---
title: "The Broken Telephone Changes Tone: Examining Nuanced"
description: "Earlier studies showed that this repetition erodes factual content and diversity, but they relied on aggregate scores. 33 references."
authors: ["Quang Minh Nguyen", "Maida Aizaz", "Braahmi Padmakumar"]
year: 2026
venue: "Proceedings of the 1st Workshop on Multilinguality in the Era of Large Language Models (MeLLM 2026)"
source: "https://aclanthology.org/2026.mellm-1.29/"
doi: "10.18653/v1/2026.mellm-1.29"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/nguyen-2026-broken-telephone-changes-tone-examining/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/nguyen-2026-broken-telephone-changes-tone-examining/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/45/) / The Broken Telephone Changes Tone: Examining Nuanced Linguistic Cues in LLM Chains-of-Translation

# The Broken Telephone Changes Tone: Examining Nuanced Linguistic Cues in LLM Chains-of-Translation

**Authors:** Quang Minh Nguyen, Maida Aizaz and Braahmi Padmakumar · **Year:** 2026 · **Venue:** [Proceedings of the 1st Workshop on Multilinguality in the Era of Large Language Models (MeLLM 2026)](https://md.research.elamur.ai/free/en/venues/mellm/) ([Workshop on Multilinguality in the Era of Large Language Models 2026](https://md.research.elamur.ai/free/en/venues/mellm/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.mellm-1.29/) · [PDF](https://aclanthology.org/2026.mellm-1.29.pdf) · [DOI 10.18653/v1/2026.mellm-1.29](https://doi.org/10.18653/v1/2026.mellm-1.29) · [OpenAlex](https://openalex.org/W7166805901)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
LLM output is increasingly re-translated and reprocessed by other models. Earlier studies showed that this repetition erodes factual content and diversity, but they relied on aggregate scores. This paper asks what happens to finer linguistic features. The authors took 600 BBC News articles from 2024 onward, 50 from each of 12 topics, and ran them through 12 rounds of English→L→English backtranslation. They varied the pivot language (Chinese, French, Korean, Arabic), the model (Qwen, Llama, Gemma) and the chain design (self-loop, two-player, multiplayer), giving 17 configurations. Using spaCy dependency parses and regex lexicons, they tracked epistemic markers, degree adverbs, grammatical voice and nominalisation, normalised per 1,000 words. Over the iterations, texts sound more certain: evidential and factive markers grow while hedges shrink. They also become more formal, with Latinate nouns replacing verbs and agent-naming passives disappearing. The size of these effects depends on the model. The study explains part of the reported factual loss, showing that uncertainty, attribution and agency cues quietly vanish in model-mediated text.

## Key points
- **Setup:** The study used Qwen3-30B-A3B-Instruct-2507, Llama-3.3-70B-Instruct and Gemma-3-27B-it via OpenRouter at temperature 1.0. The 17 runs (12 self-loop, 4 two-player, 1 multiplayer) produced 122,400 back-translated texts.
- **Factual decay:** At iteration 12, only 57% of source facts remain supported according to FActScore with a Llama judge. BLEU falls from 36.8 to 21.0, BERTScore F1 from 0.94 to 0.91, and outputs end up 1–57% shorter.
- **Epistemic markers:** Evidentials rise in every run (mean +0.53 on a baseline of 0.90). Factives rise in 15 of 17 runs, and hedges fall in 13 of 17.
- **Nominalisation and verbs:** Nominalisation density increases in all runs (mean +6.95, about 21% above the baseline of 33.0). Verb density drops in 15 of 17 runs.
- **Voice and model effects:** With-agent passives decline, most sharply for Arabic and Korean. Qwen self-loops show the largest nominalisation and booster gains, and Llama the smallest.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [Korea Advanced Institute of Science and Technology](https://md.research.elamur.ai/free/en/universities/korea-advanced-institute-of-science-and-technology/) — University, Daejeon, South Korea · 51 works · 3,167 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

## Cite

Cite as: Quang Minh Nguyen, Maida Aizaz and Braahmi Padmakumar. 2026. The Broken Telephone Changes Tone: Examining Nuanced Linguistic Cues in LLM Chains-of-Translation. Proceedings of the 1st Workshop on Multilinguality in the Era of Large Language Models (MeLLM 2026).

