From: Cervantes: Bulletin of the Cervantes Society of America 8.2 (1988): 141-58.
Copyright © 1988, The Cervantes Society of America

ARELY IN Western literature has discourse within discourse —and about discourse— resulted in such complexity as in Cervantes' double novela, El casamiento engañoso and El coloquio de los perros.² Speaking voices imitate other voices which imitate yet others, invisible hands transcribe or compose diaphanous layers of words, and at every level receptive minds question and reshape these words. While all literary discourse may be considered citation, as Graciela Reyes has maintained (9, 14, 34), in a very literal sense citation generates the

     * This electronic version of the article does not include diacritics over or under letters in Arabic words and names. The original version of this article shows the diacritics using small graphics files.
     ¹ Julio Cortázar, “Diario para un cuento,” Deshoras (Madrid: Alfaguara, 1982), 158.
     ² All references to these novelas are from volume 2 of Harry Sieber's edition of Cervantes' Novelas ejemplares (Madrid: Cátedra, 1981).

double novela in nearly all its various processes of speech and writing; outside of citation, only the barest narrative frame remains, and even this could be considered a sort of anonymous citation on the part of the writer. The Coloquio, in fact, is in its entirety a citation of the Casamiento. This essay focuses on different aspects of citation in the Casamiento / Coloquio —citation being understood as the explicit or implicit attribution of words to someone regardless of whether those words are actually repeated, altered somehow or invented. My aim is to sketch out the workings of citation theoretically in the Casamiento / Coloquio. Like Don Quixote, the Casamiento / Coloquio is one of those exceptional texts that provide extreme and sometimes bizarre examples of widespread novelistic practices, and thus invite theoretical inquiry on their own grounds.
     I am especially interested in what happens to language which, when quoted, becomes something quite other than what it was. This certainly happens when it passes from one sort of discourse to another, from one character type to another, one worldview or set of values to another, one ontological status to another, one set of circumstances to another, and so forth. In such cases words come to take on a markedly different significance and orientation every time they are quoted. When this occurs repeatedly, the ever-expanding transmission itself elaborates on its own true or fictive history of intermediate transmissive events and their various contexts; cited language thus becomes densely populated with recognizable faces, as it were. I shall characterize citation as a complex process composed of the simultaneous enactment and displaced reenactments of many communicative events. The very fact that such citation is differential, dialogic, “otherizing,” implies an active relationship between citing and cited discourses. In the Casamiento / Coloquio, discursive interaction brings about a refraction of value and viewpoint and a consequent loss of discursive authority on the part of most participants.
     In order to bring novelistic citation into relief, I shall contrast it with multiple citation in a radically different body of texts, the Islamic hadith, or “traditions.”³ Because the hadith, associated with the sacred, were popularly transmitted, it became increasingly important in the early years of Islam to protect them from alteration and forgery. Means were devised to isolate cited discourse from both discursive and

     ³ To avoid confusion, I shall use the singular form hadith (and not the plural ahadith for plural as well as singular meaning.

circumstantial context and thereby —at best— preserve its integral authority over centuries. In many ways the hadith are poles apart from Cervantes' novela with respect to citation.
     Nowhere in the Casamiento / Coloquio do strata of citation run deeper than when la Camacha, a sorceress, utters a divination concerning the fate of two puppies to which her fellow sorceress la Montiela has given birth:

La Montiela transcribes and memorizes these lines and somehow passes them on to a third witch, la Cañizares,⁵ who in turn cites them in a much longer discourse spoken to the dog Berganza, whom she identifies as one of her friend's litter. Much later Berganza quotes Cañizares' monologue at great length as he narrates his life story to his canine companion Cipión, who repeats the divination and subjects it to a skeptical critique. A delirious ensign named Campuzano, for his part, claims to have overheard the dogs' dialogue and to have transcribed it faithfully while recovering from syphilis at the Hospital de la Resurrección. His friend Peralta, after hearing Campuzano's own account of events leading to the illness (the subject of the

     ⁴ I have taken the translation from Forcione (44):

Forcione discusses the literary sources of this divination (44-48).
     There are two minor discrepancies between Cipión's version (346) and Berganza's (338). These could well be Cervantes' inadvertent errors: he is not, after all, renowned for accurate citation. Or they could be attributed to numerous others, ranging from the dogs to Cervantes' typesetters.
     5 There is an ambiguity here concerning whether la Cañizares is present at the deathbed scene and hears the divination directly, or whether she receives it from la Montiela. Because of the text's insistence on la Camacha's speaking to la Montiela (e.g., la Camacha “llamó a tu madre y le dijo . . . ;” “Esto dijo la Camacha a tu madre . . . ,” 338-39), I favor the latter. If this is the case, a further ambiguity concerns how la Cañizares comes to know the prophecy, whether by reading or listening (“Tomólo tu madre por escrito y de memoria, y yo lo fijé en la mía . . .”).

