Rationale: 
A Meditation on Print and Multimedia Editing Techniques as Acts of Representation

The Imagined Text vs. the Printed Document

In the 1881 version of his first "Calamus" poem, "In Paths Untrodden," Walt Whitman, a printer by trade, distinguishes his imagined, ideal poem from the printed text of the poem. "Clear to me now [are] standards not yet published," he proclaims (l. 6). Announcing the unmistakable clarity of an imagined set of conventions, Whitman privileges the "not yet published" text over the "published" one. Implicitly, publication of a poem is an act of representation, not an ideal transcription of the poet's intentions, for the text in the poet's mind differs from the one that he produces materially. Moreover, in casting the unrealized poem as the authoritative one, Whitman demotes the rendered version to an inferior status. "In Paths Untrodden" suggests that the ideal, authoritative poem resides in a "life that does not exhibit itself, yet contains all the rest" (l. 11). The range and scope of this imagined text is in fact too great, its "standards" too broad in their simultaneity, to fit the reductive confines of print technology.

Despite these limitations, Whitman drafted "In Paths Untrodden" in three manuscripts and printed it, in each edition of Leaves of Grass, as the first of his "Calamus" poems. Every version, arising from a set of editorial choices, emphasizes an aspect of his ideal text. Any discussion of "Calamus 1," as this poem was named in 1860, or "In Paths Untrodden," as it was subsequently named, or "By the Calamus Pond I Wander," as it was temporarily named during a period of revision, must, therefore, refer to more than one rendered poem. And each version, from the manuscript through its several printings, includes textual variants and concessions to the reductive nature of the physical act of writing and of publishing technology.

These few lines of Whitman's poem adumbrate the fundamental questions that drive this project. What is a text? How do different media affect its representations? Can the rendered text escape its limitations through the multiple possibilities of electronic presentation? Can Whitman's idealizing urge to encompass all be realized through this more expansive medium? Whitman's imagined text of the "not yet published" poem, with all its grandeur and simultaneity, resembles the imagined potential of computer editing. This expectation of a transcendent medium has a factual basis: the electronic text is nonmaterial and thus should escape the confines of materiality.

At the outset of my project, I suffered from a misconception of computer technology that resonates with Whitman's fantasy of the ideal unpublished work: I believed that the computer would allow the editor easily to represent every aspect and every critical interpretation of Whitman's poem. Yet I quickly discovered that computers operate on an "if, then" logical framework and consequently could not work as a human mind does-- simultaneously imagining a "not yet published" ideal text, yet at the same time interpreting various aspects and elements of the text.

As I show in the HTML version of my hypertext project, Whitman's distinction between the ideal and the realized texts remains. HTML cannot represent every version of the poem as Whitman might have imagined it. Hypertext editing strategies, like those for print, must focus the reader on a set of textual characteristics and discriminate against the rest--much as Whitman chose to emphasize certain characteristics in each rendering of "In Paths Untrodden." What HTML can do is represent the versions that Whitman did render; the hypertext can discuss them as different manifestations of an ideal form.

Editorial Choices and the Text

Printed texts operate as individual acts of representation and tools for interpretation. Hypertexts can also offer focused representations, insofar as the editor consciously structures these virtual documents in a way which reflects a chosen, rather than imagined, set of editorial issues. In this meditation I first discuss a printed form of the poems. I use The Portable Walt Whitman to illustrate print conventions, but also to examine the verse in its most simplistic and reductive representation. Using this text as a point of departure, I present my electronic rendering of the poems.

Each of the computer models is a structured database encoded in Standard General Markup Language (SGML). The principles of SGML are simple, the practice and usage of SGML complex. The editor begins with a plain text of the "content," in this case the letter characters of the poem "In Paths Untrodden." When the computer reads this file, it recognizes the letters alone as characters. The editor must encode words, phrases, or bibliographic structural units with a set of "tags" so the computer can recognize these features of the text as structural units. The editor then "tags" the text. Each tag becomes a "handle" that the computer can grab and interpret according to the editor's rendering instructions. These instructions are created in the form of a style sheet, which tells the computer how to format the text on the page: such instructions control font size, text color, arrangement and portrayal of structural divisions within the text, and display--that is, what parts of the text are displayed or suppressed.

Each of my models of "In Paths Untrodden" is constructed according to a different SGML Document Type Definition (DTD), which incorporates a structured set of editorial choices. These choices constitute a limited vocabulary of tags with which the editor can encode text for selection and interpretation by the computer. By governing the representative and interpretive enterprises, these encoded vocabularies focus the reader on a set of textual characteristics. Such discrimination of editorial priorities resonates with the printed book, in which textual authority or interpretive strategies are suggested by the editor's emphasis of some characteristics of the text, and his suppression of others.

