• Dutch
  • Frisian
  • Afrikaans
Show full table of contents
5.1.5.Fragment clauses

Fragment clauses cannot be immediately recognized as such because they do not contain an overt finite verb and, consequently, look like phrases of some non-verbal category. There are two types of fragment clauses: fragment wh-questions and fragment answers. Examples of the former are given in the primed examples in (248), which show that fragment wh-questions can plausibly be analyzed as phonetically reduced finite interrogative clauses.

Example 248
a. Jan heeft gisteren iemand bezocht.
speaker A
  Jan has  yesterday  someone  visited
  'Jan visited someone yesterday.'
a'. Wie heeft Jan gisteren bezocht?
speaker B
  who has Jan yesterday visited
  'Who (did he visit yesterday)?'
b. Jan heeft Marie bezocht.
speaker A
  Jan has  Marie visited
  'Jan has visited Marie'
b'. Wanneer heeft Jan Marie bezocht?
speaker B
  when  has  Jan Marie  visited
  'When (did Jan visit Marie)?'

Ross (1967) derived fragment wh-questions by means of a deletion operation that he referred to as sluicing, and fragment wh-questions are therefore also known as sluicing constructions; the suppressed information is indicated here by means of strikethrough. At first sight, the deletion seems licensed simply by the presence of some antecedent clause in the preceding discourse, which contains some (implicit) correlate of the wh-phrase constituting the fragment wh-question, but our discussion below will bear out that on closer scrutiny the situation is more complex.
      The examples in (249) show that fragment answers may arise in conversation as a response to wh-questions; the suppressed information is again indicated by strikethrough.

Example 249
a. Wat heeft Jan gisteren gekocht?
speaker A
  what  has  Jan yesterday  bought
  'What did Jan buy yesterday?'
a'. Een nieuwe computer heeft Jan gisteren gekocht.
speaker B
  a new computer  has  Jan yesterday  bought
  'A new computer (Jan bought yesterday).'
b. Wanneer heeft Jan die nieuwe computer gekocht?
speaker A
  when  has  Jan that new computer  bought
  'When did Jan buy that new computer?'
b'. Gisteren heeft Jan die nieuwe computer gekocht.
speaker B
  yesterday  has  Jan that new computer  bought
  'Yesterday (Jan bought that new computer).'

The non-reduced clauses corresponding to the fragment clauses in the examples above are grammatical but less felicitous, for reasons of economy, given that the suppressed information can easily be reconstructed from the context; usually the preceding discourse contains some antecedent clause which provides the information suppressed in the fragment clause. Nevertheless, we cannot a priori assume that the deletion analysis suggested above is correct, especially because it runs into several problems. Establishing that we are dealing with some kind of reduction will therefore be an essential part of our discussion of fragment clauses. After having established this, we will discuss the properties of fragment clauses in greater detail. Fragment wh-questions are discussed in Subsection I and fragment answers in Subsection II.

[+]  I.  Fragment wh-questions (sluicing)

The examples in (250) show that fragment wh-questions do not only occur as independent utterances but also as subparts of clauses. If we are indeed dealing with reduced clauses, this would show that sluicing can apply to matrix and embedded clauses alike.

Example 250
a. Jan heeft gisteren iemand bezocht.
speaker A
  Jan has  yesterday  someone  visited
  'Jan visited someone yesterday.'
a'. Kan je me ook zeggen wie Jan gisteren bezocht heeft?
speaker B
  can you  me also  tell  who  Jan yesterday  visited  has
  'Can you tell me who (Jan visited yesterday)?'
b. Jan heeft gisteren iemand bezocht, maar ik weet niet wie Jan gisteren bezocht heeft?
  Jan has  yesterday  someone  visited    but  I know not  who  Jan yesterday  visited  has
  'Jan visited someone yesterday, but I donʼt know who.'

The following subsections discuss fragment wh-questions in more detail, subsection A begins by showing that fragment wh-questions are indeed clauses, and that we must therefore assume that some sort of sluicing operation is at work here. This need not imply, however, that sluicing must be seen as a deletion operation, subsection B shows that there are at least two ways of analyzing sluicing, which in fact both face a number of challenges, subsection C continues by investigating to what extent the interpretatively present but phonetically non-expressed part of the fragment wh-question must be isomorphic to some antecedent clause, subsection D investigates the correlate of the wh-phrase in the antecedent clause, subsection E concludes with a number of specific examples that may involve sluicing.

