University of Virginia Library

Text as Matter, Concept, and Action
by
Peter L. Shillingsburg

Textual criticism and scholarly editing do not occupy conspicuous positions on the cutting edge of literary theory. This is because theory and practice in these disciplines have seemed largely unaffected by several fundamental propositions underlying modern literary theory, and indeed, scientific theory and philosophy, as well. Consequently, textual criticism—the science or art of detecting and removing textual error, the discipline of establishing what the author wrote or final authorial intention, the work of purifying and preserving our cultural heritage—textual criticism, I say, has appeared to occupy an intellectual backwater concerning itself with goals and a methodology challenged or abandoned by modern communication theory, principles of relativity, and concepts about the nature of knowledge. If, to the traditionalists, modern literary theory seems to have lost its moorings in reality, to the literary theorists the textual critics seem moored to a chimera.

I propose to entertain three fundamental propositions underlying recent challenges to old certainties in relation to the materials, goals, and methods of textual criticism to see whether, taken seriously, they would effect a revolution, or totally marginalize, or simply reify textual critical theory and practice. Although I can be only referential and suggestive in what I say about fields other than textual and literary criticism, I think that excursions into related fields is a way of raising a series of questions, to a few of which I want to contribute potential answers.

The first fundamental proposition of modern theories relating to factual, historical, and scientific knowledge is that objectivity is a chimera and that statements about facts, history, and truth are relative—not actually "knowable"—because of the gap in perception between object and subject (an inability to verify correspondence between mental constructs and "real" objects). This is not a new idea, of course. The second proposition is the structuralist notion that language provides the vehicle

and imposes the limits for mental constructs of "reality"; therefore, recent investigations of the nature of "facts," "history," and "truth" have been focused on the structuring effect of language. The relevance of these fundamental propositions to any form of speech or writing is quite obvious. In communication, whether or not the listener/reader receives into understanding precisely what the speaker/writer sent from intention is problematical—not ascertainable, not verifiable. The third proposition is, then, that the reader, listener, or perceiver is the most important, or some might say the only important, functional authority for meaning or understanding. That is, it is impossible to conceive of a work of art apart from a perceiver's perception of it. Moby-Dick, for example, as it "exists" between the covers of a closed book has no functional existence as a work of art, remaining potential until someone reads it. These three propositions are, I think, closely related; they may even be said to entail one another. The reading, which "creates" the functional existence of the work, is subject to the perception gap and determined by the structuring nature of language, as was the writing which created the "potential" existence of the work.[1]

Now the question I wish ultimately to tease out is how these propositions, if taken seriously, would affect specific ideas about the materials, methods, and goals of scholarly editing and scholarly reading, how they would affect the making and using of scholarly editions. A few years ago I asked two or three colleagues who claimed to have no expertise in bibliography but who were "up" on literary theory, as I was not, "What difference does it make to a deconstructive reading what text the critic starts with?" They either did not understand the question or found it irrelevant, and from some points of view they were right, for deconstruction is a means of seeing how meanings are generated from any text, not a means of detecting the "intent" of a specific text. But what follows does attempt an answer to the question. Therefore, I begin with a survey of some ways in which the principles of relativity, structuralism, and reading have affected the practice and theory of literary criticism. If textual criticism and scholarly editing are to provide texts and insights that are valuable to literary criticism, they must be conducted in the light of what literary critics find valuable to do. It seems to me that a great deal of the textual criticism of the past twenty years has been conducted in the light of literary critical practices of the 1930s to early '60s. I begin

then with an attempt to characterize some of the fundamental ideas of modern literary theory, though that field is such a seething sea of conflict that no summary can be adequate. Crucial differences in the basic assumptions literary critics and textual critics hold about texts, however, will emerge.

I. Literary Theory and the Work

In literary criticism the indeterminacy of meaning was long seen as the problem; and, until acceptance of relativity changed the aim of criticism, it was the goal of historical criticism to develop means to interpret texts so that they would be understood as they were intended. When structural linguistics began affecting the practice of literary criticism, faith in recovering intended meanings through strenuous biographical, historical and philological study began to erode. The word no longer could be used as a stable semantic unit or as access to "reality," since the word (signifier and its component phonemes and morphemes) bore an arbitrary, not "natural," relation to the signified, which was itself a concept, not the "object in reality." Attempts to link the phonemic elements of language to physiology or neurology so as to demonstrate that they were not entirely arbitrary have not proven very fruitful. Structural relations between words (syntax) was seen to govern their meaning, and the structuring aspect of language governed what could be meant. The Author as Authority for meaning lost ground to "the text itself."

Rumblings about intentional and affective fallacies focused attention on the text's meaning as opposed to the author's meaning, for the latter was both inaccessible and perhaps subverted by a failure to achieve that which was intended, though the text might well witness the success of other perhaps unintended meanings.[2] Literary theorists abandoned the author by defining texts as acquiring "determinate meaning through the interactions of the words without the intervention of an authorial will."[3] The phrase or sentence replaced the word as the irreducible semantic unit, and the intentional fallacy became something to avoid or disguise carefully. However, faith in syntax as a reliable semantic unit continued, as is evident in Beardsley's definition, and "the text itself" seemed a stable and concrete object amenable to disciplined analysis, though authorial intention seemed remote and problematic. The idea that meaning is created by the relations between the words and by the perceived choices among words that could, grammatically, have been used instead

made texts seem even more complex and at the same time apparently more able to communicate successfully. Bakhtin's ideas about a dialogical interaction between/within texts and Claude Levi-Strauss's concept of "bundles" which included societal and behavioral elements in the "sets" which defined the choices by which differentiations in structural relations were identified and understood can be seen as means by which communication can work more effectively (by narrowing the appropriate range of possible meanings) or to increase the amount of slippage (by multiplying our awareness of oblique references—i.e., we become more aware of the potential counters not chosen). In either case, the need for an author is diminished.

With post-structuralism, and particularly with focus being placed on the creative act of reading, came a second wave of reaction against the author that exceeded the new critics' distrust of the intentional fallacy. The author was proclaimed dead because meaning was seen as located and created in the readers' interaction with the text, making any meaning the text "has" or "is witness to" functional only in reading acts —the intending acts of authors having receded, so to speak, into the inaccessible past. Thus, scholars' attempts to recreate the moment of authorship were seen as futile; historical criticism had beached itself like a disoriented whale. Furthermore, faith in the semantic stability of syntax became as problematic as the meaning of the word had become with the advent of structuralism.

A strong undercurrent of thought accompanying the wave of post-structuralism and deconstruction indicated that because the past was unknowable and because speech acts, writing acts, and reading and listening acts have (or create) their meaning now in a cultural or social setting fraught with power struggles, hegemonic structures, and political agendas—including very local, perhaps even domestic ones—therefore (A) the meanings we create for a text now matter more than the supposed original or historical meanings and (B) the way in which meanings are generated and the uses to which meanings are put are a more interesting study than are the texts or the authors or the meanings they may originally have tried to produce. Overtly political forms of literary criticism, particularly marxist and feminist criticism, have received a boost of energy from this line of reasoning. Such criticism focuses on the economic power structures at work in diction and syntax, the patriarchal and class assumptions and structures—both linguistic and social—that can be seen imbedded in texts. To some literary theorists it seemed logical to conclude that this line of thought totally set aside the major concerns of textual criticism and scholarly editing as they had been understood and defended traditionally, since the most interesting aspects of texts are not

supposed authorial intentions but, rather, the unintended revelations the text is witness to.[4]

In very recent years new historicism, while profiting from the insights of structuralism and relativity theory, has resurrected some of the interests of historical criticism. The result is radically different from the old historicism, for in developing accounts of the past, the new historicist is very conscious of the absence of any means to validate the correspondence between the past (whatever it was) and the historian's account of it. Further, new historicists are often very much concerned with the structuring influences of language and the political, social, and mythological "realities" it framed. New historicism, though subject to abuses and unconvincing practice, provides several fruitful means of investigation. One is that the understanding of a text derived from even indeterminate historical investigations is often palpably different from "readings" that relate the text only to the present reader's experience. It suggests that the richness and complexity of a text (and of language) is more fully experienced by contrasting the text as a product of a partially known (that is, constructed) past with the text as free-floating in the present or as it seems to have been experienced at significant moments in intermediate times.

Another rationale for new historicism considers that structuralism's undermining of authorial autonomy and post-structural emphasis on the death, absence, or self-subverting of the author has taken the reaction against belief in objectivity about as far as it seems likely to go. Furthermore, by concentrating its efforts on the creative act of reading, deconstructive criticism provides a methodology that concerns itself with only half of the picture. A returned interest in the idea of texts "conveying meaning" from an "originator of discourse" and belief in the possibility of "fiduciary trust" between author and reader has been defended in several ways. First, it is a demonstration of interest in the workings of culture and tradition—an interest that does not necessarily entail belief in the objectivity of their reconstructions nor a nostalgic reactionary hope to "re-establish" or "restore" anything, but a genuine interest in roots and differences and a fascination with the malleability and tensile strength of histories and ideologies. Curiously, this approach considers both "history" and "the present" as current constructs, which are nevertheless useful as a means of exploring the sense of continuity and change

in human feeling and thought. Second, some new historicists attempt to explore the concepts of utterance and discourse as functional links between the verbal text and its social, economic, and material contexts. The aim of the approach is to study behavior rather than to ascertain and pin down definitive meanings or interpretations. It is in this way that new historicism can emphasize the link between the text as a material object and the meanings created from its physical format, which is one of the focuses of this paper.[5]

Deconstruction has exploited the implications of relativity by recognizing the futility of regaining or understanding intention and has focused on the independent life of the text as it is confronted by actual readers (who might, in spite of themselves be trying to conform to the roles of implied readers). To this, deconstruction adds the concept of ideological influences (structured "realities")—mostly subconscious—which make texts self-subverting in ways probably contrary to authorial intention but nevertheless very important to the reader. One should note that this way of putting it suggests that deconstruction "reinstates" authorial intention as something that can and must be inferred in order that it might be "decentered"—i.e. authorial intention is identified but not treated as an authority for meaning.[6] It should further be noted that "decentered" meanings are not dislodged by nothing, but rather by other provisionally centered meanings, each of which must be justified before it can in turn be "decentered." Some marxist and feminist critics have capitalized on these two principles to read texts politically and to value or discard texts according to the ideologies revealed in the "subtexts." As we shall see, however, release from the bondage of an impossible objectivity is not an escape from the physical object, the

book. Nor is it an escape from the consequences of using one edition or copy of the book rather than another.

