University of Zaragoza, 1990
(Online edition 2005)
0. Introduction to the levels of structural analysis of the narrative text
1. The Structure of the Fabula, I (Aristotle)
2. The Structure of the Fabula, II (Tomashevsky, Bal)
3. Time Structure in the Story (Genette)
4. Story: Distance
5. Story: Perspective
6. Narrative discourse: Narrators
More Structural Narratology
0.1. Definition of Narrative
0.2. On Definitions and Concepts
0.3. The Linguistic Approach to Narrative
0.4. Basic Directions in Analysis
0.5. Identity and Difference Between Narratives and Narratological Models
0.6. Vertical Analysis in Bal
0.7. Vertical Analysis in Aristotle
A narrative is a semiotic representation of a series of events connected in a temporal and causal way. Films, plays, comic strips, novels, newsreels, chronicles and treatises of geological history are all narratives in this widest sense. Narratives can therefore be constructed using a wide variety of semiotic media: written or spoken language, images, gestures and acting.
Since here we shall concentrate on the literary genres of the novel and the short story, we will use the word in a more restricted sense, meaning a linguistic narrative text, or the representation of a series of events by means of language. The fact that in literature the events are mostly fictional has only indirect consequences at the level of analysis we shall undertake here, and in studying the structure of a narrative text we can disregard the difference between fiction and nonfiction for the moment. We should remember, moreover, that this difference is not an absolute one. Fictional and nonfictional narrative situations can be clearly defined and distinguished in theory and for most practical purposes as different discourse activities taking place in well-defined contexts (e. g. we expect a novel to be fictional, but a news programme is supposed to give us actual facts). But in specific cases the borderline between one situation and another may be blurred, and several sets of conventions may be at work at once (e. g. in a literary biography). And beyond this communicative level in which a "fictional pact" (or other types of illocutionary pact) are established between the participants, there remains the problems of representation. On one hand, fiction is not entirely fictional in the sense that its materials are taken from reality. On the other, reality is not all that solid, since any representation involves a measure of fictinalisation. Any representation involves a point of view, a selection, a perspective on the represented object, criteria of relevance, an implicit theory of reality. Narrative structures may be at their most elaborate in artistic texts, but narrativization is one of the commonest ways of imposing an order and a perspective on experience. Even those historians or journalists which try to represent the bare facts must do so using narrative patterns. Together with other linguistic resources such as tropology, narrative acts a shuttle between formal, ideal perception and representation, and the concreteness of experience which must be given a shape. It always involves in some measure the intrusion of poetry and rhetoric upon any naive notion of purely transparent, immediate representation. In literature and other narrative arts we can study the fantasies of representation engaging the real in different ways. After a close study of narrative, we will no longer be able to speak of the real without taking into consideration at the same time the way it is narrated to us, the way we narrate it.
We have defined narrative provisionally as "a semiotic representation of a series of events linked in a temporal and causal way". There is no absolute definition of this or of any other concepts or phenomena. Definitions are a kind of translations: they allow us to grasp a phenomenon in relation to other phenomena which are already available to us and are (strategically) supposed to need no definition. Therefore, a definition of a particular object will vary according to the "language" in which we want to formulate it. A definition answers to a particular purpose --it must therefore be contextualized-- and can be more or less specific or detailed according to the contextual needs it answers. Definitions of narrative, of point of view, of plot, etc. are to be looked at in this way. It is not that we define what a plot is more accurately than Aristotle did: it is that Aristotle did not need to relate the concept of plot to so many areas of human activity (psychology, linguistics, history, sociology, politics, literature, etc.) as we do nowadays, owing to the increased specialization of discourse in our society. Definitions therefore will never be accurate in one sense, and will always be in another. The problem of their accuracy is to some extent a pseudo-problem. The real problem is their usefulness in a particular context, or rather their usefulness in helping to relate two areas of knowledge which were previously unrelated. Concepts, therefore, are tools which we use to grasp a flow of phenomena which otherwise remains unnamable according to a post-metaphysical (or Nietzschean) ontology. And narrative is precisely one of the main ways in which we organize concepts and impose order on the world. Analyzing narrative patterns is one way of getting to know the nature of that order.