Printed Texts, Scholarly Editions, and the Hypertext Critical Edition

The Portable Walt Whitman presents its version of "In Paths Untrodden" as authoritative. The book lacks any indication of the textual variants in different editions of Leaves of Grass. The editors who prepared this version selected it from many existing versions, as they heeded Whitman's endorsement of the 1881 edition. For this reason, the editors decided that this version best recovered the meaning the author had intended. Other, more practical reasons contribute to the presentation of the poem in this format. For example, as any user of the Whitman Variorum or Fredson Bowers' Parallel Text can attest, print does not easily lend itself to the representation of multiple editions. When print attempts these ends, the reading experience becomes nonlinear and cumbersome, an experience too frustrating for most readers to bear. The pile of readers' editions, the Variorum, the Blue Book and its companion textual analysis, and the parallel editions weigh more than thirty pounds and occupy, when open in use, more than six square feet of desk space.

The Portable Walt Whitman performs one task well: limiting the material to a single version of a poem, the text provides simplicity and clarity for the reader. Users' interpretive questions will focus on what the poem on the page means; the format consequently invites close-reading methods. The reader of the PWW conceivably experiences pleasure in the luxury of reading only one text, bound in a book, an object with which he or she is familiar. For the same reasons, however, the PWW forecloses on other kinds of interpretive questions and reading experiences. This volume suggests that it represents "the" poem, refined out of the productive, printed, or physical history it may have experienced throughout the several editions of Leaves of Grass. The collection also fails to call attention to the text as a constructed object, a technological product. This traditional, physical book fits so closely with the reader's experience of a text that acknowledgement of the medium remains on a culturally unconscious level. The limited presentation of the poem's history in the PWW printed version, moreover, removes evidence that would encourage the reader to ask why the poem exists in the way it does in its present state. Compared to the traditional, single- version text of the PWW, a scholarly edition that represents textual variants demands a much more complex reading. The interpretive process changes from a vertical to both a horizontal and vertical one: rather than simply interpreting for depth of meaning, the scholar confronts questions about the reasons for variants. The variants and printed text alone rarely supply such evidence. Ironically, because The Portable Walt Whitman provides the reader with less information, it encourages her to arrive at "defensible," or "true" interpretations, while the reader of the scholarly edition, which provides more information, must first face uncertainty as a result of this exercise. For example, readers studying "Calamus 1" in its manuscript iterations and textual variants will presumably ask why changes were made, and should further interrogate the text for evidence of cultural or biographical reasons for these changes. Since answers to such inquiries are not always manifest in the text, even in its complex presentation, such a reading exercise ultimately suggests the need for further research into the period and the poet.

My collection of electronic representations of "In Paths Untrodden" arises from a different set of choices than the editors of The Portable Walt Whitman or the various scholarly editions selected. In creating these editions, I aimed primarily to complicate the reader's concept of "In Paths Untrodden" as a text and as a product of technical and cultural histories. Secondly, I hoped to present these considerations without burdening the reader with multiple scholarly editions. Thirdly, I wanted to display material that even these scholarly editions could not deliver: full-color digital facsimiles of primary sources, which readers could not otherwise view without trips to several archives throughout the nation. Finally, and perhaps most importantly, in creating multiple models for a single poem, I sought to illustrate its editing as a process of representation, in which each instance provides a lens magnifying different aspects of the poem.

While my electronic model provides a scaffolding for a more complex reading of this poem than The Portable Walt Whitman version allows, and a less cumbersome apparatus than multiple scholarly editions provide, its has limitations as well. For example, any traditional close-reading exercise, such as that favored in undergraduate literature courses, would be challenged by electronic texts' presentation of various versions of the same poem. The technology itself may act as an obstacle to some users, for the physical book, as an object and as a technical practice, constitutes readers' conceptual and metaphorical framework of intellectual exchange. Reading from one or more computer screens, struggling with browsers, and waiting for the web to deliver a given page, rather than just turning it as a leaf, may frustrate or impede some users. But while these exigencies may be confronted as obstacles, they also highlight McLuhan's contention that the publishing technology, or "medium," is inseparable from the "message." Because the physical book is a familiar object, readers often overlook the role of print technology in shaping the information conveyed. Electronic media, in separating the reader from the physical book, throw into relief the characteristics and conventions of print.

Each document type definition that I have employed emphasizes different aspects of the texts, and thus represents the poems in different ways. The first of my models, constructed in Hypertext Markup Language (HTML) emphasizes the productive process of the poems; the physical embodiments of this process, in books and manuscripts; and critical responses to the poems. The model stresses, in particular, the several stages of composition of "Calamus" poems 1 and 2, or "In Paths Untrodden," and "Scented Herbage of My Breast." For the first poem, this model includes three manuscripts, the poet's corrected proofs, the poet's annotated Blue Book, and four printed editions. For the second poem, it includes one manuscript, the poet's corrected proofs, the poet's annotated Blue Book, and four printed versions of the poem. Furthermore, this hypertext presents the variant states of these poems alongside the varying states of Leaves of Grass, a book that evolved through an organic process at the hand of the poet. Images of the physical text in the form of bindings, frontispieces, and tables of contents help to contextualize the poems in this way. The entire "Calamus" cluster appears with criticism, including various kinds of readings, and with documents related to its literary history, through Fredson Bowers' discussion of theLive Oak with Moss notebook and a digital facsimile of that notebook. Although portions of this model remain under construction, at its completion it will combine the digital facsimile of the document with a transcription, strongly directing the reader to the physical text.

