Book, Body and Text: The Women Writers Project and Problems of Text Encoding

Introduction
SGML and text encoding arose from the need to
make paper disappear. Documents whose size,
complexity, or frequency of revision rendered
them unmanageable as physical texts were the first
and strongest impetus behind the development of
SGML, and perhaps provided some of the crucial
emotional component as well, so to speak. The
conception of a text as an “ordered hierarchy of
content objects,” which completely separates the
text from its physical vehicle, does so almost with
a sense of relief: the repudiation of procedural
markup and the cumbersome mechanics of pages
seems like an entry into a purer, simpler world
where the real structures of the text become tractable.
With the advent of the Text Encoding Initiative,
SGML began to look additionally like a way to
give paper a new life. Humanities text encoding
projects often focus on rare or inaccessible documents, which are encoded in order to give them
wider accessibility and greater flexibility of use.
Projects of this nature include the Canterbury Tales Project and Project Electra at Oxford; the SEENET edition of Piers Plowman; the Women Writers Project at Brown University; the Rossetti
Archive; and numerous other similar undertakings, all of which aim at reproducing existing
physical texts in electronic form. Approaches to
this reproduction vary; some provide digital images of the original text along with the encoded
version; others attempt to include aspects of the
physical structure and appearance of the text in the
encoding; still others encode only the aspects of
the physical book which are seen to have linguistic
meaning. All of them, however, are aimed primarily at audiences whose scholarly work requires
reference to the physical text; the electronic text is
a facilitator, but ultimately it only stands in front
of, in place of, the real object which is seen as
grounding its existence.
As Jerome McGann and others have pointed out,
the aims of projects like these are often difficult to
accommodate using the structures provided by
SGML, since the physical and textual structures
of the book often overlap and cannot both be
represented. A number of methods have been developed which offer workarounds and practical
solutions, but the essential conception of the
SGML document as having a single hierarchy
remains unchanged. Text encoding projects, then,
must choose a method which best suits their particular commitments and needs.
This paper places the Women Writers Project’s
own approach within the context of these alternatives, and also within a larger theoretical context;
editorial theory is rooted in notions of the status of
the physical book and its relation to the linguistic
text, and this theory is reflected in the assumptions
made by text encoding projects about the relation
between the electronic transcription and the original.
1. Editorial Theory and the Physical Text
The physical document plays a crucial practical
and metaphysical role in editorial theory. Literary
editing proposes to itself the task of consulting
physical exemplars of a text in order to determine,
by their variations, how much of the authorial
message they carry and what authority they have
within the textual tradition. In an almost ironic
way, the real goal of this activity is to transcend
the physical document and get at the textual message that it carries. The “best text” which the
resulting new edition attempts to present, and the
“copytext” which in some theories is the projected/hypothesized authorial ur-text from which all
others vary, are both in a sense ideal and disembodied texts which none of the existing documents
fully produces. The physical document, then, is
metaphysically a failure, a corrupted falling-away
from the ideal of the authorial text. For the editor
it is like a dirty window through which one struggles to see the real original which never was.
Practically, it becomes a fetish, since it replaces
the unattainable object of one’s desire; an edition
becomes authoritative by attention to every physical exemplar in their minutest details.
Emotionally, our culture has a deep attachment to
physical books, particularly to old ones, which are
made to represent an earlier age which respected
learning and textuality rather than shopping malls
and video. Accompanying the new wave of fascination with electronic texts is a vein of deep distrust at their perceived replacement of the physical book. Among the criticisms of the electronic
text is that is has nothing grounding it; that it is
changeable and hence unreliable; and that it is
potentially anarchic in allowing multiple readings
and user modification. It is important to bear in
mind that these reservations do not regard the
electronic text’s disembodiment as moving it closer to the ideal, but rather think of it as a failed
version of the physical, one which has all the
potential for corruption that books suffer from, but
none of their ability to serve as a solid point of
grounding. Thus editors who address the task of
producing an electronic scholarly edition or an
electronic version of a primary source document
are faced with a crisis of confidence; they must
endow the electronic text with an authority which
it has not as yet been accorded by the general
imagination. Furthermore, in preparing an electronic version of a primary source, they are under the
special burden of making this text function – practically and metaphysically – like its physical source. At the same time, they are under pressure to
build in the kinds of functionality that come from
foregrounding the linguistic text – enabling searches, navigation, and retrieval based on the conceptual structures of the text – since this is part of
the point of encoding the document in the first
place. These two imperatives, unfortunately, do
not always pull in harness.
