That's a good point. And it reminds me of the misnamed bibTeX "key"
field, used to sort items in a reference list that are missing author
data. Likewise a separate MODS element outside of <titleInfo> (with a
more sensible name than <key>, <sort> for example) would be useful. The
desired sort order in some applications might have nothing to do with
the title. The title might even be a made-up identifier and not really
a title at all (e.g. "personal communication" in a reference database).
You mention using pinyin to sort Chinese. For Japanese titles,
however, the sort order would be neither the original nor the
transliterated title. (I'm not sure of the exact rules, but I think
writing the title phonetically in katakana would sort properly by
Unicode code point order). An element outside of <titleInfo> would come
in handy here too, so one could set language attributes on it, e.g.
>>> [log in to unmask] 2004-01-27 12:17:57 >>>
I'd like to add another bit of info to Caroline's very clear statement
of support for the sort title element: machine sorting is only really
easy when you limit your title to the basic ASCII characters. It gets
more complex as you add accented characters, and then ramps up in
complexity through non-Latin alphabets (i.e. Arabic and Hebrew) and
non-alphabetic languages (Chinese, etc.).