If you forbid an AI from using basic words like 'the' or 'is,' it actually works harder and gets much better at solving riddles.
April 6, 2026
Original Paper
Trivial Vocabulary Bans Improve LLM Reasoning More Than Deep Linguistic Constraints
arXiv · 2604.02699
The Takeaway
It suggests that AI 'fluency' acts as a distraction that keeps models on a mindless autopilot rather than engaging in deep reasoning. By breaking the AI’s habit of using filler words, we can force it to concentrate more effectively on the task at hand.
From the abstract
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 fil