• Options
• Login
• Desktop Style

The Voynich Manuscript: Understanding the Untranslatable with Lisa Fagin Davis

• Index

  • The Voynich Ninja
  • Voynich Research
  • Voynich Talk
  • The Voynich Manuscript: Understanding the Untranslatable with Lisa Fagin Davis
• Thread Closed
• The Voynich Manuscript: Understanding the Untranslatable with Lisa Fagin Davis

• RE: The Voynich Manuscript: Understanding the Untranslatable with Lisa Fagin Davis

  eggyk > 13-09-2026, 10:29 AM

  (13-09-2026, 05:18 AM)Torsten Wrote: You are not allowed to view links. Register or Login to view.English is not my native language. I have been criticized for repetitive phrasing when I wrote in my own English and also now that I use language tools. In both cases the content is mine. Every source is verifiable. Every quote is checkable. If any of my points is factually wrong, I would welcome a correction. The style of the prose doesn't change whether Bennett 1976 exists, whether Bowern wrote 'does not prove,' or whether Reddy & Knight found no vowel-consonant separation.

  
  You are not just using the tool to formulate the english language or to just using it to translate your words, but instead to generate the entire structure and flow of each post (at least). I'm highly doubtful that you are actually writing a draft post in your native language and then asking the LLM to translate it. 
  
  But fundamentally, the issue is that it changes whether we know that we are actually speaking to you. If the prose looks identical to what would be seen if you simply prompted "write me an in-depth post laying out my *insert argument here*, using the following *insert sources here*" how can we know that it is genuinely your content? Whether that content is true or not is an entirely seperate issue; it's the principle of knowing that we are actually communicating with eachother in a forum environment that matters here. It's noteworthy that the post that I am replying to/quoting actually feels like a post written by the real you. 
  
  Plenty of the AI assisted theories that arose here before the great moderator firewall were -on the face of it- reasonable sounding theories with interesting ideas. However, the amount of time needed to check them for mistakes was so large compared to the time taken to generate them that it became a huge issue. 
  
  However, I will say that it's clear that you are checking these outputs for mistakes, and are likely supplementing many details into them to ensure they are correct. But again, I can't know that for sure.

• RE: New VMS podcast

  Torsten > 13-09-2026, 10:55 AM

  (13-09-2026, 07:17 AM)asteckley Wrote: You are not allowed to view links. Register or Login to view.Like I said, I've disagreed with her on some things too and haven't hesitated to say so. But the problem here is the persistent adversarial responses on issues that you know perfectly well remain open. And without offereng any new evidence or new reasoning on the matters.
  
  Lisa’s judgments may be debatable, but so are yours. IMO, it would be more respectable to criticize claims with an aim to clarify the evidence and move the understanding of the various issues forward, not simply to create another occasion for a public denunciation of someone who hasn't bought into all of your own assessments.

  
  Andre, you write that I should "criticize claims with an aim to clarify the evidence and move the understanding forward" and that I do so "without offering any new evidence or new reasoning."
  
  Several of the issues I raise are not open questions. Whether a vowel-consonant separation exists was tested by Guy (1991) and Reddy & Knight (2011) — it does not. Whether Zipf's Law proves meaning was addressed by Bowern — it does not. Whether the entropy is compatible with natural language was measured by Bennett (1976), Stallings (1998), and Lindemann & Bowern (2021) — it is not. These are published results, not debatable opinions.
  
  When Davis states on a podcast that "it seems to be most closely related to some kind of Indo-European language," the appropriate scholarly response is to ask for the published source. If no such source exists, that is worth documenting — regardless of how many times the question has to be asked.
  
  You are right that I scrutinize Davis's and Bowern's claims more closely than those of other researchers. The reason is proportionality. Davis publishes as Executive Director of the Medieval Academy of America, presents at international conferences, appears on television and podcasts reaching million viewers, and uses the Beinecke Library at Yale as a platform (see You are not allowed to view links. Register or Login to view.). Bowern publishes as Professor of Linguistics at Yale in high-profile journals like You are not allowed to view links. Register or Login to view.. Their claims are cited uncritically by other researchers — as your own paper demonstrates by accepting Davis 2020 without examination. In my eyes the scrutiny should match the reach and the authority behind the claim.
  
  Regarding constructive engagement: your own paper "Subtle Signs of Scribal Intent" (You are not allowed to view links. Register or Login to view.) analyzes position-dependent token distributions in the Voynich manuscript — tokens at line starts, line ends, and adjacent to drawings. These positional effects were discussed in Timm 2014 and Timm & Schinner 2020. Your paper does not cite either. Citation choices in scholarship are never neutral — they show whom one acknowledges as a participant in the conversation and whom one excludes.

• RE: New VMS podcast

  Koen G > 13-09-2026, 12:37 PM

  (13-09-2026, 07:17 AM)asteckley Wrote: You are not allowed to view links. Register or Login to view.People need to catch up with the fact that it is now a commonly used tool in virtually all fields of writing -- as common as spell checkers, if not more.

