No, I don't think Perl5ers would find the last example most approachable. Even with those superficial similarities like sigils and the sub keyword. They will go all what-the-fuck. I would find the Haskell version much easier to explain.
Regarding the mixing of languages: it's exactly the other way around! It does make sense to mix languages inline IF (and almost only if) they are NOT closely related. It makes sense if 1) they serve a different purpose and 2) can be (almost) immediately told apart! If you have to scroll up to see whether that thing you are looking at is Perl6, Perl5 using some feature you were not aware of or Perl5 with a source filter or some clever prototype hackery, then you are wasting everyone's time and introducing bugs. Bugs coming from subtle differences of the languages, from misunderstanding, from reading the wrong documentation, ...
Enoch was right!
Enjoy the last years of Rome.