Detail #296: A Restriction on Possession

In Proto-Uralic (and even possibly Proto-Finnic), possessive suffixes could not be applied to subjects. This leaves an interesting morphological consequence in modern Finnish - subjects with possessive suffixes on them are morphologically indistinguishable from objects with possessive suffixes on them, even when there's significant differences in the subject and object forms:

season:
nominative: kausi
accusative: kauden
possessed nom-acc:

1sg: kauteni
2sg: kautesi
3sg/pl: kaudensa, kauteaan
1pl: kaudemme
2pl: kaudenne

We could of course imagine a similar thing with any type of possession - note that Uralic probably permitted using nouns and pronouns in the genitive as possessors of subjects. Semantically, though, it's obvious that subjects can be possessed - they might just maybe be slightly less likely to be so than objects and such, or there might not even be any statistical difference there.

Now, we still need a method for that, and a few options appear:

• some kind of oblique argument of the verb
• some 'other type' of attribute (analogous, say, to the English 'of'-genitive or somesuch)
• not actually permitting possessed subjects, instead demoting possessed subjects to some kind of oblique position, or comparable position to the causee of a causative (the whole possessive construction with subjects could very well be perfectly analogous to causatives)

Now, we can start imagining interesting other differences: in English, for instance, "-'s" covers the same syntactical spot as "the" and other determiners. We could imagine that subjects generally either have a much more restricted set of permitted things in that position, or even none at all; maybe subjects in this language only can be subjects if they're definite, and other agents that are closely related to the verb must be in some kind of oblique position, or maybe there must be some kind of pseudo-definite dummy noun with the actual agent as an attribute of some kind.

Further, one could of course have determiners for subjects behave oddly - maybe they migrate to a position similar to that of an auxiliary, so:

dogs all_(verbal morpho) bark
dogs some_(verbal morpho) like playing fetch

But

I saw some cats