Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 4: Student Research Workshop)

Domain Adaptation of Image Encoder for Multimodal Manga Translation (2026.eacl-srw)

Challenge:	Recent studies improve visual contrastive decoding (VCD) by constructing more informative auxiliary views.
Approach:	They propose to construct an object-aligned auxiliary view that disrupts unsupported tokens and produces a stronger contrast signal.
Outcome:	Empirically, the proposed method shows consistent gains on two popular object hallucination benchmarks across two MLLMs.

Copied to clipboard

Kota Manabe, Tomoyuki Kajiwara, Takashi Ninomiya, Isao Goto, Shonosuke Ishiwatari, Hiroshi Noji

Do Multi-Agents Solve Better Than Single? Evaluating Agentic Frameworks for Diagram-Grounded Geometry Problem Solving and Reasoning (2026.eacl-srw)

Challenge:	Existing machine translation systems lack sufficient manga comprehension capabilities when utilizing image information.
Approach:	They propose a domain-adapted image encoder training method for manga . the method trains encoders to acquire visual features that consider the structural and sequential characteristics of the manga based on a Japanese-English translation task.
Outcome:	The proposed method improves translation evaluation metrics in Japanese-English translation task compared to the conventional method .

Copied to clipboard

Mahbub E Sobhani, Md. Faiyaz Abdullah Sayeedi, Mohammad Nehad Alam, Proma Hossain Progga, Swakkhar Shatabda

Reasoning

Challenge:	Diagram-grounded geometry problem solving is critical for multimodal large language models, but the benefits of multi-agent design over single-aggent remain unclear.
Approach:	They compare diagram-grounded geometry problem solving to four visual math benchmarks . they found that multi-agent pipelines provide clear benefits for open-source models .
Outcome:	Theorem-based solvers and architectural refinements improve performance on four visual math benchmarks.

Luth: Efficient French Specialization for Small Language Models and Cross-Lingual Transfer (2026.eacl-srw)

Copied to clipboard

Maxence Lasbordes, Sinoué Gad

Machine Translation for Low-Resource Languages through Monolingual Data and LLM: A Case Study of English-to-Basque (2026.eacl-srw)

Challenge:	Existing Large Language Models are predominantly English-centric, resulting in a performance gap for other major languages.
Approach:	They propose a family of French-specialized Large Language Models that address the English-centric performance gap by targeting French data.
Outcome:	The proposed models outperform open-source counterparts on multiple French benchmarks while retaining their original English capabilities.

Copied to clipboard

Nam Luu, Aitor Soroa, German Rigau, Ondřej Bojar

Rethinking the Evaluation of Alignment Methods: Insights into Diversity, Generalisation, and Safety (2026.eacl-srw)

Challenge:	Existing LLMs do not translate well from English to Basque, but they yield an acceptable performance in the reverse direction.
Approach:	They propose to use a Basque monolingual corpora to train an LLM-based MT system . they use 'sovereignty fine tuning' to generate parallel corporata, and then use preference optimization .
Outcome:	The proposed system improves translation quality in English-to-Basque direction while requiring limited data for low-resource languages.

Copied to clipboard

Denis Janiak, Julia Moska, Dawid Motyka, Karolina Seweryn, Paweł Walkowiak, Bartosz Żuk, Arkadiusz Janz

Challenge:	Existing studies focus on individual techniques or specific dimensions, lacking a holistic assessment of the inherent trade-offs.
Approach:	They propose a framework that compares LLM alignment methods across five axes . they use a validated LLM-as-judge prompt to compare the results .
Outcome:	The proposed framework compares LLM alignment methods across factuality, safety, conciseness, proactivity, diversity and safety axes . it provides insights into trade-offs of common alignment methods, guiding the development of more balanced and reliable LLMs.

Modality Matching Matters: Calibrating Language Distances for Cross-Lingual Transfer in URIEL+ (2026.eacl-srw)

Copied to clipboard

York Hay Ng, Aditya Khan, Xiang Lu, Matteo Salloum, Michael Zhou, Phuong Hanh Hoang, A. Seza Doğruöz, En-Shiun Annie Lee

Is He Extroverted? Identifying Missing Relevant Personas for Faithful User Simulation (2026.eacl-srw)

Challenge:	Existing linguistic knowledge bases such as URIEL+ lack a principled method for aggregating these signals into a single, comprehensive score.
Approach:	They propose a framework for type-matched language distances that unifies these signals into a robust, task-agnostic composite distance.
Outcome:	The proposed representations improve transfer performance when the distance type is relevant to the task, while yielding gains in most tasks.

Copied to clipboard

Weiwen Su, Yuhan Zhou, Zihan Wang, Naoki Yoshinaga, Masashi Toyoda

Natural Language Interfaces

Challenge:	Existing user simulation approaches focus on generating user-like responses in dialogue without verifying whether critical personas are supplied.
Approach:	They propose a task of identifying persona dimensions that are relevant but missing in simulating a user's reply for a given dialogue context.
Outcome:	The proposed model identifies persona dimensions that are relevant but missing in simulating a user’s response for a given dialogue context.

