[Gerald Prince] Narratology the Form and Function(BookFi.org)

August 28, 2017 | Author: Alkav19 | Category: Narration, Narrative, Storytelling, Philosophical Science, Science
Share Embed Donate

Short Description

Download [Gerald Prince] Narratology the Form and Function(BookFi.org)...



N arratology Series Maior 108

The Form and Functioning of Narrative

Studia Memoriae Nicolai van Wijk Dedicata edenda cural

Gerald Prince

C. H. van Schooneveld Indiana University

Mouton Publishers Berlin' New York ' Amslcrdarn






N arratology Series Maior 108

The Form and Functioning of Narrative

Studia Memoriae Nicolai van Wijk Dedicata edenda curat

Gerald Prince

C. H. van Schooneveld Indiana University

Mouton Publishers Ikrlin . New York' Amsterdam

! I


Introd uction

Chapter One

Library of Congress Cataloging in Publication Data Prince, Gera ld. Narratology: th e form and funtioning of narrative. (Janua Linguarum. Series Maior ; 108) Bibliography: p. Inclu des index. 1. Discourse analysis, Narrative. I. Title. 11. Series. P302. P75 808.3'0014 1 82-64 15 ISBN 90-279-309()"'2 AACR2


The Narrator I. Signs 0 f the 'I' 2. Intrusiveness, Self-Consciousness, Reliability, Distance 3. Narrator-Character 4 . Multiple Narrators

7 8 10 13 15

T he Narratee I . Signs 0 f the 'you' 2. Narratee-Character 3. Know ledge 4. Change 5. Individual Narratee and Group Narratee 6. Hierarchy of Narratees

16 17

The Narration I . Posterior, Anterior and Simultaneous Narration 2. Temporal Distance 3. Duration 4. Space 5. Origin, Medium and Interaction with the Narrator 6. Multiple Narrations

ISBN 90 279 3090 2

© Copyrigh t

1982 by Walter de Gruytcr & Co., Berlin . 1\11 rights rese rved. inctuding those of translat io n into foreign languages. N() purl uf thi .. book may be I'epro(]m:ed tn :lilY form - by pho toprint , microfilm , or any other means no r trnllSlIIllI cd nor translrll cd

int o u machine l a n 8 u a ~c with ou t wri tt en I \C II1I I~ K l o n tl ll ll! the IIl1hll shl'1'. 'l'YPCl-oct tlnR: Grcs tun (;rnph ics. 1\ bln~du n , I' rl llt l!!)!: DI IH'kl·/t,j I11111 111111111\1, 11 (11 1111 , 111Iltlln ~: Liideril /. &. 1I 11\1I~ r 1I\1 f.' h ~{\WI\ ,h t\ (:lIlh ll 1'1III It' ll ~II ( :\'/ /lllI lly .

" III I Kfl


1111111 11 K

The Presentation of the Narrated I . Ex pli cit and Implicit Informa tion 2. Presupp osed Information 3. Modes of Discourse

20 21 22

23 24

26 27 29 31

32 33 34 35

36 41





4. Order 5. Point of View 6. Speed

Chapter Two


Textual Constraints

1 11

50 54

Metanarrative Signs

1 15

The Reader







Organizations 1. Temporal Relation 2. Spatial Relations 3. Causal Relations 4. Modifications 5. Relevance 6. Aggregates of Situations and Activities 7. Character 8. Setting 9. Theme 10. Functional Relations 11. Multiple Sequences

64 64 66 66 67 68 70 71 73

Chapter Three



The Orientation of Narrative


The Point of Narrative




74 74






Subject Index


Author Index



83 83 88


The Narrating Component I. Singulary Transformations

95 97

Chapter Four


Reading Narrative

The Code of Written Narrative



The Logical Component

The Expression Component

Chapter Five

Event Description

Narrative Grammar

The Structural Component I. Kernel Narratives 2. Rewrite Rules and the Structure of Kernel Narratives 3. Generalized Transformations and the Structure of Non-Kernel Narratives



Maximal Reading , Minimal Reading, and Narratively Relevant Question s 109



In his " An Introduction to the Structural Analysis of Narrative," Roland Barthes writes: There are countless forms of narrative in the world. First of all, there is a prodigious va,fiety of genres , each of which branches ou t into a variety of media, as if all substances could be relied upon to accommodate man's stories. Among the vehicles of narrative are articulated language, whether oral or written, pictu res, still or moving, gestures, and an ordered mixture of all these sub s t ances ~ narrative is present in myth , lege nd, fables, tales, short stories, epics, history , tragedy, drame [suspense drama ] , comedy, pantomime, pain tings (in Santa Urs ula by Carpaccio, fo r instan ce), stained-glass win dows, movies, local news, conversation. Moreover, in th is infinite variety of forms , it is present at all times, in all places, in all societies ; indeed narrative starts with the very history of mankind; there is not, there has never been anywhere, any people without narrat ive; all classes, all human groups have th eir stories, and very often those stories are enjoyed by men of different and even opposi te cultural backgrounds: narrative remains largel.y unconcerned with good or bad li te rature. Like life itself. it is there , in ternational, transhisto rical. transcultural. 1

Narrative , indeed universal and infinitely varied, may be defined as the represe ntation of real or fictive even ts and situations in a time sequence. Note th at, although many - not to say all - repre sentations can be said to be linked to the dimension of time , not all of th em constitute narratives. In Roses are red! Violets are blu e! Sugar is sweet! And so are you , (2) Roses arc rcd

( I)



could be said to come in time before (3) Violets are blue However, this temporal dimension has nothing whatever to do with the objects or events represented; rather, it has to do with (the production or reception of) their representation. In the world represented, roses are not red before violets are blue and violets are not blue before sugar is sweet. With narratives, on the other hand , we can speak of temporal sequence not only at the representational level but also at the repesented one. In th e world referred to by (4) John was very rich then he began to gamble and he became very poor John's being very rich does precede in time his being very poor. Note also that, although many things (not to say anything) take time , at least some of their representations do not necessarily constitute a narrative. A fight can take a few minutes and a trip can take a few hou rs yet neither (5) There was a fight yesterday nor (6) It was a beautiful trip constitute narratives: they do not represent the fight nor the trip as a series of events but as one event. According to our definition, some messages - however trivial clearly qualify as narrative while others - however interesting - do not. For example, (4) , (7) Mary drank a glass of orange juice then she drank a glass of milk and (8) A people on the Co lumbia had no eyes or mouth. They ate by smelling the sturgeon. Coyote gave them eyes and a mouth as well as Les Trois Mousquetaires. Th e Secret Agent, or The Peiopollnesian War sa tisfy the definiti on and , in fac t, wou ld generall y be considered narrative. On th e 0 1her ha nd , ( I) .



(9) All men are mortal ; Socrates is a man; Socrates is mortal and (10)


It is 900 in New York and it is 95 in Philadelphia

as well as Language, Truth and Logic and Tractatus LogicoPhilosophicus do not satisfy the definition and, in fact, would generally not be considered narrative. But what about simple statements like (II) John got out of the room or (12)

Bill opened the door

which could be claimed to satisfy the definition? Both (II) and (12) refer us to a series of situations and events in a time sequence: after all, (I I) can be restated as (13) John was in the room , then he got out of the room, then he was not in the room and ( 12) can be restated as (14) The door was closed , then Bill opened the door, th en, as a result, the door was open Since (13) and (14) clearly satisfy the definition (in the world represented , John is in the room before he gets out of it; the door is closed before Bill opens it) , (II) and ( 12) do too and so would many statements describing a simple action. Yet , surely, there is a differen ce between such statements and narratives. It cannot merely be a difference in significance since (7), for instance, is not much more significant than (II) or (I 2), if at al l. Likewise, it is not enough to say that (II) and (12) each describe explicitly one and on ly one event, since there are statements which describe a single event and which are sometimes regarded as narratives. In fact , a statement like (15)

At 2 a.m. yesterday, the U.S. declared war on England

could even be called a news story. The difference lies somewhere else. In (II )-( 14), two of the three events and situations evoked arc prcsupposed or entailed by the third. Specifically,





(16) John was in the room and (17) John was not in the room are presupposed or entailed by (II); and (18) The door was closed and (19) The door was open are presupposed or entailed by (12).' Such is not at all the case with (7). Nor is it the case with (15): the latter functions as a story if something like the following reconstruction is made: (20) Most people thought that the U.S. would never declare war on England; then, at 2 a.m. yesterday, th e U.S. declared war on England ; then , as a result, most people were extremely surprised and it is clear that th e reconstructed passages are not necessarily presupposed or entailed by (15).3 A redefinition of narrative, taking the preceding into account is called for: narrative is the representation of at Jeas! twoJ cal oc fictive events or situations in _a tillllu,eq\lence neither of whicl)., presupposes or entails the oth g, Narratology is th e study of the fonn and functioning of narrative. Although the term is relatively new, the discipline is not and , in the Western tradition , it goes back at least to Plato and Aristotle. During the twentieth century , narratology has been considerably developed. The last ten or fifteen years, in particular, have witnessed a remarkable growth of narratological activity. The discipline has attracted numerous literary analysts and many linguists, as well as philosophers, psychologists , psychoanalysts, biblicists, scm ioticians, folklorists, anthropologists , and co mmunication theorists in many parts of the world: Denmark (the 'Copenhagen Group'), France (Barthes, Bremond, Genette, Greimas, Hamon , Kristeva, Todorov, etc.) Germany (lhwe, Schmidt , etc.), Italy (Eco , Segre), the Netherlands (van Dijk) , North America (Chatman, Colby , Doleze l, Dundes, Georges, Hendricks, Labov, Pavel, Scholes, etc.), th e U.S.S.R. (Lotman , Topo rov , Uspenski, etc.). Narratology ~x amin cs what all narrati ves have in co mmon




{D { be t.1

narratively speaking - and what allows them to be narratively different. It is therefore not so much concerned with the history of particular novels or tales, or with their meaning, or with their esthetic value, but rather with the traits which distinguish narrative from other signifying systems and with the modalities of these traits. Its corpus consists not only of all extant narratives, but of all possible ones. As for its primary task, it is the elaboratio n of instruments leading to the explicit description of narratives and the comprehension of their functioning. -J attempt in this study to answer three questions of central concern to narratologists: what are the features of narrative which allow us to characterize its possib le manifestations in pertinent terms (Chapters one and two)? ~ould a formal model accounting for these features and_manifestations look like (Chapter three)? what are the facto-rs which affect our understanding of a narrative and our evaluation of its narrativity (Chapters four and five)? In my presentation , J focus on written narrative because it is the kind I know best. However, much of what 1 say is applicable to any narrative regardless of the medium of representation. 1 do not try to discuss everything that is known about narrative nor even everything that should be known; but, at the risk of frequently stating the obvious and of repeating even more frequently what others have said very well before me (my debt to Roland Barthes, Wayne C. Booth, Gerard Genette , Tzvetan Todorov, and many more is clearly tremendous), 1 try to discuss, however briefly , most of what I think must be known. For the sake of convenience, clarity and brevity, and to emphasize the fact that the domain of narratology consists of all narratives and not merely great ones, or literary ones, or interesting ones, 1 often use as examples (parts of) narratives of which I am the author. Finally, many of the translations are my own. 1 hope that it will not be too obvious. Some of the material in this book has already appeared in Centrum 1(1) (1973), Poetique (14) (1973), Poetics Today 1(3) (1980) and my A Grammar of Stories: An Introduction, published by Mouton and Co. I should like to thank the editors and publishers for permission to reprint. I should also like to thank Ellen F. Prince for many stimulating ciiscussions.



A narrative is, among other things, a collection of signs which can be grouped into various classes. More particularly, in written narrative, certain features and combinations of the linguistic signs making up the narrative constitu te signs of the narrating (or narrating, for short): they represent the narrating activity, its origin and its destination. Other features and combinations constitute signs of the narrated (or narrated for short): they represent the events and situations recounted. Each of these two classes may in turn be divided into sub-classes. Thus, among signs of the narrating some pertain more specifically to the narrator (the one who narrates), others to his narratee (the one who is narrated to) and others still to his narration (the act of his narrating); and among signs of the narrated, some pertain more specifically to characters, for instance, others to the time during which the characters act, and others still to the space in which their actions occur.


