1It is small wonder that narratological categories should have been derived almost exclusively from fictional narratives and that, in return, they should generally have been applied to the study of narrative fiction. After all, narratology developed as a branch within poetics and its practitioners have been literary scholars rather than linguists, sociologists or even historians. Still, it will appear as unsatisfactory to many that the theoretical discipline most immediately concerned with narrative as such has more or less restricted itself to what is only a segment of the narrative domain — even if, by common consent, it is the segment most rewarding to narratological analysis. It is only in the wake of the "linguistic turn" in historiographic theory and the debate on narrativity and fictionalization in historical writing as carried on by theorists like Hayden White, Paul Ricoeur, Paul Veyne or Michel de Certeau that the relevance of narratological categories to factual narratives,1 especially to the writing of history, has begun to be investigated (Rigney, 1990; Carrard, 1992). At the same time, literary narratologists like Gérard Genette (1990) and Dorrit Cohn (1989; 1990) have employed these categories to probe the boundaries between factual and fictional narratives, drawing on pioneering work by Käte Hamburger (The Logic of Literature) and, in their own ways, on Paul Ricoeur's monumental synthesis of literary narratology and historiography (Ricoeur, 1984-88). In the long run, these efforts amount to a final abandonment of narratology's structuralist orientation and to an opening up of the discipline to the concerns of discourse typology and of cultural hermeneutics. They can further be seen as addressing two grievances, of which the second is in many ways a consequence of the first. For notwithstanding narratology's almost exclusive concern with narrative fiction, the theoretical concepts derived from narratological analyses of fictional texts have been invested, more often than not, with overall validity for fictional as well as for factual narrative. This is the background to Genette's postulate that "... fictional narratology cannot indefinitely postpone asking whether its results, that is, its methods, apply to a domain which it has never properly explored but only silently annexed, without examination or justification" (Genette, 1990, 756). Surprisingly, however, the question of the relevance of narratological criteria for a distinction between factual and fictional narrative has not been pursued any further within the framework of the discipline since what must be regarded as the seminal essays by Cohn (1990) and Genette (1990) — not to speak of the earlier philosophical radicalization of the concerns of narratology and of historiographic theory in Ricceur's Time and Narrative (Ricoeur, 1984-88). On the contrary, the latest developments in narratology (Fludernik, 1996) tend to dismiss the issue as irrelevant for narratological purposes.

2The present paper proposes to discuss the applicability of narratological categories (especially tense, mode2 and voice) to factual narrative. As a wider-ranging discussion of several types of factual narrative discourse would far transcend the scope of this paper, I shall concentrate on historiography including biography and autobiography as the fields which have provoked most of the existing work on the subject. Although Monika Fludernik has tentatively excluded historiography from the domain of narrative in her impressive Towards a 'Natural Narratology' (1996) because of what she considers its lack of "experientiality," I should very much like to retain historiography as a paradigm of factual narrative. For one thing, it seems to me that the emphasis on "experientiality" as the essential element of narrativity amounts to privileging stream of consciousness types of narrative in the same way as the traditional emphasis on plot (as criticized by Fludernik) implies privileging report and "realistic" narratives. Moreover, in many historical accounts, "experientiality" is not really lacking, since history is also concerned — even if in an oblique manner — with the motivations and expectations of historical figures. I should argue even further that history participates in the category of "experientiality," as all individual experience is embedded in historical and cultural experience. If Fludernik, by adopting a constructivist point of view, tends to align experientiality with fictionality, narrativist theories of historiography have also stressed the fictionalization of the historian's account which occurs as a result of "emplotment" (cf. White, 1973, 1978, 1987).

3Applying narratological categories to historical narrative involves advantages for both narratology and historiography. As Wolf-Dieter Stempel has argued, historiographic texts provide more favourable opportunities for an analysis of the "elementary conditions of narrative" ("die elementaren Bedingungen des Erzählens") (Stempel, 1973, 326; my translation) than fictional narratives. Thus, it appears that the application of narratological categories to factual narrative may even lead to an expansion or modification of the inventory of narratology. Historiography, on the other hand, profits alike from a mutual illumination of the two disciplines. Narratology has charted, in the domain of fictional narrative, a wide range of complex narrative strategies whose implications for the interpretation of fictional narratives can be clearly defined. These strategies are usually foregrounded in literature as central elements of fictional representation and may be anticipated by the reader on the basis of his previous knowledge of fictional narratives. Applying the narratological tool-box to historywriting may help to identify stylistic or rhetorical effects and thus bring to light ideological substrata which would otherwise pass unnoticed.

