The Problem: Voice, Textuality, Impersonation

In The Idea of the Canterbury Tales Donald R. Howard identified a perennial strain in the Chaucer criticism of the last thirty years or more—isolated it, defined it clearly, and gave it a name. Discussing the Knight's Tale , he remarks:

Chaucer . . . introduced a jocular and exaggerated element that seems to call the Knight's convictions into question. For example, while the two heroes are fighting he says "in this wise I let hem fighting dwelle" and turns his attention to Theseus:

The destinee, ministre general,
That executeth in the world over all
The purveiaunce that God hath seen biforn,
So strong it is that, though the world had sworn
The contrary of a thing by ye or nay,
Yet sometime it shall fallen on a day
That falleth nat eft within a thousand yeer.
For certainly, our appetites here,
Be it of wer, or pees, or hate, or love,
All is this ruled by the sight above.

This mene I now by mighty Theseus,
That for to hunten is so desirous,
And namely at the grete hert in May,
That in his bed there daweth him no day
That he nis clad, and redy for to ride
With hunt and horn and houndes him beside.
For in his hunting hath he swich delit
That it is all his joy and appetit
To been himself the grete hertes bane.
For after Mars he serveth now Diane.
      (1663–82)

All this machinery is intended to let us know that on a certain day Theseus took it in mind to go hunting. It is impossible not to see a mock-epic quality in such a passage, and hard not to conclude that its purpose is ironic, that it is meant to put us at a distance from the Knight's grandiose ideas of destiny and make us think about them. This humorous element in the Knight's Tale is the most controversial aspect of the tale: where one critic writes it off as an "antidote" to tragedy another puts it at the center of things, but no one denies it is there. It introduces a feature which we will experience in many a tale: we read the tale as a dramatic monologue spoken by its teller but understand that some of Chaucer's attitudes spill into it. This feature gives the tale an artistry which we cannot realistically attribute to the teller: I am going to call this unimpersonated artistry . In its simplest form it is the contingency that a tale not memorized but told impromptu is in verse. The artistry is the author's, though selected features of the pilgrim's dialect, argot, or manner may still be impersonated. In its more subtle uses it allows a gross or "low" character to use language, rhetoric, or wit above his capabilities. (Sometimes it is coupled with an impersonated lack of art, an artlessness or gaucherie which causes a character to tell a bad tale, as in Sir Thopas , or to violate literary conventions or proprieties, as in the Knight's Tale.) The effect is that of irony or parody, but this effect is Chaucer's accomplishment, not an impersonated skill for which the pilgrim who tells the tale deserves any compliments.
           (Howard, Idea of the Canterbury Tales , 230–31)

Having generated this principle, Howard goes on to apply it, at various points in the book, to the tales of the Miller, the Summoner, the Merchant, the Squire, and the Manciple. He is in good and numerous company. One thinks of Charles Muscatine's characterization of certain central monologues in Troilus and Criseyde : "The speeches must be taken as impersonal comments on the action, Chaucer's formula-

tion, not his characters'"; of Robert M. Jordan, who, having presented an impressive array of evidence for a complicated Merchant in the Merchant's Tale , argues from it, like Dryden's Panther, "that he's not there at all"; of Anne Middleton's exemption of selected passages of the Physician's Tale from the pilgrim's voicing; of Robert B. Burlin's praise of the Summoner's Tale despite its being "beyond the genius of the Summoner"; and of many other commentators on the Knight's Tale , some of whom I will return to later.^[1]

In my view this "unimpersonated artistry" is a problem, and a useful one. Howard's formulation—an attempt to describe an aspect of Chaucer's general practice—is valuable because it brings into the sharp relief of a critical and theoretical principle something that is more diffusely present in the practical criticism of a great many Chaucerians: the conviction, often unspoken, that at some point it becomes necessary to move beyond or away from the pilgrim narrators of the Canterbury Tales and to identify the poet himself as the source of meaning. If the assumption is stated this broadly, I probably agree with it myself; but Howard's way of putting it does seem to me to reflect a tendency, common among Chaucer critics, to invoke the poet's authority much too quickly. Howard helps me to focus my own discontent, not with his criticism (much of which I admire), but with a more general situation in the profession at large. If we consider "unimpersonated artistry" as a theoretical proposition, it seems open to question on both general and specific grounds; that is, it seems both to imply a rather peculiar set of assumptions to bring to the reading of any text and, at least to me, to be an inaccurate reflection of the experience of reading Chaucer in particular.

