Elamur Research⌕
☰
home / papers / 2026 / SemEval-2026 Task 2: Predicting Variation in Emotional Valence and Arousal over Time from Ecological Essays

SemEval-2026 Task 2: Predicting Variation in Emotional Valence and Arousal over Time from Ecological Essays

Paper · 2026 · Nikita Soni, H. Andrew Schwartz, Ryan L. Boyd et al.

Summary

This shared task addresses a gap in affect research. Most NLP emotion datasets rely on outside annotators who judge what a social media post seems to express, not what the writer actually felt. The organizers release a longitudinal English dataset from U.S. service-industry workers. Over 14-day phases from 2021 to 2024, participants wrote short "ecological essays" and "feeling words" up to three times daily. With each text, they placed their own mood on an affective circumplex grid, giving valence and arousal scores.

The task has three parts:

  • Subtask 1: assess valence and arousal for each text, for both seen and unseen authors.
  • Subtask 2a: forecast short-term state change in affect.
  • Subtask 2b: forecast longer-term dispositional change in affect.

Most systems fine-tuned transformer encoders. They handled valence assessment well, but arousal remained harder. Forecasting proved difficult: a plain linear baseline using only current affect scores outperformed most submissions. The resource matters because it supports modeling affect as personal, self-reported and changing over time.

Key points

  • Data: 5,285 texts from 182 authors across 7 collection phases (2021–2024). Authors contributed 72.8 texts on average (median 35). Valence ranges 0–4 and arousal 0–2, both self-reported.
  • Participation: over 200 registrations on Codabench, with official submissions from 31 teams (104 members). Of these, 28 teams (90 members) wrote system description papers.
  • Metrics: Subtask 1 uses a composite Pearson correlation that combines between-author and within-author scores. Subtasks 2a and 2b use Pearson correlation. Rankings average the valence and arousal results.
  • Methods: RoBERTa and DeBERTa regression models were common. Several teams, such as UIT and Ajman University, used a Concordance Correlation Coefficient loss to keep arousal predictions from collapsing to near-constant values.
  • Findings:
    • Feeling words carried stronger signal than essays, especially for arousal.
    • On unseen authors, valence scores dropped slightly while arousal scores often rose.
    • Good per-text assessment did not translate into good forecasting.

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

🎓Stanford UniversityStanford · United States434 works · 314,419 citations🎓University of PennsylvaniaPhiladelphia · United States221 works · 63,234 citations🎓Cornell UniversityIthaca · United States120 works · 59,940 citations🏛️National Research Council CanadaOttawa · Canada59 works · 21,677 citations🎓The University of Texas at DallasRichardson · United States61 works · 5,614 citations🎓Stony Brook UniversityStony Brook · United States43 works · 2,571 citations🎓Vanderbilt UniversityNashville · United States14 works · 1,413 citations

Also on the byline: OsloMet – Oslo Metropolitan University.

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksPsychology · 1,552 worksExperimental and Cognitive Psychology · 870 worksSocial Psychology · 306 works

🧭 Topics 1Research subjects and languages the work is about.Sentiment & emotion · 972 works

Cite

Cite as Nikita Soni, H. Andrew Schwartz, Ryan L. Boyd, Phi Long Bui, Syeda Mahwish, August Håkan Nilsson, Adithya V Ganesan, Lyle Ungar, Niranjan Balasubramanian and Saif M. Mohammad. 2026. SemEval-2026 Task 2: Predicting Variation in Emotional Valence and Arousal over Time from Ecological Essays. Proceedings of the 20th International Workshop on Semantic Evaluation (2026).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.semeval-1.451

@inproceedings{soni2026semeval,
  title = {{SemEval-2026 Task 2: Predicting Variation in Emotional Valence and Arousal over Time from Ecological Essays}},
  author = {Nikita Soni, H. Andrew Schwartz, Ryan L. Boyd, Phi Long Bui, Syeda Mahwish, August Håkan Nilsson, Adithya V Ganesan, Lyle Ungar, Niranjan Balasubramanian and Saif M. Mohammad},
  booktitle = {Proceedings of the 20th International Workshop on Semantic Evaluation (2026)},
  year = {2026},
  doi = {10.18653/v1/2026.semeval-1.451},
  url = {https://doi.org/10.18653/v1/2026.semeval-1.451}
}