4The fact that narratological categories as constituents of different theories of narrative are rarely if ever fully compatible in their detailed circumscriptions, let alone the fact that they come dressed in so many different terminologies, leaves them open to endless discussion and revision. Here, however, is not the place to enter into a discussion of narratological categories as such. Rather, I have decided upon a combination of Stanzelean and Genettean concepts (as compared in Cohn, 1981), partly in order to follow the paths indicated and tentatively explored by Dorrit Cohn in "Signposts of Fictionality. A Narratological Perspective" (1990), who refers to Stanzel's narrative situations, and by Gérard Genette in "Fictional Narrative, Factual Narrative" (1990). First, however, I shall address — with Dorrit Cohn — the distinction between story and discourse and its validity in fictional and factual narrative. Following this, I shall refer to the Genettean categories for the analysis of time (order, duration and frequency) before proceeding to narrative situations (Stanzel, [1955] 1971) and to the distinction between author and narrator (aspects covered by the Genettean categories of mode and voice). In their essays, both Cohn and Genette point to the opposing positions on fictionality of John Searle, for whom a priori "there is no textual property, syntactical or semantic that will identify a text as a work of fiction" (Searle, 1975, 325), and Käte Hamburger, for whom there exist, in the words of Genette, "incontestable textual'indices' (symptoms) of fictionality" (Genette, 1990, 757). I have included references to Hamburger's classic The Logic of Literature whenever this seems apposite, but have refrained from a more extensive discussion of her tenets.

1. Story and discourse as elementary narratological categories and their applicability to historical narrative

5The distinction between histoire and discours (Todorov, 1966, 126), or story and discourse (Chatman, 1978), which goes back to the Russian formalist critic Boris Tomashevsky's opposition of fabula and sjuzhet, has become one of the commonplaces of literary narratology,3 but has been ignored by narrativist approaches within historiographic theory. As Dorrit Cohn (1990, 778) suggests, the reason for this may well be that the story-discourse model of literary narratology, besides being a nonreferential model, is a synchronic one positing a simultaneity of the two levels. A model valid for historical narrative ought to comprise a diachronic as well as a referential level, as historiography, if one disregards (post-) structuralist or deconstructionist tenets, refers to events belonging to the real past.4 Cohn's concern — like that of Karlheinz Stierle (1973, 530-4) — lies with the re-introduction of a third, referential level into the analysis of historical narrative (in Stierle's model: all narrative), while at the same time she is cognizant of the constructivist view of historiography as propounded by narrativist theoreticians of historiography, notably by W. B. Gallie, Louis O. Mink and Hayden White. Leaving aside the theoretical and philosophical issues involved in such an extension of the story-discourse model, it must be asked, from a narratological point of view, whether the bi-level model of classical narratology is indeed only of marginal interest for the analysis of historical narratives. In other words, is the duplicity of énoncé and énonciation in fact the privilege of fictional narrative, as Paul Ricceur (1985, 61-99) has it? This seems to be refuted by historical practice, which gives us different versions of one and the same "history." However, Dorrit Cohn (1990, 778) is most certainly right in assuming a "more stable" and low-profiled relation between story and discourse in historiography as opposed to a more complex and variable one in fictional narratives, an assumption that would have to be proven by a comparative study of historical and fictional discourse drawing on insights provided by the respective theoretical disciplines: historiographic theory and (literary) narratology.

6Unfortunately, each of the two disciplines has hitherto been concerned with only one level of the model to the virtual exclusion of the other. In the case of historiographie theory, this has been the story-level of historical narratives. Hayden White's term "emplotment" refers to the cognitive pre-structuring of historical contiguity in terms of established patterns of narrative. This process, according to White, is motivated by a desire for [the] formal coherency of a story," by the "universal need not only to narrate but to give to events an aspect of narrativity" ([1980] 1987, 4). Selection comes in as a major factor, as Arthur C. Danto states in his analysis of the function of narrative structure in the creation of meaning in historiography: "... narrative is a structure imposed upon events, grouping some of them together with others, and ruling some out as lacking relevance" ([1965] 1968, 12). For White, the historian's narrative encoding of his material is based on what he calls four "master tropes" — metaphor, metonymy, synecdoche and irony-, which correlate with the archetypal plot structures defined by Northrop Frye with respect to fictional literature. This correlation is not incidental, as the narrativizing of history as described by White involves a process of fictionalization. According to White's theory, the narrative (pre)structuring of the historian's material sets in on a level prior to its exposition in the historiographical text with the conception of the past as history, i. e. as a coherent and "narrativized" sequence of events.

