Austria

The purpose of this work is to share an English-Yorùbá evaluation dataset for openbook reading comprehension with open-ended questions to assess the performance of models both in a high- and a low-resource language. The dataset contains 358 questions and answers on 338 English documents and 208 Yorùbá documents. Experiments show a consistent disparity in performance between the two languages, with Yorùbá falling behind English for automatic metrics even if documents are much shorter for this language. For a small set of documents with comparable length, performance of Yorùbá drops by 2.5 times and this comparison is validated with human evaluation. When analyzing performance by length, we observe that Yorùbá decreases performance dramatically for documents that reach 1500 words while English performance is barely affected at that length. Our dataset opens the door to showcasing if English LLM reading comprehension capabilities extend to Yorùbá, which for the evaluated LLMs is not the case.

ACL 2025

Y-NQ: English-Yorùbá Evaluation dataset for Open-Book Reading Comprehension with Open-Ended Questions

### Welcome to the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 
Continuing its mission of expanding and involving the science community of all European countries, EACL had selected the Ukrainian community for the 16th EACL, which took place online due to the COVID pandemic. Unfortunately, the ongoing war made the organisation in Kyiv impossible. Considering the importance of physical interaction among researchers, especially after the restrictions imposed by the COVID pandemic, in addition to an online mode, the conference will be held in Dubrovnik, Croatia, from 2 to 6 of May, 2023. The original aim of strengthening the connection with the Ukrainian community will not change as our program will feature a dedicated session and a workshop to highlight work on Ukrainian language technologies. As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

You need to register for the conference in order to access this site. Please visit https://2023.eacl.org/registration for more information.

EACL 2023

As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

workshop paper

### Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Message from the General Chair: 
*It is my great pleasure and honor to welcome you to the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025), held in beautiful Vienna, Austria, from July 27 to August 1, 2025. ACL2025continues our field’s 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.* 
[Read more](https://drive.google.com/file/d/1GI_hvOpjswAuYdUTromfeDiPpCcqidwg/view?usp=sharing)

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 2025". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you need to log in with the **email address you registered with**. 

Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

This paper presents Yankari, a large-scale monolingual dataset for the Yoruba language, aimed at addressing the critical gap in Natural Language Processing (NLP) resources for this important West African language. Despite being spoken by over 30 million people, Yoruba has been severely underrepresented in NLP research and applications. We detail our methodology for creating this dataset, which includes careful source selection, automated quality control, and rigorous data cleaning processes. The Yankari dataset comprises 51,407 documents from 13 diverse sources, totaling over 30 million tokens. Our approach focuses on ethical data collection practices, avoiding problematic sources and addressing issues prevalent in existing datasets. We provide thorough automated evaluations of the dataset, demonstrating its quality compared to existing resources. The Yankari dataset represents a significant advancement in Yoruba language resources, providing a foundation for developing more accurate NLP models, supporting comparative linguistic studies, and contributing to the digital accessibility of the Yoruba language.

Yankari: Monolingual Yoruba Dataset

Most of the 40+ languages spoken in Togo are severely under-represented in Natural Language Processing (NLP) resources. We present YodiV3, a comprehensive approach to developing NLP for ten Togolese languages (plus two major lingua francas) covering machine translation, speech recognition, text-to-speech, and language identification. We introduce Eyaa-Tom, a new multi-domain parallel corpus (religious, healthcare, financial, etc.) for these languages. We also propose the Lom metric, a scoring framework to quantify the AI-readiness of each language in terms of available resources. Our experiments demonstrate that leveraging large pretrained models (e.g.NLLB for translation, MMS for speech) with YodiV3 leads to significant improvements in low-resource translation and speech tasks. This work highlights the impact of integrating diverse data sources and pretrained models to bootstrap NLP for under-served languages, and outlines future steps for expanding coverage and capability.

