DynaSAND

Syntactic Atlas of Dutch Dialects – Commentary – Volume II

Table of contents Volume II

                        2.1                   Introduction

                        2.1.1                Interruption of the verbal cluster

                        2.1.1.1             Verb Projection Raising (VPR)

                        2.1.1.2             Particle Incorporation

                        2.1.1.3             Third Construction

                        2.1.1.4             Interruption by te

                        2.1.1.5             Noun Incorporation

                        2.1.1.6             Historical development

                        2.1.2                Morphosyntax and the verbal cluster

                        2.1.2.1             Infinitivus pro Participio

                        2.1.2.2             Participium pro Infinitivo

                        2.1.2.3             Imperativus pro Infinitivo and verbal hendiadys

                        2.2                   Discussion of the literature

                        2.2.1                Overview of the literature on cluster interruption

                        2.2.1.1             Nominal, adverbial and prepositional material          

                        2.2.1.2             Particles

                        2.2.1.3             Position of te

                        2.2.1.4             Noun Incorporation

                        2.2.2                Morphosyntax and the verbal cluster - overview of the literature

                        2.2.2.1             Infinitivus pro Participio

                        2.2.2.2             Participium pro Infinitivo

                        2.2.2.3             Imperativus pro Infinitivo and verbal hendiadys

                        2.3                   Discussion of the maps

                        2.3.1                Verb Projection Raising

                        2.3.1.1             Interruption of the verbal cluster: bare noun  (map 28a)

                        2.3.1.2             Interruption of the verbal cluster: adverbial   (map 28b)

                        2.3.1.3             Interruption of the verbal cluster: plural noun (map 29a)

2.3.1.4             Interruption of the verbal cluster: indefinite object (map 29b)

                        2.3.1.5             Interruption of the verbal cluster: definite object (map 29c)

                        2.3.1.6             Interruption of the verbal cluster: prepositional phrase (map 30a)

                        2.3.1.7             Interruption of the verbal cluster - synthesis I (map 30b)

                        2.3.2                Interruption by particles

                        2.3.2.1             Interruption of the verbal cluster: adverbial particle (map 31a)

                        2.3.2.2             Interruption of the verbal cluster: adpositional particle (map 31b)

                        2.3.2.3             Interruption of the verbal cluster - synthesis II (map 32a)

                        2.3.3                Third Construction (map 32b)

                        2.3.4                Interruption by te 'to'

                        2.3.4.1             Presence of te 'to' in the verbal complex (map 33a)

                        2.3.4.2             Position of te 'to' in the verbal complex (map 33b)

                        2.3.5                Noun incorporation

                        2.3.5.1             Noun incorporation and verbal hendiadys (map 34a)

                        2.3.5.2             Incorporation of noun and adposition (map 34b)

                        2.3.6                Morphosyntax and the verbal cluster

                        2.3.6.1             Infinitivus pro Participio (IPP)

                        2.3.6.1.1          Infinitivus pro Participio in the verbal cluster (map 35a)

                        2.3.6.1.2          Infinitivus pro Participio without verbal cluster (map 35b)

                        2.3.6.1.3          IPP in verbal cluster with the verb proberen 'try' (map 36a)

                        2.3.6.2             Forms of the verbs - infinitival and past participle morphology

                        2.3.6.2.1          Form of the participle of the modal verb willen 'want' (map 36b)

                        2.3.6.2.2          Form of the participle of the modal verb kunnen 'can' (map 37a)

                        2.3.6.2.3          Correlation between IPP, the presence of the prefix ge- and word order (map 37b)

                        2.3.6.3             PPI & IPI

                        2.3.6.3.1          Participium pro Infinitivo (PPI) (map 38a)

                        2.3.6.3.2          Imperativus pro Infinitivo (IPI) (map 38b)

                        2.4                   Literature on Verb-cluster interruption and verbal morphosyntax

Chapter 2

The verbal cluster: interruption and morphosyntax

2.1 Introduction

As discussed in the previous chapter, a typical property of Dutch is that verbs cluster at the end of the clause, with the exception of the finite verb in main clauses. Complex clusters such as (... dat hij dit probleem) moet willen kunnen hebben proberen op te lossen (that he this problem must want can have try to solve) are grammatical, although extremely rare in the spoken language. This pile-up of verbs is often considered to be the result of the multiple application of a process that is called Verb Raising, in which a verb is separated from its complements and attached to its matrix verb. The preceding chapter has demonstrated that a lot of variation can be observed in the precise manner in which verbs cluster, in particular with respect to the order of the verbs.

To make things even more complicated, the verbal cluster does not allow verbal material only; it can be interrupted by nominal, prepositional and adjectival material. In the first part of this chapter we discuss the various possibilities for such cluster break-ups. Various subcases of cluster interruption are distinguished in this chapter.

The phenomenon often called Verb Projection Raising is concerned with the process in which non-verbal material enters a verb cluster that is created in the context of a modal or causative matrix verb. An example is Ik denk dat Jan mij wil een boek geven (I think that John me wants a book give) in which the verbal cluster wil geven is interrupted by an indefinite direct object. The phenomenon of Verb Projection Raising is found in various West-Germanic language varieties, such as Afrikaans, German dialects and Swiss-German (Wurmbrand 2004, 2006). According to the literature (e.g., Haeseryn 1990), this phenomenon is found in the Dutch language area in Belgium predominantly, in particular in West-Vlaams. Maps 28a-30b provide an illustration of the variation that is found in this construction.

Another, possibly related, phenomenon is the case of Particle Incorporation. A verbal particle, such as op ('up') in opbellen ('call-up'), can be separated from the verb and may appear in the verbal cluster in various positions. The separation of particle and verb is found quite productively in Germanic languages such as English, German and Swedish (Haiden 2006). In the literature on Dutch, different views have been expressed with respect to the acceptability of particle incorporation in different positions. It is clear that this phenomenon is by no means restricted to Flemish dialects. The standard language allows particle incorporation as well. This issue is illustrated on the maps 31a-31b.

A third instance of cluster interruption regards the construction that is calledThird Construction or Partial Extraposition. There is a group of matrix verbs such as proberen that allow full infinitival complements to occur to their right. Some of these verbs also appear to allow constructions in which part of the complement is to the left side of the matrix verb, whereas the verb and the rest of its non-verbal complement show up to the right, as in ... dat Jan zijn broertje heeft geprobeerdnaar school te brengen (... that John his brother has tried to school to bring). This construction is explained in more detail in paragraph 2.1.1.3. The distribution of this phenomenon is shown on map 32b.

Two maps (33a-33b) are included that show the variable occurrence of the infinitival element te in the verbal cluster. Te is generally taken to be a separate functional word marking an infinitival verb. In various constructions the appearance of te seems to be optional in the verbal cluster. This is particularly evident in the construction with matrix verbs such as zitten ('sit') and staan ('stand') which can be considered to be aspectual verbs with a durative interpretation whenever they are used as auxiliary verbs.

In some of the northern Dutch dialects it is possible to incorporate a noun into the verb it is semantically closely related to. Although noun incorporation is not as clear an instance of cluster interruption as Verb Projection Raising – given the fact that the noun in the incorporation structure surfaces in a position between the infinitival marker te and the infinitival verb – noun incorporation can be considered as a special type of verb-cluster interruption. Map 34a presents the distribution of this phenomenon. In addition to noun incorporation, this map also provides the geographical distribution of a different construction: verbal hendiadys. Verbal hendiadys involves the coordination of two verbs: a finite verb of the type staan 'stand' and an infinitival verb; together they express a single activity; verbal hendiadys is comparable to the – in the Dutch language area geographically less restricted – zitten te 'sit to' construction, the difference between the two being the presence of the coordinative element en in verbal hendiadys versus the infinitival marker te in the other construction.

In addition to the variation in the possibilities of interruption of the verbal complex, the Dutch dialects also exhibit quite a lot of variation with respect to the form of the verbs. The final part of this chapter deals with morphosyntactic variation in the verbal cluster.

            Standard Dutch exhibits the Infinitivus pro Participio effect (IPP): an infinitival form appears where we normally would have expected a past participle. Maps 35a-37b illustrate (the distribution of) this phenomenon. In some varieties of Dutch the opposite pattern of the IPP is found: infinitives are replaced by participles (Participium pro Infinitivo); this phenomenon is the subject of map 38a. Finally, map 38b presents the distribution of the Imperativus pro Infinitivo construction. The IPI construction involves the conjunction of two sentences with the coordinative element en 'and', and the replacement of the infinitival verb with a clause-initial imperative in the second conjunct.

2.1.1 Interruption of the verbal cluster

2.1.1.1 Verb Projection Raising (VPR)

Verb Projection Raising is a phenomenon in which the verbal cluster is interrupted by non-verbal material related to the main verb that follows the non-verbal material. Although the name Verb Projection Raising is generally used within the generative framework, we use the name VPR here without theoretical presuppositions, only to distinguish this type of interruption from others. VPR has a number of properties that are illustrated below. First, the process of VPR only occurs if the main verb follows the auxiliary verb. In (1a,b) it can be observed that a main verb and a modal matrix verb occur in two orders, as was discussed in chapter 1. In (1c,d) it is shown that VPR is possible only if the verbs occur in the a-order.

(1)       a.         Ik         weet    dat       Jan       een       nieuwe schuur moet    bouwen.

                        I           know   that      John    a          new     barn     must    construct

                        'I know that John must construct a new barn.'

            b.         Ik         weet    dat       Jan       een       nieuwe schuur bouwen           

                        I           know   that      John    a          new     barn     construct                                 moet.

•                         must

            c.         Ik         weet    dat       Jan       moet    een       nieuwe             schuur

I           know   that      John    must    a          new                 barn     bouwen.

• construct

            d.         *Ik       weet    dat       Jan       bouwen            een       nieuwe

                        I           know   that      John    construct         a          new                                         schuur moet.

•                         barn     must

A second observation is that VPR is optional. It is always the case that alongside a VPR-sentence such as (1c), the non-VPR variant (1a) is available. A third observation is that VPR is possible only with a subset of the verbs that have infinitival complements. It shows up with auxiliary verbs such as modal verbs and the causative verb laten. A fourth observation is that the IPP-effect (cf. map 35a-36a) – the occurrence of an infinitival form instead of a past participle – is retained in VPR. Finally, VPR generally requires the interrupting phrase to be closely related to the following verb.

            We can reformulate most of these properties in a more abstract manner, as in (2).

(2)       VPR

Optionally restructure ... [X - Y] - [V₁-V₂] ... as ... [X] - [V₁ -[Y-V₂]] ...

if          - V₁ is a modal/causative/auxiliary verb

- Y is semantically/grammatically related to V₂.

In the sand questionnaire, a number of distinct types of Verb Projection Raising have been included. These differences relate to the categorial status of the interrupting element Y. The sentences in the questionnaire include the interruption by a bare noun object (brood 'bread'), an indefinite plural object (varkens 'pigs'), an indefinite singular object (een nieuwe schuur 'a new barn'), a definite singular object (de auto 'the car'), a prepositional object (naar Jef  'to Jef'), a sentence adverbial (jammergenoeg 'unfortunately') and an adverb that modifies the verb phrase (vroeg 'early').

            From the literature it is known that the phenomenon of VPR is geographically restricted. In the Dutch language area, it occurs predominantly in Belgium, and it shows up mostly in West-Vlaanderen. Moreover, it is known that the nature of the interrupting element is relevant in allowing VPR to occur.

            The data that were collected in the interviews and that are represented on the maps 28a-30a are brought together in the table in (3). This table indicates the correlation between region and interruptor type (Y in (2)).

(3)       the occurrence of VPR in Belgium (+ the Netherlands' province of Limburg)

The table above shows that the possibility of cluster interruption increases from east to west in the most southern part of the language area. From east to west, we go from Limburg, via Vlaams-Brabant and Oost-Vlaanderen, to West-Vlaanderen. It is remarkable that VPR hardly ever occurs in the province of Antwerpen. This area behaves similar to the Netherlands' part of the language area in which only scattered instances of cluster interruption are attested.

            The condition on semantic relatedness between verb and interruptor in (2) is demonstrated by the difference between a low adverb, i.e., an adverb that modifies the verb phrase (vroeg in vroeg opstaan 'early rise'), which allows VPR quite easily, and a sentence adverbial (jammergenoeg 'unfortunately') which does not occur within the verbal cluster. This observation is probably related to constituents in a verbal cluster resisting a wide scope interpretation (Haegeman & Van Riemsdijk 1986).

Another effect that might be relevant in allowing cluster interruption is the complexity of the interruptor. Bare nouns, low adverbs, and indefinite plurals are one-word interruptions, whereas an (in)definite singular noun phrase and a prepositional phrase consist of minimally two lexical items. The clear difference between indefinite plural and indefinite singular suggests that one-word interruption is considered to be more acceptable than interruption by a complex phrase.

2.1.1.2 Particle Incorporation

Verbal particles are adpositional, adverbial, adjectival or nominal elements that have a close semantic relation to the verb. Examples are [[op]_P [maken]] 'put on make-up' (literally: make up), [[weg]_Adv [maken]] 'lose' (literally: make away), [[dood]_A [maken]] 'kill' (literally: make dead) and [[huis]_N [houden]] 'play havoc' (literally: hold house). Often, the particle-verb combination is semantically untransparent in such a way that the meaning of the combination cannot be derived, or only partially so, by composition of the meaning of the parts. For instance, the verbal complex op-maken 'apply make-up' is not a transparent combination of the meanings of the verb maken and the particle op, just as the English combination 'make-up' is not transparent. There are even cases in which the particle-verb combination contains a non-existent verb, as in op-dirken 'doll up' which contains a non-existent verbal element dirken. In the literature, there has been extensive discussion of these particle-verb combinations, which are also called 'separable compound verbs' (scheidbaar samengestelde werkwoorden). The two labels of this construction – particle-verb complex vs. separable compound verb – illustrate the two views on this construction. Some linguists take particle and verb to constitute one, morphologically complex, verb, whereas others argue for a syntactic complex formed by two independent lexical elements (see par. 2.2.1.2). An important ingredient in these discussions is the fact that particle and verb must be separated by verb second, as is illustrated in (4a,b), and may be separated in verbal clusters, as in (4c,d).

(4)       a.         Marie  maakt zich     niet      op.

                        Mary   makes herself not       up

                        'Mary doesn't put on make-up.'

            b.         *Marie op-maakt        zich     niet.

                        Mary   up-makes        herself not

            c.         Marie heeft    zich     niet      willen op-maken.

                        Mary   has       herself not       want    up-make

                        'Mary has not wanted to put on make-up.'

            d.         Marie heeft    zich     niet      op        willen maken.

                        Mary   has      herself             not       up        want    make

                        ‘Mary has not wanted to put on make-up.’

