Elamur Research⌕
☰
home / papers / 2026 / SEEM-CZ: Annotation and Classification of Epistemic Markers in Czech

SEEM-CZ: Annotation and Classification of Epistemic Markers in Czech

Summary

Czech adverbs such as pravděpodobně 'probably', určitě 'certainly' and zřejmě 'apparently' signal how sure a speaker is about a claim. They often also serve pragmatic purposes like hedging, assurance or persuasion, yet resources describing them systematically are scarce. The SEEM-CZ project at Charles University addresses this gap. The authors designed a detailed annotation scheme, intended to be reusable for other languages, and applied it manually to the 40 most frequent Czech epistemic markers. The material came from Czech–English parallel texts in InterCorp, annotated with the TEITOK tool. Each occurrence is labelled for:

  • type of use and degree of certainty
  • communicative function
  • contextual factors, such as scope, negation, predicate verb, sentence position and evidence type
  • its English translation equivalent

The resulting dataset underpins SEEMLex, the first lexicon of such expressions for Czech. To test its value for NLP, the team fine-tuned XLM-RoBERTa classifiers. Type of use proved learnable fairly well, while degree of certainty remained clearly harder. The resource supports both pragmatic linguistic research and automatic modelling of certainty in text.

Key points

  • The dataset covers nearly 4,000 manually annotated occurrences of the 40 most frequent Czech epistemic markers. Data come from InterCorp (Core part, Czech originals with English translations), processed in TEITOK.
  • Type of use has six classes: epistemic (with optional evidential and confirmatory subtypes), response, autosemantic and other. Certainty is rated on a 4-point scale: high, higher medium, medium and low/doubt.
  • Communicative function has six coarse values (assertive, directive, interrogative, commissive, (dis)approval, expressive) plus an open-ended specific function label.
  • XLM-RoBERTa with 10-fold cross-validation reached 72.6% accuracy on type of use (6 classes) and 54.2% on degree of certainty (4 classes).
  • The data and scripts are released at github.com/ufal/SEEM-CZ. The SEEMLex lexicon is announced for upcoming publication.

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

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/2ewzmb73r3h8

@inproceedings{stepankova2026seem,
  title = {{SEEM-CZ: Annotation and Classification of Epistemic Markers in Czech}},
  author = {Štěpánková, Barbora and Novák, Michal and Musil, Tomáš and Polakova, Lucie},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/2ewzmb73r3h8},
  url = {https://doi.org/10.63317/2ewzmb73r3h8}
}