[+]  A.  Fragment wh-questions are clauses

This subsection reviews the evidence in favor of the claim that fragment wh-questions are really clauses. We will follow the literature in mainly discussing examples of the type in (250b), but this is not a matter of principle; similar arguments can be given on the basis of examples such as (250a').

[+]  1.  Selection restrictions

A first argument for claiming that fragment wh-questions are clauses is based on the selection restrictions imposed by the verb on its complements; embedded fragment wh-questions can only occur with predicates that select interrogative clauses. The primeless examples in (251) illustrate that verbs like weten'to know' and zien'to see' may take an interrogative clause and the primed examples show that they may likewise take an embedded fragment wh-question. Examples such as (251a') are especially telling given that the verb weten'to know' can only be combined with a severely limited set of noun phrases, and noun phrases referring to objects are certainly not part of this set (contrary to what is the case with its English counterpart to know): cf. Ik weet het antwoord/*dat boek'I know the answer/that book'.

Example 251
a. Ik weet [wat Jan gekocht heeft].
  know  what  Jan bought  has
  'I know what Jan has bought.'
a'. Jan heeft iets gekocht maar ik weet niet wat.
  Jan has  something  bought  but  know  not  what
  'Jan bought something but I donʼt know what.'
b. Ik zag [wie er wegrende].
  I saw   who  there  away-ran
  'I saw who ran away.'
b'. Er rende iemand weg en ik zag ook wie.
  there  ran  someone  away  and  saw  also  who
  'Someone ran away, and I also saw who.'

The examples in (252) show that verbs like beweren'to claim', which do not select interrogative clauses, cannot be combined with fragment wh-questions either.

Example 252
a. * Marie beweert [wat Jan gekocht heeft].
  Marie claims   what  Jan bought  has
b. * Peter denkt dat Jan iets gekocht heeft *(en Marie beweert wat).
  Peter thinks  that  Jan something  bought  has    and  Marie claims  what
[+]  2.  Coordination

A second argument for assuming that fragment wh-questions are clauses can be based on coordination: given that coordination is normally restricted to phrases of the same categorial type, the fact that full clauses fragment wh-questions can be coordinated with fragment wh-questions suggests that the first are also clauses.

Example 253
a. Jan vroeg me [[waar ik gewoond had] en [hoe lang]].
  Jan asked  me  where  lived  had  and   how long
  'Jan asked me where I had lived and for how long.'
b. Ik weet niet [[wat hij gedaan heeft] of [waarom]].
  know  not   what  he done has  or why
  'I donʼt know what he has done or why.'
[+]  3.  Case assignment

A third argument is based on case assignment: the wh-phrase constituting the overt part of the fragment wh-question in (254a) is assigned the same case as the corresponding phrase in the antecedent clause and not the case normally assigned by the embedding predicate. One must keep in mind, however, that cases like these may be misleading as they may involve N-ellipsis on top of sluicing. An argument in favor of such an analysis is that the possessive pronoun wiens in (254b) does not have a syntactic correlate in the antecedent clause, whereas the noun phrase wiens auto does.

Example 254
a. Jan heeft iemands boek gelezen, maar ik weet niet wiens.
  Jan has  someoneʼs book  read  but  know  not  whose
  'Jan has read someoneʼs book but I donʼt know whose.'
b. Er staat een auto op de stoep, maar ik weet niet wiens.
  there  stands  a car  on the pavement  but  know  not  whose
  'There is a car on the pavement but I donʼt know whose.'

Since Dutch has overt case marking on pronominal possessives only, we cannot provide any better evidence than cases such as (254), but Merchant (2001/2006) provides a number of examples from German (and other languages) that involve nominal arguments. Although the verb wissen'to know' governs accusative case, the wh-phrase that constitutes the fragment wh-question in (255) has dative case just like the complement of the verb schmeicheln'to flatter' in the antecedent clause.