7White is not at all concerned with the discourse level of historical narrative, the level which literary narratology (after earlier attempts at identifying narrative deep structures or at elaborating narrative grammars) has concentrated on in the field of narrative fiction. It must be pertinent, then, to apply narratological criteria to the discourse level of historical narratives, and the works by Rigney (1990) and Carrard (1992) are steps in this direction. While Rigney employs narratological criteria to cast new light on the ideological bias of the works she analyses and Carrard demonstrates that even in the works of the statistically orientated "new" historians narrative still holds its own, the more intricate question is of course whether we can identify discursive criteria to distinguish between fictional and factual narratives. The recent narratological investigations of the fact/fiction boundary by Cohn (1990) and Genette (1990) cautiously suggest a positive answer. Of the aspects investigated by Cohn, only focalization is firmly embedded in the level of discourse. Before I turn to focalization, however, I propose to take a look at a phenomenon which comprises both story and discourse: time (order, duration and frequency).

2. Time

8In historical and fictional narrative alike there is necessarily a tension between the chronology and — in a wider sense — the syntactic structuring of the story. Literary narratology in the wake of Günther Müller, Eberhard Lämmert and Gérard Genette has developed a refined inventory of descriptive categories for the analysis of narrative chronology that may be profitably applied to a comparative analysis of fictional and factual narratives. Eberhard Lämmert states in his Bauformen des Erzählens (1955) that "an observing and critical comparison of narrated time and the time of narration... is at first the safest way to grasp hold of the relation between the world of the narrative and its linguistic representation" ("... ein beobachtender und urteilender Vergleich von erzählter Zeit und Erzählzeit ist der zunächst sicherste Weg, das Verhältnis von erzählter Wirklichkeit und sprachlicher Wiedergabe zu fassen") (1955, 22 f.; my translation). This certainly holds true for the fictional narratives Lämmert is concerned with, but can the analysis of time provide clues for a distinction between factual and fictional narrative? Genette, in applying the categories of order, duration and frequency (as distinguished in his Narrative Discourse [1980]) to factual narrative (cf. Genette, 1990), denies this possibility. According to him, "nothing prevents factual narrative from using analepses or prolepses" (1990, 758), and a statistical comparison would reveal, as he rightly assumes, a higher degree of affinity between certain factual and fictional types (e. g., authentic diaries and diary-novels) than between all fictional types, on the one hand, and all factual types, on the other hand. It follows, therefore, that "fictional narrative and factual narrative are not to be distinguished wholesale either by their use of anachrony or by the manner in which they signal this usage" (Genette, 1990, 760). The same is valid, by way of analogy, for duration and frequency. If Hamburger, as Genette is aware, is right in including "the presence of detailed scenes, dialogues reported in extenso and verbatim, and extended descriptions" (Genette, 1990, 761) among the indices of fictionality as identified by her, it must be noted that these features are not ruled out in historical narrative, even though they may well result in a fictionalization of the historian's account.5 Indeed, analepsis, prolepsis, concentration, expansion and iteration do occur in both fictional and historical narrative, although I have a feeling that, in line with Cohn's assumption of a generally less variable relation between the "story" and "discourse" levels in history-writing, an empirical investigation would reveal more level and regular chronological patterns in historiography. At any rate, historical narratives in practice lack the extreme relations between story time and discourse time often resorted to in fiction. The prerogative of fiction thus lies in the exploitation of the creative potential arising from the dichotomy of story time and discourse time and in the foregrounding of that dichotomy for artistic purposes. To my knowledge, this prerogative is first emphasized in that famous passage from book II, ch. 1 of Tom Jones, where Fielding, styling himself "the founder of a new province of writing," declares his intention to "pursue a contrary method" to the isochronic manner of uninspired historians by making free use of scenic presentation as well as of concentration to the point of ellipsis:

