India

Emotions are a fundamental facet of human experience, varying across individuals, cultural contexts, and nationalities. Given the recent success of Large Language Models (LLMs) as role-playing agents, we examine whether LLMs exhibit emotional stereotypes when assigned nationality-specific personas. Specifically, we investigate how different countries are represented in pre-trained LLMs through emotion attributions and whether these attributions align with cultural norms. To provide a deeper interpretive lens, we incorporate four key cultural dimensions, namely Power Distance, Uncertainty Avoidance, Long-Term Orientation, and Individualism, derived from Hofstede’s cross-cultural framework. Our analysis reveals significant nationality-based differences, with emotions such as shame, fear, and joy being disproportionately assigned across regions. Furthermore, we observe notable misalignment between LLM-generated and human emotional responses, particularly for negative emotions, highlighting the presence of reductive and potentially biased stereotypes in LLM outputs.

IJCNLP-AACL 2025

From Anger to Joy: How Nationality Personas Shape Emotion Attribution in Large Language Models

cultural framework

emotion attribution

model bias/fairness evaluation

bias in llms

## 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*

EMNLP 2025

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

technical paper

### Welcome to IJCNLP-AACL 2025! 
 It is a great honor to host this joint conference in Mumbai, India, from December 20 to 24, 2025. The joint conferences of IJCNLP and AACL are organized with alternating leadership in the Asia-Pacific region. The event is run by the Asian Federation of Natural Language Processing (AFNLP) in odd years, and by AACL in even years, while it is organized solely by ACL when the annual ACL meeting is held in the region. This year, the conference is primarily organized by AFNLP. 
*Kentaro Inui
MBZUAI, UAE
General Chair, IJCNLP-AACL 2025* 
Read full message and download the Conference Handbook [**here**](https://drive.google.com/file/d/1UTwxkAqSqI-GAoJC3wE1zZt5VP1Y8GX0/view?usp=sharing).

The 14th IJCNLP & 4th AACL will be held in Mumbai, India from December 20th to December 24th, 2025.

Large Language Models (LLMs) demonstrate strong reasoning and self-correction abilities in high-resource languages like English, but their performance remains limited in low-resource languages such as Korean. In this study, we investigate whether reinforcement learning (RL) can enhance Korean reasoning abilities to a degree comparable to English. Our findings reveal that RL alone yields limited improvements when applied to models lacking inherent Korean reasoning capabilities. To address this, we explore several fine-tuning strategies and show that aligning the model’s internal reasoning processes with Korean inputs—particularly by tuning Korean-specific neurons in early layers—is key to unlocking RL's effectiveness. We introduce a self-correction code-switching dataset to facilitate this alignment and observe significant performance gains in both mathematical reasoning and self-correction tasks. Ultimately, we conclude that the crucial factor in multilingual reasoning enhancement is not injecting new linguistic knowledge, but effectively eliciting and aligning existing reasoning capabilities. Our study provides a new perspective on how internal translation and neuron-level tuning contribute to multilingual reasoning alignment in LLMs.

Do LLMs Need Inherent Reasoning Before Reinforcement Learning? A Study in Korean Self-Correction

Large language models (LLMs) exhibit strong multilingual abilities, yet the neural mechanisms behind language-specific processing remain unclear. We analyze language-specific neurons in Llama-3.1-8B, Mistral-Nemo-12B, and Aya-Expanse-8B & 32B across $\textbf{21 typologically diverse languages}$, identifying neurons that control language behavior. Using the Language Activation Probability Entropy (LAPE) method, we show that these neurons cluster in deeper layers, with non-Latin scripts showing greater specialization. Related languages share overlapping neurons, reflecting internal representations of linguistic proximity.

Through $\textbf{language arithmetics}$, i.e. systematic activation addition and multiplication, we steer models to deactivate unwanted languages and activate desired ones, outperforming established replacement approaches. These interventions effectively guide behavior across five multilingual tasks: language forcing, translation, QA, comprehension, and NLI. Manipulation is more successful for high-resource languages, while typological similarity improves effectiveness. We also demonstrate that neuron steering enhances downstream performance and reveal internal $\textbf{"fallback"}$ mechanisms for language selection when neurons are progressively deactivated. Our code is made publicly available at https://github.com/d-gurgurov/Language-Neurons-Manipulation.

Language Arithmetics: Towards Systematic Language Neuron Identification and Manipulation

The Political Compass Test (PCT) and similar surveys are commonly used to assess political bias in auto-regressive LLMs. Our rigorous statistical experiments show that while changes to standard generation parameters have minimal effect on PCT scores, prompt phrasing and fine-tuning individually and together can significantly influence results. Interestingly, fine-tuning on politically rich vs. neutral datasets does not lead to different shifts in scores. We also generalize these findings to a similar popular test called 8 Values. Humans do not change their responses to questions when prompted differently ("answer this question" vs "state your opinion"), or after exposure to politically neutral text, such as mathematical formulae. But the fact that the models do so raises concerns about the validity of these tests for measuring model bias, and paves the way for deeper exploration into how political and social views are encoded in LLMs.

A Detailed Factor Analysis for the Political Compass Test: Navigating Ideologies of Large Language Models

Lipograms are a unique form of constrained writing where all occurrences of a particular letter are excluded from the text, typified by the novel Gadsby (Wright, 1939), which daringly avoids all usage of the letter 'e'. In this study, we explore the power of modern large language models (LLMs) by transforming the novel The Great Gatsby (Fitzgerald, 1925) into a fully 'e'-less text. We experimented with a range of techniques, from baseline methods like synonym replacement to sophisticated generative models enhanced with beam search and named entity analysis. We show that excluding up to 3.6% of the most common letters (up to the letter 'u') had minimal impact on the text’s meaning, although translation fidelity rapidly and predictably decays with stronger lipogram constraints. Our work highlights the surprising flexibility of English under strict constraints.