The full impact of relativity, structuralism, and reader oriented theory has, as I averred, not affected textual criticism, but it has begun to rock the boat. Textual criticism and its "handmaidens," bibliography and paleography, have had a strong positivist tradition, which manifests itself from time to time in phrases like "the calculus of variants" and "definitive editions," and more recently as "determinate meanings" and "social contracts."[7] The fact is, however, that the discipline has made some accommodations to the "truths" of relativism: the concepts of "critical editions" and "eclectic texts" as Fredson Bowers and G. Thomas Tanselle have developed them are such accommodations, diminishing the positivist force of "solid historical research" which supposedly resulted in "established texts that will not have to be edited again." Likewise, recent discussions of multiple texts and problematic texts, which refer specifically to indeterminacy of the words and punctuation constituting the work of art in addition to acknowledging the indeterminacy of the meaning of the text, are accommodations of relativity. One could even say that the recent emphasis on "process texts" and "versioning," which Michael Warren, Donald Reiman, and Paul Eggert to name only a few have undertaken, is an accommodation of structuralism—though to my knowledge none of them has characterized it as such. The result has been a slight shift (to some editors it seems a great shift, perhaps even a sellout) in the aim of textual criticism from considering the text as an established (or establishable) locus of authoritative stability to a concentration on text as process.

What this means in practice is that the editor or critic declares an interest in multiple texts for each work rather than just in the one true or final text. It means authorial revision and production influences on texts are seen as having potential "integrity" as representations of the work at various stages in the process of composition, revision, and production. It has meant, moreover, that for some works the editor posits two or even more texts to be read and studied in tandem. It has also, however, meant that the concept of textual purity has been rescued by making it necessary to "edit" correctly each stage of the process or to make the process visible by some means that distinguishes between the various agents of change and evaluates the changes, not only according to the perceived effect made by the change but according to the "authority" of the agent of change. In short, "process editing" has not embraced deconstruction

as an approach to texts. It retains the idea of the author and of authority, though in the theory and practice of some practitioners, process editing has loosened its grip on the text as sacred icon or as the well-wrought urn.

I would suggest, incidentally, that though a great deal has been published on these subjects, and although they have radically altered the way some post-modernist works are produced, scholarly publishing itself as a technical practice seems to have resisted any influence at all that these ideas about communication might have on the nature of publishing and on the notions that publishers have about what a book is and how it should be printed. Editions continue to be published (and read) as if written works were stable, achievable, objective, tangible substances, though these are the very concepts about "reality" that have been challenged by the propositions with which I began.

II. The Hole at the Center of Theory: Textual and Literary

The weakness of much literary theory and textual criticism is that practice is based on insights which have not had the advantage of a clear taxonomy of texts. Textual critics have not had a clear enough vision of the varieties of viable answers to questions about who has the ultimate authority (or even the "functional authority") over what the text becomes, whether it is possible for a work to have a variety of "correct forms," and the extent to which the editor's decisions about the "authority" of textual variants is a function of "reader response" rather than evidence. Likewise literary critics have not had a clear enough vision of the problematic nature of physical texts and their assumptions about textual stability (e.g., that a work is a text and a text is a book and the book at hand is, therefore, the work itself).[8]

It seems to me from this survey that the "structure of reality of written works" implied by the three propositions with which I began places the writer, the reader, the text, the world, and language in certain relationships and locates the focus of experience of that reality in the reader.

This relationship has been mapped by a number of theorists, some of whom I shall discuss presently, but it seems to me that these maps reveal a gaping hole in our thinking around which swirls a number of vague and sloppily used terms that we pretend cover the situation. The lack of clear, focused thinking on this question can be seen graphically if we locate the physical materials of literary works of art in a center around which we visualize scholarly interest in Works of Art. To the West of this physical center we can place the scholarship of interest in creative acts, authorial intentions and production strategies, biography and history as it impinges on and influences authorial activities. To the East of the physical center we can place the scholarship of interest in reading and understanding, interpretation and appropriation, political and emotive uses of literature. To the North of the physical center we can place the scholarship of interest in language and speech acts, signs and semantics. All three of these segments of our map tend to treat the work of art as mental constructs or meaning units; the physical character of the work is incidental and usually transparent.[9] To the South we can place the scholarship of interest in physical materials: bibliography, book-collecting, and librarianship. Only in this last area do we detect the appearance of special attention on the Material Text, but because traditionally scholars in these fields have made a sharp distinction between the Material and the Text and because they have focused their attention on the Material as object, their work has seemed tangential to the interests of the West, North, and East.

In fact the "Southern" interest has traditionally been looked upon by the others as dull and supportive—we must have libraries and bibliographies—rather than as full-fledged fields of significant interest.[10] Textual criticism has tended to occupy itself with the concerns of the West (intention) and the South (documents), but if it took seriously the propositions underlying relativity and structuralism, it could be in the center of the "structure of reality" depicted in this graph, drawing upon all sides and informing all sides. It would not be self-defensive and apparently narrow-minded or subservient, as it has often appeared, clinging to questionable notions of objectivity and stability.[11]

It might be noted, by the way, that this particular "map" of textual concerns leaves out entirely what might be called the "data world" or that which in ordinary usage language is thought to refer to—the objective referents of language. It is because "knowledge" about that part of the picture has been removed or relativized or made objectively inaccessible by the perception gap and by the notion that knowledge of it is structured by or constructed through language. This "world view" may not be the "true" one, but it is the purpose of this paper to explore its implications to the concept of texts or works as attested by or extant in physical documents.

The specific questions I want now to raise for examination fall within a narrow band at the center of the related and interesting questions implied by this brief survey. I do not wish to be misunderstood as having raised them all or to have attempted answers to any outside that band. I am not, for example, raising any questions about what a particular text means, or what the author or other issuer of the text might have meant

by it, or even what a reader might have understood it to mean. I am supposing that the author and other purveyors of texts do mean something or somethings by them, and I am assuming that texts are understood by readers to mean certain things. The fact of these meanings is important but the meanings themselves are not my concern here. The answers to such questions lie to the West and East of my concerns. The questions I ask have to do with the mental and physical acts and the material results of acts attending the processes of composition, publication, and reception of written texts. And the questions I ask are about what these acts and results can be, not what they should be. Further, I assume that whether the author and reader understand the same thing by a text is not ascertainable. Moreover, I am not asking questions about whether an author's or publisher's "sense" of the work is individual or culturally determined, nor am I asking if the readers' reactions are culturally bound. At the moment I believe that, at least to some extent, and mostly unawares, they are. But I am not aware that any specific opinion about this notion bears significantly on the proposed taxonomy. Nor am I asking how the meanings of author and reader are generated and how they either succeed or go astray. The answers to these questions lie North of my concerns. On the other hand, I am not confining my interest to documents and books as items for bibliographical description or cataloguing for shelving.

The questions I raise are essentially those of textual criticism, but they involve all of these other fields at their margins, for texts—both as physical and mental constructs—lie at the center of any attempt to record or communicate any knowledge.[12] I wish to propose corollaries for two of the propositions that I proposed to entertain for their effects on textual theory: first, the perception gap that holds that our "knowledge" of the "real" world is restricted to our mental, inferred constructs, and, second, the view that language is the structuring tool through which "knowledge" is constructed. The corollaries of these propositions are: first, that the text of a work as found in a document (what I will call the Material Text) is the locus and source of every reader's experience of a written work of art and that regardless of what concepts of works are inferred from the evidence of the Material Text, there is no channel other than inference by which a reader may "reach out" to the mental forms of works as they may have been experienced by authors or other agents and originators of texts. The second corollary is that the mental construct of

the work derived by a reader from the Material Text in the act of reading (what I will call the Reception Text) is the only "thing" that a reader can refer to when making comments about a work.

These two fundamentals—the physical documents and the reading experience of decoding them—are the irreducible core of literary works. Without the reader, the physical documents are inert and inoperative; without the physical documents there is no reading.[13]

For most practical purposes the words "work of literary art," "book," and "text" are thought to be vaguely synonymous. But in fact there is a great deal of confusion about these words; whenever anyone means something specific by them, qualifications become necessary. So we talk about classroom texts, standard texts, established texts, inscriptions, or revised editions; and we add other concepts relative to production economics or reader response theory. It strikes me that even with these qualifications we do not have enough distinct terms for the concepts we use the words "text" and "work" for. Arguments about how to edit works are fueled by our confusions about what are or are not textual corruptions and about what aspects of book production are or are not legitimate "enhancements" of the work. And these confusions and controversies become heated to the extent that one or more parties believe there is a correct or optimum definition of "text" which is a guide to the desired good, correct, standard, or scholarly edition.

It has long seemed to me that the difficulty which we were not handling well was bridging the distance between concepts of works of art that are abstract, ideal, or mental with the material manifestations of or records of these concepts in paper and ink documents and books. One could try to put this in terms familiar to textual critics as an attempt to draw more clearly the relationship between intended texts and achieved texts, but that puts the question too narrowly (and too Westerly on my map). Or one could try to put it in the language of the English philosopher and linguist J. L. Austin as an exploration of the relationship between perlocution, illocution, and locution, but that tends to emphasize the Westerly and Northerly aspect at the expense of the physical center.[14] Most of the work upon the mental and abstract aspects of works

of art is marred by vague or coarse notions of what the material texts are. And most of the work upon the physical materials of works of art has been marred by a parochial focus of attention or adherence to notions about objective reality.

Ferdinand de Saussure did explore the relation between mental concept and physical sound-image in speech, and a good deal of thought has been applied to that relation in linguistics; so what I am proposing to do for literary works is not entirely new. But confusion arises for at least two reasons when applying Saussure's model of speech to written works. First a speech act takes place in the presence of speaker and listener as a single event in time and in a shared space and physical context. Written works do not. Second, written works, contrary to folk tradition, are not stable, singular, verbal texts. They tend to change in "transmission" (to use one of textual criticism's least elegant terms) either by revision, by editorial intervention, or by accident. I will develop the implications of these two differences between speech acts and "write acts" in due course. For the moment, however, I would like to emphasize that the alleged similarity between the two has led many practitioners of literary and textual criticism and linguistics to treat the physicalness of the written text as unitary and unproblematic.

