Elamur Research⌕
☰
home / papers / 2026 / “Newspaper Eat” Means “Not Tasty”: A Taxonomy and Benchmark for Coded Language in Real-World Chinese Online Reviews

“Newspaper Eat” Means “Not Tasty”: A Taxonomy and Benchmark for Coded Language in Real-World Chinese Online Reviews

Paper · 2026 · Ruyuan Wan, Changye Li and Ting-Hao Kenneth Huang

Summary

Chinese-speaking reviewers on Google Maps sometimes hide criticism inside reviews that look positive or neutral. They use homophones, Pinyin, emoji and other substitutions that readers in their community can decode but that machine translation and moderation tools miss. The authors treat this coded language as a real form of communication rather than noise. They built CodedLang, a collection of 7,744 Chinese Google Maps reviews of U.S. businesses. Of these, 900 carry span-level labels for coded expressions.

The data was mined from two large public review corpora through an iterative human-in-the-loop process. It began with seed examples from news and social media and grew through dictionary lookup, LLM-assisted bootstrapping and manual validation. The authors also proposed a seven-class taxonomy of encoding strategies. They then tested language models on three tasks: detecting coded language, classifying its type, and predicting review ratings. Even strong models often failed to spot or interpret the codes. A phonetic analysis of coded versus decoded forms linked pronunciation-based distortions to errors in downstream predictions. The work provides a benchmark for a hard, understudied problem in real-world NLP.

Key points

  • Dataset: CodedLang contains 7,744 reviews, 900 of them with span-level coded-language annotations. It is released at github.com/Crowd-AI-Lab/CodedLang.
  • Taxonomy: The seven classes are:
    • Ambiguous Homophone
    • Non-Lexical Homophone
    • Phonetic Substitution (Pinyin, Zhuyin, Roman letters)
    • Emoji Substitution
    • Orthographic Substitution
    • Cross-Lingual Phonetic Encoding (e.g., "Newspaper eat", which decodes to 不好吃, "not tasty")
    • Cipher
  • Sources and rarity: Reviews came from Google Maps corpora of 1.77M restaurant reviews (Li et al., 2022) and 666M local reviews (Yan et al., 2023). Filtering for Chinese characters left 5,391 and 112,521 reviews respectively. A manual pilot found coded language in only 19 of the 5,391 restaurant reviews (0.35%).
  • Construction: The process started from 23 seed examples taken from Xiaohongshu, Threads and news reports. Two authors annotated the data, working through a bootstrapping pipeline that combined dictionaries, LLMs and human checks.
  • Limitations: Coverage depends on what the seed dictionaries and LLMs could surface. The data ends in 2022 and is limited to U.S. reviews. Interpretation reflects the annotators' own linguistic background.

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

🎓University of WashingtonSeattle · United States353 works · 200,043 citations🎓Pennsylvania State UniversityState College · United States56 works · 5,360 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 1Research subjects and languages the work is about.Evaluation & benchmarks · 1,220 works

Cite

Cite as Ruyuan Wan, Changye Li and Ting-Hao Kenneth Huang. 2026. “Newspaper Eat” Means “Not Tasty”: A Taxonomy and Benchmark for Coded Language in Real-World Chinese Online Reviews. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.426

@inproceedings{wan2026newspaper,
  title = {{“Newspaper Eat” Means “Not Tasty”: A Taxonomy and Benchmark for Coded Language in Real-World Chinese Online Reviews}},
  author = {Ruyuan Wan, Changye Li and Ting-Hao Kenneth Huang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.426},
  url = {https://doi.org/10.18653/v1/2026.acl-long.426}
}

