you still can have dynamic semantics at read-time -- eg, does the macro-expand in reading the first form load new macros that affect the next form? Eg, in (progn (M) (M)), is it ever the case that the second M will dispatch to a different macro definition than the first M?
In the world of Common Lisp, yes. In the world of R6RS's more static semantics, no.
Forgive my ignorance, but aren't Lisp's semantics very compilation-friendly? Doesn't macro expansion happen at read-time?
EDIT: I think that's the case for Common Lisp. By "the Lisp tradition" did you mean prior dialects?