It is clear that the sentences in (4c,d) are directly related to the construction in the preceding paragraph. If we take Y in (2) to be a verbal particle, (4c) is a case of VPR and (4d) a case of a regular verbal cluster (cf. chapter 1). However, if we take op-maken to be a complex verb, (4c) is a case of a non-interrupted verbal cluster, and (4d) is the result of a special rule of particle-verb separation. Data on the distribution of Particle Incorporation might shed some light on the choice between the two competing analyses of this construction.

            Two sentences were included in the sand interviews that were directed towards the distribution of particle incorporation of the type in (4c,d). One sentence with a two-verb cluster and an adpositional particle (op willen eten 'up want eat') and one with a three-verb cluster and an adverbial particle (weg zou moeten gooien 'away should must throw'). The latter construction is interesting for another reason as well. In the literature it has been argued that intermediate positions are available for particles too: in addition to the orders WEG zou moeten gooien (adv-1-2-3) and zou moeten WEG gooien (1-2-adv-3), the order zou WEG moeten gooien (1-adv-2-3) is often taken to be acceptable.

2.1.1.3 Third Construction

It was indicated above that the IPP-effect is present in VPR. In fact, in those cases the past participle is obligatorily replaced by an infinitival verb (cf. par. 2.1.2.1). The fact that IPP is retained under VPR is shown in (5b). However, there is a set of verbs that allow cluster formation and cluster interruption without an IPP-effect; these are found in constructions with matrix verbs such as proberen 'try' and weigeren 'refuse'. In the standard language, these verbs allow infinitival complements as a whole to be placed at the right periphery of the clause (6a); they show up with a simplex verbal cluster, as in (6b); and they occur with cluster interruption, as in (6c).

(5)       a.  Ik    denk    dat       Jan       [Marie             naar     school]            heeft    *gewild /         

     I      think    that      John    Mary   to         school has       wanted /          willen  [brengen].

•      want           bring

            b. Ik    denk    dat       Jan       [Marie]            heeft    *gewild /          willen

                 I      think    that      John    Mary               has       wanted /          want                            [naar    school brengen].

•                  to                school bring

                 'I think that John has wanted to bring Mary to school.'

(6)       a. Ik    denk    dat       Jan       heeft    geprobeerd      /*proberen       [Marie

    I      think    that      John    has       tried                 / try                 Mary               naar     school te         brengen].

•      to                school  to         bring

            b. Ik    denk    dat       Jan       [Marie naar     school]            heeft    geprobeerd /

                 I      think    that      John    Mary   to         school has       tried /  

                 proberen     [te        brengen].        

•                  try              to         bring

            c. Ik    denk    dat       Jan       [Marie]            heeft    geprobeerd /    proberen

                 I      think    that      John    Mary               has       tried /               try

                 [naar           school  te         brengen].        

•                  to                school to         bring

                 'I think that John has wanted to bring Mary to school.'

The construction in (6c) is known as theThird Construction. This construction differs from VPR in the fact that the participle is generally retained, i.e., the IPP-effect is optional (compare 6c and 5b), and in the nature of the matrix verb. Whereas VPR is generally restricted to constructions with auxiliary verbs of different types (modal, causative), the Third Construction shows up with matrix verbs that select a full infinitival clause as their complement. These verbs require the embedded verb to be directly preceded by the infinitival marker te 'to'.

            From the literature it is known that the geographical distribution of the Third Construction differs from the distribution of VPR. The maps 28a-30b show quite clearly that VPR is mainly restricted to Belgium. Cluster interruption with a prepositional phrase (map 30a / example (5b)) is clearly marked, and does not show up north of the border between Belgium and the Netherlands. However, the Third Construction in (6c) has been argued to be available as a marked construction in the standard language. In the sand questionnaire one relevant instance of the Third Construction has been included. The distribution of this clause – with an indefinite object as interruptor – shows quite clearly that this construction has a completely different distribution from VPR. It occurs throughout the language area. This implies that the Third Construction should indeed be kept apart from VPR.

2.1.1.4 Interruption by te

The presence or absence of te as an infinitival marker shows a great deal of variation in the standard language, largely related to properties of the matrix verb. In general, the generalisation holds that matrix verbs that select a full sentential complement require te as an infinitival marker when the complement is a non-finite clause. Verbs such as proberen 'try' or beloven 'promise' take an infinitival complement with an optional complementiser (om) and the infinitival marker te. With auxiliary verbs such as modals, the causative verb laten 'let' and the verba sentiendi, among which horen 'hear' and zien 'see', the infinitival complement does not allow a complementiser and must be te-less. Examples are provided in (7).

(7)       a.         Jan       probeert /        belooft             (om)    een       artikel *(te)

                        John    tries     /           promises         for       a          paper   to                                 schrijven.

•                         write

                        'John tries / promises to write a paper.'

            b.         Jan       wil/      laat mij/           ziet      mij       (*om)  een       artikel

                        John    wants/ lets me/            sees     me       for       a          paper                           (*te)    schrijven.

•                         to         write

                        'John wants to / allows me to / sees me write a paper.'

Although the general strategy is quite clear, there is a lot of variation in the standard language. There are matrix-verbs such as helpen ('help') and leren ('learn') that allow infinitival complements with or without te, often corresponding to slightly different interpretations. Secondly, there are verbs such as durven ('dare') and hoeven ('have to') that differ among speakers of the standard language in requiring infinitival complements with or without te. Examples are provided in (8).

(8)       a.         Jan       helpt/   leert     mij       een       artikel (te)      schrijven.

                        John    helps/ teaches me       a          paper   to         write

                        'John helps me/ teaches me to write a paper.'

            b.         Jan       durft/   hoeft    mij       niet      de        waarheid          (te)     

                        John    dares/   has-to me       not       the       truth                to                                 vertellen.

•                         tell

                        'John doesn't dare / doesn't have to tell the truth.'

Finally, a rather strange phenomenon can be attested with respect to a small set of derived auxiliary verbs. It concerns verbs such as zitten ('sit') and staan ('stand') that normally function as main verbs. They can be used as auxiliary verbs as well, in which case they receive a durative interpretation. In those cases, these verbs allow te to be dropped from the infinitival complement, but only if these auxiliary verbs are infinitival themselves, which implies that te-drop is available in the verbal cluster only. This is shown in (9).

(9)       a.         Jan       zit        een       artikel *(te)     schrijven.

                        John    sits      a          paper   to         write

                        'John is writing a paper.'

            b.         Jan       zal       wel      een       artikel zitten   (te)      schrijven.

                        John    will      part    a          paper   sit        to         write

                        'John may be writing a paper.'

            c.         Jan       probeert          om       de        hele      dag       te         zitten   (te)

                        John    tries                 for       the       whole day      to         sit        to                     schrijven.

•                         write

                        'John tries to write all day long.'

In the sand questionnaire, and in particular in the telephone interviews, the latter construction is included because we wanted to know whether the attested variation is (in part) determined by geographical factors. It turns out that there appear to be no clear geographical patterns that determine the presence of te in the verbal clusters in (9b,c). The pattern that shows up most clearly is that te is preferably present in clusters such as (9b) – 184 (+te) vs 136 (-te) cases – and preferably absent in (9c) – 106 (+te) vs 159 (-te) instantiations.

2.1.1.5 Noun Incorporation

Noun incorporation is a phenomenon that some argue occurs in the northern dialects of the Netherlands. It involves the fusion of a noun and a verb, yielding a complex verb. A noun can only be incorporated into the verb when the two are semantically closely related. In Standard Dutch this phenomenon occurs, but is not productive. A transparent example is the verb/verbal complex stof-zuigen (literally: dust-suck, 'vacuum-clean'). In some dialects this phenomenon is more productive. An example of noun incorporation in a northern dialect is given in (10); in this sentence the bare noun par 'pear' forms a complex with the verb skilen 'peal'.

(10)     Marie  sit        te         par-skilen.                                           (Jorwerd)

            Mary   sits      to         pear-peel

            'Mary is peeling pears.'          

As this example shows, the noun is in between the infinitival marker te and the infinitival main verb; given that the element te is part of the verbal complex, noun incorporation of this type can be seen as another type of verb-cluster interruption.

There are some conditions the noun and the verb need to satisfy in order to constitute a well-formed noun incorporation structure (cf. Schuurman & Wierenga 1986, Dijk 1997). First, the noun needs to be interpreted non-referentially. Related to this condition is the fact that the noun cannot have a determiner or a modifier, as exemplified in the Frisian sentences in (11) below; nor can the incorporated element be a proper noun or a personal pronoun.

(11)     a.         De       buorlju             sieten   bûten               te (*de/*dy/ *sokke)  

                        the       neighbours       sat       outdoors          to  the/ that/ such                                wyn-drinken. 

•                         wine-drink

                        'The neighbours were sitting outdoors drinking wine.'

            b.         Heit     sit        te         (*grouwe)        jerappel-skilen.                      

                        father   sits      to         huge                 potato-peel    

'Father is peeling potatoes.'                            Dijk (1997:16)

Secondly, the verb needs to express an activity; this condition relates to the fact that the auxiliary verbs in noun incorporation structures express durative aspect. Therefore, noun incorporation is found most frequently with te+infinitive constructions in combination with aspectual auxiliaries like zitten 'sit' and staan 'stand', or in the context of the aan het-construction (comparable to the progressive

-ing construction in English).

            In the sand questionnaire, four sentences with noun incorporation in the te+infinitive construction are included. These involve noun incorporation with a bare noun and incorporation with a plural noun. The distribution is found to be restricted to the northeastern part of the Dutch-speaking language area. In addition to these noun incorporation sentences, a construction with incorporation of the adpositional particle voor into the verb zeggen - te voor-zeggen (before-say, 'prompt') - is included, in order to see whether there is a difference in the geographical distribution between noun incorporation and adposition incorporation. Note that we have to distinguish this type of incorporation from the particle incorporation that we have discussed in 2.1.1.2. In this case it concerns the placement of the adposition in between the infinitival marker te and the main verb. It turns out that there does not seem to be a significant difference in distribution between this type of adposition incorporation and noun incorporation.

2.1.1.6 Historical development

The interruption of the verbal cluster by nonverbal material does not appear to have a southern Dutch origin. Hoeksema (1990) shows that the interruption of the verb cluster - of the two types VPR and the Third Construction - occurred rather generally in Medieval Dutch texts and Van der Meer (1990) found that this phenomenon appeared in old Frisian as well, although not very often. In Frisian, the scarcity of VPR can be related to the fact that the regular order of verbs (3-2-1) does not provide a context for cluster interruption (cf. 2.1.1.1). Koelmans (1965) and Hoeksema (1994) – both of them discuss the historical development of data regarding interrupted verb sequences – note that the option of splitting up the members of a verb cluster gradually declines in the course of the 17^th century and is nowadays an available configuration in the southern Dutch-speaking area only. The fact that this type of interruption of the verbal cluster becomes less ordinary should be seen, according to Hoeksema (1994), as a change in language use, rather than a grammatical change.

Whereas the southern language area generally allows incorporation of non-verbal material into the verb cluster, it is often mentioned in the literature that, in the case of separable compound verbs, the southern Dutch dialects disfavour splitting particle and verb. The literature on the history of particle splitting is rather limited. Den Besten & Edmondson (1983:199) and Koelmans (1965) note that until the 17^th century, northern as well as southern forms of Dutch particle splitting are rarely found. This may imply, as also suggested by De Cubber (1968), that the tendency to split the particle and the verb has become stronger.

2.1.2 Morphosyntax and the verbal cluster

The subject of the second part of this chapter is the variation in the morphosyntactic properties of verbs in the verbal complex, since varieties of Dutch exhibit variation regarding the form of verbs in different cluster structures, in addition to variation in word order of the verbs in the cluster (chapter 1) and variation with respect to the interruption of the cluster with nonverbal material (section 2.1.1).

2.1.2.1 Infinitivus pro Participio

Standard Dutch exhibits the Infinitivus pro Participio effect (IPP): an infinitival form appears in the position where we would expect a past participle. The phenomenon is found in constructions in which an auxiliary verb in a verbal cluster is directly governed by a perfective auxiliary verb (hebben 'have' or zijn 'be'). The effect is exemplified in sentence (12b): the expected past participle gewild in the verb cluster [*gewild eten] is obligatorily replaced with the infinitival verb willen.

(12)     a.         dat       Jan       het       brood   heeft    gegeten/           *eten.

                        that      John    the       bread   has       eaten

                        'that John has eaten the bread.'

            b.         dat       Jan       het       brood   heeft    willen/*gewild eten.

                        that      John    the       bread   has       want(ed) eat

                        'that John has wanted to eat the bread.’

The literature states that there is a one-to-one relation between cluster formation and the presence of the IPP-effect. This means that in structures with extraposition or in the case of the Third Construction, the IPP-effect is predicted to be absent. In paragraph 2.1.1.3, we already saw that in the context of the verb proberen 'try' this generalisation does not seem to hold. In the sand questionnaire three sentences with the verb proberen are included – one in which proberen is part of the clause-final verb cluster, one with proberen in the Third Construction, and another sentence in which proberen occurs with extraposition. These latter two sentences are included to test the geographical distribution of the appearance of the IPP-effect in sentences without strict cluster formation.

            The occurrence of IPP also correlates with the word order of the verbs in sentence-final position. It appears to be the case that the IPP-effect is generally absent in varieties that show an order in which the main verb precedes auxiliary verbs. This may account for the fact that the dialects spoken in the northeastern part of the language area systematically lack the IPP-effect. Another perspective that has been introduced in the literature is the idea that the IPP-effect correlates with the presence of the perfective prefix ge- on the past participle: in dialects in which there is no perfective prefix ge- the IPP-effect is absent. Map 37b shows the correlation between the presence / absence of the IPP-effect, the presence / absence of the perfective marker ge- and the effect of word order. This map demonstrates that both perspectives have some truth to them and probably need to be taken together to account for all the attested variation.

2.1.2.2 Participium pro Infinitivo

In addition to the replacement of a participle by an infinitival verb (IPP), some varieties of Dutch exhibit the opposite pattern: an infinitive is replaced by a participle, or to be more precise, the presence of a perfective auxiliary triggers the appearance of more than one past participle. This phenomenon is usually called the Participium pro Infinitivo effect (PPI), and its geographical distribution is said to be restricted to Friesland.

(13)     Sol       hij        dat       daan                 kund                hebben?           (Sleen)

would he        that      done.PCP         could.PCP        have.inf          

            'Could he have done that?'

The sand material includes several sentences that each consist of a clause final three-verb cluster that varies in the order of the verbs –V3-V2-V1 or V3-V1-V2 – and their form – the verbs have participial or infinitival morphology. The geographical distribution of the PPI-effect is found to be restricted to the northeastern part of the Dutch-speaking language area: Friesland, Drenthe and Groningen.

