DynaSAND

Syntactic Atlas of Dutch Dialects – Commentary – Volume II

Table of contents Volume II

                        1                      Verb clusters

                        1.1                   Introduction

                        1.1.1                Verb positions in Standard Dutch

                        1.1.2                Two-verb clusters

                        1.1.3                Three-verb clusters

                        1.1.4                Three-verb clusters expressing irrealis

                        1.1.5                Historical development

                        1.2                   Discussion of the literature

                        1.3                   Discussion of the maps

                        1.3.1                Two-verb clusters

                        1.3.1.1             Auxiliary BE_finite – V_participle (map 14a)

                        1.3.1.2             Auxiliary HAVE_finite – V_participle I & II (maps 14b, 15a)

                        1.3.1.3             Modal_finite – V_infinitive (map 15b)

                        1.3.1.4             Two-verb clusters – synthesis (map 16a)

                        1.3.2                Three-verb clusters

                        1.3.2.1             Modal_finite – Modal_infinitive – V_infinitive (map 17a)

                        1.3.2.2             Modal_finite – HAVE_infinitive – V_participle (map 17b)

                        1.3.2.3             Auxiliary BE_finite – Aspectual/Modal – V_infinitive

                        1.3.2.3.1          BE_finite – GO_{infinitive/participle} – V_infinitive (map 18a)

                        1.3.2.3.2          Heen ‘away’, te ‘to’ and IPP (maps 18b, 19a)

                        1.3.2.3.3          Perfective auxiliary BE in three-verb clusters (map 19b)

                        1.3.2.3.4          HAVE_finite – Modal_{infinitive/participle} – V_infinitive (map 20a)

                        1.3.2.4             Three-verb clusters – irrealis (map 20b)

                        1.3.3                Potential correlations

                        1.3.3.1             Two- and three-verb clusters – perfective I (map 21a)

                        1.3.3.2             Two- and three-verb clusters – perfective II (map 22a)

                        1.3.3.3             Two- and three-verb clusters – infinitive (map 23a)

                        1.3.3.4             Uniform order in three types of verb clusters (map 24a)

                        1.3.3.5             Systems of three-verb clusters (map 25a)

                        1.4                   Literature on verb clusters

Chapter 1

Verb clusters

1.1 Introduction

1.1.1 Verb positions in Standard Dutch

The Standard Dutch main clause has three positions for verbs. In V1 clauses, such as imperatives, yes-no questions and conditionals, the finite verb occurs in the first position (1a-c). In V2 clauses, usually declaratives or Wh-questions, the finite verb directly follows the first constituent (1d). Non-finite verbs, i.e., infinitives and participles, occur in a clause-final position (V_FINAL; 1e,f). In all three cases, this verb placement is obligatory, i.e., there are no alternative positions for the verbs.

(1)       a.         Koop   nu        een       krant!                                                              V1

                        buy      now     a          newspaper

                        ‘Buy a newspaper now!’

            b.         Koop   je         vandaag           een       krant?                                      V1

                        buy      you      today               a          newspaper

                        ‘Are you going to buy a newspaper today?’

            c.         Koop   je         een       krant,   neem    dan      chocola            mee.     V1

                        buy      you      a          paper, bring    then     chocolate         with

                        ‘If you buy a newspaper please bring chocolate.’

            d.         Je         vader   koopt   vandaag           een       krant.                           V2

                        your    father   buys    today               a          newspaper

                        ‘Your father will buy a newspaper today.’

            e.         Je         moet    vandaag           een       krant    kopen.                    V_FINAL

                        you      must    today               a          paper   buy

                        ‘You should buy a newspaper today.’

            f.          Hij       heeft    vandaag           een       krant    gekocht.                  V_FINAL

                        he        has       today               a          paper   bought

                        ‘He has bought a newspaper today.’

In Standard Dutch, subordinate clauses, verbs occur in clause-final position, whether finite or non-finite (2a,b). This means that when a subordinate clause contains both a finite and a non-finite verb, these two verbs occur at the end of the clause (2c,d).

(2)       a.         (Ik denk) dat    hij        vandaag           een       krant    koopt.         V_FINAL

                        I think     that he        today               a          paper   buys

                        ‘(I think) that he will buy a newspaper today.’

            b.         (Ik probeer) om     vandaag     een       krant    te         kopen.        V_FINAL

                        I’ll try           comp today        a          paper   to         buy

                        ‘(I try) to buy a newspaper today.’

            c.         (Ik denk) dat   hij     vandaag   een       krant    wil       kopen.        V_FINAL

                        I think      that he      today      a          paper   wants buy

                        ‘(I think) that he wants to buy a newspaper today.’

d.         (Ik denk) dat   hij     vandaag    een    krant    gekocht    heeft.       V_FINAL

                        I think      that he     today       a        paper   bought     has

                        ‘(I think) that he has bought a newspaper today.’

When there is no complementiser in a finite subordinate clause, the finite verb has to occur in V1 position. This happens, e.g., in conditional clauses (3a vs 3b).

(3)       a.         Als       je         geld      hebt,    (koop dan een krant).

                        if          you      money have     buy then a newspaper

                        ‘If you have got money, (buy a newspaper.)’

            b.         Heb     je         geld,     (koop dan een krant.)

                        have     you      money buy then a newspaper

                        ‘If you have got money, (buy a newspaper.)’

Thus, complementisers and finite verbs are in complementary distribution. Summarising, the rule for verb placement in Standard Dutch can be formulated as in (4), capturing both main and subordinate clauses:

(4)       All verbs occur in clause-final position. If the clause has no complementiser, the finite verb occurs in clause-initial position (i.e., V1 or V2).

1.1.2 Two-verb clusters

Verbs in clause-final position normally follow their complements, i.e., direct objects (5a), indirect objects (5b) and predicative complements (5c,d) precede the clause-final verb.

(5)       a.         Hij    heeft    vandaag    <een krant> gekocht    <*een krant>.          DO

                        he      has       today       <a paper>     bought      <a paper>

                        ‘He has bought a newspaper today.’

            b.         Hij    heeft    <haar>     een krant       gegeven    <*haar>.                  IO

                        he      has       <her>       a      paper     given        <her>

                        ‘He has given her a newspaper.’

            c.         Hij    heeft    de   krant <op tafel>    gelegd      <*op tafel>.       PRED

                        he      has       the paper <on table>    put           <on table>

                        ‘He has put the newspaper on the table.’

            d.         Hij    is         <verliefd>       geweest <*verliefd>.                        PRED

                        he      is         <in love>                      been         <in love>

                        ‘He has been in love.’

Not all complements show this behaviour, however. Prepositional objects can both precede and follow the verb (6a), sentential objects and to-infinitives always follow the verb (6b,c).

(6)       a.         Hij wil     altijd    <over de krant>     praten    <over de krant>.        PO

                        he   wants always <about the paper>             talk     <about the paper>

                        ‘He always wants to talk about the newspaper.’

            b.         Ik    heb    <*dat ik ‘n krant koop>    gezegd <dat ik ‘n krant koop>. SO

                        I      have <that I a paper buy>      said        <that I a paper buy>

                        ‘I have said that I will buy a newspaper.’

            c.         Ik    denk dat hij   het niet    <*te lezen>  hoeft    <te lezen>.       to-inf

                        I      think             that he                 it     not    <to read>                   needs   <to read>

                        ‘I think that he does not need to read it.’

In Standard Dutch, bare infinitival complements, i.e., infinitival complements without te ‘to’, may either precede or follow the auxiliary. This can be a modal auxiliary (moeten ‘must’, kunnen ‘can’, mogen ‘may’, zullen, ‘will’, willen ‘want’), an aspectual auxiliary (gaan ‘go’, komen ‘come’, blijven ‘keep’), a causative auxiliary (laten ‘let’, doen ‘do’) or a perception verb (zien ‘see’, horen ‘hear’) (7a-d). Dutch auxiliaries vary with respect to the freedom of bare infinitive placement and there is interspeaker variation. For example, for many speakers a bare infinitive preceding a perception verb is less acceptable than the same order with a modal or an aspectual auxiliary.

(7)       a.         Ik           vind     dat je            het       <lezen>    moet    <lezen>.

                        I   think that you     it       <read>     must <read>

                        ‘I think that you should read it.’

            b.         Ik denk  dat  ik  vandaag    niet    <werken>       ga <werken>.

                        I   think that      I      today          not <work>    go        <work>

                        ‘I think that I am not going to work today.’

            c.         Ik  denk dat  ik  de   deur  <verven> laat   <verven>.

                        I think   that I   the door <paint>    let     <paint>

                        ‘I think that I let paint the door.’

            d.         toen       ik    haar     <lopen>     zag           <lopen>

                        when     I      her       <walk>       saw <walk>

                        ‘when I saw her walking’

The freedom of ordering bare infinitives is related to the absence of te ‘to’. The auxiliary hoeven ‘need’ in (6c) arguably has properties that are very similar to modal auxiliaries (e.g., cluster formation, restrictions on complementation, ambiguity between deontic and epistemic interpretations), except that it takes a to-infinitive. Apparently, this minimal difference, the presence of te ‘to’, determines the place of the infinitival complement.

In addition to bare infinitives, there is a second class of verbal complements that may occur both before and after the selecting verb, i.e., the participles. They are selected by perfective auxiliaries (hebben ‘have’, zijn ‘be’), passive auxiliaries (worden ‘become’, zijn ‘be’) and krijgen ‘get’ (8a-e).

(8)       a.         Ik           denk       dat hij   de           krant <gelezen>  heeft     <gelezen>.

                        I   think        that     he      the         paper <read>     has       <read>

                        ‘I think that he has read the newspaper.’

            b.         Ik denk  dat  hij   <verdwenen>  is  <verdwenen>.

                        I   think        that     he      <disappeared>  is             <disappeared>

                        ‘I think that he has disappeared.’

            c.         Ik denk dat hij   <gevonden>    wordt     <gevonden>.

                        I   think        that     he      <found>          become      <found>

                        ‘I think that he will be found.’

            d.         Ik denk dat hij   <gevonden>    is <gevonden>.

                        I   think        that     he      <found>            is             <found>

                        ‘I think that he has been found.’

            e.         Ik denk dat je    een cadeau <aangeboden>   krijgt <aangeboden>.

                        I   think        that     you   a             present     <offered>                   get       <offered>

                        ‘I think that they will offer you a present.’

A general intuition is that in Standard Dutch the order of auxiliary and participle is completely free whereas in combinations of an auxiliary and an infinitive the order auxiliary – infinitive would be the most natural. It has also been claimed (cf. section 1.2) that the order participle – perfective auxiliary is more common with the auxiliaries zijn ‘be’ and worden ‘become’ than with the auxiliary hebben ‘have’. This may seem to suggest that in the order participle – BE, the participle is a predicative complement. Such a predicative complement would denote the state the subject is in as a result of the process denoted by the verb. In the order BE – participle, the participle would be a verbal complement, focussing on the process. As we have seen in (5d), non-verbal predicative complements always precede the selecting verb. However, a participle preceding BE is ambiguous between a predicative and a verbal complement. It is true that the position following BE strongly favours the verbal complement status, but even there predicative verbal complements are not always completely excluded (9).

(9)       Ik  weet  dat hij      al              jaren    <geblesseerd>   is    <geblesseerd>.

            I    know       that   he   already          years   <injured>      is             <injured>

            ‘I know that he has been injured for years now.’

The distinction between verbal complements and predicative verbal complements is not relevant for this atlas, as it only includes cases of auxiliary BE with a verbal participial complement.

Sequences of one or more verbs including auxiliaries as illustrated in (7-8) are called verb clusters, because they behave as units in many varieties of Dutch. In those varieties these verb clusters cannot be interrupted by constituents such as objects, adverbials etc. For constructions and dialects that are exceptions to this adjacency requirement see chapter 2. Importantly, verb clusters are not restricted to bare infinitives and participles. In many varieties of Dutch, to-infinitives can also occur in clusters with an adjacency requirement (10a). Whether this adjacency requirement holds depends on the matrix verb. For example, in Standard Dutch a matrix verb like proberen ‘try’ allows for constituents intervening between the verbs (10a), whereas the matrix verb durven ‘dare’ does not (10b) (cf. chapter 2 for constructions like this).

