"Construction En AbÎme" and the Principle of the "Third Text"

The signpost motif is of interest from another point of view as well. We have seen that the integration of an anomalous textual moment into its context cannot always rely on just one outside source (such as the Aeneid ). It may require two, three, or even more texts, making it a case of hyperquotation. This case is far from being rare; indeed, it may well be characteristic of intertextuality as a phenomenon. A quote becomes a hyperquote whenever one source is insufficient for its integration into the fabric of a text.

The case of hyperquotation raises several further questions. The hyperquote does not just open a text up to other texts, thereby simply broadening its horizon of meaning. It places a number of texts and significations one on top of the other. The hyperquote essentially becomes a kind of semantic funnel, drawing in all the competing meanings and texts, even if the latter contradict each other and are not readily reconciled into one unitary and dominant meaning. True, the Aeneid and the poems by Baudelaire and Cocteau allowed us to integrate the signpost in Vampyr into the remainder of the film, but only at the cost of invoking a range of diverse intertexts. In the final analysis, the integration of a quote in a film is won only at a cost: a hyperquote can bring order to a text only by erecting a new Tower of Babel, a babble of meanings.

The dilemma of the hyperquote can be resolved in one of two ways. On the one hand, intertextuality can be seen as an arbitrary accumulation of associations, quotes, and voices, in the spirit of Godard's claim that people quote "whatever they like": "In my notes, where I put down everything that I might need for my film, I might even put down a phrase from Dostoevskii if I like it. Why make a fuss? If you want to say something, there's only one solution: to say it."^[78] Godard's notes, a vast repository of everything that struck his fancy along the way, are one model of intertextuality. Roland Barthes came to similar conclusions, which he formulated in the following way: "I savor the sway of formulas, the reversal of origins, the ease which brings the anterior text out of the subsequent one."^[79] Barthes values the

very randomness of this accumulation of meanings for the way it reflects the unpredictable and mobile character of life.

Nonetheless, the majority of scholars have taken a different position. In Laurent Jenny's words, "Intertextuality does not imply a confused and mysterious accumulation of influences, but the work of transformation and assimilation of several texts performed by a centering text which retains its position of leadership in meaning. . . . I propose to use the term 'intertextuality' only when one is able to recover elements in a text that were structured prior to the text itself."^[80] Jenny is here echoed by Riffaterre, who also rejects Barthes's position: "All intertextual comparisons are imposed and controlled not by lexical coincidences but by a structural identity, the text and its intertext being variants of the same structure."^[81]

Truly, only if we can detect a structural isomorphism existing between texts or parts of texts is it possible for us to unify meaning within the centering framework of a "leading text." Essentially, what is at stake is the repetition of the same in both text and intertext, although this repetition is naturally subject to a semantic reworking.

Structural isomorphism is particularly evident in paragrams, where two messages are superimposed and thereby inserted into a single textual structure. It is no coincidence that Julia Kristeva sees paragrams as performing a "centering within the framework of a unitary meaning."^[82] Yet she herself has reduced the possibility of intertextual relations to a three-fold schema: (1) complete negation (complete inversion of meaning); (2) symmetrical negation (the logic of meaning is retained, while crucial nuances are removed); and (3) partial negation (the negation of part of a text).^[83] Clearly, all these patterns become possible only if isomorphic structures are at work, in a dynamic of reciprocal influence. How else could one conceive of "symmetrical" or "partial" negation?