Quality-Aware Adversarial Ensemble for Singer Identification in 1960s Tamil Film Music (2026.eacl-srw)

Copied to clipboard

Sathiyakugan Balakrishnan, Uthayasanker Thayasivam

Robustness in NLP

Challenge:	1960s Tamil film music lacks adequate metadata identifying playback singers in archival recordings.
Approach:	They propose a quality-aware adversarial ensemble approach based on variable audio degradation and instrumentation leakage confounding singer-specific features.
Outcome:	The proposed approach achieves 96.2% accuracy and 2.0% EER on a held-out test set of 52 clips.

Thesis Proposal: Efficient KV Cache Reuse for Multi-Document Retrieval-Augmented Generation (2026.eacl-srw)

Copied to clipboard

Zhipeng Zhang, Dmitry Ilvovsky

Thesis proposal: COGNILENS: Analyzing Cognitive Decline in Language Models for Alzheimer’s Monitoring (2026.eacl-srw)

Challenge:	Retrieval-Augmented Generation (RAG) systems face efficiency bottlenecks in prefill due to attention mechanism, and traditional KV cache only accelerates decoding.
Approach:	They propose a multi-document KV cache reuse framework for multi-doc RAG workloads . they propose to resolve position and context misalignment while eliminating document-specific quadratic complexity in prefill.
Outcome:	The proposed framework solves position and context misalignment issues while eliminating document-specific quadratic complexity in prefill.

Copied to clipboard

Jonathan Guerne

Beyond One-Step Distillation: Bridging the Capacity Gap in Small Language Models via Multi-Step Knowledge Transfer (2026.eacl-srw)

Challenge:	Existing methods to detect AD and Mild Cognitive Impairment (MCI) are not effective in early stages.
Approach:	They propose to develop digital twins of Alzheimer's Disease using language models to mimic functional deficits observed in AD patients.
Outcome:	The proposed models will mimic the functional deficits observed in AD patients and evaluate their effects on brain score against the state-of-the-art models.

Copied to clipboard

Gaeun Yim, Nayoung Ko, Manasa Bharadwaj

Thesis proposal: Are We Losing Textual Diversity to Natural Language Processing? (2026.eacl-srw)

Challenge:	Large Language Models (LLMs) excel across diverse tasks but remain too large for efficient on-device deployment.
Approach:	They revisit multi-step knowledge distillation as an effective remedy . they demonstrate that MSKD improves ROUGE-L and perplexity over single-step approaches .
Outcome:	The proposed approach improves ROUGE-L and perplexity over single-step approaches . large language models are too large for efficient on-device deployment, the authors show .

Copied to clipboard

Josef Jon, Ondřej Bojar

Constructing a Dataset for Hallucination Detection in Japanese Summarization with Fine-grained Faithfulness Labels (2026.eacl-srw)

Challenge:	Using Neural Machine Translation, we examine whether the algorithms used in NMT have inherent inductive biases that are beneficial for most types of inputs but might harm the processing of untypical texts.
Approach:	They propose to use a set of measures to quantify text diversity based on its statistical properties to determine whether NMT systems struggle with maintaining the diversity of such texts.
Outcome:	The proposed approaches maintain the diversity and complexity of language and allow for better global planning of the output generation.

Copied to clipboard

Hikari Tanaka, Atsushi Keyaki, Mamoru Komachi

Comparing Text Compression Capabilities of Large Language Models with Traditional Compression Algorithms (2026.eacl-srw)

Challenge:	Large language models (LLMs) can generate fluent text, but the quality of generated content depends on its consistency with the given input.
Approach:	They constructed a Japanese evaluation dataset for hallucination detection in summarization by manually annotating sentence-level faithfulness labels in LLM-generated summaries of Japanese documents.
Outcome:	The proposed model can detect hallucinations in Japanese documents by annotating faithfulness labels in Japanese summaries.

Copied to clipboard

Mehran Haddadi, William John Teahan

Comprehensive Comparison of RAG Methods Across Multi-Domain Conversational QA (2026.eacl-srw)

Challenge:	Experimental results show that large language models outperform baselines on non-English datasets . traditional methods remained dataset-agnostic, and the results suggest that current methods are impractical for the compression task.
Approach:	They evaluate the non-English and unstructured text compression performance of Large Language Models . they compare them with traditional baselines on datasets from eight most widely spoken languages .
Outcome:	The evaluated LLM outperformed baselines on non-English datasets . the results show that the current methods are highly impractical for the compression task .

Copied to clipboard

Klejda Alushi, Jan Strich, Chris Biemann, Martin Semmann

Natural Language Interfaces

Challenge:	Existing studies evaluate RAG methods in isolation and focus on single-turn settings.
Approach:	They compare retrieval-augmented generation methods for multi-turn conversational QA with those that use dialogue history and coreference to ground large language models.
Outcome:	The proposed methods outperform vanilla RAG and advanced methods fail to yield gains and can even degrade performance below the No-RAG baseline.

LEMUR: A Corpus for Robust Fine-Tuning of Multilingual Law Embedding Models for Retrieval (2026.eacl-srw)

Copied to clipboard

Narges Baba Ahmadi, Jan Strich, Martin Semmann, Chris Biemann

Trainable, Multiword-aware Linguistic Tokenization Using Modern Neural Networks (2026.eacl-srw)

Challenge:	Existing large language models are not designed for semantic retrieval and PDF-based legislative sources introduce substantial noise due to imperfect text extraction.
Approach:	They propose a large-scale multilingual corpus of EU environmental legislation constructed from 24,953 official EUR-Lex PDF documents covering 25 languages.
Outcome:	The proposed model improves Top-k retrieval accuracy in monolingual and bilingual settings . it also improves accuracy in low- and high-resource languages .