2.1.2.3 Imperativus pro Infinitivo and verbal hendiadys

The Imperativus pro Infinitivo construction (IPI), also known as the en + imperative construction, is found in sentences containing a conjunction of two sentences by the coordination marker en 'and'. The second conjunct in an IPI-construction has the form of an imperative clause. From a historical point of view it seems to be the case that the infinitival morphology on the clause-initial verb in the second conjunct has, in the course of time, been replaced by imperative morphology. An example of the IPI-construction is given in (14).

(14)     Hoe     hellest              it          yn        e          kop      en        smyt               

            how     get-you            it          in         your    head     and      throw-2sing                mei      iten!                                                                             (Hindeloopen)

•             with     food

'How dare you fling your food about!'

There seem to be two grammatical types of the IPI-construction. The first type, illustrated in (15a), involves 'real' coordination: the two conjuncts each express a different activity, whereas in the second type, exemplified in (15b) (and (14)), we are dealing with subordination. This second type can always be replaced by a regular om+te+infinitive construction.

(15)     a.         De       plysje soe       by        him      komme            en        helje

                        the       police would at         him      come                and      get.2sing                     him      op.                                                                   (Frisian)

•                         him      up

                        'The police would come and pick him up.'    

b.         Ik         ried      jimme oan      en        drink                net       te        

            I           advise you      prt      and      drink.2sing     not       to                                 folle     kofje.                                                               (Frisian)

•             much   coffee 

            'I advise you not to drink too much coffee.'

J. Hoekstra (1987:97)

Only the IPI-construction of the subordination type is represented in the sand questionnaire. The relevant sentences in the questionnaire differ with respect to the form of the clause-initial verb following the coordinative element en: these verbs have either imperative or infinitival morphology. The data show that the IPI-effect is typically Frisian, and that the distribution of the variant with an infinitive in the first position of the second conjunct is more restricted than the IPI construction with an imperative.

A construction that also involves the coordinative element en 'and' in a subordination construction is verbal hendiadys, as illustrated in (16). Verbal hendiadys involves the coordination of two verbs – a finite verb with a durative interpretation like zitten 'sit' or staan 'stand', and an infinitival verb – by the coordination marker en 'and', although it might be questioned whether the functional element en involved should still be analysed as a coordination marker, rather than an infinitival marker in some of these cases. The two verbs in this construction express a single activity; therefore, constructions involving verbal hendiadys are comparable to zitten te ('sit to')-constructions in that they have a progressive interpretation.

• (16)     E          stoat    en        zagen.                                      (Merckeghem)

            he        stands and      nag      

'He is nagging.'

In verbal hendiadys, the direct object that belongs to the verb in the second conjunct may appear in the first conjunct, as exemplified in (17).

(17)     Marie  zit        patatn              en        schrooën.         (Wulvergem)

Mary   sits      potatoes          and      peel

Mary is peeling potatoes.'

The sand questionnaire includes a sentence with verbal hendiadys of the type in (17), with the direct object appearing in the first conjunct. The data show that this construction has a very restricted distribution; we only found a total of seven instantiations in the southern parts of West-Vlaanderen and Frans-Vlaanderen.

2.2 Discussion of the literature

2.2.1 Overview of the literature on cluster interruption

2.2.1.1 Nominal, adverbial and prepositional material

Regarding the distribution of verb-cluster interruption by nominal, adverbial and prepositional material (VPR), the literature points out that this phenomenon is specific to the southern language area; Standard Dutch and Frisian disallow interruption of the VP-final verb cluster. Interruption of the verb cluster only occurs in the northern dialects when we are dealing with stereotypical expressions, often idioms (cf. Haeseryn (1990:81)), or particles. A typical property of southern Dutch dialects often mentioned in the literature is the interruption of the verbal cluster by a stranded preposition, as in Ge zoudt ER kunnen BIJ neer vallen (literally: you may there [can at down fall]; Koelmans 1965:156). Koelmans (1965) and Vanacker (1970) put forward the generalisation that the number of cluster interruptions decreases across Belgium from west to east. Haeseryn (1990), moreover, mentions that the northern part of the province Antwerpen is an area that allows little or no interruption (cf. Stoops 1969, Bobine 1970). It should be noted that VPR is optional and that in places or regions where interruption is attested, this is by no means the most frequently used variant.

The distribution of the various lexical elements that can interrupt the verbal cluster differs quite substantially. Vanacker (1970) and Hoeksema (1994) note that interruption by a pronoun does not occur. Hoeksema (1994) further notes that interruption by clausal complements is possible, but extremely rare. Vanacker (1970) observes that interruption with a stranded preposition, with a noun(phrase) in the function of direct object, with a locative/directional prepositional phrase, with the complement of a copula and with (part of) an idiosyncratic expression are the most frequent forms of interruption. Interruption of the verbal cluster by other prepositional phrases and adverbials does occur, but significantly less than by a noun phrase or a directional prepositional phrase. For a brief discussion and a map of the interruption of a two-verb cluster by a directional PP, see Vanacker (1964:162). Haeseryn (1990) provides a summary of several studies that focus on the interruption of the verbal cluster by a noun phrase and a prepositional phrase (such as Stoops 1969, Bobine 1970, Mestdagh 1970 and De Bruycker 1972) and found, somewhat different from the results presented by Vanacker (1970), that a direct object very rarely interrupts the verb cluster, whereas a locative/directional PP does interrupt the verb cluster quite often. In two corpora with spontaneous speech from dialect speakers in all areas of Belgium, Minnaert (1998) has looked for sentences with verb-cluster interruption with various kinds of lexical material. She found that in general, the number of sentences with verb clusters that are not interrupted is higher than the number of sentences with interrupted verb sequences. The reverse holds in the case in which the interruptor is a stranded preposition only.

In generative literature, starting with Evers (1975), Verb Raising affects the verb of a non-finite complement and usually causes the verb to move to the right of the matrix verb in Dutch. In Verb Projection Raising structures, not only the verb of the embedded clause is affected, but noun phrases and other constituents of the embedded verb phrase can be incorporated into the verb cluster as well. It is important to note that VPR is different from noun incorporation (cf. Schuurman 1987 and paragraph 2.1.1.5). Den Besten & Edmondson (1983) were the first to give an analysis of VPR in terms of transformational rules. For a detailed overview of the history of VPR, see Hoeksema (1994). The most common analysis of VPR in the generative framework is the Scrambling-plus-Adjunction account (cf. Den Dikken 1989, Den Besten & Rutten 1989, Vanden Wyngaerd 1989, Haegeman 1992, Broekhuis et al. 1995) by which VPR can be analysed as the result of two interacting rules: Scrambling and Adjunction. Haegeman & Van Riemsdijk (1986) argue against an adjunction approach and favour a reanalysis account of VPR. Den Besten & Broekhuis (1992) hypothesise that Dutch is a pure VPR-language; the variation regarding the possibility to interrupt the verb cluster is accounted for by assuming that languages and dialects differ in the amount of scrambling they exhibit. Broekhuis (1993) considers the reordering in VPR constructions to be the result of the application of a phonological rule, instead of a syntactic one. For more recent analyses in the minimalist framework, see Den Dikken (1994), Zwart (1995), Broekhuis (1997) and Koopman & Szabolcsi (2000).

As has already been mentioned in paragraph 2.1.1.3, VPR cannot be equated with the Third Construction. The observation that some verbs allow verb clustering and interruption without an IPP-effect was noted by Den Besten et al. (1988). Since then, the prevailing assumption has been that the Third Construction is quite common in Standard Dutch. Hartevelt & Hoekstra (1999) examined the occurrence of the Third Construction in the Netherlands. Moreover, they wanted to see whether there is variation in the form of the verb, i.e., participle or infinitive, the latter being a manifestation of the IPP-effect. On the basis of data gathered by a written questionnaire they found that the Third Construction occurs everywhere in the Netherlands, but the Third Construction in combination with the IPP effect is only attested in Limburg and the Achterhoek occasionally. Further discussion of the Third Construction can be found in Broekhuis & Hoekstra (1990) and Hoeksema (1990).

2.2.1.2 Particles

The ANS (1997:1357-1358) notes that the separated and the non-separated variant of the particle-verb combination both show up in Standard Dutch. The choice to separate particle and verb is dependent on the area: particle splitting occurs more often in the Netherlands than in Belgium (cf. Verhasselt 1961:156-157, De Schutter & Van Hauwermeiren 1983:191 and Den Besten & Edmondson 1983:199). In the literature on (the distribution of) particles, there has been some debate on the possible positions the particle can occupy within the verb cluster. Van Riemsdijk (1978) notes that sentences in which the particle appears in an intermediate position inside the verb cluster are ungrammatical in Standard Dutch, although these sentences do occur in southern Dutch dialects. Bennis (1992) claims that the distribution of particles in the verb cluster is less restricted; he argues that the particle may appear in every position inside the verb cluster in the standard language. Haeseryn (1990) also accepts some of the intermediate cases.

As mentioned in paragraph 2.1.1.2, there are two views on particle-verb constructions. On one view, particle and verb are taken to be a single, morphologically complex verb. The other perspective assumes that the combination particle-verb consists of two independent lexical elements.

The discussion regarding the placement of particles in the verb cluster in generative literature starts with Koster (1975). He argues that the verb and the particle form a compound verb, i.e., the particle is part of a complex verb. Separation of the particle from the verb proceeds via verb movement, which implies that Dutch has no rule for particle movement. For more recent proposals along these lines see Groos (1989), Model (1991), and Hoeksema (1991). These proposals allow syntactic movement of a part of a lexical item, in violation of the condition on Lexical Integrity. This view is also defended by Neeleman (1994) and Neeleman & Weerman (1993). Van Riemsdijk (1978) adopts the second perspective and argues that the particle projects its own syntactic projection; for comparable approaches see also Bennis (1991, 1992), Vanden Wyngaerd (1994) and Koopman (2000). Based on this view, particle and verb can form a syntactic complex by assuming the existence of a particle incorporation rule. In order to be able to account for the attested positions, including intermediate positions of the particle in the verbal cluster, Bennis (1992) argues for the need to assume particle movement/incorporation in interaction with Verb Raising. An intermediate position has been adopted by Booij (2002a,b) and Blom (2005), who claim that the particle-verb combination is in between a lexical item and a syntactic complex. They refer to this combination as a 'constructional idiom'.

2.2.1.3 Position of te

With respect to the position of te in the verbal complex, the ANS (1997:970-973) notes that the presence of te can be optional in the case of a two verb cluster consisting of a verb such as zitten 'sit' and staan 'stand' in combination with an infinitive+te (see map 33a). Moreover, the position of te in these clusters is also variable to some extent: te sometimes precedes the two infinitival verbs and sometimes it occurs in the position between the verbs, as in Wim schijnt de hele les te zitten slapen and Wim schijnt de hele les zitten te slapen, ‘Wim seems in class (to) sit (to) sleep'. The first of these two sentences is the standard variant, the latter mainly occurs in the spoken language. Sometimes it is even possible to have te present in both positions at the same time as in Wim schijnt de hele les te zitten te slapen. The ANS characterises the occurrence of sentences like these doubtful and mentions that there is a tendency to avoid a sequence of two infinitives with te.

According to Meeussen (1943:50-51), Belgium can be divided into two areas with respect to the position of te in a verb cluster consisting of two infinitives preceded by om. He made use of a written questionnaire with the two following test sentences: Ge moet geld hebben om te kunnen betalen and Ge moet geld hebben om kunnen te betalen (‘You must money have for (to) can (to) pay’; 'You must have money to be able to pay') West-Vlaanderen, the west and the northeast of Oost-Vlaanderen and the north of Antwerpen uniformally have te preceding the two infinitives. The southeast of the Dutch-speaking part of Belgium is a mixed area in that it has both te preceding the verbs and te in an intermediate position. Meeussen suggests to attribute to the influence of Standard Dutch the option of te preceding the verbal cluster in these areas. Doubling of te is found twice in Oost-Vlaanderen (Wetteren and Buggenhout) and a single instance is found somewhat further to the east, namely in St.-Truiden in Belgisch Limburg. Vanacker (1969) found more or less the same pattern on the basis of a written questionnaire and recorded spontaneous speech. However, he notes that the western area is not as homogeneous as Meeussen (1943) leads us to believe, i.e., in this area te sometimes surfaces in the position between the two infinitives.

2.2.1.4 Noun Incorporation

Noun incorporation is a typically northern-Dutch phenomenon; see Ter Laan (1953), Schuurman & Wierenga (1986) and Schuurman (1987) for the Groningen dialect, J. Hoekstra (1993:141) for some observations on noun incorporation in West-Frisian, and Dijk (1997) for an overview of the distribution and an analysis of the phenomenon in Friesland. With respect to the different types of noun incorporation, Dijk (1997) notes that although noun incorporation in Frisian occurs most frequently in the context of te-infinitives and in combination with the oan it-construction, the phenomenon is also observed with participles, and to a lesser extent in the context of a finite verb. In the Groningen dialect, however, noun incorporation into finite verbs is prohibited (cf. Schuurman & Wierenga 1986). With respect to the restrictions on the incorporated noun, Dijk (1997:15) mentions that it cannot be pluralised, i.e., the noun cannot have the plural suffix -en or -s. Nouns that normally take the plural morpheme -en, may occur in noun incorporation structures in combination with a binding -e element only.

With respect to the nature of noun incorporation, there has been some controversy in the international literature: some researchers have claimed that we are dealing with a morphological process (Mithun 1984), while others claim that it is a syntactic process (Sadock 1985,1986; Schuurman 1987). Dijk (1997) argues that Frisian noun incorporation cannot be analysed in terms of syntactic movement; it should be derived in the morphological component.

2.2.2 Morphosyntax and the verbal cluster - overview of the literature

2.2.2.1 Infinitivus pro Participio

A common claim in the literature is that the Infinitivus pro Participio (IPP) effect only occurs in verbal clusters, i.e., structures in which we have cluster formation; it does not appear with extraposition or the Third Construction (e.g., Nieuwenhuijsen 1974; Bennis & Hoekstra 1989; Den Dikken 1989; Vanden Wyngaerd 1994). Although this generalisation seems to hold in the majority of the Dutch dialects, some dialects in the Achterhoek and in Limburg allow the IPP-infinitive in combination with the Third Construction and with extraposition structures (Hartevelt & Hoekstra 1999); see also map 36a. With respect to the geographical distribution of IPP, we can observe that the southern dialects all show the IPP-effect, whereas it is absent in the northeast of the Netherlands, i.e., Frisian and the dialects of Groningen (Weijnen 1966:320).