When any extraordinary scene presents itself, (as we trust will often be the case) we shall spare no pains nor paper to open it at large to our reader; but if whole years should pass without producing anything worthy his notice, we shall not be afraid of a chasm in our history; but shall hasten on to matters of consequence, and leave such periods of time totally unobserved. (1966, 88)

9The fiction-maker's freedom from the constraints of strictly chronological presentation becomes most clearly visible in the case of the life-story, as "... one of the distinctions of fictional as compared to historical narrative is that the former is able to make an entire life come to life as a unified whole in a short span of story time" (Cohn, 1989, 3). To the obvious examples cited by Cohn (Joyce's Ulysses and Woolf s Mrs Dalloway) could be added William Golding's Pincher Martin (1956), in which the hero's whole imaginative life on the island passes before his inward eye in the brief moments before his death by drowning.

10As different from chronology, temporality has been marked as a feature distinguishing historical from fictional narrative by Käte Hamburger. Central to her theory is the belief that in epic fiction "the preterite loses its grammatical function of designating what is past" (1973, 66) and becomes the a-temporal tense of (fictional) narrative. It does not refer to a real past, but instead merely indicates fictionality. Thus, the atemporality of the "epic preterite" is pivotal in her distinction between "reality statements" and "epic fiction." Even in the historical novel the preterite has nothing to do with the historical character of the novel's subject matter (cf. Hamburger, 1973, 110-16), but loses its temporality in the same way that the characters become fictionalized, undergoing an ontological transformation: "As subject-matter of a historical novel... Napoleon becomes a fictive Napoleon" (1973, 112). On the discursive level, this loss of temporality becomes manifest, she maintains, in the un-grammatical use of deictic adverbs.6 Contrary to what Hamburger implies, however, the indices of fictionality as identified by her, particularly the ungrammatical use of temporal adverbs in sentences like "Tomorrow was Christmas," are merely optional in fiction, while they can occur in factual narrative, as well, and it is not difficult to see the circularity of her argument as to the a-temporality of the preterite in fiction. Even if Hamburger's attempt to define fictionality via (a-) temporality and vice versa has proved to be a dead end, however, a sense of the difference in the value of tenses should continue to inform theoretical discussions on the fact-fiction issue. Yet, in the wake of (post-) modernist fictional practice, the focus in narratological investigations of temporality has shifted to its significance in defining narrativity (cf. Ronen, 1990; Sternberg, 1992; Fludernik, 1996, 320 ff.), the question being now how to distinguish not fictional from factual narrative, but narratives from non-narratives. Obviously, positions denying the past-ness of events as an essential prerequisite of narrativity create an even wider rift between historiographic practice, which is still informed by the past-ness of its subject matter, and fictional narratology.

11As it seems, the rendering of time in narrative fiction and historiography cannot in principle provide a criterion to distinguish between fictional and factual discourse, although in practice different patterns are very likely to emerge. The case is different when it comes to the categories of mode and voice, as will be shown in the following.

3. Mode

12"Fiction is recognizable as fiction only if and when it actualizes its focalizing potential" (Cohn, 1989, 9). Identifying focalization as a decisive criterion to distinguish factual from fictional narrative, Dorrit Cohn pursues thoughts developed by Kate Hamburger in The Logic of Literature. According to Hamburger's by now famous definition, "[e]pic fiction is the sole epistemological instance where the I-originarity (or subjectivity) of a third-person qua third person can be portrayed" (1973, 83). Simply speaking, this means that it is only in third-person fiction that characters can be presented, not as objects, but as subjects apparently existing independently from the "I" of the author (narrator). It should be added that Cohn's statement is expressly directed towards third-person fiction as the fictional pendant of historiography. In first-person fiction, (internal) focalization automatically applies. On the level of discourse, therefore, there cannot be any intrinsic criteria to distinguish a fictional autobiography of the David Copperfield type from the autobiography of a real (historical) person. The only clue, as Cohn remarks, is provided by the onomastic (non-)identity of author and narrator — a clue all too easily obfuscated by, for example, a nameless first-person narrator. Hamburger, as one will remember, banished homodiegetic narrative from the realm of "epic fiction" on the grounds that it represents a "feigned reality statement" different in its poetological ("dichtungslogische") implications from heterodiegetic narrative. As opposed to third-person narrative, according to Hamburger, first-person narrative does not permit a categorical line to be drawn between fiction and "reality statement" (e. g., factual and fictional autobiography), there existing only a gradation according to the degree of feigning: in other words, the distance of the persona of the first-person narrator from the "I" of the real author.

13Genette is prepared (with restrictions) to regard "subjectivizing constructions" as "distinctive features which differentiate one type from the other [i. e., fictional from factual narrative]" (1990, 762); contrary to Hamburger, however, who disregards this instance, Genette also regards "objective" narration of the Hemingway or Robbe-Grillet type, which is based on external focalization, as another such feature, thereby arriving at "two symmetrical forms of focalization" which "characterize fictional narrative, as opposed to the ordinary attitude of factual narrative" (1990, 762). For Genette as well as for Cohn, "mode is, at least in principle, revelatory of the factual or fictional status of a narrative and, therefore, a point of narratological divergence between the two types" (Genette, 1990, 763). Notwithstanding what has been said about the discourse level of homodiegetic narrative, this may be relevant, to some extent, for first-person narrative, too, if Philippe Lejeune (1980) is right in claiming that factual autobiography tends to emphasize the narrating self of the first-person narrator as opposed to a tendency towards emphasizing his/her experiencing self in autobiographical fiction.

14Here, I am mainly concerned with heterodiegetic narrative. Historiography, as Cohn makes certain, "cannot present past events through the eyes of a historical figure present on the scene, but only through the eyes of the forever backward-looking historian-narrator" (1990, 786). In other words, there is nothing in historiographic narrative comparable to Stanzel's figural narrative situation with its emphasis on a character's perception and point of view. That the presentation of a character's thoughts and feelings can serve as a criterion to distinguish between factual and fictional discourse was already known to Jane Austen, who has one of her heroines, an avid reader of novels, declare: "... a great deal of [history] must be invention. The speeches that are put into the heroes' mouths, their thoughts and designs — the chief of all this must be invention" (1972, 123). In historiography, statements as to a person's motivations, desires, reflections, etc. can be made only on a conjectural basis, unless there is evidence from sources such as diaries, letters or memoirs. The historical novel has colonized this dark area of historiography, although it should be kept in mind that the founder of the genre, Walter Scott, (almost) consistently restricted the presentation of consciousness to his fictive characters. Tolstoy, in War and Peace, on the other hand, freely provides us with the (undocumented) thoughts and feelings of Napoleon on the eve of the battle of Borodino. Still, Tolstoy's novel does not clash with recorded history (as do revisionist historical novels like Peter Ackroyd's recent Milton in America [1996], which has the poet seek refuge from royalist persecution among the Puritan settlers of New England in 1660), but merely supplements it. Although in practice, historians (and especially biographers) may sometimes break the rules so as to lend more interest to their narratives, it still holds that a historian's account of a person's inner life should either be accompanied by a "perhaps" or a "maybe" or, when there is reason for more conclusive conjectures, a "he/she must have thought" — or the account should be documented in a footnote. In his biography of Queen Victoria, Lytton Strachey is careful to insert a cautionary "perhaps" to mark his presentation of the thoughts of the dying monarch as conjectural:.. perhaps, in the secret chambers of her consciousness" (quoted from Cohn, 1989, 10). In his portrait of Florence Nightingale in Eminent Victorians, on the other hand, Strachey seems to be far less scrupulous, as can be seen from a passage which clearly approaches the fictional:

... dream she did. Ah! To do her duty in that state of life unto which it had pleased God to call her! Assuredly she would not be behindhand in doing her duty; but unto what state of life had it pleased God to call her?.... What was that secret voice in her ear, if it was not a call? Why had she felt, from her earliest years, those mysterious promptings towards... she hardly knew what, but certainly towards something very different from anything around her?.... Why was her head filled with queer imaginations of the country house at Embley turned, by some enchantment, into a hospital, with herself as matron moving among the beds? Why was even her vision of heaven itself filled with suffering patients to whom she was being useful? So she dreamed and wondered... (Strachey, [1918] 1986, 112)

15Florence's unuttered (and undocumented7 dreams are presented by Strachey in the form of free indirect discourse, of which Käte Hamburger notes that its "sole grammatical locus" is "narrative literature" (1973, 88). Indeed, free indirect speech, which blurs the boundaries between authorial and figural discourse, should be banned from historiography (cf. Cohn, 1989, 11). As in the case of the chronological structure of narrative, historiographic theory might wish to draw upon narratology's refined inventory of analytic criteria pertaining to the presentation of consciousness to distinguish between factual and fictional elements in historiographic discourse. Unlike chronology, however, focalization does indeed present a decisive criterion in such a distinction.

