Methodological Aspects of Developing and Managing an Etymological Lexical Resource: Introducing EtymDB-2.0

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3207–3216 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3207 Methodological Aspects of Developing and Managing an Etymological Lexical Resource: Introducing EtymDB 2.0 Clémentine Fourrier Benoît Sagot Inria {clementine.fourrier, benoit.sagot}@inria.fr Abstract Diachronic lexical information was mostly used in its natural field, historical linguistics, until recently, when promising but not yet conclusive applications to low resource languages machine translation started extending its usage to NLP. There is therefore a new need for fine-grained, large-coverage and accurate etymological lexical resources. In this paper, we propose a set of guidelines to generate such resources, for each step of the life-cycle of an etymological lexicon: creation, update, evaluation, dissemination, and exploitation. To illustrate the guidelines, we introduce EtymDB 2.0, an etymological database automatically generated from the Wiktionary, which contains 1.8 million lexemes, linked by more than 700,000 fine-grained etymological relations, across 2,536 living and dead languages. We also introduce use cases for which EtymDB 2.0 could represent a key resource, such as phylogenetic tree generation, low resource machine translation and medieval languages study. Keywords: Etymological lexicon, Lexical Resource Development, Language Resource Life-cycle, Methodology 1. Introduction Most available electronic lexical resources are synchronic (formalising a language as it is or was at a specific, al- though sometimes broad, period of time). Until recently, the few diachronic resources available were mostly used for computational historical linguistics tasks (List et al., 2018; Carling et al., 2018). However, over the last few years, diachronic resource use has found its way to more gen- eral applications, notably for tasks involving low resource languages, from machine translation using diachronic lan- guage relations (Nguyen and Chiang, 2017) or cognates and loan words sets (Grönroos et al., 2018) to bilingual lexicons generation for low resource languages (Nasution et al., 2017). Yet only a small number of multilingual etymological lexicons are available (de Melo, 2014; Sagot, 2017; Panta- leo et al., 2017; Batsuren et al., 2019), most extracted from the etymological information found in the Wiktionary.1 There is still room for improvement regarding the qual- ity, richness, lexical or language coverage and etymolo- gical granularity (differentiation between inheritance, bor- rowing, cognacy) of such resources. The work described in this paper has two parallel objec- tives. We investigate the methodological challenges under- lying the development and use of an etymological database based on the Wiktionary. At the same time, we describe how we addressed these challenges in the case of our own etymological database, EtymDB. After a description of existing diachronic lexicons, we go through the life-cycle of an etymological resource in five steps: creation, up- date, evaluation, dissemination and exploitation. We use as a case study our previous work on the development of EtymDB’s initial version, EtymDB 1.0 (Sagot, 2017), as well as the development of a new version, EtymDB 2.0, its evaluation, dissemination and exploitation. To illustrate the latter, we discuss possible applications in low resource lan- 1http://en.wiktionary.org guage studies, machine translation, and describe …

Content truncated to 500 words