• Dutch1
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
11.1. The formation of V1- and V2-clauses
quickinfo

This section discusses some general issues related to the clause-initial position, subsection I starts with a review of the operation that moves the finite verb from its clause-final position into the C-position in the left-periphery of the clause; see Chapter 10 for a more extensive discussion. Verb movement results in verb-first (henceforth: V1) structures and Subsection II will demonstrate how verb-second (henceforth: V2) clauses can be derived by subsequent topicalization or question formation, subsection III will show that the clause-initial position can be filled by at most one constituent, subsection IV will show that there are no constraints on the syntactic function of the constituent occupying the clause-initial position; it seems that virtually any clausal constituent can occupy this position. This is related to the fact, discussed in Subsection V, that the clause-initial constituent normally has a specific information-structural function, subject-initial main clauses are exceptional in this respect but Subsection VI will show that there are more reasons to set such cases apart, subsection VII concludes by showing that main and embedded clauses exhibit different behavior with respect to their initial position: for example, while the initial position of declarative main clauses is normally filled by the subject or some topicalized element, the initial position of declarative embedded clauses is normally empty.

readmore
[+]  I.  Verb movement: Verb-first/second

Since Paardekooper (1961) it has normally been assumed that complementizers in embedded clauses and finite verbs in main clauses occupy the same structural position in the clause. In the traditional version of generative grammar this is derived as depicted in (4). In embedded clauses, the complementizer dat'that' or of'if' must be inserted in the C(omplementizer)-position. In main clauses, the finite verb is moved into this position from its original VP-internal position via the intermediate T(ense)-position; note that, for theoretical reasons, it is normally also assumed that the finite verb also moves through all intermediate X-positions, but this is not depicted here. Verb-movement is blocked in embedded clauses because complementizer insertion is obligatory in this context and thus occupies the target position of the finite verb. The obligatoriness of verb-movement in main clauses follows if we assume that the C-position must be filled but that complementizer insertion is restricted to embedded clauses.

4

The claim that complementizers in embedded clauses and finite verbs in main clauses are placed in the C-position is empirically motivated by Paardekooper's observation that they display similar placement with respect to referential subject pronouns like zij'she'. Putting subject-initial main clauses aside for the moment, the examples in (5) show that such pronouns are always right-adjacent to the finite verb in main clauses or right-adjacent to the complementizer in embedded clauses.

5
a. Gisteren was zij voor zaken in Utrecht.
main clause
  yesterday  was she  on business  in Utrecht
  'Yesterday she was in Utrecht on business.'
a'. * Gisteren was voor zaken zij in Utrecht.
b. Ik dacht [dat zij voor zaken in Utrecht was].
embedded clause
  thought  that  she  on business  in Utrecht was
  'I thought that she was in Utrecht on business.'
b'. * Ik dacht dat voor zaken zij in Utrecht was.

This observation can be derived immediately if we assume that subject pronouns obligatorily occupy the regular subject position, that is, the specifier position of TP, which is indicated by "Subject" in representation (4).

[+]  II.  Topicalization and question formation

The derivation of V1 and V2-clauses is now very straightforward and simple. The clause-initial position can be identified with the specifier position of CP, indicated in (4) by the dots preceding the C-position. V1-clauses arise if this position remains empty, while V2-clauses arise if this position is filled by some constituent. Prototypical cases of V1-clauses are yes/no-questions such as (6a); whether the clause-initial position is truly empty or filled by some phonetically empty question operator is difficult to establish; we will postpone this issue to Section 11.2.1. V2-clauses arise if some constituent is moved into the specifier position of CP, that is, the clause-initial position: the movement operation involved is used to derive various different kinds of constructions like the topicalization construction in (6b) and the wh-question in (6c); the traces indicate the original position of the moved phrase.

6
a. Heeft Jan dat boek met plezier gelezen?
V1; yes/no-question
  has  Jan that book  with pleasure  read
  'Has Jan enjoyed reading that book?'
b. Dat boeki heeft Jan ti met plezier gelezen.
V2; topicalization
  that book  has  Jan  with pleasure  read
  'That book, Jan has enjoyed reading.'
c. Welk boeki heeft Jan ti met plezier gelezen?
V2; wh-question
  what book  has  Jan  with pleasure  read
  'Which book has Jan enjoyed reading?'
[+]  III.  The clause-initial position contains at most one constituent

Consider again the representation in (4), repeated below as (7). Functional elements like T and C are generally assumed to contain certain semantic and morphosyntactic features. The functional element T(ense), for example, is normally assumed to contain the feature ±finite; this verbal feature is what enables the movement of the finite verb into T, as depicted in representation (7). A positive value for this feature enables T to assign nominative case to the subject of the clause, and it is assumed that this morphosyntactic relation between T and the subject enables the latter to be moved into the specifier position of T; we refer the reader to Section 9.5 for arguments showing that the subject is base-generated in a VP-internal position.