Gatsby without the ‘E’: Creating Lipograms with LLMs

Large language models can rephrase and restructure natural language effectively, but their potential for reformulating graph encodings remains underexplored despite the significant impact of encoding choices on performance.
In this work, we introduce ReGraph, a reinforcement learning-based approach that guides language models to reformulate graph encodings for improved task alignment.
We demonstrate that reformulating graph encodings enhances reasoning and yields consistent performance gains on graph question answering tasks.
Our results show that language models often prefer specific graph encodings, even if they are suboptimal for the task at hand.

ReGraph: Learning to Reformulate Graph Encodings with Large Language Models

Over 200 million people speak Indonesian, yet the language remains significantly underrepresented in preference-based research for large language models (LLMs). Most existing multilingual datasets are derived from English translations, often resulting in content that lacks cultural and linguistic authenticity. To address this gap, we introduce IndoPref, the first fully human-authored and multi-domain Indonesian preference dataset designed to evaluate the naturalness and quality of LLM-generated text. The dataset contains 522 prompts and yields 4,099 human-annotated pairwise preferences from comparisons across five instruction-tuned LLMs. All annotations are natively written in Indonesian with strong inter-annotator agreement, measured by Krippendorff’s alpha. Our benchmark spans 10 diverse categories, enabling practitioners to identify LLMs' fine-grained strengths and weaknesses.

IndoPref: A Multi-Domain Pairwise Preference Dataset for Indonesian

We introduce a set of resampling-based methods for quantifying uncertainty and statistical precision of evaluation metrics in multilingual and/or multitask NLP benchmarks.
We show how experimental variation in performance scores arises from both model and data-related sources, and that accounting for both of them is necessary to avoid substantially underestimating the overall variability over hypothetical replications.
Using multilingual question answering, machine translation, and named entity recognition as example tasks, we also demonstrate how resampling methods are useful for quantifying the replication uncertainty of various quantities used in leaderboards such as model rankings and pairwise differences between models.

Beyond statistical significance: Quantifying uncertainty and statistical variability in multilingual and multitask NLP evaluation

Radiology report generation requires advanced medical image analysis, effective temporal reasoning, and accurate text generation. Although recent innovations, particularly multimodal large language models, have shown improved performance, their supervised fine-tuning (SFT) objective is not explicitly aligned with clinical efficacy. In this work, we introduce **EditGRPO**, a mixed-policy reinforcement learning algorithm designed specifically to optimize the generation through clinically motivated rewards. EditGRPO integrates on-policy exploration with off-policy guidance by injecting sentence-level detailed corrections during training rollouts. This mixed-policy approach addresses the exploration dilemma and sampling efficiency issues typically encountered in RL. Applied to a Qwen2.5-VL-3B, EditGRPO outperforms both SFT and vanilla GRPO baselines, achieving an average improvement of 3.4\% in clinical metrics across four major datasets. Notably, EditGRPO also demonstrates superior out-of-domain generalization, with an average performance gain of 5.9\% on unseen datasets.

EditGRPO: Reinforcement Learning with Post -Rollout Edits for Clinically Accurate Chest X-Ray Report Generation

Four billion people cannot access vital information or be heard simply because of the language they speak—a human rights crisis where technology advances have left speakers of most of the world’s 7,000+ languages digitally invisible. This talk explores why languages matter in the digital space generally and in Language AI specifically, through real-world examples from CLEAR Global’s work across health, education, and agriculture sectors. We showcase practical tools developed to inform decisions (Language Use Data Platform), collect quality-controlled voice data (TWB Voice), and demonstrate how the NLP community can actively use these resources to address critical gaps in language technology. Drawing from projects spanning crisis response, educational equity, and access to agricultural information, we illustrate the transformative impact of language-inclusive technology. From conversational AI chatbots serving COVID-19 information in Lingala, Hausa, and Kanuri, to exploring EdTech solutions for mother tongue education in the Marma community in Bangladesh, to evaluating how synthetic voice data can be used for African ASR at lower costs than traditional data collection costs—our work demonstrates that marginalized languages can achieve competitive NLP performance with appropriate data and community engagement. Throughout our work, CLEAR Global’s 100,000+ volunteer linguist community provides the foundation for quality-controlled translations, data collection, cultural validation, and ensuring that technology development remains grounded in the needs and expertise of native speakers. This talk invites the NLP community to collaborate in ensuring that the right to information and to be heard doesn’t depend on which language you speak.

Bridging the Digital Language Divide: Building NLP Solutions for the Four Billion Left Behind

Recent advances in large language models (LLMs) have revolutionized how humans and AI systems work, learn, and interact, creating new opportunities for collaboration while also raising new challenges. In this talk, we explore the evolving landscape of human–AI collaboration from three perspectives. The first part focuses on teaming by rethinking human–AI collaboration through a large-scale audit of the future of work, highlighting mismatches between what humans want and current AI capabilities. We then talk about how we can create general user modeling from computer use to support robust memory modeling and proactive AI assistance during collaboration. The last part discusses how to evaluate human–AI collaboration, moving away from exams toward studying how people work with LLMs on diverse tasks. Overall, this talk demonstrates how to develop AI systems that are not just tools, but meaningful collaborators working alongside us, helping us grow, and adapting to who we are.

Downloads

Next from IJCNLP-AACL 2025

Do LLMs Need Inherent Reasoning Before Reinforcement Learning? A Study in Korean Self-Correction

Similar lecture

Exploring Changes in Nation Perception with Nationality-Assigned Personas in LLMs