The necessity of a structuring principle within intertextual relations also forces us to confront the question of the "text within the text," which the French have called la construction en abîme (literally, construction in the form of an abyss). The term has been borrowed from the culture of heraldic insignias, where it referred originally to a coat of arms that contained a copy of itself reproduced in a smaller size. It was André Gide who first made this heraldic term current as a constructive principle, whose use he illustrated in pointing to paintings that contain a mirror, the rat-catching scene in Hamlet, or the puppet-theater in Wilhelm Meister. He was referring, in the case of the plays, to the characters theatrically enacting the situation in which they actually found themselves.^[84]

Construction en abîme is marked by a heightened level of structural likeness between the framing text and the text it incorporates. Lucien Dällenbach has noted the similarity between construction en abîme and a phenomenon Claude Lévi-Strauss named the "small-scale model [le modèle reduit ]."^[85]

What Lévi-Strauss had in mind was the reproduction of an existing object on a smaller scale, as, for example, in a painting. This change in proportion has semantic consequences: it gives the copy the qualities of a handmade object and grants the maker the experience of mastering an object, compensating "for the renunciation of its sensory dimensions by allowing for the acquisition of its intelligible dimensions."^[86] At stake is the translation of an object into some other qualitative condition: I would call the "small-scale model" a case of transition from objecthood to representation.

In his analysis of that classic case of construction en abîme, the Velázquez painting Las meninas, with its complex specular construction of space, Michel Foucault came to a similar conclusion: "And representation . . . can offer itself as representation in its pure form."^[87] A text that functions on the basis of a duplication juxtaposes fragments that have been encoded in different ways but carry similar messages, thereby making their encodedness, that is, the fact of representation, all the more palpable. Iurii Lotman has defined this phenomenon as follows: "Duplication is the simplest way of making the code a part of a text's acknowledged structure."^[88]

Ultimately, construction en abîme creates little more than an illusory play of references, constantly emphasizing one and the same thing—the play of codes, the palpability of representation, structural isomorphism, and, behind all of this, the flickering gleam of shifting meanings. To be sure, no reflection is fully accurate, always involving a variation, a transformation that is stressed by the repetition itself. The concrete production of meaning, however, is often the result of a play of reflections and is quickly submerged in this play. We appear to witness the very birth of meaning, and this is an illusion that always somehow accompanies the phenomenon of "pure representation." Existing analyses of cinematic versions of construction en abîme confirm my conclusions: the work of Viacheslav Ivanov or Christian Metz on the subject is entirely devoted to the study of systems of mirroring.^[89]

A typical case of construction en abîme involving quotes (Dällenbach calls this an instance of autotextuality or self-quotation rather than intertextuality) arises when a film displays painterly or graphic forms of representation. These can be seen as a kind of "small-scale model." Sharply differentiated from the texture of the film thanks to the way they are encoded, they seem to flaunt the fact that they are representations.

Let us look at some examples of how this form of quotation generates meaning. In the episode from Godard's Breathless discussed earlier, we are actually shown several paintings: apart from the Renoir poster there are two reproductions from Picasso. One, showing an idyllic young couple, appears at the moment Patricia says, offscreen, "I would like for us to be Romeo and Juliet." The second Picasso appears a little later: it shows a youth with the mask of an old man's face in his hands. As we see it, Michel

says, again offscreen, "It's just like telling the truth when you're playing poker. People think that you're bluffing, and you win."

Both these instances involve a duplication of meaning bordering on tautology. To show a painting of a young couple while someone is speaking of Romeo and Juliet, or the youth with the mask while someone is speaking of deception, adds nothing of substance to what is being said by the actors. And yet by virtue of being juxtaposed alongside quotations from Picasso, these passing comments do seem to gain a certain visibility and weight, acquiring something of the representational character of Picasso's works. In the given instance, meaning is not so much revealed as introduced into a context marked by heightened representativity, where it becomes all the more palpable and unitary for having been repeated. The quotes here function much like a teacher's comments in red ink.