Copied to clipboard

Clara Boesenberg, Kilian Evang

Syntactic Text Processing

Challenge:	Tokenization is a fundamental task in natural language processing that forms the first step of many pipelines.
Approach:	They propose to use a standard tokenizer trained without MWE-awareness as a baseline and a character-level SRN+CRF model to train token-level models.
Outcome:	The proposed tokenizers are based on a character-level and token-level sequence labeling problem and are consistent with the proposed pipelines.

Different Time, Different Language: Revisiting the Bias Against Non-Native Speakers in GPT Detectors (2026.eacl-srw)

Copied to clipboard

Adnan Al Ali, Jindřich Helcl, Jindřich Libovický

Call, Reward, Repeat: Advancing Dialog State Tracking with GRPO and Function Calling (2026.eacl-srw)

Challenge:	Prior work suggests that automated methods falsely flag essays from non-native speakers as generated due to their low perplexity extracted from an LLM, which is supposedly a key feature of the detectors.
Approach:	They propose to use a perplexity-based detector to detect essays from non-native speakers of Czech and a detector to examine the effects of different families.
Outcome:	The proposed methods are not biased against non-native speakers, but instead falsely flag essays from non-natural speakers as generated, compared to the English essays.

Copied to clipboard

Timur Ionov, Anna Marshalova, Valentin Malykh

Dialogue Systems & Conversational Agents

Challenge:	Recent advances in Large Language Models (LLMs) have notably enhanced task-oriented dialogue systems, particularly in Dialogue State Tracking (DST).
Approach:	They propose a group-relative policy optimization method that guides LLMs toward improved DST accuracy even under low-resource conditions.
Outcome:	The proposed method improves on established DST benchmarks while using significantly reduced out-of-domain training data.

Generalising LLM Routing using Past Performance Retrieval: A Few-Shot Router is Sufficient (2026.eacl-srw)

Copied to clipboard

Clovis Varangot-Reille, Christophe Bouvard, Antoine Gourru

CAPID: Context-Aware PII Detection for Question-Answering Systems (2026.eacl-srw)

Challenge:	Large Language Models (LLMs) use a single LLM to perform tasks.
Approach:	They propose a meta-evaluation framework that predicts per-model performance for new queries by retrieving similar past queries and reweighting model scores with lightweight attention.
Outcome:	The proposed framework matches the quality–cost trade-offs of generalisable routers across five routing benchmarks.

Copied to clipboard

Mariia Ponomarenko, Sepideh Abedini, Masoumeh Shafieinejad, D. B. Emerson, Shubhankar Mohapatra, Xi He

Natural Language Interfaces

Challenge:	Existing approaches mainly redact all PII, disregarding the fact that some may be contextually relevant to the user’s question, resulting in a degradation of response quality.
Approach:	They propose a method that fine-tunes a locally owned small language model that filters sensitive information before it is passed to LLMs for QA.
Outcome:	The proposed approach outperforms baselines in span, relevance and type accuracy while preserving significantly higher utility under anonymization.

Exploring the Semantic Space of Second Language Learners (2026.eacl-srw)

Copied to clipboard

Trisha Godara, Rui He, Wolfram Hinzen, Yan Cong

Challenge:	Using machine learning models, we compared the semantic space of university-level students learning French with native speakers' (L1) .
Approach:	They extracted semantic features from narrative text and used interpretability techniques to identify the most informative features per model.
Outcome:	The results show that the second language learners had higher semantic similarity scores than the native speakers at the token level, whereas the similarity decreased over time but did not reach native-level values.

Kahaani: A Multimodal Co-Creative Storytelling System (2026.eacl-srw)

Copied to clipboard

Samee Arif, Muhammad Saad Haroon, Aamina Jamal Khan, Taimoor Arif, Agha Ali Raza, Awais Athar

A Benchmark and Evaluation of Automated Language of Study Extraction from Computational Linguistics Publications (2026.eacl-srw)

Challenge:	Kahaani is a multimodal, co-creative storytelling system that leverages Generative Artificial Intelligence to address the challenge of sustaining engagement to foster educational narrative experiences.
Approach:	They propose a multimodal, co-creative storytelling system that leverages Generative Artificial Intelligence to help children develop their storytelling skills.
Outcome:	The proposed system combines large language models, text-to-speech, and music generation to produce a rich, immersive, and accessible storytelling experience.

Copied to clipboard

Henry Gagnier, Ashwin Kirubakaran

Who Plays Which Role? Protagonist Detection and Classification in Moral Discourse (2026.eacl-srw)

Challenge:	Language of study extraction is an aspect of computational linguistics papers that is useful for analyses of trends and diversity in computational linguists.
Approach:	They propose to benchmark and evaluate automated language of study extraction from computational linguistics papers.
Outcome:	The proposed language extraction benchmarks show that they can extract languages from papers with accuracy without high computational costs.

Copied to clipboard

Mirko Sommer, Maria Becker

Thesis Proposal: Multimodal Benchmark for Music Understanding in Large Language Models (2026.eacl-srw)

Challenge:	a systematic study of phrase-level protagonist detection and classification in moral discourse focuses on moral values rather than the actors involved.
Approach:	They propose to decompose a task into identifying protagonist mentions and classifying them by what kind of actor they are and what function they serve in the moral argument.
Outcome:	The proposed model outperforms previous models on the Moralization Corpus and fine-tuned lightweight models and prompting-based large language models.

