Google at EMNLP 2026
Google at EMNLP 2026
Google is proud to be a Diamond Sponsor for the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026), hosted at the Hungexpo Budapest Congress and Exhibition Centre in Budapest, Hungary from October 24–29, 2026. Researchers across Google, including Google Research and Google DeepMind, are excited to share 31 accepted papers, participate in 11 workshops and 2 tutorials, deliver an industry keynote and panel discussion, and host interactive demo and Q&A sessions throughout the conference.
Attending EMNLP 2026 in person? Stop by the Google booth (#D5 in Hall H) to learn more about how we’re actively pursuing the latest innovations in natural language processing, language modeling, multimodal understanding, agentic systems, and responsible AI. For real-time updates on conference activities, including live booth demos and careers sessions, follow @GoogleResearch on X and visit the Google Research LinkedIn page.
Continue below to learn more about how Google researchers are engaged at EMNLP 2026 (Google affiliations highlighted in bold, * Indicates work done while at Google).
Please note that all scheduled session times are listed in Central European Time (CET).
Google Booth Activities
All demo and Q&A sessions will be hosted at the Google booth (#D5) in Hall H.
-
Sun, Oct 25 | 3:30PM — 4:00PM
Mapping the Gaps: A Dashboard of Cultural Queries and Failures in Real-world AI UsagePresenters: Vinodkumar Prabhakaran, Piyawat L. Kumjorn, Sunipa Dev
We introduce a comprehensive evaluation framework to assess Cultural Intelligence in AI as a suite of latent capabilities spanning diverse domains, and we demonstrate operationalizing these capabilities through a series of measurable indicators designed for reliability and extensibility. We will demonstrate a real-time demo that applies this framework to real-world user queries and highlights failures.
-
Mon, Oct 26 | 1:00PM — 2:00PM
Sign Language to Text (SL2T)Presenters: Dan Garrette, Miloš Stanojević
SL2T is a massively multilingual sign-language-to-text translation model that marks a breakthrough in quality and generality. With it, we are bringing sign language AI out of the lab and into consumer products for the first time: SL2T powers sign-to-text dictation in Gboard and Live Transcribe on Pixel 11, starting with American Sign Language (ASL) to English.
-
Mon, Oct 26 | 4:15PM — 4:45PM
Recruiting Q&APresenters: Sunipa Dev, Nima Kalantari
Interactive Recruiting Q&A session covering research career pathways, student programs, and opportunities at Google.
Industry Keynote & Panel
-
Mon, Oct 26 | 9:00AM — 10:30AM (Session 5 - Industry 3)
Keynote: Toward a Science of Constitutional Alignment: Positive Alignment for Value-Based AgentsSpeaker: Verena Rieser
As AI systems become increasingly autonomous, we need a rigorous new science to ensure that more capable models make safe, reliable choices even when we aren’t looking. A promising direction is positive alignment: anchoring models to an intrinsic “north star” of high-level normative values rather than brittle rubrics or feedback signals that fail to prevent reward hacking or unsafe behaviour out-of-distribution While constitutions offer a powerful vehicle to define these foundational principles, existing approaches rely on ad-hoc heuristics. This talk makes the case for constitutional alignment as an empirical, systematic science, outlining the normative reasoning paradigms, evaluation methodologies, and governance structures required to navigate emergent capabilities and safely steer the next generation of AI agents.
-
Tue, Oct 27 | 11:00AM — 12:30PM (Session 11 - Industry 6)
Panel: AI Agents in the Real World: Capability, Trust, Values, and DeploymentPanelist: Verena Rieser
Large language models are rapidly evolving from systems that generate responses into agents that pursue goals, use tools, maintain memory, interact with people and other agents, and act across extended periods of time. Yet greater agency brings more than greater capability: it introduces new questions about reliability, control, accountability, values, and the appropriate division of labour between humans and machines. An agent may succeed on a benchmark while failing under changing, ambiguous, or adversarial real-world conditions; it may complete a task while violating the user’s underlying intent; and it may appear collaborative while manipulating beliefs or merely agreeing with its user. This panel brings together perspectives from research and industry to examine what meaningful progress in AI agents should look like, how it should be evaluated, and what technical, institutional, and societal foundations are needed before increasingly autonomous agents can be deployed responsibly at scale.
Orals
-
Sun, Oct 25 | 11:00AM — 12:30PM, Room F6
From Visual Cues to Spoken Narration: Rethinking Audio DescriptionAkshita Gupta, Aditya Arora, Federico Tombari, Marcus Rohrbach, Anna Rohrbach
-
Tue, Oct 27 | 11:00AM — 12:30PM, Room G13-14
When LLMs Benchmark Themselves: Deconstructing Self-Bias in Automated EvaluationWenda Xu, Sweta Agrawal, Vilém Zouhar, Markus Freitag, Daniel Deutsch
Demos & tutorials
-
Wed, Oct 28 | 2:00PM — 5:30PM, Room P1
Multilingual Multicultural LLMsTutorial organizer: Vinodkumar Prabhakaran
-
Wed, Oct 28 | 9:00AM — 12:30PM, Room P2
Tokenization for Large Language ModelsTutorial organizer: Elizabeth Salesky
-
Mon, Oct 26 | 11:00AM — 12:30PM, Hall H
AgentDebugX: An Open-Source Toolkit for Failure Observability, Attribution, and Recovery in LLM AgentsDemo presenters: Kunlun Zhu, Xuyan Ye, Zhiguang Han, Yuchen Zhao, Bingxuan Li, Weijia Zhang, Muxin Tian, Xiangru Tang, Pan Lu, James Zou, Jiaxuan You, Heng Ji
TACL papers
Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders
Carolina Zheng, Nicolas Beltran-Velez, Sweta Karlekar, Claudia Shi, Achille Nazaret, Asif Mallik, Amir Feder, David M. Blei
Papers
DeepSEA: A Multilingual Evaluation Benchmark for Southeast Asia
Juizhou Han*, Abhishek Bijay Mishra, Trevor Cohn, Shubham Mittal
Dual-Modality Multi-Stage Adversarial Safety Training: Robustifying Multimodal Web Agents Against Cross-Modal Attacks
Haoyu Liu*, Dingcheng Li,Lukas Rutishauser, Zeyu Zheng
From Cause to Effect: Generating Intermediate Event Chains in Videos
Yuxiang Liu*, Jiaming Luo, Eleftheria Briakou, Colin Cherry
How Coding Agents Fail Their Users: A Large-Scale Analysis of Developer-Agent Misalignment in 20,574 Real-World Sessions
Ningzhi Tang, Chaoran Chen, Gelei Xu, Yiyu Shi, Yu Huang, Collin McMillan, Tao Dong, Toby Jia-Jun Li
LifeAgentBench: Benchmarking LLMs for Long-Horizon, Cross-Dimensional Lifestyle Health Reasoning
Ye Tian, Zihao Wang, Onat Gungor, Xiaoran Fan, Tajana Rosing
NaijaS2ST: A Multi-Accent Benchmark for Speech-to-Speech Translation in Low-Resource Nigerian Languages
Marie Maltais, Yejin Jeon, Min Ma, Shamsuddeen Hassan Muhammad, Idris Abdulmumin, Maryam Ibrahim Mukhtar, Daud Abolade, Joel Okepefi, Johnson Sewedo, David Ifeoluwa Adelani
POLARIS: Guiding Small Models to Write Long Stories
Rishanth Rajendhran, Jenna Russell, Mohit Iyyer, John Wieting
SCAR: Shapley Credit Assignment for More Efficient RLHF
Meng Cao, Shuyuan Zhang, Xiao-Wen Chang, Doina Precup
SOPBench: Evaluating Language Agents at Following Standard Operating Procedures and Constraints
Zekun Li, Shinda Huang, Jiangtian Wang, Nathan Zhang, Antonis Antoniades, Wenyue Hua, Kaijie Zhu, Sirui Zeng, Chi Wang, William Yang Wang, Xifeng Yan
TaskPress: Query-Agnostic KV Cache Compression via Task-Guided Pruning
Wonpyo Park, Seung-won Hwang
Findings
AgenticTagger: Generating Structured Item Representation for Recommendation with LLM Agents
Zhouhang Xie*, Bo Peng, Zhankui He, Ziqi Chen, Alice Han, Isabella Ye, Benjamin Coleman, Noveen Sachdeva, Fernando Pereira, Julian McAuley, Wang-Cheng Kang, Derek Zhiyuan Cheng, Beidou Wang, Randolph Brown
Are Multilingual Models Improving? Revisiting Cross-Lingual Transfer
Prasoon Bajpai, Eleftheria Briakou, Colin Cherry, Preethi Jyothi, Vihari Piratla
Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training
Hejian Sang, Yuanda Xu, Zhengze Zhou, Ran He, Zhipeng Wang, Alborz Geramifard
CANVAS: Continuity-Aware Narrative Generation via Visual Agentic Storyboarding
Ishani Mondal*, Yiwen Song, Mihir Parmar, Palash Goyal, Jordan Boyd-Graber, Tomas Pfister, Yale Song