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

```bibtex
@inproceedings{nguyen2026broken,
  title = {{The Broken Telephone Changes Tone: Examining Nuanced Linguistic Cues in LLM Chains-of-Translation}},
  author = {Quang Minh Nguyen, Maida Aizaz and Braahmi Padmakumar},
  booktitle = {Proceedings of the 1st Workshop on Multilinguality in the Era of Large Language Models (MeLLM 2026)},
  year = {2026},
  doi = {10.18653/v1/2026.mellm-1.29},
  url = {https://doi.org/10.18653/v1/2026.mellm-1.29}
}
```

- [← previous The Bitter Lesson of Diffusion Language Models for Agentic Workflows: A Comprehensive Reality Check](https://md.research.elamur.ai/free/en/papers/lu-2026-bitter-lesson-diffusion-language-models/)
- [next → The Chulalongkorn Corpus of Spoken Thai (CCOST)](https://md.research.elamur.ai/free/en/papers/pittayaporn-2026-chulalongkorn-corpus-spoken-thai-ccost/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/45/)

## References

33 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Alberto Acerbi and Joseph M Stubbersfield. 2023. Large language models show human-like content biases in transmission chain experiments. Proceedings of the National Academy of Sciences, 120(44):e2313790120. [find](https://scholar.google.com/scholar?q=Alberto+Acerbi+and+Joseph+M+Stubbersfield%2E+2023%2E+Large+language+models+show+human%2Dlike+content+biases+in+transmission+chain+experiments%2E+Proceedings+of+the+National+Academy+of+Sciences%2C+120%2844%29%3Ae23137)
2. Yazan Shaker Almahameed, Khaleel Bader Al Bataineh, and Raeda Mofid George Ammari. 2022. The Use of Passive Voice in News Reports for Political Purposes. Journal of Language Teaching and Research, 13(6):1196–1202. [find](https://scholar.google.com/scholar?q=Yazan+Shaker+Almahameed%2C+Khaleel+Bader+Al+Bataineh%2C+and+Raeda+Mofid+George+Ammari%2E+2022%2E+The+Use+of+Passive+Voice+in+News+Reports+for+Political+Purposes%2E+Journal+of+Language+Teaching+and+Research%2C+13%28)
3. Mona Baker. 2018. In other words: A coursebook on translation. Routledge. [find](https://scholar.google.com/scholar?q=Mona+Baker%2E+2018%2E+In+other+words%3A+A+coursebook+on+translation%2E+Routledge%2E)
4. Douglas Biber, Stig Johansson, Geoffrey N Leech, Susan Conrad, and Edward Finegan. 2000. Grammar of spoken and written english. [find](https://scholar.google.com/scholar?q=Douglas+Biber%2C+Stig+Johansson%2C+Geoffrey+N+Leech%2C+Susan+Conrad%2C+and+Edward+Finegan%2E+2000%2E+Grammar+of+spoken+and+written+english%2E)
5. Elvis Dohmatob, Yunzhen Feng, Arjun Subramonian, and Julia Kempe. 2025. Strong Model Collapse. In The Thirteenth International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Elvis+Dohmatob%2C+Yunzhen+Feng%2C+Arjun+Subramonian%2C+and+Julia+Kempe%2E+2025%2E+Strong+Model+Collapse%2E+In+The+Thirteenth+International+Conference+on+Learning+Representations%2E)
6. Mohammed Farghal and Mohammed O Al-Shorafat. 1996. The Translation of English Passives into Arabic: An Empirical Perspective. Target. International Journal of Translation Studies, 8(1):97–118. [find](https://scholar.google.com/scholar?q=Mohammed+Farghal+and+Mohammed+O+Al%2DShorafat%2E+1996%2E+The+Translation+of+English+Passives+into+Arabic%3A+An+Empirical+Perspective%2E+Target%2E+International+Journal+of+Translation+Studies%2C+8%281%29%3A97%E2%80%93118%2E)