Theorists are, of course, greatly concerned with the complexities and problematics of "intention" and "interpretation," which precede and succeed the text, but the supposedly stable, unproblematic physical signifier between them, the written text, is simply missing from most diagrams of the problem. Paul Hernadi's adaptation of J. L. Austin's speech act theory is one of the most useful and enlightening of such diagrams. (See Chart 2.) He elaborates both ends of the author-work-reader equation and indicates relevant concerns about language as a communication system and its function in the "world as representable by verbal signs," but the center of Hernadi's chart identifies the "Work as verbal construct and locutionary act." As such it is the work of the author and a field of reader response and is described as verbal, not as physical. The paper and ink Work, as a repository of signs for the verbal construct and locutionary act, untethered from its origins does not exist on the chart.[15]

This physical absence (or transparency) is typical of speech act and literary critical formulations of the communication process. See for example Roman Jakobson's model:

What should, perhaps, be the physical text is apparently a straight line. That line, like Dr. Who's Tardis, may look small and ordinary from outside, but it is spacious and complex inside. From the outside, so to speak, written communication looks like spoken communication, but the differences are so startling as to make conclusions about speech seem simply inapplicable to writing. The problems can be easily demonstrated.

I was spring cleaning the family deepfreeze and came to three jars of frozen grape juice. The labels said: "This year's juice." When the person who canned and labelled the juice wrote the label, it was natural and perfectly unambiguous to say "This is this year's juice." Considered as a "speech act" rooted in time and place, the labelling had a "speaker," a "hearer," a place of utterance (the kitchen), a time (the year and moment of placing the juice in the freezer), a richness of social and physical context that identified the relevant "bundle" (Levi-Strauss's term) or molecule (Caldwell's term) that prevented any misunderstanding or sense of inappropriateness or inadequacy in the phrase, "This year's juice." Only when seen as a written message, a "write act," untethered from speaker, from moment and place of utterance, and from designated hearer, do we find it risible, inadequate, or frustrating to imagine this label as capable of signifying something specific at any time it happens to be read.

Another example: I was reading excerpts from some articles that had been photocopied and bound together for student use. One of the sources photocopied was itself a compilation of essays. At one point a cross-reference said: "See p. 33 of this book." When it was first written and printed "this book" was a phrase probably meant to distinguish the compilation from the original works being excerpted ("those books"). Now, in the photocopy for student use, the reference was inadequate and frustrating. The statement "This office will be closed until tomorrow" is perfectly

clear when announced to a waiting crowd, but totally ambiguous when posted on a locked door and read in the early morning. The "bundle" or "molecule" changes with reference to written material in ways never experienced in a speech act. The difference and ambiguity can be consciously exploited—as in the pub sign announcing "free beer all day tomorrow." Thus, an exploration of the relation between mental concepts (signifieds) and physical texts (signifiers) for literary works leads to problems Saussure never discussed (that I know of) and will lead to descriptions of writing and reading acts in ways that clarify some of our disagreements about what they are and how they are. Perhaps it can also defuse some of the vehemence of our disagreements about what and how they ought to be.

III. A Taxonomy of Texts[16]

In 1984 I made an attempt to delineate the gradations of concepts from the ideal to the concrete, which I thought clarified the editorial materials and goals sufficiently so that disagreements among scholarly editors about editorial policies could be understood clearly and not result from vagueness or confusion.[17] Disagreements could thus be resolved or brought to a truce in which the parties at least knew why they disagreed. I have been gratified by the response to this effort from editors who expressed feelings of relief and release from conflicts between what their common sense inclined them to think was a desirable editorial solution and what standard editorial practice and principles seemed to dictate.

Now it seems profitable to raise the question again because the arguments about what constitutes the work of art rages not merely among textual critics, but among literary critics generally. I have found inspiration to continue my 1984 attempt in the writings of Jerome McGann, D. F. McKenzie, Joseph Grigely and James McLaverty.[18] My discussion

will take the form primarily of definitions. The distinctions between concepts, and the relations I will try to show existing between them, are designed first to provide a system for describing the range of materials that are commonly referred to vaguely as books or works of literary art, and second, to provide a ground for discussing the various sorts of acts (often characterized by confusion and conflict) undertaken in response to these materials. My purpose is to enable the conflicts to be focused more clearly on substantive differences of opinion and judgment rather than on confusions about what is being said. Although taxonomies are by definition logocentric and tend to pin down concepts or objects in a conventional way, the result of the taxonomy I propose is to suggest that the drive towards arresting and codifying Works of Art is futile. Instead, it suggests that the work is partially inherent in all "copies" of it. One might say the Work is neither this, nor that, but both and none. The Work is partially in the copy of the work but is not the copy. Works are known through proliferations of texts, not through their refinement or concentration. Nearly all experiences of works are, therefore, partial. This taxonomy helps reveal what parts remain unknown or unexperienced. I have adopted the convention of capitalizing the terms I have appropriated for definition. Since I use some of these terms before I have had a chance to define them, their capitalization is an indication that I will eventually define them.

A. Methods of Classification

It is customary to speak of a Work of literary art, such as Moby-Dick or Dombey and Son, as though such titles designated something definite. That they do not is easily demonstrated by asking, "If the Mona Lisa is in the Louvre in Paris, where is Hamlet?"[19] The term Work is used to classify certain objects, so that we can say "This is a copy of Moby-Dick, but this over here is a copy of Dombey and Son." The term Work and the title Moby-Dick do not refer to a thing, an object, but rather to a class of objects. We can see this by saying, "This is a copy of Moby-Dick, and this, too, is a copy of Moby-Dick." We might try to push the limits of

this insight by defining Work of literary art as that which is implied by and bounded by its physical manifestations.[20] This statement suggests both that a Work can have forms other than that of one of its physical manifestations, and that its potential forms are limited by the forms of its physical manifestations. It suggests, further, that a Work is in important ways both plural and fragmented. These are not simple or comfortable suggestions, and the reaction of some critics and editors is to limit their attention pragmatically to the physical manifestations of works, the book in hand, as if the Book and the Work were coeval and congruent. They are not interested in abstract notions of "intention" or in fragmentary forms of the work, which they would label "pre-utterance forms" or "pre-copy-text forms" or "shavings on the workshop floor." For them, the work is the book in hand. It is simple, it is practical; it is achieved by willfully ignoring certain sets of questions about the work.

But for those who stop to think that not all copies of a work are identical (which is particularly true of well-known, often reprinted, works), and that what person X says is the work (because he holds copy X in his hand) is different from what person Y says is the work (because she holds copy Y), there is a problem worth resolving, because what person X says about the work, referring to copy X, might be nonsense to person Y, checking the references in copy Y.[21] A fruitful approach to this problem is to examine the concept that the work is implied by and limited by its physical manifestations, rather than being identical with them. This examination requires that we contemplate, if only for argument, the idea that the work is an ideal or mental construct (or constructs) separate from but represented by physical forms. We can do this without arguing that the work is either the mental construct or that it is the physical form, and we need not argue that one or the other has a greater claim. Instead we might pursue the implications of defining the work as a mental construct that can be known only through its physical forms and the

effects they create or allow.[22] Note carefully that I do not mean, by this distinction, the difference between a sign and its meaning or referent. I mean instead the difference between the physical sign sequence as recorded in copies of a work and the sign sequence a user of the copy of the work takes to be the work. The latter sign sequence is a mental construction deriving from the former with the added proviso that the user may consider the physical copy of the work to be marred by error or abridgement or to be partial by reason of revisions not recorded in that copy or even by reason of inappropriate packaging.

When two or more of these physical forms of a work disagree, it is patently obvious that, if the Work is a single ideal entity, they cannot both accurately represent it. Two possible explanations for differences between two physical manifestations of the work can be suggested. The first is that one is corrupt and thus misrepresents the Work (or both copies could be corrupt in different ways). The second is that the Work exists in two (or more) Versions each represented more or less well by one of the physical copies.[23] We can think of the Work, then, as existing in more than one Version and yet be one Work. This does not, however, help to resolve the problem of whether the Work or a Version of it is accurately represented by the physical copy held by person X or person Y.

Before pursuing that problem, there are some difficulties with the concept of Version to try to clarify. First, like the term Work, the term Version does not designate an object; it, too, is a means of classifying objects. In the same way that the Work Dombey and Son is not Moby-Dick, so too a first version is not a second, or a magazine version is not a chapter in a book, or a printed version is not a version for oral presentation. The term Version in these formulations is a means of classifying copies of a Work according to one or more concepts that help account for the variant texts or variant formats that characterize them. Second, it is not just the existence of different texts of the same Work which

leads us to imagine multiple Versions of a Work. What we know about composition also suggests Versions. And to help distinguish various concepts relating to version, I would suggest the sub-categories Potential Version, Developing Version, and Essayed Version. These categories correspond to ideas we have about composition and revision. Potential Version refers to the abstract incipient ideas about the Work as it grows in the consciousness of the author. The Potential Version has no physical manifestation, but we judge from our own experience in composing that such a version exists at least in outline and we imagine this version capable of being developed, abandoned, or changed. The Potential Version is unavailable to us except as an idea. Developing Version refers to a process that does have physical outcomes. The Potential Version processed by thought and inscription produces, in the case of many authors, drafts or notes, which when added to more thought, more inscription, and perhaps some revision results in additional drafts. When the Developing Version has progressed sufficiently and been consolidated into an inscription of the whole, we have a physical representation of what I would call an Essayed Version.

The point at which the developing version reaches sufficient wholeness to be thought of as representing the first Essayed Version is, of course, a matter of opinion and, therefore, of dispute. This problem is another demonstration of the fact that the term "Version" refers to a means of classification, not to an actual stable object. The first Essayed Version can be thought about and revised and used as a basis for producing a second Version, etc. It might also be thought of as a provisional version or a finished version, but it is a version of the work in that it represents the work. Though the Essayed Version has physical embodiment in a text, it is not the physical text. We can imagine the Essayed Version in the author's imagination as more perfect than his or her ability to record it in signs which require compromise and are liable to inscription error. Even if there is only one physical copy of the Work, one could not say that the Version it embodied was the Work, for as soon as a new Version appeared the distinction between Version and Work would become necessary again.