In grammar, a distinction is made among the first person ('I', for example), the second person ('you') and the third person ('he'). The first person is defined as the one who speaks, the second person as the one who is spoken to, and the third person as the being or object that is spoken about. Similar distinctions can be made in narratology: we can say that the narrator is a first person, the narratee a second person and the being or object narrated about a th ird person.

The Narrator



In a given sentence, an 'I' representing the speaker mayor may no t appear, of course. Consider my saying the following : (I) I am a plumber (2) Paris is the capital of France (3) Mary took her exams in July. Similarly , in a given narrative , an 'I' representing the narrator may or may not appear. Consider my narrating the following: (4) I go to th e refrigerator, take out a can of beer and drink it. (5) He goes to the refrigerator, takes out a can of beer and drinks it (6) Joan is ri ch. She meets John and she becomes very poor. There 's at least one narrator in any narrative and this narrator may or may not be explicitly designated by an 'I'. In many narratives where he is not, the 'I' may have been deleted without leaving any traces but the narrative itself: there is nothing in (5) and (6) which refers to or implies a narrating ac tivity and, th erefore , a narrating self except for the fact that they are narratives. In many narra tives where he is, the 'I' may constitute the only reference to his narrating self. Thus, in (4) , we learn nothing explicit about the narrator as such, except that he is recounting events in which he takes part: we do not know what he thinks of these events as he is narrating them ; we do not perceive what his attitude towards his narration is ; and so on and so forth. Although he describes his own actions, (4) is not really more subjective than (5) or (6). Indeed, in (4), (5), and (6) respectively, '1' , 'he' , and 'Joan' do not function in a significantly different fashion: they each act simply as a character's name. But there are also many narratives where numerous signs representing the narrator and signifying his presen ce in the narrative are evident, whether or not an 'I' designating him ever appears. I.

Signs of th e 'I'

Some of these signs may fun ction indirectl y. T hu s, any second person pronoun wh ich does not (exclu sive ly) refcr to a chara cter and is no t uttered (or "th ought ") hy h im lIlust re fe r to so me ne


whom a narrator is addressing and th erefo re constitutes a trace of the latter's presence in the narrative. Consider for example, (7)

and (8)

"All is true, - so true that you may recognize its elements in your experience , and even find its seeds within your soul" (Le Pere Coriot), As you know, John went to France then he went to Germany.

But some - we can call them signs of the 'I' - function more directly and represent the narrator and/or his spatio-temporal situation.' It is clear, for instance, that any first person plural pronoun which does not exclusively designate characters (or narratees) refers to a narrating self: (9)

"We will confess that , following the examp le of many a serious author, we have started our hero's story one year before his birth" (La Chartreuse de Parme) (10) "To sum up the points to which we have directed attention , three kinds of ravages nowadays disfigure Gothic architecture" (Notre-Dame de Paris)


Furthermore, there is a class of deictic tenn s ('now', 'here' , 'yes ter- I day', 'tomorrow', and so on) which relate to the situation of their utterance and, more particularly , to the spatio-temporal situatiOn ) of the utterer. Should one of them appear in a narrative and should it not be part of a character's utterance, it mu st be related to a narrator. In (\ I) Mary went to the beach yesterday, then she went to the movies, then she went home , (12) "Then, some sixty years ago, a sudden change took place. The gin-pits were elbowed aside by th e large mines of the fin anciers" (Sons and Lovers) and (13) J ohn came here , he got drunk and he left, 'yesterday' , 'some six ty years ago' and 'here' characterize the narrato r's spatio-tcmpora l situatio n. There is also a class of modal te rms (' perhaps' , ' unfortunately' , 'c lea rl y', and so forth) which


The Narrator


indicate a speaker's attitude about what he says. Once again, should one of them not be part of a character's utterance , it describes the narrator's position: 'perhaps' and 'clearly' in

must be led by sombre colors and solemn thoughts" (L e Pere Goriot) and (2 1)

(14) John went to the movies. Perhaps he :-vas lonely and (15) John reacted very coldly. Clearly, this was the result of his having suffered too much and too often function as signs of '1'. More generally, any sign in a narration which represents a narrator's persona, his attitude , his knowledge of worlds other than that of the narrated , or his interpretation of the events recounted and evaluation of their importance constitutes a sign of the '1'. Consider the following underlined passages, for example: (16) People are remarkable. John was poor and sick ; he kept on trying to improve his lot and managed to become rich and healthy (17) He was wearing one of those flashy ties often seen on Broadway (18) Like aU people of good taste , he drove a Cadillac (19) He must have been scared since he was sweating profu sely.2 2.


"No part of Paris is so depressing, nor we may add , so little known. The Rue Neuve Sai nte-Genevieve, above all , may be likened to an iron fram e the o nly fram e fit to hold the coming narra ti ve, to which th e rea der's mind

"Outside the arc-light shone through the bare bran ches of a tree . Nick >valked up the street beside the car-tracks and turned at the nex t arc-light down a side-street. Three houses up the street was Hirsch 's rooming house . Nick walked up the two steps and pushed the bell. A woman came to the door. " ("The Killers")

Furthermore , intru sions may have different degrees of obviousness. Given a narrative reco unting events which took place in Corsica in 1769, for example, 'Emperor' wou ld constitute an intrusion in (22)

Emperor Napoleon's birth was greeted with joy

since nobody in the world of the narrated could know the newborn baby's destiny. On the oth er hand, some intrusions are far more ev ident. In Tom Jon es the narrator even warn s his narratee of the many digressions to co me: (23)

"Reader, I think proper, before we proceed any farther together, to acquaint thee that I intend to digress through this whole history , as often as I see o ccasion, of which I am myself a better judge than any pitiful critic whatever ; and here I must desire all those critics to mind their own business . . .. "

Intrusiveness, Self-Consciousness, Reliability , Distance

Whether or not the narrator is deSignated by an '1', he may therefore be more or less intrusive, that is, more or less explicitly characterized as a narrating self. If we eliminated every narrator's intru sion from Le Pere Goriot, Eugenie Grandet or Malone meurt, we would be left with relatively little ; but if we did the same with "The Killers", "Hills like White Elephants" or L'Age de raison , we would have, relatively, quite a lot to read. Compare


Note that so me narratologists would consider the slightest "evaluative" adjective or adverb or the most discreet logical connection between events to be intrusions' Given (24) (lnd (25)

John walked elega ntly Bill was happy because he had just seen Robert

fo r example, th ey wou ld regard 'elegantly ' and 'because' as intrusive clements. Vet this is not a very convincing position; for there is n thin g in (24) and (25) which indicates that perhaps John did no t wa lk elega ntl y and perhaps Bill's happin ess was not the resul t o f his havin g see n Robert ; th at is, th ere is no thing which indicates



that the evaluation and the logical connection are the result of the narrator's interpretation, the consequence of his special knowledge, the mere product of his subjectivity rather than well-established facts in the world of the narrated. Indeed, the elegance of John's walk and the cause of Bill's happiness are given as incontrovertible and we take them as such when we read. If a narrator may be more or less intrusive, he may also be more or less self-conscious, that is, he may seem more or less aware that he is narrating: Jacques Revel in L 'Empioi du temps - who often discusses the circumstances of his writing - is a self-conscious narrator ("The day before yesterday, as I set down my recollections of the seven-months past Sunday". "That is why I now feel compelled to interrupt the pattern I had been following for the past month in my narrative", "I have used the last inch of daylight to finish rereading my account of the second week in June, written two months ago at this very table") whereas Meursault in L 'Etranger is not: not once does he comment on the fact that he is telling his own story. Note that, whereas a self-conscious narrator is always intrusive, the reverse is not true: the narrators of (17), (18) and (19) are intrusive without being self-conscious in the least. A narrator may also be more or less reliable; in other words, (parts of) his account may be more or less worthy of trust in terms of the narrative itself. When reading La Chute, we are led to conclude that Jean-Baptiste Clamence is quite unreliable: he is a confirmed liar; he constantly and systematically contradicts himself; and it becomes clear that most of what he says - if not everything - is not supposed to be taken at face value. In Le Pere Goriot, on the other hand, we are not made to question the narrator's reliablility: no reason is given us - insofar as the fictional world is concerned - for doubting the validity of his account and of his judgments. Note that a reliable narrator is not necessarily one that I - as a reader - always agree with: after all, however honest and trustworthy he may be portrayed as, I may find his values repugnant and his conclusions stupid. Conversely, I may find the attitudes of an unreliable narrator very attractive indeed. Finally, a narrator may be at a greater or lesser distance from the events narrated, from the clwractcrs presented and/or from his narratcc. The distance Illay be tl'IIlI'Or,1I (I narrate events which

The Narrator


happened yesterday or fifty years ago); it may by physical (Oskar in The Tin Drum does not address dwarves); it may be intellectual (the narrator of The Sound of the Fury is certainly more intelligent than Benjy), moral (Sade's Justine is far more virtuous than the characters populating her story), emotional (the narrator of "Un Coeur simple" is not as moved as Felicite by Virginie's death), and so forth. Of course, a given distance may vary in the course of a given narrative: at the end of Great Expectations, the narrator is temporally closer to the narrated than at the beginning; and in La Chute Jean-Baptiste Clamence's narratee resists him more and c more as the novel progresses. The intrusiveness of a given narrator, his degree of self-conscious'/ ness, his reliability, his distance from the narrated or the narratee? not only help characterize him but also affect our interpretation (" of and response to the narrative. Thus, intrusions commenting on some of the events recounted may bring out or underline their importance in a certain narrated sequence or their intrinsic interest; they may also delight us (if we find them witty, for instance) or annoy us (if we find them superfluous). Intrusions referring to the narrator or the quality of his narration may lead us to conclude that the real subject of the narrative is the rendering of certain events rather than the events themselves and that the real hero is the narrator rather than anyone of his characters. Similarly, the narrator's unreliability forces us to reinterpret many of his statements in order to arrive at a knowledge and understanding of "what really happened"; and variations in distance may entail variations in our intellectual appreciation of and emotional commitment to one character or another. 4 3.


Just as he mayor may not be explicitly designated by an 'I' and whether or not he is intrusive, self-conscious or reliable, the narrator mayor may not be a participant in the events he recounts. When he is, we usually speak of a first-person narrative because the first person narrates-- among other things -- events in which he takes part (Moll Handa\·. Great lix/,ectatiuns, The Great Gatsby).


The Narrator



We can then make a distinction between the first person as narrator and the first person as character. In (26)

I ate meat

the character T is the one who ate and the narrator T is the one who tells about the eating ; similarly in Great Expectations, the matu re Pip, who recounts the adventures of his you nger self, is different from that younger self; and in All the King's Men, the Jack Burden who narrates his own story as well as Willie Stark's is not quite like the Jack Burden who studied history in graduate school, had a couple of nervous breakdowns and worked for Willie. When the narrator is not a character, we usually speak of a thirdperson narrative, becau se the events narrated refer to third persons (Barch ester Towers, The Portrait of a Lady, L 'Education sentimentale). Sometimes, of course, the narrator may be a character yet refer to himself as to a third person - as to one character among many others - more or less frequen tly and systematically. In Thackeray's Henry Esmond the protagonist tells his own story mostly in the third person ("'Tis needless in these memoirs to go at any length into the particulars of Harry Esmond's college c&reer"; "Esmond went away only too glad to be the bearer of such good news"; "With the exceptio n of that one cruel letter which he had from his mistress, Mr. Esmo nd heard nothing from her"); and in Ca mu s' La Peste, Dr, Rieux refe rs to himself as Dr. Rieux through most of the novel. Another possibility - and a relatively seldom exploited one in fiction - is the second-person narrative, where the events narrated pertain to a second person: (27)


You were both standing in the doorway between the brigh t room and the dark room, and she was whispering these words not in your ear but against your mouth , with her lips touching yours from time to tim e (La Modification) "Sometimes, you stay three, four, five days in yo ur room, you don't know. You sleep almost co ntinuously , you wash your socks, your two shirts. You reread a mystery novel you've already read twenty times, forgotten twenty timcs" (UI/ lIomme qlli dart)


Again, the narrator may be a character yet refer to himself as 'you', and in a work like La Modification it is difficult - initially, at least - to tell whether the 'you' who is the protagonist designates a narrator-character or not. In cases where the narrator is a character, he may playa more or less considerable role in the events which he recounts. He may be the protagonist (The Confessions of Zeno, Great Expectations, Voyage au bout de la nui!, Kiss me deadly), or an important character (All the King's Men, La Porte £troite), or a minor one (A Study in Scarlet), or even a mere observer ("A Rose for Emily"). Sometimes, he may be a character in one part of his narrative but not in another (the ' I' in "Sarrasine") and sometimes, though he plays no part in the events which he himself narrates, h e may be a character in events recounted by another narrator (Scheherazade in Arabian Nights). 4.