7. Richárd Farkas, Veronika Vincze, György Móra, János Csirik, and György Szarvas. 2010. [The CoNLL-2010 Shared Task: Learning to Detect Hedges and their Scope in Natural Language Text](https://md.research.elamur.ai/free/en/papers/farkas-2010-conll-2010-shared-task-learning/). In Proceedings of the Fourteenth Conference on Computational Natural Language Learning – Shared Task, pages 1– 12, Upp…
8. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, and 542 o… — [The Llama 3 Herd of Models](https://md.research.elamur.ai/free/en/papers/grattafiori-2024-llama-3-herd-models/)
9. Matthew Honnibal, Ines Montani, Sofie Van Lan-deghem, and Adriane Boyd. 2020. spaCy: Industrial-strength Natural Language Processing in Python. [find](https://scholar.google.com/scholar?q=Matthew+Honnibal%2C+Ines+Montani%2C+Sofie+Van+Lan%2Ddeghem%2C+and+Adriane+Boyd%2E+2020%2E+spaCy%3A+Industrial%2Dstrength+Natural+Language+Processing+in+Python%2E)
10. Ken Hyland. 1998. [Hedging in scientific research articles](https://md.research.elamur.ai/free/en/papers/hyland-1998-hedging-scientific-research-articles/).
11. Lauri Karttunen. 1971. Implicative Verbs. Language, 47(2):340–358. [find](https://scholar.google.com/scholar?q=Lauri+Karttunen%2E+1971%2E+Implicative+Verbs%2E+Language%2C+47%282%29%3A340%E2%80%93358%2E)
12. Youngshin Kim. 2007. Korean-English Differences of Communicative Preferences with Focus on the Subject Position. The Journal of Translation Studies, 8(2):241–258. [find](https://scholar.google.com/scholar?q=Youngshin+Kim%2E+2007%2E+Korean%2DEnglish+Differences+of+Communicative+Preferences+with+Focus+on+the+Subject+Position%2E+The+Journal+of+Translation+Studies%2C+8%282%29%3A241%E2%80%93258%2E)
13. Paul Kiparsky and Carol Kiparsky. 1970. FACT, pages 143–173. De Gruyter Mouton, Berlin, Boston. [find](https://scholar.google.com/scholar?q=Paul+Kiparsky+and+Carol+Kiparsky%2E+1970%2E+FACT%2C+pages+143%E2%80%93173%2E+De+Gruyter+Mouton%2C+Berlin%2C+Boston%2E)
14. Dongryeol Lee, Yerin Hwang, Yongil Kim, Joonsuk Park, and Kyomin Jung. 2025. Are LLM-Judges Robust to Expressions of Uncertainty? Investigating the effect of Epistemic Markers on LLM-based Evaluation. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for… [find](https://scholar.google.com/scholar?q=Dongryeol+Lee%2C+Yerin+Hwang%2C+Yongil+Kim%2C+Joonsuk+Park%2C+and+Kyomin+Jung%2E+2025%2E+Are+LLM%2DJudges+Robust+to+Expressions+of+Uncertainty%3F+Investigating+the+effect+of+Epistemic+Markers+on+LLM%2Dbased+Evaluation%2E)
15. Yucheng Li, Frank Guerin, and Chenghua Lin. 2024. LatestEval: Addressing data contamination in language model evaluation through dynamic and time-sensitive test construction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 18600–18607. [find](https://scholar.google.com/scholar?q=Yucheng+Li%2C+Frank+Guerin%2C+and+Chenghua+Lin%2E+2024%2E+LatestEval%3A+Addressing+data+contamination+in+language+model+evaluation+through+dynamic+and+time%2Dsensitive+test+construction%2E+In+Proceedings+of+the+AAA)
16. Chin-Yew Lin. 2004. [ROUGE: A Package for Automatic Evaluation of Summaries](https://md.research.elamur.ai/free/en/papers/lin-2004-rouge-package-automatic-evaluation-summaries/). In Text Summarization Branches Out, pages 74–81, Barcelona, Spain. Association for Computational Linguistics.
17. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://md.research.elamur.ai/free/en/papers/liu-2019-roberta-robustly-optimized-bert-pretraining/). arXiv preprint. ArXiv:1907.11692 [cs].
18. Jiaming Luo, Colin Cherry, and George Foster. 2024. [To Diverge or Not to Diverge: A Morphosyntactic Perspective on Machine Translation vs Human Translation](https://md.research.elamur.ai/free/en/papers/luo-2024-diverge-or-not-diverge-morphosyntactic/). Transactions of the Association for Computational Linguistics, 12:355–371.
19. George A. Miller. 1994. [WordNet: A Lexical Database for English](https://md.research.elamur.ai/free/en/papers/miller-1994-wordnet/). In Human Language Technology: Proceedings of a Workshop held at Plainsboro, New Jersey, March 8-11, 1994.
20. Sewon Min, Kalpesh Krishna, Xinxi Lyu, Mike Lewis, Wen-tau Yih, Pang Koh, Mohit Iyyer, Luke Zettlemoyer, and Hannaneh Hajishirzi. 2023. [Factscore: Fine-grained atomic evaluation of factual precision in long form text generation](https://md.research.elamur.ai/free/en/papers/min-2023-factscore-fine-grained-atomic-evaluation/). In Proceedings of the 2023 Conference on Empirical Methods in Natural L…
21. Amr Mohamed, Mingmeng Geng, Michalis Vazirgiannis, and Guokan Shang. 2025. LLM as a Broken Telephone: Iterative Generation Distorts Information. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7493–7509. [find](https://scholar.google.com/scholar?q=Amr+Mohamed%2C+Mingmeng+Geng%2C+Michalis+Vazirgiannis%2C+and+Guokan+Shang%2E+2025%2E+LLM+as+a+Broken+Telephone%3A+Iterative+Generation+Distorts+Information%2E+In+Proceedings+of+the+63rd+Annual+Meeting+of+the+Associ)
22. Jérémy Perez, Grgur Kovač, Corentin Léger, Cédric Colas, Gaia Molinaro, Maxime Derex, Pierre-Yves Oudeyer, and Clément Moulin-Frier. 2025. When LLMs play the telephone game: Cultural attractors as conceptual tools to evaluate LLMs in multi-turn settings. In The Thirteenth International Conference o… [find](https://scholar.google.com/scholar?q=J%C3%A9r%C3%A9my+Perez%2C+Grgur+Kovac%CC%8C%2C+Corentin+L%C3%A9ger%2C+C%C3%A9dric+Colas%2C+Gaia+Molinaro%2C+Maxime+Derex%2C+Pierre%2DYves+Oudeyer%2C+and+Cl%C3%A9ment+Moulin%2DFrier%2E+2025%2E+When+LLMs+play+the+telephone+game%3A+Cultural+attractors+as+co)
23. Richard Xiao, Tony McEnery, and Yufang Qian. 2006. Passive constructions in english and chinese: A corpus-based contrastive study. Languages in contrast, 6(1):109–149. [find](https://scholar.google.com/scholar?q=Richard+Xiao%2C+Tony+McEnery%2C+and+Yufang+Qian%2E+2006%2E+Passive+constructions+in+english+and+chinese%3A+A+corpus%2Dbased+contrastive+study%2E+Languages+in+contrast%2C+6%281%29%3A109%E2%80%93149%2E)
24. Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger, and Yoav Artzi. 2020. BERTScore: Evaluating Text Generation with BERT. arXiv preprint. ArXiv:1904.09675 [cs]. [find](https://scholar.google.com/scholar?q=Tianyi+Zhang%2C+Varsha+Kishore%2C+Felix+Wu%2C+Kilian+Q%2E+Weinberger%2C+and+Yoav+Artzi%2E+2020%2E+BERTScore%3A+Evaluating+Text+Generation+with+BERT%2E+arXiv+preprint%2E+ArXiv%3A1904%2E09675+%5Bcs%5D%2E)
