University of Virginia Library

ORIENTATIONS TO TEXT, REVISITED

by Dirk Van Hulle and Peter Shillingsburg *

Our initial purpose was to add a "genetic orientation" to the list of orientations presented in Shillingsburg's Scholarly Editing in the Computer Age, 3rd ed. (1996): documentary, sociological, authorial, bibliographic, and aesthetic. How-ever, in working through them to add a genetic one, we discovered ways to improve the original design. We offer here a re-orientation to orientations to text.1

Definitions and a Rationale

The words "text," "work" and "document" are often used interchangeably. "Text" in our essay means the sequences of letters, spaces, and punctuation that do or can represent a literary work in part or in full. The text of a document, when reproduced accurately, creates a new document with the same text; but a single document can have two or more texts, initial and revised, for example. A work is represented more or less well by each of its variant texts.

Depending on one's orientation to texts, one values all texts equally or some over others. The variant texts of a work, and the form, nature, and history of its documents, constitute a complex array of facts, offering opportunity for development of inferences about the history and condition of the work, and offering support for variant critical understandings of a work. The particular way a textual scholar organizes and values complex textual factors-the scholar's orientation-affects how the work is understood.

The elemental materials, facts, and forces involved in the original production of literary works and then in their revision, reproduction, and dissemination are material, causal (agents), temporal, genetic (inventive), and performance. Another element now seldom encountered in scholarly textual discussions is aesthetic / commercial—normally excluded because it is often confused with "flavor of the month" value. An orientation identifies a perspective that reveals the relative importance of these elements for a given purpose. None can be neglected but not all are of equal importance to each reader or editor; hence the varieties of approaches to scholarly textual tasks and disagreements about how to understand

a work and how to edit it. We will discuss each orientation in section 2, below, to show how it affects what scholars understand to be the nature of texts.

In addition to shaping the way a work is identified and understood, orientations guide what we do with textual materials: differentiating ways to edit, and determining how the narrative of composition, revision, and publication is framed. An orientation to text, in addition, determines what constitutes relevant materials to consider when tracing the progression of textual changes from source texts through drafts and revisions of scenes or sentences to produce scholarly accounts other than editions, such as genetic studies or arguments to support critical interpretations.

A brief summary highlights the most significant differences among orientations. (1) The material orientation tends to focus on extant documentary material evidence asking how to represent it in an orderly fashion. It has two subsets: lexical and bibliographic. The purpose is primarily archival. (2) The causal orientation focuses on the agents responsible for the condition in which we find the evidence and on their authority to effect textual change. It divides into two subsets: authorial and social. (3) The temporal orientation focuses on when the text came into being, asking if there are aspects of the text that do not fit the period of its creation. It also divides into two subsets, whose purpose is either to establish synchronic snapshots of stages of textual completion or to trace the process of textual development diachronically. (4) The genetic orientation focuses on how textual invention happens as a creative process, including the effect of non-linguistic influences and byproducts of writing. (5) The performance orientation searches through and beyond documents to determine the staging of plays, the performance of musical works, and public or even private readings. (6) The aesthetic/commercial orientation tends to ask how the text can be improved to fulfill the wishes of an author, or a corporate aim at a historical moment, or to achieve current commercial or otherwise public purposes.

It may seem odd that we do not include a "linguistic orientation," but we think that language is constitutive for all orientations rather than an element that can be separated from other orientations. Even if one turns to A. E. Housman's "The Application of Thought to Textual Criticism," or to the most discussed aspects of George Kane and E. Talbot Donaldson's editing of Piers Plowman2 where knowledge of grammar and prosody substitute for knowledge about the author's personal habits or capacities, we find that linguistic elements turn out to be aspects of other orientations—constrained by material documents, and a sense of time and place, and/or of the personal agents of textual change.

Textual scholars reviewing this list of orientations will likely say, "We are interested in all of them." But we see a distinction between interests and orientations. What, for example, does the following hypothetical scenario tell us about the editor's concept of the work? Imagine an editor facing a work that survives

in four authoritative forms: a manuscript, a set of corrected proofs for the first edition, the first edition, and a revised edition. Many works in the modern era are more complicated than that, but this suffices to reveal what may be a problem. Say that the editor chooses as copy-text a printed form, the first edition, justifying the choice on the grounds that a social contract called for the author to "submit" (i.e., surrender) his work to a production process, and explaining that though the work was later revised, the new edition is designed to represent an early moment in the historical development of the work. But then the editor finds that an error in the manuscript was repeated in both the proofs and the first edition, but was corrected in the revised edition; the editor emends it on the grounds that the work could not have been intended to have an error, even though this one persisted in the first three forms of the work; the editor thus "furthers" the acts of socializing the text by accepting the revised (post-copy-text) edition reading. The editor then finds that the manuscript has a reading (an okay reading), but the compositors changed it in the proofs to something better (in the editor's opinion); the author, however, stubbornly marked the proofs to restore the manuscript reading; the editor reasons that the socializing process takes precedent over the authorial view because the editor likes the compositor's reading better or because of a sense of duty to follow the copy-text. However, the editor then notes that in the process of normalizing the punctuation, the compositor has run roughshod over some delicate rhetorical pointing, changing both the cadences of the prose and the meaning of the sentences; the editor decides to restore the manuscript here and there on the grounds that the socializing process occasionally did the authorial text a disservice. Finally, the editor decides that a few arcane or archaic forms of words, spellings, or punctuation would confuse readers, and so these archaic forms are slightly updated and regularized.

