Orlando: Women’s Writing in the British Isles from the Beginnings
to the Present is a literary-historical textbase comprising more
than 1,200 core entries on the lives and writing careers of
British women writers, male writers, and international women
writers; 13,000+ free-standing chronology entries providing
context; 12,000+ bibliographical listings; and more than 2
million tags embedded in 6.5 million words of born-digital
text. The XML tagset allows users to interrogate everything
from writers’ relationships with publishers to involvement in
political activities or their narrative techniques.
The current interface allows users to access entries by name
or via various criteria associated with authors; to create
chronologies by searching on tags and/or contents of dated
materials; and to search the textbase for tags, attributes, and
text, or a combination. The XML serves primarily to structure
the materials; to allow users to draw on particular tags to
bring together results sets (of one or more paragraphs
incorporating the actual ‘hit’) according to particular interests;
and to provide automatic hyperlinking of names, places,
organizations, and titles.
Recognizing both that many in our target user community of
literary students and scholars dislike tag searches, and that
our current interface has not fully plumbed the potential of
Orlando’s experimentation in structured text, we are exploring
what other kinds of enquiry and interfaces the textbase can
support. We report here on some investigations into new ways
of probing and representing the links created by the markup.
The current interface leverages the markup to provide
contexts for hyperlinks. Each author entry includes a “Links”
screen that provides hyperlinks to mentions of that author
elsewhere in the textbase. These links are sorted into groups
based on the semantic tags of which they are children, so that
users can choose, for instance, from the more than 300 links
on the George Eliot Links screen, between a link to Eliot in
the Elizabeth Stuart Phelps entry that occurs in the context of
Family or Intimate Relationships, and a link to Eliot in Simone
de Beauvoir’s entry that occurs in the discussion of features
of de Beauvoir’s writing. Contextualizing Links screens are
provided not only for writers who have entries, but also for
any person who is mentioned more than once in the textbase,
and also for titles of texts, organizations, and places. It thus
provides a unique means for users to pursue links in a less
directed and more informed way than that provided by many
Building on this work, we have been investigating how Orlando
might support queries into relationships and networking,
and present not just a single relationship but the results of
investigating an entire fi eld of interwoven relationships of
the kind that strongly interest literary historians. Rather than
beginning from a known set of networks or interconnections,
how might we exploit our markup to analyze interconnections,
reveal new links, or determine the points of densest
interrelationship? Interface design in particular, if we start to
think about visualizing relationships rather than delivering
them entirely in textual form, poses considerable challenges.
We started with the question of the degrees of separation
between different people mentioned in disparate contexts
within the textbase. Our hyperlinking tags allow us to
conceptualize links between people not only in terms of direct
contact, that is person-to-person linkages, but also in terms of
linkages through other people, places, organizations, or texts
that they have in common. Drawing on graph theory, we use
the hyperlinking tags as key indices of linkages. Two hyperlinks
coinciding within a single document—an entry or a chronology
event--were treated as vertices that form an edge, and an
algorithm was used to fi nd the shortest path between source
and destination. So, for instance, you can get from Chaucer to
Virginia Woolf in a single step in twelve different ways: eleven
other writer entries (including Woolf’s own) bring their names
together, as does the following event:
1 November 1907: The British Museum’s reading room
reopened after being cleaned and redecorated; the dome
was embellished with the names of canonical male writers,
beginning with Chaucer and ending with Browning.
Virginia Woolf’s A Room of One’s Own describes the
experience of standing under the dome “as if one were a
thought in the huge bald forehead which is so splendidly
encircled by a band of famous names.” Julia Hege in Jacob’s
Room complains that they did not leave room for an Eliot
or a Brontë.
It takes more steps to get from some writers to others: fi ve,
for instance, to get from Frances Harper to Ella Baker. But this
is very much the exception rather than the rule. Calculated according to the method described here, we have
a vast number of links: the textbase contains 74,204 vertices
with an average of 102 edges each (some, such as London at
101,936, have considerably more than others), meaning that
there are 7.5 million links in a corpus of 6.5 million words.
Working just with authors who have entries, we calculated
the number of steps between them all, excluding some of the
commonest links: the Bible, Shakespeare, England, and London.
Nevertheless, the vast majority of authors (on average 88%)
were linked by a single step (such as the example of Chaucer
and Woolf, in which the link occurs within the same source
document) or two steps (in which there is one intermediate
document between the source and destination names).