7

For our present discussion it is important to emphasize that the relation between the T-head and the subject is unique: a finite clause has (at most) one nominative argument. In the active clause in (8a) nominative case is assigned to Peter/hij and in the passive clause in (8b) it is assigned to Marie/zij, but there are no clauses with two nominative nominal arguments: *Hij bezocht zij'*He visited she'.

8
a. Peter/Hij heeft gisteren Marie/haar bezocht.
  Peter/he  has  yesterday  Marie/her  visited
  'Pete/He visited Marie/her yesterday.'
b. Marie/zij werd gisteren door Peter/hem bezocht .
  Marie/she  was  yesterday  by Peter/him  visited
  'Marie/she was visited by Peter/him yesterday.'

It is often assumed that the element C has features related to the illocutionary force of the clause: the feature ±q, for example, may determine whether we are dealing with a declarative or an interrogative clause. Contrary to ±finite, the feature ±q has no overt morphological manifestation on the verb in Dutch but it does affect the morphological form of the complementizer: the feature -q requires it to be spelled-out as dat'that' while +q requires it to be spelled-out as of'if'.

9
a. Marie zegt [CP dat[-Q] Peter het boek met plezier gelezen heeft].
  Marie  says  that  Peter the book  with pleasure  read  has
  'Marie says that Peter has enjoyed reading the book the book.'
b. Marie vraagt [CP of[+Q] Peter het boek met plezier gelezen heeft].
  Marie  asks   if  Peter the book  with pleasure  read  has
  'Marie is asking whether Peter has enjoyed reading the book the book.'

The examples in (10) show that the value of the feature ±q also determines what element may occupy the specifier position of CP in main clauses: while the (a)-examples show that it is possible to topicalize the direct object het boek or the indirect object aan Marie in the declarative clauses, the (b)- and (c)-examples show that topicalization is excluded in interrogative clauses; the feature +q only allows the specifier of CP to be filled by a wh-phrase. Note that (10b'&c') are (marginally) acceptable as echo-questions but this is of course not the reading intended here.

10
a. Dit boeki heeft Peter ti aan Marie aangeboden.
  this book  has  Peter  to Marie  prt.-offered
  'This book, Peter has offered to Marie.'
a'. Aan Mariei heeft Peter dit boek ti aangeboden.
  to Marie  has  Peter this book prt.-offered
  'To Marie, Peter has offered this book.'
b. Welk boeki heeft Peter ti aan Marie aangeboden?
  which book  has  Peter  to Marie  prt.-offered
  'Which book has Peter offered to Marie?'
b'. * Aan Mariei heeft Peter welk boek ti aangeboden?
  to Marie  has  Peter  which book  prt.-offered
c. Aan wie heeft Peter dit boek ti aangeboden?
  to who  has  Peter this book  prt.-offered
  'To whom has Peter offered this book?'
c'. * Dit boek heeft Peter ti aan wie aangeboden?
  this book  has  Peter  to who  prt.-offered

The examples in (11) further show that the specifier position of CP can contain at most one constituent; it is impossible to move more than one constituent into the clause-initial position. First, although the (a)-examples in (10) have shown that the direct and the indirect object can both be topicalized, example (11a) shows that they cannot be topicalized simultaneously. Second, although example (11b) shows that a clause may contain more than one wh-phrase, example (11b') shows that it is not possible to place more than one wh-phrase in its clause-initial position.

11
a. * Dit boeki aan Mariej heeft Jan titj aangeboden.
  this book  to Marie  has  Jan  prt.-offered
b. Welk boeki heeft Jan ti aan wie aangeboden?
  which book  has  Jan  to who prt.-offered
  'Which book did Jan offer to whom?'
b'. * Welk boeki aan wiej heeft titj Jan aangeboden?
  which book  to who  has  Jan  prt.-offered

The examples in (10) and (11) show that the specifier position of CP resembles the specifier position of T in that it can be filled by at most one constituent which is compatible with its feature specification: like the specifier of T[+finite] can only be occupied by a nominative argument, the specifier of C[+Q] can only be occupied by a wh-phrase. Note that the C-feature +q postulated in this subsection may be part of a larger set of features, as the constituents in clause-initial position may have a variety of special semantic functions; we return to this in Section 11.3.

