If you think of an array as a function that maps a linearized index to an element value, you get most of APL for free in Haskell.