HELFI: a Hebrew-Greek-Finnish Parallel Bible Corpus with Cross-Lingual Morpheme Alignment

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4229–4236 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4229 HELFI: a Hebrew-Greek-Finnish Parallel Bible Corpus with Cross-Lingual Morpheme Alignment Anssi Yli-Jyrä1,3, Josi Purhonen1,2, Matti Liljeqvist3, Arto Antturi3, Pekka Nieminen3, Kari M. Räntilä3, Valtter Luoto3 1Department of Digital Humanities, University of Helsinki, 00014, Finland anssi.yli-jyra@helsinki.fi 2Faculty of Theology, University of Helsinki, 3Raamatun Tietokirja, Aika[media] Oy Abstract Twenty-five years ago, morphologically aligned Hebrew-Finnish and Greek-Finnish bitexts (texts accompanied by a trans- lation) were constructed manually in order to create an analytical concordance (Luoto et al., 1997) for a Finnish Bible translation. The creators of the bitexts recently secured the publisher’s permission to release its fine-grained alignment, but the alignment was still dependent on proprietary, third-party resources such as a copyrighted text edition and pro- prietary morphological analyses of the source texts. In this paper, we describe a nontrivial editorial process starting from the creation of the original one-purpose database and ending with its reconstruction using only freely available text editions and annotations. This process produced an openly available dataset that contains (i) the source texts and their translations, (ii) the morphological analyses, (iii) the cross-lingual morpheme alignments. Keywords: parallel texts, word alignment, the Bible, Finnish translation, alignment guidelines, free annotation 1. Introduction Parallel editions of Bible translations have existed for 1,800 years, but now there is a steadily growing interest to attach a fine-grained alignment to the numerous translations of the Bible and other par- allel texts. The interest brings together linguistics, theology, translation studies and language engineering. Fine-grained manual text alignment produces valuable linked data sets and gold standards for automatic alignment. Tools for producing such manual an- notations range from coarse word group aligners (Melamed, 1998b) to ones that align words with confi- dence levels (Lambert et al., 2005b) and to ones that link various types of subword units (Yli-Jyrä, 1993; Yli-Jyrä, 1995). The obtained gold datasets support the evaluation of alignment algorithms (Mihalcea and Pedersen, 2003; Lambert et al., 2005a; Cysouw et al., 2007) and the development of methods for automated corpus linguistics (Szymanski, 2012). The aim of this work is to produce an openly shareable, fine-grained alignment for parallel Bibles. Such openly available language resources catalyse research, sharing, linking and tool development. 1.1. The Aim Scientific Bible editions and copyrighted translations are seldom freely shareable. Moreover, manually vali- dated annotations and fine-grained alignments are ex- pensive to produce and thus not typically freely avail- able for research and product development. The research problem of this paper is to overcome the obstacles to producing an openly shareable, manually aligned fine-grained alignment for a parallel Bible. 1.2. The Approach The paper describes a recently completed process to produce a fine-grained and open morpheme-alignment between the most important 20th century Finnish Bible translation and the most relevant source texts. The whole process covers the years 1991-2020 and con- sists of two active phases: (1) Product Development. The first phase (1991- 1997) added …

Content truncated to 500 words