Building a Knowledge Base of Morphosyntactic Terminology

D. Terence Langendoen, William Lewis, and Scott Farrar

University of Arizona

As an initial step in the development of markup recommendations for the
E-MELD project, we are constructing a knowledge base of morphosyntactic
terms. The idea for this project arose in discussion by the markup working
group at the Endangered Languages Workshop in Santa Barbara in June. The
group agreed that linguists in general, and the endangered language
communities in particular, need guidance on the use of linguistic
terminology in making language data, analysis, and tools available
electronically. Such guidance is needed not only for the preparation of
websites, but also for using them for research. One problem we face is
that certain terms are used differently by scholars working on different
language groups For example 'absolutive' in Uto-Aztecan language
descriptions typically refers to an affixed form in opposition to an
uninflected 'possessive' form, whereas 'absolutive' for many other
languages refers to a case form (often unmarked morphologically) in
opposition to 'ergative'. The situation in which different terms are used
for the same notion is also familiar. We felt that the best place to start
work on the problem of identifying and relating terms is in the area of
morphosyntax, since a great deal of morphosyntactic analysis can already
be found on the web, and such analysis is almost invariably part of any
new project that involves presenting analyzed text, dictionaries, and
grammars.

We are seeking to develop a database structure and content that can be
easily accessed, modified and extended by field linguists and native
language communities working cooperatively. At the moment, we have
organized about 900 terms into a conceptual ontology using the
Protege-2000 tool available from Stanford Medical Informatics. To begin
with, we were able to import about 500 terms from a database developed in
1997 by SIL for their Lingua Links project, and used with their
permission. We are currently adding terms by hand from lists prepared by
the DOBES group, and information found in T. Payne's Morphosyntax, D.
Crystal's Dictionary of Linguistics and Phonetics and other "standard"
sources. The real challenge will come when we try to mine primary sources
of materials that have been developed for a variety of languages world
wide. Before the conference, we will incorporate the terms used in a
variety of Uto-Aztecan language materials including the Hopi Dictionary.