DynaSAND

Syntactic Atlas of Dutch Dialects – Commentary – Volume I

Table of contents Volume I

5          Fronting

5.1        Introduction

5.1.1     Fronting

5.1.2     Relative clauses

5.1.3     Partial fronting of question words

5.1.4     Other instances of fronting

5.1.4.1  Split fronting

5.1.4.2  Preposition stranding

5.1.4.3  Fronting in subordinate clauses with inversion

5.1.4.4  Fronting in main clauses without inversion

5.1.4.5  Fronting in imperative clauses

5.1.5     Historical development

5.2        Discussion of the literature

5.3        Discussion of the maps

            5.3.1     Relative clauses

            5.3.2     Partial fronting of question words

5.3.3     Other instances of fronting

5.4        Literature on fronting

5  Fronting

5.1  Introduction

5.1.1  Fronting

Fronting is the phenomenon in which a constituent appears in the first position of the sentence instead of in its usual position. In Standard Dutch, a fronted constituent occupies a position immediately in front of the finite verb. An example is given in (1b).

(1)     a.       Hij    heeft de      krant          niet   gelezen.

                   he      has    the    newspaper not    read

                   'He did not read the newspaper.'   

b.      De     krant          heeft hij     niet   gelezen.

         the    newspaper has    he      not    read

         'The newspaper, he did not read.'

In subordinate clauses, a fronted constituent takes a position as the first constituent in the subordinate clause, directly in front of the complementiser, if present. This is shown in (2).

(2)     (Ik vroeg)   welke                   krant          (of)             hij     had    gelezen.

         I asked       which         newspaper (whether)   he      had    read

Both in main clauses and in subordinate clauses, only one constituent can be fronted. In main clauses, this yields inversion of the subject and the finite verb. In most analyses of Standard Dutch, the position of the finite verb in main clauses is taken to be structurally similar to the position of the complementiser in embedded clauses. Sentence pairs such as in (3) provide evidence for this view.

(3)     a.       Heb   je       de      kans,                    kom  dan    langs.

                   have  you   the    opportunity,       come then  along

                   'If you have the opportunity, drop by.'

b.      Als    je       de      kans           hebt, kom  dan    langs.

         if       you   the    opportunity have,        come then  along

                   'If you have the opportunity, drop by.'

In embedded clauses introduced by a complementiser, the finite verb normally is in clause-final position. Embedded clauses differ from main clauses in that they do not allow inversion of the subject and the finite verb.

         Another important difference between main clauses and subordinate clauses concerns the nature of the constituents that can be fronted. In subordinate clauses, we only find fronting of constituents with relative or question pronouns. This is demonstrated in (4).

 (4)    a.       (Ik weet)    welke                   wedstrijd    hij     gaat   zien.

                   I know       which         match         he      will   watch

b.   * (Ik weet)    die     wedstrijd    hij     gaat   zien.

         I know       that   match         he      will   watch

c.       (de wedstrijd)      die              hij     gaat   zien

         the match            which         he      will   watch

There is no such restriction for root clauses. Almost all constituents can be fronted in main clauses, as is shown in (5).

 (5)    a.       Die    wedstrijd    gaat   hij     zien.

                   that   match         will   he      watch

         b.      In      de      Arena         gaat   hij     die     wedstrijd    zien.

                   in      the    Arena         will   he      that   match         watch

c.       Meestal      gaat   hij     twee  wedstrijden         zien.

         usually       will   he      two   matches      watch

The negative adverb niet and most verbal particles are well–known counterexamples to the generalisation that all constituents can be fronted in root clauses.

(6)     a.     *Niet   heeft Jan    gewerkt.

                   not    has    John  worked

b.    *Uit     mag   je       hem   niet   lachen.

         out    may  you   him   not    laugh

         'You should not jeer at him.'

         Normally, fronting is optional. However, fronting of constituents with a question word or a relative pronoun is generally obligatory in Standard Dutch. This is illustrated in (7).

(7)     a.       Wie   heb    je       gezien?

                   who  have  you   seen

b.    *Je      hebt  wie    gezien?

         you   have who seen

c.       (de wedstrijd)      die              je       gezien         hebt

         the match            which         you   seen            have

d.    *          (de wedstrijd)      je       die             gezien         hebt

         the match            you   which         seen            have

e.       Jan    heb    ik      gezien.

         John  have  I        seen

f.       Ik      heb    Jan    gezien.

         I        have  John  seen

Sentences of the type in (7b) are grammatical, but only if they concern a so–called echo question in a special context, for example if the speaker did not hear the preceding sentence well or if he expresses his surprise.

         Due to the obligatory nature of question-word fronting, sentences that have a question-word both in the main and in the subordinate clause force both question words to be fronted in their clause. An example is given in (8b).

(8)     a.       Je      hebt Jan    verteld        dat    je       dit     gaat   doen.

                   You   have John told            that   you   this   will   do

b.      Wie   heb    je       verteld        wat    je       gaat   doen?

                   who have you   told            what you   will   do

Fronting within an embedded clause is not always acceptable. To a large extent, the acceptability of embedded fronting is determined by properties of the verb in the main clause, as is shown in (9).

(9)     a.     *(Ze denkt) wat    ze      gaat   doen.

                   she thinks what           she    will   do

b.      (Ze bedenkt)       wat    ze      gaat   doen.

         she considers       what she    will   do

If embedded fronting is impossible, the question word in the embedded clause may sometimes be fronted to the first position in the main clause, thereby creating an instance of 'long question-word fronting.'

(10)   Wat   denkt          ze      dat    ze      gaat   doen?         (cf.(9a))

         what           thinks         she    that   she    will   do

         'What does she think that she will do?'

The obligatory nature of question word fronting also becomes apparent in (11). When the main clause contains a fronted question word (wie 'who') and the main verb blocks fronting of another question word in the embedded clause (hoe 'how'), there is no way to form a grammatical sentence:

(11)   a.      *Wie  dacht          hoe    Jan    zich   gedroeg?

                   who  thought      how  John  refl  behaved

b.      *Wie  dacht          dat    Jan    zich   hoe    gedroeg?

         who  thought      that   John refl  how behaved

Multiple questions constitute the only exception to the obligatoriness of question word fronting. In a sentence such as (12), the object question word wat 'what' is not fronted and occupies the normal object position. The contrast between (11b) and (12) shows that the absence of fronting is legitimate only if both question words belong to the same subclause.

(12)   Wie   heeft wat    gedaan?

         who  has    what done

         'Who did what?'

If a constituent is fronted in the subordinate clause, e.g., waar 'where' in (13d),  long fronting of another question word (wat 'what' in 13d) is impossible.

(13)   a.       Piet   vertelde      dat    Jan    het    boek  op     de      tafel  had    gelegd.

                   Piet   told            that   John  the    book on     the    table  had    put

                   'Piet told us that John had put the book on the table

         b.      Piet   vertelde      waar          Jan    het    boek  had    neergelegd.

                   Piet   told            where         John the    book had    put

                   'Piet told us where John had put the book.'

         c.       Wat   vertelde      Piet   dat    Jan    op     de      tafel  had    neergelegd?

                   what           told            Piet   that   John on     the    table had    put

                   'What did Piet tell you that John had put on the table?'

         d.    *          Wat   vertelde      Piet   waar          Jan    had    neergelegd?

                   what           told            Piet   where         John had    put

The ungrammaticality of sentences like (13d) is found in many languages with obligatory question-word fronting and is known as a violation of the principle of Subjacency. In the generative literature, the ungrammaticality of (13d) is taken to demonstrate that the process of long fronting should be broken up in two instances of short fronting. First a constituent is preposed in the embedded clause and then in the main clause. If the first position of the embedded clause is already occupied by another constituent, fronting is impossible. Partial Wh-movement (cf. section 5.1.3/5.3.2) is often taken to support this analysis.

         Summing up, fronting in Standard Dutch is governed by the following rules:

(i)        only constituents can be fronted;

(ii)       in each subclause, only one constituent can be fronted;

(iii)      a fronted constituent immediately precedes the finite verb in main clauses and precedes the complementiser, if present, in subordinate clauses;

(iv)      fronting is obligatory for constituents with a question word or a relative pronoun, and optional in other cases;

(v)       long fronting is the result of a combination of two or more instances of short fronting.

5.1.2  Relative clauses

In relative clauses, a relative pronoun is fronted that is coreferential with an antecedent. In (14a), the relative pronoun die takes the noun phrase die film 'that movie' as its antecedent; the phrase die ik niet heb gezien is the relative clause.

(14)   a.       die     film            die              ik      niet   heb    gezien

                   that   movie         which         I        not    have seen

         b.      alles            wat    je       hebt  gedaan

                   everything  what you   have  done

The rules for selecting the correct relative pronoun in Standard Dutch are quite complex. We will discuss the rules insofar as they are relevant for the variation we have encountered in Dutch dialects. First, there must be agreement between pronoun and antecedent with respect to the features gender and number. This is shown in (15).

(15)   a.       het boek              dat /*die                                 ik      heb    gelezen

                   the book_[+neuter]   that_[+neuter] /         which_[–neuter]        I        have  read  

         b.      de film                 die/*dat                                  ik      heb    gezien

                   the movie_[–neuter] which_[–neuter]/ that_[+neuter]        I        have seen

         c.       de boeken            die /*dat                                 ik      heb    geschreven

                   the books_[+plur]    which_[+plur] / that_[–plur]            I        have  written

If the antecedent is human, wie is selected instead of die, but only if the relative pronoun has the function of indirect object, or is part of a prepositional phrase or a possessor within a noun phrase. We call relative pronouns such as wie 'who', wat 'what' and waar 'where' W-forms and relative pronouns such as die 'that' and dat 'that' D-forms. In a prepositional phrase, a non–human relative pronoun gives rise to the pronoun waar 'where'. A relativised possessor is often realised as wiens 'whose'. Examples are given in (16).

