QUD-Based Annotation of Discourse Structure and Information Structure: Tool and Evaluation

What it is about

The paper tests the QUD-tree method of Riester and colleagues: a text is cut into atomic assertions (sentences, plus splits at coordinations and before optional adjuncts), each assertion is the answer to one explicit or implicit question under discussion, and the questions form the inner nodes of a tree whose leaves are the assertions in text order. Implicit QUDs are constrained by three principles: the QUD must be answerable by the assertion it dominates (congruence), may contain only given or salient material (givenness), and should contain as much given material as possible (maximal anaphoricity); parallel assertions share a QUD whose background is their common material. From the QUD, information structure follows: focus = the part answering the QUD, background = material in the QUD, contrastive topic, and non-at-issue material (e.g. «It was clear from the President's speech that…»). Two pairs of annotators worked on an English and a German interview with a new tool (TreeAnno): agreement on tree structure is moderate (kappa .45–.53, computed on the matrix of all segment spans after Marcu et al.), on information-structure labels substantial (.67–.69 overall; focus .63–.67).

Ideas

• I1. Derive focus/background labels deterministically from Mova's QUD stack: once world/src/qud.rs attaches an assertion to a question, the words matching the question's content lemmas are background and the constituent filling the wh-slot is focus. Store the focus on the frame; the school student can then prefer, among remembered frames, the one whose focus fills the asked role. Measure on courses with world knowledge diagnose (step 5, right option outscored).
• I2. Implement maximal anaphoricity as the tie-breaker among candidate implicit QUDs in qud.rs: prefer the candidate with the most given lemmas that still passes congruence and givenness, and log the rejected ones with the violated principle, as the paper does for its examples.
• I3. Mark non-at-issue material: matrix clauses of evidence and attitude («it was clear from X that», «as everyone knows», appositives) are not what an assertion answers. In prag/src/frames.rs flag the frame of such a matrix as non-at-issue so that answer scoring weights the embedded frame. Count course questions whose answer frame sits under such a matrix.
• I4. Score QUD trees with the paper's span-matrix kappa (one cell per segment span: does a QUD span it?) when comparing qud.rs output with a teacher-made QUD tree on a few tales. It is cheap, needs no alignment of question wordings, and has a known human reference (about .5).

For Mova

29. Ellipsis, word order and information structure (focus, background, contrastive topic) and 36-discourse-relations.md §9 and §11.5 (QUD stack). Code: world/src/qud.rs, prag/src/frames.rs, world/src/knowledge.rs. Measures: span-matrix kappa against teacher trees; course answers after focus-aware scoring.

Topics: Discourse analysis

Cite

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

@inproceedings{kuthy2018qud,
  title = {{QUD-Based Annotation of Discourse Structure and Information Structure: Tool and Evaluation}},
  author = {Kuthy, Kordula De and Reiter, Nils and Riester, Arndt},
  booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)},
  year = {2018},
  url = {https://aclanthology.org/L18-1304/}
}

References

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

1. Asher, N. and Lascarides, A. (2003). Logics of Conversation. Cambridge University Press. find
2. Beaver, D. and Clark, B. (2008). Sense and Sensitivity. How Focus Determines Meaning. Wiley-Blackwell, Oxford. find
3. Büring, D. (2003). On D-trees, beans, and B-accents. Linguistics and Philosophy, 26(5):511–545. find
4. Büring, D. (2008). What’s new (and what’s given) in the theory of focus? In Proceedings of the 34 th Annual Meeting of the Berkeley Linguistics Society, pages 403– 424. find
5. Büring, D. (2016). Intonation and Meaning. Oxford University Press. find
6. Calhoun, S., Carletta, J., Brenier, J., Mayo, N., Jurafsky, D., Steedman, M., and Beaver, D. (2010). The NXT-format Switchboard corpus: A rich resource for investigating the syntax, semantics, pragmatics and prosody of dialogue. Language Resources and Evaluation, 44:387– 419. find
7. Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20(1):37–46. find
8. Krifka, M. (2008). Basic notions of information structure. Acta Linguistica Hungarica, 55(3-4):243–276. find
9. Kruijff-Korbayová, I. and Steedman, M. (2003). Discourse and information structure. Journal of Logic, Language and Information, 12:249–259. find
10. Marcu, D., Amorrortu, E., and Romera, M. (1999). Experiments in constructing a corpus of discourse trees. In Proceedings of the ACL Workshop on Standards and Tools for Discourse Tagging, pages 48–57, College Park, MD, USA. find
11. Onea, E. (2016). Potential Questions at the Semantics-Pragmatics Interface. Brill, Leiden. find
12. Poláková, L., Mı́rovský, J., Nedoluzhko, A., Jı́nová, P., Zikánová, Š., and Hajičová, E. (2013). Introducing the Prague Discourse Treebank 1.0. In Proceedings of the International Joint Conference on Natural Language Processing, pages 91–99, Nagoya. find
13. Polanyi, L., van den Berg, M., and Ahn, D. (2003). Discourse structure and sentential information structure: An initial proposal. Journal of Logic, Language and Information, 12:337–350. find
14. Polanyi, L. (1988). A formal model of the structure of discourse. Journal of Pragmatics, 12(5-6):601–638. find
15. Reyle, U. and Riester, A. (2016). Joint information structure and discourse structure analysis in an Underspecified DRT framework. In Julie Hunter, et al., editors, Proceedings of the 20 th Workshop on the Semantics and Pragmatics of Dialogue (JerSem), pages 15–24, New Brunswick, NJ, USA. find
18. Ritz, J., Dipper, S., and Götze, M. (2008). Annotation of information structure: An evaluation across different types of texts. In Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC), pages 2137–2142, Marrakesh. find
19. Roberts, C. (2012). Information structure in discourse: Towards an integrated formal theory of pragmatics. Semantics and Pragmatics, 5(6):1–69. find our card
20. Rooth, M. (1992). A theory of focus interpretation. Natural Language Semantics, 1(1):75–116. find
21. Schwarzschild, R. (1999). GIVENness, AvoidF, and other constraints on the placement of accent. Natural Language Semantics, 7(2):141–177. find
22. Schweitzer, K., Eckart, K., Gärtner, M., Faleńska, A., Riester, A., Rösiger, I., Schweitzer, A., Stehwien, S., and Kuhn, J. (2018). German Radio Interviews: The GRAIN release of the SFB732 silver standard collection. In Proceedings of the Eleventh Language Resources and Evaluation Conference (LREC), Miyazaki, JP. find
23. Stutterheim, C. v. and Klein, W. (1989). Referential movement in descriptive and narrative discourse. In Rainer Dietrich et al., editors, Language Processing in Social Context, pages 39–76. North Holland, Amsterdam. find
24. van Kuppevelt, J. (1995). Discourse structure, topicality and questioning. Journal of Linguistics, 31:109–149. find
25. Velleman, L. and Beaver, D. (2016). Question-based models of information structure. In Caroline Féry et al., editors, The Oxford Handbook of Information Structure, pages 86–107. Oxford University Press. find
26. Yimam, S. M., Gurevych, I., Eckart de Castilho, R., and Biemann, C. (2013). WebAnno: A flexible, web-based and visually supported system for distributed annotations. In Proceedings of the 51 st Annual Meeting of the Association for Computational Linguistics: System Demonstrations, pages 1–6, Sofia. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/kuthy-2018-qud-based-annotation-discourse-structure/