Casamiento) reads the manuscript of the Coloquio as fiction. All of this is conveyed by a nameless narrator, and the narration as a whole is of course Cervantes' written text. The poetic lines are thus produced and reproduced in many discursive events occurring in different times and places and on different ontological planes. Yet they only appear twice in the text we read, once within Berganza's life story told to Cipión and the other time as Cipión's citation of them back to Berganza in the same dialogue.
     Significantly, maximum depth of citation corresponds exactly to the crucial importance of the prophecy within the Coloquio. Berganza anticipates the prophecy several times as a possible key to the mystery of the dogs' origin and sudden powers of speech. Cipión singles it out for exegesis and rejection. La Cañizares (like Berganza) centers her discourse on it. La Montiela dies of grief upon hearing it along with la Camacha's confession of having performed the malicious sorcery. And la Camacha utters the divination on her own deathbed.⁶ Vital interests and discursive interests coincide to produce multiple citation.
     The following discursive levels are active precisely when the divination appears for the second time in the novela:

     ⁶ Ruth E1 Saffar analyzes both the anticipation and the centrality of the witch episode in her study Cervantes: El casamiento engañoso and El coloquio de los perros, 59-70. See also sections devoted to these novelas in her Novel to Romance, especially 72-81.

     One should bear in mind, firstly, that this artificial scheme, whose “levels” are really communicative events and their reenactment, corresponds to only one moment in the text. (I use the term “level” merely for lack of an appropriate word; there are no “levels” as such in discourse.) Just before and after Cipión quotes the divination (III), levels IV through VII disappear entirely. When Berganza utters the prophecy (IV), in contrast, all levels except number III of the above scheme are simultaneously active; just before and after his citation of the prophecy, all but levels III, VI and VII are active. Given such mutability from one moment to the next, it would be pointless to try to establish a “discursive structure” for the double novela as a whole without taking temporality into account.⁷ Outside the text the number of levels is likewise unstable, because if I quote or misquote la Camacha's divination and someone else reads or listens to this, or if we discover that the divination is itself a quotation of some extratextual source —la Camacha, after all, was tried and sentenced by the Inquisition for sorcery before Cervantes made her (or rather her legend formed by decades of citation) into a novelistic character— more levels can be added indefinitely.
     Secondly, I have arbitrarily chosen to subdivide levels I and II, since both involve author / narrator and reader / narratee (and thus also invention / transmission) distinctions that are clearly of a different order from the relationship between any consecutive “levels” as such. In levels VII through IIb —to follow the order of events— one character says or writes something that is heard or overheard or read by another character, who in turn cites part or all of this some time later in his or her own discourse, and so on; the result is a chain of verbal transmission from one character to another. The knowledge of an “omniscient” narrator unites levels I and II, embracing the entire narration and all levels of citation.
     Thirdly, level III is clearly anomalous because what Cipión says would normally be located at the same level as what Berganza says, both dogs being engaged in a dialogue that is overheard by Campuzano, but here he quotes his companion, and thus a level momentarily interposes itself between Berganza's speech and

     ⁷ José Maria Pozuelo attempts to produce such a scheme for the novela. The result is applicable neither to the entire novela, since narrative strata continually change, nor to any particular moment in the text, since both Campuzano's autobiographical tale and his Colloquy are made to occupy different levels of the hierarchy at the same time, whereas in fact they should be somehow parallel and mutually exclusive in terms of discursive time. Pozuelo also seems to be unaware of several levels of enunciation / reception.
     John Barth (54) much more successfully plots out complex citation in other works by taking temporality into account.

Campuzano's transcription of it. Whereas swallowing another's discourse more or less whole is the norm in levels VII through IIb, Cipión at level III does not include Berganza's life story in his discourse but rather interrupts that story to quote and comment upon a fragment of it.
     In terms of the reconstructed order of discursive events rather than the novela's actual unfolding, an “ascending” chain of transmission is thus established with no fewer than six links (VII to IIb before level IIa reveals the entire discursive chain “below” it to be as counterfeit as the fake gold chain Campuzano used to deceive other characters in the Casamiento engañoso (290-91). What such discourse loses in authenticity it more than makes up for by the artistry that has gone into its “forgery.” Such, at least, is the opinion of Peralta, who reads as fictive a manuscript presented to him as factually true, praising its “invención” (359). Moreover, Campuzano's discursive counterfeiting itself turns out to be counterfeit within Cervantes' fiction. The prophecy's transmission moves forwards in time through disparate communicative events, while those events are doubly invented, projected backwards into a putative past that is made for them.
     The cited prophecy undoubtedly figures among the most resonant moments in all of Western literature. John Barth, in his study “Tales within Tales within Tales,” finds no Western text with more than five discursive levels, that is, levels of citing and cited discourses active at any one moment in the text. Cervantes, with his multi-discursive finale to the Novelas ejemplares, could well be inviting his readers to take part in a profound literary game, one that reveals some of the secrets of fiction-making.
     The study of verbal transmission / invention has led me rather far afield to a cursory comparative survey of identical issues in very different texts, the Islamic hadith, which reported the sayings and actions of the Prophet Muhammad in chains of communicative events extending across the first few centuries of Islam. Much can in fact be learned from the rigorous discipline that began to emerge in the 2nd century A. H. (8th century A. D.) out of the need to distinguish true or reliable hadith from false or “weak” ones among the many thousands extant at the time. Such distinctions were essential because the reported sayings of the Prophet had, for the mainstream Sunni Muslims, become the most important source of ethical guidance apart from the Qur'an itself.
     The hadith consisted of two parts: the isnad, or chain of

transmission (literally “support,” “prop”), and the matn (“body” or “text”). In the words of one scholar, the hadith would typically adopt the following formula: “It was related to me by A, on the authority of B, on the authority of C, on the authority of D, from E (here a companion of Muhammad) that the Prophet said: ‘. . .’ —and the matn would follow (Cragg 537). The isnad could of course vary greatly in the number of transmitters. Those cited included women and men, people of all social strata, the learned and ignorant, the faithful and heretical. It should also be noted that both writing and speech were cited as authority: manuscripts could be cited orally and speech transcribed (Siddiqi, 43-44, 100, 160). The matn would most often come in the form of an injunction, proverb, aphorism, brief dialogue or anecdote whose sense might apply to a range of new contexts, as may be appreciated in Ibn Hazm's Collar de la paloma (174-76).
     Zealous concern for the accuracy and authenticity of transmission is more than understandable given the religious and cultural importance of the hadith. Interestingly, the practice of attaching an isnad to received discourse extended to other genres of literature and science in Arabic, including instances of fictive narrative in which authors would record all the extraneous facts of a story's transmission —who told it to whom, and where, and so on— before actually telling the story itself (Siddiqi 141-42). Trivial as some of these instances may be, widespread use of the isnad attests to an extraordinary preoccupation with traditio and to an unusual wariness with regard to transmitted discourse.
     So important was the specificity of transmission that when two traditions were textually identical but diverged in isnad, they were regarded as completely different hadïth. (Siddiqi 164). Moreover, because criticism of the matn was a delicate matter, nearly the full force of hadith criticism fell on the isnad: it was a matter of examining the chain of transmission. The most meticulous chronologies and tens of thousands of biographies were established to determine not only the whereabouts and contacts of transmitters at different times but more importantly to judge their character, intentions, memory, social standing and affiliations in terms of reliability. Techniques of interpreting and comparing hadith became extremely refined. The strictest criteria were thus set up to discern whether any given hadith was authentic or forged, “healthy” or “infirm,” etc. Ironically, criteria for authenticity became so acute that any new hadith conforming to them might for that very reason be highly suspect; one may surmise that after a while the science of hadith had to take this paradox into

