The documentation associated to @type on <orth> seems to indicate that it could be used (values such as "trans" or "lat") to indicate the script associated to a specific orthographic transcription. Still, xml:lang should be there to indicate part of this semantics, for instance in the case where we want to encode a Korean word with its transliteration, should we use:
<form xml:lang="ko">
<orth type="hangul">치다</orth>
<orth type="romanized">chida</orth>
</form>
or
<form>
<orth type="standard" xml:lang="ko-Hang">치다</orth>
<orth type="transliterated" xml:lang="ko-Latn">chida</orth>
</form>
or drop the @type attribute altogether?
Script information should _definitely_ be in @xml:lang, so I think your second example is the way to go. @type is still handy for project-level distinctions, as your example shows. so I don't think we should remove it.
We have only one example of <orth> with @xml:lang on the Chinese reference page, and a couple more in the Names and Dates chapter. If it's OK to use your sample encoding, it would make a good example for the main <orth> reference page. Would that be OK?
Perfect for me.
I've added your example to the ref page for <orth>. The <valDesc> has already been commented out by Lou, following this bug report:
http://purl.org/TEI/BUGS/3599062
so I'm closing this ticket.