(10)     a.         Ik           denk       dat hij      <het    huis>        probeert          <het    huis>   te         verkopen.

                        I   think        that        he      the                house       tries       the                  house   to         sell

                        ‘I think that he is trying to sell the house.’

            b.         Ik           denk       dat hij      <het    huis>        durft <*het     huis>              te         verkopen.

                        I   think        that        he      the           house   dares      the     house to         sell

                        ‘I think that he will dare to sell the house.’

In this chapter, we restrict ourselves to word order variation in verb clusters without te ‘to’. As one of the cases of intralinguistic word order variation, the relatively free ordering of verbs in two-verb clusters of the type discussed above has attracted a lot of attention. One of the unusual properties is that difference in word order apparently does not correspond to a difference in meaning here, in violation of Von Humboldt’s (1836) Isomorphy Principle according to which natural languages show a strong tendency towards a one-to-one correspondence of form and meaning.

In Germanic, only OV (object-verb) languages show cluster formation and word order variation within these clusters. Thus, we find it in varieties of Afrikaans, German, Frisian and Dutch, but not in English and the Scandinavian languages, which are all VO.

Various linguistic and non-linguistic factors favouring a particular order have been identified in the literature (see section 1.2). For this atlas, we are interested in the geographical determinants of this word order variation. Unlike much of the literature, our goal is not to establish to what extent geography contributes to the word order variation in Standard Dutch. Rather, we are interested in the word order variation in the dialects themselves. We have systematically investigated two-verb clusters consisting of a finite modal and an infinitive and clusters with a perfective auxiliary and a participle. The test sentences are given in (11).

(11)     a.         Ze   weet    niet dat  Marie  gisteren    <gestorven> is <gestorven>.

                        she  knows not that Mary yesterday died              is  died

                        ‘She does not know that Mary died yesterday.’

            b.         Dat   is     de   man   die     ze      <geroepen>  hebben   <geroepen>.

                        that   is     the man   who       they              <called>            have                 <called>

                        ‘That is the man they called.’

            c.         Dat   is     de   man          die          het    verhaal   <verteld>   heeft    <verteld>.

                        that   is     the man          who   the           story   told                        has       told

                        ‘That is the man who told the story.’

            d.         Ik    vind   dat jij    het ook    niet <zien> mag     <zien>.

                        I      think that you it     also   not see       may     see

                        ‘I think that you should not see it too.’

In the descriptions of the maps, the linear order auxiliary - main verb is referred to as 1-2, where 1 is the hierarchically higher verb selecting 2 as its dependent. Conversely, the order 2-1 refers to a linear order in which the main verb precedes the selecting auxiliary. Since Pauwels (1953), the 1-2 order is called the red order in the literature because Pauwels used red symbols for it, while she used green symbols for the 2-1 order. Our maps follow this tradition.

1.1.3 Three-verb clusters

If a verb in a two-verb cluster can precede or follow the selecting auxiliary and if a verb cluster can contain more than one auxiliary, as is the case in all varieties of Dutch, there are six logically possible word orders for three-verb clusters.

(12)     a.         Ik    vind   dat iedereen      moet       kunnen    zwemmen        1-2-3

                        I      think        that everyone    must   can.inf             swim.inf

                        ‘I think that everyone should be able to swim.’

            b.         moet    zwemmen     kunnen                                                    1-3-2

                        must    swim.inf       can.inf

            c.         kunnen     zwemmen     moet                                                   2-3-1

                        can.inf     swim.inf      must

            d.         kunnen     moet        zwemmen                                                           2-1-3

                        can.inf     must        swim.inf

            e.         zwemmen       moet      kunnen                                                           3-1-2

                        swim.inf         must      can.inf

f.         zwemmen       kunnen moet                                                   3-2-1

            swim.inf         can.inf   must

Five of the six orders in (12) were tested in all locations. The order 2-1-3 in (12d) was not tested as the postal pilot survey had shown that this order does not occur in clusters with bare infinitives. In addition to (12), we tested the sentences in (13-15).

(13)     Mod – Aux – V

Jan    weet    dat hij   voor      3 uur    de wagen   moet       hebben   gemaakt.

John  knows        that he   before    3 hour  the car        must.fin       have.inf          made.pcp

            ‘John knows that he is supposed to have the car repaired before 3 o’clock.’

(14)     Aux (BE) – Asp – V

Ik      weet  dat  hij      is           gaan         zwemmen.                              

            I        know       that   he          is.fin      gone.pcp swim.inf

            ‘I know that he has gone swimming.’

(15)     Aux (HAVE) – Mod – V

Vertel    maar    niet    wie    zij        had          kunnen   roepen.

            tell         just      not    who       she  had.fin     can.inf   call.inf

            ‘Just don’t tell her who she could have called.’

If the claim referred to above is true that auxiliary-participle combinations show complete freedom of ordering whereas auxiliary-infinitive combinations have a preference for the 1-2 order, we can expect that the cluster types in (12)-(15) show different behaviour. The cluster in (12) contains two modals. The cluster in (13) has a modal as the highest auxiliary and the perfective auxiliary HAVE as the second highest auxiliary. The cluster in (14) has the perfective auxiliary BE as the highest auxiliary, followed by an aspectual auxiliary. The cluster in (15) has the perfective auxiliary HAVE as the highest auxiliary which is followed by a modal auxiliary. A further difference between these clusters is the Infinitivus pro Participio effect in (14) and (15) (IPP; cf. chapter 2), the phenomenon that a verb that is dependent on a perfective auxiliary does not have the expected participial form but occurs as an infinitive.

The aspectual verb gaan ‘go’ in (14) shows cross-dialectal variation. In a number of eastern dialects it requires a directional particle heen ‘towards’, whereas in a number of northern dialects it requires the infinitival marker te ‘to’ on its verbal complement (see section 1.3.2.3.2)

1.1.4 Three-verb clusters expressing irrealis

While the different word orders in the verb clusters discussed above do not correspond to differences in meaning, there is a construction in which this is the case. In Standard Dutch, irrealis is usually expressed by sentences of the type in (15), with a past tense perfective auxiliary as the higher and a modal infinitive as the lower auxiliary. In some southern dialects, this meaning is expressed by sentences such as (16a), with a past tense modal as the highest auxiliary and the infinitive of a perfective auxiliary as the second highest verb. In Standard Dutch, the hierarchical order modal.past – perf-aux.inf – V is possible as well, but this normally does not have an irrealis meaning. Instead it is interpreted as the past tense of deontic or epistemic modality (16b).

(16)     a.         Vertel maar niet        wie         zij   kon                                              geroepen         hebben.

                        tell       just    not who       she could.fin                     called.pcp       have.inf

                        ‘Just don’t tell her who she could have called.’

            b.         Het was   duidelijk   dat  zij   hem   kon                           hebben

                        it     was   clear         that she  him   could.fin                  have.inf                             geroepen.

                        called.pcp

                        ‘It was clear that it was possible for her to have called him.’            

1.1.5 Historical development

The development of word order variation in verb clusters can only be fully understood if various other developments in the history of Dutch are taken into account. One development is the tendency towards a more rigid V_final order. Another possible development suggested in Duinhoven (1998) and Coupé (2007) is that the current auxiliary system is the result of grammaticalisation of combinations of main verb – participle and main verb – infinitive. In older stages of Dutch, the participles and infinitives in such combinations may possibly have functioned as predicative adjuncts, which would later have been reanalysed as aux – participle and aux – infinitive combinations. Here we will only focus on word order changes in verb clusters in various parts of the language area.

As maps 14a and 15b show, the southern part of the language area (Belgium and Frans-Vlaanderen) is quite homogeneous as regards the distribution of 1-2 and 2-1 orders in two-verb clusters of the type aux_PERFECTIVE – participle and aux_MODAL – infinitive. Participles almost always precede the selecting auxiliary (2-1), whereas infinitives almost always follow the selecting auxiliary. De Meersman (1990) shows that in the Brabantish and Flemish area the current state of affairs is the result of an evolution that can be traced back to the 13^th century. In the 13^th century, the ratio of 1-2 and 2-1 orders in clusters with a participle was roughly 50-50 in Flanders. In Brabant (the present-day provinces of Vlaams Brabant and Antwerpen), only about 25% of these clusters had the 1-2 order. During the 14^th-16^th centuries, the frequency of the 1-2 order quickly decreases in the whole southern area and becomes marginal. Frans-Vlaanderen and the environment of Ieper shows a slightly different development in that the decrease of 1-2 orders goes more slowly and that 1-2 increases again in the sixteenth century. Table 1 provides a global overview of these developments, based on data from De Meersman (1990) (the abbreviations Fr-Vl, W-Vl and O-Vl stand for Frans-Vlaanderen, West-Vlaanderen and Oost-Vlaanderen).

Table 1Proportion of 1-2 orders in Aux – Participle clusters in Belgium

For clusters with an infinitive, the development is quite different. In the 13^th century, both orders occur; the position of 1-2 is a bit stronger in the Flemish areas studied; in Brabant, roughly 70% of the orders is 2-1. During the 14^th and 15^th century, the proportion of 1-2 orders strongly decreases. In the 16^th century, the 1-2 order makes a strong come back in all areas studied, such that it becomes the preferred order. A global overview of this development is given in table 2, based on data in De Meersman (1990).

Table 2Proportion of 1-2 orders in aux - infinitive clusters in Belgium

            In the Hollandic part of the Netherlands the situation was different, as Coussé (2003) shows. Coussé studied the development of word order variation in Hollandic  verb clusters from 1275 until 1825 on the basis of a selection of texts from four periods separated by three intervals of 150 years. In two-verb clusters with a participle, the 2-1 order is dominant during the whole period. As table 3 (based on Coussé 2003) shows, the ratio 1-2 and 2-1 is approximately 20-80, with the 15^th century as an all time low for 1-2 orders.

Table 3Proportion of 1-2 orders in aux – participle clusters in Holland

Again, in clusters with an infinitive the development is different, and a bit more similar to the development in the southern area studied by De Meersman. In the 13^th and 15^th century the 2-1 order is quite dominant. In the 17^th century, the proportion of 1-2 orders radically increases, with a small decrease in the 19^th century.

Table 4Proportion of 1-2 orders in aux – infinitive clusters in Holland

It seems that word order variation in the south and in Holland develops roughly into the same direction, towards a preference for the 2-1 order in clusters with a participle and towards a preference for the 1-2 order in clusters with an infinitive.

            Coupé (2007) compares verb clusters in 15^th-century Drenthe, in the north-east of the language area, and 17^th-century Zeeland, in the south-west of the Netherlands. As far as we know, Coupé’s study is the only one that systematically pays attention to the diachronic development of three-verb clusters, in addition to two-verb clusters. In Drenthe, in two-verb clusters with a participle the order 1-2 is virtually absent during the whole 15^th century. In 17^th-century Zeeland, however, there is a dramatic change. During this century, the proportion of 1-2 orders increases from 17% in 1600 up to 89% by the end of the century. Put differently, the 2-1 order rapidly disappears. As map 14a shows, the direction of change must have been reversed later on, given the exclusive use of 2-1 orders in current day Zeeland. In Drenthe, the situation remained stable.

            In two-verb clusters with an infinitive and the modal zullen ‘will’, the order 2-1 is strongly dominant as well in Drenthe at the beginning of the 15^th century. Unlike the clusters with a participle, the 2-1 order gradually loses ground during this century, ending with a proportion of about 60% at the end of the period investigated. In 17^th-century Zeeland, the order 1-2 is stronger right from the start (58%) and it gains ground, having a proportion of more than 90% at the end of the century. This is consistent with our map 15b, where the 1-2 order has a strong position.

            Coupé’s study does not yield results for the development of three-verb clusters in 15^th-century Drenthe, as the number of three-verb clusters that occur in the corpus  is too small. For the purpose of comparison, we would like to mention that clusters of the type Modal-Aux_PERF-V show up in the 3-1-2 order in almost 90% of the cases, the other orders attested being 1-3-2 (around 10%) and 1-2-3 (exceptional). Map 18a shows that the order 3-1-2 is still quite strong in Drenthe, but the order 1-3-2 also occurs quite frequently. The orders 1-2-3 and 3-2-1 occur in present day Drenthe as well.

