"Women's Literary History by Electronic Means: the Creation and Communication of Meaning In the Orlando Project"

It would be appropriate to begin by relating D. F. McKenzie's Bibliography and the Sociology of Texts to the Orlando project, and throughout this talk we have in mind the words we have quoted in our title. Since it was launched in 1995, for the purpose of producing the first large-scale, wide-ranging and scholarly history of women's writing in the British Isles, the Orlando Project has been grappling with the construction and creation of meaning in ways that are new to the experience of all of its participants. They are new, too, among scholars in the humanities community. The Creation and Communication of Meaning in the Orlando Project has provided us with challenges we could not then have dreamed of.

But in 1995, when we set out on this project, scholarship on women's writing, of every kind, was pouring in - and had been doing so for twenty years. Now, we thought, we need the overall account of a history. We felt this need with particular acuteness because we had recently completed work on The Feminist Companion to Literature in English, a reference book that had made us aware of the many links and connections and communities among the women who were its subjects. That book, like others, had been based on individual accounts of women's lives and writing. Allan Liu saw it as exemplifying the contemporary practice of literary history and its decision not to be over-arching; it was, he said, built of a succession of "petits recits;" the "grand recit" remained unattempted. David Perkins had demanded, "Is Literary History Possible?" (which Laurence Lipking later answered with a wisecrack: "Literary history used to be impossible to write; lately it has become much harder.")

It became clear to us at the outset that our project might recreate a condition of possibility by integrating the critical and the technological. The potential we saw in this new methodology was of course being widely explored elsewhere, not only for storing vast amounts of knowledge, but also for meeting some of the new needs in - McKenzie's words - "the creation and communication of meaning."

The principle behind the choices we made is (to quote from another Orlando text) "that some of the charges levelled at traditional literary history - exclusivity, linearity, over-reliance on narrative, a certain totalizing or monologizing tendency - constitute a challenge that can be met by producing and disseminating literary history electronically. The history we are producing is formed of thousands of multiply linked and dynamic chunks of text which may be navigated along an almost infinite variety of pathways. Its organization is not linear. The text is multiply and collaboratively authored, in significant part by graduate students. It is therefore comprised of a range of authorial voices, discursive strategies, critical approaches, and theoretical emphases. The narrative which we consider necessary to literary history is multiple, parallel, and fractured, rather than continuous and singular. The electronic form of the multiple text allows it to be ordered and traversed according to various principles and interests. Perhaps most exciting of all, to us, is that the Orlando Project is using electronic text encoding to produce a map, as part of the prose we are writing, of the conceptual structures and intellectual priorities governing our history. This opens up new ways of reading and writing literary history." New ways, we think, of creating and communicating meaning.

In considering our own medium here, we shall not be focussing on the details of how our encoding works. Since the encoding is where literary history and theory most engagingly encounter technology, it will no doubt come up for discussion in the related seminar. But here our focus is on our completed text - or textbase, or composite bundle of related texts - as it will be available in the future. We shall look at the ways in which it functions like a printed book, and the ways in which it works quite differently.

The electronic text archive which we are producing (already running to hundreds of thousands of words of historical-critical prose, written and electronically encoded by the Orlando team) deals with discrete authors and with discrete historical events and processes. When the whole Project is complete, this electronic text will be accompanied by overall narrative-discursive accounts, available also in the form of printed volumes. These will provide a synthesising view of conditions, developments, movements, trends: in short, of the bigger outline. They will be available both on line and in print, as books. They will be single-authored, yet in their electronic form will be linked with the multi-authored electronic text archive. Some of our readers, then - those choosing to access both the books and the electronic text - will be moving in and between the two environments of print and screen, linking them as they choose.

This hybrid plan gives us cause to rejoice at, and to rely on, Don McKenzie's vision of print as only one form of text, and his insistence that all the new forms of non-book texts need to be taken seriously, to be studied and discussed. In our project electronics is the medium, while literary history - heretofore very largely the history of book texts - is the material. In this lecture, not for the first time, we turn aside to examine our own medium, whose impact on the kind of history we write has been formative.

