Jojajovai: A Parallel Guarani-Spanish Corpus for MT Benchmarking

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2098–2107 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2098 Jojajovai: A Parallel Guarani-Spanish Corpus for MT Benchmarking Luis Chiruzzo1, Santiago G´ongora1, Aldo Alvarez2, Gustavo Gim´enez-Lugo3, Marvin Ag¨uero-Torales4,5, Yliana Rodr´ıguez1 1Universidad de la Rep´ublica, Montevideo, Uruguay 2Universidad Nacional de Itap´ua, Encarnaci´on, Paraguay 3Universidade Tecnol´ogica Federal do Paran´a, Curitiba, PR - Brasil 4Universidad de Granada, Granada, Spain 5Internet Interdisciplinary Institute (IN3), Universitat Oberta de Catalunya, Barcelona, Spain luischir@fing.edu.uy, sgongora@fing.edu.uy, aldo.alvarez@fiuni.edu.py, gustavogl@utfpr.edu.br, maguero@correo.ugr.es, yrodriguez@fhuce.edu.uy Abstract This work presents a parallel corpus of Guarani-Spanish text aligned at sentence level. The corpus contains about 30,000 sentence pairs, and is structured as a collection of subsets from different sources, further split into training, development and test sets. A sample of sentences from the test set was manually annotated by native speakers in order to incorporate meta-linguistic annotations about the Guarani dialects present in the corpus and also the correctness of the alignment and translation. We also present some baseline MT experiments and analyze the results in terms of the subsets. We hope this corpus can be used as a benchmark for testing Guarani-Spanish MT systems, and aim to expand and improve the quality of the corpus in future iterations. Keywords: Guarani, Jopara, Jehe’a, Machine Translation, language contact, corpus linguistics 1. Introduction Guarani is one of the few indigenous languages spo- ken daily by both indigenous and non-indigenous peo- ple. Even though it is not a minority language in terms of its speakers, it is under-resourced and under- researched from a computational linguistics perspec- tive. Together with Spanish, Guarani is an official lan- guage of Paraguay, making it the only South American nation where an indigenous language has survived as a majority language spoken by the non-indigenous pop- ulation (Estigarribia, 2015). It is also spoken in Bo- livia, Argentina and Brazil, and is one of the official languages of Mercosur1. However, as no language remains immutable through- out time, the long history of contact between Guarani and Spanish in South America has resulted in many interesting language varieties, the most salient being Jopara and Jehe’a. Here we will refer to the Jehe’a vari- ant as a mixture between Guarani and Spanish where Spanish loanwords are incorporated with their mor- phology adapted to Guarani; while we will refer to the Jopara variant as a deeper mixture between Guarani and Spanish, often involving code switching and Span- ish loanwords that keep their original morphology. It is worth mentioning that language contact and its consequences are not exceptional nor new linguistic phenomena. In fact, Guarani-Spanish language mixing can be traced back to colonial times in the Jesuits notes, (e.g. Dobrizhoffer (1783)). Contact between both lan- guages started with colonization during the first initial 1Trade bloc and regional organization involving a number of countries in South America. encounters with Europeans in the 1500s. In spite of this longstanding contact between Guarani and Span- ish, there are very scarce bilingual texts to resort …

Content truncated to 500 words