Elamur Research⌕
☰
home / papers / 2026 / Meaning-Making Process and Error Dynamics in ChatGPT-Mediated Translation

Meaning-Making Process and Error Dynamics in ChatGPT-Mediated Translation

Summary

The study looks at how translation trainees handle LLM output when post-editing a specialised text, using German–Spanish, a pair largely missing from MTPE research and benchmarks. Benchmarks such as WMT APE, LangMark and DivEMT are dominated by English-centred pairs. Twenty students post-edited a ChatGPT-4.5 translation of a short German economic article on inflation into Spanish. The authors annotated the resulting errors along two dimensions: where each error came from (the model or the student's own revision) and which linguistic category it belonged to.

Half of the problems trace back to the model. A sizeable share, however, were created by students rewriting segments that were already fine. Terminology was by far the most error-prone area. The authors argue that ChatGPT's smooth, natural-sounding Spanish lowers trainees' vigilance toward meaning and domain inaccuracies, while also inviting needless reformulation.

Comparing these results with an earlier DeepL-based study, they propose that trainees check adequacy first and polish form afterwards. They also call for training in LLM literacy, explicit decision thresholds for editing, and genre-aware use of glossaries.

Key points

  • Data: 132 annotated instances from a 127-word source text post-edited by 20 students. Only 12 students produced annotated errors, and 8 had none.
  • By origin: 50.8% of instances are ChatGPT-related, 21.2% were introduced by students through over-editing, and 28.0% are acceptable preferential changes.
  • By category: terminology is the riskiest at 34.1%, followed by tense/aspect at 15.2% and style at 13.6%.
  • Recommended practice: a two-pass QA protocol covering terminology, tense/aspect and meaning transfer first, then orthotypography, punctuation and style.
  • Limitations: a single text, a single LLM (ChatGPT-4.5), and no process data. Future work should add keystroke logs or screen recordings and test other LLM and NMT systems.

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 · Machine translation

Cite

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

@inproceedings{mihalache2026meaning,
  title = {{Meaning-Making Process and Error Dynamics in ChatGPT-Mediated Translation}},
  author = {Mihalache, Iulia and Salinas, María-José Varela},
  booktitle = {Proceedings of the 26th Annual Conference of the European Association for Machine Translation (Volume 1)},
  year = {2026},
  url = {https://aclanthology.org/2026.eamt-1.40/}
}

