O'Donnell, Response to Flanders, Lynch, and Tetreault

NASSR '96 Seminar on Electronic Texts and Textuality

Response to Flanders, Lynch, and Tetreault

Brennan O'Donnell, Loyola College in Maryland

I am impressed, first, by the diversity of the projects described by Julia Flanders, Jack Lynch, and Ronald Tetreault. The use of electronic-text technology for three editions so different in kind suggests that we need to be careful not to give the medium too much credit for the message. Reading these descriptions, it is clear that electronic textuality, whatever it may turn out to be, does not exert in practice anything like the deterministic influence that some technology watchers (both cyber-jockeys and neo-luddites) imply that it does. At the same time, there is no denying that the technology of electronic text transmission is rapidly transforming our sense of what we can and should do with texts. These papers, individually and collectively, reveal some of the enormous promise and challenge involved in the new media. They are the kinds of arguments, grounded in practice, that we need to engage and develop if we're going to move beyond the "hype" of hypertext.

Each of these reports from the trenches gives an account of how the future of electronic textuality is being shaped by the diverse needs of different kinds of scholarship. The Women Writers Project aims to provide access to a large number of texts (eventually composing a new corpus), the uses of which are potentially unlimited and therefore impossible to predict. Work on this project is opening up important questions about standards of encoding and presentation of a vast array of different kinds of texts. The hypertext Frankenstein-- "perhaps [the most extensively glossed edition] of any novel ever"--will facilitate the kind of intertextual and multi-media study to which Mary Shelley's work has long lent itself. This project, like the creature at the center of Frankenstein itself, is engaging its editors in a host of problems stemming from its sheer complexity and bulk. [If I were adept at this sort of thing, I'd provide a link to Victor's account of his momentous decision to make a huge being: "As the minuteness of the parts formed a great hindrance to my speed, I resolved, contrary to my first intention, to make the being of a gigantic stature."] Jack Lynch's account suggests that he and his fellow editors have sometimes had an eerie feeling that this thing has taken on a life of its own. The electronic Lyrical Ballads envisioned by Tetreault will foreground the issue of Wordsworth's revision in a way that print could never allow, revealing the "dynamic process" underneath the "static representation" of the printed text. As Tetreault's discussion of the difficulties he is encountering in presenting so complex a process suggests, this is the kind of work that eventually will have an impact on the development of software and hardware better suited to serious scholarly uses.

I find Julia Flanders's "Argument" and its links an excellent and challenging overview of the whole topic of Electronic Text editing, as well as a tantalizing description of the work of the Women Writers Project. Among its most important contributions to this discussion is its strong advocacy of Standard Generalized Markup Language (SGML) and the Text Encoding Initiative Guidelines for Text Encoding and Interchange (TEI). The editorial commitment to what Flanders calls the "upward mobility and flexibility of the data, the avoidance at all costs of decisions which limit or cut off future possibilities," requires a markup language that allows encoding of kinds of information well beyond the alphanumeric characters of a basic text. As Flanders says in her consideration of "Text Use," electronic texts (particularly those based on existing print publications) need to allow a user to engage in a number of activities if these texts are going to be genuinely and usefully interactive. She lists the following as among those that a good electronic text would allow-- "at least":

  1. instant navigation
  2. simple word searches
  3. complex searches involving Boolean logic
  4. variant views (particularly important for historical sources, where "original readings are important for evidentiary purposes")
  5. side-by-side or computational comparison of witnesses
  6. support for analysis (such as collocation, word frequency, and usage studies).

The ability of SGML to provide variant views (no. 4) raises important questions about the responsibility of electronic text editors to the material conditions of a text's original publication. The more our experience of texts is mediated by electronic means, the more conscious we are becoming of the importance of print and print culture in the writing, dissemination, and reception of historical texts. We in eighteenth- and nineteenth-century literary studies are certainly seeing evidence of a new level of interest in what Robert Essick calls "media reflexivity," in recent studies of the impact on print technology on the rise of the "man of letters," of the aesthetics of the book, and of the business of print publication. (See recent studies by Kernan, McGann, Viscomi, and Erickson, for example.)[1] SGML can encode a great deal of information--about typefaces and shifts in typeface, lineation, pagination, margins, relationships of various texts within bound volumes, printers' errors, and the like--that could be of enormous use to future studies in this vein. In fact, it might be argued that electronic text, somewhat paradoxically, can enable unprecedented access to this sort of evidence, which hitherto has been available only to scholars fortunate enough to have regular access to major collections of eighteenth- and nineteenth-century books, or through expensive facsimile editions.

