(Return to contents)

The Editor in the Machine:

Theoretical and Editorial Issues Raised by

the Design of an HTML Literary Hypertext

   Jon Bath
Corey Owen
Peter Stoicheff
Department of English
University of Saskatchewan
    The editorial decisions involved in creating a scholarly digital version of a literary text are as crucial as they are with the printed edition, and more numerous, as we recently found when creating hypertexts of T. S. Eliot's poem "The Love Song of J. Alfred Prufrock" and Mary Shelley's novel Frankenstein: Or the Modern Prometheus.1 Yet many proponents of hypertext anticipate that the digital platform will erase the traditional separations between reader, author and editor, diminishing the editorial role in the process.2 From some perspectives this is at least partly accomplished. The multi-linearity of hypertext, for instance, permits the reader to control the reading experience to a far greater degree than does the somewhat more linear book format. Too, in the case of hypertexts that allow readers to contribute links (such as hypertexts frequently designed for university courses3) the reader becomes a co-editor of an eternally incomplete text. And, to the extent that a digital edition of a text is, at least hypothetically, infinite in content, one might expect that the traditional role of the editor of critical editions is consequently minimized. One might even go so far as to imagine a textual situation where the notion of "editor" becomes obsolete, where the apparent democracy of the hypertext environment precludes the need for any authority deciding, as is the case with the book's finite boundaries, what should and should not be included. If the digital environment has room for every edition of a novel, every intertextual source, every critical commentary, the decisions that were the responsibility of the scholarly book editor should, one might expect, be far fewer in number, and reduced in significance, if not absent altogether. In practice, though, the digital format does not abandon the need for traditional editorial acuity, or resolve the debates concerning editorial principles; if anything, it makes them more prominent. And the potentially limitless size of a hypertext edition carries with it an additional dimension of editorial problems that must be faced by anyone attempting to create a worthwhile scholarly hypertext edition.4
    We began laying the groundwork for our digital editions when the majority of hypertext resources available were multi-media versions with little, or no, scholarly value. Of the numerous web-based hypertext versions of "The Love Song of J. Alfred Prufrock," many still restrict themselves to providing intertextual links scanned from B.C. Southam's A Guide to the Selected Poems of T. S. Eliot, for instance, or offering eccentric spins on the poem that lead to other, only ironically related, web sites (such as a seafood ordering service, a link from the "sawdust restaurants with oyster shells" line in the poem's first stanza). Hypertexts of Frankenstein on the web can be equally uninformative and misleading, their compilers frequently unaware that the novel was crucially revised in 1831 by Mary Shelley after its anonymous first appearance in 1818. An enormous academic debate surrounds the questions of how to account for and interpret the differences, and how to choose which edition should be considered Frankenstein. Yet most sites devoted to the novel ignore the distinction entirely, use the 1831 edition, and often preface their 1831 text with a descriptor such as "Mary Shelley's 1818 novel."
    The proliferation of these sites between about 1995 and 1997 suggested to us that scholarly hypertext editions were needed. The appearance of George Landow's "In Memoriam Web" and "Victorian Web," among others, showed they could be done. Since then, of course, the list of scholarly digital projects has lengthened considerably to include the many at University of Virginia's IATH, the British Library Board's Electronic Beowulf and Canterbury Tales projects, University of Indiana's Victorian Women Writers project, and more. The motivations behind the creation of these digital projects varies. Some, such as the Walt Whitman archive, exploit the platform's ability to include numerous manuscripts in order to show how the "range and scope of [the Calamus] imagined text is in fact too great, its ‘standards' too broad in their simultaneity, to fit the reductive confines of print technology."5 Others, such as the Rossetti Archive, exploit the visual capabilities of the digital environment: "all texts deploy a more or less complex series of bibliographical codes, and page design -- if not page ornament and graphic illustration -- is a rich scene of textual expression. Computerized tools that deploy hypermedia networks and digitization have the means to study visual materials and the visibilities of language in ways that have not been possible before." The Canterbury Tales project has still another mandate. The Tales are in a state of textual disarray and present the reader with many questions concerning the composition history and the status of the manuscripts. Peter Robinson describes the project's origins as lying "in the perception that the advent of computer technology offers new methods, which might help us ask these questions in a new and more fruitful manner."
