Voice

Voice

GRARD GENETTE

In chapter 5 of his Narrative Discourse, under the general heading of 'Voice', Grard Genette offers a systematic analysis of the role of the 'narrating instance' (the one who narrates) as distinct from that of the 'writing instance' (the one who writes) and that of the characters (the agents of the action). Genette's use of the word 'instance' is meant to erase the ideological associations of subjectivity that accompany traditional concepts of narrator and author, and to underline the functionality of their roles, in the same way Greimas uses the term 'actant' (instead of character) to describe the functional role of the agents of the action, whether human or otherwise, at fabula level. Genette's contribution to the definition of the narrative instance involves the coining of key concepts like that of the 'time of the narrating' (that is, of the narrative's temporal position with respect to the time of the story narrated), and the notion of narrative levels, which he conceives as intangible and yet impenetrable boundaries separating the worlds of the characters and events narrated; of the narrator and his addressee, the narratee; and of the author and his addressee, the reader. The theoretical separation of these levels permits analysis of the effects produced in a narrative when the boundaries are transgressed or blurred (although Genette's scheme makes insufficient distinction between the narrative levels and other semiotic levels which may likewise involve a difference in fictional status). Finally, developing Wayne Booth objections in The Rhetoric of Fiction, Genette refines the traditional concepts of first-or third-person narrative, adding to them the more functional notions of 'homodiegetic' and 'heterodiegetic' narrations, which allow for the theoretical distinction between a narrator who participates in the action as character and a narrator who does not.

Narrative Discourse is a seminal work, presenting a systematic theoretical framework for the analysis of what Genette calls the 'constant literary forms' existing or conceivable in all kinds of narrative. He undertakes the analysis of narrative tense, studying such temporal aspects as 'order', 'duration' and 'frequency'. Another key concept introduced for the first time is 'focalization', a term which helps clarify the difference between 'perspective' (who sees) and 'point of view' (who narrates). (Incidentally, in Narrative Discourse Revisited ( 1983, trans. 1988), Genette rejected Mieke Bal's reinterpretation of his concept reprinted as chapter 6 above.) Also innovative is the parallel Genette draws between 'showing' and 'telling' in the different areas of 'narration of words' and 'narration of events'. The impact of the book, when it first appeared as 'Discours du rcit' in Genette collection Figures III ( 1972), was enormous, and marked a turning-point in the consolidation of French narratology.

The Narrating Instance

'For a long time I used to go to bed early': obviously, such a statement -- unlike, let us say, 'Water boils at one-hundred degrees Celsius' or 'The sum of the angles of a triangle is equal to two right angles' -can be interpreted only with respect to the person who utters it and the situation in which he utters it. I is identifiable only with reference to that person, and the completed past of the 'action' told is completed only in relation to the moment of utterance. [. . .] Even historical narrative of the type ' Napoleon died at Saint Helena' implies in its preterite that the story precedes the narrating, and I am not certain that the present tense in 'Water boils at one-hundred degrees' (iterative narrative) is as atemporal as it seems. Nevertheless, the importance or the relevance of these implications is essentially variable, and this variability can justify or impose distinctions and contrasts that have at least an operative value. When I read Gambara or Le Chef-d'oeuvre inconnu, I am interested in a story, and care little to know who tells it, where, and when; if I read Facino Cane, at no time can I overlook the presence of the narrator in the story he tells. [. . .]

This kind of effect is what we are going to look at under the category of voice: 'the mode of action,' says Vendrys, 'of the verb considered for its relations to the subject' -- the subject here being not only the person who carries out or submits to the action, but also the person (the same one or another) who reports it, and, if need be, all those people who participate, even though passively, in this narrating activity. [. . .] On the one hand critics restrict questions of narrative enunciating to questions of 'point of view'; on the other hand they identify the narrating instance with the instance of 'writing,' the narrator with the author, and the recipient of the narrative with the reader of the work: ¹ 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, even if assumed directly by the author, and where the supposed narrating situation can be very different from the act of writing (or of dictating) which refers to it. [. . . ] The references in Tristram Shandy to the situation of writing speak to the (fictive) act of Tristram and not the (real) one of Sterne; but in a more subtle and also more radical way, the narrator of Pre Goriot 'is' not Balzac, even if here and there he expresses Balzac's opinions, for this author-narrator is someone who 'knows' the Vauquer boardinghouse, its landlady and its lodgers, whereas all Balzac himself does is imagine them; and in this sense, of course, the narrating situation of a fictional account is never reduced to its situation of writing.

