Dissent and Collaboration

Collaboration—literally a shared work—is typically
understood to rest upon a form of agreement: about
shared goals, common projects, standards of practice. As
Aldo de Moor observes, modern-day collaborative practices
tend to emerge from self-forming teams rather than
being organized from above, and as a result this kind of
agreement constitutes an essential foundation on which
to proceed. He goes on to note that the crucial element
to successful digital collaboration has a great deal to do
with the way norms are developed and adjudicated:
The members of virtual professional communities, just
like their peers in more traditional communities, are guided
in their work by social norms. These norms guide both
the operational activities of the network and the specification
processes in which the network is defined. However,
as these networks are egalitarian by nature, the norms cannot
be imposed from above, but are to emerge from the
community as a whole. Thus, the user-driven specification
process needs to be legitimate, in the sense that specification
changes are not only meaningful but also acceptable
to all community members. A specification change is acceptable
if and only if the users for whom the particular
change is relevant have been adequately involved in
the specification process (Aldo de Moor, “Empowering
Communities: A Method for the Legitimate User-Driven
Specification of Network Information Systems”, summary
of PhD thesis, http://www.communitysense.nl/phd/
thes_summeng.html).
Relevant norms in the digital humanities context are
quite wide-ranging: they include social norms, disciplinary
norms, and technical norms. But what I would like
to focus on for purposes of this paper is a specific kind
of case in which disciplinary and technical norms overlap:
the arena of standards for digital representation of
research materials, and in particular the domain of scholarly
markup languages. Text encoding as practiced in the
digital humanities world sits at the juncture of humanities
scholarship—textually nuanced, exploratory, and
introspective—and digital technology, with its emphasis
on formalism and upward scalability. As a result its
norms carry a double weight: they must achieve some
kind of technically actionable uniformity, but they must
also express useful scholarly concepts and differentiations.
Encoding standards such as the Text Encoding Initiative
(TEI, http://www.tei-c.org/) are thus foundationally collaborative
technologies: they presume the need and the
desire to make individual insight widely communicable
in a form that permits its extension, critique, and reuse.
But the mechanisms for achieving this result in practical
terms are complex and require thoughtful balancing of
the requirements, respectively, of the individual and the
community. As de Moor observes in the material quoted
above, legitimate norms arise from a process of community
assent, but how is that assent best managed and expressed?
And in the universe of humanities disciplines,
where “the community” consists of multiple communities
with shifting boundaries, how are norms constituted
in a way that still permits intellectual growth? Even
more importantly, given the critical role that dissent and
debate play within the humanities research context, how
can these be expressed? Can we imagine a role for dissent
within a functioning technical standard, without vitiating
its power to support collaboration?
Over the past 20 years, the research of the TEI and its
user community has been centered on developing mechanisms
that address the problem of norm-setting, in both
the social and the technical sphere, in a way that I will
argue is specifically designed to accommodate dissent in
a way that actively facilitates collaboration. This paper
will critically examine the TEI’s framing motivations
and the specific mechanisms—intellectual, social, and
above all technical—through which they have been realized
during the course of the TEI’s development. In
particular I will consider the practice of schema customization,
through which the TEI manages both the representation
of the TEI language as a standard and the
processes of dissent and expansion through which it is
modified by its users. The central components of the TEI
customization process express, in effect, the relationship
between the individual and the community. The TEI
source or ODD file represents the entire landscape of the
TEI in potential terms, and can be used to generate a
maximally capacious schema that contains all possible
TEI elements and structures. The ODD customization
file represents the world of an individual user or project:
the set of choices and limitations or extensions through
which the individual adapts the TEI schema to local usage.
From these two files, with appropriate processing,
one can generate a schema that expresses the TEI landscape
as viewed through the lens of the individual application.
The ODD customization file, then, expresses
the distance between the individual and the community:
both the agreement the community has established concerning
meaning, and also (perhaps more importantly)
the degree by which the individual dissents from the
standard, while still acknowledging its centrality for the community as a whole.
Dissent is nothing new in the humanities; what is distinctive
about this mechanism is that it formalizes dissent
and allows its vector to be traversed in two directions.
The same path that leads away from the unmodified
TEI standard towards the individual application (from
generality to specificity) can also be followed back to
the center again. This traversal can be effected both by
human beings and by computer processes. Information
concerning what has been changed and why can be
expressed in human-readable form and may serve as a
valuable support in understanding the methodological
choices that underlie a project’s encoding practice. Similarly,
the ODD customization file can serve as the basis
for automated analysis that could, for example, identify
all projects from a large set that use the same set of TEI
modules or remove the same set of elements; generate a
list representing the greatest common set of values for a
given attribute across a group of projects and also identify
the values that are unique to each project; identify
the range of new elements created by each project and
their TEI equivalents. Taken as a whole, the customizable
approach taken by the TEI permits the standard to
function (both socially and technically) as an agreement
at many levels—on the intention to treat data as a sharable
and preservable resource, on the value of shared
data standards, on the descriptive utility of this particular
approach to modelling humanities texts, and on the
impossibility of creating a single descriptive model that
will satisfy all needs.
In an important sense, this customization mechanism encapsulates
the central challenge of collaborative work,
and even of language itself: that of how to balance the
urge toward individual expressiveness with the mandates
of public comprehensibility, the desire for individual
freedom of agency against the need for group action.
This paper will explore several specific examples drawn
from real TEI projects to demonstrate the different kinds
of managed dissent that can be expressed using the customization
mechanism, including the following:
• the use of shared generic structures (<div>, <seg>,
etc.) coupled with customized constraint of attribute
values to express variations on common structures
• the use of “syntactic sugar”: the creation of projectspecific
elements that are explicitly equivalent to
standard TEI elements
• the use of the TEI class system to express how a
project-specific vocabulary fits into the TEI’s structural
model
The important question emerging from these issues is
what this kind of discursive agreement and dissent actually
achieves in the collaborative domain. Several points
should be noted and will be explored in more detail in the
finished paper. First, agreement of course permits communication,
but properly formalized dissent is an equally
important dimension of communication. Language (including
formal encoding language) expresses a view
of the world, and it is essential to be able to disagree
about that view, or to express a different view. But that
disagreement must be explicit rather than covert: otherwise
tag abuse is the result. A second, complementary
question is how we can distinguish between meaningful
dissent (essential points of disciplinary difference) and
meaningless dissent (based on laziness or other social
failures). Are the mechanisms for making this distinction
solely social and human, or are there automatable
mechanisms conceivable as well? Finally, what kinds
of collaboration are possible where formalized dissent
and disagreement, rather than complete agreement, are
the result of using a standard? I will argue that for the
digital humanities community, effective management
of disagreement—rather than simply the production of
agreement—is the most important role a standard like
the TEI can play.