Multiple Narrators

Up to now, I have mainly proceeded as though there were only one narrator per narrative and , obviously, this is often the case. Consider (29) or (30) , for examp le, in which one and only one "{" recounts a series of events: (29) (30)

I was very happy , then [met Peter, then, as a result , I was very unhappy Peter was very unhappy , then he met Jane, then, as a result, he was very hap py

But there are many narratives with more than one narrator ; indeed, in a given narrative, there may be an indefinite number of narrators (two, three, ten , etc.). For instance, a narrator may introduce another narrator who in turn introduces another narrator, and so forth: (3 1)

I was having a cup of coffee in a dingy luncheonette when a stranger sat at my table and told me: .. A few years ago - I was twenty at the time - I had a very strange experience. I waS walking down the street. . .. A few years


The Narratee


later, a beautiful woman came to see be and told me : 'I

(33) She is very sick . She drinks a glass of wine and she becomes very healthy

was . .. . ''';

Or a narrator may introduce another narrator, then another one, th en another one , and so on: (3 2)

I was having a cup of coffee in a dingy luncheonette when John sat at my table and told me a story: " A few yea rs ago, I was . ... " Then Peter came and told me ano th er story: "A few days ago, I was. ... " I kept drinking coffee ....

When there are two or more narrators in a narrative, it is possible to establish a hierarchy among them. The one who ultimately introduces the entire narrative (including all the mini-narratives comprising parts of it) is the main narrator. The others are secondary narrators, or tertiary ones, etc. In (31), the first 'I' is the main narrator, the stranger is a secondary one, and the beautiful woman a tertiary one. In (32), the first 'I' is the main narrator, whereas John and Peter are secondary o nes. Note that a tertiary narrator, for examp le, may be more important or interesting than a secondary one . There are three narrators in L 'Immoraliste : the on; who provides a title for the novel an d through whom all of the narrated even ts are ultimately presented, Michel's friend and Michel himself. The latter, who is a terti ary narrator, is clearly more interesting tha n his friend, who is a secondary one. Finally, note that one narrator may be at a greater Or lesse r distance from another one, that this distance ma y be physical, or intellectual , or emotional , or moral , and that it may vary within a given narrative .


If there is at least one narrator in any narrative , there also is at least one narratee and this narratee mayor may no t be exp licitly designated by a 'you' . In many narratives where he is not, the 'you' may have been deleted without leavi ng any traces but the narrative itse lf. Th ere is nothing in


or (34) Joan is very rich. She drinks a cup of coffee and she becomes very poor for instance , which refers to or implies a narrating activity and , therefore, a narrative audience except for the fact th at they are narratives. In many narratives where he is, the 'you' may constitute the only reference to a narrative audience. Given (35)

You are very sick. You drink a glass of wine and you become very healthy

we lea rn nothing explicit about th e narratee as such, except that he takes part in the events recounted to him: we do not know what he thinks of these events as he is told th em; we do not perceive what his attitude towards the narrator and his narration is; and so on and so forth . Indeed, the 'you' does no t function differently from the 'she' and the 'Joan' in (33) and (34) respectively: each acts simply as a character's name. But there are also many narratives where numerous signs representing the narratee and signifying his presence in them are evident, whether or not a 'you' designating him ever appears. I.

Signs of the 'You'

Some of these signs may function indirectly. Thus, just as any 'y ou' designating a narratee implies a narrator, any 'I' designating a narrator implies a narratee' But some - we may call them signs of the 'you' - function more directly and rep resen t the narratee (and his si tuation). In (36)

and (37)

"We could hardly do otherwise than pluck one of its n owers and present it to the reader" (The Scarlet Leller) Bu t let the one who is liste n in g to this tale be patient. He will find o ut ~oon enou gh what fate awaited John

' I he reader' and ' th e one who is listenin g' refcr to an audience.



Similarly, first-person pronouns, for example, may designate not (only) a narrator but (also) a narratee. When Marcel writes in A la recherche du temps perdu: "Besides, most often we did not stay home, we went for a walk" the 'we' exclu des the narratee; on the other hand, when he writes (39) "In such perfect coincidences, should reality apply to what we have been dreaming for such a long time, it entirely conceals it from us"

Th e Narratee

and the narrator of Les Trois Mousquetaires ex plicitly contradicts one of his narratee's inferences : (4 3)


the ' we' and the ' us' include the person he is addressing. Furthermore, parts of a narrative may take the shape of questions or pseudo-questions. Sometimes, these do not eman ate from a character or from th e narrator, who merely seems to be repeating them. They can be attributed to the narratee. In Le Pere Coriot, it is the narratee who asks about M. Poiret's career: (40)

"What he had been? Well , possibly a clerk in the Department of Justice . ... "

Sometimes, when questions or pseudo- Active Event

For the sake of brevity and an easier handling of rules, I shall use the following set of symbols: N NScc



kernel narrative


narrative section, containing n episodes and

= =

one narrative episode cluster of one or more conjunctive features


Ep stat


Ep act


ep stat


cp act



sub CCL


e stat cad

= =

string of episodes, containing no narrative episode string of stative episodes, containing no narrative episode string of active episodes, containing no narrative episode stative episode (group of conjoined stative events belonging to the same time sequence and containing no narrative event) active episode (group of conjoined active events belonging to the same time sequence and containing no narrative event) conjunctive feature of time indicating a before-after relationship between two episodes or events proper subset of CCL, containing no conjunctive feature of time stative event active even t


= =

= =


stative narrative event stative narrative event to be modified modified stative narrative event active narrative event (modifying In Ne stat) conjunctive feature indicating that one event is the consequence of the preceding event any conjunctive feature which is not one of time or consequence

prop stat prop act N prop stat N prop statmod LT t LTc LTn

narrative episode, containing n events and one narrative event


Ne stat In Ne stat In Ne statmod Ne act CFc


= = = = = =

stative proposition active proposition stative narrative proposition modified stative narrative proposition logical term of time logical term of consequence any logical term which is not one of time or consequence

The sign + indicates the concatenation of the various symbols in a string and may be suppressed where there is no danger of confusion. Parentheses are used to enclose optionally chosen items. For the two rules A-->B A---+B+C (but not A --> C) we may write A --> B (C) Alternative replacements for a symbol, one of which may be chosen at a single application, are listed vertically within braces. Thus, for the three rules A ---+ B A-->C A ---+ 0 we may write

The Structural Component

Nan-alive Grammar

If we wish to apply a replacement for a given nonterminal symbol (a symbol appearing on the left of the arrow) in certain contexts only, we specify it in the appropriate rule. For instance, if A may be rewritten as B only when it is in initial position in any given string of symbols, we have the following rule: A ---+ B/ # -If A may be rewritten as B only when it is in final position in any given string of symbols, we have the rule A ---+ B/- - # If A may be rewritten as B only when it does not precede or immediately follow C, we have the rule: A ---+ B/C + ... + -In all cases, - - shows the place where the given replacement is allowed. The set of rules describing the structure of kernel narratives which I shall call component C, from now on - is the following:


. INsec + CCL + NSec + CCL + NSec I. N ---+ Ep stat + CCL + Ep stat Ep act + CCL + Ep act 2. NSec ---+ Ncp (CCL + Ep)/- - # (Ep + CCL) Nep Ep stat 3. Ep---+ Ep act 4. Ep stat ---+ ep stat (CCL + Ep stat) 5. Ep act ---+ ep act (CCL + Ep act) 6. CCL ---+ CFt + sub CCL/ ... +Nep+ ... +Nep+ - - +Nep CFt (sub CCL) (cp stat + sub CCL) Ne stat (sub CCL + ep stat)/--+ ... + Nep+ ... +Nep 7. Ncp Nc stat (sub CCL + ep stat)/Ne stat+ ... +Nep +

... + -(ep act + sub CCL) Ne act 8. ep stat ---+ e stat (sub CCL + ep stat) 9. ep act ---+ e act (sub CCL + ep act) 10. Nc stat ---+ In Nc stat/- - + ... +Ne stat


lin Nc statlll(xi



CF C (CFnJ/--+ In Ne statmedl (CFc)CF n \ CFc e stat - + prop stat e act - + prop act In Ne stat - + N prop stat Ne act - + prop act In Ne statmed - + N prop stat med CFt ---+ LT t CFc-+LTc CFn - + LTn

II. subCCL---+

12. 13. 14. 15. 16. 17. 18. 19.

In other terms, the rewrite rules making up component C specify that any kernel narrative consists of three narrative sections conjoined by a cluster of conjunctive features or of two conjoined strings of stative or active episodes; each narrative section contains a string of n episodes and one narrative episode; each string of episodes consists of stative or active episodes; each cluster of conjunctive features consists of at least a conjunctive feature of time; and so on and so forth. If we apply the rules of component C, we get a derivation of the structure of any kernel narrative. For example, we could get the derivation of the structure of (I), (2), (14), or (18)

John was happy, then John met Bill, then, as a result, John was unhappy

Note that in the following derivation of the structure of (18), the number at the left of each line refers to the rule of component C used in constructing that line from each preceding line:

N NSec + CCL + NSec + CCL + NSec


Nep + CCL + NSec + CCL + NSec


Nep + CCL + Nep + CCL + NSec


Ncp + CCL + Nl'I' + eeL + Ncp


Ner + ('I"I + Nel' + ('('I. + Neil



The Structural Component

Narrative Grammar

Nep + CFt + Nep + CFt + sub CCL + Nep


Ne stat + CFt + Nep + CFt + sub CCL + Nep


Ne stat + CFt + Ne act + CFt + sub CCL + Nep


Nep stat + CFt + Ne act + CFt + sub CCL + Ne stat


In Ne stat + CFt + Ne act + CFt + sub CCL + Ne stat


In Ne stat + CFt + Nc act + CFt + sub CCL + In Ne stat mod (10) In Ne stat + CFt + Ne act +CF t + CFc + In Ne stat mod

(11 )

Nprop stat + CFt + Ne act + CFt + CF c + In Ne stat mod


Nprop stat + CFt + prop act + CFt + CF c + In Ne stat mod

( 15)

Nprop stat + CFt + prop act + CFt +CF c + Nprop statmod

( 16)

Nprop stat + LT t + prop act + CFt + CF c + Nprop stat mod


Nprop stat + LT t + prop act + LT t + CF c + Nprop stat mod Nprop stat + LT t +prop act + LT t + LTc + Nprop statmod

(17) (18)

The derivation shows that three events (three propositions) are recounted by (18), The first event is stative and temporally precedes the second; the second event is active, temporally precedes the third and causes it; and the third event is stative and constitutes a modification of the first. 3,

Generalized Transformations and the Structure of Non-Kernel Narratives

Many narratives are not kernel narratives since they recount more than one modification of a situation or state of things, Specifically many narratives are constituted by the conjoining of one (kernel) narrative with another one; or by the embedding of one (kernel) narrativc into another one; or again, by the alternation of a section of one (kernel) narrative with a section of another one, For instance, a narrative like (19)

