From: Asmus Freytag (asmusf@ix.netcom.com)
Date: Thu Jan 03 2008 - 11:03:14 CST
On 1/3/2008 6:43 AM, Michael Everson wrote:
> At 06:25 -0800 2008-01-03, John Hudson wrote:
>>> On foot of the shape of the letter I would consider encoding it as
>>> the capital form of U+0261, as LATIN CAPITAL LETTER SCRIPT G.
>>
>> Surely not considering the presence of the dot.
>
This is total an utter nonsense. What one encodes is always the
*identity* of the character. In this case, the character was used for a
sound that can and is spelled with either G or J in French (and, more
importantly for the purpose) in French loanwords used commonly in
German, and therefore familiar to the reader.
The phonetic notation that this artificial letter is a part of, is an
example of a common device used for such purposes, i.e. a system that
uses, as much as possible, notations that are close to what the lay
reader is familiar with, adding a few signs and diacritics to mark
sounds not found in the readers native language.
Note that the 'sh' sound for the name of the letter 'H' is spelled out
with the German 'sch', to see what I mean.
The example for /juge/ makes clear that the Gj is used "as a /unicase/
or caseless letter" as Andreas writes. He continues "It is questionable
wether to speak of it as a ligature" and I would agree. The intent here
is not to combine the sounds of a g and a j but to denote a sound that
ordinarily is written with either a 'g' or 'j'. The choice of uppercase
G for deriving the glyph seems designed to let this letter stand out and
make both of the constituent sources of its glyph apparent to the lay
reader.
Thus, the concept of case does not apply, and, as the character has not
found its way into any existing "orthography", there's little pressure
to innovate in that direction. Whether IPA or any other phonetic
notations use glyphs based on some decoration of the letter 'g' or 'G'
for the same sound is irrelevant - unless one can show a relation by
derivation. For obvious reasons G just happens to be a natural choice to
represent such sounds.
If it's deemed important to encode the character to be able to encode
such dictionaries as originally published, then go ahead and encode
*that* identity. Don't invent something new.
A./
This archive was generated by hypermail 2.1.5 : Thu Jan 03 2008 - 11:05:58 CST