Case and PPs
5.1 Elements of the extended nominal projection above DP
Above the DP layer, the Hungarian extended nominal projection contains case markers and postpo-sitions. That case markers have their own dedicated projection in the nominal functional sequence enjoys tremendous acceptance ever since Travis and Lamontagne (1992).1
(1) KP > DP
The discussion of adpositions (Ps) in the context of the extended nominal projection is moti-vated by several factors. P and K are known to interact cross-linguistically, for instance in the form of adpositions subcategorizing for case markers, or both Ps and Ks being able to express spatial relations. Grimshaw (1991, 2000, 2005) argue that the maximal extended projection of the noun is PP rather than DP, and some researchers have proposed that adpositions and case markers belong to the same category (Fillmore, 1968; Emonds, 1985 and much recent work).
Beyond these broader theoretical issues, considerations specific to Hungarian also motivate the discussion of postpositions in the extended nominal projection. Hungarian postpositions fall into two natural classes. So-called dressed Ps take complements which have no morphologically visible case. So-called naked Ps, on the other hand, subcategorize for specific oblique cases.
(2) a the
sz´ek chair
alatt under
‘under the chair’ naked P
(3) a the
sz´ek-en chair-sup
t´ul beyond
‘beyond the chair’ dressed P
We have seen in Chapter 4 that inflecting demonstratives sit in spec, DP and must bear the same case marker (and number morphology) as the noun they modify.
(4) a. h´az-ak-at house-pl-acc
‘houses’
b. az-ok-at that-pl-acc
a the
h´az-ak-at house-pl-acc
‘those houses’
Unsurprisingly, this requirement also extends to the case marker selected by naked Ps.
(5) az-ok-hoz that-pl-allat
a the
h´az-ak-hoz house-pl-allat
k¨ozel close.to
‘close to those houses’
However, just like case markers and unlike naked Ps, dressed postpositions must also be copied onto the DP-internal demonstrative.
1See many Distributed Morphology papers for a different view, e.g. Embick and Noyer (1999, 2001); McFadden (2004); Embick and Noyer (2007).
105
(6) a. az-ok that-pl
mellett next.to
a the
h´az-ak house-pl
mellett next.to
‘next to those houses’
b. *az-ok that-pl
a the
h´az-ak house-pl
mellett next.to
‘next to those houses’
K and P thus especially strongly interact in Hungarian: one group of Ps select for cases and the other group of Ps behave like cases.
This pattern raises two fundamental questions: i) what is the relation between K and P, and ii) how to characterize the dressed versus naked P divide. The aim of this chapter is to solve these two mysteries (which, I will argue, are tightly linked to the lexicalization problem) and to shed light on some theoretical issues pertaining to case markers and adpositions.
The chapter is structured as follows. The rest of Section 1 is devoted to the inventory of case markers and Ps in Hungarian. Section 5.2 lays out the distribution of dressed and naked Ps. In Section 5.3 I propose a Nanosyntactic account of the observed patters. Section 5.4 shows how the analysis captures the distributional differences and similarities between the two types of Ps.
In Section 5.5 I turn to the nature of the relationship between P and K and refine the analysis.
Finally Section 5.6 rounds off the chapter.
5.1.1 The inventory of Hungarian case markers
Hungarian has a rich inventory of nominal suffixes. Which of these suffixes are case markers is a question that to date has not received a consensual answer in the literature. According to the strictest counts, there are 16 cases excluding the phonologically zero Nominative (Payne and Chisarik, 2000) or 17 cases including the Nominative (Antal, 1961; Kornai, 1986). Depending on how casemarkerhood is defined, other authors identify up to 30 case markers in the language.
For the purposes of this chapter, nothing crucial hinges on the exact number of case suffixes.
The main contribution of this chapter is the novel analysis of the distribution of postpositions in Section 5.3, therefore I will not enter into a discussion of what is a case marker in Hungarian.
(The case markers selected by adpositions are included even in the most restrictive lists of cases and their status as cases has not been debated, c.f. Antal, 1961; Kornai, 1986; Payne and Chisarik, 2000; Kiefer, 2006.)
For the sake of explicitness, and the reader’s convenience, however, I include here the definition of case markers from Payne and Chisarik (2000). Payne and Chisarik suggest that a nominal suffix is a case marker if it can mark noun phrases with a full range of determiners and premodifiers; and it is stranded in noun phrase ellipsis and attaches to the linearly last (originally prenominal) modifier.2 The suffixes that satisfy these criteria are given in table 5.1. For the sake of completeness, I have added Nominative to Payne and Chisarik’s list.
2This property of noun phrase ellipsis has been already discussed in Chapter 3 and will not be elaborated on here.