This hypothetical editor had a variety of interests in the work; and by listing the emendations and providing a history of variant texts, the editor enables other persons to see the text differently. But, is it okay to shift the grounds for emendation as one or another of the editor's interests arises? Although the hypothetical editor in this imagined scene is an eclectic editor, the notion of "orientation drift" can also mar the clarity of thought that anyone brings to the texts that represent a work. Even when editors reject eclecticism, their reasons for doing so can shift from one orientation to another and thus cloud the reasoning that justifies their own choices. We think that the concept of orientations and their subsets provides an ordering of interests so that a consistent approach can be taken to textual investigations. An orientation does not eliminate interests, but it does provide a consistent perspective from which to organize and explore one's interests. Its usefulness lies not in isolating and approving one approach over another, but in clarifying the elements that go into decisions about the text.

In the following section, we treat the second, causal, orientation at greater length in part because of recent decades of dispute about the role of persons and intentions in editing and in part because we wish to argue for the close connection that we see between traditionally disparate "authorial" and "social" editorial agendas. We also treat the fourth, genetic, at greater length, because it is a new

orientation. The relative proportions of the six discussions do not represent any preference or value judgment.

Orientations

1. Material Orientation

The attraction of the material orientation derives primarily from the fact that documents, regardless of one's orientation, are the "bottom line" or primary evidence for works. Efforts to "get behind the evidence" or to "sort out priorities among documents" involve critical analysis, inference, and argument, which introduce dispute and opinion. Since all documents are manufactured objects, subject to the human foibles of creation and production, any new document would, by that logic alone, take its place along with already extant documents as a document with a potentially variant text. If the new document's text is somehow inferior (or superior) to the historically extant documents, then mere documentary existence is not the only essential element for scholars who use the material argument. It is something else, such as recognition of the importance of one or more of the other textual elements, or some concept of authority that asks of the new document, "By what authority do you claim to be primary evidence?" This question must then be asked of each document including drafts and manuscripts. Or differences in the value of one document over another could be related to temporal priority. Or value might arise from the particular agents who created the more valuable document. For most documentary editors there is something beyond its status as document that raises the value or authority of the text in an historical document above that of a new text created with the tools of modern scholarship.

The material orientation is usefully divided into two subsets: (a) lexical and (b) bibliographic. Both are depersonalized approaches to the document. Neither asks, "Who did this?" But, interestingly, only the lexical subset allows, logically, for editorial work.

(a) The lexical approach distinguishes between the document and the text far enough to allow the text to be replicated but, usually, not emended because the "lexical text in the document" is the ultimate textual evidence.3 We call it lexical because what usually indicates to an editor that there is a flaw to be pointed out or corrected is its violation of lexical conventions. But authors who are known to have deliberately violated those conventions (famously Joyce, for example) pose serious difficulties for determining what is an error. The lexical text as found in a document is a historical fact—replicable, but only minimally emendable under this view.

(b) The bibliographic approach4 includes both the visual or iconic aspects of documents (which can be reproduced photographically) and the tactile or physical

structure (which can be photographed but not reproduced).5 The bibliographic approach logically allows neither replication nor emendation.

Both lexical and bibliographic are documentary in that they consider the physical document to be the basic unit of textual evidence, but where the lexical focuses on the text, the bibliographic focuses on the material object. Replication of any material aspect of a work in a new edition would entail new invention and, hence, failure to reproduce an essential aspect of the original object. For the bibliographic approach, facsimiles can produce similar effects and give some notion of the effects created by the original, but any emendation distorts the physical, historical record.