Not only our writing has been formed by it. Like others seeking to realise the potential usefulness of the new tools, we find that the medium has also had major impact on some of our institutional practices. It affects how we do our work together; how we work with collaborators who are our graduate students; how we see our authorship, our intellectual property, the future of this work when it leaves our hands. A necessary part of doing our work has been observing the impact of our medium on us and on what we are making. Does the electronic medium make our history of women's writing in the British Isles more accessible, or does it constrain complexity? Such self-referentiality is vital for anyone trying to keep up with the constant new issues and new potential which work in a new medium opens up around them.

Several more points confirm our alliance with the school of Don McKenzie. Firstly, the principles we follow in dealing with the history of women's writing relate closely to his principles for approaching texts. Like him, we see the history of texts as inseparable from the histories of broader social and cultural concerns: from, in McKenzie's words, "the composition, formal design and transmission of texts by writers, printers, and publishers; their distribution through different communities by wholesalers, retailers, and teachers; their collection and classification by librarians; their meaning for - and . . . their creative regeneration by - readers."

Also like McKenzie, we seek to get away from "our recent culture's 'single-minded obsession with book-forms.'" This obsession has been one of the factors making against women's acceptance as part of the discursive field of literature (women's texts have had, on average, a longer and more difficult passage from manuscript into printed form than those authored by men). It is essential to seek beyond the printed text if we are better to understand women's historical relationship to the act and the products of writing. McKenzie's four principles of bibliography run parallel with our own principles of literary history - especially his final principle. As bibliography is ineluctably social, so too is literary history. Our emphasis will fall on its social interactions.

One strong reason (among several) not to focus literary history on great individuals is that this assumes implausible degrees of agency for those individuals (if not in particular cases, then in principle). We see the writers' agency as always in dialogue with the forces of her material conditions and the broader conditions of her culture, as well as with the influence of preceding texts. Students of women's writing are perhaps not in danger of overestimating the agency exercised by the authors they study; but they are certainly prone to presentism, to an undervaluing of questions of specificity and the detail of cultural context.

On the one hand we set out to situate the history of women's texts in relation to the histories of women and of wider culture. On the other hand the history we are writing constitutes a web of interwoven smaller histories. We sought to embody this approach to literary history in the sets of tags we designed, early in the project, for encoding our discussion of authors, texts, and history. This kind of electronic encoding is now widely familiar in its simpler forms (the structural tags used to mark paragraphs, or acts and scenes, the tags used to mark names and dates). Our major originality lies in the detail and depth of our tagging (with, for instance, interpretive tags for concepts like intertextuality, and further distinction of different types of intertextuality).

As it happens, all our writing about texts falls into three broad categories, all identified with slightly different emphases by McKenzie in relation to bibliography: the composition and transmission of texts (which we call production), their formal design (which we call textual features), and their creative regeneration by readers (which we call reception).

So our tri-partite organization of our discussions about writing can be mapped onto McKenzie's theory of bibliography. This highlights the crucial intellectual function of text encoding for the Orlando Project: the encoding scheme is a translation of the theory by which the literary history is created. The theory is thus embedded in the text we are writing, and governs the ways in which our material will be delivered, used, and read. When we as scholars write for print media, whether monographs, essay collections, articles, or reference works, the conventions of print technology are so naturalized that we give them little thought, even as we adapt our writing to the demands of these different forms of print. In the familiar process of writing critical prose, we rely on the immense powers of nuance available in language to communicate meaning, but we need not, unless we so choose, be explicit about our fundamental intellectual priorities. To use text encoding in the process of composition, however, as the Orlando Project is doing, places entirely different demands on us as scholars.

Text markup can be performed to different ends. Translation of a pre-existing printed text into an electronic medium may be a complex operation, or, if the aim is simply to get it online in HTML for easy access and free-text searching, a quite straightforward matter of formatting it for readability. A number of online resources, such as Brewer's Dictionary of Phrase and Fable, use electronic markup for headings, paragraphs, etc., without fundamentally reshaping the printed text. Nor does the encoding offer significant information about the content of the text or the conceptual organization of the material.