  
  A habit being common does not mean that it's good. Not so long ago, smoking was as common as spell checkers are now. As late as the 80's, early 90s, manufacturers were even consciously aiming for the underage market (with cartoonish mascots etc.) in an attempt to further boost sales. Everybody's doing it, might as well light up, kid.
  
  As you tell us to get with the times and accept that AI written texts are part of everyday life now, I encounter more and more comments by people complaining that they "couldn't get through an AI-written article" with its weird language and inconsistencies. And I've been noticing that more and more VM-related searches lead to AI-written slop. We see a slop-ridden future manifest in front of our eyes.
  
  If you say AI has its place, I hear you. I would not want to stand in the way of genuine advancements AI could offer in certain fields. But you must also hear people when they complain about AI in places where it is - apparently - not appreciated. Let those places be those where we expect human redaction, like news broadcasts, books and research articles. Or where we go in search of human interaction: Facebook, Reddit, forums... These are the places where AI is violently forcing its way in, and where you see the most backlash.
  
  We want your bad English, we want your repetitive phrasing, we want your imperfections, because we want to talk to you as a human being. That's what places like these are for. If someone comes here with robotic takedowns of targeted individuals, that simply feels wrong, even when the arguments are logically valid. You see the reality: when AI-Torsten posts something, people complain about the style and we've all forgotten what the point was he's trying to make.
  
  Be real. We know this is an international forum, and we appreciate regional differences in the mastery of the lingua franca. Write in your own words in your own language and have AI translate it instead of "enhance" it, because the enhancements are nauseating to those of us who are fed up with having to figure out whether we're responding to an actual human being or not.

• RE: New VMS podcast

  Torsten > 13-09-2026, 12:55 PM

  I use AI assistance since writing a post in english is time consuming and exhausting to me. That said, the discussion about AI usage is missing the point.
  
  I compare claims against the cited literature. I document contradictions between what is said on podcasts and what is written in papers. I did this in 2020 and I do it now. The responses were ad hominem then and they are ad hominem now. My posts were rejected in 2020 and I doubt that they would be welcome now if they were written differently.
  
  If any of my points about the podcast is factually wrong, I would welcome a correction. That would be more useful than discussing how the English was formulated.

• RE: New VMS podcast

  asteckley > 13-09-2026, 01:06 PM

  (13-09-2026, 12:37 PM)Koen G Wrote: You are not allowed to view links. Register or Login to view.A habit being common...

  
  I don't disagree with anything you said there. (Except perhaps the apparent preference for poorly written content because it means no AI was used. I would rather
  read clearly stated, and well crafted posts so long as the semantic content itself is valid. It is when what is being said suffers from the use of AI that there is the problem. But the prevalence and the demonstrated value of AI as a writing tool is now well past objecting to its very use. That has become be a futile battle -- and in a remarkably short amount of time.)

• RE: New VMS podcast

  asteckley > 13-09-2026, 01:46 PM

  (13-09-2026, 10:55 AM)Torsten Wrote: You are not allowed to view links. Register or Login to view.Andre, you write that I should "criticize claims with an aim to clarify the evidence and move the understanding forward" and that I do so "without offering any new evidence or new reasoning."
  
  Several of the issues I raise are not open questions...

  It is not about the technical accuracy of what you say in your posts; it is about the how, and when, and why of them. Most of your detailed statements are true, some are not. It still comes across that your main aim is more to make a targeted attack rather than a "scholarly" response. (You know full well, for example, the many reasons why Davis holds a fairly common opinion that it seems to be Indo-European. Your response of asking for the published source was not because it was the scholarly thing to do.)
  
  As for the references in my 2024 paper, your paper from 2020 (I am not aware of any 2014 one in Cryptologia) was omitted because it was no more relevant than many other publications addressing the same general topic.  (References are not meant to be a literature survey -- not unless that is the specific purpose of the paper itself.)  You were also omitted because of where you published. That is a risk you take when you choose to publish your work in a journal that restricts its articles behind an unreasonable paywall. Unless such a paper offers high value and a unique relevance, I don't cite it.
  
  

  (13-09-2026, 10:55 AM)Torsten Wrote: You are not allowed to view links. Register or Login to view.Citation choices in scholarship are never neutral — they show whom one acknowledges as a participant in the conversation and whom one excludes. 

   
  Now those words may be more revealing of your motivations than anything else you've said. Are you concerned with the subject matter, or are you just offended that your work isn't getting accepted enough by certain people in the community?

• RE: The Voynich Manuscript: Understanding the Untranslatable with Lisa Fagin Davis

  bi3mw > 13-09-2026, 02:05 PM

  I don't understand why you'd even need AI to translate your post into English.  There are translation programs like deepl.com that serve exactly this purpose. Here the text is translated 1:1, that is, in all its imperfection.  For example, if metaphors and idioms  (which I like to use) seem nonsensical in the translation, you can always tweak them manually. Your own writing style remains recognizable in any case.