2. Causal Orientation

The idea of an authorial orientation became a particularly controversial editorial approach in the 1980s. By renaming it the causal orientation we wish to focus attention on the personal involvement of all agents of textual change. Unlike the material or temporal, the causal orientation focuses attention on the author and on other persons involved in the writing, revising, and production of texts. In determining who "did the text" and by what authority the text was created or changed, the documents are treated as merely the surviving evidence and the "moment in time" as a mere concept about a finished product's "moment of completion"—both used to help address the question of textual integrity. Literary works are intentional creations, using the medium of writing in attempts to achieve desired goals. The causal orientation's focus on human action usually acknowledges human fallibility and often seeks ways to identify and mitigate the results of fallibility. However, this orientation also has a rich tradition of facing the pitfalls awaiting those who try to identify and emend lapses in "intentions." And yet, no one disputes that authors and others have or had intentions. Nor does anyone dispute that intentions have physical and lexical consequences in the inscription or transmission of texts. The problem is that the evidence pointing to intention is almost always problematic because speculation about it is difficult and often impossible to verify beyond dispute.

One way to avoid the appearance of attempting the impossible is to shift the emphasis from intentions to effects. Asking "what difference is made by the change?" seems more accessible and may initially seem superior to asking "what

was the intention of the person(s) who made the change?" This shift in emphasis postpones or avoids the question, "whose intentions are being considered—whether the author's, the publisher's, or the reader's and editor's?" The shift also goes well with a second way to avoid the pursuit of intention, which is to say that the work we are really interested in is the product of a social collaboration and so we do not need to distinguish editorially between flawed and intended texts. The important issue is that a social complex was involved.

On the other hand, "effects of the change" is also ambiguous. If effect refers to how a reader is affected by the change, then the same problem attends assessments of the effects of textual changes as that which troubles the pursuit of intentions, for though a scholar can vouch for the effects a change has on himself or herself, it is difficult to verify that these effects were perceived or felt by other readers. For our purposes, "intention" is restricted to the intention to make a textual change, and "effect" is taken to mean the resulting textual changes—both evidenced by marks on a document and variants in inscriptions. Editors do provide narrative explanations for the effected change, and readers can dispute those explanations. It is important that textual critics restrict their speculation about intentions to the interpretation of ambiguous or illegible or highly improbable inscriptions. For the textual critic, it is a question about what the author intended the text to be—what mark was intended to be inscribed. It is not about what the author intended the text to mean or how the author intended the text to succeed—though, of course, it is impossible to separate completely the questions of what the text is from what the text means. Textual critics agree with literary critics that intentions to mean or intentions to affect are both beyond our ability to recover indisputably; and, even if we could recover them, they would not restrict our own exercise of critical judgment about the success of such intentions.

The causal orientation divides into two subsets for understanding and using textual material: (a) the authorial and (b) the social. Both the individual authorial and the social subsets see the text as a deliberately (or inadvertently) agented work which requires that students and editors address the work, not primarily as documentary or historical, but as a record of human actions. We think of the causal orientation as consisting of a continuum between an authorial approach and a social approach. At any given time and in any given document we might find the work of more than one agent of textual change, with unequal authority or skill to make changes. Should we value the work of one agent over that of others, or should we value all agents equally? It is the authorial end of the scale that is the most difficult and controversial. The social approach tends to accept the work of all who are involved in the production of each form of the work as if the work were a collaborative social and cultural endeavor among equals, whereas the authorial approach seeks to attribute the text to one or more agents whose work must be distinguished and assessed separately—agents of varying skills and authority who participated in creating and changing the text(s). Agents of textual change can fail in palpable ways in their attempts to record the text of the work as the text of a document. The text as achieved by one agent of change (say the author) may differ from the text as revised or as rendered by other agents of change (such as secretaries, editors, and compositors) each of whom was fallible (including the author) in ways that can sometimes be detected. The

causal orientation asks us to choose whether to focus on the text created by and/or desired by the author or on a text created or desired in concert with one or other or all of the production personnel. Further it asks us to decide whether to focus on the text actually achieved or, if there is sufficient evidence, to discover texts attempted but not actually achieved.

In judging or assessing the variant texts, it would seem legitimate and perhaps even necessary to judge the competence and authority of each agent of change. Are all the compositors at work on a Shakespeare text equally competent—whether apprentice, journeyman, or master typesetters? Is the publisher's editor as good as or a better judge of grammar, spelling, or public tolerance for edgy language than the author? Does the censor have the same kind of authority over the text as has the author? Can the scholarly editor restore the authorial reading when either or both the publisher and the censor have overruled the author?

(a) An authorial approach would tend to favor the work of the author over that of production personnel, but there is no reason that the author's competence should not be questioned as well as that of other agents of textual change.

(b) The social point of view, like the authorial, focuses attention on the agents of change because this view holds that literature is a socialized object. Textual agency consists of many people working in concert to make and distribute a final product. It can be argued, however, that it is not just an increase in the number of people allowed to be agents of change, but rather that the social approach entails a radically different notion of textual authority, one that exists under an agreed contract entered into, usually but not always formally, by the author and a representative of the production process—the publisher. In place of the authority of creation or authorship or of intellectual property, it is the social contract that exerts its authority over the text. Furthermore, it can be argued that this point of view respects institutions and guilds or traditions as constitutive elements in the production of literary works. As is being increasingly demonstrated by book historians, the industries that contribute the materials and skills, the laws that influence industrial practices through taxation and regulation, the laws of censorship, the customs of decency and permissiveness, and technological developments, to name a few, all have an influence on what writers write, what publishers publish, and what readers read. It follows that a social view of the literary work must deal with the actual products of these influences in order to engage with and understand the social dynamics that influenced a work.

