Mass media is often investigated for its influence on public opinion. However, media analysis often relies on measuring term prevalence, elements of framing, and determining bias. New approaches to media analysis are advantageous to the social sciences. Leveraging the German General Social Survey (GGSS), we utilize Latent Semantic Analysis (LSA) to categorize and compare discourse for key points of time (2006, 2010, 2012, 2016, 2021), with over 10,000 media articles from several German media outlets. We focus on the migration and integration of foreigners in Germany and the competing discourse narratives surrounding these events. We adapt the term “foreigner” (Ausländer) in media text; German compound variations such as Ausländerproblem (foreigner problem) and Ausländerintegration (foreigner integration) are central to the discourse analysis. Based on semantic meaning and co-occurrence, these compound terms are grouped into four categories: Administration and Policy, Social Integration, Xenophobia, and Limiting Migration. Results demonstrate that Social Integration discourse becomes more prevalent over time. A subsequent sentiment analysis reveals that Social Integration discourse is not positive but neutral – other categories reflect a negative bias. We therefore discuss computational applications for the enhancement of media analysis, as well as challenges to contextualizing survey data.