2. Encoding the Book, Encoding the Text
The WWP’s own commitment within this landscape is somewhat troubled. To begin with, our
primary goal is to create the electronic equivalent
of an archive – a repository of primary source
documents which are available to researchers for
a wide range of uses. It was the very inaccessibility
of archival materials that fueled the WWP’s beginnings; early modern women’s texts exist in
abundance but are too often only available in rare
book libraries, and our intention has been to create
another, more accessible and functional archive
which would broaden access and research. This
commitment points us naturally towards fidelity
to the physical book; our audience needs both
completeness and accuracy of description of features like pagination, line breaks, typographical
errors, front and back matter, damage and illegibility, and so on. At the same time, part of the
seduction of encoding quantities of textual material is that one has the opportunity to add function
to it, enabling users to fulfill their wildest research
dreams of searches and comparisons and easy
navigation – indeed, one could not get grant funding without conceiving of the project in these
terms. At once, then, we are thrown back towards
the other end of the spectrum, committed to encoding the conceptual structure of the text and addressing the nuances of its analytical hierarchies.
Beside our desire to create an archive – texts which
meet our expectations of their physicality – is the
equally strong desire to create texts which respond
to our expectations of their textuality.
Our accommodation of these two motives ends by
privileging the textual structure, and encoding the
physical structure using milestone tags. This is
partly because the TEI’s provision for tagging this
structure is so much more richly conceived and
nuanced, and partly because – in the absence of a
workable method of encoding both simultaneously – the textual structure facilitates more basic
activities. Users rely on notions of act, scene and
verse, of chapter and paragraph, for their basic
navigation in the document, while their use of
physical details amounts to something more like
consulting: an event rather than an ongoing process. Furthermore – and this is indicative of something quite basic about our conceptions of documents – the user is more apt to use the textual
structure to compare two versions of the same
work. One thinks of Chapter 2 of Clarissa as being
the same item from edition to edition, whereas
page 37 may be altogether different. The “work,”
we think, transcends the book precisely because it
is abstractable from it. This sense, however, may
say more about our emotional and intellectual
commitment to this notion of difference than
about the nature of things; why should this transcendence appeal to us so overwhelmingly? The
answer, I would argue, has to do with the way that
books and authors have been written into our
culture; about this I will say more in the finished
paper, concluding that cultural institutions rely on
the authority vested in notions of authorship and
transhistorical textuality.
By and large, the WWP has found that milestone
tags fill our basic needs for encoding the physical
structure of the book. However, in some cases they
introduce considerable complexity, and require
considerable ingenuity in execution. One case of
particular interest is the problem of continued
footnotes, which require, in some sense, two separate milestones indicating the end of the page. I
will outline this in more detail in the finished
paper, showing how it is necessary to think of the
footnote as orthogonal to both the textual and
physical structures in order to avoid difficulty.
Furthermore, other projects who need to record
even more information about the physical text –
projects dealing with manuscripts, for instance –
might find that using milestones is unworkably
cumbersome. Clearly, additional research on these
kinds of encoding needs will help advance our
thinking about overlapping structures – particularly those of the physical and the textual book – and
how to handle them.
Conclusions
Although it is possible to address the issue of
overlapping structures in SGML as a purely practical problem, I believe that this is a short-sighted
solution. I have tried to indicate the ways in which
85
overlapping structures can indicate important methodological issues, and also the degree to which
these methodological issues are implicated in the
larger cultural and political arena. Text encoding
projects ignore such matters at their peril; in creating data, we both create and rely on implicit
intellectual structures about which we can choose
to be naive or canny.
Bibliography
DeRose, Steven, J., David Durand, Elli Mylonas
and Allen H. Renear. “What is Text, Really?,”
Journal of Computing in Higher Education.
1:2 (1990).
Jed, Stephanie. Chaste Thinking: the Rape of Lucretia and the Birth of Humanism. Bloomington: Indiana University Press, 1989.
McGann, Jerome. “What is Critical Editing?,” The
Textual Condition. Princeton: Princeton University Press, 1991.
O’Keefe, Katherine O’Brien. “Text and Work:
Some Historical Questions on the Editing of
Old English Verse.” New Historical Literary
Study: Essays on Reproducing Texts, Representing History Jeffrey N. Cox and Larry Reynolds (eds.). Princeton: Princeton University
Press, 1993.
Shillingsburg, Peter. Scholarly Editing in the
Computer Age : Theory and Practice. Athens:
University of Georgia Press, 1984.
Tanselle, G. Thomas. Textual Criticism and Scholarly Editing. Charlottesville: University Press
of Virginia, 1990.