From Reader To Writer

In moving through Chaucer's work, we observe a development in the figure of the Narrator. The Chaucerian Narrator is always both reader and writer, yet the balance of these functions is not constant. In the earlier dream-visions, the readerly function dominates as the Narrator confronts and absorbs various discourses, texts, and experiences. In some cases a specific text is named and summarized: Ovid's Metamorphoses and Heroides, Virgil's Aeneid, and Macrobius's commentary on the Somnium Scipionis are among them. This text provokes the Narrator's dream, which, when recorded, becomes the poem at hand. The narrative stance, then, is a passive or receptive one at the start of the poet's career. Over a period of about

eighteen years, however, it shifts very perceptibly toward an active and self-conscious authorial position. The balance inclines progressively toward the writerly function, until in the Canterbury Tales we are very little aware of the Narrator as reader or scholar. Instead, we are invited to see him primarily as writer and artist, his ostensible sources no longer books but people and experience: "the book of the world," as a medieval cleric might say. In reality the Tales are of course every bit as bookish or intertextual as the earlier compositions. What I investigate here is not, however, actual literary sources, but rather the diegetic claim of the work, what the work asks us to believe about its Narrator and the processes of its own poetic production: a more speculative area of investigation than source study perhaps, but, for any writer, equally important.

It is probably well to address at the start the notion of chronology implied in these remarks. Robert O. Fayne—to whose book The Key of Remembrance every Chaucerian interested in poetics must acknowledge a debt—thought the chronology of the Chaucer canon so precarious that "it would be foolhardy and pointless ¼ to suggest any particular direction of development in [Chaucer's] poetics" (115). It is no depreclation of Payne's contribution to point out that this radical skepticism about order is exactly what his project requires, for his focus is "structural stereotypes" that cut across chronology. My hypothesis, on the other hand, while not originating in the traditional chronology, does tend to confirm it, although not rigidly (as will emerge later on). Hence the traditional chronology will be a useful, not a decisive, support for my hypothesis. There are limits to doubt, as my discussion of philosophical skepticism will note (see the second part of this chapter), and while we lack certainty about chronology, there is some probability.

In the Book of the Duchess (1368), the Narrator appears almost exclusively as reader. The setting of the dream-vision is the Narrator's chamber, where he is reading a romance in bed late at night. It is the story of Ceix and Alcyone (Halcyon) from Ovid's Metamorphoses 11, a tale whose theme of death and transfiguration anticipates the elegiac narrative to follow in the dream. After a lengthy retelling of this story, the Narrator falls asleep while reading. References by the Narrator to his authorial function are sparse and perfunctory. Lines 216–19, 226, 271, and 711 refer to the Narrator relating a story

or a dream, not writing a poem, so that it could be maintained that these places do not constitute true authorial self-reference. For actual statements of writerly function in the Narrator's voice, only two appear, and these are extremely timid, particularly in comparison with what will come in later works. There is the "I, which make this book" of line 96, and at the very end of the poem:

Thoghte I, "Thys ys so queynt a sweven
That I wol, be processe of tyme,
Fonde to put this swven in ryme
As I kan best, and that anoon."
This was my sweven; now hit ys doon.
                            (1330–34)

Now there is a living poet in the Book of the Duchess, but it is not the Narrator. It is the Black Knight, whom the Narrator meets in the forest of his dream, and who is surprised in the very act of composing "a compleynte to hymselve ¼ a lay, a maner song / Withoute noote, withoute song" (464–72). This plaint is reproduced in the Chaucerian text (475–86), and the Black Knight's subsequent elaboration of his loss—his gloss upon his own poetic composition—constitutes the body of Chaucer's poem.

What kind of poet is the Black Knight? He seems to be not only an active maker but an experienced and remarkably confident one, with, apparently, the faculty of virtually total recall of all he has ever composed. He claims to have made many songs and performed them often (1155–59); he associates himself through the modestytopos with the archetypal scriptural and classical progenitors of the art, Tubal and Pythagoras (1160–70); he remembers and repeats his very first ("altherferste") youthful composition (1175–80). The Knight is equally self-conscious in other speech acts as well. He is able to recreate in detail his first confession of love to his lady (1181–1257) with all the psychological circumstances attendant upon this performance. These include his ambivalence about speaking, his worries about how to begin, and his nervousness at the possibility of a bad recital. The fear that causes him to omit part of his script ("many a word I over-skipte" [1208]) seems less a fear of the lady herself than of performing badly: "for pure fere / lest my wordes mysset were" (1209–10). He recalls, too, indeed he quotes from memory, the begging for mercy and swearing fidelity (1221–35),

and notes his inability accurately to reproduce the lady's response except for its general tenor refusing his love (1236–44). Finally, he recounts the depression he suffered because of the lady's refusal.

The Black Knight, then, and not the Narrator, is portrayed as productive rhetor-poet in the Book of the Duchess. It is he who describes a poetic career, dwelling particularly on the launching of that career. He does so in the poem that is generally considered to have launched Chaucer's career as a courtly poet, not only because of its apparent priority in the canon, but also because of its engagement with a significant event of courtly social life, the death of John of Gaunt's Duchess Blanche. Onto the Black Knight, then, onto the fictional character, the poet has displaced active authorial function.

It is in this sense that I see the Black Knight as a kind of "alter ego" to the Narrator, rather than with respect to the issue of mourning, which has generally been the focus of criticism of the poem. Some remarks are in order here on the "alter ego" notion and on mourning. First, it seems obvious that in any literary work, every fictional character with any depth at all is to some extent a projection of its creator's inner life. In proclaiming, "Madame Bovary, c'est moi," Flaubert may have stated the principle with blunt unequivocality, but he can scarcely be considered unique in this matter in the history of Western literature. It would be ludicrous, I believe, to confine a discussion of "alter ego" to relations among characters and omit from it the author. My discussion is plainly one that in principle accepts the concept of character as at some level authorial "alter ego," although as I shall note further on, the extent of the relationship between author and narrator or author and character is neither constant nor predetermined in any Chaucerian text.

What I find odd in some recent uses of the "alter ego" notion is its application. In 1963, Bernard Huppé and D. W. Robertson, Jr., wrote of the Black Knight that he was not intended as the dream representative of John of Gaunt, but rather as a sorrowing alter ego of the speaker in the poem, like the poet himself representative of all those who have honored and loved Blanche and lost her in death (52). There is a peculiar circularity to this, which winds up negating its own initial assertion. John of Gaunt is excised; the Knight is alter ego to the Narrator, who in turn is "like" the poet, who in turn "represents" all who have lost Blanche—and would this not, after all, include her husband? In general structure this follows the pat-

tern established by Bertrand Bronson in 1952. While reminding us that John of Gaunt was out of England during most of the nine years of his marriage to Blanche, including the time of her death, Bronson Stressed the therapeutic function of the Narrator's dream. The Black Knight is the Narrator's "surrogate," his sorrow a projection or externalization of the Narrator's own "private grief" described at the beginning of the poem. A few pages later, however, the Man in Black also appears "as John of Gaunt," and the article ends with an orgiastic fusion of all personae including ourselves: "in the presence of death, Ceyx and Blanche, Gaunt and Alcyone, the Dreamer and Chaucer and his audience too, of which we now form a part, are become essentially one" (Wagenknecht, 294). A more sophisticated version of the position is offered in Judith Ferster's hermeneutically based discussion of "characters who are shadows of each other" (74), a narrator "who becomes a version of the knight" (92), and the evolution of each by his absorption of the other.

Such interpretations take a great deal for granted. In the most general, even banal, sense that we will all suffer the death of a loved one and require consolation, all the personae, including ourselves, do share something. But to focus on the issue of sorrow maintains the poem in a narrowly autobiographical mode, whether the sorrow is seen as that of John the Gaunt, the poet Chaucer, or the Chaucerian Narrator. This focus on sorrow can be called historicism, for it links the poem to a known historical event, Blanche's death in September 1368. It is a narrow historicism, though, and one that satisfies our own sentiment—indeed, our own sentimentality—by imposing on the past what we consider the "right" response to a noblewoman's death. We do this partly because we know that the real noblewoman was a "wife," and to this term there adheres a very considerable body of evaluation and response, conditioned partly by the development of family mores during the centuries after Chaucer's death. The realities of marriage and family were, however, rather different among late-medieval gentry and aristocracy, where matches often had little to do with the couple's taste, desire, or choice; where servants, tutors, and ladies- or gentlemen-in-waitmg often performed the nurturing and disciplinary functions now associated with parents; where travel abroad or among a family's domestic territories kept a noble family dispersed. Then, just as now, practice often deviated from the recommended ideal—

surely this is why the ideal had (and has) to be recommended so often; but one ought not to take one for the other. There is anecdotal evidence for less than ideal practice; one thinks, for instance, of the Pastons' occasionally harsh treatment of their daughters.^[1] As well, the "ideal" itself might be disputed or redefined from the perspective of intense religious commitment. St. Elizabeth of Hungary and St. Bridget of Sweden—both of them mothers of several children and well-known devotional models in Chaucer's England—pointedly renounced their family ties as hindrances to perfect spirituality.

A century or so after Chaucer, the anonymous secretary to Francesco Capello, Venetian ambassador to England, recorded the earliest extant Venetian account of the country. Besides English smugness, insularity, and superficiality, the young Italian deplored "the want of affection in the English ¼ manifested towards their children; for ¼ at the age of 7 or 9 years at the utmost, they put them out, both males and females, to hard service in the houses of other people."^[2] Although this and my other examples of contemporary family relations are (somewhat) later than Chaucer, and do not constitute a "scientific" survey, they suggest a climate of opinion and practice. Nobles and gentry did send their children out to noble homes to learn manners and a career—Chaucer was a page in the household of Lionel, duke of Clarence—while bourgeois and artisan families often sent theirs out to apprentice in a trade; generally the children did not return home after their training but went on to work and marry. The fact of intra-familial murder among French and English noblemen of the period, for dynastic reasons, also makes my point about a different conception of familial obligation. So does the not inconsiderable evidence of infanticide and abandonment,^[3] and so do the intra-familial lawsuits about money and property that went on at every level of society. All this is not to deny the existence of affectionate family ties—indeed the Paston letters are full of such affection as well as of friction and even abuse. It is

[1] The Pastons were a Norfolk (East Anglian) family of prosperous business people and civil servants; their letters and documents cover the period 1418–1506.