[+]  IV.  The syntactic function of the constituent in clause-initial position

The fact illustrated in (11) that the clause-initial position may contain at most one constituent underlies the standard Dutch constituency test: anything that may occur in clause-initial position can be analyzed as a constituent. The utility of this test is based on the fact that virtually all clausal constituents can occupy this position. The examples in (12), for instance, show that topicalization and question formation affect arguments and adverbial phrases alike.

12
a. Jan zal morgen dat boek lezen.
  Jan will  tomorrow  that book  read
  'Jan will read that book tomorrow.'
b. Dat boeki zal Jan morgen ti lezen.
object
  that book  will  Jan tomorrow  read
  'That book, Jan will read tomorrow.'
b'. Wati zal Jan morgen ti lezen?
  what  will  Jan tomorrow  read
  'What will Jan read tomorrow?'
c. Morgeni zal Jan ti dat boek lezen.
adverbial phrase
  tomorrow  will  Jan  that book  read
  'Tomorrow Jan will read that book.'
c'. Wanneeri zal Jan ti dat boek lezen?
  when will  Jan  that book  read
  'When will Jan read that book?'

The examples in (13) show that complementives can also be placed in clause-initial position. For the sake of brevity, (13) illustrates this for wh-questions only, but similar examples are common in topicalization constructions as well; cf., e.g., Boven mijn bed hang ik jouw schilderij'Over my bed, I will hang your painting.'

13
a. Ik wil dierenarts worden. Wati wil jij ti worden?
  want  vet  become  what  want  you  become
  'I want to be a vet. What do you want to be?'
b. Ik vond de film saai. Hoei vond jij hem ti?
  I found the movie  boring.  how  found  you  him
  'I thought the movie boring. What did you think of it?'
b'. Ik hang jouw schilderij boven mijn bed. Waari hang jij het mijne ti?
  hang  your painting  above my bed  where  hang  you  the mine
  'Iʼll hang your painting over my bed. Where will you hang mine?'

It should be noted however, that the clause-initial position is not only accessible to clausal constituents, but may sometimes also contain parts of clausal constituents. This is illustrated in the (a)-examples in (14) for so-called R-extraction: although the primeless example shows that prepositional objects are normally wh-moved as a whole, the primed example shows that they can easily be split if they have the pronominalized form waar+P; see Section P5 for extensive discussion. Example (14b) shows the same for so-called wat voor-phrases; cf. N4.2.2.

14
a. <Naar> wie zoek je <*naar>?
  for  who  look  you
  'Who are you looking for?'
a'. Waar <naar> zoek je <naar>?
  where  for  look  you     for
  'What are you looking for?'
b. Wat <voor een boek> wil je <voor een boek> lezen?
  what    for a book  want  you  read
  'What kind of a book do you want to read?'

We must therefore be aware not to jump to the conclusion that we are dealing with a clausal constituent if a certain string of words occurs in clause-initial position: all we can conclude is that we are dealing with a constituent, which may be a clausal constituent but which may also be a subpart of clausal constituent.

[+]  V.  Clause-initial constituents are semantically marked

The previous subsection has shown that there are no syntactic restrictions on the constituent in clause-initial position, that is, the specifier position of CP; in principle any clausal constituent may be placed in this position. In this respect, the specifier position of CP is of an entirely different nature than the specifier of TP, which is a designated position of the subject. The movements involved in filling these specifiers are therefore also of an entirely different nature, which is sometimes expressed by saying that there is a distinction between A- and A'-movement. A(rgument)-movement is restricted to the nominal arguments, that is, subjects and direct/indirect objects. These movements are triggered by morphosyntactic features like ±finite or ±agreement, which play a role in syntactic relations like structural case (nominative, accusative and dative) assignment and subject/object-verb agreement. A'-movements are not restricted to nominal arguments and are not triggered by morphosyntactic but by semantic features. Features that may play a role in topicalization constructions are the features ±topic and ±focus. The feature +topic introduces the clause-initial constituent as the active discourse topic. An example such as (15a) introduces the referent of the direct object as a (new) discourse topic and it is consequently likely that in a follow-up sentence more information will be provided about this referent. The feature ±focus marks the clause-initial constituent as noteworthy in some sense, which is emphasized by the fact that this constituent is normally assigned extra accent (indicated here by small caps). Example (15b), for example, contrasts the referent of the clause-initial constituent with other entities in a contextually given set.

