China

Recent advances in Large Language Models (LLMs) have significantly enhanced their capabilities, highlighting the need for comprehensive evaluation frameworks that extend beyond task-specific benchmarks. However, existing benchmarks often focus on isolated abilities, lacking a holistic framework for assessing LLM capabilities. To address this gap, we propose the Cognition-Domain-Task (CDT) framework, which comprehensively measures a model&#39;s capabilities across three dimensions. We expand the scope of model capability definitions at the cognitive level by incorporating the Cattell-Horn-Carroll cognitive theory, refining the categorization of model capabilities. We apply CDT in two directions: dataset capability evaluation and data selection. Experiments show that our capability metrics correlate well with downstream performance and can support effective dataset analysis and construction. The experiments on data selection also show significant improvements in both general and specific benchmarks, achieving scores of 44.3 and 45.4, with an increase of 1.6 and 2.2 points over the baselines, respectively. These results validate the effectiveness and practicality of CDT. Source code and models are available at https://github.com/Alessa-mo/CDT.

EMNLP 2025

CDT: A Comprehensive Capability Framework for Large Language Models Across Cognition, Domain, and Task

capability framework

cognitive theory

large language models

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) 
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

Registration Required

ACL 2026

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

Recent advances in Large Language Models (LLMs) have significantly enhanced their capabilities, highlighting the need for comprehensive evaluation frameworks that extend beyond task-specific benchmarks. However, existing benchmarks often focus on isolated abilities, lacking a holistic framework for assessing LLM capabilities. To address this gap, we propose the Cognition-Domain-Task (CDT) framework, which comprehensively measures a model's capabilities across three dimensions. We expand the scope of model capability definitions at the cognitive level by incorporating the Cattell-Horn-Carroll cognitive theory, refining the categorization of model capabilities. We apply CDT in two directions: dataset capability evaluation and data selection. Experiments show that our capability metrics correlate well with downstream performance and can support effective dataset analysis and construction. The experiments on data selection also show significant improvements in both general and specific benchmarks, achieving scores of 44.3 and 45.4, with an increase of 1.6 and 2.2 points over the baselines, respectively. These results validate the effectiveness and practicality of CDT. Source code and models are available at https://github.com/Alessa-mo/CDT.

poster

## Welcome!
"I am excited to welcome you to this year’s edition of the Conference on Empirical Methods in Natural Language Processing! Importantly, it marks the 30th edition of EMNLP. With over 8,000 submissions, more than 3,000 accepted papers, and thousands of attendees, we have come a long way from that first
workshop, which had 14 accepted papers. As the field looks ahead, Suzhou is the fitting location for celebrating this milestone: rooted in a long literary tradition, yet modern and forward-looking, and home to a large share of the NLP community." 

*Message from the General Chair, Dirk Hovy*

Celebrate 30 Years of EMNLP! 
EMNLP 2025 will be held in Suzhou, China from November 5th to November 9th, 2025.

The robustness and security of Large Language Models (LLMs) face increasing threats, especially in multilingual settings. A notable vulnerability is “jailbreaking” via translating harmful queries into rare or underrepresented languages, which often bypasses existing safeguards. In this work, we propose Multilingual Collaborative Defense (MCD), a novel learning method that optimizes a continuous soft safety prompt automatically to facilitate multilingual safeguarding of LLMs. MCD organically leverages collaborative signals from multiple languages by rotating each as the training “center,” allowing auxiliary languages to reinforce safety prompt learning and ensuring cross‑lingual consistency. As a result, MCD improves defense performance across all languages, reduces false refusals, and mitigates safety misalignment caused by corpus imbalance. To evaluate MCD, we construct multilingual versions of jailbreak benchmarks such as MaliciousInstruct and AdvBench, including zero-shot languages, to assess language transferability. Experiments show that MCD outperforms prior approaches in multilingual jailbreak defense while exhibiting strong cross-lingual generalization. Our code is available at https://anonymous.4open.science/r/MCD-0519.