In accepting the social product as the object of choice, the social approach ends up with the same "textual condition" shared by the bibliographic view: texts that cannot be edited without obscuring or contaminating the social elements of the historical object. A new edition is, of necessity, the product of a new social dynamic, not a replication of the one claimed by the social "editor" to be the object of interest. The logic of the social orientation, like that of the bibliographic, requires one to deal directly with original materials, and yet much of this type of criticism is based on derived editions and facsimiles—to the detriment of the scholarship.6 Furthermore, if all the agents of change are treated as

equal authorities over the text, such that the work is seen as "co-authored" by all of its agents of change, much of the power of this point of view is vitiated. The result of the authorial approach, on the other hand, is more often than not an eclectic edition.

A strict adherence to the dictates of either the social approach (one blanket rule covers the actions of all agents of change and we accept the results as a fact of history), on one hand, or the authorial orientation (excluding the actions of all agents of change save those whose ministrations served the author's intentions), on the other hand, might lead to unsatisfactory distortions of the craft of textual criticism. Understanding what the artist has achieved and how that achievement failed or succeeded in the production process requires that we pay attention and respect to both aspects of the causal orientation. Anyone who has studied the genesis of a work of art knows that authors sometimes welcome and sometimes deplore the help they get. Blanket rules and rigid applications of them would seem to undermine the potential achievements of textual criticism.

It could be that the best compromise, particularly in a print edition, is for the text to be edited to reflect the work of art, as an artist's work, accepting the intervention of others only when it supports the artist's conception, rejecting it when it subverts that conception; and to then construct critical, historical, and textual supplements to account for the economic, political, and industrial complex of production, showing the history of that art in the marketplace. This is a compromise well in keeping with Anglo-American editorial traditions, where the object of editorial work has always been the production of a well-edited text and provision of an apparatus that supports and explains the editorial approaches taken to develop the newly edited text.7 But it is a compromise not likely to be acceptable to the historical-critical editorial practice often associated with German traditions. The reason is understandable. The compromise between the social and authorial subsets of the causal orientation presupposes that an editor is to adjudicate among the archival textual elements that vie for the reader's attention in order to produce a clear reading text (a new, never-before extant text), with an apparatus that helps the reader see the work as the editor saw it. By contrast, in the historical-critical tradition, the object of textual criticism is to bring order to the historical evidence by representing the history of the text—not by interfering in the textual evidence to somehow rectify the aberrations of history. In the historical-critical tradition it is the purpose of textual criticism to collect and give coherent arrangement to the record of historical evidence. The narrative of that history is presented both in introductions and collations.

Information about the sources of changes should be given where possible, but adjudication among competing authorities about what the text "should be" is not admitted by historical-critical editors as part of the textual critic's remit. In fact, the historical-critical approach follows the material orientation, not the causal. By contrast with the Anglo-American tradition, the historical-critical edition has devoted its efforts to creating a comprehensive, coherent textual apparatus which is attached to a historical (lexical) text accurately presented. These are precisely the differences between the material and causal orientations.

3. Temporal Orientation

The temporal point of view abandons "documentary integrity" as the prime or essential element of text, opening a space between "document" and "temporal stage of development" for the idea that the "text of the work" and the "text of the document" might not be the same at any particular moment in the development of the work; the document can thus mis-represent the moment of its inscription. The temporal approach is often less interested in the agent of change than in the cultural moment or stage of development. In the case of classical and medieval works, where the surviving evidence often post-dates composition of the work by decades or centuries, the concepts of "period of origination" or "period of inscription" enters more plausibly than "moment in time"—editors are often keenly aware of anachronisms in scribal copies. And yet for works surviving in an array of authorial or authorized versions a sense of the moment in time could give way to ideas about duration through time, where instead of asking why a change was introduced one could focus on the question why some text was replicated without change? For works represented by compositional and revision materials, the temporal view might locate the work in a developmental path from inception to final or last form of the work either (a) as a series of production endpoints or (b) as a continuous scale of genetic development.

(a) The productive approach sees the work as a series of static or snapshot efforts to capture the work to be studied synchronically at specified moments in time, usually as fixed, more or less well, in some document.

