Annotating Discourse Connectives in Spoken Turkish

Summary

This paper reports the first effort to bring PDTB/TDB-style discourse annotation, previously applied only to written Turkish, to spoken Turkish. The authors took the 77 search tokens defined for the Turkish Discourse Bank (TDB) and marked explicit discourse connectives, along with their arguments, in the demo release of the Spoken Turkish Corpus (STC). Following TDB practice, they also annotated phrasal expressions such as onun için. To avoid building new software, they loaded the STC transcripts into the existing DATT tool and gave annotators the audio as separate files. This meant the corpus's time alignment went unused.

The authors identify several difficulties specific to speech:

pronouns that may refer either to a proposition or to a noun phrase
connectives that pick up a first argument from far back in a dialogue, including a previously unseen interrogative use of için
arguments that are cut off, inaudible or overlapping

The annotation yielded 416 relations. Frequencies of the four most common connectives, normalized per 1000 words, showed no statistically significant difference from the TDB. This provides a baseline resource for studying discourse structure in spoken Turkish.

Key points

Data: STC demo has about 20,000 words in 23 recordings (2 h 27 min). Twenty are casual conversations and service encounters (2 h 1 min); three are news-commentary broadcasts (26 min).
Tools and baseline: Annotation used DATT (Discourse Annotation Tool for Turkish). The comparison resource, TDB, is built on a 400,000-word subcorpus of the METU Turkish Corpus (MTC).
Coverage: ve, için, ama and sonra account for 217 of 416 STC relations (52.2%), versus 4951 of 8484 TDB annotations (58.3%). Ama is notably more frequent in speech (4.61 vs. 2.57 per 1000 words as a connective).
Statistics: Chi-square tests found differences in frequency and in the rate of discourse use to be non-significant. The paper reports this as p>0.5. The authors partly attribute the similarity to the TDB containing dialogues and interviews, and to the STC containing scripted broadcasts.
Future work: The authors plan to merge the discourse layer with the STC's time-aligned annotation, add a second double-blind annotation set, and compute inter-annotator agreement.

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

Cited by

4 papers of our library cite it (of the 140673 whose bibliographies we read); 2 of them have a page here, 2 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/v1/w14-4916

@inproceedings{demirsahin2014annotating,
  title = {{Annotating Discourse Connectives in Spoken Turkish}},
  author = {Demirşahin, Isin and Zeyrek, Deniz},
  booktitle = {Proceedings of LAW VIII - The 8th Linguistic Annotation Workshop},
  year = {2014},
  doi = {10.3115/v1/w14-4916},
  url = {https://doi.org/10.3115/v1/w14-4916}
}

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.

Aktaş, B., Bozsahin, C., & Zeyrek, D. 2010. Discourse relation configurations in Turkish and an annotation environment. Proceedings of the Fourth Linguistic Annotation Workshop, pp. 202-206. find
Asher, N. 1993. Reference to Abstract Objects in Discourse. Kluwer Academic Publishers. find
Demirşahin, I., Sevdik-Çallı, A., Ögel Balaban, H., Çakıcı, R. & Zeyrek, D. 2012. Turkish Discourse Bank: Ongoing Developments. Proceedings of LREC 2012 The First Turkic Languages Workshop. find
Forbes-Riley, K., Webber, B., & Joshi, A. 2006. Computing discourse semantics: The predicate-argument semantics of discourse connectives in D-LTAG. Journal of Semantics, 23(1), 55–106. find
Prasad, R., Dinesh, N., Lee, A., Miltsakaki, E., Robaldo, L., Joshi, A., and Webber, B. 2008. The Penn Discourse Treebank 2.0. In Proceedings of the Sixth International Conference on Language Resources and Evaluation. (LREC’08). find our card
Ruhi, Şükriye, Çokal Karadaş, Derya. 2009. Features for an internet accessible corpus of spoken Turkish discourse. Working Papers in Corpus-based Linguistics and Language Education 3, 311-320. find
Ruhi, Şükriye, Eröz Tuğa, Betil, Hatipoğlu, Çiler, Işık-Güler, Hale, Can, Hümeyra, Karakaş, Özlem, Acar, Güneş, Eryılmaz, Kerem (2010). Türkçe için genel amaçlı sözlü derlem oluşturmada veribilgisi, çeviriyazı ölçünleştirmesi ve derlem yönetimi. XXIV. Dilbilim Kurultayı, 17-18 Mayıs, 2010, Yuvarlak… find
Say, B., Zeyrek, D., Oflazer, K., and Özge, U. 2002. Development of a Corpus and a Treebank for Present-day Written Turkish. In Proceedings of the Eleventh International Conference on Turkish Linguistics (ICTL 2002). find
Zeyrek, D., and Webber, B. (2008). A Discourse Resource for Turkish: Annotating Discourse Connectives in the METU Turkish Corpus. In Proceedings of the 6thWorkshop on Asian Language Resources, The Third International Joint Conference on Natural Language Processing (IJNLP). find
Zeyrek, D., Demirşahin, I., Sevdik-Çallı, A. B., Çakıcı, Ruket. 2013. Turkish Discourse Bank: Porting a discourse annotation style to a morphologically rich language. Discourse and Dialogue 4 (3), 174-184. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ahin-2014-annotating-discourse-connectives-spoken-turkish/

📄 Annotating Discourse Connectives in Spoken Turkish