The question arises: Should the inclusion of such information be a priority for the editor of electronic texts based on printed books? The WWP answers this question, "yes." Its statement on "Encoding Practices" speaks of a "commitment to preserving the integrity of the text as an object which circulated in the culture of a particular historical moment." I would be interested in hearing seminar participants' views on the extent to which electronic text can or should allow access to such information, and on the likelihood that the principle set forth by the WWP will become the standard in electronic text editing. Ronald Tetreault says that the Cambridge Lyrical Ballads on CD will employ SGML encoding. Do the editors embrace the view of the WWP regarding this commitment to historical integrity of the text? Will the CD version allow a reader interested in the look of the poems on the page access to information on such things as typographical peculiarities (kinds and sizes, italics, boldface, spacing of characters and lines, printer's errors, and the like), margins, the number of the number of lines or stanzas per page, or the relation of page breaks to stanzaic patterns or blank verse paragraphing?

I would be especially interested to know whether or not the markup of the Cambridge Wordsworth will allow a reader conveniently to reconstruct the various orderings of poems in successive editions of Lyrical Ballads. (As an aside, it would seem that one of the real virtues of hypertext in studies of poetry would be to allow more of the kind of work on the organization of poetic collections that Neil Fraistat did, in The Poem and the Book. I have had students write excellent papers on poems in the context of collections, but topics have been limited by my library's modest holdings of lifetime editions and facsimiles.) Tetreault's argument, in "Electrifying Wordsworth," emphasizes the ability of the proposed edition to liberate Wordsworth from the "static representation" of print and to "represent [him] in development." Is it possible, or desirable, to achieve this aim while at the same time providing copious and accurate information about the minute particulars of Wordsworth's appearance to his own age, in the medium of that age? Whatever the nature of Wordsworth's development, after all, that development took place in the process of writing for print. And his wish for an integral self, of days "bound" each to each, surely reflects a mind deeply informed by the logic of the printed book

Another issue raised by Flanders's discussion of SGML encoding is the question of analyzability, which in turn raises questions about the extent of tagging and the level at which the tagging occurs. Here, it seems to me, the principle of not cutting off "future possibilities" for the use of electronic texts puts editors in a very challenging position. The kinds of analysis allowed by SGML are limited only by the editor's ability to imagine and anticipate those "future possibilities," yet we know that we cannot know exactly what some future community of scholars will decide is essential information about these texts. How does an editor decide how much tagging is enough? David Chisholm and David Robey (members of the TEI Work Group on Verse), for example, have developed guidelines for verse tagging that would allow analysis of prosodic structure at the level of the individual syllable, and that could allow complex analysis of interaction between syntactical and metrical structures. Of course, encoding at this level is tremendously laborious. Should we expect of electronic editions of poetry that they will allow analysis of such information? Or is this a case where it would be a better use of time and resources to concentrate efforts in developing new parsers and new electronic dictionaries that will perform by electronic means what now requires extensive manual markup (see Chisholm and Robey 100)? [2] How does the WWP handle such questions? How extensive will the markup be of the texts in the Cambridge Wordsworth? As is the case so often in these discussions--whether about the amount and kinds of annotation or about the specificity of the tagging in SGML--the freedom of the medium itself can become quite a burden. Blake's proverb "Enough! or Too much" might do very well as the electronic-text editor's motto.

The Penn Frankenstein project brings another proverb of Hell to mind: "The road of excess leads to the palace of wisdom." A very different kind of edition from the WWP and the Cambridge Wordsworth, the Penn Frankenstein aims to present the 1818 and 1831 texts, encoded in HTML, in the context of an astonishingly large web of hypertext and multi-media links. I had a great deal of fun working with the small excerpt made available to this seminar, and I look forward to having the entire "virtual library" (the equivalent of 20,000 printed pages!) available soon. I can see the edition making an immediate impact on my teaching of the text, and (oh, fond hope!) on my students' papers. My college library has a serviceable collection of primary texts and major books of criticism. Journal holdings, however, are not what one might wish. This edition will give students access they otherwise would never have to a large collection of secondary (as well as rare primary) sources. Just as important, it will provide assistance in keying these sources to relevant themes and passages in the novel.