John was poor, then he found gold in his field, then, as a result, he was rich, Then, as a result, Peter was sad, then


he found oil in his field, then, as a result, he was happy can be said to be constituted by the conjoining of (20) John was poor, then he found gold in his field, then, as a result, he was rich and (21) Peter was sad, then he found oil in his field, then, as a result, he was happy Similarly, a narrative like (22) John was rich and Joan was poor. Then Joan made money, then, as a result, she was rich, Then John lost money, then, as a result he was poor could be considered to result from the embedding of (23) Joan was poor then Joan made money, then, as a result, she was rich into (24) John was rich, then John lost money, then, as a result, he was poor

Finally, a narrative like (25) John was happy and Joan was unhappy, then John got divorced and Joan to married, then, as a result, John was unhappy and Joan was happy may be said to result from the alternation of one event from (26) John was happy, then John got divorced, then, as a result, John was unhappy and one event from (27) Joan was unhappy, then Joan got married, then, as a result, Joan was happy It is clear that component C cannot account for the structure of non-kernel narratives, In order to account for it, it is necessary to add a new set of rules to the rules which we already have, These will be transformational and will operate on two strings, provided these strings have a certain structure lO The first part of a transformational rule is a structural analysis (SA) specifying the kind of strings (in terms of their structure) to which the rule applies, A rule might apply, for example, to any two strings which can be analyzed as follows:

The Structu ral Component

Narrative Grammar



of (a): of (b):

Nep - CCL - Nep - CCL - Nep Nep - CCL - Nep - CCL - Nep

The st ru ctural analysis often co ntains symbols like X or Y, standing for any set of eleme nts. Suppose only one Nep mu st be specified in each string for the rule to operate ; the stru ctural analysis may be given as fo llows: SA:

of (a): of (b):

X - Nep - Y

X - Nep - Y

The scco nd part of the rule specifics the stru ctural change (SC) by means of numbers referri ng to t he elements in the structural analysis. Thus, given SA above , 1-3 would refer to the elements in (a) , 4-6 to the elements in (b) and the struct ural change might be: SC:

(1-2- 3; 4-5-6)


1-2-3 -4-5-6

No te that , sometimes, it is necessa ry to describe certain conditions that mu st be met in addition to those specified in the structural analysis. Sup pose , for instance, that, for the transformation above to apply, it were necessary to specify that 3 and 4 are not identjcal, we would add a condition: (where 3 '" 4) Note also that , from now on , [ shall call transformational rules operat in g on two strings generali zed transformations and J shall ca ll their output a transform. The structure of narratives such as (19) could be accounted for if we applied the following generalized transformation (GT 1 ) SA: SC:

of (a): X - NSec - Y of (b): X - NSec - Y (1-2- 3: 4-5-6) ---+ 1-2- 3-CCL-4-5- 6 (where 3 and 4 are null)

The rule indicates that a string contain ing a narrative sectio n may be conjoined with a simi lar string by a cluster of co njunctive features. Similarly, the stru ctu re of narratives such as (22) cou ld be acco unted for if we app lied th e fo ll ow ing genera lized transforma tio n(GT 2):



of (a): X - NSec - Y of (b): X - NSec - Y (1-2- 3; 4-5- 6) ---+ 1-2- CF n -4-S-6-3 (where I and 4 or 6 may be null)

The rule indicates how a string co ntain ing a narrative section m ay be embedded into a simi lar string. Finally, t he structure of narratives such as (25) could be accounted for if we app lied the foll owing generalized transform ation (GT 3): SA: SC:

of (a) : NSec - CCL - NSe c - CCL - NSec of (b): NSec - CCL - NSec - CCL - NSec ( 1-2-3-4- 5; 6-7-8-9-10) ---+ I-CF n -6-2-3CFn -8-4-5- CFn -10 (where 2 = 7;4= 9)

The rule indicates h ow narrative sections in o ne string m ay alternate with narrative sections in another string. Of course , the application of other generalized tranfo rm ations of conjoining, embedding and alternation to various pairs o f wellstructured strings would accou nt for other types of structures. Furthermore, and although in the examp les above J have shown that the structure of some narratives ca n be accoun ted for by applying a transfonnational nile once, th ere are narratives the struct ure of which can be accounted fo r on ly through the repea ted appl ication of a generalized transformation or through the use of more than.one such rule. Genera lized transformations must th erefo re operate in such a way that they can app ly not on ly to strings yiclded by component C but also to strings that have already been tra nsformed; moreover, the prod uct of a transfo rmation sho uld be capable of und ergoing furthe r changes. For instance, the structure o f a narrative sll ch as


Jo hn was poor, then he met Joan , t hen , as a resuit , he was rich. T hen, as " result, Pe ter was poor, th en he met Mary , then , as a result , Peter wa s rich. Th en, as a result ,

Jack waS poo r, th en he met Et hel, th en, as " result , Jack wa s ri cll

wo uld be "cco ulll ,'" f'orh y: (1I) ;II)ply ill g( ;T , lo l wo prope r slrin gs;

The Logical Component

Narrative Grammar


(b) applying GT I to the transfonn thus obtained and another proper string. In other words, (28) would be shown to result from th e conjoining of (29)

John was poor, then he met Joan, then , as a result , he was ri ch

and (30)

Peter was poor, th en he met Mary , then, as a result , Peter was rich. Th en, as a result, Jack was poor, then he met Ethel. then , as a result, Jack was ri ch , (30) itself resulting from th e conjo ining of (31) Peter was poor, then he met Mary , then, as a result , Pete r was ri ch and (32) Jack was poor, then he met Ethel, then, as a result, Jack was rich Finall y, note that , like the rules of component C, generalized transformations will be finite in number and may have to be (partially) ordered. At this point , however, and until we have a more thoroughly worked out grammar, it is not possible to deterfnine the order in which they must apply. The structural component of the grammar, consisting of component C plus the generalized transfonnations, accounts for the structure o f any narrative . Besides, it helps account for such feature s of narratives or narrative segments as cohesiveness (all other things being equal, a segment with n conjunctive clusters will be more cohesive than a segment with 11-1), pace (all other things being equal , a segment with n episodes will move more quickly than a segment with 11-1) and hierarchy of event relevance (narrative events, for examp le, are more relevant than non-narrative o nes). The structural compo nent also allows us to compare any two narratives in tenns of their structure . T hus, it would allow us to show that (29), (31) and (32) are structurally identical whereas (I), C), (19) and (25) are structurally different. However, there are many features of narratives that the structural component does not describe. Specifically , it tells us nothing about the propositional conte nt of narratives and does not all ow us to co mpare them in term s of that co ntent.



What I will call the logical component of the grammar allows us to do just th at. It consists o f clusters of semanti c features constituting instances of logical tenllS and propositions (or, rather, primitive arguments and pred icates combining into propositions). Fo r example, it would include such clusters as "then" (i nsta nce of LTt ) , "as a result" (instance of LTc) , or "John eats" (instance of proposition). Once we have a terminal string yield ed by component C (p lu s transformations) , we can substitute for each element ohhe string an appropriate in stan ce of it: given a string like (33)

prop act

+ LT t + prop act

we could substitute the proposition "J ohn ea ts" for th e first element, the logical tem1 "then" for the seco nd, the proposition "John sleeps" for the third, and obtain (34) "John eats" "then" "John sleeps" (34) represe nts the propositional co ntent of a narrative lik e (35) John ate then John slept Substitution obeys the following constraints: (I) it operates on the first (Ietfmost) element of the string first, then on the seco nd element, then on the third, and so on ; (2) a proposition can be a substitute for an element in a string if and on ly if: (a) it is not identical to and does not contradict a proposition preceding it and not separated from it by "then"; (b) it is not identical to a proposition preceding it and imm ed iately separated from it by "then". In other words, each proposition substituted will provid e new information in context and the string obtained after all the substitutio ns are perform ed will be consistent; (3) a proposition can be a substitute for the element N prop statmod if and only if its initial argument is id entica l to that of the proposition substituted for the co rrespo nding N prop stat and it s predicate is a modifi ca tion of that of th e latt er proposi tion :


The Narrating Component

Na"ative Grammar


"J o hn is unhappy"

or (37) "Bill is poor" can replace an N prop statmod if and only if somethin g like (38) "Jo hn is happy" or (39) "Bill is ri ch"

has rep laced the N prop stat of the sa me kern el. The logical component of th e grammar accounts for the propositi o nal co ntent of any narrative . Furthermore, like th e structural co mponent , it he lp s acco unt for features o f narrati ves o r narrative segments such as coh esiveness (all other things being equal, the more features sets of arguments o r predicates have in co mmo n, the more coheSiveness obtains) and it allows us to summarize the co ntent o f any narrative (an adequate summ ary wou ld inc.1ude the co nte nt of the narrative even ts and cou ld include the content of such even ts that are th e causes or consequences of narrative eve nts and such events that lead up to o r proceed from th ese ca uses and co nsequen ces). F inally , the logical component allows us to compare an y two narratives in terms of th eir content. Thu s, we wou ld say that (40)

John felt sick , th en he slept for twenty-four hours, th en , as a result, he felt wel l

and Jo hn was fat, th en he took a pill , then, as a result , he was thin have t he same struc ture but differe nt propositional content whereas (42) John was good and Bill was bad , th en Bill met Mary , then he became good , th en Jo hn met Joan , then he became bad and (43) Bill was bad and Joh n was good , then Bill met Mary, then J o hn met Joan, then Bi ll became good, then J o hn became bad (4 1)

have a different stru cture but t he sam e co nten t. Howeve r, th ere


are many features of narratives which neither the stru ctural no r the logica l component describes. Speci fi cally, neither tells us anyth ing abo ut the narrating in narrati ves and neither allows us to compare them in tenTIS of that narrating. "


I have already indicated that events can be recounted in th e o rder of their occurrence or in a different order. Consid er, for in sta nce (44)

and (45)

John was happ y , th en Jo hn met Mary, then. as a result, he was very unhappy John me t Mary , t hen, as a result , he was very unhap py. Before John met Mary , Jo hn had been happy

I have also indicated that the recounting of a series of events may fo llow, or precede, or be simultaneous with these events: (46) John was walking down. th e street and he saw Joan (47) John will be walk ing down the street and he will see Joan (4 8) Jo hn is now walking down the stree t and he sees Joan

Furthenno re, the same eve nt (or series of events) may be mentio ned several tim es, as in (49)

Mary was good: she helped th e poo r. Mary was good : she took care of th e sick

On th e other hand , a given event may not be ex pli citly stated . Instead of writing John was rich, th en th e Stock Marke t went down , then John became poor I may simply write (5 1) J o hn was rich th en th e Stock Market went down Although (50)


Jo hn bccamc poo r


The Narrating Component

Narrative Grammar

is not explicitly stated and although it is not logically entailed by any (set of) events in (51) - after all, the latter could go on with something like "But John remained rich ... " - (52) is certainly a (very) plausible consequence of (51).12 Note that any event in a narrative may be (said to have been) deleted if and only if (a) it is not identical to (part of) another event; (b) it is not presupposed by another event (as "John was not in the room" is presupposed by "John entered the room"); (c) it is not cotemporaneous with an event by which it is implied (as "John is European" is implied by "John is French"); (d) it is retrievable on the basis of an examination of the remainder of the narrative (and its context). Thus, in (53)

John was very handsome, then he ate meat, then he ate fish, then , as a result, he was very ugly, (54) he ate fish cannot be deleted since it would not be retrievable. Similarly, in (55) John was happy, then he ate fish , then, as a result, he became rich,

(56) he became rich cannot bc deleted: there is nothing in (57) John was happy then he ate fish which allows us to say that John's eating fish caused him to become rich. Of course, given a (cultural) context in which it is a wellknown fact that eating fish usually leads to riches. (56) may very well not be stated explicitly and be retrievable for people familiar with that (cultural) context. Indeed , in many narratives, numerous events are deleted because they can be reconstructed thanks to the context. There are even (non-trivial) narratives in which only one event is expressed: (58)

Gcrald Ford was one of the leaders of the Watergate conspiracy I can very well function as a narrative if something like the following reconstruction - providing new information - is possible: (59) Most people thought that Gerald Ford put the law above himself. then some people found out that he was one of the leaders of the Watergate conspiracy . then , as a result ,


most people thought that Gerald Ford put himself above the law. Finally, and more generally, I have indicated that a narrative consists not only of narrated but also of narrating, and that such features as point of view, signs of the T or signs of the 'you' affect its form and functioning. 1.

