From: Gregg Reynolds (unicode@arabink.com)
Date: Mon May 23 2005 - 18:04:40 CDT
Dean Snyder wrote:
> Gregg Reynolds wrote at 4:23 PM on Monday, May 23, 2005:
>
>
>>Dean Snyder wrote:
>>
>
> Latin r in no way resembles Latin z; Arabic "r" & "z" are exactly alike
> except for the tiny dot above "z".
>
There is no Arabic "r" nor "z". ;)
>
>
>>Can you tell that difference from the integers x0631 and x0632?
>
>
> You can when they're rendered; and I have been talking all along about
> loss of GLYPHIC correspondences in transliterations.
>
Not if you render them "r" and "z".
I see your point but I think it only applies if you are restricted to
pencil and paper. Even then I don't really understand where the problem
is. A transliteration/encoding is lossless (as I construe it anyway) if
you can successfully encode and decode a message using it. I'll bet if
I sent you a message using Buckwalter's scheme you could take a pencil
and decode it into written Arabic and reproduce exactly the same text.
The fact that the codepoints it uses can also be construed to denote
latin letterforms in no way affects their interpretation according to
his scheme. Hence, lossless.
Actually I see this as a flaw in Unicode's conceptual structure; it sees
codepoints as denoting "abstract characters", but it conceives of
abstract characters as graphical abstractions. I think it better to
think of them just like numbers: elements of a conceptual structure
which have no natural physical representation. I.e. x0632 denotes a
concept which can be represented in many ways (including audibly, with
smoke signals, morse code, etc.). There should be no necessary
connection between glyphs and "characters". Hence *any* physical
representation is a kind of transliteration.
-g
This archive was generated by hypermail 2.1.5 : Mon May 23 2005 - 18:05:44 CDT