Researchers from Universidad del Desarrollo in Chile, the University of Massachusetts Amherst, and Northwestern University in the U.S. have discovered that the connection between moral language and online engagement is not unlimited. Their study, published in Nature Human Behaviour, examined posts and comments from three online platforms to see how using moral language affects user engagement. The results show that while expressing moral ideas generally increases engagement, using a higher concentration of moral content—called moral density—can actually reduce engagement when the overall amount of moral language remains the same. To conduct their research, the team combined previously released data sets that included Twitter posts and engagement metrics with archived discussions from Reddit and 8chan. They analyzed about 1.62 million observations, covering Twitter and Reddit posts as well as 8chan discussions, across 13 topics and communities. Using computational language analysis, they measured how closely messages connected to moral concepts and how concentrated that language was. They then looked at how this related to retweets or replies, depending on the platform, while considering features like links and multimedia. They also confirmed their findings using traditional moral-word dictionaries. The researchers used a computational technique called Distributed Dictionary Representations to assess how closely language connected to concepts in a predefined dictionary. They used a pre-trained model to represent words as numerical vectors, based on their meaning and the contexts in which they appeared. They calculated the moral loading (the extent to which a post contained moral ideas) and moral density (how concentrated the moral content was in the words used) for each post. Their findings showed that across all three platforms, the use of moral ideas was consistently linked with higher engagement. However, when the overall moral loading remained the same, higher moral density was associated with lower engagement. The model estimated that engagement reached its peak at a moral-density score of approximately 0.30 on average across all platforms and topics, and engagement dropped as the scores increased. The researchers emphasized that their study was observational, not experimental, and therefore does not prove that using more moral language causes changes in engagement. They suggested that future research could test this hypothesis through controlled experiments. The study's findings could provide useful guidance for communicators and content creators online. One question the researchers raised is whether AI-generated messages use moral language differently from human-written messages and how readers respond to those differences. They also noted that more research is needed to understand why an overuse of moral language is linked to lower engagement and whether similar patterns apply to trust, understanding, and sustained attention.