Elamur Research⌕
☰
home / papers / 2026 / From Trial by Fire To Sleep Like a Baby: A Lexicon of Anxiety Associations for 20k English Multiword Expressions

From Trial by Fire To Sleep Like a Baby: A Lexicon of Anxiety Associations for 20k English Multiword Expressions

Paper · 2026 · Saif M. Mohammad

Summary

Lexicons that link words to anxiety already exist, but phrases larger than a single word have been mostly ignored. Saif M. Mohammad fills this gap with WorryMWEs, a crowdsourced resource that records how strongly over 20k English multiword expressions are tied to anxiety or to calmness. The paper treats an MWE simply as a sequence of two or more words, often one with some notable semantic, syntactic or functional behaviour.

The ratings are descriptive norms. They record how people actually perceive the link between a phrase and anxiety, not how language ought to be used. The author reports that the annotations are highly reliable.

The lexicon is then used for two analyses:

  • how common anxiety-linked and calm-linked expressions are, broken down by MWE type and by length (bigrams, trigrams, fourgrams);
  • how far a phrase's anxiety rating can be predicted from the ratings of its individual words.

The paper also argues at length that such a resource matters for many fields, including semantics, affective science, computational social science and public health. It notes that human-made norms remain necessary even when LLMs can produce lexicons, because LLM output still needs validation.

Key points

  • WorryMWEs covers more than 20k English MWEs, rated by crowd annotators for association with anxiety versus calmness.
  • WorryMWEs is released together with WorryWords (about 44k words) as WorryLex v2, freely available at saifmohammad.com/worrylex.html.
  • The analyses compare MWE types and n-gram lengths (two, three and four words), and test whether MWE anxiety is compositional with respect to the constituent words.
  • Related resources:
    • Muraki et al. (2023): concreteness ratings for about 62k MWEs.
    • NRC VAD Lexicon v2: valence, arousal and dominance for 10k MWEs.
    • Jochim et al. (2018): sentiment annotation of 5000 idioms.
  • The paper motivates its anxiety–calmness framing with the view of anxiety as one continuum running from high calmness to high anxiety (Siddaway et al., 2018).

Interesting ideas

  • I1. Rating items on one bipolar axis from calm to anxious, instead of a yes/no anxiety label, captures soothing phrases (e.g. sleep like a baby) as well as alarming ones.
  • I2. Compositionality is checked directly by comparing each phrase's rating with the ratings of its parts, which shows where the parts do not explain the whole.
  • I3. A human-made lexicon is used in two halves: one portion to steer LLM generations and a held-out portion to validate them.
  • I4. Expressions of emotion regulation (taking a deep breath, bottling it up) are singled out as linguistic evidence of coping strategies, not just of the emotion itself.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (world): affective states of entities in the world of a text could be stored as a signed value on such an axis, so "calmed down" and "on edge" become opposite movements of the same state.
  • I2 (frames): the same gap between phrase and word scores could flag idioms whose literal frame reading (jumping out of my skin) should be replaced by a single emotional-state frame.
  • I3 (learn): LLM teacher tiers that propose lexical or affective labels could be scored against a held-out slice of human norms, giving an external check separate from the training signal.
  • I4 (frames): pragmatics could recognise such phrases as coping acts by an experiencer, enriching who-feels-what-and-does-what-about-it frames for tales and fables.

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, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksPsychology · 1,649 worksExperimental and Cognitive Psychology · 930 worksSocial Psychology · 330 works

🧭 Topics 2Research subjects and languages the work is about.English · 2,379 worksIdioms & multiword expressions · 222 works

Cite

Cite as Saif M. Mohammad. 2026. From Trial by Fire To Sleep Like a Baby: A Lexicon of Anxiety Associations for 20k English Multiword Expressions. arXiv:2602.18692 · LREC 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2602.18692 · arXiv 2602.18692

@misc{mohammad2026trial,
  title = {{From Trial by Fire To Sleep Like a Baby: A Lexicon of Anxiety Associations for 20k English Multiword Expressions}},
  author = {Saif M. Mohammad},
  year = {2026},
  doi = {10.48550/arxiv.2602.18692},
  eprint = {2602.18692},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2602.18692}
}