Perhaps to avoid the relative didacticism of mere duplication, the construction en abîme gravitates toward triplication, or reproduction in even greater numbers. In Dreyer's Gertrud (1964), the heroine is tormented by a persistent dream. She dreams that she is naked and being pursued by a pack of dogs. In the same film we see a large Gobelin tapestry showing the naked Diana being torn apart by Acteon's dogs. The dream is thus duplicated by the tapestry. The same motif is also clearly projected onto Gertrude's life. The dream's meaning, as Pascal Bonitzer puts it, is "as transparent as it is lapidary: the dogs are the men to whose desires, as crude simulacra of the love she seeks, she abandons herself and by whom she is torn apart."^[90] The very obviousness of the situation hardly seems to call for duplication, or what Dällenbach calls autotextuality. In fact we are dealing not with a doubling but with a tripling of the same collision in different signifying systems: the words of the heroine and the tapestry. The multiplication of representation becomes analogous to the nagging persistence of dreams. Repetition here imitates the repetition compulsion generated by a trauma that has been repressed into the unconscious. These repetitions do not just reproduce a certain meaning: they point to the presence of a certain repressed semantic kernel whose significance is consistently expressed as other, different, mysterious. Repetition thus simultaneously asserts and denies that meaning is univocal.

Variations in repetition can thus lead simultaneously to a centering and a decentering of meaning. This can happen in more than one way. Let us look at one more example, from The Most Dangerous Game (1932), by Ernest B. Shoedsack and Irving Pichel. This somber Gothic film relates the adventures of two young people, Rainsford and Eva, who are shipwrecked off the shore of an island belonging to the sinister maniac Count Zaroff. The count agrees to release his two unwilling guests if they survive a hunt against them that he himself will lead. Working for the count is a terrifying bearded servant named Ivan. The film begins with a depiction of the door

to Zaroff's castle. A close-up shot shows us the strange knocker to the castle door in the shape of a centaur pierced by an arrow with a young woman, no longer able to resist, in his hands. Then from off-camera we see Rainsford's arm appear, lift the knocker, and knock thrice on the door. The centaur holding the woman in his arms appears again on the Gobelin tapestry that adorns the castle's staircase. This time we clearly notice the similarity between the centaur's face and the servant Ivan's sinister visage. Belaboring the point further, the directors at one point deliberately place Ivan against the tapestry, making him stand immobile until the similarity becomes unavoidable. Not only does the representation of the centaur holding the woman appear twice in the film (on the knocker and on the tapestry); it also establishes a link with one of the film's characters.

The construction en abîme once again becomes triple in character, although it is the tapestry that develops the plot in greatest detail: in the overgrowth we also see the young man who shoots the centaur with an arrow—Rainsford, of course, who will soon emerge victorious in the struggle to save Eva and himself. The triple character of these autotextual relations creates a more complex space of references and reciprocal mirroring, and with it the illusion of semantic depth, here particularly effective thanks to the specific variations in the distribution of motifs.

In a powerful psychoanalytic reading of the film, T. Kuntzel writes: "What is articulated in the very body of the centaur are man and beast, and this mixture, in all its varied forms (savage/civilized, nature/culture, game/hunter), constitutes the film's problematic, the field of its operations."^[91] In this context an autotextual repetition acts to center the film, emphasizing its principal theme without adding anything new. Yet the very image of the centaur, and the mythic layer related to it, allows us to widen the range of the film's references considerably. The figure of the predatory centaur belongs to classical Greek mythology. Centaurs abducted the wives of the Lapithae, the centaur Eurytion tried to abduct the bride of Pirithous, and Nestor attacked Hercules' wife Deianira. Dante placed the centaurs Nessus, Chiron, and Pholus in hell, where they were compelled to pursue sinners in an endless hunt:

Dintorno al fosso vanno a mille a mille
saettando qual anima si svelle
del sangue piú che sua colpa sortille.

(And many thousands wheel around the moat,
their arrows aimed at any soul that thrusts
above the blood more than its guilt allots.)^[92]

The Dante quote alone is enough to give the film an added metaphoric meaning that allows us to reconsider the shipwreck, the menacing swamps on the island, and the very meaning of the hunt.

