A direct-inverse language, Wikipedia claims, is one which
involve[s] different grammar for transitive predications according to the relative positions of their "subject" and their "object" on a person hierarchy, which, in turn, is some combination of saliency and animacy specific to a given language.
First person generally outranks second person, and these two outrank non-speech participant arguments. When the subject outranks to the object, the verbs are generally unmarked. Thus, in a language with free word order, without case marking, and without the marking of verbal arguments on the verb, one would expect the gloss
1SG 2SG killed
to translate as
"I killed you."
The gloss of
1SG 2SG killed.INV,
with "killed" morphologically marked for an "inverse voice" of sorts, would, in turn, correspond to the non-prototypical situation in which the lower ranking argument acts on the higher ranking one. In this case, the expected translation would be
"You killed me."
All the literature I've seen investigates languages in which the inverse morphology appears on the verb. Nonetheless, inverse marking on the verbal arguments seems at least conceptually sound to me. In a hypothetical language with this sort of alignment, there would exist an "inverse case" slapped onto, let's say, the subject if it is lower ranking than the object. In such a language the former gloss would remain as it is, but the latter would be as follows:
1SG 2SG.INV killed.
Why is the "inverse alignment" typologically absent? Is there something incoherent about the very idea, is it an accidental gap, or is it a result of broader typological tendencies found among the direct-inverse languages?
