• Options
• Login
• Desktop Style

Token-level rules applied to EVA transcripts – reproducible code example

• Index

  • The Voynich Ninja
  • Voynich Research
  • Theories & Solutions
  • ChatGPTrash
  • Token-level rules applied to EVA transcripts – reproducible code example
• Thread Closed
• Token-level rules applied to EVA transcripts – reproducible code example

• RE: Token-level rules applied to EVA transcripts – reproducible code example

  fran9262 > 24-09-2025, 09:28 PM

  (24-09-2025, 09:20 PM)Mauro Wrote: You are not allowed to view links. Register or Login to view.If I understood, your proposal is that the VMS is a kind of You are not allowed to view links. Register or Login to view.. Which it might well be!
  
  But the problem of your approach (and of innumerable others) to the 'translation' is that it's way too easy to assign some arbitrary meaning to some tokens and then build a whole theory upon that. In the last few months we have had here at least a dozen different solutions, going from slavic languages to celtic languages to musical notation to alchemical procedures. All of them were internally 'coherent', but did any of them hit the mark? No: they were all speculations without evidence (albeit coherent speculations).

  
  
  Thanks — I think you’ve understood me correctly, and I do see the risk you’re pointing out. It is easy to assign arbitrary meanings to tokens and spin a theory that looks coherent but isn’t testable. That’s why I’m trying to present this as a rule-based framework, not a narrative translation.
  
  What I’m aiming to demonstrate is:
  
  Transparency: Every substitution is tied to a prefix/suffix/infix rule that can be checked against any EVA line. If the rule fails, the output is marked [?], not filled in with a guess.
  
  Consistency across folios: Labels like chedy, olkeeody, and olkarar recur dozens of times. My goal is to show they always map to the same semantic category (plant, liquid vessel, tub) regardless of folio or context.
  
  Alignment with imagery: I deliberately started with labels tied to clear illustrations (tubs, pipes, vessels) because that provides an external anchor for testing whether “herbal vessel” or “liquid container” is a plausible fit.
  
  So I’m not claiming a solved language, but I am trying to move away from arbitrary assignment toward something reproducible. If someone else applies the same rules to the same labels, they should get the same outputs — including the same unresolved tokens.
  
  I know that doesn’t resolve the larger question of whether this is “the solution,” but I’m hoping to at least contribute a testable framework rather than another closed theory.

• RE: Token-level rules applied to EVA transcripts – reproducible code example

  fran9262 > 24-09-2025, 09:39 PM

  (24-09-2025, 09:20 PM)Mauro Wrote: You are not allowed to view links. Register or Login to view.If I understood, your proposal is that the VMS is a kind of You are not allowed to view links. Register or Login to view.. Which it might well be!
  
  But the problem of your approach (and of innumerable others) to the 'translation' is that it's way too easy to assign some arbitrary meaning to some tokens and then build a whole theory upon that. In the last few months we have had here at least a dozen different solutions, going from slavic languages to celtic languages to musical notation to alchemical procedures. All of them were internally 'coherent', but did any of them hit the mark? No: they were all speculations without evidence (albeit coherent speculations).

  
  What I’m showing here is not just an interpretation of the Voynich words, but how they fit into a documented medieval system of shorthand writing, specifically the compressed notation used in alchemical manuscripts.
  
  In the 14th–15th centuries, alchemists and physicians often wrote in what we’d call alchemical shorthand: Latin words were stripped down, abbreviated, or reduced to symbols so that complex instructions could be written quickly and concisely. For example, coquere (“to cook/boil”) was usually abbreviated as “coq.”, materia (“base matter”) as “mat.”, and ignis (“fire”) as “ign.” or even just a flame sign.
  
  The Voynich tokens line up with these same practices:
  
  qok- mirrors the contraction of coquere → coq. → qok- (boil/infuse).
  
  -ody plays the same role as materia → mat. → ody (base matter).
  
  she- acts like the shorthand ignis → ign. (fire/calcination).
  
  In other words, the Voynich text behaves like a compressed technical script, not like an ordinary language. Each token is made of modular parts (prefix, suffix, infix), which can be cross-checked across multiple folios.
  
  This is what separates the approach from arbitrary “translations”: the same rules apply wherever the tokens appear, and they match known shorthand forms in contemporary alchemical literature (Aurora Consurgens, Liber de Alchimia, pseudo-Lullian works).

• RE: Token-level rules applied to EVA transcripts – reproducible code example

  tavie > 24-09-2025, 09:50 PM

  OK, I'm closing this too. Please don't start a third thread for the same thing.  I'll respond to your PM soon but for now please read You are not allowed to view links. Register or Login to view. about why you cannot trust LLMs.
• Next Oldest Next Newest