1 Background

Refining Greenberg’s (1963) Universal 20, Cinque (2005) and Dryer (2009) show that the typological distribution of the order of the four elements Demonstrative, Numeral, Adjective, and Noun is extremely uneven. There are 24 possible permutations of these elements, but only 5 of them constitute the dominant orders across languages (1i). Another 9 orders are attested, but are less frequent (1ii). The final 10 word orders are either unattested or extremely rare, accounting altogether for less than 1% of languages (1iii).

(1)

graphic

Cinque’s syntactic account of these generalizations is based on the assumption that the base order is as shown in (2), and on a constraint on movement.

(2)

  • a.

    [WP Dem [XP Num [YP Adj [NP N]]]]

  • b.

    Cinque’s Constraint

    Only movement of the overt NP and phrases containing it is allowed, with different movements having different costs.

In this proposal, neither head movement nor the movement of any constituent that does not contain the NP is allowed. Different types of permitted movements are assigned different costs, expressed as degrees of markedness. Roll-up movement with pied-piping, like (3a), is unmarked. So if NP moves above Adj, then the [YP [NP N] Adj tNP] moves above Num, and so on, yielding a full mirror order, there is no cost to the computational system. If, on the other hand, the NP moves all the way up to a higher specifier without pied-piping, as in (3b), the movement is somewhat marked, so word orders requiring this movement will be less frequent. Moving an XP containing the NP, but without moving the NP up to XP’s specifier first, as in (3c), is more marked, so word orders requiring it are even less frequent. Finally, splitting the NP out of a moved constituent and moving it to a higher position as in (3d) is highly marked, so word orders requiring it will be infrequent.

(3)

graphic

Word orders requiring any other type of movement—namely, head movement or movement of a phrase not containing the overt NP—are underivable, so they are expected to be completely unattested. Typologically, that expectation is verified (Cinque 2005).

2 Examining Intralinguistic Variation: Lebanese Arabic

While Cinque’s proposal is only intended to explain the typological generalization, we test the ability of this proposal to predict the grammaticality of the 24 possible permutations of Demonstrative, Numeral, Adjective, and Noun in the DP in Lebanese Arabic. As it turns out, 11 of these 24 orders are grammatical in Lebanese Arabic, when tested using nonintersective adjectives, and 13 are ungrammatical (see the online appendix at http://archive-ouverte.unige.ch/unige:86360).

We start with the ungrammatical orders. Of the 13 ungrammatical orders, 8 require moving a phrase not containing the NP and are therefore correctly predicted to be ungrammatical. These are listed in (4).

(4)

graphic

The other 5 orders can be derived in Cinque’s system and are attested typologically, but they are ungrammatical in Lebanese Arabic. In all these orders, listed in (5), the adjective precedes the noun.

(5)

graphic

We explain their ungrammaticality by the proposal—motivated in Shlonsky 2012—that agreement between a noun and a modifier (at least in Arabic and Hebrew) must involve NP-movement above the agreeing modifier. Since adjectives agree with nouns in Lebanese Arabic, they cannot precede the noun. So these orders cannot be grammatical.

Moving now to the 11 grammatical orders, we note that 9 of them are derivable by moving only the overt NP or phrases containing it (i.e., respecting Cinque’s Constraint). These 9 constitute a subset of the 14 word orders predicted by Cinque to be typologically frequent (see (1i) and (1ii)). They are listed in (6).1

(6)

graphic

This leaves 2 orders that are grammatical, but not derivable. These orders require violating Cinque’s Constraint.

(7)

  • a.

    Num-N-Dem-A

  • b.

    N-Num-Dem-A

Rather than abandoning Cinque’s Constraint and exploring other options to rule out the impossible orders, we will show that the constraint is valid and that the problematic orders in (6) are due to other conditions. To do so, we turn to orders involving a second adjective and demonstrate that Cinque’s constraint makes the correct predictions concerning the acceptable and inacceptable orders. We show that the grammatical orders that are not predicted by Cinque in DPs with two adjectives receive the same explanation as the orders in (6).

3 Extending the Paradigm: Adding an Adjective

We examine the grammaticality of the different word orders of DPs containing five items: a demonstrative, a numeral, an intersective adjective (

graphic
awiil/
graphic
waal ‘tall.MSG/MPL’, lebneeni/lebneeniyiin ‘Lebanese.MSG/MPL’), a nonintersective adjective (seebeʔ/seebʔiin ‘former.MSG/MPL’), and a noun. (For examples with ‘tall’, see the online appendix.) There are 5! (= 120) possible permutations of these elements, and 117 of them are not problematic: 80 orders are ungrammatical and are ruled out because they have at least one (agreeing) adjective preceding the noun (Shlonsky 2012). Another 9 orders are highly marked and are ruled out because the adjectives are in the wrong order with respect to each other (e.g., a nonintersective adjective is incorrectly placed farther from the noun than the intersective adjective, an order that is highly marked and difficult to interpret2 also see Cinque 2010). A further 14 orders are ungrammatical and are ruled out by Cinque’s Constraint because they require moving phrases not containing the overt NP.

(8)

graphic

Finally, 14 orders are good and can be derived by moving only phrases containing the overt NP.

(9)

graphic

Of the 120 possible orders, there remain 3 problematic word orders that are grammatical in Lebanese Arabic, but that Cinque’s Constraint, barring the movement of constituents not containing the overt NP, rules out. These are listed in (10) and exemplified in (11).3 Crucially, the adjective following the demonstrative is not a reduced relative, as it can be nonintersective (“low” in Cinque’s (2010) terms), like ‘former’.

(10)

  • a.

    Num-N-A-Dem-A

  • b.

    N-Num-A-Dem-A

  • c.

    N-A-Num-Dem-A

(11)

  • a.

    ✓t-tlat mhandsiin l-madaniyyiin hol

    the-three4 engineers the-civil these

    l-lebneeniyiin/s-seebʔiin

    the-Lebanese/the-former

  • b.

    ✓/?l-mhandsiin t-tleeteh l-madaniyyiin hol

    the-engineers the-three the-civil these

    l-lebneeniyiin/s-seebʔiin

    the-Lebanese/the-former

  • c.

    ✓l-mhandsiin l-madaniyyiin t-tleeteh hol

    the-engineers the-civil the-three these

    l-lebneeniyiin/s-seebʔiin

    the-Lebanese/the-former

    ‘these three Lebanese former civil engineers’

Deriving the order Num-N-A-Dem-A in (11a) from the base order Dem-Num-A-A-N requires moving A out of the constituent [YP Num [YP [NP N] Adja [YP2 Adjb tNP ]]] prior to moving YP above Dem. This step in the derivation, illustrated in (12c), is barred by Cinque’s Constraint. The moved element in each step of the derivation in (12) is boldfaced.

(12)

  • a.

    [XP Dem [YP Num [ZP Adja [ZP2 Adjb [NP Noun]]]]]

  • b.

    [XP Dem [YP Num [ZP Adja [ZP2[NPNoun] Adjb tNP]]]]]

  • c.

    [XP Dem [Adja [YP Num [ZP tAdja [ZP2 [NP Noun] Adjb tNP ]]]]]

  • d.

    [XP[YPNum [ZPtAdja[ZP2[NPNoun] AdjbtNP]]] Dem [Adja tYP]]

Similarly, deriving N-Num-A-Dem-A in (11b) from Dem-Num-A- A-N requires the operation in (13c), which violates Cinque’s Constraint. And deriving the order N-A-Num-Dem-A requires the barred movement in (14d).

(13)

  • a.

    [XP Dem [YP Num [ZP Adja [ZP2 Adjb [NP Noun]]]]]

  • b.

    [XP Dem [YP[NPNoun] Num [ZP Adja [ZP2 Adjb tNP ]]]]

  • c.

    [XP Dem [Adja [YP [NP Noun] Num [ZPtAdja [ZP2 Adjb tNP ]]]]]

  • d.

    [XP[YP[NPNoun] Num [ZPtAdja[ZP2AdjbtNP]]] Dem [Adja tYP]]

(14)

  • a.

