[XeTeX] IPA superscript diacritics

[Kevin Russell]

I often use the Unicode characters for IPA superscript diacritics (like 
aspiration [ʰ], palatalization [ʲ], velarization [ʷ]) in the source of 
my XeLaTeX documents.

While the symbols *look* fine in the resulting PDF, they're no longer 
the correct Unicode characters.  For example, apiration [ʰ] is no longer 
u02B0, but a regular "h" (u0068) printed in superscript.  This makes it 
impossible to copy and paste transcriptions from the PDF and still have 
them be correct transcriptions.  More problematically for me right now, 
it feeds incorrect text to the screenreading software of our blind students.

Where along the chain are these characters getting changed and how do I 
get it to stop?

It happens with the xelatex in both MacTeX2008 and MacTeX2010.  It 
happens whether the PDF is viewed with Adobe Reader 9, Adobe Acrobat CS4 
(both for Mac), or with the Mac Preview application.  It happens with 
every Unicode font I've used (Gentium, DejuVu, Linux Libertine).

Thanks for any advice.

-- Kevin Russell