VMs: Re: Paradigms Regained

[Gabriel Landini <G.Landini@xxxxxxxxxx>]

On Saturday 19 Oct 2002 2:57 am, Dennis wrote: 
> I must be dense. Could you be more specific? E.g.,
> French, Italian, etc. syllables don't have the 
> paradigmatic nature of Voynichese words, don't have the 
> relative small number of endings that lead one to think 
> of Chinese, have too many/too few open syllables, or 
> something I can't think of?

This is not the same problem as the type of language according to the number 
of characters. There is a nice table in Coe's "Breaking the Maya code", 
Chapter 1 (p 43 in the edition I have). There he shows that writing systems 
with 20-36 characters they are alphabetic or consonantal, between 40-87 they 
are "pure" syllabic, and for hundreds of them (the smallest is 497 to >5000) 
they are logographic. What we are talking here is to encode syllabically (?) 
a language which is not  normally written in a syllabic script.
In EVA, the vms has about 8200 words. I think that it would be extremely 
difficult to write (and read) a text this way, but this may be irrelevant as 
we do not know how long it took the author to write the book and if anybody 
ever read it back.

One would need to know: How many syllables is one expected to find in a text 
of the size of the vms (assuming that it is written is language X)? I have no 
idea of what this number may be (in any language), but I have the feeling 
that 8200 is a bit too large.

Anyway, let's suppose that we are set to do a "syllable encoding" from 
scratch. One could start by making a table of the syllables with the 
"encoding" as one goes along the text. Also let's assume that each "Voynich 
word" (i.e. each syllable) is represented as a number that increases with 
each new entry (yes I know that this is assuming too much, but bear with me). 
The more common syllables would have their entries early in the table (small 
numbers) and rarer ones would appear later on (larger numbers).

There must be some relation between the length of the text and the structure 
of the table since by Zipf's law we know that words have different 
frequencies and therefore syllables will have different frequencies too. I 
think that the earlier parts of the text would have more common syllables 
(small numbers) while the final parts of the text would have more entries 
from the final part of the table (larger numbers). (I have the feeling that 
this would be somewhat similar to the structure of the table in the Ziv and 
Lempel's universal algorithm for sequential data compression (Jim R?)).

So would looking at the rate of appearance of new syllables (=Voynich words) 
across the text and comparing this to known texts written as words and as 
syllables reveal anything? Of course this implies that the dissecting of 
words into syllables was done systematically, and still one should have to 
explain what the character sequences and the labels (syllables on their own?) 
stand for... 
Sorry if this was too long.

Cheers,

Gabriel