[mb-users] Artist sort names for non latin scripts

[Philip Jägenstedt]

Greetings fellow metadataists!

I've recently been losing some sleep (no not really) over the state of
the sort names for Chinese artists and want to lift the discussion to
this list. Some discussion is close to the bottom of
http://wiki.musicbrainz.org/SortNameStyleDiscussion (signature foolip)
and some more in a recent edit:
http://musicbrainz.org/show/edit/?editid=7491761

The issue is this: most editors use the "English" names of japanese
artists as the sort name. From a sorting perspective, this isn't all
bad, because most of the time the surname doesn't change ("Chou, Jie
Lun" is "Chou, Jay"). The style guidelines
(http://wiki.musicbrainz.org/SortNameStyle) however state that the
name should be transliterated into latin script. Should real
transliterations be used as sort names it would make searching for
artists by the string of latin characters which they are actually know
very hard, as only the original name and the sort name show in search
results. Would anyone who doesn't know Chinese know that 周杰倫/Chou Jie
Lun is actually Jay Chou? So, proper sortnames make it more difficult
for international users.

I would argue that sort names (correct or not) are useless for
*sorting* Chinese releases as well, and sometimes make it harder to
find that which shouldn't be that hard to find. The reason is that
Chinese names can be transliterated at least with Hanyu pinyin
(mainland China), Wade-Giles (Taiwan) or by the Cantonese
pronounciation (I don't know the name of the system). The surname 周
can be written in at least 3 different ways: Zhou/Chou/Chow. The first
two both Mandarin, just two different romanization systems! This is in
no way helpful as one would expect artists by the same surname in the
same language to be in the same place...

So, what to do about this? To be honest, I've never understood what
sort names are good for, so I might be missing something important.
It's said that it's how things would be sorted in a record store, but
I seriously doubt any record store would consider sorting music of
foreign script under the latin alphabet. More likely, they would sort
the Chinese stuff separately by a Chinese sorting system and the same
for Japanese and so on...

This may have to wait until NGS, but it would make sense to simply
have one artist for each variant name and have
translation/transcription ARs between them. Ideally the artist name
would also have a language/script so that taggers could automatically
choose a better alias.

Are there any moved to solve this problem on the way? In the meantime,
what would you like to be done for Chinese artist sort names? I've
actually changed a few sort names from the widely known English to the
correct but never heard of romanization, but have stopped because I
can't imagine it makes anyone happy.

Thoughts?

Philip