Internet Shakespeare Editions

Hypertexting

Silver Currents, page 2

O, two such silver currents, when they join,
Do glorify the banks that bound them in.

King John, 2.1.442-3

It has been my pleasure in recent months to work with a group of younger scholars, each contributing to the Internet Shakespeare Editions, in co-editing a special issue of EMLS, with Eric Rasmussen, "Monitoring Electronic Shakespeares."[Note 1] The issue, which will be published within a few weeks of the circulation of this paper, provides something of a snapshot of the way a new generation of editors is bringing Shakespeare to the computer screen. Each scholar is grappling with the question of what the text has become as it has moved from paper to pixels. In my introduction to the special issue, I quote one of the essays, where the writer speaks in terms that sum up the confidence of his peers: "My intent here is not to restate the argument for an affinity between Shakespearean content and electronic media, but, assuming that argument's validity, to explore its consequences in the practical terms of humanities computing and textual studies" (Galey 2). Collectively the articles suggest that bringing Shakespeare to screen will be a complex and demanding exercise, one in which the editorial enterprise has an opportunity to bring new life to the texts as they are made available to an audience of unprecedented size. My own intent in this paper is to allow the voices of these scholars to instruct us in their experience as they explore the potential of the hypertext edition, and to suggest further ways in which computer-mediated editions may modify our sense of text.

E-editors spend their time in much the same way as p[rint]-editors. They establish a text, collate relevant editions, annotate, write introductory materials, and provide supporting documentation. But each of these activities changes in both magnitude and process in the electronic medium.

The text

Especially with complex texts, the e-edition seems to have it easier than print, since there is room for separate, interconnected editions of the Early Modern variants. Patrick Finn, however, suggests that there is some need for caution. He argues that e-editors need to construct environments that are hospitable: the potential spaciousness of the e-text can lead to a complexity that is self-defeating, as the reader clicks from node to node of a network with no necessary sense of direction, and where a reader can enter the text at any point from a search engine not knowing anything about the structure of the edition as a whole. Thus, despite the current tendency towards deconflation, the editors of the Cambridge CD ROM King Lear in Performance chose to use what they called a Finder text -- a conflation -- as the central text from which others could be accessed. The ISE editions will perforce require a similar central text. It may be less important what kind of central text it is than that its principles are clear, and the choices it embodies capable of being fully traced by a reader who has transcriptions and images of all relevant originals available through links.[Note 2] The challenge is both to create a familiar and scholarly finder text and to make the structures of display hospitable enough that navigation does not become confusing. Internet users -- even scholarly ones -- like to move quickly to the information they seek, and will not spend hours trying to work out a complex interface.

Collation

As part of the discipline of deciding on their modern text, editors collate originals and editions of significant importance with the most careful attention to detail as they try to establish its provenance, even in a briskly sceptical discipline where there is considerable debate about even the possibility of deciding what kind of manuscript lay behind the print version. It may seem that the ready access to representations of the originals reduces the need for collation, but in the design of the ISE editions the editors are asked to create highly accurate diplomatic transcriptions of those originals, and to encode the text with information that both describes its physical appearance and its conceptual structure. The rationale for these transcriptions is principally to make it possible for a visitor to the site to search the originals, but the information thus encoded may also provide important information about the habits of compositors, and more accurate information about the patterns of Shakespeare's language than is available in modernized texts.

To transcribe, even as a diplomatic transcription, is also to interpret. The ISE transcriptions are encoded to quite a high level of detail. For example, they record type-forms that cannot yet be fully displayed in modern browsers: the long-s, and the many ligatures that compositors used.[Note 3] Alan Galey's article in EMLS explores ways in which the practice of encoding the Renaissance text inevitably involves changing the nature of the text, not simply as it is translated from one medium to another, but in our construction and understanding of the literary work. As an example of the effects of making encoding decisions, Galey considers the moment at which a reader of Much Ado becomes aware that Don John is identified as a bastard. Those who read the original text, or who view the play without helpful program notes, will not be made aware of his illegitimacy until the fourth act; those reading a print version with its list of dramatis personae, or encountering the play through encoding practices that require metadata that normalize speech prefixes, will know of his origin from the beginning of the play. The potential significance of the moment that Don John's illegitimacy is revealed is lost.

