This study investigates shifts in blame attribution within the Danish Parliament over a nearly three-decade period, spanning from 1997 to 2026. To analyze this longitudinal data, the researchers developed BlameBERT, a specialized classifier designed for low-to-mid resource languages, which achieved an F1 score of 0.80 using an annotation-efficient pipeline. By applying multilevel statistical modeling to the parliamentary records, the authors identified a "banana-shaped" trajectory in political rhetoric; blame attribution levels declined until approximately 2016 but have since experienced a significant and sustained increase between 2019 and 2026. The research highlights a phenomenon termed "political contrasting," where opposition parties assign blame at substantially higher rates than governing parties. This effect is heavily moderated by political ideology. Specifically, the tendency for governing status to dampen blame is less evident among right-wing parties, and ideological extremity has been found to amplify blame more aggressively on the right side of the spectrum. The findings suggest that the contemporary rise in hostile political language is not a uniform trend but rather an ideologically asymmetric hardening of discourse, concentrated primarily among right-wing ideological extremists. Sensitivity analyses confirmed that these conclusions remain robust regardless of the classification thresholds applied to the BlameBERT model.