[2] In Sneyd, 24–25. For other information in this paragraph, see Gies and Gies.

[3] Although the nature of medieval family life and attitudes is sharply disputed among social historians, nonetheless the sombre realities of infanticide and abandonment have been documented by Kellum, Trexler, Helmholz, and Boswell.

intended, however, to caution against automatic assumptions about the quality of a given marital or familial relation, and against the imposition upon medieval courtly life of the sentiments of our own domestic lives.

There is, therefore, a certain naïveté—it may alternatively be referred to as "the sincerity fallacy"—in assuming that the author of a courtly elegy is himself really in pain, or indeed that anyone is in pain, including the dead woman's husband. In the middle of the next century, the chronicler John Hardyng's dry comment on the marriage of John to Blanche of Lancaster was, "The duchy by hir had, men saied he had well sped" (330); and while Hardyng had a political-dynastic axe to grind, I doubt his estimate can have been unique. That John of Gaunt built an architectural monument for his deceased wife is no evidence to the contrary, notwithstanding that Donald Howard bids us remember, in support of his evocation of Gaunt's "towering" grief, the effigy erected on Blanche's grave. (The critic's perhaps unconscious wordplay with "towering" is very much in the Chaucerian spirit, as Chapter 3 will demonstrate).^[4] The project proves nothing about John's feelings for his wife, although it certainly tells us something about his sense of social status and public display.

It is equally gratuitous to assume that the real death was perceived by the poet as anything other than an opportunity to write. Robert Burlin mentions the "ecstatic mutuality" that is Chaucer's "finest compliment to his lost Duchess" (63). Despite the homophonic allusion to Browning (there seems to be an irresistible pull to wordplay even among scholars concerned to establish sincerity of sorrow), I am not convinced that the intensity of the poem has anything to do with Blanche. I find Earle Birney's comment a refreshing antidote:

The Duchess is not an ironical poem, yet is there any other elegy in the language with such playful passages and with such a general effect of lightheartedness? From what we know of Gaunt's character, and specifically of his readiness to remarry, it is likely of course that the husband's grief was more chivalric than intense; certainly Chaucer's lament seems delicately adapted to just such polite mourning.

[4] Howard, Chaucer, 123. In "Chaucer the Man," Howard also writes of shedding "light on the Dark Lady."

I use the term "alter ego," then, in gingerly fashion, and in the context of the production of fiction rather than in that of loss and consolation. This might, I suppose, be considered equally autobiographical, although in another register than the immediately and perhaps mechanically historical. I do not claim that the Narrator "is" Chaucer. The Narrator is a fictional character among others. Onto the Narrator, and sometimes onto other characters, the poet displaces certain of his own activities, particularly the reading and writing that are my concern here. From the proportionality of this displacement, we may be able to infer something about Chaucer's sense of himself as a developing poet.

That the Black Knight should be so strongly characterized as maker, and the Narrator play so depreciated a role with respect to writing, is no mere flattery of a recently bereaved patron. Surely we may also read in this displacement of function the real poet's ambivalence or uncertainty about his own social role and his burgeoning poetic powers. For Chaucer was still between twenty-five and thirty years of age; he had recently married the royal mourner's future sister-in-law, Philippa Payne de Roet, a woman far above her husband in social rank. He was a relative newcomer to the king's service, although he had been trained for it since early adolescence. With the Book of the Duchess he was making a bid to become something more at court than a minor diplomat: both an acknowledged poet and a lucratively rewarded courtier. For the moment, however, the future was uncertain.

With the House of Fame, we enter the heart of uncertainty. This begins with the date of the poem, which has been estimated as between 1374 and 1385. It is generally agreed to be later than the Book of the Duchess and probably dates from about 1375. Although much else is at issue in the House of Fame, Chaucer has at least now incorporated the role of writer into his first-person narrative persona. The Narrator is both reader and writer from the start of the poem, but the readerly role continues to dominate and the writerly role is marked by uncertainty.

The 65-line Prologue reflects the Narrator's experience as reader. His puzzlement about conflicting theories of dream produces an agnostic stance ultimately eased by the fideistic transfer of responsibility to divine power:

For I of noon opinion
Nyl as now make mensyon,
But oonly that the holy roode
Turne us every drem to goode!
                      (55–58)

No such easy route will be available in the writerly role, where responsibility for error is untransferable (although, as the Retraction to the Canterbury Tales acknowledges, good effects may be attributed to divine influence). Self-referential comments about authorial function in the Proems and Invocations to each of the three divisions of the poem are timid, self-doubting, and modest. They reveal anxiety about accurate narration of the dream (79 and 527), about the technical ability "to endite and ryme" (520), about the difficulty of reproducing one's idea (1101–3). These are, of course, among the important questions for any artist, but they usually loom largest at a relatively early stage of a writer's career. They must be resolved before another set of important questions can be dealt with, such as the artist's social and moral responsibilities, attitudes toward love and nature, place in a tradition, and so on. The curiously vehement blessing and curse on the good and bad audience (81–108) suggest the defensive maker, unsure of his reception and despising in advance any potential malicious or philistine misinterpretation of his work.

The content of the dream-vision is highly literary and traditional: again it is a scholar's dream, this time centering on the story of Dido and Aeneas. Yet if the sources of the main story are undeniably literary—derived from those two authoritative (and conflicting) texts, Virgil's Aeneid and Ovid's Heroides —nonetheless the diegetic representation of this story is far from being straightforwardly literary. There is in fact a very odd mixture and confusion of media in this section of the poem; surely one of its most disorienting features. Ekphrasis, the representation of one medium in another, is an important technique in Ovid's Metamorphoses, as Eleanor Leach has shown. But it is most probably from Virgil that Chaucer borrowed it, for when Aeneas first arrives at Carthage, he sees, upon the walls of the Temple of Juno, the visual representation of his own history and the fall of Troy (Aeneid 1.453–93). Whether the medium is tapestry, painting, or relief is not specified, but the verbal represen-

tation of the visual representation is at least clear and consistent, and if supplementary explanation of a scene appears, this can be attributed to the viewer's—Aeneas's—intimate knowledge of the persons and events portrayed (e.g., the interjection about Troilus, "infelix puer atque inpar congressus Achilli": "unfortunate boy, and unequal to the meeting with Achilles" [1.475]). Nonetheless, both Virgil and Chaucer offer, in a sense, a double ekphrasis. Neither of them verbally represents a simple three-dimensional art object; rather they represent something ornately decorated that—like Achilles's shield in Iliad 18 and Keats's urn—in turn represents narrative. If ekphrasis is, to borrow Murray Krieger's phrase, language attempting to "freeze itself into a spatial form" (10), then the ekphrases discussed here re-freeze a spatial form that has already frozen into spatial form the language-exchanges and the actions it represents.

Not all classical or later ekphrasis is as clear as Virgil's. Discussing the representation of maps in painting, Svetlana Alpers notes that the only Greek word the Alexandrian astronomer Ptolemy (2d century C.E.) had available for referring to a maker of pictures was graphikos, a suffix denoting one who writes, draws, or records. Later,

where the word description is used by Renaissance geographers, it calls attention not to the power of words, but to the sense in which images are drawn or inscribed like something written. It calls attention, in short, not to the persuasive power of words but to a mode of pictorial representation¼ . When we look back at Ptolemy now we have to say that his term grapho was opened up to suggest both picture and writing. (Alpers, 136)

Closer to home, the medieval commentaries on classical myth often employed ekphrasis in their verbal representation of visual representation. The operative word is pingitur, from Latin pingo, to represent pictorially whether in drawing, painting, embroidery, or tapestry. Beryl Smalley notes that Fulgentius, writing in the late fifth century, occasionally uses the word; Alberic of London about 1200 uses it more often, and the fourteenth-century commentator John Ridevall "never omits it." Smalley goes on to observe that "Ridevall's 'pictures' did not lend themselves to visual representation and the results were as clumsy as might have been expected. Ridevall de-

scribed Juno as 'redolent of unguents'. The poor artist could only set an open flask beside her head" (English Friars, 112–13). What we encounter in these cases seems to be an aspect of manuscript culture and early print culture, where modes of apprehension interpenetrate. Part of this is doubtless due to the relative novelty and marginality of writing itself, an art whose boundaries had not yet become rigidly fixed.

The Chaucerian Narrator, like Aeneas, also finds a visual representation on the walls of Venus's temple; it is not only the history of Troy but of the entire Aeneid. The treatment of this motif, however, is very different from that in the classical source. To begin with, Virgil's succinct comment that the Temple of Juno is donis opulentum ("opulent with gifts" [447]) is amplified into an architectural nightmare. The Chaucerian Temple of Venus is an unusually bizarre and grotesque example of high Gothic architecture (121–27): it is made of glass and profusely filled with images, niches, altars, towers, statuettes, and paintings. The setting is thus as chaotic as the representation of the ancient story will shortly prove to be. About this puzzling edifice the Narrator knows only that it is the Temple of Venus, for he sees a painted portrait of Venus surrounded by her usual iconography (129–39). Next he finds a brass tablet on which is written a translation of the opening lines of the Aeneid (140–48): "And tho began the story anoon, / As I shal telle yow echon" (149–50). This transitional couplet leads us to expect a smooth passage into the rest of the story, and to expect that it will appear to the Narrator in the same medium as what he has already seen—that is, as words written on a brass tablet. Yet such is not evidently the case. At no time, for instance, does the Narrator use the word "read" as we might expect with a written account; it is always "see," more appropriate to visual than to verbal representation. On the other hand, what the Narrator "sees" includes motivation, lament, and emotion—in short, narrative with all its rhetorical features. The frequent recurrence of the phrase "sawgh I graven" and other forms of the verb "graven" only intensifies our uncertainty, for the word may mean to carve, as in statuary (a "graven image" is an idol), or it may mean to incise either words or images on metal or stone.