The practice of encoding the originals will also, inevitably, reduce the information that is available in the physical original or its graphic image. I mentioned in the previous paragraph that the transcriptions are encoded to "quite" a high level of detail. ISE texts cannot reproduce all the characteristics of any one original. Where would one stop? Should stains on the paper be encoded? Should damaged type be identified? In realistic terms, there must be a stage at which the law of diminishing returns sets in, and for the sake of our editors the encoding adopts some relatively arbitrary limits on information. In describing her experience of editing Cymbeline for the ISE, Jennifer Forsyth records her delight in discovering that the use of spaces around semicolons can be used as a way to differentiate between Folio Compositors B and E (Forsyth 2004) -- but for the sake of simplicity (and perhaps the sanity of our editors) ISE transcriptions normalize spacing, so this information is lost in the transcription. The advantages of having accurate transcriptions of quarto and Folio texts are considerable, but libraries are not yet out of business: the analogue version always encodes more data than the digital.

One attractive by-product of collation in an electronic edition is that the information it provides will allow for the development of texts that respond to a reader's choice in providing additional information. Bernice Kliman's Enfolded Hamlet is one example of a text that uses various symbols and colour coding to reveal different readings from different texts; in an ISE edition it will be possible for a reader to choose to view a display that highlights the origins of specific readings.[Note 4] Two years ago at the meeting of the SAA, I made a suggestion, which at the time I thought more mischievous than serious, that in some cases where modernization reduces the "semantic field" of meaning of the Early Modern text, it would be possible to restore something of that meaning through the very Web-inspired device of animation.[Note 5] Somewhat to my surprise, some of my colleagues have found some promise (and pleasure) in the possibility that animation offers. Some of you may have seen Sonia Massai's paper at the most recent meeting of the SAA, where she uses this technique to dramatize speech prefixes that are ambiguous in the original (see Massai 2004 and associated links).

Annotation

Print editions of Shakespeare continue to experiment with different formats for annotation. The traditional text puts them at the bottom of the page, keyed by line number; one variant of this practice is to indicate in the text, or margin, whether there is an associated note. Some student editions put notes on the facing page. Norton texts put brief glosses in the margin, with somewhat more detailed comment at the foot of the page. While different editions will provide widely varying detail in their annotation, all are limited by the physical constraints of the white space available on the page. The e-annotation is not so limited, and the temptation might be to annotate everything in great detail. So often a simple note is the result of hours of digging, and it would be gratifying to an editor to be able to demonstrate how much work went into a single annotation by including all the materials that had been referenced. Forsyth raises some interesting questions about annotation in the electronic medium, looking both at the possibilities and temptations of copia, and at the largely unexamined convention that the modern editor should present at least a mask of objectivity in annotation (Forsyth 2004). Seminars at the SAA amply demonstrate that editors are passionate people, and it seems odd that they should be required to disguise their undoubtedly strong feelings about the texts they devote so much time to.

One way of making extensive annotation hospitable is to provide a user with some meta-information about it. ISE annotation is divided into three levels: basic glossorial explanations, more detailed scholarly references (compare the kinds of annotation that a typical Arden edition will provide), and fully-fledged discussion of cruxes of the kind that might be included in a print edition as an appendix. Editors categorize their annotations by these three levels, so the published version can indicate by various signals which a user will be invoking with a mouseover or click. Massai's paper in the EMLS issue demonstrates examples of the three levels of annotation, with linked illustrations of possible ways in which they might be displayed. The current displays are not by any means indicative of how the editions will actually be displayed, however. All ISE texts are encoded in a standard, generic code that allows for display to be modified without changing the underlying texts; one important process the texts will undergo is familiar in the computer world, but rarely used in print: beta-testing. We will be testing a number of alternative ways of displaying annotation, with the aim, again, of creating the most hospitable environment for the user.

There is an additional kind of beta-testing that would benefit all editions, print and electronic. What do readers want to be annotated? Readers will be frustrated both by notes that tell them things they know already, and by passages they don't understand, but which are not annotated. As cultures, language, and educational systems change, readers will need different kinds of assistance. In the Guidelines to editors of ISE texts I remark that editors would be "wise to assume that the current generation of university students is fundamentally ungodly," and I have been reminded recently of a change in language that makes it necessary for to point out to my students that when Lysander claims that Demetrius "[m]ade love to Nedar's daughter, Helena" (1.1.107), the passage does not necessarily mean that they have had sex. Conversely, I find I don't have to explain so many bawdy passages to the current knowing generation. Although different cultures will require different kinds of information, it would be a fascinating study to test students, say, at entrance to university, to see what they do and do not understand.

