The 'Chinese' Theory: For and Against

[Jorge_Stolfi]

The problem, however, is that this approach allows you to retroactively label any “disturbing” form as an error. In this way, almost any finding can be dismissed.

Not so much, because the errors must be plausible.  

• Errors that were incurred by the Author while taking dictation should be mostly omission of a phoneme (glyph of diglyph), replacement of a phoneme by another with a similar sound, or undue duplication of a glyph that can be duplicated (like writing cheedy for chedy or daiiin for daiin).
  
• Errors made by the Author when rewriting and editing his dictation notes could additionally be sequence errors like skipping a line, or omitting or repeating a string of one or more words. They could also include substitution of a letter by a letter that had a similar shape in his own handwriting (like an s by an r, or a iin by a ir) or accidental omission or truncation of a stroke (like turning y into a or o, Sh into Ch, or Ch into ee).
  
• Errors by the Scribe when copying the Author's draft to vellum could be all of the above; but the confusion of letters with similar shape may have been be much more frequent.  The Scribe may have also omitted many spaces or inserted many bogus ones.
  
• Errors by the Global Retracer should be mostly omission of strokes or glyphs that had completely faded away (like turning a t into a k, or a d into an s), or mutation of glyphs into other glyphs or weirdos with roughly similar shape (like turning an a into an ei, or a Ch into an Ih).
  
• Errors by the modern transcribers should mostly be incorrect guessing of glyphs with ambiguous shapes (like halfway between r and s, or between a and o), omission of spaces, or transcription of bogus spaces.
  

This list may seem rather large, but it is in fact a very small subset of the set of all theoretically possible errors.   For instance, it does not let the person making some claim to shrug with "dictation or scribal error" if the text has a t where his claim implies should be an o (since the word structure models strongly imply that t and o are very different phonemes), or has chedy chedy where his claim says that there should be a daiin.

Here is another way of stating your objection above: Since it is possible that 5% or more of the VMS tokens and spaces are erroneous, one should not attach much significance to any statistical peculiarity that could be a consequence of those errors.

But I do realize that these sh/ch structures I’m pointing out here at least have the theoretical potential to call your Chinese theory into question.

I do not see that yet.   As I wrote, your numbers are saying that some Voynichese tokens that begin with Sh or Ch are often followed by some tokens that start with qo (more often than what would be expected if each token was chosen independently of the preceding one.)  

But that is not incompatible with the Chinese Origin theory (that the text of the VMS is a phonetic transcription of a text in some East Asian monosyllabic language).  On the contrary, the COT predicts that there will be several "remote effects" like that one.

I have not checked, but I am assuming that in Shakespeare's Hamlet a token that begins with "th" is often followed by one that begins with "i", more often than expected by chance.  Would that be evidence that the text of Hamlet is not in a natural language? 

in the REVA analysis (Eva reverse), this phenomenon ... doesn’t occur at the token level here, but within words and when read backward, it sort of dissolves.

I don't get it.  It you reverse the glyph order in each token, the same "remote influence" effect should still be there: namely, a token that ends with -Sh or -Ch should often be followed by a token that ends in -oq.  

And if you also reverse the order of the tokens in each line or parag, you will probably find that a token that ends in -oq will often be followed by a token that ends in -Sh or -Ch. Isn't that the case?

All the best, --stolfi