Moreover, there is a striking similarity in the distribution of the
number of steps required to get from one person to another,
regardless of whether one moves via personal names, places,
organizations, or titles. 10.6% of entries are directly linked, that
is the two authors are mentioned in the same source entry or
event. Depending on the hyperlinking tag used, one can get to
the destination author with just one intermediate step, or two
degrees of separation, in 72.2% to 79.6% of cases. Instances of
greater numbers of steps decline sharply, so that there are 5
degrees of separation in only 0.6% of name linkage pages, and
none at all for places. Six degrees of separation does not exist
in Orlando between authors with entries, although there are a
few “islands”, in the order of from 1.6% to 3.2%, depending on
the link involved, of authors who do not link to others.
These results raise a number of questions. As Albert-Lászlo
Barabási reported of social networks generally, one isn’t dealing
with degrees of separation so much as degrees of proximity.
However, in this case, dealing not with actual social relations
but the partial representation in Orlando of a network of social
relations from the past, what do particular patterns such as
these mean? What do the outliers—people such as Ella Baker
or Frances Harper who are not densely interlinked with
others—and islands signify? They are at least partly related
to the brevity of some entries, which can result either from
paucity of information, or decisions about depth of treatment,
or both. But might they sometimes also represent distance
from literary and social establishments? Furthermore, some
linkages are more meaningful, in a literary historical sense,
than others. For instance, the Oxford Dictionary of National
Biography is a common link because it is frequently cited by
title, not because it indicates a historical link between people.
Such incidental links can’t be weeded out automatically. So we
are investigating the possibility of using the relative number
of single- or double-step links between two authors to
determine how linked they ‘really’ are. For instance, Elizabeth
Gaskell is connected to William Makepeace Thackeray, Charles
Dickens, and George Eliot by 25, 35, and 53 single-step links,
respectively, but to Margaret Atwood, Gertrude Stein, and
Toni Morrison by 2, 1, and 1. Such contrasts suggest the likely
utility of such an approach to distinguishing meaningful from
The biggest question invited by these inquiries into linkages
is: how might new modes of inquiry into, or representation
of, literary history, emerge from such investigations? One
way to address this question is through interfaces. We have
developed a prototype web application for querying degrees of
separation in Orlando, for which we are developing an interface.
Relationships or associations are conventionally represented
by a network diagram, where the entities are shown as nodes
and the relationships as lines connecting the nodes. Depending
on the content, these kinds of fi gures are also referred to as
directed graphs, link-node diagrams, entity-relationship (ER)
diagrams, and topic maps. Such diagrams scale poorly, since
the proliferation of items results in a tangle of intersecting
lines. Many layout algorithms position the nodes to reduce the
number of crisscrossing lines, resulting in images misleading to
people who assume that location is meaningful.
In the case of Orlando, two additional complexities must be
addressed. First, many inter-linkages are dense: there are often
50 distinct routes between two people. A conventional ER
diagram of this data would be too complex to be useful as
an interactive tool, unless we can allow the user to simplify
the diagram. Second, the Orlando data differs from the kind
of data that would support “distant reading” (Moretti 1), so
our readers will need to access the text that the diagram
references. How, then, connect the diagram to a reading view?
We will present our preliminary responses to these challenges
in an interface for degree of separation queries and results. We
are also experimenting with the Mandala browser (Cheypesh
et al. 2006) for XML structures as a means of exploring
embedded relationships. The current Mandala prototype
cannot accommodate the amount of data and number of tags
in Orlando, so we will present the results of experimenting
with a subset of the hyperlinking tags as another means of
visualizing the dense network of associations in Orlando’s
representations of literary history.
Barabási, Albert-Lászlo. Linked: The New Science of Networks.
Cambridge, MA: Perseus Publishing, 2002.
Brown, Susan, Patricia Clements, and Isobel Grundy, ed.
Orlando: Women’s Writing in the British Isles from the Beginnings
to the Present. Cambridge: Cambridge Online, 2006.
Cheypesh, Oksana, Constanza Pacher, Sandra Gabriele,
Stéfan Sinclair, Drew Paulin and Stan Ruecker. “Centering the
mind and calming the heart: mandalas as interfaces.” Paper
presented at the Society for Digital Humanities (SDH/SEMI)
conference. York University, Toronto. May 29-31, 2006.
Mandala Rich Prospect Browser. Dir. Stéfan Sinclair and Stan
Ruecker. http://mandala.humviz.org Accessed 22 November
Moretti, Franco. Graphs, Maps, Trees: Abstract Models for a
Literary Theory. London: Verso, 2005.
If this content appears in violation of your intellectual property rights, or you see errors or omissions, please reach out to Scott B. Weingart to discuss removing or amending the materials.