    [XP Dem [YP Num [ZP Adja [ZP2 Adjb [NP Noun]]]]]

  • b.

    [XP Dem [YP Num [ZP Adja [ZP2[NPNoun] Adjb tNP ]]]]

  • c.

    [XP Dem [YP[ZP2[NPNoun] AdjbtNP] Num [ZP Adja tZP2 ]]]

  • d.

    [XP Dem [Adja [YP [ZP2 [NP Noun] Adjb tNP] Num [ZP tAdja tZP2 ]]]]

  • e.

    [XP[YP[ZP2[NPNoun] AdjbtNP] Num [ZPtAdjatZP2]] Dem [Adja] tYP]

Like (6a–b), (11a–c) require violating Cinque’s Constraint. Simply abandoning this constraint, however would overgenerate: if the movement of constituents not containing the overt NP were allowed, the desirable results of Cinque’s (2005) proposal would be lost (but see Steddy and Samek-Lodovici 2011 for a possible alternative), and we would no longer be able to rule out the typologically unattested orders that are also ungrammatical in Lebanese Arabic (see (4)). We would also need to find other explanations for ruling out the 14 ungrammatical orders in (8) (with two adjectives), which are straightforwardly ruled out by Cinque’s Constraint. We therefore propose maintaining Cinque’s Constraint and changing the base order.

4 Proposal: Changing the Base Order

There are two ways to derive (11) by changing the base order. The first is to posit an optional additional position for Dem, lower in the structure, as illustrated in (15), and to assume that either Dem1 or Dem2 can be active. Using the low Dem position would derive (11) through permitted movements, as illustrated in (16)–(18).

(15) [WP Dem1 [XP Num [YP Adj [ZP Dem2 [YP2 Adj [NP Noun]]]]]]

(16)

graphic

(17)

graphic

(18)

graphic

Such an analysis would find its place among those proposed by Brugè (1996, 2002), Guardiano (2010), and Roberts (2011), who provide crosslinguistic evidence for a low Dem. It is also supported by the optional cooccurrence of a pre- and a postnominal demonstrative in Lebanese Arabic (19). WALS also lists 17 languages in which this cooccurrence is obligatory (Dryer 2013). However, as a Linguistic Inquiry reviewer correctly points out, a low Dem—configured below Num and above Adj—overgenerates, yielding for example the unacceptable order in (1iiie). Abels and Neeleman (2012) provide substantial arguments against a low demonstrative. Also relevant to this discussion are Adger 2013 and Belk and Neeleman 2017.

(19)

  • ha l-mhandsiin hol

  • DEM the-engineers DEM.PL

  • ‘these engineers’ (not redundant)

Another alternative is to retain a unique position for demonstratives, as Cinque does, but posit a second, lower position for numerals, as in (20). The movements in (21)-(23) would derive the orders in (11).

(20) [WP Dem [XPNum [YP Adj [ZPNum [YP2 Adj [NP Noun]]]]]]

(21)

graphic

(22)

graphic

(23)

graphic

Both options derive the three desirable word orders. A low demonstrative finds support in the literature but leads to overgeneration. A low numeral position does not lead to language-internal overgeneration. In addition, there are multiple reasons militating for a low numeral position, to which we now turn.

4.1 Reason 1: Case Marking

Standard Arabic DP-internal case patterns militate in favor of a low numeral. In the absence of numerals, case is manifested on the head noun, and adjectives show concord with the noun in case (24). When a DP contains a numeral, the numeral bears the DP case, and the noun appears in the genitive (25).

(24)

  • ʃaahada al-mudiir-u as-saabiq-u barnaamaj-an

  • watched the-boss-NOM the-former-NOM program-ACC

  • ‘The former boss watched a program.’

(25)

  • ʃaahada θalaaθat-u mudaraa-in xamsat-a

  • watched three-NOM boss.PL-GEN five-ACC

  • baraamij-in

  • program.PL-GEN

  • ‘Three bosses watched five programs.’

When an adjective is added to an example like (25), it can match either the noun or the numeral in case (26).

