From Num to D
4.1 The landscape of the DP between numerals and D
4.1.1 Quantifiers have their own projection
Quantifiers are often lumped together with and placed into the same projection as numerals. This is motivated by the intuition that quantifiers essentially have the same function as cardinals: they both express some quantity. In this view, the only difference between numerals and quantifiers is how specific that quantity is: while numerals express a specific number, quantifiers express some non-specific quantity. Analyses that do not assume different positions for numerals and quantifiers include Pereltsvaig (2006); Rutkowski and Maliszewska (2007) as well as Borer (2005).
The fact that numerals often share morphological properties with quantifiers fits naturally into this view. In Hungarian, for instance, both numerals and quantifiers require the quantified noun to be morphologically singular.
(7) a. h´et seven
katona-(*k) soldier-pl
‘seven soldiers’
b. sok many
katona-(*k) soldier-pl
‘many soldiers’
69
In Russian, the numeralsfive throughone hundredand the quantifiersmnogo ‘many’,nemnogo ‘a little’,stol’ko ‘so much’ andskol’ko ‘how much’ require the quantified noun to bear Genitive plural morphology in Nominative environments (Pesetsky, 2009).1
(8) a. pjat’
‘many(nom) problems’ (Bailyn, 2004, p. 5., ex. 9.) Russian
These facts can be given a unified account if numerals and quantifiers are introduced in the specifier of the same head (call it Num, Q, or #), and the morphological requirement on the noun is attributed to this head.
Numerals and (at least some) quantifiers are often assigned to the same position in Hungarian as well. Kornai (1989), for instance, proposes that the unified treatment should be extended to all quantifiers of the language. ´E. Kiss (2002), on the other hand, has a different opinion and suggests that this is justified only for a subset of Hungarian quantifiers, such as n´eh´any ‘some’, sz´amos
‘several’ andsok ‘many’ .
The distribution ofsok ‘many’,kev´es ‘few’, n´eh´any‘some’ etc. is indeed like that of numerals, and they plausibly occupy spec, NumP. However, other quantifiers have a rather different distri-bution, and the existence of separate QP designated for these elements is well motivated (Bartos, 1999; ´E. Kiss, 2002). In the literature QP has been argued to be the merge-in site of two kinds of elements. Bartos (1999) suggests that the reason why the quantifierminden ‘every’ can co-occur with numerals is that it is merged in spec, QP, which is higher than the position of numerals in spec, NumP.2
‘We give 5 bonus points after every eight/fiteen/hundred forints.’
In addition, quantifiers ending in the‘unique identifying suffix’3 morpheme-ik (or -ik quantifiers for short) have been argued to occupy spec, QP in ´E. Kiss (2002). The-ik morpheme has a diverse distribution: it is found at the end of certain quantifiers such asmind-egy-ik ‘each’ (lit. all-one-ik),b´ar-mely-ik ‘any’ (lit. any-which-ik),vala-mely-ik ‘a certain’ (lit. some-which-ik), as well as in ordinals (¨ot ‘five’, ¨ot-¨od ‘fifth’ as a fraction,¨ot-¨od-ik ‘fifth’ as an ordinal) and in expressions like egyik ‘the one’ (lit: one-ik) and m´as-ik ‘the other’ (lit: other-ik). ´E. Kiss argues that unlike quantifiers likesok ‘many’ or‘n´eh´any’‘some’, which sit in spec, NumP,-ik quantifiers are in spec, QP.
An important property shared by -ik quantifiers and minden ‘every’ is that they cannot be string-adjacent to the definite article.
(10) a. (*a)
‘any soldier / certain soldiers’
This is also true of the quantifier valamennyi ‘each’.4 For this reason, I will treat valamennyi
‘each’ as a spec, QP quantifier as well.
(11) (*a)
1Numerals belowfiverequire the noun to bear Genitive singular morphology.
2That the interpretation of (9) is [every [eight forints] rather than [[every eight] forints] also supports this conclusion.
3The term is taken from Kenesei et al. (1997).