Copied to clipboard

Tomáš Sourada

Communication as a Complex System: Modeling the Feedback Dynamics of Trust and Credibility (2026.eacl-srw)

Challenge:	Existing music-focused benchmarks are fragmented, largely single-modality, Western-centric . existing methods for evaluating MLLMs are lacking reproducibility and reliability .
Approach:	They propose to develop a musically multimodal benchmark that will integrate music into the benchmark.
Outcome:	The proposed benchmark will integrate culturally diverse musical material beyond the dominant Western canon.

Copied to clipboard

Swaptik Chowdhury, Samuel D. Allen, Jung Hee Hyun

The Clinical Fingerprint: Comparing the Rhetorical Integrity and Epistemic Safety of Human Physicians and Large Language Models (2026.eacl-srw)

Challenge:	A scoping literature review synthesized disparate communication models from media studies, science communication, psychology, and information science to identify a shared set of system variables.
Approach:	The study synthesized disparate communication models from media studies, science communication, psychology, and information science to identify a shared set of system variables.
Outcome:	The proposed framework provides a foundation for future system dynamics modeling to examine how interventions in transparency, media literacy, or platform governance may influence public trust over time.

Copied to clipboard

Bayram Ayadi

Acceleration of Backpropagation in Linear Layers of Transformer Models Based on Gradient Structure (2026.eacl-srw)

Challenge:	Large Language Models demonstrate expert proficiency on medical benchmarks, but clinical encounter requires a sophisticated rhetorical performance of care.
Approach:	They compare the rhetorical performance of large language models with human physicians . they find that generic models often bury critical advice under layers of linguistic recursion .
Outcome:	The proposed models lack the ethical integrity needed to deliver clinical advice, the authors argue . they show that generic models often bury critical advice under layers of complex linguistic recursion .

Copied to clipboard

Dmitrii Topchii, Alexander Panchenko, Viktoriia A. Chekalina

Chronocept: Instilling a Sense of Time in Machines (2026.eacl-srw)

Challenge:	a sparsity-exploiting backward pass is a memory-efficient way to accelerate LLM fine-tuning.
Approach:	They propose a method that exploits padding-induced gradient sparsity to accelerate backward computation.
Outcome:	The proposed method achieves a backward pass speedup of 2.15x on GLUE and 1.99x on reasoning benchmarks while maintaining memory usage identical to the regular PyTorch fine-tuning.

Copied to clipboard

Krish Goel, Sanskar Pandey, KS Mahadevan, Harsh Kumar, Vishesh Khadaria

Reasoning

Challenge:	Human cognition is deeply intertwined with a sense of time, known as Chronoception, which allows us to judge how long facts remain valid and when knowledge becomes outdated.
Approach:	They propose a model that captures nuanced patterns of emergence, decay, and peak relevance using skew-normal curves fitted along semantically decomposed temporal axes.
Outcome:	The proposed model captures nuanced patterns of emergence, decay, and peak relevance in two datasets.

When Prompt Optimization Becomes Jailbreaking: Adaptive Red-Teaming of Large Language Models (2026.eacl-srw)

Copied to clipboard

Zafir Shamsi, Nikhil Chekuru, Zachary Guzman, Shivank Garg

GraphRAG-Rad: Concept-Aware Radiology Report Generation via Latent Visual-Semantic Retrieval (2026.eacl-srw)

Challenge:	Existing safety evaluations rely on fixed collections of harmful prompts . such attacks span single-shot prompts, multi-turn interactions, cross-lingual settings .
Approach:	They propose to use black-box prompt optimization techniques to search for safety failures . they use GPT-5.1 to optimize for a continuous danger score .
Outcome:	The proposed approach reduces effective safeguards for large language models . the average danger score of Qwen 3 8B increases from 0.09 in its baseline setting to 0.79 after optimization.

Copied to clipboard

Faezeh Safari, Hang Dong, Zeyu Fu, Aline Villavicencio

Token Pruning for Improving Graph-Generating State Space Model Performance (2026.eacl-srw)

Challenge:	Existing encoder-decoder models suffer from hallucinations, generating plausible but incorrect medical findings.
Approach:	They propose a novel architecture that integrates biomedical knowledge through a latent visual-semantic retrieval approach.
Outcome:	The proposed architecture achieves competitive performance with strong results across multiple metrics.

Copied to clipboard

Monish Beegamudre, Jack Zheng, Margaret Capetz

Scale Is All You Need: Analyzing Modality Interaction and Speaker Intent Without Fine-Tuning (2026.eacl-srw)

Challenge:	State Space Models (SSMs) are efficient alternatives to Transformers for sequence modeling, but extending them to two-dimensional vision tasks remains challenging.
Approach:	They propose a leaf-guided pruning strategy that accelerates GG-SSM inference . they selectively scales or bypasses secondary refinement computations associated with leaf nodes .
Outcome:	The proposed pruning strategy accelerates GG-SSM inference without modifying graph topology . the proposed pruning method achieves throughput improvements with controlled accuracy degradation .