References

45 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Ashraf Allawama, Aseel Zibin, Abdel Rahman Altakhaineh, et al. 2025. Idioms as gateways to emotional expressions of sadness and joy in french. Journal of Intercultural Communication, 25(1):83–97. ↗
  2. Melissa Bateson, Ben Brilot, and Daniel Nettle. 2011. Anxiety: an evolutionary approach. The Canadian Journal of Psychiatry, 56(12):707– 715. find
  3. Marc Brysbaert and Boris New. 2009. Moving beyond kučera and francis: A critical evaluation of current word frequency norms and the introduction of a new and improved word frequency measure for american english. Behavior research methods, 41(4):977–990. ↗
  4. Marc Brysbaert, Boris New, and Emmanuel Keuleers. 2012. Adding part-of-speech information to the subtlex-us word frequencies. Behavior research methods, 44:991–997. ↗
  5. Piotr Cap. 2016. The language of fear: Communicating threat in public discourse. Springer. find
  6. Francesca MM Citron, Cristina Cacciari, Jakob M Funcke, Chun-Ting Hsu, and Arthur M Jacobs. 2019. Idiomatic expressions evoke stronger emotional responses in the brain than literal sentences. Neuropsychologia, 131:233–248. ↗
  7. Pamela M Cole, Laura Marie Armstrong, and Caroline K Pemberton. 2010. The role of language in the development of emotion regulation. Child development at the intersection of emotion and cognition. ↗
  8. Mathieu Constant, Gülşen Eryiǧit, Johanna Monti, Lonneke van der Plas, Carlos Ramisch, Michael Rosner, and Amalia Todirascu. 2017. Survey: Multiword expression processing: A Survey. Computational Linguistics, 43(4):837–892.
  9. Nina Heinrichs, Ronald M Rapee, Lynn A Alden, Susan Bögels, Stefan G Hofmann, Kyung Ja Oh, and Yuji Sakano. 2006. Cultural differences in perceived social norms and social anxiety. Behaviour research and therapy, 44(8):1187–1197. ↗
  10. Hossam S Ibrahim, Sherif M Abdou, and Mervat Gheith. 2015. Idioms-proverbs lexicon for modern standard arabic and colloquial sentiment analysis. arXiv preprint arXiv:1506.01906. ↗
  11. Ray Jackendoff. 1997. Twistin’the night away. Language, pages 534–559. find
  12. Charles Jochim, Francesca Bonin, Roy Bar-Haim, and Noam Slonim. 2018. SLIDE - a sentiment lexicon of common idioms. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. European Language Resources Association (ELRA). find
  13. Natalie A Kacinik. 2014. Sticking your neck out and burying the hatchet: what idioms reveal about embodied simulation. Frontiers in human neuroscience, 8:689. ↗
  14. Jean Kim and Jack Gorman. 2005. The psychobiology of anxiety. Clinical Neuroscience Research, 4(5-6):335–347. ↗
  15. Sophia Yat Mei Lee. 2017. Figurative language in emotion expressions. In Workshop on Chinese Lexical Semantics, pages 408–419. Springer. find
  16. Abdullah Mazhar, Zuhair hasan shaik, Aseem Srivastava, Polly Ruhnke, Lavanya Vaddavalli, Sri Keshav Katragadda, Shweta Yadav, and Md Shad Akhtar. 2025. Figurative-cum-commonsense knowledge infusion for multimodal mental health meme classification. find
  17. Sandra Mitrović, Oscar William Lithgow-Serrano, and Carlo Schillaci. 2024. Comparing panic and anxiety on a dataset collected from social media. In Proceedings of the 9th Workshop on Computational Linguistics and Clinical Psychology (CLPsych 2024), pages 153–165, St. Julians, Malta. Association for Computational Linguistics. find
  18. Saif Mohammad. 2018. Obtaining reliable human ratings of valence, arousal, and dominance for 20,000 English words. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 174–184, Melbourne, Australia. Association for Computational Linguistics.
  19. Saif Mohammad. 2025. Breaking bad: Norms for valence, arousal, and dominance for over 10k english multiword expressions. In Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics, pages 1964–1988. ↗
  20. Saif M. Mohammad. 2022. Ethics sheet for automatic emotion recognition and sentiment analysis. Computational Linguistics, 48(2):239–278. ↗
  21. Saif M. Mohammad. 2023. Best practices in the creation and use of emotion lexicons. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, Dubrovnik, Croatia. Association for Computational Linguistics. ↗
  22. Saif M. Mohammad. 2024a. Worrywords: Norms of anxiety association for 44,450 english words. In Proceedings of The Annual Conference of the Empirical Methods on Natural Language Processing (EMNLP 2024, main), Miami, FL. find
  23. Saif M. Mohammad. 2024b. WorryWords: Norms of anxiety association for over 44k English words. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 16261–16278, Miami, Florida, USA. Association for Computational Linguistics.