In line 211 appears the phrase "peynted on the wal." This seems unequivocal enough, except that the appearance of the word "peynte" a few lines later (246) to mean verbal rhetoric once again

confuses the picture. Perhaps it is well that we are forced to problematize appearance, since the question of deceptive appearance is highlighted as the core of the love story (263–92). Next, the Narrator appears to abandon the fiction of visual art in referring us to books if we want to know the details and the analogues of the love story (375–432). Then the ambiguous "sawgh I grave" formula reappears several times. Finally, the Narrator reflects on the "noblesse / Of ymages" (471–72) he has seen and his ignorance of "whoo did hem wirche" (474), reverting to the imagery of the visual arts. The episode ends with the Narrator as confused as his reader. Indeed, it is the Narrator-as-reader who is confused, with no one to advise or inform him ("rede or wisse" [491]) about the unfamiliar desert landscape he now confronts and is unable to interpret.

And, after all, interpretation has been the problem all along. What we find in the Dido and Aeneas story is something analogous in literature to the portrayal in painting of a written inscription. Mieczyslaw Wallis has called this, rather than ekphrasis, a "semantic enclave": part of a work of art that consists of signs of a different kind or from a different system than the signs of which the main body of the work consists. This might be, in a text, the insertion of a different language for part of the text, as in macaronic verse; or it might be the use of illustration, as in an illuminated manuscript. In a painting, the autonomous semantic enclave might be the representation within the painting of a map, a musical score, a heraldic device, or an inscription, whether free-floating or placed on an object (e.g., a book, banner, robe, etc.). Cubist, dadaist, or surrealist collage would be modern examples.

While the text at hand remains integrated in the sense that it is all in the same language, nonetheless it seems to me that the constant shifting of ground between visual and verbal registers in this portion of the House of Fame operates as an oscillation between semantic enclaves, reminding us, as it must remind the Narrator, of the proliferation of languages, texts, and meanings: the unreliability of communication generally that is the lesson to be learnt at the houses of Fame and Rumor.

For, of course, it is not only the desert landscape the Narrator-asreader is unable to gloss, but his experience in the Temple of Venus. The dual Virgilian-Ovidian tradition of the Dido and Aeneas story is quite as problematic for a reader as the multimedia representation

of it the Narrator has just given. In that sense, the Dido-Aeneas story is paradigmatic of other written traditions to be encountered by the Narrator, traditions in science, myth, and history whose internal contradiction will pose a similar problem to that of the classical love story and be resolved, as I have argued in Poetics of Skeptical Fideism, in a similar way.

The inner structure of the House of Fame thus centers on the Narrator as reader: more generally, as observer-interpreter. Its plot, however, focuses on his role as writer, and one notes that it is in this text that, for the first time, the author names his Narrator with his own—the author's—proper name: "Geffrey, thou wost ryght wel this" (729) are the words with which the kindly Eagle begins his lecture on sound. Geffrey's journey to the houses of Fame and Rumor is his reward for perseverance in poetic making. But when the great eagle, Geffrey's psychopomp, says to his passenger,

when thy labour doon al ys ¼ .
Thou goost hom to thy hous anoon,
And, also domb as any stoon,
Thou sittest at another book
Tyl fully daswed ys thy look ¼
                           (652–58)

we do not know whether it is as reader or as writer that Geffrey dazes himself in front of another book. The ambiguity encapsulates my point about the poem's dual representation of narratorial roles.

The aim of the aerial trip is to reward Geffrey as poet by providing him with "tydynges" (information) about "Loves folk" that will amuse and instruct: it will be a diversion, and the implication is that the tidings may furnish some vicarious experience of love or even material for further making, although these latter purposes are not specified. The emphasis is firmly on hearing, not on utterance. Accordingly, the Eagle's farewell benediction to Geffrey is not that he write well about his journey, but that he have the grace "Some good to lernen in this place" (1088). The true center of the poem is not yet the production of discourse but its reception.

In the Proem to book 3, the Narrator at first appears as writer, placing himself under Apollo's aegis. Nonetheless he quickly reverts to the receptive-scholarly position as he confronts, in the narrative, a central readerly problem: the unreliability of fama (fame,

reputation, tradition) and especially the coexistence of mutually contradictory versions of the same events. Given the evanescence and ambiguity of tidings, it is scarcely surprising that none can be produced as a climax to the poem despite the scene of frenetic anticipation with which it ends. Even with its intermittent glimmers of authorial self-assertion, the House of Fame remains essentially a book about reading, with writing submerged as a kind of by-product of reading.

The equilibrium of roles in the Parliament of Fowls (dated between 1374 and 1380) is fairly similar to that in the House of Fame. John Fisher has remarked that although the conventional placement of the House of Fame before the Parliament is "satisfactory, ¼ the reverse order would do just as well" (208), and so it would for my present purpose. Again the Narrator enters the poem as a reader; again the dream-vision is stimulated by his reading of a classical literary source; again he requests the ability to rhyme and "endite" (119); again the dream-journey is a reward for labor, although here it is the scholarly labor of reading Macrobius's commentary on Cicero's Somnium Scipionis (109–12) rather than of writing love songs. The purpose of this journey is not simply to acquire information but now explicitly to acquire poetic material, for the psychopomp Africanus says, "I shal the shewe mater of to write" (168). Yet this offer of poetic material is undercut by the snide comment of Africanus: "And if thow haddest connyng fort' endite ¼ " (167). So an ambivalence prevails with respect to the writerly role.

