ISO code, according to the W3C Working Group Note PDF16 (http://www.w3.org/TR/WCAG20-TECHS/PDF16.html):
"Acrobat includes 16 preset language selections. If you need to specify a language that is not on the list, such as Russian, you must type the ISO 639 code for the language, not its name."
I'm not sure what version of Acrobat the W3C was using at the time. There are more than 16 languages listed in v. 10, incuding Russian, but I am pretty confident "the ISO 639 code for the language, not its name" still applies.
That's what my research found, as well, I just wondered if IF Vietnamese would work. Wouldn't it be nice if you could see a human-readable version of the underlying code of a PDF? To see where the program translates the language choice into ISO, if that's what happens, for example.
I wonder how might I test the two language label choices? Anybody out there had any experience with this?
There really isn't any choice. What you put in goes directly into the PDF so it must be what the PDF standard specifies. That's going to be the ISO 639 code. It isn't checked because Acrobat doesn't contain a full list, so it has no way to know if what you enter is valid.