A HOP TO THE LEFT: DISLOCATION, RELATIVES AND RESUMPTIVES IN ROMANI

Emily Manetta

emily.manetta@uvm.edu

​

February 28, 2023

University of South Carolina

1INTRODUCTION

Left dislocation structures involve two distinct elements referring to one and the same individual: an XP situated at the left periphery of a clause, and a proform within.

1. My aunt Vicky, she used to get bird poop on her every year. (Manetta 2007: (2))
2. kodole muršes, ame diam les mašina nevi.  ROMANI

that.obl guy.obl we give.pst him.obl car new

‘That guy, we gave him a new car.’

​

Longstanding & important questions:

• What is the origin of the XP on the left edge?
• How is the dependency between the left dislocated material and the pronoun structured?
• Do the manifestations of this construction within and across languages represent diverse distinct processes or guises of a unified phenomenon?
• What is the nature of the pronoun (“resumptive pronoun”) to which the left dislocated material is linked?

​

2

1INTRODUCTION

Accounts representing answers to these questions have fallen into two families:

• Movement: the XP found on the left edge in the surface string is a constituent of the correlate clause at some point in the derivation and is dislocated to the left edge (a prominent formulation of this proposal in Grohmann (2003), see recent discussion in e.g. den Dikken and Surányi (2017); for Hindi-Urdu correlatives it is Bhatt 2003)
• Non-movement: the XP found on the left edge is base generated in or near that position (e.g. clausal juxtaposition + ellipsis in Ott 2014) and is construed as linked to the clause-internal position by another mechanism (e.g. variable binding as in Srivastav’s (1991)/Dayal’s (1996) correlative account)

​

3

1INTRODUCTION

Where does that leave the pronoun?

• Under non-movement accounts, this is fairly straightforward – it is a regular pronoun in its criterial position
• But under movement accounts, this resumptive pronoun is mysterious. It seems to sit in a position previously occupied during the derivation. Several questions arise:

- Is the pronoun part of the numeration (that is, a regular pronoun)?

- If so, how does it share a clause-internal position with the left dislocated material?

- If not, why is it a pronoun (and not a gap, or a full copy of the NP, or some other

dedicated lexical item) which appears here?

• What if multiple options from this logical hypothesis space are realized within or across languages?

(for an elegant unpacking of these theoretical questions, see McCloskey 2017)

​

4

1INTRODUCTION

HINDI-URDU

(3) vo aurat, amy jaanti he ki john usse pyaar

that woman.nom, Amy knows aux that john her love

kaarta he

do aux

‘That woman, Amy knows John loves her’ (Chandra 2011)

(5) [jo T-shirt sale-par hai] Maya us-ko khareed-egi.

cor T-shirt sale-on is, Maya DEM-ACC buy-fut.f

‘Which T is on sale, Maya will by that.’ (Bhatt (2003)

(7) Vo kitaab [jo __ sale-par hai], achchii hai.

That book [dem sale-on is good is.

‘The book that is on sale is good’ (Bhatt 2003)

ROMANI

(4) kodole muršes, ame diam les mašina nevi. 

that.OBL guy.OBL we give.PST him.OBL car new

‘That guy, we gave him a new car.’

(6) [o murš savo avilo kaj nunta] si les mašina nevi

DEF man who come.pst to wedding is him.OBL car new.

‘The guy who came to the wedding, to him is a new car.’

(Adamou and Matras: (45b))

(8) dikhlj-om o kher [kaj bori-es-as. andar les-te]

saw-1SG DEF house which talk-2S-rem about it-LOC.

‘I saw the house you talked about’ (M & T 2016: (108))

5

​

A puzzle from Indic: are these pronouns/gaps the same?

