Content uploaded by David Kolb
Author content
All content in this area was uploaded by David Kolb on Jun 25, 2019
Content may be subject to copyright.
Scholarly Hypertext:
Self-Represented Complexity
David Kolb
Department of Philosophy
Bates College
Lewiston ME 04240 USA
Tel: 207 786 6308
E-mail: dkolb@bates.edu
ABSTRACT
Scholarly hypertexts involve argument and explicit self-
questioning, and can be distinguished from both informa-
tional and literary hypertexts. After making these distinc-
tions the essay presents general principles about attention,
some suggestions for self-representational multi-level
structures that would enhance scholarly inquiry, and a
wish list of software capabilities to support such structures.
The essay concludes with a discussion of possible conflicts
between scholarly inquiry and hypertext.
KEYWORDS: Hypertext rhetoric, argument, scholarship,
typed nodes, typed links, self-representation
1 SCHOLARLY HYPERTEXTS
If we classified all the hypertexts in use, most would be
networks of information that are either in the process of
being organized or are being consulted. Tremendous effort
has gone into discovering manual and automatic linking
strategies to organize information and make it efficiently
manipulable.
On the other hand, there are literary hypertexts whose
goals differ from the organization and delivery of
information. These texts experiment with new modes of
writing; they break traditional linearities of narrative and
form; they offer adventuresome modes of encounter and
ongoing textual re-definition. Theorizing about and
production of literary hypertexts offers the most
compelling evidence for Landow's suggestion that
hypertext bears important relations to views about writing
and culture found in recent critical theory [20].
There has been less discussion of what I'll call 'scholarly'
or 'inquiry' hypertexts. If talked about at all, these tend to
be equated with informational hypertexts, but they are
significantly different. The neglect of scholarly inquiry
hypertexts may be because literary writers equate them
with linear critical essays, while scientific writers presume
that scholarly writing will be handled as the Web was
originally designed to do: individual lexia will be scientific
papers or their equivalents, with links offering cross
reference to evidence and to related
papers. Also, many interested in electronic writing view
hypertext as a possible escape from moribund scholarly
writing seen as dedicated to the preservation of hierarchy
and rigid trivialization. "Expert language is a prison for
knowledge and understanding. A prison for intellectually
significant relationships. It is time to move beyond the
institutional practices of triviledge" [41].
I want to argue that there should be a mode of hypertext
writing that is neither avant-garde literature nor organized
information, but more akin to scholarly inquiry in the hu-
manities and philosophy.1 This has unique formal features
of its own that may highlight new possibilities for
hypertext in general.2 Some of these features could be
better implemented in hypertext. Doing so might both
inspire new hypertext forms and enliven scholarly inquiry.
1 We should question the universal applicability of
Schneiderman's Golden Rules of hypertext ("(1) A large
body of information is organized into numerous fragments,
(2) The fragments relate to each other, (3) The user needs
only a small fraction at any time" [32]). These presuppose
that the hypertext is to make available units of information
that can be put into one node each. Neither literary nor
scholarly hypertexts fit this model.
2 My threefold division of hypertexts is somewhat
tendentious. Stand-alone hypertexts can be classified ac-
cording to their primary use. But as hypertexts get larger
and the environment allows more interconnection, links
are likely to be established among large hypertexts so that
one may be part of another for a given reading. My
distinctions could also be attacked in an imperialistic way,
claiming that all text is basically literary [34]. But this
turns on an argument "in principle" that does not contest
functional differences between different types of text in
daily use.
However, 'scholarly' is not always a positive appellation.
Scholarly writing is often stigmatized as dead, linear, lack-
ing imagination, protective of boundaries, using rigid
methodology to destroy creativity. Nor is there any lack of
examples that fit the accusations. But over time inquiry
does change. A community of inquirers forms around
agreement on methods, a set of questions, and relevant
options and methods for answering them. This is akin to
what Kuhn called "normal science" [18]. Such
normalization accounts for the deadening effect of much
scholarly writing, but also for its progress. For, as in
science, these normalizations eventually produce
revolutions. People come to question their presuppositions,
principles, and criteria. A field like philosophy or literary
criticism may be more or less permanently in revolution,
but in all fields inquiry continues and the revolutions
come. In the humanistic studies the revolutions usually
cause a divergence from rather than a replacement of the
older discussions. So scholarly inquiry writing has never
been as linear as people claim, for its real existence is not
as the isolated article but as the contentious library. There
it is already inter-linked in complex and rival ways
through bibliographies, dueling evaluative summaries,
competing authorities, and so on.
So let us not conceive the scholar too narrowly. The
paradigmatic scholar, the classicist studying Greek and
Roman literature or history knew the books, could trace
the references, make the connections. What becomes of
such a scholar when automatic concordances and
automated searches start to replace the scholarly memory
and perhaps automated linking makes the connections?
Memory was never all of scholarship; there were also
those who reflected on what was being said, who changed
the templates, asked new questions, saw the texts in new
ways, found new relations and made connections across
borders. Those skills are still needed and they are not best
described as the management of information, because they
debate the criteria for what is to count as information. How
do we get that debate into hypertext?
While I mean to raise issues concerning hypertext in gen-
eral, I am speaking as someone who creates neither know-
ledge bases nor literary texts, but rather argumentative
inquiry. In particular, I come from philosophy, which has
an uncertain relation to the scholarly essay and a long feud
with literature. Philosophy has always been given to self-
reflection and self-criticism of its own presuppositions,
and this influences how I describe scholarly activity. In
philosophy it is very evident that 'inquiry' exists as a
multiplicity of contending voices mutually including one
another and constantly turning back on themselves in an
always incomplete self-reference.
The obvious features of scholarly inquiry are questions, as-
sertions, argumentation, evidence, and a community of in-
quiry to which the writing is submitted for judgment. This
model is familiar to us from science, and the differences
between humanistic discussion and science are themselves
matters of dispute. At the least, there is more continual
self-reflection and more global self-criticism, more
divergence and less straightforward addition or subtraction
from a knowledge base. And less agreement. Important
formal features are on-going self-representation and debate
about criteria for argument and organization, without any
clear or stable hierarchy within which to locate such
debate.
Assertion and argument make inquiry different from litera-
ture. Stuart Moulthrop says that "the act of reading in
hypertext is constituted as struggle: a chapter of chances, a
chain of detours, a series of revealing failures in com-
mitment out of which come the pleasures of the text" [29].
Reading literary hypertexts creates contours of meaning
that emerges from the field of the text. This is not enough
for assertive inquiry. The pleasure of the text is not the
only goal. Literary productions do not relate to one another
as assertions do in inquiry. The question whether or not
there is a basis for a universalistic ethics is asked for more
than the pleasure of the text. As are: Might Quine be right
about the analytic/synthetic distinction? Did Van Gogh
paint this canvas? And there are other questions where
most of the discussion might be about what criteria could
ever certify a possible answer: What is the value of art? or
What can we learn from the fall of Rome? Writing about
such questions involves claims and judgment, as well as
ongoing debate about the form and criteria for the
discussion as we engage in it. These are different from
both the management of information and the pleasure of
the text. The landscape this writing creates is complexly
self-connected and we seek not only to explore it but to
create new dimensions, new stakes.
I want to ask how in hypertext we might allow not just
connection but assertion, self-representation, and debate
about criteria. How do we perform these in hypertext, and
how make them available for discussion and judgment?
Hypertext theorists, if they think of scholarly inquiry,
often see hypertext's task as the organization of a new and
improved library as a web with live cross-references, links
back and forth between articles and their predecessors and
successors. But such a hyper-library could just assemble
current scholarly forms, which are best suited to a different
institutional framework and rhythm than what is now
developing. A hyper-library is not yet a hyper-text.1
1 Producing the hypertext library is surely
important, and we understand the difficulties of indexing,
of automatic link creation, of classification, issues
librarians have studied for years, not to mention the
digitalization of unthinkable amounts of past data--which
is important since humanistic inquiry does not outpace its
The hyper-library is not what I want to deal with in this es-
say. I want rather to ask what else hypertext might do for
communal inquiry. Could it affect not just the linkages
among works in the library but the works themselves?
We are here discussing such issues in linear conference es-
says because it is efficient to do so. But our discussions are
beginning to live in a new context that will alter their ecol-
ogy, and there will be new genres. Can hypertext be one of
them? Is there a way to do inquiry in a "native hyper-
text"[28]?
Or is native hypertext to be restricted to either the manage-
ment of information or associative modes of writing?
There should be other alternatives. We should not leave
argument, assertion, critique, self-reflection and self-
criticism behind as we move into the new media. We need
to find out how to do such actions in new ways.1
What interests me most is whether we might create new
modes of reflection, new ways to organize inquiry, and
new intellectual objects. I wonder about new forms of
writing which would be like literary hypertext in that some
discursive effects would occur in the landscape and its
traversal rather than inside individual lexias or across one-
step links.2 Might new intellectual moves be possible in
these freer spaces? Perhaps even a new structure for
inquiry that was not just the point-counterpoint of
traditional debate.3
past as does scientific literature, and thus needs references
to old texts.
1 For discussion and tentative examples of such
moves see [16]. About forty percent of the long hypertext
in [16] was published as an essay in [22]. Some of the
background assumptions can be found in [15, 17].
2 My interest in such possibilities was sparked by
my philosophical work with Hegel and with other thinkers,
such as Derrida, for whom philosophy should lead not to
the presentation of isolated results but to an understanding
embodied in a movement that cannot be confined to one
static proposition or argument.
3 This issue is difficult to conceptualize. Must the
overall form of inquiry be the confrontational point-
counterpoint of standard scholarly dialogue? There are
forms hypertext could assume that would not be essays or
monographs but would still assert and judge in dialogue.
Could there also be forms in hypertext that give the overall
inquiry a different structure than point-counterpoint?
