> > glad to learn that; can you perhaps tell us how you resolved the
 > > problem of either killing ligatures and kerns or alternatively
 > > making the wrong choice at the begining of an \halign cell that
 > > afterwards (!) jumps into math mode?
 >
 > One limited possibility would be
 > \mathcode`\ä="8000

yes, that is the sort of only out one would have, but as far as we could see
it isn't going to allow a resolution.

point is

 a) this doesn't help for LICR objects like \" as they potentially execute
 \accent and that bombs in math

 b) and even if \"a expands to a \chardef then the chardef number depends on
 the outer fontencoding so might vary from case to case

 c) and of course that would mean that inputenc mapping would be called upon
 so to make that sfe you would need to swap inputenc encoding in and out ...

 > If the letter then ended up being typeset in math mode instead of text
 > mode, it would get reevaluated.  Unfortunately, this particular trick
 > would require that input encoding and font encoding match...

exactly, which bring us back to square one: Don's big mistake of not
separating the two in the first place

so i doubt that this is helping
frank