Multilingual Collaborative Defense for Large Language Models

Multimodal Large Language Models (MLLMs) have shown substantial capabilities in integrating visual and textual information, yet frequently rely on spurious correlations, undermining their robustness and generalization in complex multimodal reasoning tasks. This paper addresses the critical challenge of superficial correlation bias in MLLMs through a novel causal mediation-based debiasing framework. Specially, we distinguishing core semantics from spurious textual and visual contexts via counterfactual examples to activate training-stage debiasing and employ a Mixture-of-Experts (MoE) architecture with dynamic routing to selectively engages modality-specific debiasing experts. Empirical evaluation on multimodal sarcasm detection and sentiment analysis tasks demonstrates that our framework significantly surpasses unimodal debiasing strategies and existing state-of-the-art models.

Beyond Spurious Signals: Debiasing Multimodal Large Language Models via Counterfactual Inference and Adaptive Expert Routing

This paper addresses the challenge of handling unseen modalities and dynamic modality combinations at test time with our proposed text-centric alignment method. This training-free alignment approach unifies different input modalities into a single semantic text representation by leveraging in-context learning with Large Language Models and uni-modal foundation models. Our method significantly enhances the ability to manage unseen, diverse, and unpredictable modality combinations, making it suitable for both generative and discriminative models to adopt on top. Our extensive experiments primarily evaluate on discriminative tasks, demonstrating that our approach is essential for LLMs to achieve strong modality alignment performance. It also surpasses the limitations of traditional fixed-modality frameworks in embedding representations. This study contributes to the field by offering a flexible and effective solution for real-world applications where modality availability is dynamic and uncertain.

Text-centric Alignment for Bridging Test-time Unseen Modality

Execution failures are common in daily life when individuals perform procedural tasks, such as cooking or handicrafts making. Retrieving relevant procedural documents that align closely with both the content of steps and the overall execution sequence can help correct these failures with fewer modifications. However, existing retrieval methods, which primarily focus on declarative knowledge, often neglect the execution sequence structures inherent in procedural documents. To tackle this challenge, we introduce a new dataset Procedural Questions, and propose a retrieval model Graph-Fusion Procedural Document Retriever (GFPDR) which integrates procedural graphs with document representations. Extensive experiments demonstrate the effectiveness of GFPDR, highlighting its superior performance in procedural document retrieval compared to existing models.

Sequence Structure Aware Retriever for Procedural Document Retrieval: A New Dataset and Baseline

Prior research diverges on language diversity in LLM fine-tuning: Some studies report benefits while others find no advantages. Through controlled fine-tuning experiments across 132 translation directions, we systematically resolve these disparities. We find that expanding language diversity during fine-tuning improves translation quality for both unsupervised and---surprisingly---supervised pairs, despite less diverse models being fine-tuned exclusively on these supervised pairs. However, benefits plateau or decrease beyond a certain diversity threshold. We show that increased language diversity creates more language-agnostic representations. These representational adaptations help explain the improved performance in models fine-tuned with greater diversity.

The Effect of Language Diversity When Fine-Tuning Large Language Models for Translation

In recent years, the need for natural language interfaces to knowledge graphs has become increasingly important since they enable easy and efficient access to the information contained in them. In particular, property graphs (PGs) have seen increased adoption as a means of representing complex structured information. Despite their growing popularity in industry, PGs remain relatively underrepresented in semantic parsing research with a lack of resources for evaluation. To address this gap, we introduce ZOGRASCOPE, a benchmark designed specifically for PGs and queries written in Cypher. Our benchmark includes a diverse set of manually annotated queries of varying complexity and is organized into three partitions: iid, compositional and length. We complement this paper with a set of experiments that test the performance of different LLMs in a variety of learning settings.

ZOGRASCOPE: A New Benchmark for Semantic Parsing over Property Graphs

Partially Relevant Video Retrieval (PRVR) aims to retrieve untrimmed videos partially relevant to a given query. The core challenge lies in learning robust query-video alignment against spurious semantic correlations arising from inherent data uncertainty: 1) query ambiguity, where the query incompletely characterizes the target video and often contains uninformative tokens, and 2) partial video relevance, where abundant query-irrelevant segments introduce contextual noise in cross-modal alignment. Existing methods often focus on enhancing multi-scale clip representations and retrieving the most relevant clip. However, the inherent data uncertainty in PRVR renders them vulnerable to distractor videos with spurious similarities, leading to suboptimal performance. To fill this research gap, we propose Robust Alignment Learning (RAL) framework, which explicitly models the uncertainty in data. Key innovations include: 1) we pioneer probabilistic modeling for PRVR by encoding videos and queries as multivariate Gaussian distributions. This not only quantifies data uncertainty but also enables proxy-level matching to capture the variability in cross-modal correspondences; 2) we consider the heterogeneous informativeness of query words and introduce learnable confidence gates to dynamically weight similarity. As a plug-and-play solution, RAL can be seamlessly integrated into the existing architectures. Extensive experiments across diverse retrieval backbones demonstrate its superiority, \eg, achieving a +9.7 SumR improvement over the previous SOTA GMMFormer v2 on TVR.