Could there be other movements that are equally as
thoughtful and inquiring as confrontation and
commentary? The Socratic and Peircean answer affirming
the ultimacy of question-and-answer argument is generally
accepted but it has been questioned not only by anti-
foundationalists since Nietzsche [33], but also by
Hypertext can do better some features of scholarly
inquiry.4 Hypertext offers space for ever new dimensions
of writing. It could offer new ways for self-representation
and self-criticism. And it offers these together with a
structural refusal of authoritative meta-positions, through
their endless proliferation and mutual inclusion. This one
feature alone already fights against the rigidification of
inquiry.5
In the next two sections I discuss two prerequisites for de-
veloping such new intellectual objects and discursive
moves: regions and multiple self-representations.
2 ATTENTION AND STRUCTURAL EVENTS
Hypertext theory often talks as if the unlimited dimensions
of hypertextual composition could offer endless reading.
But our time here is limited. And attention is a scarce
resource. Not only does our attention flag, but action and
policy cannot be postponed indefinitely. We can create
hypertextual structures sprawling as far as we wish, but in
reading them we have scarce time and limited attention.
James Joyce may have wanted perpetual insomniacs to
devote their life to reading Finnegans Wake but most of us
will find less dedicated readers. We have to think about
structural emphasis and movement in hypertext, but print
conventions are inadequate to the possibilities of the
medium. How do we compose hypertexts that can do
inquiry that has no simple limits, that turns on itself, that
always has other questions, and yet remains readable?
Every node cannot demand equal attention to itself, or at-
tention will fail. If everything has equal emphasis, nothing
has much emphasis.
A hypertext must be more than a sequence of random
associative links. If anything at all can follow this node,
there will be no play with expectation, so attention will get
more difficult to sustain except by ever more extravagant
moves.
foundationalists in the dialectical and phenomenological
traditions.
4 As Bernstein comments with regard to the
opening chapters of Thucydides,"Linearity was never an
option for historical writing; hypertextuality can make
complex structure concrete, clear and responsive to both
the author and reader" [2].
5 "As Kaplan and I have observed in working with
students, electronic writing complicates the work of
literary criticism. A critical project set up within a
hypertextual network becomes an intimate and integral
part of the work it tries to anatomize. In its root sense,
"criticism" implies a separation of one discourse from
another; but in hypertext this primary agenda runs into
difficulties" [29, 30]. Versions of this failure to establish a
secure meta-discourse are common in philosophical
inquiry, though not through direct hypertextual inclusion.
If emphasis cannot fall equally on all nodes, then hypertext
rhetoric must be more than a sequence of one-step links.
Emphasis should occur on different scales. Besides
individual nodes, patterns may extend over many links,
and may themselves be the objects of emphasis and self-
description.
Investigations in music indicate that sequences of
randomly chosen notes do not hold the listener's attention.
Sequences ("brown music") where the choice of each note
is somewhat constrained by neighboring notes sound
better, but still lack the attraction of more composed music
where there are complex relations among more distant
parts of the piece.1
Hypertext suffers this problem when the horizon of
reading is too close. Musical compositions go beyond this
kind of linkage to create rich and complex temporal and
formal and thematic connections to other sections of the
piece in many different levels and sizes. Hypertext could
do the same with sequences and nodes that are not just
influenced by their immediate neighbors. There should be
large structures, echoes, returning themes, transformations
and recapitulations and variations, but without a fixed
linear framework of reading. Literary hypertexts strive for
this; scholarly hypertexts must learn how these kind of
structures might be appropriate for their inquiries.
Hypertext thus offers possibilities for exciting complexity.
Compare this with the movement of entertainment multi-
media that emphasize serial intensities and must become
more and more 'loud' to keep attention engaged. Such
structures are inadequate for inquiry. Replaying a video
game brings increased familiarity and skill, but little new
insight. Rereading a hypertext should do more for thought
and feeling. We should not substitute association for all
kinds of questioning and discursive moves.
3 REGIONS AND SELF-REPRESENTATIONS
To manage emphasis and attention in extended thought,
then, hypertext needs what music has: different kinds of
unities on many levels that interact with each other in
complex ways. The single node should not stand alone, nor
should a single level of linking. There should be larger
structures and discursive moves as well as ways to become
aware of them and their relations and links. We are
familiar with discursive moves such as making an
assertion, giving backing, offering alternatives, contesting
a question, expanding a topic. Less familiar are moves
might be undermining a duality, raising questions about
criteria, ironic parody, showing internal tensions within a
set of concepts, and the like. Are there new moves possible
1 I remember these results as reported in Scientific
American but its indices contain only [11], which mentions
similar issues but is not the article I recall.
in hypertext that might take advantage of more expansive
and self-reflective linking?2
Linkage and inclusion are not new in inquiry; the library is
full of links and quotations. What's new is the density of
links, with miscegenational linkage and full-scale mutual
inclusion that crosses borders. What's new is the tempo of
writing and connection, and the ability to include any
number of self-representations and self-commentaries in
different dimensions, and yet, despite the distance one may
have traveled, to bring these to bear with a single link
back.
