The « Bibliothèques Virtuelles Humanistes » (Virtual Humanistic Libraries in Tours): a Collection, or a Corpus?

The goal of the Bibliothèques Virtuelles Humanistes
(BVH, or Virtual Humanist Libraries, a digitization
project begun in Tours in 2003, http://www.bvh.univtours.
fr) is to offer two types of digital representations of
a selection of books printed during the Renaissance or of
manuscripts: the image of a copy (its “facsimile”) on the
one hand, and its transcription on the other hand, without
additions besides corrections or variations that are essential
for understanding the text, and TEI encoding. These
two goals necessitate the combined efforts of two communities
whose objectives, methods and formulations
are very different. The difference is further complicated
as the elaboration of a corpus for this period presents additional
difficulties: librarians and book historians work
with “image processing” computer programs, whereas
literature and language specialists use linguistic systems
(Hyperbase, Weblex, Philologic...). Currently, technical
progress has allowed these different approaches to come
together, although they have not yet been combined: libraries
prefer to offer text-only versions of documents,
obtained via OCR or by manual transcription encoded in
order to show a readable text alone or along with its facsimile.
Linguistic corpora and text databases for works
before 1800 are often constituted of modern editions,
which are under copyright and impossible to show next
to “their” facsimiles—which often do not exist, as they
were established from several different reference editions
and do not respect the physical presentation. These
editions have the obvious advantage of easily lending
themselves to searches for data and to detailed encoding.
The goal of libraries of text images is entirely different,
as research is generally done only on metadata
and, at best, on the table of contents which constitutes
a minimal indexation, and sometimes is done through a
quick round of OCR. In light of these two traditions, and
taking into account the reading habits and requests that
recent navigational tools have encouraged, the BVH are
devoted to conducting research on the two-pronged front
of indexing text in image mode, extracting images from
images of the pages, classifying and indexing them, and
acquiring a significant corpus of transcribed texts. But
does the collection of texts constitute a corpus?
2. It is revealing that this proposition of intervention is
on the fence between the two initial themes of the congress
and we defend the name of “corpus” for the imageand-
text combination, although it was imposed upon us,
so as not to frighten financiers and the general public, to
call it a “library”. We will discuss the difficulties that this
grouping engendered, difficulties both theoretical and
technical. The digital libraries as they are comprised at
the CESR would not be a “corpus”, but rather a “collection”
as their only commonality is their period of publication,
from 1470-1650, identified as the “Renaissance”
in the largest sense, including Antique or Medieval texts
edited during that period. Even if the collection includes
broad categories (classics of the Renaissance—sources
of science—legal and political history—philosophy
and theology), a priori each book on any shelf could
fall under one of these categories, all the more as a fifth
category, “particular projects‚“ allows additions to one
or several subcategories (like the “Rabelais” database,
the dictionaries, etc.), the only ones in fact that merit
the name of “corpus”, the rest being composed of standalone
books or curiosities. This is how the members of
the jury of the National Research Agency (ANR) understood
the word “corpus” in France, as it is obvious from
the projects chosen that the notion was relatively vague,
even as they excluded a collection of digitized texts in
image mode alone. In order for there to be a corpus, there
must be text.
3. Going against this analysis, we would defend:
3. 1. The idea that each work chosen corresponds to an
analysis of its form and its content. The researchers who
are charged with this (under the supervision of Toshinori
Uetani, CESR, with the collaboration of Marie-Elisabeth
Boutroue, IRHT) examine its interest from the point of
view of the history of the book and of the directions of
research that obviously reflect the options of researchers
at the Center or their colleagues. They hope to render
the object upon which they are working accessible to the
scientific community, in order to share knowledge, going
against the traditional editorial process that consists of
offering a “definitive” paper edition once the establishment
has finished it. Selection is therefore a tool of anonymous
collaboration and results from a step that is one of
a researcher, using available funds (as it happens, those
of the Région Centre and its partner establishments) that
they enrich in so doing: the library of the Museum of
Sologne, in Romorantin, possesses a copy of the 1580
edition of Montaigne’s Essais: although it is not extraordinarily
rare, this state of the text merited being offered
to the public in order to compare it with that of the BnF’s copy (Gallica).
3. 2. The exploitation of image mode, in particular illustrated
elements, is specific to the digitized collection:
the AGORA program, developed by the computer sciences
laboratory of the University of Tours, allows semiautomatic
extraction of illustrations, graphs, portraits,
typographic material and initials, providing particular
indexed and searchable databases(notably thanks to
Thesaurus Iconclass). These “processed” works assuredly
constitute a corpus which can be subject to precise
searches.
4. As for the texts, the standard is to accept the name of
“corpus” for a database that gathers together transcriptions
of Renaissance texts in French.
4. 1. But what about texts in Latin, Italian, English,
Spanish, etc., that are not excluded a priori? A multilingual
corpus is all the more conceivable as many Renaissance
texts contain large fragments of text in Latin
(the Essais, for example). Even if we initially preferred
texts in French, originals or translations, leaving to researchers
from other linguistic regions the care of developing
their own corpora, the fluidity of TEI encoding
allows the possibility of making a “Renaissance” corpus
that would not be limited to French and would contain
even bilingual or multilingual “alined” corpora. Thanks
to the “TEI Renaissance and modern times” application
(Tours, July 2008), appropriate encoding of different
versions of a text makes a model available that renders
the physical description of a text compatible with its logical
description (Nicole Dufournaud, CESR, Jean-Daniel
Fekete, INRIA). These procedures are taught in a specialized
Master’s program and during workshops open
to students, researchers and librarians.
4. 2. The progress of “Renaissance” OCR (RETRO) also
developed in Tours by Jean-Yves Ramel allows the acquisition
of text from difficult-to-read printed matter, and
it allows correction thanks to form dictionaries compiled
from transcriptions and in different languages. Even if
acquisition in text mode with an accuracy rate of over
97% still represents a considerable cost for these early
printed books (post-correction is always necessary), it
allows incrementation of a collection of texts with highly
varying written forms, processed by an annotation tool
(Analog, developed in Poitiers and at the ENS-LSH by
Marie-Hélène Lay) or a “dissimilation” tool (Thierry
Vincent, Poitiers); these collections in turn facilitate the
acquisition of new texts and allow for linguistic analyses
about the uses found with double-checking the text: in
the context of course, and in the facsimile. It will also be
possible to search the texts for keywords that constitute
the Iconclass thesaurus, offering in this way a selection
of topoi present in the texts, in their tables of contents
(all transcribed) and in the illustrated elements.
5. In this way, the BVH is a “Renaissance” corpus that
is still in the process of being built, a unique but double-
headed corpus, one that is dissymmetric: there are
obviously many more facsimiles than transcribed texts.
Such a corpus is unified by the metadata that harmonize
the catalogue of the work and that of the transcribed text
thanks to appropriate descriptors (TEI headers, Dublin
Core, OAI protocol) and by the search tools that are used
on the plain text, but also thanks to the keywords associated
with images and with texts, always offering as
an anchor the facsimile and the reference to the original
work.