We should pause for a moment here, suspended in the ethereal realm of ideal forms, to observe that the idea that "a Work is implied by a series of Versions" is based on ideas about composition, revision, and editorial interventions. That is, I have developed these ideas by imagining the processes of composition, not by starting with finished copies of the Work and inferring the processes "backward" from them. To think in this way about a work entails also believing that each new version has integrity or "entity" as an Utterance of the Work. If two copies of a work

differ in ways that are explained by "infelicities in transmission" then one does not need a concept of Version to explain the differences. But if each is thought to be desirable or "authoritative" in its own way, then the concept of Version is useful for classification. One could think of a Version, then, as the conception or aim of the Work at a point of Utterance. But Version is a very complex and slippery concept I will define and discuss in detail later. Where there is a well-established convention for using the term Work to distinguish between Vanity Fair and Jane Eyre, there is not an established convention for distinguishing Versions of Vanity Fair.

Of the problems concerning the concept of Versions which must be discussed in detail later there are two which should be mentioned before moving on to definitions of Text and making clearer the connections between ideal concepts of works and their physical manifestations. The first is the problem of determining when the Essayed Version has stopped being the Developing Version so that it can be thought of as coalesced into a Version that can be identified and read as such. The second is the problem of determining if and when a second version has coalesced that should be considered as separate from the first. To discuss these problems we need several related concepts I will develop later: Time, Content, Function, and Material. One should also note that concepts of Intention and Authority are crucial to the idea of Versions; neither of these concepts is simple.[24] Needless to say, I think the idea of Versions is a very useful one, in spite of its problems.

B. Texts: Conceptual, Semiotic, and Physical

Although I have deferred discussion of some of the problems with the term Version, I need here to imagine the writer composing a Version of the Work in order to pursue the taxonomy through various concepts that are too often hidden in the use of terms such as Work or Version. One should note, then, that an Essayed Version is a conceptual entity not a physical entity; it is not equivalent with the physical embodiment of it, because its embodiment can be and usually is an imperfect representation of the Version. The contortions of that last sentence bear witness to the fact that Version is being used in two ways: it is a classification system for those texts that represent Version X as opposed to those that represent Version Y, and it is a Conceptual Text which copies of Version X or Y represent. This latter notion, the Conceptual Text, is not a system of classification but more like an ideal form of the Work. But it is not a Platonic ideal, for it develops and changes, and probably does not

"pre-exist" as an ideal, and it probably does not last very long either. The imperfections of physical texts are of various origins, including failures of creative imagination, failures of inscriptional skill or care, use of elisions and abbreviations to be filled in later, or unhappy interventions by scribal assistants. The Essayed Conceptual Text is always manifested in a physical form, but it is not a physical or Material Text, for the Conceptual Text that is Essayed remains (as the author's mental concept) invisible and probably not stable; but the embodiment of the Conceptual Text is visible and fixed in a material medium. The concept of "fixing" suggests another reason the Material Text may misrepresent or at least only suggest the work: Version (Potential, Developing, and Essayed) is fluid conceptual process, but the material text is physically static, fixed. However, since the Essayed Conceptual Text cannot be known except through a Material Text, people tend to equate them for practical purposes. But the Material Text can misrepresent the Essayed Conceptual Text and hence that equation is not exact. The ways in which the Material Text can misrepresent the Conceptual Text are many and often are indeterminate but some might be revealed in the drafts or by violations of syntax, grammar or orthography that cannot be justified as accurate representations of the ideal Version.[25]

It is common, at least among textual critics, to think of a text as consisting of words and punctuation in a particular order. I would like to call this concept of texts the Linguistic Text.[26] It refers to the semiotic dimension of Texts—the specific signs for words and word markers that stand for the Work (or the Version of the work). Linguistic Texts have three forms: Conceptual, Semiotic, and Material. The author's Conceptual Linguistic Text consisted of the signs he "intended to inscribe." A Semiotic Text consists of the signs found recorded in a physical form of the work. If a Version represents the conception or aim of the Work at a point of Utterance, the Linguistic Text is the execution or achievement of that Version, first as a Conceptual Text (thought) then as a Semiotic Text (sign), and then as a Material Text (paper and ink or some other physical inscription or production), at that point of Utterance. The Material Text is the evidence that a Conceptual Text was

formed and Uttered as a representation of a Version of the Work—in short, if there is no Material Text there is no Linguistic Text and hence no Version available to a reader. The Conceptual Text can be Materialized in spoken or written form, and it can be recorded in a mechanical or electronic way. It follows that the Linguistic Text can have more than one Semiotic form—spoken, written, electronic, and Braille, for instance. The Linguistic Text is not, therefore, physical; it is a sequence of words and word markers, conceived before spoken or written, and taking its semiotic form, when written, from the sign system used to indicate the language in which it is composed. We must also distinguish between the Linguistic Text and the Documents that preserve them, for as long as the sequence of words and markers is the same, the Linguistic Text is one, regardless of the number of copies or number of forms it is manifested in. All accurate copies, whether facsimiles, transcriptions, or encodings are the same single Linguistic Text. An inaccurate copy, however, is a different Linguistic Text for it is a different sequence of words and word markers, though it might still represent the same Version. The new Linguistic Text might represent the Essayed Conceptual Text more faithfully or less faithfully.

It should be noted that the Linguistic Texts representing an Essayed Version (the ideal aim of Utterance) run the risk of error at each transformation in production both through a failure of articulation (we've heard authors complain that they just couldn't put what they wanted into words) and because the author or a scribe failed to inscribe it accurately or completely. The Linguistic Text, therefore, corresponds to the Essayed Version only to the extent that its production was perfect. Editors (particularly "authorial intention" editors) have understood their job to be the production of a newly edited Linguistic Text that accurately represents the author's intentions for the final Version. Put in the terms defined here, the traditional "intention" of scholarly editing has been to create a new Material Text, the Linguistic Text of which coincides with the Essayed Conceptual Text. But because the author's Essayed Text is available to the editor only through material evidence for it, the editor can do no more than construct a new conceptualization of it (i.e., the editor does not in fact "recover" the author's Conceptual Text). The resulting edition is then a forward construction rather than a "backward" restoration.

To speak of the Linguistic Text as a sequence of words and word markers is to emphasize a distinction already made but that is of primary importance: that the Linguistic Text, being composed of signs, is a representation of the work and is not the work itself. It represents a Version, it is not the Version itself. It is the result of an encoding process

undertaken by the author or the author and his assistants. The Linguistic Text is, therefore, a sign and not an object, though it is always manifested in an object. To speak this way about the Linguistic Text is also to emphasize the act of decoding which is necessary before another person can be said to have seen or experienced the work of art. It should be equally evident that such a decoding experience cannot take place without a physical manifestation of the text as a starting point.[27]

The word Document can be used to refer to the physical "container" of the Linguistic Text. It might be paper and ink or a recording of some sort, including for example a Braille transcript which can be just paper. Records, tape recordings, microforms, and computer disks are also documents, though decoding such documents requires mechanical or electronic equipment. Documents are physical, material objects that can be held in the hand. Each new copy of the Linguistic Text is in a new document. Two documents containing the same Linguistic Text are still two separate entities but only one Linguistic Text. This physical form not only provides a "fixing medium" (to borrow a concept from photography) but it inevitably provides an immediate context and texture for the Linguistic Text. It will be useful therefore to have a term for the union of Linguistic Text and Document. I call it the Material Text. It seems clear that a reader reacts not just to the Linguistic Text when reading but to the Material Text, though it be subconsciously, taking in impressions about paper and ink quality, typographic design, size, weight, and length of document, and style and quality of binding, and perhaps from all these together some sense of authority or integrity (or lack thereof) for the text. These aspects of the Material Text carry indications of date and origin, and social and economic provenance and status, which can influence the reader's understanding of and reaction to the Linguistic Text.[28] (See Chart 3.)

We should pause again for a moment, this time with our feet firmly planted in the material realm. A Material Text, any Material Text, is

the reader's only access route to the Work. A Linguistic Text cannot exist for anyone (who does not already hold it in memory) without a material medium;[29] the Linguistic Text and its medium are the Material Text with all the implications of that union. Material Texts are the production of Utterance. The first Material Text (says the manuscript) is the first attempted union of the Essayed Version and a Document. There might be a problem in distinguishing that first Material Text from draft fragments, and it might be possible to "reconstruct" archaeologically a Version buried in drafts in early manuscripts or in the cancelled and altered passages in a manuscript or typescript whose final revisions represent Essayed Version one. Material Texts numbers 2-n are transcriptions made by anyone including the author. These Material Texts might incorporate the results of revisions, editorial interventions, or errors, or they might be accurate transcripts.

It would appear from the concept of Material Texts that when an editor has extracted or edited the Linguistic Text which he believes best represents the Version he is editing, he must embody that Linguistic Text in a new document which will be a new Material Text with implications all its own. He cannot reincorporate a new Linguistic Text into an old Document to present a "restored" Material Text. The force of this idea came to me while reading Jerome McGann's explanation of the work as a product of social contract in which the production process was described as an integral and inevitable aspect of the concept of the work of art.[30]

C. Texts Again: Physical, Semiotic, and Conceptual

The terms Version, Text, and Document have brought us in the life of a literary work of art only through the down-swing of the pendulum from the "mind of the author" to the concrete manifestations of the work in Material Texts (i.e., books). And it should perhaps be emphasized once again that a Work may be "implied" by more than one Version and by more than one swing of the pendulum. But now we must face the Material Text in the absence of the Author and with a realization that as we approach the Material Text we are not before a verbal construct and that we cannot see prelocution, perlocution or illocution, or even intention or meaning. What we have before us is molecules compounded in paper and ink. Everything else must be inferred from that, beginning with the recognition of the sign shapes, which the ink shapes materially represent. I have mentioned that the Material Text is the starting point for further processes, the up-swing, necessary before the Work can be perceived, for the Material Text is not equivalent with the Work but is instead merely a coded representation or sign of the Work. Furthermore, the Material Text before us is only a single instance of many possible manifestations of the Work. Not all Material Texts are necessarily representative of the same Version of the Work, nor are they all equally accurate representatives of the Work. Nevertheless, a Material Text is where the reader begins the process of perceiving or experiencing the work of art. This process is one of decoding or dematerializing the Material Text into some mental construct of it. It is in this decoding process that the Work can be said to function.[31]

IV. Textual Performances as Write Acts

It seems useful here to add the term Performance to our taxonomy of concepts related to Works of literary art.[32] Performance is an act, an event. Performances take place in time and space. They are not material objects, though they might produce results that are material and that can be used as records of the performance. However, these outcomes of performance are not the performances themselves. It will be useful to distinguish between at least three types: Creative Performance, Production Performance, and Reception Performance.