Yet this is not all. More significant is the fact that this multiplication of representations strips the characters of reality, dissolving them into reflections and symbols. Marc Vernet, in his analysis of the motif of the mystery portrait in film, has concluded that the pictorial duplication of the hero "derives its substance from the real," or, as Lévi-Strauss might have put it, gives it an "intelligible" form. Reality becomes the bearer of a meaning that cannot always itself be formulated. Pascal Bonitzer has rightly observed: "The shot of a painting always provokes a doubling of vision and gives a sense of mystery to the image, a mystery that can be understood in religious terms, or like a detective mystery that has to be solved."^[93] Indeed, what is at stake here is the creation of mystery as such, the affirmation that meaning, as it emerges, is an enigma. Hence the necessity to create doubles and accumulate more and more references. Nor is it a coincidence that the film The Most Dangerous Game effectively combines a criminal case with some kind of religious mystery (if we take into account the reference to Dante).

We discover another kind of construction en abîme in D. W. Griffith's A Drunkard's Reformation (1909), a film we shall have occasion to return to in chapter 2. The film was conceived against the backdrop of a campaign launched toward the end of 1908 by New York's Mayor George McClellan for the moral regeneration of the cinema industry.^[94] It was a true didactic film, intended to show the horrors of alcoholism. In the first scene, we see the head of the family returning home drunk and terrorizing his wife and eight-year-old daughter. He is invited by his daughter to the theater, where he effectively sees the first scene in the film reenacted onstage: a degenerate alcoholic organizes a drunken orgy and then terrorizes his wife and daughter. What he has seen onstage awakens the father's conscience: he goes back home, gives up his drinking habit, and is seen in the film's final episode in an idyllic family reunion by the hearth.

The construction en abîme deployed in the film—the duplication of the plot onstage—was noted at the time of the film's making in a publicity statement released by Biograph studios: "The whole construction of the picture is most novel, showing, as it does, a play within a play."^[95] The specular nature of the "text-within-text" construction is particularly emphasized by the alternating character of the montage, which shifts from the stage to the audience, where the father, in his daughter's embrace, reacts vigorously to the play, pointing his finger at the stage and then at himself, as if to underline the fact that the play and his life are one and the same. This duplication of the plot in a theatrical performance is a classical trope; the repetition gives the film a didactic flavor, whereby the plot is viewed in the mirror of theatrical representation.

The publicity release put out by Biograph contains one highly significant detail. The play-within-the-play seen in the film is said to be an adap-

tation of Zola's L'Assommoir. Of course the general theme of alcoholism is common to the play and Zola's novel. Yet it is hard to believe that the film's viewers would have necessarily identified L'Assommoir as the text behind the play. The film does not acknowledge the novel in its credits, and the plot of the play has little in common with the novel, which has less to do with the decline of the male drunkard (in Zola's novel he is called Coupeau) than with the fall of a woman, Gervaise, herself a heavy drinker as well as a streetwalker who cheats on Coupeau with another man. There is none of this in A Drunkard's Reformation, and the analogue to Gervaise in the play, like the heroine of the film itself, is a pure, innocent, and all-suffering woman fallen victim to a husband in decline.

We might even dismiss the reference to Zola in Biograph's publicity as a simple error were it not for the fact that the whole structure of A Drunkard's Reformation reproduces an entirely analogous construction en abîme in another Zola novel, La Curée. In this novel there is a description of a performance of Phèdre, attended by the story's heroine, Renée, and Maxime, her stepson by her husband's prior marriage, for whom she is consumed by an incestuous passion. The relationship between Renée and Maxime is mirrored by those enacted onstage between Phaedra and Hyppolitus. Moreover this performance, like Griffith's play-within-the-play, imposes an absolute identification between its own protagonists and the heroes of the novel who constitute its audience, leaving Renée deeply shaken. As Théramène is about to begin his monologue onstage, Renée hysterically equates Racine's play with her own life's drama:

