cs.CL · 2026-05-27 · No. 10

Computation and Language, 2026-05-27.

30 new papers in cs.CL. Titles, authors, abstracts. Links to arXiv. Want this in your inbox every morning? Subscribe →

01 — The papers

30 entries
  1. 01

    It's Not Always Sycophancy: Measuring LLM Conformity as a Function of Epistemic Uncertainty

    Kevin H. Guo, Chao Yan, Avinash Baidya, Katherine Brown, Xiang Gao, Juming Xiong, Zhijun Yin, Bradley A. Malin

    cs.CL · cs.AI · cs.LG

    Large language models (LLMs) are known to abandon their initial stance to conform to user pushback. While prior research largely attributes this behavior to sycophancy learned during reinforcement learning from human feedback, we hypothesize that conformity is also driven by a model's epistemic uncertainty at inference time. In this paper, we introduce MUSE, a two-stage evaluation framework to disentangle the mechanisms driving LLM...

    arxiv.org/abs/2605.27288 · PDF

  2. 02

    Lost in Sampling: Assessing Lexical Reachability in LLMs via the Word Coverage Score (WCS)

    Samer Awad, Javier Conde, Carlos Arriaga, Tairan Fu, Javier Coronado-Blázquez, Pedro Reviriego

    cs.CL · cs.AI

    Modern Large Language Models (LLMs) are often criticized for producing repetitive and homogeneous text, despite possessing vast latent vocabularies. While previous research has focused on model knowledge and training data, we investigate the role of decoding mechanics in suppressing linguistic diversity. We introduce the Word Coverage Score (WCS), a metric that quantifies the extent to which contextually appropriate human vocabulary is...

    arxiv.org/abs/2605.27268 · PDF

  3. 03

    Pair-In, Pair-Out: Latent Multi-Token Prediction for Efficient LLMs

    Wenhui Tan, Minghao Li, Xiaoqian Ma, Siqi Fan, Xiusheng Huang, Liujie Zhang, Ruihua Song, Weihang Chen

    cs.CL · cs.AI

    Long chain-of-thought reasoning has made autoregressive decoding the dominant inference cost of modern large language models. Existing methods target either the input side (latent compression) or the output side (speculative decoding and multi-token prediction, MTP), but the two lines of work have been pursued independently. Moreover, output-side methods must incur an expensive verifier pass to validate the unreliable draft tokens predicted...

    arxiv.org/abs/2605.27255 · PDF

  4. 04

    Not All Tokens Matter Equally: Dynamic In-context Vector Distillation with Decisive-Token Supervision for Long-form Medical Report Generation

    Ning Wu, Rui Liu, Xinkun Lin, Weixing Chen, Jinxi Xiang, Tao Wei, Lina Yao, Mingjie Li

    cs.CL · cs.CV · cs.LG

    Distilling demonstration effects into hidden-space interventions offers a lightweight alternative to full finetuning. However, existing multimodal variants are mostly evaluated on short-form tasks, where outputs end after a few tokens. Extending these methods to long-form generation exposes a fundamental yet underexamined limitation: token-level distillation implicitly treats all output tokens as equally informative, but long-form outputs are...

    arxiv.org/abs/2605.27194 · PDF

  5. 05

    Learning When to Think While Listening in Large Audio-Language Models

    Zhiyuan Song, Weici Zhao, Yang Xiao, Suhao Yu, Cheng Zhu, Jiatao Gu

    cs.CL · cs.AI · cs.LG · cs.SD

    Recent advances in Large Audio-Language Models (LALMs) have made real-time, streaming spoken interaction increasingly practical. In this setting, reasoning quality and responsiveness are tightly coupled: delaying reasoning until the speech endpoint can improve answer quality but moves deliberation into user-visible response delay, while answering too early risks committing before decisive evidence arrives. We introduce a learnable...

    arxiv.org/abs/2605.27190 · PDF

  6. 06

    Beyond Binary: Speech Representations Across the Cognitive Score Hierarchy

    Serli Kopar, Roshan Prakash Rane, Christian Mychajliw, Lydia Federmann, Gerhard Eschweiler, Daniela Berg, Sam...