A complex discursive move could be made by a locality
composed of many nodes. There could be a region that is
an explorable landscape whose links and transitions are
meaningful in themselves but also contribute to the
regional effect. Such units of meaning could have complex
internal structures and external relationships. There could
also be partial localities. Or a region might itself be a node
in some larger move or gesture. As in nature, units of
meaning would occur on many scales.
Jim Rosenberg has argued for the importance of hypertext
"episodes" [37]. When I speak of a "region" I mean some-
thing different. Rosenberg's episodes take form in the
mind of the reader, and embrace nodes that need not be
related "in the text." I am concerned with structure that is
not necessarily gathered in one reader's mind but is there
in the link or relation patterns. Not the hypertext as read
but as readable. On the other hand, Rosenberg's idea of
"the episode as a virtual document" suggests regions as
units with sub-units, and I agree strongly with his claim
that "meaning is not just a function of the lexia, but
happens as we move through the links" and his plea for
"hypertext as a medium of thought . . . [not] as a medium
for organizing . . . linear thoughts which are not
themselves hypertexts."3
2 What I here call "discursive moves" might be
termed in philosophy "speech acts," though there are some
difficulties with the individuation of speech acts (see the
debate in [5]). Motion is not pure passage; this is one
reason why I have sometimes used the word "gesture" for
the discursive moves in hypertext.
3 The discursive moves and gestures I have in mind
mostly involve relations among already syntactically com-
plete propositions. I am concerned with expanding and
making explicit the types available for discursive moves.
The "simultaneities" that Rosenberg discusses and creates
in [35, 36, 38] attempt a related expansion in another way,
through an open juxtaposition that refuses to type the
relations among groups of words as it refuses syntax
within the groups.
Figure 1: Regions. These could show Aquanet layouts
instead of node-and-link structures.
Regions come closer, perhaps, to Michael Joyce's notion
of "contour" [13, 14, 2]. For contours are in the landscape
as well as assembled by my eye. The experience of reading
a hypertext with complexly related regions would involve
a sense of changing contour, of inquiry altering its
horizons as in the shift of ground that happens when new
language is deployed, or a question is discredited, a
presupposition challenged or a duality shown not to be
exclusive, or criteria changed. A region would not
necessarily offer a single contour but perhaps occasion for
a family of contours whose mutual transitions might be a
discursive move.
Emphasizing the importance of regions does not mean,
however, that every node must belong to some unique
locality. A single node might be ingredient in several
localities, and there could be nodes or sequences of nodes
that remain alone or compose lines of flight outside of any
larger unity.
There are no closed forms in hypertext, just as there can be
no truly isolated or finished works. (There are no truly iso-
lated works in print, either. Connectivity is a condition of
meaning even without explicit links.) A whole set of nodes
in one region might also appear in another but linked quite
differently. Even if it had a relatively closed form a region
would contains nodes that led outside or intersected differ-
ent regions, and there should be other ways to move than
by explicit links.
Figure 2: Intersecting Regions
Intersecting regions need not be hierarchically nested nor
need the text make a hierarchical whole. Since hypertext
always provides the possibility of more dimensions of
movement and connection, authoritative metapositions can
be undone by mutual inclusions, interactions among levels,
and multiple and conflicting totalizations.
An ability to perform complex linking across multiple lev-
els of description and abstraction would provide the possi-
bility for creating new intellectual objects and discursive
moves.
Besides these kinds of units in hypertext, we could have
many kinds of relations among the units. A complex
hypertext must be more than a sequence of self-enclosed
localities, as are many video games. Hypertext can take
advantage of the possibilities for links that move from a
member to its region as a whole, or from a region to
another region, or from a region to abstract representations
of the forms and relations of itself or other regions.
Such self-representations provide more than a rhetoric of
departure and arrival [19]. If hypertexts are to serve
inquiry, they need to question and comment on their own
and others' form and relations, so they must be able to
represent their own form and movement.1
1 Having available more means to construct
different self-representations and so to question one's
criteria and schemes of organization would help avoid the
situation Joyce fears where hypertext becomes an endless
commentary on authoritatively pre-given structures and
categories [14].
How inquiry even in well-defined tasks can turn back upon
its own schemes and organization can be seen in the
adventures of Aquanet from the optimism of its
announcement [23] through the unexpected observations
about informal and implicit patterning [24] that lead to
creation of VIKI [25, 26, 40]. In the flow of scholarly
literature, writers must be necessarily reluctant to commit
themselves to fixed types and patterns in advance. Thus
we find a multiplication of overviews and tentative
patternings. Some means of amplifying and representing
this multiplicity and debate about form and criteria would
be necessary for scholarly hypertext. My suggestion is to
allow the construction of and discussion about self-
representations of many types on many levels, none of
them final or authoritative.