(26)

  • a.

    wa

    graphic
    ala [θalaaθat-u [[muhandisiin] saabiqiin]]

    arrived.3MSG three-NOM engineer.GEN former.GEN

  • b.

    wa

    graphic
    ala [[θalaaθat-u [muhandisiin]] saabiquun]

    arrived.3MSG three-NOM engineer.GEN former.NOM

    ‘Three former engineers arrived.’

In the presence of more than one adjective, one adjective can match the noun’s case and another adjective can match the numeral’s case. But if the adjective closest to the noun matches the numeral in case, the second adjective must too.5

(27)

  • a.

    wa

    graphic
    ala [[θalaaθat-u [[muhandisiin] saabiqiin]]

    arrived.3MSG three-NOM engineer.GEN former.GEN

    lubnaaniyuun]

    Lebanese.NOM

  • b.

    *wa

    graphic
    ala θalaaθat-u muhandisiin saabiquun

    arrived.3MSG three-NOM engineer.GEN former.NOM

    lubnaaniyiin

    Lebanese.GEN

  • c.

    wa

    graphic
    ala [[[θalaaθat-u muhandisiin] saabiquun]

    arrived.3MSG three-NOM engineer.GEN former.NOM

    lubnaaniyuun]

    Lebanese.NOM

    ‘Three Lebanese former engineers arrived.’

Let us assume that adjectives show case concord with the case-marked complement they immediately c-command prior to any DP-internal movement. The variation in (26a–b) can be attributed to different positions of the numeral. When the base order is [Num [Adj* [N]]] (the Kleene star indicates the possible occurrence of multiple adjectives), the adjective c-commands and hence manifests concord with the noun, which is genitive (26a). NP-movement over ‘former’ yields the observed order. When the numeral is merged lower and the base order is [Adj* [Num [N]]], the adjective manifests concord with the nominative-bearing Num (26b), and the surface order is derived by moving Num + NP above the adjectives, with pied-piping. In (27a), the numeral is base-generated in a position that is straddled by the adjectives, with ‘former’ immediately modifying ‘engineer’ and ‘Lebanese’ modifying ‘three former engineers’. The surface order is attained by moving NP above ‘former’, and then moving [NumP three [XP[NP engineer] former]] above ‘Lebanese’. In (27c), the numeral is base-generated in a position under both adjectives; hence, both adjectives modify [NumP three [NP engineer]] and agree with it in case. Finally, the ungrammaticality of (26b) shows that case is not arbitrarily assigned: an adjective manifesting concord with the genitive noun cannot be separated from the noun by the numeral (and adjectives in concord with the numeral). An alternative, at odds with typological generalizations, is that numerals are adjectives and can intersperse freely among different adjectives. But numerals assign or determine case on their complement, so they cannot be adjectives.

4.2 Reason 2: Cooccurrence

Like demonstratives, a numeral can occur in Lebanese Arabic before or after the noun, and two numerals may cooccur, the result being semantically odd, but grammatical. The examples in (27a–b) are respectively redundant and contradictory, suggesting that two numerals are indeed at play.

(28)

  • a.

    t-tlatt wleed t-tleeteh

    the-three kids the-three

    ‘the three kids’ (redundant)

  • b.

    l-xams qaarraat l-arba henneh tleeteh . . .

    the-five continents the-four they three . . .

    ‘The five continents, which are four, are three . . .’

    (contradictory)

More specifically: Assuming that there are, as we claim, two positions for numerals, then each instance of the numeral is a true numeral, so the occurrence of both is expected to be redundant. If there is only one position for a demonstrative, the occurrence of two demonstratives cannot be two true occurrences of demonstratives, and must instead be a case of concord, so the cooccurrence is not expected to be redundant. And indeed, (19), which contains two demonstratives, does not express a redundancy.

4.3 Reason 3: Agreement

The two positions for numerals are motivated in Ouwayda’s (2013, 2014) work on transdecimal numerals in Lebanese Arabic. As Ouwayda observes, nouns occur in the singular when they follow transdecimal numerals. Adjectives, however, can either agree with the noun and be singular-marked (29a), or agree with the whole noun phrase and be plural-marked (29b).6

(29)

  • a.

    tleetiin mudiir seebeʔ

    thirty director former.

    graphic

    ‘thirty former directors’

  • b.

    tleetiin mudiir seebʔ-iin

    thirty director former-PL

    ‘thirty former directors’

Assuming that adjectives agree directly with the noun they modify again suggests that adjectives merge below the numeral when they are singular, and above the numeral when they are plural. Supporting this idea, Ouwayda shows that when the adjective agrees with the noun, the only interpretation is distributive. When the adjective agrees with the numeral + noun, a collective interpretation becomes available.7

(30)

  • a.

    tleetiin walad mnazzam

    thirty child organized.

    graphic

    ‘thirty organized kids’

  • b.

    tleetiin walad mnazzam-iin

    thirty child organized-PL

    ‘an organized group of thirty kids’

    or ‘thirty organized kids’

Since the lower position for numerals is semantically motivated, with a clear semantic difference between (30a) and (30b), we argue that positing such a position does not vitiate the force of the predictions engendered by Cinque’s proposal. More specifically, since the order in (30a) is the less-marked one, it is safe to assume the equivalent of (30a) will be the dominant order in any given language.

If we assume, following Ouwayda (2013, 2014), that (30a) involves a base structure in which the numeral initially merges higher than the adjective, and (30b) involves the numeral merging below the adjective, it makes sense to assume that for any given language, the base order Dem-Num-A-N will be the dominant one, and any other possible base orders will be less frequent. Therefore, the typological data, which only reflect the dominant order in each language, will not be affected by the option of a lower Merge position for numerals or demonstratives (but see Abels and Neeleman 2012 for the suggestion that a low Num position could explain some of the typological facts as well).

Examining again the ungrammatical orders that are underivable from the order Dem-Num-Adj-N (and setting aside the orders that are derivable from the base order Dem-Num-A-N, as they are not contingent on the low Num base order), we note that they are all ruled out either because they are also not derivable from the base order Dem- A-Num-N or because they violate Shlonsky’s (2012) generalization that DPs in Semitic languages must involve NP-movement above agreeing modifiers.

(31)

  • a.

    Num-N-Dem-A

    (is grammatical in Lebanese Arabic; cf. (7))

  • b.

    Dem-A-Num-N

    (A precedes N)

  • c.

    Num-Dem-A-N

    (requires Num to move above Dem alone)

  • d.

    N-Num-Dem-A

    (is grammatical in Lebanese Arabic; cf. (7))

  • e.

    Num-Dem-N-A

    (requires Num to move above Dem alone)

  • f.

    A-Dem-Num-N

    (requires A to move above Dem alone)

    g. A-Dem-N-Num

    (requires A to move above Dem alone)

    h. Num-A-Dem-N

    (requires Num + A to move above Dem without N)

    i. A-Num-Dem-N

    (requires A + Num to move above Dem alone)

    j. A-Num-N-Dem

    (A precedes N)

Similarly, for five-element DPs, the orders not excluded by Shlonsky’s requirement are these:

(32)

  • a.

    Num-Dem-N-Ani-Ai

    (requires Num to move above Dem alone)

  • b.

    Num-N-A-Dem-A

    (is grammatical in Lebanese Arabic; cf. (11))

  • c.

    N-Num-A-Dem-A

    (is grammatical in Lebanese Arabic; cf. (11))

5 Conclusion

In this squib, we made novel observations showing that Cinque’s (2005) phrasal-movement proposal correctly predicts the grammaticality of 22 of the 24 different word orders in Lebanese Arabic DPs containing a demonstrative, a numeral, an adjective, and a noun. We showed that there are 2 recalcitrant orders that are grammatical but cannot be derived given Cinque’s Constraint, and that adding a second adjective confirms the recalcitrant orders and yields yet more grammatical orders that Cinque’s (2005) proposal cannot derive. We illustrated how assuming an optional additional Merge position either for demonstratives or for numerals derives these desirable orders, and we argued for the low numeral position. Since the low numeral position is semantically marked, we note that a structure containing a low numeral will never be the dominant DP structure in a language. Thus, while it is important to acknowledge the need for this position, Cinque’s typological proposal, which only concerns dominant orders crosslinguistically, need not be extended. We therefore conclude that these assumptions do not compromise the typological predictions of the phrasal-movement proposal.

