EACL 2026 Main Conference

Please visit this section regularly for any updates and important information. Enjoy the Conference!

Live Session Recordings Are Now Available!

Live session recordings, including our yesterday's keynotes, are now available for viewing on Underline.

Enjoy watching, and don’t forget - you can still reach out to paper authors using the chat feature on the paper page on Underline, or via the messaging feature (top right corner).

Live session recordings, including our yesterday's keynote and Business Meeting, are now available for viewing on Underline: https://underline.io/events/519/sessions?searchGroup=all

Enjoy watching, and don’t forget - you can still reach out to paper authors using the chat feature on the paper page on Underline, or via the messaging feature (top right corner).

Live session recordings, including our yesterday's keynote, are now available for viewing on Underline: https://underline.io/events/519/sessions?searchGroup=all

Enjoy watching, and don’t forget - you can still reach out to paper authors using the chat feature on the paper page on Underline, or via the messaging feature (top right corner).

Loved a talk, poster, or paper? Head to the Voting tab on the lecture page on Underline and cast your vote! Your picks decide the Audience Award, announced live at the Closing Session. Every vote counts - show support to the authors who inspired you!

*The system will exclude any votes cast by authors on their own papers.

Join us in congratulating the Best Demo and Best SRW Awarded Papers! Attend their presentations on-site or remotely on Underline. You can also leave comments or vote for your favorite papers there.

Best System Demonstration Award

• SciTrue: Evidence-Grounded Claim Verification in Science

SciTrue is recognized for combining a strong scientific contribution with a highly polished and impactful system demonstration, addressing the timely challenge of evidence-grounded verification of scientific claims.

SRW Best Paper Award

• Probabilistic Bilingual Subword Segmentation with Latent Subword Alignment

This paper is recognised for its high-quality research, novelty, and the high praise it received from reviewers. It addresses an important problem in sub-word tokenisation and presents a novel yet intuitive algorithm to improve consistency in subword segmentation between source and target languages in bilingual machine translation settings.

Join us in congratulating the Awarded Papers! Attend their presentations on-site or remotely on Underline. You can also leave comments or vote for your favorite papers there.

Best Paper Award

• Humans and transformer LMs: Abstraction drives language learning

A strong, original, and timely paper that addresses an important theoretical question at the intersection of linguistics, psychology, and language modeling. Its innovative use of behavioral-style testing on GPT-2 to investigate abstraction versus exemplar learning is methodologically clever and offers meaningful, potentially influential insights.

Outstanding Paper Award

• Mind the Gap: Benchmarking LLM Uncertainty and Calibration with Specialty-Aware Clinical QA and Reasoning-Based Behavioural Features

This paper tackles the societally critical problem of quantifying predictive uncertainty in LLMs for clinical question answering, benchmarking multiple uncertainty estimation methods across 10 models, 2 datasets, and 11 medical specialties. It introduces a novel annotation scheme and proposes behavioral signal-based uncertainty techniques, offering clear methodological contributions and insightful findings with relevance beyond the clinical domain.

• Emotionally Charged, Logically Blurred: AI-driven Emotional Framing Impairs Human Fallacy Detection

This paper makes a significant social impact contribution by showing that AI-driven emotional framing can measurably impair human fallacy detection, highlighting concrete risks related to misinformation and persuasive manipulation with LLMs. Its combination of controlled LLM interventions and a large-scale human study is methodologically rigorous and broadly relevant

• DivMerge: A divergence-based model merging method for multi-tasking

This paper presents a theoretically grounded, reference-free framework for model merging based on minimizing Jensen–Shannon divergence, offering a practical and resource-efficient alternative to traditional fine-tuning with strong empirical results across diverse tasks.

Evaluation and Model Insight Award

• Are All Prompt Components Value-Neutral? Understanding the Heterogeneous Adversarial Robustness of Dissected Prompt in Large Language Models

This paper provides rigorous and actionable insights into LLM robustness by systematically decomposing prompts and revealing how different components contribute unequally to adversarial vulnerability, offering practical guidance for safer deployment. Its broad evaluation and widely applicable findings make it a solid and award-worthy contribution.

Resource Paper Award

• HarfoSokhan: A Comprehensive Parallel Dataset for Transitions between Persian Colloquial and Formal Variations

This paper makes an original and socially valuable contribution by introducing linguistically informed resources for low-resource Persian, notably a high-quality parallel corpus of colloquial-formal Persian that is likely to have lasting impact.

Social Impact Award

• DIVINE: Coordinating Multimodal Disentangled Representations for Oro-Facial Neurological Disorder Assessment

This paper proposes DIVINE, a multimodal framework that disentangles speech and facial video representations to support more objective and consistent assessment of oro-facial neurological disorders, addressing known variability in subjective ratings. By producing interpretable, scalable assessment signals, the approach has strong potential to broaden access to screening and support more equitable monitoring workflows for patients with ALS and stroke.

Dear Authors, congratulations on behalf of Underline!

You can upload or update your slides using this link.

Whether you are joining us in Morocco or online, here are a few final items to check off before the event:

• Log into Underline, or create your Underline account, to access the detailed schedule, view content, and find links for joining remotely.
• Build your personal agenda so you don’t miss any presentations.
• Start voting for the papers that sparked your interest.
• Review the user guides to ensure a smooth experience.

Still have questions? Please visit our Help Desk

We look forward to your participation!

We are pleased to announce that the EACL 2026 Proceedings are now available. Please visit each paper’s lecture page to find a direct link to the EACL 2026 Anthology.

If you encounter any issues, please visit our Help Desk section.

As we get closer to the conference, we encourage all participants to ensure they are fully prepared to take their part in the conference - whether you’re joining us as an attendee, presenter, or session chair.

To help you prepare, we’ve put together User Guides that answer common questions and provide helpful tips for making the most of your EACL 2026 experience.
Be sure to read the Before You Go section for useful information!

We look forward to seeing you soon in Morocco and online!

• Tuesday, March 24: 10:30 - 19:30
• Wednesday, March 25: 07:30 - 17:00
• Thursday, March 26: 08:30 - 17:00
• Friday, March 27: 08:30 - 16:30
• Saturday, March 28: 08:00 - 16:30
• Sunday, March 29: 08:00 - 16:30