If Lisp is meant to be a formal model of computation ( a quote from the Bel article by pg), then other formal models of computation should also be examined; perhaps 'compare and contrast' of two different perspectives will lead to a new thought or even a sort of amalgamation.

Lisp and Bel's basic data structure is the list. APL's is the array, which can be seen as a list or in multidimensional form as a list of lists.

Lisp uses a form of notation that is quite different from the APL notation.

If there are advantages to both notations then it makes sense to study both of them.

Bel's description of functions might benefit from looking at APL's niladic, monadic and dyadic functions, as another example.

In another comment above, pg wrote: "language A is better than language B if programs are shorter in A"

APL's terseness due to weird characters aside, since he did say that's not what he meant, APL can do A LOT by chaining a small count of operators together, which, to me, does seem to fit with the above quote rather well.

I would personally also suggest a Forth-derivative like Factor as another language that would meet the criteria of being small in parse-tree terms (and incidentally also in tokens).

As for what it would bring into Lisp.. I dunno. Maybe nothing :)