4Valamennyi is ambiguous between ‘each’ and ‘some’. In this chapter I am only concerned with the ‘each’
meaning.
This property, however, does not extend to numerals and quantifiers such assok‘many’ or‘n´eh´any’
‘some’. These happily occur right after the definite article.
(12) a. a
Following Bartos (1999) and ´E. Kiss (2002), I conclude that in addition to NumP Hungarian also has a QP. This projection houses-ik quantifiers as well as minden ‘every’ and valamennyi
‘each’.
(13) QP > NumP > NP
QP, the quantifiers in its specifier and the ban on their adjacency with the definite article will figure prominently in Section 4.2 in the discussion of the lexicalization of the D projection.
4.1.2 Demonstratives
Hungarian demonstratives exhibit a binary distance contrast between proximal and distal, and fall into two morphological classes. So-called inflecting demonstratives obligatorily take the same number and case suffix that the noun does. When the noun occurs with a so-called case-like postposition (also termed ‘dressed’ postposition in the Hungarian adpositional literature), then this adposition, too, must occur on the demonstrative as well.5 Inflecting demonstratives must be immediately followed by the definite article. Demonstratives in this class are proximalez ‘this’ and distalaz ‘that’, as well as compounds formed with these demonstratives, such asugyan-ez ‘same (proximal)’,ugyan-az ‘same (distal)’, mind-ez ‘all this’,mind-az ‘all that’,emez ‘this other one’
andamaz ‘that other one’.
(14) ez-ek-et
‘those exact same students’
(16) mind-az-ok
‘under all those bridges’
So-called non-inflecting demonstratives, on the other hand, cannot take the suffixes of the noun and cannot occur contiguously with the definite article. Demonstratives in this class are proximal e, eme andezen, all of which mean‘this’, and distal ama andazon, both of which mean‘that’.6 (17) eme
5‘Postpositional agreement’ as in (16) is cross-linguistically rare. Adpositions in general and the example in (16) will be a major concern of mine in Chapter 5.
6The proximale has no distal counterparta. A tanul´o-k-at is grammatical but it means‘the students’, where a(z)is the definite article.
These demonstratives are somewhat archaic, poetic or represent a highly elevated register. In contemporary Hungarian, they are used far less frequently than inflecting demonstratives.
There is evidence that inflecting and non-inflecting demonstratives differ not only with respect to their morphological properties, but also occupy different positions in the functional sequence.
As evidenced by (14) and (15), inflecting demonstratives always precede the definite article. Non-inflecting demonstratives cannot be contiguous with a definite article. However, the inclusion of a non-finite restrictive relative clause or a Nominative possessor in (17) through (21) makes the article visible: it obligatorily appears with these modifiers and it precedes the non-inflecting demonstrative.7
‘this letter received from you’
(´E. Kiss, 2002, ch. 7., ex. 7. b.)
‘these hats of mine’
(Szabolcsi, 1994, ex. 24.)
Note that Hungarian personal pronouns cannot take the definite article (´en vs. *az ´en ‘I’).
Therefore the definite article in (23) definitely belongs to the head noun, as depicted in (24). The structure in (25), where the article forms a constituent with the possessor can be safely excluded.
(24) [DP az [ ´en [ ezen [N P kalapjaim ]]]] (25) *[[ az ´en ] [ ezen [N P kalapjaim ]]]]
While inflecting and non-inflecting demonstratives do not co-occur in the same DP, based on their relative ordering with respect to the definite article we can conclude that the surface order is as in (26).
(26) inflecting Dem > definite article > non-inflecting Dem
Cross-linguistically, demonstratives often co-occur with the definite article. When they both pre-cede the noun, then both Dem > Art > N and Art > Dem > N orders are attested.8 Some examples featuring the Dem > Art > N order are provided in (27) through (34).