Enhancing Partially Relevant Video Retrieval with Robust Alignment Learning

Fine tuning pretrained language models for downstream NLP tasks, while effective, can be costly when the model size and the number of tasks increase, as it requires full parameter updates and a separate model served for each task. Parameter-efficient tuning (PET) addresses the issue by keeping the pretrained parameters fixed while introducing minimal task-specific parameters. There are two essential PET paradigms: prompt-based tuning and adapter-based tuning, each with distinct limitations. Prompt-based methods suffer from increased input lengths and sensitivity to weight initialization, whereas adapter approaches can substantially increase inference time. To overcome these limitations, we propose prompt adaptive tuning (PrAd), a general prompt-based tuning framework for decode-only models that delivers strong performance with high efficiency, even in multi-task scenarios. Unlike conventional prompt-based tuning which uses soft tokens to "wrap" inputs, PrAd employs adapters for flexible input transformation. While traditional adapter-based tuning adapts both the prompt and decoded tokens, PrAd only adapts the prompt. PrAd enables the creation of diverse prompt-based approaches while providing critical advantages for real-world use: (1) it can maintain original input lengths with easy initialization during training, like adapter-based methods; (2) it can reduce management costs while facilitating deployment and efficient batch inference of different tasks, like prompt-based tuning.; and (3) it introduces no additional inference latency in the decoding phase even when serving multiple tasks concurrently. Experiments on six diverse tasks demonstrate that PrAd can consistently attain comparable or better performance and higher inference efficiency.

PrAd: Prompt Adaptive Tuning for Decoder-only Language Models

Evaluating large language models (LLMs) in medicine is crucial because medical applications require high accuracy with little room for error. Current medical benchmarks have three main types: medical exam-based, comprehensive medical, and specialized assessments. However, these benchmarks have limitations in question design (mostly multiple-choice), data sources (often not derived from real clinical scenarios), and evaluation methods (poor assessment of complex reasoning). To address these issues, we present LLMEval-Medicine, a new benchmark covering five core medical areas, including 2,996 questions created from real-world electronic health records and expert-designed clinical scenarios. We also design an automated evaluation pipeline, incorporating expert-developed checklists into our LLM-as-Judge framework. Furthermore, our methodology validates machine scoring through human-machine agreement analysis, dynamically refining checklists and prompts based on expert feedback to ensure reliability. We evaluate 13 LLMs across three categories (specialized medical models, open-source models, and closed-source models) on LLMEval-Med, providing valuable insights for the safe and effective deployment of LLMs in medical domains.

LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician Validation

Cross-lingual consistency should be considered to assess cross-lingual transferability, maintain the factuality of the model knowledge across languages, and preserve the parity of language model performance. We are thus interested in analyzing, evaluating, and interpreting cross-lingual consistency for factual knowledge. We examine code-mixed coreferential statements conveyed identical knowledge across languages to study cross-lingual knowledge consistency. We use some interpretability approaches to analyze the behavior of a model in cross-lingual contexts, discovering that multilingual models show different levels of consistency, subject to language families, linguistic factors, and a bottleneck in cross-lingual consistency on a particular layer. In addition, we evaluate common strategies aimed at improving multilingual performance to observe whether these strategies can improve knowledge consistency at the same time. While knowledge is not cross-lingual consistent in many cases, code-switching training and cross-lingual word alignment objectives show the most promising results, emphasizing the noteworthiness of cross-lingual alignment supervision and code-switching training for both multilingual performance and cross-lingual consistency enhancement.

Downloads

Next from EMNLP 2025

Multilingual Collaborative Defense for Large Language Models

Similar lecture

Deriving Character Logic from Storyline as Codified Decision Trees

Stay up to date with the latest Underline news!

PRODUCT

COMPANY

RESOURCES