Re: VMs: Strange or not?

[Koontz John E <John.Koontz@xxxxxxxxxxxx>]

On Mon, 3 Jan 2005, Eric wrote:
> The heavy ch <-> sh similarity has been noted in different ways before
> ...
>
> Back in the archives I read where Jacques was arguing [for] dialect
> variations ...
>
> It could also simply be a homophonic substitution where ch and sh both
> represent the same underlying letter.

These are not the only cases where two EVA letters have been argued to be
the same on distributional grounds combined with a similarity in
appearence, e.g., various gallows characters.  However, it would be wise
to remember that distributional grounds can be used to equate any two
different letters that happen to occur in the same orthographic contexts,
or, more or less equivalently, represent different sounds that can occur
in the same canonical context.  For example, in native English sh and ch
have rather similar distributions, or many vowels, or r and l, and so on.
Given any tendency for similarity between different letters, and some
scripts have strong tendencies in that direction, you can combine
coincidences of the two to argue for identity.  For example, e, o, and
a are not the same, even though there are many words in many languages
that differ only in terms of one of more of these letters.

What controls this sort of distributional argument in standard linguistic
situations is the information that two forms involving these two letters
or sounds have different meanings - what is called a minimal pair.
Naturally the whole concept of minimal pairs fails if the text is
encyphered in a way that varies the letter used to encode a given clear
letter.  However, if a scheme like that is in use, I wonder if one would
expect to find much in the way of repeated words at all, especially if the
scheme was a strong one.  A counter exception here would be that if
multi-letter sequences are being used as the basic units in the encyphered
text they might easily reappear, even in different senses, e.g.,
cheol/sheol might be one letter one time and another another time.

Obviously it is rather difficult to come up with minimal pairs in the case
of a document of unknown meaning.  One way to attempt to control for this
lack would be to look at the distribution of apparently interchangeable
forms.   Looking at sh and ch in larger syntactic contexts is one way.
For example, in Knox Mix's examples, the only one in which the
similarities extend beyond the immediate word or word pair is:

> 1875       ody ch*k*es otal/sol sheeol  OL CHEEY  os sheky sheol or
> 1876      sheol or shear oly/lcheol ol  OL SHEEY  olsheey shol keey
                               *********************************

Notice also that ch and sh occur in both examples, so it's not a dialect
difference, or, at least dialects are freely mixed.

Another comparison that can be made is to determine whether particular
variants have any particular distribution in the whole text.  I don't know
the answer on this.

One other strategy that occurs to me is that occurrence of different
variants in labels of different graphics might suggest different meaning,
even if the meaning was not known.  In other words, different graphics
would suggest, but obviously not prove, different meanings.

======

I was tempted to try to work in some sort of play on Knox Mix and Machts
Nichts, but decided to settle for admitting that I have been wondering.

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