            In 17^th-century Zeeland all types of three-verb clusters show a strong preference for the 1-2-3 order. For clusters with two modals, this is almost 100%, the other possible order being 1-3-2. For clusters with 1Aux_PERF-2Modal-3V it is 100%, and for clusters of 1Modal-2Aux_PERF-3V, 1-2-3 it rises from 10% to 80% during the century, the other possible orders being 1-3-2 and in particular 3-1-2. Map 17a shows that in present-day Zeeland in clusters with two modals, 1-2-3 is dominant, but 3-1-2 occurs as well. In clusters with 1Aux_PERF-2Modal-3V, 1-2-3 and 2-3-1 occur (map 18a), and in clusters of the type 1Modal-2Aux_PERF-3V, 3-1-2 is most popular. In addition, the 1-2-3 order is quite frequent and 1-3-2 order also occurs.

1.2 Discussion of the literature

The literature on verb clusters is so overwhelming that it is impossible to do justice to it in this short overview. We have tried to provide a comprehensive list of references in 1.5. We will highlight a number of publications that are particularly relevant for this atlas. Diachronic literature was discussed in the previous section. Important dialect-geographical studies are Van den Berg (1949), Pauwels (1953), Stroop (1970) and Gerritsen (1991), because they try to establish the distribution of different variants for large parts of the language area.

Van den Berg provides a map of the whole language area on the basis of two-verb clusters in, among other texts, Winkler’s dialecticon (1874), a collection of translations of the parable of The Prodigal Son in various dialects of Dutch. His map roughly coincides with our map 14a: 2-1 is dominant in the north-eastern provinces and North-Holland above the Noordzeekanaal, and it is also dominant in the majority of places in Vlaanderen, Brabant and Belgisch Limburg. The 1-2 order occurs in the rest of the language area, but there are cases of 2-1 as well.

Pauwels (1953) studies the distribution of word order variants in two-verb clusters in the Dutch language area. Her data come from translation tasks in a written questionnaire. Although her methodology has been criticised (Stroop 1970), the resulting maps show many similarities with our results (see the descriptions of the individual maps in section 1.4).

Stroop (1970) restricts himself to the Netherlands. His maps are based on an analysis of recordings of spontaneous conversations in 162 locations made between 1954 and 1970 (Dialect Archive Meertens Institute). Stroop compares his maps with Pauwels’ maps, which also show many similarities (cf. section 1.4 for further discussion).

In a discussion of the Infinitivus pro Participio effect (cf. chapter 2), De Schutter (1995) provides a sketch of the geographical distribution of different orders in three-verb clusters, but no map is provided.

A recent more detailed description of the distribution of word orders in three-verb clusters based on the sand-material is Barbiers (2005), which also provides an explanation for the general absence of 2-1-3 orders (cf. also Wurmbrand 2005) and the impossibility of 2-3-1 orders in clusters with a modal as the highest auxiliary.

De Schutter (1995) mentions two cases that seem to contradict the conclusion that 2-1-3 is impossible. One involves a Standard Dutch cluster with the aspectual verb beginnen ‘begin’ embedded under a perfective auxiliary. Since this aspectual verb selects a to-infinitive and since to -infinitives have a distribution which is different from bare infinitives, such examples are irrelevant for the present generalisation, which only holds for bare infinitives. Another potential counterexample comes from the dialects of Stellingwerf (in the north-east of the language area) and is also cited in Hoekstra (1994). This one does involve bare infinitives, but here we always find perception verbs as the second highest verb. Presumably, the infinitival complements of perception verbs are more complex than those of modals and aspectuals, which may give rise to different ordering possibilities.

            There is also a vast literature on the linguistic and non-linguistic factors that may favour the choice of one order over another in two-verb clusters containing a participle. Examples of such studies are Haeseryn (1990), De Schutter (1976, 1996), De Sutter et al. (2003), De Sutter et al. (2005), De Sutter (in press), Cornips & Ribbert (2006). Obviously, such studies are only relevant for varieties that allow various word order options per cluster type. This literature primarily concentrates on two-verb clusters with a participle in standard varieties, such as Standard Dutch spoken in The Netherlands, in Belgium, and in Heerlen (Dutch Limburg). In addition to geographical area, some of the factors explored include:

• Register; the spoken register favours the order 2-1;
• Prestige; the 1-2 order is the more prestigious variant and is favoured by normative grammarians (cf. Sassen 1963; Sturm 1990);
• The number of words in the middle field, i.e., between the V2 position and the V_FINAL position; the more words the more the order 1-2 is favoured;
• The type of auxiliary; the passive auxiliary worden ‘be’ strongly favours 2-1, this influence is less strong for the perfective auxiliary zijn ‘be’ and relatively speaking the weakest for hebben ‘have’.
• Morphology; non-compositional verbs favour 2-1 orders and compositional verbs favour 1-2 orders.
• Rhythm; according to De Schutter (1976, 1996) and Swerts (1998) verb clusters may give rise to prosodically less optimal configurations. De Schutter argues that adjacency of a primary and secondary accent as found, e.g., in sequences such as predicative complement₁ – main verb₂ – auxiliary (where 1 and 2 indicate stress) lead to a preference to place the main verb after the auxiliary.
• Sociolinguistic factors such as gender and age do not seem to play an important role.

In the generative framework, the most extensive and insightful recent overview of the empirical facts, generalizations, theoretical debate and literature is Wurmbrand (2005), which covers all relevant Germanic languages including Afrikaans, varieties of Dutch, varieties of German (in Germany, Austria and Switzerland), and Frisian. For Dutch the debate starts with Evers (1975). Among the central theoretical issues concerning verb clusters are the following:

• Are the various orders derived from one underlying order?
• Is this underlying order left-headed (1-2; VO) or right-headed (2-1; OV)?
• Does the reordering rule deriving the alternative orders, known in the generative literature as verb raising, apply in syntax or at the level of phonological form?
• Does the reordering rule apply to constituents (VPs) or heads (V)?
• Does the reordering rule move the verb (phrase) leftward or rightward?
• Why is it impossible for constituents larger than a word to interrupt the cluster in some varieties but not in others?

This debate has yielded a wealth of new data and insights. It seems to be clear that at least one cluster order, 3-1-2, must be the result of a reordering rule, as it cannot be generated directly if hierarchical structure is to be respected (cf. Barbiers 2005 for discussion). Another conclusion is that the 2-1-3 order is non-existent because it destroys the original hierarchical structure.  For many of the other issues it is too early to provide decisive answers, as Wurmbrand (2005) rightly concludes.

            Three-verb clusters expressing irrealis (cf. 1.1.4) are discussed in Sassen (1985), Wierenga (1986) and in particular in Hoekstra (1990) and De Rooij (1991).

Hoekstra shows that the irrealis construction with a modal as the highest verb is quite frequent in 19^th century Frisian. In the 20^th century it is on its way out, and is replaced by the perfective-auxiliary-first construction. According to Hoekstra, this is a general development in Germanic. In Standard Dutch, the modal-first construction is infrequent, but in peripheral areas such as West- and Oost-Vlaanderen and Zeeland it is still strong. According to De Rooij, who looks at Standard Dutch, both constructions are attested in the whole of the language area, but different regions show different frequencies, the modal-first construction being much more frequent in the south.

            The obligatory occurrence of the particle heen ‘towards’ on verbal complements of aspectual gaan ‘go’ has been discussed most recently in Haslinger (2007). Further literature can be found in Tuerlincx (1865), Naarding (1951), Michels (1952), Nuijtens (1962) and Van Bree (2000). The status of heen and the constructions in which it occurs is unclear (see further 1.3.2.3.2). The occurrence of the infinitival marker te ‘to’ on complements of aspectual gaan ‘go’ is discussed in Hoekstra (1997). Perfective be (participle or infinitive) as the second highest verb in three-verb clusters is discussed in Weijnen (1937), De Schutter (1974) and Vanden Wyngaerd (1994). Weijnen provides a map of Noord-Brabant that shows the distribution of the form weeste(n) which has both participial and infinitival characteristics. De Schutter discusses the geographical distribution of the various variants and provides a diachronic scenario. Vanden Wyngaerd discusses the southern variant with weeste(n) to provide an analysis of the Infinitivus Pro Participio (IPP) effect.

1.3 Discussion of the maps

1.3.1 Two-verb clusters

1.3.1.1 Auxiliary BE_finite – V_participle (map 14a) (map in dynaSAND)

Sentence (17) contains a subordinate clause dat Marie gisteren gestorven is ‘that Mary yesterday died is’ ending with the verb cluster gestorven is ‘died is’ (Dutch uses the perfective auxiliary BE with the verb DIE). An alternative order attested is is gestorven ‘is died’ with the same meaning. Sentence (17) was presented to the informants in the order gestorven is and they were asked to translate it. Subsequently, they were asked if the order of verbs they had given in their translation could be inverted.

(17)     Ze     weet    niet    dat    Marie gisteren    gestorven   is.

            she    knows not    that   Mary yesterday died            is

            ‘She does not know that Mary died yesterday.’

Map 14a shows the distribution of the orders gestorven is (2-1; green) and the order is gestorven (1-2; red). The geographical distribution of the two orders is very clear. There are three major areas.

The first area is the northern part of the language area, which includes the provinces Friesland, Groningen, Drenthe, and North-Holland to the north of the Noordzeekanaal. In this part the order gestorven is ‘died is’ (2-1; green) is the only one possible.

The second main area is the middle and southeastern part of the language area, which allows both the gestorven is ‘died is’ order (2-1, green) and the order is gestorven ‘is died’ (1-2; red). This area includes the southern part of North-Holland, below the Noordzeekanaal, Zuid-Holland, Utrecht, Overijssel, Gelderland, Noord-Brabant, and Dutch and Belgisch Limburg. There are three reasons that strongly suggest that in this area as well, the green 2-1 order is the dominant order. First, there are very few locations (10) that only allow the red 1-2 order and they are scattered across the area. Secondly, there are very few speakers that spontaneously altered the presented 2-1 order into a 1-2 order in their first translation. Thirdly, in those cases where informants were asked explicitly which of the two orders they preferred, this was almost always the 2-1 order.

The third major area is in the south and only has the order gestorven is ‘died is’ (2-1; green). This area includes from west to east: Frans-Vlaanderen, West-Vlaanderen, Oost-Vlaanderen, and, if we ignore a few exceptions, also Vlaams Brabant and Antwerpen. The islands of Zeeland only have the order gestorven is ‘died is’ (green; 2-1), whereas Zeeuws-Vlaanderen allows both orders.

Let us compare these results for The Netherlands with the results of Pauwels (1953) and Gerritsen (1991). The similiarities with Pauwels’ map of die de bedelaar gezien hadden ‘who the beggar seen had.pl’ are striking, even though Pauwels offered the test sentence in the red 1-2 order in a written translation task. Friesland, Groningen, Drenthe and Noord-Holland pattern alike in a detailed manner, even the part of Noord-Holland directly to the north of the Noordzeekanaal that allows both orders is found both on Pauwels and our map. Also, the dominance of the order 2-1 in Zeeland is found on both maps. On Pauwels map, an eastern area reaching from Twente (Overijssel) via eastern Gelderland to the Dutch province of Limburg predominantly shows 1-2. On our map this area is one which allows both orders. Map 24a in Gerritsen (1991), which also includes Dutch speaking Belgium, shows a very similar pattern to ours as well. Gerritsen offered the two alternative orders in a written questionnaire in Standard Dutch, asked whether these orders occurred and asked for preference. Given that the studies compared used different methodologies, different informants and different time slices over a period of more than 50 years, it can be concluded that the given pattern is very reliable and stable. It may be the case, however, that the very dominant 2-1 order does not belong to all of the dialects in which we find the red 1-2 order. Possibly, this is a case in which the informants were influenced by the standard language, but this cannot be decided on the basis of the available data.

1.3.1.2 Auxiliary HAVE_finite – V_participle I & II (map 14b, 15a) (maps in dynaSAND for participle I and participle II)

In this section, we consider two-verb clusters with HAVE instead of BE as the perfective auxiliary, and we make a methodological comparison. The relevant sentences are given in (18).