Notes

1 The crossed-out orders are those that are derivable and typologically frequent, but ungrammatical in Lebanese Arabic (see (5)).

2 For independent reasons that we do not go into, NP must pied-pipe a lower adjective if it moves above a higher adjective (Shlonsky 2004), though it can strand an adjective if it moves above a numeral. This rules out high (e.g., intersective) adjectives appearing closer to the noun after movement than low (e.g., nonintersective) adjectives, thus enforcing the “mirror order” for adjectives in Lebanese Arabic. Clearly, this is a language-specific restriction, and one predicts the existence of grammars similar to Lebanese Arabic, in which adjective pied-piping is not enforced.

3 Modifiers in Lebanese Arabic all agree with the noun in definiteness. Cinque (2010) suggests that determiner agreement indicates relative clause modification. This is not likely in Lebanese Arabic, because adjectives are direct modifiers (e.g., ‘former’) and are therefore very “low” in the DP, too low for Cinque-type relative clauses. As a reviewer points out, it is conceivable that the head noun is modified by a reduced relative clause with a null predicative noun linked to the head noun, and the low adjective ‘former’ modifying that null noun (Dimitrova-Vulchanova and Giusti 1998). However, it is not clear that the interpretation would turn out as desired: a [[former

graphic
engineer] dancer] would need to be someone who is still a dancer in some sense—which is not true, for instance, of a former dancer who not only no longer dances but also no longer has the skill or ability to dance.

4 In Lebanese Arabic, prenominal numerals appear in a reduced form, and postnominal numerals appear in full form. There is no agreement on numerals in either position.

5 Note that the low Num position is only higher than “very low” adjectives that form a sort of compound with the noun, like ‘civil’ in ‘civil engineers’. Like intersective adjectives, nonintersective adjectives like ‘former’ can merge below or above the low numeral. This is possible because nonintersective adjectives need not directly take N as their argument: in cases of stacked nonintersective adjectives, like ‘the former beautiful dancer’, ‘former’ takes as an argument ‘beautiful dancer’, which is not the head noun alone, resulting in the interpretation ‘former dancer who was beautiful as a dancer’. Similarly, in (27c), ‘former’ modifies ‘three engineers’. In this case, the DP denotes a former cohort of three engineers. This is not possible for “very low” adjectives like madani ‘civil’ in mhandes madani ‘civil engineer’; this can be seen from the impossibility of the nominative case on this adjective when it is modified by a numeral (see Asarina 2009, Ouwayda 2013, Pesetsky 2013, and previously circulated manuscripts by the same authors for similar discussion on gender marking in Russian and number marking in Lebanese Arabic)..

(i) wasala θalaaθat-u muhandisiin madaniyiin/*madaniyuun

arrived three-NOM engineers.GEN civil.GEN /*civil.NOM

6 A reviewer wonders whether (29) can be explained either by multiple Num positions or by allowing adjectives to merge higher. But as Ouwayda (2014) points out, the multiple Num positions are visible even in the absence of an adjective, with the verb agreeing with the DP either in the singular, corresponding to a high Merge position for Num, or in the plural, corresponding to a low Merge position for Num. Ouwayda provides semantic and syntactic arguments for this position.

(i) tleetiin walad wesel/wesel-u

thirty child arrived.

graphic
/ arrived-PL

‘Thirty children arrived.’

Similarly, adjective fronting related to information structure, as in Cinque 2010, fails to explain either the case-marking facts in (27) or the number agreement facts in (29) and (30), as the marking of case and of number is independent of any information-structural fronting.

7 A collective interpretation is also available for (29b), where it denotes the members of a former cycle of directors (e.g., if directors are hired into office in cycles, like parliament members). We use ‘organized’ in (30) to show the clearer truth-conditional distinction between the collective and the distributive interpretation (which may be less clear with ‘former’, where the collective interpretation also entails the distributive interpretation): while all members of a former cycle of directors are also by entailment all individually former directors as well, not all members of an organized group are necessarily individually organized themselves.

