect discourse and psycho-narration belong to level N2 (p. 28). 1 In spite of these theoretical difficulties regarding levels Ni and N2, 1. Compare, however, the standard dual-voice account a page later: "Der Erzahlgegenstand `BewuBtsein der Figuren' kann erzdhltechnisch mit verschiedenen Verfahren dargestellt werden, von denen nur psycho-narration eindeutig ganz dem Erzdhler zugeschrieben wird, wãhrend andere Techniken der BewuBtseinsdarstellung etwa quoted oder narrated monologue graduell Elemente der Figurenrede enthalten" (The narrative matter referred to as "characters' consciousness" can be represented by a number of different narrational techniques. Among these only psycho-narration is unambiguously referable to the narrator, while other techniques for the representation of consciousness, such as quoted or narrated monologue, display a gradable amount of characters' language) (p. 29). There is no need to insist that psycho-narration cannot be said to belong exclusively to the narrator's language.

7 734 Poetics Today 14:4 which are, pragmatically at least, uncontroversial, Nanning's model merits serious consideration for its formulation of levels N3 and N4. Like Chatman, Nanning asserts the existence of a narrative instance even in those texts lacking an explicit (overt) narrator. The German term Erzahlinstanz necessarily reduces that speaker on level N2 (= S2) to the source of fictional utterances. (I will return to this point below in connection with Chatman's cinematic narrator.) All the narrative models that I know of, however, collapse Nanning's N3 and N4 levels into the single narrative instance of the "implied author" and comparable categories. In Nanning's formulation N3 is derived from purely textual criteria and comprises the system of textual norms as inferable from the "words of the text"; N4, on the other hand, designates the projected interpretive inferences elicted by the text, in particular the implied "meaning" of the text as an "argument" of the fiction. What N4 therefore seems to correspond to is precisely the time-honored. "authors" reflected by such sentences as "In Hard Times Dickens inveighs against the heartlessness of industrialization." It is N3 that constitutes a valuable refinement of the theory, although it also raises some problems in comparison with current models. Since N3 is a purely textual category, it seems to include (or project?) N2 as its signifie, while at the same time implying N4 as its global signifii. 2 Nanning's model makes a neat box-within-box structure, along the lines of Chatman's or Bal's model, impossible. Nanning's five levels pragmatically link unreliability to the "empirical" author/"empirical" addressee level (N4). His model therefore marks an important advance with regard to the workings of textual inference, especially since Nanning clearly distinguishes between inferences drawn on the basis of the text as a whole and those made by weighing inferences from prior textual levels (Ni through N3). Besides these theoretical innovations which, I believe, could be improved on even further Nanning is to be commended for a very neat yet comprehensive categorization of narratorial functions on level N2.3 The narrative instance or speaker on N2 (= S2) much in line with Chatman's covert versus overt narrator can be analyzed on a scale between a narrative medium, also called a "neutral instance of transmission" (p. 61), and an explicit (personalized) narrator. In accordance with Bal's model, on which Nanning relies for his conception 2. On page 33, Nanning calls N3 a "semantic" content level of a text, which contradicts the textual nature of N3 as illustrated by the characteristics listed on pages 37 to 39. In particular, Nanning's remarks on diverse kinds of irony and reliability are too condensed to be really useful. Instead, a clear characterization of N4 would have been more valuable here. 3. Nanning's initial characterization of narratorial properties (pp ) is less felicitous than his later, more extended categorization and explanation.

8 Fludernik - Narratology 735 of focalization, the neutral narrative medium can be correlated with external focalization. The narrative medium has only so-called technical functions in the narrative (erziihltechnische Funktionen), whereas an explicit narrator (S2) also has speaker functions. This scale between neutral and explicit narration coincides with other scales of tendentially graded characteristics of the narrative function (see Nanning's table on p. 84). I cannot dwell here on the many functions of the narrative medium and the narrator that are explicated by Nanning his presentation seems to be the first complete account of all the various well-known. functions.4 Suffice it to say that the neutral narrative medium's technical functions basically amount to constituting the plot and characters as well as their space and time (pp ). The explicit narrative speaker, on the other hand, can serve a great variety of functions (pp ). These discursive functions are opposed to the neutral diegetic ones. Where this schema of Nanning's is less satisfactory is in its failure to interrelate the various speaker functions of the narrator. Some of these are more compatible with a neutral narrative medium than others. In particular, certain diegetic restrictions follow from the personalization of a heterodiegetic narrator, though fewer than those affecting a homodiegetic narrator. Nanning is apparently less concerned with these possibilities because he has developed the schema to account for George Eliot's fiction, where the narrator is at all times omniscient, if not omnicommunicative (Sternberg 1978). will now turn to Chatman's Coming to Terms, a study that departs from his classic Story and Discourse (Chatman 1978) in three basic directions. First, Chatman refines and modifies his earlier paradigm by making a lengthy plea for an implied-author position to be included in the narrative communication schema. Secondly, he offers a radical terminological revision of the point-of-view problematic. Thirdly, and most importantly, Chatman attempts to integrate narrative with other text types, namely, argument and description, documenting the peculiarities of each text type in both fiction and film. Although there are a few references to drama, Chatman's new book clearly calls for a complementary study that would integrate this basic narrative genre. Let me start with the more narrow narratological problems and leave the text-type theory for a transition into pragmatics later on. Chatman, after reviewing the intentionalist issue,5 defines the implied 4. Most books on narratology concentrate on only some of these, shying away from a potentially boring enumeration of all the possibilities. 5. The question of authorial intention is closely linked to the postulation of the implied author (see Juhl 1980; as well as Booth 1983 [1961] and Bronzwaer 1978).

9 736 Poetics Today 14:4 author as the "record of textual invention," the repository of the text's "codes and conventions": "As an inscribed principle of invention and intent, the implied author is the reader's source of instruction about how to read the text and how to account for the selection and ordering of its components" (pp ). The implied-author position is to Chatman an indispensable theoretical construct because one needs to distinguish between the "act of a producer, a real author," and "the product of that act, the text" (p. 83). Unlike the narrator, the implied author has "no 'voice,' " but instead "empowers others to 'speak' " (p. 85); it is a fictional "agency that does not personally tell or show but puts into the narrator's mouth the language that tells or shows" (ibid.). Although not all texts require the projection of an implied author, Chatman assumes the construct to be as firm a category as the narrator (a narrative instance that may also exist covertly). From our previous discussion of Nanning, it should be very apparent that this definition of the implied author merges Nanning's levels N3 and N4, and one can also clearly see why Chatman has to personalize the implied author, at least terminologically since the reader (Nanning's addressee on level N4, i.e., the "empirical" reader qua recipient of the text) constructs a source or origin of intention and invention which can be filled only by an agency and not by a completely abstract level of analysis. Nanning's N4 is therefore a worthwhile complement to Chatman's implied author because it allows the more abstract codes and conventions (which require no speaker or source) to be described on a non-agency narrative level (such as Niinning's N3). I would even suggest that, for the purposes of analysis, the text in its entirety should be represented as a theoretical level. Chatman takes care of this with his discourse level (L2), but insofar as this is supposed to correlate with the narrator qua fictional enunciator, there are some problems with including titles, epigraphs, and so forth. Now, for Chatman's new model, this constitutes no real difficulty because the abstract-narrator concept generates the text as a whole, including a possibly overt narrator persona, just as the cinematic narrator simply presents the film. In Nanning's schema, however, N2 is an exclusively communicative level of diegesis or utterance, and the "text" would therefore come to be situated between levels N2 and N3. Ultimately, N3 would no longer have a speaker or addressee because it is a purely abstract, reconstructed level, which would effectively dispose of the implied narratee, or implied reader. On level N4, by contrast, the addressee is the real (empirical) reader in her function as reader. Whereas the intentions and meanings of the empirical author have to be gauged from the constructed textual implications (N3) and their possible inconsistencies (resolved by the active interpretive process of the empirical reader on level N4), the reader is, of course, an actual

10 Fludernik Narratology 737 agent within the reading process and a construct only from the literary critic's perspective. This accords with Chatman's analysis, as he agrees that the work acquires a life of its own once it has been produced, undergoing subsequent interpretive violations of its originally intended "meaning." (That such a perspective does not sanction the complete neglect of the circumstances of literary production and the historical reactions of the text's original audience is forcefully demonstrated by Jerome McGann in his contribution to the Sell volume, also under review here.) Nanning's schema, therefore, is a potent response to those critics of the implied author who have always pointed to the residuum of authorial intentionality in the concept. Both Chatman and Nanning also deserve commendation for integrating the actual communicative process in their schemas. Thus, it seems to me, the two models, on the level of reading, accord to a large extent even though they split up the spectrum in slightly different ways and use different terminology. With respect to Chatman's basic story/discourse dichotomy and the positing of a cinematic narrator, the kind of narratology that Chatman promotes in Coming to Terms was implicit in his earlier Story and Discourse. Discourse, in Chatman's schema, is very much a nonlinguistic function, inextricably bound up with the rearrangement of events in the course of narration and equivalent to the process of narrative transmission in the respective narrative media. In this respect, Chatman radically departs from positions held by Genette, Stanzel, and Bal, all of whom restrict the notion of narration (and of a narrator) to linguistic narrating and hence to fiction (and, possibly, oral narrative). Stanzel and Genette are both very explicit about this, emphasizing the verbal medium. Stanzel, reflecting the German narratological tradition, constitutes narrative mediation or mediacy on the opposition between narrative and drama. In this framework drama becomes the equivalent of pure mimesis, while fictional narrative is seen as an intrinsically mixed (impure) genre in which diegesis (not necessarily linked to a narrator-persona) dominates. (Mimetic dialogue and interior monologue are regularly embedded in the narrative diegesis.) Although Bal includes the medium of film in her narrative model, she is careful to define her story level as a reordering and emplotment of the fabula, including focalization, and she reserves the third surface level of the text for the narration and the narrator. Like Genette, Bal therefore links narration with enunciation, leaving the question of filmic narration (i.e., of the text level in films) somewhat up in the air. Chatman entirely reevaluates the relationship between drama and fiction, although he chooses to concentrate on film rather than the theater. In his view, all narrative, qua an artistic form of telling a story, is presented by a narrator, even though the narration may consist en-

11 738 Poetics Today 14:4 tirely of showing. As in Story and Discourse, Chatman therefore locates narrative by definition in the story/discourse dichotomy of temporal rearrangement and presentation (in whatever medium), whereas the standard accounts link diegesis or narrative to actual (linguistic) narration, without bothering to specifically define the connections with film or drama. The incompatibility of these two views, however, is not the result of including drama and film the deep structure of these forms has been acknowledged to be narrative (i.e., plot-oriented) by all major narratologists, the crucial stumbling block, rather, is Chatman's identifying the narrator as an intrinsic element of the discourse. The traditional argument would go something like "Of course, drama and film are narrative genres, but only fiction has narration, diegesis" and then there might be some equivocation about voice-overs or lengthy stage directions. Chatman, on the other hand, asserts that there is narration and hence a narrator in all narrative. Since the cinematic narrator obviously presents a series of shots, that is, "shows" rather than "tells," this stance forces Chatman to redefine diegesis as "presenting," with telling and showing as two equal and interchangeable forms of diegesis or narration.6 Secondly, Chatm.an finds himself forced to withdraw his earlier position on non-narrated events: the presentation of a character's actions, even in the most inconspicuous linguistic form, no longer simulates a disguise of the narrational act; it in fact constitutes the most basic function of narrative presentation. This restructuring of narrative levels and interdependencies has obvious theoretical advantages since it allows Chatman to align the level of the surface structure's temporal dispositions (in the processes of viewing or reading the narrative) with the notion of an instance of narration that "performs" this aesthetic presentation. However, the advantage of these parallels and the theoretical gains of their restructuring are outweighed by the theoretical problems that emerge in their wake. If only linguistic narrative has a personalized narratorpersona, then a question arises as to whether this singularity is due to the function of telling since film or drama cannot tell but merely show. However, telling is not necessarily linked to a narrator-persona, but can be equally performed by a narrative medium as impersonal as the invisible cinematic narrator. Since telling does not require a narrator figure, perhaps it is the medium of language itself that invariably produces a teller-figure; not so: narration can be performed in language purely by means of interior monologue without a teller-figure. 6. Here Chatman comes very close to Stanzel's splitting up of mediacy into mediation by means of a narrator versus mediation by means of a reflector character. Note, however, that Stanzel's distinction is based on linguistic narration or diegesis, of which telling and showing are subcategories.

12 Fludernik Narratology 739 One must therefore conclude that the narrative function, if bestowed on an agent of human or quasi-human constitution, can take only the shape of linguistic telling (or showing). If an agent performs the narration, it must be linguistic due to the cognitive invariants and schemata that relate to human existence and agency. This rule also holds true for the cinema and the theater: if there is a narrative agent, s/he tells in language, and in film and drama this agent can be only an embedded voice-over or dramatic narrator (as in Brecht's plays or Wilder's Our Town). In the theater such a teller closely resembles a frame narrator in fiction because s/he appears on stage ( just like any of the other characters), but occupies a fictional world adjacent to or "above" the world of the dramatic story (with possible overlaps in cases where this narrator is one of the characters of the main story). In cinema, on the other hand, voice-over, if heterodiegetic, comes close to being a vocal frame of oral narration (telling), where the film presentation itself continues the narration in cinematic form and the voice-over can therefore be aligned to the preparatory narrative text that antedates the cinematic showing: "In.... [date] the kingdom of France has been without a. king for several months. The nobility therefore announce that who ever wins a tournament to be held on the following holiday in Paris will be elected king." Homodiegetic voice-over, on the other hand, with its personalized narrator-teller, is always clearly a vocal narrative. Such homodiegetic voice-over can comment both on the events and on the experiencing self, and it does so within a double-layered structure of embedding in which the narrated story is presented in cinematic discourse (showing), but the commentary functions are reserved for the homodiegetic narrative voice-over. This brings me to the next point: Because cinematic narration cannot tell but only show, it is deprived of certain narratorial functions, such as commentary, evaluation, explanation, or address to an (extradiegetic) narratee. The narrative functions that can (easily) be replicated by showing are the basic diegetic ones (Nanning's "erzdhltechnische Funktionen") as well as the rnetadiegetic, self-referential functions. The latter are discussed with great perspicacity in Chatman's tenth chapter, where he analyzes the film adaptation of The French Lieutenant's Woman. Some of the commentary functions that are missing from filmic narrative can be recuperated by screen commentary in the form of printed text on the screen (a feature borrowed from silent film) and, of course, by cinematic voice-over. The function of address, necessarily linked to a "voice" (language) and a speaker, cannot, however, be replicated cinematically. I would suggest, therefore, that if narration is necessarily linked to a narrator (cinematic or otherwise), the term "narrator" makes sense only if this instance can be personalized and if there is a possible narratee. Chatman's model,

13 740 Poetics Today 14:4 it seems to me, would gain much from a relegation of the speaker function to fiction, or linguistic narrative, thus allowing the basic transmissive function to be defined as a deep-structural characteristic of narrative in general. It needs to be noted, however, that such a rewriting would contrary to Chatman's reevaluation of his earlier nonnarrated events locate the specifically narrative (fictional) utterance exclusively in a personalized narrative agent and relegate the simple presentation of the story (in whatever medium) to a "lower" discourse level. One would then end up with the Genettean model of a separate plane of enunciation (with which Bal concurs), although the simple presentation of events would then perhaps have to be excluded from the level of enunciation. That such a course creates even more problems if one has a narrator of the Tom Jones kind will be apparent: How could one possibly separate his enunciation from the narrational function? As long as one postulates the existence of narrational instances and narrational levels, such conundrums are inevitable. As already hinted, Chatman's model also causes some serious problems for the communicative function of (deep-structural) discourse. There is no problem with a separate narrational level where the enunciative process requires an (at least potential) addressee, but there simply cannot be a required narratee for discourse in just any medium. Things work much better, of course, on the implied reader/implied narrator level. Nanning's decision to have no communicative structure on his N3 level (the "work as a whole") very aptly reflects the need to dispose of communication when no personalized narrative agent is implied. Bal (1985) is probably the most consistent and convincing here: narration (in the linguistic medium) goes on the top level, with focalization already determined on the lower, story (Chatman's discourse) plane. Chatman sometimes seems to imply a similar point of view, but nowhere clarifies what precisely happens between discourse and the actual text. Chatman's narratological refinements of focalization have already been published in an earlier article in this journal (Chatman 1986). In Coming to Terms, Chatman proposes to replace Genette's distinction of "who speaks" and "who sees" by the terms slant and filter (p. 144), respectively. If, as is frequently claimed, in certain passages the narrator "sees" through the eyes of a character, then we have the case of a filter, naming the "much wider range of mental activity experienced by characters in the story world perceptions, cognitions, attitudes, emotions, memories, fantasies, and the like" (p. 143). On the other hand, "it makes no sense to say that a story is told 'through' the narrator's perceptions since he/she/it is precisely narrating, which is not an act of perception but of presentation or representation, of transmitting story events and existents through words or images" (p. 142).

14 Fludernik - Narratology 741 Chatrnan therefore proposes the term slant to "name the narrator's attitudes and other mental nuances appropriate to the report function of discourse" (p. 143). Slant therefore equals Nanning's perspective a borrowing from Pfister (1988 [1977]). A filter the presentation by the narrator of characters' thoughts, perceptions, or feelings has to be distinguished from what Chatman, somewhat misleadingly, calls a center. A center is roughly equivalent to a character of central importance, but the term does not imply that one gains access to this character's psyche. 7 Chatman's center cannot, therefore, be equated with a Jamesian center of consciousness. Contrary to Chatman's definition, his later uses of the term slant do not seem to apply merely to ideological perspective, but also to external focalization. Consider, for instance, the following statement: "Camera-eye narration, then, is simply slant without filtration and without narrator's commentary" (p. 149). This seems to imply that slant in connection with the filter function results in psycho-narration. (from the narrator's external perspective), whereas consistent filtration without slant corresponds to reflector narrative, the Jamesian center-of-consciousness technique. Chatman very usefully goes on to distinguish between unreliable narrators and fallible filters. One needs to note that fallibility in a filter (i.e., the character's psychological attitudes) can become narratologically "unreliable" only if unaccompanied by a corrective narrative slant. While the narrator can thus expose the character's fallibility explicitly, the narrator's own unreliability can only be implied since it involves the level of implied author and narratee (p. 153). These distinctions come to fruition in Chatman's discussion of cinematic slant and filtering. Here, the cinematic narrator's presentation is ruled by the camera's perceptual slant, whereas a character's perceptions and sometimes feelings can be implied by a number of well-known cinematic techniques (pp ). Although one might quibble over Chatman's proposed terminology and wish for some additional explication, his distinctions are extremely useful, particularly in discussing film. I now finally turn to Chatman's new integration of narrative with a theory of text types, which deserves special attention. Chatman postulates that there are three types of texts, arguments (i.e., argumentative prose), descriptions, and narratives. Each of these types can occur in mixed forms, with arguments using narrative and description, descriptions using narrative (as in character sketches), and, of course, narratives using both description (to establish the existants) and argument (narratorial commentary). Argument correlates with suasion, 7. Chatman also proposes the term interest-focus for minor characters.

15 742 Poetics Today 14:4 description with a presentation of the properties of things, narrative with action and actants and the formal discourse/story distinction. In his final chapter, Chatman returns to the textual functions of such text types, arguing that the rhetoric of narrative has both an aesthetic and an argumentative side: narrative rhetoric attempts to convince the implied reader (Niinning's empirical addressee) to "accept the form of the work," and it also "argues" for a "certain view of how things are in the real world" (p. 203). Neither of these rhetorical functions is necessarily served by argument (as a text type), but rather by the implied author's imaginative use of codes and narrative conventions. Chatman's tripartite text-type theory allows for a neat recuperation of narrative and other text types as speech acts between the (implied?) author and his audience and here Niinning's empirical author and reader would definitely be valuable theoretical assets. The suggested. text types reorganize the traditional genres into quite different compartments and particularly invalidate the fiction/nonfiction boundary as a text-type criterion. Thus guidebook directions can be pure description (with subsidiary narrative, not arguments), while some poetry (e.g., "Ozymandias") is an argument (sic transit gloria mundi) in descriptive form (p. 10). History in this schema would, I suppose, be nonfictional narrative within a largely argumentative framework. The literary genre that one misses most in this model is poetry. According to Chatman's model, it would always have to be either argumentative (e.g., La Fontaine's Fables [pp ]), descriptive, or narrative. Presumably, poetry can always be interpreted as (philosophical) meditation and hence argument, but this solution is far less satisfactory for poetry than for the other genres. Furthermore, the category of description seems to be much less constitutive of traditional genres than either argument or narrative, and the examples provided for pure description can be reduced to an implicit argumentative macro-context. A guidebook itinerary is, after all, a proposal to do one's sightseeing in a certain way. Even more importantly, Chatman's schema obviously created for written texts entirely skirts the issue of naturally occurring (oral) conversation, in which argument, description, and narrative constitute only some of the possible discursive units or speech acts. Chatman, of course, does not use the term "speech act" or identify his text types with speech acts. Yet if spoken discourse is to be included in a general text theory such as his and I believe that it must be then the theory would have to account for the communicative and phatic functions of discourse. These conundrums are very similar to the problems faced by speechact theory once the constative or assertive utterance is integrated. within a system of speech acts. Constative utterances (indirectly?) perform a number of interactional moves, such as those of explanation,

16 Fludernik Narratology 743 excuse, rebuttal, denial, refusal, exemplification, agreement, and so forth. With description in particular, the question that then arises is whether, like affirmative sentences, they should not be regarded as surface-level phenomena employed for a variety of illocutionary purposes. One might argue that description conveys information about the properties and situation of existants to the same extent that narrative conveys information about their action. It is true, of course, that Chatman did not conceive of his text types as speech acts, so the deliberations above are primarily meant to suggest some directions for further theorizing. Whatever possible amendments one might make to the text-type categories proposed here, Chatman's attempt to conceptualize narration as one text type among others has the enormous advantage of allowing the rhetorical suasion of narrative to be fruitfully compared to the rhetorical strategies of argument in other text types, and this comparison convincingly rescues narrative from its customary vacuum of nonreferential irrelevance. I am now ready to take up the second focus of this review, that is, (narrative) pragmatics. As I emphasized above, Niinning and Chatman, by their respective stipulations of the empirical author/reader and the theory of text types, have broken away from a purely formalist narrative analysis and moved toward one oriented by the reading process, the real-world effects of the text, and the consideration of literature as a social act. Sandy Petrey's Speech Acts and Literary Theory, while coming from an entirely different perspective, covers some of the same ground and implicitly suggests further developments along similar lines. Petrey's handy volume is eminently lucid and readable. It provides an excellent introduction to the J. L. Austin of How to Do Things with Words (Austin 1980 [1955]: pt. I), then applies the Austinian speech-act model to literature (ibid.: pt. 2), and finally addresses the deconstructionist criticism of John Searle by showing how Austinian literary criticism, which Petrey holds to be the genuine, unadulterated model of speech-act theory, could in large measure accord with deconstructionist tenets as well as where, and why, there are also some irreconcilable presuppositions in those two theoretical universes. Austin's great discovery, according to Petrey, was the fact that one can do things by merely saying, or uttering, certain words. This discovery originally formulated in the constative/performative dichotomy was later complemented by the almost deconstructionist demonstration that constative utterances are themselves intrinsically performative, that is, also speech acts. Petrey situates Austin between two extremist views on language. According to the first, language con-

17 744 Poetics Today 14:4 sists just of (or in?) words. Several grammatical models, from Benveniste's to Jerrold Katz's, are discussed in this context. The second view described by Petrey sees language as mere action (or perlocution). Austin's stroke of genius, according to Petrey, was his creative functional alignment of locution with perlocution on the basis of a set of conventions linking the two, an alignment that produced the now familiar concepts of illocution and felicity conditions. Petrey has a tougher nut to crack when he tries to sell Austin's speech-act theory as a literary-critical methodology. In order to put Austin to such a use, one must deconstruct his pronouncements on literary language (which as is well known he characterizes as "nonserious," "parasitic," and an "etiolation" of language [ibid.: 22]). Austin himself has suggested that the "abnormality" of literature could be regarded as a convention, and Petrey proposes, for instance, that one could then understand commands in poetry (e.g., "Go and catch a falling star") as an invitation to interpret such utterances (p. 52). Since literary texts become what collectivities make of them (p. 65), literature can be said to have illocutionary force. The example given here is the debacle over Salman Rushdie's Satanic Verses, which, by Western conventions, reads as a literary utterance, but by an entirely different set of conventions reads as blasphemy. Illocutionary force in such a context thus becomes fundamentally and intrinsically dependent on contextually or culturally determined conventions and demonstrably independent of authorial intent. (The issue of intentionality is discussed more fully in Petrey's analysis of Searle's position in chapter 4.) Writing, like speaking, therefore "performs" (p. 56). Petrey then backtracks to discuss various adaptations of speechact theory. In Richard Ohmann's (1971b) "Speech Acts and the Definition of Literature," for example, literature is presented as "quasispeech acts" in which fictional personae pretend to refer to a real world, but do not actually refer. Ohmann suggests that literature's illocutionary force can be interpreted as mimetic an explanation that may work for realist narrative or drama, but seems singularly unsuited to poetry. Ohmann (1973) extends this viewpoint in "Literature as Act" from individual sentences as quasi-speech acts to the global speech act of a literary work, which "participate[s] in the imaginative construction of the world" (ibid. : 107). Mary Louise Pratt (1977), on the other hand, attempts to erase the difference between literary and nonliterary language, a move that necessarily renders literature without its illocutionary force. For Pratt, literature is to be understood as part of a social convention and as comparable to quasi-speech acts in nonliterary language. However, Pratt then ends up locating the literary speech act that occurs between author and reader as equivalent to any other act of communication, a

18 Fludernik Narratology 745 theoretically untenable position with regard to the intentional fallacy exposed earlier by Petrey. More important still, at this point Petrey makes by far the most significant comment of the book: One speech-act insight is that, because language performs its users' collective engagement with each other, looking at language means looking at what words are doing with, for, and to human beings. That insight does not, however, force all forms of communication into the model supplied by oral exchanges between two individuals in the same space. That literature does indeed "take place in a context" need not lead to the assumption that its context is the same as those that put speaker and audience in one another's presence. (P. 77 [my emphasis]) This statement implies that written texts can be speech acts, that is to say, conventionally ordered effects produced by (written language) between writer and audience, without at the same time necessarily involving such communication as occurs in face-to-face conversation.. Unfortunately, Petrey never elaborates on this crucial insight in relation to literary texts. He does, however, make the very important point that any kind of written communication is necessarily removed from direct authorial control since the text will be read according to the conventions of its audiences, not those of its production. Hence, in writing, "the speaker-hearer model breaks down even though speechact theory does not" (p. 78). Petrey then moves on to consider the use of poetic language to create an illusion of a specifically "poetic" language that is different from ordinary usage (the "Poetic Language Fallacy" in Pratt's model), reinterpreting this illusion as the intended effect of the "Poetic Language Convention" (p. 80). It is therefore within the conventional speech act of literature (representation) that individual literary speech acts acquire (rather than lose) illocutionary force when used felicitously in accordance with literary conventions (p. 79). These suggestions constitute an immense advance over earlier proposals on how to apply speech-act analysis to literature. Petrey's concern here is with the text as a whole. Although what he says about textual effects and conventions can only be applauded, there are nevertheless some interesting inconsistencies in his application. Like Pratt, Petrey jumps from the literary sentence to the literary utterance, that is, to text. This move was linked in Pratt's account to the explicit comparison of conversational "turns," in which narratives are said to constitute one such "turn." Such an approach works very well for conversational narrative although one would have to disregard the listener's encouraging or appreciative phatic contributions; the model does not, however, work as well for poetry, novels, or drama since even during a dramatic performance in the presence of an audience, no standard face-to-face communication can be said to take place. Petrey says as