(18)     a.         Dat   is de      man   die     het     verhaal   heeft         verteld.

                        that is    the     man   who the     story     has.fin     told.pcp

                        ‘That is the man who told the story.’

b.         Dat is    de      man   die     ze      geroepen    hebben.

                        that is    the     man   who they called.pcp   have.fin

                        ‘That is the man who they called.’

Map 14b is based on test sentence (18a). Informants were asked to translate the sentence. Nothing was asked about the order of the verbs in the cluster. This means that informants who came up with a green 2-1 order in their translation spontaneously altered the order that was offered to them. This is a clear sign that the 1-2 order is strongly dispreferred. Map 14b shows that BE (cf. map 14a) and HAVE do not give very different results. Again, we see a northern and a southern area which exclusively allow 2-1 orders, and a central and southeastern area that allows both orders. There are some differences, however. We find more locations with a 1-2 order in Noord-Holland, Zeeland and Vlaams Brabant than on map 14a. Also, the number of locations that can only have the 1-2 order is considerably higher than with auxiliary BE.

            It is methodologically interesting to compare (18a) with (18b). In (18b) as well, informants were asked to translate the sentence, and nothing was asked about the order of the verbs in the cluster. However, in this test sentence the verbs were presented in the green 2-1 order, which as we have seen is the dominant order that occurs in the entire language area. As map 15a shows, in this case only 6 informants came up with the alternative 1-2 order, giving the false impression that the 1-2 order is almost absent. We can conclude from this that word order variation should be investigated by explicitly testing alternative orderings. Translations may give unreliable results as the order that is offered may influence the order that is returned. Similarly, corpus studies do not necessarily reveal the possibility of alternative orderings.

1.3.1.3 Modal_finite – V_infinitive (map 15b) (map in dynaSAND)

The sentence in (19) contains a clause-final two-verb cluster consisting of a finite modal mag ‘may’ and an infinitive zien ‘see’. The sentence was presented in the red 1-2 order and the task was to translate it. Subsequently, the interviewer asked if the order of verbs in the translation could be inverted.

(19)     Niemand  mag   het  zien   dus ik  vind   dat  jij    het  ook niet mag       zien.

            nobody    may it     see    so   I   find   that you it     also not may.fin see.inf

            ‘Nobody is allowed to see it so I think you are not allowed to either.’

We see that the geographical distribution of word orders in clusters with a modal and an infinitive is very different from that of clusters with a perfective auxiliary and a participle (map 14a), although we can distinguish the same three areas on map 15b that we saw on map 14a. There is a northern area consisting of northern Noord-Holland, Friesland, Groningen and Drenthe that has a strong preference for the 2-1 order. However, this northern area is less homogeneous than in the case of clusters with a participle. In particular, there are more locations in Noord-Holland, Groningen and Drenthe that also allow the 1-2 order. In the middle and southeastern area, the majority of locations allow both orders, but there are a number of locations that only have the red 1-2 order, in particular in Utrecht, Zuid-Holland and Noord-Brabant.  The 1-2 order is usually considered to be the Standard Dutch order. The third area is the Dutch-speaking part of Belgium, where the green order 2-1 is almost completely absent, exactly the opposite of the case of the clusters with a participle. In sum, except for Friesland, the whole language area has the 1-2 order for clusters with a modal, and the 2-1 order is a possible alternative in the northern, middle and eastern part of the language area.

1.3.1.4 Two-verb clusters – synthesis (map 16a)

Map 16a is a synthesis of maps 14a and 15b. The three areas mentioned in the previous descriptions are clearly discernible: a northern area with a strong preference for the green 2-1 order in both types of two-verb clusters, a southern area with a strong preference for 2-1 in clusters with a participle and 1-2 in clusters with an infinitive, and a central and eastern area which allows both orders in both types of clusters.  

1.3.2 Three-verb clusters

In this section we will discuss four types of clusters: (i) modal – modal – Vinfinitive; (ii) modal – perfective auxiliary – Vparticiple; (iii) perfective auxiliary – aspectual auxiliary – Vinfinitive; (iv) perfective auxiliary – modal – Vinfinitive. As we have seen in the previous sections, the geographical distribution of word orders in two-verb clusters covaries with the type of auxiliary (or, alternatively, with the presence of a participial or infinitival form). In this section, we will see that the same holds for three-verb clusters.

1.3.2.1 Modal_finite – Modal_infinitive – V_infinitive (map 17a) (map in dynaSAND)

The cluster type modal – modal – Vinfinitive was tested for the five different orders of the verb cluster given in (20a,b,d-f). The sixth logically possible order (20c) was not tested, as it was completely absent in the results of the postal pilot survey and had been reported in the literature to be non-existent. Each variant was presented to the informant in the local dialect with the question whether it occurred in that dialect. In the oral interviews the informants were not asked about their preferences for one or more of the occurring orders. Readers interested in such preferences may want to consult the data of the postal pilot survey in Dynasand.

(20)     a.         Ik vind dat iedereen   moet           kunnen    zwemmen.       V1-V2-V3

                        I   find   that             everyone         must.fin can.inf             swim.inf

                        ‘In my opinion, everyone should be able to swim.’

            b.         moet             zwemmen     kunnen                                          V1-V3-V2

                        must.fin       swim.inf      can.inf

            c.         kunnen moet           zwemmen                                               V2-V1-V3

                        can.inf   must.fin     swim.inf.       

d.         kunnen zwemmen   moet                                                       V2-V3-V1

                        can.inf   swim.inf    must.fin

            e.         zwemmen        moet                kunnen                                                V3-V1-V2

swim.inf         must.fin          can.inf
f.          zwemmen        kunnen            moet                                        V3-V2-V1

swim.inf         can.inf             must.fin

The results for the test sentences in (20) are summarized in table 1.

Table 1Modal – Modal – V

Map 17a shows that the 1-2-3 order, which is also the Standard Dutch order, is highly dominant in the entire language area except in Friesland, where most locations only allow the mirror image order 3-2-1. This 3-2-1 order also occurs in Noord-Holland to the north of the Noordzeekanaal, Groningen en Drenthe, but in the relevant locations it is never the only option.

In a large part of the south, the 1-2-3 order even is the only order attested. This southern area includes Vlaanderen (i.e., Frans-Vlaanderen, Zeeuws-Vlaanderen (the southern part of the Dutch province of Zeeland), West- and Oost-Vlaanderen), Brabant (i.e., the provinces of Vlaams Brabant, Antwerpen and Noord-Brabant) and a large part of Belgisch Limburg. In view of what we have found for two-verb clusters with a modal, the behaviour of three-verb clusters with two modals in Friesland and the southern area just described is largely as expected.

A third area that can be distinguished is a northern area that has the order 1-3-2, though nowhere as the only option. There are 30 occurrences out of 34 in an area to the north of a line that runs from IJmuiden, at the mouth of the Noordzeekanaal in Noord-Holland, to Arnhem in the south-east of Gelderland. The remaining 4 occurrences are located in Dutch Limburg.

The fourth area revealed by this map is determined by the 3-1-2 order which is quite frequent (83 occurrences) but never the only option. This order is virtually restricted to the Netherlands, with the exception of some locations in the south of Belgisch Limburg. It has been suggested that this order involves nominalisation of the main verb (cf. Den Besten & Broekhuis 1989).

It is worth mentioning that there are various locations that allow all four orders. Such locations are found in Noord-Holland, Friesland, Groningen and Dutch Limburg. In the same provinces and in the eastern part of Overijssel we find dialects that allow three orders. A final important observation is that the 2-3-1 order is not attested for this cluster type.

1.3.2.2 Modal_finite – HAVE_infinitive – V_participle (map 17b) (map in dynaSAND)

The cluster type modal – HAVE – Vparticiplewas tested for the five different orders given in (21a-e). The sixth logically possible order 2-1-3 was not tested as, here too, the postal pilot survey had shown that this order was non-existent for this cluster type. For these test sentences, exactly the same methodology was used as for the sentences in (20) in the previous section. All sentences were presented in the local dialect and the informants had to indicate whether they occurred in their dialect.

(21)     a.         Jan       weet    dat       hij        voor     3          uur       de        wagen

                        Jan       knows that      he        before 3          hour     the       car                               moet                hebben             gemaakt.                                 V1-V2-V3

                        must.fin          have.inf           made.pcp

                        ‘John knows that he is supposed to have the car repaired by 3 o’clock.’

            b.         moet         gemaakt      hebben                                                  V1-V3-V2

                        must.fin   made.pcp   have.inf

            c.         hebben     gemaakt      moet                                                     V2-V3-V1

                        have.inf   made.pcp   must.fin

            d.         gemaakt    moet           hebben                                                  V3-V1-V2

                        made.pcp must.fin     have.inf

            e.         gemaakt    hebben       moet                                                     V3-V2-V1

                        made.pcp have.inf      must.fin

This cluster type shows the same ordering possibilities as the type with two modals. The results for this cluster are summarised in table 2.

Table 2Modal – Auxiliary – V

The geographical distribution of the various orders provided by map 17b shows some similarity with the two-modals cluster, but the differences are more striking.

Let us start with a similarity. The 3-2-1 order is largely restricted to the northern area consisting of Noord-Holland north of the Noordzeekanaal, Friesland (including the Wadden islands), Groningen and Drenthe, but we also find some locations along the entire eastern border of the language area. In Friesland, 3-2-1 usually is the only order, outside Friesland this is rarely the case.

            The distribution of the 1-2-3 order with this cluster is clearly different from the two-modal cluster. Whereas in the latter cluster 1-2-3 was dominant in the whole language area except Friesland, involving 242 occurrences, the 1-2-3 order in the current cluster type is much less frequent (91 occurrences) and almost completely absent in Belgium, except for Belgisch-Limburg. The 1-2-3 order is found in all Dutch provinces.

            The 1-3-2 order, which is relatively infrequent in two-modal clusters (34 occurrences) and restricted there to the area above the line IJmuiden – Arnhem, is quite popular for clusters with a modal and HAVE (163 occurrences). In Dutch-speaking Belgium, Frans-Vlaanderen and part of the Dutch province of Noord-Brabant there is a strong preference for this order, which is not surprising given that this preference exists in the same southern area for the 2-1 order in two-verb clusters with a perfective auxiliary. Apparently, in the relevant southern dialects participles must precede perfective auxiliaries both in two- and in three-verb clusters. In the rest of the Netherlands, we find the 1-3-2 order scattered across the entire language area, but seldom in Noord-Holland to the north of the Noordzeekanaal, Zuid-Holland and Friesland.

            The 3-1-2 order is also much more frequent for this type of cluster, as compared to the two-modal cluster (186, which is twice as many occurrences). Whereas this order is virtually absent in two-modal clusters in the southern participle-first area just described, quite a number of southern dialects allow it as a second order with modal-HAVE clusters. The order is particularly frequent in Oost-Vlaanderen, Antwerpen and Belgisch Limburg. The 3-1-2 order is also found in the Dutch provinces, but rarely in Friesland.

            For this cluster type, there are hardly any dialects that allow all four orders, but there are fairly many dialects with three orders, in particular in Dutch Limburg, Noord-Brabant, Utrecht, Gelderland, Overijssel and Drenthe. As was the case with the previous cluster type, the order 2-3-1 is not attested.

1.3.2.3 Auxiliary BE_finite – Aspectual/Modal_inf – V_infinitive

There are four maps for this cluster type. Map 18a shows the distribution of the different verb orders for clusters with aspectual GO. Map 18b depicts three variables that complicate the interpretation of map 18a. Map 19a shows that the 3-1-2 order occurs much less often with this cluster type than map 18a seems to suggest. Map 19b shows the distribution of an alternative construction with BE instead of aspectual GO. Finally map 20a shows the distribution of different word orders for a cluster with a modal instead of an aspectual auxiliary as the second highest verb.

1.3.2.3.1 BE_finite – GO_{infinitive/participle} – V_infinitive (map 18a) (map in dynaSAND)

The cluster BE – GO – Vinfinitive as given in (22a) differs from the previous one in three respects: (i) it contains an aspectual instead of a modal auxiliary, (ii) the perfective auxiliary BE is the hierarchically highest instead of the second highest auxiliary; (iii) in many dialects of Dutch, this hierarchy gives rise to the well-known Infinitivus pro Participio effect, the occurrence of an infinitive instead of a participle as a dependent of the perfective auxiliary.