    Our intentions at the outset of our two projects were slightly different again. With the T. S. Eliot project we envisioned allowing any user to move toward becoming Barthes' ideal reader, "the space on which all the quotations that make up a writing are inscribed without any of them being lost" (Barthes, "Death" 148 ) -- that is, a reader who, through the hypertext, is given extensive access to the many intertextual influences that comprise the poem and to interpretive materials. The goal of the Frankenstein project was to place the novel not so much within the context of its influential literary predecessors, as with the Prufrock site, but within the context of its later cultural (primarily cinematic) legacy. However, in neither case did we want to appeal only to a specialized scholarly audience; although hypertext is now emerging as a useful medium for highly specialized textual renditions, we hoped to achieve a wide academic audience while maintaining scholarly rigor. As a result, we developed our hypertexts to be used at the university undergraduate level from first to fourth year. This required us to create editions that were both academically sound and relatively wide in appeal, and allowed us to test the work in progress by using it in our undergraduate classes.
* * *
    Interwoven with this issue of identifying the reader is the designation of limits for the work. Traditionally, the finite dimensions of the book, the size of the printed page and the cost of printing have silently set the limits for the editor of a scholarly book, but in the potentially infinite digital space these boundaries are widened. The space that "never presumes to close itself off" (Bolter 149) quickly becomes an illusion for the hypertext editor, however, who soon recognizes that several factors define the parameters of a hypertext.
    For instance, just as the amount of information in a book is dependent on the number of pages within it, a hypertext is limited by the amount of storage space available on the editor's Web server or CD-ROM.  As the archive grows, and especially if audio or video are added, it soon becomes clear that the writing space is not "infinite." As technology advances, these limits continue to expand, but it is important to remember that the larger a hypertext's memory requirement the more it will cost to produce, an increase that will inevitably be passed on to the consumer in the form of a higher purchase price. Copyright is also a limiting factor in any form of published discourse, but the possible size of a hypertext archive and the disdain for and fear many print publishers have of electronic media make copyright the single greatest limiting factor for the hypertext editor. Often resources will simply not be available, as shown in Stuart Curran's inability to include Mary Shelley's manuscripts in his Frankenstein hypertext edition, to be published alongside the 1998 University of Pennsylvania edition of Frankenstein.6 Or the price of use for a given item may outweigh its worth to the project. Often the price of individual items is reasonable, but as a whole they may overwhelm the available budget, forcing the editor to decide which resources are truly necessary. As with storage capacity, generally the larger the hypertext the more it will cost to produce and the more expensive the finished product will be.
    The restrictions of size in traditional book publishing make it possible, and academically necessary, for the intellectual market to contain and frequently produce different editions of the same text. However, a well designed hypertext edition can cover a vast range of topics, include a large number of critical essays, display textual variations, and even offer (as in the case of Mary Shelley's novel, for instance) two or more complete editions. These capabilities reduce, though do not erase, the need for many other digital editions of the same text. The physical limitations of a book edition help determine its editorial shape; usually, for instance, a system of annotations suffices for alerting the reader to variants between editions, and the inclusion of two complete versions of the text is not possible. The wider boundaries of the hypertext environment reduce such practical limitations to a degree, thus minimizing the number of necessary editions of a text in the academic marketplace.
    The editorial decisions that distinguish the Norton Critical Edition of Frankenstein from the Broadview edition, or either of those texts from the St. Martin's Press Bedford Books edition (concerning what critical articles to include, or which version of the novel to use) could not realistically generate three hypertext editions of the novel; one hypertext could embrace all those choices.7 Differences in hypertext editions, therefore, need to be conceptually significant. Curran's forthcoming edition of Frankenstein works comprehensively with the novel's prehistory (intertextual influences, Mary Shelley's biography, the geographical context) and with its critical heritage; but it leaves aside the cultural appropriations of the novel since its appearance -- primarily the film versions. Our edition includes less of the former material, and of critical