Our markup occupies the other end of the spectrum: it organizes the text we are writing not only according to the kind of visual organization so familiar to us from print culture - headings, subheadings, paragraphs, citation of sources - but also according to our intellectual priorities as literary historians. The demands of the computer have involved us in what scholars in humanities computing (Allen Renear, John Unsworth, and others) are speaking of as "knowledge representation." What this means essentially is that because we want computers to be able to process our writing in particular ways, and because computers still have a lot to learn in understanding the semantic content of language, we have to be as explicit as we can about what we are saying by means of our encoding schemes. The pressures in production of scholarly electronic text have taken creation and communication of meaning in a new direction.

As writers of a text encoding scheme and the text that employs it, we are creating prose that retains its full potential for nuance, but the embedded encoding permits new ways of manipulating that text - not only of searching but of selecting and recombining - and thus new ways of reading it: of making sense, of finding new senses. The text thus becomes multivalent in new ways. The encoding system, or tagging, or markup (all three terms are used) can achieve this multivalency without being obtrusive. It does not need to be visible. What you will see, as readers, is the prose, plus the underlying markup only if you so choose.

Words, or sentences, or paragraphs, are marked up (enclosed in tags) so that the computer can recognise them. Whatever the eventual appearance of the Orlando text on the screen (something else to which we are devoting a lot of thought) it will look like paragraphs of varying length in prose, without the boxy, or columnar, or systematised appearance typical of the database.

To think of our (future) completed, complex text, is to think of what it will be like to read and use it. We hope that our readers, or users, will engage with it in several distinct ways, including at one end of the spectrum very familiar reading practices, and at the other end practices which, while related to things that readers already do with books, are very substantially new. These new practices are supported by the mapping and structuring function of the tags.

We speak of readers, because the Project will offer a wealth of reading matter, and also of users, who will exploit the potentiality of the new medium. Not that we present this as a neat dichotomy. An idea of "users" does not alter our text's alignment with books, only aligns it particularly with reference books like the OED or DNB. Even narrative histories are more often dipped into and read in sections than they are read from cover to cover. That fact does not diminish the pleasure they offer by means of traditional, continuous reading. We too shall afford such pleasure, while almost certainly rendering a cover-to-cover reading impossible, and while hugely enhancing potential for other, non-continuous kinds of reading.

You will be able to read in the traditional manner those sections (we call them documents) devoted to specific authors. In print these would run from under a page to something like 30 or 40 pages: the length of a reference book entry or critical essay. But also readers will be able to select and recombine parts of documents in a way you can only do with a book by taking scissors to it. And as the documents range from hundreds of words to thousands, so recombined selections of text will range from the equivalent of a footnote to, once again, a substantial essay. People will do this selection and recombining in the same spirit and for the same purposes that currently we use an index to locate pieces of needed information, but also in order to produce newly combined texts for consecutive reading. These texts will be new, and so is the form of looking up, or browsing, enabled by our markup. Browsing texts - shifting, around, flipping back and forth, looking things up, whether in the OED, or library catalogues, or the internet - is nothing new. But browsing in a deeply-encoded electronic text is something still almost untried.

The transition from printed book to electronic screen is often likened to the shift from oral to written culture. That nicely catches the magnitude of the change; but in other ways a better parallel might be other technological leaps forward in the history of writing. When the bound book superseded the scroll, again when movable type superseded manuscript transcription, and again when the alphabetical index was invented, it was on each occasion the management of knowledge which took a great leap forward: expressive or creative writing was less profoundly changed by the new technology.

Now we are working at another great technological faultline: with the electronic. Again, we perceive less change in creative writing and traditional reading than in knowledge representation through markup or encoding. The Orlando textbase will facilitate easy movement back and forth between our electronic text and printed books - whether they are the sources we cite or the texts we analyse - and between our electronic text and other electronic texts - the genuinely scholarly and reliable ones, such as the Corvey and Perdita Projects, which are fortunately on the increase.

