ABSTRACT How do political blocs form, consolidate, and dissolve? And can we trace these dynamics through the language of parliamentary debate? We address these questions by analyzing speeches in the Swedish Riksdag, using supervised machine learning to examine changes in rhetorical alignment among parties. Adapting classification methods developed for two‐party systems to a multiparty context, we assess rhetorical alignment between and within Sweden's political blocs across nine legislative periods (1991–2023). The results closely match expectations: rhetorical cohesion between parties was strongest during the peak of formal alliances (2006–2014) and declined during subsequent periods of political realignment. These findings demonstrate that parliamentary rhetoric offers a valuable lens for tracking the rise and fall of political blocs. More broadly, the study shows how supervised machine‐learning classification can capture long‐term shifts in party‐system structure within multiparty democracies.
Wising et al. (Fri,) studied this question.