FindingMarch 2026Draft
Cross-Sector Convergence in European Institutional Language
Organisations from structurally unrelated sectors — luxury goods, central banking, and EU governance — are converging on statistically indistinguishable linguistic profiles. PLI scores across 11 years show compressing variance and rising homogenisation indices from 2021 onwards.
homogenisationcross-sector2015–2025
FindingMarch 2026Draft
The AI Signature: Perplexity Compression Across 600 Organisations
Perplexity scores — a measure of text predictability — have fallen sharply since 2022 across every sector in the corpus. The pattern holds regardless of language, country, or industry. Burstiness decline is concurrent. The statistical signature is consistent with large-scale AI-assisted drafting.
perplexityburstinessAI influenceL1
Methodology2026Draft
PRISM™: A Statistical Framework for Measuring AI Influence on Institutional Language
Full specification of the PRISM™ analytical framework. Covers all ten L1 atomic dimensions, four L2 composite detectors, and the TEMPORAL and HOMOGENISATION indices at L3. Statistical approach, corpus design, normalisation methodology, and validation results.
PRISMmethodologyL1–L3statistical framework
Dataset2026Draft
PLI Observatory Corpus v1: 70,000+ Institutional Documents, 2015–2025
Specification and access information for the PLI Observatory's public corpus. 70,000+ scored documents across 9 languages, 1,030+ European organisations, 11 years. Collection methodology, QA criteria, and language distribution.
corpusopen datamultilingual