4 A WISH LIST
Constructing such hypertexts demands strong representa-
tional and linking capabilities. A complex hypertext needs
some way to modulate and focus emphasis on different
levels of organization. For inquiry, everything in the text
needs to be open to question and discussion: content, form,
progression, criteria, and so on. In native hypertext inquiry
the software should enable discursive moves that happen
within a landscape or region rather than in a single node or
link, and have ways to refer to regions that are related
without hierarchy and in many dimensions.
In using the node-and-link conception of hypertext in
framing these suggestions I do not mean to ignore the
criticisms of "the tyranny of the link" [10]. Indeed I would
add a criticism of my own, that the model encourages the
notion that a hypertext must consist of linked facts or bits
of information.
However, explicit author-made links are important in
scholarly inquiry. For one thing, given the state of AI
today it is unlikely that structures and connections
computed on the fly could handle the kinds of relations
needed in inquiries where what is at stake may be the very
criteria of relevance and connection presumed by the
search. For instance, automatic link creation would have
limited success in arguments over the worth of a new
metaphor, or over alternate translations, or over whether
thinker X really influenced thinker Y who uses a totally
different vocabulary. Similar difficulties would arise with
mechanisms for automatically calculating link inheritance
and simplification [4, 1, 39]. Even author guided content
search would be of limited value in such cases.
Scholarly hypertexts would need multiply typed nodes that
can be collected or referenced as having various kinds of
content, belonging to various regions, or standing in
various relations.
Regions could perhaps be implemented as saved
composites in the Dexter-based framework outlined by
Grønbaek [8]. These would be structured collections of
unrestricted types of components. Since the same
component could belong to more than one region, a
composite would probably reference rather than include its
components. However a two-way relation is needed,
despite the computational overhead, so that components
could be aware of their region(s) and discuss or dispute
their membership.
Scholarly hypertexts also need links that are first class ob-
jects which can themselves be linked to and discussed.
Given the need to refer to and to gather links, there should
be an open-ended capability for multiply typing links (for
instance, something like the keyword system that allows
users to give multiple types to Storyspace nodes). Non-
dyadic links, or Aquanet relations offer the possibility of
more complex moves, especially if objects of any kind or
level can enter into the relations.
Link types pose problems since pre-defined types would
not likely be used consistently, while user-defined types
would proliferate beyond easy usefulness. But such
proliferation would be preferable to a lack of types. If we
let types proliferate but make them objects that can be
referred to and discussed, then the discussion of link types
and rival collections of links into new types would be one
way to embody disagreement over criteria and appropriate
moves.
Scholarly hypertexts would need to create (and link to and
comment on) patterns of links or patterns of regions in
one's own text or in texts referenced or included,
emphasizing different sorts of connectivities on different
levels of abstraction. This would allow reference to larger-
scale features. The familiar planar or nested map of boxes
and arrows (as in Intermedia or Storyspace) could be
enriched with colors and labels and styles of nodes. It
could have capabilities for displaying only certain chosen
link patterns or types, or the relations of larger regions.
This would involve bringing items from what could be
widely distanced areas on the full view into various
abstracted views that create new localities. There could be
many types of overview and filtered lists, with the author
specifying parameters such as the type of objects or links
being mapped. In a complex multi-authored hypertext
there might be competing overviews, and discussions
about their adequacy; deciding the criteria for an abstract
mapping is not a neutral act. A planar map has difficulty
representing the many levels of abstraction and cross-
reference that might be involved, so perhaps the map
would be a three-dimensional space in which regions
occupied planes that were allowed to intersect or overlap.
Such views would be temporary updatable maps, but also
be able to be made into permanent discussible nodes in the
system, with added labels and comments. In that case auto-
matic updating would be user-specifiable.
There might be a combination of author-created and
system-generated views, as with VIKI's algorithms for
locating implicit structures. The problems here are similar
to those mentioned above for automatic content search, but
one could imagine that as users produced maps or
diagrams with spatial arrangement of tokens for regions or
discursive moves, the system compared rival maps for
similarities and differences, or noted common formal
structures in maps of different areas.
Such self-representations could, like regions, be imple-
mented as composites in the Dexter-based framework,
much as Grønbaek speaks of browsers and table tops as
objects. They would be structured collections referring to
their components rather than including them, and available
either globally or within the one object. These
representations could be either temporary or saved. They
would be either computed or created by hand, but they
would need more kinds of structure than the hierarchical
structuring discussed in [8], and more kinds of editing
operations, so that they could receive comments or visual
annotations that might propose connections, indicate
divergencies, and the like.
Special self-representations of structure and content could
be created to meet particular needs. These might be cross-
roads documents such as the overviews found in George
Landow's work [21]. Or the author-created representations
might involve special tools such as the structural views in
SEPIA [9, 32] or the Aquanet type editor and display [23].