Singulary Transfomlations

In order to account for the narrating, it is necessary to devise a new set of rules constituting what I will call the narrating component. The rules will be transformational and will allow us to perform certain changes in certain strings or transforms provided these have certain stmcture. Note that whereas generalized transformations operate on two strings, the new rules operate on a single string and are singulary transformations. The first part of a singulary transfomlation is a structural analysis specifying the kind of string (in temlS of its structure) to which the rule applies. A rule might apply , for example , to any string analyzed as follows: e act + eFt + e act The second part of the rule specifies the structural change by means of numbers referring to the elements in the structural analysis. Given SA above , 1,2 and 3 would refer to e act, eFt and e act respectively and the structural change might be: se: 1-2-3


3- before 3-1

Like in generalized transformations, the structural analysis may contain symbols such as X or Y, standing for any set of elements. Furthermore , it is sometimes necessary to describe certain conditions that must be met in addition to those specified in the structural analysis. Th e differen ce betwee n narrated and narrating orders in a story such as (45) could th en be accounted for if we applied to a specific strin g in its stru ctural deri v,-lti on th e fo ll owin g singulary trans-

fo rmati o n (ST , ):

.' 98


Narrative Grammar


X - Str ep - CF t -Str ep - Y 1-2-3-4-5 --> 1-4- BEFORE 4-2-5 (where Str ep is any string of conjoined episodes and/or narrative episodes; I and 5 are null)

The rule shows that a string of conjoined episodes and/or narrative episodes may appear after another string of conjoined episodes and/or narrative episodes even though the first string precedes the second in time, provided that no episodes or narrative episodes precede the first string or follow the second and that the constant BEFORE is introduced to indicate the original order of elements. Similarly, the fact that narration is posterior in (46), anterior in (47) and simultaneous in (48) would be accounted for if we applied to a specific string in their structural derivation ST 2, ST 3 and ST 4 respectively: ST 2 :


ST 3



ST 4



X - Str ev - Y 1-2-3 --> 1-2- POST-3 (where Str ev is any string of conjoined events; I and 3 may be null) X - Str ev - Y 1-2-3 --> 1-2- ANT -3 (where Str ev is any string of conjoined events; 1 and 3 may be null) X - Str ev-Y 1-2-3 --> 1-2- SIM-3 (where Str ev is any string of conjoined events; 1 and 3 may be null)

The rules show that a narration may follow, precede, or be simultaneous with any string of conjoined events provided that the constants POST. ANT, or SIM are introduced to indicate this fact." Furthermore, the multiple mention of Mary's goodness in (49) could be accounted for with the application of ST 5 : SA: SC:

X - Str ev - Y 1-2-3 --> 1-2-3-REP 2 (where Str ev is any string of conjoined events; 1 and 3 may be null)

The Narrating Component


ST 5 shows that a series of conjoined events can be recounted more than once provided that the constant REP is introduced to indicate the repetition. On the other hand, the deletion of something like (60) from (61)

Peter became rich Peter was poor, then he inherited a gigantic fortune, then, as a result, Peter became rich

would be accounted for with the application of ST 6 SA: SC:


X - CCL -Ne stat - Y 1-2-3-4 --> 1-2 0 -3 0 -4 (where 2 and 3 are retrievable from the string; subscript o indicates that any element it is attached to is not explicitly stated in the narrative)

Of course, the application of other sigulary transformations to various well-formed strings would account for other features of the narrating (point of view, foregrounding, intrusiveness of narrator, signs of the narratee, and so on). Besides, and although in the examples above I have only shown that certain features of the narrating in some narratives can be accounted for by the single application of a transformational rule, there are many cases in which the narrating can be accounted for only through the repeated application of a singulary transformation or through the use of more than one such rule. Singulary transformations must therefore operate in such a way that they can apply to strings that have already been transformed by the narrating component; moreover, the product of a singulary transfonnation should be capable of undergoing further changes. Consider for instance, (63)

John found oil in his field, then, as a result, he became very rich. Before John found oil, he was very poor

and suppose that the structural component had yielded a derivation of its structure and that its propositional content had been inserted. Its ]l alT;1I i]lg would hc (part Iy) accolinted for hy: (a) applying ST! ; (h) a p ply ing ST 2 to 1iiI' I ra ]lSforll! I h II s oil Ia i]l('d. I,' illally , note tltat, likv till' nill's or Illl' sl rlH'lllr;!l C()llq)(lI11'J1t, silll',tllary transformations


Narrative Grammar

will be finit e in number and may have to be (partially) ordered. At this point, however, and until we have a more thoroughly worked out grammar, it is not possible to determine the order in which they must apply. Th e narrating component of the grammar allows us to compare any two narratives (or narrative segments) in terms of thei r narrating. Thus, it wo uld allow us to sho w that (62) and John was very poor, th en he found oil in his fi eld, then, as a result , he became very rich have the same structure and propositional content but a differe nt (63)

narrating; or that

(64) He went to th e theatcr then he went to the movies an d (65) He will go to th e theate r then he will go to the movies have the same structure but a different co nten t and narrating; o r th at (66) Mary ate then she went out and (67) Joan took a shower , then , as a result , she felt good have the sa me narrat ing but a different stru cture and content.


The three co mponents I have describ ed so fa r cannot yield a given narrative; th ey can o nly yield th e stru ctu re, content and narratin g dimension of that narrative. To obtain the latter, an exp ressio n compon ent is necessary. This com po nent will be equivalent to a given language - say. writte n Engl ish - or. rather, to its gramm ar an d it will all ow us to rewrite in that language th e informatio n provided by the oth er components. In other words, if we co nside ra particular o ut put of th e latter as a set of in stru ctions, the expression comp o nent carries out these in stru ctions in writte n English. Should there be any difference between any two sets, two d ifferent narratives arc y ield ed. 14 Suppose. for exa mple . that a de ri vation with th e fo ll ow in g te rmin al strin g has been yielded by t he st ructura l co mponent:

The Expression Component


N prop stat + LT t + prop act + LT t + LTc + N prop stat mod Suppose also that the propositional content for this stri ng is "J ohn is happy " "then" "John meets Mary" "then" "as a result" " John is unhapp y" Finally, suppose that a singulary transfo rm atio n ST 2 has operated on th e string of events recounted . The ex pression component wo uld yield (69) John was happy, then John met Mary, then , as a result, John was unh ap py On the other hand , suppose that the propositi onal content had been (70) "John feels sick" "then" "John takes a showe r" "then" "as a result" "John feels healthy" and suppose that transfonnations ST, and ST 2 had operated , th e ex pression component would have y ielded (71) John took a shower, then , as a result , J ohn fe lt healthy. Before taking a shower , John felt sick (68)

T he narrative grammar I have presented co nsists of four maj or co mponents: (I) a finite set of rewrite rules and generalized transfonna tions accoun ting for all and only narrative structures; (2) a component accounti ng for the propositional co nte nt of any narrative; (3) a finite set of singulary transfonnati o ns accounting for narrating; and (4) a com po nent capable o f translating the instructio ns of the other components into (a signifying system such as) written English. As it stands, the grammar is clearly in need of mu ch elaborati on and its ultimate co nstruction is not for th e immediate future. In particular, to be fully operative , the logical co mponent and th e ex pression co mponent depend on an adequate se mantic th eo ry and an adequate grammar of English, neither o f which is now available. IS As it sta nds, however, the grammar is capab le of assigning a structural descrip tio n to any narrative, of capturing ma ny significant feat ures of th e narrating , of characterizing - to a certain ex tent - th e natu re of the pro'positional co ntent and the way it Ill ay va ry, and o f allowin g for th e (limited) compariso n of any two narratives. In deed , as it stand s, t he gra mll1>II' co nstit utes not o nl y a relal' ivc ly adequat.e descriptive and exp lanatory dev ice but also


Narrative Grammar

a heuristi c one which allows us to ask well-defined questions conce rning narrative and may help us answer them : Is the stuctu re of folktal es sign ificantly different from that of more "sophisticated" stories? What kind of stories - in te nn s o f stru cture and narrating does a given society favor? wh y are some stru ctures which are theoreti ca ll y possible according to the gram mar rarely, if ever, encountercd in practice? what stages does a child go through in developing his ability to narrate? does an emotionally disturbed child or adult consistently favor certai n patterns that a normal child or adult would not favor? The grammar can thu s deepen not only our understanding of narrative but also our und erstanding of man. Of course , eve n if it should be thoroughly worked out, the grammar would be unable to tell us everything (or, sometimes, any thing!) we may want to know about such important aspects of a narrativc as its characters, its th emes, or its point. Although it would indica te who the characters in a narrative might be - we need o nly isolatc thc sets of proposi tions having a logical participant in common and contain ing th e feature "+ human" - and althou gh it would yield a lot of information about the makeup of these se ts, it cou ld not provide information about the con notations th ey would be taken to have or the roles they would be considered to play (se nder o r receiver? helper or o pponent? infonmer or concealer?) Similarly , although it would indicate that certain (sets of) semantic features are more prcvalent than others, it would have no way of specifying the general though ts or ideas they would be t"ken to illu strate. Finally , it would be incapa ble o f describing the point of a narrative sin ce th at point depend s on the contex t in which th e narrative is received (wh at is pointless in certain circumstan ces may be very significant in others; what has a given point on one occasion may have 'a very diffe re nt point on another occasion). In other words, such aspects as th eme, symbol or point arc not th e domain of narrative gramm ar. Beca use they have to be explain ed at le ast partly in terms of a receiver, they are the domain of a theo ry of reading.


Reading Narrative

In recent years, the study of literature in general and narrative in particular has been shifting from a concern with the author or with the tex t to a concern with th e reader. Instead of cstablishing the meaning of a given tex t in term s of an author's intentions or a se t of tex tual patterns, for instan ce, students of literature have fo cused more and more frequ ently on the ways in which readers, armed with expectations and interpretive conventions, structure a text and give it meaning. Ideal readers, virtual readers , implied readers, infonmed readers, competent readers , experienced readers, superreaders, archreaders, average readers, and plain old rcaders now abound in literary criticism and we seem to have entercd an age in which the writer, the writing and the written arc less important than the read , the reading and th e reader. I But what is a reader and what is reading? Very generally spcaking, read ing may be defined as an activity presupposing a text (a set of visually presented linguisti c symbols from whi ch meaning can be ex tracted), a reader (an agen t capable of ex tracting meaning from that set) and an interaction between th e text and the reader such that the latter is able to answer correctly at least some questions about the meaning of the fonm er. Indeed, reading a .text ma y be said to be grossly equivale nt to processing textual d ata gradually by asking questions of the text and answering them on the basis of it. Note that, accord ing to this d efinitio n, reading a tex t and a reading o f a tex t need not bc equivale nt : th e latte r may consist in (and very o ft en docs co nsist in ) a selecti o n, develop me nt and reo rd erin g of th e answers rcac hed durin g th e former. Sim ilarly,


reading a text and responding to it need not amount to the same thing at all. Given a Sllbtext like (1)