Creative Performance refers primarily to acts of authority over linguistic texts, determining what shall be encoded as the inscription representing a Version.[33] Creative Performance includes all that was indicated above by the terms Potential Version, Developing Version, and Essayed Version. Creative Performance is primarily inventive but usually involves some sort of mechanical work to inscribe through writing, typing or dictating. This mechanical aspect should perhaps more properly be called Production Performance, but when the author is inscribing new material it is clearly primarily a creative activity. One might say, however, that when the author makes a mechanical error in inscription, it might be a failure of production rather than of creation. To a casual reader this difference makes no difference, but to the editor who holds production authority over the work, it makes a significant difference, since he will correct a production error but not a creative failure (creative "errors" might, by the way, be creative innovations the editor has failed to understand).

Production Performance refers primarily to acts of authority over Material Texts, determining what material form the Linguistic Text shall have and re-inscribing it in those forms for public distribution. Production Performance can have a variety of methods and outcomes;

they can be nurturing or negligent, skillful or clumsy, well-intentioned and wise or well-intentioned but ignorant. Production Performance often affects the Linguistic Text and always affects the Material Text, but it differs from Creative Performance in that its primary purpose is the transmission and preservation and formal (not substantive) improvement of the Linguistic Text. It is a process of transcription, not one of revision. Creative Performance and Production Performance are often carried out simultaneously by the same person, but traditionally Creative Performance has been associated with authoring the Linguistic Text and Production Performance has been associated with manufacture and publishing the Material Text. In practice these two processes are not always easily separable, for authors occasionally perform production acts and publishers, printers, and editors quite often perform "authoring" acts. The results of these crossings are sometimes "happy" and sometimes not—often the judgment depends on who is judging.

Reception Performance refers to acts of decoding Linguistic Texts and "conceptualizing" the Material Text; that is what we do when reading and analyzing. Reception Performance differs from Production Performance in that its primary purpose is not the reproduction of the Linguistic Text in a new material form, but the construction of and interaction with the Linguistic Text in the form of a Conceptual Text. Readers do not normally distinguish consciously between the Material Text and their Conceptualized Text derived from it. They are also often unconscious of the ways in which the Material Text is more than just the Linguistic Text of the Work so that their Conceptual Text is formed under the influence of material contexts that did not attend the process by which the author materialized his Conceptual Text by inscribing it. To put this in a simple model, the author's Essayed Conceptual Text takes form as a Material Text which the reader uses to construct the Reception Conceptual Text. If we imagine, then, that the specific copy of the Work that reader X is using is Material Text X, that copy with its textual limitations and errors is what the reader is reading. It is a Material Text, not the Work, though the Work can only be known through a Material Text. It need not, however, be known through this particular copy; the imperfections of the particular Linguistic Text as well as the implications of the particular Material Text contribute to the uniqueness of this particular representation of the Work. Furthermore, it is not the Work itself that is known through the Material Text but the reader's reconstruction of the Work that is known, the "reader's Conceptual Linguistic Text as mediated by the Material Text," or, in short, the Reception Text. It should be noted that the Reception Text is still what Saussure calls a "signifier," for it is no more than the Linguistic

Text in internalized Semiotic form. It is then reacted to in a variety of ways and according to a variety of principles of interpretation which taken all together can be called the Reception Performance. The point to emphasize and then to elaborate is that these reactions are to the Reception Text not to the Material Text. (See Chart 4, where critics Q and R read the same copy of the Work and may disagree about interpretation because of their different skills in performing the Reception Text,

because their experiences of life and reading differ, or because they employ different interpretational principles. Critics S and T, who read two different editions reproducing more or less well the same Version of the Work, may disagree about interpretation for any of the same reasons Q and R disagree, and because the Material Texts in which they encounter the Work differ. They may also, fortuitously, agree with each other if one or both have managed to ignore or "misread" the Material Text. Critics X and Y, who read different copies of the work, each representing a different Version, may disagree for any of the same reasons affecting Q, R, S, and T and also because the Linguistic Texts they are reading are different. To the extent that Q, R, S, T, X and Y think their copy of the work is the Work, their disagreements will seem unaccountable, irresolvable, or evidences of inadequacy in the others as critics.)

It might be useful to describe the process of Reception Performance by adapting some terms used by I. A. Richards to describe his experiments in practical criticism in the 1920s. Several "perusals" of a text at one sitting constituted an "attack" on the work of art. Several "attacks" spread over a short period of time, say a week, constituted a "reading." The reader's commentary on the work—the record of his reading and reaction—was called a "protocol." We sometimes call interpretations of works "readings," but the word is vague and overworked; we should call them something else such as protocols or records of the Reception Performance. I think it can be said that Richards was interested in this process as a process of interpretation of meaning, effect, and tone suggested by the words as grouped into sentences and paragraphs, and that he was not concerned with the problematic nature of the Material Texts he and his students used. Nor was he concerned with the problematic nature of the dematerialization of the text signs for words and punctuation. That is, he was interested in what the Text said, not in what the Work was. This is a common strategy of literary critics to avoid the problems of "authorial intentions." What I have called the Reception Text is in part the reader's decoding of the Linguistic Text as embodied in the Material Text at hand, but it also includes the reader's semiotic reconstruction or reading of the Material Text as a totality and to the environment in which the reader has undertaken the Reception Performance. Anything the reader says or writes about his experience of the work is a "protocol." The rules by which protocols are produced and judged are as numerous as there are games to be played in the Performance Field. (See summary of terms in Appendix B.)

We have in these three performances a key to why observations made about speech acts go awry when applied to writing. A speech act or

spoken utterance is one event with three basic elements: the utterer's mental concept, the physical medium of utterance, and the listener's mental concept. These three elements always exist together in the context of time and place when and where the utterance is spoken. In written works all three of these elements exist also, but the context of time and place is fragmented, so that the writer's utterance takes place, so to speak, in the presence of an absent reader, and the reader's reception or construction of utterance takes place in the presence of an absent writer. Therefore, each utterance takes place in a context of time and place that is unknown to the other party and adventitious meanings are the highly likely result, for the "bundle" or "molecule" has been broken, modified or replaced. Finally, to complicate things even more, the writer's writing is seldom seen by the reader who usually has instead the printer's printing. So a written work entails at least three separate events (performances) whereas the spoken work is one event.[34] Experienced writers are, of course, aware of this and compensate by a multitude of strategies. That is one reason it is normal to think of writing as more formal and requiring more care than spoken communication. There are also many other reasons that written language must be made clearer, among them the fact that punctuation is a coarse substitute for intonation and gesture.

V. Hermeneutics or Theory

The terms and definitions provided thus far represent a "structuring of reality," which cannot be "true" but can perhaps be useful. It can be used in two fundamentally different ways that correspond to hermeneutics and literary critical theory. Photography again provides a good though not perfect analogy, since photographs can also be treated in these two ways. One could say that the Material Text is not the Work, but like a photograph, it represents the work, so that a person contemplating the Material Text (photograph) can create a mental construct corresponding to that which was photographed. There are, in this view of the matter, two reception images: one of the photograph or Material Text itself and another of that which was photographed or the "created world" in the work of art.[35] Another useful analogy to emphasize this hermeneutical use of the taxonomy is musical. The Material Text is like

the score of music; the Reception Performance is like the concert which created the musical event when the score is played. Thus the documentary musical score and the musical performances it enables are separate though related referents; likewise the Material Text and the Reception Performances it enables are separate though related. Reception Performances actually have two distinguishable parts—the first is the construction of a Reception Text, the second is the interpretation of and response to it. In the analogy with music, constructing the Reception Text is like an orchestra playing the music. Response to the Reception Text is analogous to listening to and responding to the played music. The reader is orchestra and audience rolled into one for a literary work. This distinction is readily seen as applicable to drama. Not only is there a difference between the written text of a play and the performances on stage—both of which precede Reception Performance by the audience in a theater—but there are often differences between the written text for performance and the written text for publication, the later intended to be performed as a reading (closet drama) rather than as a stage performance.[36] It is important to note that the reader's responses are not to the "work itself" but to the Reception Text. The quality of the Reception Text depends in part on the Material Text used as a basis and in part on the skill and quality of its rendition as performance; halting or inattentive readers are not likely to perform technically "good" Reception Texts, but this will not keep them from responding to them.

The taxonomy I have suggested, then, shows not only how interpretation or other protocols depending on a specific Material Text to reconstruct a Work might differ from other protocols depending on other Material Texts with the same title, but it shows how interpretations of the same Material Text by two persons or by the same person on two different occasions might also be different—each experience of the Work is based on a separately constructed Reception Text in a separate Reception Performance using a particular (and partial) representation of the work.

The second way of looking at this taxonomy is analogous to literary critical theory, focusing attention on the phenomenon of works embodied in Material Texts rather than focusing on the interpretation of those texts. In the analogy with photography, it is an examination of the photograph as photograph and photography as a process rather than seeing the photograph as representation of something else. The photograph becomes, in a way, an end in itself with its own meaning and ways of meaning regardless of any correspondence between itself and some other object or objects towards which the camera was aimed when the exposure was made. The critic's attention might be focused on technique, composition, texture and any number of other things relevant to the photograph as object and as result of process, all of which are unconcerned with "realism" or accuracy of representation. In photography each newly developed picture from a single negative is a material object which may be considered as a separate work; viewers of more than one copy might comment on the differences and prefer one to another for various technical and aesthetic reasons. We can use this taxonomy, therefore, to illuminate a variety of interests relating to process rather than to interpretation.