            In view of the desire to compare minimal pairs, it is a bit unfortunate that the cluster contains an aspectual instead of a modal auxiliary, as this gives rise to a number of complications (see section 1.3.2.3.2 and map 18b). Despite these complications, there is reason to think that the word order properties of this cluster type to a large extent carry over to clusters with a modal as the second highest auxiliary. We therefore abstract away from the complications and concentrate on the word order variation.

(22)     a.         Ik weet    dat hij        is          gaan    zwemmen.                   V1-V2-V3

                        I   know   that     he   is.fin    go.inf  swim.inf

                        ‘I know that he went for a swim.’

            b.         gaan          zwemmen  is                                                          V2-V3-V1

                        go.inf       swim.inf    is.fin

            c.         zwemmen            gaan       is                                                               V3-V2-V1

                        swim.inf         go.inf  is.fin   

d.         zwemmen            is           gaan                                                           V3-V1-V2

                        swim.inf         is.fin   go.inf

For map 18a, we tested the four orders in (22), 1-2-3, 2-3-1, 3-2-1 and 3-1-2. Each order was presented in a sentence in the local dialect and the informants were asked whether such as sentence would occur in their dialects. As in the case of the previous two cluster types, the order 2-1-3 was not tested as it did not occur in the postal pilot survey or the literature (cf. also below). Nevertheless, there are two spontaneous cases of 2-1-3, one on the island of Schiermonnikoog and one in the Groningen village of Een. Both involve a cluster with te ‘to’ preceding the main verb. Since TO-infinitives have a different syntactic distribution, the generalisation that 2-1-3 is excluded for bare infinitival clusters can be safely upheld. Like the 2-1-3 order, the 1-3-2 order was not tested for this cluster type because it was absent in the results of the postal pilot. In the results of the oral interviews, there is one spontaneous attestation in Limburg. As we will see in the discussion of map 20a, there is reason to think that this order is indeed not completely excluded in clusters with perfective auxiliaries as the highest verb. The results for the test sentences in (22) are summarised in table 3.

Table 3 Aux – Asp – V

For the four orders that were tested, map 18a shows that the language area is again divided in three parts. There is a northern area (Noord-Holland north of the Noordzeekanaal, Friesland, Groningen, Drenthe and the northern part of Overijssel) in which many dialects have a 3-2-1 order. Although this is consistent with the northern order in two- and three-verb clusters discussed above, it should be noted that the northern area is less homogeneous for the cluster under discussion. In all the relevant provinces, even in Friesland, we find 1-2-3 as an alternative, and the same holds for the 2-3-1 order.

            In the middle and eastern area, including Limburg, we find 1-2-3 as the predominant order. Zeeland, Zuid-Holland, the northwestern part of Gelderland and Belgisch Limburg have 2-3-1 as an alternative order. Most locations in Noord-Brabant only allow the 1-2-3 order, with some exceptions in the western part where 2-3-1 occurs.

            In a southern area consisting of all Dutch-speaking Belgian provinces the 2-3-1 order is highly dominant. Oost-Vlaanderen and Brabant are the most homogeneous, with few alternative orders. In the western part of West-Vlaanderen, we find some locations that have 1-2-3 as an alternative, and the same holds for the western part of Antwerpen.

            Finally, the 3-1-2 order is attested 18 times along the eastern border of the language area, in particular in the eastern part of Overijssel and in Dutch Limburg. As for 11 locations in Overijssel, Drenthe and Gelderland, these all involve dialects that have the heen ‘towards’ particle with gaan ‘go’. In most of the 3-1-2 orders in these locations heen also shows up, so it may be necessary to discount them, as is argued in the next section.

1.3.2.3.2 Heen ‘away’, te ‘to’ and IPP (maps 18b, 19a) (map in dynaSAND)

As was noted in the introduction to this section, the cluster type BE – GO – Vinfinitive gives rise to some complications that need to be discussed. We will discuss three complications, all depicted on map 18b, in this section. The fourth, with its own map, will be discussed in the next section.

The first complication is that in some dialects the aspectual auxiliary gaan ‘go’ co-occurs with the directional particle heen ‘towards’. An example is given in (23).

(23)     Ik    wete      dat     hij   heen zwemn        egaan        is.    (Groenlo, Overijssel)

            I      know     that   he   away swim.inf    gone.pcp    is.fin

            ‘I know that he went out for a swim.’

Map 18b shows that these dialects are primarily located in eastern Overijssel and a few neighbouring places in Drenthe and Gelderland. In this area, heen occurs with clusters that have a 3-1-2 or a 3-2-1 order. It may be that this order is forced by the presence of heen and hence is impossible with aspectual or modal auxiliaries that do not select heen. Although the precise status of heen ‘towards’ is unclear, there is a good candidate for an analysis that could explain its influence on the order in the verb cluster. The particle and the main verb could form a predicative complement together. Predicative complements always precede their selecting verbs in Dutch. On this account, heen would be comparable to Standard Dutch uit ‘out’, which forces the main verb to precede the rest of the cluster as well (24a). (24b-d) show the effects of the presence of uit ‘out’ on word order in the three-verb cluster. The main verb can only precede the other verbs if uit ‘out’ is present (24b,c) and it can only follow the other verbs if uit ‘out’ is absent (the star in (24b,c) means that this cluster order is only possible when uit ‘out’ is present; the star in (24d) means that the 1-2-3 order is only possible if uit ‘out’ is absent).

(24)     a.         Ik      weet    dat    hij   uit     <zwemmen>     is <*zwemmen>.

                        I        know   that   he   out    swim.inf           is swim.inf

                        ‘I know that he went out for a swim.’

            b.         dat    hij   *(uit)     zwemmen   is gegaan.

                        that   he    out         swim.inf    is gone.pcp

                        ‘that he has gone out for a swim.’

            c.         dat    hij   *(uit)     zwemmen   gegaan            is.

                        that   he    out         swim.inf    gone.pcp         is

                        ‘that he has gone out for a swim.’

d.         dat    hij   (*uit)     is gaan         zwemmen.

                        that   he    (out)        is gone.pcp swim.inf

                        ‘that he has gone out for a swim.’

            e.         Het   is     jammer  dat  FC Twente niet winnen   heeft kunnen/gekund.

                        it       is     pity       that FC Twente   not  win.inf has      can.inf/pcp

            ‘It is a pity that FC Twente has not been able to win.’

f.         Sommige  mensen gingen      heen en   deden   de   blinden   dicht.

                        some             people   went      pcl        and    did    the    shutters closed

                        ‘Some people started to close the shutters.’

If this line of reasoning is correct, the Overijssel 3-1-2 cases are in fact two-verb clusters with a verbal predicative complement. This would mean, then, that the possibility of 3-1-2 orders with clusters of the type BE – GO – Vinfinitive in Overijssel and surroundings does not tell us anything about the general possibility of 3-1-2 order in clusters of the type perfective auxiliary – aspectual/modal auxiliary - Vinfinitive.

We thus did a telephone survey among the informants accepting 3-1-2 orders for BE – GO – Vinfinitive, and tested the sentence in (24e). As map 19a shows, of the 17 locations (one informant could not be reached) with a 3-1-2 order with BE – GO – Vinfinitive, only 5 allow the 3-1-2 order for clusters of the type HAVE – modal – Vinfinitive, 4 of which are located close to the German border in the north of Overijssel and the south of Drenthe. Thus, although it cannot be concluded that the 3-1-2 order in clusters of the type perfective auxiliary – aspectual/modal auxiliary – Vinfinitive is impossible, it can be concluded that it is extremely rare and that the response for the cluster BE – GO – Vinfinitive may give a false picture. In Overijssel, this is due to the presence of the particle heen.

The geographical distribution of heen gaan differs from the description given in Naarding (1951). He found it in Noord-Holland (Zaans), Friesland, Groningen and Drenthe, and suggests that it also occurs in Overijssel and Gelderland, although he does not have data for the latter two provinces. In all of these areas, heen gaan usually occurs in a (pseudo-)coordinative construction such as (24f), unlike the cases we have found in Overijssel. Since the instances of heen gaan on map 19a are spontaneous occurrences and the construction was not explicitly tested, it is impossible to tell on the basis of the SAND data whether heen gaan still occurs in the areas mentioned by Naarding.

The second complication involves the presence of the infinitival marker te ‘to’ on the verbal complement of gaan ‘go’. An example is given in (25).

(25)     Ik    weet    dat     hij te  zwemmen   gaan    is. (Hippolytushoef, Noord-Holland)

            I      know   that   he to swim.inf    go.inf  is.fin

            ‘I know that he went out for a swim.’

The to-marker may also have an influence on the word order. Recall that in Standard Dutch TO-infinitives are different from bare infinitives in that they must follow the selecting verb. As map 18b shows, the occurrence of te ‘to’ is clearly restricted to a northern area consisting of Friesland (including the Wadden islands) and the northern part of Noord-Holland.

The third complication is the morphology of gaan ‘go’ in this cluster. Some dialects have an infinitive (IPP), others have a participle (no IPP). Map 18b shows that many Limburg and central-eastern dialects have a participle. Since the possibility of a participle was not explicitly tested in all dialects, it may be that the number of locations that have it is larger than indicated on the map.

Since the three properties discussed above are independent of each other, they can be combined in one dialect in one construction, which is why one location can have more than one square. For each relevant location we also indicated the word order for this cluster.

1.3.2.3.3 Perfective auxiliary BE in three-verb clusters (map 19b) (map in dynaSAND)

In a number of locations in Belgium a variant of the three-verb cluster perfective auxiliary GO–V was also presented with BE instead of GO. This may be considered to be the perfective of the absentive construction, illustrated in (26b) for Standard Dutch, which is very close in meaning to BE – GO – V. The difference is that in the case of perfective auxiliary – BE – V the event denoted by V has been completed in the past, while this is not necessarily so for the variant with GO. The variant that is particulary relevant for Belgium is one in which the form weeste(n) occurs instead of wezen, a form that looks like a participle without the prefix ge-. Since this form only occurs in the absentive construction, it could be that synchronically it should not be analysed as the participle of the verb BE.

(26)     a.         dat    hij      zwemmen   is.

                        that   he      swim.inf    is.fin

                        ‘that he went out for a swim.’

            b.         dat    hij      is wezen    zwemmen.

                        that   he      is be.inf    swim.inf

                        ‘that he has been out for a swim.’

            c.         Ik      wete dat ij       weeste zwemmen   eet .  (Oosteekloo; O-Vlaanderen)

                        I   know that he   been       swim          has

                        ‘I know that he has been out for a swim.’

This variant was tested in West- and Oost-Vlaanderen and the western part of Antwerpen. Map 19b shows that there is a concentration of locations in Oost-Vlaanderen and the central eastern part of West-Vlaanderen with this form in the order 2-3-1, being the unmarked order for clusters of this type. Note that the relevant dialects all have HAVE as the perfective auxiliary. There are also some spontaneous attestations in the eastern part of Antwerpen that have BE–weeste. On De Schutter’s (1974) map, the construction also occurs in the southern part of Antwerpen, an area for which we do not have enough data. Four linguists from Vlaams-Brabant report (p.c.) that the construction does not occur in this province. A striking difference with the map in De Schutter (1974) is the low number of dialects in West-Vlaanderen that have weeste in this construction. What we do find in West-Vlaanderen is five locations with a four-verb cluster: weeste-GO-V-BE. The same holds for two locations in Frans-Vlaanderen.

1.3.2.3.4 HAVE_finite – Modal_{infinitive/participle} – V_infinitive (map 20a) (map in dynaSAND)

The sentence in (27) contains a three-verb cluster that differs from the one in the previous section in that it has HAVE as the highest auxiliary and a modal instead of an aspectual as the second highest verb. Also, HAVE has past tense morphology and expresses irrealis.

(27)     Vertel  mij    eens       wie      zij      had               kunnen       roepen.

            tell       me     once       who     she    had.fin can.inf call.inf

            ‘Just tell me who she could have called.’

