In some languages, nouns low on the animacy hierarchy, particularly inanimates cannot surface as A, and if a situation arises where they are underlyingly A, some reparative strategy such as a passive must be used. Some theorise that this was the case in early PIE, and a friend of mine mentioned Algonquian and Japhung Qiangic in the conversation that lead me to consider this:
Are there any languages where a similar rule is found at the other end of the animacy hierarchy, such that the 1st person is defective and cannot be O, and an underlying 1st person O requires a syntactic derivation to be used, moving the first person either to S (a passive) or oblique status (an antipassive)?
Chukchi seems to come close, as it has a verbal prefix (i)ne- ~ (e)na-, which Comrie(1979) analyses as a "de-ergativising prefix" which in some cases acts like an antipassive is required with 1st person singular Os when the A is not 3rd person plural (in which case a 1sg.S/O suffix is used), however it's not quite a full antipassive as in these cases, as while there is a change to verbal person agreement (absolutive forms are used to mark underlying A) there is not change to nominal case marking, and furthermore, in the "present II"-tense it's used with a much larger set of A/P combinations. Further examples of borderline cases like Chukchi would be fine, but what I'm really interested in is if there are any more clear-cut examples where the construction involved is clearly a syntactic derivation.
Comrie, B. 1979: Degrees of Ergativity: Some Chukchee evidence, pp. 219-40 in Plank, F. 1979 (ed.): Ergativity - Towards a Theory of Grammatical Relations, Academic Press