[1] Muscatine, Chaucer and the French Tradition , 264–65. In a more general statement on the Canterbury Tales , Muscatine observes: "No medieval poet would have sacrificed all the rich technical means at his disposal merely to make a story sound as if such and such a character were actually telling it. The Miller's Tale , to name but one of many, would have been thus impossible" (171–72). See also Lawlor, Chaucer , chap. 5, and Jordan, Shape of Creation . Jordan is, on theoretical and historical grounds, the most thoroughgoing and principled opponent of the notion of consistent impersonation in Chaucer's work. In this connection his book deserves to be read in its entirety; see also Jordan, "Chaucer's Sense of Illusion." Middleton,"Physician's Tale and Love's Martyrs." Burlin, Chaucerian Fiction , 165. Elizabeth Salter's reading of the Knight's Tale , in her introduction to Chaucer, The Knight's Tale and the Clerk's Tale , is perhaps the most consistently developed in terms of the "two voices" of the poet. See also Thurston, Artistic Ambivalence .

"Unimpersonated artistry" implies a technique, or perhaps an experience, of reading something like this: we assume that the Canterbury Tales are, as they say, "fitted to their tellers," that they are potentially dramatic monologues, or, to adopt what I hope is a less loaded term, that they are instances of impersonated artistry, the utterances of particular pilgrims. After all, we like to read Chaucer this way, to point out the suitability of the tales to their fictional tellers, and most of us, even Robert Jordan, would agree that at least some of the tales, and certainly the Canterbury frame, encourage this sort of interpretation.^[2] We read along, then, with this assumption in mind until it seems to break down, until we come across a passage that we have difficulty reconciling with the sensibility—the temperament, training, or intelligence—of the pilgrim in question. At that point, alas, I think we too often give up. This passage, we say, must be the work of Chaucer the poet, speaking over the head or from behind the mask of the Knight or the Miller or the Physician, creating ironies, setting us straight on doctrine, pointing us "the righte weye." Unfortunately, these occasions are seldom as unequivocal as the one case of genuine broken impersonation I know of in the tales, the general narrator's "quod she" in the middle of a stanza of the Prioress's Tale (VII, 1771). Different critics find the poet in different passages of the same tale, and they often have great difficulty in deciphering his message once they have found him—a difficulty that seems odd if Chaucer thought the message worth disrupting the fiction.

Thus Howard, whose observations on the critical disagreement over the humorous element in the Knight's Tale are well taken, offers an interpretation of "The destinee, ministre general" that is in fact uncommon. His account of the ironic tone of these lines in context is at least more attentive to the effect of the language than are the numerous readings that take the passage relatively straight. Even within this group, however, the range of proposed answers to the question, Who's talking here? is sufficiently various to raise the issue I am interested in. To mention only those who discuss this particular passage, the work of William Frost, Paul G. Ruggiers, and P. M. Kean is representative of the large body of criticism that remains relatively inattentive to the whole question of voicing in the tale.^[3] They share a

[2] Howard states the position admirably (Idea of the Canterbury Tales , 123–24).

[3] Frost, "Interpretation of Chaucer's Knight's Tale"; Ruggiers, "Some Philosophi-cal Aspects"; Kean, "The Knight's Tale ," in Kean, Chaucer and the Making of English Poetry , 2:1–52.

view of the passage as a piece of the poem's doctrine, to be taken seriously as part of an argument about man's place in the cosmos. Of those who, like Howard, find something odd about the passage, Burlin thinks that the speaker is Chaucer, who intends to suggest by it that Theseus is a man superior to Fortune but unaware of Providence (Chaucerian Fiction , 108), whereas Richard Neuse, the only critic to attribute the speech unequivocally to the Knight, maintains that it differentiates the latter's implicitly Christian view of the story from Theseus's more limited vision.^[4] Who is talking here, and to what end? What are the consequences for interpretation if we concede both that the passage makes gentle fun of the machinery of destiny, at least as applied to so trivial an event, and that it is the Knight himself who is interested in obtaining this effect? Howard's suggestion notwithstanding, the passage is not really directed at Theseus's hunting but at the improbably fortuitous meeting in the glade of Theseus, Palamon, and Arcite described in the lines that immediately follow (I, 1683–1713).^[5] This encounter is one of many features in the first half of the tale that show that most of the plot, far from being the product of portentous cosmic forces (Palamon and Arcite are consistently made to look silly for taking this view), is generated by human actions and choices, not least by those of the narrating Knight in conspicuously rigging events and manipulating coincidences. The Knight, as Neuse points out ("The Knight," 300), is adapting an "olde storie" for the present occasion, and the irony here reflects his opinion of the style of those "olde bookes." To him that style embodies a dangerous evasion of human responsibility for maintaining order in self and society by unconsciously projecting the responsibility onto gods and destinies.