This sentence was primarily intended to provide information about the complementiser system. The informants had to translate this sentence from Standard Dutch into their own dialect. There were no explicit questions about the word order in the verb cluster. As was discussed above for the two-verb clusters, if an informant in his translation spontaneously changes the verb order that was presented to him/her, we can safely assume that the resulting order is the one he/she prefers.

Map 20a shows that the large majority of informants retain the 1-2-3 order in the translation, which is not surprising as we have seen on map 18a that this is the most frequent order for the whole language area. There are three areas where alternative orders show up.

In the south, Oost-Vlaanderen, Zeeuws-Vlaanderen and small parts of Zeeland, Antwerpen and Vlaams-Brabant have the 2-3-1 order. This is a proper subpart of the area that has 2-3-1 on map 18a, where West-Vlaanderen, Brabant and Belgisch Limburg show a frequent occurrence of 2-3-1 and where 2-3-1 also occurs in a number of Dutch provinces.

In the north, there is an area consisting of Friesland and Groningen that has a 3-2-1 order, as expected given map 18a. Notice that dialects differ with respect to the form of the modal: in Groningen it is a participle, in Friesland we find dialects with a participle, with an infinitive or both. The third area that can be distinguished has the order 1-3-2. This is found eight times in an area roughly surrounding the former Zuiderzee (currently the IJsselmeer and the IJsselmeerpolders, between Noord-Holland in the west and Friesland, Overijssel and Gelderland in the east). As was mentioned in the discussion in section 1.3.2.3.1, this order was not tested for map 18a given its absence in the results of the postal pilot. The fact that it shows up eight times here as a correction of the order offered strongly suggests that this order is possible in clusters of the type perfective auxiliary – aspectual/modal – Vinfinitive.

The final order that occurs on the map is the 3-1-2 order, which occurs five times, twice with the modal as an infinitive and three times as a participle. Two of these occur in Overijssel, the province in which we found this order for the cluster HAVE – modal – Vinfinitive. We reach the same conclusion as in 1.3.2.3.2, namely that the order 3-1-2 seems to exist but is extremely rare.

As in all other three-verb clusters so far, the order 2-1-3 does not show up, supporting the generalisation that clusters with a bare infinitive do not allow that order.

1.3.2.4 Three-verb clusters – irrealis (map 20b)

The sentences in (28) are two ways to express irrealis. The Standard Dutch variant in (28a) has a three-verb cluster with the past tense perfect auxiliary as the highest verb and a modal infinitive as the second auxiliary, whereas the southern variant (28b) has a three-verb cluster with a past tense modal as the highest verb and an infinitival perfective auxiliary as the second auxiliary. There is an independent difference between (28a) and (28b). The main verb precedes the second auxiliary in the b-example because this is the dominant order in the relevant southern area.

(28)     a.         Vertel    maar    niet      wie    zij     had               kunnen     roepen.

                        tell         just           not who   she    had.past  can.inf     call.inf

                        ‘Just don’t tell her who she could have called.’

            b.         Vertel    maar    niet      wie    zij     kon                    geroepen           hebben.

                        tell         just           not who   she    could.past    called.pcp     have.inf

                        ‘Just don’t tell her who she could have called.’

            c.         Vertel    maar    niet      wie    zij     zou                    geroepen           hebben.

                        tell         just           not who   she    would.past   called.pcp     have.inf

                        ‘Just don’t tell her who she could have called.’

Our postal pilot showed, unlike De Rooij’s study (1991; cf. the discussion in section 1.2) that the irrealis of type (28b) is almost completely absent in The Netherlands. In the oral questionnaire, this test sentence did not have the highest priority. It was only asked in Belgium, except for West-Vlaanderen. Informants were asked if this construction occurred in their dialect. To make the picture as complete as possible, we asked some West-Flemish colleagues for their judgements. Their answers suggest that the construction is rare is West-Vlaanderen, but not always excluded. Map 20b shows that Oost-Vlaanderen is the home of this irrealis type, but it occurs in the other Dutch-speaking Belgian provinces as well. This squares with the map in De Rooy (1991). In some locations, informants produce (28c) with zou ‘would’ instead of kon ‘could’ as the most natural way to express the irrealis.

1.3.3 Potential correlations

The goal of this section is to discover which verb cluster systems there are in the Dutch language area. We do this by looking at correlations between the word order in two-verb clusters with a perfective auxiliary and the word order in the two types of three-verb clusters that contain a perfective auxiliary, and by looking at the correlation between the word orders in the three types of three-verb clusters.

1.3.3.1 Two- and three-verb clusters – perfective I (map 21a)

Map 21a combines map 14a and map 18a. The goal is to show that a perfective auxiliary as the highest verb in the cluster has the same effect on word order in both two and three-verb clusters. This yields a very coherent picture. In the south, the Belgian provinces with the exception of the eastern part of Belgisch Limburg all require the perfective auxiliary to follow the rest of the cluster, i.e., 2-1 and 2-3-1. The Dutch province of Zeeland joins up with Dutch-speaking Belgium, except for Zeeuws-Vlaanderen. This we may call the Belgian system.

Noord-Brabant, the northern-part of Dutch Limburg, Utrecht, Zuid-Holland, the southern part of Noord-Holland and part of Gelderland constitute a homogeneous area having the perfective auxiliary in first position and an ascending order for 2-3, i.e., 1-2 and 1-2-3. This we can call the Netherlandic system.

There is a large northern area including Noord-Holland, Friesland, Groningen and Drenthe that, like Belgium, always has the perfective auxiliary in final position. The difference with Belgium is that the second auxiliary and the main verb also invert, so we get 2-1 and 3-2-1. This we can call the Frisian system.

The eastern dialects that have 1-2 in two-verb clusters and 3-1-2 in three-verb clusters can be said to be consistent in having the perfective auxiliary preceding the second auxiliary. However, as we have seen in the discussion of map 18a and 20a there are reasons to question the existence of the 3-1-2 order for this cluster type.

The remaining parts of the language area show a mixed behaviour. We find combinations of the Dutch and Belgium system in the Netherlands, but also combinations of the Belgian and Frisian system in Drenthe and Belgisch Limburg.

1.3.3.2 Two- and three-verb clusters – perfective II (map 22a)

Map 22a combines maps 14b and 17b. The perfective auxiliary is again in the focus of our attention. This time, we would like to know in particular whether dialects with a 2-1 order for clusters with a perfective auxiliary have a 1-3-2 order in clusters with a perfective auxiliary as the second highest verb, such that the main verb always precedes the perfective auxiliary. As map 22a shows, this is indeed the case for most dialects of Dutch in Belgium. There are also many dialects in Belgium that have the system 2-1 and 3-1-2, mostly as an alternative to 2-1 and 1-3-2. In this case as well, the main verb always precedes the perfective auxiliary. As expected, we find the system 2-1 and 3-2-1 in the north of the language area, which quite consistently uses the mirror image of the Standard Dutch order. A quite frequent pattern in the rest of the Netherlands involves 1-2 with 3-1-2. Slightly less frequent are the patterns 1-2 with 1-2-3, and 1-2 with 1-3-2, both mainly occurring in the Netherlands.

1.3.3.3 Two- and three-verb clusters – infinitive (map 23a)

Map 23a combines maps 15b and 17a. The map shows the correlations between the order in two-verb clusters with a modal and the order in three-verb clusters with two modals. The most dominant pattern is 1-2 with 1-2-3, which is almost the only option in Belgium and the Dutch province of Noord-Brabant. In this pattern, the linear order from left-to-right directly reflects the dependency relations between the three verbs. In Belgisch Limburg, some dialects have 1-2 with 3-1-2 as an additional option, where the main verb precedes the two auxiliaries in three-verb-clusters. This system can be found quite frequently in the Netherlands, except for Friesland and Noord-Brabant. The combination 2-1 with 3-2-1 is found in the northern part of the language area. This is the mirror image of the Standard Dutch system 1-2 and 1-2-3. A quite frequent system is 2-1 with 1-2-3, which is found in the Netherlands, again with the exception of Friesland and Noord-Brabant.

1.3.3.4 Uniform order in three types of verb clusters (map 24a)

Map 24a shows which dialects can have a uniform order for all three different types of verb clusters. There are three areas. Starting in the south, it is clear that a uniform ordering across cluster types hardly exists in Belgium. In The Netherlands, we find a large area where three-verb clusters can always have an ascending order (1-2-3). This area includes all Dutch provinces except Friesland. There is a third area including north Noord-Holland, Friesland, Groningen and Drenthe where we find uniformly descending orders (3-2-1). Finally, there are dialects along the eastern border, in particular in Overijssel and Limburg, where 3-1-2 seems to be the uniform order for all clusters. However, as we have seen above (sections 1.3.2.3.3 and 1.3.2.3.4), there is reason to doubt the existence of 3-1-2 orders with a perfective auxiliary as the highest verb.

1.3.3.5 Systems of three-verb clusters (map 25a)

Map 25a shows which three-verb cluster systems can be found in the Dutch language area, regardless of whether the order is uniform across cluster types. The uniform systems were discussed in the previous section. Two non-uniform systems show an interesting geographical pattern. Dutch-speaking Belgium quite consistently has the following system: 1-2-3 (modal.fin – modal.inf – V.inf), 3-1-2 (V.pcp – modal.fin – HAVE.inf),  2-3-1 (GO.inf – V.inf – BE.fin). This system is rare but certainly not absent in The Netherlands. The most frequent system (140 locations) has 1-2-3 (modal.fin – modal.inf – V.inf), 3-1-2 (V.pcp – modal.fin – HAVE.inf) and 1-2-3 (BE.fin – GO.inf – V.inf). This system is found all over The Netherlands with the exception of Friesland. It can also be found in Belgium, in particular in Belgisch Limburg.

1.4 Literature on Verb clusters

Ackema, Peter (2004). ‘Do Preverbs Climb?’ In K. É. Kiss & H. van Riemsdijk     (eds.),  Verb Clusters: A Study of Hungarian, German and Dutch.             Amsterdam/Philadelphia: John Benjamins. 359-393.

Arfs, M. (2007). Rood of groen?: de interne woordvolgorde in tweeledige

            werkwoordelijke eindgroepen met een voltooid deelword en een       hulpwerkwoord in bijzinnen in het hedendaags Nederlands. Proefschrift             Göteborg. Acta Universitatis Gothoburgensis, 2007.

Barbiers, S. (2005). ‘Word order variation in three-verb clusters and the division of            labour between generative linguistics and sociolinguistics.’ In L. Cornips & K. Corrigan (eds.), Syntax and Variation. Reconciling the Biological and the     Social. Amsterdam/Philadelphia: John Benjamins. 233-264.

Bech, Gunnar (1955). Studien zum Deutschen Verbum Infinitum. Tübingen: Max    Niemeyer.

Bennis, H. & T. Hoekstra (1989). ‘Why Kaatje Wasn’t Heard Sing a Song.’ In D. Jaspers, W. Klooster, Y. Putseys & P. Seuren (eds.), Sentential             Complementation and the Lexicon: Studies in Honour of Wim de Geest.       Dordrecht: Foris. 21-40.

Bloemhoff, H. (1977). Enkele volgordeverschijnselen in het Stellingwerfs in verband         met V-raising. Doctoraalscriptie Nederlands Instituut (RUG). Unpublished.

Bouma, G. & G.J. van Noord (1996). ‘Word Order Constraints on Verb Clusters in           German and Dutch.’ Groningen University. Unpublished.

Broekhuis, H. & Hoekstra, K. (1990). ‘Verb Raising, Extrapositie en controle.’ TABU       20, 153-174.

Broekhuis, H. (1993). ‘Verb Projection Raising.’ Spektator 22, 28-47.

—  (1997). ‘Nogmaals Verb Projection Raising.’ TABU 27, 1-27.

Broekhuis, H., H. den Besten & K. Hoekstra (1995). ‘Infinitival Complementation in        Dutch: On Remnant Extraposition.’ The Linguistic Review (12), 93-122.

Broekman, H.W. (1995). ‘Verb Clusters in Germanic: The Non-Existence of the     Third Construction’. In T.F. Shannon & J.P. Snapper (eds.), The Berkeley          Conference on Dutch Linguistics, Dutch Linguistics in a Changing Europe.             Lanham: University Press of America. 117-130.

