16di,   PDF
Negation Topicalisation 31

Marcel den Dikken

Negation Topicalisation*

0 Introduction Particles which indicate the polarity of a proposition have the entire proposition in their scope. Yet they typically occur clause-medially, not in sentence-initial position. Consider the case of Dutch, the focus of this paper: (1)

a. wel zou hij haar gekust hebben AFF would he her kissed have ‘he IS said to have kissed her’ b. * niet zou hij haar gekust hebben NEG would he her kissed have

(1a) is grammatical but forces a contrastive reading: something in the sentence must serve as a contrastive focus. Thus, (1a) works in a context in which it is preceded by a statement such as hij heeft haar niet betast ‘he has not fondled her’ (with contrastive focus in (1a) on gekust) or hij heeft hem niet gekust ‘he has not kissed him’ (with contrastive focus in (1a) on haar). Sentence-initial wel cannot be used for the purpose of simple denial of a statement with the opposite polarity: in the context of (2A), (2Bʹ) is impossible. But even on a contrastive reading, (1b) is unacceptable. (2)

A:

hij heeft haar niet gekust he has her not kissed

B:

jawel, hij heeft haar wél gekust! yes.AFF he has her AFF kissed ‘he did kiss her’

Bʹ: *jawel, wél heeft hij haar gekust! yes.AFF AFF is he been there

Versions of this paper were presented at the Dutch department of Eötvös Loránd University, Brussels University, Göttingen University, and the first Budapest–Potsdam–Lund Linguistics Colloquium at the Research Institute for Linguistics of the Hungarian Academy of Sciences. I thank the audiences present on these occasions as well as a reviewer for the The Even Yearbook for their very helpful feedback. This paper is part of a larger manuscript (available from the author upon request) in which contrastive negation topicalisation is placed in the wider context of the typology of negation and affirmation constructions, and in which the contrast between (1a) and (1b) is explained. In the present piece, I can only address (1b). The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 32

It is not strictly impossible, however, to place the negation particle niet in sentence-initial position by itself: sentences such as (3a,b) are well-formed (though typically restricted to elevated styles such as legal documents). Here we find a contrastive focus effect similar to that discovered for positive (1a). Thus, (3a) presupposes that there has been discussion of all sorts of other issues, and asserts specifically that the refugee crisis was not discussed. (3)

a. niet is gesproken over het vluchtelingenprobleem not is spoken about the refugee.problem ‘there has been no discussion of the refugee crisis’ b. niet is de rechtbank gebleken dat een strafbaar feit is gepleegd not is the court appeared that a criminal offence is committed ‘to the court, there are no indications that a criminal offence has been committed’

As in the case of (1a), it is impossible to use (3) as a simple denial of a statement with opposite polarity: for (3a), we see this in (4). For (3a) to be felicitous, it wants a preceding context in which it is affirmed that some other topics were discussed at the meeting: er is wel over de eurocrisis en de situatie in Syrië gesproken ‘there WAS discussion about the euro crisis and the situation in Syria’. But though (1a) and (3) share their contrastiveness requirement, preposing of negative niet is subject to a more stringent restriction than fronting of positive wel: in niet-fronting cases, the contrastive focus can only be a phrasal constituent in clause-final position. (4) A: er is gesproken over het vluchtelingenprobleem there is spoken about the refugee.problem ‘there has been discussion of the refugee crisis’ B: nee, er is niet gesproken over het vluchtelingenprobleem no there is not spoken about the refugee.problem Bʹ: *nee, niet is (er) gesproken over het vluchtelingenprobleem no not is there spoken about the refugee.problem From a comparative perspective, (3) is quite unusual. English neither and nor can occupy the sentence-initial position, immediately followed by the finite verb; but under no circumstances can English front the simple negative polarity particle not all by itself, as we see in (5). Though Insular Scandinavian languages can place the negation particle in clause-initial position quite The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 33

freely, the stylistic inversion process that gives rise to this word order has properties that are diametrically opposed to the conditions under which Dutch (3) is grammatical: (6a) is characterised precisely by the fact that it does not have a contrastive or focus effect. Icelandic (6b), which is not derived via stylistic inversion, seems subject to a special information-structural restriction, judging from what Holmberg (2000) says; but in light of what Brandtler (2006) reports about an apparently similar example from Swedish, (6b) is also quite different from Dutch (3). (5) a. it has not been demonstrated that a criminal offence has been committed b. * not has it been demonstrated that a criminal offence has been committed c. * not it has been demonstrated that a criminal offence has been committed (6) a. þetta ær glæpamaðurinn sem ekki hefur verið dæmdur this is the.criminal that not has been convicted ‘this is the criminal who has not been convicted’ b. ekki kann ég að not can I INF ‘I certainly can’t speak Russian’