account and establish new criteria to detect forgeries that satisfied its old criteria. Faulty memory, “story-telling” (qisas), piety at the expense of truth, heresy, and sectarianism on the part of the transmitters of hadith were considered principal causes undermining the validity of traditions (Siddiqi 52-59, 127-29). Hence a sophisticated discipline emerged whose object of inquiry, regardless of its specific scope and practical purposes, was nothing other than verbal transmission and invention.
     Under ideal circumstances, the chain of corroboration would presumably preserve the integrity of the matn across vast expanses of time: nothing would be changed in the matn, no voices would be added to it, no accents, no traces, no circumstances of telling, except inevitably those of the most immediate source. A strong link in such a chain would serve as a partial guarantee of the transmitted words: the stronger the link the less it would interfere with the message itself. The importance of who speaks or writes, and who listens or reads, would ultimately balance on the crux of reliability, for once this is decided upon, the isnad has fulfilled its footnoting function for better or for worse, and the quoted passage is judged on a linear scale of authenticity. The where, when, how and why of transmission likewise are mainly of interest to the extent that they strengthen or weaken the claims of authority implicit in any hadith. The basic formula for the isnad of course suppresses such particulars in favor of a chain of names, and in doing so it dissolves the verbal and extraverbal context in which the matn was reportedly quoted. This context, in which someone previously found the hadith worth citing and placed it in some wider discourse or compilation or life circumstance, disappears at every instance of re-transmission unless the teller or writer deviates from the formula. Thus the circumstances of reception are effaced at the moment of retransmission.
     The bare form of the isnad therefore not only serves to authenticate the hadith as a whole but also, by taking the citation continually out of context, preserves it as something set within but detachable from the context in which it was uttered or written, and ensures against any lasting intervention by discourses or circumstances alien to it. The isnad puts many pairs of quotation marks around the matn without allowing any of these quotation marks to frame the quoted words with additional text. It presents the matn as coextensively quoted time and again, adding only a name for each communicative event. Discursive levels representing the sequence of transmissive events may be reconstructed here, but since the matn is

taken to be practically identical at every stage, only a pair of names, those of the addresser and the addressee, would distinguish one level from another. Previous transmitters are dispossessed of the discourse in which they cite the tradition, with the result that the hadith provided it is thought to be sound, retains its original authority: the alleged words of the prophet or a witness's account of his actions come to inhabit the present without bringing with them the many discursive contexts in which they have been transmitted. The isnad thus isolates its citation from much of the refraction that inevitably occurs when wider discourses enclose it. The transmitters play a sort of guardian's role with regard to the words they cite: their function is to deliver them intact, and then disappear; the words themselves come from a source whose authority they can never attain.
     When the isnad was believed to be weak or apocryphal, the hadith as a whole would obviously lose all credibility and be separated from the canon. The loss of authority nullified the tradition entirely. The science of hadith so geared to discovering the reasons why people invented hadith, could abandon its investigation at this point. Understandably, once a hadith was considered to be fiction, it became devoid of interest except perhaps insofar as it might aid in detecting other false traditions.
     In most respects, the multiple citation of Cervantes' double novela behaves very differently from that of hadith. Above all, there tends to be an ever-widening sphere of discursive contexts with each new quotation of a quotation. An essential part of this expansion involves recontextualizing others' discourses in one's own, framing them in such a way that they make sense in new circumstances. Berganza, for example, tells Cipión about his encounter with la Cañizares both before and after his extensive citations of her discourse. Campuzano, for his part, tells his friend Peralta about the dogs before handing him the manuscript of the Coloquio, which supposedly cites everything the dogs talked about, and everything they cited. Through this snowballing, or what I would call augmentative citation as opposed to coextensive citation, alleged acts of communication themselves become part of the narration: tellers, listeners, writers and readers in turn all become figured within and around the discourse they have produced as objects of ever new discourses. Those involved in transmitting hadith, in contrast, become a string of names outside the citation (matn).
     In the Casamiento / Coloquio, the previous telling becomes told with each citation: discours, while remaining discours with respect to the

language it governs, turns into histoire when objectified by another discourse. Narration becomes a narrated event which in turn becomes the narrated event of a narrated event, and so on to about the ninth power.
     Citation temporally aligns communicative events word for word to produce a complex communicative event. Although a cited discourse belongs to the past of any framing discourse, citation itself involves transposing the once here-and-now of the cited discourse into the present here-and-now. Thus discrete communicative processes belonging to different times are made simultaneous when represented through direct discourse. As we read la Camacha's divination, the almost unfathomable time-frames of transmission (or invention), reception and reference artificially coincide with our own reading time. This also occurs in the hadith, except that the isnad, by stripping away discursive contexts, all but mutes awareness of intermediate communicative events and their time-frames.
     To complicate matters further, all the basic types of temporal relations between discours and histoire —as defined by Genette— function simultaneously in Cervantes' text (Figures III 228-34). Characters narrate events after they happen, while they happen (e.g., when ;he dogs speak about what they are doing: speaking), before they happen (e.g., the divination itself), and in the interstices between which they happen. Hence a series of pasts, presents, futures and in-betweens, all projected from different discursive presents, are superimposed one on another. Distinct values and viewpoints of each process of transmission / invention or reception come into play with those of other such processes. For instance, despite his own limited understanding, Berganza takes la Cañizares' already complex understanding of events into account to the extent that his discourse incorporates, allows room for and interacts with her cited discourse. The interplay of understanding within single characters should not be overlooked in this regard: Berganza, as listener to la Cañizares, as teller of his own life-story, and as listener to Cipión, behaves variously according to the three discursive situations.
     So much discursive activity means, of course, that an extraordinary number of minds, both real and imaginary, are simultaneously active in the text, engaged in reading, writing, speaking, listening. There is a compounding of minds, each caught in distinct vital circumstances, each viewing things differently, each with its own active memory and sense of anticipation, each involved in communicative activity. Discourse itself becomes highly intensified, transcending

