Exploring Partisan Interest in Language of Judicial Opinions

Although court partisan influence is somehow intrinsic, it remains unclear the degree of influence partisan justice nomination has in the language of judicial opinions and the main subject of the influence of attention by the given parties. In this article, we explore current computational methods to assess and determine the main subject topics that the partisans provide a particular focus on in large-scale annotated corpora. Making use of a publicly available dataset with court opinions from the \gls{usa}, we conduct an exploratory analysis and estimate word embeddings found in legal opinions conditioned to the identified partisans to identify what the subjects were partisan are keener to give attention, and in combination with trained models identify the particular author of the individual opinion. In an exceeding series of exploratory analyses and word embeddings, we discover strong evidence that dominant topics such as economics and land are addressed mainly by right-wing partisan nominated justices. At the same time, the left-wing focuses on social and racial disparities in society, with partisan nomination showing clear evidence that the produced opinions among republicans nominated justices rely on a well-defined set of topics, suggesting the existence of a common language and topics produced by this subgroup of justices. The results show some implications for the neutrality of the language for understanding the entrenchment of biased decisions on specific partisan and political topics of interest.