1
*Ellipsis in negative fragment answers
Lobke Aelbrecht
Catholic University of Brussels/ CRISSP
lobke.aelbrecht@kubrussel.ac.be
1. INTRODUCTION.
In certain dialects of Belgian Dutch, for instance Asse Dutch, fragment answers of the type in (1B) occur:
(1) A: Wie eit‟em gezien? - B: Niemand nie.
who has.he seen no-one not
„Who did he see?‟ – „No-one.‟ [Asse Dutch]
This example displays negative concord: B‟s answer contains two negative elements (niemand „no-one‟ and nie
„not‟), but it expresses only one semantic negation.
Merchant (2004) analyzes fragment answers as involving movement of the fragment to the left periphery of the clause, followed by ellipsis of TP. I claim that the fragment answer niemand nie „no-one not‟ is derived in a
similar manner from the sentence in (2), as is schematically represented in (3) and (4):
(2) B: Ij ee niemand nie gezien.
he has no-one not seen
„He didn‟t see anyone.‟ [Asse Dutch]
(3) STEP 1: fronting of niemand nie
[[Niemand nie] [ ij ee t gezien]] TPniemand nie
no-one not he has seen
(4) STEP 2: ellipsis of TP
[[Niemand nie] [ ij ee gezien]] TP
For this analysis to work the fragment must be a constituent, however, and I will show later on, in section four, that this is indeed the case. First, the next section introduces the basic data, while section 3 provides some background on negative concord. The actual analysis of negative fragment answers is presented in section 5. In the sixth section this analysis is extended to another kind of negative concord called negative spread, and I round off with some open questions and problems (section7) and a conclusion (section 8).
* I would like to thank Marijke De Belder, Jeroen van Craenenbroeck, Dany Jaspers, Luis Vicente and Guido Vanden Wyngaerd for their useful comments and their support, as well the audience of the Bilbao/Deusto Student Conference in Linguistics 2006 for their questions. All errors are mine.
2
2. THE BASIC DATA.
N-words are words expressing negation, like no-one, nothing or not. I will call not a negative marker or negator here, however, because it only marks negation, in contrast with the other n-words, which also express person
(no-one) or time (never) for instance. As can be seen in the examples above, these n-words sometimes co-occur.
The combination of the n-word niemand „no-one‟ and the negator nie „not‟ is not restricted to fragment answers,
however. It can also occur IP-internally and sentence-initially:
(5) a. Ij ee niemand nie gezien.
he has no-one not seen
„He didn‟t see anyone.‟
b. NIEMAND NIE eit‟em gezien.
no-one not has.he seen
„He hasn‟t seen anyone.‟ [Asse Dutch]
When niemand nie occurs at the beginning of a sentence, it has to be stressed. In the Standard Dutch variant
with only niemand, however, stress is also needed. The same pattern is observed with other n-words, such as the
negative adverbs nerges „nowhere‟ and noet „never‟.
(6) a. A: Waor zijde geweist? - B: Nerges nie.
where are.you been nowhere not
„Where have you been?‟ – „Nowhere.‟
b. Ik zen nerges nie geweist.
I am nowhere not been
„I haven‟t been anywhere.‟
c. NERGES NIE zen ik geweist.
nowhere not am I been
„I haven‟t been anywhere.‟
(7) a. A: Wanneir got‟em na ‟tschoel? - B: Noet nie.
when goes.he to the.school never not
„When does he go to school?‟ – „Never.‟
b. Ij got noet nie na ‟tschoel.
he goes never not to the.school
„He never goes to school.‟
c. NOET NIE got‟em na ‟tschoel.
never not goes.he to the.school
„He never goes to school.‟ [Asse Dutch]
In all these examples the n-word and the nie-negator express a single semantic negation. This phenomenon, generally known as negative concord, is present in a lot of natural languages and it is widely discussed in the
literature. The next section provides some background on this matter.