​

1INTRODUCTION

Based on a new investigation of left dislocation in Romani I will argue here for:

• A movement approach to left dislocated material (following Grohmann 2003; Bhatt 2003); the left-dislocated XP is a constituent of the correlate clause at some point in the derivation and is dislocated to the left edge
• A unified approach to resumptive pronominals in Romani -- those found in left dislocated structures and within relative clauses – as morphophonological material inserted late in the derivation following an instance of movement (in the tradition of Engdahl 1986; Sichel 2014)
• An explanation for the contrasts we see in the properties of these pronominals and their associated XPs in left dislocated, extraposed relative/correlative, and externally headed relative clauses in Romani and Hindi-Urdu

​

6

1INTRODUCTION

This work investigates a core property of syntax: the capacity for profound mismatches in form and interpretation. In particular the larger project contributes to:

• The typology of left dislocated structures and the role of the left periphery in form/meaning mismatch
• The diverse roles of resumption (and alternation with gaps) in unbounded dependency constructions
• Bring forward a novel pattern occurring in the incredibly understudied language Romani; while much is known about left dislocation, left dislocated relative clauses, and resumption in more studied languages, it remains vital to pursue questions of variation/microvariation in lesser-studied contexts

7

1INTRODUCTION

Structure of the talk:

1. Introduction
2. Romani language: background and basics
3. Left Dislocation three ways
4. Correlatives, dislocated relatives, and their relatives in Indic
5. An account of resumptives in Romani and Hindi-Urdu
6. Conclusions and open questions

8

2 ROMANI: BACKGROUND AND BASICS

• Romani is the name for a set of varieties of language (macrolanguage) of the Central Indic group brought across Eurasia from India by the migration(s) of its speakers over 1000 ya and spoken from Wales to Anatolia; first documented in the 15^th century
• The language’s Indic origins were first formally proposed within the European scientific community in Rüdiger (1777)
• Roma are not conventionally ‘nomadic’ (Lemon 2000; Silverman 2012; Fraser 1992) or inherently migratory, and in most European countries inhabit communities of long settlement (Andrea Scala, p.c.), but in the wake of the Holocaust, forms of social regulation across Europe, and systematic persecution have engendered creative modes of mobility and flexible emplacement.

​

​

9

2 ROMANI: BACKGROUND AND BASICS

• Despite a rich body of descriptive and historical work (for reviews see Bakker and Matras 2003; University of Manchester’s Romani Project Bibliography), this state of social and political affairs contributes to the relative dearth of theoretical linguistic scholarship on Romani language even in comparison to languages and language families with many fewer speakers and with far more narrow domains of circulation.
• Work on the modern analytical syntax of Romani (beyond passing mention for exemplification or comparison) can largely be attributed to Dana McDaniel, including her dissertation (1986) and her critical contribution to the literature on wh-scope-marking structures (1989)

​

​

10

2 ROMANI: BACKGROUND AND BASICS

Unusual among Indic languages, Romani language is SVO and features obligatory wh-movement to the clause edge in matrix clauses, and overt displacement of wh-material out of embedded clauses for the purposes of achieving wide scope.

​

(9) Kas o Demìri dikhlâ?

who.obl def Demir see.pst

‘Who did Demir see?’ (McDaniel 1986:45)

(10) Kon kerola akaja kera?

who construct.pst these houses

‘Who constructed these houses?’ (Maljoku, in Eynard 2018:46)

(11) Kas o Demìri mislinol so i Arìfa dikhlâ?

who.obl def Demir think that def Arifa see.pst

‘Who does Demir think that Arifa saw?’ (McDaniel 1989:(8a))

​

11

2 ROMANI: BACKGROUND AND BASICS

On the other hand, much like other related Indic languages, Romani features a strategy of wh-scope-marking, in which the embedded wh-phrase may remain at the edge of the embedded clause, and the particle so ‘what’ may be found at the left edge of the clause in which the wh-phrase is interpreted.

​

(12) So o Demìri mislinol kas i Arìfa dikhlâ?

expl DEF Demir think who.obl def Arifa see.pst

‘Who does Demir think that Arifa saw?’ (McDaniel 1989:(8b))

​

These facts set serve as background for the exploration of left dislocated constituents in what follows.