The structure of the HTML hypertext suggests from the outset that there is no single authoritative version of the poem, that the poem grows through an organic process rather than simply emerges as a static, completed work, and that each text is built. At the same time the presence of these documents suggests that each one expresses aspects of Whitman's imagined, "not yet published" poem. That Whitman created these various iterations implies his dissatisfaction with his ability to register his vision upon paper, as well as his willingness to make the compromises necessary for writing or printing these works.

Three major and distinct revisional transactions simultaneously take place in this representation of Whitman's process. First, the transaction between the poet and the manuscript page constitutes the poet's own act of representation to himself. Second, the transaction between the poet and reader takes place through the vehicle of the printed page. Third, the transaction between the editor and the reader takes place through the medium of the hypertext. As an act of representation, my model initially breaks down the first transaction into three stages of manual revision in the form of three manuscripts, and then follows this process through the corrected proofs and the Blue Book. This arrangement of the printed poems and manuscripts displays a traceable dialogue among the individual documents, as if Leaves of Grass--the larger, imagined poem--is made material in each of the poet's attempts to render it on paper. By reading across these instances of the poem, the user encounters frozen stages of a living process, similar to the frames of an animation.

Departing from the Manuscript: An Alternate Reading Strategy

The first manuscript in my hypertext, from the Trent collection, distinctly differs from the text of the printed editions, as well as subsequent manuscript iterations of the same poem. An examination of these differences shows the ways in which conventional verbal information--a comparative close reading of the various manuscripts and printed versions--inflects the poem's meanings. From the outset, the Trent manuscript emphasizes a homoerotic space of camaraderie, which is linked to death; the Barrett manuscript reinforces this connection. The Trent manuscript begins, ""And now I care not to | walk the earth unless | a comrade walk by my | side" (ll. 1-4). Later printed editions, as well as The Portable Walt Whitman (or 1881 Leaves of Grass) version, first establish the "Calamus" as a sequestered imaginative space, "In paths untrodden | by margins of pond-waters," before alluding to the desire for comrades, thus subordinating the prominence of homoerotic desire (ll. 1-2). In the Trent manuscript, Whitman also proclaims that through these songs he will "proceed to | enter that substantial | final life" (l. 4). In the printed versions of the poem, Whitman withholds this evocation of death until the second "Calamus" poem, "Scented Herbage of My Breast." The earlier allusion to death in the Trent manuscript resonates with yet another manuscript fragment to be found in my hypertext, the wrapper for the twelve-poem Live Oak With Moss notebook found in the Berg Collection of the New York Public Library. On this wrapper the poet writes that the poems are "Fit to be perused during the | Days of the approach | of Death" (l. 3). Together, the documents suggest that Whitman associated spiritual death and rebirth with homoerotic companionship. The Barrett Manuscript alludes to this connection as well: Whitman begins with an external, visible life from which he has departed, a death of sorts. He states: "Long I was held by the life | that exhibits itself" (l. 1). A later line in the manuscript suggests that this death is a passage to a larger, all-encompassing life. The speaker states: "But for now I know a life | which does not exhibit | itself, yet contains | all the | rest, | And now, escaping, I celebrate | that concealed but substantial | life. . . " (ll. 4-5). While these two manuscripts, as well as the printed versions, begin very differently, all of them conclude with the profession of a "need for comrades." This common ending, in juxtaposition with the various beginnings, suggests that Whitman ultimately aimed to introduce the idea of cohesion, yet felt ambivalent about its direct presentation. This presentation represented a type of death and rebirth for the poem, and the poet's waffling through various manuscripts implies that he did not enter into this process lightly.

Readings of the poem's various titles over time also point to Whitman's conception of the poem's setting as pastoral and phallic. In the Blue Book version of this poem, Whitman changes the 1860 title, "1," to "In Paths Untrodden," then draws a line through this title, replacing it with "By the Calamus Pond I Wander." In the 1867 printing of the poem, however, he returns to the title that he had deleted, "In Paths Untrodden." Both titles suggest that, as well as associating this cluster of poems with the phallic calamus plant, the poet imagined "Calamus" as a marginal space, probably rural or pastoral, away from the urban landscapes he usually depicted in his poetry. Furthermore, the text of "Calamus"--a space of meditation on death and desire for comrades--suggests that the poem itself acts as a place of transition, an imaginative liminal zone for the writer. This space for the positive process of rebirth also acts as a space for the passive process of surrender to external forces, forces that contributed to the poet's ambivalence toward cohesiveness. At the end of this poem in the Blue Book, which Whitman carried during his days as a civil war nurse, he inserted lines that suggest surrender to his lover. He replaces "I proceed, for all who are, or have been, young men," with "I consent at last, for all who are, or have been, young men" (l. 16; l. 16). Yet the poet's failure to incorporate this change in the 1867 printed version suggests his ambivalence about such a surrender.