(b) A diachronic alternative sees the work as a creative development through time always progressively approaching (not necessarily reaching) a final endgame so that stages along the way are just provisional. Since for both of these views the "moment in time" or "sequence of development"—not the document itself—is the central issue, it follows that the record of that temporality, as captured in the document, is potentially faulty and therefore liable to correction or emendation. Here Pietro Beltrami's notion of the "witness text" enters to suggest that for any given time, the nearest document might be flawed.8 Here, too, Tanselle's distinction between the text of the work and the text of the document performs the same

office.9 The temporal orientation, like the documentary, is depersonalized—does not discriminate primarily among those persons responsible for the text. The primary question is what was the text supposed by its producers to be at some moment in time?

4. Genetic Orientation

The genetic orientation offers special opportunities and problems. In Daniel Ferrer's terms, genetic criticism can be defined as the science of written invention ["la science de l'invention écrite"]10; it focuses on "invention" rather than on "repetition."11 Keeping in mind that an orientation organizes what we understand about the factors that produced the documents and guides what we do with textual materials, the special focus that serves this organizing purpose in the genetic orientation is creative invention. Genetic criticism is applied to various art forms, including film, music, architecture, visual arts, as well as to the creative act in science. When applied to literature, the raw material with which the artist works to creatively invent a fictional world is language. However, the composition process is not exclusively linguistic in nature. Drawings of characters, sketches of the setting, schematic representations of the plot, doodles, etc. are important pre- or metalinguistic elements in the process of creation. Still, any literary project entails linguistic composition in some form or another.

The genetic orientation has a natural affinity with the documentary and temporal orientations, for it focuses its attention on the documents as sources of evidence of textual development and change through time. It also has an affinity with the causal orientation for its purposes require assessment of changes in accord with the individual agents of change. In particular there is a link with the social approach, for the writer may ask a friend, a wife or husband to copy the manuscript; a publisher's editor prepares the manuscript for typesetting; the compositor may misread the manuscript and cause the author to revise the misreading to something new; a censor may delete a passage; the genesis of a work may continue after the first publication and often involves more than one agent. From the viewpoint of the social subset of the causal orientation, for which authority resides in the institutional unit of author and publisher, this multiple agency is an aspect of the past that needs to be secured; from the genetic viewpoint, the discrete acts of invention by discrete individuals must be unpacked for study. So, in spite of the affinities, there is a difference in focus, also with reference to the material and temporal orientations. Whereas for a material orientation, authority resides in the historical document and for a temporal orientation it resides in periods of inscription or re-inscription, the genetic orientation's focus is on the actions of invention as implied by the chronological succession of documents and the transformations within and between them.

A genetic edition, according to French genetic critics such as Almuth Grésillon,12 contains the complete facsimile reproduction of the documents pertaining to the genesis of a work. There is, however, a variant form which is referred to as an "édition critico-génétique," rather than an "édition génétique." This type of edition tends to present the transcriptions of the manuscripts at the service of a critically edited text ne varietur. The link with this edited text bothers committed genetic critics. After all, one of the merits of genetic criticism is its suggestion that the study of modern manuscripts does not need to be reduced to a subservient role at the service of the establishment of an edited text ne varietur. Whereas, in scholarly editing, manuscript analysis is often seen as a means to an end, that is, a tool to make an edition, genetic criticism reverses these roles and sees the making of an edition as a tool to facilitate manuscript analysis.

The publishing market's demand for a single reading text is hard to reconcile with a theory of the fluid13 or mutable text, but such a reconciliation is not impossible, as for instance John Bryant and Haskell Springer have shown by interrupting the reading text with 'revision narratives' in their edition of Melville's Moby Dick).14 Even though many genetic critics tend to insist on the difference between textual criticism and critique génétique, it is possible to establish a rapprochement because the definition of orientations to text is not necessarily geared towards the production of an edited text, but conceived as a way of identifying how we read the materials in order to understand the work.

A genetic orientation focuses on the trajectory or trajectories of creative development, including sources of inspiration. It does not have to, but may involve a form of scholarly editing. If it does, it implies an editorial strategy that displays creativity in motion rather than settling on a final version as the main object of editing. It aims to produce a narrative of genetic development. And unlike many scholarly editions, the genetic orientation includes source texts and their transformation into the work as a part of the process of invention. It comprises all the stages and facets of a work's genesis, for which Raymonde Debray Genette15 coined the terms "exogenesis" and "endogenesis," and to which a third category could be added: "epigenesis."16 These concepts will be discussed in the following paragraphs.

