RE: <span> within a word any issue for screen readers?

I feel that this topic hasn't been specified enough in either WCAG, UAAG, 
nor the HTML and CSS specifications - or at least there are no documented 
best practices for authors, tool vendors, and most importantly screen 
reader vendors.  I also feel that a complete solution requires a way for 
the author to specify how they want to influence the screen reader to 
behave when rendering the visual text in voice. (and what about Braille 
too?) 

Let's take Lois' example:
"In user documentation, it is common to
underline the letter that activates a menu or other screen control in
conjunction with a modifier key: e.g. "Use the Print Pre<u>v</u>iew menu 
to
see what your file will look like" (or one could better use a styled span 
to
achieve same)."

In this case, a windows screen reader would say something like"Print 
Preview Control V"  when reading a menu list of options.  But, how should 
this be spoken inside a web page? A "Control V" won't do anything for 
Internet Explorer, Fire Fox, or HPR on the Windows platform.  Should the 
screen reader say "Print Preview Underlined V" .  And how does the author 
hint to the screen reader that this is or isn't a shortcut key?  Use 
accesskey?  and where is that best practice implemented?

My point here is that there are some immediate bugs in screen reader 
behavior that could be fixed.  Simple examples like:
Acc<span>e</span>ssibility 
should not be pronounced as though it were written: Acc e ssibility (ie, 
as three distinct words).
But there are also some missing specifications from the WCAG, UAAG, and 
HTML & CSS specifications, and more importantly, a missing specification 
that describes the way for the author to specify how to pronounce words 
that have all this nonstructural markup in the middle.  Geoff wrote:
        If SPAN element is empty of attributes ignore
        else apply rules appropriately.
What rules?   If <span> element has some visual styling, should it be 
ignored?  How should it be ignored?  Should bold's be ignored, but 
underlined's be spoken after wards?  Where is that a rule - or best 
practice written?

Lot's of questions with no answers. Lots of specification but few best 
practices.  And I agree that works needs to be done here.  A working group 
could be started to understand it better and make some recommendations, 
but please, don't hold up WCAG 2.0 over it.

Regards,
Phill Jenkins
IBM Worldwide Accessibility Center
http://www.ibm.com/able