tala speak

rússnesku Russian

Barbiers (2002) and Zeijlstra (2013) were the first to address some of the properties of Dutch sentences of the type in (3). Their examples of nietfronting are reproduced in (7) and (8) (the latter slightly simplified from the original, to save space). (7) ik had wel gezien dat Jan aankwam, maar niet had ik gezien dat Ed vertrok I had AFF seen that Jan arrived but NEG had I seen ‘I did see that Jan arrived, but I didn’t see that Ed left’

that Ed left

moeten worden aangekruist de planten die je al hebt NEG must become PRT.crossed the plants that you already have ‘you must not mark the plants that you already have’

(8) niet

Barbiers’ and Zeijlstra’s papers address (7) and (8) obliquely, and only touch upon the tip of the iceberg. I will examine the properties of (3) in more detail, against the background of the syntax of negation and association to focus.

1 Signature properties of contrastive negation particle fronting I will start by inventorying the signature properties of sentences of the type in (3) in more detail. The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 34

1.1 Obligatory extraposition An important property of both (7) and (8), and also of the examples in (3), is that the focus associate of the sentence-initial negation particle is in extraposed position, at the very end of the sentence. For clausal arguments, as in (3b) or Barbiers’ (7), this is not particularly remarkable: clausal arguments are almost always extraposed in Dutch. But for prepositional objects, as in (3a), and particularly for nominal ones, as in (8), the obligatory extraposition seen in these examples is unusual. PP-complements are normally quite flexible in their placement in Dutch: extraposition is easy — but except in sentences like (3a) (adapted as (9b)), it is never forced. The contrast in (9) brings this out. (9) a. er

zal niet <over het vluchtelingenprobleem> worden gesproken there will not about the refugee.problem become spoken <over het vluchtelingenprobleem> about the refugee.problem ‘the refugee crisis will not be discussed’

b. niet zal <*over het vluchtelingenprobleem> worden gesproken not will about the refugee.problem <over het vluchtelingenprobleem> about the refugee.problem

become spoken

Zeijlstra’s (8), repeated as (10b), confronts the fact that the focus associate of niet is in extraposed position with the fact that nominal objects do not otherwise extrapose very easily in Dutch (see (10a)). The marked nature of extraposition of nominal objects in Dutch lends a degree of unnaturalness to Zeijlstra’s example. But while somewhat awkward, (10b) is much better than the alternative in (10b′), with the object in clause-internal position. (10) a.

je moet niet <de planten die je al hebt> you must NEG the plants that you already have <?*de planten die je al hebt> the plants that you already have

aankruisen PRT.cross

b. niet moeten worden aangekruist de planten die je al hebt NEG must become PRT.crossed the plants that you already have worden aangekruist b′. *niet moeten de planten die je al hebt NEG must the plants that you already have become PRT.crossed

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 35

1.2 The phrasal nature of the fronted material There can be no doubt that sentence-initial niet is phrasal: it can, for instance, be coordinated with other phrasal material (nauwelijks ‘hardly’ in (11a)), or intensified (with volstrekt ‘absolutely’ or in het geheel ‘totally’ in (11b)). (11) a. niet of nauwelijks is gesproken over het vluchtelingenprobleem not or hardly is spoken about the refugee.problem b. volstrekt niet/in het geheel niet is gesproken over absolutely not/in the whole not is spoken about het vluchtelingenprobleem the refugee.problem From a theoretical perspective as well, the phrasality of initial niet is beyond dispute: the position occupied by fronted niet must be a specifier position in the high left periphery, in a run-of-the-mill Verb Second construction. 1.3 Sentential negation is involved The point of Zeijlstra’s (2013) paper is precisely to exclude, for languages that have Verb Second, the fronting of markers of sentential negation by themselves into sentence-initial position. Zeijlstra advocates an analysis of constructions with preposed niet as cases of constituent negation rather than sentential negation. But sentential negation must be involved. What we see in (12) is that sentence-initial niet can license the negative polarity items meer ‘anymore’, ooit ‘ever’ and enig ‘any’. Unlike sentential negation, pure constituent negation does not license NPIs, as (13b) tells us. (12) a. niet valt meer te ontkennen dat het probleem not falls anymore to deny that the problem uit de hand loopt out the hand runs ‘it cannot be denied anymore that the problem is getting out of hand’

b. niet is ooit/op enig moment gebleken/aangetoond dat not is ever/at any time appeared/demonstrated that een strafbaar feit is gepleegd a criminal offence is committed ‘it has never turned out to be the case/been demonstrated that a criminal offence has been committed’