References

28 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Jialin Chen. 2025. Coded language emerges in taiwanese online reviews. A Taiwanese user’s highly encrypted review of an overseas restaurant attracts 1.61 million viewers. find
  2. Won Ik Cho and Soomin Kim. 2021. Google-trickers, yaminjeongeum, and leetspeak: An empirical taxonomy for intentionally noisy user-generated text. In Proceedings of the seventh workshop on noisy user-generated text (W-NUT 2021), pages 56–61. find
  3. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, Luke Marris, Sam Petulla, Colin Gaffney, Asaf Aharoni, Nathan Lintz, Tiago Cardal Pais, Henrik Jacobsson, Idan Szpektor, Nan-Jiang Jiang, and 3416 other…
  4. DeepSeek-AI, Aixin Liu, Aoxue Mei, Bangcai Lin, Bing Xue, Bingxuan Wang, Bingzheng Xu, Bochao Wu, Bowei Zhang, Chaofan Lin, Chen Dong, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenhao Xu, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, and 245 others. 2025. Deepseek-v3.2: Pushing the frontier of open…
  5. Mia Fan. 2024. Disappointed chinese diners are writing google maps business reviews in secret codes. Accessed May 15, 2025. find
  6. Casey Fiesler and Nicholas Proferes. 2018. “participant” perceptions of twitter research ethics. Social Media + Society, 4(1):2056305118763366. find
  7. Jan Fillies and Adrian Paschke. 2024. Simple llm based approach to counter algospeak. In Proceedings of the 8th Workshop on Online Abuse and Harms (WOAH 2024), pages 136–145. find
  8. Huang Huang, Xiaoquan Kong, and 1 others. 2025. pypinyin: A python library for converting chinese characters to pinyin. https://github.com/ mozillazg/python-pinyin. Version 0.55.0. ↗
  9. Heng Ji and Kevin Knight. 2018. Creative language encoding under censorship. In Proceedings of the first workshop on natural language processing for internet freedom, pages 23–33. find
  10. Julie Kallini, Dan Jurafsky, Christopher Potts, and Martijn Bartelds. 2025. False friends are not foes: Investigating vocabulary overlap in multilingual language models. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 21138– 21154. find
  11. Soomin Kim, Changhoon Oh, Won Ik Cho, Donghoon Shin, Bongwon Suh, and Joonhwan Lee. 2021. Tr-kic g00gle: Why and how users game translation algorithms. Proceedings of the ACM on Human-Computer Interaction, 5(CSCW2):1–24. find
  12. Jiacheng Li, Jingbo Shang, and Julian McAuley. 2022. Uctopic: Unsupervised contrastive learning for phrase representations and topic mining. arXiv preprint arXiv:2202.13469. find
  13. Jianfei Ma, Zhaoxin Feng, Emmanuele Chersoni, Huacheng Song, and Ziqi Zhang. 2025a. Phonothink: Improving large language models’ reasoning on chinese phonological ambiguities. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 19018–19033. find
  14. Jianfei Ma, Zhaoxin Feng, Huacheng Song, Emmanuele Chersoni, and Zheng Chen. 2025b. Reasoning or memorization? investigating llms’ capability in restoring chinese internet homophones. In Proceedings of the 3rd Workshop on Towards Knowledgeable Foundation Models (KnowFM), pages 120–139. find
  15. Julia Mendelsohn, Ronan Le Bras, Yejin Choi, and Maarten Sap. 2023. From dogwhistles to bullhorns: Unveiling coded rhetoric with language models. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 15162–15180. find
  16. Meta. 2024. Llama-3.1-8b-instruct. https://huggingface.co/meta-llama/Llama-3. 1-8B-Instruct. Accessed: 2026-04-20. ↗
  17. Andrew Morris, Viktoria Maier, and Phil Green. 2004. From wer and ril to mer and wil: improved evaluation measures for connected speech recognition. find
  18. OpenAI. 2025. Gpt-5 mini model documentation. https://platform.openai.com/docs/ models/gpt-5-mini. Accessed: 2026-01. ↗
  19. Zhu Qing. 2024. ’Newspaper eat:’ How Chinese students hide warnings on Google reviews. Accessed May 15, 2025. find
  20. Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jingren Zhou, and 25 others. 2025. Qwen2.5 technical report. Preprint, arXiv:2412.15115. find
  21. Itsumi Saito, Kugatsu Sadamitsu, Hisako Asano, and Yoshihiro Matsuo. 2014. Morphological analysis for japanese noisy text based on character-level and word-level normalization. In Proceedings of COL-ING 2014, the 25th International Conference on Computational Linguistics: Technical Papers, pages 177… find
  22. Stefan Taubert. 2025. pinyin-to-ipa: A tool for converting pinyin to ipa. https://github.com/ stefantaubert/pinyin-to-ipa. Version 1.0.0. ↗
  23. Ruyuan Wan, Lingbo Tong, Tiffany Knearem, Toby Jia-Jun Li, Ting-Hao’Kenneth’ Huang, and Qunfang Wu. 2025. Hashtag re-appropriation for audience control on recommendation-driven social media xiaohongshu (rednote). In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, pages… find
  24. Yunze Xiao, Yujia Hu, Kenny Choo, and Roy Lee. 2024. Toxicloakcn: Evaluating robustness of offensive language detection in chinese with cloaking perturbations. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 6012–6025. find
  25. Canwen Xu, Wangchunshu Zhou, Tao Ge, Ke Xu, Julian McAuley, and Furu Wei. 2021. Blow the dog whistle: A chinese dataset for cant understanding with common sense and world knowledge. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics:… find
  26. An Yan, Zhankui He, Jiacheng Li, Tianyang Zhang, and Julian McAuley. 2023. Personalized showcases: Generating multi-modal explanations for recommendations. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 2251–2255. find
  27. Boliang Zhang, Hongzhao Huang, Xiaoman Pan, Heng Ji, Kevin Knight, Zhen Wen, Yizhou Sun, Jiawei Han, and Bulent Yener. 2014. Be appropriate and funny: Automatic entity morph encoding. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers),… find
  28. Boliang Zhang, Hongzhao Huang, Xiaoman Pan, Sujian Li, Chin-Yew Lin, Heng Ji, Kevin Knight, Zhen Wen, Yizhou Sun, Jiawei Han, and 1 others. 2015. Context-aware entity morph decoding. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wan-2026-newspaper-eat-means-not-tasty/