The monologue continued interminably. She was in the greenhouse, under the ardent foliage, and she dreamed that her husband was walking in and catching her in the arms of her son. She suffered horribly, losing consciousness, when the last death rattle of Phaedra, repentant and dying in the throes of poison, made her open her eyes once more. The curtain was falling. Would she have the strength to take poison one day? How petty and shameful her drama was, when compared to the ancient epic!^[96]

Griffith's film, we recall, reproduces not only the general structure of the construction en abîme but also the specific nature of the protagonists' reaction to what they see. In La Curée this device is used twice, although with variations. The novel also describes the performance of a series of tableaux vivants based on the amorous exploits of Narcissus and the nymph Echo. This time, however, the actors are Renée and Maxime themselves, who effectively enact their own relationship before a public. The system of mirrors is here reversed. For Zola it was essential that the theatrical performance depict a situation of falsehood, since it was his conviction that the theater, at least in its classical form, was a place of falsehood, in opposition to the naturalist novel: "Here one always has to lie," he observed.^[97] Yet

theatrical representation is not only the repository of falsehood in Zola's novel: it can also expose falsehood and hence assert the truth. A difference in the way the text is coded is thus projected onto the novel's own drama of deception. As will become clearer from my argument in chapter 2, the director Griffith also came to view the theater as a place of falsehood, deceit, and vice. These views, however, would become explicit somewhat later: at this point they appear in a muted way, through the intertextual links we have reconstructed between A Drunkard's Reformation and Zola's La Curée.

With the preceding example, we have seen how the construction en abîme functions. On the one hand, it duplicates the text and thereby emphasizes its message as unambiguously didactic. Meaning is unified and thereby "narrowed." Yet the introduction of a third text (in this case the Zola reference and the parallel construction en abîme in La Curée, which may have given Griffith the idea of inserting a "text-within-a-text" in his own film) acts to undermine the one-dimensionality of the text's didactic content. Meaning is opened up, made more enigmatic, ambiguous, even to the point of being reversed. There is, at the same time, a dissolution of unitary meaning in a form that acquires increasing complexity.

The third text, by layering form on form, helps to overcome the univocal nature of content. For this reason, the third text works in a manner that is diametrically opposed to the second text. Such threefold (and even more complex) constructions en abîme thus simultaneously widen and narrow the range of meaning, creating different potential perspectives in reading. The accumulation of textual doubles impedes the freezing of meaning in specific representational systems, such as theater and painting, even as these very systems are absorbed into the fabric of the film.

The theory of the sign elaborated by Charles Sanders Peirce sheds some light on the work of the "third text." Peirce, we know, asserted that a third element, the interpretant, needs to be added to the traditional dichotomy of sign and object. The third element, for Peirce, was essential to the emergence of meaning as such. This element inserts a difference into the tautology, a difference that allows for the creation of meaning: "Three things east, west and up are required to define the difference between right and left."^[98] In Peirce, the interpretant is a secondary sign created by the representamen in the human mind. It is precisely this secondary sign, when it is joined to the relation between the representamen and the object, that allows for the manipulation of signs, and hence meaning, via the arrangement of signs into chains. Peirce referred the interpretant to the sphere of "pure rhetoric," whose task was to "ascertain the laws by which in every scientific intelligence one sign gives birth to another, and especially one thought brings forth another."^[99]

Paul de Man, departing from Peirce's triadic model, has pointed out that rhetoric considers the "third element" responsible for the creation of

figural meaning, which, when superimposed onto the literal meaning, enters into complex relations with it. This creates a situation in which "it is impossible to decide by grammatical or other linguistic devices which of the two meanings (that can be entirely compatible) prevails. Rhetoric radically suspends logic and opens up vertiginous possibilities of referential aberration."^[100]

This triadic principle has been applied to the problem of intertextuality by Michael Riffaterre, who has suggested that the third text be given the Peircean name of interpretant (in the following quote, T = text, T' = intertext, and I = interpretant): "Intertextuality does not function, and consequently a text is not a text, unless the reading passes from T to T' through I, and the interpretation of the text in the light of the intertext is a function of the interpretant."^[101]

