20-02-2026, 08:32 PM
(19-02-2026, 12:57 PM)ReneZ wrote: You are not allowed to view links. Register or Login to view.While I don't disagree, I think that it works in the opposite direction. The observation that ed (in fact e) does not appear at the start of words suggests that e is something attached to the preceding character.
Definitely, the various word models, including mine, were deduced from bigram frequencies.
But those models already incorporate a lot of information about bigrams, and much more. So simple bigram statistics on one hand may be partially wasted word, because it will discover things that the models already say (like "ed cannot appear at the start of a word"), and/or provide information that cannot be easily interpreted. If one adopts my list of elements, it would be more useful to compute the distribution of the elements that can precede d in each language.
IIUC, in language B the {d} element can be preceded by some or all of E = {che} {she} {ee} {eee} or F = {ke} {te} {ckhe} {cthe} (forget the puffs and other rare combos for now). Which ones, exactly? (Wild guess: ee is much less common than eee in that context.)
But it seems none of the E and F elements can occur before {d} in language A. Can we say that some of the elements of E or F that precede {d} in language B get replaced by other specific elements in language A?
Quote:Word statistics are also useful, but we do not know that Voynich word represent plain text words
Even if they are encrypted, the Zipf plots suggest that the encryption is mostly one-to-one on words. So word statistics and word-level dependency analysis could help us understand the "grammar" (or absence thereof) of the underlying language.
Quote:word statistics are much more affected by spelling and spacing uncertainties.
Yes, when analyzing word statistics one must always keep in mind that possibly 10% or more of the words in any transcription file may be wrong.
(But glyphs may be wrong too --- I bet that 10% or more of the transcribed r were in fact s, 10% of the o were a, 10% of the ee should have been Ch, etc. Or vice versa.)
All the best, --stolfi