So it is this narrating instance that we have still to look at, according to the traces it has left -- the traces it is considered to have left -- in the narrative discourse it is considered to have produced. But it goes without saying that the instance does not necessarily remain identical and invariable in the course of a single narrative work. Most of Manon Lescaut is told by Des Grieux, but some pages revert to M. de Renoncourt; inversely, most of the Odyssey is told by ' Homer,' but Books IX-XII revert to Ulysses; and the baroque novel, The Thousand and One Nights, and Lord Jim have accustomed us to much more complex situations. ² [. . .]

A narrating situation is, like any other, a complex whole within which analysis, or simply description, cannot differentiate except by ripping apart a tight web of connections among the narrating act, its protagonists, its spatio-temporal determinations, its relationship to the other narrating situations involved in the same narrative, etc. [. . .] Therefore, we will look successively at elements of definition whose actual functioning is simultaneous: we will attach these elements, for the most part, to the categories of time of the narrating, narrative level, and 'person' (that is, relations between the narrator -- plus, should the occasion arise, his or their narratee[s] ³ -- and the story he tells).

Time of the Narrating

By a dissymmetry whose underlying reasons escape us but which is inscribed in the very structures of language (or at the very least of the main 'languages of civilization' of Western culture), I can very well tell a story without specifying the place where it happens, and whether this place is more or less distant from the place where I am telling it; nevertheless, it is almost impossible for me not to locate the story in time with respect to my narrating act, since I must necessarily tell the story in a present, past, or future tense. ⁴ This is perhaps why the temporal determinations of the narrating instance are manifestly more important than its spatial determinations. With the exception of second-degree narratings, whose setting is generally indicated by the diegetic context ( Ulysses with the Phaeacians, the landlady of Jacques le fataliste in her inn), the narrating place is very rarely specified, and is almost never relevant. [. . .]

The chief temporal determination of the narrating instance is obviously its position relative to the story. It seems evident that the narrating can only be subsequent to what it tells, but this obviousness has been belied for many centuries by the existence of 'predictive' narrative ⁵ in its various forms (prophetic, apocalyptic, oracular, astrological, chiromantic, cartomantic, oneiromantic, etc.), whose origin is lost in the darkness of time -- and has been belied also, at least since Les Lauriers sont coups, by the use of narrative in the present tense. We must consider, further, that a past-tense narrating can to some extent be split up and inserted between the various moments of the story, much like a 'live' running commentary ⁶ -- a common practice with correspondence and private diary, and therefore with the 'novel by letters' or the narrative in the form of a journal ( Wuthering Heights, Journal d'un cur de campagne). It is therefore necessary, merely from the point of view of temporal position, to differentiate four types of narrating: subsequent (the classical position of the past-tense narrative, undoubtedly far and away the most frequent); prior (predictive narrative, generally in the future tense, but not prohibited from being conjugated in the present, like Jocabel dream in Moyse sauv); simultaneous (narrative in the present contemporaneous with the action); and interpolated (between the moments of the action).

The last type is a priori the most complex, since it involves a narrating with several instances, and since the story and the narrating can become entangled in such a way that the latter has an effect on the former. This is what happens particularly in the epistolary novel with several correspondents where, as we know, the letter is at the same time both a medium of the narrative and an element in the plot. This type of narrating can also be the most delicate, indeed, the one most refractory to analysis, as for example when the journal form loosens up to result in a sort of monologue after the event, with an indefinite, even incoherent, temporal position: attentive readers of L'Etranger have not missed these uncertainties, which are one of the audacities -- perhaps unintentional -- of that narrative. Finally, the extreme closeness of story to narrating produces here, most often, a very subtle effect of friction (if I may call it that) between the slight temporal displacement of the narrative of events ('Here is what happened to me today') and the complete simultaneousness in the report of thoughts and feelings ('Here is what I think about it this evening'). The journal and the epistolary confidence constantly combine what in broadcasting language is called the live and the prerecorded account, the quasi-interior monologue and the account after the event. Here, the narrator is at one and the same time still the hero and already someone else: the events of the day are already in the past, and the 'point of view' may have been modified since then; the feelings of the evening or the next day are fully of the present, and here focalization through the narrator is at the same time focalization through the hero. [. . .] We know how the eighteenth-century novel, from Pamela to Obermann, exploited that narrative situation propitious to the most subtle and the most 'irritating' counterpoints: the situation of the tiniest temporal interval.