Example 255
Er will jemandemdat schmeicheln, aber sie wissen nicht wemdat/*wenacc.
  he wants someone  flatter  but  they  know  not who/who
'He wants to flatter someone, but they donʼt know who.'
[+]  4.  Syntactic distribution/placement of fragment wh-questions

The most important argument for claiming that fragment wh-questions are clauses involves the syntactic distribution of embedded fragment wh-questions like Wie?'Who?' or Wat?'What?'. If such fragment wh-questions were noun phrases, we would expect them to have the distribution of nominal phrases and hence to appear before the clause-final verbs. If, on the other hand, such fragment wh-questions are clauses, we expect them to occur in the normal position of clauses, that is, after the clause-final verbs. The examples in (256) therefore unambiguously show that fragment wh-questions are clauses.

Example 256
a. Jan heeft iets gekocht en ik denk dat ik weet wat.
  Jan has  something  bought  and  think  that  know  what
  'Jan has bought something and I think that I know what.'
b. * Jan heeft iets gekocht en ik denk dat ik wat weet.
  Jan has  something  bought  and  think  that  what  know

The examples in (257) show that, like regular object clauses, fragment wh-questions functioning as direct object can only occur to the left of the clause-final verbs if they are topicalized or left-dislocated. The relevant sluicing construction is given in the second conjunct of (257b).

Example 257
a. [Wat hij gekocht heeft] (dat) weet ik niet.
  what  he  bought  has   that  know  not
  'What he bought, (that) I donʼt know.'
b. Hij heeft iets gekocht, maar wat (dat) weet ik niet.
  he  has  something  bought  but  what   that  know  not
  'He bought something but what (that) I donʼt know.'
[+]  5.  The anticipatory pronoun het

Yet another argument involves the distribution of the anticipatory pronoun het. We would expect this pronoun to be possible if fragment wh-questions are clauses, but not if they are some non-verbal category. The examples in (258) show that the results are somewhat mixed: the (a)-examples show that fragment wh-questions functioning as objects cannot co-occur with the anticipatory pronoun het, whereas the (b)-examples show that fragment wh-questions functioning as subjects can.

Example 258
a. Ik weet (het) nog niet [wie er morgen komt].
  know    it  yet  not   who  there  tomorrow  comes
  'I donʼt know yet who is coming tomorrow.'
a'. Er komt morgen iemand, maar ik weet (*het) nog niet wie.
  there  comes  tomorrow  someone  but  know     it  yet  not  who
  'Someone will be coming tomorrow, but I donʼt know yet who.'
b. Het is nog niet duidelijk [wie er morgen komt].
  it  is yet  not  clear   who  there  tomorrow  comes
  'It isnʼt clear yet who will come tomorrow.'
b'. Er komt morgen iemand, maar het is nog niet duidelijk wie.
  there  comes  tomorrow  someone  but  it  is yet  not  clear  who
  'Someone will be coming tomorrow, but it isnʼt clear yet who.'

A possible account for the contrast between the two primed examples in (258) may be that fragment wh-questions are always part of the focus (new information) of the clause, as is clear from the fact that they are always assigned contrastive accent. Section 5.1.1, sub III, has shown that the anticipatory object pronoun het tends to trigger a presuppositional reading of the object clause; so it may be that combining it with a fragment wh-question results in an incoherent information structure, which may account for the judgment given in (258a'). Although Section 5.1.3, sub III, has shown that the anticipatory subject pronoun het can sometimes likewise trigger a presuppositional reading of the subject clause, there are also many cases in which this effect does not arise; this means that the information structure of example (258b') may be fully coherent, regardless of whether the anticipatory pronoun is present or not. We leave it to future research to establish whether this account of the contrast between the two primed examples in (258) is tenable, but conclude for the moment that the acceptability of the anticipatory pronoun het in examples such as (258b') provides support for the claim that fragment wh-questions are clauses.

[+]  6.  Left dislocation

The argument on the basis of the anticipatory pronoun can be replicated in a slightly more straightforward form on the basis of left-dislocation constructions such as (259); the primed examples show that the resumptive pronoun dat'that' is possible with fragment wh-questions, irrespective of the latter's function.