12

3 LD THREE WAYS

In this section we explore new data from Romani concerning the nature of left dislocation and the linked pronoun.

• The full set of leftward dislocation patterns in Germanic, Italian, and Arabic is famous, fascinating, and thoroughly analyzed (Cinque 1990; Aoun and Benmamoun 1998; Aoun et al 2001; Grohmann 2003 i.a.); it behooves us to consider new Romani patterns alongside some better-understood patterns, as well as Hindi-Urdu.
• Due to constraints of space and time, suffice it to say that Romani left dislocation is most like what has traditionally been termed Left Dislocation (LD) or Contrastive Left Dislocation (CLD) in Germanic languages, and so I will constrain our focus to that operation and its properties here.
• While they share certain features, we also find that the properties of Romani LD and Germanic CLD/Hindi-Urdu LD diverge in intriguing ways.

​

​

​

​

​

13

3 LD THREE WAYS

Case Connectivity:

As in Germanic CLD, left dislocated DPs in Romani are linked to a demonstrative pronoun within the clause in Romani, and the two must match in case. This contrasts with Hindi-Urdu, in which the fronted DP must be nominative.

(13) kodole muršes/ ^✻ kodo murš, ame diam les mašina nevi. 

that.obl guy.obl/that.nom guy.nom I gave him.obl car new

‘That guy, I gave him a new car.’

(14) vo aurat/ ^✻ us aurat-ko, amy jaanti he ki john usse pyaar kartaa he.

that woman-nom/that.obl woman-acc, Amy knows aux that john her.obl love do aux

‘That woman, Amy knows John loves her’ (Chandra 2003: (8))

​

​

​

​

​

14

3 LD THREE WAYS

Position of the demonstrative pronoun:

In Germanic CLD, the demonstrative pronoun (d-pronoun) is typically found in the left periphery (den Dikken and Surányi 2017), but in Romani and in Hindi-Urdu, the pronoun is typically found in situ (e.g. in its argument position in the clause).

(15) kodole muršes ame diam les mašina nevi. 

that.obl guy.obl I gave him.obl new car

‘That guy, I gave him a new car.’

(16) vo aurat john usse pyaar kartaa he.

that woman-nom john her.obl love do aux

‘That woman, John loves her’

​

​

​

​

​

15

3 LD THREE WAYS

Embedded demonstratives:

The demonstrative pronoun may also be found in an embedded clause in Romani and in Hindi-Urdu as in Germanic; case-matching remains a requirement in Romani.

(17) kodole muršes/^✻ kodo murš, Sonja ashundas [kaj diam les mašina nevi] 

that.obl guy.obl/that.nom guy.nom Sonja heard that gave him.obl new car

‘That guy, Sonja heard we gave him a new car.’

(18) vo aurat amy jaanti he [ki john usse pyaar kartaa he]

that woman-nom Amy knows aux that john her.obl love do aux

‘That woman, Amy knows John loves her’ (Chandra 2003: (8))

​

​

​

​

​

​

16

3 LD THREE WAYS

Island sensitivity:

A-bar movement in Romani in general is island-sensitive. McDaniel (1989) shows that wh-extraction is impossible from relative and adverbial clauses:

(19) ^✻ Kas_i jane jîkas [_CP ko_j t_j kalol t_i ]

who you know who marry

‘Who do you know someone who will marry?’

(20) ^✻ Kas_i o Demiri stalno rovol [angle te kosol t_i ]

who the Demir always cry before he scolds

‘Who does Demir always cry before he scolds?’ (McDaniel 1989: (32a-b))

​

​

​

​

​

17

3 LD THREE WAYS

Island sensitivity:

The same is true for left dislocated DPs in Romani and in Hindi-Urdu.

