VMs: Re: Weirdoes, ligatures (long)

[Gabriel Landini <G.Landini@xxxxxxxxxx>]

Hi all,
Like Rene said, I am not trying to conver anybody to EVA either.
What strikes me is that most of the differences between EVA and a new alphabet 
are not that great.

So far if one wants to code the ms in a new alphabet, then you have to read 
and transcribe all of it.
There are simpler ways to do it. If one thinks that the plume in the <c'h> (or 
<sh>) has different meanings, then the only task is to review all the <sh> 
but not all the ms. There are about 4500 of those but about 233500 
characters. This is about 50 times less work to do.

On Saturday 05 Oct 2002 6:41 am, GC wrote:
> I separate
> the two by calling one {9} and the other {7}, but there is no
> individual representation for {7} in the EVA set as I read it.

Again, there are 17800 instance of <y>, so if one corrects only all <y> and 
<sh>, there is less than 10 times the amount of work that would take reading 
the entire ms.

> Is <ii> in the middle of a word the same
> glyph as <in> at the end of a word, and a hundred more questions
> that beg answers.

But one cannot say that <ii> is the same as <in> because the glyphs look 
morphologically different. They may *mean* the same, but they do not look the 
same.

> If I write:
> <sshol.shecthy.qokaiin.chkedy.rchey.dair.chey.qokaiin>, 
> {2Woe.WcI9.4oham.whc89.swc9.8aS.wc9.4oham}

This could have been done with Bitrans from any file in EVA that may be 
already available. The table would be more or less:

s 2
sh W
e c
l e
cth l
iin m
q 4
o o
ch w
d 8
y 9
ir S
a a

Note how Bitrans *can tell* when translating <s> alone or as part of <sh>.
The translation is also reversible (you can go from your transcription scheme 
to EVA).

If you want to code {9} and {7}  and different variations of <sh>, then EVA is 
a lossy encoding, but it is easy to make a Bitrans table (let's say EVA2Glen 
to convert the available files to your new scheme and correct only the lossy 
EVA characters, than transcribing everything from scratch. You would 
save lots of time.

> Is the VMS written in glyphs, or merely a compilation of
> penstrokes?  

To get the text in a file, this may not be really very important at this 
stage. You can decompose all the characters that stand on their own later on.

> My
> experience has been that while the EVA design is impressive, the
> resulting implementation has been remarkably disappointing. 

The small "detail" is that the proper EVMT file has not been finished yet. The 
"EVA" file is not been made yet available because it is still in the works.
The interlinear files are available, and these more or less follow the EVA 
format, but have not been corrected or transcribed with the EVA format in 
mind (for example ligatures, weirdoes, etc). (except the Takahashi's 
transcription which is in basic EVA if I recall correct).

> If you view these forms as individual glyphs, you're
> privately answering the questions I'm asking publicly, and relying
> on your own observations in absence of group consciousness to
> formulate your theories.  So let's talk about this.  Let's look at
> what most of us are seeing, and try to gain some valuable feedback
> in the process.

I do not think that this is so simple.
Lets take <n>. Is this part of <in> and <iin>? Many think so.
Then, <n> would not apear after any other character except <i>.
But it does, in the duplets <an> (119 times), <en>  (7 times), <on> (9 times), 
<sn> (once) and on its own (at least 3 times).
So coding n as part of another character leaves about 139 instance of the 
character without proper coding. Does one make yet another character to code 
for it? Or shoud we code those as <i>?... but <n> does not "look" like <i> at 
all... So why not to code what looks like <n> as <n>?

Cheers,

Gabriel