Burridge, K, (1993). Syntactic Change in Germanic. Amsterdam: John Benjamins.

Coppen, P-A. & M. Klein (1992). ‘Het Einde van Verb Raising.’ In M. Klein        (ed.), Nieuwe Eskapades in de Neerlandistiek.. Groningen: Wolters-            Noordhoff.

Cornips, L. & A. Ribbert (2006). ‘Variatie in tweeledige verbale clusters in het                   Heerlens Algemeen Nederlands.’ Nederlandse Taalkunde 11 (1), 59-76.

Coupé, G. (2007). Verb clusters in Late Middle Dutch and Early Modern Dutch.    Radboud University Nijmegen. Unpublished.

Coussé, E. (2003). ‘Volgordevariatie en herinterpretatie in de tweeledige     werkwoordelijke eindgroep in de bijzin.’ Taal & Tongval 55, 138-156.

— (2006). ‘De historische wortels van volgordevariatie in het hebben-perfectum.’ Taal & Tongval 58 (2), 250-277.

Daems, Fr. (1981). ‘Een eigen syntactische norm in Vlaanderen? Volgordevariatie in           werkwoordgroepen’. Handelingen 36e Ned. Filologencongres 36, 119-125.

De Cubber, W. (1986). ‘"...wat er den dag te voren is voorgevallen te Lokeren...",   Bijdragen tot de studie van de dubbele werkwoordgroep’. In  M. Devos en J.     Taeldeman (eds.), Vruchten van z’n akker: opstellen van (oud-)medewerkers         en oud-studenten voor Prof. V.F. Vanacker, hem aangeboden bij zijn afscheid    van de Rijksuniversiteit Gent. Gent: Seminarie voor Nederlandse Taalkunde            en Vlaamse Dialectologie, Rijksuniversiteit Gent,  63-78.

De Haan, G. (1992). ‘The Verbal Complex in Frisian.’ Tydskrift foar Frisistyk 41: 59-92.

— (1993). ‘The Third Construction in Frisian.’ In W. Abraham & J. Bayer (eds.)   Dialektsyntax. Opladen: Westdeutscher Verlag. 117-130.

— (1996). ‘Recent Changes in the Verbal Complex of Frisian.’ In A. Petersen & H. F.        Nielsen (eds.), A Frisian and Germanic Miscellany: Published in Honour of       Nils Århammar on his Sixty-fifth Birthday, 7 August 1996. Bredsted: Odense           University Press. 171-184.

De Meersman, F. (1990). ‘Bijzinsvolgorde in tweeledige ww-groepen.’       Dialectsyntaxis – Themanummer 3 van Taal & Tongval. 152-185.

Den Besten, H. den, & H. Broekhuis (1992). ‘Woordvolgorde in de werkwoordelijke         eindreeks’. GLOT 12, 79-137.

Den Besten, H., J. Rutten & T. Veenstra (1988). ‘Verb Raising, extrapositie en de derde constructie. Verslag van de onderzoeksgroep “Werkwoordclusters en        ‘vrije woordvolgorde”.’ University of Amsterdam. Unpublished.

Den Besten, H. & H. Broekhuis (1992). ‘Verb Projection Raising in het Nederlands.’         Spektator 21: 21-34.

Den Besten, H. & J. Rutten (1989). ‘On Verb Raising, Extraposition and Free Word          Order in Dutch.’ In D. Jaspers, W. Klooster, Y. Putseys & P. Seuren             (eds.). Sentential Complementation and the Lexicon: Studies in Honour of   Wim de Geest. Dordrecht: Foris. 41-56.

Den Besten, H. & J. A. Edmondson (1983). ‘The Verbal Complex in Continental   West Germanic.’ In Werner Abraham (ed.). On the Formal Syntax of the             Westgermania. Amsterdam: John Benjamins. 155-216.

Den Dikken, M. (1989). ‘Verb Projection Raising en de analyse van het IPP-effect’.          TABU 19, 59-75.

— (1994). ‘Minimalist Verb (Projection) Raising’. In J.-W. Zwart et al. (eds.), GAGL 37. Minimalism and Kayne’s Asymmetry Hypothesis. Groningen. 71-88.

— (1995). ‘Verb (Projection) Raising, Scope, and Uniform Phrase Structure.’ In J. N.         Beckman (ed.), Proceedings of the North Eastern Linguistics Society 25.       Amherst: Graduate Linguistics Student Association. 95-110.

— (1996). ‘The Minimal Links of Verb (Projection) Raising’. In W. Abraham, S.D.            Epstein, H. Thrainsson & C.J.-W. Zwart (eds.), Minimal Ideas: Syntactic          Studies in the Minimalist Framework. Amsterdam: Benjamins. 67-96.

De Rooij, J. (1986). ‘Waarom het naamwoordelijk gezegde groen blijft.’ Onze Taal             55, 142-144.

— (1991). ‘Ik had al moeten eindigen vs. Ik moest al geëindigd hebben.’ In Gramma           15, 235-246

De Schutter, G. (1964). ‘De dubbele woordgroep in het Nederlands.’ Studia           Germanica Gandensia 6, 45-84.

— (1974). ‘Wezen vissen. Dialektgeografie van een konstruktie.’ Taal & Tongval 26,         70-85.

— (1976). ‘De bouw van de Nederlandse zin. Beschrijving en voorstel tot beregeling.’        Verslagen en Mededelingen van de Koninklijke Vlaamse Academie, 165-282.

— (1991). ‘Syntactische verklaringen voor werkwoordvolgordes’. De Nieuwe         Taalgids 82, 312-319.

— (1995). ‘Werkwoordvolgorde en het IPP-effect in het Nederlandse taalgebied’. Taal       & Tongval, 31-39.

— (1996). ‘De volgorde in tweeledige werkwoordelijke eindgroepen met voltooid   deelwoord in spreek- en schrijftaal.’ Nederlandse Taalkunde 1 (3), 207-220.

De Sutter, G., D. Speelman, D. Geeraerts & S. Grondelaers (2003).             ‘Woordvolgordevariatie in tweeledige werkwoordelijke eindgroepen. Naar een             statistische evaluatie van zes factoren.’ In T. Koole, J. Nortier en B. Tahitu            (eds.) Artikelen van de vierde sociolinguistische conferentie. Delft: Eburon.             111-121

De Sutter, G., D. Speelman & D. Geeraerts (2005). ‘Regionale en stilistische effecten        op de woordvolgorde in werkwoordelijke eindgroepen.’ Nederlandse    Taalkunde 10 (2), 97-128.

De Sutter, G. (in press). ‘Naar een corpusgebaseerde, cognitief-functionele verklaring         van de woordvolgordevariatie in tweeledige werkwoordelijke eindgroepen.’ Nederlandse Taalkunde.

Duinhoven, A.M. (1985). ‘De deelwoorden vroeger en nu.’ Voortgang. Jaarboek   voor de Neerlandistiek VI, 97-138.

— (1988). Middelnederlandse Syntaxis – synchroon en diachroon, deel 2 (De          werkwoordgroep). Leiden: Martinus Nijhoff.

— (1998). ‘Concurrerende volgordepatronen in de werkwoordgroep.’ Nederlandse             Taalkunde 3 (2), 96-119.

Evers, A. (1982). ‘Twee functionele principes voor de regel “Verschuif het             werkwoord”’. GLOT 1, 11-30.

— (1975). The Transformational Cycle in Dutch and German. Bloomington: Indiana          University Linguistics Club.

Fanselow, G. (1989). ‘Coherent Infinitives in German.’ In C. Bhatt, E. Löbel, & C.            Schmidt (eds.), Syntactic Phrase Structure Phenomena in Noun Phrases and    Sentences. Amsterdam: John Benjamins. 1-16.

Frings, T. & J. Vandenheuvel (1921). Die südniederlåndischen Mundarten. Texte, Untersuchungen, Karten. Marburg, Elwert’sche Verlagsbuchhandlung.

Gerritsen, M. (1991). Atlas van de Nederlandse dialectsyntaxis (AND). (= Publikaties        van het P.J. Meertens Instituut voor dialectologie, volkskunde en naamkunde      van de Koninklijke Nederlandse Akademie van Wetenschappen 17).            Amsterdam, KNAW.

Guéron, J. & T. Hoekstra (1988). ‘T-Chains and the Constituent Structure of        Auxiliaries.’ In A. Cardinaletti, G. Giusti, & G. Cinque (eds.), Constituent      Structure. Dordrecht: Foris . 35-100.

Haegeman, L. (1990). ‘The Syntax of Motional Goan in WF.’ In R. Bok-Bennema &         P. Coopmans (eds.), Linguistics in the Netherlands 1990. Dordrecht: Foris. 81-  90.

— (1992). Theory and Description in Generative Grammar: A Case Study in West             Flemish. Cambridge: Cambridge University Press.

— (1994). ‘Verb Raising as Verb Projection Raising: Some Empirical Problems.’     Linguistic Inquiry 25 : 509-522.

— (1995). ‘IPP Constructions and V-Movement in West Flemish.’ In M. Starke, E.           Haeberli & C. Laenzlinger (eds.), Geneva Generative Papers 3.1. Geneva:           University of Geneva. 50-76.

— (1998a). ‘‘Extraposed’ Clauses in the West Germanic SOV Languages.’ In M.    Starke & E. Haeberli (eds.), Geneva Generative Papers 6.1. Geneva:             University of Geneva. 61-70.

— (1998b). ‘Verb Movement in Embedded Clauses in West Flemish.’ Linguistic    Inquiry 29 : 631-656.

— (1998c). ‘V-Positions and the Middle Field in West Flemish.’ Syntax (1): 259-299.

Haegeman, L. & H. van Riemsdijk (1986). ‘Verb Projection Raising, Scope and the             Typology of Rules Affecting Verbs.’ Linguistic Inquiry 17 : 417-466.

Haeseryn, W. (1990). Syntactische normen in het Nederlands. Een empirisch          onderzoek naar volgordevariatie in de werkwoordelijke eindgroep. PhD             Dissertation Nijmegen.

Haider, H. (1986). ‘Nicht-Sententiale Infinitive.’ Groninger Arbeiten zur    Germanistischen Linguistik 28 : 73-114.

— (1991). ‘Fakultativ Kohärente Infinitivkonstruktionen.’ Working Papers of the Sonderforschungsbereich 340.

Haslinger, I. (2007). The Syntactic Location of Events. Aspects of Verbal      Complementation in Dutch. PhD Dissertation Tilburg University.

Hinterhölzl, R. (1997). ‘A VP-Based Approach to Verb Raising.’ In K. Kusumoto            (ed.), Proceedings of the North Eastern Linguistics Society 27. Amherst:         Graduate Linguistics Student Association. 187-201.

— (1999). Restructuring Infinitives and the Theory of Complementation. PhD         dissertation, Los Angeles, University of Southern California.

Hoeksema, J. (1988). ‘A Constraint on Governors in the West Germanic Verb       Cluster’. In M. Everaert, A. Evers, R. Huybrechts & M. Trommelen (eds.), Morphology and Modularity. Dordrecht: Foris. 147-161.

Hoekstra, E. (1994). ‘Woordvolgorde en het Infinitivus-pro-Participio-Effect in het

            Zaans.’ Taal & Tongval 46, 132-141.

Hoekstra, E. & E. Blom (1996). ‘IPP en werkwoordsvolgorde in het Achterhoeks’.            Taal & Tongval 48, 72-83.

Hoekstra, E. (1998). ‘Analysing Linear Asymmetries in the Verb Clusters of Dutch           and Frisian and their Dialects’. In D. Beerman, D. LeBlanc & H. van      Riemsdijk (eds.), Rightward Movement. Amsterdam: Benjamins. 153-169.

— (1998). ‘Iets over de werkwoordsvolgorde in "dat ie komen kijken is"‘. TABU 27,          171-177.

Hoekstra, E. & W. Taanman (1996). Een West-Friese Gradatie van het Infinitivus-            Pro-Participio Effect. Nederlandse Taalkunde (1): 13-25.