The third type (simultaneous narrating), by contrast, is in principle the simplest, since the rigorous simultaneousness of story and narrating eliminates any sort of interference or temporal game. We must observe, however, that the blending of the instances can function here in two opposite directions, according to whether the emphasis is put on the story or on the narrative discourse. A presenttense narrative which is 'behaviorist' in type and strictly of the moment can seem like the height of objectivity, since the last trace of enunciating that still subsisted in the Hemingway-style narrative (the mark of temporal interval between story and narrating, which the use of the preterite unavoidably comprises) now disappears in a total transparency of the narrative, which finally fades away in favor of the story. That is how the works that come under the heading of the French 'new novel,' and especially Robbe-Grillet's early novels, ⁷ have generally been received: 'objective literature,' 'school of the look' -- these designations express well the sense of the narrating's absolute transitivity which a generalized use of the present tense promotes. But inversely, if the emphasis rests on the narrating itself, as in narratives of 'interior monologue,' the simultaneousness operates in favor of the discourse; and then it is the action that seems reduced to the condition of simple pretext, and ultimately abolished. This effect was already noticeable in Dujardin, and became more marked in a Beckett, a Claude Simon, a Roger Laporte. So it is as if use of the present tense, bringing the instances together, had the effect of unbalancing their equilibrium and allowing the whole of the narrative to tip, according to the slightest shifting of emphasis, either onto the side of the story or onto the side of the narrating, that is, the discourse. And the facility with which the French novel in recent years has passed from one extreme to the other perhaps illustrates this ambivalence and reversibility.

The second type (prior narrating) has until now enjoyed a much smaller literary investment than the others, and certainly even novels of anticipation, from Wells to Bradbury -- which nevertheless belong fully to the prophetic genre -- almost always postdate their narrating instances, making them implicitly subsequent to their stories (which indeed illustrates the autonomy of this fictive instance with respect to the moment of actual writing). Predictive narrative hardly appears at all in the literary corpus except on the second level. [. . .] The common characteristic of these second narratives is obviously that they are predictive in relation to the immediate narrating instance ( Aaron, Jocabel's dream) but not in relation to the final instance (the implied author of Moyse sauv, who explicitly identifies himself with Saint-Amant): clear examples of prediction after the event.

Subsequent narrating (the first type) is what presides over the immense majority of the narratives produced to this day. The use of a past tense is enough to make a narrative subsequent, although without indicating the temporal interval which separates the moment of the narrating from the moment of the story. In classical 'third-person' narrative, this interval appears generally interminate, and the question irrelevant, the preterite marking a sort of ageless past: the story can be dated, as it often is in Balzac, without the narrating being so. It sometimes happens, however, that a relative contemporaneity of story time and narrating time is disclosed by the use of the present tense, either at the beginning, as in Tom Jones or Le Pre Goriot, or at the end, as in Eugnie Grandet or Madame Bovary. These effects of final convergence (the most striking of the two types) play on the fact that the very length of the story gradually lessens the interval separating it from the moment of the narrating. But the power of these final convergences results from their unexpected disclosure of a temporal isotopy (which, being temporal, is also to a certain extent diegetic) between the story and its narrator, an isotopy which until then was hidden (or, in the case of Bovary, long forgotten). In 'first-person' narrative, on the other hand, this isotopy is evident from the beginning, where the narrator is presented right away as a character in the story, and where the final convergence is the rule, ⁸ in accordance with a mode that the last paragraph of Robinson Crusoe can furnish us with a paradigm of: 'And here, resolving to harrass my self no more, I am preparing for a longer Journey than all these, having liv'd 72 Years, a Life of infinite Variety, and learn'd sufficiently to know the Value of Retirement, and the Blessing of ending our Days in Peace.' ⁹ No dramatic effect here, unless the final situation should itself be a violent denouement, as in Double Indemnity, in which the hero writes the last line of his confessionnarrative before slipping with his accomplice into the ocean where a shark awaits them: 'I didn't hear the stateroom door open, but she's beside me now while I'm writing. I can feel her./The moon.' ¹⁰

In order for the story to overtake the narrating in this way, the duration of the latter must of course not exceed the duration of the former. Take Tristram's comic aporia: in one year of writing having succeeded in telling only the first day of his life, he observes that he has gotten 364 days behind, that he has therefore moved backward rather than forward, and that, living 364 times faster than he writes, it follows that the more he writes the more there remains for him to write; that, in short, his undertaking is hopeless. ¹¹ Faultless reasoning, whose premises are not at all absurd. Telling takes time ( Scheherazade's life hangs by that one thread), and when a novelist puts on his stage an oral narrating in the second degree, he rarely fails to take that into account. [. . .] Flaubert needed almost five years to write Madame Bovary. Nevertheless -- and this is finally very odd -- the fictive narrating of that narrative, as with almost all the novels in the world except Tristram Shandy, is considered to have no duration; or, more exactly, everything takes place as if the question of its duration had no relevance. One of the fictions of literary narrating -- perhaps the most powerful one, because it passes unnoticed, so to speak -- is that the narrating involves an instantaneous action, without a temporal dimension. Sometimes it is dated, but it is never measured. [. . .] Contrary to simultaneous or interpolated narrating, which exist through their duration and the relations between that duration and the story's, subsequent narrating exists through this paradox: it possesses at the same time a temporal situation (with respect to the past story) and an atemporal essence (since it has no duration proper). ¹² [. . .]

