The French Hypothesis, Etc.

[Dennis <tsalagi@xxxxxxxx>]

	I was peeved that my name was not mentioned, and my
ideas cited almost not at all, in the New Scientist
article.  After that, I decided that perhaps my ideas
need rethinking. 

	A quick idea first.  The most recent statistical
studies - Rene's "Character Entropy to Word Entropy",
Mark's LSC, and Gabriel's spectral analysis -  seem to
show that a Voynichese word has at least as much
information content as a word in a natural language. 
Taking that, and also the fact that the VMs would have
been deciphered long ago if it were a in a natural
language in a cipher of even the 20th century, it seems
most likely that the VMs is in an artificial language,
like the philosophical languages of Dalgarno, Wilkins,
etc.   

	Having said that, I want to go over my French
hypothesis.   

*The French Hypothesis*

	1) the underlying language of the VMs is medieval
French, and 

	2) Voynichese word divisions are actually French
syllable divisions.  

*Consider:*

	1)  Voynichese words are short, and yet they might
have to be in a verbose cipher to show the low
character entropies that we see.  Syllables would
explain that.  (So would a monosyllabic language like
Chinese; much of this also applies to the Chinese
theory.)

	2)  French was a common language; indeed it had become
the European upper classes' language of communication
even before the Renaissance.  So we do not need some
odd route for a Sino-Tibetan or Polynesian language to
reach Europe and N. Italy during the Middle Ages.

	3)  Word divisions are not marked out by a stress
accent in spoken French.  Lines of poetry are defined
by the number of syllables, not the number of stress or
quantity feet.  Interestingly enough, Louis XIV's Royal
Cipher, finally broken by the great late-19th-century
crippie Etienne Bazeries, enciphered syllables, rather
than letters.  So French is the European language most
likely to be decomposed into syllables.  


*What's Wrong with It?*

	1) Syllables would presumably not show as much
information content as words, yet our most recent
studies say that Voynichese words do have as much
information content (but see the next point).

	2)  280 Voynichese words constitute 80% of the Voynich
text, as Robert Firth showed.  Yet the remainder is
8000 words!!!  Are even natural languages like that? 
Gabriel's various Zipf's Law studies indicate not;
Voynichese follows the Zipf's Laws, but the y-intercept
of the Voynichese curve is anomalous.  How much more a
problem for syllables!  The French hypothesis might say
that this is due to a lax orthography and clumps of two
or three syllables, but it seems a stretch.  I recall
that Gabriel ran some of his tests on the Canterbury
Tales, with their lax English orthography, and it
didn't make a difference of that magnitude. 

	3)  The French hypothesis probably would not explain
the tripartite paradigms of Stolfi or even the
bipartite paradigms of Tiltman and Firth, unless such a
paradigm were as simple as the CVC, CVCC, CCVC, etc.
for most syllables.  

..... so the French hypothesis doesn't look so good. 
Even so, I'm tempted to try it.  We spend our time
studying the VMs, but we'll never solve it unless
*someone* makes a hypothesis and tries to get some
language out of it. Perhaps I need to try it just to
encourage others.  ;-) 

	If it really is a philosophical language, it sounds
really tough to me.  Perhaps that's the tripartite
paradigm: the combination of three groups of qualities,
like Ramon Lull's wheels of correspondences.  But how
do you solve a whole a priori language?  It would have
to at least be related to the ones we know about.

Comments, anyone?

Dennis