References

Abels,
Klaus
, and
Ad
Neeleman
.
2012
.
Linear asymmetries and the LCA
.
Syntax
15
:
25
74
.
Adger,
David
.
2013
.
A syntax of substance
.
Cambridge, MA
:
MIT Press
.
Asarina,
Alya
.
2009
.
Gender and adjective agreement in Russian
.
Handout from the 4th annual meeting of the Slavic Linguistics Society, University of Zadar, Croatia
.
Belk,
Zoë
, and
Ad
Neeleman
.
2017
.
AP Adjacency as a precedence constraint
.
Linguistic Inquiry
48
:
1
45
.
Brugè,
Laura
.
1996
.
Demonstrative movement in Spanish: A comparative approach
.
University of Venice Working Papers in Linguistics
6
:
1
53
.
Brugè,
Laura
.
2002
.
The positions of demonstratives in the extended nominal projections
. In
Functional structure in DP and IP: The cartography of syntactic structures
,
vol. 1
, ed. by
Guglielmo
Cinque
,
15
53
.
New York
:
Oxford University Press
.
Cinque,
Guglielmo
.
2005
.
Deriving Greenberg’s Universal 20 and its exceptions
.
Linguistic Inquiry
36
:
315
332
.
Cinque,
Guglielmo
.
2010
.
The syntax of adjectives: A comparative study
.
Cambridge, MA
:
MIT Press
.
Dimitrova-Vulchanova,
Mila
, and
Giuliana
Giusti
.
1998
.
Fragments of Balkan nominal structure
.
University of Venice Working Papers in Linguistics
8
:
141
172
.
Dryer,
Matthew
.
2009
.
On the order of demonstrative, numeral, adjective, and noun: An alternative to Cinque
.
Handout from Conference on Theoretical Approaches to Disharmonic Word Orders
. Available at http://attach.matita.net/caterinamauri/sitovecchio/1898313034/cinqueH09.pdf.
Dryer,
Matthew
.
2013
.
Order of Demonstrative and Noun
. In
The world atlas of language structures online
, ed. by
Matthew S.
Dryer
and
Martin
Haspelmath
.
Leipzig
:
Max Planck Institute for Evolutionary Anthropology
. Available at http://wals.info/chapter/88. Accessed on 12 June 2014.
Greenberg,
Joseph
.
1963
.
Some universals of grammar with particular reference to the order of meaningful elements
. In
Universals of language
, ed. by
Joseph
Greenberg
,
58
90
.
Cambridge, MA
:
MIT Press
.
Guardiano,
Cristina
.
2010
.
Demonstratives and the structure of the DP: Crosslinguistic remarks
.
Paper presented at the Workshop on Disharmony in Nominals at the annual meeting of the Linguistics Association of Great Britain, Leeds University, September 2010
.
Ouwayda,
Sarah
.
2013
.
Where plurality is: Agreement and DP structure
. In
NELS 42
, ed. by
Stefan
Keine
and
Shayne
Sloggett
,
2
:
423
436
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Ouwayda,
Sarah
.
2014
.
Where number lies: Plural marking, numerals, and the collective-distributive distinction
.
Doctoral dissertation, University of Southern California, Los Angeles
.
Pesetsky,
David
.
2013
.
Russian case morphology and the syntactic categories
.
Cambridge, MA
:
MIT Press
.
Roberts,
Ian
.
2011
.
FOFC in DP: Universal 20 and the nature of demonstratives
. Available at http://ling.auf.net/lingbuzz/001502.
Shlonsky,
Ur
.
2004
.
The form of Semitic noun phrases
.
Lingua
114
:
1465
1526
.
Shlonsky,
Ur
.
2012
.
On some properties of nominals in Hebrew and Arabic, the construct state and the mechanisms of AGREE and MOVE
.
Italian Journal of Linguistics
24
:
267
286
.
Steddy,
Sam
, and
Vieri
Samek-Lodovici
.
2011
.
On the ungrammaticality of remnant movement in the derivation of Greenberg’s Universal 20
.
Linguistic Inquiry
42
:
445
469
.