(16)   a.       de man       wie / die      we     dat    hebben       gevraagd

                   the man      REL           we     that   have           asked

         b.      de man       *wie / die    we     hebben       gezien

                   the man      REL           we     have           seen

         c.       de man       met   wie /*met die /?waarmee       zij     binnenkwam

                   the man      with  REL/ withREL/REL–with     she    came in

         d.      de man       wie /*die     zijn   boeken       ik      heb    gekocht

                   the man      REL           his     books         I        have bought

         e.       de man       wiens                   boeken       ik      heb    gekocht

                   the man      whose        books         I        have  bought

• The variation in the form of the relative pronoun is thus dependent on the grammatical function of the relativised constituent. In the SAND, we have compared variation in form between subject and object relatives, keeping the antecedent constant: de man 'the man'. The distribution of the attested variants is shown on the maps 82a–84a. In addition to this, we have looked at variation in locative relatives of the type de bank waarop zij zaten 'the couch where–upon they sat' (map 88a–88b) and variation in possessive relatives such as de jongen wiens moeder 'the boy whose mother' (map 89a–90a).

In the standard language, the relative pronoun cannot be followed by a lexical complementiser (17a) and it is impossible to replace the relative pronoun by the complementiser dat 'that', as is possible in English (17b,c) (we assume here that dat and that are complementisers in such cases).

(17)   a.    * de man       die     dat    ze      hebben       geroepen

                   the man      REL that   they have           called

         b.    *          de man       dat    ze      hebben       geroepen

                   the man      that   they have           called

         c.       the man      that   they called

In various dialects of Dutch, these options are available (cf. maps 82–84a; and see chapter 1, map 16b/17a).

         Unlike Standard Dutch, the dialects sometimes show variation in a form which is dependent on the depth of embedding of the clause in which a relative pronoun receives its grammatical function. In Oostende (West–Vlaanderen), for example, the relative clause starts with the relative pronoun die if the pronoun has the grammatical function of (human) subject of the relative clause (18a). However, if the relative clause has an embedded clause and the grammatical subject of that embedded clause is relativised, the relative clause starts with da (18b). The variation for long relativisation is depicted on the maps 85a–87b.

(18)   a.       Das            die vent      die     da      graptje        et      verteld. (Oostende)

                   that–is        the man      REL that   joke           has    told

b.      Das     die vent    da–nk         peizn          die     da      komieksje   et      verteld.

         that–is the man   COMP–I   think REL that   joke            has    told

Relative clauses without antecedent, the so–called ‘free relatives’, have the forms wie and wat as relative pronouns, in contrast with die and dat in relative clauses with an antecedent. This is shown in (19).

 (19)  a.       Wie   geld            heeft,          moet mij    wat             geven.

                   REL  money        has,   must me     somewhat  give

                   'Anybody who has got money should give me some.'

         b.      Wat   hij     zegt,  is       onzin.

                   REL  he      says, is       nonsense

The sentence in (19a) was part of the SAND questionnaire. The variation we have found, concerns the use of die instead of wie and the occurrence of a complementiser after the relative pronoun.

In summary, we have found various types of variation in the domain of relativisation:

(i)      w–pronoun (wie, wat, waar, ...) instead of d–pronoun (die, dat, daar, ...);

(ii)     relative pronoun followed by complementiser;

(iii)    variation in the form of the complementiser;

(iv)    complementiser instead of relative pronoun, depending on the grammatical function of the relative pronoun;

(v)     complementiser instead of relative pronoun, depending on depth of embedding of the clause in which the relative pronoun has its grammatical function.

5.1.3  Partial fronting of question words

The view that long fronting should be broken up into two or more instances of short fronting (cf. 5.1.1) is corroborated by the fact that doubling of question words is found in child language and in languages other than Standard Dutch. Children that are raised with Standard Dutch produce sentences such as (20) spontaneously, although these sentences are claimed to be unacceptable in the standard language.

(20)   a.       In      welk           huis   denk  je       waar          ik      woon?

                   in      which         house         think you   where         I        live

         b.      Wat   denk  je       wat    ik     zie?

                   what           think you   what           I        see

Question word doubling of the type in (20b) is also found in adult languages such as Afrikaans, German and Frisian.

(21)   a.       Wer   glaubst       du     wer   nach  Hause         geht?                    German

                   who  believe        you   who  to      house                   goes

                   'Who do you believe will go home?'

         b.      Wêr            tinke jo      wêr't           Jan    wennet?                        Frisian

                   where         think you   where-that John  lives

                   'Where do you think John is living?'

In German and Frisian, sentences like (22) occur as well.

(22)   a.       Was  glaubst       du     wen   Peter getroffen    hat?                     German

                   what believe        you   who  Peter met             has

                   'Who do you believe Peter has met?'

         b.      Wat   tinkst                   wêr't           se      wennet?                                 Frisian

                   what think-you   where-that she    lives

                   'Where do you think she is living?'

Sentences such as (21) and (22) are generally referred to as cases with partial Wh-movement. It is remarkable that the 'real' question word in (22) is fronted within the subordinate clause. Was and wat in (22) are considered to be expletive question words which are inserted in the first position of the main clause, instead of the 'real' question word which remains in the subordinate clause. These expletive elements do not contribute to the interpretation of the clause and are comparable to the expletive there in there is a man in the garden.

         Whereas partial question-word fronting does not appear in Standard Dutch, the construction is attested in many dialects of Dutch. Map 91a demonstrates the distribution of partial question-word fronting of the expletive type (cf. (22)) and map 91b shows the distribution of partial question-word fronting of the doubling type (cf. (21)) in sentences with the question word wie 'who'.

         We indicated above that partial question-word fronting can be argued to corroborate the view that long question-word fronting should be broken up in two or more instances of short fronting. An alternative view is that partial question-word fronting does not involve a special case of long question-word fronting, but should be analysed as fronting of a question word from the main clause. The subordinate clause should then be taken to be a relative clause with the main clause question word as the antecedent. The fact that there are cases of partial question-word fronting in dialects of Dutch in which the subordinate clause is introduced by the relative pronoun die 'which', seems to support such an analysis at least for some cases.

5.1.4  Other instances of fronting

5.1.4.1  Split fronting

Normally, a constituent is fronted as a whole. Fronting of a part of the constituent gives rise to ungrammaticality. For Standard Dutch, this is illustrated in (23).

(23)   a.       Welke         wedstrijd    heb    je       gezien?

                   which         match         have you   watched

         b.     *Wedstrijd   heb    je       welke                   gezien?

                   match         have  you   which         watched

         c.       Hij    heeft drie   boeken       geschreven.

                   he      has    three books         written

         d.     *Boeken      heeft hij     drie   geschreven.

                   books         has    he      three written

The construction in (23d) occurs in various Dutch dialects, as is shown on map 93a. However, we do not know for sure that the fronted element is really a part of a split constituent. In Noord–Brabant, we find constructions such as (24), which look like cases of split fronting, but which cannot be analysed as such, given the ungrammaticality of the noun phrase *die geen 'those none'.

(24)   (Over goede voornemens:)     Die    heb    ik      geen.                    (Schijndel)

         (about good intentions)         those have  I        none

                                                        'I do not have any.'

5.1.4.2  Preposition stranding

Contrary to English, Standard Dutch does not allow fronting of the object of a preposition, thereby stranding the preposition. Preposition stranding is only allowed in Standard Dutch in the case of the so–called R–pronouns (i.e., one of the words er 'there', daar 'there', hier 'here', waar 'where', ergens 'somewhere', nergens 'nowhere', overal 'everywhere'). This is shown in (25).

(25)   a.       I talked with that boy.

b.      That boy I talked with.

c.       Ik      ben    met    die     jongen        naar   de      markt                   geweest.

         I        have  with  that   boy            to      the    market        been

         d.    *          Die    jongen        ben    ik      mee   naar   de      markt         geweest.

                   that   boy            am     I        with  to      the    market        been

         e.       (Die jongen)        daar  ben    ik      mee   naar   de      markt                   geweest.

                   that boy              there am     I        with to      the    market        been

         f.       In      de      tuin            (daar)        zit     een    eekhoorn.

                   in      the    garden        there                    sits    a        squirrel

On map 93b, it is illustrated that sentences such as (25d) are possible in a number of Dutch dialects. It is not immediately clear, though, that these dialects are similar to English in this respect. It might also be the case that (25d) is the result of leaving out the R–pronoun daar 'there' from a sentence such as (25e), comparable to the situation in (26f) in which a locative R–pronoun can be omitted in Standard Dutch.

5.1.4.3  Fronting in subordinate clauses with inversion

In Standard Dutch, only constituents with question words or relative pronouns can be fronted in subordinate clauses. In those cases there is no inversion. In Standard German embedded topicalisation with inversion is possible, as is demonstrated in (26b).

(26)   a.     *Ik      geloof         Frits  vinden        veel   mensen       aardig.

                   I        believe        Frits  find            many          people        nice

b.      Ich    glaube         den    Fritz  mögen        viele.

