This paper introduces the IRIS method (Interpretive Rotation on Invariant Signals) — a procedurally explicit analytical protocol for making visible the interpretive architectures through which shared evidence is processed into divergent conclusions. The method treats interpretive frameworks as grammars: rule-bound processing systems defined by five explicit parameters (V, S, W, C, A). The analyst holds the empirical input constant, rotates the grammar, and observes what changes in the output. The paper specifies the method's formal architecture, demonstrates it through a worked example (Citizens United v. FEC), summarises a preliminary convergence programme involving 15 distinctively prompted large-language-model instances across 20 case studies, and positions the method relative to discourse analysis, frame analysis, Q-methodology, analytical pluralism, argumentation theory, and the sociology of knowledge. Working paper / preprint.
Rodrigo Jazinski (2026) studied this question.