The narrating instance of the Recherche obviously corresponds to this last type. We know that Proust spent more than ten years writing his novel, but Marcel's act of narrating bears no mark of duration, or of division: it is instantaneous. The narrator's present, which on almost every page we find mingled with the hero's various pasts, is a single moment without progression. [. . .]

Narrative Levels

When Des Grieux, having reached the end of his narrative, states that he has just sailed from New Orleans to Havre-de-Grace, then from Havre to Calais to meet his brother who is waiting for him several miles away, the temporal (and spatial) interval that until then separated the reported action from the narrating act becomes gradually smaller until it is finally reduced to zero: the narrative has reached the here and the now, the story has overtaken the narrating. Yet a distance still exists between the final episodes of the Chevalier's loves and the room in the 'Lion d'or' with its occupants, including the Chevalier himself and his host, where after supper he recounts these episodes to the Marquis de Renoncourt: the distance between episodes and inn lies neither in time nor in space, but in the difference between the relations which both the episodes and the inn maintain at that point with Des Grieux's narrative. We will distinguish those relations in a rough and necessarily inadequate way by saying that the episodes of the Chevalier's loves are inside (meaning inside the narrative) and the inn with its occupants is outside. What separates them is less a distance than a sort of threshold represented by the narrating itself, a difference of level. The 'Lion d'or,' the Marquis, the Chevalier in his function as narrator are for us inside a particular narrative, not Des Grieux's but the Marquis's, the Mmoires d'un homme de qualit; the return from Louisiana, the trip from Havre to Calais, the Chevalier in his function as hero are inside another narrative, this one Des Grieux's, which is contained within the first one, not only in the sense that the first frames it with a preamble and a conclusion (although the latter is missing here), but also in the sense that the narrator of the second narrative is already a character in the first one, and that the act of narrating which produces the second narrative is an event recounted in the first one.

We will define this difference in level by saying that any event a narrative recounts is at a diegetic level immediately higher than the level at which the narrating act producing this narrative is placed. M. de Renoncourt's writing of his fictive Mmoires is a (literary) act carried out at a first level, which we will call extradiegetic; the events told in those Mmoires (including Des Grieux's narrating act) are inside this first narrative, so we will describe them as diegetic, or intradiegetic; the events told in Des Grieux's narrative, a narrative in the second degree, we will call metadiegetic. ¹³ In the same way, M. de Renoncourt as 'author' of the Mmoires is extradiegetic: although fictive, he addresses the actual public, just like Rousseau or Michelet; the same Marquis as hero of the same Mmoires is diegetic, or intradiegetic, and so also is Des Grieux the narrator at the 'Lion d'or,' as well as the Manon noticed by the Marquis at the first meeting in Pacy; but Des Grieux the hero of his own narrative, and Manon the heroine, and his brother, and the minor characters, are metadiegetic. These terms (metadiegetic, etc.) designate, not individuals, but relative situations and functions. ¹⁴

The narrating instance of a first narrative is therefore extradiegetic by definition, as the narrating instance of a second (metadiegetic) narrative is diegetic by definition, etc. Let us emphasize the fact that the possibly fictive nature of the first instance does not modify this state of affairs any more than the possibly 'real' nature of the subsequent instances does: M. de Renoncourt is not a 'character' in a narrative taken charge of by the Abb Prvost; he is the fictive author of Mmoires, whose real author, of course, is Prvost, just as Robinson Crusoe is the fictive author of the novel by Defoe that bears his name; subsequently, each of them (the Marquis and Crusoe) becomes a character in his own narrative. Neither Prvost nor Defoe enters the space of our inquiry, which, let us recall, bears on the narrating instance, not on the literary instance. M. de Renoncourt and Crusoe are authornarrators, and as such they are at the same narrative level as their public -- that is, as you and me. This is not the case with Des Grieux, who never addresses himself to us, but only to the patient Marquis; and inversely, even if this fictive Marquis had met a real person at Calais (say, Sterne on a journey), this person would nonetheless be diegetic, even though real -- just like Richelieu in Dumas, Napoleon in Balzac, or the Princesse Mathilde in Proust. In short, we shall not confound extradiegetic with real historical existence nor diegetic (or even metadiegetic) status with fiction. [ . . .]

But not every extradiegetic narrating is necessarily taken up as a literary work with its protagonist an author-narrator in a position to address himself, like the Marquis de Renoncourt, to a public termed such. ¹⁵ A novel in the form of a diary (like the Journal d'un cur de campagne or the Symphonie pastorale) does not in principle aim at any public or any reader, and it is the same with an epistolary novel, whether it include a single letter writer (like Pamela, Werther, or Obermann, often described as journals disguised as correspondence) ¹⁶ or several (like La Nouvelle Hlose or Les Liaisons dangereuses). Bernanos, Gide, Richardson, Goethe, Senancour, Rousseau, and Laclos present themselves here simply as 'editors,' but the fictive authors of these diaries or 'letters collected and published by . . .' -- as distinct from Renoncourt, or Crusoe, or Gil Blas -obviously did not look on themselves as 'authors.' What is more, extradiegetic narrating is not even necessarily handled as written narrating: nothing claims that Meursault or The Unnamable wrote the texts we read as their interior monologues, and it goes without saying that the text of the Lauriers sont coups cannot be anything but a 'stream of consciousness' -- not written, or even spoken -mysteriously caught and transcribed by Dujardin. It is the nature of immediate speech to preclude any formal determination of the narrating instance which it constitutes.

Inversely, every intradiegetic narrating does not necessarily produce, like Des Grieux's, an oral narrative. It can consist of a written text, like the memoir with no recipient written by Adolphe, or even a fictive literary text, a work within a work, like the 'story' of the Curious Impertinent discovered in a cloak bag by the curate in Don Quixote. [. . .] But the second narrative can also be neither oral nor written, and can present itself, openly or not, as an inward narrative (for instance, Jocabel dream in Moyse sauv) or (more frequently and less supernaturally) as any kind of recollection that a character has (in a dream or not). [. . .] Finally, the second narrative can be handled as a nonverbal representation (most often visual), a sort of iconographic document, which the narrator converts into a narrative by describing it himself (the print representing the desertion of Ariadne, in The Nuptial Song of Peleus and Thetis, or the tapestry of the flood in Moyse sauv), or, more rarely, by having another character describe it (like the tableaux of Joseph's life commented on by Amram, also in Moyse sauv).

Metadiegetic Narrative

Second-degree narrative is a form that goes back to the very origins of epic narrating, since Books IX-XII of the Odyssey, as we know, are devoted to the narrative Ulysses makes to the assembled Phaeacians. [. . .] The formal and historical study of this technique would go well beyond our intention, but for the sake of what follows it is necessary here at least to differentiate the main types of relationships that can connect the metadiegetic narrative to the first narrative, into which it is inserted.

The first type of relationship is direct causality between the events of the metadiegesis and those of the diegesis, conferring on the second narrative an explanatory function. It is the Balzacian 'this is why,' but taken on here by a character, whether the story he tells is someone else's or, more often, his own. All these narratives answer, explicitly or not, a question of the type 'What events have led to the present situation?' Most often, the curiosity of the intradiegetic listener is only a pretext for replying to the curiosity of the reader. [. . .]

The second type consists of a purely thematic relationship, therefore implying no spatio-temporal continuity between metadiegesis and diegesis: a relationship of contrast (the deserted Ariadne's unhappiness, in the midst of Thetis'joyous wedding) or of analogy (as when Jocabel, in Moyse sauv, he6sitates to execute the divine command and Amram tells her the story of Abraham's sacrifice). The famous structure en abyme, not long ago so prized by the 'new novel' of the 1960s, is obviously an extreme form of this relationship of analogy, pushed to the limits of identity. Thematic relationship can, moreover, when it is perceived by the audience, exert an influence on the diegetic situation: Amram's narrative has as its immediate effect (and, moreover, as its aim) to convince Jocabel; it is an exemplum with a function of persuading. We know that regular genres, like the parable or the apologue (the fable), are based on that monitory effect of analogy. [. . .]

The third type involves no explicit relationship between the two story levels: it is the act of narrating itself that fulfills a function in the diegesis, independently of the metadiegetic content -- a function of distraction, for example, and/or of obstruction. Surely the most illustrious example is found in the Thousand and One Nights, where Scheherazade holds off death with renewed narratives, whatever they might be (provided they interest the sultan). We notice that, from the first type to the third, the importance of the narrating instance only grows. [. . .]