As mentioned in paragraph 2.1.2.1, there has been some discussion about the nature of IPP and the question to which phenomenon IPP is, or should be, related. Several researchers have suggested that the IPP-effect correlates with word order in the verbal cluster (Hoekstra 1994; De Schutter 1995); more specifically, the generalisation has been put forward that IPP is systematically absent in head-final constructions, i.e., language varieties that have predominantly the order V3-V2-V1 (such as Frisian; see chapter 1). However, data from West-Frisian and Zaans (Hoekstra 1993, 1994, 1998; Pannekeet 1995:385-392; Hoekstra & Taanman 1996) indicate that this generalisation does not hold. West-Frisian has a prevailing head-final wordorder but exhibits the IPP-effect. Hoekstra (1994) proposes that the IPP-effect in Zaans might be related to the word order in which V1 precedes V2. Although this seems to hold for Zaans, it cannot account for the whole Dutch-speaking language area. Also Wolf (1997) notes, on the basis of data from Interference Frisian – a modern variant of Frisian – that inversion, i.e., word order, and IPP are two independent phenomena: inversion and IPP do not necessarily go hand in hand. De Schutter (1995) elaborates on Hoekstra's (1994) generalisation and reformulates it as follows: only if the complex V2-V1 is respected is IPP absent, in all the other possible orders IPP is possible, but crucially not necessary. However, Hoekstra & Blom (1996) show that the dialects of the Achterhoek actually allow the word order V3-V2-V1 in combination with the IPP-effect. Once more, this indicates that the morphological properties of the verbs are not directly related to word order in the verbal complex. De Schutter (2000) emphasises the need to take into account the interference between (current) language varieties in order to give an analysis of the complexity of the data regarding (the relation between) the IPP-effect and word order.

Another suggestion with respect to the relation between the IPP-effect and other phenomena is that the presence of the perfective prefix ge- is the factor that correlates with the IPP (cf. Hoeksema 1980:240-243; Lange 1981; Vanden Wyngaerd 1994). In other words, when a dialect has a perfective prefix to mark the past participle, it also exhibits the IPP and vice versa (regardless of the order of the verbs). According to Lange (1981:64) the double infinitive construction, i.e., the IPP effect, appears in all the Germanic languages in which the perfective participle has the prefix ge-, and in those languages in which the participle is not marked with ge- the IPP-effect is absent. With respect to the Dutch-speaking language area, Van Loey (1965:161) notes that the prefix ge- is absent in the north of North-Holland, Friesland, Groningen, Drenthe, the east of Twente, and the east of the Achterhoek; see also Weijnen (1966:286). However, the 'prefix generalisation' does not seem to be able to account for all the attested data. Hoekstra & Blom (1996) show that the IPP-effect is not obligatory in the dialects of the Achterhoek that have a perfective prefix (in the form of a schwa). Wolf (1996) notes that the prefix-generalisation cannot account for the data from Interference Frisian; Interference Frisian does not have a prefix to mark the past participle, but it has optional IPP. The same holds for West-Frisian: West-Frisian does not have a perfective prefix (cf. Pannekeet 1995:178), but does allow IPP. Hoekstra & Taanman (1996) try to save the prefix generalisation by arguing that the apparent violation in West-Frisian is due to bilingualism of the dialect speakers.

Most analyses of the IPP-effect are based on one of the two perspectives discussed above. For approaches that focus on IPP in connection with word order phenomena, see Den Dikken (1989), Hoekstra (1994) and De Schutter (1995); analyses that try to explain the IPP in terms of a relation with the perfective prefix ge are found in Vanden Wyngaerd (1994) and Hoekstra (1998). Some observations on the historical development of the IPP-construction can be found in Ponten (1971) and Duinhoven (1975). Van der Meer (1990) also provides a brief overview of the history of the IPP-effect; he argues that the existence of the IPP-effect is strongly related to semantic factors: the IPP expresses a particular meaning (cf. Pardoen 1986). Minimalist approaches to the IPP-phenomenon have been provided by Zwart (1995), Haegeman (1998) and Hoekstra (1998).

2.2.2.2 Participium pro Infinitivo

Regarding the geographical distribution of the Participium pro Infinitivo effect (PPI) – or parasitic participles, as the effect is called by Den Dikken & Hoekstra (1997) – Van der Meer (1990) notes that it is a common phenomenon in Swedish and also to some extent in Norwegian. PPI constructions are ungrammatical in Standard Dutch; they are mainly found in Frisian and in some northeastern dialects, including Stellingwerfs, spoken on the border of Friesland, Overijssel and Drenthe (Bloemhoff 1979). In the literature we find two types of PPI construction in Frisian, one in which the order of the verbs is V3-V2-V1 and the modal (V1) takes scope over the perfective auxiliary (V2), and one in which the order of the verbs is V3-V1-V2 and the perfective auxiliary (V1) takes scope over the modal (V2) (cf. map 38a). The two constructions differ in their word order and in their semantics. Moreover, Den Dikken & Hoekstra (1997:1068) observe that it is possible in Frisian to have structures with three participles – two of which are 'parasitic' – in constructions with multiple modals.

The literature on the PPI-effect is very scarce. Hoeksema (1990:183) notes that the PPI might be an assimilation phenomenon, just as the IPP effect, with the difference between the two being the direction of assimilation. Den Dikken & Hoekstra (1997) argue that the PPI cannot be analysed in terms of phonological assimilation - or participle harmony - since the PPI construction can appear without the two participles being adjacent; see also map 38a. They provide a rather minimalistic analysis of the PPI phenomenon based on the claim that PPI constructions need to be analysed in terms of syntactic movement. The PPI-effect is optional, that is to say, PPI constructions always have a counterpart with the same word order and without PPI. This might be an indication that word order instead of morphosyntactic form is a relevant criterion for PPI to appear, suggesting that we need a syntactic analysis. The difference in distribution of the PPI-construction between Standard Dutch (no PPI) and Frisian (PPI) might be attributed to the differences in word order in the verbal cluster between the two languages.

2.2.2.3 Imperativus pro Infinitivo and verbal hendiadys

The Imperativus pro Infinitivo (IPI) – sometimes called "the en + imperative construction" – has been claimed to be a typical Frisian phenomenon (cf. De Waart 1971; Hoekema 1971,1975; De Haan 1990) and it is probably one of the best studied morphosyntactic phenomena in that language. The verb in the second conjunct of the IPI construction has the form of an imperative, although it evolved from an infinitive: the second conjunct in older varieties of Frisian made use of an infinitival verb in initial position, hence the name of the construction (cf. Hoekstra 1987,1997). With respect to the evolution of the construction, Hoekstra (1987,1997) proposes that due to the markedness of clause-initial infinitives, the infinitive was replaced with an imperative, a verb that usually surfaces in clause-initial position. De Waart (1971) was the first to argue that there are two grammatical types of IPI: a subordination and a coordination type. The subordinate IPI-construction evolved from the coordination construction; that is to say, in Old Frisian and in Medieval and 17^th-century Frisian we only find instances of the coordination IPI construction (cf. Hoekstra 1987,1997). De Haan (1990) proposes to abandon the dichotomy between coordination- and subordination-IPI and argues that we are in fact dealing with one grammatical subordination construction of which there are two types: an adjunct- and an argument-IPI-type. This approach implies that we no longer have to assume a rather drastic grammatical transition from coordination to subordination with respect to the historical development of the IPI construction.

As for its geographical distribution, Van der Meer (1975) notes that the IPI construction is not as typically Frisian as has been claimed; see also map 38b. Current Standard Dutch and southern Dutch dialects, however, do not exhibit the IPI construction. There is some controversy regarding the analysis of the IPI effect and its development. Van der Meer (1975) focuses on the coordination IPI type and proposes an analysis in terms of 'conjunction reduction', meaning that the word order in the second conjunct gets deleted and is replaced by the default VO-order. Hoekstra (1987) provides arguments against this analysis and proposes an account in terms of Verb-Second phenomena. Recently, Postma (2005) has argued that the (subordination) IPI-construction actually does not resemble the om+te+infinitive construction so closely as always has been assumed (cf. 2.1.2.3); rather, Postma proposes to relate the IPI construction to the subjunctive construction (both diachronically and synchronically).

Regarding the phenomenon of verbal hendiadys, it has been mentioned in the literature that the verb following the coordinator en need not necessarily be an infinitival verb; it can also be a finite verb (cf. Paardekooper 1993). On map 40 in Gerritsen (1991), which presents data on noun incorporation and verbal hendiadys, only one instance is found of verbal hendiadys in combination with a finite verb; the test sentence was Marie zit aardappelen en schilt (literally: ‘Mary sits potatoes and peels’), in which the direct object of the verb schilt appears in the first conjunct. Verbal hendiadys in combination with an infinitival verb is found sporadically and the instantiations are scattered over the language area, the exception being three occurrences in the southern part of West-Vlaanderen; see also map 34a. Paardekooper (1993) provides an overview of the distribution of verbal hendiadys in the Dutch-speaking language area after 1600 on the basis of a written corpus mostly consisting of literary texts from the 17^th century (see also Strengholt 1970), and focuses (i) on the emergence of the construction and (ii) on the question why and when the construction disappeared. He concludes that en + finite verb were more frequently replaced by te+infinitive in Amsterdam in the course of the 17^th century, and eventually no longer existed. Recently, Haslinger & Van Koppen (2003) have argued that verbal hendiadys is not really a coordination configuration; they rather take it to be an instance of subordination – the fact that the IPP effect occurs with verbal hendiadys and that the direct object may appear in the first conjunct were some of their reasons for this analysis. Regarding the status of en in the verbal hendiadys construction, Haslinger & Van Koppen (2003) note that it cannot be the case that en is a dialectal variant of the Standard Dutch subordination te, since en in verbal hendiadys cannot always be replaced with te.

2.3 Discussion of the maps

2.3.1 Verb Projection Raising

2.3.1.1 Interruption of the verbal cluster: bare noun (map 28a) (map in dynaSAND)

Map 28a sketches the geographical distribution of the sentence in (18), in which the verbal complex in the declarative subordinate clause, consisting of a modal and an infinitive, is interrupted by the bare noun brood 'bread' which functions as the direct object of the verb eten 'eat'. The sentence has been included in the oral interviews in almost every sampling point, yielding a total of 257 locations. The informants were asked whether the sentence occurred and if so, how it should be translated into their dialect.

(18)     Ik         weet    dat       Eddy   morgen            [wil      brood   eten].  

            I           know   that      Eddy   tomorrow        wants bread   eat

            'I know that Eddy wants to eat bread tomorrow.'

As the map shows, this type of verb-cluster interruption is found mainly in West-Vlaanderen and the south-eastern and north-western part of Oost-Vlaanderen. The construction does not occur in the Waasland and the province of Antwerpen, except for two locations, 's Gravenwezel and Zandhoven. In Vlaams-Brabant the occurrences of the construction are rather diffuse, with the exception that it does not occur in the southwestern part of this province. Furthermore, interruption by a bare noun has a single occurrence in Frans-Vlaanderen, in Sint-Marie-Cappel, and is attested only incidentally in the southern part of Limburg. This general pattern correlates with the generalisation made by Koelmans (1965) and Vanacker (1970) that the appearance of verb-cluster interruption decreases when going from west to east. In the Netherlands the construction only occurs sporadically and its instantiations are rather scattered.

Interestingly, interruption of the verb cluster by a bare noun is the only type of VPR interruption of which occurrences are found in the Netherlands. This may lead us to suggest that verb-cluster interruption by a bare noun is distinct from other types of interruption since the results appear to indicate that Belgium and the Netherlands have different strategies regarding non-verbal material occurring in the verb cluster.

The construction in which the bare noun is inserted between the verbs in a VPR cluster appears to be related to the phenomenon of noun incorporation, which is argued to occur in the northern dialects of the Netherlands (cf. 2.1.1.5): a bare noun can be incorporated into the verb(cluster) when the two are semantically closely related. Schuurman (1987:185) argues that in the example in (19) from the Groningen dialect, the bare noun bouk 'book' is incorporated into the verb.

(19)     K         heurde,                        dat       zai       hom     bouk    lezen    luit.

            I           heard               that      she       him      book    read     made

            'I heard that she made him read a book.'

Other examples of noun incorporation are sentences in which the incorporated noun follows te or aan het and precedes the infinitive main verb (cf. maps 34a and 34b). We might argue that we are dealing with noun incorporation in (18) as well, although the semantic relation between bread and eat in (18) seems to be less strong than between piano and play or between book and read in (19).  However, there are at least two arguments to consider these constructions different. First of all, bare-noun-VPR causes the noun to appear in front of the infinitival marker te (if present), whereas noun incorporation of the type in (19) causes the noun to appear in between te and the infinitival verb. Moreover, bare-noun incorporation of the type in (18) shows up in Belgium and in scattered locations in the Netherlands whereas noun incorporation of the type in (19) is found in the northern dialects in the Netherlands only (cf. maps 34a,b).

            Alternatively, we may take the occurrences of bare noun incorporation of the VPR type in the Netherlands as instances of particle incorporation. As has been shown in 2.1.1.2, Dutch allows nominal particle-verb constructions of the type huis-houden ('play havoc') and pianospelen ('play the piano'). The nominal particles are non-referential, bare nouns. The major difference between bare-noun-VPR and nominal-particle incorporation is the strong interpretive relatedness of noun and verb in the particle construction. We may claim that the VPR strategy (Belgium only) and the particle strategy overlap in the case of bare nouns. Given the fact that the Netherlands allows particle incorporation, we may expect (18) to show up, while the other cases of VPR do not appear in the Netherlands.

2.3.1.2 Interruption of the verbal cluster: adverbial (map 28b)(map in dynaSAND)

Map 28b presents the distribution of sentence (20), a declarative main clause in which the adverb surfaces between the verbs of the verb cluster kunnen opstaan, consisting of a modal verb and an infinitival main verb. The adverb vroeg 'early' modifies the verb and precedes the verbal cluster in Standard Dutch. The sentence was tested in almost every sampling point (244 places in total) by requesting the informants to translate the sentence into their dialect.

(20)     Eddy   moet    [kunnen           vroeg   opstaan].                                            

      Eddy   must    can.inf             early    on-stand.inf

      'Eddy has to be able to wake up early.'

Verb-cluster interruption by an adverb is found exclusively in Belgium; in almost every sampling point in West-Vlaanderen and in parts of Oost-Vlaanderen: in the northwest and in the area to the south of the Schelde river. The map furthermore illustrates that the construction frequently occurs in Vlaams-Brabant and only occasionally in Limburg. In the Waasland no instances are found and the sentence appears twice in the province of Antwerpen, one of which is found in Willebroek, a sampling point on the border with Vlaams-Brabant. The construction is attested four times in Frans-Vlaanderen. What we see as a general pattern is an increase in the cases of verb-cluster interruption from east to west.

The sand material also includes a sentence in which a 'high adverb' interrupts the verb cluster. It involves an adverb that modifies the whole sentence instead of the verb(phrase) only (102: Ik weet dat Jan moet jammergenoeg vertrekken, 'I know Jan has to leave, unfortunately'). The map of this construction is not included since only 14 instantiations were found, scattered over the whole language area. Moreover, most informants who accepted the sentence noted that it is a very unusual construction. This outcome correlates with, and provides further evidence for the distinction between low and high adverbs in terms of their scope. It also ties in with the observation that VPR requires the interruptor to be directly related to the main verb semantically (cf. (2)). The non-occurrence of verb clusters interrupted by higher adverbs is corroborated by the results of test sentence 107 in which the sentence negation niet 'not' breaks up the verb cluster: Ik weet dat Hans mag niet komen 'I know Hans is not allowed to come.' This order was not found in our data.

