Elamur Research⌕
☰
home / papers / 2026 / Harder than Finding the Lost Sheep? Towards Automatically Suggesting Deliberate Metaphor Annotations in German Sermons

Harder than Finding the Lost Sheep? Towards Automatically Suggesting Deliberate Metaphor Annotations in German Sermons

Summary

The paper examines how German-language sermons can be automatically scanned for deliberate metaphors. Deliberate metaphors are used with an intended communicative effect, unlike ordinary conventionalized ones. They are rare, and labeled German data is scarce. The goal is a pre-annotation tool for the CRC "Metaphors of Religion" that proposes candidate tokens for human annotators, so the authors optimize mainly for recall.

They frame the task as binary token-level sequence labeling and fine-tune several multilingual transformers: mBERT, XLM-RoBERTa, two sentence transformers and the longformer XLM-Long. They compare every combination of four training corpora (German sermons and TEDx, English Reddit and VUA) and context sizes from single sentences up to whole documents.

The best recall, .70 with a precision of only .10, came from multilingual SBERT trained on Reddit plus VUA. Adding in-domain religious data helped even though it was in English and followed other guidelines. Contrary to expectations about extended metaphors, sentence-level input worked better than larger contexts. The work matters as a first step toward assisted annotation of deliberate metaphors in a language other than English.

Key points

  • Datasets: Sermon (de), split into folds A and B for two-fold cross-validation; TEDx (de), annotated under the same guidelines; Reddit (en), Christian subreddits with DMIP labels; VUA (en), the VUAMC with DMIP deliberate-metaphor labels. All original labels were collapsed to binary.
  • Rarity: Metaphorical tokens are only 0.58–2.63% of all tokens across the corpora.
  • Best result: Multilingual SBERT trained on Reddit+VUA reached recall .70, precision .10 and F1 .17. Sentence transformers beat XLM-RoBERTa on recall and needed fewer training datasets.
  • Context size: Sentence-level input generally outperformed 50/100-token windows and whole documents. This is plausibly because metaphor density is highest at the sentence level.
  • Common false positives:
    • words in quotation marks;
    • coordinations and enumerations;
    • Bible quotations, which the guidelines exclude from annotation.

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 · History & humanities

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.latechclfl-1.16

@inproceedings{laarmannquante2026harder,
  title = {{Harder than Finding the Lost Sheep? Towards Automatically Suggesting Deliberate Metaphor Annotations in German Sermons}},
  author = {Laarmann-Quante, Ronja and Dipper, Stefanie},
  booktitle = {Proceedings of the 10th Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature 2026},
  year = {2026},
  doi = {10.18653/v1/2026.latechclfl-1.16},
  url = {https://doi.org/10.18653/v1/2026.latechclfl-1.16}
}