Such systems provide facilities for representing a structure
of argument or data that is the object of discussion and de-
velopment, but they do not provide a easy medium for
recording the stages in such discussion, the proposed
alternative organizations, or the issues that are disputed in
various suggested formalisms or spatial organizations. For
instance, Aquanet offers different views of the same
knowledge structure but not multiple structures. However,
imagine a three-dimensional space in which different
Aquanet arrangements appeared as floating planes that
could be compared and linked to, and this map was itself
something that could be referred to and discussed.
Something like this could be arranged with the flat
Aquanet or VIKI plane if the objects were duplicated and
arrayed in rival arrangements on different regions of the
plane, and some selective overview provided to compare
the regions.
Although spatial arrangement allows structure to remain
implicit and to grow while being discussed, the syntax of
spatial arrangement, even complemented with visual
typing by color and shape and font, seems too limited to
represent the many kinds of discursive moves in inquiry
and argument. Aquanet relations with their recursive
capabilities seem to offer more possibilities in this regard.1
Scholarly hypertexts would need to be able to make links
in all directions among types and levels of nodes, refusing
any hierarchical separation of text from representations of
the text.
Since regions and their discursive moves are important it
might be a helpful option for a backtrack capability to
move by larger regions in reverse order, but within a
region to repeat the nodes in the order in which they were
visited or in some other way that would preserve the
moves made by linking and choices.
Figure 3: Self-representations and links.
5 POSSIBLE CONFLICTS
Whatever one may think about this wish list, there are fur-
ther questions about the possibility of scholarly hypertext.2
1 Given the reluctance users showed for the
machinery of relation-creation in Aquanet [24], it is a
legitimate question whether the elaborate typing and self-
representing facilities I am suggesting would actually be
used. But scholarly inquiry already trains people to value
overviews of options and self-critical analysis, so that if
the facilities were relatively easy to use and returned value
because they allowed the raising of new questions, there
would be an impetus to use them.
2 One could object that there is no point in
discussing hypertext form "in advance." Should we try to
legislate the form of future poetry? Or even of
scholarship? Of course we should let forms evolve as
needed (or as unneeded!), but this essay discusses
capabilities in the underlying technology that would bring
the possibility for new kinds of form
A node in A
discussing
the form of A
One node in B
discussing B as
a whole
A node outside A and B
referring to one link in A
and to B as a whole
Region A
Region B
One node in A
discussing B
as a whole
A node in C
comparing
the forms of
A and B
A node in B objecting to
the comparison of A
and B made in region C
Region C
A node in C
discussing
a map of
certain typed
links in A
We could start by asking whether modes of scholarly pre-
sentation really could be adapted to the presentation of
scholarly structures of argument, evidence, and inquiry.
Perhaps long prose passages of such inquiry could be
linked in the hypertext library, but the inquiry itself still
demand a more linear mode of presentation inside each
unit. "If strict control of information is paramount [or if
argument, assertion and self-questioning are paramount],
why trade print for hypertext in the first place?" [31].
It is true that when things must be presented in a single un-
varying fixed order network linking is not appropriate,
though hypertext can mimic an article or book by using
large nodes and linear linking. However, very little writing
needs to be presented in a single fixed order, considering
the varying ways in which scholarly moves have been
presented in prose and how readers skip around in linear
books and essays.1 There should be advantages to
presenting ideas and assertions in regions that are multiply
explorable landscapes located in complex relations to
others. Imagine even as small a text as this present essay
presented in several regions with multiple links on
different levels of detail instead of the paragraphs plus
digressive footnotes. Other links could lead to self-
reflective remarks such as this.
On the other hand, the issue could be reversed, asking
whether my emphasis on complexity and self-
representation creates self-totalizing texts that present and
control their own form, and so deflect the natural tendency
of hypertext to associative and borderless thinking. Taylor
and Saarinen say that we must move "from edifice to
improvization" [41].
We are not caught between linear and associative-
interstitial writing. Nor is there one essentialist way to use
hypertext; it is a technology, not a literary genre. However
what I have been urging does involve more unity than
theories of literary hypertext might recommend. The unity
stems from the focus on inquiry and the need for self-
representation. In Joyce's terms, scholarly hypertexts
would be more exploratory than constructive [13].
However the overall inquiry would be a constructive
endeavor that kept adding and changing regions and
dimensions of relations, and so altering the older contours.
As a result the whole complex would not be a totalized
unity, though it would contain competing self-represen-
tations.
1 Would exploration be compromised by the self-
representational capabilities I am suggesting? Not if they
are used in multiple ways. We should distinguish
psychological motivations to explore, because we don't
know what is coming next, from the structural gradients
within the text. The latter remain when we reread a novel
or a poem. Jane Yellowlees Douglas 'explored' Michael
Joyce's story 'afternoon' uncounted times [6, 12, 7].
For many regions or sets of regions within hypertexts of
the sort I am imagining there would likely be some
complex abstract armature "behind" the text, perhaps a set
of ideas and arguments, a conflict of points of view, or
rival interpretations of evidence. If that is so, why not just
present the abstract structure straight out? First, because it
is not evident that complex argument is always best
presented in step-by-step fashion, since it is frequently
presented otherwise even in linear books or essays.
