Experimental analysis reveals significant differences in semantic entropy of translatable and untranslatable concepts, suggesting important implications for language processing.
This repository contains the paper, experimental code, raw data, and curated database demonstrating that semantic entropy , the dissipative cost of expressing a meaning in the absence of a stable lexical compression, is a measurable physical quantity. Three convergent embedding-space measures show highly significant differences between untranslatable and translatable concepts (p < 0.000001, Cohen's d = 1.57–6.13). Results are replicated with a purpose-built Sentence-Transformer model. A curated periphrasis database confirms S = k ln(W) at the inter-group level.
No takes yet. Share an insight, caveat, or question.
Massimo Lacchè (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: