Abstract
<p>Background: Japan lacks statutory legislation for euthanasia, physician-assisted suicide, or treatment withdrawal; case law from the 1962 Nagoya Yamauchi judgment through the 1995 Tokai University four-criteria framework, the 2009 Kawasaki Supreme Court decision, and the 2024 Kyoto ALS convictions has established the de facto normative criteria.Methods: We present the first systematic computational analysis of Japanese euthanasia-related judicial discourse (N = 113 bibliographically verified judgments, 1964–2024; sensitivity N = 119) using a 14-layer robustness hierarchy that pairs lexicon-based modelling with Tohoku-BERT v3 fine-tuning, leave-one-out sensitivity, a scramble-label negative control, and three-LLM inter-model reliability.Results: Relative to the three-factor public benchmark from Watamura, Hunt, and Ioku (2026, Ethics &amp; Behavior; Euthanasia Attitude Scale, Tordella and Neutens 1979; analytic N = 890), judicial discourse shows between-scale standardized discrepancies (not within-population Cohen's d) of −3.21 on openness-to-ending-life, −4.26 on dignified self-determination, and +1.49 on family-and-social-order, under primary scaling convention pooled-SD, supported under three of four conventions and across all 113 leave-one-out drops.Conclusions: The pattern admits an institutional-thoughtlessness two-axis reading (judicial discourse de-emphasizes both patient autonomy and the broad openness-to-ending-life construct that F1 captures, of which suffering-termination is one component) and an institutional-prudence counter-reading anchored in Japanese communitarian bioethics; Stage 2 is designed to adjudicate.</p>