Elamur Research⌕
☰
home / papers / 2006 / Talbanken05: A Swedish Treebank with Phrase Structure and Dependency Annotation

Talbanken05: A Swedish Treebank with Phrase Structure and Dependency Annotation

Summary

For a long time Swedish had no large publicly available treebank, even though one of the very first syntactically annotated corpora, the Talbanken of the 1970s, was Swedish. Its annotation, following the MAMBA scheme, consists of flat part-of-speech and grammatical-function tags attached to each token. Modern parsers and tools handle this format poorly. The authors wrote a program that automatically converts this corpus (referred to in the paper as Talbanken76) into Talbanken05. The conversion has four stages. First, an unlabeled constituent tree is built from the multi-layered tags. Then manually ordered rules assign categories (S, NP, PP, etc.) to the nodes. Next, the tree is deepened with additional nodes. Finally, head-finding rules yield a dependency structure in which the arcs carry MAMBA function labels. The result is three consistent versions of the annotation in XML (TIGER-XML and Malt-XML). The authors did not carry out a large-scale evaluation of the conversion, but preliminary checks indicate that it is reliable. The resource is free for research and teaching. It provides Swedish language technology with a large treebank for training and evaluating parsers.

Key points

  • The source corpus contains nearly 300 thousand tokens. Roughly half consists of written texts: professional prose (P) and essays by upper-secondary school students (G). The other half is spoken language: interviews (IB), conversations and debates (SD).
  • Talbanken05 is available in three new formats. Two versions of the constituent trees with grammatical functions, flat and deepened, are encoded in TIGER-XML following the model of the German TIGER Treebank and allow discontinuous constituents. The dependency version is encoded in Malt-XML for MaltParser and allows non-projective trees.
  • Node categories are assigned by ordered rules in the form of quadruples (C, P, L, N). A rule takes into account the functions of the child nodes, the function of the node itself, and the lexical categories of the children, while N is the label it assigns.
  • The head is determined as follows. In coordinate constructions, the leftmost conjunct becomes the head. At the clause level there is no explicit head, since MAMBA provides a topological analysis there, so a priority list of functions applies: HD, FV, IV, SP, then subjects and objects, and so on down to punctuation. Ties are resolved by left-to-right order.
  • The resource is freely distributed for research and education, provided the authors are properly cited.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics Treebanks Syntax & grammar formalisms

Cite

Cite as Joakim Nivre, Jens Nilsson and Johan Hall. 2006. Talbanken05: A Swedish Treebank with Phrase Structure and Dependency Annotation. Proceedings of the Fifth International Conference on Language Resources and Evaluation (LREC'06).

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{nivre2006talbanken05,
  title = {{Talbanken05: A Swedish Treebank with Phrase Structure and Dependency Annotation}},
  author = {Joakim Nivre, Jens Nilsson and Johan Hall},
  booktitle = {Proceedings of the Fifth International Conference on Language Resources and Evaluation (LREC'06)},
  year = {2006},
  doi = {10.63317/25vs6zxb65s7},
  url = {https://doi.org/10.63317/25vs6zxb65s7}
}

References

10 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. S. Brants, S. Dipper, S. Hansen, W. Lezius, and G. Smith. 2002. TIGER treebank. In Erhard Hinrichs and Kiril Simov, editors, Proceedings of the First Workshop on Treebanks and Linguistic Theories (TLT), pages 24–42. find
  2. Michael Collins. 1996. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Annual Meeting of the Association for Computational Linguistics (ACL), pages 184–191. find
  3. Jan Einarsson. 1976a. Talbankens skriftspråkskonkor-dans. Lund University, Department of Scandinavian Languages. find
  4. Jan Einarsson. 1976b. Talbankens talspråkskonkordans. Lund University, Department of Scandinavian Languages. find
  5. A. Gelbukh, H. Calvo, and S. Torres. 2005. Transforming a constituency treebank into a dependency treebank. In Procesamiento Lenguaje Natural, Spain. find
  6. Jan Hajič, Barbora Vidova Hladka, Jarmila Panevová, Eva Hajičová, Petr Sgall, and Petr Pajas. 2001. Prague Dependency Treebank 1.0. LDC, 2001T10. find
  7. Jerker Järborg. 1986. Manual för syntaggning. Technical report, Göteborg University, Department of Swedish. find
  8. David M. Magerman. 1995. Statistical decision-tree models for parsing. In Proceedings of the 33rd Annual Meeting of the Association for Computational Linguistics (ACL), pages 276–283. our card
  9. Joakim Nivre and Johan Hall. 2005. MaltParser: A language-independent system for data-driven dependency parsing. In Proceedings of the Fourth Workshop on Treebanks and Linguistic Theories (TLT). find
  10. Ulf Teleman. 1974. Manual för grammatisk beskrivning av talad och skriven svenska. Studentlitteratur. find

This page in Markdown → https://md.research.elamur.ai/free/en/papers/nivre-2006-talbanken05-swedish-treebank-phrase-structure/