References

55 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Omar AlZoubi, Saja Khaled Tawalbeh, and AL-Smadi Mohammad. 2022. Affect detection from arabic tweets using ensemble and deep learning techniques. Journal of King Saud University-Computer and Information Sciences, 34(6):2529–2539. find
  2. Troy Arthur, Aidan Kelley, and Sierra Reschke. 2026. NLPGroup8 at SemEval-2026 Task 2: Diverse Ensembles and Hierarchical Transformers for Emotional State Prediction. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  3. Nabiha Asghar. 2016. Yelp dataset challenge: Review rating prediction. arXiv preprint arXiv:1605.05362. — [arXiv:1605.05362](https://arxiv.org/abs/1605.05362) find
  4. Abdessamad Benlahbib, Zouhir Essalmani, Achraf Boumhidi, Anass Fahfouh, and Hamza Alami. 2026. find
  5. NLP-FSDM at SemEval-2026 Task 2: Temporal Smoothing and CCC-MAE Optimization for Balanced Longitudinal Affect Assessment. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  6. Fumika Beppu, Hiroki Takushima, Aiswariya Kumar Manoj, Daichi Yamaga, Yuki Shibata, and Takayuki Hori. 2026. CuriosAI at SemEval-2026 Task 2: Predicting Emotion using RoBERTa-large model. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  7. Niall Bolger, Angelina Davis, and Eshkol Rafaeli. 2003a. Diary methods: Capturing life as it is lived. Annual Review of Psychology, 54:579–616. find
  8. Niall Bolger, Angelina Davis, and Eshkol Rafaeli. 2003b. Diary methods: Capturing life as it is lived. Annual review of psychology, 54(1):579–616. find
  9. John T. Cacioppo and Wendi L. Gardner. 1999. Emotion. Annual Review of Psychology, 50(1):191–214. find
  10. Jiyu Chen, necva bölücü, Sarvnaz Karimi, Diego Molla, and Cecile L. Paris. 2026. CSIRO-LT at SemEval-2026 Task 2: In-the-Wild Valence and Arousal Forecasting on Ecological Text Time Series. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  11. G. Coppersmith, C. Hilland, O. Frieder, and R. Leary. 2017. Scalable mental health analysis in the clinical whitespace via natural language processing. In 2017 IEEE EMBS International Conference on Biomedical Health Informatics (BHI), pages 393–396. find
  12. E. Diener. 2000. Subjective well-being: The science of happiness and a proposal for a national index. The American Psychologist, 55(1):34–43. find
  13. Nam Dinh. 2026. One and Only at SemEval-2026 Task 2: Evaluating Zero-Shot Autonomous LLM Agents and Heuristic Proxies in Ecological Affect Forecasting. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  14. Johannes C Eichstaedt, H. Andrew Schwartz, Margaret L Kern, Gregory Park, Darwin R Labarthe, Raina M Merchant, Sneha Jha, Megha Agrawal, Lukasz A Dziurzynski, Maarten Sap, Christopher Weeg, Emily E Larson, Lyle H Ungar, and Martin E P Seligman. 2015. Psychological language on Twitter predicts county… find
  15. Adithya V Ganesan, Vasudha Varadarajan, Juhi Mittal, Shashanka Subrahmanya, Matthew Matero, Nikita Soni, Sharath Chandra Guntuku, Johannes Eichstaedt, and H Andrew Schwartz. 2022. Wwbp-sqt-lite: Multi-level models and difference embeddings for moments of change identification in mental health forums… find
  16. Somdev Ganguli, Vibhan Dutta, Romit Datta, Amit Barman, and Sudip Kumar Naskar. 2026. LexMachina at SemEval-2026 Task 2: Predicting Variation in Emotional Valence and Arousal over Time from Ecological Essays. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  17. Debanjan Ghosh, Kai North, Ekaterina Kochmar, Mamoru Komachi, and Marcos Zampieri, editors. 2026. Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). Association for Computational Linguistics, San Diego, United States. find
  18. Zhuojun Gu, Katarina Kjell, H. Andrew Schwartz, and Oscar Kjell. 2025. Natural language response formats for assessing depression and worry with large language models: A sequential evaluation with model pre-registration. Assessment. find
  19. Hatice Gunes and Maja Pantic. 2010. Automatic, dimensional and continuous emotion recognition. International Journal of Synthetic Emotions (IJSE), 1(1):68– 99. find