(27) had
‘this boy’ (Shlonsky, 2004, p. 1489., ex. 55.) Moroccan Arabic
(28) w9y that.one
´a-j7ab art-girl
‘that girl’ (Rijkhoff, 2002) Abkhaz
(29) dan this
il-ktieb the-book
‘this book’ (Plank, 1992, p. 454.) Maltese
(30) afto
‘this house’ (Ezcurdia, 1996) Greek
(31) s´a
‘the old horse’ (Lohndal, 2007, p. 288. ex. 1.a.) Old Norse
7This phenomenon has been extensively discussed in the literature, and it will be the focus of Section 2 of this chapter.
8Demonstratives co-occur with the specificity marker in postnominal position in Gungbe (Aboh, 2004a, 2010a) and (optionally) with the definite article in Zina Kotoko (Demeke, 2002). As is well known, in several languages demonstratives co-occur with the definite article but they are on different sides of the noun. Examples include Welsh, Scottish Gaelic, Irish, Breton, (Roberts, 2011), Spanish (Bernstein, 1997), Kana (Dryer, 2005), Galela (Rijkhoff, 2002), and this is also an option in Greek (Giusti, 1997), Moroccan Arabic (Shlonsky, 2004) and Samoan (Rijkhoff, 2002). Demonstratives co-occur with enclitic articles on different sides of the noun in Norwegian, Albanian (Dimitrova-Vulchanova and Giusti, 1998) and it is also a possibility in Gulf Arabic (Dryer, 2005). Postnominal demonstratives co-occur with an enclitic article in Roumanian (Dimitrova-Vulchanova and Giusti, 1998; Brug`e, 2002). The space of options to derive these orders is substantially bigger than what is necessary to derive Dem >
Art > N and Art > Dem > N.
(32) y`a
‘this house’ (Dryer, 2005, ex. 16.) Ngiti
(33) t´a
‘this new dog’ (Leu, 2008, p. 30. ex. 33. a.) Colloquial Slovenian
(34) ika
‘this child’ (Bernstein, 1997, p. 93., ex. 14. d) Javanese
Examples from languages with the Art > Dem > N order are given in (35) to (38).
(35) der
‘this man’ (Ezcurdia, 1996) Yiddish
(36) ke-ia the-this
kanaka person
‘this man’ (Ezcurdia, 1996) Hawaiian
(37) s´a-ma-y art-dem-lk
ap´ok
grandchild.my
‘my grandchild, i.e. that grandchild of mine’
(Diessel, 1999, p. 92. ex. 23. a.) Pangasinan
(38) si-n¯a art-that
pua’a pig
‘that poor pig’ (note thatsi has an emotional quality)
(Rijkhoff, 2002, p. 183., ex. 25.) Samoan
Thus both the inflecting Dem > D > N and the D > non-inflecting Dem > N orders of Hungarian find parallels in other languages.
The consensus view in the literature is that the definite article is in D, which yields the order in (39).
(39) inflecting Dem > D > non-inflecting Dem
At this juncture two analytical decisions need to be made. First, we need to decide whether inflecting and non-inflecting demonstratives are heads or phrases, and second, we need to be explicit about which functional projections they sit in.
There is an absolute consensus in the literature that inflecting demonstratives are phrasal (Kenesei, 1992, 1994; Bartos, 1999; ´E. Kiss, 2002; Bartos, 2001a; Ihsane and Pusk´as, 2001). This view is supported by four types of evidence. First, inflecting demonstratives can be used as a short answer to a question. An example of this is furnished by (40).
(40) a. Ez
‘Are you interested in this house?’
b. Nem,
‘No, (I am interested in) this one / that one / this other one / that other one.’
Second, inflecting demonstratives can be used on their own without an associated noun. This point is similar to the previous one, but while (40-b) could potentially be taken to be an elliptical DP, (41) cannot.
‘This / that ran away.’
Third, they can be modified by appositive modifiers and can be used as appositives themselves.
‘this, that is, the house’
(43) a
‘the house, this one’
Finally, Bartos (1999, 2000) argue that case markers and the possessive anaphor are phrasal affixes in Hungarian: they morphosyntactically attach to a phrase rather than to the nominal head.
The fact that these affixes appear on inflecting demonstratives lends further support to the claim that they are phrasal in nature.