         I        believe        the    Frits  like             many                  

Map 94b shows which dialects of Dutch behave like German in this respect.

5.1.4.4  Fronting in main clauses without inversion

In Standard Dutch, inversion of subject and finite verb is obligatory if a non–subject is fronted. With the exception of a number of well–known cases, French and English do not have this type of inversion.

(27)   a.       En temps de pluie, on ne voit pas grand chose.

         b.      With bad weather, you do not see much.

In some Dutch dialects in Frans-Vlaanderen and West–Vlaanderen, inversion can be absent in main clauses, as is illustrated on map 85a.

5.1.4.5  Fronting in imperative clauses

Modern Standard Dutch is different from Middle Dutch and German in that it does not allow fronting in imperative clauses.

(28)   a.       Nu     sit     weder         op     u       ors!                               Middle Dutch

                   now sit     again           on     your horse

         b.      Das   Buch lies    besser         nicht!                                                        German

                   the    book read   better         not

         c.     *Nu    ga      weer  op     uw    paard          zitten!

                   now go      again on      your horse sit

         d.     *Dat   boek  lees   maar  niet!

                   that   book read   better         not

Fronting in imperatives can be seen in some Dutch dialects, as is shown on map 95b. It mostly involves fronting of the pronoun dat 'that'.

(29)   Zal ik koken? Ja, dat doe maar.                                   (Noord–Brabant)

         shall I cook? yes, that do

A number of dialects also allow long fronting in imperatives (map 95b), which is possible in German as well.

5.1.5  Historic development

As in Dutch dialects, relative constructions that diverge from the modern standard language are often attested in Middle Dutch. For instance, the complementiser dat could follow a relative pronoun (Stoett 1923: 34; e.g., In een uutgehouwen graf, in dien dat noch niemant gheleit was 'in a dig grave, in which that still nobody put was'). According to Verwijs & Verdam (1889; WNT II: 84), this construction belonged to substandard Dutch in 1889. In Middle Dutch, there are also cases in which dat 'that' is used where die must be used in modern Dutch (Stoett: 35; Om sinen misdaet, dat hi ghedaen had 'due to his crime, that he done had'). This phenomenon is now restricted to Vlaanderen. Relatives with long fronting were quite common in Middle Dutch. In Stoett (1923), several examples of long fronting of subject and object relatives are given. They all utilise die as the relative pronoun of the whole relative clause and dat for introducing the embedded one, just as in Standard Dutch (e.g., Desen, dien wie waenden, dat Keye was 'this one, who we thought that was Keye'). Whether the other variants of long fronting in relative clauses were available in Middle Dutch is unclear. What is clear, however, is that long fronting may occur with a resumptive subject or object pronoun (Stoett 1923: 45). In Middle Dutch, we find both d–forms and w–forms as relative pronouns. According to Bennis (2001) and Van der Sijs (2004) the d–forms are gradually replaced by w–forms. For instance, it was quite common in Middle Dutch to use daar as the locative relative pronoun, as in de plaats daar zij woont 'the place there she lives'. At the end of the 19th century the pronoun waar had taken over completely. However, we often find locative daar used as a relative pronoun in dialects in Zeeland, Noord–Brabant and Friesland (map 88b). In possessive relative clauses the replacement of die zijn 'which his' by wie zijn is almost completed. However, in the entire language area, we encounter dialects that use die zijn in addition to wie zijn (map 90a). In modern Standard Dutch, a w–form is required in free relatives. According to Van der Sijs (2004:491), die was still in use in the 16th century, although the rise of wie had begun. Today, many dialects allow die in addition to wie, and in Vlaanderen there are areas that only take the d–form (map 90b). Fronting in imperative clauses is attested in Middle Dutch (cf. 5.1.4.5). Further observations on the issues discussed here can be found in Duinhoven (1988) and Van den Toorn et al. (1997). The diachronic development of the other fronting constructions has not been the topic of much research.

5.2  Discussion of the literature

Much literature on fronting can be found within the generative framework. Key publications are Ross (1967), Chomsky (1977, 1986), Cheng (1991), Richards (1997). The literature on relative clauses is extensive. Overviews of relevant literature can be found in Smits (1988) and De Vries (2002). In recent literature, it is assumed that two positions are available at the front of relative clauses: one for relative pronouns and another one for complementisers. Which of these positions can or must be (un)filled varies and depends on the construction and the language. For a description of some varieties of Dutch, see Pauwels (1958; Aarschots) and Verwijs & Verdam (1889; Middle Dutch), Zwart (2000). The (im)possibility of a relative pronoun followed by a complementiser has been discussed in Chomsky & Lasnik (1997). Pesetsky (1997), Dekkers (1999) and Broekhuis & Dekkers (2000) discuss the absence of relative pronoun and/or complementiser in Aarschots, English, French, and Dutch in the framework of Optimality Theory. For the choice between the relative pronouns die and dat in short relatives in West-Flemish, see Vandekerckhove (2003). The selection of relative pronoun/complementiser in long relatives has received less attention. Bennis & Haegeman (1984) discuss long relatives in West–Flemish (the dialect of Lapscheure). For a discussion of differences between long subject and object relatives in English, see Perlmutter (1971) and Chomsky & Lasnik (1997). For Dutch, see Maling & Zaenen (1978) and Bennis (1983). A discussion of long subject fronting in a number of languages can be found in Van Der Auwera (1984). Free relatives are discussed in Groos & van Riemsdijk (1981). Possessive relatives are discussed extensively in De Vries (2002). One of the earliest discussions of partial question word fronting is found in Van Riemsdijk (1983). For other discussion on this topic, see Dayal (1994) and Horvath (1997). A volume with articles on this subject is Lutz et al. (2000). As far as we know, this issue has not been discussed for Dutch dialects. Split topicalisation in Brabant Dutch is discussed in Van Hoof (1997). For German, see Fanselow (1988) and Kuhn (2001). Conditions on preposition stranding in Dutch have been thoroughly investigated in Van Riemsdijk (1978). For fronting with inversion in subordinate clauses, we refer to Den Besten (1989) and Müller & Sternefeld (1993). Fronting in imperative clauses is described for Brabant Dutch (Kempenland) by De Bont (1962) and Gerritsen (1991).

         The relative clause is an important issue in typological research, with seminal works such as Keenan & Comrie (1977) and Lehmann (1984). However, the fronting of relative pronouns, which is typical for Dutch, is not a prominent issue in typology. The reason is that relative pronouns are largely restricted to European languages, as is shown in the World Atlas of Language Structures (Haspelmath et al 2005). This atlas further shows that the obligatory fronting of question words is typical for European languages too, but is also frequently attested outside Europe. Hawkins (1986) studies fronting in English and German from a typological perspective and is indirectly relevant for Dutch. In the typologically oriented Functional Grammar of Dik (1997), which is also applied to Dutch, fronting is treated in terms of language specific availability of special positions in word order patterns.

5.3  Discussion of the maps

5.3.1  Relative clauses

5.3.1.1  Short subject relativisation  (map 82a) (map in dynaSAND)

The sentence "Dat is de man die het verhaal heeft verteld" ('That is the man who told the story') was part of the questionnaire in the oral interviews. The informants had to translate the sentence into their dialect. The subject of the relative clause is relativised. The variation in the form of the relative pronoun is remarkable. For the representation of relative clauses, a distinction has been made between two types of dialects. The first type has relative clauses that are introduced by a relative pronoun, followed by a complementiser. It leads to phrases such as de man die dat het verhaal heeft verteld ('the man who that the story has told'). This type of dialect is discussed in 5.3.1.4 (map 84a). The dialects belonging to this class (mainly in Friesland and the province of Antwerpen) are not represented on map 82a, as this would make the map difficult to interpret. In the second type of dialects, only one element is used to introduce relative clauses: a relative pronoun or a complementiser. We have attested five different forms: die, wie, dat, der, and den. Since a translation task provides insight into possible forms of the relative pronoun in a particular location, but does not indicate which forms are impossible, the three most frequent forms were investigated in the telephonic questionnaire. The sentence given above was presented with each of the three frequent forms as relative pronoun: die, dat and wie. The informant was asked to indicate whether these sentences occur in his/her dialect. There is a large amount of agreement between the results of the oral and the telephonic interviews in this respect.

         Map 82a combines the results of the oral and the telephonic interviews for dialects that allow only one element to introduce a relative clause. The standard Dutch form die is dominant in the whole language area. After die, we find wie and dat most frequently, at 35 and 34 measuring points respectively. The core area of dat is the area between Oost–Vlaanderen and Vlaams-Brabant. Some occurrences of dat are attested in the east of the Netherlands. Comparing these results with short object relatives (cf. 5.3.1.2), it is important to observe that dat does not occur in West–Vlaanderen (with one exception). The form wie is not clearly demarcated geographically. The majority of wie–dialects are found along the border with Germany. The fact that we hardly find measuring points at which wie is the only available variant, seems to suggest that wie is a secondary form. The forms der and den are attested along the German border only. Der is found mainly in Noord–Limburg and Zuid–Limburg. The form den is found somewhat more to the north.

5.3.1.2  Short object relativisation  (map 82b) (map in dynaSAND)

The sentence "Dat is de man die ze geroepen hebben" ('That is the man which they have called') was tested in the same way as the short subject relative clause: in the oral interview the sentence was translated by the informants, and in the telephonic interview the occurrence of the most frequent forms was determined by a judgement task. Dialects that require or allow a relative pronoun followed by a complementiser are discussed in 5.3.1.4 (map 84a). In this paragraph, we discuss cases in which the relative clause is introduced by one element only, a relative pronoun or a complementiser.

