mlsanityData to 14 Oct 2026

Trivial Vocabulary Bans Improve LLM Reasoning More Than Deep Linguistic Constraints

arXiv 2604.02699 ↗First posted 3 Apr 2026cs.CL cs.AIcited by 1

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

A previous study reported that E-Prime (English without the verb "to be") selectively altered reasoning in language models, with cross-model correlations suggesting a structural signature tied to which vocabulary was removed. I designed a replication with active controls to test the proposed mechanism: cognitive restructuring through specific vocabulary-cognition mappings. The experiment tested five conditions (unconstrained control, E-Prime, No-Have, elaborated metacognitive prompt, neutral filler-word ban) across six models and seven reasoning tasks (N=15,600 trials, 11,919 after compliance filtering). Every prediction from the cognitive restructuring hypothesis was disconfirmed. All four treatments outperformed the control (83.0%), including both active controls predicted to show null effects. The neutral filler-word ban, banning words like "very" and "just" with no role in logical inference, produced the largest improvement (+6.7 pp), while E-Prime produced the smallest (+3.7 pp). The four conditions ranked in perfect inverse order of theoretical depth. The cross-model correlation signature did not replicate (mean r=0.005). These results are consistent with a simpler mechanism: any constraint that forces a model off its default generation path acts as an output regularizer, improving reasoning by disrupting fluent but shallow response patterns. The shallowest constraints work best because they impose monitoring load with minimal conceptual disruption. I present these findings as a case study in discovery through disconfirmation.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 6 of its 12 references that are in the map. First posted on arXiv 3 Apr 2026.

Builds on

6 of 12 references.

References · 6

  1. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  2. 2022Large Language Models are Zero-Shot Reasonerscited by 4,374
  3. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
  4. 2023Mitigating the Alignment Tax of RLHFcited by 183
  5. 2023FollowBench: A Multi-level Fine-grained Constraints Following Benchmark for Large Language Modelscited by 117
  6. 2026Umwelt Engineering: Designing the Cognitive Worlds of Linguistic Agentscited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Decoding-Level Taboo: A Diagnostic Stress Test for LLM Robustnesscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Don't Think of the White Bear: Ironic Negation in Transformer Models Under Cognitive Load
  2. 2026Evaluation of Deontic Conditional Reasoning in Large Language Models: The Case of Wason's Selection Task
  3. 2026Thinking Inertia: LLMs Keep Thinking When Told Not To
  4. 2026Does Verbose Chain-of-Thought Really Help? In-Distribution Evidence that Content, Not Length, Matters
  5. 2026Logical Judgments Under Pressure: Diagnosing Syllogistic Stability with Learned Soft Prefixes
  6. 2026When Decodability Is Not Enough: Logical Validity Representations, Behavioral Dissociation, and Causal Tests in Language Models
  7. 2026Lexical Perturbations Disrupt LLM Reasoning: An Empirical Study of Attention Diversion
  8. 2025Under the Shadow of Babel: How Language Shapes Reasoning in LLMs