    cs.CL · cs.LG · cs.SD · eess.AS · q-bio.NC

    This study examines the relationship between speech representations and the hierarchical structure of cognitive assessment in mild cognitive impairment. Utilizing 5,754 German neuropsychological assessment recordings, we evaluate six cognitive tasks across three score levels: task, domain, and global levels. We compare hand-crafted acoustic features with self-supervised learning (SSL) embeddings. Results show that although SSL representations...

    arxiv.org/abs/2605.27189 · PDF

  7. 07

    Grounding Text Embeddings in Stakeholder Associations

    Jonathan Rystrøm, Sofie Burgos-Thorsen, Zihao Fu, Johan Irving Søltoft, Kenneth C. Enevoldsen, Chris Russell

    cs.CL · cs.AI · cs.CY

    Text embeddings are widely used to analyse large corpora of complex texts. However, it is unclear whether the embeddings capture the same semantic distances as the human experts using them. Ensuring alignment between embedding representations and human intentions is essential for valid analyses. We present the Stakeholder Grounding Exercise, a method for making expert associations explicit and grounding embedding model results in human...

    arxiv.org/abs/2605.27168 · PDF

  8. 08

    LitSeg: Narrative-Aware Document Segmentation for Literary RAG

    Ruikang Zhang, Zhanni Chen, Yiqiao Cai, Qi Su

    cs.CL · cs.AI

    Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by incorporating external knowledge, particularly for long-tail domains such as literary works. However, the critical step of document segmentation in RAG remains largely underexplored. Existing strategies are typically semantically blind and overlook the complicated narrative structures of literary works, often resulting in fragmented plots and unclear references that...

    arxiv.org/abs/2605.27156 · PDF

  9. 09

    LLMs Are Already Good Tutors: Training-Free Prompt Optimization for Pedagogical Math Tutoring

    Unggi Lee, Minchul Shin, Yeil Jeong, Sookbun Lee, Jeongsu Moon, Kyungtae Joo, Eunjoo Lee, Hoilym Kwon

    cs.CL · cs.LG

    Aligning LLMs for math tutoring typically requires RL-based training with multi-GPU infrastructure. We investigate whether training-free prompt optimization-evolving only the system prompt via API calls-can serve as a practical alternative. We adapt 7 published methods and propose 5 education-specialized methods, evaluating these 12 methods under 5 conditions on 2 OOD benchmark suites. All 12 best-per-method configurations surpass the...

    arxiv.org/abs/2605.27088 · PDF

  10. 10

    E3: Issue-Level Backtesting for Automated Research Critique

    Yashwardhan Chaudhuri, Sanyam Jain, Paridhi Mundra

    cs.CL · cs.AI

    We present E3, an automated review assistant that augments reviewers and engineering teams by identifying decision-relevant technical concerns in research papers. For each concern, E3 reports its nature, its location, its bearing on the contribution, and the analysis or evidence that would resolve it, covering unsupported claims, missing ablations, weak baselines, hidden assumptions, threats to validity, and leakage risks. To evaluate E3...

    arxiv.org/abs/2605.27072 · PDF

  11. 11

    QUACK: Questioning, Understanding, and Auditing Communicated Knowledge in Multimodal Social Deduction Agents

    Ye Yuan, Rui Song, Weien Li, Zeyu Li, Haochen Liu, Xiangyu Kong, Changjiang Han, Yonghan Yang, Zichen Zhao, Zixuan...

    cs.CL · cs.AI · cs.MA

    Social deduction games have become a popular testbed for probing reasoning, deception, coordination, and belief modeling in Large Language Model (LLM) agents. However, most environments are scored only by game outcomes such as win rates and largely remain to text-only interaction, making it difficult to tell whether an agent's language is actually grounded in what it perceived and did, or to identify the failure modes underlying its behavior....

    arxiv.org/abs/2605.27068 · PDF

  12. 12

    FalAR: A Large-scale Speaker-Annotated European Portuguese Speech Corpus of Parliamentary Sessions

    Francisco Teixeira, Carlos Carvalho, Mariana Julião, Catarina Botelho, Rubén Solera-Ureña, Sérgio Paulo, Thomas...

    cs.CL · cs.LG

    State-of-the-art performance for Automatic Speech Recognition (ASR) largely depends on the availability of large-scale labeled corpora. This creates a demand for increased data collection efforts, particularly for under-represented languages and dialectal varieties. Due to having considerably fewer speakers (around 11 million), European Portuguese (EP) is overshadowed by Brazilian Portuguese (BP) (around 200 million speakers) in currently...