Metalepses

The transition from one narrative level to another can in principle be achieved only by the narrating, the act that consists precisely of introducing into one situation, by means of a discourse, the knowledge of another situation. Any other form of transit is, if not always impossible, at any rate always transgressive. Cortzar tells the story of a man assassinated by one of the characters in the novel he is reading; ¹⁷ this is an inverse (and extreme) form of the narrative figure the classics called author's metalepsis, which consists of pretending that the poet 'himself brings about the effects he celebrates,' ¹⁸ as when we say that Virgil 'has Dido die' in Book IV of the Aeneid, or when Diderot, more equivocally, writes in Jacques le fataliste: 'What would prevent me from getting the Master married and making him a cuckold?' or even, addressing the reader, 'If it gives you pleasure, let us set the peasant girl back in the saddle behind her escort, let us let them go and let us come back to our two travelers.' ¹⁹

Sterne pushed the thing so far as to entreat the intervention of the reader, whom he beseeched to close the door or help Mr. Shandy get back to his bed, but the principle is the same: any intrusion by the extradiegetic narrator or narratee into the diegetic universe (or by diegetic characters into a metadiegetic universe, etc.), or the inverse (as in Cortzar), produces an effect of strangeness that is either comical (when, as in Sterne or Diderot, it is presented in a joking tone) or fantastic.

We will extend the term narrative metalepsis ²⁰ to all these transgressions. Some of them, as ordinary and innocent as those of classical rhetoric, play on the double temporality of the story and the narrating. Here, for example, is Balzac, in a passage from Illusions perdues: 'While the venerable churchman climbs the ramps of Angoulme, it is not useless to explain . . .,' as if the narrating were contemporaneous with the story and had to fill up the latter's dead spaces. [. . .] Sterne's temporal games, of course, are a bit bolder, a bit more literal, in other words, as when the digressions of Tristram the (extradiegetic) narrator require his father (in the diegesis) to prolong his nap by more than an hour, ²¹ but here, too, the principle is the same. In a certain way, the Pirandello manner of Six Characters in Search of an Author or Tonight We Improvise, where the same actors are in turn characters and players, is nothing but a vast expansion of metalepsis. [. . .] All these games, by the intensity of their effects, demonstrate the importance of the boundary they tax their ingenuity to overstep, in defiance of verisimilitude -- a boundary that is precisely the narrating (or the performance) itself: a shifting but sacred frontier between two worlds, the world in which one tells, the world of which one tells. Whence the uneasiness Borges so well put his finger on: 'Such inversions suggest that if the characters in a story can be readers or spectators, then we, their readers or spectators, can be fictitious.' ²² The most troubling thing about metalepsis indeed lies in this unacceptable and insistent hypothesis, that the extradiegetic is perhaps always diegetic, and that the narrator and his narratees -- you and I -- perhaps belong to some narrative.

Person

Readers may have noticed that until now we have used the terms 'first-person -- or third-person -- narrative' only when paired with quotation marks of protest. Indeed, these common locutions seem to me inadequate, in that they stress variation in the element of the narrative situation that is in fact invariant -- to wit, the presence

(explicit or implicit) of the 'person' of the narrator. This presence is invariant because the narrator can be in his narrative (like every subject of an enunciating in his enunciated statement) only in the 'first person' -- except for an enallage of convention as in Caesar Commentaries; and stressing 'person' leads one to think that the choice the narrator has to make -- a purely grammatical and rhetorical choice -- is always of the same order as Caesar's in deciding to write his Memoirs 'in' one or another person. In fact, of course, this is not the issue. The novelist's choice, unlike the narrator's, is not between two grammatical forms, but between two narrative postures (whose grammatical forms are simply an automatic consequence): to have the story told by one of its 'characters,' ²³ or to have it told by a narrator outside of the story. The presence of firstperson verbs in a narrative text can therefore refer to two very different situations which grammar renders identical but which narrative analysis must distinguish: the narrator's own designation of himself as such, as when Virgil writes 'I sing of arms and the man . . .,' or else the identity of person between the narrator and one of the characters in the story, as when Crusoe writes 'I was born in the year 1632, in the city of York. . . .' The term 'first-person narrative' refers, quite obviously, only to the second of these situations, and this dissymmetry confirms its unfitness. Insofar as the narrator can at any instant intervene as such in the narrative, every narrating is, by definition, to all intents and purposes presented in the first person (even if in the editorial plural, as when Stendhal writes, 'We will confess that . . . we have begun the story of our hero . . .'). The real question is whether or not the narrator can use the first person to designate one of his characters. We will therefore distinguish here two types of narrative: one with the narrator absent from the story he tells (example: Homer in the Iliad, or Flaubert in L'Education sentimentale), the other with the narrator present as a character in the story he tells (example: Gil Blas, or Wuthering Heights). I call the first type, for obvious reasons, heterodiegetic, and the second type homodiegetic.