All these forms of electronic browsing are immensely easier to do than browsing exclusively in books. The recombinations of text of which we have been speaking are of course not impossible with books. That process of juggling a dozen books which you have annotated or bookmarked - that is a familiar condition to scholars. The books themselves, it sometimes seems, actively resist such work, but out of it, with effort and skill, can be drawn a document using related excerpts from all those books. Orlando readers will still be doing that work - when they go back and forth between our text and its sources, say. But within our multiple text they will find the process a lot easier.

An electronic text generously offers itself for copying and recombination. Therefore the habit has already developed of cutting and pasting titles from the electronic library catalogue rather than court error in transcribing. The Orlando electronic text, unlike the Bodleian catalogue, is being written with the express purpose of later selection and recombination. It is collaboratively-authored, as we have said, both in the sense that different documents have been initially drafted by different people, and also that every document bears traces of the work of several team members. But it is equally important that the text is collaborative with its eventual readers and users (McKenzie's creative regeneration). When you read our ready-written documents on authors, you assess, in traditional fashion, what we have written for you. When you do a search which brings together successive passages from different documents, they recombine in a new document which, in this form, the Orlando team did not write and may not have read. In that case, there is a real sense in which the reader constructs her own text.

The first available way of relating to Orlando, the familiar process of consecutive reading, centres on individual authors. It is, in this electronic textbase, modified by embodying the potential for back-and-forth switching of attention. Should you be reading the life of Mary Russell Mitford in the Orlando textbase, you will be able to click on the name of her teacher, Frances Arabella Rowden, and discover that Rowden (herself a published writer in several genres) was a significant force in the forming of four other published women writers. Or you might want to see what else was going on in the London theatre world during the years of Mitford's tragedy-writing. That is the equivalent of traditional consecutive reading punctuated by index use or consultation of other books.

Our focus on individual authors reflects our conviction that the individual's history and subject position is the irreducible starting-point of literary history; that the forces of cultural change and the transmission of ideas function through their effects on individual sensibilities; that the production of texts is a creative activity of individuals or of combinations of individuals - no matter how far away these individuals are from enjoying unrestricted agency.

Author documents offer fairly extensive accounts of events and stages in lives and in writing careers. Meanwhile parts of those documents (either brief statements or whole passages) can be lifted out in groups and recombined.

The second way of reading the Orlando textbase is the reading of passages selected and combined according to the principle of chronology. Any statement in any part of the textbase which has been written to attach to a particular date is available for lifting out and instant recombining into the form of a timeline. When anyone on the Orlando team, in the course of writing a continuous document, makes such a dated statement, they write it in such a way that it will be able to make sense on its own, away from its parent document. Statements in the Mary Russell Mitford documents are available as contributions to a timeline on Hampshire (which begins with a thirteenth-century abbess interested in drainage works, and ends lectures by Kathleen Raine and P. D. James, and on the road protest at Twyford Down).

This chronological function is the means you would use to refer, as we said, from the documents on Mary Russell Mitford to a sketch of the London theatre when she was writing for it - or from the documents on Nancy Cunard to a quick timeline on Paris, and from Caryl Churchill to one on the Royal Court Theatre. Chronologies offer a consultation or reference process, a digression from the consecutive reading of an individual document.

But equally, such timelines are documents which can themselves be read in a consecutive manner - although they have not been traditionally composed, not written as a whole by any one member of the Orlando team, but are in a real sense composed by the reader who evinces a wish for them. Of course a timeline on a British woman writer - say, Elizabeth Barrett Browning - comes largely from the documents on that writer, and has been planned by the team member who wrote those documents. The chronology of a life is now a familiar genre: the Orlando chronology on Elizabeth Barrett Browning constitutes a summary or abstract of the documents on her, plus a few statements from documents on other writers (Mary Russell Mitford, Elizabeth Siddal, et al.).

An Orlando timeline for a male writer, however, will differ significantly from the chronology supplied in a volume of that man's works. It will, predictably, include dates for birth, death, and a selective number of major publications. But it will very largely focus on those writers' interaction with women's literary history. T. S. Eliot's personal and professional relations with Virginia Woolf will bulk larger than in other chronologies of Eliot. Here we are back at the social aspect of the creation of literature.

