Constructing Climate Change Discourse in the Digital Public Sphere: A Computational Analysis of Social Representations

Public discourse on climate change is highly complex and rapidly evolving, especially in digital spaces. On social media, people attribute responsibility by linking abstract accountability concepts to concrete entities such as governments, corporations, or nations, thereby creating shared understandings of who is responsible and how to address the issue. Understanding how these collective meanings are built, maintained, and challenged through everyday communication is a key concern for discourse-focused social research. Web platforms intensify this challenge by producing large volumes of text, where representations develop dynamically and often fragment across communities, making systematic analysis difficult both theoretically and methodologically. Social Representation Theory (SRT) provides a solid framework for this, viewing shared knowledge as emerging from communicative processes that make unfamiliar phenomena understandable. SRT relies on anchoring, which integrates new info into existing categories, and objectification, which turns abstract ideas into tangible images, helping circulate these ideas socially. These processes are especially prominent in contentious areas like climate change, where scientific uncertainty, moral values, and political views intersect. However, applying SRT to large digital datasets requires tools that operationalise these concepts without oversimplifying their interpretive richness. To fill this need, we present ThemeScope, an analysis pipeline that detects and interprets social representations in large text datasets. It combines network-based community detection and visualisation, linking qualitative theory with quantitative text analysis. To implement SRT core mechanisms, we use a Prototypical Salience Index to measure the importance and centrality of concepts via anchoring, and a Concreteness Score to assess how well themes are expressed through tangible, experiential indicators of objectification. Plotting themes in a space defined by these two measures helps identify core representations, peripheral ideas, emergent discourses, and hidden patterns. We demonstrate its use with a large set of Reddit comments on climate change, showing how different discursive communities shape, negotiate, and stabilise diverse social representations, and illustrating how computational tools can enhance SRT analysis of digital public spheres.