Tag Archives | Word Embeddings

Although many argue that American society has become overly politicized, social scientists have yet to systematically measure “politicization”—the process by which issues are framed as political through their association with recognizable political divisions in public discourse. This article develops a new measure of politicization using à la carte word embedding models trained on 1.35 million articles from three national newspapers. I trace the historical politicization of 25 issue domains in the United States from 1980 to 2024 by examining the extent to which specific issues are discursively associated with ideological labels (liberal and conservative) and partisan markers (Democrat and Republican). The analysis yields three main findings. First, politicization in the American public sphere has increased substantially over the past four decades, and this increase has extended beyond traditionally political issues into previously less politicized aspects of social life. Second, the discursive locus of political conflict has shifted over time: race and immigration, gender and family, and legal institutions have become more politicized than economic issues, which had been among the most politicized domains in the 1980s. Third, the boundary between the political and the less-political remains consequential, as some important areas of social life such as the home, financial markets, and outer space remain relatively unpoliticized in public discourse today. By studying politicization as a phenomenon in its own right, this research contributes to cultural and political sociology by treating “politics” as a dynamic category whose boundaries are continuously renegotiated, while also reorienting scholars of political division to shift their focus from public opinion and formal institutions toward the “public culture” through which social conflict and cohesion are represented and experienced.

This work is licensed under a Creative Commons Attribution 4.0 International License.