MrKrov wrote:
So position of agreement affixes: Agent before patient and both prefixes, or patient before agent and one a prefix, other a suffix?
If your conlang's ancestor-language was SOV (or APV) then it should be Agent-before-Patient-both-Prefixes.
If your conlang's ancestor-language was SVO (or AVP) then it should be Agent-Prefix, Patient-Suffix.
OTOH if your conlang's ancestor-language was VOS (or VPA) then it should be Patient-before-Agent-both-Suffixes.
Or at least that's my guess, which is based on complete ignorance of your con-diachronics.
Assuming the pattern I guessed above holds, a Patient-Prefix, Agent-Suffix system would imply an ancestor that was OVS (or PVA); and those are rare among real-life natlangs. But they do exist; and besides, I don't even know if your intend your conlang to be a human conlang.
http://wals.info/chapter/104 has 192 languages in which both the person of A and the person of P are marked on the verb.
In 20 of them the AP-person marker is a fused marker (like Marc Okrand's "pronominal prefix" for Klingon verbs).
In another 19, both orders of A and P person-markers occur. Among the other 153, most (96 of them) have the A person-marker before the P person-marker; but that includes situations in which they are both prefixes (so that the P marker comes closer to the verb-stem), situations in which they are both suffixes (so that the A marker comes closer to the verb-stem), and situations in which the A marker is a prefix and the P marker is a suffix.
The remaining 57 have the P person-marker before the A person-marker; but, again, that includes three possibilities; both prefixes (so the A is closest to the stem), both suffixes (so the P is closest to the stem), and a P prefix but an A suffix.
I do not know how to find out whether A-P-stem is more common than P-stem-A or not.