The point is that a notion like unimpersonated artistry, by dividing speakers into parts and denying them the full import of their speaking, puts us in the difficult position of trying to decide which parts of a single narrative are to be assigned to the pilgrim teller and which to the "author"; in these circumstances it is not surprising that different

[4] Neuse, "The Knight," 312–13.

[5] References to Chaucer are from Benson, ed., The Riverside Chaucer , 3d ed.,which is based on Robinson, ed., The Works of Geoffrey Chaucer , 2d ed. References to the Canterbury Tales give fragment number and line numbers.

critics make the cut in different places. All such formulations involve finding or creating two speakers (or even more) in a narrative situation where it would appear simpler to deal with only one.^[6] The procedure seems to me theoretically questionable because it is unparsimonious or inelegant logically: it creates extra work and it leads to distraction. Narrative entities are multiplied to the point where they become subjects of concern in their own right and require some sort of systematic or historical justification such as unimpersonated artistry or the deficiency of medieval ideas of personality;^[7] before long we are so busy trying to save the appearances of the epicyclic constructs we ourselves have created that we are no longer attending to the poems that the constructs were originally intended to explain. Therefore I would like to preface my more detailed opposition to unimpersonated artistry with a general caveat. I call it Leicester's razor: narratores non sunt multiplicandi sine absolute necessitate .

Naturally I do not intend to let the matter rest with this general and essentially negative formula, though I think its application clears up a lot of difficulties. I want to use the space my principle gives me to argue that the Canterbury tales are individually voiced, and radically so— that each of the tales is primarily (in the sense of "first," that is, the place where one starts) an expression of its teller's personality and outlook as embodied in the unfolding "now" of the telling. I am aware

[6] See, for example, Jordan, Shape of Creation , 181, where what is apparently envisioned is Chaucer the poet projecting Chaucer the pilgrim as the (intermittent?) narrator of the Knight's story. For an instance of how far this sort of thing can go, see Campbell, "Chaucer's 'Retraction.'"

[7] Jordan once again provides the clearest example of this historicist form of argument, but D. W. Robertson also uses it, as in the introduction to his Chaucer's London , 1–11, where he both specifies and generalizes statements he made previously in his Preface to Chaucer such as the following: "The actions of Duke Theseus in the Knight's Tale are thus, like the actions of the figures we see in the visual arts of the fourteenth century, symbolic actions. They are directed toward the establishment and maintenance of those traditional hierarchies which were dear to the medieval mind. They have nothing to do with 'psychology' or with 'character' in the modern sense, but are instead functions of attributes which are, in this instance, inherited from the traditions of medieval humanistic culture" (265–66; see also the discussion of the Friar's Tale that immediately follows). This whole line of argument probably originated with Leo Spitzer's "Note on the Poetic and the Empirical 'I' in Medieval Authors." Spitzer's argument is drawn from particular textual investigations and is relatively tentative about its conclusions. Judging from his remarks on Boccaccio, I am not at all sure that Spitzer would see Chaucer as a representative user of the "poetic 'I,'" but in any case I think his successors, unlike him, are arguing from "history" to texts, not the other way around. Spitzer's formulation has become fossilized in those Chaucerians.

that something like this idea is all too familiar. Going back, in modern times, at least as far as Kittredge's characterization of the Canterbury Tales as a human comedy, with the pilgrims as dramatis personae, it reaches its high point in Lumiansky's Of Sondry Folk (and apparently its dead end as well: no one since has attempted to apply the concept systematically to the entire poem).^[8] Moreover, as I said before, we are all given to this sort of reading now and then. I think one reason the idea has never been pushed as hard or as far as I would like to take it is that the voicing of individual tales has almost always been interpreted on the basis of something external to them, usually either some aspect of the historical background of the poems (what we know from other sources about knights, millers, lawyers, nuns) or the descriptions of the speakers given in the Canterbury frame, especially in the General Prologue . Such materials are combined in various ways to construct an image of a given pilgrim outside his or her tale, and each tale is then read as a product of the figure who tells it, a product whose interpretation is constrained by the limitations we conceive the pilgrim to have.

