At 12:46 -0400 2005-08-20, Peter T. Daniels wrote:

> > There is a lot of interesting information about Vai in it.
>
>It's a rather large download. Time vs. intrinsic interest argues against
>it

It is a large download because it's 29 pages long, with a good many
charts and illustrations. I am on slow dialup internet myself, and in
my case, I had to *upload* many drafts of the document, so I can
sympathize. John Singler thinks that it is a very good document, I
should say, and it has a deeper analysis than in the Scripts of West
Africa section which he authored for the WWS. It would be easier to
discuss Vai with you if you were able to read the document.

> > It was necessary for us to determine a linear order for Vai letters
> > for sorting.
>
>For _you_ to do sorting.

No, for _computers_ to do sorting so that Vai-speaking users can find
documents.

When encoding a script in Unicode, the characters have to go into the
code chart in some order. They could be thrown in at random; it
doesn't matter, in principle, because the coding in a coded character
set is simply having a unique computer identifier for each character.
But of course, organizing the chart in a sensible way is more helpful
to the user.

Real sorting of characters is also something that the Unicode
Standard specifies, because it's something that *has* to happen.
Let's say a Vai user creates 25 documents and names them in Vai and
puts them in a folder on his computer. He opens the window and clicks
on the "sort by name" tab. Well, now, the computer *has* to have
instructions as to what to do about it.

For this reason, we studied the range of possible choices for
specifying an ordering for Vai, and we settled on the one I have
described. Our Vai-speaking informants agreed with our proposal, and
now the Vai user's data will be ordered in a sensible fashion.

>Is there evidence that it's ever been used for "sorting" previously?
>If the symbols had numerical values, that would be prima facie
>evidence.

We have not seen a glossary; there is a short sorted list of biblical
names in the back of the Vai New Testament, but these are ordered
according to their Latin transliterations.

We do have many syllable charts, made by linguists and made by
educators. Typically, these organize the columns in terms of place of
articulation, grouping the labials, then the dentals, then the
velars, with some stragglers toward the end. (Charts organized by
Latin transliteration, like Massaquoi's, don't do this, and the
systematic relationship between glyphs is conceiled.)

Where those charts do differ is typically in the order of the vowel
columns. We find I-A-U-EE-E-O-OO, A-EE-E-I-OO-O-U, and the order we
prefer, EE-I-A-OO-U-O-E, which is used in alphabet chants in Liberia.

> > The Vai script's glyphs have relations between them, where groups of
>> syllables with related consonants are clustered together, within
>> final-vowel classes. This is different from the structure of some
>> other scripts, like Canadian Syllabics, where the glyphs
>> relationships are consonant-based.
>
>"Related consonants"? "Clustered together"? If you haven't been able to
>determine the original order, how can you say what's "related" and
>what's "clustered"?

Some consonants are related to others in terms of their articulation.
In the range of syllables ending in -EE, the glyphs used for PEE and
BHEE are similar. The glyphs used for BEE and MBEE are similar. The
glyphs used for KPEE and MGBEE and GBEE are similar. The glyphs used
for FEE and VEE are similar. The glyphs used for TEE, THEE, DHEE,
DHHEE, LEE, and REE are similar. The glyphs used for DEE and NDEE are
similar. The glyphs used for SEE, SHEE, ZE, ZHEE are similar. The
glyphs used for CHEE and JEE are similar. The glyphs used for NJEE
and YEE are similar. The glyphs used for KEE and NGGEE are similar.

Syllabary charts cluster these together.

>We already know that there's no correlation between graphics and sounds.

There is some correlation between graphics and sounds in Vai, as I
have explained already, and is shown very clearly in
http://std.dkuug.dk/jtc1/sc2/wg2/docs/n2948.pdf

> > It hasn't got one, apart from sets like I/J and U/V/W. Its order is
>> inherited from the Phoenician. Vai, on the other hand, shows an
>> internal structure.
>
>If you don't know the original order, how can you determine its structure?

By observing the data and analysing it, of course.
--
Michael Everson * http://www.evertype.com