16That literary practice (especially post-modernist literary practice) is always apt to outwit theoretical efforts at compartmentalization does not in principle harm the validity of narratological analysis, as may be illustrated by the much-discussed example of Marbot, Wolfgang Hildesheimer's ingenious fake biography of a nineteenth-century English aesthete and art-critic who never really existed (cf. Cohn, 1992). Hildesheimer's narrator styles himself as a conscientious biographer, "adhering to the truth, to his best knowledge and belief, by discriminating clearly and rigorously between the documented and the undocumented, between fact and conjecture" ("mit bestem Wissen und Gewissen bei der Wahrheit [bleibend], indem er zwischen Bewiesenem und Unbewiesenem, zwischen Faktum und Konjektur, deutlich und streng unterscheide") (Hildesheimer, 1981, 189; my translation). If Genette (1990) cites Marbot as an illustration of the fact that fictional narratives may well do without any indices of their fictionality, this somewhat understates the case: Marbot, rather than merely refraining from marking its fictionality, flaunts its purported factuality in a way that may well arouse the suspicions of a sceptical reader. It is certainly a matter of no little import that, for instance, Hildesheimer suppresses indicative insights into the mind of his protagonist with a consistency which in all likelihood surpasses even that of the most scrupulous factual biographer. For, as Genette himself concedes, "[in] actual practice... there exists neither pure fiction nor history [nor, in this case, biography] so rigorous as to abstain from all'plotting'and all novelistic devices whatsoever..." (1990, 772). Because Hildesheimer painstakingly refrains from internal analysis, an analysis of Marbot carried out along strictly narratological lines will of necessity produce the wrong results as to the factual or fictional nature — not of its discourse, but of the work itself. Even if this is the case, however, one must admit that only narratology, in its turn, can provide us with an explanation of why Hildesheimers trompe l'oeil succeeds in the first place.

4. Voice

17Genette's cautious statement in Narrative Discourse that the identification of the narrator with the author of a given work is a "confusion that is perhaps legitimate in the case of a historical narrative or a real autobiography, but not when we are dealing with a narrative of fiction, where the role of narrator is itself fictive" (1980, 213) was reformulated in more decisive terms in his "Fictional Narrative, Factual Narrative" (1990, 764): "It seems to me that [the] rigorous identification [of author and narrator] (A = N), to the degree that this can be established, defines factual narrative.... Conversely, their dissociation (Α # Ν) defines fiction." With this, Genette addresses a distinction that has been valid in literary studies since Wolfgang Kayser's dictum that "[t]he narrator is always an imagined, a fictive figure, which is inseparable from the fictional text as a whole" ("Der Erzähler ist immer eine gedichtete, eine fiktive Gestalt, die in das Ganze der Dichtung hineingehört"; my translation) ([1954] 1961, 17). It must not be forgotten, however, that the role of the narrator in fiction has undergone some modification in that the "overt narrator," to use Chatman's (1978) term, of eighteenth-and nineteenth-century works has since given way to a more "covert" or objective narrative voice.

18According to Marie-Laure Ryan (1980, 409), the narrator is a "substitute speaker," an "impersonation" of the author belonging to the alternative, fictional world of the text. In the possible worlds theories of Ryan (1991), Thomas Pavel (1986) or Lubomir Dolezel (1980), the worlds of fictional narratives are brought forth by the speech act of an authoritative source — in the widest sense, the narrator-, who is invested with what Dolezel calls "authentication authority" (1980, passim). Dolezel develops a triadic model of the narrator's authenticating function which roughly corresponds to the three "typical narrative situations" proposed by Franz K. Stanzel ([1955] 1971, [1979] 1984). Both models, however, are restricted to fictional narrative, and in both models the narrative voice in fiction represents the result of an imaginative projection on the part of the real author. Be it overt or covert, the narrator's presence is thus the outcome of a psychological transformation as manifested in the text, while at the same time it indicates an ontological distinction between fictional narrator and real author. What about historiography? Paul Hernadi may be right in claiming that "fictional narratives demand, historical narratives preclude, a distinction between the narrator and the implied author" and that "[t]he distinction between the implied and the actual author is in turn essential only with regard to historical narratives" (1976, 252). Yet, the applicability of his formula to practical analysis seems to me to be much impaired by his retention of that eminently superfluous category, the "implied author," as he thereby confuses distinctions pertaining to the textual level, on the one hand (the distinction between narrator and implied author as inscribed in the text), and to the ontological level, on the other hand (the distinction between fictional narrator and real author). Eliminating the "implied author," Genette's (1990, 766 ff.) comparative analysis of factual and fictional narrative concentrates on the (non-)identity of real author (A), narrator (N) and protagonist (P), as represented by the various constellations possible within a triangular scheme, with

