• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
9.3.The clause-initial position

Section 9.2 has shown that finite verbs occupy the second position in main clauses, that is, that they can be preceded by at most one constituent. This constituent can be the subject of the clause or a topicalized phrase in declarative clauses, or a wh-phrase in interrogative clauses.

a. Mijn zuster heeft dit boek gelezen.
  my sister  has  this book  read
  'My sister has read this book.'
b. Dit boek heeft mijn zuster gelezen.
  this book  has  my sister  read
  'This book, my sister has read.'
c. Welk boek heeft mijn zuster gelezen?
  which book  has  my sister  read
  'Which book has my sister read?'

The standard generative analysis of examples such as (31) is that they all involve movement of some constituent from a clause-internal position into the specifier of CP, that is, the position preceding the finite verb in the C-position in the structure in (32). By assuming that specifier positions of any projection (that is, the positions to the immediate left of the heads C, T, X and V) can contain at most one constituent, we derive the verb-second effect.


The following subsections will briefly discuss the three construction types in (31) both in main and in embedded clauses. This discussion will lead to a slightly revised version of the proposal in (32).

[+]  I.  Wh-movement

There are two types of questions: so-called yes/no-questions such as (33a), which request the addressee to provide the speaker with information about the truth of the proposition expressed by the clause, and wh-questions such as (33b), which request the addressee to provide the speaker with some piece of missing information related to the proposition. The clause-initial position of yes/no-questions remains phonetically empty (although it is perhaps lexically filled by a phonetically empty question operator). In wh-questions, the wh-phrase is normally moved into clause-initial position.

Main clauses
a. Heeft mijn zuster dit boek gelezen?
  has  my sister  this book  read
  'Has my sister read this book?'
b. Wanneer heeft mijn zuster dit boek gelezen?
  when  has  my sister  this book read
  'When did my sister read this book?'

The hypothesis in (32), that the wh-phrase is moved into the specifier of CP, leads to the prediction that wh-phrases also precede the C-position in embedded questions. Although in the more formal registers complementizers are normally not phonetically realized in embedded wh-questions, it is easily possible to do so in colloquial speech. Example (34a) first shows that embedded yes/no-questions differ from embedded declarative clauses in that the complementizer does not have the form dat'that' but the form of'whether'. The (b)-examples in (34) show that this complementizer can be optionally realized in embedded wh-questions, and must then follow the wh-phrase in clause-initial position; see Barbiers (2005: Section, where it is also shown that in some regions of finds an alternative realization as of dat or dat; see also Hoekstra & Zwart (1994), Sturm (1996) and Zwart & Hoekstra (1997) on the question as to whether of dat should be analyzed as a compound or as two separate words.

Embedded clauses
a. Jan vroeg [CP of [TP mijn zuster dit boek gelezen heeft]].
  Jan asked  comp  my sister  this book  read  has
  'Jan asked whether my sister has read this book.'
b. Jan vroeg [CP wiei (of) [TPti dit boek gelezen heeft]].
  Jan asked  who  comp  this book  read  has
  'Jan asked who has read this book.'
b'. Jan vroeg [CP wati (of) [TP mijn zuster ti gelezen heeft]].
  Jan asked  what  comp  my sister  read  has
  'Jan asked what my sister has read.'

Example (35a) shows that wh-movement need not necessarily target the clause-initial position of the embedded clause, but that it is also possible to move a wh-phrase from the embedded clause into the clause-initial position of the sentence; we will refer to this as longwh-movement. This is excluded, however, if the embedded clause is itself an embedded question: examples (35b&c) show that both yes/no- and wh-questions constitute a so-called island for wh-extraction from the embedded clause; note that some (but not all) speakers report a slight acceptability contrast between the two examples in that (35b) is slightly less degraded than (35c).

Long wh-movement
a. Wati denk je [dat mijn zuster ti gelezen heeft]?
  what  think  you  comp  my sister  read  has
  'What do you think that my sister has read?'
b. * Wati vroeg Jan [of mijn zuster ti gelezen heeft]]?
  what  asked Jan  comp  my sister  read  has
c. * Watj vroeg Jan [CP wiei (of) [TPtitj gelezen heeft]]?
  what  asked Jan  who  comp  read  has

The examples in (35) are normally taken to show that wh-phrases originating in embedded clauses cannot be moved into the sentence-initial position in one fell swoop; they can only be extracted from embedded clauses via the specifier position of the embedded CP, which thus functions as an "escape hatch". As a result, "long" movement can be reinterpreted as a series of movements that apply in a local/clause-bound fashion; cf. the schematic representation in (36), and Chomsky (1977) for detailed discussion. The claim is that this escape hatch is only available when the embedded clause is declarative: the position must be filled syntactically by a phonetically empty question operator (or perhaps remain empty) in yes/no-questions and be filled by some other interrogative phrase in wh-questions.