Name of case spell-out meaning
Nominative -∅ subject
Accusative -t object
Dative -nak/nek possessor, goal, recipient Instrumental -val/vel with, instrument/companion Illative -ba/be to interior
Inessive -ban/ben at interior Elative -b´ol/b˝ol from interior Subative -ra/re to exterior Superessive -on/en/¨on at exterior Delative -r´ol/r˝ol from interior Allative -hoz/hez/h¨oz to proximity Adessive -n´al/n´el at proximity Ablative -t´ol/t˝ol from proximity
Causal -´ert for the sake of
Terminative -ig until, as far as Translative -v´a/v´e into (change of state) Temporal -kor at (temporal only)
Table 5.1: Hungarian case markers
Genitive case is conspicuous by its absence in table 5.1. Hungarian uses Dative case in many instances where Genitive would be expected, for instance on possessors, and the consensual view is that Hungarian lacks Genitive case altogether. In Chapter 8, however, I am going to argue that the so-called possessive anaphor-´eis the Genitive case marker, in fact (c.f. also Bartos, 2001a).
5.1.2 The inventory of Hungarian postpositions
As already mentioned before, Hungarian has two classes of adpositions: dressed Ps have comple-ments without visible case and naked Ps have oblique complecomple-ments.
(7) a the
h´az house
el˝ott in.front.of
‘in front of the house’ dressed P
(8) a the
h´az-on house-sup
kereszt¨ul through
‘through the house’ naked P
The terms ‘dressed P’ and‘naked P’ come from Mar´acz (1986) and were meant to suggest that dressed Ps have something that naked Ps don’t. Hungarian postpositions agree with pronominal DP-complements. Mar´acz’s original observation is that with dressed Ps the agreement is suffixed to the postposition itself, while with naked Ps the agreement is suffixed to the case-marker (leaving the P agreementless, or naked). I will use these labels because they are well-known in the literature on Hungarian PPs. In addition, these terms are appropriate for my analysis, too, as I will suggest that dressed Ps have a K feature that naked Ps do not. That is, one can think of dressed Ps as
‘wearing’ a K feature as an additional garment in addition to what naked Ps have.
The two types of postpositions are listed in tables 5.2 and 5.3.3 Note that postpositions are not partitioned into the dressed and naked classes based on their semantics. One finds Place-denoting, Path-denoting and non-spatial postpositions in both groups. Hungarian is thus different from German, for instance, where the choice of case correlates with the Place/Path distinction (Dative case goes with Place and Accusative case goes with Path).
3The list has been compiled on the basis of Kenesei et al. (1997) and Asbury (2008a), the former claims to be near-exhaustive. I have modified the glosses to reflect the three-way distinction of at,to and from marked postpositions. The reason for shading some of the rows in both tables will be clarified later on.
postposition meaning case agreement with pronouns
alul below superessive doesn’t co-occur with a pronoun
bel¨ul inside of superessive doesn’t co-occur with a pronoun fel¨ul over superessive doesn’t co-occur with a pronoun innen on this side of superessive doesn’t co-occur with a pronoun k´ıv¨ul-re outside-to, beside-to superessive doesn’t co-occur with a pronoun k´ıv¨ul-r˝ol outside-from superessive doesn’t co-occur with a pronoun t´ul-ra beyond-to superessive doesn’t co-occur with a pronoun t´ul-r´ol beyond-from superessive doesn’t co-occur with a pronoun (fogva) as a result of adessive doesn’t co-occur with a pronoun (fogva) from (time) ablative doesn’t co-occur with a pronoun (kezdve) beginning from ablative doesn’t co-occur with a pronoun (kiv´eve) except for accusative yes, on the case-marker
(n´ezve) regarding sublative yes, on the case-marker
´
at through, across superessive yes, on the case-marker egy¨utt together instrumental yes, on the case-marker hasonl´oan similarly to allative yes, on the case-marker k´epest compared to allative yes, on the case-marker kereszt¨ul through superessive yes, on the case-marker k´ıv¨ul outside, beside superessive yes, on the case-marker
k¨ozel close to allative yes, on the case-marker
szembe opposite.to instrumental yes, on the case-marker szemben opposite.at instrumental yes, on the case-marker szemb˝ol opposite.from instrumental yes, on the case-marker szemk¨ozt opposite.at instrumental yes, on the case-marker
t´ul beyond superessive yes, on the case-marker
v´egig (along) to the end of superessive yes, on the case-marker Table 5.2: Naked postpositions
postposition meaning agreement with pronouns
alatt under.at yes, on the P
al´a under.to yes, on the P
al´ol under.from yes, on the P
el˝ott in.front.of.at yes, on the P
el´e in.front.of.to yes, on the P
el˝ol in.front.of.from yes, on the P
felett/f¨ol¨ott above.at, over.at yes, on the P f¨ol´e above.to, over.to yes, on the P f¨ol¨ul above.from, over.from yes, on the P
k¨or´e arond.to yes, on the P
k¨or¨ul around.at yes, on the P
k¨oz¨ott between.at, among.at yes, on the P k¨oz´e between.to, among.to yes, on the P k¨oz¨ul between.from, among.from yes, on the P
mellett near.at yes, on the P
mell´e near.to yes, on the P
mell˝ol near.from yes, on the P
m¨og¨ott behind.at yes, on the P
m¨og´e behind.to yes, on the P
m¨og¨ul behind.from yes, on the P
