Computers, Writing, Rhetoric and Literature
Prev. | Next | Main
A Critical Look at How Some Scholarly Hypermedia Sites Deal with the Problems of the Medium
In practice, nearly all hypertext projects are experiments in defining a genre. This section looks at several scholarly hypertexts to see what their structure and content reveal about their creators' assumptions about the functions of the medium.
WORP (Women of the Romantic Period) Project
Which is the best version of a text to use? If we offer variants, will one serve as a main text off which the others feed or will they all be equally important? If we are comparing different texts, how do we assign value to each? Hypertext offers the opportunities to break away from some of the linear and chronological constraints of printed text, but only if we pay attention to the way we structure and order our hypertexts.
The WORP Project at the University of Texas offers a playful solution to the problems of privileging of texts in a multi-text environment through the creative use of scrolling frames. WORP's hypertext uses Polwhele's "The Unsex'd Females" as a door to studying the work of the female writers criticized in the poem.
The page is split into two vertical frames. On the left appears the text of the poem, with linked footnotes. When clicked, these footnotes appear in the right-hand frame. The text of these footnotes can lead to further discussion of the female authors they refer to. Alternatively, users can view a linked list of all the writers referred to in the text.
The page's authors note that, while at first glance the structure seems to privilege a male-dominated discourse, with Polwhele's sexist text as an entryway at left and the women writers grouped in a list at right, this structure is quickly subverted. Following a link to an on-line work by a particular author causes that work to appear on the left-hand side of the screen, supplanting Polwhele's text. "The hypertextual construction of WORP, then, doesn't attempt to obviate issues of privilege and status. In fact, it makes them more obvious by actualizing privileging and elisions which are often discussed only in the abstract."
WORP does not attempt to provide any sense of what Polwhele's original text looked like, as does the University of Virginia's version, which provides both a digitized facsimile of the original and an e-text version which has page breaks that match the original. For scholars interested in doing bibliographic work, or those concerned with how Polwhele's audience might have reacted to his text, the Virginia site has some advantages. As Jerome McGann notes, we need all versions of a text in order to fully assess it for ourselves. The WORP site betrays some prejudices about the value of "The Unsex'd Females" in its construction and use of the poem.
Interestingly enough, however, WORP's text version is easier to read than Virginia's, as it appears on screen as a continual text. In this case, sacrificing "realism" results in a superior research tool at the level of the text itself. The WORP site highlights the problems of balancing the needs of users in creating a scholarly hypertext.
The Dante G. Rossetti Hypertext Archive
Jerome McGann's Rossetti Archive offers an intriguing approach to the problems of documenting the work of an author working in multiple mediums with multiple textual variants: use everything!
For Rossetti, it seems to be a reasonable solution. Not only did Dante Rossetti work in both written and visual mediums, but his written and visual works often commented upon each other. Notes McGann,
- Rossetti's work habits were such that these structural complexities of his work get vastly extended. Rossetti was an obsessive reviser of his written work, and these revisions are carried out at every level of the writing: he works and reworks words, phrases, passages, and he rearranges "finished" units into dizzying sets of variant organizational units.... All these features of Rossetti's work pose a complex and hitherto unsolvable editorial problem. One cannot properly study or appreciate Rossetti's work without having access to all of it.
And the archive attempts to provide access. For the poem
"The Blessed Damozel,"
for example, the archive offers eight text-versions from two manuscripts and six printed versions, with color
facsimiles of each where appropriate or legally possible, plus commentaries on the poem's textual history, printing history, historical background, literary significance, pictorial references, autobiographical references, and a bibliography.
While the archive manages to condense what would otherwise be an unwieldy amount of material into a very concise and elegantly designed set of pages, there are few links between the individual elements making up the content of a particular page. This is especially problematic when it comes to comparing textual variants. No way exists to compare two versions side-by-side, nor are variant readings offered within a single version of a poem, nor are in-line or linked annotations available. While McGann hopes to show the text's "nervous structural features," it is only with a good deal of searching and restructuring of the data at home that this can be accomplished. The Rossetti archive thus fails to make use of one of the most valuable aspects of hypertext. Perhaps the advent of frames will allow for some of these problems to be overcome.
Pride and Prejudice Hypertext
H. Churchyard's Pride and Prejudice Hypertext is an interesting attempt to make use of the linking abilities of hypertext to perform complex indexing and searching of a textual work. Nearly every item of interest in the text is linked to one of five indexes: a list of characters, a list of events in chronological order, comments on random topics, an index to the motifs of "pride" and "prejudice," and a list of important places, with a map. All links are bi-directional (it is amazing how often hypertext creators do not include this simple but necessary navigation tool), so that the user can scroll down the list of character names, say, jump to a passage in the text, and then continue from the text back to the list or to another index. The indexes themselves are extremely useful, especially with regard to plotting characters -- every character appearance, dialogue, and description is accounted for.
In many ways, this is what a hypertext should be. The editor has given care to determine which structural elements are important and has gone about indexing them. On the other hand, this pre-selection process does not give users choices in the matter of how they will use the material. At this stage in the development of hypermedia and html, such a case is, perhaps, inevitable, but, as Jerome McGann reminds us, we must think of ways to offer users the ability to make use of the material, even in ways we have not yet envisioned.
Churchyard's indexes provide an excellent way into the text, but, unfortunately, they overwhelm the text,
making it a nightmare to read or even scan. Every third line has a link. Every occurrence of a proper name
is highlighted. No indication is provided as to where a link will send readers, whether it be to a particular
index or a useful gloss. There is no way to distinguish between different categories of links nor to
dispense with unhelpful ones.
Compounding these problems is the sheer size of these pages, due to (1) unnecessary pictures, which make the document pretty but do nothing for textual searching, and (2) the vast amount of data per page. For example, all the characters named in the character index have their text references on the same page. There is not a page in the entire web site less than 50K, and many are much larger.
Some of these problems could be solved with the use of frames, of course, and with more careful breaking up of pages. However, the nature of this site raises one of the central issues in hypertext design -- the need for compromise in nearly every step of the development process. The usability of this site makes for an nearly unreadable text. Satisfying both these needs is a significant challenge of hypertext.
Finnegans Wake
Jorn Barger's Finnegan's Wake project shows off some of the possibilities of the medium. It is an annotated version of one paragraph from Finnegans Wake (chapter 4, paragraph 1). The annotations were contributed by members of the Finnegans Wake mailing list in 1991.
The amount of material contributed here is phenomenal. A single phrase may produce twenty different and useful glosses; the complete project takes up 71K, or about 12 full pages of text.
This is the sort of information one would like to have on hand when reading a hypertext. However, its organization -- which is in part constrained by the limitations of the medium -- leaves something to be desired. Because every few words provides the source for annotations, the text is broken up into individual sentences, albeit long ones. Each sentence appears twice, first as an unlinked bit of straight text, then underneath as linked text. However, because there was no way to distinguish between separate though adjacent links at the time (and still no satisfying, systematic way now), each annotated fragment appears on its own line.
The fragmenting this method causes renders the hypertext almost useless as a textual resource. It is just too difficult to scan the text for useful information.
As hypertexts evolve, annotations will become much more sophisticated. We need to find ways to distinguish among (1) different classes of links (color coding has its limits), (2) adjacent links, (3) links of multiple classes, etc., so as to make use of our annotating potential.
The next section looks at why Ulysses is the perfect hypertext subject.
Prev. | Next | Main