2.3.1.3 Interruption of the verbal cluster: plural noun (map 29a)(map in dynaSAND)

In addition to a sentence with a bare noun interrupting the final verb sequence, the sand material also included a sentence in which a single plural noun breaks up the verb cluster. This sentence, exemplified in (21), was only included in the interviews in those places where either interruption by a bare noun or interruption by an adverb occurred (see maps 28a and 28b); this means that the construction was tested in 91 sampling points, mainly in Belgium.

(21)     Ik         weet    dat       Jan       [wil      varkens            kopen].                       

            I           know   that      John    wants pigs                 buy

            'I know that John wants to buy pigs.'

Map 29a shows the geographical distribution of the sentence in (21), in which the plural noun varkens 'pigs' follows the modal and precedes the infinitival main verb. This construction is mainly found in West- and Oost-Vlaanderen, with the exception of the Waasland and the eastern part of West-Vlaanderen. In Vlaams-Brabant the construction occurs frequently, whereas it is rare in the province Antwerpen and Frans-Vlaanderen. In Limburg, the interruption of the verbal cluster by a bare plural is found three times in the east and only once in the north.

If we compare this map with map 28a, which presents the occurrence of interrupted verb sequences by a bare noun, we find that the distribution of the latter is less restricted. The apparent difference between the two constructions is the number marking on the direct object. This indicates that marking a noun for number limits its distribution with respect to its potential occurrence inside a verb cluster.

2.3.1.4 Interruption of the verbal cluster: indefinite object (map 29b) (map in dynaSAND)

The verb cluster in (22), consisting of an inflected modal and an infinitival main verb, is interrupted by the indefinite object een nieuwe schuur 'a new barn'. This sentence was only presented to informants who allowed interruption of the verb cluster by either a bare noun or an adverb (see maps 28a and 28b); the total number of locations was 89.

(22)     Ik         weet    dat       Jan       [moet   een       nieuwe             schuur bouwen].                    

            I           know   that      John    must    a          new                 barn     build

            'I know Jan has to build a new barn.'

As map 29b shows, the distribution of this construction is rather restricted. It is frequently found in West-Vlaanderen. In Oost-Vlaanderen the construction is mainly found in the area around and to the south of the Schelde river, and a single instantiation is found in the west of Oost-Vlaanderen on the border with West-Vlaanderen, in Aalter. An isolated instance of the construction is found in Limburg (Itteren) and the construction occurs sporadically in Vlaams-Brabant.

Comparing this configuration with the maps 28a, 28b and 29a – on which the distribution of verb-cluster interruption is presented for a bare noun, a low adverb and a plural noun interruptor respectively – we observe a striking difference between the number of cases in West-Vlaanderen and in the rest of Belgium. In West-Vlaanderen, there appears to exist no real difference between the number of cases in which the verb cluster is broken up by a single adverb or noun, or by a complex indefinite object. In Oost-Vlaanderen, Vlaams-Brabant and Limburg, however, the number of break-up cases by an indefinite object drastically decreases relative to whether the interruptor is an adverb or a noun.

In general, we can conclude that the fact that interruption of the verb sequence by an indefinite object occurs significantly less than interruption by a noun or an adverb seems to indicate that interruption of the verb cluster becomes harder when we are dealing with a more complex phrase such as een nieuwe schuur 'a new barn'.

2.3.1.5 Interruption of the verbal cluster: definite object (map 29c) (map in dynaSAND)

In addition to the question about the occurrence of verb-cluster interruption by an indefinite object, the interviews also included a question about the possibility of interruption by a direct object with a definite determiner. The sentence, given in (23), was part of the questionnaire in Frans-Vlaanderen, West-Vlaanderen and Oost-Vlaanderen with the exception of the Waasland, yielding a total of 51 locations.

(23)     Ik         zei       dat       Willy   [moest de        auto     verkopen].                  

            I           said      that      Willy   must    the       car       sell

            'I said that Willy should sell the car.'

The geographical distribution of the construction in (23) is presented on map 29c. As this map shows, it is mainly spread across West-Vlaanderen and it is occasionally found in Oost-Vlaanderen: we find the sentence in two sampling points in the east, and two occurrences in the north of Oost-Vlaanderen. The map furthermore shows a single instance in Frans-Vlaanderen. We see that the geographical distribution of interruption by a definite object differs from the distribution of interruption by an indefinite object. Definite-object interruption is less attested than interruption by an indefinite object (compare map 29b): 15 versus 27 locations. This result suggests that the definiteness of the noun phrase plays a role in the distribution of objects interrupting the verb sequence. This observation is corroborated by the fact that definite pronouns do not occur in the verbal cluster at all.

2.3.1.6 Interruption of the verbal cluster: prepositional phrase (map 30a) (map in dynaSAND)

Map 30a presents the geographical distribution of the interruption of a two-verb cluster by the prepositional phrase naar Jef  'to Jef'. The sentence in (24) was only provided in those sampling points where interruption of the verb cluster by either a bare noun or an adverb had been noticed; the construction was included in the interview in 85 locations. The informants had to indicate whether the sentence occurred in their dialect.

(24)     Ik         vind     dat       Marie [moet   naar     Jef       bellen].                       

            I           think    that      Mary   must    to         Jef       call

            'I think that Mary has to call Jef.'

In West-Vlaanderen the construction occurs rather frequently, with the exception of the eastern area. In Oost-Vlaanderen it is found to the south of the Schelde. In Vlaams-Brabant the construction mainly occurs in the southwest and the sentence is attested three times in Limburg. Interruption of the verb cluster by a prepositional phrase does not occur in Frans-Vlaanderen. These results parallel the data in the Reeks Nederlandse Dialectatlassen ('Series of Netherlands' Dialect Atlasses'), as summarised by Vanacker (1964), with respect to the test sentence Zijn vader heeft hem zes jaar lang laten naar school gaan 'His father let him go to school for six years', in which the verb cluster, consisting of two infinitives, is interrupted by the prepositional phrase naar school 'to school'.

2.3.1.7 Interruption of the verbal cluster - synthesis I (map 30b)

The maps discussed earlier in this chapter document the various types of verb-cluster interruption. Map 30b provides an overview of the variation in distribution of this phenomenon by presenting together the data of the six test sentences that were used to investigate the distribution of verb-cluster interruption of the VPR type by different non-verbal material.

The map indicates that the only kind of interruption that is attested in areas other than Belgium is interruption by a bare noun, and even then the occurrence of cluster interruption is quite restricted: we have only found thirteen isolated instantiations in the Netherlands. As has already been mentioned earlier (par. 2.3.1.1), this result suggests that interruption by a bare noun might be analysed as different from the other types of VPR interruption. It could be the case that instead of VPR, which seems to be the prevailing strategy in Belgium, we are dealing with particle incorporation in the Netherlands. Such a line of reasoning would explain the non-occurrence of the other types of verb-cluster interruption in the Netherlands. Further research should show whether this suggestion is on the right track.

VPR interruption of the verb cluster thus occurs almost exclusively in Belgium. The phenomenon shows a high density in West-Vlaanderen, in the Oost-Vlaanderen area below the Schelde and in the northwestern part of Oost-Vlaanderen. Interruption occurs most frequently and systematically in these three areas; that is to say, the type of material that breaks up the verb sequence does not seem to be of great influence since almost all types of interrupting material are allowed. This suggests that these areas exhibit a clear and productive VPR-strategy. In Vlaams-Brabant the phenomenon occurs fairly frequently as well; especially interruption by an adverb is well represented in this province. Hardly any cases are found in the province of Antwerpen, and interruption of the verb cluster is found incidentally in Limburg (the exception being Itteren). The Waasland, the northeastern part of Oost-Vlaanderen, clearly joins Zeeland or the province of Antwerpen in that it disallows verb-cluster interruption generally. The pattern we thus establish correlates with the generalisation widely found in the literature, i.e., that verb-cluster interruption is a typically southern Dutch phenomenon and that the amount of interruption decreases from west to east (cf. Koelmans 1965, Vanacker 1970).

Looking at the entire Dutch-speaking area, we find that interruption with a bare noun is the most frequent type of interruption; it is attested 62 times. However, if we focus on Belgium only, interruption by an adverb is found equally frequently: 54 instantiations versus 48 instantiations of interruption by a bare noun. The number of instantiations of verb-cluster interruption by a noun marked for number is comparable to the amount of interruption by a bare noun or an adverb (n=46). Interruption by an indefinite object and a prepositional phrase substantially decreases relative to the other types of interruption (cf. (3)). When we insert a definite object, the number is even lower (n=15). We may propose the following generalisation: the more 'complex' the element that interrupts the verbal cluster and the more it contains 'old information', the more restricted its geographical distribution.

2.3.2 Interruption by particles                               

2.3.2.1 Interruption of the verbal cluster: adverbial particle (map 31a) (map in dynaSAND)

The verb cluster in (25) consists of three verbs: two modals and the main verb gooien 'throw'. The adverbial particle weg 'away' may appear in different positions within this cluster, as is illustrated in (25). This sentence was offered to the informants three times, the only variation being the position of the particle.

(25)     Ik         vind     dat       je         veel      [(weg) zou      (weg)   moeten           

            I           think    that      you      much   (away) should (away) must                (weg)               gooien].

•             (away)             throw

            'I think that you should throw away a lot.'

Map 31a sketches the distribution of the position of the particle weg 'away' in a three-verb cluster for 230 sampling points. Since the particle always precedes the main verb, the map only presents the distribution of the particle in those dialects that have the order V1-V2-V3 – i.e., the order in which the main verb (V3) is the final verb in the cluster (cf. chapter 1). This explains the absence of symbols in Friesland, since Friesland generally has the order V3-V2-V1.

As map 31a shows, the total number of occurrences of weg at the beginning of the verb cluster does not differ much from the total number of occurrences of weg adjacent to the main verb, 152 and 182 respectively. This indicates that in general there does not seem to be a great preference for the presence or absence of splitting of particle and verb. If we look at the distribution of these sentences, we find that the variant with the particle adjacent to the main verb occurs in the whole of the Dutch-speaking area. However, this variant is predominantly chosen as the preferred one in Belgium. In the Netherlands, on the other hand, the reverse appears to hold. This finding correlates with a statement often found in the literature that particle splitting occurs more often in Holland than in Belgium, cf. Verhasselt (1961:156-157), De Schutter & Van Hauwermeiren (1983:191) and Den Besten & Edmondson (1983:199). This difference between the Netherlands and Belgium suggests that the two areas have different strategies regarding verb-cluster interruption by a particle. Belgium seems to focus on a VPR strategy, especially since particle and verb are closely related semantically. The VPR strategy is lacking in the Netherlands, as was shown above. For the interpretation of the situation in the Netherlands there are two alternatives (cf. paragraph 2.2.1.2 above). One option is to start from the perspective that particle and verb are independent lexical items. If so, it should be possible to incorporate the particle into V, creating a particle-verb complex. Apparently, the process of particle incorporation is optional, although the absence of incorporation appears to be the preferred option, especially in Noord-Brabant and Zeeland. The other option is that particle and verb are part of one complex lexical item. If so, it should be able to separate particle and verb in order to account for orders in which particle and verb are non-adjacent.

Let us call the three different particle positions P1, P2 and P3 for expository reasons, P1 referring to the construction in which the particle surfaces at the beginning of the verb cluster, P2 to the situation in which the particle occupies an intermediate position in the cluster, and P3 to the construction in which the particle is adjacent to the verb. The Dutch-speaking area appears to be roughly divided in three areas. The first area consists of West-Vlaanderen and the southeastern part of Oost-Vlaanderen. In this area P3 is always chosen, sometimes in combination with P1, but P2 is almost never an option. Note that this is precisely the area in which we frequently and systematically find verb-cluster interruption (see map 30b). The rest of Belgium more or less conforms to this strategy, with the rider that the more we move to the east, the more we note that P1 and P2 are available as well. Leaving aside Zeeuws-Vlaanderen, a second relevant area consists of Noord-Brabant and Zeeland. In this area, P1 is always chosen, regularly in combination with P3, and sometimes in combination with P2 and P3. A third area consists of the more northern dialects for which Drenthe and Overijssel are representative. Here it seems to be the case that every order is possible, although P2 is less present than P1 or P3.

2.3.2.2 Interruption of the verbal cluster: adpositional particle (map 31b) (map in dynaSAND)

The particle op 'up' of the separable compound verb op-eten 'eat up' can appear separated from the verb eten 'eat'. In the sentence in (26) we are dealing with a verb cluster consisting of two sentence-final infinitives. The particle can either precede this cluster, in which case it shows up separated from the verb it semantically belongs to, or it can be adjacent to the verb, the incorporation variant. Sentence (26) was only offered to the informants in the order willen op-eten (want up-eat): the variant in which particle and verb are adjacent. Since the Frisian dialects do not exhibit the verb order willen eten, symbols in this area are missing on the map. The sentence was offered to the informants and they were asked to translate it into their dialect. Map 31b shows the distribution of the particle op in the two-verb cluster in a total of 224 sampling points.

(26)     Jan       had      het       hele      brood   wel [(op)         willen (op)      eten].  

            John    had      the       whole  bread   aff (up)          want    (up)     eat

            ‘John would have liked to eat the whole loaf.'

As is very clear on the map, Belgian and Dutch Limburg uniformly choose the variant in which the particle is not separated from the verb. The only exception is Sint-Jozef-Olen in the province of Antwerpen where the variant without incorporation is chosen. In a number of places in the south and the northeast of West-Vlaanderen, a variant is attested in which the verb cluster is also interrupted by the noun phrase het hele brood (compare map 29c). With the exception of three isolated locations, these complex interruptions co-occurred with the variant in which the particle alone interrupts the verbal cluster.

In Zeeland and Noord-Brabant the variant with particle splitting is attested almost exclusively. Since the test sentence was only offered in the variant in which particle and verb are not separated, this would seem to indicate that there is a strong preference for the variant with particle splitting in these areas. In the northern dialects of the Netherlands there is no clear pattern, both the variant in which the particle is adjacent to the verb and the variant with separation of particle and verb are found, although the second variant has more instantiations than the first.

In general the results presented on this map are comparable to the results regarding the distribution of the position of the particle weg in a three-verb cluster as presented on map 31a. The language area can roughly be divided into three areas that seem to exploit different strategies. We might have expected that the area in which particle incorporation is optional, would occur in between the area in which incorporation is preferred and the area in which there is little or no incorporation. However, this is not what we find. The area in which incorporation is optional is the northern part of the language area, the area showing incorporation is Belgium and in between these areas we find the area in which there is not much incorporation (Noord-Brabant and Zeeland).