Example 259
a. [Wie er morgen komt] dat weet ik nog niet.
  who  there  tomorrow  comes  that  know  not  yet
  'Who is coming tomorrow, that I donʼt know yet.'
a'. Er komt morgen iemand, maar wie dat weet ik nog niet.
  there  comes  tomorrow  someone  but  who  that  know  yet  not
  'Someone will be coming tomorrow, but who, that I donʼt know yet.'
b. [Wie er morgen komt] dat is nog niet duidelijk.
  who  there  tomorrow  comes  that  is yet  not  clear
  'Who is coming tomorrow, that isnʼt clear yet.'
b'. Er komt morgen iemand, maar wie dat is nog niet duidelijk.
  there  comes  tomorrow  someone  but  who  that  is yet  not  clear
  'Someone will be coming tomorrow, but who, that isnʼt clear yet.'

It should be noted that the possibility of left dislocation strongly disfavors the nominal analysis of fragment wh-questions. First, example (260) shows that left dislocation is normally excluded with wh-phrases.

Example 260
a. Wat (*dat) wil je kopen?
  what     that  want  you  buy
  'What do you want to buy?'
b. Welke boeken (*die) wil je kopen?
  which books   these  want  you  buy
  'Which books do you want to buy?'

Second, the primeless examples in (261) show that resumptive pronouns normally exhibit number agreement with left-dislocated noun phrases, whereas the primed examples show that left dislocation of fragment wh-clauses involves the invariant form dat'that', that is, the form normally found with left-dislocated clauses.

Example 261
a. Het boek, dat wil ik kopen.
  the book  that  want  buy
a'. Jan wil een boek kopen, maar welksg dat weet ik niet.
  Jan  wants  a book  buy  but  which  that  know  not
b. De boeken, die/*dat wil Jan kopen.
  the books  those/that  want  Jan buy
b'. Jan wil wat boeken kopen, maar welkepl dat/*die weet ik niet.
  Jan wants  some books  buy,  but  which  that/these  know  not
[+]  7.  Nominalization

Nominalization also provides evidence for the claim that fragment wh-questions are clauses. First, the (a)-examples in (262) show that nominal objects of verbs normally appear as van-PPs in the corresponding nominalizations; cf. N2.2.3.2. Second, the (b)-examples show that object clauses are never preceded by a preposition. The fact that the nominalization in (262c') does not contain the preposition van thus shows that fragment clauses are not nominal, but clausal.

Example 262
a. Jan rookt sigaren.
  Jan smokes  cigars
a'. [Het roken *(van) sigaren] is ongezond.
  the  smoking     of  cigars  is unhealthy
b. Marie vroeg [waarom Jan sigaren rookt].
  Marie asked   why  Jan cigars  smokes
  'Marie asked why Jan smokes cigars.'
b'. Marie vroeg waarom.
  Marie asked  why
c. de vraag [waarom Jan sigaren rookt]
  the question   why  Jan cigars  smokes
  'the question as to why Jan smokes cigars'
c'. de vraag waarom
  the question  why
[+]  8.  Subject-verb agreement

The final argument again pertains to fragment wh-questions functioning as subjects. If fragment wh-questions are really clauses, we expect finite verbs to exhibit (default) singular agreement throughout, whereas we would expect finite verbs to agree in number with nominal fragment wh-questions if they are not. The examples in (263) show that the former prediction is the correct one; finite verbs are always singular even if the fragment wh-question has the form of a plural noun phrase.

Example 263
a. Het is niet duidelijk [welke boeken Jan wil hebben].
  it  is not  clear   which books  Jan wants.to  have
  'It isnʼt clear which books Jan wants to have.'
a'. Jan wil wat boeken hebben, maar het is/*zijn niet duidelijk welke.
  Jan wants.to  some books  have  but  it  is/are  not  clear  which
  'Jan wants to have some books, but it isnʼt clear which.'
b. [Welke boeken Jan wil hebben] is niet duidelijk.
  which books  Jan wants.to  have  is not  clear
  'Which books Jan wants to have isnʼt clear.'
b'. Jan wil wat boeken hebben, maar welke is/*zijn niet duidelijk.
  Jan wants  some books  have  but  which  is/are  not  clear
  'Jan wants to have some books, but which ones isnʼt clear.'
[+]  B.  What is Sluicing?

