Using Sketch Engine to explore two sets of data, a corpus of US Supreme Court opinions and a corpus of opinions from Poland’s Trybunał Konstytucyjny (in literal translation: Constitutional Tribunal), this paper explores the use of evaluative language in the context of judicial justification. Adopting a corpus-driven approach, the analysis has shown that a number of 3-4-grams are found in co-occurrence patterns with value-laden lexis in both the SCOTUS and the Constitutional Tribunal justifications. In semantic terms, these expressions have been found to serve as pointers to evaluation and as clues to the textual segments where argumentation unfolds. The scrutiny of the relevant co-texts has revealed that these phrases tend to be utilized as building blocks of judicial discourse to help frame interpretive and argumentative concerns.