The question of historical presuppositions, the feeling that medieval men and women could not have thought or spoken in certain ways, I will address in a moment; but the problem of the Canterbury frame has been the more immediate obstacle to reading the tales as examples of impersonated artistry. Since I do not mean by that phrase what either the critics or the defenders of similar notions appear to have meant in the past, the topic is worth pausing over. The issue is generally joined concerning the question of verisimilitude, the consistency with which the fiction of the tales sustains a dramatic illusion of real people taking part in real and present interaction with one another. The critic who has most consistently taken this dramatic view of the poem is Lumiansky, who locates both the "reality" of the pilgrims and the "drama" of their relations with one another outside the tales themselves, preeminently in the frame. He ordinarily begins his discussion of a given tale and its dramatic context with a character sketch of the pilgrim drawn from the General Prologue (and from any relevant links) and then treats the tale itself as an exemplification and extension of the traits and situations in the frame. He is attentive to such details as direct

[8] Kittredge, Chaucer and His Poetry , chap. 5; the famous phrasesare on 154–55.

addresses to the pilgrim audience within a tale (such as the Knight's "lat se now who shal the soper wynne," I, 891) and, to a degree, to the ways tales respond to one another, as in fragment III or the marriage group. This approach leads to an account of the poem as a whole that doubles the overt narrative of the frame and, in effect, allows the frame to tyrannize the individual tales: what does not fit the model of actual, preexisting pilgrims really present to one another is not relevant to the enterprise and is variously ignored or dismissed. Other critics have not been slow to point out that this procedure neglects a great deal.^[9]

The objection to this "dramatic" model that I would particularly like to single out is its disregard for the poem's constant and intermittently insistent textuality , for the way the work repeatedly breaks the fiction of spoken discourse and the illusion of the frame to call attention to itself as a written thing. Such interruptions as the injunction in the Miller's Prologue to "Turne over the leef and chese another tale" (I, 3177) and the moment in the Knight's Tale when the supposedly oral narrator remarks, "But of that storie list me nat to write" (I, 1201) not only destroy verisimilitude but draw attention to what Howard has called the bookness of the poem (Idea of the Canterbury Tales , 63–67), as do, less vibrantly, incipits and explicits, the patently incomplete state of the text, or "the contingency that a tale not memorized but told impromptu is in verse" (Idea of the Canterbury Tales , 231). Now this conspicuous textuality (by which I mean that Chaucer not only produces written texts but does so self-consciously and calls attention to his writing) certainly militates strongly against the illusion of drama as living presence. It is no doubt this realization, coupled with the counterperception that some tales do seem "fitted to the teller," that has led Howard and others to adopt formulations like unimpersonated artistry in order to stay responsive to the apparent range of the poem's effects. Such a notion allows the critic to hover between bookness (more commonly called writing nowadays), which always implies absence , and the logocentrism that Howard calls voiceness —the fiction of presence we feel when "the author addresses us directly and himself rehearses tales told aloud by others: we seem to hear his and the pilgrims' voices, we presume oral delivery" (Idea of

[9] Jordan is particularly good at evoking the element of "the girlhood of Shakespeare's heroines" that often finds its way into this sort of interpretation; see "Chaucer's Sense of Illusion," especially 24–26.

the Canterbury Tales , 66).If we cannot have presence fully, we can at least have it partly. But when and where exactly, and, above all, whose? As I have tried to suggest, a notion like unimpersonated artistry—which is an intermittent phenomenon—tries to save the feeling that someone is present at the cost of rendering us permanently uncertain about who is speaking at any given moment in (or of) the text: the pilgrim, the poet, or that interesting mediate entity Chaucer the pilgrim.

It seems to me that the roadside-drama approach, the criticisms of that approach, and compromise positions (whether explicitly worked out like Howard's or more intuitive) have in common a central confusion: the confusion of voice with presence .^[10] All these views demand that the voice in a text be traceable to a person behind the language, an individual controlling and limiting, and thereby guaranteeing, the meaning of what is expressed. The language of a given tale, or indeed of a given moment in a tale, is thus the end point of that person's activity, the point at which he or she delivers a self that existed prior to the text. For this reason all these approaches keep circling back to the ambiguous traces of such an external self—in the frame, in the poet, in the facts of history, or in the "medieval mind." But what I mean by impersonated artistry—and indeed what I mean by voice—does not necessarily involve an external self.