(21) ^✻ Kodole muršes, jane jîkas [_CP ko_j t_j kalol les]

That guy you know who marry him

‘That guy – you know who will marry him’

(22) ^✻ kodole chaves, o Demiri stalno rovol [_CP angle te kosol les ]

that boy the Demir always cry before he scolds

‘That boy, Demir always cries before he scolds him’

​

​

​

​

​

18

3 LD THREE WAYS

Island sensitivity:

The same is true for left dislocated DPs in Romani and in Hindi-Urdu.

(23) ^✻ voh tasvir, Mary [yeh khabar ki John vo pasand he] jaanti thii.

That picture, Mary the fact that John DEM like aux know aux.f

‘That picture, Mary knows the fact that John likes it.

(24) ^✻ voh tasvir, Mary ro rahii he [kyuuki John-ne us-ko toR diyaa]

that picture, Mary cry prog.f aux because John-erg Dem-obl tear give.perf

‘That picture, Mary is crying because John ripped it.’ (Chandra 2003)

​

​

​

​

​

19

3 LD THREE WAYS

Romani exhibits weak crossover effects generally (quantifiers must c-command the pronouns they bind):

(25) [svako chavoro]_i kamel [peski]_i daj. 

every boy loves his mother

‘Every boy loves his mother.’

(26) ^✻ [leski]_i daj kamel [svako chavoro]_I

‘His mother loves every boy’

But not in the context of LD, suggesting the potential for the LD-d content to be interpreted at the position the pronoun occupies on the surface.

(27) [Leski_i daj]_k, [svako chavoro]_i vakerja kaj ov kamel la_k.

his mother every boy said that he loves her

Romani, Hindi-Urdu, and Germanic CLD pattern alike with respect to this diagnostic.

​

​

20

3 LD THREE WAYS

Comparison Chart: these standard diagnostics, taken together, suggest that the left dislocated DP in Romani undergoes movement. Remaining mysterious is the difference in case connectivity between the two Indic languages as well as the positions of the pronoun within the clause. What does this tell us about the resumptive pronoun in each case?

​

​

21

4CORRELATIVES AND THEIR RELATIVES

It turns out that a similar pattern of contrasts emerge in examining another iconic type of left dislocated constituents in Indic: simple correlative clauses in Hindi-Urdu (Srivastav (1991)/Dayal 1996; Bhatt 2003) .

​

Correlatives are relative clauses which appear to the left of a full CP (external to the clause), coreferent with a (possibly null) proform in argument position.

(31) [[CP1 wh ... ]_i [CP2 ...proform_i ... ]] (where CP1 is a free relative)

(32) [jo T-shirt sale-par hai] Maya us-ko khareed-egi.

COR T-shirt sale-on is, Maya DEM-ACC buy-fut.f

‘Which T-shirt is on sale, Maya will by that one.’ (Bhatt 2003)

​

Romani no longer exhibits traditional correlative structures (Matras and Tenser 2016), but Romani left dislocation shares many properties with correlative leftward movement.

​

​

22

4CORRELATIVES AND THEIR RELATIVES

Bhatt (2003) provides evidence that correlatives in Hindi-Urdu have the following properties in contrast with correlative clauses in South Slavic (Izvorski 1996) (see also Lipták 2012 for similar facts in Hungarian)

23

4CORRELATIVES AND THEIR RELATIVES

• Izvorski (1996) suggests that the contrast between South Slavic correlatives and Indic correlatives involves overt/covert distinction in movement of the demonstrative pronoun and links this to the wh-in-situ property of Hindi-Urdu.
• Bhatt (2003) counters that this can’t be right since Hindi-Urdu covert wh-movement is clause-bound (scope limited), but the demonstrative may remain embedded. The wh-fronting Indic language Kashmiri reveals that the inverse claim (wh-movement language -> overt demonstrative pronominal fronting) also does not hold. (Romani LD might make the same point)

​

24

4CORRELATIVES AND THEIR RELATIVES

Bhatt proposes a movement-based account for single-headed correlative clauses: that is, they undergo leftward movement to a position on the periphery from a base position adjoined to their linked demonstrative pronoun (which does not itself move). This demonstrative pronoun is therefore just that: a regular pronoun.