The dream-content is at first extremely static, emblematic, and mythological. It demands interpretation or at least a syntax: its self-evident metonymies (Bacchus/Ceres/Venus, Cupid/Will, Behest/ Art et al.) require to be ordered by the observer into some sort of structured, intelligible statement about love. But this scenario shortly gives way to the dynamic, colloquial, and rapid-paced birdparliament in Nature's Park, where utterance, albeit not the Narrator's utterance, is the center of interest.

As in the Book of the Duchess, the Parliament offers another poem within a poem, the concluding rondel. As in the Duchess, performance is ascribed to fictional characters (here, the birds) rather than to the Narrator. Composition is left anonymous, both for words and music: "The note, I trowe, imaked was in Fraunce, / The wordes were swiche as ye may heer fynde" (677–78). While the Duchess

ends with the promise of writing, the Parliament promises further reading and hopes for more dreams that will in turn effectuate improved experience:

I hope, ywis, to rede so som day
That I shal mete som thyng for to fare
The bet, and thus to rede I nyl nat spare.
                               (697–99)

We know that dream operates as a metaphor for composition, but composition itself is effaced in this ending.

Anelida and Arcite (of uncertain date) seems, despite its brevity and lack of poetic distinction, to suggest a more even distribution of functions than appears in the House of Fame, with the writerly coming more into prominence. (My reading thus tends to confirm that of J. Norton-Smith, who links the work with the Legend both chronologically and thematically.) To begin with, Chaucer plays fast and loose with sources. He claims to be translating an old Latin story (8–10) but no such text has been identified. He claims to follow "Stace, and after him Corynne" (21) and even if scholars were agreed on the identity of Corynne, he or she has left no literary remains. The Thebaid of Statius may have provided some details at the beginning of the poem, but not its body; and Boccaccio's Teseida, also used, is not mentioned at all. "Scholars are loath to credit anything to pure invention," observed F. N. Robinson in his introductory remarks to the poem. To credit invention here is to acknowledge an important step beyond the House of Fame, for it means we are now dealing with an author who accepts the autonomy of fiction. A more stable equilibrium of readerly and writerly functions has been achieved than in the earlier work. Thus, according to the Invocation, the Narrator-reader has found material that the Narrator-writer will now present.

Besides a source, the poet devises a purpose, indeed, a mission both patriotic and socially responsible: to translate into English an old story that age has already nearly effaced from memory:

For hit ful depe is sonken in my mynde,
With pitous hert in Englyssh to endyte
This olde storie, in Latyn which I fynde,
Of quene Anelida and fals Arcite,
That elde, which that al can frete and bite,

As hit hath freten mony a noble storie,
Hath nygh devoured out of oure memorie.
                                 (8–14)

The preservation of old stories is a theme that will reappear in the Prologue to the Legend of Good Women. Here, the (evidently nonexistent) "nygh"-forgotten ancient material is assumed to be without complications: unambiguous and therefore writable. The skeptical abyss is temporarily healed over, disbelief is for the moment suspended. What is read can be assimilated and in turn produced.

It is with Troilus and Criseyde (1385 or after) that a decisive shift occurs in the earlier proportioning of readerly and writerly functions. The Narrator cannot be entirely divested of his readerly stance, because he is working with an older story, which he claims to have translated. This source is Boccaccio's Filostrato, but Chaucer effaces Boccaccio from his text, substituting the nonexistent "Lollius." Evidently he believed in good faith that there was a Lollius, but the substitution itself was no error: it was a conscious fiction. This takes us a step beyond "Anelida," where a real author (Statius) was used to cover invented material. Here an invented author is used to cover real material and to displace its known author. Hence the treatment of source evinces authorial initiative rather than scholarly receptivity. Why does Chaucer need Lollius at all? I think it is somewhat more complicated than the medieval "respect for authority" such as is generally said to have led Chrétien de Troyes and Wolfram von Eschenbach to invent andent books as sources for their chivalric romances. (Perhaps it was more complicated for them also.) Chaucer has not at this point arrived at a definitive author-narrator distinction; in my view, he never fully achieves this, despite valiant efforts to do so. The closest he can come at this point, I suggest, is an author-translator distinction: a more limited version of the impulse to the autonomy of fiction. To this end he requires a text to translate, an old text in an authoritative language. Boccaccio is simply too close to provide the necessary distancing. As a recent contemporary writing in the vernacular (for Filostrato in any case), Boccaccio was simply neither distant nor, therefore, different enough to justify a credible disjunction between author and translator.^[5]

[5] David Wallace offers another and intriguing possibility: that Chaucer wanted to avoid being compared with Boccaccio, an author whose "poetic evocations ofcortesia actually furthered the debasement of courtly language¼ . Perhaps this is why Chaucer was anxious to publicise his admiration for Petrarch and Dante whilst passing over Boccaccio in silence" (152). This does not contradict my hypothesis, for careerist and poetic motives might well overlap and reinforce one another. In his psychoanalytically oriented article on the same topic ("Chaucerian Authority"), A. C. Spearing also suggests that Boccaccio is "dangerously close, a father rather than a distant ancestor," but he does not fully account for why Chaucer tends to suppress his father-figures.