One of the areas of our course into which we aim to translate the concepts of narratology will be the area of semiotics. This is not just one more discursive area. It is more like a universal coinage where the contributions of different theories can be compared to one another. From a semiotic point of view, we can study narrative in its syntactic, its semantic or its pragmatic aspect, just as we study any other linguistic phenomenon.
For our first approach it will be useful to consider a narrative as an expanded phrase, which can be analyzed at the levels of description used in linguistics and other semiotic disciplines: the levels of syntax, semantics and pragmatics.
á Narrative cannot really be defined syntactically if we take the term syntax at its face value, since a narrative structure is generally acknowledged to be a suprasentential phenomenon. Syntax as we usually understand it is a formal manifestation of semantic relationships which have become standardized or congealed at sentence level (f. i. the case of the passive voice, or of theme and rheme structures). But there are some narrative germs in syntactic structures. We can establish some syntactic patterns which look more narrative than others: a narrative typically consists of a subject and a verb --"John came." The narrative becomes more interesting if the number of actions and participants increases, if there are objects as well (direct and indirect) and if the circumstances of place, time, cause, etc. are specified: "Against his better judgment, John forced himself to accept Cartwright's suggestion; he gave him the password." The passive voice is in a certain way a narrative with a perceptible manipulation of point of view and action role: "Someone is watching Tom" / "Tom is being watched." Imperatives, on the other hand, are not narratives; they are more like drama, if anything. Impersonal constructions make boring narratives: "It is raining", so what. And attributive sentences are not narrations, but descriptions. These linguistic analogies are of course reelaborated when the sentences actually work within a text, but the relative frequency of certain types of linguistic constructions can be subjected to stylistic analysis and shown to create particular effects. Likewise, whole narrative texts might be compared to basic sentence constructions and shown to be rather "passive" than "active" or more "attributive" than "predicative." But of course from the moment we abandon the simplicity of abstract syntactic analysis these analogies become more and more metaphorical, since they are based on semantic as well as on purely syntactic considerations.
á Semantically, the definition of narrative is a definition of narrativity. Several issues can be raised in this connection to circumscribe an area of discussion:
- Does narrative involve human or at least animate agents?
- Does it necessarily involve action or process verbs?
- What about verbs of cognition or perceptual activity? Or modal verbs?
- Can there be one-phrase narratives?
Different theorists work with different assumptions as to what is or is not narrative with respect to these or other criteria. Several degrees of narrativity will be defined according to which of the questions we answer in the affirmative. The highest degree of narrativity is defined by the presence of human agents involved in deliberate and intentional action, articulated in a well-defined temporal and causal connection. The traditional patterns of myth or folk narrative (for instance, the hero's quest as described by Campbell or Propp), or the rules of classical stagecraft set down by Aristotle, Corneille or Freytag would serve equally well to illustrate this ideal of maximum narrativity. The classical narrative ideal involves not only a connected sequence of actions, but also the construction of a unifying pattern which establishes a maximum of connexion between the individual actions.
According to Genette's influential theory ("Frontieres du recit"), the borders of narrative are threefold:
-The opposition between narration and representation (here he deals with Plato's notion of mimesis as opposed to diegesis, a question we shall be concerned with next week).
-Description is another limit: while narration is concerned with time, description extends itself in space. Genette conceives of description as an indispensable element within narration itself. The relationship between narration and description is paradoxical: while description can be conceived to appear without narration, whenever it appears in narration it is an ancillary mode, subordinate to the narrative element. But such an account would give too rash a view of the function of narrative structures in literary texts. Quite often, in lyrical poetry for instance, the narrative element is insignificant, a mere prop to sustain symbolic, descriptive or meditative elements. These texts can be considered to have narrative elements, but they are not primarily narratives. This is one of the criteria traditionally used, for instance, to classify a poem as lyric or epic.