(a) Exogenesis. The work's genesis often starts in the margins of other authors' books. This link with external sources of information is referred to as "exogenesis." The stereotypical river metaphor, suggesting that a text developed from a single source, or went through a sequential and orderly set of changes to reach

a final product, requires deeper exploration, for many creative histories have more resemblance to a complex fluvial system arising from many tributaries and branching into a delta of many outlets. To facilitate manuscript research, a genetic edition could offer an author's personal library, where it is still extant. It is evident that this type of edition benefits hugely from new possibilities created by digital media. Since the limitations of print demanded that (long) works be represented by only one text at a time, the establishment of a single reading text was presupposed in the original definition of orientations to texts (although notable exceptions were taken into account, such as Wordsworth's two Preludes or Shakespeare's King Lears). Since "the business of editing" was seen by many editors of print editions as "the elimination of error and the selection of the authoritative readings for the new text, which supersede all other authoritative alternatives,"17 a selection needed to be made by attributing authority, located variously according to the editor's critical orientation. This selection can still be very useful, but in the meantime, electronic forms of presentation have eased the limitations so that it is possible to present several edited texts alongside each other, and combine them with digital facsimiles and transcriptions of all the paralipomena, notes and drafts preceding the first publication, including marginalia and reading notes.

(b) Endogenesis. The genetic orientation's special fascination is with documents that bear the evidence of multiple efforts, often by more than one agent, over extended periods of time. It is less absorbed in the documentary aspect than in what can be extracted from it about the sequence of discoveries, inventions, and actions of the writer(s). Again, the digital medium offers new possibilities to produce a display of textual history for a work that will help focus attention on the acts of writing—the acts of invention—so that they can be studied. This genetic textual instrument does not substitute for other kinds of editions but serves as a tool for studying the work in the genetic critical way. At the level of the endogenesis, a rapprochement between scholarly editing and genetic criticism can be mutually beneficial.

From an editorial perspective, the focus on invention underlies the rationale of a genetic edition, designed to show process but taking product into account as an often crucial element in the dynamics that drive the process of invention. Even if a particular writing project was eventually not finished, the author usually did need a projected end to write toward, in order to start and continue writing in the first place. The difference of merely one letter between "invention" and "intention" is deceptively small.18 Writing a text generally implies an intention, but the study of manuscripts usually reveals a plurality of intentions. Similarly, the notion of invention often implies a succession of inventions. But the main difference

is that the notion of invention is directly related to the traces of this creative act on the extant manuscripts, whereas the link between an intention and the materialized result is somewhat less direct and it is notoriously hard to figure out if or to what extent this result corresponds with an intention. The documents, warts and all, bear witness to the efforts of an agent of change (usually the writer), in the first place, as a reader of him- or herself and as an active participant in a process of rewriting. Interpreting this process of invention constitutes the "critical" component of critique génétique. The "genetic" component focuses on the genetic dossier, which can now be conveniently presented in a digital environment, as the recent addition to the Text Encoding Initiative's (TEI) guidelines for genetic editions testifies.

From a genetic perspective, a shift of emphasis from motives to effects also involves an attempt to regard variants not necessarily as textual elements that necessitate emendation or purging, but to understand them as elements of inventive development. For that reason, French genetic critics often prefer to speak of "rewritings" (réécritures), instead of variants.19 The argument is that one can only speak of "variants" if there is an "invariant" against which they can be compared. Some authors' writing methods, such as Marcel Proust's, proceed in relatively large textual units that were shuffled around and for which a hypertextual electronic architecture might be a suitable form of presentation. In such cases, it is indeed difficult to work with the notion of a variant. Other writing methods, however, do show patterns of relative stability in terms of the text's structure. In these cases, one does not need to insist on a rejection of the term "variants." Some works' genetic dossiers do consist of versions that can be compared, not merely in general terms, but even at the level of small textual units such as a segment or sentence. In such a case, it is possible to speak of compositional variants, since every version can serve as a temporary invariant for comparison against the subsequent version. Such a relative calibration can be applied in an electronic genetic edition by identifying segments that correspond between different versions, allowing the user to compare a segment in one version to the same segment in another version or in all the other versions simultaneously (synoptically). In such a synoptic survey of all the extant versions of a small textual unit (for instance a sentence), the syntactical context of each segment can remain intact, which results in a form of "versioning" that allows for detailed comparison of stages in the writing process.

Depending on the author's writing method, for instance in the case of Samuel Beckett's late writings (www.beckettarchive.org), it is possible to allow for a form of digitally supported collation20 by highlighting the variants between versions. If a manuscript or typescript shows clear signs of different writing tools or different handwritings, a version can be subdivided into several stages, which would allow for an even more refined collation of compositional variants or réécritures.