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 36

c. niet is vast komen te staan dat hij hiermee enige bemoeienis not is fixed come to stand that he herewith any involvement heeft gehad has had ‘it has not been determined that he has had anything to do with this’

(13) a. ik wil niet meer met hem samenwerken I want not anymore with him collaborate b. ik wil [niet dit jaar maar volgend jaar] (*meer) met hem I want not this year but next year anymore with him samenwerken collaborate And in (14a,b) we see that predicates like zeker ‘certain’ license interrogative clauses only in the presence of sentential negation. Constituent negation does not license an interrogative complement clause below zeker, as (15) shows. The niet-fronting construction in (14c) again behaves like a sentential negation construction in this respect. (14) a. het is zeker {dat/*of} het probleem verholpen kan worden it is certain that/if the problem fixed can become ‘it is certain {that/*if} the problem can be fixed’ b. het is niet zeker of het probleem verholpen kan worden it is not certain if the problem fixed can be ‘it is not certain if the problem can be fixed’ c. niet is zeker of het probleem verholpen kan worden not is certain if the problem fixed can be (15) a. het is [niet zeker maar waarschijnlijk] {dat/*of} het probleem it is not certain but likely that/if the problem verholpen kan worden fixed can become b. ik weet [niet 60 maar 75%] zeker {dat/*of} het probleem I know not 60 but 75% certain that/if the problem verholpen kan worden fixed can become In principle, we should also be able to check that sentential negation is involved in niet-fronting constructions on the basis of the distribution of The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 37

affixes/clitics specifically marking sentential negation in negative concord varieties of Dutch. Unfortunately, bringing this distribution to light is difficult because of the formal register that sentence-initial niet is characteristic of: the construction type exemplified by (3), (7) and (8) does not occur naturally in negative concord dialects. But Liliane Haegeman (p.c.) points out that to the extent that she can say these things in her dialect at all, she certainly would include the marker en, the particle found only in sentential negation constructions — which goes along with the conclusion that niet-fronting constructions involve sentential negation. 1.4 Predicate pied-piping Usually, not only does the negation particle niet fail to prepose by itself (16a), niet also cannot be taken along by fronting of the predicate (16b). (16) a. *niet mag je huilen not may you cry b. *[niet huilen] mag je not cry may you But the niet-preposing constructions in focus in this paper allow both of these things. In all of the b–examples in (17)–(19), niet forms a constituent with the predicate of the extraposed material. (17) a. niet is gesproken over het vluchtelingenprobleem not is spoken about the refugee.problem b. [niet gesproken] is over het vluchtelingenprobleem not spoken is about the refugee.problem both: ‘(though other topics HAVE been discussed) there has been NO discussion of the refugee crisis’ (18) a. niet is vastgesteld dat een strafbaar feit is gepleegd not is firm.put that a criminal offence is committed b. [niet vastgesteld] is dat een strafbaar feit is gepleegd not firm.put is that a criminal offence is committed both: ‘(though other things HAVE been determined) it has NOT been determined that a criminal offence has been committed’

