Re: VMs: Voynichese as an Abugida

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

On Mon, 26 Jul 2004, william edmondson wrote:
> Interesting post.  I too have wondered about syllabic representations.

Thanks.  That's what I was wondering - whether anyone else has pursued an
approach like this and could save me some trouble by poking holes in it!

> A few thoughts of encouragement.  Don't be put off by the apparently
> large number of syllables in English.  The figures may be high for text
> but nothing like so high for actual sounds.

Even if the approach to the analysis of sounds was based on a standard
Roman orthography, you can simplify matters extensively by using the
abugida approach, which eliminates the need to code closed syllables.
Some kinds of closings or codas can be included, of course.  For example,
philosopher's stone, can be encoded in representations for fi-lo-so-fe-r-s
s-to-n or fi-lo-so-fer-s s-ton.  The second version assumes that the -r
and -n codas are included within the representational units.  Of course,
as I've pointed out, unless I'm missing something, it doesn't look like
there are enough consonants or vowels to accomodate English.  Even
orthographic English requires five vowels, even if we ignore diphthongs.

If the encoding of the representations involved discrete, character-like
entities, and these were written in the opposite of production order,
e.g., as if-ol-os-ef-r-s s-ot-n, it would certainly complicate matters for
folks who were looking for a character-based encoding.

For that matter, when did pig latin and such word games first appear?
I've heard of people who could rattle off pig latin versions of English
at conversational speeds.

> Couple this with the fact that naive intuitions regarding syllables can
> be good - the written form for Cherokee, if I recall correctly, was
> invented by a non-linguist and serves well - and it seems likely that
> someone of, say, Kelly's abilities would have no trouble devising a
> serviceable syllabary for, say, Latin or other European 'phonetic'
> languages.

I don't believe Sequoiah was even literate before he developed the
Cherokee syllabary.  He did understand the concept of literacy, clearly,
and produced a nice workmanlike reverse engineering of the technology in a
new context.  Brilliant, and a lot of hard work.

> We should note that the consonant vowel distinctions in semitic
> languages are managed independently for morphological reasons (patterns
> of vowels are morphemes, and interleave with patterns of consonants, as
> morphemes).  We would not expect to find that in VMS if it is a
> rendering of anything other than Hebrew/Arabic.

I believe four grades in a series (zero, a, i, u) would suffice to
represent most Semitic languages.  Or Greelandic Eskimo, for that matter.
However, Semitic languages have relatively large consonant sets, so the
number of series I've suggested or at least recognized seems insufficient.
Greenlandic, however, has fewer consonants, something more like the number
of series in Voynichese - p t k q v s g r m n ng l, if I recall
correctly.  Of course, I'm not suggesting Voynichese is Greenlandic!
For one thing the maximum word length would be a lot more than 10.

It occurs to me that you could represent a Semitic language by placing the
consonant frame and prefix/suffix/vocalic pattern in sequence separately,
rather than intermingling them, if you had a philological turn of mind.
That, too would be confusing to decode if the wrong assumptions were made
about the nature of the input string.

> I'd go for consonants with distinct symbols, plus some sort of simple
> minded coding for vowel sounds (or even their omission).

Me, too.  Of course the acid test is a decoding ...
______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list