Readers will come to our textbase with their own research interests; we cannot possibly foresee the kinds of searches they will want to do, the kinds of new documents they will want to compile. We anticipate interest in timelines on historical figures, on institutions, places, etc.. Timelines on words - like witch (as in witchcraft) or railway - produce a fascinating mixture of fairly predictable historical milestones with details from writings by women which may well be new even to an expert on the history of witchcraft or of railways.

Timelines on an institution - parliament, say, or Oxford University - incorporate a strong concentration on the relations of those organisations to women as a group and to issues regarding gender. They nicely exemplify the kind of multiple mini-histories which make up our composite history. Parliament has a fairly continuous story in its Orlando timeline, from the beginnings of the Poor Law to the abolition of the House of Lords as we knew it, but there is much more detail given for the period of the struggle for women's suffrage.

Oxford University gets a good deal of coverage as colleges were founded during the thirteenth to the sixteenth centuries, and such coverage is shared with standard reference sources. But printed histories of its serenely masculinist centuries are no doubt oblivious of its being mocked in 1691 by a local satirist, Alicia D'Anvers, or of jokes floated in the eighteenth-century periodical press which have their basis in the fact that the university was closed to women. This is unquestionably Oxford as seen in women's literary history. A century after those eighteenth-century jokes we reach the long struggle of women for admission as equal members of the university community, and in its train an army of women writers equipping themselves with Oxford BAs, most of them in English.

Nobody on the Orlando Project sat down to compose a timeline for Oxford, and nobody sat down to compose one for the year 1928: the year that Virginia Woolf's Orlando appeared, our namesake and inspiration, and a highly significant year for literary history even apart from that fact. The Orlando timeline you would find for that year has been composed by accretion of dated statements in authors' documents and from coverage of historical topics. For instance, some important research in Germany into the menstrual cycle would have been contributed to the 1928 timeline by a graduate student working on the history of medicine.

The chronological function constitutes (as well as a handy reference tool) a source of innumerable discrete documents having their own validity as reading material. Just as the provision of documents on individual authors reflects a belief in the importance of the individual career as a component of literary history, so the provision of the timeline function reflects our belief in the importance of chronological sequence and context for understanding literary history. It was by working on individual writers on the one hand, and on topics in history on the other - history of science and medicine, history of transport, history of education and publishing and domestic objects as well as power and politics - that we have built up the material for potential recombination. The chronological capacity in our encoding exploits the potential of time as an ordering principle in human experience.

If author documents offer one form of reading in Orlando, and chronologies offer another, a third kind of reading is offered by newly recombined documents constructed on the basis of theme (very broadly and variously defined). That is, sections are picked out and recombined by means of our encoding tags. We hope to give you some idea of the potential results, without going too far into detailed explanation of our markup system.

Turning back to the Mary Russell Mitford document, we find it devotes several paragraphs to financial issues - a huge lottery win when she was ten, her father's progress from great spender to imprisoned debtor, and so forth. Each of these paragraphs is enclosed in a tag we call Wealth (though it tends to contain more material about poverty). A reader interested in the finances of literary women might lay beside those paragraphs the Wealth paragraphs out of other biography documents from the same period. This would be using the encoding tags to generate a free-standing, not-previously-composed document to read: a pithy, granular, somewhat jerky account of women's relations with money in the early nineteenth century.

A new document of greater complexity could be created by combining a search on the Wealth tag in biography documents with searches on various tags in writing-career documents (like Material Conditions and Motive). Then both Mitford's cramped workspace in her tiny crumbling cottage, and her passionate pursuit of literary earnings, could be combined with her family's financial vicissitudes into one part of a complex story bringing in the financial pressures felt by other writers of the period, like L.E.L., Caroline Norton, and Marguerite Blessington.

