[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Re: VMs: voynich dice game ... sunday thoughts

On Wed, 15 Sep 2004, Marzio De Biasi wrote:
> >13/09/2004 9:39:51 PM, Marzio De Biasi <voynich@xxxxxxxxxxxx> wrote:
> > >Theorem:
> > >spaces are not randomly inserted over a plain unencoded text (or over a
> > >text encoded with a "space unaware" schema):
> >
> > >Proof:
> > >if they were inserted randomly the frequencies distributions of these sets:
> >
> > >A={the characters in VM}, B = {the first characters of each word}, C = {the
> > >last characters of each word}
> >
> > >should be almost the same ... but they are completely different. CVD
> >
> >Possibly. It seems rational enough. But wait... what of Arabic?
> >Spaces are not inserted at word breaks, but are the effect of some
> >letters being able to connect to the next, or to the previous, and
> >some not being able to. The same Arabic text, in Roman spelling
> >(as Maltese, an Arabic dialect, is spelt) have spaces between words.
> >By "Arabic" here, I also mean any writing system where some letters
> >connect and some not, forcing the use of spaces inside words.

However, though certain characters occur more frequently or less
frequently initially in Voynichese, I think only q is (almost) strictly
associated with initial position, and it seems to be an additional element
in that position.  So, while the Arabic situation warns us that spaces -
or at least interruptions in the flow of connection of glyphs - can be
inserted by rules other than "word division," it doesn't seem to apply
specifically to Voynich script's word divisions.

It does potentially apply directly to the issue of the "flourishes" or
"terminals," in the sense that, if they are separate glyphs, they may
apparently only be inserted by direct attachment to a preceding element.
Thus, EVA l, for example, seems to consist of an "i" with an attached
"loop back and across from the top" entity that cannot occur without a
preceding i, or e, if this is the same element that forms the second part
of a y.  EVA represents l and y as single glyphs, and not pairs of glyphs
because it tries to associate a single transcription character with each
"letter space" or "stroke discontinuity" delimited unit in the text.
This is perfectly reasonable for transcriptional purposes, of course, and
I don't know that we really have anything much more than pattern coherence
to suggest that the flourishes are really separate glyphs.

> >Space insertion in such systems is not random, but it tells us
> >nothing about what constitutes a word.

Paraphrasing, "stroke discontinuities" in such systems are not random, but
they tell us nothing about what constitutes a glyph (or character).

> But my poor and UNUSEFUL theorem tells only that:
> IF one is convinced that spaces ARE inserted randomly on an existing text
> (like I was a few days ago :-) THEN, when he removes 'em, he must deal with
> a piece of text that cannot be a simple unencoded text or a text encoded
> with a "space unaware" schema.
> My theorem really says nothing in favour of random/unrandom spaces in VM.

My conclusion from the theorem relative to the Fincher Window hypothesis
set is that this shows that spaces are either not added to the text after
a sequence is generated, or, that if they are, there is something
preferential about the way they are added.

I have another argument against the Fincher Windows hypothesis set in any
simple form, and this arises from the glyph lists in D'Imperio's Figures
18 and 20, listing of unusual combinations of flourishes and ligature
lines with more familiar characters (Fig. 18) and unusually elaborated
gallows characters (Fig. 20).  I take it that these are at least two
subclasses of the weirdoes that are mentioned from time to time by the

It strikes me that neither set of entities is statistically likely to
arise from copying strings from a master text being sampled with a window.
Nor do most of these entities strike me as scribal alternatives to simpler
forms of standard EVA sequences per se, unless we assume that most of the
ligaturing and flourishing is arbitrary and ornamental whimsy.  In short,
these are things that seem unlikely to me to occur in second order text
that is being copied from first order elsewhere that lacks them, while at
the same time they seem to me quite likely to occur in first order text
being generated from scratch.  If they are being added to second order
text, then the relationship between the first order text and the second
order text is likely to be quite loose.
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list