The Printed Text as a "Made" Object

In addition to providing enriching, even corrective, readings of these iterative stages across the productive process, the manuscripts, the Blue Book revisions, andPoet's Corrected Proofs also display the mechanical and "made" aspects of Whitman's verse. Invisible in the standard printed work, the organic process of the poetry is displayed in this edition both through page images and through the "physical text" section of the project. The critical essays in this hypertext that discuss the origins of the "Calamus" poems compliment the visual evidence of the manuscript and proof facsimiles assembled here. For example, the student interested in the Blue Book revisions in juxtaposition with the Barrett manuscripts can easily call up the essays by Arthur Golden and Fredson Bowers in order to learn about the context of these revisions. To explore further changes in the poem, such as the variants in the printed texts, the user may consult Harold Blodgett's commentary on each poem, reproduced from the Comprehensive Reader's edition.

The arrangement of this model illustrates the book as a constructed, physical object both at the level of the finished work and in the countless manual stages Whitman went through to create it. The poet's instructions to his printers on the first leaf of "In Paths Untrodden"--"Calamus p. 124 to 147 inclusive," and "on odd pages | running head & from pages 124 to 147 is 'Calamus'"-- illustrate his involvement as a craftsman of the physical object as well as the idea. While the Romantic stereotype of the poet, emblematized in Coleridge, continues to collapse the conceptions of the "ideal" and the rendered poem into a perfect transparency, these documents reveal the uncertainty and opacity of the writing and revision process, emblematized by a poet such as Whitman.

These facsimiles and variants provide insights into Whitman's evolving relationship to his audience, through his representation of the imagined poem on the printed page. It has been suggested, for instance, that Whitman followed an impulse to respond to criticism of his erotic themes in revisions of "Calamus," and in fact that he attempted to "clean up" the cluster for the rendering of the "Centennial" 1876 edition in order to emphasize national unity instead of sex. Because this hypertext emphasizes the productive process, arranges the "Calamus" poems in juxtaposition with one another, and offers selected criticism, a cross-reading of poems supports an argument against this position. The formal changes in the first "Calamus" poem are benign adjustments of language and meter. These variants contrast with thematic changes in the second poem. A reading of the deletion of the 1860 line of "Scented Herbage of My Breast," appearing after line 7 ("O burning and throbbing surely all one day will be accomplished;") suggests that Whitman deleted particularly explicit erotic locutions such as these in his revision process. But as Arthur Golden argues in his essay on the "'Calamus' and 'Infans d'Adam,'" the poet passed up ample opportunities to conform to Victorian heteronormative tenets in the revisions in the Blue Book. The digital facsimiles of the Blue Book pages which I have provided illustrate Golden's assertion, allowing the student to retrace the critic's argument in a scholarly hypertext essay.

HTML: A Weak Fragment of SGML

Even in its best, completed state, when each facsimile is accompanied with a transcription and a descriptive bibliography some months from now, the information in my HTML model will remain, in a theoretical and technical sense, relatively unrefined. This edition is curatorial, rather than tutorial, because it does not narrowly focus on a limited set of characteristics in each document or image. The HTML model constitutes a critical edition only in the sense that selected critical works are represented, and that information is hierarchically organized at the level of individual files, which are rendered as "pages". The model cannot be searched, except by keywords and at the level of the file. Organization of information by structure or content within each file is impossible, and thus any representation of printing and productive processes is also impossible, except by means of explicit discursive discussion. The HTML DTD allows the editor the necessary freedom to represent a broad range of materials with the least expenditure of time and labor, and a realistic amount of training for the undergraduate student. Yet HTML's inherent bibliographic anarchy hampers any highly articulate organization of this information. This anarchy, however, allows the editor to represent many issues simultaneously, as in the case of my creation of a digital facsimile of the Live Oak with Moss notebook, which contains several manuscripts that carry more than one poem.

The shortcomings of the HTML DTD arise from its primary function as a presentation tool rather than a means of representing documentary structure and content. The HTML DTD is a subset of the larger, more descriptive tagging language, SGML, which is designed to configure and describe the structure, as well as the form and content, of documents. Because of the greater capacity of its description of content and structure, SGML is relatively demanding of accuracy. HTML, in contrast, is designed for simplicity of use and freedom from issues of structure and content. The HTML DTD, in contrast to the TEI DTD, emphasizes issues of presentation. For example, titles or subject headings of an HTML document can be tagged with a small variety of "heading" tags which control the font size of the lines they surround. Titles tagged with "H1" are rendered large, while running heads tagged "H2" are not as large. But the practical usage of HTML does not support structural applications of these header tags, so does not require that these tags be applied only in instances of headers or titles. For example, if the editor chooses the "H1" tag can surround a line of text in the middle of a paragraph simply to make that text larger. While such usage might render the document as the editor envisioned it, this rendering is divorced from any corollary in the bibliographic structure of the document. In an TEI document of the same text as the example I have given for HTML, the titles would be surrounded with the "title" tag, and the running heads would be surrounded with "head" tags. The "title" and "head" tags represent different structural units of the text, and at the same time name the content of those lines; everything within the "title" tag is a title, and can be nothing else (whereas everything within the HTML "H1" tags could be a title or a running head, or text of any nature rendered in large font). If the editor's desire is to render the text within the "title" tags larger than the rest, or the text within the "head" tags of a different size, these rendering instructions are provided within the separate style sheet document. Furthermore each of these tagged documentary units, because of their discrepancy in a tagging which distinguishes content, can be accessed separately via electronic search tools. In a large archive of the works of Walt Whitman, such a function could be invaluable for anyone searching for titles only.

