Possessive agreement and appositives
7.3 Possessive agreement defies the Mirror Principle
7.3.4 An order that doesn’t look Mirror, but it is
Finally let me turn to an apparent ordering inconsistency that turns out to conform to the ex-pectations if examined carefully. Consider the nominal phrases in (54) and (55): the former with a possessed noun, and the latter with a case marked personal pronoun. In Chapter 5 I argued that the agreement on (55) is possessive agreement. In light of this, the order of case markers and possessive agreement appears to be dependent on whether the nominal base is an R expression or a pronoun.
(54) szem-em-ben eye-poss.1sg-iness
‘in my eye’
(55) (´en)-benn-em I-iness-poss.1sg
‘in me’
14Further, with a third person singular pronoun, the agreement cannot co-occur with the case marker at all (in my idiolect, at least):
Agreement only (i) el´e-je,
to.in.front-3pl
‘to in front of him’
Illative case only (ii) el´e-be,
to.in.front-3pl-illat
‘to in front of him’
Agreement plus Illative case (iii) *el´e-be-je
to.in.front-illat-3pl
‘to in front of him’
(iv) *el´e-je-be
to.in.front-3pl-illat
‘to in front of him’
Let me point out at the start that it has been repeatedly argued in the literature that with case-inflected pronouns like (55), the real stem is apro(or the case marker), and the overt pronoun is just an appositive-like modifier. While surprising at first glance, this analysis is supported by several considerations. Firstly, the pronoun is entirely optional. (56) is just as good:
(56) benn-em iness-1sg
‘in me’
Secondly, the case marker attached to a pronoun does not show vowel harmony, it has a fixed value for each case marker (front for some, including Inessive, and back for others, like Adessive). Vowel harmony is is obligatory with R expressions.
(57) a. a the
szem-ben eye-iness
‘in the eye’
b. *a the
szem-ban eye-iness
‘in the eye’
(58) a. ´en-benn-em I-iness-1sg
‘in me’
b. *´en-bann-am I-iness-1sg
‘in me’
(59) a. a the
szem-n´el eye-adess
‘at the eye’
b. *a the
szem-n´al eye-adess
‘at the eye’
(60) a. *´en-n´el-em I-adess-1sg
‘at me’
b. ´en-n´al-am I-adess-1sg
‘at me’
Finally, the −v of the comitative case suffix -val/vel does not undergo assimilation to the preceding consonant with a pronominal base, but this is obligatory with an R expression base.
(61) a. a the
szem-mel eye-comit
‘with the eye’
b. *a the
szem-vel eye-comit
‘with the eye’
(62) a. *´en-nel-em I-comit-1sg
‘with me’
b. ´en-vel-em I-comit-1sg
‘with me’
These arguments against treating the overt pronoun as the actual stem of (55) are convincing, and I will treat (55) as a nominal expression headed bypro. This, however, has no bearing on the fact that the possessive agreement and the case marker appear in a different order in (54) and (55);
this fact still needs to be accounted for.
In Chapter 5 I argued that case markers spell out, among other P-heads, a silent Nplace. Nplace
participates in a possession relationship: it is the possessee, and its complement KP (the Ground) is the possessor. The proposed structure (with a DP-proas the Ground) is shown in (64).
(63) (´en)-benn-em I-iness-1sg
‘in me’
(64) PathP PlaceP
AxPartP Nplace
KP DP pro
K
Nplace
AxPart
Place Path
-benn
The order of agreement and case markers is thus consistent in (54) and (55) in a relative sense:
the agreement always spells out on the possessee. However, the ordering can be shown to be consistent in absolute terms, too. To wit, the agreement morphemes in (54) and (55) occur at different places because they mark two different possession relationships.
The agreement in (55), as we have seen, is a reflex of the possession relationship between KP, which functions as the Ground, and the silent Nplace. The agreement in (54) (repated here as (65)), however, is a reflex of a possessive relationship internal to the Ground KP (it cross-references the pronominal possessor of the head noun eye). In other words, in (65) the possessor of the silent Nplace is itself a possessed noun phrase,my eye.
(65) szem-em-ben eye-poss.1sg-iness
‘in my eye’
(66) [ in [possessee place[possessor my eye ]]] = [ in [ theplace[ of my eye ]]]
(67)
PathP
PlaceP
AxPartP
NplaceP
KP
Poss2P
pro(1sg)
Poss2 -em
NumP
Num PossP
tpro(1sg)
Poss NP szem
K
Nplace
AxPart Place
Path
-ben
The apparent ordering difference between (63) and (65) is thus epiphenomenal: these examples feature two different agreements in two different phrases (one internal to KP, another internal to the PP). The more pertinent question is thus why (65) does not display the agreement at both places, as in (68); i.e. once below the case suffix to mark the possessedness ofbone, and once after the case suffix to mark the possessedness of Nplace.
(68) *csont-om-ban-agr
bone-poss.1sg-iness-agr
‘in my bone’
The answer to this question follows from the general make-up of possessed noun phrases. Recall from Section 7.2.1 that only pronominal possessors have theirφ-features cross-referenced on the possessee, and there is no agreement (overt or covert) with R-expression possessors. In (63) the Ground has no possession relationship internal to it, the Ground is a simple pronoun. Given the pronominal nature of the Ground, there is agreement between the pronoun and the silent Nplace
possessee. In (65), on the other hand, the possessor of the silent Nplace ismy eye, an R-expression.
The lack of agreement after the case suffix in (65) is thus reducible to the category of the possessor.
One last issue remains to be addressed. For possessive agreement between a possessor and a possessee, I assumed that the agreement features reside in Poss2, the phrase that hosts nominative possessors in its specifier on the surface, and which is located above NumP in the hierarchy. If we now look back at the structure of case markers and postpositions, like in (55), there is no Poss2P.
The agreement features thus reside somewhere else. It is clear that they reside somewhere in the PP sequence, between the silent noun Nplace and the Place node. I suggest that they are on the silent noun Nplace, as NplaceP is the only phrase in this sequence that has anything to do with possession. That the uNumber and uPerson features of possessive agreement can reside in both