As Jack Lynch's candid overview of the project suggests, the chief virtues and chief problems of the edition stem from the same source--its abundance. I find especially interesting Lynch's discussion of questions stemming from the editor's role as the creator of links: "Ten thousand files can be linked to one another nearly a hundred million ways, and the editor's job is to determine which of these almost limitless possibilities is likely to prove useful to readers." Choosing "fifty or sixty thousand" of these links means excluding "the other 99.95 million." The task, Lynch says, made the editors feel alternately like tyrants and anarchists.

Like the SGML encoder, whose commitment to flexibility involves an attempt to anticipate the unforeseeable analytic needs of future users, the hypertext editor is engaged in an attempt to make the text useable in ways and for purposes that are not necessarily his or her ways and purposes. The reader, theoretically, is in control. The paradox is that the more abundant the resources included on a CD hypertext, the more exclusive the editor must be in the act of establishing links. As I worked with Robert Walton's first letter in the Penn Frankenstein, I found this difficulty facing me quite directly in my visceral response to the relationship of black and blue type on the screen. I may be revealing here nothing so much as my own perversity, but my response to the sheer abundance of "hot spots" was to feel frustrated that any word (with the possible exception of articles, prepositions, and conjunctions) should be left cold, black and inert. Why is "Inspirited by this wind" a gateway to exploration (it takes us to the Prelude, book 1, and to a discussion of the "correspondent breeze") but not Walton's self description as experiencing "the joy a child feels when he embarks in a little boat"? (Blake intrudes yet again: "Less than All cannot satisfy Man"!) Such uneasiness persists in following the links, as one sometimes gets the feeling, in the midst of all of this "anarchy," of being guided quite firmly through issues that have been identified as thematic.

A couple of other questions emerged from my experiments with this portion of the Frankenstein project. Some of the problems I encountered no doubt are the result of "bugs" in the demo alone (and not in the complete, CD, version); nevertheless, I offer them here in response to Lynch's request for feedback. I found the lack of some sort of prominent identification of the text I was viewing (1818 or 1831) a little disorienting when I would click back to the annotated text after viewing variants. Could a running header or footer be included? I am curious to know whether or not massive annotations, such as the full texts of Paradise Lost or of Prometheus Unbound, will be searchable and navigable. I received no directions about a search engine in the demo. Also, the text of Paradise Lost was not identified: which edition? why this one? Finally, the font used for Paradise Lost is decidedly less attractive and readable than the font used for the texts of Frankenstein and in the editorial commentary. Will this be the case in the "real" edition?

I offer the foregoing remarks from the point of view of a potential user of these editions--the man on the street pulled in for a focus group. I am very much interested in putting the high-priced hardware on my office desk to good use in my teaching and scholarship. I've read enough by the electronic-text theorists to be hopeful about the possibilities on the horizon. At the same time, my experience with what is available, though quite limited, has been uneven enough to make me skeptical about many of the claims made by the enthusiasts. Having the opportunity to think a bit about these three fine and diverse projects in combination, I come away genuinely enthusiastic about the practical applications of the new technology for serious scholarly and pedagogical uses.


1 Robert N. Essick, William Blake and the Language of Adam (Oxford, 1989); Alvin Kernan, Printing Technology, Letters, & Samuel Johnson (Princeton, 1987); Jerome McGann, "'A Thing to Mind': The Materialist Aesthetic of William Morris," Huntington Library Quarterly 55 (1992): 55-74; Joseph Viscomi, Blake and the Idea of the Book (Princeton, 1993); Lee Erickson, The Economy of Literary Form: English Literature and the Industrialization of Publishing, 1800-1850 (Johns Hopkins, 1996).

2 Chisholm and Robey , "Encoding Verse Texts." Computers in the Humanities 29 (1995): 99-111.