Automating Environmental Deception Detection: An NLP-Driven Diagnostic Framework for Measuring Linguistic Distortions and Selective Disclosure in Corporate Sustainability Reports

• Asher Noah

  covenant University

  Author

The proliferation of corporate sustainability reporting has created an urgent need for robust mechanisms capable of detecting environmental misrepresentation—a practice commonly termed "greenwashing." Despite growing regulatory scrutiny and stakeholder demand for transparent environmental, social, and governance (ESG) disclosures, existing detection approaches remain fragmented, often relying on manual content analysis or simplistic lexicon-based filters that fail to capture the sophisticated linguistic strategies employed in selective disclosure. This study addresses this gap by developing and validating a Natural Language Processing (NLP)-driven diagnostic framework designed to systematically measure linguistic distortions and selective disclosure patterns in corporate sustainability reports. The methodology integrates a hybrid computational architecture combining fine-tuned transformer models (RoBERTa-Large and ClimateBERT) with a novel composite metric—the Green Authenticity Index (GAI)—which evaluates claims along dimensions of certainty, specificity, and external evidence alignment. Analysis of 975 sustainability reports spanning six sectors from 100 publicly listed companies (2020–2024) demonstrates that the proposed framework achieves 89.4% accuracy in identifying deceptive environmental claims, significantly outperforming traditional lexicon-based approaches (72.1%, p<0.001). The GAI shows strong correlation (r=0.76) with expert human assessments, while feature importance analysis identifies five key linguistic predictors—ambiguity density, verifiability deficit, temporal hedging, performative framing, and selective emphasis—as primary indicators of greenwashing. The framework contributes a replicable, automated diagnostic tool that enables regulators, investors, and practitioners to systematically evaluate the veracity of corporate environmental claims. These findings have significant implications for enhancing accountability in corporate sustainability reporting and advancing the theoretical understanding of linguistic deception in organizational communication.

• pdf

Copyright (c) 2026 Asher Noah (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.