commentary, and conceptually diverges from Curran's by including and emphasizing the novel's cinematic and cultural legacy. Curran's assumption is that "cinema has done a gross injustice to Mary Shelley's novel." Our assumptions are that this century's relentless cinematic attention alerts us to the novel's importance; that cinematic versions are themselves interpretations; that their restlessness with Shelley's text is as significant (and culturally revealing) as the textually faithful academic readings; and that the contemporary image's obsession with Shelley's narrative of technological obsession couldn't find a better platform than digital hypertext. For a treatment of the novel's cultural appropriations to be accurate, however, the scholarly material and rigor must be there -- otherwise, the ironic distance between, let us say, the novel's creature murdering "because I am miserable," and the 1931 Universal Films' creature murdering because he is mistakenly given a "criminal brain," would be lost.
* * *
    Such practical hurdles limit the hypertext editorial enterprise; their consequences are different from the physical limitations imposed by the book format, but of the same degree. For instance, if access to all material written on a given subject can be digitally contained in hypertext, should it all be included? The editor of a book-format text is limited by space and can only include a selection of the critical material that is otherwise available, but the editor of a hypertext can potentially include it all. While to do so would be the most democratic (and the least "editorial"), the hypertext editor must still face the question of the relative significance and contribution of each. Although it seemed unnecessary to us, for example, to include the entire text of The Vicar of Wakefield because Frankenstein contains a one-line joke from it, another editor might have found it worthwhile. Curran is accurate on this point: "The capacities of the [hypertext format] would . . . allow one to attach the entire work. But, not only would such a recourse seem a silly affront to common sense; it would also counter the clear strategy with which the entire intertextual library needs to be presented. . . . [T]he intertextual dimension of [his] edition is thus restricted to what are texts with clear relevance to the intellectual dynamics of Frankenstein . . ." (46-7).
    There is also the question of whether the hypertext edition should remain self-contained or be linked to other hypertexts on the Internet, either by placing it (possibly in a reduced version) on the World Wide Web or by creating external links on a CD-ROM version. We decided it was most advantageous to create the edition for CD-ROM use, and to program external links in it. For example, a Hamlet reference in "The Love Song of J. Alfred Prufrock" is linked to "The Shakespeare Web." Because we have no hand in designing that Shakespeare site, by using it as part of our own work we reduce our editorial control. At the same time, we retain the ability to decide which sites should be included within our own work, thereby introducing a different sort of editorial power that now permits access to other archives. Undoubtedly, as hypertexts attain greater levels of popularity, and revenue is dependent upon the number of times a site is accessed, publishing companies will ensure their hypertext of Hamlet is linked to their hypertext on T. S. Eliot, or will try to ensure that someone linking to a Shakespeare site chooses theirs and not another company's. Thus, while hypertext increases the volume of material an editor can include in a work, it may introduce hidden editorial influence of a different sort that controls the routes a reader travels from text to text.
* * *
    Hypertext editors, then, share with editors of scholarly editions the tasks of envisioning an audience, defining the limits of their work and selecting material for inclusion. The electronic medium alters the results of those considerations, but does not absolve the editor from facing them. In addition, editors of hypertext encounter several problems unique to their medium. One of the most obvious is the designation of links (by colour, for instance) within the text, a question which raises larger issues of what constitutes textual significance and how it operates. One strategy for conceiving of hypertext link design is to assume that a text is supersaturated with meaning, that each component of the text warrants interpretation (including any combination of intertextual, critical, definitional, etc.) and that, consequently, links need not be marked; the reader can click on any part of the text and a link will inevitably occur.8 It may also be that a link which is differentiated from the text focuses readers on the selected passages, causing them to ignore or skim over unlinked passages. (A familiar example of how notation changes the reading experience is a library book which bears the scribblings of generations of  readers; marginal notes, underlined text, or even doodles identify the most used passages and attract the reader to them.) Even if a hypertext editor does not draw readers' attention to a selected passage, however, the editor is still controlling the reading experience by deciding where the links lead.