The benefits of the SGML structured document approach reside in the possibilities for articulation. While HTML works well as a presentation DTD, its options for presentation are severely limited because these rendering options reside within the files of the browser program (such as Netscape), rather than in a style sheet which can be distributed to the browser along with the SGML file, or rendered inot HTM "on the fly," as the document is transmitted from the server. Having experimented with representing everything at once with HTML, I have illustrated in the second portion of my hypertext the discrimination with which an editor can represent texts using SGML. Here I have adapted a vocabulary of tags to the specific editing problems presented by Whitman documents. I have created as a sub-chapter of this project a model poem for the TEI, or Text Encoding Initiative DTD--a general purpose tagset designed for use with military and government documents--in order to experiment with alternate means of representation, that is, to offer a choice between the clear text and the textual variants of "In Paths Untrodden."

Using the UVA Electronic Text Center system of representing texts, I have created a text of the entire 1860 "Calamus" cluster of poems. The first example consists of the TEI header and the body of the text, and is divided according to the structural divisions of the literary work. The rendering of this document in HTML at the client-end of transmission is governed by a filter, which provides rendering instructions upon retrieval of the document and thus replaces any style sheet. In the second TEI example, I have adapted one tag of this DTD to represent the variant words of the first "Calamus" poem, "In Paths Untrodden." This model edition, like The Portable Walt Whitman, fulfills a few aims well instead of attempting to represent all the information about the verse and its productive history simultaneously. Each variant state of this poem consists of two parts, the "header" and the "body." The SGML header, a necessary part of any electronic text in SGML, lists commonly retrieved bibliographic information such as the author's name, the date and place of publication, the printing information, the library location and call number, the editor's name, and the changes made to the text in the process of electronic preparation. Each of these issues is surrounded with an SGML tag that a search engine will recognize. Therefore, a researcher looking for the publishing place of the 1860 edition need only submit "1860 PubPlace" to the search engine. In an HTML version, the reader would have to know the name of the city, and search for "Boston," using a keyword search. Furthermore, in the SGML version, the search could be done across the several editions I have represented. By requesting "PubPlace," the user would receive a list of locations on the screen. I have employed a tagging strategy similar to that used in the header to represent textual variants. These tags interface with the search tool to locate variants.

In addition to making variants searchable, the TEI model I have created visually represents the variant words or phrases, if the reader chooses to see them. I have achieved this feature through the creation of more than one style sheet, the list that provides rendering instructions to the computer. For "In Paths Untrodden," from the 1867 edition, for example, the reader is offered a choice of clear text or a red text, which emphasizes variants. Through differentiation of text color, the variant words appear in the text of the poem, rather than in a list outside the text, such as in the Variorum.

The chief strength of the TEI model is the singularity with which it represents the textual issue of variant words. The HTML model achieves the same purpose, yet does so less effectively because of the limitations on form and format. In my HTML variants of "Calamus 1," the tagging scheme allows for only one option of font and font color, courier black typeface. Since variant words cannot be rendered a different color, the visual scanning of the several versions for their location is impossible, and thus weakens the potential reading experience. The rendering of variants in red in the TEI example, on the other hand, makes plainly visible one aspect of the productive process of revision that is invisibly absorbed by The Portable Walt Whitman.

The second of my SGML models employs the Rossetti Archive Document DTD, or Rossetti Archive Document (RAD), and like the TEI example, emphasizes a highly focused, carefully selected set of editorial concerns. Moreover, this model renders visible several characteristics of the physical book that The Portable Walt Whitman does not even mention. This model, like the TEI, demonstrates the first "Calamus" poem in its variant states. The RAD differs from the TEI, however, in that it allows the editor to describe the printed state of each individual poem as a separate document. The SGML header of this model differs from the TEI header because, in addition to cataloguing information, it contains commentary, as well as the physical characteristics of the book from which the document originates. The commentary fields allow the editor to contextualize the document among the larger set of Whitman documents--in this case, the other poems of Leaves of Grass, in their various editions. The physical description field of the header translates the physical experience of holding the respective edition of Leaves of Grass into representation. This field, which includes paper type, font and type face, binding, page lines, and columns, brings the attributes of the printed object to the reader's attention. Even in the act of reading the originals--a rare event--readers might not notice these attributes of the text. In addition to representing these characteristics visually, these fields are also searchable, so that the scholar interested in comparing Whitman's use of specific publishing materials can find them across various editions.

While time constraints did not allow me to equip these SGML files with images of the documents they represent, such an inclusion would be the next logical step for the editor to take in preparing an SGML critical edition. With such an arrangement, the juxtaposition of the facsimile and the descriptive interpretation in the physical description field reinforces the visual and physical aspects of the act of reading and uncovers bibliographic information that might affect the reading of the poem.

As in the HTML model, this collection of variants within the context of the physical work allows for a different, and more complex, reading experience than The Portable Walt Whitman offers. For example, the richly ornate binding of the 1860 edition appears luxurious and expensive next to the more economical, stiff wrapper of the 1871 edition. Observing such a difference, in light of the linguistic informality of the 1871 variants, might suggest that Whitman's departure from normative decorum involved economic as well as aesthetic aspects. Whitman's earlier binding further reflects his desire to produce a "new Bible" in 1860, while the cheaper 1871 edition shows a greater desire for egalitarian presentation and discursive national unity than for egoistic grandeur. At a more developed stage of completion of this hypertext model, such observations might be more deeply explored and explained in the commentary field of each document file. Yet even without such conclusive commentary, this model in its present state implies that the interpretive act itself produces uncertainty.

Despite SGML's many positive features, its also has disadvantages in describing Whitman's poems. First, the representation of some Whitman documents, in which two poems appear on one manuscript page, introduces a structural problem that lies outside the general document definition upon which SGML is based. Some of the "Live Oak with Moss" manuscript pages contain more than one poem. Because SGML regards documents as discrete objects that are synonymous with the title of the document, and because the computer processes information on an "if, then" basis, SGML cannot clearly represent these structures unless they are divided into grid-like maps, with each sector of the grid represented as a separate document onto which the editor must impose a document name. Second, the naming of documents, line numbers, and deletions is difficult in such cases of overlap. In editing the poem I used for my model editions, "Calamus 1," later renamed "In Paths Untrodden," I could not both represent these poems by their correct titles and make them searchable as synonymous entities. Similarly, when lines of variant poems do not match, as in the case of the deleted line which comes after line 7 of "Scented Herbage of My Breast," the editor must create phantom lines numbers to represent the deletion in subsequent editions. The problem of overlap is most troublesome, however, when the editor attempts to mark up deletions and additions on a manuscript. In the case of Whitman manuscripts such as "Scented Herbage of My Breast," or the Blue Book, both of which contain complex revisions, some deletions are also additions because of Whitman's multilayered revisions. SGML tags work well only if a word or phrase is tagged as only one entity. Since a computer cannot make decisions or represent simultaneously two attributes of a given entity, a word or phrase cannot be tagged as two elements at once. In cases such as these, the editor is faced with a difficult decision about what will and will not be represented in the SGML model.

The Metaphor of the Text

The Whitman documents that I have worked with, especially the manuscripts, are extremely interesting because of the complexity of their revisions: it would be a loss not to represent them all. As a collection, they suggest the intricacy of Whitman's "not yet published" imagined poem. In order to represent all the textual issues of a given manuscript, an editor would have to develop multiple DTD's and encoding strategies. The complex product of this endeavor, however, would conceivably present a multitude of problems to the reader that would approach the level of difficulty experienced by the user of the Variorum and of multiple scholarly editions.

The problems faced by hypertext users stem from the conceptual basis from which we approach a text and are complicated by advances in information technologies. As George Lakeoff argues in Metaphors We Live by, practical experience governs conceptual understanding: people develop metaphors for new experience through everyday practice. In our metaphorical understanding, the process of reading is intimately connected with the idea of holding the physical object of the book and turning pages. How do we define hypertexts, in light of this practice? Lakeoff says that as people strive for more complex or deeper meanings, they depart from simple metaphorical constructions by mixing or extending the metaphors that they already have for familiar objects. The individually engine powered vehicle, for instance, was first named through an opposition to the vehicle it replaced: "the horseless carriage." Thus in unfamiliar information structure of the "hypertext," we have the mixture of the metaphor of the book, or "text", and the metaphorical understanding of the less familiar energy of electronic media, or "hyper." If, then, the term hypertext helps us to understand the object in relation to unfamiliar objects, how do we better understand the actual experience of using a hypertext?

Certainly hypertexts have evolved more quickly than most readers have been able to gain understanding of how to operate them and thus gather metaphor-producing experience with them. In addition to the hypertext, we now have the "hyperstack format," the "hyperbook," the "electronic text," and Electronic Book Technologies Inc. While the nomenclature of these formats still rely on allusion to the book, none of them actually functions the way a book does. Many students and scholars are only beginning to use basic electronic forms; consequently, very few, at least initially, would use a model that represented "Calamus 1," "In Paths Untrodden," in a hundred different encoding strategies. This limitation parallels that of the Variorum, Comprehensive Reader's Edition, and Bowers' Parallel Text, whose complicated and cumbersome apparatuses demand a degree of commitment possessed only by specialists.

Each of these printed textual studies of Whitman signifies a departure from the metaphor of the book. Like the hypertext, they initially interrupt our common expectations of the text while delivering a purposefully focused reading of the poet's work. The Variorum represents variant words in a chartlike structure at the bottom of the page which produces a nonlinear reading experience. Bowers' Parallel Text of Leaves of Grass arranges the manuscript transcription on the left page and the 1860 poem on the facing page, with the descriptive bibliography and notes about the productive process across the bottom, in a different nonlinear arrangement than the Variorum uses. Blodgett, in The Comprehensive Reader's Edition, displays each poem over its literary history, placed across the bottom of the page; major commentaries appear at the beginning of every cluster. In placing these studies within the printed book, the publishers embrace the reader's metaphorical expectations of the book, but the internal information structures mix that understanding with revolutionary dissemination systems. Even the title of the seminal work that they present, Leaves of Grass, combines the metaphor of the book with a plant, suggesting that the poet found the book a constrictive medium that invited broadening experimentation.