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 38

(19) a. niet is komen vast te staan dat een strafbaar feit is not is come firm to stand that a criminal offence is gepleegd committed b. [niet vast] is komen te staan dat een strafbaar feit is not firm is come to stand that a criminal offence is gepleegd committed both: ‘(though other things HAVE become clear) it has NOT become clear that a criminal offence has been committed’ The examples in (18b) and (19b) are particularly interesting, especially when considered in tandem. In (19b), niet forms a constituent with an adjectival predicate (vast ‘firm’) that is quite deeply embedded, under is komen te staan ‘has come to stand’, in a semi-idiomatic construction best rendered in English as has become clear or has come to be determined. Semantically, (19) is very much like (18). But in (18), though it is possible to front niet together with vastgesteld, as in (18b), it is impossible to front just niet vast ‘not firm’, stranding the participle gesteld: (18bʹ) is ungrammatical. is gepleegd (18) bʹ. *[niet vast] is gesteld dat een strafbaar feit not firm is put that a criminal offence is committed The contrast between (19b) and (18bʹ) can be related to another significant difference between vaststaan and vaststellen. The adjectival predicate vast that is part of vaststaan can be intensified, with an adverbial modifier or in a compound, as in (20b); but such intensification fails with the vast of vaststellen, as (20a) shows. The ill-formedness of (20a) suggests that vaststellen is a complex predicate, a lexical unit, whereas vast in vaststaan projects its own adjectival phrase. If this is correct, the dat-clause in (20a) is an argument of the complex predicate vaststellen as a whole, whereas in (20b) it is an argument of just the AP projected by vast. (20) a. *het is volkomen vast/muurvast gesteld dat S it is totally firm/wall.firm put that S b. het staat volkomen vast/muurvast dat S it stands totally firm/wall.firm that S ‘it is totally clear/rock-solid that S’

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 39

This now leads me directly to my proposal for the syntax of niet-preposing constructions.

2 Proposal What I propose as the analysis of niet-preposing constructions is summarised schematically as in (21): (21) a. [NegP ¬ [Neg [TP [RP [niet eci] [R [PREDICATE]]]]]] [XP C-FOCUS]i b. [CP [niet eci]C-TOPIC [C [NegP ¬ [Neg [TP [RP [niet eci] [R [PREDICATE]]]]]]]] [XP C-FOCUS]i

The negative marker niet originates as a subpart of a constituent occupying an A-position related (by a RELATOR, ‘R’) to a predicate. Like a focus particle, niet modifies a contrastively marked constituent — but it does not do so directly: the contrastive constituent is a focus (C-FOCUS) in a position at the right edge of the sentence, linked to its predicate via the empty category (ec) to which niet is attached. In niet-fronting constructions, the constituent [niet ec] is what undergoes preposing to sentence-initial position, as a contrastive topic (C-TOPIC). Niet-fronting constructions are sentential negation constructions. The source of sentential negation is an abstract negation operator ¬ in a position scoping over the proposition as a whole. The particle niet indirectly forms a constituent with the focus, but it is not a constituent negation: the morphological form niet here is a focus particle that does not itself contribute negative semantics; it is the silent ¬ operator that delivers sentential negation.

3 Two key ingredients of the analysis under the microscope 3.1 Support for niet attaching to an empty category in an A-position If (21) is right, the contrast between (18b,bʹ) and (19b) falls out immediately. niet attaches to an empty category in the argument position of the predicate of the extraposed clause. For (18), that argument position finds itself outside the projection of vastgesteld ‘firm.put, determined’, which forms a complex predicate; for (19), on the other hand, this A-position is right outside the projection of vast ‘firm’. From (22b), we get a constituent niet vast, which is eligible for fronting, as in (19b). From (22a), we get no such constituent: the smallest constituent including both niet and vast also contains gesteld; that constituent is frontable, as in (18b), but it is impossible to put just niet+vast in sentence-initial position, as in (18bʹ), because that string is not a syntactic unit.

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 40

(22) a. [RP [niet eci] [R [vastgesteld]]] [CP dat S]i b. [VP [RP [niet eci] [R [vast]]] staan] [CP dat S]i 3.2 Support for the presence of an abstract sentential negation operator Of the same general type as the examples in (17)–(19) are the alternations in (23)–(25). These introduce a novel problem, however, thanks to the fact that they include a modal. is (23) a. niet valt te controleren in hoeverre een strafbaar feit not falls to check in how.far a criminal offence is gepleegd committed b. [niet te controleren] valt in hoeverre een strafbaar feit is not to check falls in how.far a criminal offence is gepleegd committed ‘it cannot be checked to what extent a criminal offence has been committed’ (24) a. niet mag worden uitgesloten dat een strafbaar feit is not may become excluded that a criminal offence is gepleegd committed b. [niet uitgesloten] mag worden dat een strafbaar feit is not excluded may become that a criminal offence is gepleegd committed ‘it cannot be ruled out that a criminal offence has been committed’ (25) a. niet kan worden volstaan met een samenvatting not can become sufficed with a summary b. [niet] volstaan kan worden met een samenvatting not sufficed can become with a summary ‘a summary will not suffice’ In all these examples, niet has scope over the modal (vallen+te-infinitive is a modal construction; mogen ‘may’ and kunnen ‘can’ are lexical modals). Thus, (23) conveys that it is impossible to check: ¬◊. This means that niet does not belong below the modal. Nonetheless, niet can form a surface constituent with The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 41

