A Universal Dependencies Treebank of Ancient Hebrew

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 2353–2361 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 2353 A Universal Dependencies Treebank of Ancient Hebrew Daniel G. Swanson, Francis M. Tyers Department of Linguistics, Indiana University, {dangswan,ftyers}@iu.edu Abstract In this paper we present the initial construction of a Universal Dependencies treebank with morphological annotations of Ancient Hebrew containing portions of the Hebrew Scriptures (1579 sentences, 27K tokens) for use in comparative study with ancient translations and for analysis of the development of Hebrew syntax. We construct this treebank by applying a rule-based parser (300 rules) to an existing morphologically-annotated corpus with minimal constituency structure and manually verifying the output and present the results of this semi-automated annotation process and some of the annotation decisions made in the process of applying the UD guidelines to a new language. Keywords: treebank, UD, Hebrew 1. Introduction The Hebrew Scriptures are a collection of 39 books primarily written in the first millennium BC in An- cient Hebrew (with a few passages in Aramaic) which were arranged and codified in their current form over the course of the first millennium AD. They are also known as the Tanakh, an acronym of the Hebrew names of the 3 main divisions: �הרת/torah/ “law”1, ��יאבנ /nevi’im/ “prophets” (a category which also includes several books of narrative history), and ��יבתכ/ketuvim/ “writings”. The Universal Dependencies (UD) project (Nivre et al., 2020) is a collaborative effort to create a collection of treebanks in a single cross-linguistically consistent an- notation scheme so as to better facilitate studying syn- tax in multiple languages. In this paper we present a UD treebank containing the books of Genesis and Ruth with the intent that, like other ancient language treebanks such as PROIEL (Haug and Jøhndal, 2008; Eckhoff et al., 2018), it can serve as a resource for studying both the Tanakh as a document and also the development of Hebrew syn- tax through the centuries. In addition, we chose to begin with these books in part because they can be compared with the Peshitta, a Coptic translation of the Bible which is partially available in UD (Zeldes and Abrams, 2018). Section 2 describes the text and morphological anno- tations used, Section 3 presents how the treebank was created, Section 4 discusses some of the more challeng- ing constructions to annotate, Section 5 provides statis- tics about treebank produced, and Section 6 concludes. 2. The Hebrew Corpus The Biblia Hebraica Stuttgartensia Amstelodamensis (BHSA) is a complete copy of the text of the Tanakh 1Instances of Hebrew script in this paper are followed by a transliteration in slashes according to the ALA-LC scheme (Barry, 1997) and an English translation in quotes. with morphological annotations which is maintained by the Eep Talstra Centre for Bible and Computer (Peursen et al., 2015). The corpus is stored as a table, where each row is a syntactic node, whether a word, phrase, clause, or sen- tence. The columns of the table …

Content truncated to 500 words