The interpretant is important to us for several reasons. First, it allows us to overcome the notion of the link between text and intertext as the link between the source and its text-successor; the related notions of "borrowing" and "influence" can also be abandoned. Second, it allows us to account far more convincingly for the work that goes into the creation of meaning, with all its displacements and transformations. Third, the interpretant, as the third text in an intertextual triangle, is responsible for the appearance of what Mikhail Bakhtin called "semantic hybrids," which practically all artistic works are to some extent.^[102] Finally, the interpretant allows us to understand how parody works, since the third text usually turns out to be the "distorting mirror" that parodically critiques the structure of the intertext within a text. The interpretant is responsible for the parodic status of a text to the extent that I and T(pr) are unable to affect each other without conflict or contradiction. The residual traces of these contradictions in a text can be seen as parody.

To conclude this chapter, I should like to look at one more film that shows how structural isomorphism between intertextually connected texts can create meaning as an enigma or mystery, and how the interpretant is involved in the creation of this enigma. I am thinking of Orson Welles's Citizen Kane (1941). Let us recall the basic plot: the film begins with the hero, Charles Foster Kane, dying in his castle, Xanadu. His last word before dying is "Rosebud." As Kane dies, a glass ball containing a miniaturized winter landscape drops free of his hand and breaks. The newspaper and newsreel magnate Rawlston orders the journalist Thompson to investigate the meaning of Kane's last word, which he believes to be the key to the dead man's life and personality. The film is a kind of mosaic—and puzzle—made up of the visualized stories about Kane that Thompson hears during the course of his investigation but that finally do not help him get any closer to the secret of Rosebud. The viewer, however, is given the answer. As the film ends, the workers who are burning the accumulated

rubbish from Xanadu throw the sled Kane had as a child into the incinerator. A picture of a rosebud, with a caption provided, adorns the sled.

The film repeatedly insists on the link between the word Rosebud and the glass ball, a children's toy that has apparently hypnotized Kane. The ball is a kind of mirror that duplicates the world and draws the viewer into a construction en abîme. The ball-as-mirror refers us to a visionary space, like the crystal balls used in spiritist seánces or the Salvator Mundi genre of painting, in which a glass ball is meant to provide a vision of the world.^[103] The film itself, however, does make one more explicit reference, to a literary text linked to the motif of vision. Kane's Xanadu naturally recalls Coleridge's poetic fragment "Kubla Khan: Or, a Vision in a Dream," whose first verses are quoted in the film. Kane's very name seems to recall "Khan," and the film is filled with motifs taken from the Coleridge poem. The River Alph (cf. alphabet) becomes the stream of newspapers owned by the press magnate Kane; the singing Abyssinian maid is parodied in the figure of Kane's second wife, a singer by the name of Susan Alexander who performs an "oriental opera" in the film; the construction of the palace out of song and music is reflected in the building of the opera house for Susan; and the paradisial garden surrounding Xanadu is echoed by Kane's gigantic zoo, "the largest private zoo since Noah."

Yet these individual motifs are overshadowed by a more important fact. Coleridge had maintained that the vision of Kubla Khan had appeared to him in a dream induced by opium. Waking, the author had begun to write down his vision, only to be interrupted by a visitor: "And on his return to his room, [he] found, to his no small surprise and mortification, that though he still retained some vague and dim recollection of the general purport of the vision, yet, with the exception of some eight or ten scattered lines and images, all the rest had passed away like the images on the surface of a stream into which a stone has been cast, but, alas! without the after restoration of the latter!"^[104] "Kubla Khan" thus appears as an enigma, a mysterious visionary fragment with a conclusion that has been lost forever.