         Just as with subject relatives, we find a rich array of forms. This is reflected on map 82b. The relative pronoun die is present in the whole language area, with the exception of West–Vlaanderen. The distribution area of dat is considerably larger than in the case of short subject relatives (map 82a). On map 82a, dat is mainly restricted to the border area between Oost–Vlaanderen and the western part of Vlaams–Brabant / Antwerpen. On map 82b on the other hand, the majority of Belgian dialects show the appearance of dat, with the exception of Limburg. The distribution of wie in short object relatives is more or less similar to the occurrence of wie in short subject relatives: along the eastern border of the language area. On map 82b, the number of occurrences of wie is somewhat larger, but it appears to be the case that wie is a secondary form on map 82b as well. The form der is found in Dutch Limburg. These dialects show a similarity with Standard German relatives, since they choose a form of the determiner as a relative pronoun, rather than a demonstrative pronoun as in Standard Dutch. The form den is found in Overijssel and once in Groningen. In Zuid–Limburg, the form dem is attested twice, in Kerkrade and Vaals. Finally, the form as is found in Frans-Vlaanderen twice, whereas this form is absent in subject relatives.

5.3.1.3  Short subject and object relativisation; synthesis  (map 83a) (map in dynaSAND)      

An overview of the various forms that are attested in subject and object relatives is given on map 83a. Five different relative systems can be distinguished. Most widespread is the system in which both subject and object relatives are introduced by die. This system, which is also the system of the standard language, is found in 144 locations in the Dutch language area, with the exception of parts of Limburg. In the Belgian language area, it is found in Limburg and the eastern part of Vlaams–Brabant. Second most frequent is the system in which subject relatives are introduced by die and object relatives by dat. This system is found in 55 locations. The centre of this system is located in West–Vlaanderen, and, slightly less frequently, in Oost–Vlaanderen. This system is comparable to the qui-que alternation found in French relatives. The third system has dat in subject and object relatives. It occurs in 31 locations, in an area which consists of Oost–Vlaanderen and western Vlaams–Brabant. Fourthly, we find a system in which wie is available as a relative pronoun in subject and object relatives. There is no clear area for this secondary system. The fifth system concerns the seven dialects in Dutch Limburg that have der in both types of relatives. Finally, there are three dialects in Overijssel that take den as the introducing element for the two relatives: Gramsbergen, Vriezenveen and Ootmarsum. The last two of these locations allow the system der–den as well. This latter system points towards a case–related system. As indicated above, we have not included on this map the cases in which a relative pronoun is followed by a complementiser (cf. map 84a). These dialects generally take die as the relative pronoun and thus are linked to the first system.

5.3.1.4  Short subject and object relativisation + complementiser  (map 84a) (map in dynaSAND)  

On map 84a, the distribution of a relative pronoun followed by a complementiser is demonstrated. Subject relatives and object relatives behave the same in this respect. There are two areas to be distinguished: a northern area with die–t and a southern area with die as/at/da. The combination die–t is found in most locations in Friesland, with subject and object relatives. The combination die as/at/da can be attested in the northeast of Oost–Vlaanderen (Land van Waas) and the western region of the province of Antwerpen. We also find this construction in Zeeland, and in particular in Zeeuws–Vlaanderen. The related combination dien at/da is found with subject relatives in a few dialects in Belgian Limburg and Brabant. In the subject relatives of Belgian Limburg, we find two occurrences of the complementiser wat preceded by a relative pronoun, die in Eigenbilzen and dat in Lauw. For a comparison of the distribution of complementisers in subject relatives, object relatives, locative relatives and free relatives, see map 16b. For a comparison with complementisers following question words, see map 17a.

5.3.1.5  Free relatives + complementiser  (map 84b) (map in dynaSAND)                

In relatives without an antecedent, the so–called free relatives, a complementiser sometimes follows the relative pronoun, as in Wie da geld heeft, moet mij maar wat geven. The sentence "Wie geld heeft, moet mij maar wat geven" ('Whoever has money, should give some to me') was presented in the standard language for translation. The original target was to find out the distribution of wie and die in free relatives (cf. 5.3.1.16 and map 90b). The cases of relative pronouns with complementiser were spontaneous attestations and are not exhaustively tested. However, we find a northern and a southern area, just as on map 84a. In Friesland, the forms wie–t, waa–t and die–t occur in free relatives. Wie–t is also attested in Aalten (Gelderland). The southern area only partially overlaps with the complementiser area on map 84a. The core area is found in Zeeland and surrounding regions; there are two variants: die as/a/an in Zeeuws–Vlaanderen and wie as/a/at on Walcheren, Zuid–Beveland, Tholen, and in Waasland. The latter form is also found twice in Friesland and in Overijssel. In West–Vlaanderen and Oost–Vlaanderen and occasionally in Antwerpen, we find the variant wie dat.

5.3.1.6  Long subject relativisation  (map 85a) (map in dynaSAND)       

In the sentence "Dat is de man die ik denk dat het verhaal verteld heeft" ('That is the man who I think that has told the story'), the noun phrase de man 'the man' is modified by the relative clause die ik denk dat het verhaal verteld heeft. This relative clause contains an embedded clause dat het verhaal verteld heeft. The most deeply embedded clause is introduced by the complementiser dat 'that' in Standard Dutch. The entire relative clause is introduced by the relative pronoun die 'which', which has the grammatical function of subject of the most deeply embedded clause. The variation attested for this sentence can be divided into three parts: (i) the form of the word that introduces the most embedded clause; (ii) the form of the word that introduces the entire relative clause; and (iii) the absence of the subject in the most deeply embedded clause.

The informants were asked to translate the sentence into their dialect. Four of the attested variants were presented as a judgement task in the telephonic interviews in order to check the results. It concerned the following variants: de man die ik denk dat ... (die...dat), de man die ik denk die...(die...die), de man dat ik denk die... (dat...die) and de man dat ik denk dat...(dat...dat). The results of the oral and the telephonic interviews showed a large of amount of agreement for the core areas of the various variants, but differed substantially in transition areas. The telephonic interviews showed a larger distribution for each of the four variants. Assuming that the oral interviews provided a more reliable picture, we have presented the results of the oral interviews on map 85a. In the representation and discussion, we have abstracted away from cases in which a relative pronoun is immediately followed by a complementiser. We find instances of relative pronouns followed by a complementiser in long relatives in the same dialects as with short relatives (map 84a). Dialects with die dat – dat have been treated as dialects with die – dat, dialects with die dat – die as dialects with die – die, etc. We have also abstracted away from the forms der and den as relative pronouns, which phenomenon is predominantly found in Limburg. We have presented these forms as die–forms on map 85a.

The Standard Dutch variant de man die ik denk dat ...(die...dat) is often attested in dialects of Dutch (64 locations), but it is hardly found at all in the provinces of Drenthe, Overijssel, Utrecht, West–Vlaanderen and Vlaams–Brabant. A variant of this construction has a lexical subject in the most embedded clause: de man die ik denk dat hij/die... 'the man which I think that HE ...'. This variant is attested in 21 locations, primarily north of the line Bergen op Zoom – Enschede. The variant de man die ik denk die... (die...die) is found at 35 measuring points and appears to be the major variant in Oost–Vlaanderen (14 attestations). It is also found along the eastern border of the language area, in particular in the neighbourhood of Nijmegen and Arnhem. The variant de man dat ik denk die... (dat...die) is found 36 times, mostly in West–Vlaanderen. We also find dat...die in Oost–Vlaanderen, especially in the southeastern part of this province. The variant de man dat ik denk dat... (dat...dat) is found along the border with French speaking Belgium, and more specifically, in the south of West–Vlaanderen and Oost–Vlaanderen and in Vlaams–Brabant. In southeast Oost–Vlaanderen, the die...die and dat...dat areas overlap. The dat...dat area shows ten occurrences of a lexical subject in the most embedded clause: de man dat ik denk dat hij/die...

The variant de man waarvan ik denk dat hij/die het verhaal heeft verteld 'the man of–which I think that he ...' occurs in 70 locations and is the most frequent variant. It is however questionable whether it concerns a subject relative. It might be derived from Ik denk van die man dat hij het verhaal verteld heeft 'I think of that man that he has told the story' by relativisation of the prepositional phrase van die man 'of that man'. It appears to be the case that long relativisation is problematic in many dialects. The prepositional construction is a way to circumvent long relativisation. The map shows where the informants spontaneously came up with this variant. This does not imply that this variant is excluded in other locations. The majority of attestations of this variant is found along the eastern border with Germany, in particular in Limburg, the eastern part of Gelderland and Overijssel, in Drenthe and Friesland. A variant of the prepositional variant is the construction de man waarvan ik denk dat het verhaal heeft verteld 'the man of–which I think that has told the story'. The subject of the embedded clause is absent, which suggests that it concerns long subject relativisation. This variant is found in nine locations, mostly in the northeast of the Netherlands (Drenthe and Overijssel).

5.3.1.7  Long object relativisation  (map 85b) (map in dynaSAND)         

The relative clause in the sentence "Dit is de man die ik denk dat ze geroepen hebben" ('This is the man which I think that they have called') contains an embedded clause from which the object is relativised. In these cases of long relativisation, the object in the most deeply embedded clause is absent in Standard Dutch. The embedded clause is introduced by the complementiser dat 'that', while the entire relative clause is introduced by die 'which'. As in the case of long subject relativisation, there are three variable properties that are relevant in the Dutch dialects: (i) the form of the word that introduces the most deeply embedded clause; (ii) the form of the word that introduces the relative clause; and (iii) the absence/presence of the subject in the most embedded clause. For this test sentence, the same methodological considerations and abstractions hold as in the case of long subject relatives (cf. 5.3.1.6).

