Another reference is Kirby’s [Theory of Concatenative Combinators](http://tunes.org/~iepos/joy.html).

Reading over this and Kleffner’s thesis Juan linked (its only 36 pages long) I’m a bit puzzled how to tease a category out. From Juan’s thesis the type signatures of concatenative combinators always seem to have some side condition.

I think it might be that you could model concatenative languages with a Kleisli category, but that feels like cheating.