Linguistic mechanisms for countering fake content in digital media

Abstract

This study aims to develop and empirically verify a linguistic model for detecting fake content in digital media environments. The research proposes a comprehensive approach integrating corpus linguistics, critical discourse analysis, and psycholinguistic methods. The methodological framework encompasses corpus analysis of textual materials from social networks and regional media, a psycholinguistic experiment assessing the perception of manipulative markers, and comparative testing of the model's effectiveness against state-of-the-art artificial intelligence algorithms. The results reveal a consistent triad of linguistic indicators of fake discourse: abnormally high emotional load, systemic coherence violations, and contradictions in modal constructions. The developed model demonstrated significant superiority over neural network counterparts in identifying unreliable content. Practical significance is confirmed through successful implementation in media literacy education programs, which recorded substantially reduced trust in misinformation, and in editorial system algorithms that decreased news verification time. The study concludes that linguistic methods not only ensure high-precision fake detection but also establish cognitive barriers against manipulative strategies, creating a foundation for digital resilience amid exponential growth of synthetic content. The findings contribute to media linguistics, digital journalism, and educational practices, offering tools to counteract information threats in the generative AI era.