[abrudz]

APL brings another couple of ternaries (besides for those mentioned with regards to J):

  x f[k] y

This one is universal among traditional APLs (though some modern APLs, including J, remove it). However, it is traditionally only used for a concept of application along an axis of APL's multidimentional arrays, and only for a strictly limited set of built-in "operators". That said, GNU APL and NARS2000 both allow it on user-defined functions (which have infix syntax, just like "operators").

  x n.f y

This is infix application of the function f from the namespace n, so n must have namespace value, but it is still a proper argument because n isn't just an identifier, but can also be an expression:

  x (expression).f y

In fact, the expression can be an entire (multidimentional!) array of namespaces:

  x (namespace1 namespace2).f y

is equivalent to

  (x[1] namespace1.f y[1])(x[2] namespace2.f y[2])

Furthermore, f can be a built-in "operator" which appears to exist in every namespace, but which also takes the current namespace into consideration. For example, x⍳y is the position of y in x and if io0 is a namespace wherein indexing is 0-based and io1 is a namespace wherein indexing is 1-based, we can write:

  'AX' 'BX' (ns0 ns1).⍳ 'X'

and get the result 1 2 because X is at position 1 in AX with 0-based indexing but at position 2 of BX with 1-based indexing. (I'm not saying you should write code like this, but you could.)