John was Jim's brother

I may fantasize all sorts of things about John and Jim - that they both were tall, dark and handsome, that they both liked to play cards, that they both excelled at sports - and I may respond to them accordingly. However, that fantasizing (and response) is not part of my reading (1) - it does not involve any extraction of meaning - even though it may occur while I am reading (1) and even though it may give rise to some of the questions I ask about the rest of the text and to some of the answers I fonnulate. Moreover, it is clear that not any set of visually presented linguistic symbols can be read: some such sets - a series of randomly picked letters scaltered on a page, for example, may not constitute a text. No meaning can be extracted from them. They do not make sense or, at best, they merely trace some of the limits between sense and nonsense. Likewise, it is clear that it is not enough to recognize visually presented symbols as linguistic in order to be a reader. Identifying a series of symbols as specific graphemes (corresponding to specific sounds) is not the same as extracting meaning from them and I would not say, except as a joke, that I read German (or Rumanian, or Russian) very well but that I did not understand it. Furthermore, my reading a text implies that the meaning which I extract from it is at least partly conditioned by it. Given (2) and

The Code of Written Narrative

Reading narrative

John was happy

(3) John was old it would be difficult for me to read them as meaning respectively (4) John was unhappy and (5) John was young

In other words, the answers I bring to the questions I fonnulate must not contradict the text. Finally. reading a text implies that the questions asked arc relevant. The notion 01" relevance demands much more attention thon I can give it here. I will simply note that


a question is relevant if its possible answer is relevant. that is, if its possible answer carries old and new information pertaining to the topic(s) developed by the text. Some questions are not relevant because they have nothing to do with the extraction of meaning: to ask how many consonants there are on the first page of The Sun Also Rises will not prove helpful for reading that page, for understanding it; other questions, as the following mock-riddle underlines, are not relevant because they cannot be answered on the basis of the text: (6)

The third deck of a ship is 600 feet long and 200 feet wide. Howald is the captain?

and still other questions are not relevant because we already know their answer. Of course, reading a text in no way implies that all the relevant questions are asked and all the possible answers found. Indeed, it frequently implies the opposite. The set of relevant questions (and answers) is often a very large one and, as I read (in order to keep on reading!), I have to select certain questions rather than others. Of course too, learning how to read is - among other things - learning how to ask more and more relevant questions. An ingenious reader is not only one who can find new answers to old questions but also one who can think of new questions.


Relevant questions may pertain to the denotational meaning of the symbols making up a (narrative) text, their connotational meaning, their thematic or symbolic meaning, their functional meaning, their significance in temlS of other textual or nontextual worlds, and the connections that can be made among the answers arrived at. Now, it is obvious that to read and understand a narrative, to ask questions and answer them, we must know much of the code in which it is framed. We would not be able to read a novel in English, for instance, if we did not know any English. To a certain extent, the code of any written narrative is thus linguistic in nature. lIowever. it is not monolithic. It conjoins, combines and orders u set of codes or sub-codes, of ~roups of


Reading narrative

norms, co nstraints and rules in terms of which the narrative is (more or less) decipherable and understand able. Suppose , for example, that th e following sentence appeared in a tale: (7) He shook his head from left to right several times. In o rd er to interpret it , in order to answer a question like (8) What did his shaking his head thus mean ? it is not enough that I know (the code constituted by) English ; I mu st also know the mea ning of th e action related by th e sentence. Acco rding to so me cu ltural cod es, this action implies affirmation or approval; and according to others, it implies negation, denial or disapproval. My co mprehension depends, th erefore, on the cultural code which I take to be framing the tale , th e code in terms of which the actions, events and situations related by a set of sente nces mean some thing in a certain cultural contex t. But th ere is more. Many, if not all , narratives can be considered to lead from one or several questions or mysteries (why was the you ng man murd ered? who was th e murderer? how was he caught?) to thei r answer or solution (th e young man was murdered for mo ney; the murderer was his best friend; he was caught thanks to an in ge nious stra tagem). Similarly, many narratives present situatio ns and activities which we ca n group into se ts having certain names because we know how such situatio ns and activities combine to yield larger ones. Indeed , as I have already indicated , this is partly what allows us to summarize a given novel or story. Moreover, many narratives co ntain various ele ments wh ich may fun ction sy mbo lically (we take them for what they arc and also for something greater. more general and /or more fundamental which they rep rese nt): given the appropriate co ntext , the account of a trip from France to Algeria may also act as that of a spiritual quest (L'I II/II/oraliste) , the description of a garden may evoke paradise (Candide) and references to snow may fu nction as references to salvation (La Chute). Roland Barth es consequently theorized that , whenever we read a narrative, we organize it and interpret it according to several codes o r sub-codes : a linguistic code and a cu lt ural code, of course; but also a herm c neuti c codc , in te rms of which ce rtain parts of a given tex t functi o n as an enigma to be so lved and ce rtain ot hers as a so luti on to that e nigma , o r th e beginnin g o f a

The Code of Written Narrative


solution , or a false solutio n ; a proaire tic code, th anks to which we group certain narrative ac tions into one sequence, ce rtain o thers in to a diffe rent sequence, etc; a sym bolic code, acco rding to which we perceive the symbolic dimensions o f variou s passages; and so on and so forth.' Th e code of written narrative is a combination of all these codes. Note th at , whereas we know quite a lot - though by no means everything - about th e nature and fun ctioning of the linguistic code, we know very littl e about oth er codes or sub-codes framing a narrative. Co nsider th e following , for in stan ce : (9) It was nine o'clock. The young woman was sta nding on a dark and filthy corn er of 17 th street . At ten past nin e, it started to snow. Yet the woman did not move. She was a spy and she had to mee t her boss at twent y past nin e. She had deci ded to kill him because she hated him but she did not know how she was going to do it and she was nervous. At twenty past nine, the boss appeared. It had stopped snowing. She took out her gun and shot him dead. For many receivers of this narrative message, the seco nd senten ce no doubt leads to certain questions (who is this young woman? what is she doing on that dark and fi lth y corn er?) ; th e fourth sente nce fun ctions in a similar way (why doesn't she move eve n though it 's snowing?); the fifth sente nce represents an an swer to the qu estions raised ; the sixth sentence leads to new qu estions (w ill she be able to kill him? how will she do it?); and the last se nte nce answers them. But the first , third , seventh and eighth se nten ces in no way co nstitute or imply enigmas or solutio ns to these en igmas.' Wh y is it that certain narrative passages are thus mad e to fu nction in te rms of a hermeneutic code and certain others are no t? We can, at best , offer only the sketch o f a sol utio n. A passage can functio n herm eneuticall y if it suggests or asse rts that there is a mystery to be solved; if it fo rmulates this mystery; if it announ ces a (possible) so lution ; and if it co nstitutes that so luti on , contributes to it or represe nts ll n obstacle to it. To put it d iffe re ntl y, a passage can fun ction herm e neuti call y if it e nco urages qu estio ns abo ut th e id entity of so meo ne 0 " so me thin g (w ho is it ? what is it ?), fo r examp le ilHou gh a te ,'II' whose ref ' re nce is un know n ; qu esti o ns


Maximal Reading, Minimal Reading, and Questions

Reading narrative

about the state of someone or something (how is it that?), for instance by implying that a given set of circumstances is abnormal and demands an explanation; questions about the outcome of an action or situation and the way this outcome is effected (how will it end up? how will it come about?); or if it provides (partial and/ or possible) answers to thcse questions. Problems pertaining to the organization of a given narrative in terms of a proairetic code, a cultural code, or a symbolic one are also difficult to solve. Why is it, for example, that certain passages are viewed according to a proairetic code and certain others are not? Onec again, I will only provide a general answer and say that events can function as (parts of) proairetic units if they introduce (intcgral elements of) an initial situation to be modificd in the world of the narratcd, or if they constitute activities (which may combine into larger activities) modifying the (modified) initial solution, or if they are the causes or consequences of such activities or in any way relevant to them, or if they constitute (integral elements of) the final state of the initial solution" The very complexity of the narrative code explains in part the variety of responses to and interpretations of a given text. In any narrative communicaton, it is not the narrative code as a whole whatever it may be - which intervenes but rather what the sender and the receiver have assimilated of the code and, more particularly, what each has selected from his personal stock to encode or decode the message. These subsets of the code have more or less in common but need not be identicaL 5 Given a narrative message, the sender does not always know everything that it says to the receiver, everything that it carrics for him, everything that it reveals or betrays, and vice versa. Two individuals - or the same one on two separate occasions - may therefore interpret that message differently. Though they may both view various passages as constituting riddles, the passages may not be the same; moreover, though they may both group certain events into similar sequences and give the sequences similar names, they may group other events very differently or use different names for tile same sequenees (as I have pointed out. we do not all summarize a story-line in exactly the same way); and though they may, at times, perceive similar thematic or symholic dimensions, they may, at other times, differ


greatly as to the themes or symbolism involved. Besides, one of them may consider a given passage in terms of a proairetic code, for instance, and the other one in terms of a hermeneutic code. A passage such as (10) John entered a small cafe, asked for the telephone, dialed a number, spoke rather briefly, then hung up and left could be viewed as a sequence "Telephone Call" or could lead to such questions as (II) Who did John caU? (12) Why did he call from a cafe? (13) Whom did he speak to? and so on 6


The relevant questions that may be asked while reading are clearly varied in kind and their number, in the case of some texts at least, may be infinite. For example, given the following passage from Perrault's Le Petit Chaperon rouge: (14)

"Little Red Riding Hood left immediately to go to her grandmother, who lived in another village. Passing through a wood, she met Brother Wolf who fclt very much like eating her; but he did not dare because of some woodcutters who were in the forest He asked her where she was going. The poor child, who did not know that it was dangerous to stop and listen to a wolf, told him: '[ am going to see my grandmother and bring her a girdle-cake with a little pot of butter that my mother is sending her . .. '"

[ may ask such questions as (15) (16) (17)

Where did Little R"d Riding lIood go? What for? Why didn't Brolhn Wolf,'al lin')


( 18) ( 19) (20) (2 1) (22) (23) (24) (25) (26) (2 7)

Textual Constrain ts

Reading narrative

Will he ge t another chance? Will he succeed? Wh at didn't the poor ch ild know? What is a girdle-cake? Does it have any special connotatio ns? What are th e connotations of the name ' Brother Wolf'? What abou t ' Little Red Riding Hood'? Is Brother Wolfs desire to be understood as sex ual? Will th e child 's gra ndmother protect her? What is the grandmother's name?

and so forth. Naturally, even if all of my qu estions have a certain relevan ce as I am reading, I may not find any answers to them: . (27) , fo r instan ce, will always remain unanswered . In view of the above , it is not easy to determine what reading max im ally wo uld be: we can on ly say wit h certai nt y that , for reasons which I have already indica ted in passing, it does not always consist in asking all th e relevant questions and coming up with all the right answers.' Nor does it always make very mu ch sense to speak of the total reading (as finished product rather than o ngoing pro cess) of a given tex t sin ce the se t of relevant questions and answers pert aining to the meaning of th at text may be infinite. On th e other hand , it is not easy to determine what reading minim ally would be either: we ca n o nly say that it entails the understanding o f th e linguisti c meaning o r, to put it in other term s, it ent ails th e capacity to parap hrase and summ arize the denotat io nal conten t of the text (and of its constituent parts). Note that , in the case of narrati ve texts , some questions are more narratively relevant than others: th ey spec ifi cally pertain to features characteri sti c of narratives rather than non-narratives. Questions about the plot. for example, questions abo ut the chrono logy of the eve nts presented, quest ions about what has happened and what will happen are narrati vely relevant whereas questions about the conn o tat ive meaning o r symbolic significance of a given eve nt arc not (at least, not necessa rily ). Ind eed . reading a tex t narrati ve ly (reading it "for the story") means asking above all question s th at have narra t ive releva nce - q Ut!s t io ns generall y referrin g

ba ck to the proa ireti c dim ensio n and th e sto ry-line

and findin g


answers to them . If attempting to read a narrative maximally involves questions and answers abo ut any and all of its meaningful aspects, reading it minim ally involves questions an d answers abo ut what happens. Given Le Petit Chapero n rouge , for example. and even though I may have gathered a lot o f interesting data about the simil arities between the mother and the grandmother, th e symbolism of the wolf and the heroine's Elec tra complex , [ will not have read it narratively if I have not p rocessed that the wolf eats the grandmother, gets into he r bed, then ea ts th e granddaughter too. On the contrary, I will have read it narratively merely by focusing on the chro nologica l sequence of events and understanding it.