References

46 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Ali Haif Abbas. 2022. Politicizing COVID-19 vaccines in the press: A critical discourse analysis. International Journal for the Semiotics of Law - Revue internationale de Sémiotique juridique, 35(3):1167–1185. find
  2. Karin Aijmer. 2002. Modality in advanced Swedish learners’ written interlanguage, pages 55–76. John Benjamins Publishing Company. find
  3. Karin Aijmer, Ad Foolen, and Anne-Marie Simon-Vandenbergen. 2006. Pragmatic markers in translation: a methodological proposal. In Approaches to Discourse Particles, pages 101–114. Brill. find
  4. Jean-Yves Antoine, Jeanne Villaneau, and Anaïs Lefeuvre. 2014. Weighted Krippendorff’s alpha is a more reliable metrics for multi-coders ordinal annotations: Experimental studies on emotion, opinion and coreference annotation. In Proceedings of the 14th Conference of the European Chapter of the Asso… find
  5. Dawn Archer, Jonathan Culpeper, and Matthew Davies. 2008. Pragmatic annotation. In Anke Lüdeling and Merja Kytö, editors, Corpus Linguistics: An International Handbook, pages 613–641. Mouton de Gruyter. find
  6. Marta Carretero and Juan Rafael Zamorano-Mansilla. 2019. Disentangling epistemic modality, neighbouring categories and pragmatic uses: the case of english epistemic modal adverbs. InQuinze études de cas sur les modalités linguistiques/Fifteen case studies on types of linguistic modalities, pages 131… find
  7. Jorge Carrillo-de Albornoz and Laura Plaza. 2013. An emotion-based model of negation, intensifiers, and modality for polarity and intensity classification. Journal of the American Society for Information Science and Technology, 64(8):1618–1633. find
  8. Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2020. Unsupervised cross-lingual representation learning at scale. In Proceedings of the 58th Annual Meeting of the Association fo… find
  9. František Čermák and Alexandr Rosen. 2012. The case of InterCorp, a multilingual parallel corpus. International Journal of Corpus Linguistics, 13(3):411–427. find
  10. Ferdinand De Haan. 2001. The Relation between Modality and Evidentiality. Linguistische Berichte, 9:201–216. find
  11. Liesbeth Degand and Jacqueline Evers-Vermeul. 2015. Grammaticalization or pragmaticalization of discourse markers? more than a terminological issue. Journal of Historical Pragmatics, 16(1):59–85. find
  12. Gabriele Diewald. 2011. Pragmaticalization (defined) as grammaticalization of discourse functions. Linguistics, 49(2):365–390. find
  13. Gabriele Diewald. 2013. "Same same but different" – Modal particles, discourse markers and the art (and purpose) of categorization, pages 19–46. John Benjamins Publishing Company. find
  14. Zi-Yi Dou and Graham Neubig. 2021. Word alignment by fine-tuning embeddings on parallel corpora. In Conference of the European Chapter of the Association for Computational Linguistics (EACL). find
  15. Bruce Fraser. 1996. Pragmatic markers. Pragmatics; Vol 6, No 2 (1996), 6. find
  16. Montserrat González, Paolo Roseano, Joan Borràs-Comes, and Pilar Prieto. 2017. Epistemic and evidential marking in discourse: Effects of register and debatability. Lingua, 186-187:68–87. Essays on evidentiality. find
  17. Miloslav Grepl and Radek Šimík. 2017. Epistemická modalita. In CzechEncy - Nový encyklopedický slovník češtiny. find
  18. Miroslav Grepl. 2017. Komunikační funkce výpovědi. In Petr Karlík, Marek Nekula, and Jana Pleskalová, editors, CzechEncy – Nový encyklopedický slovník češtiny. CzechEncy. Poslední přístup: 16. 10. 2025. find
  19. Maciej Grochowski, Anna Kisiel, and Magdalena Żabowska. 2014. Słownik gniazdowy par-tykuł polskich. Polska Akademia Umiejętności, Kraków. find
  20. Jan Hajič, Eduard Bejček, Jaroslava Hlaváčová, Marie Mikulová, Milan Straka, Jan Štěpánek, and Barbora Štěpánková. 2020. Prague Dependency Treebank - Consolidated 1.0. In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 5208–5218, Marseille, France. European Language Re… find
  21. Agnes Helbig, Gerhard Helbig. 1993. Lexikon deutscher Modalwörter, 2., durchges. aufl. edition. Langenscheidt, Leipzig , Berlin [u.a.]. find
  22. Leo Hoye. 2014. Adverbs and modality in English. Routledge. find
  23. Maarten Janssen. 2016. TEITOK: Text-faithful annotated corpora. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC’16), pages 4037–4043, Portorož, Slovenia. European Language Resources Association (ELRA). find
