Reimagining the Dictionary, or Why Lexicography Needs Digital Humanities

The promise of eLexicography stems not only
from the transformation of the production
medium, but also from the technological
feasibility of representing linguistic complexity.
Even though modern lexicography is
unimaginable without computer technology
(Hockey, 2000a; Knowles, 1989; Meijs, 1992),
the sheer use of computers in producing
a dictionary or delivering it electronically
does not automatically transform a dictionary
from "a simple artefact" to a "more complex
lexical architecture," to use Sinclair's (2000)
formulations.
Calling dictionaries “simple artefacts” is itself a
rhetorical oversimplification: there is certainly
nothing simple about a dictionary — whether we
look at it as a material object, cultural product
or a model of language. Yet the overall structure
of dictionaries as extended word lists has not
changed in centuries (Hausmann et al., 1989;
Fontenelle, 2008; Atkins and Rundell, 2008).
Admittedly, a great deal of factual information
is packed into a prototypical lexicographic
entry, but a defined term often remains
in isolation and insufficiently connected or
embedded into the language system as a whole.
This is what Miller refers to as the “woeful
incompleteness” (Miller at al.) of a traditional
dictionary entry, and what Shvedova sees as
its “paradoxical nature” — dictionary entries
tend to be “lexicocentric” while language itself is
“class-centric” (Шведова, 1988).
Furthermore, the advances in digital
humanities, textual studies and postmodern
literary theory do not seem to have had a
profound effect on the way we theorize or
produce dictionaries. Surely, many important
lexicographic projects have been digitalized
and gone online; web-portals increasingly
offer cumulative searches across different
dictionaries; and eLexicography is a thriving
field (Lemberg et al., 2001; Hockey, 2000a; de
Schryver; Hass, 2005; Nielsen, 2009; Rundell,
2009; Hass, 2005), yet dictionaries — often
commercial enterprises which are guided by
predominantly economic concerns — remain by
far and large discrete objects: no more and no
less than digitalized versions of stable, print
editions. We still consult dictionaries by going to
a particular web site. Dictionaries do not come
to us.
The time is ripe to ask — both in theoretical
and practical terms — a new set of questions:
how has the electronic text changed our notion
of what a dictionary is (and ought to be); how
have the methods of digital humanities and the
advances made in digital libraries altered our
idea of what a dictionary can (and should) do?
And, finally, where do we go from here?
The dictionary is a kind of text. In print
culture, the dictionary, like every other text,
had its material and semantic dimension. The
semantic dimension was represented on its
visible surface, whereas its depth was in the
mind of the reader, or what Eco refers to as
the "encyclopedia of the reader." (Eco et al.,
1992; Eco, 1979). Yet if we — as we should
— start thinking of the dictionary as a kind of
electronic text, the way Kathrine Hayles and
others have done for electronic literature, we
will have no choice but to strip the dictionary
of its finality and its "object-ness" and see in it,
instead, only one possible manifestation of the
database in which it is stored (Hayles, 2003;
Hayles, 2006; Folsom, 2007). A digital text
can be not only edited, transformed, cut and
pasted — as part of our computational textual
kinetics — but is always part of other activities:
search, downloading, surfing. In other words,
an electronic text is unimaginable without its
context (Aarseth, 1997; DeRose et al., 1990;
Hockey, 2000b).
The dictionary, then, should be seen as a
kind of semantic potential that can be realized
through its use. But in order to truly fulfill
this potential, the dictionary needs to be
embedded in the digital flow of our textual
production and reception. That is why we
cannot think of dictionaries any more without
thinking about digital libraries and the status

2
which electronic texts have in them (Andrews
and Law, 2004; Candela et al., 2007; Kruk
and McDaniel, 2009; Maness, 2006; Miller,
2005; Novotny, 2006). To be truly useful
for any kind of textual studies, the digital
library must "explode" the text (by providing
full-content searchability, concordances and
indexes, metadata, hyperlinks, critical markup
etc.) instead of "freezing" it as an image,
which, albeit digital, is computationally neither
intelligible nor modifiable as text. In smart
digital libraries, a text should not only be an
object but a service; not a static entity but an
interactive method (Tasovac, forthcoming). The
text should be computationally exploitable so
that it can be sampled and used, not simply
reproduced in its entirety. This kind of atomic
approach to textuality poses a host of challenges
(legal, ethical, technical and intellectual, to
name just a few), but it opens up the possibility
of creative engagement with the digital text
in literary studies (text mining, statistical text
comparison, data visualization, hypertextual
systems etc.).
The consequence of this "explosive" nature of
the electronic text is of paramount importance
for eLexcicography and the reformulation of
the dictionary not as an object, but a service.
We should start thinking of and building
dictionaries as fully embeddable modules in
digital libraries, or, to put it differently, build
digital libraries which integrate dictionaries as
part of their fundamental infrastructure and
allow an ever-expandable process of associating
words in an electronic text with an equally
changeable record in a textual database. The
changeability of the dictionary entry will, in
turn, defer
ad infinitum
the notion of a
particular dictionary edition — other than
as temporary snapshot of the database. The
dictionary as an evolving process will be in a
permanent beta state.
The future of electronic dictionaries
undoubtedly lies in their detachability from
physical media (CD, DVD, desktop applications)
and static locations (web portals). If we think of
the dictionary as a service with an API
1
that can
be called from any Web page, we can actually
start thinking about any (electronic) text as a
direct entry point to the dictionary. If every word
in a digital library is a link to a particular entry
in the dictionary, electronic textuality as such
becomes an extension of lexicography: the text
begins to contain the dictionary in the same way
that the dictionary contains the text.
The Center for Digital Humanities (Belgrade,
Serbia) is putting these theoretical
considerations into practice while working on
its flagship
Transpoetika Project
(Tasovac,
2009).
Transpoetika
(see Figure 1) is
a collaborative, class-centric, bilingualized
Serbian-English learner‘s dictionary based on
the architecturally complex, machine-readable
semantic network of the Princeton Wordnet
(Fellbaum, 1998; Vossen, 1998; Stamou et
al., 2002; Tufis et al., 2004). It is part of
a scalable, web-based, digital framework for
editing and publishing annotated, fully-glossed
study editions of literary works in the Serbian
language, primarily aimed and students of
Serbian as a second or inherited language.
Transpoetika
has been designed to be deployed
as a web service and therefore linked from
and applied to a variety of textual sources
online. Portions of the project, such as the
Serbian Morpho-Syntactic Database (SMS)
already function as a web service internally
and will also be made public and free once
the sufficient funding for the project has been
secured.
Transpoetika
can also interact with
other web services: by using Flickr as a source
of illustrations, and Twitter as a source of
"live quotes" in the entries, the
Transpoetika
Dictionary explores the role of serendipity in a
lexicographic text.
The overarching goal of the Belgrade Center
for Digital Humanities (CDHN) is to produce
a pluggable, service-based, meta-lexicographic
platform for the Serbian language, which
will interact with various Web-based digital
libraries, and contain not only our own
bilingualized Serbian Wordnet, but also
historical Serbian dictionaries that the CDHN
is digitalizing, such as, for instance, the
classic Serbian-German-Latin Dictionary by
Vuk Stefanović-Karadžić (1818 and 1852). The
platform could, in theory, be extended to
include and consolidate a number of other, more
specialized, lexicons. This is, in any case, the
general direction we would like to take.
I would like to conclude with a
hysteron-
proter+E43on
, which, in Samuel Johnson's
Dictionary of the English language was defined