• RE: New VMS podcast

  Torsten > 13-09-2026, 02:19 PM

  (13-09-2026, 01:46 PM)asteckley Wrote: You are not allowed to view links. Register or Login to view....

  
  You write that I "know full well the many reasons why Davis holds a fairly common opinion that it seems to be Indo-European." I would be interested to hear what those many reasons are and where they are published.
  
  You write that my observation about citation choices reveals my "motivations." It was a general observation about scholarly practice. Your response — that you omitted my work because of its publication venue — illustrates the point rather well.

• RE: The Voynich Manuscript: Understanding the Untranslatable with Lisa Fagin Davis

  tavie > 13-09-2026, 02:36 PM

  (13-09-2026, 12:55 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.I use AI assistance since writing a post in english is time consuming and exhausting to me. That said, the discussion about AI usage is missing the point.

  
  But, Torsten, why not write it in your native German and then ask AI for a strict translation?  
  
  Otherwise, you risk the impression of it looking like you've just told an LLM "Write me a post for Voynich Ninja criticizing this paper" and put it as a response to the thread.  It doesn't feel like we are engaging with you if we reply.  I can confirm this feeling in practice.  You provided a lengthy AI written critique of someone else's work a while ago.  One of your arguments seemed odd to me, but I may have had the wrong end of the stick.  Pre LLM days, I would have simply asked you.  But I ended up choosing not to bother: since your post was so clearly LLM written, I thought I would just be talking to ChatGPT, and I'm not interested in doing that.  I get enough LLM written correspondence as it is from sloppers claiming they aren't using LLMs.

• RE: New VMS podcast

  asteckley > 13-09-2026, 03:01 PM

  (13-09-2026, 02:19 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.You write that I "know full well the many reasons why Davis holds a fairly common opinion that it seems to be Indo-European." I would be interested to hear what those many reasons are and where they are published.

  Really?  I mean are you really interested?  
  Ok, let's see what our AI friend has to say about why Davis' and many others might hold that opinion:
  

  • Historical and codicological prior. The parchment dates to the early fifteenth century, and Yale describes the manuscript as a Central European production. Its physical construction, script direction, page design, subject divisions, zodiacal imagery, and herbal, medical, astronomical, and pharmaceutical genres place it in a European manuscript environment. Most written vernaculars in that environment were Indo-European, as was Latin, the transregional scholarly language.
    

  • Alphabetic and scribal appearance. The text runs left to right, is divided by spaces into word-like units, and uses an inventory of roughly alphabetic scale. Several glyph components resemble Latin scribal forms or abbreviation signs, although the script as a whole has not been derived securely from any known alphabet.
    

  • Language-like organization. Word frequencies, repeated forms, positional regularities, page-level vocabularies, topic structure, and the association between morphological similarity and co-occurrence are difficult to reduce to independent random marks. These findings support an organized message or at least a highly structured generative system. They do not by themselves identify Indo-European.
    

  • Morphological profile. Luke Lindemann’s broad comparison of Voynichese with 160 languages found that the closest family-level profiles were mostly Indo-European. Iranian ranked first for the full manuscript and for both Currier varieties; Germanic ranked second for the full text and Currier B; Indic and Romance also ranked highly. This is the most direct quantitative reason for preferring an Indo-European candidate, though it rests on only two surface measures and cannot identify a particular language.
    

  • Inflection-like families of forms. Voynichese tokens often appear as related series with recurrent beginnings, endings, and internal slots. Their distribution resembles a moderately morphologically complex language more than a radically isolating or extremely agglutinative one. That profile is compatible with many Indo-European languages, especially if the script, spelling, or cipher suppresses distinctions.
    

  • European-language traces around the main script. The zodiac pages carry later month names in a Romance variety, and marginal annotations use Latin letters and have repeatedly been read through Latin, German, or Romance. These additions reveal the linguistic milieu through which the codex passed; they are not a direct identification of the main text.
    

  • Feasibility under transformation. Recent cipher experiments show that Latin or Italian can be converted by a historically plausible verbose homophonic system into text with several Voynich-like statistics. This does not decipher the manuscript, but it answers one objection: an Indo-European plaintext need not retain ordinary Indo-European character entropy after encipherment.
    

  
  To be clear, I don't myself consider the question of Indo-European origins to be settled, despite all of the above -- there are also counter-arguments. But let's not pretend that after so many years of researching the Voynich and studying the material as you have done, that anything in the above list is actually new information to you. As I said, "you know full well".
  
  

  (13-09-2026, 02:19 PM)Torsten Wrote: You are not allowed to view links. Register or Login to view.You write that my observation about citation choices reveals my "motivations." It was a general observation about scholarly practice. Your response — that you omitted my work because of its publication venue — illustrates the point rather well.

  I assume you did not use AI in writing that. Because whatever point you are trying to make there is unclear.
• Next Oldest Next Newest