    This seemingly minor and isolated example suggests how difficult it is to create a hypertextual environment in which the reader is free from editorial influence. Consequently, we made no attempt to hide our presence as editors in this regard, and decided instead to identify our linked passages. Not every passage of our editions contains a link, and we did not want our readers to be frustrated by clicking on words that "do nothing." In keeping with our intention to design our hypertext editions as scholarly teaching tools, we wanted readers to be able to find the information they required quickly and without searching through every link. This necessitated links which are not only differentiated from the text, but which are also categorized and differentiated from each other. Our original idea was to alert readers to different interpretative links (e.g. intertextual sources, textual variations, feminist criticism, Marxist criticism, etc.) by colour coding the text, but changing link colour within a single document was, and apparently still is, impossible within the current state of internet technology (we used HTML 3.2 for the majority of our hypertext coding).9  Until colours are available we have decided to use different fonts to identify various types of links. In the cases where links overlapped (for example, when a passage in the text was linked both to an intertextual source and to a scene from a film) an intermediate screen was created that allowed readers to choose which link they wanted.

    The hypertext editor must also deal with the problem of creating links to large bodies of text. When linking from a text to another large work (for example, from Frankenstein to Volney's Ruins) should the reader be guided to the beginning of the linked text or directly to the intertextual passage within the linked text? Situating readers at the beginning of the linked text gives them the opportunity to read it in its entirety and draw their own conclusions, but also commits them to a time-consuming process. Taking readers directly to the intertextual passage eliminates this time expenditure, but reduces any contextual understanding of the passage: they may then need to scroll to the top of the linked document to read the title, and subsequently may not be able to find the precise intertextual passage easily. To solve this problem, we settled on an intermediary link containing the bibliographic information and a short selection from the linked-to work, with the option of pulling the entire document up in a separate window. This intermediary link allows for the convenient insertion of annotations or editorial comments.

    The amount of such annotation becomes a particularly difficult problem for the hypertext editor. Whereas the absence of annotation conceivably permits readers to become editors and to decide for themselves the significance of links, the hypertext editor implies an annotation of sorts merely by programming a link and alerting the reader to the connection. For example, linking the penultimate line of "The Love Song of J. Alfred Prufrock" ("By sea-girls wreathed with sea-weed red and brown")10 to John Masefield's "Cardigan Bay" makes it obvious we feel Masefield's poem operates intertextually with Eliot's. In our editions we have incorporated little in the way of "explicit" annotation, but we are very aware of the implicit annotation within our link structure and hope readers will question some (or all) of our links and determine for themselves the value of the links instead of just accepting them.
    The issue gets to the heart of the hypertext enterprise. A hypertext edition not only makes available a substantial amount of information "outside" or "extrinsic to" the text (such as Masefield's "Cardigan Bay," or Paradise Lost in the case of Frankenstein); in the process it questions the concept of "text" altogether, and the boundaries that have allowed metaphors like "inside" and "outside" to be used.11 Paradise Lost permeates Frankenstein in crucial ways that Volney's Ruins does not: as Curran puts it, Paradise Lost "governs the cosmology" of the novel (Curran 46). Yet this is an interpretive, and now editorial, call that sanctions the significance of Milton's text to Shelley's (in effect drawing it within the wider hypertext boundary) and distances Volney's. Enough scholarship has been done in recent years to substantiate that particular decision, but in the case of T. S. Eliot's poem the issue of relative influence, of whether Dante's Inferno, say, permeates the very spirit of the text or merely influences a discrete portion of it from a distance, is debatable. Since much of the hypertext editor's activity involves developing a strategy for intertextual display, the question quickly becomes a problem of interpretive and theoretical position, requiring a new alignment of intertextual and editorial theory.12 Unlike the codex, the digital environment does not prevent the inclusion of any intertext, either in its entirety or in the form of a representative passage. The intertextual can, in the technological sense, become infinite, and requires a theoretical position to shape its possibilities the way the physical limits of the book edition otherwise do, lending necessity to Jerome McGann's call for closing "the schism between textual and interpretive studies, opened so long ago" (Critique, p. 11).
* * *
    If hypertext has the capacity to contain and simultaneously display all available information concerning composition and publication history, textual variants, intertextual links, the history of critical reception and so on, the resulting text is a fluid and, for the reader, constantly self-organizing archive for what used to be a single text. Yet this constantly shifting space is still editorially shaped. As McGann puts it in "The Rationale of HyperText,"
To say that a HyperText is not centrally organized does not mean -- at least does not mean to me -- that the HyperText structure has no governing order(s), even at a theoretical level. Clearly such a structure has many ordered parts and sections, and the entirety of the structure is organized for directed searches and analytic operations. In these respects the HyperText is always structured according to some initial set of design plans that are keyed to the specific materials in the HyperText, and the imagined needs of the users of those materials.
Hypertext does not reduce editorial influences; instead, it transforms them and generates new ones, in the process blurring the codex notion of "copy-text." In place of an edition's opinion of copy-text that follows the principles of a Fredson Bowers, W. W. Greg or G. Thomas Tanselle, the hypertext edition permits a variety of copy-texts, reminiscent of Morse Peckham's argument that the codex editor does not try to exemplify an author's intentions, but instead "adds to the group [of printed and manuscript discourses] another version" (Peckham, p.155). The medium's new editorial practicalities create a new kind of interpretive "copy-text" which still asserts that Paradise Lost is indispensable to a reading of Frankenstein while a reading of  Volney's Ruins is not, that Dante's Inferno is "inside" T. S. Eliot's "The Love Song of J. Alfred Prufrock" and "Masefield's "Cardigan Bay" "outside" it, and so on.
    Examples of this intersection of editing and interpretation abound in the hypertext enterprise, and a crucial one returns us to the earlier issue of the distinction between the 1818 and 1831 editions of Frankenstein. Is it best to offer the 1818 edition, "beyond" which the "textual variants" of the 1831 edition are accessible? Or should both editions be included in their entirety, with an intricate system of links between them alerting the reader to the differences? We chose the former, largely in response to recent critical privileging of that text, and created links to variants in the 1831 edition wherever they occurred. A repetition of identical portions of the text (which constitute about 95% of each) seemed intuitively redundant, even if the digital environment easily permits it to be done.13
    It appears, therefore, that while the hypertext edition grants a greater multiplicity of readings than non-electronic editions, it does not escape the role of an "authoritative" edition.14 In place of the codex copy-text the hypertext edition assembles the constituents of a (finite) number of otherwise potentially authoritative copy-texts. Yet these are nevertheless collected and displayed as a result of inescapable editorial decisions, requiring the thoughtful alignment of editorial and interpretive theory that accompanies that role.