         The large majority of dialects on map 85b shows the variant of the standard language: Dit is de man die ik denk dat ze geroepen hebben (die...dat). We find this construction in 101 locations in the entire language area, with the exception of West–Vlaanderen, Vlaams–Brabant and Utrecht. In six dialects, we find a variant of this construction with a pronoun in the object position of the most embedded clause: Dit is de man die ik denk dat ze hem/die geroepen hebben. These dialects are found in the north. The variant Dit is de man dat ik denk dat ze geroepen hebben (dat...dat) is attested in 53 locations, considerably more than the occurrences of the dat...dat–variant in the case of long subject relatives. Almost all locations with dat...dat are found in West–Vlaanderen, Oost–Vlaanderen and Vlaams–Brabant. The die...die–variant is found somewhat less frequently than with long subject relativisation and the area is less clearly demarcated. Possible core areas are Oost–Vlaanderen and surroundings, the shore of the former Zuiderzee in North–Holland, and Dutch Limburg. The variant Dit is de man dat ik denk die ze geroepen hebben (dat...die) is found only eight times and occurs without a demarcated area. This contrasts with the subject relatives, in which the dat...die–variant was found 36 times, particularly in West–Vlaanderen. It suggests that the dat...die–variant is irrelevant for object relativisation. Finally, we find the prepositional variant Dit is de man waarvan ik denk dat ze hem/die geroepen hebben 'This is the man of who I think that they have called him'. The map shows in which locations this variant was given as the only option or as an alternative. It is possible that this variant also occurs in other locations. This construction is attested 49 times, much less than in the case of subject relatives. It is mainly found in peripheral areas: along the border with French speaking Belgium and with Germany, and on the Wadden islands. Some occurrences are attested in Drenthe and in Utrecht.

5.3.1.8  Long subject and object relativisation; synthesis  (map 86a) (map in dynaSAND)       

The similarities and differences between long subject and long object relativisation are demonstrated on map 86a. There are six systems that are most prominent. System I has dat...die in subject relatives and dat...dat in object relatives. Almost all 23 instances of system I are found in West–Vlaanderen. System II has die...die in subject relatives and die...dat in object relatives. This system is attested in 22 locations, predominantly in Oost–Vlaanderen. It is also found in the east of the Netherlands. System III has dat...dat in subject and object relatives. Vlaams–Brabant appears to be the core area of this system, but it is also found in the east of Oost–Vlaanderen and in the south and on the coast of West–Vlaanderen. System III is found in 17 locations. System IV has the prepositional relative clause (waarvan...dat) with a lexical subject/object. It is found in 31 locations, often in the east of the Netherlands, with a concentration in Limburg and the northwest of Drenthe. It is not found in West–Vlaanderen and Oost–Vlaanderen and hardly at all in Vlaams–Brabant (one occurrence) and Belgian Limburg (three occurrences). System V is the system of the standard language: die...dat in subject and object relativisation. This system is attested most often, in 45 locations throughout the entire language area, with the exception of the provinces of Drenthe, Overijssel, Utrecht and West–Vlaanderen. There is no clear areal demarcation of this system. The final system, system VI, has waarvan...dat in subject relatives and die...dat in object relatives. This system is found at 23 measuring points. These locations are mainly found in Zeeland, Noord–Brabant and Limburg, and along the eastern border.

5.3.1.9  Long subject relativisation + lexical subject  (map 87a)

Of the eight different variants that we encountered in the case of long subject relativisation on map 85a, five variants have a subject in the most embedded clause. We count the variants dat...die and die...die among the cases with a subject since the subject relative pronoun die stays within the most embedded clause. From the literature it is known that the absence of the embedded subject in long subject relatives, long questioning and long topicalisation leads to ungrammaticality in many languages. It is therefore interesting to determine whether subjectless long relatives can geographically be distinguished from relatives with an embedded subject. Map 87a shows that the tendency to fill the embedded subject position is present in the entire language area. This tendency is somewhat weaker in the provinces of Vlaams–Brabant and Antwerpen, but even there, we find many instances of long subject relativisation with a lexical embedded subject.

A white diamond indicates that the dialect requires an embedded subject, a black diamond indicates absence of an embedded subject, whereas a diamond with a "Y" points at optionality of an embedded subject. The colours on the map indicate the likelihood that a particular variant is attested in that area (cf. Introduction, section Colour maps for a description of this map type).

5.3.1.10  Long object relativisation + lexical object  (map 87b)

In four of the six variants of long object relativisation, the object is present within the most deeply embedded clause. According to the literature, many languages show a contrast between the absence of the subject and the absence of the object in long relativisation, question fronting and topicalisation. Absence of the subject quite often gives rise to ungrammaticality (cf. map 87a), whereas absence of the object is acceptable in many cases. Map 87b corroborates this view. In the south (Vlaanderen, Zeeland, Noord–Brabant) and in the west, the absence of the object in long relativisation is unproblematic. If we put together dialects in which the object can be optionally present (indicated by a “Y” within a diamond) and dialects in which the object must be absent, there are just a few dialects remaining in which an absent object in the most embedded clause gives rise to ungrammaticality. These dialects are scattered over the language area. A comparison between maps 87a and 87b thus confirms the contrast between fronting a subject and fronting an object.

5.3.1.11  Locative relativisation + complementiser  (map 88a) (map in dynaSAND)

The sentence "De bank waar ze op zaten was pas geverfd" ('The couch where they sat on was just painted') was presented in the oral interviews for translation into the dialect of the informant. Missing data are added in the telephonic interviews. On maps 88a and 88b, the data from the oral and telephonic interviews are put together. The test sentence shows variations in the form of the relative pronoun, the presence and the form of the complementiser and the separation of relative pronoun and preposition. Schematically, we find the following variants (leaving out variants that occur once or twice).

– de bank waar ze op zaten            (waar...op)

– de bank waar dat ze op zaten      (waar dat...op)

– de bank waar at/as ze op zaten    (waar at/as...op)

– de bank wer–t ze op zaten            (waar–t...op)

– de bank waarop ze zaten             (waarop)

– de bank waarop dat ze zaten       (waarop dat)

– de bank daar ze op zaten             (daar...op)

– de bank der–t ze op zaten            (daar–t...op)

– de bank daarop ze zaten              (daarop)

– de bank dat ze op zaten                (dat...op)

Almost all dialects with waarop 'where–on' and daarop 'there–on' allow a separation of preposition and relative pronoun. The non–split cases are thus considered to be variants of the split ones. These are not marked separately on map 88a.

         On map 88a, we notice that locative relatives allow the addition of a complementiser more easily than in the cases of subject, object and free relatives (cf. maps 84a/b). The entire Belgian area, Dutch Limburg, Zeeland, Friesland, and the eastern border allow this construction. The variant with a complementiser instead of a relative pronoun (dat...op) is found primarily in West–Vlaanderen and Oost–Vlaanderen, and in the neighbourhood of Nijmegen. The complementisers as and at appear in two core areas: the first area comprises the south of Zeeland, West–Brabant, Waasland, and the western part of the province of Antwerpen. The second core area is found in Overijssel. Finally, the variant with the complementiser suffix –t is found almost exclusively in Friesland.

5.3.1.12  Locative relativisation: d–forms vs w–forms  (map 88b) (map in dynaSAND)

On map 88b, all variants that have a relative pronoun starting with a w– are taken together and compared with variants which have a relative pronoun that starts with a d–. The variant without relative pronoun (dat...op) is left out. We observe that D–forms are almost absent in Belgium. In the Netherlands, D–forms are mostly found in Zeeland, Noord–Brabant and Friesland. It is remarkable that dialects that require a D–form are hardly attested. It seems that D–forms are secondary.

5.3.1.13  Possessive relativisation + complementiser  (map 89a) (map in dynaSAND)

In the sentence "De jongen wiens moeder gisteren hertrouwd is" (stond achter me) ('The boy whose mother is remarried yesterday, was standing behind me') the possessor of the embedded subject is relativised. The variation we have found concerns variation in the form of the relative pronoun, the presence of a complementiser and the presence of the possessive pronoun zijn 'his'. The test sentence was presented as a translation task. Some of the variants that were available from the answers on the written questionnaire were presented to the informants with the question whether these variants occur in their dialects. The data were completed and supplemented with data from the telephonic interviews. We have attested a large amount of variation in this construction.

         On map 89a, we see the distribution of variants allowing the relative pronoun to be followed by a complementiser. We have abstracted away from the specific form of the complementiser: the forms dat, da, as and a are taken together. Possessive relatives with a complementiser occur frequently in the Belgian language area. In West–Vlaanderen, the variant de jongen van wie dat zijn/de moeder (...of who that...) is dominant. The variant de jongen waarvan dat zijn/de moeder (...where–of that...) is also found in West–Vlaanderen, and in Vlaams–Brabant and Antwerpen. The variant

de jongen die dat zijn/de moeder (...which that...) is attested at the eastern border of Oost–Vlaanderen. The variant de jongen wie dat de/zijn moeder (...who that...) occurs in Vlaams–Brabant and Antwerpen, and, to a lesser extent, in Zeeland and the islands of Zuid–Holland. Finally, we find the construction with a complementiser instead of a relative pronoun – de jongen dat zijn moeder – along the southern border of the language area and, somewhat more often, in West–Vlaanderen.