Second, there may be abstract structures that are better
presented in hypertext than in linear prose. Third, there
may be discursive moves and gestures that are not the
exemplification of static structures and cannot be
performed in linear sequences of argument or proof.2
Fourth, to say that there is some abstract structure does not
mean that comprehending that structure is the only goal of
reading.
Finally, we might ask why we should care about such
complex scholarly hypertexts? Why might they be
important? The main reason is that they could enrich
inquiry and scholarship. But there is another reason. Many
current worries about the social and intellectual impact of
new technologies in the wired world depend on an image
of the net as delivering seductive serial intensities, as in a
series of video games or music videos. These are
conceived as destroying literacy and thoughtful reflection.
But the opposite could be the case if hypertext could help
create a new literacy by using in more complex and self-
referential ways those very features of the new
technologies that are most feared. Linkage and transitions,
moving attention, transgressing borders, and non-linear
reading could be complexly structured and encourage
increased attention and reflection.3 Could hypertext then
lead to a renewed literacy of inquiry and discussion?
Could it broaden participation in such pursuits?
That depends on whether such complex texts could ever be
written and read. An anonymous reviewer of an earlier
version of this essay put the question sharply: "We still
don't know what motivates changes in scholarly forms.
What would motivate a turn to hypertext? What is the
2 This touches some of the philosophical issues that
divide "analytic" and "continental" thought: what is the
relation of thinking to formal structures and is there a
thinking that is not the creation or analysis of structure?
3 "In answer to McLuhan's second question--what
does hypertext render obsolete?--the best answer is not
'literacy' but rather 'post-literacy'. As Nelson foresees, the
development of hypertext systems implies a revival of
typographic culture (albeit it in a dynamic, truly paperless
environment). That forecast may seem recklessly naive or
emptily prophetic, but it is quite likely valid. Hypertext
means the end of the death of literature" [27].
scholarly value-added (to be rather crass about it) to
writing and publishing in hypertext?"
What does hypertext have to offer, besides the efficiencies
of the hypertext library that need not contain native hyper-
texts? What about more complex structures? New
discursive moves that occur in non-linear sequences? New
forms of self-representation and ways to bring them to
bear on issues. More access to others work and easier
inclusion of others' texts. Attempts to assert one dominant
scheme of categorization both facilitated and resisted.
Elimination of the fiction of the final metaposition. A
realistic sense of not controlling the dialogue, combined
with new possibilities of surveying relevant regions. More
access to the context of discussion. Hypertext may be truer
to the real context and process of discourse and thought
than are tidy books.
The last time that major changes occurred in the media of
humanistic and historical inquiry was during the
nineteenth century when universities secured a near
monopoly on scholarly journals, which had previously
been published by private or royal associations. This
multiplied the number and influence of the journals. Their
hold on scholarly communication is now breaking down.
As self-publication becomes a genuine possibility, quality
control may take a new form as surveyors and pointers
rather than gatekeepers to the media. Then the way would
be open for more hypertextual forms. Perhaps the Web
could be the bridge: imagine web items written less like
papers to be linked and more in native and interpenetrating
hypertext, developing into regions of regions, new lands
for us to explore and build.
REFERENCES
1. Allan, J. Automatic Hypertext Link Typing.
Proceedings of Hypertext 1996. ACM, New York,
1996.
2. Bernstein, M., Joyce, M., and Levine, D. Contours of
Constructive Hypertexts. ECHT '92 Proceeding of the
ACM Conference on Hypertext. ACM, New York,
1992.
3. Bernstein, M. Hypertext and the Linearity of History.
Hypertext Now. January, 1996.
http://www.eastgate.com
/HypertextNow/archives/History.html
4. Cleary, C., and Bareis, R. Practical Methods for
Automating Linking in Structured Hypermedia
Systems. Proceedings of Hypertext 1996. ACM, New
York, 1996.
5. Derrida, J. Limited Inc. Evanston: Northwestern
University Press, 1988.
6. Douglas, J. Y. Understanding the Act of Reading: the
WOE Beginner's Guide to Dissection.Writing on the
Edge, Volume 2, Number 2 (1991).
7. Douglas, J. Y. "How Do I Stop This Thing?": Closure
and Indeterminacy in Interactive Narratives. In [22].
8. Grønbaek, K. Composites in a Dexter-Based
Hypermedia Framework. ECHT '94 Proceedings. New
York: ACM, 1994: p. 59-69.
9. Haake, J. M., and Wilson. B. Supporting collaboratiave
writing of hyperdocuments in SEPIA. Proceedings of
the ACM CSCW '92. ACM, New York, 1992.
10. Halasz, F. G. "Seven Issues" Revisited. Proceedings of
Hypertext '91. ACM, New York, 1991.
11. Hillier, L. A., Jr. Computer Music. Scientific
American, December 1959: p. 109-120.
12. Joyce, M. Afternoon, a story. Watertown: Eastgate
Systems, 1992.