Since this will become relevant in the following subsections, we note here that Dutch shows a marked difference from English in that it allows subjects to be extracted from embedded clauses introduced by a complementizer; cf. Bennis (1986:ch.3). This is illustrated in (37). If the whole embedded clause expresses new information, as in (37a), subject extraction normally requires the presence of the expletive element er; this expletive is optional when the embedded clause contains some presupposed phrase, as dit boek'this book' in (37b), and gives rise to a degraded result when the presupposed phrase is pronominal, as het'it' in (37c).

a. Wiei denk je [dat er ti komt]?
  who  think  you   that  there  comes
  'Who do you think (*that) is coming?'
b. Wiei denk je [dat (er) ti dit boek gelezen heeft].
  who  think  you   that  there  this book  read  has
  'Who do you think (*that) has read this book?'
c. Wiei denk je [dat (*er) ti het gelezen heeft].
  who  think  you   that  there  it  read  has
  'Who do you think (*that) has read this book?'
[+]  II.  Topicalization

Topicalization is typically restricted to main clauses in Standard Dutch. The examples in (38) show that it is excluded in embedded clauses, regardless of whether the complementizer is phonetically realized or whether the topicalized phrase precedes or follows the declarative complementizer.

a. * Jan zei [CP dit boeki (dat) [mijn zuster ti gelezen had]].
  Jan said  this book  comp   my sister  read  has
b. * Jan zei [CP (dat) dit boeki mijn zuster ti gelezen had]].
  Jan said  comp  this book  my sister  read  had

That topicalization is not possible in embedded clauses in Standard Dutch is clearly related to the fact that it does not allow embedded verb-second: German, as well as a large subset of the Dutch varieties that do allow embedded verb-second, also allows embedded topicalization: see Haider (1985/2010) for German and Barbiers (2005: Section for the relevant non-standard Dutch varieties. Note in passing that Dutch topicalization seems rather different from English topicalization, which can give rise to English examples of the type in (38b): cf. I believe that this book you should read, taken from Lasnik & Saito (1992:76).
      The cases in (39) show that, although topicalization is not possible within embedded clauses, it is possible to topicalize constituents from embedded clauses by placing them into sentence-initial position. The fact that example (39a) is possible (although perhaps somewhat marked) shows again that subjects can be extracted from embedded declarative clauses introduced by a complementizer.

a. Mijn zusteri zei Jan [dat ti dit boek gelezen had].
  my sister  said  Jan comp  this book  read  had
b. Dit boeki zei Jan [dat mijn zuster ti gelezen had].
  this book  said  Jan   that  my sister  read  has

The examples in (40) show that topicalization is impossible if the embedded clause is interrogative; this suggests that, just as in the case of wh-movement, topicalization of some element from the embedded clause into sentence-initial position must proceed via the specifier position of the embedded CP; cf. the schematic representation in (36).

a. * Mijn zusteri vroeg Jan zich af [welk boekj (of) titj gelezen had].
  my sister  wondered  Jan refl  prt.  which book comp  read  had
b. * Dit boekj vroeg Jan zich af [wiei (of) titj gelezen had].
  this book  wondered  Jan refl  prt.   who comp  read  has
[+]  III.  The position of the subject

The representation in (41b) sketches the standard generative analysis of subject-initial declarative main clauses such as (41a). First, it is assumed that the specifier position of TP is the canonical subject position; it is the position where the subject is traditionally taken to be assigned nominative case by the feature +finite of T. Second, since verb-second places the finite verb in C and C precedes the regular subject position, the subject must be topicalized into the specifier of CP in order to precede the finite verb.

a. Mijn zuster/Zij/Ze had dit boek gelezen.
  my sister/she/she  had this book  read
  'My sister/she had read this book.'