  20. Duc Ho, Khanh Dao Duy Bui, Daniela Teodorescu, and Grzegorz Kondrak. 2026. UAlberta at SemEval-2026 Task 2: Temporal Fusion Models for Predicting Affect Over Time. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  21. Darya Hryhoryeva, Amaia Zurinaga, Hamidreza Jamalabadi, and Iryna Gurevych. 2026. UKP_Psycontrol at SemEval-2026 Task 2: Modeling Valence and Arousal Dynamics from Text. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  22. Haseebullah Jumakhan, Soud Assad, Seyed Abdullah, and Mahmoud Al-Ayyoub. 2026. Ajman University at SemEval-2026 Task 2: Overcoming Scale Collapse in Temporal Emotion Modeling via Residual Learning. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  23. Diya Satish Kumar and Om Sujal Joshi. 2026. EcoAffectTrack at SemEval-2026 Task 2: A Hierarchical DeBERTa-Transformer Framework with CCC Optimization for Longitudinal Affect Modeling. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  24. Xin Lan, Jin Wang, and Xuejie Zhang. 2026. YNU-HPCC at SemEval-2026 Task 2: Contrastive Calibration and Temporal Modeling for Continuous Valence-Arousal Prediction. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  25. Huy Le, Truong Thien Phu, Trung Tran, Nga Nguyen, and Monojit Choudhury. 2026. VAP-GameController at SemEval-2026 Task 2: Lexical-based and Emotion-Aware Approaches for Longti-tudinal Emotion Prediction. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  26. Jing-Jun Lin. 2026. cclin at SemEval-2026 Task 2 : SLM-Enhanced Lightweight Multi-BERT Ensemble for Longitudinal Affect Assessment. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  27. Andrea Lolli, Chiara Lunazzi, Riccardo Coppola, and Flavio Giobergia. 2026. ES4MLL at SemEval-2026 Task 2: Set Attention Aggregation and Recurrent Temporal Modeling for Longitudinal Affect Prediction. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  28. Siddharth Mangalik, Johannes C Eichstaedt, Salvatore Giorgi, Jihu Mun, Farhan Ahmed, Gilvir Gill, Adithya V. Ganesan, Shashanka Subrahmanya, Nikita Soni, Sean AP Clouston, and 1 others. 2024. Robust language-based mental health assessments in time and space through social media. NPJ Digital Medicine… find
  29. Matthew Matero, Nikita Soni, Niranjan Balasubramanian, and H Andrew Schwartz. 2021. Melt: Message-level transformer with masked document representations as pre-training for stance detection. In Findings of the association for computational linguistics: EMNLP 2021, pages 2959–2966. find
  30. Julian John McAuley and Jure Leskovec. 2013. From amateurs to connoisseurs: modeling the evolution of user expertise through online reviews. In Proceedings of the 22nd international conference on World Wide Web, pages 897–908. find
  31. Saif Mohammad. 2022. Ethics sheets for AI tasks. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8368–8379, Dublin, Ireland. Association for Computational Linguistics. find
  32. Saif Mohammad, Felipe Bravo-Marquez, Mohammad Salameh, and Svetlana Kiritchenko. 2018. SemEval-2018 task 1: Affect in tweets. In Proceedings of the 12th International Workshop on Semantic Evaluation, pages 1–17, New Orleans, Louisiana. Association for Computational Linguistics.
  33. Melissa Andrea Moreno, Juan Carlos Martinez Santos, and Edwin Puertas. 2026. VerbaNex AI at SemEval-2026 Task 2: DeBERTa for Longitudinal Valence and Arousal Prediction. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  34. Supriya S. Nadiger, SUNIL SAUMYA, Rahul Pujari, VEERESH S. HIREMATH, Kiran A. Chikaraddi, and Anoop U. Kadkol. 2026. Momentum at SemEval-2026 Task 2: LongVA-RoBERTa, a transformer-Based Longitudinal Valence and Arousal Modeling. In Proceedings of the 20th International Workshop on Semantic Evaluatio…
  35. LAM ANH NGUYEN. 2026. lamanhnguyen at SemEval-2026 Task 2: Uncovering Lexical Bias and Momentum Lag in Longitudinal Emotion Prediction using Multi-task DeBERTa. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  36. August Håkan Nilsson, Ryan L. Boyd, Adithya V. Ganesan, Oscar Kjell, Syeda Mahwish, Haitao Huang, Richard N. Rosenthal, Lyle Ungar, and H. Andrew Schwartz. 2025. Language-based assessments for experienced well-being: Accuracy and external validity across behaviors, traits, and states. Manuscript und… find
  37. Sadia Noor and Mehwish Fatima. 2026. Emo-tica at SemEval-2026 Task 2: Trait?State Affect Forecaster for Longitudinal Valence and Arousal. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  38. Cetta Reswara Parahita. 2026. Cherish at SemEval-2026 Task 2: Enhancing RoBERTa-Based Models for Emotional Valence and Arousal Prediction in Ecological Essays with Personalized PLoRA and Temporal Embeddings. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  39. Son The Phuong, My Thuy Tra Ngo, Tri Minh Dao, and Duc-Vu Nguyen. 2026. CITD@UIT at SemEval-2026 Task 2: Temporal Mixture-of-Experts for Longitudinal Valence and Arousal Prediction from Ecological Essays. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  40. Jonathan Posner, James A Russell, and Bradley S Peterson. 2005. The circumplex model of affect: An integrative approach to affective neuroscience, cognitive development, and psychopathology. Developmental Psychopathology, 17(3):715–734. find
  41. Daniel Preoţiuc-Pietro, H. Andrew Schwartz, Gregory Park, Johannes Eichstaedt, Margaret Kern, Lyle Ungar, and Elisabeth Shulman. 2016. Modelling valence and arousal in Facebook posts. In Proceedings of the 7th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis… find
  42. Harsh Rathva. 2026. "AGI” Team at SemEval-2026 Task 2: Predicting Variation in Emotional Valence and Arousal over Time from Ecological Essays. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  43. James A. Russell and Lisa Feldman Barrett. 1999. Core affect, prototypical emotional episodes, and other things called emotion: Dissecting the elephant. Journal of Personality and Social Psychology, 76(5):805– 819. Place: US. find
  44. Araj Shah, Utsav Shah, and Saurav K. Aryal. 2026. AI4PC-Howard University at SemEval-2026 Task 2: Fine-Tuning DistilBERT, DeBERTa and Modern-BERT for Valence?Arousal Prediction and Change Estimation. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026). find
  45. Khushboo Singh, Vasudha Varadarajan, Adithya V. Ganesan, August Håkan Nilsson, Nikita Soni, Syeda Mahwish, Pranav Chitale, Ryan L. Boyd, Lyle Ungar, Richard N. Rosenthal, and H. Andrew Schwartz. 2025. Systematic evaluation of autoencoding and large language model representations for capturing author… find
  46. Rajalakshmi Sivanaiah, Angel Deborah S, Krishna Varun R, and Krishnaraj N. 2026. Draken at SemEval-2026 Task 2: Frozen BERT Embeddings with Ridge Regression for Predicting Emotional Valence and Arousal. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  47. Nikita Soni, Niranjan Balasubramanian, H Andrew Schwartz, and Dirk Hovy. 2024a. Comparing pre-trained human language models: Is it better with human context as groups, individual traits, or both? In Proceedings of the 14th Workshop on Computational Approaches to Subjectivity, Sentiment, & Social Med… find
  48. Nikita Soni, Pranav Chitale, Khushboo Singh, Niranjan Balasubramanian, and H. Andrew Schwartz. 2025a. Evaluation of LLMs-based hidden states as author representations for psychological human-centered NLP tasks. In Findings of the Association for Computational Linguistics: NAACL 2025, pages 7673–7682… find
  49. Nikita Soni, Dhruv Vijay Kunjadiya, Pratham Piyush Shah, Dikshya Mohanty, H Andrew Schwartz, and Niranjan Balasubramanian. 2026. Addressing the ecological fallacy in larger lms with human context. arXiv preprint arXiv:2603.05928.
  50. Nikita Soni, Matthew Matero, Niranjan Balasubramanian, and H. Andrew Schwartz. 2022. Human language modeling. In Findings of the Association for Computational Linguistics: ACL 2022, pages 622– 636, Dublin, Ireland. Association for Computational Linguistics. find
  51. Nikita Soni, August Håkan Nilsson, Syeda Mahwish, Vasudha Varadarajan, H Andrew Schwartz, and Ryan L Boyd. 2025b. Who we are, where we are: Mental health at the intersection of person, situation, and large language models. In Proceedings of the 10th Workshop on Computational Linguistics and Clinical… find
  52. Nikita Soni, H. Andrew Schwartz, João Sedoc, and Niranjan Balasubramanian. 2024b. Large human language models: A need and the challenges. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long… find
  53. Eleale Nusi Tee. 2026. Khaleesiyali at SemEval-2026 Task 2: Lexicon-Augmented RoBERTa for Valence?Arousal Regression on Ecological Essays. In Proceedings of the 20th International Workshop on Semantic Evaluation (SemEval-2026).
  54. Shrika SP Thota, Lakshmi Priya Swaminatha Rao, Shiv-aanee SK, Thirumurugan RA, Vishal Muralidharan, and Dhannya Santhakumari Madhavan. 2026. AFourP at SemEval-2026 Task 2: Predicting Variation in Emotional Valence and Arousal over Time from Ecological Essays. In Proceedings of the 20th International…
  55. Debra Trampe, Jordi Quoidbach, and Maxime Taquet. 2015. Emotions in everyday life. PLOS ONE, 10(12):1–15. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/soni-2026-semeval-2026-task-2-predicting/