    arxiv.org/abs/2605.27062 · PDF

  13. 13

    BhashaSetu: A Data-Centric Approach to Low-Resource Machine Translation

    Param Thakkar, Anushka Yadav, Michael Tiemann, Abhi Mehta, Akshita Bhasin, Shrinivas Khedkar

    cs.CL · cs.LG

    We present BhashaSetu, a linguistically enriched English--Marathi parallel dataset addressing persistent data limitations in low-resource neural machine translation (NMT). Marathi, spoken by over 95 million people, remains underrepresented in high-quality parallel corpora across diverse domains. Our dataset comprises 2.78 million sentence pairs from heterogeneous sources including news, politics, healthcare, literature, and culture, with...

    arxiv.org/abs/2605.27050 · PDF

  14. 14

    Tracing Computation Density in LLMs

    Corentin Kervadec, Iuliia Lysova, Iuri Macocco, Marco Baroni, Gemma Boleda

    cs.CL · cs.AI · cs.LG

    Transformer-based large language models (LLMs) are comprised of billions of parameters arranged in deep and wide computational graphs, but it is not clear that they exploit their full capacity for all inputs. We introduce the s-Trace method to efficiently estimate the subgraph of size s that best approximates a full model output. With this method, we find the computation in a variety of LLMs to be organized in two distinct phases. A small...

    arxiv.org/abs/2605.27033 · PDF

  15. 15

    Evaluating the Relevance of Uncertainty Estimators for LLM Hallucination

    Yedidia Agnimo, Anna Korba, Annabelle Blangero, Nicolas Chesneau, Karteek Alahari

    cs.CL · cs.AI · cs.LG · stat.ML

    Large language models (LLMs) are prone to hallucinations, i.e., statements unsupported by the input or training data, hindering reliable deployment. In parallel, numerous uncertainty estimation (UE) methods have been proposed to quantify model confidence and are often implicitly treated as proxies for model failure. However, the relationship between uncertainty and hallucinations remains insufficiently characterized. We present a systematic...

    arxiv.org/abs/2605.27016 · PDF

  16. 16

    Cast a Wider Net: Coordinated Pass@K Policy Optimization for Code Reasoning

    Yilong Li, Suman Banerjee, Tong Che

    cs.CL · cs.AI

    Repeated sampling with a verifier is the standard way to allocate test-time compute for code generation, with pass@$K$ as the canonical metric. Yet the standard policy class draws $K$ independent samples from a single answer distribution, so attempts often collapse onto near-duplicate reasoning paths and waste the budget on redundant rollouts. This failure is costly in competitive programming, where many problems admit multiple distinct...

    arxiv.org/abs/2605.27000 · PDF

  17. 17

    Recon: Reconstruction-Guided Reasoning Synthesis for User Modeling

    Alan Zhu, Mihran Miroyan, Carolyn Wang, Andrew Zhou, Lisa Dunlap, Narges Norouzi, Joseph E. Gonzalez

    cs.CL · cs.AI

    User modeling aims to use language models (LMs) to mimic an individual's behavior from a corpus of past context-action pairs (e.g., conversation turns), enabling the simulation of users in settings like behavioral science, human-AI collaboration, and market research. Recent approaches augment these corpora with synthesized reasoning traces, typically generated by conditioning on both context and action. However, such conditioning constitutes...

    arxiv.org/abs/2605.26969 · PDF

  18. 18

    Tournament-GRPO: Group-Wise Tournament Rewards for Reinforcement Learning in Open-Ended Long-Form Generation

    Zixuan Yang, Yiqun Chen, Wei Yang, Erhan Zhang, Zihan Shen, Xiaochi Wei, Yan Gao, Yi Wu, Yao Hu, Jiaxin Mao

    cs.CL · cs.AI

    Reinforcement learning in open-ended long-form generation is challenging because reliable reference answers and automatic metrics are often unavailable. Existing rubric-based methods typically rely on pointwise LLM-as-a-judge scoring, but absolute scores are difficult to calibrate across complex responses, may provide weak discrimination among same-query rollouts, and can become saturated during optimization. We propose Tournament-GRPO, a...

    arxiv.org/abs/2605.26958 · PDF

  19. 19

    JuICE: A Benchmark for Evaluating LLM-Judge in Identifying Cultural Errors

    Jiho Jin, Junho Myung, Juhyun Oh, Junyeong Park, Rifki Afina Putri, Sunipa Dev, Vinodkumar Prabhakaran, Alice Oh

    cs.CL · cs.AI

    As large language models (LLMs) are increasingly deployed to users around the world, they are integrated into everyday tasks across diverse cultural contexts, from drafting personal communications to brainstorming creative ideas. These tasks are inherently cultural: they require contextual appropriateness, symbolic resonance, and tacit cultural expectations that native speakers draw on instinctively, meaning that a response can be factually...

    arxiv.org/abs/2605.26955 · PDF

  20. 20

    Beyond Questions: Evaluating What Large Language Models (Actually) Know

    Luca Giordano, Simon Razniewski

    cs.CL · cs.AI

    Parametric knowledge in large language models (LLMs) is a cornerstone of their success, yet remains poorly understood. Existing knowledge benchmarks typically rely on predefined questions (e.g., "What is the birth date of M.L. King?"), evaluating only knowledge that benchmark designers explicitly choose to query, a problematic availability bias. In this paper, we introduce open knowledge evaluation, a new paradigm for LLM knowledge...

    arxiv.org/abs/2605.26937 · PDF

  21. 21

    Reasoning Depth and Environment Complexity: A Controlled Study of RLVR Data Allocation across Logical Reasoning Tasks

    Yihua Zhu, Qianying Liu, Fei Cheng, Jiaxin Wang, Akiko Aizawa, Sadao Kurohashi, Hidetoshi Shimodaira

    cs.CL · cs.AI

    Reinforcement learning with verifiable rewards (RLVR) has become central to post-training reasoning models, yet a key limitation of existing studies is their narrow view of the reasoning space: difficulty is treated as reasoning depth alone, and reward is concentrated on forward deductive state tracking. We instead characterize the reasoning space along two dimensions. Difficulty. Beyond reasoning depth, we study environment complexity, where...

    arxiv.org/abs/2605.26934 · PDF

  22. 22

    GeoFaith: A Spatio-Temporal Dual View of Faithful Chain-of-Thought

    Weijiang Lv, Wentong Zhao, Jiayu Wang, Yuhao Wu, Jiaheng Wei, Xiaobo Xia

    cs.CL · cs.AI

    Chain-of-Thought (CoT) reasoning has advanced large language models (LLMs), but outcome-based supervision leads to pervasive post-hoc rationalization, producing plausible yet unfaithful reasoning chains. Most prior faithfulness assessment methods are either unscalable, expensive, or unreliable. We propose GeoFaith, a spatio-temporal framework that leverages latent geometric structure and entropy dynamics to diagnose and enforce faithful...

    arxiv.org/abs/2605.26893 · PDF

  23. 23

    ContextGuard: Structured Self-Auditing for Context Learning in Language Models

    Hongbo Jin, Chi Wang, Haoran Tang, Zhongjing Du, Xu Jiang, Jingqi Tian, Qiaoman Zhang, Jiayu Ding

    cs.CL · cs.AI

    Recent benchmarks reveal that despite strong reasoning capabilities, large language models (LLMs) still struggle to faithfully apply complex contextual knowledge. These failures are often not wholesale reasoning collapses: in context-rich tasks, models may follow the central reasoning path while missing peripheral, persistent, or format-sensitive requirements.

    arxiv.org/abs/2605.26827 · PDF

  24. 24

    SeDT: Sentence-Transformer Decision-Transformer Conditioning for Multi-Turn Conversation Reliability

    Ramakrishna Vamsi Setti, Jagadeesh Rachapudi, Sachin Chaudhary, Praful Hambarde, Amit Shukla

    cs.CL · cs.AI

    Large language models (LLMs) achieve impressive performance when a task is fully specified in a single turn, yet the same models lose up to 39% of that performance when the identical task is revealed incrementally across multiple turns, a phenomenon documented at scale as Lost in Conversation. Crucially, this collapse is almost entirely a reliability failure; the best case, the aptitude only falls 16%, while the unreliability more than...

