Ideals and Reality: Cynicism Towards Diversity Discourse in US College Campuses

Abstract

Student-run media offers a unique window into the evolving mindsets of future leaders and the sociocultural climate on college campuses. This research applies state-of-the-art embedding-based text analysis to student opinion articles from eight leading U.S. universities from 2010 to 2024, tracing shifts in sentiment and cynicism alongside changes in the diversity of discourse. Using a novel combination of sentiment analysis, semantic vector embeddings, lexicon-based scoring, and large language model classification grounded in Moral Foundations Theory, we detect a significant rise in cynical tone and negativity over time, alongside a measurable decline in the diversity of viewpoints within campus discourse. Our findings suggest that student writing has become increasingly homogenized and skeptical of institutional authority, reflecting broader undercurrents of eroding institutional trust and changing values. We discuss how correlating these linguistic trends with current events provides insight into potential drivers of sentiment shifts across colleges. This interdisciplinary study highlights how AI techniques can uncover hidden campus dynamics in text by bridging computational analytics with insights from psychology, sociology, and political science. Our results demonstrate a novel approach for technologists, educators, and policy professionals to monitor public trust, civic attitudes, and the health of discourse within institutions.