In maintaining that the Canterbury Tales is a collection of individually voiced texts, I want rather to begin with the fact of their textuality, to insist that there is nobody there, that there is only the text. But if a written text implies and enforces the absence of the self, the real living person outside the text who may or may not have expressed himself or herself in producing it, the same absence is emphatically not true of the voice in the text, which I might also call the voice (or subject) of the text. In writing, voice is first of all a function not of persons but of language, of the linguistic codes and conventions that make it possible for an "I" to appear.^[11] But this possibility has interpretive conse-

[10] In what follows I ought to acknowledge a general obligation to the work of Jacques Derrida, perhaps more to its spirit than to any specific essay or formulation. For a representative discussion of the problem of presence and a typical critique of logocentric metaphysics, see "Writing Before the Letter," part 1 of Derrida, Of Grammatology , 1–93.

[11] See Benveniste, Problems in General Linguistics . Chapters 18 and 20 are especially helpful, but the whole section (chaps. 18–23) is of value.

quences that are not always noted since it means that we can assign an "I" to any statement. Because language is positional it is inherently dramatic: it always states or implies a first person, the grammatical subject, not only in grammatical relation but also in potential dramatic relation to the other grammatical persons, and it does so structurally—as language—and regardless of the presence or absence of any actual speaking person. Thus any text, by its nature as a linguistic phenomenon, generates its own set of rhetorical inflections of the grammatical subject , what in literary texts is often called its speaker. The speaker is a subject created by the text itself as a structure of linguistic and semantic relationships, and the character, or subjectivity, of the speaker is a function of the specific deployment of those relationships in a particular case to produce the voice of the text. This kind of voiceness is a property of any text, and it is therefore theoretically possible to read any text in a way that elicits its particular voice, its individual first-person subject. Such a reading would, for example, try to attend consistently to the "I" of the text, expressed or implied, and would make the referential aspects of the discourse functions of the "I." To put it another way, a voice-oriented reading would treat the second and third persons of a discourse (respectively the audience and the world), expressed or implied, primarily as indications of what the speaker maintains about audience and world and would examine how these elements are reflexively constituted as evidence of the character of this particular subject. It would ask what sort of first person notices these details rather than others and what sort conceives of an audience in such a way that he or she addresses it in this particular tone, and so forth.

Although any text can be read in a way that elicits its voice, some texts actively engage the phenomenon of voice, exploit it, make it the center of their discourse—in fact, make it their content. This sort of text is about its speaker, and I contend that the Canterbury Tales , especially the individual tales, is such a text. The tales are examples of impersonated artistry because they concentrate not on the way preexisting persons create language but on the way language creates people.^[12] They detail how a fictional teller's text im-personates him or her

[12] A number of tales, including the Prioress's and the Shipman's, suggest how this happens regardless of whether the speaker intends it.

by creating a personality, that is, a textual subject that acts like, rather than is, a person. What this textual impersonation implies for the concrete interpretation of the poem is that the relation I have been questioning between the tales and the frame, or between the tales and their historical or social background, needs to be reversed. The voicing of any tale—the personality of any pilgrim—is not given in advance by the prologue portrait or the facts of history, nor is it dependent on them. The personality has to be worked out by analyzing and defining the voice created by each tale. It is this personality in the foreground, in his or her intensive and detailed textual life, that supplies a guide to the weighting of details and emphasis, the interpretation , of the background, whether portrait or history. To say, for example, that the Miller's Tale is not fitted to its teller because it is "too good" for him, because a miller would not be educated enough or intelligent enough to produce it, is to move in exactly the wrong direction. In fact, it is just this sort of social typing that irritates and troubles the Miller himself, especially since both the Host and the general narrator typed him long before any Chaucer critic did (I, 3128–31, 3167–69, 3182). The characters in his tale repeatedly indulge in social typing, as does the Miller.^[13] The Miller's handling of this practice makes it an issue in the tale, something he has opinions and feelings about. The end of the tale demonstrates how the maimed, uncomfortably sympathetic carpenter is sacrificed to the mirth of the townsfolk and the pilgrims; he is shouted down by the class solidarity of Nicholas's brethren: "For every clerk anonright heeld with oother" (I, 3847). One could go on to show how the representation of the Miller's sensibility in the tale retrospectively and decisively revises the portrait of him in the General Prologue into something quite different from what it appears to be in prospect, but the same point can be suggested more economically concerning the Physician. When we read in the General Prologue that "His studie was but litel on the Bible" (I, 438), the line sounds condemnatory in an absolute, moral way. Reconsidered from the perspective of the tale, however, the detail takes on a new and more intensive individual life in the light of the Physician's singularly inept use of the