Copied to clipboard

Animesh Gurjar, Nikhil Krishnaswamy

Sentiment Analysis

Challenge:	Recent work on sarcasm and humor detection uses large multimodal Transformers, but they are computationally expensive and opaque.
Approach:	They propose a lightweight framework for multimodal sarcasm detection that combines frozen text, audio, and visual embeddings from pretrained encoders through compact fusion heads.
Outcome:	The proposed framework improves on the best unimodal baseline by combining text, audio, and visual embeddings from pretrained encoders with compact fusion heads.

Plasticity vs. Rigidity: The Impact of Low-Rank Adapters on Reasoning on a Micro-Budget (2026.eacl-srw)

Copied to clipboard

Zohaib Khan, Omer Tafveez, Zoha Hayat Bhatti

In-Image Machine Translation. A Preliminary Modular Approach (2026.eacl-srw)

Challenge:	Recent advances in mathematical reasoning typically rely on massive scale . yet, can strong reasoning capabilities be induced in small language models under extreme constraints?
Approach:	They train small language models with a single GPU for under 24 hours . they find that adapters unlock significant plasticity in standard instruction-tuned models .
Outcome:	The proposed model training on a single GPU (48GB) achieves 40% Pass@1 on AIME 24 (an 11.1% improvement over baseline) the model training results show that the adapter capacity and initialization are critical factors.

Copied to clipboard

Sergio Gomez Gonzalez, Miguel Domingo, Francisco Casacuberta

Text-to-Text Automatic Story Generation: A Survey (2026.eacl-srw)

Challenge:	In-image machine translation is a sub-task of Image-Based Machine Translation that aims to substitute text embedded in images with its translation into another language.
Approach:	They propose a simple task that renders parallel text over a plain background and a pipeline that obtains the transcript of the original image, translates it, and generates a new image similar to the original one.
Outcome:	The proposed approach outperforms existing models including an end-to-end approach and is competitive with other similar approaches.

Copied to clipboard

Yuan Ma, Richard Susilo, Patrik Haslum, Hanna Suominen

Probabilistic Bilingual Subword Segmentation with Latent Subword Alignment (2026.eacl-srw)

Challenge:	Automated story generation aims to produce coherent, engaging, and contextually consistent narratives with minimal or no human involvement . despite advances in large language models, maintaining narrative coherence, character consistency, storyline diversity, and plot controllability in generating stories is still challenging.
Approach:	They propose to develop new evaluation metrics and better data sets to support automatic story generation.
Outcome:	The proposed evaluation metrics and better datasets will improve narrative coherence and consistency and explore practical applications of story generation.

Copied to clipboard

Shoto Nishida, Daiki Matsui, Takashi Ninomiya, Isao Goto, Akihiro Tamura

Thesis Proposal: Development of End-to-End Speech Translation Models for Indian Languages (2026.eacl-srw)

Challenge:	Existing methods do not consider parallel relationships, preventing translation model training.
Approach:	They propose a method for learning subword correspondences in parallel sentence pairs using the EM algorithm.
Outcome:	The proposed method improves translation accuracy for many tasks.

Copied to clipboard

null Jamaluddin

Towards Singable Lyrics Translation Using Large Language Models (2026.eacl-srw)

Challenge:	Existing approaches to speech-to-speech translation rely on cascaded pipelines . current approaches rely only on text representations, but they suffer from errors and latency . a new direct speech translation framework is proposed to bridge linguistic gaps .
Approach:	They propose a sequence-to-sequence direct speech translation framework that can translate speech from one Indian language to another without relying on intermediate text representations.
Outcome:	The proposed framework can translate speech from one Indian language to another without relying on intermediate text representations.

Copied to clipboard

Liu Hanze, Yusuke Sakai, Taro Watanabe

Evaluating the Impact of SAE-based Language Steering on LLM Performance (2026.eacl-srw)

Challenge:	Existing studies on lyrics translation have relied on fine-tuning open-source language models.
Approach:	They examine a multilingual lyrics translation dataset and apply prompting methods to large language models to evaluate singability.
Outcome:	The proposed methods improve singability and naturalness, compared to naive translation, the authors show . human evaluations using songs created from translated lyrics show that complex prompting strategies improve singable naturalness .

Copied to clipboard

Sebastian Zwirner, Wentao Hu, Koshiro Aoki, Daisuke Kawahara

Annotation-Efficient Vision-Language Model Adaptation to the Polish Language Using the LLaVA Framework (2026.eacl-srw)

Challenge:	Recent advances in Sparse Autoencoders (SAEs) have revealed interpretable features within large language models (LLMs) however, the impact of SAE-based language steering on output quality and task performance remains unclear.
Approach:	They apply language-specific SAE feature steering to three LLMs from two model families and evaluate it on a translation task and a multilingual question-answering task.
Outcome:	The proposed approach outperforms prompting and language neuron-based steering on translation and multilingual question-answering tasks.

Copied to clipboard

Grzegorz Statkiewicz, Alicja Dobrzeniecka, Karolina Seweryn, Aleksandra Krasnodębska, Karolina Piosek, Katarzyna Bogusz, Sebastian Cygert, Wojciech Kusa

A Computational Forensic Linguistic Analysis of Narrative and Question-Answer Structures in Italian Police Interrogation Transcripts (2026.eacl-srw)

Challenge:	Currently, most vision-language models are trained on English-centric data, limiting their usability for non-English-speaking users.
Approach:	They reproduce and adapt LLaVA-Next methodology to create Polish VLMs . they use a fully automated pipeline for translating and filtering existing multimodal datasets based on Polish data for OCR and culturally specific tasks.
Outcome:	The proposed model improves on a Polish-adapted model and shows higher quality captions in generative evaluations.

