The Voice But Not the Song: A Shorthand Hypothesis and the Statistical Fingerprint of the Voynich Manuscript

Abstract

We report a computational analysis of the Voynich Manuscript (Beinecke MS 408) using two independent pipelines (one treating each character as a letter, the other as a syllable). Both produce consistent structural conclusions: the source language is Romance (a mix of Latin and Italian), the content is medieval medical/herbal, two distinct subsystems coexist, and the morphological structure is genuine.

We identify Italian syllabic tachygraphy, a medieval shorthand tradition documented in northern Italian notarial archives, as the encoding mechanism most consistent with the manuscript's statistical fingerprint, discriminated from 13 alternatives including the Naibbe dice cipher. A signal isolation method adapted from Ventris's decipherment of Linear B identifies 56 decoded words as genuine under permutation testing against a merged Latin-Italian dictionary. The real assignment table produces significantly more signal words than 1,000 random consonant-vowel (CV) mappings (p = 0.001). Only 1.1% of random tables reproduce the linguistic coherence of these words: Italian verb conjugations, a complete Romance function-word inventory, and pharmaceutical terminology appearing simultaneously (p = 0.011; p = 0.006 under the CVC coda model). A separate permutation test validates 22 word-level content identifications (pharmaceutical Latin: ratione, coralli, diasene, stercora; p = 0.009). Individual words and syllables have been decoded, but connected readable text has not yet been achieved.