From: Hans Aberg (haberg@math.su.se)
Date: Mon Apr 25 2005 - 10:38:47 CST
At 15:48 +0200 2005/04/25, Otto Stolz wrote:
>you have written:
>>The Swedish language symbol ä (a with two dots
>>above) is a separate letter, not to be viewed
>>as an alteration of the letter a. So it is
>>atomic. It is reasonable to enter it as a
>>separate character. In German, however it is an
>>umlaut, alteration of the letter a.
>
>Not quite so: It has its own phonetic value (almost equal to its
>Swedish sibling, IIRC), and is taugh as seperate character in schools
>(believe me, I am German and interested in linguistic issues, and my
>wife is a teacher at an elementary school).
>
>The term "Umlaut" for a class of characters does not render these
>umlauts as non-characters. There is a similar term, "Ablaut", e. g.
>for the "a" and "o" in "barst" and "geborsten" (from "bersten") --
>yet, this does not qualify "a" and "o" as non-characters, alterations
>of "e".
Let's take it easy: I attempted to make a formal
definition of the notion of an abstract
character, not to be confused with the many
possible intuitive notions of a character. When
defining an abstract character, I suggested that
it should be a linguistic semantic unit that in
some sense or another is atomic. There, the point
is that symbols like ä can be atomized in more
ways than one: It could be viewed as a whole,
indivisible unit, or a composite of more than one
characters. The choice may depend on the context.
The second point, though, is that the preference
for larger symbols be viewed as a single
character, as regards to computer software,
probably is due to limitations of this computer
software. It would probably be better, computer
implementationwise, to always represent symbols
like ä as a combination of smaller, abstract
characters, as a sufficiently smart computer
program always can recognize the Swedish or
German letter ä, and give it the proper handling,
and as we now know that the representing of
characters in a single or a bibyte will not
suffice anyhow.
-- Hans Aberg
This archive was generated by hypermail 2.1.5 : Mon Apr 25 2005 - 10:40:19 CST