Copied to clipboard

Romane Werner, Thomas François, Sonja Bitzer

Challenge:	linguistic profiling of police interrogation transcripts is rarely done in judicial contexts . despite their evidential centrality, transcription practices vary considerably across jurisdictions - despite being subject to systematic evaluation .
Approach:	They propose to analyze police transcripts using a multi-genre italian reference corpus to clarify how transcription formats shape evidential interpretation in judicial contexts.
Outcome:	The proposed models show that transcription formats shape evidential interpretation in judicial contexts and that they are linguistically informed and well-structured.

Thesis Proposal: A Multi-Agent System for Ontology-Based Perspective-Aware Knowledge Extraction (2026.eacl-srw)

Copied to clipboard

Luiz Do Valle Miranda, Grzegorz J. Nalepa

Fake News Detection Strategies under Dataset Bias: Using Large-scale Coarse-grained Labels (2026.eacl-srw)

Challenge:	Current AI systems consolidate multiple perspectives into singular, decontextualized schemas, introducing representational bias and information loss.
Approach:	They propose a framework to operationalize perspective-aware knowledge extraction using ontologies and Large Language Models.
Outcome:	The proposed framework can operationalize perspective-aware knowledge extraction without representational bias and information loss.

Copied to clipboard

Yuki Kishi, Yuji Arima, Hitoshi Iyatomi

DRAGOn: Designing RAG On Periodically Updated Corpus (2026.eacl-srw)

Challenge:	Existing datasets differ substantially in content distributions and annotation policies, complicating fair evaluation and generalization assessment.
Approach:	They quantitatively analyze dataset bias across multiple public fake news datasets with different annotation granularities, including article-level and publisher-level labels.
Outcome:	The proposed approach improves detection performance under in-dataset and cross-data set evaluation settings.

Copied to clipboard

Fedor Chernogorskii, Sergei Averkiev, Liliya Kudraleeva, Zaven Martirosian, Maria Tikhonova, Valentin Malykh, Alena Fenogenova

Efficient Low-Resource Language Models Using Tokenizer Transfer (2026.eacl-srw)

Challenge:	Existing methods for evaluating RAG systems are labor-intensive and difficult to maintain.
Approach:	They propose a method to design a RAG benchmark on a regularly updated corpus.
Outcome:	The proposed method uses a regularly updated corpus to evaluate RAG models.

Copied to clipboard

Gustaf Gren, Murathan Kurfali

Learning Nested Named Entity Recognition from Flat Annotations (2026.eacl-srw)

Challenge:	Tokenizer transfer allows training a model for low-resource languages without full retraining . a study of pre-trained tokenizers shows that they are more efficient than traditional training methods.
Approach:	They evaluate tokenizer transfer on models trained on language-specific corpora, Orthogonal Mapping Pursuit and Fast Vocabulary Transfer.
Outcome:	The proposed model adapts to a pre-trained model without full retraining and improves cross-lingual applicability.

Copied to clipboard

Igor Rozhkov, Natalia V Loukachevitch

Information Extraction & Text Mining

Challenge:	Named entity recognition (NER) requires expensive multi-level annotation.
Approach:	They evaluate four approaches to learning nested structure from flat annotations alone . on NEREL, a Russian benchmark, they find the best method achieves 26.37% inner F1 .
Outcome:	The proposed method closes 40% of the gap to full nested supervision on a Russian benchmark with 29 entity types where 21% of entities are nest.

Analysing LLM Persona Generation and Fairness Interpretation in Polarised Geopolitical Contexts (2026.eacl-srw)

Copied to clipboard

Maida Aizaz, Quang Minh Nguyen

Beyond Bias Scores: Unmasking Vacuous Neutrality in Small Language Models (2026.eacl-srw)

Challenge:	Large language models are increasingly used for social simulation and persona generation.
Approach:	They analysed personas generated for Palestinian and Israeli identities by popular LLMs across 640 experimental conditions, varying context and assigned roles.
Outcome:	The results show that large language models are increasingly utilised for social simulation and persona generation.

Copied to clipboard

Sumanth Manduru, Carlotta Domeniconi

From Detection to Explanation: Modeling Fine-Grained Emotional Social Influence Techniques with LLMs and Human Preferences (2026.eacl-srw)

Challenge:	Large Language Models (LLMs) are expensive to deploy locally and can reproduce harmful social biases in high-stakes settings such as healthcare and education.
Approach:	They propose a multi-dimensional evaluation paradigm to assess SLM fairness prior to deployment.
Outcome:	The proposed framework examines model robustness across four stages - biases, utility, ambiguity handling, and positional bias over diverse social bias categories.

Copied to clipboard

Maciej Markiewicz, Wiktoria Mieleszczenko-Kowszewicz, Beata Bajcar, Tomasz Adamczyk, Aleksander Szczęsny, Jolanta Babiak, Przemyslaw Kazienko

Dialogue Systems & Conversational Agents

Challenge:	Using Large Language Models (LLMs) is becoming increasingly important in communication and persuasion.
Approach:	They investigate the capabilities of Large Language Models (LLMs) to detect and explain emotional social influence techniques in textual dialogues.
Outcome:	The proposed models perform poorly on two tasks: detecting emotional social influence techniques and identifying text spans corresponding to specific techniques.