(c) Epigenesis. The genesis often continues after publication. About changes made by the author after publication, Pierre-Marc de Biasi once noted: "How-ever

ever important, these modifications (or "variants") [...] do not have exactly the same status as the transformations to be observed in the genetic documents" because "the mutations of the avant-texte took place in a private writing domain," whereas "postpublication modifications are made in a public sphere where the book's reality cannot be ignored."21 In broad outlines, such a division between an individual authorial creative trajectory and a social production trajectory may be useful, but often the transition from "private" to "public" sphere is marked by a continuum rather than a clear boundary. Before publication, an author may "try out" a particular chapter by reading it to a circle of friends; or the political climate may be so dangerous that it enters the private sphere in the sense that it may lead to self-censorship during the composition process, to the extent that certain creative ideas do not even reach the stage of the roughest draft.

In 1972, while genetic criticism was only just establishing itself in France as a discipline in its own right, James Thorpe already drew attention to "the status of multiple versions of a work of art," referring to important examples, such as Valéry's urge to keep modifying his texts and Yeats' "drive to revise."22 Four years later, Thomas Tanselle suggested that "two types of revision must be distinguished: that which aims at altering the purpose, direction, or character of a work, thus attempting to make a different sort of work out of it; and that which aims at intensifying, refining, or improving the work as then conceived (whether or not it succeeds in doing so), thus altering the work in degree but not in kind."23 The first type was called "vertical revision," the second "horizontal." In some cases, it may be hard to decide whether a modification alters or intensifies the work, i.e., to make a distinction between "vertical" and "horizontal" revisions, but what is important to note in the context of textual orientations is that this distinction does not coincide with the distinction between endo- and epigenetics. The latter distinction corresponds more to the example of Tennyson, which Jerome McGann refers to in A Critique of Modern Textual Criticism: "This was a poet who frequently revised his work on the basis of the responses he received from a small circle of friends (at the trial proof stage) as well as from reviewers and the larger audience (at the publication stage)."24 The first type of revisions is endogenetic, the second type is epigenetic; in both cases the modifications by Tennyson were inspired through confrontation with an audience.

In his typology of genetic documentation, Pierre-Marc de Biasi ("What Is a Literary Draft") gives due prominence to the moment the author decides the text is ready to be printed and presented to the public (bon à tirer). By the time of the bon à tirer moment, however, the initial target may have shifted during the process of writing, and once the target is reached, the process of writing is

not necessarily finished. If the author is a self-translator, for instance, the work becomes a "source text" that is translated into a new target, which in translation studies is referred to as a "target text." In the case of playwrights who direct their own plays, their practice as directors may lead to additions or cuts during rehearsals. Almuth Grésillon drew attention to the special treatment of "genèses théâtrales"25 and she wrote a foundational article with Jean-Marie Thomasseau in a special issue of the journal Genesis devoted to the topic.26 Thomasseau, who focuses on non-contemporary theatre, duly points out that "non-contemporary theatrical genetics discards the idea of a single author by focusing as much on the process of transforming a play, and on those who perform it, as on the author who wrote it. It begins with the recognition that in the theatrical process the authorial role necessarily fractures and becomes shared."27 The genetic study of theatrical performances involves different kinds of documents, such as visual and aural recordings of rehearsals, director's notes, lighting designs, prompt books, actors' notes, set designs, sketches, scale-models, stage managers' notebooks. These traces may be of help in analysing the dynamics both of the theatrical performance and of the dramatic text.

The epigenetic changes require an adequate editorial approach that treats variants as réécritures. For although scholarly editors may frown at de Biasi's observations on the term "variants," the term réécritures does invite them to look at variants, not only in terms of emendation, but also as elements of inventive development. On the other hand, genetic critics should also realize that, to some extent, the rejection of the term "variants" was part of the discourse of the early days of genetic criticism, when it strongly felt the need to distinguish itself from textual criticism. In the meantime, genetic criticism has clearly established itself as a discipline and we have arrived at a point of mutual, interdisciplinary respect for, and openness to, each other's different priorities, techniques, and approaches.

5. Performance Orientation

It is fairly easy to see that plays and music and also dance notation and architectural plans are less than complete until they have been transformed or "interpreted" as performances or transferred from their notational (textual) conceptions into their fulfillments in events or buildings or landscapes. Thus, if the endgame of the work is seen as a performance or as the construction of something for which the text is a set of instructions, a recipe, or a score, it is likely that the drafts, fragments, plans, and printed texts for the work will be looked

at in a radically different way from that which is adopted when a reader sees in a text a means of experiencing the work by merely reading it. The line between these two modes of textual function is not always very clear, however, because a poem might be seen as something to be read aloud, making it more like a song than a physical or symbolic object to be contemplated for itself. But beyond that, it seems plausible also to speak of a novel as an "object" that can only be experienced through the performing act of reading, even though silently to one's self. The point, however, of these observations is that a performative orientation toward the textual materials for a work might focus attention on their value in helping us to imagine how the work has been performed in the past or in helping us to imagine how the work could or should be performed now.