Works Cited

Barthes, Roland. "The Death of the Author." Image-Music-Text. Ed. and trans. Stephen Heath. New York: Hill and Wang, 1977. 142-48.
---------. "The Structural Analysis of Narratives." A Barthes Reader. Ed.  Susan Sontag. New York: Hill and Wang, 1982. 251-95.
Bath, Jon, Corey Owen and Peter Stoicheff. A Frankenstein Hypertext.
----------. The Prufrock Papers.
Boaz, John K. and Mildred M. Boaz. "T.S. Eliot on a CD-ROM: A Narrative  of the Production of a CD." Computers and the Humanities. 30, 1996.  131-38.
Bolter, Jay David. Writing Space: The Computer, Hypertext, and the History  of Writing. Hillsdale, new Jersey: Lawrence Earlbaum Associates.  1991.
British Library Board. The Electronic Beowulf Project.  <http://www.uky.edu/~kiernan/eBeowulf/guide.htm>
------------. The Canterbury Tales Project.  <http://www.shef.ac.uk/uni/projects/ctp/index.html>
Curran, Stuart. "Frankenstein: The University of Pennsylvania Electronic  Edition." Keats-Shelley Journal. XLVI, 1997. 44-9.
Groden, Michael. "Contemporary Textual and Literary Theory."  Representing Modernist Texts: Editing as Interpretation. Ed. George  Bornstein. Ann Arbor: University of Michigan Press. 1991.
Katz, "Current Uses of Hypertext in Teaching Literature." Computers and  the Humanities. 30, 1996. 139-48.
Landow, George P. Hypertext: The Convergence of Contemporary Critical  Theory and Technology. Baltimore: Johns Hopkins University Press.  1992.
McGann, Jerome. A Critique of Modern Textual Criticism. Charlottesville:  The University Press of Virginia. 1992. Rpt. Chicago: University of  Chicago Press, 1983.
---------. "Rationale of HyperText." TEXT. 9, 1996. 11-32.
---------. The Rossetti Archive.
Peckham, Morse. "Reflections on the Foundations of Modern Textual  Editing." Proof: The Yearbook of American Bibliographic and Textual  Studies. 1, 1971. 122-55.
Price, Kenneth M. and Ed Folsom, eds. The Walt Whitman Hypertext Archive.  <http://jefferson.village.virginia.edu/whitman>
Ross, Charles. "The Electronic Text and the Death of the Critical Edition."  The Literary Text in the Digital Age. Ed. Richard J. Finneran. Ann  Arbor: University of Michigan Press., 1996. 225-31.
Willett, Perry, ed. The Victorian Women Writer's Project.

1See A Frankenstein Hypertext and The Prufrock Papers.

2By "hypertext" in this paper we mean an electronic text containing internal and external links, coded, in our case, in HTML 3.2.  While the hypertext projects we discuss could also use SGML, informed by the TEI guidelines, to create a system of data retrieval points in the texts, that would determine and involve a set of editorial issues different from those we discuss here, which are related to a text programmed in HTML.

3For an analysis of hypertext as a teaching tool see Seth R. Katz, "Current Uses of Hypertext in Teaching Literature."

4In their article "T. S. Eliot on a CD-ROM: A Narrative of the Production of a CD," John K. Boaz and Mildred M. Boaz describe the financial, administrative and technological challenges they faced when creating a hypertext edition (in their case T. S. Eliot's The Waste Land).  We experienced many of the same impediments during the course of our project, and in fact were somewhat prepared due to the description they gave in that article.

5As stated in the rationale for the project.

6For an account of this hypertext's genesis and production see Stuart Curran, "Frankenstein: The Pennsylvania Electronic Edition."  In his talk "Frankenstein in Hypertext" at the "Mary Wollstonecraft Shelley In Her Times" conference (May 23, 1997) he stated that copyright problems prevented him from incorporating the manuscripts in his hypertext edition.

7Mary Shelley, Frankenstein: The 1818 Text, Contexts, Nineteenth-Century Responses, Modern Criticism, ed. J. Paul Hunter (New York: W. W. Norton & Company, 1996); Mary Shelley, Frankenstein: Complete Authoritative Text with Biographical and Historical Contexts, Critical History, and Essays from Five Contemporary Critical Perspectives, ed. Johanna M. Smith (Boston: Bedford Bokks of St. Martin's Press, 1992).

8Roland Barthes' idea that "art is without noise" ("Introduction to the Structural Analysis of Narratives" 261) suggests there needs to be no special identification of links because all text in a "work of art" will signify beyond itself.

9The situation is not unlike William Faulkner's in his unused publication design for his 1929 The Sound and the Fury: he wanted to distinguish between flashbacks in Benjamin's section with different colors of type, but was thwarted by the high cost of such a strategy, particularly at the beginning of the Great Depression.

10Eliot's line "By sea-girls wreathed with seaweed red and brown" has John Masefield's "Delicate, cool, sea-weeds, green and amber-brown" in "Cardigan Bay" as one of its possible intertextual echoes.

11George P. Landow makes a similar point: "Hypertext . . . blurs the distinction between what is 'inside' and what is 'outside' a text" (Hypertext 63).

12In "Contemporary Textual and Literary Theory," Michael Groden argues that instead of maintaining a comfortable separation between editorial and interpretive theory, scholars should consider developments in both when approaching a text.  Groden notes, however, that as literary theorists we often fail to consider seriously our roles as editors, and the effects that our editing has on the reader.

13The University of Pennsylvania edition includes both texts, and a third "in which both the deletions and additions are simultaneously represented in distinctive fonts" (Curran, "Frankenstein: The University of Pennsylvania Electronic Edition" 45).  As Curran writes of the decision: "The hard part . . . is how far to chart this ground for the reader."  Annotations unique to each edition are included, as well as cross-references, reinforcing "the level of complexity that hypertext, with its disdain for customary borders, routinely invites . . . .  At some obscure point, doubtless, I can expect a collision of opposite demands that will require finesse (or maybe a throwing up of hands that pretends to finesse)" (45).

14Others, such as Charles L. Ross, argue differently: "the birth of the reader-as-editor" that hypertext will create "must be at the cost of the death of the critical edition" ("The Electronic Text and the Death of the Critical Edition," 225).  We believe that hypertext does not truly release hypertext editors from regarding their projects as critical editions.

(Return to top of page)