Agreement and the plural on the possessee

In Standard Hungarian the order of nominal suffixes is possessedness marker followed by the plural (-i allomorph) followed by agreement (and case).

(21) Standard Hungarian:

noun – plural – possessive agreement – case

(22) a the

mi we

kert-je-i-nk

garden-poss-pl-poss.1pl

‘our gardens’

This is the same morpheme order that we have observed for Turkish, Kolyma Yukaghir and Bolivian Quechua in Chapter 6.

In some varieties of Hungarian the order of the possessive agreement and the plural is reversed with respect to the Standard order.

(23) some varieties of non-standard Hungarian:

noun – possessive agreement – plural – case

Example (24) is from the G¨ocsej dialect, while example (25) is representative of the Southwestern and ˝Ors´eg dialects (K´alm´an, 1966, p. 53). Note that plural’s-k allomorph is used instead of-i, as here the plural does not immediately follow a possessedness suffix-ja/-je/-a/-e.

(24) a. h´az-ank-ok

house-poss.1pl-pl

‘our houses’

b. kert-¨unk-¨ok

garden-poss.1pl-pl

‘our gardens’

(25) ¨okr-¨ot¨ok-ek ox-poss.2pl-pl

‘your oxen’

In chapter 6 we have seen that this order is standard in Chuvash, Huallaga Quechua and Kharia.

To capture the order of suffixes, I suggest that in these nonstandard varieties the uPerson and uNumber agreement features are added to a functional head below NumP. This conclusion is supported by the lack of a distinct possessedness suffix-ja/-je/-a/-e, too. Recall from (6) that in Standard Hungarian, if the possessee is singular, then in first and second person singular and plural, there is only one suffix on the possessed noun. Bartos (1999) argues that this is because in these forms the possessedness marker and the agreement end up adjacent to each other, and undergo Fusion. Fusion is not possible if the possessum is plural, because the plural marker intervenes between the possessedness marker and agreement.⁶ Compare now the possessed forms of kert

‘garden’ in Standard Hungarian and G¨ocsej Hungarian.

(26) kert-je-i-nk

garden-poss-pl-poss.1pl

‘our gardens’ Standard Hungarian

(27) kert-¨unk-¨ok

garden-poss.1pl-pl

‘our gardens’ G¨ocsej Hungarian

Crucially, in (27) there is no distinct possessedness marker, we don’t get something like (28).

This is because the agreement features being merged low, they are adjacent to the possessedness marker-ja/-je/-a/-e (the exponent of the Poss head). As the plural does not intervene between the possessedness marker and the agreement features any more, they can undergo Fusion/co-spellout.⁷

6This explanation can be directly transposed into the Nanosyntactic spell-out algorithm used in this dissertation.

Suppose that the agreement suffixes span Poss, Num with a singular specification and Poss2 with the uPerson and uNumber features. In a structure with Poss2 > Num(sg) > Poss, all three heads are spelled out by the agreement morpheme (the Maximize Span principle). But in a structure like Poss2 > Num(pl) > Poss, the agreement cannot spell out all three heads because it is not a good match for Num(pl). This forces Poss2 (with the agreement features) and Poss to be spelled out by two separate morphemes. This is the same intervention phenomenon that we have seen elsewhere in the dissertation (c.f. the analysis of non-inflecting demonstratives in Chapter 4). I will refrain from fleshing out this analysis in detail, because it would add nothing substantial to Bartos’ original insight.

7In (i) there may be a separate possessedness suffix, though. Based on the standard dialect, we would expect (ii), not (i).

(i) h´az-ank-ok house-poss.1pl-pl

‘our houses’

(ii) *h´az-unk-ok house-poss.1pl-pl

‘our houses’

(28) *kert-je-¨unk-¨ok

garden-poss-poss.1pl-pl

‘our gardens’

In G¨ocsej/ ˝Ors´eg/Southwestern Hungarian, the functional head that bears the uPerson and uNum-ber features is possibly Poss.

(29) Ors´eg, G¨ocsej, Southwestern Hungarian˝ Poss2P

Poss2 NumP

Num -(¨o)k

PossP

tpossessor

Poss

(uPerson, uNumber⇒-¨unk) NP kert

It is also possible, however, that uPerson, uNumber are added to Num and spell out as prefixes to it, as in (30).

(30) Poss2P

Poss2 NumP

Num -(¨o)k

(uPerson, uNumber⇒-¨unk)

PossP Poss NP

kert

I will not attempt to make a principled choice between these options here because that would not contribute substantially to the main point here, which is the possibility of linearizing the agreement markers at different points in different varieties of Hungarian.

Note that the G¨ocsej/ ˝Ors´eg/Southwestern dialects provide evidence against the conjecture that the phrase hosting pronominal nominative possessors in its specifier (‘AgrP’, my Poss2P) is projected by the agreement morpheme itself. I do not have data from G¨ocsej/ ˝Ors´eg/Southwestern Hungarian in which an overt nominative possessor pronoun co-occurs with the possessed nouns in (24) and (25), and I don’t speak this variety myself. It is fair to assume, however, that this co-occurrence is possible – there is no reason to suspect otherwise, and the ban on an overt pronominal possessor would be strange indeed. As in G¨ocsej/ ˝Ors´eg/Southwestern Hungarian the agreement

Therefore the segmentation may be the following, with separate possessedness and agreement markers:

(iii) h´az-a-nk-ok

house-poss-poss.1pl-pl

‘our houses’

But this is not necessarily so: we may simply see a phonological quirk of the given dialect here, whereby its linking vowel is different from the one used in the standard variety. In any case, in (24-b) there is clearly no separate possessedness suffix.

marker precedes the plural, it is either in Num or below Num. Consequently it could not possibly project the phrase hosting the nominative pronominal possessor, which is above NumP. Note that this reasoning is independent of whether agreement markers project their own phrase or not. Even under an analysis in which they do, (24) and (25) strongly argue against a direct correlation between their projected AgrP and the surface position of Nominative pronominal possessors.