Genette argues that description is not opposed to narration in its representative technique, since it too must be submitted to the successivity of language. But the successive nature of the descriptive discourse does not blur the frontiers between narration and description only: any other structure which is linguistically conveyed will have to travel in time with the chain of speech. And what if we consider the narrative versus the descriptive elements in a picture, where there is no temporal development in the signifier? The opposition between a descriptive element, such as the colour of a figure, and a narrative element (the mythological anecdote of a Renaissance picture, for instance) is still pertinent. Therefore, the difference between narration and description is to be regarded as a difference between the represented signifieds. In the case of narration, the basic articulations of the text are events; in the case of description, the structure is a non-sequential pattern of traits. The opposition between events and traits absolute precisely because it is not an opposition between given elements; it is conceptual and therefore ideal.
The third frontier of narrative described by Genette is no longer semantic. Since it concerns the modes of enunciation, we shall deal with it in a short while under the heading of narrative as a pragmatic phenomenon.
- Pragmatically, narrative is a communicative phenomenon. The scheme of the communicative situation devised by Jakobson is well known:
Sender ......................... Message .................................. Addressee
From the 1960s on, the philosophy of language has extended its attention in the direction of pragmatics, the contextual use of language and the specific norms it creates. Language use is not a chaotic parole, as the Saussurian model would lead us to think. There are norms of discursive interaction beyond the level of the sentence. The speech act is the simple language action that is effected through the communicative use of a proposition (e. g. in order to make a statement, or give an order). The discursive act is hierarchically is a web of such conventionalized speech acts, and it involves not a single proposition, but an actual discourse event --in the case of literature, the writing or reading of a literary text.
In his study of speech acts, Austin has proposed some fundamental concepts. The first basic distinction is the opposition between locutionary, illocutionary and perlocutionary acts. A locutionary act is a speech act considered from the point of view of classical linguistics: as an abstract phrase consisting of a form and a meaning, devoid of context (or rather with an in-built conventional context) and considered in its morphological, syntactic or semantic aspects. When the locution is uttered in a particular speech situation, it becomes an illocution, a communicative act of some kind whereby a social interaction is established between the speaker and the addressee (an affirmation, an order, a promise). In order to occur, a locutionary act must be identified as such by the hearer. A perlocution or perlocutionary act is the non-conventionalized effect the illocutionary act provokes on the hearer. Recognizing that a threat is a threat is an illocutionary maneuver; the fear or laughter which may result from the threat are perlocutionary effects.
According to Austin, a classification of illocutionary speech acts could establish the following types:
(1) Verdictives (giving of a verdict by a jury, arbitrator, etc.).
(2) Exercitives (exercising powers, rights, etc. E. g, voting)
(3) Commissives (they commit the speaker; promise)
(4) Behabitives (attitudes of social interaction, e. g. apologizing)
(5) Expositives (they make clear our use of words; e. g. "I hold that . . . ").
As such, none of these categories is adequate to describe narration. Simple, true narration of real facts would seem to be a kind of expositive speech act, derived from the simple statement, affirmation or telling of a fact (cf. How to do Things with Words 162). Other kinds of narration could be derived from this basic type of narration.
Austin and Searle study speech acts in a very abstract and simplified way, as if they always consisted of one-sentence utterances occurring in conventional contexts. But in fact speech acts occur only in discursive activity. What we might call primitive speech acts are linked, combined and transformed as they are used in wider units of social interaction. These units can be called discourses, discourse acts or discursive activities. A piece of discourse can often be described as a macro-speech act, a wide-ranging speech act to which the micro-speech acts occurring at sentence level are subordintated. Micro-speech acts are therefore instrumental in the configuration of macro-speech acts or discursive acts, which in turn are determined by the structure of social relationships.
