Classifying Interpretive Canons at the Sentence Level: A Benchmark from the German Federal Constitutional Court

Interpretive canons play an important role in statutory interpretation, especially in civil-law systems. This paper contributes a benchmark for evaluating the ability of LLMs to classify interpretive canons as articulated by Larenz in the tradition of Savigny, a core part of legal methodology in many jurisdictions, using decisions of the German Federal Constitutional Court. Our contributions are threefold. First, we operationalize this conception of the canons into granular criteria. Second, we provide a dataset of fifteen decisions annotated exhaustively at the sentence level, supplemented by selective annotation of further decisions to boost the positive count for rarer classes. Third, we report baseline evaluations of LLMs under three sets of prompts: one hand-written by a PhD-level expert and two optimized with Genetic-Pareto (GEPA).