the infinitive or participle in the modal’s complement, as in the b–examples. Despite the fact that the negation scopes over the modal, it can be included in the fronted constituent that originates inside the modal’s complement. This contrasts with what we see in (26): a negation that scopes over a modal normally cannot be pied-piped by fronting of the modal’s complement (as in (26b)); it must be stranded (as in (26a)). (26) a. huilen mag je niet cry may you not ‘you mustn’t cry’ b. *[niet huilen] mag je not cry may you

(= (16b))

These facts can be explained as follows. Ordinary, non-focus-related sentential negation involves the negation marker niet associated directly with ¬, the negation operator. The negation particle in this context is dependent on a c-command relation with the negation operator. A niet associated with the negation operator may not end up outside the c-command domain of the operator, which explains the ungrammaticality of (26b). In focus-related sentential negation constructions, on the other hand, niet is associated with the focus alone, as a focus particle. The scope of negation is marked by the abstract negation operator. The negation particle niet is licensed through its association with the extraposed focus, not through a dependency on the negation operator. When focus-associated niet is fronted out of the c-command domain of the negation operator, no ungrammaticality results. And because the negation operator is structurally higher than the modal, it scopes over the modal, even in the b–sentences in (23)–(25), where the negation particle is inside the complement of the modal. So both the grammaticality and the scope readings of (23)–(25) fall into place thanks to the abstract negation operator ¬ in the structure in (21).

4 Ellipsis In niet-fronting constructions, everything between preposed niet and the contrastive focus to which it is linked must be discourse-old and unaccented:

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 42

(27) a. ik had wel gezien dat Jan aankwam, maar zij had niet gezien I had AFF seen that Jan arrived but she had not seen dat Ed vertrok that Ed left ‘I HAD seen that Jan arrived, but she had NOT seen that Ed left’ b. *ik had wel gezien dat Jan aankwam, maar niet had zij gezien I had AFF seen that Jan arrived but not had she seen dat Ed vertrok that Ed left Unaccented topic material is eminently suitable for ellipsis. And indeed, from (28a), the entire string had ik gezien can easily be removed, producing (28b). (28) a. ik had wel gezien dat Jan aankwam, maar niet had ik gezien I had AFF seen that Jan arrived but NEG had I seen dat Ed vertrok that Ed left ‘I HAD seen that Jan arrived, but I HADN’T seen that Ed left’ b. ik had wel gezien dat Jan aankwam, maar niet dat Ed vertrok I had AFF seen that Jan arrived but NEG that Ed left Sentences of the type in (28b) raise the interesting question of how best to analyse the ellipsis process involved in them. An assumption that has been standard in the ellipsis literature for quite some time now is that ellipsis is an operation that generally targets only whole syntactic constituents (rather than contiguous PF strings). If this is correct, the question arises as to how ellipsis constructions such as (28b) could be derived. Sparing the negation while eliding the subject, finite verb, and past participle would seem to be difficult with the negation in its usual clause-medial position: in (29a), there is no single constituent that could be targeted by the ellipsis operation. (29) a. …, maar ik had niet gezien dat Ed vertrok b. …, maar [CP [niet ec] [IP ik had t gezien]] [CP dat Ed vertrok] If, on the other hand, the underlier for (28b) is (28a), constituent ellipsis is straightforwardly possible, as shown in (29b). [niet ec] preposes to SpecCP. As is usual in sluicing constructions, I-to-C raising does not take place (so had The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 43

stays in IP). The extraposed focus is outside the clause. So (28b) can be derived by IP-ellipsis, immediately following a contrastive topic. On an IP-ellipsis analysis of (28b), the incidence of niet-preposing suddenly increases dramatically: while niet-preposing in non-elliptical constructions is certainly quite marked and fairly rare, ellipsis cases of the type in (28b) are perfectly common.