the normal one-word-at-a-time constraints of written and spoken discourse while seeming to conform to them. Though one word follows another on the page, citation provides a means whereby every word becomes many words.
     The processes of discourse themselves come into such powerful focus in the Casamiento / Coloquio that they turn into more than a medium of expression: they are to an important extent the object of expression in this novelistic discourse about discourse. Nearly every character draws attention explicitly to the ongoing discourse he or she is engaged in by affirming its truth and accuracy or by expressing criticism, doubts, amazement, etc. At the level of the dogs, in particular, this self-consciousness intervenes and monitors speech at every turn. Such awareness and self-referentiality on the part of characters regarding their own and others' discourse, together with our knowledge that the text we read is being read by a character while the fictitious author sleeps, are bound to make us uncannily self-conscious while reading and perhaps even ontologically giddy.
     As one reads Cervantes' deeply orchestrated text, one thus becomes aware of many processes, chronotopes and viewpoints operant at the same time. Augmentative citation arranges, condenses and polyphonizes discourse. Readers are likely to find their attention here and now divided into many elsewheres and other times in which the discourses and their objects are imagined to take place. Their divided focus distributes itself unevenly among many discursive processes, for some of these intrinsically call more attention to themselves than others. Indeed, as occurs in hadith, some processes are almost entirely hidden, such as la Montiela's transcription of the divination. Others stand out strongly. Among the determining criteria involved here are the extent to which anyone's discourse focuses on material other than what it cites, the degree to which any character refers to any other character and his or her discourse, and the ways in which one discourse sets itself apart from —or integrates itself with— the discourse it cites. Character groups themselves, made up of (1) the witches / sorceresses, (2) the dogs and (3) the ensign / licenciate, zone discourse in their own peculiar ways: each group or pair manifests a social coherence and shares a worldview —and hence an orientation towards language— very different from that of the others.
     Changes of medium also accentuate contours between one discourse and another since these set up radically different relationships of transmission and reception. Again and again

textuality incribes orality, which in turn frames textuality by speaking around it and about it. As in the hadith there is no inevitable or irreversible switch from one to the other, though “transmission” does pass through a definitive textual stage in both the hadith and the Casamiento / Coloquio in the form of the canonical compilations and the novela, respectively. Yet if there is one sort of discursive activity that remains in the dark, it is writing. One sees the results of writing, but there are no witnesses to the activity itself. When characters in the novela refer to writing (their own or others'), they do so in the most shorthand way as though it were a simple act. Even Cervantes, in his prologue to the Novelas ejemplares, is remarkably reticent with regard to the process of writing.
     If the simplest act of quotation sets up a two-way relationship between discourses, the interrelationships of multiple citation increase geometrically. Moreover, the discourses in the double novela most often relate to each other through other discourses, whose mediation destabilizes the text to the point of vertigo. Our only access to la Camacha's prophecy is through the language of so many mediators, all of whom necessarily recreate everything they cite.
     Qualitatively, too, citation in the Casamiento / Coloquio brings about surprisingly complex relationships due to diverse ways in which discourses interact. I find that in order to characterize these relationships with even minimum adequacy, one has to resort to a wide range of “principles” and metaphors, many of them somehow anthropomorphic: citation is after all one of the most human activities after laughter. Antoine Compagnon, in his resourceful book on citation, invokes a plethora of such metaphors, often in the form of activities —accommodating, working, playing, exchanging, possessing, tailoring, paper-cutting, and many more culled from various fields of action and knowledge —and yet he might be the first to acknowledge the inexhaustibility of metaphors applicable to citation. Here, then, are a few of the ways I would characterize citation in the Casamiento / Coloquio.
     Most obviously, as in any act of citation, the quoting discourse claims to repeat another that has already unfolded. Accordingly, direct responsibility for the content of the quotation would fall mainly on whoever wrote or uttered it in the first place. The citer's role would be that of a conveyor of language to which a distinct mind and voice, distinct circumstances and intentions, could be assigned. Multiple citation would involve tracing words back to an originating discourse. This is precisely how citation in reliable hadith is regarded.