What is a Text?

In constructing this model of the "Calamus" poems, I have engaged with broader questions about textual acts of representation, exploring issues not only of electronic representation but also of conventional editing. In conclusion, I would like to present some of the theoretical premises that inform this edition and some of the theoretical questions that it poses. This study proceeds from two central assumptions about the text itself. First, I define the text as a series of interpretive acts rather than as a singular, ideal entity that can be recovered. Conventionally, critics have understood the text as a discrete object that possesses a true state, which may be read for authorial intention and historical truth. Yet as Jerome McGann writes in The Textual Condition, "the study of texts begins with readings of the texts . . . but those readings--like those which stand before them--are materially and socially defined. The readings, as Derrida has shown, are structured philosophically--and historically actuated--as writings" (7-8). This view implies that there is no singular authoritative version as defined by the writer's intentions or even the historical moment. Rather, such a version is one among many texts, each one of which is "perfect" in its extension of the original work. This perfection, McGann argues, is the finest attainable level of achievement given the constraints of the human, physical, and mortal condition.

Second, I follow McGann in defining the text as a mappable, long-standing dialogue between authors and readers that carries with it important cultural assumptions. These premises, which inflect meanings, consciously or unconsciously, upon a text, are reflected in editing decisions as well as elements of material embodiment. In the various collections of Whitman, editors display their interpretive assumptions in the way that they select and arrange individual poems--emphasizing, for example, the poet as a symbol of national unity rather than of homoerotic physicality. The physical properties of these texts, such as the information structures of their page layout and design, offer further insight into editors' priorities and choices. In the case of scholarly editions, they also reflect departures from the "normal" disciplines of linear print technology. Such departures are especially significant because they attest to the concrete relationship between textual material and intellectual enterprise. Critical works and collections also display the social "value" assigned to a poet's work, through indications such as intended distribution numbers, paper types, and publishing houses.

Consequently, I interpret the conversation between audience and author as a communal one, which includes media as forms of historical expression. From these conversations the scholar can ascertain cultural assumptions and philosophical agreements or disagreements, not only about the abstract "meaning" of the text, but also about the ways in which these meanings are shaped by their material or technological embodiment. Walt Whitman's production and revisions of Leaves of Grass, in the form of manuscripts, first editions, and corrected proofs, as well as the critical responses this creative process has elicited, present just such a dialogue as an extension of the work.

McGann's understanding of each version of the text as "perfect" in this sense contrasts with what Thomas Tanselle would call the "damaging materiality" of a work. According to Tanselle, an "aporia," or gap, separates the material work from any attempts to recover its true meaning. Tanselle assumes that each text holds a meaning inscribed by the author, which the critic must extract; critics, he argues, can approximate this recovery only imperfectly. McGann, in contrast, argues for "removal of the stigma of loss" such as Tanselle describes. My deemphasis on textual materiality as "damage" through employment of a "collective text" erases notions of the site of loss between the work and the interpretive act in which meaning is either located or missed, yet at the same time focuses upon the limitations of representation through various technologies.

What remains in this approach are the work and critical responses to it, transfigured into perfect entities of the conversation--what Galvano Delevolpe calls "quids." Quids, the elements of the conversation between authors and their audience, include the technological media employed as their metaphorical "containers." Quids are not considered damaged, because they function as perfect responses in a dialogue. Nevertheless, because quids are published, they are subjected to the communicative limitations of print. As acts of representation, these enterprises can be analyzed for the information they deliver as well as the constraints placed upon them by their particular publishing technology. The text is a created object and its physical form follows or breaks a set of coded conventions.

As I have established, the dialogue between authors and audiences takes on a third party through the media of expression. The medium of the electronic text, as I have also discussed, is an emerging form whose experiential metaphors have not yet fully developed and whose conventions have not yet been fully theorized. Roland Barthes, in his description of the text as a semiotic aggregate, and Edward Soja, in his exploration of textual spaces, offer insights usefully applied to the theory and metaphor of hypertext. Their observations have influenced my thinking in the practical construction of this model.

Barthes employs an idea similar to Lakeoff's metaphorical containers when he defines the text as a series of codes. He develops this definition by arguing that these codes in effect "choose" readers by admitting or denying them access and pleasure: "[T]he text is a fetish object, and this fetish desires [the reader]. The text chooses [the reader], by a whole disposition of invisible screens, selective baffles: vocabulary, references, readability, etc. and, lost in the midst of a text (not behind it, like a deus ex machina) there is always the other, the author" (27). While Barthes understands these codes as purely semantic, they can be applied usefully to the material codes presented by different media. His "selective baffles" might be compared to the paper types, fonts, page layout, or ornamental page breaks that Whitman chose for his third edition of Leaves of Grass, the absence of some of these elements in the fourth edition, the denial of their existence in The Portable Walt Whitman, or their self-conscious representation in the SGML Rossetti-style hypertext.

