Distinguishing AI-Generated and Human-Written Text Through Psycholinguistic Analysis

TL;DR

The work addresses the rising challenge of distinguishing AI-generated from human-written text in academic contexts by proposing a theory-grounded framework that integrates stylometric analysis with psycholinguistic theory. It maps $31$ stylometric features to cognitive processes, spanning $6$ feature categories, and includes $12$ novel metrics to enhance interpretability. This mapping links concrete linguistic signals to underlying processes such as $Cognitive Load Theory$, metacognition, lexical retrieval, and discourse planning, offering explanations for why features discriminate authorship. The framework aims to improve transparency and reliability of authorship detection tools, supporting integrity in education and other domains where authentic human writing matters.

Abstract

The increasing sophistication of AI-generated texts highlights the urgent need for accurate and transparent detection tools, especially in educational settings, where verifying authorship is essential. Existing literature has demonstrated that the application of stylometric features with machine learning classifiers can yield excellent results. Building on this foundation, this study proposes a comprehensive framework that integrates stylometric analysis with psycholinguistic theories, offering a clear and interpretable approach to distinguishing between AI-generated and human-written texts. This research specifically maps 31 distinct stylometric features to cognitive processes such as lexical retrieval, discourse planning, cognitive load management, and metacognitive self-monitoring. In doing so, it highlights the unique psycholinguistic patterns found in human writing. Through the intersection of computational linguistics and cognitive science, this framework contributes to the development of reliable tools aimed at preserving academic integrity in the era of generative AI.