Linguistic (including literary) narrative is one variety of discursive act: more specific varieties can be established according to the needs of the classification. This approach can offer a new perspective on the literary genres and modes of discourse. For instance, we can use it to rethink the third frontier of narrative drawn by Genette.
This last distinction refers to the opposition between histoire and discours first introduced by Benveniste in his discussion of the tense system in French. In histoire, there are no signs of enunciation; in discours there are references to the speaking subject with first-person pronouns, to the time or place of enunciation by means of deictics, etc. Discours is the natural mode of speech, while histoire is defined by a series of exclusions. But it can never completely exclude the marks of enunciation. The reason is not, however, that a certain quantitative proportion of enunciation always seeps in, as Genette seems to think in this passage. Every narrative text is also and primarily an enunciation, an act of (literary) discourse. Only by means of an abstraction can we consider it as an histoire, a story.
The conceptual distinctions we shall use in our analysis (fabula, story , text) are to be understood in this same way: what we have direct access to are texts: we construct stories from them, and we construct fabulas from the stories. But let us return to our definition of narrative in order to have a closer look at these concepts.
In which sense can we analyze narrative? How can we begin? From the definition of narrative we have proposed, "the representation of a series of events", we already see that narratives are composite entities in a number of senses. In the definition we mention a series of events: therefore, the narrative can be analyzed into the events that compose it. Also, these events can be studied according to their temporal sections. In a series of events some are at the beginning, some in the middle, some at the end. So, in our first approach, a narrative consists of a number of successive parts: it has a longitudinal structure of time and actions.
A narrative is a compound in other senses, too. Let us note that in our definition a narrative is not "a series of events", but "the representation of a series of events". Here the composite nature of narrative appears not as a number of succesive parts, in length or horizontally, but, as it were, vertically, in depth: the narrative is not that which it seems to be; it is only a sign. What we get in a narrative text are not events as such, but representations of events. Here an infinite amount of complexity begins to appear. In which way are the events represented? How is the narrative similar or different from the events it represents? The following chapters will largely consist of possible answers to these questions.
We see, then, that the very definition of narrative leads us into the beginning of analysis, and in several directions at once. We shall examine different theories which analyze narratives either horizontally, or vertically, or both. As far as horizontal analysis is concerned, we have spoken so far of beginning, middle, and end. Other concepts will complicate this simple account of parts. As far as vertical analysis is concerned, we may speak of levels of analysis. Our definition distinguishes at least two levels: if narrative is a semiotic representation of a series of events, one level of analysis will examine the events represented. The structure of the representation involves study at another level. We shall find that the narratological theories often differ when it comes to defining these levels. Some theorists will distinguish two levels of analysis, others speak of three or four levels. Mieke Bal tells us that there are three basic levels of analysis of narrative: fabula, story, and text; Tomashevski only speaks of two, fabula and siuzhet. In fact, this problem appears in all areas of literary study. Theories which are presumably about "the same" often turn out to be different. Let us concentrate for one moment on this issue of identity and difference.
Identity and difference are not absolute concepts; they involve an abstraction and need a reference point. Saussure pointed this out with reference to semiotic phenomena as a whole, using the example of the 8.45 PM express train from Geneva to Paris. Is it the same train from one day to the next? It depends on how we see it. Saussure used the train as a convenient example to illustrate the structural functioning of language. Language is a system where the relationship between the signifier and the signified is arbitrary. It is not the intrinsic nature of the signifier which determines its signification, but only the place assigned to it within a system of differences. Semiotic identity, that is, is determined by the relationships created within a signifying system, not by the materiality of the signifying phenomenon. In Saussure's words, "The linguistic mechanism rests entirely upon identities and differences, the latter being merely the counterpart of the former" (151). Thus the possibility of meaning, the origin of the sign, is founded on the possibility of identity through repetition: different material phenomena will be the manifestations of only one sign, because they are interpreted according to a particular convention which disregards their differences.
Saussure's example works both ways. Trains, just as language, are from one point of view semiotic phenomena. The 8.45 PM Geneva-Paris is to some extent an abstract entity: it may be composed of completely different coaches and a different locomotive each day, but its identity remains the same for the practical purposes of the passengers.
This Saussurean example is a convenient way of coming back to the question of the levels of analysis of a text. Signs may be signs in several senses:
á They may be signs of a referent; that is, we may study their referential function, their connection to the objects or events we are referring to whenever we use a sign.
á Signs may also be signs of themselves insofar as they are signs, that is, they are signs of their meaning, their signified.
á They may also signify themselves insofar as they are signifiers, that is, they may be signs of their form.
A sign, therefore, is a complex entity which may be read at different levels, in the manifold relationships which tie it to the world. Narratives, being complex signs or structures of signs, can also be considered at several levels of abstraction. This amounts to reading the text according to different interpretive conventions. There can be many kinds of such interpretive conventions, and all need not be narratologically significant.
For instance, according to Scholastic hermeneutics, the sacred texts could be submitted to a fourfould interpretation, to yield a variety of meanings. The literal sense was opposed to the mystical sense. And there were three possible ways of reading a mystical sense in a sacred text: moral readings, allegorical readings, anagogic readings. Each kind of reading consisted in relating the text, translating it meaningfully into a particular realm of experience: moral action, history (understood as the unfolding of God's scheme of salvation), and theology. We could call these four types of reading "levels of analysis" of a text, as well --Northrop Frye makes use of such levels of analysis in his Anatomy of Criticism. All this should make us keep in mind that there is no such thing as absolute levels of analysis: that these and other narratological concepts are interpretive constructions developed for specific interpretive needs. The levels of analysis we will use here attempt to examine the specificically narrative characteristics of a text. The levels of analysis distinguished by Frye, for instance, are not specifically narratological: they can apply to any literary work, be it a narrative or not.
When we speak of two narratives dealing with the same story, we are of course using "the same" in a relative fashion. The text is "the same" with respect to our immediate purposes of analysis, just as the Geneva-Paris express is "the same" from one day to another only for certain purposes. If two narratives dealt with exactly "the same" story, we would not have two narratives, but one. While we speak of two stories having "the same fabula", we still can recognize that the stories, not to speak of the texts in which they are conveyed, are different. When we speak of "the same story" in two narratives, we are implying that they still are different texts. And we may even speak of "the same text" to refer to a work and its translation, if the language issue is not relevant for our immediate purposes. "The same text" may also be handwritten, registered on tape or printed in different types, if these differences are not relevant for our analysis.
It follows, too, that considering a text as a narrative is also the result of a methodological choice: it amounts to focusing on certain aspects of its structure and possibly disregarding others. Narratological analysis is therefore not a variety of criticism; it is a conceptual instrument used by criticism. While it enhances awareness of the textual structures, it also furthers certain directions in criticism and interpretation. But the mere analysis of a text using the concepts of narratology is not a critique of that text; there are whole areas of literary study that such an analysis completely ignores, and which have to be taken into account if an interpretation is to be balanced and well informed.
We see, then, that the instruments of narratology are not "neutral" or "aseptic," in the sense that they are the result of an interpretive choice and lead to further interpretive decisions. While the conceptual distinctions may be clear-cut, these instruments do not have absolute values in practical analysis, since each interpertive act defines the mode of their application. The same heuristic proviso applies to narratological models. Narratological models, too, are not all the same. In the long run, they all have an axe to grind, and they offer different perspectives of the structure of a narrative text. None of them is completely right, and therefore the differences between them must be considered as meaningful and meaning-producing, and not (generally speaking) as blunders or errors of the critics.