15
a. Peteri heb ik nog niet ti gesproken. Hij is nog op vakantie.
  Peter  have  not  yet  spoken  he  is still  on holiday
  'As for Peter, I havenʼt spoken to him yet. Heʼs still on holiday.'
b. Peteri heb ik nog niet ti gesproken (maar de anderen wel).
  Peter  have  not  yet  spoken   but  the others  aff
  'Peter, I havenʼt spoken to yet, but I did speak to the others.'

      The fact that topicalization does not occur in embedded clauses suggests that the features ±topic and ±focus can be found on the C-heads of main clauses only. This does not hold for the feature ±Q that we find on the C-heads of interrogative clauses, as is clear from the fact illustrated in (16) that such clauses can also be embedded.

16
a. Ik weet niet [CP of[+Q] ik dit boek zal lezen].
  know  not  if  this book  will  read
  'I donʼt know if Iʼll read this book.'
b. Ik weet niet [CP welk boek (of[+Q]) ik ti zal lezen].
  know  not  which book   if  will  read
  'I donʼt know which book Iʼll read.'

Observe that the interrogative complementizer of'if' is optional in examples such as (16b), which is related to the fact that there is a certain preference for not pronouncing the complementizer if the clause-initial position is filled. This phenomenon is also found in other languages; see, e.g., Chomsky & Lasnik (1977), who account for this by means of the so-called doubly-filled-compfilter, and Pesetsky (1997/1998), who provides an account in terms of optimality theory.
      There are also features like ±relative that occur in embedded clauses only. This feature creates relative clauses and can be held responsible for the movement of relative pronouns into clause-initial position. The percentage signs in the examples in (17) express that the complementizer dat[+rel] is normally not pronounced in Standard Dutch but that it was possible in Middle Dutch and is still possible in various present-day Dutch dialects; see, e.g., Pauwels (1958), Dekkers (1999:ch.3) and references cited there.

17
a. de brief [CP diei (%dat[+rel]) ik gisteren ti ontvangen heb].
  the letter  which     that  yesterday  received  have
  'the letter which I received yesterday'
b. de plaats [CP waari (%dat[+rel]) ik ga ti slapen]
  the place  where      that  go  sleep
  'the place where Iʼm going to sleep'
[+]  VI.  Subject-initial sentences are special

The previous subsection has shown that clause-initial constituents normally play a specific information-structural role (wh-phrase, topic, focus, etc.) in the clause. This was confirmed by the results of a recent corpus-study: "Non-subject material in the Vorfeld (= clause-initial position) is characterized by its (relative) importance." (Bouma 2008). The reason for providing this quote is that Bouma also found that this general characterization does not extend to subject-initial clauses: these are special in that they are normally the most unmarked way of asserting a proposition. That subject-initial clauses are unmarked is clear from the fact that they are generally used if the full sentence consists of new information: the word order in example (18a) is the one we typically get as an answer to the question Wat is er gebeurd?'What has happened?'. This raises the question as to whether subject-initial main clauses have the same overall structure as other V2-constructions, as is assumed in the more traditional versions of generative grammar where an example such as (18a) is derived as in (18a).

18
a. Marie heeft haar boek verkocht.
  Marie  has  her book  sold
  'Marie has sold her book.'
b.

If the movement into the specifier of CP is indeed motivated by some semantic feature, the fact that (18a) is the unmarked way of expressing the proposition have read(Marie,this book) would be quite surprising. Furthermore, Section 9.3 has shown that there are various other conspicuous differences between clause-initial subjects and other topicalized phrases. The most conspicuous difference is that the former can be a phonetically reduced pronoun, but the latter cannot. Consider the examples in (19). The primeless examples show that the subject can be clause-initial regardless of its form: it can be a full noun phrase like Marie, a full pronoun like zij or a phonetically reduced pronoun like ze. The primed examples show that topicalized objects are different: topicalization is possible if it has the form of a full noun phrase like Peter or a full pronoun like hem, but not if it has the form of the weak (phonetically reduced) pronoun 'm. The reason for this is that while topicalized objects must be accented (which is indicated by means of small caps) clause-initial subjects can remain unstressed.

19
a. Marie helpt Peter/hem/'m.
  Marie helps  Peter/him/him
  'Marie is helping Peter/him.'
a'. Peter helpt Marie/zij/ze.
  Peter  helps  Marie/she/she
  'Peter, Marie/she is helping.'
b. Zij helpt Peter/hem/'m
  she  helps  Peter/him/him.
  'Sheʼs helping Peter/him.'
b'. Hem helpt Marie/zij/ze.
  him  helps  Marie/she/she
  'Him, Marie/she is helping.'
c. Ze helpt Peter/hem/'m
  she  helps  Peter/him/him.
  'Sheʼs helping Peter/him.'
c'. * ʼM helpt Marie/zij/ze.
  him  helps  Marie/she/she

That topicalized phrases must be accented can also be illustrated by the examples in (20). The (a)-examples show that while the adverbial pro-form daar'there' can readily be topicalized, the phonetically reduced form er cannot. The (b)-examples illustrate the same thing for cases in which these elements function as the nominal part of a pronominal PP.

