{"total":13,"items":[{"citing_arxiv_id":"2606.30815","ref_index":214,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"When transformers learn \"impossible\" languages, what do they learn?","primary_cat":"cs.CL","submitted_at":"2026-06-29T18:42:03+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Transformers on impossible-language variants show gradual grammatical sensitivity loss but sharp long-sentence generation failures, supporting generative deficiency as a link to non-attestation.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.29904","ref_index":60,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Timesteps of Mamba Align with Human Reading Times","primary_cat":"cs.CL","submitted_at":"2026-06-29T07:40:58+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Mamba's per-word timesteps significantly predict human reading times beyond GPT-2 surprisal in a naturalistic dataset.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.27206","ref_index":90,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Syntactic Belief Update as the Driver of Garden Path Processing Difficulty","primary_cat":"cs.CL","submitted_at":"2026-06-25T16:02:14+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Syntactic belief update via generalized Rényi divergence on syntactic trees predicts garden path reading times better than lexical surprisal.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.21203","ref_index":23,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"When Context Misleads: Surprisal, Energy and Attention Entropy as Metrics of Coherence Illusions in LLMs","primary_cat":"cs.CL","submitted_at":"2026-06-19T08:16:49+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Dutch LLMs display coherence illusions tracked by surprisal, with attention entropy identifying affected heads and a new energy metric quantifying discourse coherence.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.05087","ref_index":18,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Light or Full Verb? A Minimal-Pair Dataset for Probing Phraseological Competence in Language Models","primary_cat":"cs.CL","submitted_at":"2026-06-03T16:51:27+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Presents a minimal-pair dataset and reports that probing experiments show language models differentiate light-verb from full-verb uses even in minimal contexts with separable patterns by object type.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.03817","ref_index":63,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Rethinking the Idiomaticity Decomposability Hypothesis: Evidence from Distributional Learning","primary_cat":"cs.CL","submitted_at":"2026-06-02T15:59:22+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Language models show idiom decomposability correlates weakly with human judgments, negatively with syntactic flexibility, and contributes most strongly to representation stabilization during training alongside surprisal and frequency.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.28616","ref_index":33,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Measuring Form and Function in Language Models","primary_cat":"cs.CL","submitted_at":"2026-05-27T15:27:16+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Proposes CAC prompting to benchmark language models on syntactic and discourse properties of determiners against child acquisition data, finding large models approach but do not match human performance on both.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.23035","ref_index":23,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography","primary_cat":"cs.CL","submitted_at":"2026-05-21T21:00:24+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Sparse autoencoders applied to GPT-2 and Llama models recover semantic features accounting for 94% of peak brain encoding performance and map onto distinct cortical semantic regions across three languages.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.21403","ref_index":73,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Quantifying the cross-linguistic effects of syncretism on agreement attraction","primary_cat":"cs.CL","submitted_at":"2026-05-20T17:02:17+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"LLM surprisal and attention entropy replicate syncretism modulation of agreement attraction in English and German, align with null results in Turkish, and partially match Russian patterns.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.00506","ref_index":72,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Surprisal Minimisation over Goal-directed Alternatives Predicts Production Choice in Dialogue","primary_cat":"cs.CL","submitted_at":"2026-05-01T08:32:13+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Surprisal minimization over goal-directed alternatives generated by language models provides the strongest account of production choices in open-ended dialogue compared to uniform information density or length-based costs.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.23985","ref_index":2,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Representational Curvature Modulates Behavioral Uncertainty in Large Language Models","primary_cat":"cs.AI","submitted_at":"2026-04-27T03:00:47+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Contextual curvature of LLM representational trajectories correlates with and causally modulates next-token entropy.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.18563","ref_index":42,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Dual Alignment Between Language Model Layers and Human Sentence Processing","primary_cat":"cs.CL","submitted_at":"2026-04-20T17:51:26+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Later LLM layers align better with human cognitive effort in syntactic ambiguity than early layers do, indicating dual processing modes and complementary benefits from multi-layer probability updates.","context_count":1,"top_context_role":"background","top_context_polarity":"support","context_text":"modeling, but leave a full analysis as a direction for future research. 2 Background 2.1 Surprisal theory Humans exhibit different cognitive load (e.g., mea- sured by reading time) for different interest ar- eas (e.g., words or tokens) in a text during read- ing. Surprisal has proven to be a robust predic- tor of reading time across languages and exper- imental paradigms (Levy, 2008a; Demberg and Keller, 2008; Wilcox et al., 2023). The sur- prisal (Cover, 1999) of a word wt ∈W in context w<t := [w 0,· · ·, w t−1]⊤ is defined as −logP t(W=w t|w<t), where Pt :W→[0,1] is a family of conditional distributions assigning a probability to a word w at time step t given its prefix. Thus, the more unexpected wt is, the more costly it is for humans to process."},{"citing_arxiv_id":"2211.09110","ref_index":6,"ref_count":1,"confidence":0.88,"is_internal_anchor":false,"paper_title":"Holistic Evaluation of Language Models","primary_cat":"cs.CL","submitted_at":"2022-11-16T18:51:34+00:00","verdict":"ACCEPT","verdict_confidence":"UNKNOWN","novelty_score":7.0,"formal_verification":"none","one_line_summary":"HELM establishes a multi-metric evaluation covering 30 language models on 42 scenarios (16 core) to raise average scenario coverage from 17.9% to 96% under uniform conditions while releasing all prompts, completions, and a toolkit.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null}],"limit":50,"offset":0}