The previous subsection has shown that there is overwhelming evidence in favor of the claim that fragment wh-questions are clausal in nature, and hence that something like sluicing must exist. Let us assume the standard generative claim discussed in Section 9.1 that embedded finite interrogative clauses have the CP/TP structure in (264a), and that the wh-element occupies the position preceding the (phonetically empty) complementizer indicated by C. Sluicing can then be derived in at least two ways: the phonetic content of TP might be deleted under identity with its antecedent clause in the preceding discourse, or the TP might be phonetically empty right from the start and function as a pro-form that can be assigned an interpretation on the basis of its antecedent clause. The two options have been indicated in the (b)-examples in (264), in which strikethrough stands for deletion of the phonetic content of the TP and e for an empty pro-form replacing TP.

Example 264
a. Ik weet niet [CP wati C [TP Jan gekocht ti heeft]].
  I know not  what  Jan bought  has
  'I donʼt know what Jan has bought.'
b. Ik weet niet [CP wati C [TPJan gekocht ti heeft]].
b'. Ik weet niet [CP wat C [TP e ]].

We will not attempt to compare the two analyses here, but confine ourselves to mentioning a series of problems that must be solved by any proposal that claims that fragment wh-questions are CPs with a phonetically empty TP; readers who are interested in a comparison of the two analyses are referred to Merchant (2001/2006), who also discusses a number of other proposals, such as the idea that fragment wh-questions are reduced wh-cleft-constructions: Wat is het dat Jan gekocht heeft'What is it that Jan has bought?'. Because it is easier for reasons of exposition, we will follow Merchant's (2001/2006) wh-movement + TP deletion approach in (264b) in our structural representations, without intending to imply, however, that we consider this approach superior or inferior to the TP pro-form approach.

[+]  1.  Sluicing is possible in wh-questions only

A first problem that should be accounted for is that sluicing is generally impossible outside the domain of fragment wh-questions. This is illustrated in the (a)-examples in (265): the first conjunct Jan is hier may not give rise to sluicing in the declarative object clause in the second conjunct, although it can be pronominalized by means of the pronoun het/dat. The same thing is illustrated in the (b)-examples which involve an embedded yes/no question. The unacceptability of the primeless examples shows that we need to formulate certain non-trivial conditions on the application of sluicing to ensure that it gives rise to fragment wh-questions only.

Example 265
a. * Jan is hier maar Peter mag niet weten [dat hij hier is]
  Jan is here  but  Peter is.supposed.to  not  know   that  he  here  is
a'. Jan is hier maar Peter mag het/dat niet weten
  Jan is here  but  Peter  is.supposed.to  it/that  not  know
  'Jan is here but Peter isnʼt supposed to know it/that.'
b. * Jan komt misschien maar niemand weet zeker [of hij komt].
  Jan comes  maybe  but  nobody knows  for.sure  whether  he  comes
b'. Jan komt misschien maar niemand weet het/dat zeker.
  Jan comes  maybe  but  nobody knows  it/that  for.sure
  'Jan may be coming but nobody knows it/that for certain.'

For completeness' sake, example (266b) shows that sluicing is not possible in the domain of wh-exclamatives either.

Example 266
a. Het is ongelooflijk [wat een boeken Els geschreven heeft]!
  it  is incredible  what a books  Els  written  has
  'It is incredible how many books Els has written!'
b. * Els heeft veel geschreven; het is vooral ongelooflijk wat een boeken.
  Els has  a.lot  written  it  is especially  incredible  what a books
[+]  2.  The overt part does not include elements in the C-position

A second problem that should be solved is that fragment wh-questions normally cannot contain material that is not part of the wh-phrase. Some speakers of Dutch allow the overt realization of the complementizer of in embedded clauses, but, contrary to what is to be expected on the basis of the analyses in (264), example (267a') shows that the complementizer does not surface in embedded fragment wh-questions. The (b)-examples further show that, under the standard analysis that finite verbs occupy the C-position in interrogative main clauses, it would predict wrongly that non-embedded fragment wh-questions like Wat?'What?' should contain a finite verb.

