Moral Language Boosts Online Engagement Only Up to a Point
Researchers analyzed posts and comments from three online platforms to test how moral language relates to engagement. They found that engagement rises with moral content only up to a point, then declines as moral language becomes more concentrated. The study, published in Nature Human Behaviour, distinguishes a message's moral relevance from its saturation with moral wording.
A team from Universidad del Desarrollo, UMass Amherst, and Northwestern examined roughly 1.62 million Twitter and Reddit posts plus 8chan discussions across 13 topics. Published in Nature Human Behaviour, the work separated a message's moral relevance from how densely moral wording was packed into it.
Using Distributed Dictionary Representations, the researchers mapped words as numerical vectors and estimated each post's moral loading and moral density. Engagement reportedly increased with moral content only until a threshold; beyond it, reposts and replies tended to fall.
This finding may influence how social media users, community moderators, and platform designers think about moral rhetoric. If engagement peaks before moral language becomes saturated, communicators could adjust tone, while platforms may reconsider ranking signals. Broader publics might benefit from less polarized exchanges, though effects depend on context and platform norms.