Concept-Conditional Cross-Tradition Binding in Semantic Embedding Space: A Method and an Application to Mysticism

An NLP methodology paper introducing concept-conditional cross-tradition binding (CCB), a bias-aware embedding-based statistic for testing whether textual passages from unconnected source traditions are more similar when conditioned on shared structural concepts than when not. Stress-tested on the 65-year-old cross-cultural mysticism convergence debate (Stace 1960; Katz 1978; Forman 1990; Hood 1975).

This release archives the paper (Draft 5), the corpora (Phase 0: 143 passages across 23 traditions; Phase 1a: 920 chunks sampled from 20 whole books across 11 traditions), the full analysis pipeline (passage- and sentence-level CCB, OpenAI text-embedding-3-large and ONNX MiniLM backends, vocabulary substitution, paraphrase-exclusion robustness, technical-only-tagger variant), and the result tables.

Released MIT-licensed for replication, extension, adversarial reuse, and application to other convergent-claim test cases.