In this sense it is typologically similar in form to Citizen Kane, which is also constructed around an enigmatic vision. In fact Citizen Kane points to yet another work of literature, in a way that is evident if less persistent. Several times during the course of the film, the hypothesis is advanced that Rosebud refers to a woman's name. Rosebud is in fact the name of the heroine of the Dickens novel The Mystery of Edwin Drood. Since the significance of the name remains a mystery to the very end of the film, the viewer is compelled to seek some basis for including Dickens among the various subtexts of Citizen Kane.

The Dickens novel bears a significant resemblance to both "Kubla Khan" and Citizen Kane. The Mystery of Edwin Drood, like Coleridge's poem, was never completed by its author, and the keys to its investigative mystery

have been lost. The novel also betrays several immediate parallels to the film. Like Citizen Kane, the novel involves an attempt to decipher the hero's last words: there is a scene in which futile attempts are made to unravel the words uttered by an opium smoker during his trance. These opium-induced hallucinations of the hero, Jasper, like Coleridge's visions in his poem, take up a considerable portion of the novel. The novel begins with Jasper taking up the hermeneutic challenge of deciphering an "opium-text": "When any distinct word has been flung into the air, it has had no sense or sequence. Wherefore 'unintelligible!' is again the comment of the watcher."^[105] The enigma of the novel as a whole is mirrored in this passage.

Significantly, the story of the "Abyssinian maid," Susan, is also echoed by Dickens's novel. Kane makes a dogged effort to make a singer out of her. One of these lessons ends with Susan getting hysterical and refusing to sing. Dickens also has Jasper mercilessly hounding Rosebud to sing: "As Jasper watched the pretty lips, and over and again hinted the one note, as though it were a low whisper from himself, the voice became less steady, until all at once a singer broke into a burst of tears, and shrieked out, with her hands over her eyes: 'I can't bear this! I am frightened! Take me away!'"^[106] Clearly, this scene closely parallels the analogous episode in the film.

Interestingly, the entire subplot involving the talentless Susan's compulsory singing lessons also has another literary source—George du Maurier's novel Trilby, in which the hypnotist Svengali lulls the tone-deaf Trilby into a trance and induces her to sing wondrously. One of the novel's crucial scenes involves the wounded Svengali in a theater box, inducing Trilby to sing under hypnosis. When Svengali suddenly loses consciousness, Trilby comes out of her trance and is unable to continue singing. What follows is a complete disaster: "Indeed she had tried to sing 'Ben Bolt,' but had sung it in her old way—as she used to sing it in the quartier latin—the most lamentably grotesque performance ever heard out of a human throat!"^[107]

Kane is also a kind of parodic Svengali—he is convinced that if he succeeds in hypnotizing Susan, then he could also hypnotize others. There is a moment in the film when Kane is in the opera box, casting demonic glances at the unhappy Susan, who emits some grotesque notes: this scene alone allows us to view Trilby as another subtext to Citizen Kane, all the more so since the du Maurier novel also evolves as an enigma that is solved only after Svengali's death. Trilby, then, readily serves as an interpretant, transposing the serious Welles/Coleridge parallel onto a parodic register. Interestingly, Trilby had already been used by Joyce in Ulysses for a similar parodic reversal: in chapter 15, Bloom appears as a fake hypnotist—"Bloom (In Svengali's fur overcoat, with folded arms and Napoleon's forelock frowns in ventriloquial exorcism with piercing eagle glance towards the door").^[108]

In fact any of the texts mentioned here can function as an interpretant, since all of them serve to distort the relations between Citizen Kane and the texts that form its intertextual environment. This distortion is complemented on each occasion by the marked introduction of some element of mystery, a missing key or fragment. Each intertext consistently reproduces the basic enigma of the film as a whole. N. Carroll has observed that "throughout the film, the heterogeneity of Kane's collections continues as an objective correlative ofKane's personhood."^[109] Kane does collect everything, from the sleds he had as a child to Greek and Egyptian statuary. The sheer motleyness of his collection reflects the irreducibility of his personality to any unifying center. As Borges once said, Kane is a "chaos of appearances," a "labyrinth with no center" and hence offering no way out.^[110] Intertexts also enter the composition of Citizen Kane as things from Kane's private collection. Without providing any real resolutions, they intensify the sense of mystery and deepen the film's meaning: this is the real significance of their distorting function as interpretants.