    arxiv.org/abs/2605.26788 · PDF

  25. 25

    EmoDistill: Offline Emotion Skill Distillation for Language Model Agents in Adversarial Negotiation

    Yunbo Long, Haolang Zhao, Lukas Beckenbauer, Liming Xu, Alexandra Brintrup

    cs.CL · cs.AI

    Post-trained LLMs are often optimized to align responses with human preferences, making them safe, polite, and conversationally appropriate. In adversarial negotiation, however, this alignment can become a vulnerability: emotionally framed language may steer agents toward the counterparty's interests. Using GoEmotions-based affective prompting, we show that emotion substantially shifts negotiation outcomes, suggesting that emotion is a...

    arxiv.org/abs/2605.26785 · PDF

  26. 26

    The Need for an External Observer Formalizing the Sufficiency Gap: A Mathematical Extension of Mixture Identifiability and Contextual Grounding in Sequence Models

    Francesco Corielli

    cs.CL · cs.LG

    We construct a binary mixed-regime process with one deterministic textual regime and one random regime governed by an unobserved latent state. Even an ideal infinite-capacity sequence predictor that exactly recovers the text-only marginal law can become overconfident when the observed prefix is compatible with the wrong latent regime. The resulting entropy difference is not an ordinary optimization error; it is a sufficiency gap caused by...

    arxiv.org/abs/2605.26711 · PDF

  27. 27

    An In-Vitro Study on Cross-Lingual Generalization in Language Models

    Adrian Cosma

    cs.CL · cs.AI

    Cross-lingual transfer in language models is difficult to study in natural corpora because lexical overlap, morphology, data imbalance, and tokenization are entangled. We introduce an in-vitro framework with two procedurally generated languages that share the same ontology, typed grammar, and compositional structure, but differ in surface realization. This lets us independently vary lexical distance, minority-language proportion, tokenizer...

    arxiv.org/abs/2605.26683 · PDF

  28. 28

    The Labyrinth and the Thread: Rethinking Regularizations in Sequential Knowledge Editing for Large Language Models

    Zheng Wang, Kaixuan Zhang, Wanfang Chen, Jingwen Zhang, Xiaonan Lu

    cs.CL · cs.AI

    Sequential editing of structured knowledge in large language models allows targeted factual updates without retraining, yet existing methods often rely on complex regularization or constraint mechanisms whose necessity remains unclear. In this work, we systematically investigate the mechanisms underlying effective and stable sequential editing. Specifically, we first analyze the empirical success of AlphaEdit and establish, via a rigorous...

    arxiv.org/abs/2605.26670 · PDF

  29. 29

    AI evaluation may bias perceptions: The importance of context in interpreting academic writing

    Shang Wu, Randol Yao

    cs.CL · cs.AI · econ.GN

    This paper examines how estimates of AI use in scientific writing can be biased when evaluation methods ignore contextual differences across countries and fields. Using large-scale data on journal publications from Dimensions, we construct AI-likeness benchmarks based on differences between human-written and LLM-rephrased abstracts. We show that a pooled benchmark may confound pre-existing stylistic variation with AI-generated text, producing...

    arxiv.org/abs/2605.26662 · PDF

  30. 30

    Why Prompt Optimization Works, and Why It Sometimes Doesn't: A Causal-Inspired Edit-Level Analysis

    Shuzhi Gong, Hechuan Wen

    cs.CL · cs.LG · cs.NE

    Automated prompt optimization methods (e.g., DSpy, TextGrad) can substantially improve the performance of large language model (LLM), however, their generalization ability across different tasks remains underperformed. In practice, the superiority of the optimized prompt on one benchmark often fails to transfer to another, and this limitation persists even when switching across different LLM backbones. To investigate the underexplored sources...

    arxiv.org/abs/2605.26655 · PDF

This edition is part of The Daily Abstract — cs.CL archive. Subscribe to receive these in your inbox each morning, automatically translated to Spanish, with reply-to-PDF: arxivdaily.ignorelist.com.

Colophon Set in Georgia, with system sans for interface chrome and a monospaced stack for code and paper identifiers. Sole accent: amber #D99C5E. Built and served on an always-free VM. The masthead is set 14% letterspaced because newspapers do that and it works.