But from the examples selected no doubt a dissymmetry in the status of these two types already emerges. Homer and Flaubert are both totally, and therefore equally, absent from the two narratives in question; on the other hand, we cannot say that Gil Blas and Lockwood are equally present in their respective narratives: Gil Blas is incontestably the hero of the story he tells, Lockwood is incontestably not (and we could easily find examples of even weaker 'presence'; I will come back to this momentarily). Absence is absolute, but presence has degrees. So we will have to differentiate within the homodiegetic type at least two varieties: one where the narrator is the hero of his narrative ( Gil Blas) and one where he plays only a secondary role, which almost always turns out to be a role as observer and witness: Lockwood, the anonymous narrator of Louis Lambert, Ishmael in Moby Dick, Marlow in Lord Jim, Carraway in The Great Gatsby, Zeitblom in Doctor Faustus -- not to mention the most illustrious and most representative one of all, the transparent (but inquisitive) Dr. Watson of Conan Doyle. ²⁴ It is as if the narrator cannot be an ordinary walk-on in his narrative: he can be only the star, or else a mere bystander. For the first variety (which to some extent represents the strong degree of the homodiegetic) we will reserve the unavoidable term autodiegetic.

Defined this way, the narrator's relationship to the story is in principle invariable: even when Gil Blas and Watson momentarily disappear as characters, we know that they belong to the diegetic universe of their narrative and that they will reappear sooner or later. So the reader unfailingly takes the transition from one status to the other -- when he perceives it -- as an infraction of an implicit norm: for instance the (discreet) disappearance of the initial witnessnarrator of the Rouge or Bovary, or the (noisier) one of the narrator of Lamiel, who openly leaves the diegesis 'in order to become a man of letters. Thus, O benevolent reader, farewell; you will hear nothing more of me.' ²⁵ An even more glaring violation is the shift in grammatical person to designate the same character: for instance, in Autre tude de femme, Bianchon moves all of a sudden from 'I' to 'he', ²⁶ as if he were unexpectedly abandoning the role of narrator; for instance, in Jean Santeuil, the hero moves inversely from 'he' to 'I.' ²⁷ In the field of the classical novel, and still in Proust, such effects obviously result from a sort of narrative pathology, explicable by lastminute reshufflings and states of textual incompleteness. But we know that the contemporary novel has passed that limit, as it has so many others, and does not hesitate to establish between narrator and character(s) a variable or floating relationship, a pronominal vertigo in tune with a freer logic and a more complex conception of 'personality.' The most advanced forms of this emancipation ²⁸ are perhaps not the most perceptible ones, because the classical attributes of 'character' -- proper name, physical and moral 'nature' -have disappeared and along with them the signs that direct grammatical (pronominal) traffic.

If in every narrative we define the narrator's status both by its narrative level (extra- or intradiegetic) and by its relationship to the story (hetero- or homodiegetic), we can represent the four basic types of narrator's status as follows: (1) extradiegetic-heterodiegetic -paradigm: Homer, a narrator in the first degree who tells a story he is absent from; (2) extradiegetic-homodiegetic -- paradigm: Gil Blas, a narrator in the first degree who tells his own story; (3) intradiegeticheterodiegetic -- paradigm: Scheherazade, a narrator in the second degree who tells stories she is on the whole absent from; (4) intradiegetic-homodiegetic -- paradigm: Ulysses in Books IX-XII, a narrator in the second degree who tells his own story. [. . .]

The Narratee

Like the narrator, the narratee is one of the elements in the narrating situation, and he is necessarily located at the same diegetic level; that is, he does not merge a priori with the reader (even an implied reader) any more than the narrator necessarily merges with the author.

To an intradiegetic narrator corresponds an intradiegetic narratee. [. . .] We, the readers, cannot identify ourselves with those fictive narratees anymore than those intradiegetic narrators can address themselves to us, or even assume our existence. ²⁹ [. . .]

