Finding Patterns Across Multiple Time-Series Datasets

This study investigates the contextual variation of nouns and adjectives associated with ‘democracy’ in twentieth-century Britain, Sweden and Finland. By quantitatively analysing parliamentary debates and conservative and liberal press sources from three countries, we explore the interplay between political and media discourses on democracy. Our methodology involves visualising the use of democratic terms over time and applying correlation analysis to detect patterns between multiple datasets. We use the Pearson correlation coefficient to quantify similarities between word frequency time series, identifying strong and weak relationships. Our findings reveal the strongest correlations between similar political terms within the same dataset and between identical terms in different national datasets from the same country. Notably, there is a statistically strong correlation between media and parliamentary discourses, partly explained by entanglements between journalism and political parties. Although transnational correlations are weaker than intra-national ones, they still suggest common trends across nations. Close reading highlights key periods where democratic discourse peaked, corresponding to global events like the challenge of totalitarianism in the 1930s, the rise of the left in the 1940s, the emergence of social movements and the extension of democracy after 1968, and the dissolution of the Eastern bloc in the early 1990s. This study introduces time-series methods to quantitative and comparative conceptual history. We demonstrate empirically a strong linkage between political discourses in parliament and the press, showing that parliamentary speech has often been in sync with broader societal debates – at least in the case of democracy and the quality press.