Building the Spanish-Croatian Parallel Corpus

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3932–3936 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3932 Building the Spanish-Croatian Parallel Corpus Bojana Mikelenić, Marko Tadić Faculty of Humanities and Social Sciences, University of Zagreb Ivana Lučića 3, 10000 Zagreb bmikelen@ffzg.hr, marko.tadic@ffzg.hr Abstract This paper describes the building of the first Spanish-Croatian unidirectional parallel corpus, which has been constructed at the Faculty of Humanities and Social Sciences of the University of Zagreb. The corpus is comprised of eleven Spanish novels and their translations to Croatian done by six different professional translators. All the texts were published between 1999 and 2012. The corpus has more than 2 Mw, with approximately 1 Mw for each language. It was automatically sentence segmented and aligned, as well as manually post-corrected, and contains 71,778 translation units. In order to protect the copyright and to make the corpus available under permissive CC-BY licence, the aligned translation units are shuffled. This limits the usability of the corpus for research of language units at sentence and lower language levels only. There are two versions of the corpus in TMX format that will be available for download through META-SHARE and CLARIN ERIC infrastructure. The former contains plain TMX, while the latter is lemmatised and POS-tagged and stored in the aTMX format. Keywords: written corpus, parallel corpus, Spanish, Croatian 1. Introduction A parallel or translation corpus can be bilingual or multilingual (texts in one language and their translation to one or more languages) and unidirectional (source language→target language), bidirectional (language 1↔language 2) or multidirectional (source language→multiple target languages) (McEnery, Xiao and Tono, 2006, 48). This paper describes the process of building a bilingual unidirectional (Spanish→Croatian) parallel corpus, where a well-resourced language (Spanish) is paired with an under-resourced language (Croatian). The possible uses of parallel corpora in linguistic and NLP research are well known and span from contrastive and translation analyses to bilingual lexica extraction, and they could also be used for training statistical or neural machine translation models. The corpus we are describing here was built in col- laboration between two departments of the Faculty of Humanities and Social Sciences of the University of Zagreb, namely Department of Romance Languages and Literatures and Department of Linguistics. The paper is organised as follows: Section 2 summarizes previous work on parallel corpora that includes this language pair; in Section 3 we give a brief overview of the corpus parameters; Section 4 is dedicated to the corpus composition, including sampling procedure (4.1) and output format (4.2); Section 5 describes the corpus processing and Section 6 gives a conclusion and indicates possible future research directions. 2. Previous work To the best of our knowledge, this is the first direct Spanish-Croatian parallel corpus collected as such and the only previous work we can make a reference to are larger multilingual text collections, which allow for extraction of parallel bitexts in these two languages:  OpenSubs (Tiedemann, 2012), issues 20161 and 20182, the latter being …

Content truncated to 500 words