Our attention in this introductory section will be devoted above all to the narratological concepts proposed by each of the theorists, but we can test the efficiency of these or the range of their useful application in a variety of narrative texts: literary or historical texts, newspaper stories, films, advertisements, paintings, comic-books, etc. Since every narrative theory is contextual (it answers a particular aim) it is to be expected that each will be devised to deal with a specific kind of narrative objects, and will prove less useful when applied to other kinds of objects. Therefore, when studying texts on narrative theory, we should remember that it, too, is a biased conceptual construct. We should summarize it, find the most characteristic emphasis of the text, its main innovations to earlier texts, its dialogue with them--what the text includes, what it leaves out. We should observe the discursive activity it takes as its object, and the level of its theoretical formulations--which is as well a study of the audience and the historical context of the text. A good way to test the theory is to focus on some specific question of our own interest and evaluate the theory's treatment of the issue, or to compare which aspects of narrative structure are given greater prominence, and why (e.g. narrative time in Genette's model, duplicitous function in Barthes's).
It is therefore useful to see the differences and the similarities between narrative theories, and it is absolutely necessary to be able to translate one model into each other's terms as far as possible by focusing on the conceptual distinctions, and to tell them apart from those differences which are merely terminological. Differences in vocabulary when referring to the same concepts may be irritating, but they are intrinsic to the critical activity. Critics are to some extent free to coin their own terms, and many of them have done so. We may see a practical example by studying the differences between some basic analytical concepts of one of the earliest narratologists, Aristotle, and of a contemporary author, Mieke Bal.
Mieke Bal proposes an framework of three levels of "vertical" analysis of the narrative text: as text, as story, and as fabula. In the case of Robinson Crusoe, the text is the linguistic artifact that we can buy and read, written de facto by Defoe and supposedly by Robinson; the fabula, which we may also call the action, is whatever happened to Robinson in his travels and his island. The story is the precise way in which that action is conveyed, the way the fabula is arranged into a specific cognitive structure of information. In Bal's terms,
A narrative text is a text in which an agent relates a narrative. A story is a fabula that is presented in a certain manner. A fabula series of logically and chronologically related events that are caused or experienced by actors. (Narratology 5)
In this quotation, each term is defined in a neat analytical way in terms of the next, until we reach the atomic concepts of the theory, which are "to cause", "to experience", "state", "transition", "actor" etc., of which more later on. Apparently, the definition of narrative text should be "a text in which an agent relates a story," in order to preserve the neatness of the conceptual chain. In an earlier (and theoretically subtler) account of her theory, Bal had defined these concepts as follows:
1. A text is a finite and structured set of linguistic signs.
1.1. A narrative text is a text in which an agent relates a story . . . .
2. A story is the signified of a narrative text. A story signifies in its turn a fabula.
We may represent these levels of signification by means of the following diagram:
The fabula is, according to Bal, a bare scheme of the narrative happenings without taking into account any specific traits which individualize actors or actions into characters and happenings. In the description of the fabula or action we should also neglect any temporal or perspectival distortions: there are no flashbacks or variations in point of view at this level of analysis. It is obvious that Bal's conception of the fabula is actually action-scheme: it is an abstraction, not the concrete, full-blown action that we construct when reading or watching a narrative. It is confusing that this second concept of fabula, the fabula as concrete action, is used by other theorists. Of course we shall preserve both concepts, since both are analytically significant: we shall oppose the full-blown action or fabula to a more abstract and reduced fabula-scheme or action scheme. So a modified diagram would picture the vertical levels of analysis in the following way:
Of course, we could also provide a scheme of story structures; in the case of the text, the word summary would seem to be more appropriate to name the equivalent, "reduced" version. Fabula-schemes, story-schemes and summaries are used continuously as critical tools.
Let us further specify the concept of story. A story is a fabula which has been given a shape: a specific point of view and temporal scheme have been introduced, and we deal with individualized characters, no longer mere actors in an actional scheme. We could say that a story is a fabula as it is presented in a text. But the text is not the story: "story" is still an abstraction we effect on the text. A text is a piece of language, while a story is a cognitive structure of happenings. The same story can give rise to a number of texts: for instance, when Kafka wrote The Castle in the first person and then rewrote it in the third person, the story remained the same, but the text became a different one. The same story could in principle be told by means of different texts: a film, a comic book or a novel.