(a) (b)

​

​

​

​

25

4CORRELATIVES AND THEIR RELATIVES

Chandra (2011), proposes a not dissimilar structure for Hindi-Urdu left dislocation, the ultimately left-dislocated DP is adjoined to the DP containing the d-pronoun. This configuration means that the pronoun is a regular pronoun, and that the LDd NP will not be assigned oblique case.

​

​

​

​

​

​

​

*Chandra (2004) following Dayal (1996), early proposed a structure in which the d-pron takes the LD DP as a complement.

26

4CORRELATIVES AND THEIR RELATIVES

Less investigated than correlatives are what de Vries (2002) calls “hanging free relatives”, which consist of a relative clause in matrix clause-initial position linked with a demonstrative pronoun in the matrix clause (which must be in the left periphery). He concludes they are a form of left dislocation (in (33b)).

(33) a.Wat jij van. oma kreeg, dat heeft hij gestolen.� what you from grandma got, that has he stolen� b. Dat ding, dat heeft hij gestolen.� that thing, that has he stolen. (Dutch, de Vries: (65))

​

​

​

​

​

​

27

4CORRELATIVES AND THEIR RELATIVES

Romani speakers seem to find hanging free relatives marginally acceptable, but prefer the version in which a postnominal externally headed relative is found on the left edge.

(34) ? [[savo kaj / kon] avilo ki nunta], kodoles kamav but.

who that / who went to wedding, them love.1sg much

‘Whoever went to that wedding, I love them a lot.’

(35)[o murš savo avilo ki nunta ] kodoles kamav but

DEF man who come.pst to wedding him love.1sg much.

‘The guy who came to the wedding, I love him a lot.’

​

​

​

​

28

4CORRELATIVES AND THEIR RELATIVES

Belayev and Haug (2020: (8)) support previous proposals that hanging relatives are analyzed as left-dislocated nominals. We can add Romani hanging relatives (and their externally headed relatives) to our inventory of left dislocated constituents.

​

​

​

29

4CORRELATIVES AND THEIR RELATIVES

Interim summary (sections 3-4):

• The features of Romani left dislocated structures are consistent with a movement-based account (that the left-dislocated XP has moved from a clause-internal position to arrive at its surface position)
• Romani left dislocated structures look much like their Hindi-Urdu counterparts except for case-connectivity
• Current accounts of Hindi-Urdu left dislocation and correlative clause structures (Bhatt 2003, 2015; Chandra 2011) propose movement from a clause-internal position within the DP containing the demonstrative pronominal (which is just that, a regular pronoun)
• While Romani lacks correlatives, it exhibits left adjoined free/headed relatives linked with a clause-internal pronoun, which can receive the same analysis as LD
• So what about the resumptive pronoun in Romani?

​

​

30

5 ROMANI RESUMPTIVES

• Recall that Romani also exhibits resumptive pronouns within relative clause structures
• Unlike resumptives in English (traditionally termed intrusive (Sells 1984)), Romani resumptives are typically obligatory for relativized non-subject positions (Matras 2002; Matras and Tenser 2016; Manetta 2020)

(36) Morels are the mushrooms that every chef wants to use them/___. (JD, p.c.)

(37) Ake o chavo so arakhlûm ^✻(lesko) lil

Here def boy that found.1sg.pst 3sg.poss book

‘Here is the boy that I found his book.’ (McDaniel 1986:47)

​

Hindi-Urdu, by contrast, does not permit resumptives within relative clauses at all.

(38) Vo kitaab [jo (^✻wo) sale-par hai], achchii hai.

That book dem it sale-on is good is.