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. Maria Berger, Nieke Kiwitt, and Sebastian Reimann. 2024. Applying transfer learning to German metaphor prediction. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 1383–1392, Torino, Italia. ELRA and ICCL. find
  2. 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 for Computational Linguistics, pages 8440– 8451, Online. Association for Computational Linguistics. ↗
  3. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North, pages 4171–4186, Minneapolis, Minnesota. Association for Computational Linguistics. find
  4. Stefanie Dipper. 2025. Metaphorical heads and literal dependents: Syntactic properties of metaphors in German. In Proceedings of the 23th International Workshop on Treebanks and Linguistic Theories (TLT, SyntaxFest 2025), page 81–90, Ljubljana, Slovenia. find
  5. Stefanie Dipper, Adam Roussel, Alexandra Wiemann, Won Kim, and Tra-my Nguyen. 2024. Guidelines for the annotation of deliberate linguistic metaphor. In Proceedings of the 4th Workshop on Figurative Language Processing (FigLang 2024), pages 53–58, NAACL, Mexico City, Mexico (Hybrid). Association for Computational Linguistics. find
  6. Stefanie Dipper, Alexandra Wiemann, and Adam Roussel. 2025. Guidelines zur Annotation von deliberaten Metaphern. Metaphor Papers, 27. ↗
  7. Markus Egg and Valia Kordoni. 2022. Metaphor annotation for German. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 2556–2562, Marseille, France. European Language Resources Association. find
  8. Volkhard Krech, Tim Karis, and Frederik Elwert. 2024. Metaphors of religion: A conceptual framework. Metaphor Papers, 1. ↗
  9. George Lakoff and Mark Johnson. 1980. Metaphors We Live By. The University of Chicago Press, Chicago, London. find
  10. Chee Wee (Ben) Leong, Beata Beigman Klebanov, Chris Hamill, Egon Stemle, Rutuja Ubale, and Xianyang Chen. 2020. A report on the 2020 VUA and TOEFL Metaphor Detection Shared Task. In Proceedings of the Second Workshop on Figurative Language Processing, pages 18–29, Online. Association for Computational Linguistics. find
  11. Chee Wee (Ben) Leong, Beata Beigman Klebanov, and Ekaterina Shutova. 2018. A report on the 2018 VUA Metaphor Detection Shared Task. In Proceedings of the Workshop on Figurative Language Processing, pages 56–66, New Orleans, Louisiana. Association for Computational Linguistics. find
  12. Xiaofei Lu and Ben Pin-Yun Wang. 2017. Towards a metaphor-annotated corpus of Mandarin Chinese. Language Resources and Evaluation, 51(3):663–694. ↗
  13. Arthur Neidlein, Philip Wiesenbach, and Katja Markert. 2020. An analysis of language models for metaphor recognition. In Proceedings of the 28th International Conference on Computational Linguistics, pages 3722–3736, Barcelona, Spain (Online). International Committee on Computational Linguistics. find
  14. Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, Jake Vanderplas, Alexandre Passos, David Cournapeau, Matthieu Brucher, Matthieu Perrot, and Édouard Duchesnay. 2011. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12:2825–2830. ↗
  15. W. Gudrun Reijnierse, Christian Burgers, Tina Krennmayr, and Gerard J. Steen. 2018. DMIP: A method for identifying potentially deliberate metaphor in language use. Corpus Pragmatics, 2(2):129–147. ↗
  16. W. Gudrun Reijnierse, Christian Burgers, Tina Krennmayr, and Gerard J. Steen. 2019. Metaphor in communication: the distribution of potentially deliberate metaphor across register and word class. Corpora, 14(3):301–326. ↗
  17. Sebastian Reimann and Tatjana Scheffler. 2024. Metaphors in online religious communication: A detailed dataset and cross-genre metaphor detection. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 11236–11246, Torino, Italia. ELRA and ICCL. find
  18. Nils Reimers and Iryna Gurevych. 2019. Sentence-BERT: Sentence embeddings using siamese BERT-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics. find
  19. Nils Reimers and Iryna Gurevych. 2020. Making monolingual sentence embeddings multilingual using knowledge distillation. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing. Association for Computational Linguistics. find our card
  20. Elisa Sanchez-Bayona and Rodrigo Agerri. 2022. Leveraging a new Spanish corpus for multilingual and cross-lingual metaphor detection. In Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL), pages 228–240, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics. find
  21. Gerard J. Steen. 2011. From three dimensions to five steps: the value of deliberate metaphor. metaphorik.de, 21. find
  22. Gerard J. Steen. 2023. Thinking by metaphor, fast and slow: Deliberate Metaphor Theory offers a new model for metaphor and its comprehension. Frontiers in Psychology, 14. ↗
  23. Gerard J. Steen, Aletta G. Dorst, J. Berenike Herrmann, Anna A. Kaal, Tina Krennmayr, and Trijntje Pasma. 2010. A Method for Linguistic Metaphor Identification. From MIP to MIPVU. Number 14 in Converging Evidence in Language and Communication Research. John Benjamins, Amsterdam. ↗
  24. Lennart Wachowiak, Dagmar Gromann, and Chao Xu. 2022. Drum up SUPPORT: Systematic analysis of image-schematic conceptual metaphors. In Proceedings of the 3rd Workshop on Figurative Language Processing (FLP), pages 44–53, Abu Dhabi, United Arab Emirates (Hybrid). Association for Computational Linguistics. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/quante-2026-harder-than-finding-lost-sheep/