How Do Lexical Senses Correspond Between Spoken German and German Sign Language? (2026.eacl-srw)

Copied to clipboard

Melis Çelikkol, Wei Zhao

Challenge:	Existing dictionaries do not capture the full range of polysemous and homonymous words corresponding to different signs across contexts.
Approach:	They analyze 1,404 word use–to–sign ID mappings from German and German Sign Language . they identify three correspondence types: Type 1 (one-to-many), Type 2 (many-to-1), and Type 3 (one to one)
Outcome:	The proposed method outperforms existing methods using Exact Match and Semantic Similarity.

Evaluating Cost-Efficiency of LLMs in a RAG Setup on Polish Wikipedia: Quality vs. Energy Consumption (2026.eacl-srw)

Copied to clipboard

Patrycja Smits, Tomasz Walkowiak

Thesis Proposal: Measuring Prejudice at Scale (2026.eacl-srw)

Challenge:	Retrieval-augmented generation systems are a dominant paradigm for knowledge-intensive applications.
Approach:	They evaluate language models from 4B to 70B parameters within a Polish Wikipedia-based RAG pipeline.
Outcome:	The proposed model selection process reduces energy consumption by 83% and improves quality.

Copied to clipboard

Zoran Fijavž, Senja Pollak, Veronika Bajt

Text Classification

Challenge:	Existing methods for NLP fail to confirm construct validity, limiting the validity of the model.
Approach:	They propose to shift from categorical classification to comparative scaling of grounded constructs by using prompt optimization and distillation approaches.
Outcome:	The proposed pipeline is scalable for moving from categorical classification to theoretically grounded comparative measurement.

Energy Matching based Preference Learning for Diffusion Language Models (2026.eacl-srw)

Copied to clipboard

Shiv Shankar

Thesis Proposal: Stability-Aware, Evidence-Grounded Knowledge Graph for Substance Use Disorders and Social Determinants of Health (2026.eacl-srw)

Challenge:	Existing methods for RL are not compatible with diffusion language models due to the difficulty of likelihood estimation.
Approach:	They propose a framework that reformulates KL-regularized RL as an energy-based distribution matching problem.
Outcome:	The proposed framework matches or surpasses the performance of diffu-GRPO and related baselines on multiple benchmarks in both online and offline setting.

Copied to clipboard

Gautham Vijay Kumar

Structured Data in NLP

Challenge:	Existing methods for NER and RE annotation are costly and difficult to scale.
Approach:	They propose a semantic stability framework for constructing explainable KGs using NER and RE annotations.
Outcome:	The proposed framework supports multi-hop reasoning, triadic SUD–SDOH–SUD mediation patterns, and feedback loop analysis.

Detecting Overflow in Compressed Token Representations for Retrieval-Augmented Generation (2026.eacl-srw)

Copied to clipboard

Julia Belikova, Danila Rozhevskii, Dennis Svirin, Konstantin Polev, Alexander Panchenko

Representation Learning

Challenge:	Efficient long-context processing remains a challenge for large language models (LLMs) however, the limits of compressibility remain underexplored.
Approach:	They propose a method to characterize and detect token overflow in xRAG soft-compression by mapping long contexts into dense vectors that can be directly consumed by the model.
Outcome:	The proposed method identifies token overflow with query-agnostic saturation statistics but lacks the capability to detect it.

lrnnx: A library for Linear RNNs (2026.eacl-srw)

Copied to clipboard

Karan Bania, Soham Kalburgi, Manit Tanwar, null Dhruthi, Aditya Nagarsekar, Harshvardhan Mestha, Naman Chibber, Raj Deshmukh, Anish Sathyanarayanan, Aarush Rathore, Pratham Chheda

Challenge:	Existing implementations of linear recurrent neural networks are fragmented across different software frameworks . existing implementations often require custom CUDA kernels or lack publicly available code altogether .
Approach:	a unified software library implements several modern LRNN architectures under a common interface. lrnnx aims to improve accessibility, reproducibility, and extensibility of LRnn research and applications.
Outcome:	lrnnx aims to improve accessibility, reproducibility, and extensibility of LRNN research and applications.

Automatic Generation of a Compositional QA Benchmark for Geospatial Reasoning under Spatial and Entity Constraints (2026.eacl-srw)

Copied to clipboard

Tetsuhisa Suizu, Shohei Higashiyama, Hiroyuki Shindo, Hiroki Ouchi, Sakriani Sakti

Reasoning

Challenge:	Recent advances in large language models have enhanced their ability to perform reasoning tasks that integrate linguistic, visual, and factual information.
Approach:	They propose a method for constructing compositional geographic question answering datasets that jointly consider spatial and entity constraints.
Outcome:	The proposed method performs well on questions involving rich entity grounding, but its accuracy drops on quantitative spatial reasoning questions.

Thesis Proposal: Comparing Human and Model Perception of Writing Style under Controlled Perturbations (2026.eacl-srw)

Copied to clipboard

Ewelina Paulina Księżniak

Challenge:	This thesis examines how humans and models perceive writing style under controlled perturbations.
Approach:	They examine how humans and models perceive writing style under controlled perturbations . they also examine whether perturbations that reduce algorithmic recognition obscure stylistic identity .
Outcome:	The proposed research compares models and humans to find out how linguistic cues affect writing style . it will clarify how linguistic cue contributes differently to human and algorithmic perception of style - a cnn.com article argues .

