Abstract This study examines the evolution of moral expressions in popular music lyrics over six decades. Using the WASABI dataset (1960-2010) and Billboard year-end charting songs (1960-2023), we analyse temporal trends in moral narratives across artist genders and musical genres. Transformer-based language models fine-tuned for moral foundation prediction were applied to quantify ten moral dimensions derived from Moral Foundations Theory. Our findings reveal an increase in moral vices (e.g., Harm, Cheating, Subversion), and a decline in expressions of moral virtues (e.g., Care, Purity), alongside a rise in negative sentiment, anger, and disgust. We show that moral dimensions can be inferred by lyrical cues, such as thematic content, sentiment, and emotion, with predictive accuracy improving notably when models are trained on specific music genres. These findings reveal that shifts in lyrical morality co-occur with broader societal changes in values and identity, underscoring how popular music may serve as a cultural barometer for evolving moral norms.
Preniqi et al. (Wed,) studied this question.