     Yet by taking possession of quoted discourse, the quoter appropriates that other discourse, including it within his or her own as something “other,” something exogenous. The quoter defers to the citation, letting it unfold as though intact, preserving its alleged verbal integrity, while the quoting discourse appears to suspend its own internal development. At the same time, however, the quoter contextualizes the citation within his or her own discourse, adding voice and directionality to it and generally yoking it to certain purposes, as Bakhtin has so convincingly argued (e.g., Bakhtin 276-94; Volosinov [and Bakhtin] 228-34). The quoting context orients and infiltrates into the quotation, while the latter finds itself surprised, as it were, in an alien context, mimicked by someone else for an alien audience. It goes without saying that even when words are quoted verbatim, as in Pierre Menard's version of Don Quixote,⁸ they are by no means the same words as before.
     When nonexistent discourses are “cited,” as in unreliable hadith they become almost pure functions of the citing discourses, yet retain their exogenous status (a fictive status) through the convention of attribution, as well as through any internal phenomena (e.g., stylistic) that somehow set the quoted discourse apart from the quoting one. In the Casamiento / Coloquio, the various nonexistent discourses of the witches and dogs reveal themselves as exponents of Campuzano's fiction-making, and his discourse along with everything else in the double novela turns out to be an exponent of another authorial imagination. Each discourse, then, is bounded by linguistic / ideological markers and by the character to whom it is attributed, yet entirely infiltrated by intentionalities from “above.”
     The etymology of the verb to cite —to set in motion— suggests another essential aspect of quotation, that of a dual or multiple process in which a citing discourse revives another and sets it in motion, with the result that all discourses involved necessarily “happen” simultaneously. Such inert terms as “structure,” “embedding,” “frame” or even “level,” though difficult to avoid when talking about citation, tend to deny this movement by spatializing —and to that extent falsifying— the essentially temporal, processual nature of quotation. These terms are of course a misleading legacy of structuralism, which characteristically abstracted process out of

     ⁸ Jorge Luis Borges, “Pierre Menard, autor del Quijote,” in Prosa completa (Barcelona: Bruguera, 1980), 1:425-33.

narrative, even when temporality was the object of inquiry; failure to recognize their artificiality and distortion precludes an adequate conceptualization of citation.⁹ To cite is to be active in discourse and activate another discourse, setting it in motion at the same time as one's own and in the same direction, so to speak. Even the minimal formula of the isnad, often reducible to “A said (qala) that B said that C said . . .” concatenates simultaneous discursive processes. As I have already suggested, the Casamiento / Coloquio may be seen as a complex event in which discourses mobilize other discourses to their own purposes. How and why they do so is of key importance.
     Furthermore, quotation affects authority of discourse. In this respect, the hadith and the Casamiento / Coloquio are poles apart. The authority of the hadith, as we have seen, is already immanent in the attribution of speech to the most authoritative human speaker, the Prophet, but depends on the reliability of transmitters; when any of these is considered suspect, the validity of the hadith as a whole is placed in jeopardy. In the Casamiento / Coloquio, the founding discourse is to be found not in what would be the most distant source —la Camacha's divination, for example— but rather in the most immediate source, the novels themselves. Authority and authenticity are completely at odds, though Cervantes naturally makes use of the conventions of authenticity, of verbal transmission, as he undermines them.
     As sorceresses / witches or as dogs, the first five transmitters of the divination establish one of the most feeble chains of corroboration imaginable: witches and dogs in Cervantes' (and Peralta's) world possess little or no discursive authority. Yet judged according to how they verbalize experience, la Cañizares, Berganza and Cipión would register remarkably high on the scale of reliability despite their self-doubts and close involvement with the events narrated. Where reliability breaks down entirely is in the delirious consciousness of the convalescing Campuzano, who has already in his own tale shown himself to be capable of delusion and thus of fiction-making; the

     ⁹ The classic structuralist analyses of narrative all elude questions concerning process, preferring to deal with a more stable entity, the textual product (Levi-Strauss, Genette, Todorov . . .). Oddly enough, recent narratology seems not to have taken up the challenge of rethinking “levels” in terms of the dynamic processes involved (e.g., Bal, Rimmon-Kenan). Deconstruction, for its part, seems to be as entrapped as ever in this sort of terminology.

