Abstract
Language documentation projects typically invest a lot of effort in creating digitized lexical resources, which are used in the creation of dictionaries and in the glossing of collected texts. We present and evaluate a methodology for repurposing such a lexical resource developed for Chintang (ISO639-3: ctn), a language of Nepal, for use with a precision implemented grammar developed in the DELPH-IN formalism. The target lexicon, when combined with a set of morphological rules, achieves 57% type-level coverage and 50% token-level coverage of held-out texts, while maintaining a feature-level accuracy F-measure of 70%. As lexicon development is typically one of the most expensive aspects of creating a precision grammar, this represents a significant savings of effort.