Literary critical theory has focused its attention primarily, though not exclusively on the portion of this taxonomy indicated by the terms Linguistic Text, Creative Performance, and Reception Performance. These correspond to the West, North, and East in the "map" of related interests described in section II. Literary critical theory has tended to neglect the portions indicated by Versions, Material Texts, and Production Performance. These correspond to the South and, astonishingly, the center of the "map." Marxist criticism has paid attention to Material Texts and Production Performance, but has tended to neglect or down-play Creative Performance and the mental constructs of Works and Versions. Textual criticism, until recently focused primarily on the Linguistic Texts and Creative Performance, has tended to neglect or deemphasize Production Performance and Reception Performance. To the

extent that textual critics neglect Reception Performance, they turn a blind eye to the role of their own Reception Performance in "reconstructing" authorial intention.

VI. Contexts and Conventions

I round out this Taxonomy with three more terms in a somewhat cursory attempt to indicate the relevance of contexts and conventions to concepts of Text: Material Base, Social Context, and Performance Field.

Material Base—The world of sense data to which biological authors and readers and Material Texts (manuscript, book, etc.) belong. The Material Base includes all the raw materials of historicism. The term is used by J. Hillis Miller to refer not only to the book and earth and cities and to agricultural and industrial products, but he extends it to include institutions, conventions, and mores. He does so to present the view that there can be a visceral (i.e., physical) connection between "a people" and "the land" that is part of the textual complex.[37] Note that the Material Base is the "world" that the book and reader are a part of, not the "world" to which the work could be said to refer or to portray. Both of these worlds are, according to the "beliefs of the time" with which I began this paper, constructs or structured "texts of reality," but the "world the work refers to" is one construction removed from the Material Base.

Social Context—My preferred term for the complex of institutions, conventions, and mores whose expectations and habits are reflected in the Material Text and which form the extended field of inquiry along with the Material Base for new historicism. In particular with regard to Texts it should be noted that these conventions include relations and contracts and expectations between author and typists; author and publisher; publisher and editor, compositor, printer, binder, bookseller; and between author, publisher, bookseller and reader. I separate the Social Contexts from the Material Base because, as Thomas Carlyle pointed out in Sartor Resartus with more panache than originality, institutions, skills, and conventions are aeriform—they are constructs, not found objects. The principles of relativity and structuralism tell us that "found objects" are also constructs, but that insight does not smooth out the distinction between material constructs and social constructs.

Performance Field—This is where the Performance Text is "played" according to the rules of the reader's particular game of textual interaction and further limited by the performer's capabilities and resources. This is a term suggested by Roland Barthes's definition of Text, which he distinguishes from "Work" (by which he means Book or Material

Text) but which he does not distinguish from Linguistic Text or Performance Text. Performance Field is Barthes's field of play, though every reader performs or plays by the rules of his or her own league.

VII. Problems with Concepts of Versions

A. One Version or Several?

Earlier, I passed rather quickly over two crucial but complicated issues to which I promised to return. The first was the question of how one determines whether an inscribed text accurately reflects the work it was meant to represent. The second, even more complex, had to do with determining, first, when a version had coalesced as a finished form and, second, when another version, differing from the first, can be distinguished as a separate entity.

Textual criticism from earliest times has been primarily concerned with the first of these problems; one might say its history has been one of obsession with the problem of textual corruption. I am not going to try to provide a primer on the subject here. What I said was that when two copies of a work, both bearing the same title and purporting to be the same work, contain variant Linguistic Texts, one explanation could be that one or both texts misrepresented the work. And, if the work was a single thing, then at least one of the variant texts had to be wrong. The point was that a work could be misrepresented by a copy of it. And it follows, therefore, that the work might be misrepresented by every copy of it. From this observation we must conclude that the work and the copies of the work are separate entities. It has been the business of textual criticism to do what it could about such misrepresentations. I will say no more here about how that can be done except to note that the textual critic's concept of Authority for the work is central to his task.[38]

The other explanation offered for textual variation between copies of a work was that the work might exist in two versions each represented by one of the variant copies each of which could be correct. Now it is a commonplace that authors revise their works, and mere revision has seldom been taken as proof that a separate version of the work exists. But if variant forms of the work are legitimate (i.e., not the result of corruption or inattention), and if reader X disagrees with reader Y because they are not reading identical texts, then something significant has occurred,

which impels us to think the concept of Versions of a Work might be useful.

But for several reasons, the problem of Versions is not simple. Literary works of art come in Material Texts which are linear and single in form. Variant texts, therefore, are difficult to present and assimilate; they are not easily experienced simultaneously or side by side. Thus most reading experiences are restricted to interaction with one Material Text representing better or worse one Version of the Work. Publishers are committed to perpetuating this form of experience and resist multiple text editions. Perhaps that is why there has been a tendency to consider revisions as a single continuum of creative efforts made to improve the work. The process is said to be over only when the last revision is made—and even then the process might have just stopped unfinished. This is a fundamental principle for "final intention" editions. James McLaverty calls this a Whig interpretation of revision, which often disregards meaning and effect in favor of a predisposition to credit revision with improvement.[39] The "Whig" view is convenient, for it maintains that the Work is singular and revisions are all part of a grand design toward which the author works from beginning to end. Variant texts, according to this view, either contain errors or represent incomplete revision. With this view it would be considered a reading Utopia if all Material Texts in circulation were accurate renditions of the "final intention text"—a Utopia of logo-centrism.[40]

It is tempting to dismiss such views of the Work as oversimplifications, but not only do such views characterize most readers' habitual attitudes towards the texts they use, there are powerful influences in our culture, at least in the present, to accept and even to enforce such a view. The alternatives might be more honest or more sophisticated or more intellectually rigorous, but is it art? Is it the real thing? These are questions about authority and authenticity. In painting, the questions are, Is it a copy? Is it a fake? In literature one hardly ever thinks of a fake novel. But we can say of a poem that it is "only a copy, and not a very good one at that," by which we probably mean that its "authority" has been

compromised through textual variation from "unauthorized" sources. I have discussed the question of authority elsewhere, and so will not develop the idea here.[41] Suffice it to say that authority is not found in works but is attributed to them according to how the user defines authority. There are currently four common definitions of authority, some with a marvelous variety of subdivisions which feed astonishing controversies about which is the proper work of art and what is the proper goal of editing. Authority is a system of valuations relating to the Work for the purpose of distinguishing between what is the Work and what is not the Work.[42] In the hands of Whigs—those who want the Work to be one identifiable, real thing—authority is used to discountenance all Versions but the "true one." Exceptions are sometimes made for Works that have two or even more "true versions" such as Wordsworth's Prelude or Marianne Moore's "Poetry," but these are quite exceptional cases. Anyone who admits the possibility of more than one version, however, needs concepts other than "authority" to distinguish them. It stands to reason that if two people disagree on the definition or application of the concept of Authority, they will not be able to agree about Version. As we shall see, I think Version, like Authority, is not so much found in the textual material as it is put there. The ways in which Versions are identified, then, become an important matter to discover.

B. Structuring Multiple Versions

"Post-Whig" ways of gauging the significance (i.e., meaning and effect, and thence importance) of revision involve a concept of Versions identified or delimited with reference to one or more of "four unities": Time, Content, Function, and Material. The main point here is that a concept of Versions requires a way to identify something that can be "perceived" only through potentially misleading physical representations of it. A concept of Version has to be able to identify Version by distinguishing it both from other Versions of the Work and from the physical manifestations of it, which might be corrupt or which might actually mix text from more than one Version. It must also be able to distinguish between texts which differ because they represent different Versions and texts which differ merely because one or both contain errors. Unlike any of the distinctions between terms referring to the forms of Texts (Conceptual, Linguistic, Semiotic, Material, and Reception Texts), decisions about what constitutes a Version are matters of judgment

and will depend entirely on the relative weight that the judge places on one or more of "four unities" in making that judgment.

(1) Utterances

Before discussing the "four unities" and how they have been used to identify Versions, we need to look again at the concept of Utterance, a term borrowed from speech act theory and literary theory.[43] The term is a problem, however, because it has been defined in several ways and applied to the acts of persons other than the author. Utterance can be the act of formulating the conception of the Work we call a Version into a Linguistic Text. If we define it so, however, we come very close to saying that each Utterance is a work of art and we might hesitate to accept this idea. Utterance can also be defined as the act of making a Version available or making it public. Here again, several acts can be referred to by the term. Making a Version available might be done by writing it down, or by giving it to a typist, or by submitting it to a publisher, or by reading and approving final proofs, or by publishing the printed book. Each of these acts might be thought of as a moment of Utterance which gives the Linguistic Text involved status as representing a distinct Version. Further, Utterance can be defined as what the author said or wrote, what the production process produced and published, or what the reader heard or read.

In order for Utterance to be a useful term we must not only distinguish it from Version and our other terms but show how it is helpful in describing or organizing them. We might say that Version is the aim of Utterance but that an Utterance might not succeed or might only partially succeed in its aim. But Utterance is not merely the production of a Material Text that might or might not accurately represent a Version. Utterance not only refers to the Performances of Works but to the circumstances, the contexts of those actions which influence and contain (i.e., keep from running wild) the meaning and help indicate what meanings are operable. This is a relatively simple concept in speech, as I have already noted, where the speaker and listener and circumstance are all together interacting at the moment of speech. But with written or recorded language, the Utterance of the author, of the various members of the production crew, and of the reader are each separated in time and circumstances so that meaning at every stage in the life of the written word is influenced by different milieux. It is not absurd, therefore, to

conceive not only of Creative Versions resulting from authorial acts (which has been the focus thus far) but of Production Versions (publication utterances), and Reception Versions (reader utterances). Now this would be complicated enough if the Work were one thing which could be rationalized into one universal text out of this proliferation of "versions" (which would not be true versions but simple imperfections). But, in fact, there is no agreement among scholars or artists about what one thing the Work is or ought to be (i.e., there is no universal definition of authority). Each person has a notion about what it ought to be, but the possibilities are quite numerous. Some concept of authority (to identify the authentic elements of the Work) and the four unities (to distinguish Versions of the Work) are the means by which readers impose order on this cacophony of utterances. That is, how an individual student of the Work understands Versions and how he reacts to Material Text X in his hand will depend on the specific Utterance selected and defined for use. And that selection and definition depends on the values given to the "four unities": Content, Time, Function, and Material. Needless to say, these values are usually selected and applied without conscious thought—in which case the Material Text becomes transparent and the only "text" that matters is purely and simply the Reception Text in the reader's Performance Field.