5.3.1.14  Possessive relativisation, possessive pronouns  (map 89b) (map in dynaSAND)

On map 89b, we have compared dialects that make use of the possessive pronoun zijn 'his' (die zijn, wie zijn, van wie zijn, wie dat zijn, die dat zijn, van wie dat zijn, dat zijn) with dialects that do not (diens, wiens). It is remarkable that the variant of the standard language (wiens) does not appear in Frans-Vlaanderen, Vlaams–Brabant, Limburg, Zuid–Holland and Zeeland; it occurs rather frequently in the rest of the language area. Diens is found in a few dialect areas: three times in the north of Noord–Holland, six times in the east of Gelderland, and six times along the Belgian–Dutch border. It can be deduced from map 89b that almost all dialects with diens or wiens also allow a variant with in which the possessive pronoun zijn is present.

5.3.1.15  Possessive relativisation with die and wie  (map 90a) (map in dynaSAND)

The dialects that allow a variant of this construction with the W-pronoun wie – wie zijn, wie dat zijn, van wie zijn/de, van wie dat zijn/de – are compared with the dialects that use the D-pronoun die – die zijn, die dat zijn (the variants with wiens or diens are ignored here). The majority of dialects allow both a wie–variant and a die–variant. In this construction, the wie–forms and the die–forms appear to show free variation. Areas that seem to prefer wie–variants are West–Vlaanderen and central Antwerpen. Areas that choose for die–variants are found in Oost–Vlaanderen, Belgian Limburg, the north of Noord–Holland, the area along the IJssel, and the north of the province of Groningen.

5.3.1.16  Free relatives with die and wie  (map 90b)

Free relatives differ from other relative clauses since the relative pronoun has no antecedent. Our target was to find out which dialects make use of die as a relative pronoun and which dialects use wie. Again, we have attested complementisers in this construction (cf. 5.3.1.5, map 84b). Informants were asked to translate the sentence "Wie geld heeft, moet mij maar wat geven" ('Who has money, should give me some') into their dialect. If there were indications that more than one variant might be possible, these variants were presented as a judgement task. These sentences were not a part of the telephonic interviews.

         On map 90b, we see the distribution of die and wie. It appears that most of the dialects have both variants. The northwest of the Netherlands predominantly opts for wie. In the east of Overijssel and Gelderland, we find a second area that allows wie only. The Belgian language area can be divided into three regions. In West– and Frans-Vlaanderen, most dialects select die as the pronoun in free relatives. Some of these dialects prefer the complex form den dieën die instead of die. In Belgian Limburg, the form die is generally the only possibility, with the exception of the dialects close to the Dutch Limburg border. The third area goes from the southeast of Zeeland, via Oost–Vlaanderen and the western part of Antwerpen to the centre of Vlaams–Brabant. In this area, there are many dialects that select wie as the only available candidate.

5.3.2  Partial fronting of question words

5.3.2.1  Expletive question words  (map 91a)

The sentence "Wat denk je wie ik in de stad heb gezien" ('What do you think who I saw in town') was presented in Standard Dutch at all measuring points as a judgement task. If the sentence occurred in the dialect, the informant was asked to translate the sentence. In Standard Dutch, this sentence would be: Wie denk je dat ik in de stad heb gezien? lit. who think you that I in the city have seen 'Who do you think I saw in the city?' In the test sentence wie 'who' is preposed in the embedded clause; the first position in the main clause is taken by the question word wat 'what', which does not contribute to the interpretation of this sentence.

At first sight, it seems as if the construction with wat...wie occurs in the whole language area on map 91a (for an explanation of symbols and colours, see the Introduction). This might indicate that this phenomenon represents a distinction between formal and informal use of the language. However, on closer inspection, regional differences can be observed. In the province of Drenthe, and in particular in the centre of Drenthe, the construction is generally absent. The same is true for the Randstad (i.e., the south of Noord–Holland, the northwest of Zuid–Holland, and Utrecht), with the exception of Katwijk. The dialects spoken in Holland are similar to Standard Dutch in this respect. In the Belgian language area, we find the construction considerably less than in the Netherlands. The construction is especially lacking in Frans– and West–Vlaanderen. The provinces of Oost–Vlaanderen and Antwerpen are mixed zones. The west of Oost–Vlaanderen patterns with West–Vlaanderen, whereas the construction is found more often in the centre and the eastern part. The centre of the province of Antwerpen appears to allow the expletive question construction whereas it is quite often absent in peripheral areas. In Vlaams–Brabant and the south of Belgian Limburg the construction doesn't occur, except in a few isolated cases.

5.3.2.2  Doubling of question words  (map 91b)

The sentence "Wie denk je wie ik in de stad gezien heb" ('Who do you think who I saw in town') was presented as a judgement task and as a translation task. In Standard Dutch, this sentence would be: Wie denk je dat ik in de stad heb gezien? lit. who think you that I in the town have seen 'Who do you think I saw in town?' In the test sentence we find doubling of wie 'who', once in the main clause, once in the embedded clause. These two instances of wie 'who' are interpreted as one argument of the verb zien 'see'.

Map 91b shows the distribution of question word doubling. Question word doubling has been attested in 109 dialects. It is a phenomenon that occurs predominantly in the Netherlands. In the Belgian area, it is absent in most dialects. Only in the eastern part of Oost–Vlaanderen and the western part of Antwerpen and Vlaams–Brabant, doubling of question words is found. In few dialects in Belgian Limburg, the construction has been attested. In the Netherlands, the doubling phenomenon is found on a large scale in the north and the east, i.e., the provinces of Noord–Holland, Friesland, Groningen, Drenthe, and Limburg, and everywhere along the German border. In the Randstad, doubling does not occur, whereas the informants in Noord–Brabant appear to disagree on the test sentence.

5.3.2.3  Partial fronting in questions + relative pronoun  (map 92a) (map in dynaSAND)

Sentences such as Wat denk je die ik in de stad gezien heb? ('What do you think which I saw in town') and Wie denk je die ik in de stad gezien heb ('Who do you think which I saw in town') have a relative pronoun in the embedded clause, instead of a question word. Sentences such as these were occasionally produced spontaneously during the interviews. This is depicted on map 92a. Four attestations of the type wat ... die have been attested, in Beverwijk (Noord–Holland), Staphorst, Hooghalen and Coevorden in the north–east. In three of these, the construction wat – wie is found as well; wat ... die may thus be analysed as a secondary form. The wie – die construction is found three times in the south of Noord–Holland (Beverwijk, Amsterdam and Weesp), four times in the neighbourhood of Nijmegen (Doesburg, Nijmegen, Cuijk and Liempde), and once in Drenthe (Hooghalen). Seven of these dialects also allow the wie ... wie construction. This may indicate again that the construction with a relative pronoun is secondary.

5.3.2.4  Correlation expletive question words and doubling  (map 92b)

Map 92b correlates the occurrence of wat-wie constructions and wie-wie constructions. Three areas allow both variants of partial fronting to occur: (i) the area that consists of Oost–Vlaanderen and the western part of Vlaams-Brabant and Antwerpen, (ii) Zuid-Limburg, and (iii) the eastern part of the language area, including Friesland and Groningen. In West–Vlaanderen, Antwerpen, Noord–Holland, Gelderland, and Overijssel, only the wat-wie construction is found. Only wie-wie is found in West-Vlaanderen and in Drenthe. Areas that appear to disallow both constructions are French Vlaanderen, parts of West-Vlaanderen, Zuid-Holland, and Belgian Limburg.

5.3.3  Other instances of fronting

5.3.3.1  Split topicalisation  (map 93a) (map in dynaSAND)

The sentence "Boeken heeft Jan drie" ('Books, John has three') was part of the oral interviews as a judgement task. Map 93a shows that only 15 dialects allow split topicalisation. An influence of German might be relevant since 12 dialects are in the eastern part of the language area, along the German border. However, such an hypothesis faces the problem that these dialects are scattered and that the majority of the eastern dialects do not allow split topicalisation.

5.3.3.2  Preposition stranding  (map 93b) (map in dynaSAND)

The sentence "Die rare jongen ben ik mee naar de markt geweest" ('That strange boy I accompanied to the marketplace') was presented as a judgement task. As demonstrated on map 93b, fronting of a noun phrase leaving behind the related preposition, i.e., prepostion stranding is not found in Belgium, with the exception of two isolated dialects (Koewacht in Oost-Vlaanderen and Lauw in Belgian Limburg). In the south of the Netherlands, this type of fronting is seldom allowed. It was attested in two locations in Noord-Brabant (Vlijmen and Vorstenbosch) and two locations in Zuid-Limburg (Itteren and Ubachsberg). In Holland, with the exception of the islands of Zuid-Holland, it hardly occurs. This type of preposition stranding is quite common in the northeast and central-eastern part of the language area, in particular in Friesland, Drenthe and Overijssel. It is interesting to observe that there is a coherent area with eight dialects on the border of the Gooimeer (Utrecht) and the Markermeer (Noord-Holland) in which this construction is available.