‘The book that is on sale is good’ (Bhatt 2003)

​

31

5 ROMANI RESUMPTIVES

• In recent work (Manetta 2020) I argued that that Sichel’s (2014) account of relative-clause-internal resumption in Hebrew could be extended to Romani (despite the fact that the intricacies of relative clause resumption are different in the two languages)
• Her proposal in a nutshell is that gaps and proforms are in competition for realizing the tail of a movement chain, and when a null copy is (for some reason) impossible, the proform emerges
• Because the proform is actually the morpho-phonological spellout of the bottommost copy, it will necessarily take the form dictated by the phi-features/case of the dislocated DP
• Expanding this account to resumption in left dislocation structures in Romani is fairly straightforward, and follows proposals in Grohmann (2003)/den Dikken and Surányi (2017) as well.

32

5 ROMANI RESUMPTIVES

Romani resumptives in LD and relative clause structures

33

(a) LD

(b) relative

5 ROMANI RESUMPTIVES

Looking at Hindi-Urdu and Romani LD side-by-side:

(a) Romani (b) Hindi-Urdu (Chandra 2011)

​

​

​

​

​

​

​

​

​

…and this provides an explanation for the contrast in case connectivity in LD structures between Romani and Hindi-Urdu. Both dislocated phrases arrive at their surface position via movement; the pronominals originate in different ways. With this, resumptives in the two Indic languages fall into place.

​

34

6 CONCLUSION AND OPEN QUESTIONS

Our findings:

• The properties of Romani left dislocation are consistent with an approach in which it is derived via A-bar movement of the left dislocated DP from its point of original merge within the clause to the clause edge
• The overt pronoun pronounced in that original position of the DP is the morphophonological spellout of the tail of the movement chain (the bottommost copy), and for this reason takes a form consistent with the phi/case-features of the displaced DP
• We can derive both simple left-dislocated DPs and left-dislocated externally headed relatives via these mechanisms
• We can also thus establish a unified analysis for two different instantiations of resumptive pronouns in Romani
• Adopting Chandra’s (2011) account of Hindi-Urdu left dislocation, in which the LD DP is also derived by movement, but the clause-internal pronoun is present in the numeration and the LDd DP is adjoined to it further allows us to explain the contrast in case connectivity

35

6 CONCLUSIONS AND OPEN QUESTIONS

36

​

Mere inspection of the string is insufficient to identify the distinct nature of these pronominals and their distinct relations to the moved element.

​

HINDI-URDU

(3) vo aurat, amy jaanti he ki john usse pyaar kartaa he.

woman-nom, Amy knows aux that john her love do aux

‘That woman, Amy knows John loves her’ (Chandra 2004)

(5) [jo T-shirt sale-par hai] Maya us-ko khareed-egi.

cor T-shirt sale-on is, Maya DEM-ACC buy-fut.f

‘Which T is on sale, Maya will by that.’ (Bhatt 2003)

(7) Kitaab [jo __ sale-par hai], achchii hai.

book [dem sale-on is good is.

‘The book that is on sale is good’ (Bhatt 2003)

ROMANI

(4) kodole muršes, ame diam les mašina nevi. 

that.OBL guy.OBL we give.PST him.OBL car new

‘That guy, we gave him a new car.’

(6) o murš savo avilo ki nunta si les mašina nevi

DEF man who come.pst to wedding is him.OBL car new.

‘The guy who came to the wedding, to him is a new car.’

(Adamou and Matras: (45b))

(8) dikhlj-om o kher kaj bori-es-as. andar les-te

saw-1SG DEF house which talk-2S-rem about him-LOC.

