Moderation or Amplification? Mapping Speech Distinctiveness Among Green Parties Using a Machine Learning Approach

ABSTRACT

How do issue‐owning parties adapt their rhetoric when transitioning from opposition to government? Focusing on the Green parties in Germany and Ireland, we analyze legislative debates to trace how these parties seek to maintain ownership of environmental issues. Using more than 1.7 million parliamentary speeches from the German Bundestag (1991–2018) and the Irish Dáil (1992–2013), we apply a fictitious prediction problem approach implemented as a multiclass text‐classification task to measure distinctiveness in parties' legislative speeches. Our findings show that both parties increased their rhetorical distinctiveness on environmental issues during their first time in government, but with key differences: the Irish Greens maintained a distinct profile throughout their term, while the German Greens' distinctiveness increased as elections neared. No such patterns emerge for welfare, a non‐owned issue. Our study offers a framework and a transferable method for studying how smaller parties navigate between coalition influence and visibility through the issues they own.