2.3.2.3 Interruption of the verbal cluster - synthesis II (map 32a)

As presented earlier, different kinds of direct objects, prepositional phrases, adverbs and particles can interrupt the verb cluster. Map 32a presents the number of different types of verb-cluster interruptions that are attested for almost every sampling point (total 225). It combines the data from the maps 28a-30a, 31a and 31b. This map quite clearly demonstrates that in the language area covering the Netherlands interruption is basically confined to particle incorporation, which accounts for the large number of locations in which one or two types of interruption are possible. Spread out over the Netherlands we find some instances of noun incorporation (cf. map 28a), which gives rise to the occasional location with three types of cluster interruption in the Netherlands.

2.3.3 Third Construction (map 32b) (map in dynaSAND)

Map 32b shows the distribution of test sentence 531, given in (27), in which we have an example of the Third Construction. In this sentence the indefinite direct object een cadeau 'a present' surfaces in a position between the participle geprobeerd 'tried' and the infinitive te geven 'give', whereas the indirect object Els is to the left of the verb cluster. The test sentence was included in the oral interview in almost every sampling point (n=253) and the informants were asked whether the sentence occurred in their dialect.

(27)     Wim    dacht               dat       ik         Els       [had     geprobeerd      een      

            Wim    thought            that      I           Els       had      tried.perf        a          cadeau             te         geven].

•             present            to         give

            'Wim thought that I had tried to give Els a present.'

The construction is attested 81 times, but there is no clear geographical distribution. The instantiations seem to be distributed almost randomly over the map. It appears to be the case that the Third Construction is less available in Belgium than in the Netherlands, which supports our assumption that we have to distinguish the Third Construction from the phenomenon of VPR (cf. 2.1.1.3), since VPR is generally restricted to Belgium.

2.3.4 Interruption by te 'to'             

2.3.4.1 Presence of te 'to' in the verbal complex (map 33a) (map in dynaSAND)

Sentence (28) contains a two-verb cluster consisting of the verb staan 'stand' in its use as an aspectual auxiliary with a durative interpretation, and the main verb zeuren 'whine'.

(28)     Hij       zal       wel      weer    [staan x          zeuren].

            he        will      aff       again    stand   to         whine 

            ‘He will be whining, again.’

Position x in (28) can optionally be occupied by te, a function word marking an infinitival verb. The reason to include this sentence in the sand material was to look for variation in the presence and position of te. The informants were requested to translate it into their dialect. Since the two variants – one with and one without te – were not part of the oral interviews in all locations, map 33a presents the results of the telephone interviews (245 locations). The variant with te is attested in 184 sampling points; it is found in almost every sampling point in the Netherlands. In two sampling points in Noord-Holland and one in Drenthe, the construction without te is found without a corresponding variant with te; in the rest of the Netherlands the te-less variant always co-occurs with the variant with te. In the western part of Belgium it appears to be the case that the variant without te is preferred, but the more one moves to the east, the more instances of the variant with te are found as well.

2.3.4.2 Position of te 'to' in the verbal complex (map 33b) (map in dynaSAND)

The element te can surface in different positions in a two-verb cluster consisting of infinitives of which the first is zitten 'sit', a verb that may function as an aspectual auxiliary with a durative interpretation. The sand interviews included three versions of sentence (29): a variant with te preceding the verbal complex, a version with te interspersed between the two verbs, and a variant in which te is spelled out twice. In the telephone interviews the informants were asked to judge whether the sentences were possible in their dialect. Map 33b shows the geographical distribution of the three sentences presented in 239 sampling points; data from Frans-Vlaanderen are missing. In the standard language the variant with a cluster-initial te only is the preferred one.

(29)     Jan       vindt    het       prettig                         om       de        hele      dag       [(te)

            John    finds    it          pleasant           for       the       whole  day      (to)                  zitten   (te)      werken].         

•             sit        (to)      work

            John likes to work the entire day.'

The most frequent variant of (29) is the one in which te precedes the verbal complex (159 instantiations). In West-Vlaanderen and the area to the north of the Schelde in Oost-Vlaanderen and the main part of the province of Antwerpen, this is more or less the only variant. These results coincide with those presented in Meeussen (1943) and Vanacker (1969). The distribution of the sentence in which te appears within the verb cluster is more restricted; we find 90 cases scattered over the language area. This construction is mainly found in the southern dialects: Limburg, Vlaams-Brabant and to the south of the Schelde in Oost-Vlaanderen. In these areas this construction is frequently found to be the only occurring variant of sentence (29). The construction in which te is inserted between the two infinitives frequently occurs in the northern dialects as well. In these dialects, however, this variant is almost always accompanied by the variant in which te precedes the verbal complex. There does not seem to be a clear pattern in the rest of the Netherlands: the form with te preceding the verbs occurs more often, but the variant with te within the verb cluster and the variant with two instances of te are found occasionally. Interestingly, in Ossendrecht, a sampling point in the southwest of Noord-Brabant, the doubling configuration is the only variant of (29) that is considered acceptable in the dialect.

2.3.5 Noun incorporation

2.3.5.1 Noun incorporation and verbal hendiadys            (map 34a) (map in dynaSAND)

Map 34a presents the distribution of noun incorporation, as illustrated in sentences (30a/b) in which the noun stoofperen / stoofpeer 'cooking pear(s)' gets incorporated into the verb it is semantically closely related to, schillen 'peel' in these sentences. So, instead of the noun surfacing in a position preceding the element te, as in Standard Dutch, the noun is between the infinitival marker te and the main verb. Given that no constituents are allowed to show up between these lexical elements – it can even be argued that te is a prefix, rather than a functional word – this appears to be a clear instance of incorporation of the noun into a verb. The only difference between sentence (30a) and (30b) is the number marking on the direct object; in (30a) we are dealing with a plural noun, whereas in (30b) we have a bare noun. Sentence (30c) exhibits verbal hendiadys, a phenomenon by which two verbs get co-ordinated by the coordination marker en 'and', but still express a single activity. In this particular sentence, the direct object stoofperen 'cooking pears', which belongs to the verb schillen 'peel', appears in the first conjunct. Neither noun incorporation nor verbal hendiadys occur in Standard Dutch; in the standard language the sentence would be Marie zit stoofperen te schillen (lit: ‘Mary sits cooking pears to peel’). The informants were asked whether the sentences in (30) occurred in their dialect. Sentence (30a) was offered in a total of 119 sampling points – in all the sampling points to the north of the River Delta in the Netherlands, in Vlaams-Brabant, Vlaams-Limburg and occasionally in Antwerpen and West-Vlaanderen. Sentence (30b) was presented to the informants in the telephone interviews in 85 sampling points to the north of the River Delta in the Netherlands. Sentence (30c), finally, was offered in all the sampling points in the central and northern part of the Netherlands, in Zeeland, and everywhere in Belgium except for Brabant, Limburg and the east of the province of Antwerpen (168 locations in total).

(30)     a.         Marie zit        te         stoofperen       schillen.                                  

                        Mary   sits      to         cooking pears  peel

            b.         Marie zit        te         stoofpeer         schillen.                      

                        Mary   sits      to         cooking pear    peel

            c.         Marie zit        stoofperen       en        schillen.                      

                        Mary   sits      cooking pears  and      peel

‘Mary is peeling cooking pears.'

As becomes clear immediately from the map, verbal hendiadys (30c) only occurs in the southern part of West-Vlaanderen and Frans-Vlaanderen; this result is not in complete agreement with the data on this construction as presented in Gerritsen (1991), in which the geographical distribution of verbal hendiadys was found not to be restricted to West-Vlaanderen and Frans-Vlaanderen. Noun incorporation, on the other hand, seems to be a typically northern Dutch phenomenon. The number of cases of noun incorporation with a plural noun is comparable to the amount of noun incorporation with a bare noun, 21 and 20 instantiations respectively. Noun incorporation with a bare noun is attested in almost every sampling point in Friesland, it occurs occasionally in Drenthe and Noord-Holland, and a single specimen is found in Groningen. The geographical distribution of noun incorporation with a plural noun is less restricted. This construction occurs only occasionally in Friesland, whereas it occurs frequently in Groningen, Drenthe and Noord-Holland, and is even found in the most northern sampling point of Utrecht. It seems to be the case that noun incorporation with a bare noun is a typically Frisian phenomenon, whereas incorporation with a plural noun is less clearly so. As mentioned earlier, Dijk (1997) notes that the incorporated noun in Frisian cannot be pluralised, i.e., it cannot have the suffix -en. However, in several locations in Friesland – among which Hindeloopen and Appelscha – we found occurrences of the incorporated noun with the suffix -en, indicating that, at least for some varieties of Frisian, the generalisation that incorporated nouns cannot pluralise, does not hold. It is interesting to see that in general only one of the two constructions – noun incorporation with a plural noun or with a bare noun – is found in a single sampling point. Only five times do we see both constructions showing up in one sampling point.

2.3.5.2 Incorporation of noun and adposition (map 34b) (map in dynaSAND)

The sentences in (31) consist of an auxiliary zitten-construction and show incorporation into the verb of an element that is semantically closely related to it. In the a- and b-sentences we have once more incorporation of a bare noun (compare map 34a), whereas sentence (31c) consists of a verb in which the adposition is adjacent to the verb, instead of being separated from it, as would be the case in the corresponding separable compound verb in Standard Dutch: Hij zit weer voor te zeggen. This type of adposition incorporation has to be distinguished from particle incorporation, as discussed in chapters 2.1.1.2 and 2.3.2. In this case the adposition is found in the exceptional position in between the infinitival marker te and the verb, clearly suggesting incorporation in the verb, rather than in the verbal cluster. Map 34b presents the data that were gathered in the telephone interviews. The sentences were presented to the informants in 85 sampling points, all located in the Netherlands to the north of the River Delta.

(31)     a.         Marie  zit        te         stoofpeer         schillen.                                  

                        Mary   sits      to         cooking pear    peel

                        ‘Mary is peeling cooking pears.'

            b.         Marie zit        te         piano   spelen.                                    

                        Mary   sits      to         piano   play

                        ‘Mary is playing the piano.'

c.         Hij       zit        weer    te         voorzeggen.                                        

                        he        sits      again    to         prompt

                        'He is prompting again.'

The geographical distribution of sentence (31a) has been discussed in the previous paragraph. We found 27 instantiations of sentence (31b): frequently in Friesland, Groningen and Drenthe, occasionally in the north of Noord-Holland, and sporadically in Overijssel and Utrecht. The map shows that te piano spelen is the most frequent variant of noun incorporation; it is geographically more widespread than sentence (31a). One might suggest that this is due to the very strong semantic relation between the noun piano and the verb spelen. That is to say, in Standard Dutch the sentence Hij is aan het piano spelen 'He is playing the piano' is acceptable and demonstrates noun incorporation in the progressive aan het INF-construction whereas Hij is aan het stoofpeer schillen 'He is peeling cooking pears' is less acceptable for most speakers. With respect to the distribution of sentence (31c), we observe that the number of cases is comparable to that of noun incorporation. We found 24 instantiations, mainly in the northern Dutch language area – in Friesland, Groningen and Drenthe – but occasionally in the north of Noord-Holland and in Overijssel.

2.3.6 Morphosyntax and the verbal cluster

2.3.6.1 Infinitivus pro Participio (IPP)

2.3.6.1.1 Infinitivus pro Participio in the verbal cluster (map 35a) (map in dynaSAND)

The sentence in (32) contains a clause-final two-verb cluster consisting of an infinitival modal and the infinitival main verb. This sentence exhibits the IPP effect; instead of the expected perfective participle gewild, the infinitive willen appears. In the oral interviews the informants were asked to translate sentence (32) into their local dialect.

(32)     Jan       had      het       hele      brood   wel      [willen op        eten].              

            John    had      the       whole bread   aff      want    up        eat

            ‘John would have liked to eat the entire loaf.'

Abstracting away from the interruption of the verb cluster by the particle op, as has been discussed in the text that accompanies map 31b – and focussing on the order and form of the two verbs – we observe a clear pattern. Except for some of the northern dialects, including the northern and eastern part of Drenthe, the east of Friesland and the northwest of Groningen, all dialects exhibit the IPP effect, irrespective of whether the order of the verbs is 1-2, as is the case in the southern dialects, or 2-1 as is the case in the northern dialects; the area that shows mixed behaviour, i.e., the area in which the 2-1 order and the 1-2 order occur side by side, also shows the IPP effect (compare map 4, that presents the variation in the word order of a two-verb cluster consisting of a finite modal and an infinitival main verb). Mainly in Groningen and Drenthe, we find instantiations of the sentence without the IPP-effect as the only variant.

Interestingly, the dialects that do not show the IPP effect all have the 2-1 order, except for a single specimen in Frans-Vlaanderen. This result seems to correspond to the generalisation often mentioned in the literature that there is a correlation between the IPP effect and the order of the verbs in a cluster (cf. Hoekstra 1994, De Schutter 1995). In the literature it has also been suggested that the absence of the IPP effect occurs in dialects only that lack a perfective prefix to mark the past participle (cf. Hoeksema 1980:240-243; Lange 1981; Vanden Wyngaerd 1994). For the correlation between the IPP effect, the absence/presence of ge- and word order, see map 37b.

2.3.6.1.2 Infinitivus pro Participio without verbal cluster (map 35b) (map in dynaSAND)

Sentence (33) contains the perfective auxiliary heeft 'has', and the verbs willen 'want' and kunnen 'can' with past participle morphology. In the oral interviews, this sentence was presented to the informants with the question how it should be translated into their dialects. The goal of questioning this sentence was to see whether varieties of Dutch would manifest the IPP effect in sentences without cluster formation of the verbs, as has been claimed in the literature to be non-existing (e.g., Nieuwenhuijsen 1974, Den Dikken 1989). Map 35b presents the geographical distribution of the IPP effect in (33) and shows the various forms of the infinitive encountered – of both willen 'want' and kunnen 'can'.

(33)     Niemand          heeft    dat       ooit      [gewild            of         gekund].         

            nobody            has       that      ever     wanted            or         could

            'Nobody has ever wanted or been able to do that.'

On map 35b we observe that the distribution of the IPP effect in sentence (33) is severely restricted: it only occurs in the northwestern part of the Netherlands, including Friesland except for the eastern area, the north of Noord-Holland and the Waddeneilanden. We found two isolated occurrences, one in the east of Overijssel and the other in West-Vlaanderen. In three locations the IPP effect seems to be manifested in only one of the two verbs of sentence (33). In Bergum in Friesland, the informant translated sentence (33) with Net ien heeft dat wollen of kind, with the infinitival form of willen and the participle form of kunnen. When explicitly asked whether the latter could also be an infinitival form, the informant provided a negative response. We appear to find the same in Hooglede in West-Vlaanderen, where only the verb willen shows the IPP effect. The reverse is true in Lutjebroek, Noord-Holland, where the verb kunnen, but not willen, has the infinitival form.

