A Resource for Studying Chatino Verbal Morphology

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 2827–2831 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 2827 A Resource for Studying Chatino Verbal Morphology Hilaria Cruz1, Antonios Anastasopoulos2, and Gregory Stump3 1University of Louisville, 2211 South Brook, Louisville KY 40292, USA, 2Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh PA 15213, USA, 3University of Kentucky, 13964 West 144th Court, Olathe KS 66062, USA hilaria.cruz@louisville.edu, aanastas@cs.cmu.edu, gregorystump76@gmail.com Abstract We present the first resource focusing on the verbal inflectional morphology of San Juan Quiahije Chatino, a tonal mesoamerican language spoken in Mexico. We provide a collection of complete inflection tables of 198 lemmata, with morphological tags based on the UniMorph schema. We also provide baseline results on three core NLP tasks: morphological analysis, lemmatization, and morphological inflection. Keywords: Chatino, Endangered Languages, Morphology 1. Introduction The recent years have seen unprecedented forward steps for Natural Language Processing (NLP) over almost every NLP subtask, relying on the advent of large data collec- tions that can be leveraged to train deep neural networks. However, this progress has solely been observed in lan- guages with significant data resources, while low-resource languages are left behind. The situation for endangered languages is usually even worse, as the focus of the scientific community mostly re- lies in language documentation. The typical endangered language documentation process typically includes the cre- ation of language resources in the form of word lists, audio and video recordings, notes, or grammar fragments, with the created resources then stored into large online linguis- tics archives. This process is often hindered by the so-called Transcription Bottleneck, but recent advances ( ´Cavar et al., 2016; Adams et al., 2018) provide promising directions to- wards a solution for this issue. However, language documentation and linguistic descrip- tion, although extremely important itself, does not mean- ingfully contribute to language conservation, which aims to ensure that the language stays in use. We believe that a ma- jor avenue towards continual language use is by further cre- ating language technologies for endangered languages, es- sentially elevating them to the same level as high-resource, economically or politically stronger languages. The majority of the world’s languages are categorized as synthetic, meaning that they have rich morphology, be it fusional, agglutinative, polysynthetic, or a mixture thereof. As Natural Language Processing (NLP) keeps expanding its frontiers to encompass more and more languages, mod- eling of the grammatical functions that guide language gen- eration is of utmost importance. It follows, then, that the next crucial step for expanding NLP research on endan- gered languages is creating benchmarks for standard NLP tasks in such languages. With this work we take a small first step towards this direc- tion. We present a resource that allows for benchmarking two NLP tasks in San Juan Quiahije Chatino, an endan- gered language spoken in southern Mexico: morphological All Train Dev Test Paradigms 198 Verb Classes 29 Forms 4716 3774 471 471 Table 1: Basic Statistics of our resource. analysis and morphological …

Content truncated to 500 words