Hoekstra, J. (1997). The syntax of infinitives in Frisian. PhD Dissertation, Fryske Akademy.

— (1990). ‘Trijetiidwurdkonstruksjes.’ It Beaken 52, 59-95.

Hoekstra, T. (1984). ‘Government and the Distribution of Sentential Complements in       Dutch.’ In W. de Geest & Y. Putseys (eds.) Sentential Complementation.       Dordrecht: Foris. 105-116.

Hoop, H. de en P. Smabers (1987). ‘Rood of groen? Een normenonderzoek naar     enkele stijlverschillen tussen gesproken en geschreven taal.’ De Nieuwe             Taalgids 80, 287-302.

Humboldt, W. von (1836). Ueber die Verschiedenheit des menschlichen Sprachbaus           und seinen Einfluß auf die geistige Entwicklung des Menschengeschlechts.            Berlin: L. Schneider.

IJbema, Aniek (1997). ‘Der IPP-Effekt im Deutschen und Niederländischen.’         Groninger Arbeiten zur Germanistischen Linguistik (40). 137-163.

Kaan, E. (1992). ‘A Minimalist Approach to Extraposition of CP and Verb            (Projection) Raising.’ In D. Gilbers & S. Looyenga (eds.), Language and             Cognition 2. Groningen: Groningen University. 169-179.

Kathol, A. (1996). ‘Order Variability in German and Dutch Verb Clusters.’ In W. Daelemans, G. Durieux, & S. Gillis (eds.) Computational Linguistics in the       Netherlands 1995.Antwerp: University of Antwerp. 147-164.

— (1998). ‘Constituency and Linearization of Verbal Complexes.’ In E. Hinrichs, A.         Kathol, & T. Nakazawa (eds.) Complex Predicates in Nonderivational Syntax.          San Diego/New York: Academic Press. 221-270.

— (1998b). ‘Linearization of Verb Clusters in West Germanic.’ In V. Samiian (ed.),            Proceedings of the Sixteenth Western Conference on Linguistics. Fresno:            CSU. 149-161.

Kern, J.H. (1912). ‘De met het Participium Praeteriti omschreven werkwoordsvormen      in ‘t Nederlands.’  Verhandelingen der Koninklijke Akademie van             Wetenschappen te Amsterdam, Nieuwe Reeks, dl. XII, nr. 2. Amsterdam:     KNAW.
Kiss, T. (1995). Infinite Komplementation: Neue Studien zum Deutschen Verbum   Infinitum. Tübingen: Max Niemeyer.

Koeneman, O. & E. Postma (2006). ‘Veranderingen in Friese werkwoordsclusters.’ In       Nederlandse Taalkunde 11 (2), 124-145.

Koopman, H. & A. Szabolcsi (2000). Verbal Complexes. Cambridge: MIT Press .

Kroch, A.S. & B. Santorini (1987). ‘The Derived Constituent Structure of the West           Germanic Verb Raising Construction’. In R. Freidin (ed.), Principles and             Parameters in Comparative Grammar. Cambridge: MIT Press. 269-338.

Lattewitz, K. (1993). ‘A Minimalist View on Verb Raising’. In D. Gilbers & S.      Looyinga (eds.), Language and Cognition 3. 159-168.

— (1997). ‘Movement of Verbal Complements’. In J.R. Black & V. Motapanyane             (eds.), Clitics, Pronouns and Movement. Amsterdam: Benjamins. 151-174.

Meeuwsen, A.E. & Vanacker, V.F. (1951). ‘De dubbele werkwoordgroep’. De       Nieuwe Taalgids 44, 36-50.

Meurers, W. D. (1999). Lexical Generalizations in the Syntax of German Non-Finite          Constructions. PhD dissertation, University of Tübingen.

Michels, L.C. (1952). ‘Heengaan als hulpwerkwoord van aspect.’ De Nieuwe         Taalgids 45, 113-114.

Minnaert, N. (1998). Een onderzoek naar de regionale spreiding van de volgorde en         doorbreking van de werkwoordelijke eindgroep in de Zuid-Nederlandse    dialecten. Licentiaatsverhandeling Ghent University. Unpublished.

Naarding, J. (1952). ‘Hij ging en kocht een zevenschot.’ De Nieuwe Taalgids 44,    342-344.

Nuijtens, E. (1962). De tweetalige mens: een taalsociologisch onderzoek naar het gebruik van dialect en cultuurtaal in Borne. Assen: Van Gorcum.

Overdiep, G.  (1931-35). Zeventiende-eeuwsche syntaxis. Groningen.

Pardoen, J. (1986). ‘Werkwoordclustering in de voltooide tijd’. Voortgang,             Jaarboek voor de neerlandistiek VII. 49-76.

— (1991). ‘De interpretatie van zinnen met de rode en groene volgorde.’ Forum der           Letteren 32, 1-22.

Pauwels, A. (1953). De plaats van hulpwerkwoord, verleden deelwoord en infinitief in       de Nederlandse bijzin. Leuven: Symons.

Pauwels, J.L. (1950). ‘De plaats van vervoegd hulpwerkwoord, verleden deelwoord           en infinitief in de Aartschotse bijzin’. Taal & Tongval 2, 55-59.

— (1970). ‘Statistisch onderzoek van de Nederlandse zinsbouw.’ De Nieuwe Taalgids,      Van Haeringennummer, 93-100.

Robbers, K. (1997). Non-Finite Verbal Complements in Afrikaans: A Comparative            Approach. PhD dissertation, University of Amsterdam.

Rosen, S. (1990). ‘Restructuring Verbs are Light Verbs.’ In A. Halpern (ed.),          Proceedings of the Ninth West Coast Conference on Formal Linguistics.             Stanford: CSLI Publications. 477-491.

Rosengren, I. (1992). ‘Zum Problem der Kohärenten Verben im Deutschen.’ In P. Suchsland (ed.) Biologische und Soziale Grundlagen der Sprachen. Tübingen:        Max Niemeyer. 265-296.

Ross, J. R. (1986). Infinite Syntax! Norwood, NJ: Ablex .

Sabel, J. (1996). Restrukturierung und Lokalität: Universelle Beschränkungen für Wortstellungsvariationen. Berlin: Akademie Verlag .

Sassen, A. (1963). ‘Endogeen en exogeen taalgebruik.’ De Nieuwe Taalgids 56, 10- 21.

— (1985). ‘Een syntactisch Ingvaeonisme?’ Tabu 15, 156-159.

Schonenberger, M. (1995). ‘Constituent Order in the VP: Verb Raising and Verb    Projection Raising’. In Z. Penner (ed.), Topics in Swiss German Syntax. Bern:   Peter Lang. 347-411.

Schuurman, I. & A. Wierenga (1990). ‘Het Gronings: ‘Verb-raising’ in soorten en   maten’. Taal & Tongval, Themanummer 3, 74-86.

Stroop, J. (1970). ‘Systeem in gesproken werkwoordgroepen.’ Taal & Tongval 22,           128-147.

Sturm, A. (1990). ‘Een syntactische verklaring voor de rode en groene volgorde.’ De          Nieuwe Taalgids 83, 385-404.

Swerts, M. (1998). ‘Ritme als verklarende factor voor de keuze tussen groene en rode        werkwoordvolgorde in het Nederlands.’ Nederlandse Taalkunde 3 (4), 299-          308.

Tuerlincx, J. (1865). Bijdrage tot een Hagelandsch idioticon. Gent: Uitgave der       Zuidnederlandsche Maatschappij van Taalkunde.

Van Bree, C. (2000). ‘Nordniederländische und Niederdeutsche Syntax: ein             Friesisches Substrat?’ In D. Boutkan & A. Quak (eds.), Amsterdamer Beitrage           zur älteren Germanistik (Band 54). Amsterdam: Atlanta. 41-74.

Van Brussel, F. (1982). De dubbele werkwoordgroep in het proza van enkele Wase           schrijvers. Licentiaatsverhandeling Ghent University. Unpublished.

Van den Berg, B. (1949). ‘De plaats van het hulpwerkwoord in de voltooide tijden in         de Nederlandse bijzin.’ Taal & Tongval 1, 155-165.

Van der Horst. J. (1994). Kleine Middelnederlandsche syntaxis. Amsterdam: Huis aan        de Drie Grachten.

Van der Meer, G. (1990). ‘De Infinitivus Pro Participio: Een Nieuwe Verklaring.’ Tabu (20) : 29-48.

Vanden Wyngaerd, G. (1989). ‘Verb Projection Raising and the Status of Infinitival           Complements.’ In D. Jaspers, W. Klooster, Y. Putseys & P. Seuren (eds.),              Sentential Complementation and the Lexicon: Studies in Honour of Wim de             Geest. Dordrecht: Foris. 423-438.

— (1994). ‘IPP and the structure of participles’. In C.J.W. Zwart (ed.) Minimalism and    Kayne’s Asymmetry Hypothesis. Groninger Arbeiten zur germanistischen           Linguistik. 265-276.

Van der Wal, M. (1986). Passiefproblemen in oudere taalfasen. PhD Dissertation Leiden.

Van de Velde, M. (1986). ‘De volgorde binnen de drieledige werkwoordgroep.’ In B.          Callebaut et al (eds.) Linguistische en socio-culturele aspecten van het       taalonderwijs. Handelingen van het tweede Fakulteitscolloquium (1982),     Ghent University. 139-148.

— (1986). ‘De volgorde binnen de drieledige werkwoordgroep. Een vervolg.’ In  M.           Devos en J. Taeldeman (eds.) Vruchten van z’n akker: opstellen van (oud-)       medewerkers en oud-studenten voor Prof. V.F. Vanacker, hem aangeboden bij       zijn afscheid van de Rijksuniversiteit Gent. Gent: Seminarie voor Nederlandse    Taalkunde en Vlaamse Dialectologie, Rijksuniversiteit Gent, 379-395.

• Van Leuvensteijn, J.A., (1987). ‘De werkwoordelijke eindgroep in Gouds taalgebruik         uit de 14de en de 16de eeuw.’ Taal & Tongval 39, 71-94.

Verhasselt, J. (1961). ‘Verschil tussen Noord en Zuid inzake de volgorde    hulpwerkwoord-hoofdwerkwoord.’ Taal & Tongval 13, 153-157.

Weijnen, A. (1937). Onderzoek naar de dialectgrenzen in Noord-Brabant. 137-139.

Wierenga, A. (1986).  ‘Reactie op Een syntactisch Ingvaeonisme?’ Tabu 16, 38-39.

Williams, E. (2004). ‘A Molecular Theory of Head to Head Movement with Special          Reference to Verb Clusters.’ In K. É. Kiss & H. van Riemsdijk (eds.) Verb      Clusters: A Study of Hungarian, German and Dutch.            Amsterdam/Philadelphia: John Benjamins.

Winkler, J. (1874). Algemeen Nederduitsch en Friesch dialecticon. Den Haag:         Martinus Nijhoff.

Wurmbrand, S. (2001). Infinitives: Restructuring and Clause Structure. Berlin/New            York: Mouton de Gruyter .

— (2004). ‘Verb Clusters in West-Germanic: The Empirical Domain.’ In K. É. Kiss &       H. van Riemsdijk (eds.). Verb Clusters: A Study of Hungarian, German and    Dutch. Amsterdam/Philadelphia: John Benjamins.

— (2005). ‘Verb Clusters, Verb Raising, and Restructuring." In M. Everaert & H. van        Riemsdijk (eds.). The Blackwell Companion to Syntax. Oxford: Blackwell        Publishing.

Zuckerman, S. (2001). The acquisition of ‘optional’ movement. (Groningen             dissertations in linguistics 34) . Enschede: Ipskamp.

Zwart, C.J.-W. (1995). ‘A note on verb clusters in the Stellingwerf dialect’. In M.             den Dikken & K. Hengeveld (eds.), Linguistics in the Netherlands 1995.   Amsterdam: Benjamins. 215-226.

— (1996). ‘Verb Clusters in Continental West Germanic Dialects’. In J. Black & V.           Motapanyane (eds.), Microparametric Syntax and Dialect Variation.        Amsterdam: Benjamins. 229-258.

— (1997). The Morphosyntax of Verb Movement. Dordrecht: Kluwer .