References

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

  1. Angelone, E. 2010. Uncertainty, uncertainty management and metacognitive problem solving in the translation task. In G. M. Shreve & E. Angelone (Eds.), Translation and Cognition. John Benjamins, 17–40. ↗
  2. Baumgarten, Stefan (2025). Welcome to the Machine? Prolegomena zu einer kritischen Translationswissenschaft in Zeiten ungebremster Maschinisierung. Yearbook of Translational Hermeneutics 5.1, 59-85.Bohm, David. 1980/2002. Wholeness and the Implicate Order. Ark. find
  3. Bhattacharyya, P., Chatterjee, R., Freitag, M., Kanojia, D., Negri, M., & Turchi, M. (2022). Findings of the WMT 2022 Shared Task on Automatic Post-Editing. In Proceedings of the Seventh Conference on Machine Translation (WMT) (pp. 109–117). Association for Computational Linguistics. https://aclan-thology.org/2022.wmt-1.5 ↗
  4. Beninatto, R., and L. K. Varga (2024). The Language Industry Curve. Nimdzi Insights. https://www.nim-dzi.com/the-language-industry-curve/. ↗
  5. Brynjolfsson, Erik, Danielle Li, and Lindsey Raymond. 2025. Generative AI at work. The Quarterly Journal of Economics, 140(2):889–942. ↗
  6. Castillo Bernal, P. (2022). La traducción literaria asistida por ordenador aplicada a la novela histórica (alemán-español): entrenamiento y comparación de sistemas de traducción automática. Quaderns de Filologia: Estudis Lingüístics, 27, 41–58. https://doi.org/10.7203/qf.0.24624 ↗
  7. Chatterjee, R., de Souza, J. G. C., Negri, M., & Turchi, M. (2016). The FBK Participation in the WMT 2016 Automatic Post-editing Shared Task. In Proceedings of the First Conference on Machine Translation (Vol. 2, pp. 745–750). Association for Computational Linguistics. https://doi.org/10.18653/v1/W16-2377 ↗
  8. Chatterjee, R., Federmann, C., Negri, M., & Turchi, M. (2019). Findings of the WMT 2019 Shared Task on Automatic Post-Editing. In Proceedings of the Fourth Conference on Machine Translation (Vol. 3, pp. 11–28). Association for Computational Linguistics. https://doi.org/10.18653/v1/W19-5402 ↗ our card
  9. Chen, S., & Zhou, T. (2025). The scale-integration paradox: A contrastive bibliometric analysis of global and Chinese machine translation post-editing research. Cogent Arts & Humanities, 12(1), Article 2584420. https://doi.org/10.1080/23311983.2025.2584420 ↗
  10. Ginovart Cid, C., & Oliver, A. (2020). The post-editor’s skill set according to industry, trainers and linguists. Portsiel J, editor. Maschinelle Übersetzung für Übersetzungsprofis. Berlin: BDÜ Weiterbildungs-und Fachverlagsgesellschaft mbh; 2020. Online at: https://repositori.upf.edu/bits-tream/handle/10230/46285/ginovart_bdu_postedi-tor.pdf?sequence=1&isAllowed=y. ↗
  11. Guerberof Arenas, A. (2020). Pre-editing and post-editing. In E. Angelone, M. Ehrensberger-Dow, & G. Massey (Eds.), The Bloomsbury Companion to Language Industry Studies, London/New York/Oxford/New Delhi/Sydney, Bloomsbury Academic, p. 333–360. ↗
  12. Guerberof Arenas, A., Toral, A., Bisazza, A., & Sarti, G. (2022). DivEMT: Neural machine translation post-editing effort across typologically diverse languages. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP) (pp. 7795–7816). Association for Computational Linguistics. https://doi.org/10.18653/v1/2022.emnlp-main.532 ↗
  13. Hassani, Ghodrat, and Mohammad Malekshahi. 2024. Navigating the continuum: Exploring the value of pluralism in translation uncertainty research. Journal of Language Horizons, 8(1):93–114. find
  14. Holgado-Sáez, C., & Martínez Martínez, S. (2025). Análisis de errores terminológicos de unidades léxicas aisladas en el lenguaje jurídico-administrativo nacionalsocialista: traducción automática neuronal vs. traducción humana (alemán-español). Estudios de Lingüística de la Universidad de Alicante (ELUA), 44. https://doi.org/10.14198/elua.27618 ↗
  15. Kristeva, Julia. 1984. Revolution in Poetic Language. Translated by Leon S. Roudiez. Columbia University Press, New York. find
  16. Krüger, Ralph. 2018. Technologieinduzierte Verschiebungen in der Tektonik der Translationskompetenz. trans-kom, 11(1):101–137. find
  17. Kußmaul, Paul. 1986. Übersetzen als Entscheidungs-prozess. Die Rolle der Fehleranalyse in der Übersetzungsdidaktik. In M. Snell-Hornby (Ed.), Übersetzungswissenschaft. Eine Neuorientierung. Francke, 206–229. find
