Dynamic Changes in Cognitive Processes Among Online Fraud Victims in China: Evidence from Text Analysis

Grounded in theories of deception and victim cognition, this study examines how victims’ language reflects the dynamic evolution of cognitive processes during online fraud. Using large-scale text analysis, we analyzed self-reported accounts from 1,217 victims on Chinese social media and quantified seven LIWC-based cognitive-semantic dimensions (Insight, Causation, Differentiation, Discrepancy, Tentative, Certainty, and first-person singular pronouns) across three stages (early perceived plausibility, middle active engagement, and final fraud realization). In the early stage, victims used more tentative and exploratory language, indicating uncertainty and cognitive vigilance. In the middle stage, greater use of logical reasoning and emotionally engaged expressions reflected deeper cognitive involvement and affective investment. In the final stage, increased certainty and emotional resolution signaled recognition of deception and cognitive-emotional repair. Fraud type also moderated these patterns: economically motivated scams (such as online job and investment fraud) elicited more rationalizing and self-justifying language, whereas trust-dependent scams (such as romance and acquaintance fraud) involved more socially and emotionally oriented expressions. The findings show that linguistic indicators effectively track victims’ psychological trajectories during fraudulent interactions, providing psycholinguistic evidence for fraud detection and informing prevention, psychological assessment, and language-based monitoring systems.