The extradiegetic narrator, on the other hand, can aim only at an extradiegetic narratee, who merges with the implied reader and with whom each real reader can identify. This implied reader is in principle undefined, although Balzac does turn particularly sometimes toward a reader from the provinces, sometimes toward a Parisian reader, and Sterne sometimes calls him Madam or Sir Critick. The extradiegetic narrator can also pretend, like Meursault, to address no one, but this posture -- fairly widespread in the contemporary novel -- obviously cannot change the fact that a narrative, like every discourse, is necessarily addressed to someone and always contains below the surface an appeal to the receiver. And if the existence of an intradiegetic narratee has the effect of keeping us at a distance, since he is always interposed between the narrator and us [. . .] it is also true that the more transparent the receiving instance and the more silent its evocation in the narrative, so undoubtedly the easier, or rather the more irresistible, each real reader's identification with or substitution for that implied instance will be. [. . .]

Notes

Robinson Crusoe ( Oxford: Blackwell, 1928), III, 220.

JAMES M. CAIN, 'Double Indemnity', in Cain X3 ( New York: Knopf, 1969), p. 465.

LAURENCE STERNE, Tristram Shandy, IV, chap. 13.

Temporal indications of the kind 'we have already said' and 'we will see later,' etc., do not in fact refer to the temporality of the narrating, but to the space of the text ( = we have said above, we will see further on. . .) and to the temporality of reading.

These terms have already been put forth in my Figures II ( Paris: Seuil, 1969), p. 202. The prefix meta- obviously connotes here, as in 'metalanguage,' the transition to the second degree: the metanarrative is a narrative within the narrative, the metadiegesis is the universe of this second narrative, as the diegesis (according to a now widespread usage) designates the universe of the first narrative. We must admit, however, that this term functions in a way opposite to that of its model in logic and linguistics: metalanguage is a language in which one speaks of another language, so metanarrative should be the first narrative, within which one would tell a second narrative. But it seemed to me that it was better to keep the simplest and most common designation of the first degree, and thus to reverse the direction of interlocking. Naturally, the eventual third degree will be a meta-metanarrative, with its meta-metadiegesis, etc.

The same character can, moreover, assume two identical (parallel) narrative functions at different levels: for example, in Sarrasine, the extradiegetic narrator himself becomes intradiegetic narrator when he tells his companion the story of Zambinella. Thus he tells us that he tells this story -- a story of which he is not the hero: this situation is the exact opposite of the (much more common) one of Manon, where the first narrator becomes on the second level the listener of another character who tells his own story. The situation of a double narrator occurs only, to my knowledge, in Sarrasine.

See the 'Notes by the Author' published at the head of Manon Lescaut.

There remains, however, an appreciable difference between these 'epistolary monodies,' as Rousset calls them, and a diary: the difference is the existence of a receiver (even a mute one), and his traces in the text.

JULIO CORTZAR, ''Continuidad de los Parques,'' in Final del Juego ( Madrid: Alfuguara, 1956).

PIERRE FONTANIER, Commentaire raisonn sur 'Les Tropes' de Dumarsais, vol. 2 of Dumarsais' Les Tropes ( 1818; repr. Geneva: Slatkine Reprints, 1967), p. 116.

DENIS DIDEROT, Jacques le fataliste et son maitre ( Paris: Gamier), pp. 495 and 497.

Metalepsis here forms a system with prolepsis, analepsis, syllepsis, and paralepsis, with this specific sense: 'taking hold of (telling) by changing level.'

STERNE, Tristram Shandy, III, chap. 38, and IV, chap. 2.

JORGE LUIS BORGES, Other Inquisitions, 1937-1952, trans. R. Simms ( Austin, Tex., 1964), p. 46.

This term [personnages] is used here for lack of a more neutral or more extensive term which would not unduly connote, as this one does, the 'humanness' of the narrative agent, even though in fiction nothing prevents us from entrusting

A variant of this type is the narrative with a collective witness as narrator: the crew of The Nigger of the 'Narcissus,' the inhabitants of the small town in 'A Rose for Emily.' We remember that the opening pages of Bovary are written in this mode.

STENDRAL, Lamiel ( Paris: Divan, 1948), p. 43. The inverse case, the sudden appearance of an autodiegetic 'I' in a heterodiegetic narrative, seems more rare. The Stendhalian 'I believe' ( Leuwen, p. 117, Chartreuse, p. 76) can belong to the narrator as such.

HONOR DE BALZAC, Autre tude de femme ( Geneva: Skira), pp. 75-7.

Jean Santeuil, Les Plaisirs et les jours, ed. PIERRE CLARAC and YVES SANDRE (Bibliothque de la Pliade) ( Paris: Gallimard, 1971), p. 319, trans. Hopkins, pp. 118-19.

See for example J. L. BAUDRY, Personnes ( Paris: Seuil, 1967).

A special case is the metadiegetic literary work, of the Curious Impertinent or Jean Santeuil kind, which can possibly aim at a reader, but a reader who in principle is himself fictive.