Medical practice is a multi-layered phenomenon characterized by systemic complexity, epistemic uncertainty, and outcome‑oriented nature. However, legal discourse in medical litigation often reduces this complexity by employing active/passive grammatical constructions, framing clinician– patient interactions in terms of a simplistic offender/victim binary. In response, this article introduces the concept of the middle-voice, defined by two grammatical features: (1) the internal involvement of human and non-human elements in the occurrence of action, and (2) the dynamic emergence of action through contextual interactions. It proposes a middle-voice structure that incorporates antecedent-condition clauses, inanimate subject constructions, and the Korean “-ge doeda” form (equivalent to the English verb “become”). This framework supports the development of middlevoice justice, which moves beyond adversarial models and promotes a more inclusive culture of medical accountability and safety. By offering a theoretical basis for institutional applications, such as in the design of apology laws, this approach enables more accurate descriptions of medical accidents, more balanced attributions of responsibility, and reconsideration of the institutional foundations necessary for a just medical culture.