19 746 Poetics Today 14:4 much, but fails to draw the obvious conclusions. If communication is not the basic model for literary speech acts, how does one justify applying speech-act theory to it? The problem, as far as I can see, is not the result of using the performative convention to characterize literary reading processes there can be no doubt that a published piece of writing constitutes an act that becomes interpretable only within specific societal conventions; rather, what is at issue here is whether one can extrapolate from speech-act theory that concentrates on utterances of very minor length (a few sentences at most) to complete texts, which if one follows Petrey's "communication fallacy" to its logical conclusion are, precisely, not utterances. This conundrum may be responsible for the fact that most successful applications of speechact theory to literary texts have been made in the realm of drama or fictional dialogue, where characters' utterances like characters' psyches in one method of psychoanalytic literary criticism can be profitably subjected to discourse-analytical investigation. Petrey presents Stanley Fish's (1980) essay "How to Do Things with Austin and Searle" as an example of such a successful application of speech-act theory to literature. Fish's interpretation of Coriolanus underlines the infelicity of Coriolanus's noninteractive locutions, which therefore fail to acquire the necessary illocutionary force. By contrast, Shoshana Felman's (1983) Austinian reading of Don juan's cavalier disregard for promises, which attempts to combine speech-act theory with Lacanian psychoanalytic literary criticism, is severely criticized by Petrey for its failure to account for conventions and for its regressive reinstatement of the constative I performative dichotomy. Petrey redefines global literary communication in terms of audiences' readings, which operate according to conventions. Where this becomes questionable is in the comparison of the whole text with a speech act, a comparison that derives from the "turn"-equals-text equivalence mentioned above. More recent analyses in the linguistic study of conversation suggest that conversational strategies cannot be described in terms of sentences or "turns." All speech-act analysts, including Grice (who is unaccountably missing from Petrey's study), use sentences for their examples, but actual speech acts may span several sentences, even including an entire turn. Moreover, sentences are syntactic units of the langue, and actual conversation has to be divided into tone units, idea units, or similar unconventional categories Discourse analysis, in its close observation of actual conversational output, has had to invent a new unit of parole. Wallace Chafe (1979) calls these idea units, and Jan Svartvik proposes, for instance, in the English corpus of conversation (Svartvik and Quirk 1979), tone units. Jan W. F. Mulder and other collaborators in the Freiburg project on sentential ("sententielle") analysis (organized by Professor Herbert Pilch) have devised even more specific models (Halford 1989; Halford and Pilch 1990).

20 Fludernik Narratology 747 There also exists no definition of "speech act" that would allow an uncontroversial categorization of a given conversation into individual speech acts except maybe--on a sentence-by-sentence basis. If, as some have suggested (see, e.g., Ducrot 1984), the linguistic analysis of utterances should be separated from their speech-act performance, then larger units of, say, argumentation, counter-argumentation, rebuttal, exemplification, persuasion, self-presentation, insinuation, not to mention phatic agreement, bolstering the interlocutor's self-esteem, and whatnot, might come into focus. It is unclear in the classic version of speech-act theory how one would deal with such issues, whether on the level of indirect speech acts or with respect to speech acts on a suprasegmental level. It is quite a different matter to talk about promises voiced by a variety of simple utterances than it is to discuss a text which tells a story, stimulating the reader to construct an imaginary world. And what about the thematic and ideological concerns raised by the story or its telling, or the text's task of entertaining the reader and providing some aesthetic satisfaction? Although positing such a collection of suprasegmental speech acts within the literary product as a whole goes a long way toward justifying such an application of speech-act theory, it in no way justifies identifying the text as one (and only one) speech act. Such deliberations do not invalidate the relevance of conventional interpretive strategies for analyzing literary texts in terms of Grice's maxims or Sperber and Wilson's relevancy conditions. Those theories are, in fact, much more promising models for an application to literature since they see language production and processing as a continuing interactive process rather than as one global act. The question of literary communication, which is equally raised by the Gricean model, can also be handled in much more suitable terms. In individual speech acts, as outlined by Austin and Searle, the emphasis always falls on the transmission of some sort of information (e.g., about the speaker's intention), or on the attempt to elicit a response, or on the speech act's even changing the current state of affairs. Play, pretense, deliberate insincerity, and the like as Petrey notes, too can then only be handled as performative flaws. Within a cooperative, relevanceoriented approach, on the other hand, one can consider such cases as conventionally accepted strategies of the language game, and one need not even strip literature of all communicative features; part of the game can be the implicit and explicit handling of ideas and other truly referential matters. Such a view also moves more easily from smaller to larger, and from nonfictional to fictional, texts since the cooperative principles or conventions can change according to a text's length, genre, and degree of fictionality as well as in accordance with narrative level. For example, there are interesting differences in the available choices for violating maxims of cooperation between the nar-

21 748 Poetics Today 14:4 rator and the (implied) author. As Pratt (1977: chap. 5) has noted, the implied author can only flout maxims (i.e., violate one maxim in the interests of another), whereas the narrator can be unreliable and can therefore actually violate maxims without any possible recuperation by way of flouting. Although this is what Petrey suggests on the basis of speech-act theory, it is hard to see how these desirable results could be achieved on the basis of Austinian theory without making a major argumentative leap of the kind that I have noted. Before moving on to Roger Sell's Literary Pragmatics, a few words are due on the remainder of Petrey's book. The problems with a speechact analysis of literature become most striking when Petrey attempts to apply that theory to fiction and poetry. His discussion of realist fiction (most of chapter 7) is organized around Balzac's 1830 work, Adieu, and covers Pierre Gascar's realist and Shoshana Felman's ideological readings of that novel (Gascar 1974; Felman 1975). The debate centers on the issues of historically accurate representation literature as history (Gascar) and the textual properties of intertextuality and the failure to achieve reference (Felman). In Adieu the protagonist, Philippe, attempts to recreate the traumatic historical scene during which his beloved Stephanie had lost her mind, but overly successful realism has fatal consequences when Stephanie suffers a repetition of her trauma and dies. Philippe's cure was too radical and killed the patient. Petrey praises Felman's debunking of the illusion of realism, but notes the historical situatedness of literary convention and therefore of the literary speech act of realism: "Language performs in history even though it deconstructs itself outside it" (p. 128). Petrey also briefly considers Whitman's Leaves of Grass, but, curiously, fails to mention any text-internal speaker function. Although Petrey's phrasing is admittedly careful, the text's illocutionary force is nowhere attributed to a fictional speaker (created by Whitman) but always to apparently the "empirical" author. Thus, Petrey refers to "Whitman's performative lines" and declares that "Whitman's remarkable illocutionary lexicon effects the same democratic spirit he obsessively extols" (p. 112). The latter claim suggests that Petrey is here referring to an illocutionary act by Whitman and not to that of a fictional speaker after all, the anaphoric he can only be coherently linked to the antecedent NP Whitman. With respect to the Balzac novel, the absence of any discussion of literary communication strikes me as even more baffling, and it is here that Nanning's and Chatman's books provide an important corrective to Petrey's literary applications. In particular, Petrey's speech-act theory seems to coincide with communication on Nanning's N4 level, that is, the communication between the empirical author and the empirical reader, since Petrey's analysis concerns itself with the literary opus in its totality (that is to say,