19and

20applying in historical narrative (including biography) and heterodiegetic fiction respectively.8 According to his formula (Α # Ν), Genette is disposed to regard autobiography in the third person as "closer to fiction than to factual narrative" (765). This may confirm our intuitions as to the partial fictionality (if seen from a pragmatic angle) of most autobiographies as such, but may appear as too formalistic an approach, even though Genette is certainly aware of the gradations in impersonation between Caesar's Commentaries and Gertrude Stein's The Autobiography of Alice B. Toklas. Another conundrum is presented by a (hetero-or homodiegetic) narrator who presents a "manifestly fictional story" (767), but who is identified — either explicitly or implicitly — with the real author, as in the case, e. g., of Fielding's Tom Jones or of Dante's Divine Comedy. Faced with these borderline cases, Genette resorts to Searlean concepts of factual and fictional discourse. The formulae for historical narrative and autobiography do not apply, as onomastic identity, in these special cases, is not at the same time a functional identity. In Searlean terms: the narratives in Tom Jones and The Divine Comedy do not constitute serious assertions for which the author assumes full responsibility, but pretences at serious speech acts, which for Searle, represents the paradigmatic condition of fiction.

21The identity of author and narrator in historical narrative is the reason why unreliable narration, a phenomenon that has lately come into the focus of narratological interest again, cannot apply in this case. In Booth's famous definition, a narrator is "reliable when he speaks for or acts in accordance with the norms of the work (which is to say the implied author's norms), unreliable when he does not" ([1961] 1983, 158 f.). Clearly, the notion of "unreliable narration" in the Boothian sense rests on a dichotomy between (implied) author and narrator, which, according to Genette, distinguishes fictional from factual narratives, and must thus be restricted to the former. Even if it may well be possible for a work of historiography to contradict the ethical or ideological norms it seems to be based on by the very style of its presentation (irony), this would involve an act of dissociation on the part of the author which would virtually amount to the creation of a fictional narrative persona. Recently, Ansgar Nünning has suggested a distinction between unreliable narration" (in the Boothian sense) and "untrustworthy narration" (pertaining to the facticity of the narrative) (1997, 89; see also Nünnings contribution to the present volume). It is self-evident that "untrustworthy narration" in fiction and the (deliberate) disregard or even distortion of the facts in historiography are essentially different in that the novelist may employ for artistic effect devices which the historian would be rightly reproached with for using.

22As we have seen in the section on mode, the potential omniscience of a heterodiegetic narrator in fiction (of an authorial narrator, in Stanzel's terminology) with access to the consciousness of his characters is essentially different from the "pluri-qualification" (Carrard, 1992, 105) of the historian, who remains dependent on his sources. A subtle strategy of authentication occasionally used in heterodiegetic novels therefore lies in styling comments on characters' thoughts, feelings and motivations in the conjectural manner of historiographic discourse in order to make the narrator appear as a historian dealing with real-life personages, fashioning his report from source material. In other words, the narrator (temporarily) relinquishes the omiscience which is his/hers in principle and feigns ignorance as to a character's inner life. George Eliot, who analyses the minutest thoughts and feelings of her characters with great ease in the use of the authorial mode, at times denies her narrator full access to their inner state: "It is difficult to say whether there was or was not a little wilfulness in [Dorothea's] continuing blind to the possibility that another sort of choice [i. e. marrying Casaubon] was in question in relation to her" (Eliot, [1871/72] 1988, 29). In the case of first-person fiction, a similar strategy may consist in foregrounding the act of remembrance and the dynamics of memory and oblivion in order to enhance the illusion of a work's being (factual) autobiography and thus the product of reminiscence instead of invention.9 It may be interesting to note that strategies of authentication centring on aspects of mode are also employed, although on a higher level of sophistication, in post-modernist fiction. John Fowles, in his programmatic chapter 13 of The French Lieutenant's Woman, postulates in the best post-modernist manner the all-pervasiveness of fictionality: "Fiction is woven into all" (1969, 82). On the other hand, the novel's narrative discourse makes repeated efforts at explicitly segregating fact from fiction. In the same chapter, Fowles makes it clear that