         As was demonstrated in 5.1.4.2, there are two potential analyses of this construction: as 'real' preposition stranding of the English type or as a left-dislocated noun phrase with a deleted R-pronoun in fronted position (Die rare jongen (daar) ben ik mee naar de markt geweest). To answer the question which of these two options is present in Dutch dialects, the sentence Dit denk ik niet aan ('This I do not think about') was presented in the oral interviews as a judgement task. This sentence is possible in English. However, it is not very plausible that the Dutch sentence is derived through the deletion of a fronted R-pronoun, given the fact that the corresponding sentence with an R-pronoun present gives rise to ungrammaticality, at least in Standard Dutch: *Dit daar denk ik niet aan ('This there I do not think about'). At no measuring point was the occurrence of Dit denk it niet aan attested. This seems to indicate that the type of preposition stranding discussed here is a case of deletion of a fronted R-pronoun, rather than the fronting of the object of a preposition.

5.3.3.3  Topicalisation with inversion in embedded clauses  (map 94a) (map in dynaSAND)

The sentence "Ik geloof deze jongen vinden ze allemaal wel aardig" ('I believe this boy, they all like') was presented to all informants in order to determine which dialects allow inversion of subject and finite verb in a subordinate clause. The informants were asked whether this sentence would be possible in their dialect and they were requested to translate this sentence. To prevent the informants from interpreting the subordinate clause as a main-clause quotation, the verb geloven 'believe' was selected, since it does not occur with a direct-speech complement. Moreover, the test sentence was presented in a context which did not readily allow direct speech: Ik heb nog nooit iemand kwaad zien worden op deze jongen. Dus ik zeg: ik geloof deze jongen vinden ze allemaal wel aardig ('I never saw anyone getting angry at this boy. So I say: I believe this boy, they all like'). Finally, the sentence was presented as a prosodic unit without a break after the word geloof.

         In many cases, the translation involved a construction different from the one that was targeted here. In these cases, the possibility cannot be excluded that embedded topicalisation + inversion is indeed possible. On map 94a, only the measuring points have been included at which a positive answer was given to the judgement question and at which the translation did not include a pause after the word geloof. Four of these points came from the telephonic interviews. Most dialects with embedded topicalisation + inversion are found in the eastern part of the language area, and in particular in Friesland, eastern Overijssel and Gelderland, and the south of Dutch Limburg. Some occurrences are attested in Drenthe, Groningen, Utrecht and Zeeland. The construction is absent in the Belgium and is almost totally absent in Noord-Brabant and the north of Limburg.

5.3.3.4  Correlation embedded topicalisation and preposition stranding  (map 94b)

The distribution of preposition stranding (map 93b) and embedded topicalisation (map 94a) shows many similarities. The construction themselves show some similarities, too. Both are cases of fronting of a constituent that is not a question-word or a relative pronoun, and both could be cases of left dislocation with omission of daar 'there' and die 'that'.

The number of dialects with embedded topicalisation + inversion is somewhat smaller than the number of dialects with preposition stranding, but these dialects all occur in approximately the same area. Especially in the east, we find dialects that have both possibilities. However, the rest of the language area shows that these two constructions do not necessarily co-occur.

5.3.3.5  Topicalisation without inversion  (map 95a) (map in dynaSAND)

The sentence "Met zulk weer je kunt niet veel doen" ('With such bad weather, you cannot do much') was translated into the dialects and the informants were asked to judge the sentence in their dialect. In this sentence, the phenomenon of fronting did not give rise to the inversion of subject and finite verb, as is obligatory in the standard language. This sentence was included in the written questionnaire. It transpired that the construction was only attested in the western part of the language area and in particular, in West–Vlaanderen. Due to this fact, the construction was tested only in 107 western locations, as is shown on the map. In 11 instances, the construction was indeed attested. These locations are found in West–Vlaanderen and especially in Frans–Vlaanderen. This might be due to the fact that this type of topicalisation can be found in French.

5.3.3.6  Fronting in imperative clauses  (map 95b) (map in dynaSAND)

The following sentences were part of the oral interviews: "Dat doe maar!" ('That, do!'), "Dat boek beloof mij dat je nooit meer zult verstoppen!" ('That book, promise me that you will never hide anymore!'), "Wat zeg mij dat je gekocht hebt!" ('What, say me that you bought!'). These sentences were presented in a translated version and the informants were asked to judge their acceptability in their dialect. The sentence Dat doe maar was presented in the following context; Speaker A: Zal ik vanavond koken? ('Shall I cook tonight?'), Speaker B: Dat doe maar. The context makes the sentence adhortative or admissive, rather than imperative. Given the context, the demonstrative pronoun dat has an explicit antecedent (to cook tonight), and interference with the rather similar sentence Dat doeT maar (in which case the pronoun is the subject of the verb doet) is avoided. The distribution of Dat doe maar is interesting. It is attested along the eastern border, with a clear concentration in Noord-Brabant and, to a lesser extent, in the south of Dutch Limburg. No occurrences are found in Belgium. The two locations in the west of the Netherlands (Eemnes and Knollendam) are isolated. As observed in 5.3.3.6, fronting in imperative clauses is found in Standard German. This might explain the almost exclusive occurrence of this construction along the German border.

         The two other examples of fronting in imperatives concern sentences in which the fronted constituents interpretatively belong to the embedded clause. Moreover, the fronted constituent is a full nominal phrase (dat boek) or a question pronoun (wat) instead of a demonstrative pronoun. From the literature, it is known that such sentences are attested in dialects of German. In the Dutch language area, these hardly occur. Fronting of dat boek is attested in Aalter (Oost-Vlaanderen), Itteren (Zuid-Limburg), Sibculo (Overijssel) and Nieuw-Scheemda (Groningen). The fact that these dialects are widely separated indicates that this construction does not really exist in Dutch dialects. The fronting of a question word is found in Itteren (Zuid-Limburg), Zoutleeuw (Vlaams-Brabant), Didam en Brummen (Gelderland) and Waskemeer (Friesland). These dialects are found in the east; however, the number and the dispersion of the measuring points raise doubt as to the existence of this construction in Dutch dialects.

5.4  Literature on fronting

Bayer, J. (1984). 'Towards an explanation of certain that-t phenomena: The COMP-

node in Bavarian.' In W. de Geest & Y. Putseys (eds.) Sentential complementation. Dordrecht/Cinnaminson: Foris Publications. 23–32.

Bennis, H. (1983). 'Subjektloze zinnen.' Spektator 12, 409–427.

—     (2001). Tegengestelde krachten in taal. Oratie Universiteit van

Amsterdam. Vossiuspers.

Bennis, H. & L. Haegeman (1984). 'On the status of agreement and relative clauses

in West-Flemish.' In W. de Geest & Y. Putseys (eds.) Sentential complementation. Dordrecht/Cinnaminson: Foris Publications. 33–53.

Bos, G.F. (1964). Het probleem van de samengestelde zin. Den Haag: Mouton & Co.

Bouwman, A.C. (1918). Bijdrage tot de syntaxis der <<dat>>-zinnen in het

Germaansch. Utrecht: Electrische Drukkerij L.E. Bosch & zoon.

Brachin, P. (1973). 'Een in dubbel opzicht “vreemde” constructie?' De Nieuwe

 Taalgids 66, 59–60.

—     (1974). 'Een in dubbel opzicht “vreemde” constructie.' De Nieuwe

Taalgids 67. 228–234.

Broekhuis, J. & J. Dekkers (2000). 'The Minimalist Program and Optimality Theory:

Derivations and Evaluations.' In J. Dekkers et al (eds.) Optimality Theory: Phonology, Syntax and Acquisition. Oxford: Oxford University Press.

Burridge, K. (1993). Syntactic change in Germanic: Aspects of language change in

Germanic with particular reference to Middle Dutch. Amsterdam: John Benjamins.

Cheng. L. (1991). On the typology of Wh-questions. Diss. MIT.

Chomsky, N. (1977). 'On Wh-movement.' In P. Culicover et al. (eds.) Formal syntax.

New York: Academic Press. 71–132.

Chomsky, N. & H. Lasnik (1977). 'Filters and Control.' Linguistic Inquiry 8:3.

425–504.

Cinque, G. (1990). Types of A-bar-dependencies. LI Monograph 17. Cambridge:

 MIT Press.

Cole, P. & G. Hermon (2000). 'Partial Wh-Movement: Evidence from Malay.' In U.

Lutz, G. Müller & A. von Stechow (eds.) Wh-Scope Marking. Linguistik Aktuell 37. Amsterdam: John Benjamins. 101–130.

Dayal, V. S. (1994). 'Scope Marking as Indirect wh-Dependency.' Natural Language

Semantics 2. 137–170.

De Bont, A.P. (1962). Dialekt van Kempenland. Meer in het bijzonder d'Oerse taol.

Assen: Van Gorkum.

De Rooij, J. (1967). 'Lummel dat je bent.' Tijdschrift voor Nederlandse Taal- en

Letterkunde 83. 108–121.

Dekkers, J. (1999). Derivations and evaluations. Diss. Universiteit van

Amsterdam. LOT Dissertations 21.

Den Besten, H. (1989). Studies in West Germanic Syntax. Amsterdam/Atlanta:

Rodopi.

De Visser, M. & A. Goeman (1979). 'Voegwoord, relatief partikel en persoonsvorm

in een dialect.' Taal en Tongval 31. 222–241.

De Vries, M. (2002). The syntax of relativization. Diss. Universiteit van

Amsterdam. LOT Dissertations 53.

Dik, S. (1997). The Theory of functional Grammar. Berlijn: Mouton de Gruyter.

Duinhoven, A.M. (1988). Middelnederlandse syntaxis. Synchroon en diachroon I.

Leiden: Martinus Nijhoff.

