I think maybe the "main issue" you have to deal with when translating
monads to a dynamically-type language is that, in order to write
"generic" functions which work for multiple monads, you need to provide
a way to dispatch to the particular monad you're in.  That is, whereas
a Haskell function might just mention "return" and ">>=" and use the
type inferencer to resolve the overloading, in a dynamically-typed
language, you'd probably have to make all the "(Monad m)=>" functions
take an extra parameter (the dictionary of monad methods).

Note that this issue is not dealt with in the paper at the URL above;
in much of the expository writing on monads, the writer either (1)
assumes you will only be working in one monad (as above), or (2)
provides differently-named operations for each monad (e.g. unitA/bindA
for monad A, unitB/bindB for monad B, etc.).

That's my take, anyway; I don't use dynamically-typed languages very
often, so I may be off.