Soft Power in the Multimodal Age

Cinema plays a central role in the construction of cultural meaning, national narratives and international perception through the concerted use of text, imagery and sound. However, current methods in the field of media and communication studies (and the majority of computational multimodal models) tend to analyses these modalities separately or tend to focus on predictive performance rather than interpretability, which limits understanding of the systematic alignment of ideological meaning across modalities. This study presents an Ideology-Focused Multimodal Transformer (IF-MMT) for analyzing the construction of narratives of soft power in state-sponsored cinema. Unlike conventional multimodal architectures, IF-MMT explicitly models ideological alignment between textual, visual and audio streams, which allows both identification of ideological stance and estimation of persuasive intensity. The framework is evaluated on a curated corpus of 120 Chinese state-sponsored movies and achieves improved performance compared to the state-of-the-art multimodal models in the ideology’s classification and cross modality alignment consistency. These results suggest that the influence of ideologies in cinema is coordinated structure-wise rather than additively communicated. By combining computational modelling with media and cultural analysis, this work is a replicable methodological contribution to research in the humanities and social sciences and new tools to analyze the role of multimodal cultural products in the construction of ideological meaning and soft power in the contemporary media landscape.