It is also important to note that the labyrinthine nature of intertextuality imbues the structure of the film itself with a paradoxical quality. Kane, that is, a protagonist who exists within the film's diegesis, is responsible for the creation of his own enigma. Yet this enigma quickly leads us to a range of texts that are to varying degrees formally identical to the film itself (and its discourse). All these texts appear to "describe" the fragmentary and unfinished nature of Citizen Kane itself. In other words, Kane as a protagonist of the film is aware not only of his own biography but also of the form that his own biography will take posthumously. Kane knows something of Welles's own discourse. It is for this reason that a character in the film poses a quandary that cannot be resolved from within his own life story. This quandary must be solved by someone who is aware of the posthumous treatment of his biography, someone like the viewer of the film rather than the investigative reporter Thompson. At the end of the film, Thompson nevertheless has the perspicacity to observe that it would not be possible to unlock the "meaning" of Kane just by deciphering the word Rosebud. This statement is immediately followed by a shot of the sled, which solves the puzzle for the viewer.

This solution, revealed only to the viewer, is doubtless partly fictitious. This becomes all the more evident when we consider the tautological, metadescriptive nature of the choice of all three intertexts: all three are "quoted" because they also involve puzzles. The real status of Kane cannot be established if only for the reason that Kane functions here as a character and as an author who knows something about the formal properties of the text that describes him. The discourse about him becomes his own discourse.

Kane's mystery, then, has no solution. The enigmatic literary subtexts serve another purpose. They destroy the clarity of the narrative mode and create a structure that allows for a slippage from the diegetic level (the level of narrative) to the discursive level (the level of the formal organization of the story). (Curiously, Borges was to deploy the poem "Kubla Khan" in an analogous way in his essay "Coleridge's Dream.")^[111] The link between these two levels of the text is established by the literary subtext, or rather by the puzzle involved in intertextual relations that is first posed by the author using Kane's persona and then confirmed by the hermeneutical contest between the viewer and the film's characters (Thompson, Rawlston). In this struggle for meaning, it would appear as if the viewer wins: after all, it is the viewer who sees the sled in the final scene. Nonetheless, the real victory belongs to the characters in the film, since they are the ones who know that there is in fact no real solution.

We have seen that the structure of intertextual relations is largely responsible for the labyrinthine movement of meanings in Citizen Kane, meanings that are never distilled into the text's one final meaning. As these meanings sabotage each other, a bottomless semantic funnel is created, which is typical of all constructions en abîme. The final meaning is displaced in the process of searching for it. The intertext constitutes meaning as the work involved in seeking it. The very dynamic of slippage from the diegetic to the discursive level, as seen in Citizen Kane, plays an important role in this work. When David Gray in Vampyr hears the cry of a baby that is not a part of the sound track, he is in fact hearing a cry that resounds not in Vampyr but out of Virgil's epic. While still situated within his own story, Gray—like Kane—unexpectedly penetrates the intertext in which the film that tells his story has been inserted. In this sense Gray hears (and knows) something that only the director Dreyer could have had access to. The character takes the place of the author.

This shift cannot be simply understood or assimilated: it necessarily creates a puzzle that resists solution. Intertextuality, then, while resolving certain contradictions within the text, at the same time creates others that are in fact irresolvable. The intertext functions as the resolution of some (resolvable) contradictions even as it creates others that are irresolvable. Understanding is thus accompanied by the appearance of a mystery. From this perspective meaning itself can be seen as an act of understanding that is shrouded in mystery.