Example 267
a. Ik weet niet [CP wati of [TP hij ti zei]].
  know  not  what  comp  he  said
  'I donʼt know what he said.'
a'. * Hij zei iets maar ik weet niet [CP wati of [TP hij ti zei]].
  he  said something  but  I know not  what  comp  he  said
  'He said something but I donʼt know what.'
b. Hij zei iets.
  he said  something
b'. * [CP Wat [C zei] [TP hij titv]]?
  what  said  he

Naturally, the ungrammaticality of the primed examples in (267) may be solved by assuming that sluicing affects the sequence C + TP, but this assumption is less desirable given that deletion and pronominalization normally involve maximal projections. If we want to stick to this standard assumption, the analyses in (264) require additional stipulations to be made; see cf. Merchant (2001:281ff.).

[+]  3.  The overt part sometimes includes TP-internal material

The third problem is in a sense the reverse of the second one: if sluicing involves deletion or pronominalization of the TP projection, we wrongly predict that TP-internal material will never surface. A first case that proves that this is wrong has to do with multiple questions. Example (268a) shows that, like in English, Dutch multiple questions allow at most one wh-phrase in the CP projection, which predicts that fragment wh-questions also consist of at most one wh-phrase. The (b)-examples in (268) show, however, that the presumed TP-internal wh-phrase in multiple wh-questions must be overtly expressed in fragment questions: leaving it out leads to unacceptability; we refer the reader to Merchant (2001:285ff.) for a more detailed discussion of this issue.

Example 268
a. Ik weet [CP wiei C [TPti gisteren wat las]].
  know  who  yesterday  what  read
  'I know who read what yesterday.'
b. Iedereen las gisteren iets maar ik weet niet wie wat.
  everyone  read  yesterday  something  but  know  not  who  what
b'. * Iedereen las gisteren iets maar ik weet niet wie.
  everyone  read  yesterday  something  but  know  not  who

It may be relevant in this connection that although multiple questions can be straightforward main clauses, non-embedded multiple fragment wh-questions are very marked. This is illustrated in the examples in (269).

Example 269
a. Wie heeft wat gelezen?
  who  has  what  read
  'Who read what?'
b. A. Iedereen heeft iets gelezen. B. *?Wie wat?
  A. everyone  has  something  read  B.   who  what
  'Everyone has read something. Who what?'

A second case, not mentioned in Merchant (2001/2006), involves constructions with floating quantifiers. The examples in (270) show that although the Dutch floating quantifiers nog meer'else' and allemaal'all' must appear TP-internally in embedded interrogative clauses, they nevertheless seem to survive sluicing. Note that the problem does not occur in English given that the quantifiers may be adjacent to the wh-phrase in regular wh-questions; cf. Merchant (2006:122).

Example 270
a. Ik ben vergeten [CP wie <*nog meer> C [TP er <nog meer> waren]].
  am  forgotten  who    else  there  were
  'Iʼve forgotten who else were there.'
a'. Jan was er, maar ik ben vergeten wie nog meer.
  Jan was there  but  am  forgotten  who  else
  'Jan was there but Iʼve forgotten who else.'
b. Ik ben vergeten [CP wie <*allemaal> C [TP er <allemaal> waren]].
  am  forgotten  who      all  there  were
  'Iʼve forgotten who all were there.'
b'. Er waren veel mensen, maar ik ben vergeten wie allemaal.
  there  were  many people  but  am  forgotten  who  all
  'There were many people but Iʼve forgotten who all.'

The primed (a)- and (b)-examples in (271) show that we find the same facts in main clauses. In fact, the primed (c)-examples seem to show that it is even possible in such cases to construct fragment wh-questions that contain adverbial-like material.

Example 271
a. Wie <*nog meer> waren er <nog meer>?
  who     else  were  there
a'. A. Jan was er. B. Leuk! Wie nog meer?
  A. Jan was there  B. nice  who else
b. Wie <*allemaal> waren er <allemaal>?
  who     all  were  there
b'. A. Er waren veel mensen.