Anybody got experience with non-Latin-1 properties files?

We've just hit a stumbling block in a project involving
Turkish translation. It seems that we've got our
EBCDIC-to-Unicode translation routines working perfectly
with Turkish EBCDIC, but now, on the client side, we've
run into a problem with properties files. It seems that no
matter what we do, whether we save them as UTF-8 or as
Unicode, our properties files get treated as 8-bit Latin
1, both by our client-side program, and by JBuilder. (We
can, however, read and write them fine from Windoze
Notepad; go figure.)

We've found that using backslash unicode escapes does work, but that
renders the properties files a bit hard to read and update.

Can anybody suggest a more elegant solution? Or even just a
fully-unicode-escape-aware properties file editor?

--
James H. H. Lampert
Professional Dilettante