Building an Endangered Language Resource in the Classroom: Universal Dependencies for Kakataibo

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 3840–3851 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 3840 Building an Endangered Language Resource in the Classroom: Universal Dependencies for Kakataibo Roberto Zariquieyρ, Claudia Alvaradoρ, Ximena Echevarriaρ, Luisa Gomezρ, Rosa Gonzalesρ , Mariana Illescasρ , Sabina Oportoρ, Frederic Blumβ, Arturo Oncevayϵ, Javier Veraυ ρPontificia Universidad Cat´olica del Per´u, Peru βHumboldt-Universit¨at zu Berlin and Leibniz-Zentrum Allgemeine Sprachwissenschaft ϵUniversity of Edinburgh, Scotland υPontificia Universidad Cat´olica de Valpara´ıso, Chile {rzariquiey, claudia.alvarado, ximena.echevarria, luisa.gomez, a20175617, m.illescasb, sabina.oporto}@pucp.edu.pe frederic.blum@hu-berlin.de, a.oncevay@ed.ac.uk, javier.vera@pucv.cl Abstract In this paper, we launch a new Universal Dependencies treebank for an endangered language from Amazonia: Kakataibo, a Panoan language spoken in Peru. We first discuss the collaborative methodology implemented, which proved effective to create a treebank in the context of a Computational Linguistic course for undergraduates. Then, we describe the general details of the treebank and the language-specific considerations implemented for the proposed annotation. We finally conduct some experiments on part-of-speech tagging and syntactic dependency parsing. We focus on monolingual and transfer learning settings, where we study the impact of a Shipibo-Konibo treebank, another Panoan language resource. Keywords: Universal Dependencies, Treebank, Collaborative Methodology, Kakataibo, Endangered Languages, Panoan, Amazonia, Peru 1. Introduction Kakataibo is a language that belongs to the Panoan family spoken by around 3,000 native speakers in the Amazonian region of Peru. This paper describes the methodology implemented in the context of a regu- lar undergraduate Computational Linguistics course to create a UD treebank for this language, as a strategy to develop significant learning settings and at the same time contribute to the computarization of this endan- gered language of Peruvian Amazon. The Kakataibo UD treebank would enhance the future development of an NLP toolkit for this language, since it is the main requirement to train a dependency parser. By taking advantage of the preexistence of a UD treebank for another Panoan language, Shipibo-Konibo (Vasquez et al., 2018), in this paper, we conduct some experiments in both monolingual and transfer learning settings. The paper is organized as follows. First, §2 presents some background information on the Kakataibo lan- guage. Then, §3 describes the methodology imple- mented in the classroom. §4 introduces the Kakataibo UD treebank. §5 presents the experimentation con- ducted on part-of-speech tagging and syntactic depen- dency parsing in both monolingual and transfer learn- ing settings (using the Shipibo-Konibo as a baseline). Finally, §6 summarises the conclusions of this paper. 2. The Kakataibo Language Kakataibo (cbr) is a Panoan language spoken by approximately 3,000 people in the Peruvian depart- ments of Hu´anuco and Ucayali. The Kakataibo peo- ple live in various communities along the Aguayt´ıa, San Alejandro, Shamboyacu, Sungaroyacu and Pisqui Rivers, where the language remains vital despite dif- ferent degrees of contact between Kakataibo people and non-indigenous populations. Zariquiey (2011) dis- tinguishes four living Kakataibo varieties: the Lower Aguayt´ıa/Shamboyacu, Upper Aguayt´ıa, Sungaroyacu and San Alejandro dialects. Nokam´an, a variety named and minimally documented by Tessmann (1930), was a fifth variety, …

Content truncated to 500 words