Another proposal made by Forsyth is that ISE editions could capitalize on the habits of Web users by creating discussion areas where visitors to the site could contribute information about the plays. Traditional scholarship is nervous about texts that change and appear unstable, but there is a wonderful opportunity here to interact with readers so that the supporting commentary of an edition would be shaped by their needs. The design of the ISE site allows for informal discussion (in the Annex) that would not impinge on the peer-reviewed materials in the Library, where the edition of Cymbeline is published. But the editor could modify the edition as a result of comments both from readers and academic review.

Introductory materials and supporting documentation

Print editions of Shakespeare will typically have quite substantial introductory, or "front matter": title page, table of contents, acknowledgements, preface, list of illustrations, and an extensive critical discussion that will often discuss the text and performance history of the play. The structure of hypertext changes the emphasis especially of the essay: since it exists as a separate file, or group of related files, the critical introduction is not given the privileged position of front matter. In fact, given that the majority of Web users these days reach their preferred destinations through search engines, the concept of front matter has shrunk to the title page and table of contents. Sonia Massai's discussion illustrates the importance of this kind of organization of the edition, as it is from the table of contents that a reader will access both the edition and its multifarious supporting documentation.

It is in the area of supporting documentation that the ISE editor will face both the greatest opportunity and challenge of the edition. The electronic space is such that it is possible to create what in effect is an archive of materials: contemporary documents, passages from earlier editions, exempla of critical approaches to the plays, and virtually unlimited opportunity for illustration from performance. The two latter categories, of course, will be subject to limitations because of copyright restriction, but otherwise there is no effective limit on the amount of data that can be included with the edition.[Note 6] The challenges here will be twofold: to define reasonable limits to the kinds of materials that can be considered relevant, and to make navigation through them hospitable.

A careful structure can make a hypertext edition inviting by providing consistent ways of negotiating large amounts of data. But the editor's choices of linked material will not always conform to the interests of the reader. Ray Siemens has recently been arguing that links within electronic texts can be created dynamically by the reader by using reader-generated requests for information, much as Web users create their own paths through the vast amounts of data online through Google (Siemens 1999). Modern search engines, however, can not only find and list the items they find, but, through complex algorithms, can rank the probable relevance of their findings. In effect the computer "reads" the text it searches, guessing what items the searcher is most likely to be interested in by looking for occurrences of the search terms close together, or in headings rather than body text, and so on. As the planned database of Shakespeare in performance becomes a reality on the ISE site, readers of the texts will be able to explore performance independently of the links the editor has set up to illustrate a specific point in the commentary.

The striking success of Google has signalled a change in the way many people access information on the Web. In January of this year, the total number of requests for pages from the ISE site was over 670,000,[Note 7] and over 50% of visitors came directly from a search engine; most came from Google.com (25%) but there were also many visitors from Yahoo (11%), Ask Jeeves, MSN and AOL networks, and the Google sites in Canada, the UK, Australia, Germany, New Zealand, Italy, France and Brazil. Although a reliance on major Internet search engines can be problematic, since their algorithms emphasize the popularity of the sites they index rather than their quality, searches set up wholly within a site will not suffer from this problem; the value to the user is that links generated by them are interactively created rather than pre-selected by the creator of a site; in effect, users create their own access to the information on the site dynamically.

top

Notes

[1] The papers which form the issue were originally presented at two recent conferences: The SAA meeting in Victoria, 2003, and a subsequent paper session devoted to the Internet Shakespeare Editions at the meeting of the Pacific Northwest Renaissance Society in Nanaimo, BC. [Back]

[2] One interesting choice in the current edition being prepared for Othello is that it will use Q2 as the finder text. [Back]

[3] Oddly enough, for all its thousands of available characters, Unicode does not yet have a full character set of Renaissance type forms, so even Unicode-compliant browsers like Mozilla/Firebird or Safari are unable to display all the ligatures. [Back]

[4] Back in 1997, when browsers were just starting to be able to display coloured text, I created some sample pages from Othello that showed how different colours could indicate words and phrases that originated in Q1, F, or in the current edition. See http://web.uvic.ca/shakespeare/Annex/DraftTxt/Oth/OthSample.html. [Back]

[5] The paper, with its illustration, is to be published by College English in a necessarily less animated form; in its original form it may be seen at http://web.uvic.ca/shakespeare/Annex/Articles/SAA2002/index.html; see particularly the page at http://web.uvic.ca/shakespeare/Annex/Articles/SAA2002/rich4.html. [Back]

[6] The Cambridge King Lear in Performance CD ROM describes itself accurately as "a performance archive." [Back]

[7] Requests for pages are a more reliable measure of use than "hits," since hits refer to each request for a file, and a graphic-intensive page will request may files. [Back]

Previous Table of Contents The Annex Top