Semantic Non-Identifiability in the Voynich Manuscript: Negative-Controlled Tests of Meaning-Slot Assignment

This record contains the manuscript: **Semantic Non-Identifiability in the Voynich Manuscript: Negative-Controlled Tests of Meaning-Slot Assignment** This study examines whether internally derived structural evidence from the Voynich Manuscript is sufficient to promote candidate token roles into stable semantic meaning slots. Using the Zandbergen-Landini EVA transcription (ZL3b), the paper evaluates a small strongest-case candidate queue through dependency audit, score recalibration, and strict matched-holdout testing. The main result is a negative-control boundary. Initial apparent support was high, but after alignment-dependent components were removed and candidates were compared against matched controls, zero of seven candidates survived as robust semantic promotions. The result supports a distinction between structural constraint and semantic identification: Voynich token behavior may be strongly structured while remaining semantically non-identifiable under internal evidence alone. This paper does **not** claim decipherment, translation, plaintext recovery, source-language identification, concrete referent identification, or confirmed lexical meanings. It is intended as a methodological contribution for evaluating future semantic claims about the Voynich Manuscript. The release includes the manuscript only. Exploratory materials, token-role dictionaries, unpublished interpretation notes, candidate-generation materials, and reproduction packages are intentionally excluded.