There is also a near-absolute consensus that inflecting demonstratives are harboured in spec, DP (only Ihsane and Pusk´as, 2001 place them higher, but only they assume an articulated left periphery for DP). I will follow the consensus view here and place these demonstratives into spec, DP. The fact that they are always left-adjacent to the definite article and nothing can intervene between them suggests that these demonstratives are in a specifier-head configuration with the article, and thus sit in spec, DP. Finally, it is often suggested that inflecting demonstratives are of category DP themselves (Kenesei, 1992; Bartos, 1999). In Chapter 8 I will provide support for this position from compositional semantics.
In contrast to inflecting demonstratives, non-inflecting demonstratives have received far less attention in the literature: they are little noted or poorly treated in most works (except for the phenomenon of apparent article deletion in front of them, c.f. (21), (22) and (23)). Given that in (22) and (23) they follow the definite article, it is generally assumed that they are generated lower than D. There is, however, no definitive resolution as to whether they are heads or phrases. While Szabolcsi (1994) and Kenesei (2006) analyze them as heads, Kenesei (1992) and ´E. Kiss (2002) suggest that they are phrasal elements.
Szabolcsi (1994) identifies the position of non-inflecting determiners as the Det head. In her structure, DetP is situated between NumP and the phrase hosing nominative possessors. Kene-sei (2006) follows Szabolcsi’s suggestion and places non-inflecting demonstratives into the same position.
Kenesei (1992) uses a coarse-grained bi-partite NP structure (DP > NP) and places non-inflecting demonstratives into spec, NP, while ´E. Kiss (2002) in her more fine-grained decomposition base-generates them in spec, DemP. Her structure is shown in (44).
(44) [DemP e
E. Kiss (2002) further suggests that when the presence of a participial clause or a possessive does´ not prevent this, non-inflecting demonstratives raise to spec, DP, and thus their surface position is the same as that of inflecting demonstratives.
I agree with Szabolcsi (1994) that non-inflecting demonstratives are best analyzed as heads.
Szabolcsi makes this assumption without argument, but it is easy to see why it is natural to view these demonstratives as heads. Non-inflecting demonstratives cannot be used alone without an accompanying noun, which is unexpected if they are phrases.
(45) *Eme
Further, these demonstratives cannot be modified by appositives and cannot be appositives them-selves either, as (46) and (47) show.
(46) *ama
‘that, that is, the house’
(47) *a
‘the house, this one’
Hungarian has a requirement that appositive modifiers (except for clauses) share the case and number marking of the noun.9 The fact that non-inflecting demonstratives cannot bear nominal suffixes in prenominal position does not automatically entail that they cannot function as appos-itives. Numerals and adjectives cannot share the suffixes of the noun in a prenominal attributive position either, but they are perfectly capable of doing this as appositives. (48) illustrates this for numerals.
(48) a. h´arom-(*-at) three-acc
h´az-at house-acc
‘three houses’
b. h´az-at, house-acc
h´arom-*(at) three-acc
‘house, three ones’
Therefore non-inflecting demonstratives are inherently not able to support nominal suffixes.
This is expected if they are heads. Finally it seems to me that non-inflecting demonstratives do not make for good short answers either.
(49) a. Is this house the best?
b. Igen, yes,
*e this
/ /
*ezen.
this.
‘Yes, this one.’
In sum, I have not found convincing evidence for the claim that non-inflecting demonstratives are phrases. I am going to assume with Szabolcsi that they are heads, and following ´E. Kiss’
labeling, I will call their phrase DemP. (50) summarizes the proposed positions for demonstratives in the functional sequence.
