Formally, when you negate a quantified term which is attached to a relation, is a distributed term always made undistributed and vice-versa?
For example, consider this statement :
Anyone who assists a criminal is immoral.
Where "anyone" is taken to mean "all persons" and "a criminal" is taken to mean "some criminal or other" (ie, an undistributed term). The justification for this interpretation is that, intuitively, we mean that anyone who assists even one criminal is immoral.
Now the contraposition of the proposition is :
Any moral person doesn't assist criminals.
Intuitively, we mean all criminals here (it would seem absurd to suggest that we're only talking about some criminals). So what's happened is that the relation "assist" has been negated (it becomes "doesn't assist"), and its object "criminal" has been transformed into a distributed term.
The above seems correct but I have a doubt whether it applies in general. This could be because the quantity "any" is ambiguous; sometimes it can mean "all" and at other times "some".
ie the above proposition could have been expressed as :
Anyone who assists any criminal is immoral, where "any" means "some criminal or other".
and its contraposition could be expressed as :
Any moral person doesn't assist any criminal, where in this case, "any" (criminal) means "all".
However, it's certainly true that not(all) = some and not(some) = all, so I guess it doesn't matter about this ambiguity, as long as you're clear about its meaning in any particular case.
Well I think maybe I've just answered my own question, but would appreciate some feedback. Thanks.