CoReflect: A Reflective Co-Evolution Framework for Improving Conversational Evaluation
Yunzhe Li*, Richie Yueqi Feng, Tianxin Wei*, Chin-Chia Hsu
Cross-Lingual Exploration for Parametric Knowledge
Elisha Diskind, Itamar Trainin, Uri Shaham, Leshem Choshen, Idan Szpektor, Omri Abend
Detecting Post-Generation Edits to Watermarked LLM Outputs via Error-Correcting Codes
Liyan Xie, Muhammad Siddeek, Mohamed Seif, Andrea J. Goldsmith, Mengdi Wang
FinanceHarness: Autonomous Financial Deep Research Framework
Yijia Xiao*, Rujun Han, Yanfei Chen, Zifeng Wang, Ke Jiang, Zhongying CuiZhu, Vishy Tirumalashetty, Wei Wang, Burak Gokturk, Tomas Pfister, Chen-Yu Lee
FORGE: Forensic Reasoning with Grounded Evidence
Rohit Kundu, Shan Jia, Vishal Mohanty, Athula Balachandran, Amit K. Roy-Chowdhury
GroupDPO: Memory-Efficient Group-Wise Direct Preference Optimization
Jixuan Leng*, Si Si, Hsiang-Fu Yu, Vinod Raman, Inderjit S. Dhillon
How Calibration Content Shapes Attention-Based Reranking
Petros Karypis*, Hossein Rajaby Faghihi, Peter Chen, Rui Zhu, Noveen Sachdeva, Yan Zhu, Julian McAuley
Judging with Confidence: Calibrating Autoraters to Preference Distributions
Zhuohang Li*, Xiaowei Li,Chengyu Huang*, Guowang Li, Katayoon Goshvadi, Bo Dai, Dale Schuurmans, Paul Zhou, Hamid Palangi, Yiwen Song, Palash Goyal,Murat Kantarcioglu, Bradley A. Malin, Yuan Xue*
Large Reasoning Models Struggle to Transfer Parametric Knowledge Across Scripts
Lucas Bandarkar*, Alan Ansell, Trevor Cohn
MVEB: Massive Video Embedding Benchmark
Adnan El Assadi, Roman Solomatin, Isaac Chung, Chenghao Xiao, Deep Shah, Manan Dey, Shriya Sudhakar, Zacharie Bugaud, Wissam Siblini, Ayush Sunil Munot, Yashwanth Devavarapu, Rakshitha Ireddi, Michelle Yang, Márton Kardos, Niklas Muennighoff, Kenneth Enevoldsen
What Do Reward Models Memorize?
Ivo Verhoeven, Pushkar Mishra, Ekaterina Shutova
What Limits Us? Analyzing Self-Reported Limitations in NLP Research
Tawan Thaepprasit, Peeranuth Kehasukcharoen, Ding Wang, Remi Denton, Peerapon Vateekul, Piyawat Lertvittayakumjorn
XLGoBench: Detecting Cross-Lingual Skill Gaps with Algorithmic Tasks
Purvam Jain, Preethi Jyothi, Vihari Piratla, Suvrat Raju
Workshops
Wednesday, October 28
8:30AM — 5:30PM, Room F8
Automated Knowledge Base Construction (AKBC)
Speakers: Alon Halevy, Amir Globerson
Organizer: Andrew McCallum
9:00AM — 5:00PM, Room F3
Multilingual Representation Learning (MRL)
Speaker: Alham Fikri Aji
Organizer: Tyler Chang
9:00AM — 5:30PM, Room G13-14
Natural Language Processing Meets Climate Change (ClimateNLP)
Speaker & Organizer: Markus Leippold
9:00AM — 5:15PM, Room F9-10
Speech and Audio Language Models (SALMA)
Organizer: Sreyan Ghosh
9:00AM — 5:30PM, Room G6
Widening NLP (WiNLP)
Advisory Board: Alham Fikri Aji
Thursday, October 29
9:15AM — 5:30PM, Room F4-5
Accelerating Language Modeling Research with Cognitively Plausible Datasets (BabyLM)
Organizer: Tal Linzen
9:00AM — 5:15PM, Room F3
Document Intelligence and Understanding (DocInsights)
Speaker: Julian Eisenschlos
8:30AM — 5:30PM, Room P2
Grounding Language Models: Learning Faithfully and Efficiently (GLM)
Panelist: Krisztian Balog
9:00AM — 5:30PM, Room F7
Insights from Negative Results in NLP (Insights)
Organizer: Arjun Akula, Adrian Benton
9:00AM — 5:00PM, Room G13-14
Pluralistic AI & NLP: Diversity-Aware, Sociotechnical and Responsible Alignment (PANDORA)
Speaker: Vinodkumar Prabhakaran
9:00AM — 5:30PM, Room F6
Uncertainty-Aware NLP (UncertaiNLP)
Speaker: Gintare Karolina Dziugaite
Organizer: Jonathan Berant
Board & Organizing Committee
-
Sunipa Dev
- Program Chair
-
Sweta Agrawal
- Demonstration Chair
-
Raj Dabre
- Website Chair