Despite various references to "myn auctor" and such phrases as "I fynde ek in stories" or "as I rede," the dominant image of the Narrator in Troilus is that of writer. This image is built up in numerous ways. To begin with, there are the long, important, and poetically intense invocations to each of the five books of the poem. In these the Narrator represents himself as the transparent "instrument" (1.10) of his material and single-minded servant of the servants of love (1.15–51); as disinterested translator of his Latin source (2.12–21) and cultural relativist (2.22–42); as bard of universal principles of love (3.1–49); as sympathetic moralist (4.1–14). Beside the invocations, with their acute awareness of the production of literature, there are other lengthy reflections on changing language as a problem for the modern writer (2.22–49 and 5.1793–98). At another level, there is the evolving view of Criseyde, whom we see early on as auditor of the romance of Thebes (2.80–108), but who by the end (5.806–26) has become the iconic figure in a new romance, a new tradition that bears her name and that of Troilus. This placing or distancing can have been accomplished only through the active intervention of a new author.

Finally, and especially, there is the profoundly self-conscious coda to the poem (5.1765–1869), which pulls the text definitively into range of mature authorial concern. Here Chaucer or his Narrator (the distinction is not significant for my purpose at this point) addresses himself to narrative proportion (1765–71), audience response (1772–78), authorial motive and intent (1779–85), future authorial production (1786–88), relation to tradition (1789–92), changing language and the transcription of manuscripts (1793–96), the moral implications of his own work (1828–55), and commendation of the text to friends (1856–62).

The story itself continually emphasizes the importance of discourse, both verbal and written: not simply the reception of discourse, as in the House of Fame, but its production as a psychological and ethical phenomenon. We are required to come to terms with the

frightening power of the word as it emanates from various characters in monologue, dialogue, songs, and letters; in persuasion, promise, threat, praise, self-justification, lament, blasphemy, pun—an immense variety of speech-acts. We are required to consider the relation of discourse to external circumstance and to the utterer's will. Alfred David asserts, in another but related connection, that "in Troilus and Criseyde the narrator at last steps into the pulpit that is the poet's rightful vantage point" (27).^[6] To extend this point, I propose that in Troilus Chaucer is able for the first time definitively to appropriate to his first-person Narrator the active authorial role, grasping the nettle to achieve the mature (although not necessarily unambiguous) voice that we recognize as the voice of the Canterbury Tales.

Although the tales were composed at various times during Chaucer's career, the General Prologue was evidently done in 1387. As a group, the pilgrims are fairly active readers (or, in some cases, hearers) of assorted texts. The up-to-date Man of Law has read the poetry of Chaucer; others read Aristotle, medical texts, romances, misogynistic anecdotes, Breton lais, Petrarch and Ovid, Marie de France's fables, saints' lives and miracles of the Virgin, Roman history, the Bible, treatises of moral edification, and tragedies of the fall of great men. They are also narrators and performers, many of them quite self-conscious about the sources, style, content, and interpretation of their own and each other's recitals. But with the exception of Pardoner and Parson, they are not professional rhetors: the fiction we are asked to accept is that these ordinary folk, amateurs, speak spontaneously out of their experience and their reading.

The Narrator, too, has done his reading, as his tales show: he has read in "bourgeois romance" and in compendia of ethical advice. Nonetheless he is presented primarily as speaker, reporter, and maker. Lest in our fascination with other characters we forget his shaping role, it is emphasized often, in the Narrator's own voice, as organizational interjection (GP 35–42, 715–24), opinion (GP 183, 385, 691, Canon's Yeoman's Prologue VIII.568–73), disclaimer (GP 284,

[6] David's emphasis here, and mine, is in contrast to Howard's in "Chaucer the Man," which stresses that the Narrator of Troilus "is a reader" (39; italics in original).

330), and apologia (GP 725–46, Miller's Tale I.316–86). In this poem foregrounding the production of discourse, the Narrator is constantly kept before us as the primary producer, the filter through whom all the others are apprehended. What we find, then, is a reversal of the proportions I have noticed in the Book of the Duchess: not a complete or exclusive reversal, but a general shift. There the first-person Narrator functions within the fiction primarily as reader, with discursive activity displaced onto another fictional character, the Black Knight. Here the first-person Narrator functions within the fiction primarily as producer of discourse, with readerly activity displaced onto the fictional characters, the pilgrims.

This development in the assertion of authorial function might interestingly be approached from the standpoint of naming, for a parallel evolution can be traced through the poet's career. In the Book of the Duchess, the Narrator is nameless, and likewise in the Parliament, even though there exists in the latter text an interlocutor—Africanus—whose dominant role and condescending tone might have allowed him to name the Narrator. In the House of Fame, the Narrator is for the first time named—by his interlocutor, the chatty Eagle—and with the poet's real proper name (729). He also expresses a nervous concern that "no wight have my name in honde" (1877). In Troilus, the Narrator is again unnamed: there is no interlocutor who might address him, but then an occasion for naming could have been invented, or a narratorial name included with those of Gower and Strode at the end. In any case, the pronounced authorial self-consciousness in Troilus is not expressed in naming. Nor is it in the Legend —not, at least, directly. There is a fair amount of naming in the Legend: various sources and authors are named throughout, and the naming of Alceste is an important feature of the Prologue: she names herself (F 432, G 422), she is named by Cupid (F 511, G 499), and the Narrator acknowledges her by name (F 518, G 506). Moreover, though the Narrator is not named, his previous works—which also happen to be the works we recognize as Chaucer's—explicitly and profusely are both named and evaluated. What seems to happen is that personal self-referentiality is displaced onto Alceste, while the poet himself exists as maker of his works and as object of a critical discourse. Whatever his personal fate, the poet seems to realize that he will live as the author of works

that, unlike so much medieval literature before (and even after) him, will not be untitled or anonymous; in short, his name (albeit here suppressed) is going to have a function—the one, I suggest, that Michel Foucault has called an "author-function."