Note in passing that we accept the widely supported claim (from Travis 1984:131) that the verb moves to C via all intermediate head positions, for which reason we will from now on speak of V-to-C, V-to-T, V-to-X, etc. Verb movement via the intermediate T-position is generally motivated by stating that this movement can be triggered by the tense and/or agreement features in this position. The movement of the verb via the (as yet undetermined) X-position depicted in (41b) is provided for theory-internal reasons but need not concern us now; for this reason we will not include this movement in the representations in Subsection IV; the availability of V-to-C, however, will become crucial in the discussion given there.
      If the derivation in (41) is correct, we would expect the placement of subjects to be subject to similar restrictions as regular topicalization. At first sight, this expectation seems to be borne out, given that Subsection II has already shown that embedded subjects like mijn zuster'my sister' may be placed in sentence-initial position; cf. (42a). However, this cannot be an across-the-board conclusion as weak pronominal subjects show a conspicuously different behavior; the examples in (42b&c) show that, although topicalization of embedded subject pronouns seems possible if they are strong (that is, phonetically non-reduced) and contrastively stressed, it is clearly excluded when they are weak (phonetically reduced).

a. Mijn zusteri zei Jan [dat ti dit boek gelezen had].
  my sister  said  Jan  comp  this book  read  had
b. (?) Ziji zei Jan [dat ti dit boek gelezen had].
  she  said  Jan comp  this book  read  had
c. * Zei zei Jan [dat ti dit boek gelezen had].
  she  said  Jan comp  this book  read  had

The topicalization behavior of subject pronouns thus strongly resembles that of object pronouns: whereas strong object pronouns do allow topicalization when they are contrastively stressed, weak object pronouns do not; cf. Huybregts (1991).

a. Marie/Ze heeft Peter/hem/ʼm gekust.
  Marie/she  has  Peter/him/him  kissed
b. Peter/Hem/*ʼm heeft Marie/ze ti gekust.
  him/him/him  has  Marie/she  kissed

Since example (41a) has shown that weak subject pronouns of main clauses are perfectly acceptable in sentence-initial position, the discussion above suggests that the topicalization approach to subject-initial clauses cannot be (fully) correct; let us consider an alternative approach in the following subsection.

[+]  IV.  An alternative analysis

The previous subsections have shown that the different types of sentence-initial elements in main clauses exhibit different syntactic behavior when extraction from embedded clausal complements is taken into account. The main findings are summarized in Table (44); this subsection especially focuses on the fact that subjects can only be extracted from embedded clauses and placed in sentence-initial position if they are non-pronominal or contrastively stressed; weak embedded subject pronouns do not occur sentence-initially.

The syntactic distribution of interrogative, topicalized and subject phrases
  sentence-initial embedded clauses
    extraction clause-initial
interrogative phrases + + +
topicalized phrases + +
subjects + non-pronominal: +
stressed pronouns: (?)
weak pronouns: —

Table (44) strongly suggests that the standard assumption that subject-initial sentences are derived by means of topicalization of the subject, as in (41), is not correct. However, if we adopt the structure in (10), repeated in a somewhat revised form in (45), we can readily account for the difference in extraction behavior of pronominal subjects on the one hand, and interrogative and topicalized phrases on the other, by assuming that subject-initial sentences are not CPs but TPs (which is the traditional standard assumption for English).

[CP ... C [TP Subject T [XP ... X [VP ... V ...]]]]

The verb-second property of Dutch can then be derived by assuming the analyses in (46); cf. Travis (1984) and Zwart (1997). The V-to-T movement in the subject-initial sentence in (46a) can be motivated by appealing to the earlier assumption that T contains the tense and/or agreement features of the verb. The subsequent T-to-C movement of the verb into the C-position in (46b) can be motivated by assuming that C contains certain illocutionary features. By assuming that declarative force is assigned as a default value, the absence of the CP-layer in subject-initial clauses such as (46a) can also be accounted for.

a. Subject-initial sentence
b. Topicalization and question formation

Obviously, the analysis in (46) raises the question as to why the verb does not move to T in embedded clauses, thus giving rise to a word order (found in English) in which the subject is sandwiched between the complementizer and the finite verb: *dat mijn broer heeft dit boek gelezen. The assumption that verb movement is forced by the language-specific surface condition that the highest functional head in an extended projection must be lexically filled would solve this. It predicts that when the C-position is filled by the complementizer, the verb can remain in its original position within the lexical domain. If this assumption is acceptable, verb movement can be functionally motivated by saying that each clause must be marked as such by a complementizer or a finite verb in second position. Since further discussion would take us into theory-internal argumentation, we will not elaborate here but refer the reader to Zwart (2001) and Broekhuis (2008: Section 4.1) for further discussion.
      We should point out, however, that accepting the two structures in (46) would make it possible to account for the contrast in verbal inflection in the examples in (47) by making the form of the finite verb sensitive to the position it occupies; if the verb is in T, as in (47a), second person singular agreement is realized by means of a -t ending, but if it is in C, as in (47b&c), it is realized by means of a null morpheme.