Bring the Apple, Not the Sofa: Impact of Irrelevant Context in Embodied AI Commands on VLA Models (2026.eacl-srw)

Copied to clipboard

Andrey Moskalenko, Daria Pugacheva, Denis Shepelev, Andrey Kuznetsov, Vlad Shakhuro, Elena Tutubalina

Robustness in NLP

Challenge:	Embodied AI is undergoing rapid development, with robots increasingly exhibiting practical utility in everyday environments.
Approach:	They evaluate the robustness of vision language action models under linguistic perturbations . they categorize irrelevant contexts into two groups according to their length and proximity to robot commands .
Outcome:	The proposed model can exhibit relative robustness to random context, with a performance drop within 10%, the authors show . human paraphrases of instructions lead to a drop of nearly 20%, the study shows .

An Evaluation of Classifiers for Mapping Generative LLM Responses to Answer Options of Multiple-choice Questionnaires (2026.eacl-srw)

Copied to clipboard

Alisea Stroligo, Anna Shamray, Julian Schelb, Andreas Spitz

Information Retrieval

Challenge:	Large language models are used to generate multiple-choice style questionnaires that were originally intended for humans in persona simulations.
Approach:	They investigate the performance of smaller LLMs for mapping LLM outputs into the available answer options of multiple-choice questionnaires.
Outcome:	The proposed model underperforms on three datasets with differing answer option complexity.

Pioneering Bot Detection on Polish Reddit at the Comment Level (2026.eacl-srw)

Copied to clipboard

Karmela Matyjaszek

Challenge:	40,000 comments, 58% bot-comment prevalence, are used for comment-level bot detection within Polish Reddit communities.
Approach:	They construct a dataset with 40,000 comments, 58% bot-comment prevalence, which provides labels for the subsequent model training.
Outcome:	The proposed model trains on a Polish Reddit dataset with a linguistically mixed dataset and achieves strong performance and temporal generalization to 2025.

What the Router Sees Matters: Funnel Pooling for Fast, Content Driven Expert Routing (2026.eacl-srw)

Copied to clipboard

Josef Pichlmeier, Sebastian Nicolas Mueller, Jakob Sturm, Josef Dräxl, Andre Luckow

TimeRes: A Turkish Benchmark For Evaluating Temporal Understanding of Large Language Models (2026.eacl-srw)

Challenge:	Modern large language model (LLM) systems often route inputs to specialized experts to improve accuracy, efficiency, and robustness.
Approach:	They propose a lightweight trainable in-model readout that constructs the routing vector directly from token-level hidden states.
Outcome:	The proposed model reduces language and source-dataset driven clustering and results in more topic-aligned domains.

Copied to clipboard

Habib Yağız Demir, Ümit Atlamaz, Susan Üsküdarlı

Hospitality-VQA: Decision-Oriented Informativeness Evaluation for Vision–Language Models (2026.eacl-srw)

Challenge:	Existing benchmarks focus on English and underexplore how linguistic structure contributes to temporal meaning.
Approach:	They propose a Turkish benchmark to evaluate temporal understanding of Large Language Models (LLMs) their benchmark examines Reichenbach’s temporal points and reported speech through date arithmetic .
Outcome:	The proposed model fails to resolve reported speech and fails to generalize across word order variations.

Copied to clipboard

Jeongwoo Lee, Baek Duhyeong, Eungyeol Han, Soyeon Shin, Gukin Han, Seungduk Kim, Jaehyun Jeon, Taewoo Jeong

Visual Data in NLP

Challenge:	Existing VQA benchmarks focus on factual correctness but rarely capture what information users actually find useful.
Approach:	They propose a framework to quantify how much information an image–question pair provides . they conduct experiments with several state-of-the-art VLMs to determine their reliability .
Outcome:	The proposed framework quantifies how much information an image–question pair provides in hospitality contexts.

Colorism in Multimodal AI: An Empirical Exploration of Socioeconomic Linguistic Bias in Text-to-Image Generation (2026.eacl-srw)

Copied to clipboard

Raj Gaurav Maurya, Vaibhav Shukla, Sreedath Panat

Active Learning for Corpus Refinement: Cost-Effective Preprocessing to Improve Validity of Applied Quantitative Text Analysis (2026.eacl-srw)

Challenge:	Socioeconomic inequalities worldwide are deeply linked to ethnoracial hierarchies and stereotypes, argues a new study.
Approach:	They use a Monk Skin Tone scale to benchmark VLMs and annotators . they then use linguistic cues to vary skin-tone representations in text-to-image generation .
Outcome:	The study compares 3 small VLMs and 60 human annotators on the monk skin tone scale with 210 occupations and produces over 2,500 portraits across 3 large VLM models.

Copied to clipboard

Jakob Steglich, Stephan Poppe

From Sentences to Proof Trees: Leveraging Language Models for Structured Reasoning (2026.eacl-srw)

Challenge:	Quantitative text analysis relies on high-quality corpora, but keyword-based collection often retrieves irrelevant material, undermining validity.
Approach:	They propose to use a transformer-based classifier to iteratively refine corpora by excluding irrelevant documents.
Outcome:	The proposed method outperforms random sampling and weakly supervised sampling and outperformed random sampling.

Copied to clipboard

Aayushee Gupta