i've also tried learning J and can't really say i've managed it yet, but i think the idea is that everything is an array. in J an array can contain a box, and a box might contain an array, so you have nested arrays, (different from multidimensional arrays.)

lisp baffled me for a while because, in my mind, a list is different from a tree, but in lisp a list element might be itseld a list, and so a lisp list might be a tree. similar situation with J arrays.

i'm not sure how APL does it. i /think/ it uses a flat array model, but i'm not sure what that means exactly, or if it's even applicable to modern APL, let alone what it might imply about bending APL arrays into other data structures.