3
3. BACKGROUND: NEGATIVE CONCORD.
A lot has been said about negative concord already, among others by Giannakidou (2000, 2002), Haegeman (1995) and references cited there. Giannakidou (2000:458) defines the term as follows: „situations where negation is interpreted just once although it seems to be expressed more than once in the clause‟, as was the case
in the examples discussed above in that they contained a combination of an n-word plus the negation marker nie
„not‟.
There are two varieties of negative concord. The first type is negative concord proper, where an n-word obligatory co-occurs with a negative marker. The Hungarian sentences in (8) illustrate this:
(8) a. Balázs *( nem) látott semmit.
Balázs not saw.3sg nothing
„Balázs didn‟t see anything.‟
b. Senki *( nem) jött el.
no-one not came PREVERB
„No-one came along.‟
c. Balázs *( nem) beszélt senkivel semmiröl.
Balázs not spoke no-one nothing
„Balázs didn‟t talk about anything to anyone.‟ [Hungarian, Giannakidou 2000: 458-461]
A second type of negative concord is called negative spread. Here the negative meaning is „spread‟ over two or more n-words, but the negative marker is absent.
(9) NADA (*no) le ha dado Juan a nadie.
nothing not cl has given Juan to no-one
„Juan has not given anything to anyone.‟ [Spanish, Luis Vicente p.c.]
I will come back to this second kind of negative concord in section 6. Now I turn to the main prerequisite for my analysis of negative fragment answers.
4. PREREQUISITE FOR THE ANALYSIS: „N-WORD + NIE‟ = 1 CONSTITUENT.
Following Merchant (2004) I will claim that the fragment answer niemand nie has moved to a position in the left
periphery, with subsequent ellipsis of the rest of the clause. This is only possible, however, if the fragment forms one constituent. Several arguments, which are presented below, confirm that this is indeed the case. 4.1 Supporting evidence.
4.1.1. Verb second
The first piece of evidence is provided by verb second sentences. Since Dutch is a verb second language, the finite verb always sits in the second position in declarative main clauses. Exactly one constituent must precede it, as is shown in (10).
(10) a. Ik em em gisteren gezien.
I have him yesterday seen
„I saw him yesterday.‟
4
b. Gisteren em ik em gezien.
yesterday have I him seen
„I saw him yesterday.‟
c. * Gisteren ik em em gezien.
yesterday I have him seen
d. * Ik gisteren em em gezien.
I yesterday have him seen [Asse Dutch]
In sentences (10)c,d two constituents precede the finite verb em „have‟, yielding an ungrammatical result. As is illustrated in (11), however, the n-word and nie can co-occur to the left of the finite verb, a clear indication that
they form one constituent.
(11) a. Niemand nie em ik gisteren gezien.
no-one not have I yesterday seen
„I didn‟t see anyone yesterday.‟
b. Noet nie zou ik da doen.
never not would I that do
„I would never do that!‟ [Asse Dutch]
One could claim that (11) is just a case of movement of niemand or noet with a sentence negator base generated in a high position. The example in (12)b, however, shows that this analysis must be rejected: the
sentence has a non-negative matrix clause. Both niemand „no-one‟ and the negator nie „not‟ have undergone long distance movement from their base position in the subclause to the left-peripheral position of the matrix
clause:
(12) a. Ij zou dat‟em niemand nie gezien aa
he said that.he no-one not seen had
„He said that he hadn‟t seen anyone.‟
b. Niemand nie zou‟em dat‟em t gezien aa. niemand nie
no-one not said.he that.he seen had
„He said that he hadn‟t seen anyone.‟ [Asse Dutch]
4.1.2. Coordination
Another argument in favor of the claim that „n-word + nie’ is one constituent involves coordination. Two sequences of an n-word plus nie can be coordinated, as can be seen in (13). As only (like) constituents can be
coordinated (cf. Chomsky 1957, see also Williams 1978), this once again suggests that „n-word + nie’ is one constituent.