The key word in performance is "enactment."28 What distinguishes the performance orientation from others is its focus on how a work's instructions for performance are carried out. For texts that are "merely read," performative acts are individual and vary from one reader to another depending on skill and reading style. Editors can have little to do with that kind of performance, though historians of reading find an endless fascination in tracing evidence of reading habits. In this vein, reviews of works can be seen as reports on reading performances, but again editors may be justified in ignoring this kind of evidence when considering what it is that constitutes works to be read. However, that is not the case with plays, orations, public poetry, music, or dance, because in each of these cases each performance of the work constitutes a crucial ontological form of the work. Because performances in these genres were unrecorded before the invention of recording devices, our only access to performed works of earlier periods is reports and reviews of their enactments and internal evidence. According to researchers such as Jean-Marie Thomasseau, traditional history of the theatre often reduced the theatrical to the literary. Instead, he proposes to think differently of "the theatre as an object within historical determinants, without reducing it to a mere analysis of the different manuscript states of the text leading to the edition."29 While Thomasseau's suggestions deal with "non-contemporary" theatre, contemporary trends such as post-dramatic theater30 pose yet another set of challenges to editing. As Almuth Grésillon notes, the emphasis on performance

also raises interesting questions with regard to other semiotic systems, such as: who is the "author" of a movie? where does the genesis of an architectural project end? how does one establish the "text" of a piece of free jazz? how does one define the "work" of an installation in performance art?31

From an editorial point of view, the only thing that can be edited is the text of the surviving documents and of any stage directions or notations designed to influence the manner of performance. Such documents can be edited from the material, temporal, causal, or genetic orientations, but when edited from a performance orientation, significant attention will be devoted to the "work" that is being indicated beyond the material text—that which is the "end product" intended or envisioned by the work's creator(s). It would require a dexterous argument to defend restoring to a performance text an aspect of the work that had been judged unperformable, even if that judgment were made by some "unauthorized" but experienced performer rather than the author.

It is a further aspect of the performance orientation that a far keener interest attends the history of performance through extended periods including the present than attends the posthumous repackaging of reading texts for commercial sale. Book historians are, of course, interested in the history of a work's repackaging, but editors will frequently ignore editions produced after the death of the author unless new authorial materials were involved or unless they are particularly interested in adaptations. In editing performance texts, on the other hand, it is easier to justify the effort to create histories of variant performances as part of the scholarly edition.

6. Aesthetic/Commercial Orientation

Although, as noted above, the aesthetic approach is seldom countenanced any more in scholarly editing, it might be worth pointing out that many editors and readers are attracted to their tasks initially for aesthetic reasons, if aesthetics is defined to include pleasure and the appreciation of skill in language, narrative, or forms—which are not directly related to utility or exposition. Consequently, we believe that aesthetics is relevant to the task of editing, but we acknowledge that it is a difficult aspect upon which to reach consensus and is especially vulnerable to the charge of the eccentricities of personal taste. However, one can posit a difference between an approach to editing that seeks to respect the known aesthetic principles of an author or of a historical publisher, on one hand, and the very different approach, on the other hand, that asks simply, How can we make this old-fashioned work better so that it can have renewed commercial life? The former has a historical dimension for which evidence (not the editor's personal taste) can be adduced. And finally, we believe it is worth pointing out that all the orientations, with their scholarly attention to evidence and the inferences from the evidence that are supported by argument, cannot, for all their efforts, be totally objective or free from the sense that judging amongst variants involves aesthetic principles. However, as previously noted, this subject is too complicated to pursue in this essay.

Conclusions

As the previous sections indicate implicitly, this re-orientation to "orientations to text" was also inspired in part by possibilities recently developed for electronic editing. The flexibility of digital media removes some of the narrow bands constraining a scholar's use of orientations: combinations are now possible that before tended to confuse the work. We began with the notion that an orientation to texts helps one identify the character or nature of the work one is studying and to focus attention on various aspects of a work's inspiration, creation, production, and reception–aspects that help us understand the work as an event or series of events in history. Those understandings of the work also help to give a perspective to editorial activity. In the digital world, one is always working at one remove from the primary materials, which is a disadvantage one can too easily forget, but the advantages are great. A digital scholarly edition does not need to be organized exclusively for the purpose of establishing a single edited text produced according to the priorities of one particular orientation. Several edited texts can be presented as equal alternatives serving different goals. This implies that it is perhaps even more important for editors than it already was to be aware of the distinctive elements and priorities of different orientations. What the concept of orientations also shows is that, important as edited texts are, the object of textual study and of digital presentation does not need to be an edited text. A scholar's orientation to text can organize the facts about the work such that process rather than product is the main focus. An awareness of orientations to text can therefore be useful, not only to the individual scholarly editor who tries to be consistent in his or her approach to reading and editing, but also to editors as a community of scholars with an openness to understanding and respecting other editors' approaches and priorities.