The story, then, can be looked on as a fabula which has undergone a further structuration. It can be defined as the result of a series of modifications to which the fabula is subjected. These modifications can be relative to time or to informational selection and distribution (mode). We shall deal with them in sections 2, 3, and 4. In sections 1 and 2 we shall study those structures which occur at fabula level.
As often happens with the basic concepts of literary criticism, the basic notion of analysis of a text through a series of levels of abstraction can be traced back to Aristotle's Poetics. Aristotle is concerned most of all with the structure, the organization of the literary text, not so much with the reactions of the audience to the text or with the creative inspiration of the author. Concentrating on tragedy as the epitome of a poetical work, Aristotle considers it as a compound formed of six elements: "Every tragedy . . . must have six parts, which parts determine its quality¾namely, plot, character, diction, thought, spectacle, song" (51, VI.7). Plot is the most common translation of mythos. A tragedy is not all plot: we might as well say that it is all spectacle or all character. Therefore, considering the plot of a tragedy is considering only a possible aspect of the tragedy. The difference between "plot" and "tragedy" is therefore a difference in level of analysis. Aristotle defines plot as follows: "plot is the imitation of the action¾for by plot I here mean the arrangement of the incidents" (51, VI.6); "the structure of the incidents" (52, VI.8). Plot is tragedy considered insofar as it is the imitation of an action, tragedy insofar as it consists in a series of incidents.
So far we have two levels of analysis: on one hand, the whole thing, the tragedy; on the other, the various aspects under which it can be examined. From a narratological point of view this amounts to a difference between the story and the text:
1st level: tragedy (narrative text)
2nd level plot (story)
But let us look closer at the Aristotelian definition of plot, "the imitation of an action." Tragedy as a whole (the narrative text) is also described by Aristotle as "the imitation of an action" (51, VI.5). Aristotle seems to mean that a tragedy is the imitation of an action insofar as it contains a plot. Be as it may, it is clear that the dynamic element at the bottom of the narrative, the source of narrative movement is action, and it is different from plot. Plot is not action, it is only the representation, the "imitation" of an action. We are here close to Bal's model of three levels of analysis:
1st level: tragedy (narrative text)
2nd level: plot (story)
3rd level: action (fabula)
Aristotle says that plot is a part of the tragedy, while he does not say that action is a part of the plot. The relationship between the action and the plot is not one of part to whole, but one of imitation (mimesis): the plot is the imitation of an action.
But this imitation is not natural in any simple way: it is a construction, something which has to be carefully crafted by the poet. Two opposite perspectives, literature-as-reality and literature-as-construction can thus be traced back to Aristotle's poetics. Some ages have stressed the first, others the second. Now we seem to be in an age in which careful construction is more appreciated than the aim to represent reality in a faithful way. Nowadays we do not believe in faithful representations of things as they are: when we watch a documentary on mountain climbing, we tend to ask "where is the camera? how did they get to film this?" Our contemporary culture, which has invented the name of semiotics, is interested in the structure of representation, and will not be content if it is given merely the represented object. It is perhaps because of this that narratology has developed more during the twentieth century than for ages before. All the critics we shall study see narrative as a rhetorical construct whose strategies can be uncovered and examined. But we should remember that these theories are themselves perspectives on reality, interpretations. And it is the fate of all interpretations to be interpreted again.
 Cf. Halliday's article on Golding's The Inheritors.
 1. Un texte est un ensemble fini et structure de signes linguistiques. 1.1. Un texte narratif est un texte dans lequel une instance raconte un recit . . . . 2. Un recit est le signifie d'un texte narratif. Un recit signifie a son tour une histoire. (Mieke Bal, Narratologie 4. Translation mine)
 A simplified version of the one in Narratologie 33.