Certain (sub) texts allow on ly one co rrect answe r to some of the questions asked. For in stance, a text like (28) John was twenty-five allows on ly one correct answer to (29) Howald was John? a text like (30) Jo hn had no siblings allo ws only one correct answer to (31) How many siblings did John have? and a text like (32) John was very tall allows on ly one correct answer to (33) Was John very sho rt? Should anybody looking at (28), (30), and (32) answer (29) , (3 I), and (33) wit h (34) Jo hn was seventy-eigh t (35) Jo hn had three broth ers and t ln ee sisters and (36) Yes. John waS ve ry sho rt respect ivcly. we wou ld Illost probab ly no t co nclude that he was read ing (28). (30), all d (3 2) ill a llighl y idi osy ncra ti c mann er but .


Reading Narrative

rath er, that he was misreading them or not reading them at all. Thus to a certain extent , at least, and as [ have already suggested , the text [ read ac ts as a con straint on my read ing. Note th at the tex t may allow only one correct answer to a given qu estion without spelling the answer out. Given (37)

Harry was five years older than Joan and Joan was (wen ty-five


All professors are crazy and Mary was a professor

or it is obvio us th at Harry is thirty and Mary is crazy. Note also that so metimes the tex t not o nly provides answe rs to variou s questions but explicitly asks qu es tions that a reader himself might have askeel anyway. [n A la recherche du temps perdu, for example , an ex quisite pleas ure in vades Marcel's se nses when he tastes a petite madeleine soaked in tea and several questions are raised in relation to th is ex traordinary event: "Whence co uld it have come to me, th is all po werful joy? [ was conscious that it was connected with the taste of tea anel cake, but that it infinitely tran scended those savours. co uld not. ind ecd. be of the same nature as theirs. WMnce did it come? What did it signify? How could [ se ize upon and define it?": in Les Thibault , th e narrator finds yo ung Jacques' power over Daniel de Fontanin remarkable and writes: "Why didn't this big thoughtful boy rebel again st the urchin 's influence? Didn ' t his education and th e fre edom he enjoyed give him an indi sp utable droit c/'all/esse over Jacques?": and in Le Pere Coriot , Poiret's appearance see ms to require ex planation : "What kind of work could haw thus shrivc led him up? what kind o f passion had darkened his bulbous fac e ... ?" Ind eed . there is at least one modern novel , Robe rt Pingel's L'I"'1uisitaire. which large ly consists of such explicit questions and an swers to them. If th e text constrain s my reading by the unequivoca l answers it brings to some of m y qu es tions, it al so con strain s it in variolls o th er ways. Thus. it nwy an swer my questions (or th e questions it itse lf asks) more or less quickly. In (39)

John was gett in g impati ent. He had been trying to reach Jim for over an hOllr now,

Textual Constraints



Why was John getting impatient?

is answered immediately ; but in th e case of A la recherche du temps perdu , a reader reading the novel from the first page to the last has to wait for a very long time until his (and Marcel 's !) question s abo ut the petite madeleine are answered. In fa ct, many narratives can be viewed as spaces stretching between a question and its answer and their unfolding is partly characte rized by the kinds of delays they bring to the answering of the question ' In a classical de tective story, for instance, the most important early question often centers around the identity of the murderer and the correct answer usually comes only after several other suggested answers have proven unsatisfactory. F urthermore , the text may force me to update more or less frequentl y t he infonnation [ gather as I read by introdu cing data which make some of the answers I have reached (and some of the questions [ have asked) obsolete. Consider the following: (4 1)

John had many friend s but then he committed a crime and lost them all (42) Joan very much wanted to go to ·the party then she changed her mind (43) Jim was twenty-three and he was desperately in love with Mary but she wo uldn' t even look at him. Three years passed , three years of end less humiliation and despair. One day, as Jim was walking down the street , he saw Mary sitting dejectedly on the curb Questions (44) Did John have many friend s? (45) Did Joan want to go to the party ? and (46) Howald was Jim would get different answers at d ifferent points in my reading of (4 1), (42) and (43) respec tively. Of course, th e upd ati ng of information is particularly imp o rtant whi le reading narrative texts sin ce th eir tem poral d im ensio n o ft en e ntail s very many changes in the situati o ns and characters prese nted . Note that , so me tim es, a tex t pro vid es an un eq uivoc al answer

\ 114

Metanarrative Signs

Reading Narrative

which it later modifies because it had been the wrong o ne. Suppose, for ex ample, that the iilformatio n was supplied by a narrator who lied then decided to tell th e truth or by one wh o thought that he understood a situation thell found out that he did not. I may be told that John is twenty-th ree, then that he actually is twenty-seven , th en that he is only sixtee n but looks old for his age; or that Mary loved Joan then th at she really hated h er but disguise d her hatred very well. In such cases, I may feel th at I have been misled, especially if I think that the mo dification indicates the narrator's bad faith rath er than his igno rance. There are , o f course, other reading circumstan ces in wh ich I may feel cheated : instead of giving me wro ng infonll ati on, the narrator may omit information th at is esse ntial, or he may give me too mu ch informati on and lead me on a tangent; or again , he may allow me to reach ce rtain conclusions o nly to tcllme later th at th ese conclusions, though most plausible, arc not correct.

On th e other hand , a tex t may prove to be parti cularly helpful rathe r than dece ptive . It may remind us of information it had given us previo usly if this information is necessary to the understandin g of some new event or situation : think o f such sentences as "The reader will recall t hat. . . ," " It is important to remember tilUt. ... " or "As we po inted out earlier. ... " It may explain how newly provided data , seemingly conflicting with what data we have already processe d , is actually not at all in consistent with it. Whe n, in A la r ech erche dll temps perdu , Swann , who had been portra yed as most delicate, modest and disc ree t, ac ts in a vulgar manner, the narrato r quickl y notes that there is no contradiction: after all , "who has no t seen very unpretentious ro yal princesses ado pt spontan eou sly th e language of old bores?" ; and , in fournal d'lIl1 Cllre d e campagne , when the protagonist. who is inept in dealin g with people , suddenly gets the uppe rhand in his confrontation with the Countess , there is no inconsistency either: the text makes it clear that God is o n his side. Through th e use of metanarrativ~ signs. the tex t ca n al so summarize for LI S a long series of eve nts, or give us the gist o f a complex argume nt , o r indicate the relative significan ce o f vari ous actions, or revea l th e sy mboli c ' implicati o ns of diffe rent situati ons. In fac t , a text can co mm ent


appropriately on any aspect of its constituent parts and partially do the reader's work for him.


When the subj ect of a discourse is language, we sometim es say that the discourse is metalinguisti e. Similarly , when the subjec t o f a discourse is narrative, we may say that th e discourse is metanarrative. According to this ve ry general de finition of the term , there are man y kinds of discourse which may be me tan arrative : a philosophical essay on the ontology o f narration, for instance, a history o f the Russian novel, or the present study. Obvio usly, a verbal narrative itself may be metanarrative : a given tal e may refe r to other tales; it may comment on narrators and narratees; or it may discuss th e act of narratio n. Ju st as obviously , a parti cular narrative may re fer to itself and to th ose elements by which it is constituted and communicated. Consider th e fo llowing, for ex ample: (4 7)

"There was in all this, as may have bee n obse rved, one perso nage concern ed , o f wh om, notwithstanding his precario us position, we have appeared to take but very littl e notice; this personage in M. Bonacieux, the res pectabl e martyr of the political and amorous intrigues whi ch entangled themselves so nicely together at this gall ant and chivalric period. Fortunately , the reader may remember, o r may not remember, fortunately , that we promi sed not to lose sight of him. " (Les Tro is MUlIsquetaires) (4 8) "Perh aps I shall eliminate th e preceding chapter. Among other reasons, there is, in th e last few lin es, something th at might be construed as an error on my part." (Epitaph of a Small Winner) (49) "Thus, ge ntle reader, I have given thee a faithful histo ry of my travels fo r sixteen years an d above seven months : wherein I have not been so studi o us o f o rnament as o f t ru t h . I co uld , perh aps, li ke o th ers, have asto nished thee wit h stran ge impro bab le 'a les; bu t I rath er chose to re late


Reading Narrative

plain matter of fact, in the simplest manner and sty le; because my principal design was to inform and not to amuse thee ." (Gulliver's Travels) These sel f-referen tial aspects of narrati ve have attracted quite a lot of attention recentl y and some theorists have successfull y argued tha t many a n arrative ultimately discusses itself and actually constitutes a metanarrative. 9 There is anothe r possible definition of the term metanarrative , a stri cter and perhaps more meaningful one. In a famous statement on linguistics and poetics, Roman Jakobson presented a rapid survey of the constitutive factors in any act of verbal communication: The addresser sends a message to the addressee. To be operative the message

requires a context referred to (,referent' in another somewhat ambiguous nomenclature) , seizable by the addressee, and either verbal or capable of being verbalized; a code fully , or at least partially, common to the addresser and e2, e3" .e n occurring at time t or at times tl, t2, t3' .. t n respeetively, we speak of eUipsis when one of the events is not mentioned. Cr. Gerard Geneue, Figures Ill : 128 ff. and Seymour Chatman, "Gene tte's Ana lysis of Narrative Time Relatio ns, L '£'sprit Crcatcur XIV (Win ter 1974): 353- 368. The term 'st retch' is Chatmun's. Of course, we do not usu ally process a text lIS e lliptica l when the possible breaks o r lacunae have no COnSC!llI cnccs. Cf. Gurure! Genettc, FiNllrcl' /II : 14 5- 148.



2. 3.

4. 5.








For a similar proposal, see Gerald Prince, "Towards a Nonnative Criticism of the Nove l," Genre 1/ (1969): 8, and Charles T. Scott, "On Defining the Riddle: The Problem of a Structural Unit," Genre /I (1969): 137. For a discussion of transforms of elementary strings, sec Noam Chomsky, Syntactic Slntctures (The Hague, 1957) and, by the same author, "A Transformational Approach to Syntax" in Proceedings of the 1958 Conference on Problems of Linguistic Analysis in English, A.A. Hill, cd. (Aust in, Texas, 1962), pp. 124- 158. Note that in the course of this study and for the sake of convenience, I may represent a proposition by a sentence which is not the transform of a single elementary string. From now on, 1 will take 'event' to mean 'even t or situation'. On this subject, see Ellen r. Prince, "Be-ing: A Synchronic and Diachronic Study," Transformations and Discourse Analysis Papers (81) (University of Pennsylvania, 1970). I am not making distinctions between states and processes, happenings and actions, etc., because such distinctions arC not relevant to my discussion. For a detailed discussion of exposition, sec Meir Sternberg, Expositional Modes and Temporal Ordering in Fiction (Baltimore, 1978). E.M. forster, Aspects of the Novel (London, 1927), p. 130. Modifications have been studied in detail by Tzvetan Todorov, who calls them 'narrative transformations.' See his Grammaire du D~cam~ron (The Hague, 1970) and "Les Transformations narratives," PcXtique (3) (1970): 322-333. Cf. Tzvetan Todorov, Poitique: p. 82. For another defi ni tion of sequence, see Roland Barthes, "An Introduction t o the Stru ctural Analysis of Narrative," 252 ff. On the notion of character, sec, among others, Sarin Alexandrescu, Logique du personnage (Paris, 1974); Roland Barthes, S/Z (Paris, 1970): 101-102; Claude Bremond, Logiqlle du redt (Paris, 1973); Seymour Chatman, "On the FormalistStructuralist Theory of Character," Journal of Literary Semann'cs (I) (1972): 57-79; E.M. Forster, Aspects of tile Novel, pp. 69-125; Northrop Frye, Anatomy of Oiticism (Prin ccton, 1957); Jam es Carvey, "Charactcrization in Narrative," Poetics VII (1978): 63-78; Philippe Hamon, "Pour un statut scmiologique du pcrsonnage," Litttrature (6) (1970): 86-110; W.J. Harvey. Character and the Novel (London, 1976); Tzvetan Todorov, Grammaire du Dtcamtron; and Michel Zeraffa, Personne e t personnage (Paris, 1969). According to the models of Propp and Greimas, more than one character may fulfill the same function (thcre cou ld be three opponents and four helpers, say). Besides, the same charactcr may fu lfill more than one function (he may be both subject and receiver, or object and opponent, and so on.) Finally, not cvery function need bc fulfilled by a character (various social forces may constitute the sender, or the object, or the opponent, and so forth). Sce Vladimir Propp, Mor· phology of the Folktale (Bloomington, 1958) and, by A.J. Greimas, Stmantique stnlcturale (Paris, 1966), pp. 172- 191 ; Du Seils (Paris, 1970), pp. 249- 270; and "Les Actants, les acteu rs et les figures" in Claude Chabrol, cd., Semiotique narra· tive et tex tuellc (Paris, 1973), pp. 161-176. Cf. Phillippe Hamon, "Qu 'cst-ce qu'unc description?" Poerique (12) (1972): 465~ 485. .