presence of speaking dogs renders his text even more suspicious. At this point the manuscript of the Coloquio becomes revalued as invention, as artifice, intended for the mind's recreation, as Peralta points out (359). The breakdown of internal authority heightens the reader's awareness of the fictional status of the double novela as a whole.
     This collapse is one of isnad, since it happens that Campuzano, whether he admits it or not, whether he knows it or not, has invented the colloquy he claims to have transcribed. Yet contrary to the transmission and criticism of hadith, there is also a direct attack on the matn itself —the divination— as well as on the sorceress who uttered it, in the form of Cipión's critique. At the moment of utterance, the divination derives its authority from the proven power of the sorceress, from the highly specialized language of divinatory verse, and from the hold it claims to have over the future of Montiela's offspring. Yet its ambivalent terms and dubious contingency clause render it more than suspect to Cipión: even if it were to come true, who could ever be sure that it had done so? After considering implausible figurative and literal interpretations of the lines, the dog rejects the “text” itself as malicious nonsense and attributes this to the character and profession of the sorceress and her colleagues (“la Camacha fue burladora falsa, y la Cañizares embustera, y la Montiela tonta, maliciosa y bellaca . . .” —347). Not only here but with each new speaker and listener, writer and reader, la Camacha's unverifiable divination loses in prophetic authority and becomes enriched in meaning, adulterated in intentionality. Citation undermines the prophecy's personal authority, poetic inviolability and control over the future. Even Cipión's critique becomes text, undergoing a similar loss of authority as it is subordinated to alien values and intentions. Through “transmission” with its widening texts and contexts, speakers and writers in the Casamiento / Coloquio lose control of their own meanings and of their own being as authority dissipates outwards. Characters are unable to control what happens to their own discourse once someone else has appropriated it. In particular, Campuzano's eavesdropping of discourse not intended for him, as well as his “ghostwriting” of a text not directly intended for us, demonstrates discursivity getting out of hand.
     Citation figures so insistently in the Casamiento / Coloquio that it ceases to be merely a technique: every word in the double novela somehow resonates from discursive interplay. Whereas citation in hadith literature strips away discursive contexts and strives to retain

the authority of the transmitted word, citation in the Cervantine text relativizes, decenters, undermines, recontextualizes, ironizes, enriches meaning —to mention only a few of its effects. Any attempt to deal with the complex issues of value and meaning in the Casamiento / Coloquio therefore has to take into account the destabilizing effects of citation —of transmission and invention, and corresponding reception. For this reason, simple moralistic statements in the criticism of the novela tend to fall flat. To deceive others or oneself, to act hypocritically, to engage in tropelía (making one thing appear to be another (337), to participate in witches' Sabbaths, to practice sorcery, to imagine talking dogs, to dream, to write fiction, to read it —all of these are analogous activities in the novela, some positively and some negatively charged. I would argue that in the context of the Casamiento / Coloquio, none of these important themes can be adequately discussed without reference to their subtle interaction with the others. This means taking the complex discursive interplay of the novela fully into account.
     I would further hypothesize that although various episodes in the Casamiento / Coloquio resolve themselves, not one of the many discourses operant at the time that Cipión quotes the prophecy even approaches any kind of resolution. Nor does one discourse solve the problems of another discourse. Although writing may be a therapeutic diversion for Campuzano, his fiction can hardly be said to illuminate his own past or present circumstances. The various communicative encounters end inconclusively and give way to a present as openended as an unfulfilled and suspect prophecy. Augmentative citation contributes to this openendedness, since the recontextualization of discourses revives and reorients them; it predisposes one to view discourse as uncontained and unfinished, as a praxis rather than a product.
     The citation of hadith is also unfinished, since what was once new (the words and / or deeds of the Prophet) is made new again with each telling. Indeed, the root of the word, hdth, meaning “to happen” (form I) and “to tell a happening” (form II), points at the same time to a primary event and its renewal through citation. In order to keep that happening new while externally supporting its authority, the isnad dissolves the intermediate events of citation with all their context, retaining only the insoluble names of the tellers. Thus, in the case of strong hadith, the matn is felt to be both immediate and verifiable despite the dubious medium of popular transmission over generations. The hadith typically allows one to ask what the Prophet said or