(13) a. Ij ei mou noet nie of nerges nie golpen.
he has me never not or nowhere not helped
„He didn‟t ever help me anywhere.‟
b. Noet nie of nerges nie eit em mou golpen.
never not or nowhere not has he me helped
„He didn‟t ever help me anywhere.‟ [Asse Dutch]
5
4.1.3. Complement of extraposed PPs
The negator nie „not‟ cannot by itself occur in extraposed position, as is illustrated in (14)a, but when a PP containing niemand „no-one‟ is extraposed, nie is extraposed with it (cf. (14)b). This shows that nie must be part of the DP complement of PP, or at least be part of the PP.
(14) a.*Ij zoet da veu niemand doen nie.
he would that for no-one do not
b. Vruuger deet‟em da nog veu zen vrienden, ma nou zoet‟em da doen
previously did.he that still for his friends but now would.he that do
veu niemand nie.
for no-one not
„He used to do that for his friends, but now he wouldn‟t do it for anyone.‟ [Asse Dutch] 4.1.4. Position to the left of definite DP arguments In Dutch the sentence negator nie „not‟ cannot easily occur to the left of definite DP arguments, unless the DP
has a contrastive reading (De Hoop 1992, Haegeman 1995).
(15) a. Ik peis dat‟em den auto nie gekocht eit.
I think that.he the car not bought has
„I think that he didn‟t buy the car.‟
b. Ik peis dat‟em nie den auto gekocht eit, *?( ma ‟t ois).
I think that.he not the car bought has but the house
„I think that he bought not the car, but the house.‟ [Asse Dutch]
In (15)a the sentence negator sits in its regular position right above the VP, while in (15)b it precedes the
definite DP argument. Without the contrastive reading „not the car, but the house‟, the sentence is odd. The
string „n-word + nie’, however, can occur to the left of a DP argument without a contrastive reading.
(16) Ik peis da niemand nie den auto gekocht eit.
I think that no-one not the.MASC car bought has
„I think that no-one has bought the car.‟ [Asse Dutch]
I claim that nie is not a sentence negator in this case, but that it is part of the subject DP containing the n-word.
Thus, niemand and nie form one constituent.
4.1.5. Co-occurrence with indefinite DP arguments A final argument in favor of the „one constituent‟-hypothesis is the co-occurrence of niemand nie with indefinite DP arguments. Normally, when the Dutch sentence negator nie co-occurs with an indefinite DP argument, nie
1„not‟ and the indefinite article nen „a‟ “fuse”, forming gien „no‟.
(17) a. * Ik peis dat‟em nie nen boek eit geliezen.
I think that.he not a.masc book has read
b. Ik peis dat‟em gienen boek eit geliezen.
I think that.he no.masc book has read
„I think that he didn‟t read a book.‟ [Asse Dutch]
6
The nie which appears in the sequence niemand nie in (18), however, does not obligatorily “fuse” with the
2indefinite article. This suggests that this nie is not a sentence negator, but that it is part of the DP containing niemand.
(18) Ik peis da niemand nie nen boek eit geliezen.
I think that no-one not a book has read
„I think that no-one has read a book.‟ [Asse Dutch]
4.2 The syntactic structure of niemand nie
3Haegeman (1995) analyzes the niemand nie sequence in Lapscheure Dutch (cf.(19)) as in (20).
(19) …da Valère ier niemand nie (en)-kent
…that Valère here no-one not (NEG)-knows‟
„…that Valère doesn‟t know anyone here.‟ [Lapscheure Dutch]
(20) …
CP
C‟
da FP
Valère F‟
F NegP
niemand NegP
[Neg]
‟ nie Neg
0 Neg VP/PredP
[Neg]
t (en)-kent niemand
In the tree in (20) nie is the first specifier of a NegP dominating the VP or PredP, and the n-word niemand
moves to a second specifier position of NegP to check its Neg-feature. In a second step the n-word and nie
undergo Neg-absorption, thus yielding a single semantic negation (see Haegeman 1995 for more details).
Haegeman‟s account does not carry over to Asse Dutch, however. In her analysis the n-word and nie do not
form one constituent, in contrast to what the data discussed in section 4.1 suggest. Furthermore, this analysis
cannot offer a straightforward account of negative fragment answers such as niemand nie in example (1) above.
As a result, I propose a new analysis in which niemand and nie do form one constituent. I claim that there is a NegP inside the DP, with nie as its head. As in Haegeman‟s (1995) structure, negative elements move to the
7
specifier position of NegP to check their Neg-feature. For the DP niemand nie this means that niemand moves
from its base position to the specifier position of the DP-internal NegP, which results in the order niemand nie.
4This analysis is illustrated in the tree in (21).
(21) DP
D‟
D NegP
niemand Neg‟
[Neg]
0 Neg NP
nie
[Neg] t niemand
5. THE ANALYSIS: FRAGMENT ANSWERS.
Merchant (2004) assumes fragment answers to be derived from full sentential structures by ellipsis. In his
account the fragment first moves from its base position to the left periphery of the clause, followed by ellipsis of
TP, so that only the fragment remains. The two steps of the derivation of the fragment answer in (22) are shown
5in (23).
(22) A: Who did she see?
B: John [ she saw t]. TP John
(23) FP
[ John] F‟ DP2 step 2: ellipsis F TP
she saw t2 step 1: movement
I analyze negative fragment answers such as niemand nie in (1), repeated in (24), in the same way: since the n-word and nie form one constituent (cf. section 4), they can move to [spec,FP] together. In a second step TP is
elided.
(24) A:Wie eit‟em gezien? - B: Niemand nie.
who has.he seen no-one not
„Who did he see?‟ – „No-one.‟ [Asse Dutch]
8
Step 1:
(25) [Ij eit niemand nie gezien]
he has no-one not seen
6 ; [Niemand nie [ij eit t gezien]] TP niemand nie
no-one not he has seen
Step 2:
(26) [Niemand nie [ ij eit t gezien]] IPniemand nie
(27) FP
[ niemand nie] F‟ step 2: ellipsis DP F TP
ij eit tgezien DP step 1: movement
Some of Merchant‟s (2004) arguments in favor of his analysis of fragment answers also apply to these niemand
nie cases, for instance island sensitivity. As fragments move to the left periphery, the prediction is that they are
sensitive to islands. This is not easy to show, however: normally, the question which the fragment would be an answer to would itself involve movement of the wh-element out of an island and thus be ungrammatical. Therefore Merchant uses questions with an intonation rise on the questioned constituent in situ, such as the
7example in (28).
(28) A: Eit‟em me EUR geklapt? - B: Nië, me NIEMAND NIE
has.he with her.FOC spoken - no with no-one not
„Did he talk to HER?‟ – „No, to NO-ONE.‟ [Asse Dutch]
Since there is no need to move a wh-element in these questions, the accented constituent can be embedded in an island, providing a test for island sensitivity of the fragment answers: if fragment answers involve movement of the fragment to the left periphery, a fragment answer to a question containing an island should be impossible,
8while the full sentential answer is grammatical. This prediction is borne out, as (29) and (30) illustrate. As can be seen, niemand nie and other fragment answers behave alike.
(29) A: Eit‟em geklapt me da maske da THOMAS graug eit?
has.he talked with the girl that Thomas.FOC gladly has
„Did he talk to the girl THOMAS likes?‟
B: *Nië, {BERT/ NIEMAND NIE}.
no Bert no-one not
B: Nië, ij eit geklapt me da maske da {BERT/ NIEMAND NIE} graug eit.
no he has talked with the girl that Bert no-one not gladly has
„No, he talked to the girl {BERT/NO-ONE} likes.
(30) A: Is ze weggegon omda LUKA me eur wou dansen?
is she away.gone because Luka.FOC with her wanted dance
9
„Did she leave because LUKA wanted to dance with her?‟
B: * Nië, {STIJN/NIEMAND NIE}.
no Stijn no-one not
B: Nië, ze is weggegon omda {STIJN/ NIEMAND NIE} me eur wou dansen.
no she is away.gone because Stijn no-one not with her wanted dance
„No, she left because {STIJN/NO-ONE} wanted to dance with her.‟ [Asse Dutch]
6. EXTENSION OF THE ANALYSIS: NEGATIVE SPREAD.
In this section I extend my analysis of niemand nie to the phenomenon of negative spread , which is also attested in certain dialects of Belgian Dutch. In a first subsection I take a look at some data, comparing the negative
spread cases to that of niemand nie. From these data it follows that the string „n-word + nie’ is structurally
ambiguous in a number of contexts. Subsection 6.2 provides an analysis for this structural ambiguity. 6.1 Negative spread in southern Dutch
Negative spread is distinguished from negative concord proper by Giannakidou (2000, 2002). In negative spread
contexts two or more n-words co-occur, expressing one semantic negation. Normally, a negation marker does not occur in negative spread, but in certain Southern Dutch dialects nie „not‟ optionally appears together with
two or more n-words:
(31) Ik em niemand niks (nie) gegeiven.
I have no-one nothing not given
„I didn‟t give anything to anyone.‟ [Asse Dutch]
The string niemand niks nie „no-one nothing not‟ in (31) differs from the combination „n-word + nie’ discussed
above. First of all, it cannot occur to the left of the finite verb in a declarative matrix clause:
(32) * Niemand niks (nie) em ik gegeiven.
no-one nothing not have I given [Asse Dutch]
A second difference between the two constructions has to do with coordination. Sequences of two n-words plus
nie cannot be coordinated.
(33) * Ij eit noet niks (nie) en nerges niks (nie) gezien.
he has never nothing not and nowhere nothing not seen
(34) * Ij eit noet niemand (nie) en nerges niks (nie) gezien.
he has never no-one not and nowhere nothing not seen [Asse Dutch] Furthermore, niemand niks nie is disallowed as the complement of an extraposed PP (35) and cannot easily precede definite DP arguments (36), unlike niemand nie:
(35) *Ij zoet t doen veu niemand niks (nie). voor niemand niks nie
he would do for no-one nothing not [Asse Dutch]
(36) ?? Ij eit noet niemand nie dienen boek gegeiven.
he has never no-one not that.MASC book given [Asse Dutch]
The fifth argument that showed niemand and nie to be one constituent in section 4.1 once again yields the opposite result for niemand niks nie either: when the string occurs with an indefinite DP argument, nie must
„fuse‟ with the indefinite article to form gien „no‟.
10
(37) a.*Ij eit noet niemand nie nen boek gegeiven. (NC)
he has never no-one not a.MASC book given
b. Ij eit noet niemand gienen boek gegeiven.
he has never no-one no.MASC book given
„He never gave anyone a book.‟ [Asse Dutch]
Finally, the string with two n-words and nie is disallowed as a fragment answer:
9(38) A: Wie eit‟em wa gegeiven? - B: *Niemand niks nie.
who has.he what given no-one nothing not [Asse Dutch]
These 6 empirical tests show that the string niemand niks nie does not form a constituent, and that nie is a
sentence negator here, which I assume occupies a position in a NegP preceding the verb phrase. I therefore
adopt Haegeman‟s (1995) proposal for these negative spread phenomena. In her analysis both n-words move
from their base position to a specifier position of NegP dominating the VP, checking their Neg-feature:
(39) …
FP
ik F‟
F NegP
em
niemand Neg‟
[Neg]
‟ niks Neg
[Neg]
nie Neg‟
0 Neg VP
[Neg]
tt t gegeiven em niemandniks
6.2 The structural ambiguity of ‘n-word + nie‟
It is clear from the previous subsection that a string with two n-words and an optional nie has a different
structure than the niemand nie we find in fragment answers. The latter is one constituent, a DP with an internal NegP, while in the former the n-words and nie are all specifiers of a NegP dominating the VP. This account predicts, however, that the structure used for two or more n-words should in principle also be available for
sentences with only one n-word plus nie. Nothing in the structure in (39) forces the sentence to contain more than one n-word. That means that a sentence such as Ik em niemand nie gezien „I have no-one not seen‟ is