22 Fludernik Narratology 749 with what Niinning calls his N3 level). Intrafictional communication, on the other hand, definitely reflects a different kind of communication, or indeed different kinds of communicative models. One cannot possibly ignore the narrator-narratee communication, modeled directly on the everyday communicational system, nor does Petrey offer any insights on the vexed "narratorless narrative" question, where the issue of fictional communication becomes even more critical. Irrespective of the positing of a narrator for texts without a distinct narrator-persona always a theoretical conundrum in the absence of an agreed-upon definition of "narrator" intrafictional communication invariably and demonstrably occurs whenever the text has such a fictionalized narrator-persona. Speech-act theoretical rejections of the narrator are noted by Chatman in Coming to Terms, and he has also previously analyzed the speech acts ancillary to narration (Chatman 1978: ). It seems obvious to me that this is the place to start with a speech-act theory of fiction, and one can only hope to see some future efforts made in this direction. In his last two chapters, Petrey arbitrates among Derrida, Searle, and Austin, demonstrating that while deconstruction deliberately ignores (social) context, Austin can be faulted for ignoring the iterability of the performative act as one of its constitutive features. In particular, Petrey points out how Derrida never really concerns himself with illocution (he is more interested in the signifying process that results in locution), and how the felicity of speech acts, namely, successful illocution, does not figure in his debate with Searle. Petrey's analysis here is a highly commendable attempt to demonstrate that the arguments on both sides of this famous debate fall flat, each party being concerned with the reverse side of the same coin. For Searle (and Austin), language produces perlocutionary effects (such as reference) by means of illocutionary force (granted by convention); Derrida, on the other hand, examines the conditions under which such effects and forces can be initiated in the first place, in the study of signification processes which result in locutions. Without locutions there can be no speech acts, but as Petrey correctly insists locutions are also never used outside a system of conventional utterances a system that, most of the time, ensures successful speech acts as well as (I would add) less conventionally circumscribed, multiple interpretations of complex written language acts. If Petrey used a global method to apply speech-act theory to literature, Roger Sell's collected volume of papers from a 1988 conference in Abo, Finland (at Sell's own university), presents a number of very different approaches, both theoretical (Engler, Enkvist, Hess- Liittich, Pilkington, Schaar, Sell, Steen, Van Peer) and practical (Stern-

23 750 Poetics Today 14:4 berg, Ben-Porat, McGann, Verdonk, Watts). Literary Pragmatics includes quite a few excellent contributions, among which Gerard Steen's piece on metaphor and Ziva Ben-Porat's analysis of Jerusalem cliches in Ivrit poetry probably take pride of place. I will discuss the general methodology advocated in the volume, pointing out comparable tendencies elsewhere and briefly addressing one or two examples. Within linguistics, pragmatics is that area of language study which concentrates on the uses, rather than the formal properties, of language qua Saussurean parole. Although such an orientation would seem to take as its analytical starting point the everyday language that we all speak, which according to Chomsky is performatively inadequate to the rigorous standards of linguistic competence, there has actually been a tendency in pragmatics to look for system-related properties of utterances, an orientation that has resulted in the adumbration of a langue of utterance. (Such a study of the systemic properties of the parole has already been proposed by Josef Vachek [1964] and, of course, correlates well with Louis Hjelmslev's concept of the "form" here the "form" of the parole.) Pragmatics spans a very broad spectrum of linguistic phenomena, some of which overlap with such affiliated disciplines as sociolinguistics and discourse analysis (for the standard introduction to pragmatics, see Levinson 1983). One major concern of pragmatics has always been the context of the utterance, that is, the effects of the communicative situation on the formal and semantic properties of language. For example, Peter Cole's two edited volumes of essays on pragmatics (Cole 1978, 1981) include pieces on reference, deixis, presupposition, inference and implicature, irony, and specific problems, such as the uses of (and presuppositions activated by) the lexeme almost. In addition to these more formal approaches, many recent studies on tense and aspect locate themselves in pragmatics (e.g., Hopper 1982), and the Journal of Pragmatics offers yet another range of methods and applications. What are the possible connections between pragmatics and literature? This is a question faced squarely in Sell's introduction to the volume and subsequently elaborated in the first essay (an excellent theoretical statement by Nils Enkvist, to whom the book is dedicated) and by the contributions of Richard J. Watts, Adrian Pilkington, Balz Engler, and E. W. B. Hess-Liittich. These essays view the pragmatic approach to literature as superseding the formalist paradigm, in which linguistics was used descriptively to define the literary properties of the text. Literary pragmatics, by contrast, holds that the quality of literariness is not an essential property of a text, but a dynamic attribution effect circumscribed by the conventions under which members of a specific culture operate when reading, interpreting, or talking about

24 Fludernik Narratology 751 literature. Whereas prior linguistic analyses of literature on the formalist model had emphasized the self-sufficiency of the literary work of art, pragmatics brings the real author and the real reader into view again, so some rethinking of the intentional and affective fallacies becomes necessary. Literary pragmatics also claims to combine linguistics and literature as true partners of roughly equal status, whereas for malist literary criticism either treated linguistics as a handmaiden to the literary project or allowed linguistics to usurp literary concerns. Despite the initial disappointments over the results of using linguistics to explicate literature from the famous Fowler/Bateson debate (Bateson 1966a, 1966b, 1967, 1968; Fowler 1967, 1968; cf. Fowler 1977, 1979), through the various analyses of "Les Chats" (Jakob son and Levi-Strauss 1962; Riffaterre 1966; Posner 1982; see also Werth 1976), and on up to the New Stylistics and its failings (see Mair 1985) linguistic approaches to literature have continued to flourish unabated, and with increasingly attractive results. This breakthrough is largely due to recent developments within linguistics itself. The exclusively formalist phonological or syntactic models that played such a prominent role in early "applications" to literature were particularly ill-suited to literary investigation, especially when rigidly employed as a ruling methodology. With the advent of discourse analysis and pragmatics, a much less cumbersome terminology and a wide range of pluralistic methods and approaches became available, which could be adapted to practically all levels of the literary text sentences, paragraphs, the narrator's presentation, or the characters' interaction. In the wake of Grice's formulations and textual studies on cohesion., coherence, presupposition, and relevance,9 the suprasegmental strategies of both (creative) writing and textual processing (reading/listening comprehension) could be described in a linguistic framework that helped to demonstrate how texts rhetorically manipulate their readings by implying specific presuppositions, establishing topics, or flouting cooperative maxims. As Sell argues in his introduction, such analyses restore the reader's status as more than a purely theoretical "implied instance" of ideal qualities. Within a linguistic theoretical framework, inference and implicature can become normative conventions of a pragmatic langue that determines the actual utterance (or text), and such a system of pragmatic conventions can be interposed between the linguistic medium, on the one hand, and the preprogrammed responses and interpretations of the actual reader, on the other. Nils Enkvist, for instance ("On the Interpretability of Texts in General and of Literary Texts in Particular"), proposes the notion of inter- 9. M. A. K. Halliday and Ruquaiya Hasan's (1976) lucid volume has been highly influential.