(2) Unities as Structural Glue

The unity of Content is the place to begin.[44] It is because the content, particularly the Linguistic Texts, of copy X and copy Y were not identical that this discussion began. If they had been identical there would appear to be no problem. The idea that one copy is accurate and the other inaccurate does not explain cases of revision. The idea that one copy represents an early incomplete stage of the work and the other represents a completed or improved stage does not explain cases where the revisions appear to mean contradictory things or to have palpably different but individually satisfying effects. But the problem here is to calculate first whether the content had a sufficient stability as an "entity" to be called a Version, and next to calculate how much of a change or what kind of change in content is required before a different Version, rather than an improved Version, results. The most radical answer to this question was offered by Hans Zeller, when he described the work as a network of relationships between its parts. He reasoned that any change

in any part would change the nature of the network; and so, every textual change produces, logically, a changed work.[45] There is an empirical attractiveness in this view because it equates the work with the copy of the work; each variant copy is a new work, but there is an unsatisfying or disturbing implication in this view because it makes editing a work a nearly impossible task. A scholarly edition would have to incorporate whole texts of every authoritative source. G. Thomas Tanselle offers a compromise through his distinction between "horizontal and vertical" revision. Horizontal changes merely improve a presentation or intention already achieved more or less well in the original text; vertical changes alter the intention by changing the meaning or direction of the work.[46] In order to gauge the type of change, Tanselle uses also the unity of Function, so that not all changes in Content signal changes in Version. He suggests also that differences in the Time of revision might be a useful factor, but he does not anticipate the case of an accumulation of "horizontal" improvements having a "vertical" effect.[47]

The unity of Time derives from the idea that the person changes with time so that if an effort of creation is separated from an effort of revision it is likely or at least possible that the revision effort will reflect changes in the person and thus follow its own line of inspiration rather than that which informed the first. But the problem here is how to calculate how much time must elapse between engagements with the text for the lapse of time to be deemed significant and the resulting effort to be seen as a separate Version.[48] Among modern textual critics, the most radical view of Version that depends on the unity of Time is the one presented by Hershel Parker in Flawed Texts and Verbal Icons where he argues from a phychological model of creativity that authors lose their authority over a work after a certain period and that revision often not only violates the creativity of the original effort but can end in confusion which might make a text unreadable.[49] A good deal of my own 1984 recommendation concerning identification of Versions depends on the

unity of Time. I would no longer rely so heavily on this one aspect. Even for an editor who is concerned with presenting only one somehow "best" version of the work, the unity of time is sometimes used to reject undesirable authorial revisions made later in time on the grounds that the passage of time had deprived the author of the inspiration (or at least continuity of thought or purpose) that informed the work now being revised without inspiration.

The unity of Function relates to the purpose for which the work is designed. Is it for a magazine; is it a chapter in a book; is it a play adaptation, a translation, a revised edition aimed at a new market? Each new function constitutes the potential for a new version. Revisions undertaken to adapt the work to a new function should not, according to this unity, be confused with revisions undertaken to enhance the success of the same function served by the unrevised text. This criterion requires that the revision be for a different purpose, not just a better fulfillment of an old purpose. Fredson Bowers has written considerably about this aspect of the identity of Versions, but in practice Bowers has tended to see new functions as superseding old functions (as long as they are authorial); so that, while he admits that the previous Versions have "authority" he tends to see new Versions as having "superior authority."[50] This is an example of what McLaverty calls the Whig interpretation of revision, the idea that revisions are better because it is absurd to think that an author would deliberately revise his work to make it worse.

The unity of Material relates to production efforts. In this concept the word Material means the physical object or document that bears the Linguistic Text. It equates, in effect, the concept of Version of the Work with the Material Text. The Material Text is, after all, the place where all the Performances and all the component aspects of a Work are brought together. The Creative Performance resulting in a Linguistic Text is united to the Production Performance resulting in a Material Text, which is where the Reception Performance must begin. The Material Text can be seen then as a social, economic and artistic unit and is the entity necessary for the full functioning of literary art. The primary proponents of this point of view are Jerome McGann and D. F. McKenzie.[51] The most obvious shortcoming of this position seems to be

its rigidity, its rather helpless acceptance of determinacy in the Material Text. There is a grand sense of coalescence in the view, but the Linguistic Text involved in many cases will strike some readers as having been over-determined or perhaps over-packaged. Production processes notoriously tamper with a Linguistic Text in ways both beneficial and detrimental to it as a representation of the Essayed Version. And it does not sit well with some people that the economic necessities and accidents of Production Performances should be allowed to shape (sometimes to shape out of existence) the subtleties of the Creative Performance.

VIII. The Reader as Author

The conflicting claims of these four "unities" have been forcefully and amusingly presented in James McLaverty's rehearsal of the old analogy of Theseus' ship, except that the distinction between Content and Material is fudged.[52] Suppose that Theseus sets sail, says McLaverty, in a ship which after a while undergoes repairs. Say further that the ship eventually is so repaired as to have had all its material parts replaced. Is it still the same ship? The unity of Function would say yes, but the unities of Content, Material, and Time would say no. Suppose further, that someone gathered up the discarded parts of Theseus' ship and reconstructed the original ship; would the old reconstructed ship be Theseus' ship or the new repaired one? The unity of Function, if considered in the light of Continuity, would say the repaired one was the ship; but in the light of identity of Function it would say the old reconstructed ship had as good a claim. The unity of Material would support the old ship's claim, while the unity of Time would insist that three versions of Theseus' ship existed: the old one that had been Theseus' ship but which no longer exists, the repaired one Theseus now sails, and the reconstructed one which is not identical with the original ship though made of the same materials in the same configuration.[53] What happens if Theseus sells his ship and it becomes a cargo boat? Or what if a new owner also named Theseus uses it for the same purpose it originally served? Is that possible? Etc.

The main difficulty with the analogy is that in the case of the ship, the material being is the ship, while in the case of literary works, the physical document—the Material Text—is but a representation of the

Work. If Theseus were to say, "No, that ship reconstructed from old material is not my ship," we would believe him only if we vested authority for the naming of the ship in Theseus (the author, as it were). The Material Text—whether it is the Author's or the publisher's, whether it represents Version 1 or n, whether it is accurate or corrupt—is a necessary representation without which the Work cannot be experienced, but it is not identical with the Work, for no particular copy of the Work is needed for the Work to be experienced.

We see in this way of putting the problem the beginnings of an answer to the question, "What difference to a deconstructive reading of the Work does it make what Text is used?" For we see that the structure of the Work (that which the reader takes to be the work) depends upon the structure of the Material Text (the Linguistic Text in Semiotic form contained in a particular document). And we see further that the Material Text is not one unproblematic transparent "voice" but the "spoor" so to speak of a multitude of speakers working at various times and places combining their efforts in the Material Text. Which of these voices is to be centered in order that it may be decentered? Which self is to be focused on as self-subverting? The voice that is foregrounded by one particular Material Text may not be the same as that foregrounded in another. A deconstructive reader might not care for one reading more than another; indeed, he could not and remain true to his principles, but his understanding of the process of constructing meaning for a given text will depend to some extent on the particular Material Text he is exposed to. One can see, therefore, that a concept of authority underlies the way in which the Unities are employed to identify a "voice" or a "text" for the work by identifying an Utterance.

McLaverty, following the lead of James Thorpe[54] and Jerome McGann, places coalescence or Utterance of the work at publication, referring to stages of composition as moments of "pre-utterance." He is led to this, I believe, by the analogy of Theseus' ship. The parts of the boat removed and replaced are no longer the boat because they do not function as boat. And the boat does not function as boat until it is completed and launched. I suppose with boats, one knows if they function by putting them in the water. But works of literary art are not boats. The question of coalescence of a Work as a Version is a matter of opinion and judgment. Versions are not facts to be discovered about works; they are, rather, concepts created and put there by readers as a means of ordering (or as justification for valuing) textual variants. To say that an editor who had edited from the manuscripts a pre-publication version

(such as Hans Gabler's edition of Joyce's Ulysses) has prepared a "preutterance text" is to evaluate his work by a particular conjunction of the four unities and condemn it a priori. One may do that, but it is not a purpose of this taxonomy to encourage such judgments. Rather its purpose is to describe more accurately what judgments are being and can be made concerning the identity of copies of a work. Perhaps the most useful way to use the term Utterance is to say that each utterance has its intentions and its "social life" (i.e. the purpose for which it was released, and the context, moment and form in which it was designed and launched). Regardless of what decisions are made about the identity of a Version, when that tenuous issue is settled there remains the original question concerning works and versions: How accurately does the Material Text represent the Version or Work? Are there errors in the text?

All of this points, it seems to me, to an overwhelming conclusion concerning the concepts of Works and Versions—a conclusion that is consonant with the three fundamental propositions of this study. The concept of Work and, even more so, the concept of Version depend on Reception Performance just as much as on Creative Performance. If the reader must decide whether a Version is in fact a Version, its functional existence is determined by the reader. Creative performances, the idea of the Work as an Utterance of the author, and the idea of intended meanings are all Reader constructs. All the reader has is the Material Texts and whatever information about their provenance and alternative embodiments he has managed to scrounge up. The term Work helps him separate this mass of material from other masses relating to other works; the term Version helps him sort through this mass of material. He does so by classifying Material Texts according to the structure of Versioning that, to his perception, best accounts for the materials at hand. In short, the reader becomes the "functional authority" for the Work and its Versions. It would seem, therefore, that ideally the reader should have ready access to the evidence that would fully inform his or her decisions. Hence the importance of scholarly editions that foreground rather than submerge the evidence for Versions. T. H. Howard-Hill, contending that "uncertainty is a condition of mind and is not inherent in circumstances," recommends that "editors move with confidence and resolution like Tamburlaine," not like "Hamlets, shillying and shallying between this and that."[55] But if in his resolve an editor unwittingly stands in for all readers, making the decisions and producing a single reading text that purports to have reduced the Work to a book (the text that is the work), that editor has misrepresented the work, not refined and purified it.

