Creating a Parallel Icelandic Dependency Treebank from Raw Text to Universal Dependencies

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 2924–2931 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 2924 Creating a Parallel Icelandic Dependency Treebank from Raw Text to Universal Dependencies Hildur Jónsdóttir, Anton Karl Ingason University of Iceland Sæmundargötu 2, 101 Reykjavík hildur.jonsdottir@gmail.com, antoni@hi.is Abstract Making the low-resource language, Icelandic, accessible and usable in Language Technology is a work in progress and is supported by the Icelandic government. Creating resources and suitable training data (e.g., a dependency treebank) is a fundamental part of that work. We describe work on a parallel Icelandic dependency treebank based on Universal Dependencies (UD). This is important because it is the first parallel treebank resource for the language and since several other languages already have a resource based on the same text. Two Icelandic treebanks based on phrase-structure grammar have been built and ongoing work aims to convert them to UD. Previously, limited work has been done on dependency grammar for Icelandic. The current project aims to ameliorate this situation by creating a small dependency treebank from scratch. Creating a treebank is a laborious task so the process was implemented in an accessible manner using freely available tools and resources. The parallel data in the UD project was chosen as a source because this would furthermore give us the first parallel treebank for Icelandic. The Icelandic parallel UD corpus will be published as part of UD version 2.6. Keywords: low-resource languages, parallel treebanks, Universal Dependencies, Icelandic 1. Introduction In order to survive the competition with a global English in various technology-associated domains, the Icelandic lan- guage, spoken by 350,000 people, must meet the challenges brought on by developments in Language Technology. Al- though it is not yet considered to be in imminent dan- ger (Rögnvaldsson et al., 2012b), a number of efforts are currently underway to address this situation. One of the core projects that the Icelandic government is supporting to achieve this is to build treebanks and especially depen- dency treebanks (Nikulásdóttir et al., 2017). In recent years, The Universal Dependencies project (Nivre et al., 2016) has been a leading force in parsing and cross-lingual research and becoming a part of it can make Icelandic Language Technology more viable. A treebank based on this type of an annotation scheme could become a foundation for further Icelandic parser development because treebanks are the essential training data for natural language data-driven parsers. The widespread interest that the UD project has received may also generate more interest in working on Icelandic Language Technology solutions in general once Icelandic UD resources are available. No dependency parser has yet been developed for Icelandic. However three phrase structure parsers are available. These are IceParser, a shallow phrase-structure parser which is a part of the IceNLP toolkit (Loftsson and Rögnvaldsson, 2007), Greynir, a rule-based parser based on context-free grammar (Þorsteinsson et al., 2019), and a parsing pipeline built on the IcePaHC treebank and the Berkeley Parser (Jökulsdóttir et al., 2019). A …

Content truncated to 500 words