(M&T 2016: (108))

6 CONCLUSIONS AND OPEN QUESTIONS

Open questions:

• What conditions the availability of these very heavy DPs in a language like Hindi-Urdu (with adjoined clausal or nominal constituents)? Is there empirical evidence we could bring to bear on their finer structure?
• What are the conditioning factors for crosslinguistic variation in the presence of gaps vs. pronouns in the spellout of the bottommost copy of a movement chain, and can they be derived from some more general principles? (note that Sichel (2014) does not make particular claims about this)
• Are ‘high’ d-pronouns spellouts of mid-positions in the LD movement chains (Grohmann 2003)? What conditions variation between low/high?
• If resumptive pronouns are what the tail of a movement chain sounds like when it isn’t a gap, then are wh-scope markers something like what the top of a movement chain sounds like when they aren’t realized by the full XP? To fully understand movement-derived form/interpretation mismatches, we should probe the nature of these functional morphophonological spellouts of otherwise invisible syntactic dependencies.

​

37

ACKNOWLEDGEMENTS

For their patience, kindness, and unflagging enthusiasm for discussing Romani language, I am deeply indebted to Ian Hancock (Vlax/other varieties), Slavka Radenez (Bulgarian Xoraxané) and the Radenez family, Sacsha Zanko and the Zanko family, Santino Spinelli, and all of those involved with the MuCEM Barvalo council of experts. For our many and varied conversations about Hindi-Urdu, I’m especially grateful to Rajesh Bhatt and Khushboo Jain.

​

38

REFERENCES

Adamou, Evangelia and Yaron Matras. 2016. Romani Syntactic Typology. In Yaron Matras; Anton Tenser (eds). The Palgrave Handbook of Romani Language and Linguistics, Springer, pp.187-227.

Aoun, J., L. Choueiri, and N. Hornstein. 2001. Resumption, movement, and derivational economy. Linguistic Inquiry 32, 371-403.

​

Chandra, Pritha 2011. Left dislocation in Hindi-Urdu: Movement or construal? In: Omkar Nath Koul (ed.): Indo-Aryan linguistics, 123–34. CIIL, Manasagangothri.

Dayal, Veneeta. 1996. Locality in Wh-Quantification: Questions and Relative Clauses in

Hindi, Studies in Linguistics and Philosophy, No. 62, Kluwer, Dordrecht.

den Dikken, Marcel and Balázs Surányi. 2017. Contrasting Contrastive Left-Dislocation

Explications. Linguistic Inquiry 48(4): 543–584.

Belyaev, Oleg and Dag Haug. The genesis and typology of correlatives. Language, vol. 96 no. 4, 2020, p. 874-907.

Bhatt, Rajesh. 2003. Locality in Correlatives. Natural Language and Linguistic Theory 21,

485-541.

Grohmann, Kleanthes. 2003. Prolific domains: On the anti-locality of movement dependencies.

Amsterdam: John Benjamins.

​

Izvorski, Roumyana. 1996. 'The Syntax and Semantics of Correlative Proforms', in K. Kusumoto(ed.), Proceedingsof NELS26, GLSA Amherst, Massachusetts,pp. 133-147.

​

​

39

REFERENCES

40

Matras, Yaron and Tenser, Anton. "Complementizers in Romani". Complementizer Semantics in European Languages, edited by Kasper Boye and Petar Kehayov, Berlin, Boston: De Gruyter Mouton, 2016, pp. 341-376.

McCloskey J. 2017. Resumption. In the Wiley Blackwell Companion to Syntax, ed. M Everaert, HC van Riemsdijk, pp. 1–30. Hoboken, NJ: Wiley.

McDaniel, Dana. 1986. Conditions on Wh-chains. Ph.D. dissertation, City University of New York

McDaniel, Dana. 1989. Partial and multiple WH-movement. Natural Theory 7:565–604. Language and Linguistic Theory. 7: 565-604.

Srivastav, Veneeta.1991. The Syntax and Semantics of Correlatives. Natural Language

and Linguistic Theory 9, 637-686.

Sichel, Ivy. 2014. Resumptive Pronouns and Competition. Linguistic Inquiry. 45 (4): 655–693.

Vries, Mark de. 2002. The syntax of relativization. Doctoral dissertation, University of Amsterdam.

​

​