See, fo r example, Barbara Lco ndar, "Hatching Plots: Genesis of Story making" in Dav id Perkin and Barbara Leondar, cds., 71le Arts and Cognition (Baltimore, 1976) and Peter F. Neumayer, "The Child as Storyteller: Teaching Literature through Tacit Knowledge," College English XXX (1969): 515 - 517. See Vladimir Propp, Morphology of the Folktale and Alan Dundes, The Mor· phology of North American Indian Folktales (Helsink i, 1964). On requirements for grammarS, see Noam Chomsky, Syntactic Structures and, by the same author, "O n the Notion 'Ru le of Grammar'," Proceedillgs of the Twelfth Symposium in Applied Mathematics XII (1961): 6-24; "Some Methodological Remarks on Generative Grammar" Word XVII (1961): 219-223; "A Transformational Approach to Syntax"; Aspects of the Theory of Syntax (Cambridge, Mass., 1965). See also Emmon Bach, All Introduction to Transformational Grammars (New York, 1964) and Paul Postal, COIIStituefit Structure. A Study of Syntactic Description (The Hague, 1964). On degrees of grammaticalness, see Noa m Chomsky, "Some Methodological Remarks on Generative ,Grammar." ' See, for instance, Claude Bremond, Logique dll recit; Benjamin N. Colby , " A Partial Grammar of Eskimo folktales," American Anthropologist LXXV (1973): 645-662; Tcun A. van Dijk , Some Aspects of Text Grammars: A Study of Theor· etical Linguistics and Poetics (The "lague, 1972) and "Narrative Macro-Struct ures. Logical and Cognitive Foundations," PTL 1 (1976): 547-568; Lubomir DoleteI, "from Motifemes to Motifs," Poetics. (4) (1972): 55-90; Gcra rd Ge not Problemes de calcul du redt (Universite Paris X-Nan terre, CRLU (l0) (1976) and Problemes de calcul du redt II (Universite Paris X-Nan terre, CRLU (12) (1977); Robert A. Georges, "Structure in Folktales: A Gencr~ tive-Transformational Approach,'" The Conch II (2) (1970): 4-17; Claude Uvi-Strau ss. Anthropologie structurale (Paris, 1958) ; Thomas G. Pavel, La Syntaxe narrative des tragtdies de Comeille (Paris, 1976) ; Maric-La ure Ryan, "Narration, gencration, transformation: La Grande Breteclie de Balzac," L'Esprit Cr~ateur XVII (Fall, 1977): 195-210; Tzvetan Todorov, "La Grammaire du rccit,"'.angages (12) (1968): 94-102 and Grammaire du Decameron; Picter Dirk van der Yen, From Narrative Text to Narrative Structure (Dordrecht, February, 1978). ror a general discussion of the field, see Jonathan Culler, Structuralist Poetics. Structuralism, l.illguistics, and the Study of Literature (Ithaca, 1975); William O. Hendricks, Essays on Semiolinguistics and Verbal Art (The Hague-Paris, 1973); Gerald Prince, "Narrative Signs and Tangents," Diacritics (Fall, 1974): 2- 8; and Robert Scho les, Stnlcturalism ill Literature. See, for example, Claude Chabrol, cd., S~miotique narrative et texnlelle. For detailed critiques of various narrative grammars, sec, among others, Claude Bremond, Logique de rceil , pp. 9- 128; Bertel Nathhorst, Formal or Structural Studies of Traditional Tales, (Stockh olm, 1969); and Marie-Laure Ryan , "Growing Texts on a T ree," Diacritics VII (4) (Win ter 1977) : 34-46. Th is gralllma r i .~ a great ly modified version of the onc presented in Gerald Prince, ;1 Grammar of Stories (The 1I :lglll.J- I)aris, 1973) . llccau se I am main ly intcrested in verbal (writt en) narrative lind tor th e sake o f convenience, th e narratives I use as ex amples art.l ull ve rbal. Thill duct! no l l1l eu n lhal the rules I propose cannot possibly be u!lplled to no n·vc,hlll 1l1lr l[lll vl1~ UI Ihul Ih e co ncepl s th ey ex press arc not tmll SICluhle tn t htllli .

;: ..

170 7. 8.



11. 12.




,, '


Notes Cf. Louis Hjelmslev, "La Stratification du langage," Word X (1954): 163-188. No more than what I defined as a minimal story in A Grammar of Stories, pp. 1637. As the structural component will show, although any story is a narrative, not any narrative is a story. In constructing my grammar, I follow Chomsky's Syntactic Structures, "On the Notion 'Rule of Grammar','" and "'A Transformational Approach to Syntax"; Emmon Bach, An Introduction to Transformational Grammars; and Paul Postal, Constituent Structures. In my elaboration of transformational rules, I once again follow Naaru Chomsky and Emman Bach. For an excellent discussion of the use of the word transfor" mation in narratology, see Thomas Pavel, La Syntaxe narrative des tragedies de Corneille, pp. 131-147. Note that onc could easily account for conjoining and embedding in terms of rewrite rules. So far, however, I have been unable to account for alternation without the help of transformations. Therefore, and for the sake of consistency, 1 prefer to account for any complex structure with transformational rules. Besides, a bipartite structural component has the advantage of underlining the fact that many a narrative can be considered to be made up of smaller narratives, Together, the structural and logical components account for the narrated. Given a stative cvent A followed by an active evcnt B, we take B to lead to a modification of A unless the text explicitly indicates otherwise. Moreover, in the absence of any other information, we will select the most plausiblc modification of A as the one obtaining. If ST 2 , ST 3 , or ST4 are not applied, we get a narrative in which it is impossible to determine the relationship between time of narration and time of narrated. The expression component could be equivalent to a non-linguistic signifying system. Furthermore, it is quite conceivable that some such systems could not carry out all of the instructions. Recent work in text-grammar and narratology shows that the obstacle is not insuperable and that perhaps what is needed most is patience. See, for example, Teun A. van Dijk, "Philosophy of Action and Theory of Narrative,'" Poetics V (1976): 287-338; Lubom·ir Dolezel, "Narrative Semantics," PTL 1(1976): 129151; Thomas G. Pavel, '''Possible Worlds' in Literary Semantics," The Journal of Aesthetics and Art Criticism XXXIV (1976): 165-176; and Jinos S. Petofi, Vers une theorie partielle du texte (Hamburg, 1975).


2. 3.



6. 7.

8. 9.







See, for instance Roland Barthes, S/Z; David Bleich, Readings and Feelings; Introduction to Subjective Criticism (Urbana, 1975); Wayne C. Booth, The Rhetoric of Fiction; Michel Charles, Rhitorique de la lecture (Paris, 1977); Jonathan Culler, Structuralist Poetics; Stanley Fish, "Literature in the Reader: Affective Stylistics,'" New Literary History 11(1970): 123-162; Norman Holland, 5 Readers Reading (New Haven, 1975); Wolfgang lser, The Implied Reader: Patterns of Communication in Prose Fiction from Bunyan to Beckett (Baltimore, 1974); Walter 1. Ong, "The Writer's Audience is Always a Fiction," PMIA XC (January 1975): 9:....21; Michael Riffaterre, Essais de stylistique strnctura[c (Parh, 1971); Walter Slatoff, With Respect to Readers: Dimensions of I.iterary U{'Sf1VI1S(, (Ithaca, 1970). Note


16. 17.

I H.


that much of what I will say about reading applies to receiving and interpreting narratives which do not adopt written language as a medium. See Roland Barthes, S/Z: 23-28 and, by the same author, "Analyse textuelle d'un conte d'Edgar Poe" in Semiotique narrative et textuelle, pp. 29-54. Of course, some receivers may wonder what the significance of the time or the snow is in the overall narrative strategy; but it is not the same thing. On the proairetic and hermeneutic codes, see Roland Barthes, S/Z; Jonathan Culler, Strncturalist Poetics, pp. 205-224; and Josue V. Harari, "The Maximum Narrative: An Introduction to Barthes' Recent Criticism," Style VIII (Winter 1974): 56-77. The same is true of any act of verbal communication. Many non-narrative messages too are susceptible of various interpretations; but not messages framed in monolithic codes (the code of traffic lights, for example). Of course, psychologists may be able to tell us how much information we can process at a time and to help us clarify the notions of maximal reading, See Roland Barthes, S/Z: 81-83. See, for instance, Roland Barthes, S/Z: 219; William Cass, Fiction and the Figures of Ufe (New York, 1970), pp. 24-25; Tzvetan Todorov, Poetique de la prose (Paris. 1971), pp. 66-91. Roman Jakobson, "Closing Statement: Linguistics and Poetics" in Style and Language, Thomas Sebeok, ed. (Cambridge, Mass, 1960). p. 353. Some scholars prefer to speak of seven factors: Dell Hymes, for example, divides context into topic and setting. See "The Ethnography of Speaking" in Readings in the Sociology of Language, Joshua A. Fishman, ed. (The Hague, 1970), pp. 110-113. Cf. Roman Jakobson, "Closing Statement: Linguistics and Poetics," pp. 353-357. Of course, a verbal act may have more than one major function. For a good discussion of metalinguistic statements and signs, sec Josette ReyDebove, Etude linguistique et semiotique des dictionnaires fran~ais contemporains (The Hague-Paris, 1971), pp. 43-52. For a similar definition, see Gerald Prince, "Ramarques sur les signes metanarratifs,: Degres (11 ~ 12) (1977): el-el0. See also Philippe Hamon, "Textc littcraire et metalangage," Poetique (31) (1977): 261-284 and Pierre van den Heuvel, "Le narrateur narrataire ou Ie narrateur lecteur dc son propre discours," Agora (1415) (1977): 53-77. In other words, a narrator's intrusion or an explanation does not necessarily constitute a metanarrative sign. Note that all of the explanations by the narrator (including non-metanarrative ones) similarly function as indications on his relationship with his narratce. More generally, all of the explanations in any text (including non-narrative texts) provide information on the relationship between the addresser and the addressee. My description of what a reader brings to a text is, of course, very incomplete. On the legibility and readability of texts, sec Roland Barthes, S/Z; Jonathan Culler, Structuralist Poetics; Philippe Hamon, "Un Discours constraint," Poetique (16) (1973): 411-445 and "Note sur Ie texte lisible" in Missions et demarches de fa critique. MflallKcs offerts ali Professeur J.A, Vier (Rennes, 1973), pp. 827-842; Tzvetan Todorov, "line Complication ue textc: Les Illuminations," POCtiquc (34) (197H): 241-253 and "I.a I.ecluf(' COllllllt' consfrllc1iot]," /'ot-/iqul' (24) (1975): 417-425. J 0I1il 1~~uvjll': asidl' llll' IUl"lur o! knj.llh ~dlll(Hlgh il j.~
View more...


Copyright © 2017 KUPDF Inc.