20
a. Jan heeft daar/er gewandeld.
  Jan has  there/there  walked
  'Jan has walked there.'
a'. Daar/*Er heeft Jan gewandeld.
  there/there  has  Jan walked
b. Jan heeft daar/er mee gespeeld.
  Jan has there/there with played
  'Jan has played with that/it.'
b'. Daar/*Er heeft Jan mee gespeeld.
  there/there  has  Jan with  played

We can readily account for these differences between subject-initial main clauses and other types of V2-clauses if we assume that these have the two different representations in (21): if sentence-initial subjects are not topicalized, there is no reason to expect that such constructions give rise to a marked interpretation or require a special intonation pattern.

21
a. Subject-initial main clause
b. Topicalization and question formation in main clause

Accepting the two structures in (21) would also make it possible to account for the contrast in verbal inflection in the examples in (22) by making the form of the finite verb sensitive to the position it occupies; if the verb is in T, as in (22a), second person singular agreement is realized by means of a -t ending, but when it is in C, as in (22b&c), it is realized by means of a null morpheme.

22
a. Jij/Je loop-t niet erg snel.
  you/you  walk-2sg  not  very fast
  'You donʼt walk very fast.'
b. Erg snel loop-Ø jij/je niet.
  very fast  walk-2sg  you/you  not
  'You donʼt walk very fast.'
c. Hoe snel loop-Ø jij/je?
  how fast walk-2sg  you/you
  'How fast do you walk?'

The discussion above thus shows that there are good reasons not to follow the traditional generative view that Dutch main clauses are always CPs; subject-initial main clauses may be special in that they are TPs. This hypothesis may help us to account for the following facts: (i) subject-initial clauses are unmarked assertions, (ii) sentence-initial subjects can be a reduced pronoun and (iii) subject-verb agreement may be sensitive to the position of the subject.

[+]  VII.  Main versus embedded clauses

There is a conspicuous difference between the clause-initial positions of main and embedded finite declarative clauses: the examples in (23) show that while the former are normally filled by the subject or some topicalized phrase, the latter are normally empty.

23
a. Jan heeft Vaslav van Arthur Japin gelezen.
  Jan has  Vaslav by Arthur Japin  read
  'Jan has read Vaslav by Arthur Japin.'
a'. Vaslav van Arthur Japin heeft Jan gelezen.
  Vaslav by Arthur Japin  has  Jan read
  'Vaslav by Arthur Japin Jan has read.'
a''. * Ø heeft Jan Vaslav van Arthur Japin gelezen.
  has  Jan  Vaslav by Arthur Japin  read
b. Ik denk [CP Ø dat [TP Jan Vaslav van Arthur Japin gelezen heeft]].
  think  that  Jan  Vaslav by Arthur Japin  read  has
  'I think that Jan has read Vaslav by Arthur Japin.'
b'. * Ik denk [CP Jani (dat) [TPti Vaslav van Arthur Japin gelezen heeft]].
  think  Jan  that  Vaslav by Arthur Japin  read  has
b'' * Ik denk [CP Vaslav van Arthur Japini (dat) [TP Jan ti gelezen heeft]].
  think  Vaslav by Arthur Japin   that  Jan  read  has

Observe that we placed the complementizers in the primed (b)-examples within parentheses because we have seen that the phonetic content of a complementizer is often omitted if the specifier of CP is filled by phonetic material; see the discussion of the doubly-filled-compfilter in Subsection V.
      Such a difference between finite main and dependent clauses does not arise in the case of interrogative clauses. The examples in (24) show that the initial position is phonetically empty in yes/no-questions but filled by some wh-phrase both in main and in embedded clauses.

24
a. Ø Heeft Jan Vaslav van Arthur Japin gelezen?
  has  Jan  Vaslav by Arthur Japin  read
  'Has Jan read Vaslav by Arthur Japin?'
a'. Wati heeft Jan ti gelezen?
  what  has  Jan  read
  'What has Jan read?'
b. Ik weet niet [CP Ø of [TP Jan Vaslav van Arthur Japin gelezen heeft]].