5 English versus Dutch The kind of ellipsis illustrated in (28b) is also possible in English, as we see in (30b). But unlike Dutch, English does not allow the negation particle not to prepose by itself: (30a) is impossible. (30) a. *I saw that John arrived; not did I see that Ed left b. I saw that John arrived; not that Ed left I would like to argue that the fact that English does not allow notfronting, not even in the limited environments in which Dutch does, is related to the fact that constituent negation in English always wants to be directly attached to what it belongs to. Extraposition of just the associate of the negation particle, stranding the particle itself, as in (31a), is bad; but including not in the extraposed constituent is fine: (31b). The opposite is true for Dutch constituent negation: extraposition of a constituent-negated PP strands but cannot pied-pipe the negation. We see this in (32). (31) a. *there was a lot of discussion not at the meeting yesterday [about the euro crisis but about the refugee crisis] b. there was a lot of discussion at the meeting yesterday [not about the euro crisis but about the refugee crisis] (32) a. er is niet gesproken [over de eurocrisis maar over het there is spoken not about the euro.crisis but about the vluchtelingenprobleem] refugee.probleem b. *er is gesproken [niet over de eurocrisis maar over het there is spoken not about the euro.crisis but about the vluchtelingenprobleem] refugee.problem

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 44

English for some reason cannot adjoin not to an empty category linked to an extraposed associate: since it cannot make (31b), it cannot make (30a) either. English CAN, by contrast, associate its negation particle directly with the contrastive constituent itself. In light of this, it seems reasonable to suppose that examples such as (30b) come about differently in English from the way they arise in Dutch: while in Dutch (28b) the string niet dat Ed vertrok is not a single constituent, it is likely that in English (30b) not that Ed left IS one constituent, similar to not even two years ago in (33a) (which, tellingly, Dutch can only render by stranding the negation particle clause-internally, as in (33b)).1 (33) a. [not even two years ago] could you swim there b. [zelfs <*niet> twee jaar geleden] kon even zwemmen swim

two

year ago

je hier <niet> could you here not

Preposing the string not that Ed left as a unit will then provide the basis for ellipsis in (30b). A reviewer suggests that the difference between Dutch and English, at least in the realm of niet/not+clause constructions, may have to do with the position of the particle niet/not: whereas in Dutch niet is in an adjunction position, and may thus associate with the ec linked to an extraposed clause, in English not that S constructions the particle not finds itself in SpecCP, reducing the string that S to a mere C′, ineligible for extraposition. The reviewer relates to this suggestion the fact that when not immediately precedes a clause, the complementiser that is obligatory in contexts in which it would have been optional in the absence of not (I know/saw (that) John arrived but not *(that) Ed left): this, the reviewer points out, could be a residual Verb Second effect (i.e., C must be filled because SpecCP is filled, by not). I find these ideas very interesting, but must leave a fuller evaluation (and the possibility of an extension to the non-clausal examples) for another occasion.

6 Concluding remarks I have discussed a pattern of negation found in Dutch in which the negation particle is in sentence-initial position followed by the finite verb, in a Verb 1

A reviewer points out that constructions of the type not that it’s any of my business, but… might support the text hypothesis that we are dealing in (30b) with a single constituent not that Ed left. What muddies the water is that Dutch has this construction type as well: niet dat het me wat aangaat, maar… ‘not that it concerns me anything, but…’.

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken

Negation Topicalisation 45

Second pattern. This construction type, in which the negation is a contrastive topic that must be associated with a contrastive focus in extraposed position, has things in common both with sentential negation (in particular, the presence of a logical negation operator) and with constituent negation (its contrastiveness) — but it does not directly reduce to either. References Barbiers, Sjef. 2003. Microvariation in negation in varieties of Dutch. In Sjef Barbiers, Leonie Cornips & Susanne van der Kleij (eds), Syntactic microvariation. Meertens Institute Electronic Publications in Linguistics. 13–40. Brandtler, Johan. 2006. On Aristotle and baldness: Topic, reference, presupposition of existence and negation. Working Papers in Scandinavian Syntax 77. 177– 204. Holmberg, Anders. 2000. Scandinavian stylistic fronting: How any category can become an expletive. Linguistic Inquiry 31. 445–83. Zeijlstra, Hedde. 2013. Not in the first place. Natural Language & Linguistic Theory 31. 865–900.

Marcel den Dikken Department of English Linguistics, Eötvös Loránd University & Research Institute for Linguistics, Hungarian Academy of Sciences marcel.den.dikken@nytud.mta.hu

The Even Yearbook 12 (2016), Department of English Linguistics, Eötvös Loránd University, Budapest ISSN 2061–490X, http://seas3.elte.hu/even, © 2016, Marcel den Dikken