This paper adopts a pluralist conceptualization of the construct of moral understanding, in particular within the context of audiences’ verbalization patterns in response to watching morally-laden narratives on YouTube. We propose eight unique dimensions along which expressions of moral understanding can be measured in naturalistic settings and demonstrate the pragmatic feasibility of leveraging Large Language Models for moral content analysis. Our findings suggest a heightened prevalence of dimensions including emotional response, moral awareness, moral contextualization, moral reasoning, and aesthetic perception in audience responses, suggesting the multifaceted ways in which audiences cognitively and affectively interpret moral themes as constructed within narratives. Additionally, we find that specific moral content dimensions, i.e., harm and loyalty, differentially activate multiple moral understanding dimensions. In doing so, this study provides initial evidence of the existence of robust linkages between moral content, as exemplified within popular short-film narratives, and the saliency of audiences’ expressed moral understanding.