YodiV3: NLP for Togolese Languages with Eyaa-Tom Dataset and the Lom Metric

Large-scale multilingual evaluations, such as MEGA, often include only a handful of African languages due to the scarcity of high-quality evaluation data and the limited discoverability of existing African datasets. This lack of representation hinders comprehensive LLM evaluation across a diverse range of languages and tasks. To address these challenges, we introduce AFROBENCH—a multi-task benchmark for evaluating the performance of LLMs across 64 African languages, 15 tasks and 22 datasets. AFROBENCH consists of nine natural language understanding datasets, six text generation datasets, six knowledge and question answering tasks, and one mathematical reasoning task. We present results comparing the performance of prompting LLMs to fine-tuned baselines based on BERT and T5-style models. Our results suggest large gaps in performance between high-resource languages, such as English, and African languages across most tasks; but performance also varies based on the availability of monolingual data resources. Our findings confirm that performance on African languages continues to remain a hurdle for current LLMs, underscoring the need for additional efforts to close this gap.

AfroBench: How Good are Large Language Models on African Languages?

The fact that everyone with a social media account can create and share content, and the increasing public reliance on social media platforms as a news and information source bring about significant challenges such as misinformation, fake news, harmful content, etc. Although human content moderation may be useful to an extent and used by these platforms to flag posted materials, the use of AI models provides a more sustainable, scalable, and effective way to mitigate these harmful contents. However, low-resourced languages such as the Somali language face limitations in AI automation, including scarce annotated training datasets and lack of language models tailored to their unique linguistic characteristics. This paper presents part of our ongoing research work to bridge some of these gaps for the Somali language. In particular, we created two human-annotated social-media-sourced Somali datasets for two downstream applications, fake news & toxicity classification, and developed a transformer-based monolingual Somali language model (named SomBERTa) -- the first of its kind to the best of our knowledge. SomBERTa is then fine-tuned and evaluated on toxic content, fake news and news topic classification datasets. Comparative evaluation analysis of the proposed model against related multilingual models (e.g., AfriBERTa, AfroXLMR, etc) demonstrated that SomBERTa consistently outperformed these comparators in both fake news and toxic content classification tasks while achieving the best average accuracy (87.99%) across all tasks. This research contributes to Somali NLP by offering a foundational language model and a replicable framework for other low-resource languages, promoting digital and AI inclusivity and linguistic diversity.

Detection of Somali-written Fake News and Toxic Messages on the Social Media Using Transformer-based Language Models

Large language models (LLMs) are effective in predicting the labels of unseen target instances if instructed for the task and training instances via the prompt. LLMs generate a text with higher probability if the prompt contains text with similar characteristics, a phenomenon, called priming, that especially affects argumentation. An open question in NLP is how to systematically exploit priming to choose a set of instances suitable for a given task. For stance classification, LLMs may be primed with few-shot instances prior to identifying whether a given argument is pro or con a topic. In this paper, we explore two priming strategies for few-shot stance classification: one takes those instances that are most semantically similar, and the other chooses those that are most stance-similar. Experiments on three common stance datasets suggest that priming an LLM with stance-similar instances is particularly effective in few-shot stance classification compared to baseline strategies, and behaves largely consistently across different LLM variants.

Exploring LLM Priming Strategies for Few-Shot Stance Classification

In this position paper, we advocate for the development of conversational technology that is inherently designed to support and facilitate argumentative processes. We argue that, at present, large language models (LLMs) are inadequate for this purpose, and we propose an ideal technology design aimed at enhancing argumentative skills. This involves re-framing LLMs as tools to exercise our critical thinking skills rather than replacing them. We introduce the concept of reasonable parrots that embody the fundamental principles of relevance, responsibility, and freedom, and that interact through argumentative dialogical moves. These principles and moves arise out of millennia of work in argumentation theory and should serve as the starting point for LLM-based technology that incorporates basic principles of argumentation.

