Computer-Assisted Textual Analysis: Interpreting a Medieval Text Across Languages and Time

Composed in multilingual twelfth-century England, the influential
_Lais_ of Marie de France have been read by medievalists and translated
into languages around the world (Maréchal 2003). The context in which
Marie's _lais_ were composed and transcribed can be characterized as
follows: Marie is believed to have lived and written in twelfth-century
England, which was characterized by trilingualism or triglossia (i.e.,
English, French, Latin), with a Celtic substratum. She composed in a
dialect of Old French tales that she had heard from the "ancient Bretons."
The favored manuscript of Marie's collection (British Library MS Harley
978) exhibits features of Anglo-Norman or Anglo-French. Interpretation of
such a poetic text depends upon a deep knowledge of semantic and
grammatical information, especially in connection with those key terms that
are "charged" with polysemy (Burnshaw 1976). Additionally, like many other
medieval works, this text abounds with homographs, homophones, and
paranomasia, as well as a lack of consistent orthography. Moreover, in the
introduction to her collection of tales or lais, Marie de France emphasizes
that in transmitting older works ("des anciens") subsequent generations
must "gloser la letre," adding a "surplus de sens" ("Prologue," _Lais_).
Implicit in this description of reading, interpreting, and translating is
the idea of a secondary creation and a recognition that texts and glosses
are submitted to a continual transformation over time (Dragonetti 1993).
Given this invitation by the author, Marie's text seems well suited to
computer-assisted textual analysis and translation.
This paper describes a prototype project, currently under
development, consisting of a three-part computer-assisted model for
providing close analysis and translation of an Old French text, composed in
England and transcribed by an Anglo-Norman scribe.
This model utilizes three computer components: digitized parallel texts of
the original Old French and of a translation into modern English;
hypertextual links from the
original text and the translation to an extensive glossary and to a guide
illustrating the grammatical function of the individual term; a relational
database that offers several kinds of lexicographic information (e.g.,
semantic, etymological, and contextual).
These three computational supplements can guide, assist, and enhance
readings and interpretations in the original and in translation:
1. The separately scrolling digitized (and therefore searchable) text in
Old French and the relatively literal translation into modern English can
be read individually or in tandem for comparing parallel passages.
2. Hypertextual links that illustrate the syntactic function of the
original offer a simplified version of the grammatical relationships among
words (for instance, illustrating the effects of the reduced but still
operational declensional system).
3. The relational lexicographic database enables a user to compare
definitions from a wide array of dictionaries and glossaries (see, for
example, Hockey 2000). Queries can show all word instances (some with
variant spellings of the same lemma), homographs, the range of definitions
of the word in English, etymology and word derivation, terms related to
specialized registers, and matrices of linguistic relationships (e.g., Old
French, Modern French, Old English, Middle English, Latin).
This project is inherently collaborative and interdisciplinary, requiring
at least a small team with complementary areas of expertise: specialists in
Old French, Old and Middle English and Modern English; a person with
database skills, and an interface programmer. In this project, the process
entails transcribing the Old French text from its manuscript form; gleaning
definitions and other relevant information from multiple specialized
dictionaries (e.g., Tobler- Lommatzsch's _Altfranzösisches Wörterbuch_,
Godefroy's _Lexique de l'ancien français_, the _D.E.A.F._, the _Old English
Dictionary_, the _Middle English Dictionary_, the _Anglo-Norman
Dictionary_, the Oxford English Dictionary, the _Robert_, dictionaries of
classical and medieval Latin, and glossaries of editions of related works);
inputting lexical forms, definitions and grammatical information; adapting
the database; designing interfaces for ease in entering information and
making queries. Once the necessary information has been digitized and
organized into the various constituent parts, the computer performs the
support tasks that it does best (e.g., filtering, remembering, recognizing
patterns, and providing a specialized knowledge base), thereby enabling
experienced as well as less-skilled readers, critics, or translators to
focus on the complex decisions that undergird interpretation or
translation.