Dyk, S. & J. Hoekstra (1987). 'I oersjoch fan 'e studzje fan 'e Fryske syntaksis.' In

S. Dyk & J. Hoekstra (eds.) Ta de Fryske syntaksis. Leeuwarden: Fryske Akademy. 7–37.

Fanselow, G. (1988). 'Aufspaltung von NPs und das Problem der 'freien'

Wortstellung.' Linguistische Berichte 114. 91–113.

Fokkema, B. & H.T.J. Miedema (1962). 'De konjunksje at.' Pompeblêdden Snits 33.

156–158.

Gerritsen, M. (ed.)(1991). Atlas van de Nederlandse Dialectsyntaxis (AND). Deel 1.

 Amsterdam: P.J. Meertensinstituut.

Grosu, A. & F. Landman (1998). 'Strange relatives of the third kind.' Natural

Language Semantics 6. 125–170.

Haeseryn, W. et al. (1997). Algemene Nederlandse Spraakkunst. Groningen: Wolters-

Noordhoff. 2e druk.

Haspelmath, M. et al. (2005). World Atlas of Language Structures. Oxford: Oxford

         University Press.

Hawkins, J.A. (1986). A comparative typology of English and German. Unifying the

 contrasts. London: Croom Helm.

Hoekema, T. (1990). 'De bynwurden as, at, dat, of, oft.' Us Wurk 39. 64–76.

Hoekstra, E. (1991). Licensing conditions on phrase structure. Diss. Universiteit

van Groningen.

Höhle, T. (1996). 'The W-…W-Construction: Appositive or Scope Indicating.' In U.

Lutz, G. Müller & A. von Stechow (eds.) Wh-Scope Marking. Linguistik Aktuell 37. Amsterdam: John Benjamins. 37–58.

Horvath, J. (1997). 'The status of 'wh-expletives' and the partial wh-movement

construction of Hungarian.' Natural Language and Linguistic Theory 15.

509–572.

Jacobson, P. (1995). 'On the quantificational force of English free relatives.' In E.

Bach et al. (eds.) Quantification in Natural Languages. Dordrecht: Kluwer. 451–486.

Kahrel, P. (1985). 'Indirect questions and relators.' In A.M. Bolkestein et al. (eds.)

Predicates and terms in functional grammar. Dordrecht: Foris Publications. 165–181.

Keenan, E.L. & B. Comrie (1997). 'Noun phrase accessibility and universal

grammar.' Linguistic Inquiry 8. 63–99.

Kuhn, J. (2001). 'Resource sensitivity in the syntax-semantics interface and the

German split NP construction.' In T. Kiss & D. Meurers (eds.) Constraint-Based Approaches to Germanic Syntax. Stanford: CSLI Publications. 177–215.

Lasnik, H. & M. Saito (1992). Move a. Conditions on its application and output.

Cambridge: MIT Press.

Lehmann, C. (1984). Der Relativsatz. Tübingen: Gunther Narr Verlag.

Lutz, U., G. Müller & A. von Stechow (eds.)(2000). Wh-Scope Marking. Linguistik      Aktuell/Linguistics Today 37. Amsterdam/Philadelphia: John Benjamins.

Maling, J. & A. Zaenen (1978). 'On the non-universality of a surface filter.'

Linguistic Inquiry 9. 475–97.

Müller, G. & W. Sternefeld (1993). 'Improper movement and unambiguous binding.'

         Linguistic Inquiry 24:3. 461–507.

Overdiep, G.S. (1949). Stilistische Grammatica van het Moderne Nederlandsch.

Paardekooper, P.C. (1963). 'Stommeling dat je bent.' De Nieuwe Taalgids 56.

160–164.

—     (1963). Beknopte ABN-syntaxis. Den Bosch: Malmberg.

Pauwels, J. (1958). Het dialect van Aarschot en omstreken. Belgisch Interuniversitair

Centrum voor Neerlandistiek.

Perlmutter, D. (1971). Deep and surface structure constraints in syntax. New York:

Holt, Rinehart and Winston.

Pesetsky, D. (1982). 'Complementizer-Trace Phenomena and the Nominative Island

         Condition.' The Linguistic Review 1. 297–343.

—     (1997). 'Optimality Theory and Syntax: Movement and Pronunciation.'

In D. Archangeli & D. Langendoen (eds.) Optimality Theory, An overview. Oxford: Blackwell.

Reis, M. & I. Rosengren (1992). 'What do Wh-imperatives tell us about Wh-

movement. Natural Language and Linguistic Theory 10. 79–118.

Richards, N. W. (1997) What moves where when in which language. Diss. MIT.

Rizzi, L. (1990). Relativized Minimality. LI Monograph 16. Cambridge: MIT Press.

Ross, J.R. (1967). Constraints on variable in syntax. Diss. MIT.

Smits, R. (1988). The relative and cleft constructions of the Germanic and Romance

languages. Dordrecht: Foris.

Sternefeld, W. (2002). 'Wh-Expletives and Partial Wh-Movement: Two

Non-Existing Concepts?' In W. Abraham & C.J.-W. Zwart (eds.) Issues in Formal German(ic) Typology. Linguistik Aktuel. Amsterdam/Philadelphia: John Benjamins. 285–305.

Tappe, H.T. (1984). 'On infinitival clauses without COMP.' In W. de Geest & Y.

Putseys (eds.) Sentential complementation. Dordrecht/Cinnaminson: Foris Publications. 227–236.

Vanacker, V.F. (1970). 'Een paar punten uit een vergelijkend syntaxisonderzoek van

enkele Zuidnederlandse dialecten.' In [geen ed.] Zijn akker is de taal. Faculteitenreeks 11. Den Haag: Bert Bakker. 291–303.

Van Calcar, W. (1973). 'Het voegwoord 'of'.' Spektator 3. 95–113.

Van Coetsem, F. (1960). 'De Friese relativa en conjuncties met t.' In J.H. Brouwer &

K. Heeroma (eds.) Fryske Studzjes oanbean oan Prof. Dr. J.H. Brouwer op syn sechstichste jierdei . Assen: Van Gorcum & Comp. N.V. 327–334.

Vandekerckhove, R. (2003). 'De relativa die en dat in het dialect en de standaardtaal

van West-Vlamingen.'

Van den Berg, B. (1963). Onderzoekingen betreffende de zinsbouw in het

Nederlands. Den Haag.

Van den Toorn, M.C. (1977). Nederlandse grammatica. Groningen.

Van den Toorn, M.C. et al. (eds.)(1997). Geschiedenis van de Nederlandse Taal.

Amsterdam: Amsterdam University Press.

Van der Auwera, J. (1984). 'Subject vs. non-subject asymmetries in the relativization

of embedded NP's.' In W. de Geest & Y. Putseys (eds.) Sentential complementation. Dordrecht/Cinnaminson: Foris Publications. 257–269.

—     (1985). 'Relative that: A centennial dispute.' Journal of Linguistics 21.

         149–197.

Van der Horst, J.M. (1988). 'Over relatief dat en wat.' De nieuwe taalgids 81.

194–205.

Van der Horst, J.M. & R. Storm (1991). 'Over de geschiedenis van het betrekkelijk

voornaamwoordelijk bijwoord.' Tijdschrift voor Nederlandse Taal- en Letterkunde 107. 105–119.

Van der Woude, G. (1960). 'Oer it gebrûk fan ‘t by bynwurden.' In J.H. Brouwer &

K. Heeroma (eds.) Fryske Studzjes oanbean oan Prof. Dr. J.H. Brouwer op syn sechstichste jierdei. Assen: Van Gorcum & Comp. N.V. 335–343.

Van der Sijs, N. (2004). Taal als mensenwerk. Het ontstaan van het ABN. Den Haag:

SDU Uitgevers.

Van Hoof, H. (1997). 'Left dislocation and split topicalization in Brabant Dutch.' In E.

          Anagnostopoulou, H. van Riemsdijk & F. Zwarts (eds.) Materials on Left

Dislocation. Linguistik aktuell/Linguistics Today 14. Amsterdam: John Benjamins. 275–305.

Van Langendonck, W. (1995). 'De funktie van relativa in een Zuidbrabants dialect.'

In J. Cajot, L. Kremer & H. Niebaum (eds.) Lingua theodisca: Beiträge zur Sprach- und Literaturwissenschaft: Jan Goossens zum 65. Geburtstag. Münster: Zentrum für Niederlande Studien. 433–440.

Van Riemsdijk, H. (1978). A case study in syntactic markedness. Lisse: Peter de

Ridder Press.

—     (1983). 'Correspondence Effects and the Empty Category Principle.' In Y. Otsu

& H. van Riemsdijk (eds.) Studies in Generative Grammar and Language Acquisition: a Report on Recent Trends in Linguistics. Tokyo: Monbusho Grant for Scientific Research. 5–16.

Verwijs, E. & J. Verdam (1889). Middelnederlandsch Woordenboek (2). Den Haag:

Martinus Nijhoff.

Von Stechow, A. (2000). 'Partial Wh-movement, Scope Marking, and Transparent

Logical Form.' In U. Lutz, G. Müller & A. von Stechow (eds.) Wh-Scope Marking. Linguistik Aktuell 37. Amsterdam: John Benjamins. 409–446.

Zwart, C. J-W. (2000). 'A head raising analysis of relative clauses in Dutch.' In A.

         Alexiadou et al (eds.) The syntax of relative clauses. Amsterdam/Philadelphia:

John Benjamins. 349–385.