25. Kaitlyn Zhou, Dan Jurafsky, and Tatsunori Hashimoto. 2023. Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 5506–5524, Singapore. Association for Comput… [find](https://scholar.google.com/scholar?q=Kaitlyn+Zhou%2C+Dan+Jurafsky%2C+and+Tatsunori+Hashimoto%2E+2023%2E+Navigating+the+Grey+Area%3A+How+Expressions+of+Uncertainty+and+Overconfidence+Affect+Language+Models%2E+In+Proceedings+of+the+2023+Conference+on+)
26. A Andrew J Peterson. 2025. AI and the problem of knowledge collapse. AI & SOCIETY, 40(5):3249–3269. [find](https://scholar.google.com/scholar?q=A+Andrew+J+Peterson%2E+2025%2E+AI+and+the+problem+of+knowledge+collapse%2E+AI+%26+SOCIETY%2C+40%285%29%3A3249%E2%80%933269%2E)
27. Matt Post. 2018. [A Call for Clarity in Reporting BLEU Scores](https://md.research.elamur.ai/free/en/papers/post-2018-call-clarity-reporting-bleu-scores/). In Proceedings of the Third Conference on Machine Translation: Research Papers, pages 186– 191, Brussels, Belgium. Association for Computational Linguistics.
28. Neil Rathi, Dan Jurafsky, and Kaitlyn Zhou. 2025. Humans overrely on overconfident language models, across languages. In Second Conference on Language Modeling. [find](https://scholar.google.com/scholar?q=Neil+Rathi%2C+Dan+Jurafsky%2C+and+Kaitlyn+Zhou%2E+2025%2E+Humans+overrely+on+overconfident+language+models%2C+across+languages%2E+In+Second+Conference+on+Language+Modeling%2E)
29. Ilia Shumailov, Zakhar Shumaylov, Yiren Zhao, Nicolas Papernot, Ross Anderson, and Yarin Gal. 2024. Ai models collapse when trained on recursively generated data. Nature, 631(8022):755–759. [find](https://scholar.google.com/scholar?q=Ilia+Shumailov%2C+Zakhar+Shumaylov%2C+Yiren+Zhao%2C+Nicolas+Papernot%2C+Ross+Anderson%2C+and+Yarin+Gal%2E+2024%2E+Ai+models+collapse+when+trained+on+recursively+generated+data%2E+Nature%2C+631%288022%29%3A755%E2%80%93759%2E)
30. Anna Siewierska. 2024. The passive: A comparative linguistic analysis. Routledge. [find](https://scholar.google.com/scholar?q=Anna+Siewierska%2E+2024%2E+The+passive%3A+A+comparative+linguistic+analysis%2E+Routledge%2E)
31. Gemma Team, Aishwarya Kamath, Johan Ferret, Shreya Pathak, Nino Vieillard, Ramona Merhej, Sarah Perrin, Tatiana Matejovicova, Alexandre Ramé, Morgane Rivière, Louis Rouillard, Thomas Mesnard, Geoffrey Cideron, Jean-bastien Grill, Sabela Ramos, Edouard Yvinec, Michelle Casbon, Etienne Pot, Ivo Penche… [find](https://scholar.google.com/scholar?q=Gemma+Team%2C+Aishwarya+Kamath%2C+Johan+Ferret%2C+Shreya+Pathak%2C+Nino+Vieillard%2C+Ramona+Merhej%2C+Sarah+Perrin%2C+Tatiana+Matejovicova%2C+Alexandre+Ram%C3%A9%2C+Morgane+Rivi%C3%A8re%2C+Louis+Rouillard%2C+Thomas+Mesnard%2C+Geoffrey)
32. Qwen Team. 2025. [Qwen3 technical report](https://md.research.elamur.ai/free/en/papers/yang-2025-qwen3-technical-report/). Preprint, arXiv:2505.09388.
33. Dustin Wright, Sarah Masud, Jared Moore, Srishti Yadav, Maria Antoniak, Peter Ebert Christensen, Chan Young Park, and Isabelle Augenstein. 2025. Epistemic Diversity and Knowledge Collapse in Large Language Models. arXiv preprint arXiv:2510.04226. — [arXiv:2510.04226](https://arxiv.org/abs/2510.04226) [find](https://scholar.google.com/scholar?q=Dustin+Wright%2C+Sarah+Masud%2C+Jared+Moore%2C+Srishti+Yadav%2C+Maria+Antoniak%2C+Peter+Ebert+Christensen%2C+Chan+Young+Park%2C+and+Isabelle+Augenstein%2E+2025%2E+Epistemic+Diversity+and+Knowledge+Collapse+in+Large+Lan)

*Updated 1 November 2026*