4arXiv cs.CL (Computation and Language)·
Artificial Epanorthosis: LLMs systematically overuse a classical rhetorical self-correction figure, study finds
A new arXiv paper identifies that large language models systematically overuse 'epanorthosis' — a rhetorical self-correction figure (e.g., 'This is not a course. It is a journey of transformation') — attributing the pattern to training data rich in promotional prose and RLHF preference tuning that rewards emphatic phrasing. The authors introduce an Epanorthosis Index to measure deviation from genre-specific human baselines, finding models overshoot in oratory (up to threefold in Italian) and undershoot in informal Q&A. Mitigation experiments show a one-line instruction cuts the figure by 50–75%, while a LoRA fine-tuning adapter nearly eliminates it, with the stated goal being calibration to human rates rather than elimination.
Related guides (3)
Related events (8)
Benchmark study reveals how linguistic framing of user beliefs shifts LLM context-following behavior
A new arXiv paper introduces a typology of 17 linguistically motivated expression-of-belief (EoB) types—spanning form, evidentiality, epistemic stance, and tone—to evaluate how phrasing affects whether LLMs defer to user-stated beliefs or their own prior knowledge. The authors benchmark 16 LLMs across Llama 3, Qwen3, and Gemma3 families at scales from 1B to 30B parameters, finding that larger and instruction-tuned models are systematically less context-following than smaller or base models. Specific linguistic framings (e.g., presuppositions, certainty markers) are identified as statistically more persuasive, with implications for prompt robustness and sycophancy research.
Empirical taxonomy of factual errors in human-written text reveals LLM detection gaps
A new arXiv paper introduces a taxonomy of factual error types in human-written text, derived from analysis of newspaper article corrections, identifying categories like kanji misconversions and numeral classifier errors absent from existing hallucination benchmarks. The authors evaluate several LLMs on Factual Error Detection (FED) tasks using both synthetic and real correction data. Even high-performance models like GPT-5.4 achieve only ~52% word-level F1 on synthetic data, underscoring the difficulty of detecting human-induced factual errors versus LLM hallucinations. The work highlights a neglected subproblem in factual accuracy research as the field has shifted focus toward LLM-generated hallucinations.
Parameterized framework for measuring sycophantic praise in language models
A new arXiv paper argues that sycophantic praise and flattery constitute a distinct alignment problem separate from the more commonly studied excessive agreement. The authors introduce a parameterized framework that measures whether praise is excessive relative to contribution quality and expected user ability, outperforming generic LLM judges on human annotation agreement. Key finding: sycophantic praise occurs far more frequently in social and interpretive domains than in objective reasoning settings, positioning praise calibration as a distinct alignment challenge.
Cross-lingual evaluation framework reveals LLMs redistribute cultural narrative structure while preserving semantic meaning
A new arXiv preprint introduces a multilingual evaluation framework using 414 proverbs across 15 languages to assess whether LLMs preserve culturally grounded meaning when generating narratives. Using four LLMs to produce 13k narratives, the study finds that cross-lingual prompting preserves proverb-level semantic meaning but systematically redistributes agency, social positioning, and narrative structure. Strong inter-model convergence across architectures suggests multilingual LLMs rely on shared semantic abstractions. The authors argue that semantic similarity metrics alone overestimate cultural preservation in multilingual evaluations.
Instruction-tuned LLMs show stronger syntactic convergence toward human interlocutors than humans do
A new arXiv preprint measures syntactic convergence in 16 open-weight Llama and Gemma models (1B–70B, pretrained and instruction-tuned) by substituting model outputs into pre-existing human dialogues and measuring context-free grammar rule reuse. Every model showed greater CFG-rule overlap with the preceding human turn than a random baseline, and instruction-tuned models exceeded the syntactic convergence of the human responses they replaced. Instruction tuning increased local syntactic and semantic alignment with interlocutors, though it also raised baseline overlap with unrelated primes, suggesting a more globally accommodating output style rather than purely targeted priming.
Study: LLM-Derived Error Highlights and APE Suggestions in MT Post-Editing
Researchers conducted a controlled study with professional En-Nl translators comparing post-editing (PE) workflows augmented with LLM-derived error highlights and automatic post-editing (APE) correction suggestions against regular PE and QE-derived highlights. No condition produced measurable productivity or quality gains over standard PE. However, APE-derived highlights were preferred over QE-derived highlights, and correction suggestions improved subjective user experience.
LLM judges over-credit incorrect answers in no-reference evaluation settings
A new arXiv paper investigates the reliability of LLM-as-judge evaluation in no-reference settings, finding that judge models systematically over-credit incorrect answers when no ground-truth is provided. Sensitivity experiments across three languages show that adding reference answers to prompts flips correct/incorrect decisions by up to 85% in some settings, with these changes aligning with human annotations. The authors propose a calibration methodology—testing judge knowledge with reference-aware samples before deploying in reference-free setups—as a blueprint for practitioners.
Paper examines how LLMs reproduce standard language ideologies and marginalize non-dominant Englishes
A new arXiv paper analyzes how AI systems reinforce dominant language ideologies that privilege Inner Circle (Global North) English norms and marginalize World Englishes, tracing this reproduction across training data, design protocols, evaluation benchmarks, and user feedback. The paper uses the public controversy over AI-associated vocabulary (e.g., the word 'delve') to illustrate how Global North speakers police English norms of Global South users. It identifies a 'standardisation paradox' where generative AI simultaneously homogenizes English toward standard forms while potentially pluralizing it through diverse corpora. The authors argue for more inclusive AI design approaches that recognize the plurality of Englishes.