Automatic Reconstruction of Missing Romanian Cognates and Unattested Latin Words

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3226–3231 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3226 Automatic Reconstruction of Missing Romanian Cognates and Unattested Latin Words Alina Maria Ciobanu, Liviu P. Dinu, Laurentiu Zoicas University of Bucharest alina.ciobanu@my.fmi.unibuc.ro, ldinu@fmi.unibuc.ro, l zoicas@yahoo.fr Abstract Producing related words is a key concern in historical linguistics. Given an input word, the task is to automatically produce either its proto-word, a cognate pair or a modern word derived from it. In this paper, we apply a method for producing related words based on sequence labeling, aiming to fill in the gaps in incomplete cognate sets in Romance languages with Latin etymology (producing Romanian cognates that are missing) and to reconstruct uncertified Latin words. We further investigate an ensemble-based aggregation for combining and re-ranking the word productions of multiple languages. Keywords: cognates, word production, sequence labeling, multilinguality 1. Introduction The transition from Latin to Romance languages was fol- lowed by a series of vocabulary transformations, whose re- sults are found in all Romance languages (Sala, 1998). The main trend was the simplification of the vocabulary, which consisted mainly in eliminating archaisms and anomalies in favor of regular norms (Sala, 1998), reducing synonyms and deleting sense nuances. These tendencies have led to a diminishing of the vocabulary, making every modern Ro- mance language borrow from Latin about the same num- ber of words – around 2,000. Out of these 2,000 words, about 500 were transmitted to all Romance languages (gen- erally common, important, high-frequency terms), but most of them were preserved only in a subset of the Romance languages. There are, for example, words preserved only in the lateral areas of the Roman Empire, Romanian and the Ibero-Romance languages, such as frumos (Ro) - her- moso (Es) (meaning beautiful). Other words were pre- served in only one language (for example, there are about 100 words preserved only in Romanian). On the other hand, there are about 200 words that evolved from Latin in all Romance languages, but are not found in Romanian. The corresponding Romanian words have probably entered the language at a later stage, as borrow- ings from other languages. Out of these, many are com- mon words and it is unlikely that they were not part of the Romanian lexicon. Fischer (1985) identified a num- ber of causes that led to the disappearance of these words and to their substitution with words with different etymolo- gies: external causes, of socio-economic nature, the change of people’s occupations, the interruption of Romania’s con- tact with the Western world, the development of the Roma- nian language away from the Romance kernel, and so on. For some fundamental words there are no unanimously ac- cepted explanations. For example, there are no Latin forms for a iubi (to love), drag (dear), cocos¸ (rooster). In this paper we propose a computational approach (the first of this type, to the best of our knowledge) to reconstruct these words, starting from …

Content truncated to 500 words