Computer Glossing as a Tool in Convergence Research

Abstract

This paper reports the use of computer glossing to assist in the generation of parallel texts for an ongoing research project on Dravidianization in Sourashtra (Tamil Nadu Gujarati).

The initial data base is a corpus of 35 hours of recorded conversations in Sourashtra. The machine glossing system inputs a Sourashtra text and, after performing some preliminary morphological analysis, generates a Tamil gloss on a simple unit-by-unit basis from a computerized glossary. The machine output is given to bilingual assistants as a basis for their translation. The approach is particularly suited to work employing linguistically untrained assistants.