  24. Saif M. Mohammad. 2026a. From composure to catastrophe: Norms of calmness–anxiety associations for 64,000 english words and multiword expressions. PsyArXiv. find
  25. Saif M. Mohammad. 2026b. When are we worried? temporal trends of anxiety and what they reveal about us. ↗
  26. Saif M. Mohammad and Peter D. Turney. 2010. Emotions evoked by common words and phrases: Using Mechanical Turk to create an emotion lexicon. In Proceedings of the NAACL-HLT Workshop on Computational Approaches to Analysis and Generation of Emotion in Text, LA, California.
  27. Saif M. Mohammad and Peter D. Turney. 2013. Crowdsourcing a word–emotion association lexicon. Computational Intelligence, 29(3):436–465. ↗
  28. Agnes Moors, Jan De Houwer, Dirk Hermans, Sabine Wanmaker, Kevin Van Schie, Anne-Laura Van Harmelen, Maarten De Schryver, Jeffrey De Winne, and Marc Brysbaert. 2013. Norms of valence, arousal, dominance, and age of acquisition for 4,300 dutch words. Behavior research methods, 45(1):169–177. ↗
  29. Emiko J Muraki, Summer Abdalla, Marc Brysbaert, and Penny M Pexman. 2023a. Concreteness ratings for 62,000 english multiword expressions. Behavior research methods, 55(5):2522—2531. ↗
  30. Emiko J Muraki, Summer Abdalla, Marc Brysbaert, and Penny M Pexman. 2023b. Concreteness ratings for 62,000 english multiword expressions. Behavior research methods, 55(5):2522–2531. ↗
  31. Mark Nichter. 2010. Idioms of distress revisited. Culture, Medicine, and Psychiatry, 34(2):401– 416. ↗
  32. Carlos Ramisch. 2015. Multiword expressions acquisition. A Generic and Open Framework. Cham: Springer International Publishing. ↗
  33. Aryan Rastogi, Liu Qian, and Erik Cambria. 2022. Stress detection from social media articles: New dataset benchmark and analytical study. In 2022 IEEE World Conference of Computational Intelligence (WCCI). find
  34. Ivan A Sag, Timothy Baldwin, Francis Bond, Ann Copestake, and Dan Flickinger. 2002. Multiword expressions: A pain in the neck for nlp. In International conference on intelligent text processing and computational linguistics, pages 1–15. Springer.
  35. Andy P Siddaway, Peter J Taylor, and Alex M Wood. 2018. Reconceptualizing anxiety as a continuum that ranges from high calmness to high anxiety: The joint importance of reducing distress and increasing well-being. Journal of personality and social psychology, 114(2):e1. ↗
  36. Eva Smolka and Sabine Schulte im Walde. 2020. The role of constituents in multiword expressions: An interdisciplinary, cross-lingual perspective (Volume 4). Language Science Press. find
  37. Masahito Takahashi, Toshifumi Tanabe, Jack Halpern, and Kosho Shudo. 2024. A comprehensive japanese mwe lexicon: Jmwel. In Recent Advances in Multiword Units in Machine Translation and Translation Technology, pages 218–242. John Benjamins Publishing Company. ↗
  38. Xiaoyu Tong, Rochelle Choenni, Martha Lewis, and Ekaterina Shutova. 2024. Metaphor understanding challenge dataset for llms. arXiv preprint arXiv:2403.11810. ↗
  39. Elsbeth Turcan and Kathy McKeown. 2019. Dreaddit: A Reddit dataset for stress analysis in social media. In Proceedings of the Tenth International Workshop on Health Text Mining and Information Analysis (LOUHI 2019), pages 97–107, Hong Kong. Association for Computational Linguistics. ↗
  40. Aline Villavicencio, Ann Copestake, Benjamin Waldron, and Fabre Lambeau. 2004. Lexical encoding of mwes. In Proceedings of the Workshop on Multiword Expressions: Integrating Processing, pages 80–87. find
  41. Melissa LH Võ, Markus Conrad, Lars Kuchinke, Karolina Urton, Markus J Hofmann, and Arthur M Jacobs. 2009. The berlin affective word list reloaded (bawl-r). Behavior research methods, 41(2):534–538. ↗
  42. Amy Beth Warriner, Victor Kuperman, and Marc Brysbaert. 2013. Norms of valence, arousal, and dominance for 13,915 English lemmas. Behavior Research Methods, 45(4):1191–1207. ↗
  43. Joseph P Weir. 2005. Quantifying test-retest reliability using the intraclass correlation coefficient and the sem. The Journal of Strength & Conditioning Research, 19(1):231–240. ↗
  44. Víctor Yeste and Paolo Rosso. 2026. Do schwartz higher-order values help sentence-level human value detection? when hard gating hurts. arXiv preprint arXiv:2602.00913. ↗
  45. Ladislav Zgusta. 1967. Multiword lexical units. Word, 23(1-3):578–587. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/mohammad-2026-trial-fire-sleep-like-baby-lexicon/