RE: VMs: [ha] [hb] not different languages

[Nick Pelling <incoming@xxxxxxxxxxxxxxxxx>]

Hi GC,

At 11:56 02/08/2003 -0500, GC wrote:
> Nick wrote:
> > To my eyes, the entire VMS is expressed in a single systematic way - the
> > same pair-like structures (qo, ol, al, or, ar, dy, of, ot, ee, air, ain,
> > etc) I've been pointing to are evident on every single page of VMS text,
> > and any statistical differences we're talking about appear as deviations
> > from (or embellishments to) this homogeneous basic structure.
>
> I would have to disagree with you here, especially on just about everything
> you said in the upper paragraph :-)
>
> 1.  The entire VMS is *not* expressed in a single systematic way - as Rene
> has demonstrated quite conclusively, there are rather clear and
> well-established demarcations between the statistics of any given section
> that make them separate from the others in the VMS.  Rene gives a minimum of
> six distinct classifications, expanded into a longer initial list. You'll
> find this a worthy read, at http://www.voynich.nu/extra/lang.html

I've read that page many times, as I'm sure you have too. However, as 
perhaps the only person here who openly admits to thinking that the 
apparent words we see are deliberately misleading (ie that the real word 
boundaries almost certainly fall inside these apparent words), I have 
difficulty drawing the kind of statistical word-based inferences that you 
and Rene do. To be precise, I think these are caused by the differences in 
the languages, not the actual causes of them.

I'm much more sympathetic to Mark Perakh's idea that perhaps A is more 
heavily abbreviated than B - my last major post on this covered this quite 
fully. IMO, any analysis that might serve to differentiate between these 
languages *without* recourse to the apparent spaces (and hence apparent 
words) would be a very much more substantial result. For example, the 
presence of the free-standing <l> (ie, not part of an <ol> or <al> pair) is 
a fairly good indicator of language, IIRC?

> 2.  The same pair-like structures *do not* occur on every every single
> page - there are common pairs that do, but for the most part there are very
> clear statistical differences between the sections.  You appear to be
> focusing on a small handful of structures, but in pairs that make up word
> endings, there are 294 different pairs that serve as word endings.  When the
> larger portion is viewed, the numerous 'deviations' cannot be dismissed as
> mere 'embellishments', but must be accounted for in any viable hypothesis or
> theory.

According to my view of the VMS' coding system, one key feature of the 
back-end pair cipher used is in-built error-correction - it's robust enough 
to withstand many types of scribal copying errors. I'm sure that very many 
of the different rare pairs can be accounted for in this way. I also don't 
think that pairs account for the whole system - the {ch|sh}{e|ee|eee}{o} 
regular expression is another combinatorial device, which I think comprises 
abbreviation and word suffixes.

> Taken in the sense that the few 'pairified groups' form 'fundamental
> building blocks', I've an open mind on that concept, as long as it
> progresses to the point of ordering and structuring the rest of the 'blocks'
> as it goes along.

But of course. :-)  FWIW, I think that <ol> and <al> (and the free-standing 
<l>) are particularly suggestive of there being some kind of a pairified 
basis at the core of the system... though I'm quite sure you have your own 
ideas on this. :-)

> You and I have been speaking of cryptographic systems in our posts.  There
> are a few universally bantered general words in this field, but "language"
> almost always refers to the underlying cleartext language.  Statistical
> differences in a system are not referred to as "languages" in any sense, nor
> are they called "dialects".  "Dialect" usually refers to the specific (and
> often unusual) use of decrypted cleartext words in the underlying cleartext,
> not statistical differences in cipher groups.  But that's just my thinking -
> that specific terms allow the reader to follow your line of thinking in
> terms specific to your approach, without much ambiguity.

This is all very well, but please remember my aim is - just as much as 
yours - to achieve clarity of expression and discussion, so that we can all 
move forward.

In the VMS, we have a unique object which apparently does not conform to 
the 20th Century cryptological mindset: if we are collectively to build up 
an understanding of it, we must agree metaphors which capture its unique 
facets. Where words fail us, we must make (or appropriate) new words. Where 
we appropriate words, we must make sure that we are not being trapped by 
their normal nuances.

For example: is there a universal system underlying the VMS' "languages" as 
observed, or are they completely independent of each other? I'm sure of the 
latter, but the use of the term "language" for each seems to be a 
manifestation of the former. If this is an unchallenged assumption, all I 
want to do is challenge it. :-)

Also note that I'm currently making no assumptions about the language(s) in 
the plaintext, as I simply don't think we're yet in a position to make 
stats-based predictions about this. This is perhaps another reason to be 
somewhat uneasy about the use of the word "language" in discussion of the 
VMS - perhaps we're actually talking about "expressive mode A" or "trope B"?

> > For me, "dialect" in the context of the VMS is neither a cryptological
> > (code) nor a linguistic (language) construct, but rather a *behavioural*
> > (statistical) one - it's what we observe the VMS "doing", how we
> > observe it varying locally.
>
> I detect a smattering of psychology-101 creeping in as well! :-)

Errrmm.... looking at the statistics as expressions of a underlying 
behavioural construct [ie without wanting to posit a causal underlying 
theory] would be rather closer to sociology-101 (or perhaps some American 
schools of philosophy, IIRC) than psychology, I believe. :-o

Cheers, .....Nick Pelling..... 


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