Re: VMs: left & right word entropy

[Gabriel Landini <G.Landini@xxxxxxxxxx>]

Hi,

On Wednesday 24 Dec 2003 12:41 am, Akinori Ito wrote:
> I have three interpretations about my finding. One is that Voynichese
> word-like strings are not actually words, that perhaps means inter-word
> spaces are fakes, or a Voynichese word corresponds to a phrase of ordinary
> languages.

This is possible, but I am not very keen on this for 2 reasons. 1. If one 
collapses all spaces, still the mean character correlation funtion has a peak 
at the same length that with spaces. Other languages I tested (English and 
Latin) show this effect too. 
2. The labels

> The second interpretation is, as Frogguy pointed out, Voynichese
> is word-order-free language.

Which is quite interesting. Could this be used to rule out some languages? 
Just note that one does not have to have the alphabet right for this to hold 
(I mean whether <iin> is 1, 2 or 3 chars).

> The last one is that Voynichese word order has
> no meaning (therefore VMS have no grammar).

That is what firtst came to my mind, yet then I wondered whether lossy 
encoding would hide any real differences in left-right entropies by assigning 
ambiguous abbreviations to different words. (So there you have a 4th 
possiblility).


> I feel that this phenomenon (left entropy=right entropy) must be explained
> with the fact that VMS has more than two times as many distinct words as
> English.

I am not sure what you meant. The proportion of words approximates Zipf's law. 
For the size of the text doesn't this guarantee that the number of words is 
about right?

Thanks for any further comments.
Merry Christmas to all.

Gabriel





______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list