  18. Levý, Jiří. 2000 [1967]. Translation as a decision process. In L. Venuti (Ed.), The Translation Studies Reader. Routledge, 148–159. find
  19. Lu, Y. (2022). A bibliometric analysis of machine translation post-editing from 2012 to 2021. In 2022 International Conference on Electronics, Computers and Natural Language Processing in Information Retrieval (ECNLPIR) (pp. 65–69). IEEE. https://doi.org/10.1109/ECNL-PIR57021.2022.00020 ↗
  20. Martín de León, Celia, and José M. Cardona Guerra. 2022. Spoiled for choice?: Uncertainty facing options in translation. Translation & Interpreting, 14(2):50–67. ↗
  21. Mossop, Brian. 2020. Revising and Editing for Translators. Routledge. ↗
  22. Naikar, Neelam, Robert Hoffman, Emilie M. Roth, Gary Klein, Laura G. Militello, and Cindy Dominguez (2025). “Should we Make AI More Tool-like or Teammate-Like?” Journal of Cognitive Engineering and Decision Making, vol. 0(0) 1–21. ↗
  23. Nitzke, J., & Hansen-Schirra, S. (2021). A short guide to post-editing (Translation and Multilingual Natural Language Processing 16). Berlin: Language Science Press. find
  24. Pym, Anthony. 2010. Exploring Translation Theories. Routledge, London. ↗
  25. Pym, A. (2011). Democratizing translation technologies – The role of humanistic research. Paper presented at the Luspio Translation Automation Conference, Rome, April 5, 2011. find
  26. Robinson, Douglas. 1991. The Translator’s Turn. The Johns Hopkins University Press, Baltimore and London. ↗
  27. Roiss, S. (2021). Y las máquinas rompieron a traducir... Consideraciones didácticas en relación con la traducción automática de referencias culturales en el ámbito jurídico. TRANS: Revista de Traductología, 25, 491–505. https://doi.org/10.24310/trans.2021.v1i25.11978 ↗
  28. Sarasso, Pietro, Gianni Francesetti, Jan Roubal, Michela Gecele, Irene Ronga, Marco Neppi-Modona, and Katiuscia Sacco. 2022. Beauty and uncertainty as transformative factors. Frontiers in Human Neuroscience, 16:906188. find
  29. Setkowicz-Ryszka, Anna. 2025. Post-editing of Machine Translation Output in the Training of Legal Translators. Doctoral dissertation, University of Łódź. find
  30. Sinitsyna, Daria. 2025. Generative AI for translation in 2025. Intento blog, 30 March 2025. https://in-ten.to/blog/generative-ai-for-translation-in-2025/. ↗
  31. Vaccaro, Michelle, Abdullah Almaatouq, and Thomas Malone. 2024. When combinations of humans and AI are useful: A systematic review and meta-analysis. Nature Human Behaviour, 8:2293–2303. find
  32. Varela Salinas, M. J. , & Burbat, R. (2023). Google Translate and DeepL: Breaking taboos in translator training. Observational study and analysis. Ibérica, (45), 243-266. ↗
  33. Vasconcellos, M. (1987). “Postediting on-screen: machine translation from Spanish into English”. In A profession on the move: proceedings of translating and the computer 8, ed. C. Picken, 133-146. London: Aslib. find
  34. Velazquez, D., Grace, M., Karageorgos, K., Carin, L., Schliem, A., Zaikis, D., & Wechsler, R. (2025). LangMark: A multilingual dataset for automatic post-editing. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (ACL). Association for Computational Linguistics. https://doi.org/10.18653/v1/2025.acl-long.1569 ↗
  35. Veselovsky, V., et al. 2021. [Full bibliographic reference to be completed — cited in Section 2 of the original paper but missing from its reference list.] find
  36. Vinchon, Florent, V. Gironnay, and Todd Lubart. 2024. GenAI creativity in narrative tasks. Journal of Intelligence, 12(12):125. find
  37. Wang, Pei. 2008. What do you mean by “AI”? In P. Wang, B. Goertzel, and S. Franklin (Eds.), Artificial General Intelligence 2008. IOS Press, 362–373. find
  38. Wang, S. (2023). Towards English-centric zero-shot neural machine translation: The analysis and solution. In 2023 International Conference on Sensors, Electronics and Computer Engineering (ICSECE) (pp. 1577–1582). IEEE. https://doi.org/10.1109/ICSECE58870.2023.10263 323 ↗
  39. Wilss, Wolfram. 2007. The impact of uncertainty on translation. In R. Jääskeläinen, T. Puurtinen, and H. Stotesbury (Eds.), Text, Processes, and Corpora. Joensuu University, 163–170. find
  40. Winner, L. (2014). Technologies as Forms of Life. In: Sandler, R.L. (eds), Ethics and Emerging Technologies. Palgrave Macmillan, London. ↗

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/mihalache-2026-meaning-making-process-error-dynamics/