The Education tag provides another example of the way that encoding generates new pathways through our material, and of how this process, like the use of individual writers or of dates, reflects some of our beliefs about literary history. Every biographical document contains a paragraph or several about its subject's education, enclosed by an Education tag. Naomi Haldane (later Mitchison) would appear in a timeline of Oxford University, since she spent some time as an Oxford Home Student, without taking a degree. To understand the significance of her experience, it is helpful to lay the Education paragraphs from Mitchison's biography document alongside some Education paragraphs from semi-contemporary women writers; for one individual's experience of university lectures is just one detail in the broader picture of educational opportunities - or lack of them - open to women before the first world war.

Documents created by searching on markup make less smooth and pleasant reading than a traditionally-composed author document (or than a timeline), because their component parts were written neither to mesh together (as in the author document) nor to be self-sufficient (as in the timeline). It will always be possible to move directly into the complete document for fuller context. Such searching is intended first and foremost as a way of selecting and lining up documents for the Orlando user to turn to and to read in their entirety. But it produces also another alternative among the project's ways of reading: an alternative avenue into the material.

Searches resulting in screens to read are not the only kind which have value. Many chronological searches yield no extended series, but only a finite nugget of data. Sometimes highly selective results are the desired goal. A user may not want a general picture of education for women in Naomi Haldane's day, but may want to see what modern writers we have with a BSc rather than a BA. Or you might have no general interest in intertextuality but want only to check who it was who wrote that prequel to Jane Eyre.

Having tried to speak of the users' future experience, we switch briefly to our own, for something about the strange and severe discipline of writing prose and simultaneously encoding it. After a while tagging goes in tandem with thinking. The need to tag - to reveal to future users the concepts and issues in what you are writing - comes to affect the way you write. For instance, the issues raised by a seventeenth-century Quaker who goes on missionary preaching journeys and gets thrown into jail. A writer of prose for print might put this in a single sentence, but if you wish the encoding to stress the various aspects of the incident's Location, Occupation, Politics and Religion, a single sentence is impossible and the discussion must be fragmented across several tags. These fragments, you are aware, can be recombined in ways you cannot foresee, not by chance, but according to the structures of meaning inherent in markup. But this makes you aware that you are a part of the whole, not the whole itself, and this severe discipline develops a habit of thinking constantly about the relation of the part to the whole - or at least to the other parts. What at first feels like regimentation comes to feel instead like an invaluable aid to clarity and to perceiving the implications of what you write.

Is this the same thing as a full analytical index? Of course there is some affinity. Yet it would be a rare index that could go to such depth, and especially across such a range of topics. Orlando's sensitivity to context and its ability to allow users to modify or contextualise searches cannot be produced in even the best index. Even if the index includes organisational layering, a term is either in it or not. And there is this fundamental difference: that an index is necessarily composed after the text is complete, as an add-on, whereas Orlando's encoding is integral to the production of text. Encoding provides, as we have said, a fusing of content and tool, of message and medium.

This lecture addresses the electronic Orlando text, the product of our many-partnered collaborative process, the outcome of the ongoing negotiation between the work in language, literature, and history on the one hand, and electronic possibilities, limitations, and developments on the other. It does not deal at length with the books which are another essential part of the Orlando history. Three individually authored historical accounts will be published as our history of women's writing in the British Isles. Their narrative accounts will be linked in critical ways with the text base, offering to the reader the possibility of deepening or expanding the narrative by choice. The volumes, then, will serve as both as overview to the electronic text and as key to it. They will be available in two forms: the traditional, portable, printed volumes which you can read in an easy chair or on top of a bus; and in electronic form where readers can exploit the encoding function to digress from the overview into the granular detail from which it draws.

In his lecture on "The dialectics of bibliography now," D.A.McKenzie spoke of two kinds of text. "One," he said, "is authorially sanctioned, contained, and historically definable. The other is the text as always incomplete, and therefore open, unstable, subject to a perpetual re-making by its readers, performers, or audience." We have organized the unstable, wide-open text of Orlando to ensure that readers who take the time to learn to use it may generate research results not known to or even suspected by the team who made the text. New readers will bring it new meanings. For that to happen, we have one major task before us. We must complete and perfect a delivery system both technologically and intellectually capable of communicating this potential to the Project's final collaborators. Their "creative regeneration" of the encoded material in the Orlando Project will expand our present collaboration to make Orlando the truly sociable text.