(50) [DP infl.Dem [D′ def.art [DemP [Dem′ non-infl.Dem ]]]]
This entails that the surface position of inflecting and non-inflecting demonstratives cannot be the same: the former are hosted in a specifier, the latter are harboured in a head. ´E. Kiss (2002) suggests that non-inflecting demonstratives do not co-occur contiguously with the definite article because the demonstrative raises to spec, DP, and a doubly-filled comp filter type restriction prevents both the specifier and the head of DP being overtly filled. But as she suggests that inflecting demonstratives are also in spec, DP, she has to assume that these are an exception to the rule (recall that inflecting demonstratives must co-occur with the definite article). This exceptionality does not follow from anything and remains a re-statement of the facts. If, however, non-inflecting demonstratives are heads and they undergo head-movement to D, then it becomes immediately clear why they don’t co-occur with the definite article contiguously. When they can, they raise to D and this leaves no space for the article; and when they cannot raise, then the definite article can fill D.
This analysis leaves one open question. Why should participial relative clauses and nominative possessors block head movement from Dem to D? Both types of constituents are phrases.10 I assume that they are introduced in specifiers of functional projections, and the heads of these projections are phonologically empty. While the movement does not have an overt blocker, it could be argued that these functional heads are not radically empty, rather they are filled by a phonologically null element. This covert head then could block head-movement from Dem to D if it is not affixal in nature. The non-affixal nature of the phonologically empty head could not be supported from independent evidence, however.
9As we have seen, inflecting demonstratives also have to share these nominal suffixes. While this makes them look like appositive modifiers, on closer scrutiny they turn out to show a good number of properties that cannot be accommodated into the hypothesis that they are appositives. I will return to this issue in more detail in Chapter 8.
10Note that the blocking of the movement is not entirely accounted for in ´E. Kiss (2002) either. She suggests that non-restrictive relatives are adjoined to QP or DemP. If they are adjoined to DemP, they structurally intervene between the base position of the non-inflecting demonstrative and D, but it does not follow that as adjuncts they should block the movement. Non-inflecting demonstratives do not obviously share features with participial relatives (or possessors), thus it is difficult to argue that a relativized minimality violation is involved in these cases. Rather, in ´E. Kiss’ proposal participial relatives and nominative possessors look like defective interveners: elements that cannot do some job (in this case, the job of raising to D) themselves but at the same time they prevent other, lower elements from doing so.
In Section 4.2 of this chapter I will take a different tack and I will argue that non-inflecting demonstratives do not co-occur contiguously to the definite article because these demonstratives, in fact, span Dem and D. The introduction of the functional head related to relatives and possessors in between them disrupts the adjacency of Dem and D, and this forces these heads to be spelled out separately. With this, the pattern falls out as a by-product of the way lexicalization works, and add-ons like the non-affixal nature of the intervening heads are not required.
Let us now return to the surface position of demonstratives in (50), repeated here as (51), and ask whether these are also the merge-in position of these elements.
(51) [DP infl.Dem [D′ def.art [DemP [Dem′ non-infl.Dem ]]]]
There is no immediately obvious evidence internal to Hungarian that either inflecting or non-inflecting demonstratives originate lower in the structure than where they appear on the surface.
I suggest, however, that inflecting demonstratives are merged in spec, DemP and their position in spec, DP is a derived one. This is motivated by the assumed syntax-semantics mapping. In Chapter 1 I committed to the idea that each and every piece of semantics is available at exactly one projection, and that specifiers must be semantically compatible with and share the interpretation of every head they get into a local configuration with. The D projection in Hungarian has no semantic import of deicticity, the definite article that spells out D is completely neutral in this respect. The Dem projection, however, does have a semantic import of deicticity. The meaning of non-inflecting demonstratives, which sit in Dem0, and the meaning of inflecting demonstratives are not only related but in fact indistinguishable. The proximal inflecting demonstrativeez has no discernible meaning difference from either of the proximal non-inflecting demonstrativeseme, ezen, e. The same holds for the inflecting distalaz and the non-inflecting distalsazon andama. If every piece of semantics is available in one position, and specifiers share the interpretation of every head they get into a local configuration with, then inflecting demonstratives must have merged in (or passed through) spec, DemP. This is the only way they can assume a deictic interpretation.
Then spec, DP must be a derived position for them. For now I will take this to be correct on the
Then spec, DP must be a derived position for them. For now I will take this to be correct on the