2.3.6.1.3 IPP in verbal cluster with the verb proberen 'try' (map 36a) (map in dynaSAND)

Map 36a presents the distribution of the IPP effect in sentences with the verb proberen 'try' on the basis of the following sentences.

(34)     a.         Marie  zei       dat       jij         Piet      een       boek    hebt    

                        Mary   said      that      you      Pete     a          book    have                                         geprobeerd      / proberen        te         verkopen

•                         tried                 / try                 to         sell

                        'Marie said that you have tried to sell a book to Pete.'

            b.         Wim    dacht               dat       ik         Els       had      geprobeerd/

                        Bill      thought            that      I           Els       had      tried/                                        proberen          een       cadeau             te         geven.

•                         try                   a          present            to         give

                        ‘Bill thought that I had tried to give Els a present.'

c.         Karel               weet    dat       jij         hebt     geprobeerd/     proberen

                        Charles            knows             that      you      have     tried/                try                               Marie een       boek    te         verkopen.

•                         Mary   a          book    to         sell

                        ‘Charles knows that you have tried to sell a book to Mary.'

Sentence (34a) contains a clause-final verb cluster consisting of the auxiliary hebt 'have', the perfective participle geprobeerd 'tried' and the infinitival main verb te verkopen 'buy'. Sentences (34b) and (34c) differ from the first sentence in that they have one (34b) or two (34c) objects intervening between the IPP-infinitive and the te+infinitive. Sentence (34b) is an example of the Third Construction, see also map 32b, and sentence (34c) is an example of Extraposition, a phenomenon by which the whole complement appears to the right of the matrix verb, in this example geprobeerd. In Standard Dutch the IPP effect is normally absent in Extraposition constructions and in the Third Construction (cf. Vanden Wyngaerd 1994:266). The sentences in (34) were offered to the informants with both the form geprobeerd without IPP, and the infinitival form proberen as a manifestation of the IPP effect; the informants were asked whether these sentences occur in their dialect.

It becomes immediately clear from the map that in the case we are dealing with a non-interrupted clause final verb cluster, the IPP effect with the verb proberen can be present quite easily, compared to the presence of IPP in combination with the Third Construction or Extraposition, 113 versus 22 or 13 instantiations. The IPP effect in sentence (34a) is frequently found in Belgium and Limburg, and occasionally in the Netherlands scattered over the language area, with the exception of the western part.

The IPP effect in combination with the Third Construction is found in the northeast of the Netherlands (Friesland, Drenthe, and Overijssel). In Belgium, IPP + Third Construction is attested in Oost-Vlaanderen, the west of Brabant and in Limburg. We observe that this construction is generally attested in locations where sentence (34a) is found with IPP as well; an exception is Drenthe, where we find three locations in which (34b), but not (34a), is found with IPP. The distribution of sentences with the IPP effect in an extraposition context (34c) is even more restricted; it is found in those sampling points in Friesland where the other two sentences (34a,b) also show an IPP effect; construction (34c) with IPP shows up in Belgium sporadically.

2.3.6.2 Forms of the verbs - infinitival and past participle morphology

2.3.6.2.1 Form of the participle of the modal verb willen 'want' (map 36b) (map in dynaSAND)

Map 36b shows the variation in form of the participle of the verb willen, based on the translations of the sentence Niemand heeft dat ooit gewild of gekund (literally: ‘Nobody had that ever wanted or could’). On this map, the infinitival and participial forms of the verb encountered are presented, i.e., sentences with and sentences without the IPP effect are taken together. We see the forms gewild, ewild, wild, wold, wuld, wul, gewillen, willen, wollen.

            The general pattern that emerges is that the northern dialects – the provinces Friesland, Groningen, Drenthe, the northern part of Noord-Holland and the eastern part of Overijssel – lack a perfective prefix on the participle; we have found forms such as wild, wold, wul and wuld. In the remaining part of Overijssel and the main part of Gelderland, we have found a form with a prefix in the form of a schwa: ewild. The rest of the Dutch-speaking language area tends to make use of the prefix ge- as a perfective marker. The pattern that we get is that the southern dialects generally have the prefix ge-; the more we move northwards, the more easily this prefix disappears.

2.3.6.2.2 Form of the participle of the modal verb kunnen 'can' (map 37a) (map in dynaSAND)

Map 37a is based on the various translations of the participle of the verb kunnen 'can' in the test sentence Niemand heeft dat ooit gewild of gekund (literally: ‘Nobody had that ever wanted or could’;, 'Nobody has ever wanted or been able to do that.'). On this map, the infinitival and participial forms of the verb encountered are presented, i.e., sentences with and sentences without the IPP effect. What we have encountered are the forms gekund, gekind, gekend, ekund, kund, kind, kend, gekost, gekunnen, gekannen, ekunnen, kunnen, kennen, kinnen, kannen.

We observe the same pattern as we have found for the verb willen 'want' (cf. map 36b). The northern dialects do not have a perfective marker in the form of a prefix, whereas the southern dialects generally have the prefix ge-. In the main part of Overijssel and Gelderland, the perfective prefix may have the form of a schwa, as in ekund and ekunnen.

2.3.6.2.3 Correlation between IPP, the presence of the prefix ge- and word order (map 37b) (map in dynaSAND)

Map 37b shows the correlation between the absence of the IPP-effect, the absence of the perfective prefix ge- on the past participle and the order V3-V2, derived from sentence (35). This sentence contains a clause final three-verb cluster consisting of the auxiliary had 'had', the modal kunnen 'can' (as a manifestation of the IPP effect) and the infinitival verb roepen 'call'. It thus involves verbal clusters such as had roepen kund. Sentence (35) was offered in every sampling point; the informants were asked whether the sentence occurred in their local dialects and if so, how it should be translated.

(35)     Vertel  mij      eens     wie      (of)      (dat)    zij        [had                 kunnen            roepen].

            tell       me       once     who     (if)      (that)   she       had.fin                        can.inf            roepen].

•             call.inf

            'Tell me who she could have called.'

The map shows a very clear picture: IPP is absent in dialects in which there is no perfective prefix on the participle (see maps 36b and 37a) and when the order of the verbs is V3-V2. This finding seems to provide evidence for the suggestion that the IPP is related to word order (cf. Hoekstra 1994; De Schutter 1995) and for the claim that the appearance of the IPP effect is related to the availability of the perfective prefix (cf. Hoeksema 1980; Lange 1981; Vanden Wyngaerd 1994). Note that given the data presented on the map, it cannot be argued that the order V3-V2 forces the IPP effect to be absent, since we find several instantiations of the V3-V2 order in West-Frisian with IPP (cf. Hoekstra 1993, 1994, 1998 and Hoekstra & Taanman 1996). Whether the prefix-generalisation on its own may provide an account for the presence/absence of the IPP effect cannot be decided on the basis of the data presented on this map.

2.3.6.3 PPI AND IPI

2.3.6.3.1 Participium pro Infinitivo (PPI)   (map 38a) (map in dynaSAND)

The Standard Dutch sentence in (36) contains a clause final three-verb cluster, consisting of an infinitival perfect auxiliary, an infinitival modal (IPP) and the infinitival main verb. This sentence exhibits the IPP effect: no past participle shows up, notwithstanding the presence of the perfective auxiliary hebben. As has become clear from maps 35a and 37b, some northern dialects lack the IPP effect. In some varieties, it even appears that two participles are possible: in addition to the presence of the participle of the modal (gekund 'could'), the main verb may also show up as a participial verb (i.e., participial morphology doubling). This is shown in the test sentences in (37). In sentence (37a), the three verbs show up in the order V3-V2-V1 and the modal verb has participle morphology. Sentence (37b) and (37c) are clear examples of the Participium pro Infinitivo construction. In sentence (37b), the verbs appear in the order V3-V2-V1 again, but now the modal verb kunnen and the main verb doen 'do' both have participle forms. In sentence (37c), the verb doen has a participle form again, but in this case the verbs show up in the order V3-V1-V2.

(36)                 Zou     hij        dat       hebben             kunnen            doen?                           1aux - 2modal - 3Vinf

            would  he       that      have.inf           can.ipp                        do.inf

                        ‘Could he have done that?’

(37)     a.         Zou     hij        dat       doen    gekund             hebben?                                                           V3-V2-V1      

                        would he        that      do.inf could.partpcp           have.inf                                  

                        'Could he have done that?'

b.         Zou     hij        dat       gedaan             gekund             hebben?           V                                             3-V2-V1         

                        would he        that      have.partpcp done.partpcp           could have.inf            

c.         Zou     hij        dat       gedaan             hebben             gekund?           V                                             3-V1-V2         

                        would he        that      done.partpcp            have.inf           could.partpcp                       

In the oral interviews, the sentences in (37) were presented to the informants with the question whether they occur in their dialects and how they should be translated. Test sentences (37b) and (37c) were offered to the informants in almost every sampling point, 228 and 256 locations respectively. Test sentence (37a), which lacks the IPP effect, was only presented in the northern provinces of the Netherlands, i.e., in Friesland, Groningen, Drenthe and Overijssel, in the eastern parts of Gelderland and Noord-Brabant, and in every sampling point in Limburg (96 locations in total). Map 38a shows the geographical distribution of the PPI effect, and the variation attested in order and form of the different verbs in the clause-final verb cluster.  

            Sentence (37a) has 16 instantiations and is found in the northern dialects only: Friesland and occasionally Groningen and Drenthe – see also maps 35a and 37b. The distribution of the PPI construction as given in sentence (37c) is comparable to that of sentence (37a), except for the fact that its occurrence is found more frequently in Groningen and Drenthe, and there are some isolated occurrences of the construction in Overijssel, Gelderland and even in Limburg. Sentence (37b) is found more often than (37c), 34 versus 21 instantiations. This type of PPI construction seems to be a typical northeastern phenomenon: it occurs frequently in Friesland, Groningen and Drenthe and it is occasionally found in Overijssel and Gelderland.

In sum, the map shows a clear pattern with respect to the distribution of the PPI effect. The PPI effect is a typically northeastern phenomenon: occurrences of sentences with PPI in which the word order is 3-1-2 – sentence (37c) – are found in the northeastern language area, consisting of Friesland, Drenthe and Groningen; when the word order is 3-2-1, the distribution of the sentence with PPI is less restricted: we also find occurrences in Overijssel and Gelderland. It is interesting to observe that the PPI variant (37b) is found more often than the corresponding variant without PPI (37a).

2.3.6.3.2 Imperativus pro Infinitivo (IPI) (map 38b) (map in dynaSAND)          

Map (38b) presents the distribution of the Imperativus pro Infinitivo effect, based on the following four test sentences which the informants were asked to translate.

(38)     a.         Ik         heb      geen     zin       en        voeren                         de        koeien.                                    

                        I           have     no        sense   and      feed.inf           the      cows

                        'I don't feel like feeding the cows.'

            b.         Hij       wou                 nog      snel      even     bij        de        bakker naar binnen en koop een broodje.   

he        wanted            PRTprt         quick   just      at         the       bakery naar     binnen             en        koop                een       broodje.           

• to         into      and      buy.2sing      a          sandwich        

'He just wanted to go get to the bakery quickly to buy a sandwich.'

(39)     a.         Hoe     haal      je         het       in         je         hoofd   en        gooi met eten!                        

                        how     get      you      it          in         your    head     and     

•                         gooi                 met      eten!                           
•                         throw.2sing   with     food

            b.         Hoe     haal      je         het       in         je         hoofd   en        gooien

                        how     get       you      it          in         your    head     and      throw.inf                   met      eten!

•                         with     food

'How dare you fling your food about!'

Test sentence (38a) was questioned in almost every sampling point (256 in total); sentence (38b) was offered to the informants in 119 locations: in Belgium, with the exception of West-Vlaanderen, Frans-Vlaanderen and the west of Antwerpen, and in the northwestern part of the Netherlands, i.e., in Friesland and the north of Noord-Holland; in the rest of the Netherlands, sentence (38b) was offered to the informants sporadically. With respect to these two sentences, map 38b is based on the data from the oral interviews. Sentence (39a) and (39b) were only provided in the northeastern part of the language area, to the north of the River Delta in the Netherlands (i.e., Friesland, Groningen, Drenthe, Overijssel and Gelderland), yielding a total of 72 locations. Map 38b presents the data from the telephone interviews for these two sentences.

The map shows that the IPI effect is a typically Frisian phenomenon; it is found in almost every sampling point in Friesland. Furthermore, it shows up occasionally in Drenthe and sporadically in Groningen, Overijssel and Gelderland. If we focus on the nature of the IPI construction, we observe that the distribution of the variant with an infinitive ­– test sentences (38a) and (39b) – is more restricted than the variant with an imperative, i.e., test sentences (38b) and (39a). This correlates with the assumption that the imperative in initial position in the second conjunct developed from an infinitive in this position (Hoekstra 1987:101). The variant with an infinitival verb in the second conjunct is mainly found in Friesland (38a) and Drenthe (39b). It might be the case that the sentences with an infinitival form represent an earlier stage in the development of this construction.  

2.4 Literature on verb-cluster interruption and verbal morphosyntax

ANS(1997). W. Haeseryn et al. (eds.) Algemene Nederlandse Spraakkunst.

Groningen: Wolters-Noordhoff. 2nd edition.

Bennis, H. (1991). 'Theoretische aspekten van partikelvooropplaatsing II.' Tabu 21:3.

89-96.

— Bennis, H. (1992). 'Long Head Movement: The Position of Particles in the Verbal

Cluster in Dutch.' In: R. Bok-Bennema & R. van Hout (eds.) Linguistics in the Netherlands 1992. Amsterdam: Benjamins. 37-47.

— Bennis, H. (1993). 'Morfologie bestaat niet? Over de verhouding tussen zinsbouw en

woordvorming.' Tabu 23:1. 15-27.

Bennis, H. & T. Hoekstra (1989). 'Why Kaatje was not 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. (1979). 'Heranalyse van een Stellingwerver oppervlaktestructuur.' Us

Wurk 28: Scripta Frisica. Tinkbondel foar Anne Spenter (special issue: H. Århammer and T. Hoeksema (eds.)). 31-38.

Blom, C. (2005). Complex Predicates in Dutch; Synchrony and Diachrony. PhD-

dissertation. Free University of Amsterdam.

Bobine, H. (1970). 'Woordschikking.' Nu nog XVIII. 76-77.

Booij, G. (2002a). The Morphology of Dutch. Oxford: Oxford University Press.

— Booij, G. (2002b). 'Constructural idioms, morphology, and the Dutch lexicon.'

Journal of Germanic Linguistics 14. 301-329.

Broekhuis, H. (1993). 'Verb Projection Raising.' Spektator 22:1. 28-47.

— Broekhuis, H. (1997). 'Nogmaals Verb Projection Raising.' Tabu 27. 1-27.