13. Joyce, M. Of Two Minds: Hypertext Pedagogy and
Poetics. Ann Arbor: The University of Michigan Press,
1995.
14. Joyce, M. (1995a) A Feel for Prose: Interstitial Links
and the Contours of Hypertext. In [13].
15. Kolb, D. Postmodern Sophistications. Chicago:
University of Chicago Press, 1990.
16. Kolb, D. Socrates in the Labyrinth. Watertown:
Eastgate Systems, 1994.
17. Kolb, D. Communicating Across Links. In Ess, C., ed.
Philosophical Perspectives on Computer Mediated
Communication. Albany: SUNY Press, 1996.
18. Kuhn, T. The Structure of Scientific Revolutions.
Chicago: University of Chicago Press, 1970.
19. Landow, G. P. The Rhetoric of Hypermedia: Some
Rules for Authors. In Hypermedia and Literary
Studies, ed. Paul Delany and George Landow.
Cambridge, MA: MIT Press, 1991.
20. Landow, G. P. Hypertext: The Convergence of
Contemporary Critical Theory and Technology.
Baltimore: Johns Hopkins University Press, 1992.
21. Landow, G. P. The Dickens Web. Watertown: Eastgate
Systems, 1993.
22. Landow, G P., ed. Hyper/Text/Theory. Baltimore:
Johns Hopkins University Press, 1994.
23. Marshall, C. C., Halasz, F. G., Rogers, R. A., and
Jansen, W. C. Aquanet: a hypertext tool to hold your
knowledge in place. Proceedings of Hypertext 91, San
Antonio, Texas, (Dec. 16-18, 1991): p. 261-275.
24. Marshall, C. C. and Rogers, R. A. Two Years before
the Mist: Experiences with Aquanet. Proceedings of
the 1992 European Conference on Hypertext (ECHT
'92), Milan, Italy, (Dec. 1-4,1992): p. 53-62.
25. Marshall, C. C. and Shipman, F. M. Searching for the
Missing Link: Discovering Implicit Structure in Spatial
Hypertext. Proceedings of Hypertext 93, Seattle,
Washington, (Nov. 14-18, 1993): p. 217-230.
26. Marshall, C. C., Shipman, F. M., and Coombs, J.
VIKI: Spatial Hypertext Supporting Emergent
Structure. Proceedings of 1994 European Conference
on Hypermedia Technology (ECHT '94), 1994: p. 13-
23.
27. Moulthrop, S. You Say You Want A Revolution?
Hypertext And The Laws Of Media. Postmodern
Culture v.1, n.3 (1991).
28. Moulthrop, S. Toward a Rhetoric of Informating Texts.
ECHT '92 Proceedings of the ACM Conference on
Hypertext. Milan, Italy: p. 71-180.
29. Moulthrop, S. Traveling in the Breakdown Lane A
Principle of Resistance for Hypertext. Mosaic 28/4,
December, 1995: p. 55-77.
30. Moulthrop, S., and Kaplan, N. They Became What
They Beheld: The Futility of Resistance in the Space of
Electronic Writing. In Selfe, C. and Hilligoss, S., edd.
Literacy and Computers. New York: Modern Language
Association, 1994.
31. Moulthrop, S., and Kaplan, N. Citescapes: Supporting
Knowledge Construction on the Web.
http://raven.ubalt.
edu/staff/moulthrop/essays/citescapes/citescapes.html
32. Nielsen, J. Multimedia and Hypertext: the Internet and
Beyond. New York: Harcourt Brace, 1995.
33. Nietzsche, F. The Gay Science. New York: Vintage,
1974.
34. Rorty, R. Contingency, Irony, Solidarity. Cambridge:
Cambridge University Press, 1989.
35. Rosenberg, J. Navigating Nowhere / Hypertext
Infrawhere, SIGLINK Newsletter 3, 3, December
1994.
36. Rosenberg, J. A Conversation with Jim Rosenberg on
the Interactive Art Conference on Arts Wire, January
1996. http://www.artswire.org/Artswire/interactive
/www/jim.html
37. Rosenberg, J. The Structure of Hypertext Activity.
Proceedings of Hypertext 1996. ACM, New York.
38. Rosenberg, J. Foldings, to the Chord Trellis Relate-
Shape Draw, The Barrier Frames, Diffractions
Through. Eastgate Quarterly Review of Hypertext, vol
2, no. 3 (Summer 1996).
39. Salton, G., Singhal, A., Buckley, C., and Mandar, M..
Automatic Text Decomposition Using Text Segments
and Text Themes. Proceedings of Hypertext 1996.
ACM, New York: p. 53-65.
40. Shipman, F. M., Marshall, C. C., and Moran, T. P.
Finding and Using Implicit Structure in Human-
Organized Spatial Layouts of Information. Proceedings
of CHI `95 (May 7-11, Denver, Colorado). ACM, New
York, 1995.
41. Taylor, M., and Saarinen, E. Imagologies. New York:
Routledge, 1994.
