Using Guarani Verbal Morphology on Guarani-Spanish Machine Translation Experiments

Resumen

This paper shows the results of a project for building computational tools and resources for processing the Guarani language, an under-researched language in the NLP community. We developed a baseline machine translation system for the Guarani-Spanish pair, and then performed a series of experiments trying to improve its quality using morphological information. In this work we focus on the analysis of verbs, which is the most complex part of speech in Guarani. We report the results of the different tools implemented for verbs analysis and detection in Guarani, as well as the experiments on machine translation carried on using different versions of the corpus augmented with morphological features.