did, according to whom, and what applicability this could have to life now.
     In the Casamiento / Coloquio, Cervantes pushes the techniques of citation about as far in the opposite direction as anyone has in Western literature, showing not their limits but their limitlessness despite discursive constraints. He exploits one of prose fiction's integral features and richest resources —citation— and mobilizes it fully in service of the wider aims of the double novela. Who, how, why, what for, for whom: these are some of the questions that emerge most insistently about every transmissive event. Citation in the Casamiento / Coloquio accumulates the circumstantiality of discourses about discourses so as to bring about a simultaneously mediated discursive process, heightening readers' awareness of the multiple significance and intentionalities of the words cited. It show how citing and inventing in the novel are sometimes indistinguishable. It shows, moreover, how people or characters articulate their own discourses obliquely through the discourses of others —and how novels cite their own invented discourses.

Selected Bibliography

Bakhtin, M. M. The Dialogic Imagination: Four Essays. Trans. Caryl Emerson and Michael Holquist. Ed. Michael Holquist. Austin, Texas: U of Texas P, 1981.

Bal, Mieke. “Notes on Narrative Embedding.” Poetics Today 2 (1981): 41-59.

——. Narratology: Introduction to the Theory of Narrative. Toronto: Toronto UP, 1985.

Barth, John. “Tales within Tales within Tales.” Antaeus 43 (1981): 45-63. Cervantes Saavedra, Miguel de. Novelas ejemplares. Ed. Harry Sieber. Madrid: Cátedra, 1981. 2 vols.

Chambers, Ross. “The Artist as Performing Dog.” Comparative Literature 23 (1971): 312-24.

Compagnon, Antoine. La Seconde Main ou le travail de la citation. Paris: Seuil, 1979.

Cragg Albert Kenneth. “Hadith.” Encyclopaedia Britannica: Macropaedia. 1984.

El Saffar, Ruth. Novel to Romance: A Study of Cervantes' Novelas ejemplares. Baltimore: Johns Hopkins UP, 1974.

——. Cervantes: El casamiento engañoso and El coloquio de los perros. London: Grant & Cutler, 1976.

Forcione, Alban K. Cervantes and the Mystery of Lawlessness: A Study of El casamiento engañoso y El coloquio de los perros. Princeton, New Jersey: Princeton UP, 1984.

Genette, Gérard. Figures III. Paris: Seuil, 1972.

——. Nouveau Discours du récit. Paris: Seuil, 1983.

González Echevarría, Roberto. “The Life and Adventures of Cipión: Cervantes and the Picaresque.” Diacritics 10 (Fall, 1980), 15-26.

Huerga, Alvaro. “El proceso inquisitorial contra La Camacha. Cervantes: Su obra y su mundo. Actas del I Congreso Internacional sobre Cervantes. Ed. Manuel Criado de Val. Madrid: EDI, 1981. 453-62.

Ibn Hazm. El collar de la paloma. Trans. Emilio García Gómez. Madrid: Sociedad de Estudios y Publicaciones, 1967.

Marawski, Stefan. “Quotation in Art.” Inquiries into the Fundamentals of Aesthetics. Cambridge, Massachusetts: MIT Press, 1974. 341-61.

Meyer, Herman. The Poetics of Quotation in the European Novel. Trans. Theodore and Yetta Ziolkowski. Princeton, New Jersey: Princeton UP, 1968.

Pozuelo Yvancos, José María. “Enunciación y recepción en el Casamiento-Coloquio.” Cervantes: Su obra y su mundo. Actas del I Congreso Internacional sobre Cervantes. Ed. Manuel Criado de Val. Madrid: EDI, 1981. 423-35.

Reyes, Graciela. Polifonía textual: La citación en el relato literario. Madrid: Gredos, 1984.

Rimmon-Kenan, Shlomith. Narrative Fiction: Contemporary Poetics. London: Methuen, 1983.

Siddiqi, Muhammad Zubayr. Hadith Literature: Its Origins, Development, Special Features and Criticism. Calcutta: Calcutta UP, 1961.

Siddiqui, Abdul Hamid. Selection from hadith. Kuwait: Islamic Book Publishers, 1979.

Stanzel, Frank K. “Teller Characters and Reflector Characters in Narrative Theory.” Poetics Today 2 (1981): 5-15.

Volosinov, V. N. [and M. M. Bakhtin]. Marxism and the Philosophy of Language. Trans. Ladislaw Matejka and I. R. Titunik. Cambridge, Massachusetts: Harvard UP, 1986.