25 752 Poetics Today 14:4 pretability and supplements it with contextual acceptability or appropriateness (p. 7) to include the more specific cultural norms of certain contexts. Interpretability relies in turn on intelligibility (i.e., the recognition of phonological, lexical, and morphemo-syntactic structures), comprehensibility (i.e., the assignment of meaning), and intelligibility proper, which incorporates the utterance/text into a scenario, or a. state of affairs within which an utterance "makes sense" (ibid.). Interpretability is highly useful as a concept for language in general since it makes explicit what many syntacticians have implicitly presupposed when asking people to evaluate the acceptability of constructed sentences: sentences that are acceptable allow for the construction of a. possible context within which their utterance would make sense. Interpretability also readily explains our ability to make sense of exchanges only when we are aware of the topic under discussion much over heard conversation remains completely opaque unless one is informed of the relevant topic, and even written instructions can be downright unreadable without an accompanying diagram or a topic heading to guide the uninitiated. Or think of having to "translate" a legal document in which many words just don't make sense on a literal level, but only if one knows the jargon or is privy to the relevant circumstances. Knowing how to relate a situation to the text and to create a situation from a text requires a continual see-sawing that comes very close to the processing that Roger C. Schank and Robert P. Abelson (1977) have analyzed within what is now called frame theory. Within ordinary language use, Enkvist proposes, the situations evoked by a text will be of manageable proportions if there is ambiguity, only a few alternative interpretations will be possible, and the overt context will usually make one, and only one, relevant. If ambiguity persists, attempts are made to resolve it: "When you said... did you mean to imply...?" In literature, on the other hand, several noncontradictory interpretations may be equally appropriate and relevant in the circumstances (see also Pilkington, pp ), and (obviously) one cannot simply ask the writer (or the narrator!) to clarify. Pragmatics has also rescued a historical engagement with literature from its present-day irrelevance and promises some very interesting structural observations as well. The effect of aesthetic pleasure derived from reading, for example, can be seen as the result of a process of matching and recognition, or as the result of unexpected turns, the repetition of the known, and so forth (Enkvist, p. 25). Other pragmatic projects include the investigation of the politeness levels of texts (cf. Sell's essay in this volume and earlier works by him), a textual level that also correlates with textual opacity and the concomitant challenge to the reader's interpretive persistence and ingenuity. Literary pragmatics has already established itself as a separate field of inquiry

26 Fludernik Norratology 753 (Hickey 1989; Sell 1991) and promises to contribute a number of incisive insights on the workings of literary texts. I will conclude by briefly touching on some of the applications of these pragmatic models to literary texts. Meir Sternberg's "How Indirect Discourse Means: Syntax, Semantics, Poetics, Pragmatics" delivers one more shrewd assessment of the recontextualizing features of narrative quotation in addition to his earlier articles (Sternberg 1982a, 1982b). With his well-known agility, Sternberg keeps his balance on the tightrope between form and function, between opaque and transparent readings of the indirect. He discusses the conventional implications drawn from the syntactic forms available to the quoter and provides some new counterexamples to Ann Banfield's (1982) claims about indirect discourse. Gerard Steen's excellent "Understanding Metaphor in Literature: Towards an Empirical Study" provides some surprising insights into the workings of metaphor. Metaphor identification is apparently "dependent on both the reader's actions and his attitudes and knowledge" (p. 115). On the basis of Tanya Reinhart's (1976) distinction between focus and vehicle interpretation in metaphor, in which literary metaphors are said to have a tendency toward explicit vehicle interpretation, Steen suggests that such vehicle prominence may in fact constitute a cue for explicit metaphor interpretation, whereas focus-centered, metaphor is not perceived as metaphorical. Thus metaphorical reading can be linked to the mediations of interpretive communities and the reading attitudes they propose. Steen bases his observations on the results of the empirical study of literature 10 initiated by S. J. Schmidt (1980, 1982a, 1982b) and outlines some experiments that could enrich the results already available from a number of empirical studies analyzing metaphor interpretation.' 1 I will conclude with what I believe to be the best essay of the volume (regrettably skipping other valuable contributions, such as Jerome McGann's interesting discussion of the publishing circumstances of Byron's "Fare Thee Well"). Ziva Ben-Porat's "Two-Way Pragmatics: From Word to Text and Back" analyzes the characterization of "Jerusalem the City" in nineteenth-century (Eastern) European Hebrew literature and its adaptation to the historical facts of twentieth-century immigration to Israel, that is, the confrontation between the mythic description of Jerusalem and the reality of the twentieth-century place. Although some very limited modifications of the symbolic "canon" 10. This is very confusingly abbreviated as ESL, an acronym generally used for "English as a Second Language." 11. Another study along these lines is that of Willie van Peer (1984), who analyzes the empirical validity of the poetics of deviation.

27 754 Poetics Today 14:4 can be shown to occur in contemporary poetry, these are quantitatively negligible in comparison with the continued evocation of themes that are quite inappropriate to the present experience. Thus, Naomi Shemer's "Jerusalem the Golden" introduces the environmental ingredients of clear mountain air and the scent of pine trees to the representational components, but these have little historical (allegorical) significance. On the other hand, cisterns and the marketplace (absent from the classical paradigm) are introduced not in the interests of realism (i.e., representation) the marketplace is "empty" and the wells are "dry" but in order to replace the ruined walls and destroyed palaces of the canon (p. 150). Even Byron's "Jerusalem" becomes affected in translation by the powerful concepts of the traditional paradigm. Ben-Porat supplements her analysis with results from empirical experiments on associational reflexes among both adults and school children. At issue were people's associations with the autumn season. In reality, there is no observable seasonal difference between Jerusalem's autumn and winter, and fall if conceptualized as an experience-- represents relief from the summer heat and a consequent reawakening of nature. Yet in Ivrit poetry the European experience of autumn as the dying of nature, as the season of harvest, with its literary associations of "sweet sadness" (pp ), persists. Hebrew poets not only developed linguistic ways of distinguishing between autumn and winter, they also took over the European literary paradigm according to which autumn signifies sadness and decay. Thus the poplar, which does not grow in Israel, nevertheless flourishes in Israeli poetry as an image of autumnal sadness (p. 155), and adults have been found to associate the season with these poetic topoi. The most frequent association with autumn is the withering of leaves, followed by winds, clouds, sadness, migrating birds, mists, parting, shorter days, and, much less frequently, the high holidays, apples, and citrus fruit associations that would correspond to the physical reality of autumn in Israel today. Children, on the other hand, associate autumn with largely realistic items and events, such as the flowers called "squills," coolness, the Jewish high holidays, the beginning of the school year, the first rain, wigtails (birds that winter in Israel), withering leaves, clouds, cotton fields, apple picking, and citrus fruit. However, even the children's most frequent association with autumn squills turns out to be a literary allusion, since squills occur in a common rhymed maxim which every child learns in kindergarten and which pairs autumn and squills. Their principal association therefore relies on a mnemonic collocation of literary origin (pp ). The link between literature and life could not be closer. Vaheed K. Ramazani's Free Indirect Mode: Flaubert and the Poetics of Irony can also be regarded here as another variation on the pragmatic analy-

