The Norwegian Dialect Corpus Treebank

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 4827–4832 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 4827 The Norwegian Dialect Corpus Treebank Andre K˚asen†, Kristin Hagen∗, Anders Nøklestad∗, Joel Priestley∗, Per Erik Solberg† and Dag Trygve Truslew Haug∗ †National Library of Norway {andre.kasen, per.solberg}@nb.no ∗Department of Linguistics and Scandinavian Studies {kristiha, noklesta, joeljp, daghaug}@uio.no Abstract This paper presents the NDC Treebank of spoken Norwegian dialects in the Bokm˚al variety of Norwegian. It consists of dialect recordings made between 2006 and 2012 which have been digitised, segmented, transcribed and subsequently annotated with morphological and syntactic analysis. The nature of the spoken data gives rise to various challenges both in segmentation and annotation. We follow earlier efforts for Norwegian, in particular the LIA Treebank of spoken dialects transcribed in the Nynorsk variety of Norwegian, in the annotation principles to ensure interusability of the resources. We have developed a spoken language parser on the basis of the annotated material and report on its accuracy both on a test set across the dialects and by holding out single dialects. Keywords: treebanks, spoken language, dialects, Norwegian, spoken language parser 1. Introduction In this article, we present the Norwegian Dialect Cor- pus Treebank – a treebank of spoken Norwegian di- alects transcribed in the Bokm˚al variety of Norwegian from The Nordic Dialect corpus (NDC; Johannessen et al. (2009)). The project has been carried out within the CLARINO+ project,1 a Norwegian project in the pan-European CLARIN infrastructure. In this chapter we will start to introduce the NDC Tree- bank and relate our work to the two other dependency treebanks for spoken and written Norwegian. In chap- ter 2 we will go in to different aspects of the mor- phosyntactic annotation in more detail and describe how the work was undertaken. In chapter 3 we report on experiments with training a spoken language parser on the resulting treebank. Chapter 4 describes the ac- cessibility of the treebank in the search interface Glossa and how the treebank can be downloaded. The NDC Treebank consists of 4587 speech segments, overall 66009 tokens, from 17 different Norwegian di- alects from south, west, east and north of Norway. The geographical distribution of the dialects is visualized in Figure 1 below. The recordings in The Nordic Di- alect Corpus were made between 2006 and 2012 and comprise both interviews and more informal conversa- tions between pairs of speakers. The transcriptions are therefore filled with spoken language phenomena, such as overlaps, pauses and various types of disfluencies. The NDC Treebank is annotated with morphological information and dependency-style syntactic analysis. The NDC Treebank project is related to the two other dependency treebanks made for Norwegian: The Nor- wegian Dependency Treebank (NDT; Solberg et al. (2014)) with mostly written texts and The LIA Tree- 1https://clarin.w.uib.no/about/ bank of Spoken Norwegian Dialects (Øvrelid et al., 2018). The NDC Treebank differs from the other spoken lan- guage treebank, LIA, in two major ways. First, the LIA Treebank is …

Content truncated to 500 words