We introduce the Structural Hermeneutic Drift (SHD) framework, a computablemodel for decomposing and measuring the process by which the interpretive architecturesurrounding a culturally stable lexical marker is structurally replaced. Unlike diachronicword-embedding methods, which measure distributional distance between temporal snapshots,SHD takes the drift event as its unit of analysis: a phased trajectory from original framethrough friction, amplification, stabilization, and condensation into a replacement frame.The framework derives a suite of descriptive metrics—including Institutional AuthorityConcentration (ICA), Frame Compression Depth (FCD), and Drift Success Index (DSI)—from expert-coded historical inputs.We evaluate the framework’s internal consistency against eleven canonical historical cases(seven positive, three negative, one borderline), including two timestamped pre-coding heldoutevaluations (Judas Iscariot, Methuselah). The coding logic produces perfect ordinalseparation between POSITIVE and NEGATIVE cases (Mann-Whitney U = 21/21, exactp = 0.008), demonstrating strong internal coherence and establishing a foundation for largerscalevalidation. A general-purpose sentence-embedding baseline applied to the same textsillustrates the risk of conflating structural frame replacement with lexical variation. Acomparative analysis of PCA, t-SNE, and UMAP projections of OpenAI embeddings showspersistent spatial overlap between cases of structural drift (POSITIVE) and cases without drift(NEGATIVE), illustrating that latent geometry alone cannot recover the historical structureof frame replacement. Ultimately, SHD does not replace embedding-based semantic analysis;it constrains and contextualizes it through historical structural decomposition, providing aformal, auditable vocabulary for the comparative analysis of interpretive transformation.
Ruben Abella (Sat,) studied this question.