[13] For example, "A clerk hadde litherly biset his whyle,/But if he koude a carpenter bigyle"; "What! thynk on God, as we doon, men that swynke"; "She was a prymerole, a piggesnye,/For any lord to leggen in his bedde,/Or yet for any good yeman to wedde" (I, 3299–300, 3491, 3268–70).

exemplum of Jephthah's daughter (VI, 238–50). Retrospectively the poet's comment characterizes a man of irreproachable, if conventional, morality, whose profession channels his reading into medical texts rather than sacred ones and who uses such biblical knowledge as he has for pathetic effect at the expense of narrative consistency: he forgets, or at any rate suppresses, that Jephthah's daughter asked for time to bewail her virginity, whereas Virginia is being killed to preserve hers. The situation in the tale is a good deal more complex than this, but I think the general point is clear enough: it is the tale that specifies the portrait, not the other way around.^[14]

The technique of impersonation as I am considering it here has no necessary connection with the question of the integration of a given tale in the Canterbury frame. The Knight's mention of writing in his tale is indeed an anomalous detail in the context of the pilgrimage. It is often regarded as a sign of the incomplete revision of the (hypothetical) "Palamon and Arcite," supposedly written before Chaucer had the idea of the Canterbury Tales and afterward inserted in its present position in fragment I. The reference to writing is taken as evidence that the Knight was not the speaker "originally" and, in a reading like Howard's, that sometimes he still isn't.^[15] As far as it goes, the argument about the chronology of composition may well be valid, but it has nothing to do with whether the tale is impersonated, a question that can and should be separated, at least initially, from the fiction of the pilgrimage. Details like the Knight's mention of writing are not immediately relevant because they do not affect the intention to create a speaker (they may become relevant at a different level of analysis later). Impersonation, the controlled use of voicing to direct us to what a narrative tells us about its narrator, precedes dramatization of the Canterbury sort in Chaucer, analytically and no doubt sometimes chronologically. The proper method is to ascribe the entire narration, in all its details, to a single speaker (on the authority of Leicester's razor) and use it as evidence in constructing that speaker's subjectivity, keeping the question of the speaker's "identity" open until the analysis is complete. It is convenient and harmless to accept the framing fiction that the Knight's Tale is the tale the Knight tells, as long as we recog-

[14] See Kempton, "Physician's Tale ."

[15] See David, Strumpet Muse , 77–89.

nize that it merely gives us something to call the speaker and tells us nothing reliable about him in advance.

That is the method of reading that will be employed in this book. The first of its leading notions is that the tales must be treated not as the performances of preexisting selves but as texts. They are not written to be spoken, like a play, but written to be read as if they were spoken. They are literary imitations of oral performance, in which the medium, writing, makes all the difference.^[16] Second, the texts of the tales are to be read with a view to analyzing the individual subject, the voice-of-the-text, that each tale constructs. I will adopt as my central hypothesis the assumption that the speaker of each tale—the pilgrim who tells it—is in both senses the subject of all of its details. I will focus on the world of a given tale as evidence for a characterization of its speaker and examine how that speaker's telling creates him or her in the course of the narration. To find out who each pilgrim "is" is the end point or goal, not the beginning, of the investigation. It can only be approached by looking at what the pilgrim does in telling. In line with these assumptions I will generally refrain from drawing on portraits of the individual pilgrims in the General Prologue for advance characterizations of them, though I will occasionally make use of information in the portraits if a given tale seems to authorize it. I will also refrain from drawing on ideas of history and culture taken from outside the poems as a way of preconceiving (or limiting) either the characters of the pilgrims or the nature of their society. This constraint does not mean, however, that I will ignore historical and cultural materials as they are represented in the text. In fact, I am concerned to suggest not only that the pilgrims are the products of their tales, rather than the producers, but also that Chaucer's fiction may explain, rather than be explained by, the "facts" (or, better, the institutions) of the fourteenth century and its social history.

[16] Although my assumptions about the relations between literary and oral cultures in Chaucer's poetry start from a position nearly opposite to that of Bertrand H. Bronson (see especially his In Search of Chaucer , 25–33), I agree with him that the problem of performance in Chaucer is worth further study; in fact, I think it is a central theme throughout the poet's career. In the Canterbury Tales the frame exists precisely to provide a literary representation of the ordinarily extratextual and tacit dimensions of storytelling in writing. The poem presents not merely stories but stories told to an audience that is part of the fiction, and this circumstance allows Chaucer to register the effects of a range of performance conditions.