Visualizing Emptiness and Beyond: A Stylometric and Adversarial Analysis of Mahayana Buddhist Concept Illustrations as Intersemiotic Translation

This study investigates how core Mahayana Buddhist concepts—including but not limited to emptiness (空), enlightenment (菩提), nirvana (涅槃), prajñā (般若), middle path (中道), dharma realm (法界), Buddha-nature (如來藏), bodhisattva stages (十地), true vs. deluded mind (真/妄心), and great compassion (大悲)—are visually rendered across seven fully illustrated Buddhist volumes. Treating these illustrations as intersemiotic translations of doctrinal teachings, we construct a multimodal visual-textual corpus comprising over 600 images. Stylometric profiling (via CLIP embeddings), adversarial modeling, and entropy-based metrics are used to test whether doctrinal styles possess computationally detectable “visual fingerprints.” We also explore the role of adversarial classifiers in flagging doctrinal inauthenticity in modern adaptations. This project advances Translation Studies by introducing a novel paradigm to evaluate sacred image translation through both machine learning and doctrinally informed visual hermeneutics.