Broekhuis, H. & K. Hoekstra (1990). 'Verb Raising, Extraposition en Controle.'

Tabu 20. 153-174.

Broekhuis, H., H. den Besten, K. Hoekstra & J. Rutten. (1995). 'Infinitival

Complementation in Dutch: On Remnant Extraposition.' Linguistic Review 12. 93-122.

Colleman, T. (to appear). 'Ik zie hem morgen (te) komen; een 'echte' accusativus-

cum-infinitivo met zien.' Taal en & Tongval.

Cubber, W. de. (1968). 'De splitsing van scheidbaar samengestelde werkwoorden in

hedendaags proza.' Dissertation, University of Gent. Unpublished.

De Bruycker, G. (1972). 'Is dat "algemeen" Zuidnederlands?' Persoon en

Gemeenschap XXIV. 249-254.

De Haan, G. (1990). 'De Friese Imperativus-pro-Infinitivo.' Taal en & Tongval

themanummer 3. 87-107.

Den Besten, H. & J. Edmondson (1983). 'The Verbal Complex in Continental West-

Germanic.' In: W. Abraham (ed.) On the Formal Syntax of the West Germania. Amsterdam: Benjamins, 47-138.

Den Besten, H., J. Rutten, T. Veenstra & J. Veld (1988). 'Verb Raising,

Extrapositie en de Derde Constructie.' University of Amsterdam. Unpublished.

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. & H. Broekhuis (1992). 'Woordvolgorde in de werkwoordelijke

eindreeks.' GLOT 12. 79-137.

Den Dikken, M. (1989). 'Verb Projection Raising en de analyse van het IPP-effect.'

Tabu 19. 59-75.

Den Dikken, M. (1994). 'Minimalist Verb (Projection) Raising.' In: J.-W. Zwart (ed.),

Groninger Arbeiten zur Germanistischen Linguistik 37: Minimalism and Kayne's Asymmetry Hypothesis. Groningen: Rijksuniversiteit. 71-88.

Den Dikken, M. & E. Hoekstra (1997). 'Parasitic participles.' Linguistics 35. 1057-

1089.

De Schutter, G, (1995). 'Werkwoordvolgorde en het IPP-effect in het Nederlandse

taalgebied.' Taal en Tongval. 31-39.

— De Schutter, G. (2000). 'Systeem en ontlening in taal: nog eens het IPP-effect.' Taal &

Tongval 52:1. 208-229.

De Schutter, G. & P. van Hauwermeiren (1983). De structuur van het Nederlands.

Taalbeschouwelijke grammatica. Malle: De Sikkel.

De Waart, A.A.J. (1971). 'Constructies met en + 'imperatiefzin' in het moderne

Westerlauwerse Fries.' In: Stûdzjekonferinsje Frysk. Ljouwert. 3-31.

Dijk, S. (1997). Noun Incorporation in Frisian. Leeuwarden: Fryske Akademy.

PhD-dissertation, University of Groningen.

Duinhoven, A.M. (1975). 'Nobel die coninc hadde ghedaen sijn hof crayeren ouer al;

De ontwikkeling van participium + infi­nitief tot infinitief + infinitief.' Spel van zinnen. Album A. van Loey. Brussel. 89-99.

Evers, A. (1975). The transformational Cycle in Dutch and German. PhD-

dissertation, University of Utrecht.

Fischer, O. (1994). 'The fortunes of the Latin-type Accusative and Infinitive

Construction in Dutch and English compared.' In: T. Swan (ed.) Language Change and Language Structure. Older Germanic languages in a comparative perspective. Berlin: Mouton. 91-133.

— Fischer, O. (1997). 'The grammaticalization of infinitival to in English compared with

German and Dutch.' In: R. Hickey & S. Puppel (eds.) Language history and linguistic modelling. Berlin: Mouton. 265-280.

Gerritsen, M. (1987). Syntactische verandering in kontrolezinnen; een socio-

linguistische studie van het Brugs van de 13e tot de 17e eeuw. PhD-dissertation, University of Leiden.

— Gerritsen, M. (1991). Atlas van de Nederlandse Dialectsyntaxis (AND), part I (tekst)

& II (maps). Amsterdam: P.J. Meertens-Instituut.

Groos, A. (1989). 'Particle-verbs and adjunction.' In H. Bennis & A. van Kemenade

(eds.), Linguistics in the Netherlands 1989. Dordrecht: Foris. 51-60.

Haegeman, L. (1992). Theory and description in generative syntax: A case study in

West Flemish. Cambridge: Cambridge University Press.

— Haegeman, L. (1998). 'Verb movement in embedded clauses in West Flemish.'

Linguistic Inquiry 29. 631-656.

Haegeman, L. & H. van Riemsdijk (1986). 'Verb Projection Raising, Scope and the

Typology of Rules Affecting Verbs.' Linguistic Inquiry 17:3. 417-466.

Haeseryn, W. (1990). Syntactische normen in het Nederlands, Een empirisch

onderzoek naar volgordevariatie in de werkwoordelijke eindgroep. PhD-dissertation, Catholic University Nijmegen.

Haiden, M. (2006). 'Verb Particle Constructions.' In: M. Everaert & H. van Riemsdijk

            (eds.) The Blackwell Companion to Syntax. Volume V. Blackwell. 344-375.

Hartevelt, A. & E. Hoekstra (1999). 'De Derde Constructie.' Tabu 29:4. 179-184.

Haslinger, I. & M. van Koppen (2003). 'De verbale hendiadys als pseudocoördinatie.'

Taal en & Tongval 15-16. 102-122.

Hoekema, T.B. (1971). 'Sentences with Imperativus pro Infinitivo.' Us Wurk 20.

61-73.

— Hoekema, T.B. (1975). 'Sentences with Imperativus pro Infinitivo II.' Us Wurk 24.

35-48.

Hoeksema, J. (1980). 'Verbale verstrengeling ontstrengeld.' Spektator 10. 221-249.

— Hoeksema, J. (1990). 'Partiële extrapositie van infinitiefkomplementen in het

Middelnederlands.' Tabu 20-3. 175-186.

— Hoeksema, J. (1991). 'Theoretische aspecten van partikelvooropplaatsing'. Tabu 21.

18-26.

— Hoeksema, J. (1994). 'The history of Dutch verb projection raising.' Unpublished.

Hoekstra, E. (1993). 'Over de implicaties van enkele morfosyntactische

eigenaardigheden in Westfriese dialecten.' Taal en & Tongval 45. 135-154.

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

Zaans.' Taal en & Tongval 46. 132-141.

— Hoekstra, E. (1998). 'Analysing Linear Assymetries 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.

— Hoekstra, E. (1999). 'De opkomst van het aspectuele werkwoord liggen.' Tabu 29:1.

43-47.

Hoekstra, E. & E. Blom (1996). 'IPP en werkwoordsvolgorde in het Achterhoeks.'

Taal en & Tongval 48. 72-83.

Hoekstra, E. & W. Taanman (1996). 'Een West-Friese gradatie van het Infinitivus-

pro-Participio effect.' Nederlandse Taalkunde 1. 13-25.

Hoekstra, J. (1987). ''Verb second' en de 'imperativus pro infinitivo' in het Fries.' Tabu

17. 96-121.

— Hoekstra, J. (1997). The syntax of infinitives in Frisian. Leeuwarden: Fryske

Akademy. PhD-          dissertation, University of Groningen.

Koelmans, L. (1965). 'Iets over de woordorde bij samengestelde predikaten in het

Nederlands.' De Nieuwe Taalgids 58. 156-165.

Koopman, H. (2000). 'Prepositions, postpositions, circumpositions and particles.' In:

H. Koopman (ed.) The Syntax of Specifiers and Heads. London: Routledge. 204-260.

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

Koster, J. (1975). 'Dutch as a SOV language.' In: A. Kraak (ed.) Linguistics in the

Netherlands 1972-1973. Assen: Van Gorcum. 165-177.

Lange, K.-P. (1981). 'Warum Ersatzinfinitiv?' Groninger Arbeiten zur

Germanistischen Linguistik 19. 62-81.

Meeussen, A. E. (1943). 'Vier Isotagmen.' Leuvense Bijdragen XXXV. 47-55.

Mestdagh, M. (1970). 'Woordschikking.' Nu nog XVIII. 114-115.

Minnaert, N. (1998). Een onderzoek naar de regionale spreiding van de volgorde en

doorbreking van de werkwoordelijke eindgroep in de Zuid-Nederlandse dialecten. Dissertation, University of Gent. Unpublished.

Mithun, M. (1984). 'The evolution of noun-incorporation.' Language 60. 847-894.

Model, J. (1991). 'Incorporatie in het Nederlands.' Gramma 15:1. 57-88.

Neeleman, A. (1994). Complex Predicates. PhD-dissertation, Utrecht University

(OTS).

Neeleman, A. & F. Weerman (1993). 'The balance between syntax and morphology:

Dutch particles and resultatives.' Natural Language and Linguistic Theory 11. 433-475.

Nieuwenhuijsen, P. (1974). 'Infinitief in plaats van deelwoord.' Spektator 3:6. 477-

479.

Paardekooper, P. (1993). 'Een waterlantsche Trijn zat eens ajuin en schelde.' Leuvense

Bijdragen 82. 145-171.

Pannekeet, J. (1995). Het Westfries: inventarisatie van dialectkenmerken.

Wormerveer: Noord-Holland.

Pardoen, J. (1986). 'Werkwoordclustering in de voltooide tijd.' Voortgang, Jaarboek

van de neerlandistiek VII. 49-76.

Ponten, J.P. (1971). 'De 'Infinitief pro Participio.' Een verschijnsel uit de Duitse

syntaxis, belicht vanuit het Nederlands.' Handelingen van de Koninklijke

Zuidnederlandse Maatschappij voor Taal- en Letterkunde en Geschiedenis 25. 229-243.

Postma, G.-J. (2005). 'Infinitivale V2 in het Middeldrents en de subjunctief.' Taal en &

Tongval 57:2. 126-166.

Sadock, J.M. (1985). 'Autolexical syntax: a proposal for the treatment of noun-

incorporation and similar phenomena.' Natural Language and Linguistic Theory 3. 379-439.

— Sadock, J.M. (1986). 'Some notes on noun-incorporation.' Language 62. 19-31.

Schuurman, I. (1987). 'Incorporation in the Groningen dialect.' In: F. Beukema & P.

Coopmans (eds.) Linguistics in the Netherlands 1987. Dordrecht: Foris. 185-194.

Schuurman, I. & A. Wierenga (1986). 'Syntactische nomen-incorporatie bij

infinitieven en deelwoorden.' In: C. Hoppenbrouwers, J. Houtman, I. Schuurman & F. Zwarts (eds.) Proeven van Taalwetenschap (TABU-special). Groningen. 339-350.

Stoops, Y. (1969). 'Algemeen Zuidnederlands of niet?' Nu nog XVII. 105-107.

Strengholt, L. (1970). 'Waterlandsche Trijn heeft nog een appeltje te schillen.'

Nieuwe Taalgids 63:2. 127-131.

Taeldeman, J. (1986). 'Nog een beetje te: "Ik zie hem vandaag nog (te) komen".' Taal

en Tongval 38:1/2. 79-82.

Ter Laan, K. (1953). Proeve van een Groninger Spraakkunst. Winschoten: Van der

Veen.

Vanacker, V.F. (1964). 'Is het algemeen Zuidnederlands?' Taal en & Tongval 16. 159-

164.

— (1969). 'Zuidnederlands dialektmateriaal op de band en enkele

dubbele werkwoordgroepen in te-positie.' Taal en & Tongval 21:3/4. 239-244.

— (1970). 'Een "Zuidnederlandse" konstruktie in een paar

Zuidnederlandse dialekten.' De Nieuwe Taalgids; Van Haeringennummer. 140-147.

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.

Vanden Wyngaerd, G. (1994). 'Partikelwerkwoorden in het Nederlands.' Tabu 24.

141-148.

Vanden Wyngaerd, G. (1994). 'IPP and the Structure of Participles.' In: J.-W. Zwart

(ed.) Groninger Arbeiten zur Germanistischen Linguistik 37:Minimalism and Kayne's Asymmetry Hypothesis. Groningen: Rijksuniversiteit. 265-275.

Van der Auwera, J. & B. Bultinck (2000). 'Over de infinitief, de lange en de korte.' In:

S. Gillis, J. Nuyts & J. Taeldeman (red.) Met taal om de tuin geleid; opstellen voor Georges de Schutter. Antwerpen. 379-385.

Van der Horst, J.M. (1997). 'Over en naar aanleiding van Zuid-Nederlandse

doorbrekingen.' In: A. van Santen & M. van der Wal (red.) Taal in tijd en ruimte. Leiden: Stichting Neerlandistiek. 299-307.

—  (2002/2003). 'Geschiedenis van een taalnorm: de doorbroken

werkwoordelijke eindgroep.' Delen I en II. Verslagen en mededelingen van de KANTL 112: 529-556 and 113: 141-161.

Van der Meer, G. (1975). 'The Imperativus pro Imperativo reconsidered.' Us Wurk

24. 19-34.

Van der Meer, G. (1990). 'On the position of Old Frisian verbs and pronouns.' In:

R.H. Bremmer Jr., G. van der Meer & O. Vries (eds.) Amsterdammer Beiträge zur Älteren Germanistik, Band 31-32: 'Aspects of Old Frisian Philology.' Amsterdam & Atlanta: Rodopi / Groningen: SSFYRUG. 311-335.

— (1990). 'De Infinitivus pro Participio: Een nieuwe verklaring.' Tabu

20:1. 29-48.

Van Loey, A. (1965). Schönfelds Historische Grammatica van het Nederlands.

Zutphen: Thieme.

Van Riemsdijk, H. (1978). A Case Study in Syntactic Markedness. Dordrecht: Foris.

Verhasselt, J. (1961). 'Verschillen tussen Noord en Zuid inzake de volgorde

hulpwerkwoord-hoofdwerkwoord.' Taal en & Tongval 13. 153-157.

Weijnen, A. (1966). Nederlandse dialectkunde. Assen: Van Gorcum.

Wolf, H. (1996). 'IPP en morfologische markering.' Tabu 26:1. 33-40.

Wolf, H. (1997). 'Default-morphological forms in Interference Frisian.' Groninger

Arbeiten zur Germanistischen Linguistik 40. 249-256.

Wurmbrand, S. (2004). 'West Germanic Clusters: The Empirical Domain.' In: K.E.

Kiss & H. van Riemsdijk (eds.) Verb Clusters: A study of Hungarian, German and Dutch. Linguistics Today 69. John Benjamins. 43-85.

—  (2006). 'Verb Clusters, Verb Raising, and Reconstruction.' In: M.

Everaert & H. van Riemsdijk (eds.) The Blackwell Companion to Syntax. Volume V. Blackwell. 229-343.

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.