28 Fludernik Narratology 755 sis of literature. This is not a study of free indirect discourse as such., and the formal definition of the mode is clearly inadequate by linguistic standards. (The bibliography fails to list any studies of FID other than the most traditional ones. In fact, Ramazani's manuscript seems to have been completed as early as 1985, and he has therefore missed some interesting recent contributions to narratology, among them Bal's [1985] Narratology.) Ramazani's slim volume is less concerned with the formal delimitation of free indirect discourse than with its uses in Flaubert's Madame Bovary and L'Education sentimentale. This is also a study of irony, and although Ramazani offers little in the way of precise definition, he nevertheless makes some remarkably good points about Flaubertian irony. Thus he notes that although Emma is the victim of Flaubert's (i.e., the implied author's) irony, the object of irony is not Emma, but rather the kind of romantic cliché that rules her life; additionally, Ramazani proposes that besides being the victim of irony, Emma also functions as the medium through which that irony attains its object in the novel (pp. 2-3). Such formulations strike me as decidedly superior to most explanations of irony on the basis of contradiction alone, and they could be profitably refined by integration with the triangular models proposed by Ross Chambers (1989) and W.-D. Stempel (1976). Catherine Kerbrat-Orecchioni's (1976, 1977, 1980) concept of irony signals, which Ramazani discusses at length (pp ), is intelligently supplemented here by a more contextual, genrerelated frame of reference, but the very clear parallels between the status of her irony signals in the co-text and the signals of free indirect discourse in that co(n)text unfortunately remain unaddressed one of the consequences of Ramazani's lack of interest in the form of free indirect discourse. Ramazani is at his best in chapters 3 and 4, where the functions of free indirect discourse are closely related to the properties of Flaubertian style and irony, as described in Jonathan Culler's (1974) and Naomi Schor and Henry F. Majewski's (1985) books on Flaubert. In particular, Ramazani manages to bring literary, pragmatic, and philosophical concepts to bear on the unique place of free indirect discourse within narrative, as the following passages show: The free indirect mode therefore inscribes its reversible pretense of empathy within an overarching metairony [sic] of narration and reading. The hierarchy of reflector and narrator, subjective and objective, necessarily subsists at the level of corrective irony. But in foregrounding both a uniformity of registers and a disparity of ideologies, the disembodied voice of the free indirect mode also gives rise to textual self-consciousness, or romantic irony. (P. 115) Thus, it is in the free indirect mode that romantic irony best objectifies the dissonance between meaning and experience.... Ultimately, Flaubert's response to this fall into language will be to fully embrace the alienating

29 756 Poetics Today 14:4 materiality of words not to lower poetry, but to raise the positivity of speech to its highest power. The poetic qualities of Flaubert's overdetermined style are potentially most perceptible when irony in the free indirect mode dramatizes the autonomy of the signifier. Reading becomes a. dialectic between radically agonistic readings, an esthetic pantheism alternating between mimesis and metaphoricity, culture and nature, self and self-destruction. (Pp ) It is in the linking of irony with textuality that Ramazani comes closest to a pragmatic approach to literature, and yet, at the same time, the deconstructionist tenor of his remarks seems to conflict unresolvably with the spirit of literary pragmatics. Ramazani therefore returns us to Petrey's comparison between Derrida and Searle, and one can see that Ramazani is obviously much more interested in language texture and a text-internal reading strategy than in actual readers, historical or generic influences, or a precise linguistic description of literary texts. Nevertheless, in his study of irony, of romantic cliche and stylistic infelicity, Ramazani is addressing issues that equally concern the more empirically minded literary pragmaticists, who are grappling with them in a more down-to-earth, commonsensical manner. If Flaubertian irony remains elusive with a verigeance this is due less to any lack of awareness of the pragmatic parameters than to Ramazani's resistance to the rigorous empirical tools of narratology and/or linguistics. Since I have covered a number of widely disparate topics, some recapitulation may be in order at this point. I have grouped this selection of recent publications according to two related topics, communication and applications of pragmatics to literary texts. Since questions of deixis and the context of utterances are central to linguistic pragmatics, the issue of communication can be regarded as subsidiary within the general pragmatic frame. However, as I tried to explain in the first part of this article, communication in literature occurs on various textual and nontextual levels. Moreover, one cannot simply lump all of these communicative processes within a single general model of faceto-face communication that takes linguistic enunciation or the pattern of information transmittal between a sender and a receiver as its paradigmatic frame. Communication in narrative, for instance, can neither be reduced to the level of intrafictional dialogue, or the dynamics of the narrator-narratee relationship, nor be entirely identified with a "reader-response" model in its standard accounts. Speech-act theory, a linguistic precursor of pragmatics, can help to clarify the rhetorical and macrostructural qualities of literary "communication," and Chatman's groundbreaking application of such an analysis to literary and cinematic narrative should stimulate more extended analyses in the

30 Fludernik - Narratology 757 same direction. On the text-internal, microstructural level, however, such speech-act theoretical concerns lose their relevance and need to be replaced with more traditional narratological models, models that will also have to reflect further consideration of the mimetic status of narratological concepts of communication. Literary pragmatics, on the other hand, may become a valuable ally of narratology precisely in the murky circumference of narratology proper, in the area of generic, diachronic, empirical framing within which texts operate. Seemingly antipragmatic approaches to literature, such as textualist antihistorical readings in the manner of deconstruction, can in fact be recuperated within a literary pragmatics to the extent that such readings operate within generic conventions, and. one possibility for reading is to read in a textualist manner. A pragmatic analysis of literature therefore allows one to accommodate the empirical effects of deconstructionist readings rather than excluding all contextualizing frames for this common activity. I trust that this presentation of recent work in narratology and its applications has left the reader with the measured optimism that I myself feel. The introduction of pragmatics into literary studies seems to promise excellent opportunities for a fruitful dialogue between a number of related disciplines. I have additionally suggested that more conventional narratological frameworks will continue to be of great practical use and that they can profit immeasurably from the pragmatic shift within linguistics and literary study. In contrast to other current evaluations of the narratological climate, mine would forecast less invariably rainy weather than "partly cloudy" conditions. Like all such forecasts, this may prove overly optimistic, but we won't know until the next report is in. References and Related Works Austin, John L Philosophical Papers, edited by J. 0. Urmson and G. J. Warnock (Oxford: Clarendon Press) [1955] How to Do Things with Words. The William James Lectures Delivered at Harvard University in d ed., edited by J. 0. Urmson and Marina Sbisa. (London: Oxford University Press). Bal, Mieke 1981 "Notes on Narrative Embedding," Poetics Today 2(2): Narratology: Introduction to the Theory of Narrative (Toronto: University of Toronto Press) "The Point of Narratology," Poetics Today 11(4): Banfield, Ann 1982 Unspeakable Sentences: Narration and Representation in the Language of Fiction (Boston: Routledge and Kegan Paul). Bateson, F. W. 1966a "Editorial Postscript," Essays in Criticism 16: b "Editorial Postscript," Essays in Criticism 16: