Detail #154: Some Experiments with Inverse Alignment

Inverse alignment is a system wherein nouns are ordered on a hierarchy of animacy  or somesuch (often also person, such that i.e. 1 > 2 > proximative > obviative or similar, as in Ojibwe). Objects and subjects are not explicitly marked as such by congruence on the verb nor by case marking. Instead, verbs have a marker that can invert the order, so that the lower-ranked noun is subject instead of the higher-ranked one. Unlike a passive, this does not reduce transitivity.

However, could we have some neat other things going on? How about involving the indirect object in some way? It's quite natural that the indirect object will be higher ranked than the direct object; might we simply have two suffixes, "-INV1" and "-INV2", where INV1 switches subject and object, and INV2 switches indirect object and object.

Now, of course, this ends up somewhat incomplete. We can now generate the meanings that follow in English. Note that the examples are dumb, but it's pretty hard to come up with an example where three nouns that all could be objects or recipients or subjects occur. Let's assume a culture where kingdoms are legal entities distinct from kings, and where slavery is permitted so as to make sense of it all.

assumed hierarchy: the woman > a man > a kingdom (note, the = prox, a = obv in this example, and not the usual def. vs. indef)

the woman gave a man a kingdom (no inversion)
the kingdom gave a man a woman (INV1)
the woman gave a kingdom a man (INV2)
the kingdom gave the woman a man or the kingdom (INV1+INV2)

I'm making some odd assumptions here: 1) inverse switches the highest and lowest. This seems unnatural - it is more likely for the highest and next highest to switch places. Let's fix that. Now INV1 switches the two "highest", and INV2 switches the two lowest.

the woman gave a man a kingdom (no inversion)
a man gave the woman a kingdom (INV1)
the woman gave a kingdom a man (INV2) 

We still lack a few possible meanings. I don't like the idea of repeating operations like these - seems too much like some kind of algebra. An INV3 that switches the end-point arguments would probably be a bit better than repeated operations. We'd still lack one possible meaning:

the woman gave a man a kingdom (no inversion)
a kingdom gave a man the woman (INV3)
a man gave the woman a kingdom (INV1)
the woman gave a kingdom a man (INV2)
?a kingdom gave the woman a man (lacking)
?a man gave a kingdom the woman

Some languages do not really distinguish direct and indirect objects, so we could possible accept conflating some of them. Having all six possible permutations as independent morphemes seems somewhat boring. Combining the INV-suffixes could work - and it would be interesting to do some weird stuff where the combinations are not entirely straightforward.