The print-based metaphorical concept of the Rossetti Archive hypertext format, and Barthes' geographical "midst of a text," where the author ultimately resides, further suggest spatial arrangements of our understanding of a text. Edward Soja develops the spatial dimension of Barthes' semantic discussion through three conceptual divisions: "spatial practice," "representation of spaces," and "spaces of representation." The initial "spatial practice" of the book arrives as the "materialized, socially produced, empirically perceived space" of the text. Readers of books take for granted the actuality of physical text, for reception of this object begins with a set of unacknowledged assumptions about what a book is and does. Soja's "secondspace," like Barthes' semiotic text, is constructed of preconceptual expectations and readerly expectations. Soja defines the "spaces of representation," or "thirdspace," as the practical juncture of the firstspace of the book and the secondspace of readerly selection where "space is directly lived" (67).

The conceived space, or secondspace, "is also tied to the relations of production and, especially, to the order or design that they impose," according to Soja. He continues: "Such order is constituted via control of knowledge, signs, and codes: over the means of deciphering spacial practice and hence over the production of spatial knowledge" (66-69). Secondspace includes the reader's metaphorical concept of the book, through a set of readerly expectations, which, as Barthes puts it, "desire[s]" particular readers through its selections of various semantic codes and screens. In the Leaves of Grass excerpts published in The Portable Walt Whitman, some of these screens--such as the revisions Whitman made of these poems before and after their initial publication--actually disappear, to widen the "desires" of the text. In the "secondspace," of this collection, even the paper and type become invisible to the reader. The HTML critical models, in contrast, display the raw materials of the "secondspace" that are rendered invisible by the printed reproduction. The Rossetti method renders these materials according to the individual "work," which can include several documentary states of the same piece of literature, and links various works according to either reader selection or editorial structure, at the user's choice. This edition manipulates readers' spatial concept of the book: in addition to its unlimited potential for spatial representation, the hypertext has no spatial orientation--up, down, left or right-- unless of course these concepts are imagined by the reader and reinforced by the editor.

Barthes presents a less spatial, but nonetheless material, metaphor for Soja's productive process when he describes the text as a dynamic tissue that both creates and obscures meaning: "[T]ext means tissue," he asserts, "but whereas hitherto we have always taken this tissue as a product, a ready-made veil, behind which lies, more or less hidden, meaning (truth), we are now emphasizing, in the tissue, the generative idea that the text is made, is worked out in a perpetual interweaving; lost in this tissue . . . dissolving in the constructive secretions of its web. Were we fond of neologisms, we might define the theory of the text as an hyphology (hyphos is the tissue of the spider's web" (64). Barthes' understanding of the text resonates with the initial assumptions of the hypertext editor since, in its electronic, rather than material state, the "tissue" of the hypertext is an "interweaving" of readerly assumptions and visual and verbal codes. Thus Whitman's selection of a gold-stamped binding or ornamental page breaks within the space of his third edition may be "read" as grand gestures that are consistent with the egoistic thematic of his poems. For Soja, such elements of the mode of production act as a "storehouse of epistemological power" (67).

In Whitman, secondspace is also the primary space of utopian thought and vision of the semiotician, decoder, or the encoder of hypertexts. Such a space, of the unconstructed text, resonates with utopian ideals of the hypertext which, at the outset of my experiments with representation of editing issues, seemed limitless. Such naive assumptions were disproven in the "thirdspace" of creating the models. In the printed collections of Whitman, the collective critical response to Leaves of Grass, or in hyperediting projects, thirdspace arrives at the linking of the reader and the text itself. Spaces of representation embody, like works of art, "complex symbolisms, sometimes coded, sometimes not" (67).

Leaves of Grass: The Book in a Mixed Metaphor

Walt Whitman's primary literary achievement, Leaves of Grass, represents a departure from the conceived space of the book into a lived textual space. The poet delivers his collection as a dynamic, changing organic growth. His title resonates with the dominant metaphor of the hypertext, as a rhysome. A rhysome, a "decentered" plant such as grass, produces "nodes" of growth that stand in an individual, yet equal, relationship to each other. Whitman's choice of "leaves," rather than blades, provides a rhysomic metaphor for the self-generating parts of a book which stand in equal relation to the whole. Thus Whitman's poetry collection, in its original six editions, manuscripts, and corrected proofs, can be regarded as equal parts of a single process of organic growth. Whitman did not stop at his title in exploring conceptual departures from the book. In the 1860 and subsequent editions, he arranged his poetry in "clusters" that emulate the growth structure of the calamus plant. Throughout subsequent editions of Leaves of Grass, Whitman continued using the cluster arrangement.

As a rhysome, Whitman's "leaves of grass" metaphor suggests that even critical works are an extension of the poet's original work. Such an imaginative extension supports McGann's view of critical works as conversational quids. Leaves gather as well as distribute nourishment in a two-way conversation with the environment. These metaphorical implications invite a critical dialogue that contributes to the growth of the original work.

---

El diseño de la página y las imágenes son
© 1996-2000 Grupo "mmm"
Comentarios a: © Dr.Vicent Fores
València  15th September 2000