fel´e towards yes, on the P
fel˝ol from the direction of yes, on the P
´
altal by yes, on the P
postposition meaning agreement with pronouns
ellen against yes, on the P
helyett instead of yes, on the P
ir´ant towards yes, on the P
miatt because of yes, on the P
n´elk¨ul without yes, on the P
szerint according to, in the opinion of yes, on the P
ut´an after yes, on the P
jav´ara in favour of yes, even with a lexical DP kedv´e´ert for the sake of yes, even with a lexical DP l´et´ere despite being yes, even with a lexical DP
r´esz´ere for (DAT) yes, even with a lexical DP
r´ev´en through, by means of yes, even with a lexical DP sz´am´ara for (DAT) yes, even with a lexical DP
ellen´ere despite doesn’t co-occur with a pronoun
eset´en in case of doesn’t co-occur with a pronoun
folyt´an as a consequence of doesn’t co-occur with a pronoun
gyan´ant as doesn’t co-occur with a pronoun
m´ulva in, after (time) doesn’t co-occur with a pronoun
nyom´an based on doesn’t co-occur with a pronoun
´
ota since (point of time) doesn’t co-occur with a pronoun sor´an in the course of doesn’t co-occur with a pronoun t´ajban/t´ajt around (point in time) doesn’t co-occur with a pronoun
´
utj´an by way of doesn’t co-occur with a pronoun
v´egett with the aim of co-occurs with a pronoun only in nonstandard Hungarian, then yes
Table 5.3: Dressed postpositions
Let us now turn out attention to those items in table 5.2 that appear in the shaded rows.
E. Kiss (1999, 2002) observe that these naked postpositions are derived from verbs by adding the´ -va/-ve suffix and argues that these are not true postpositions, but participles subcategorizing for case-marked nouns, in fact. I agree with her and exclude these items from the discussion.
This already leads to a significant simplification of the pattern. Above, naked Ps were defined as postpositions taking nouns in some oblique case. The only exception is kiv´eve ‘except for’, which takes an Accusative-marked complement. Nowkiv´eve is transparently built up of the verbal particleki ‘out’, the verb vesz ‘take’ and the participial suffix -va. Compositionally, this should
give‘taking X out (from the discussion)’, which is exactly whatexcept f or means. Naked Ps thus
can be accurately defined as Ps taking oblique complements.
Antal (1961); ´E. Kiss (1999, 2002) and Trommer (2008) do not categorize the rest of the items in table 5.2 as postpositions either. ´E. Kiss (2002), for instance, treats them as adverbs instead. The arguments for not including naked Ps in the class of adpositions seem to be very strong at first. In particular, i) naked Ps can be separated from their complements by various movement operations while dressed Ps cannot, ii) naked Ps can be intransitive while dressed Ps cannot, and iii) dressed Ps bear agreement and copy onto demonstratives while naked Ps do not. However, contra Antal (1961); ´E. Kiss (2002) and Trommer (2008) I will argue that naked Ps are nevertheless true Ps.
I will show that all the differences boil down to the size of the lexical items involved: dressed Ps spell out a larger piece of structure than naked Ps do. Consequently, I will place naked Ps inside an extended PP in syntax. Mar´acz (1989, 1986, 1984); Payne and Chisarik (2000); K´ad´ar (2009) also treat naked Ps as postpositions. For detailed arguments concerning the categorial identity of naked Ps and adpositions, see K´ad´ar (2009) (she suggests that the label PP could be extended to adverbs in general); Asbury (2005); Asbury et al. (2007); Asbury (2008a,b) (she argues that naked Ps are categorially adpositions and that‘adverb’ is a function rather than a category); and Heged˝us (2006, 2007) (she proposes that naked Ps originate in the Path head that also houses directional case-markers and dressed Ps).
As for the lexical items in the shaded rows of table 5.3, they take a complement without visible case but bear agreement with a full DP complement, too. Agreement with a full DP is atypical for a postposition but standard in possessive constructions. This raises the question whether these words could be analyzed as NP possessees. As it turns out, there is some evidence for this conjecture.
All of them are transparently multi-morphemic, consisting of a noun, an agreement marker and a locative case-marker (10). This is the same as the order of morphemes in possessive constructions (9).
(9) kert-em-ben
garden-poss.1sg-iness
‘in my garden’
(10) a. l´et-em-re
existence-poss.1sg-sublat
‘despite me being’
b. r´esz-ed-re
share-poss.2sg-sublat
‘for you’
c. r´ev-´e-n
ferry-poss.3sg-sup
‘by means of him’
d. sz´am-unk-ra
number-poss.1pl-sublat
‘for us’
Given that the morphological make-up of these words is exactly like that of possessive con-structions and that they agree with full DP complements, I will follow Kenesei (1992) and ´E. Kiss (1999) in treating them as possessive-marked DPs. This allows us to maintain the generalization that postpositions only agree with pronominal complements.
As for the items in table 5.3 that don’t co-occur with a pronoun, most of them are transparently multi-morphemic: several of them appear to have a possessive structure, similarly to (10), and some of them have the-va adverbial suffix. I will exclude these from the discussion.