Elamur Research⌕
☰
home / papers / 2010 / The Leeds Arabic Discourse Treebank: Annotating Discourse Connectives for Arabic

The Leeds Arabic Discourse Treebank: Annotating Discourse Connectives for Arabic

Summary

Before this work, Arabic corpora had morphological and syntactic annotation but no discourse layer. Earlier Arabic studies of connectives were small, mixed discourse connectives with ordinary prepositions, and measured no agreement. The authors built the Leeds Arabic Discourse Treebank (LADTB) on top of the Penn Arabic Treebank. Following the theory-neutral PDTB approach, they mark each explicit connective, the relation it signals and its two arguments (Arg1/Arg2).

Several adaptations reflect Arabic structure. Clitics count as connectives, prepositions governing nominalizations (Al-Mazdar) are included, and certain noun sentences can serve as arguments. Two annotators worked on 537 news texts using a dedicated annotation tool.

Agreement on finding connectives was very high. Agreement on relation labels was moderate overall, mainly because news texts conventionally begin sentences with و (w/and), often without any real discourse function. When that connective is excluded, relation agreement becomes good. The result is the first principled, reliability-tested discourse resource for Arabic, intended for training connective and relation classifiers.

Key points

  • Connective lexicon: 91 basic connectives plus 16 modified forms gives 107 types, comparable to about 100 in the PDTB. Only 68 of these appeared in the final corpus.
  • Relation scheme: the 4 PDTB top-level classes are kept, with 18 subclasses. Two subclasses are new, EXPANSION.Background and COMPARISON.Similarity. Annotators also used 22 combinations of multiple relations.
  • Connective identification (Task I): over 23,331 candidate tokens, observed agreement was 0.95 with kappa 0.88. Only about 5,586 of the candidates were true connectives, largely because of ambiguous clitics.
  • Relation assignment (Task II): agreement was kappa 0.57 and alpha 0.58 overall. Without و it rose to 0.80 observed, kappa 0.77 and alpha 0.80.
  • Corpus size: after a 121-text pilot, the final LADTB covers 5,651 connectives in 537 files, about 75% of ATB part 1. Disagreed paragraph-initial و instances were set to Expansion.Conjunction by default.

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: Discourse analysis · Treebanks

Cited by

31 papers of our library cite it (of the 140673 whose bibliographies we read); 14 of them have a page here, 17 not yet.

Cite

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

@inproceedings{alsaif2010leeds,
  title = {{The Leeds Arabic Discourse Treebank: Annotating Discourse Connectives for Arabic}},
  author = {Al-Saif, Amal and Markert, Katja},
  booktitle = {Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC'10)},
  year = {2010},
  url = {https://aclanthology.org/L10-1332/}
}

References

14 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Amal Al-Saif, Katja Markert, and Hussein Abdul-Raof. 2009. Corpus-Based Study: Extensive Collection of Discourse Connectives For Arabic. In Proceedings of The Saudi International Conference 2009 (SIC09), Surrey, UK. find
  2. W. Al-Sanie, A. Touir, and H. Mathkour. 2005. Towards a rhetorical parsing of Arabic text. In The International Conference on Intelligent Agents, Web Technology and Internet Commerce (IAWTIC05). find
  3. N.N. Al-Warraki and A.T. Hasanayn. 1994. The connectors in modern standard Arabic. American University in Cairo Press. find
  4. I.H. Alansari. 1985. Mogny Allabib. Dar Alfekur, Beirut. find
  5. H. Alfarabi. 1990. Ketab AlHorof. Dar AlMashreg, Lebnan. find
  6. R. Artstein and M. Poesio. 2008. Inter-coder agreement for computational linguistics (survey article). Computational Limnguistics, 34(4):555–596. ↗
  7. K. Dukes and N. Habas. 2010. Morphological annotation of quranic arabic. In International Conference on Language Resources and Evaluation (LREC 2010). find our card
  8. J. Hajic, O. Smrz, P. Zemánek, J. Šnaidauf, and E. Beška. 2004. Prague Arabic dependency treebank: Development in data and tools. In Proc. of the NEMLAR Intern. Conf. on Arabic Language Resources and Tools, pages 110–117. Citeseer. find
  9. M.A.K. Halliday and R. Hasan. 1976. Cohesion in English. Longman London. find
  10. J.R. Hobbs. 1985. On the coherence and structure of discourse. Center for the Study of Language and Information, Stanford, Calif. find
  11. E.H. Hovy. 1993. Automated discourse generation using discourse structure relations. Artificial intelligence, 63(1-2):341–385. ↗
  12. A. Knott and T. Sanders. 1998. The classification of coherence relations and their linguistic markers: An exploration of two languages. Journal of Pragmatics, 30(2):135–175. ↗
  13. M. Maamouri and A. Bies. 2004. Developing an Arabic treebank: Methods, guidelines, procedures, and tools. In Proceedings of the Workshop on Computational Approaches to Arabic Script-based Languages (COLING), Geneva. find our card
  14. W.C. Mann and S.A. Thompson. 1988. Rhetorical structure theory: Toward a functional theory of text organization. Text, 8(3):243–281. Number of files Connective types Discourse relation types Potential connective tokens Real discourse connectives Most frequent connective Least frequent connectives ↗ our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/saif-2010-leeds-arabic-discourse-treebank-annotating/