The Northern Khanty third person singular possessive suffix -əλ [POSS.3SG] apart from its typical possessive function (1) also appears in contexts where an entity is mentioned that is part of a contextually given group of entities (2).
(1) λʉw kătˊ-eλ moś-λ
(s)he cat-POSS.3SG purr-NPST[3SG]
‘His/her cat purrs’.
(2) Context: — How can I help you? — See the two cups on the table?
i an-əλ mij-e
one cup-POSS.3SG give-IMP.SG.SG
‘Give me a cup [one of these]’.
In this talk I will argue that in (2) a different, independent marker is used which is distinguished from the proper possessive by its morphosyntax and semantics. It looks like a proper possessive, but it does not function like one.
This particular unpossessive — a marker homonymous with, but independent from some possessive — appears specifically in contexts like (2) marking partitivity, so one may reasonably call it an (unpossessive) partitive article.
Unlike a proper possessive (1), it does not admit an explicit possessor. Thus, adding a personal pronoun possessor to a partitive marked noun phrase (NP) yields a possessive interpretation (3).
(3) #λʉw i an-əλ / i λʉw an-əλ mij-e
(s)he one cup-POSS.3SG / one (s)he cup-POSS.3SG give-IMP.SG.SG
‘Give me one of his cups’.
Intended: [same as (2)].
Unlike a proper possessive which indexes the person-number features of the possessor, the partitive article is not consistent in person-number features with its antecedent (“possessor”) NP. Thus, it looks like a POSS.3SG even with a plural antecedent (2) (and the expected POSS.3DU or POSS.3PL are infelicitous in (2)).
One may retort that such disagreement is due to the marker agreeing with a “plurality as a single entity”, but evidence from other anaphoric expressions of Northern Khanty speaks against this hypothesis. It appears from the field texts available to me and from my own elicited examples, that no other anaphoric expression of Northern Khanty admits such disagreement in number in similar contexts.
To take a minimal pair, the anaphoric demonstrative śit [DEM] must still be marked plural in this context, since it refers to a plurality of cups from the preceding context (4).
(4) Context: [same as (2)].
a. #śit-en mij-e
DEM-POSS.2SG дать-IMP.SG.SG
Intended: give me these [cups].
b. śit-ŋəλ-an mij-a-λi
DEM-DU-POSS.2SG дать-IMP-NSG.OBJ
‘Give me these [cups]’.
I take these arguments as showing that the partitive article should indeed be diagnosed “unpossessive” — a marker that is homonymous with, but independent from some possessive — in this case the POSS.3SG. (In the talk I will also consider contexts like (2) with an animate antecedent and like (4) with a personal pronoun.)
Similar arguments can be (and, in fact, have been) put forth for other Uralic languages with a partitive POSS.3SG, as well as, most likely, for multiple other cases of extended, non-possessive uses of possessive markers elsewhere.