Toward Reasonable Parrots: Why Large Language Models Should Argue with Us by Design

Through manual annotation or automated argument mining processes, arguments can be represented not only as text, but also in structured formats like graphs. When searching for relevant arguments, this additional information about the relationship between their elementary units allows for the formulation of fine-grained structural constraints by using graphs as queries. Then, a retrieval can be performed by computing the similarity between the query and all available arguments. Previous works employed Graph Edit Distance (GED) algorithms such as A* search to compute mappings between nodes and edges for determining the similarity, which is rather expensive. In this paper, we propose an alternative based on Vision Transformers where arguments are rendered as images to obtain dense embeddings. We propose multiple space-filling visualizations and evaluate the retrieval performance of the vision-based approach against an existing A* search-based method. We find that our technique runs orders of magnitude faster than A* search and scales well on larger argument graphs while achieving competitive results.

Retrieving Argument Graphs Using Vision Transformers

Online discussions can either bridge differences through constructive dialogue or amplify divisions through destructive interactions. This paper proposes a computational approach to analyze dialogical relation patterns in YouTube comments, offering a fine-grained framework for controversy detection, enabling also analysis of individual contributions. Our experiments demonstrate that shallow learning methods, when equipped with these theoretically-grounded features, consistently outperform more complex language models in characterizing discourse quality at both comment-pair and conversation-chain levels. Ablation studies confirm that divisive rhetorical techniques serve as strong predictors of destructive communication patterns. This work advances understanding of how communicative choices shape online discourse, moving beyond engagement metrics toward nuanced examination of constructive versus destructive dialogue patterns.

Old but Gold: LLM-Based Features and Shallow Learning Methods for Fine-Grained Controversy Analysis in YouTube Comments

Implicit premise is central to argumentative coherence and faithfulness, yet remain elusive in traditional single-pass computational models. We introduce a multi-agent framework that casts implicit premise recovery as a dialogic reasoning task between two LLM agents. Through structured rounds of debate, agents critically evaluate competing premises and converge on the most contextually appropriate interpretation. Evaluated on a controlled binary classification benchmark for premise selection, our approach achieves state-of-the-art accuracy, outperforming both neural baselines and single-agent LLMs. We find that accuracy gains stem not from repeated generation, but from agents refining their predictions in response to opposing views. Moreover, we show that forcing models to defend assigned stances degrades performance—engendering rhetorical rigidity to flawed reasoning. These results underscore the value of interactive debate in revealing pragmatic components of argument structure.

Multi-Agent LLM Debate Unveils the Premise Left Unsaid

Argument Mining (AM) aims at detecting argumentation structures (i.e., premises and claims linked by attack and support relations) in text. A natural application domain is political debates, where uncovering the hidden dynamics of a politician's argumentation strategies can help the public to identify fallacious and propagandist arguments. Despite the few approaches proposed in the literature to apply AM to political debates, this application scenario is still challenging, and, more precisely, concerning the task of predicting the relation holding between two argument components. Most of AM relation prediction approaches only consider the textual content of the argument component to identify and classify the argumentative relation holding among them (i.e., support, attack), and they mostly ignore the structural knowledge that arises from the overall argumentation graph. In this paper, we propose to address the relation prediction task in AM by combining the structural knowledge provided by a Knowledge Graph Embedding Model with the contextual knowledge provided by a fine-tuned Language Model. Our experimental setting is grounded on a standard AM benchmark of televised political debates of the US presidential campaigns from 1960 to 2020. Our extensive experimental setting demonstrates that integrating these two distinct forms of knowledge (i.e., the textual content of the argument component and the structural knowledge of the argumentation graph) leads to novel pathways that outperform existing approaches in the literature on this benchmark and enhance the accuracy of the predictions.

Downloads

Next from ACL 2025

Yankari: Monolingual Yoruba Dataset

Similar lecture

Ɛ KÚ [MASK]: Integrating Yorùbá cultural greetings into machine translation

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES