(Trans-)Local Integration in Urban Twitter Issue Discourses: A Semantic Network Analysis Approach

The question of whether digital discourses are integrated or fragmented has received great scholarly and public attention. Fragmentation between different political camps, different social groups, or different geographical regions are seen as threats to social cohesion. Network analytical approaches are often utilized to understand the patterns of integration or fragmentation in digital public discourses. Usually, researchers conceptualize integration and fragmentation at the actor level. They utilize social network analysis to understand who communicates with whom and where holes and disconnections emerge. Yet, integration and fragmentation are also located at the content level. By communicatively connecting certain issues, actors, or places, speakers highlight them as belonging together. This paper therefore uses semantic network analysis for a new perspective on discursive integration. Because of the outsized role of the local environment in fostering social cohesion, it draws attention to local, urban digital discourses. Specifically, it asks: What patterns of (trans-)local discursive integration/fragmentation emerge in digital discourses around urban political issues? To address this question, the use of discursive references to places is studied. When events in different localities are discussed in conjunction with each other, they become integrated. In the public imagination, they face shared concerns and are constructed as belonging to a common ‘issue space.’ The paper uses Twitter data around four urban policy issues, spanning two German cities (Berlin, Frankfurt am Main) and two issues (housing, cycling infrastructure). References to districts, streets, and neighborhoods in the cities were identified using semi-automated content analysis. Semantic networks were then constructed by identifying instances where multiple places were referenced within the same message. Places serve as nodes, while ties indicate co-occurrences. Across all case studies, the networks are dense, but with largely weak ties, showing that places all over the cities are discursively integrated, but at a low level. The strongest integration emerges between places within the urban center. Peripheral places are less densely integrated and, if they are discussed, it is most likely in connection to the center. At the same time, the networks are locally clustered, which indicates that geographically proximate places are more integrated with each other. Exponential Random Graph Models (ERGMs) demonstrate that (trans-)local integration can be predicted through properties of the places: Places where the issue is more prevalent (e.g., rents rise more quickly) are more discursively integrated. Moreover, places with more communicative resources (e.g., socioeconomic resources of the population, community organizations) are more integrated. The paper offers a perspective on the question of integration in digital discourses which is focused on semantic connections, rather than social relations. It draws attention to the local scale and shows that highly consistent patterns emerge across different discourses. Finally, it shows that discursive integration, visibility, and belonging in these digital issue spaces are contingent upon the resources local populations can rely on for public communication and organizing.