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?'

Given that Dutch exhibits morphological alternations like these with second-person singular subjects only, we will not digress on this point here, but refer the reader to Zwart (1997), Postma (2011) and Barbiers (2013) for a discussion of language varieties which more generally exhibit similar contrasts in inflection.

[+]  V.  Conclusion

This section has discussed the clause-initial position, which can be filled by means of topicalization and wh-movement. The two movement types differ, however, in that topicalization always targets the sentence-initial position, whereas wh-movement may also target the initial position of embedded clauses. Traditionally, subject-initial main clauses are also analyzed as topicalization constructions; the verb is moved into the C-position of the clause and the subject must therefore be subsequently moved into the specifier of CP. The fact that topicalization of weak (phonetically reduced) pronouns is normally not possible sheds doubt on this view, given that weak subject pronouns can readily occur sentence-initially, thus giving rise to the claim that subject-initial main clauses can be TPs.

  • Barbiers, Sjef2013Geography and cartography of the left periphery. The case of Dutch and German imperativesCarrilho, Ernestina, Alverez, Xose & Magro, Catarina (eds.)Current approaches to limits and areas in dialectologyNewcastle upon TyneCambridge Scholars Publishing267-292
  • Barbiers, Sjef, Bennis, Hans, Vogelaer, Gunther de, Devos, Magda & Ham, Margreet van de2005Syntactic atlas of the Dutch dialectsAmsterdamAmsterdam University Press
  • Barbiers, Sjef, Bennis, Hans, Vogelaer, Gunther de, Devos, Magda & Ham, Margreet van de2005Syntactic atlas of the Dutch dialectsAmsterdamAmsterdam University Press
  • Bennis, Hans1986Gaps and dummiesDordrechtForis Publications
  • Broekhuis, Hans2008Derivations and evaluations: object shift in the Germanic languagesStudies in Generative GrammarBerlin/New YorkMouton de Gruyter
  • Chomsky, Noam1977On <i>wh</i>-movementCulicover, Peter W., Wasow, Thomas & Akmajian, Adrian (eds.)Formal syntaxNew YorkAcademic Press71-132
  • Haider, Hubert1985V-second in GermanHaider, Hubert & Prinzhorn, Martin (eds.)Verb second phenomena in Germanic languagesDordrecht/RivertonForis Publications49-75
  • Haider, Hubert2010The syntax of GermanCambridgeCambridge University Press
  • Hoekstra, Eric & Zwart, Jan-Wouter1994De structuur van CP. Functionele projecties voor topics en vraagwoorden in het NederlandsSpektator23191-212
  • Huybregts, Riny1991CliticsModel, Jan (ed.)Grammatische analyseDordrechtICG Printing
  • Lasnik, Howard & Saito, Mamoru1992Move α. Conditions on its applications and outputLinguistic Inquiry MonographsCambridge, MA/London
  • Postma, Gertjan2011Het verval van het pronomen <i>du</i>- dialectgeografie en de historische syntaxisNederlandse Taalkunde1656-87
  • Sturm, Arie1996Over functionele projectiesNederlandse Taalkunde1191-206
  • Travis, Lisa1984Parameters and effects of word order variationMITThesis
  • Travis, Lisa1984Parameters and effects of word order variationMITThesis
  • Zwart, Jan-Wouter1997Morphosyntax of verb movement. A minimalist approach to the syntax of DutchDordrechtKluwer Academic Publishers
  • Zwart, Jan-Wouter1997Morphosyntax of verb movement. A minimalist approach to the syntax of DutchDordrechtKluwer Academic Publishers
  • Zwart, Jan-Wouter2001Syntactic and phonological verb movementSyntax434-62
  • Zwart, Jan-Wouter & Hoekstra, Eric1997Weer functionele projectiesNederlandse Taalkunde2121-132
report errorprintcite