IX. Conclusions

This exercise in naming leads, I believe, to a number of conclusions about literary works of art.

One is that the word Work conveys both a singular and a plural meaning. A work is one thing: all the Versions of Henry James's Roderick Hudson are subsumed under this one title. Simultaneously a work is a thing of internal diversity. It exists wherever a copy of the work exists. Each copy is a more or less accurate representation of one Version of the Work.[56]

Another conclusion is that attempts to repair or restore original or pure texts of a work or to revise and improve them tend to proliferate texts rather than to refine them. If one thinks of proliferating refined copies, one must remember not only that the "unrefined" copies have not been changed but that, by the unities of Material, Time, and perhaps Function, the refined copies represent new and therefore different Material Texts, complete with all that that entails.

A further conclusion is that the crucial act in relation to a Work of literary art is not writing, or publishing, or editing it, but reading it. Of course, without the first of these there will be no reading, but without reading the first seems incomplete or lacks fulfillment. Several observations about reading arise from this taxonomy. First, to read Material Text X is to decode a Work (i.e., that which is implied by its various Versions) from interaction with only one of its many static forms. Reading, therefore, is almost always a partial interaction with the Work. Second, if Material Text X is taken as a transparent window on the Work, there is no question asked about Versions or about errors. Third, if Material Text X is taken as the result of a single, prolonged production effort, subject only to human error, there is no question asked about Versions, just about accuracy. Fourth, if Material Text X is taken "for what it is"—one of many representations of a version of a work—there are questions of both accuracy and Version. Questions of Version include questions about the agents of change (author, editor, etc.), and about time, function or motive, and material. Material Texts are not, in other words, transparent.

Since editing and publishing tend to increase the number of copies of a work, not just in numbers but in variant forms, it seems useful to devise a graphic system to identify and categorize the Material Texts which represent the work (see Chart 3). Thus when person X reads and remarks upon Material Text X and person Y evaluates those remarks in

relation to Material Text Y, difficulties arise from several false assumptions: that both MTx and MTy contain the same Linguistic Text, that both are equivalent Material Texts, that a Linguistic Text as embodied in a document is a full rather than partial representation of the Work, and that the Work is represented adequately and equally well by any Material Text (or at least by MTx and MTy). Persons X and Y may disagree about the Work because they are not discussing a work but two unlike manifestations of the work. However, if X and Y understood the relation between the Material Text in hand and the Work, they might temper their judgments and remarks about it in the light of that understanding. Finally, if X and Y understand how each is developing a sense of Version by applying various mixes of the four unities, they might at least disagree with clarity about the issues in dispute.

For example, in the disputes between those who say the work of art is a social product finding its "true" form in the Material Text and those who consider the production process as unfortunately corrupting—but why be abstract? In the disparate views represented by Jerome McGann and Hershel Parker about the moment of coalescence for a work, McGann placing it in the Material Text and Parker placing it in the Linguistic Text at the moment of greatest creative control by the author, we have, I think, a disagreement that becomes clarified and a bit nonsensical. While many people have a gut feeling that "authorial authority" or Creative Performance is more interesting than "production authority" or Production Performance, the plain facts are that authors do some things badly and production does some things well.[57] If we take the view that the inscription which the author is finally satisfied to relinquish to a publisher is the closest representation of a Version, we are likely to take it as the basis for a new edition. But authors often show or relinquish manuscripts they know will be or must be changed. Would we be willing to say that the Essayed Version as embodied in the printer's copy (author's fair copy) is a form of the work which the critic can use as the basis for a "reading" of the work? Does the scholarly editor have a Production Performance task parallel to that given the work by its original publisher? It has seemed wise to say that the materials of the editorial project will dictate which answer is the most appropriate, but if that were true there would be no disputes. Disputes arise not only because, for example, McGann worked on Byron (who gratefully left the details of punctuation to those who cared and knew about such things)

and Parker worked on Twain (who claimed to have telegraphed instructions to have a compositor shot for tampering with his punctuation), but because they define Version and Work by differing valuations of authority and of the "unities." I suppose the final irony is that any edition Parker produced would be, after all, a new production performance and that any edition McGann produced would undoubtedly be read by many as establishing the author's intentions. In short, the problem is not one of editors' shillying and shallying over uncertainties in their minds, but, first, of a cacophony of voices "in" written texts to be selected from, and, second, of a world of readers who habitually treat books as if they fully represented the one voice that matters (each reader, of course, identifying that voice as seems right in his own eyes).

Another conclusion that might be drawn, tentatively at least, is that the idea of "conveying meaning" might be a misleading way to think about how texts function. The processes of encoding meaning (by authoring) and repackaging the coding (by publishing) and decoding (by reading) are perhaps too complex and fraught with "noise" to allow for "conveying," and our experience is rife with instances of meaning being apparently misconveyed (misconstrued is a more accurate and more frequent term, as it should be) either by accident or by deliberate appropriation. This taxonomy suggests that texts influence, rather than control, Reception Performance. All the work of Creative Performance and Production Performance is ostensibly geared toward influencing Reception Performance.[58] The only chance that an author has to influence the Reception Performance is so to arrange the Linguistic Text that he will have the best chance possible of influencing the reading and thus be said to have been understood rather than to have been misconstrued. The Reception Performance is, however, influenced not only by the Linguistic Text but by a great deal besides, much of which is subconscious and fortuitous. When the Reader has produced a Reception Text, its coherence is usually considered satisfactory proof that the performance has succeeded. Dissatisfaction with that coherence can only come when a second reading or someone else's description of a reading appears more satisfyingly coherent. There is, of course, no way to verify any correspondence between Reception Performance and Creative Performance.

All of this seems to confirm a conclusion bruited among some literary theorists: that the community of scholarship (or any community of

readership—our sense of cultural heritage and values) derives its power and cohesiveness from arbitrary agreements to use certain conventions as standards of behavior regarding the interpretation of works and the relevance of history and perceived hegemonic structures in commentaries upon literary works. All of these conventions and standards are convenient constructs, not natural truths, and are deemed convenient as long as we agree to find them so. I no longer find it convenient to consider the Material Text an original, stable, or transparent sign source for an entity called the Literary Work of Art.

APPENDIX A

Occasionally textual critics have referred to theories from related disciplines that seem suggestive as clearer explanations of some aspect of textual criticism or of some problem relevant to editorial practice. The following is intended as a tentative "commentary" on two of them: J. L. Austin and Michael Hancher.

J. L. Austin:

Michael Hancher:

It will be noted that these schemata treat the Linguistic Text as a totally transparent entity. That is, all the analysis is expended on intention to mean something and to cause effects on readers. They do not consider the Material Text as a focus or object of intention itself. They do not describe the author's intention to write a paragraph or spell a word or use a point of punctuation. The technical construction of the text is not seen as problematic. The result is that little attention is given to the particular characteristics of the Material Text in hand being subjected to analysis. Further, none consider the work of book-designers (type fonts, page formats, density of lines, width of margins, quality of paper and binding) or of price and methods of marketing

and distribution as "intentions" of Production Performances or as influences on the total "perlocutionary impact" of works.

APPENDIX B
Definition of Terms

Texts as Matter:

• Material Text = The union of Linguistic Text and Document: A Sign Sequence held in a medium of display. The Material Text has "meanings" additional to, and perhaps complementary to, the Linguistic Text.
• Document = The Material medium of display (paper, ink, etc.).
• Protocol = A written or otherwise verbalized response to and commentary on a Work.

Texts as Concepts:

• Work = That which is manifested in and implied by the material and linguistic forms of texts thought to be variant forms of a single literary entity. The term Work incorporates concepts of Versions (Ideations) which are made concrete or Material by a Production Performance and then reconceptualized by Reception Performance.
Version = A concept by which Material Texts (such as manuscripts, drafts, proofs, first editions, revised editions, etc.) are classified as representative of:
• Potential—abstract conceptual texts in the mind of an author.
• Developing—abstract conceptual texts as evidenced by trial drafts in some material form.
• Essayed—finished (at least temporarily) versions as evidenced by completed manuscripts or revised texts.
• Linguistic Text = A Sign Sequence for an Essayed Version displayed in a Document.
• Reception Text = The Performed Text conceptualized by the reader in the act of reading; the decoded Material Text.
• Conceptual Text = Any text that is "held" in the mind or contemplated by a person. Conceptual texts are the only kind that can be experienced, though Material Texts are where they begin.
• Semiotic Text = The signs used to represent any given Linguistic Text.

Texts as Action:

• Speech Act = The whole event of creation, production, and reception of a communication at a specified time and place.
• Write Act = The complex, never closed, serial event encompassing the creations, productions, and receptions at any and all places and times in which a written work is created, produced and received.
• Utterance = A whole Speech Act or a coherent selection of "speaker," medium, hearer, time(s), and place(s) employed with regard to a Write Act within which "an understanding" of the Work is achieved.
• Creative Performance = The authorial development of Essayed Versions

  ---

  82

  
  
  
  resulting in Linguistic Texts as found in manuscripts and authorially revised texts.
• Production Performance = The scribal and publication development of Material Texts resulting in typists' copies, proofs, and printed books.
• Reception Performance = The development of a conceptualized Reception Text in the act of reading.

Texts as Units, the integrity of which is defined by:

• Time—The Work as it existed at some significant time.
• Content—The Work as represented by one Linguistic Text as opposed to variants of it.
• Function—The Work as designed for a particular purpose or appearance.
• Material—The Work as embodied in one or another particular physical format.

Textual Contexts:

• Material Base = The world of sense data to which the Material Text (manuscript, book, etc.) belongs.
• Social Context = The complex of institutions, conventions, and mores whose expectations and habits are reflected in the Material Text.
• Performance Field = An ambiguously conceived abstraction that provides the illusion of stability or a "locus of meaning." It is where the Text is "played" according to the rules of the reader's particular game of textual interaction and limited by the performer's capabilities and resources. The "game" metaphor here accounts in a different way for the "play" or looseness in the "machine" model of interpretation, in which play is equivalent to tolerance, a concept about how much variation in interpretation is allowable before communication "breaks down."

Notes