Well, introducing the new term “Abelian” certainly avoids overloading terminology in a confusing way :-)
“Pair invariant” IMO, isn’t good, certainly worse than “Pairing invariant”. “Parentheses invariant” might work, but of course would get just as confusing/incorrect as “pair invariant” once one moves from groups to fields.
Perhaps pair-independent is a distinctive name for associative...