  24. Anu Koivunen, Anu Kanner, Maciej Janicki, Auli Harju, Jussi Hokkanen, and Essi Mäkelä. 2021. Emotive, evaluative, epistemic: A linguistic analysis of affectivity in news journalism. Journalism, 22(5):1190–1206. find
  25. Miroslav Komárek, Jan Kořenský, and Jan Petr. 1986. Mluvnice češtiny 2. Academia. find
  26. Klaus Krippendorff. 2013. Content Analysis: An Introduction to Its Methodology (third edition). Sage Publications, Thousand Oaks, CA. find
  27. Yang Liu, Xiaohui Yu, Zhongshuai Chen, and Bing Liu. 2013. Sentiment analysis of sentences with modalities. UnstructureNLP ’13, page 39–44, New York, NY, USA. Association for Computing Machinery. find
  28. Bolei Ma, Yuting Li, Wei Zhou, Ziwei Gong, Yang Janet Liu, Katja Jasinskaja, Annemarie Friedrich, Julia Hirschberg, Frauke Kreuter, and Barbara Plank. 2025. Pragmatics in the era of large language models: A survey on datasets, evaluation, opportunities and challenges. In Proceedings of the 63rd Annu… find
  29. Roser Morante and Caroline Sporleder. 2012. Modality and negation: An introduction to the special issue. Computational Linguistics, 38(2):223– 260. find
  30. Vítor Míguez. 2022. On epistemic modality and discourse strategy: Evidence from galician adverbs. Journal of Pragmatics, 201:32–42. find
  31. Jan Nuyts. 1999. Epistemic modality, language, and conceptualization. John Benjamins. find
  32. Juri Opitz, Shira Wein, and Nathan Schneider. 2025. Natural language processing relies on linguistics. Computational Linguistics, 51(3):1009–1032. find
  33. Frank Robert Palmer. 2001. Mood and Modality. Cambridge University Press. find
  34. Paola Pietrandrea. 2018. Epistemic constructions at work: A corpus study on spoken italian dialogues. Journal of Pragmatics, 129:171–191. find
  35. Lucie Poláková, Jana Šindlerová, and Barbora Štěpánková. 2025. No doubt, but... on connective functions of epistemic markers. Jazykovedný časopis / Journal of Linguistics, 76(1):75–84. find
  36. Agata Rozumko. 2016. Linguistic concepts across languages: The category of epistemic adverbs in english and polish. Yearbook of the Poznan Linguistic Meeting, 2. find
  37. Agata Rozumko. 2022. Textual functions of low confidence adverbs: The case of perhaps. Lingua, 268:103191. find
  38. Victoria L. Rubin. 2007. Stating with certainty or stating with doubt: Intercoder reliability results for manual annotation of epistemically modalized statements. In Human Language Technologies 2007: The Conference of the North American Chapter of the Association for Computational Linguistics; Compa… find
  39. Daisuke Suzuki. 2018. Variation between modal adverbs in british english: The cases of maybe and perhaps. Functions of Language, 25:392– 412. find
  40. Jana Šindlerová, Barbora Štěpánková, and Ingrid Andrén. 2023. Epistemická částice zřejmě pohle-dem paralelního korpusu. Korpus – gramatika – axiologie, (27):37–52. find
  41. Barbora Štěpánková, Lucie Poláková, Jana Šindlerová, and Michal Novák. 2024. What can dictionaries tell us about pragmatic markers – building the lexicon of epistemic and evidential markers in czech. In Lexicography and Semantics. Proceedings of the XXI EURALEX International Congress., pages 728–741… find
  42. Barbora Štěpánková, Jana Šindlerová, and Lucie Poláková. 2023. The epistemic marker určitě in the light of corpus data. Jazykovedný časopis / Journal of Linguistics, 74(1):130–139. find
  43. Mizuho Tamaji. 2009. What L2 processing strategy reveals about the prototypical relationship: A case of Japanese modal markers of possibility. In Proceedings of the 23rd Pacific Asia Conference on Language, Information and Computation, Volume 2, pages 540–548, Hong Kong. City University of Hong Kong… find
  44. Elizabeth Traugott. 1995. The role of discourse markers in a theory of grammaticalization. find
  45. Björn Wiemer and Katerina Stathi. 2010. The database of evidential markers in european languages: A bird’s eye view of the conception of the database. Language Typology and Universals, 63:275–289. 11. Language Resource References Klégr, Aleš and Kubánek, Michal and Malá, Markéta. Korpus InterCorp –… find
  46. Rosen, Alexandr and Vavřín, Martin and Zasina, Adrian. Korpus InterCorp – čeština. PID http://www.korpus.cz. ↗

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/stepankova-2026-seem-cz-annotation-classification-epistemic/