The Voynich Manuscript: A 1,609-Line English Translation with a Cross-Validated Procedural Grammar
Year: 2026
Type: preprint
Abstract: This work presents a 1,609-line English translation of a constrained corpus of the Voynich Manuscript derived from a frozen compositional grammar developed through iterative structural, morphological, positional, and held-out validation. The accompanying Translation Edition renders the manuscript into readable English while preserving explicitly unresolved lexical material in brackets. The underlying corpus contains 34,776 normalized tokens across 4,084 lines. The frozen structural classifier licenses 30,534 tokens (87.802%) under the canonical Zandbergen–Landini ZL3b normalization. A fully constrained 1,609-line benchmark contains 12,495 token positions, of which 12,294 (98.3914%) receive a defensible functional-role interpretation, leaving 201 explicitly opaque positions. Post-freeze adversarial replication using the RF1b reference transliteration and alternate uncertain-space treatments shows that the principal grammatical and functional-role architecture survives transcription changes, although exact whole-corpus coverage falls to 86.065% on RF1b. The deposit therefore distinguishes the robustness of the recovered architecture from transcription-dependent surface coverage. The release includes the English Translation Edition, controlled translation audit, frozen role lexicon, classifier data, adversarial audit, R1/R2 replication results, reproducibility code, manifests, and checksums. The work does not claim recovery of the manuscript's historical phonology or a complete dictionary mapping of every lexical root.
Author: Dominik, Matthew
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7203884108
Cites:
Cited by:
Related to: