Turkic subjects of nominalised clauses: nmod:poss or nsubj:poss?

[jonorthwash]

In nominalised clauses in Turkic languages, the subject can often be in genitive case, in what looks like an nmod:poss relation to the nominalised form of the verb. For example (Kyrgyz):

(1) Досумдун            алыс кеткени        жаман болду.
    dos-um-dun          alıs ket-ken-i      jaman bol-du
    friend-POSS.1SG-GEN far leave-VN-POSS.3 bad   be-PST.3
    "It was bad that my friend went far away." / "My friend('s) going far away was bad."

In adjectival clauses, the subject of the relativised clause can also be in genitive case, in what looks like an nmod:poss relation to the nominalised form of the head noun, or in Turkish sometimes the verb. For example (Kyrgyz):

(2) Досумдун             барган  жери         алыс болду.
    Dos-um-dun           bar-ğan jer-i        alıs bol-du
    friend-POSS.1SG-GEN  go-VADJ place-POSS.3 far  be-PST.3
    "The place my friend went was far."

In UD-Turkish_BOUN, these subjects are marked nsubj:poss dependents of the verb. In UD-Kazakh_KTB, these are marked nsubj dependents of the verb. They also feel a little bit like the subject has been raised out of the verbal clause so that the genitive noun c-commands the head of the DP that the verbal noun or the RC head occupies, in which case nmod:poss on the verb (in 1) or RC head noun (in 2) might be appropriate. However, having nsubj:poss for these makes them easy to find and examine as a distinct phenomenon.

Note that in both sentences in Kyrgyz, a nominative subject is also grammatical; in (2), the head noun (жери) would need to not be possessed (жер).