This story I am telling is all imagination. These characters I create never existed outside my own mind. If I have pretended until now to know my characters' minds and innermost thoughts, it is because I am writing in... a convention universally accepted at the time of my story [i. e., the Victorian Age]: that the novelist stands next to God. He may not know all, yet he tries to pretend that he does. (1969, 81)

23Leaving aside the fact that there is no pretence involved in an author's rendering the "minds and innermost thoughts" of his fictional creations, as they are precisely that — his creations (there would be pretence in doing so in the case of historical characters, as we have seen) —, Fowles frequently undermines the fictional status of his characters as declared in this passage by presenting their inner lives mostly in the form of conjectures or inferences ("I think..."; "he/she must have felt..."). When he does so, his characters are implicitly designated as historical, and despite the author's declarations to the contrary, the tangle of fact and fiction remains unresolved.

24The question of authorial interpolations in the narrative as illustrated by the passages quoted above indeed seems to be of particular relevance to our subject. According to Hamburger (1973, 155 ff.), authorial intrusions amount to a breaking of the aesthetic illusion created by a narrative fiction and thus to an emphasis on the fictionality of the work. In her terminology, authorial interpolations cause a rupture within the "field of fiction," with the real "I-Origo" of the author taking over, "whose narrating — for this moment — is not fictional, but historical" (155). This amounts to the assumption of the possibility of a partial fictionality of texts (a novel can contain non-fictional passages), which contradicts Hamburger's insistence, otherwise, on a categorical separation of fiction and "reality statement": in a novel, she claims, everything (including historical characters, events or settings) is drawn within a "field of fiction" and is thus marked as fictional. Theoretically, therefore, a historical novel could feature both a historical Napoleon (where he is the object of such nonfictional intrusions by the author and thus, in Hamburger's terms, a "statement-object") and a fictional Napoleon (as a character in the novel, i. e., as a fictive I-Origo) side by side — a state of affairs which may appear as somewhat ambivalent from a poetological point of view. Stanzel's concept of an authorial narrator, a fictional persona created by the author, whose repertoire of rôles includes, among others, that of commentator to whom such discursive interpolations in the narrative can be attributed, provides us with a basis for a much more consistent explanation of this phenomenon: the narrator poses as a historian. This is corroborated by Lämmert's contention that discursive interpolations, even if they are apt to break the illusion of the narrative's factuality, will create an illusion of a different kind, namely that of the "personal reality and proximity of the author" ("der persönlichen Wirklichkeit und Nähe des Autors") (1955, 69; my translation). In sum, voice is a narratological category which allows for a distinction between factual and fictional narrative mainly on a theoretical level. Unlike focalization, it does not necessarily become distinctive on the level of discourse.

25As I have tried to show, narratological categories may in some cases provide decisive criteria to distinguish between "factuality" and "fictionality." At any rate, they can provide valuable guidelines for drawing a theoretical boundary between factual and fictional narratives — a boundary, however, which may be transgressed and which has indeed been transgressed by postmodernist fiction. One could thus argue that, as postmodernist literary practices aim at blurring the boundaries between fact and fiction, the need for narratological criteria allowing — at least in theory — for a distinction between factual and fictional narrative discourse, such as certain aspects of mode and voice, is increasing. The reason for this is not any desire for compartmentalization, but to come to a fuller understanding of the interpretative relevance of narrative techniques used by authors like Fowles, Hildesheimer and many others, who have shown themselves "theoretically conscious" in a striking manner. In spite of prognostications to the contrary, there is thus a continuing relevance of structural narratology and narratological analysis in literary and sociohistorical studies. And one cannot help feeling (especially in the face of post-modernist pan-fictionality) that it would be a pity if narratology gave away its competence to contribute to a distinction between factual and fictional discourse by dismissing the relevance of this distinction for its own concerns.