I do not want to claim for Chaucer the fullest sense of this Foucauldian term, for it is unlikely that any poet or belletrist can found a discursive practice that will be "embodied in technical processes, in institutions, in patterns for general behavior, in forms for transmission and diffusion, and in pedagogical forms which, at once, impose and maintain them" (Language, 200). What I do want to get at, though, is the process we know Chaucer witnessed already in his life, a process that the Prologue to the Legend fictionally represents, namely, the hardening of himself and his work into a figure, something beyond the merely personal or autobiographical. As Howard observes, "Chaucer was recognized as a major poet in his own lifetime. He was praised over and over as a 'philosophical' poet, a great rhetorician and translator; he was imitated by lesser poets like Usk and Hoccleve" (Chaucer, 524). In this light, it is legitimate to infer a developing attitude toward his own work. "The comparisons we make, the traits we extract as pertinent, the continuities we assign, or the exclusions we practise" (Foucault, Language, 127): these aspects of the author-function are surely recognizable as what Alceste and Eros do to Geffrey and his writing, and it is an experience at once exhilarating and frightening.

The effect, I believe, is that the poet is becoming aware of producing not only works but texts, a discourse in a receptive field (that is, a field of divergent receptions). It is a discourse because it is a body of work with a recognizable style and subject matter and an ideological position. It can generate both imitators and opponents—sometimes both in the same person, as in the case of Osbern Bokenham. The maker, I suggest, is concerned about the status of his discourse in much the same way that Criseyde sees herself becoming frozen into a tradition "rolled ¼ on many a tongue," hated especially by women for dishonoring them (5.1054–68). It is precisely the charge laid against the poet-Narrator of the Legend. To his readers—the two of them represented in the Legend—he protests, as it were, that it is more complicated than they think. He proceeds to show them how complicated it is by writing yet another book, this one about writing books.

It is as if that act frees the poet (relatively) from this obsessive concern with his role to produce a work that looks at the world rather than at its own feet: the Canterbury Tales. There, the bumbling, scholarly, versifying "Geffrey" no longer exists. Instead, we have "Chaucer" (Man of Law's Prologue II.47)—the formal, dignified patronymic that will survive through history and set the poet apart from any other Geoffrey. (It is not irrelevant to recall in this connection that when Mary Serjeantson edited Bokenham's legendary for EETS, she mistakenly identified "Galfryd of ynglond" as Geoffrey Chaucer rather than as Geoffroi de Vinsauf. So far from being subsumed by any other Geoffrey, Chaucer has come to subsume all the others!) In the Man of Law's recital, Chaucer's works are named and discussed, and by a far more sympathetic reader than Alceste. This naming, in context of a multi-class pilgrimage, suggests a selfconfident awareness on the real author's part that such reference would not be completely inappropriate on such an occasion: that his name might be dropped by a lawyer, and that it might be recognized by at least some of his companions.

We may like to assume that this named Chaucer is identical with the "I" who narrates the entire set of stories and who, within it, tells the tales of Melibee and Thopas. This was the scribal assumption (and it is worth remembering that all extant manuscripts of the tales postdate the poet's death). It is duplicated in incipits and explicits that make their way into editions, it is taken for granted by every scholar, and it makes for delicious ironies; but there is no actual evidence within the poetic text for such an identification. The Narrator evinces no response either to the Man of Law's praise or to the Host's denunciation of his tale that might suggest real authorial status at stake. As far as textual evidence goes, the Chauceriannairatorial "'je' est un autre" (as Rimbaud put it), and the absence of indication of originary authorial intent permits the speculation that we may see the Narrator as not-Chaucer. Strictly speaking, there is no "Chaucer the pilgrim" on the pilgrimage. Once again we have an unnamed Narrator. He is distanced, as far as present evidence lets us judge, in a number of ways from the poet designated "Chaucer," among them his (the Narrator's) apparent poetic ineptitude. If the first-person Narrators of the earliest works were nameless because a lack of poetic security prevented the maker from naming himself, the first-person Narrator of the Tales is name-

less because the maker's name is borne by someone else, someone outside the fiction who is already famous. David Lawton rightly emphasizes the rhetorical, not characterological, function of Chaucer's narrators, writing of "not one device but a scale, a register, of different ones" ^[7] . My own image for the relation of author to narrator is that of the U.S. and Canadian dollars: sometimes so close in value as to be exchangeable at par, at other times very far out of phase. It is because of this flexibility, this shifting signifier that is the narratorial persona, that virtually everyone who writes about Chaucer's narrative voice is right—at least in places.