| Home | Skriuwsels | taalkunde | Impressum |
| changes Changes in the Frisian verbal complex
Henk Wolf
0 Introduction
The linguistic situation in the Dutch province of Frisia (Fryslân, Friesland) has changed considerably over the past decades. In the first decades of the century Frisia was mainly diglossic, Frisian being the spoken language of almost the entire rural population, and Dutch the language for writing and education. These days, Dutch too has become a language spoken daily by a substantial part of the people. An important cause of that is the immigration of people from outside the province to Frisia in the sixties and seventies, also because of relatively low housing expenses at the time. A second cause is the fact that many people from the traditionally non-Frisian speaking cities moved to the countryside. The linguistic situation amongst the speakers of Frisian themselves has changed too. Partly as a consequence of the rise of the mass media and of children spending more years at school, their contacts with Dutch have become much more intensive. The result is that practically all speakers of Frisian are now bilinguals. Besides, they are often expected to make active use of Dutch, even within Frisia. So, in Frisia bilingualism has come up both at the personal level and at the level of the community. Speakers of Frisian do not only come into contact with speakers of Dutch frequently, they are also part of the Dutch speaking community themselves. A result of this intensified language contact is that Frisian is not only used less often, but also that its internal structure is undergoing a number of far-reaching changes, especially within the youngest generation of speakers. Those changes are taking place at all levels of grammar. This new variety of Frisian is mostly referred to as Dutchified Frisian or Interference Frisian (IF). I will be using the latter term from now on, as opposed to the traditional variant that I will be referring to as Standard Frisian (SF). Frisian syntax too has undergone a number of striking changes during the last decades. Especially the changes that have taken place in the verbal complex have drawn attention in syntactic literature. An inventory of syntactic (im)possibilities in IF is lacking, though. This article contains a description of a survey conducted in the autumn and winter of 1994/'95, in which I have attempted to make a first step towards such an inventory. First I will give a description of the results of this survey and then I will try to account for the constructions I have found from both a diachronic and a synchronic perspective.
The verbal complex
The Continental West Germanic SOV-languages are characterized by the formation of sentence-final verbal complexes. This generalization covers a fairly wide range of differences between the separate languages, though. In the following sections I will give a description of four syntactic differences between SF and Dutch. These four `points of attention' have served as the base of this survey.
Inversion
A first important difference between SF and Dutch is demonstrated in (1) and (2).
(1) a Pyt kin dat bêst dien ha
p can that well done have
b * Pyt kin dat bêst ha dien
p can that well have done
(2) a Piet kan dat best gedaan hebben
p can that well done have
b Piet kan dat best hebben gedaan
p can that well have done
(1) shows that in the Frisian verbal complex there is only one word order possible. In an SOV analysis that is the underlying order. In other words: the complement precedes the head. In Dutch, inversion is always possible, and sometimes even obligatory, as is shown in (3).
(3) a * Ik kan de acteur spelen zien
i can the actor play see
b Ik kan de acteur zien spelen
i can the actor see play
In the literature it is usually assumed that IF has also got inversion, but it is unclear when exactly it occurs and whether it is obligatory in those cases. Therefore the possibility of inversion is one of the points of attention in the survey.
Inversion with a to-infinitive
In Dutch inversion is obligatory when the highest verb is a so-called to-infinitive (Frisian and Dutch: te). In SF here too inversion is impossible. The element te directly precedes its complement in the verbal complex, as illustrated in (4) and (5).
(4) a Ik gean nei it teater om de akteur spyljen te sjeni go to the theatre for the actor play to see b * Ik gean nei it teater om de akteur te sjen spyljeni go to the theatre for the actor to see play
(5) a * Ik ga naar het theater om de acteur spelen te zieni go to the theatre for the actor play to see b Ik ga naar het theater om de acteur te zien speleni go to the theatre for the actor to see play
I will also look at the behaviour of to-infinitives in this type of verbal complexes in IF.
The IPP effect
Where a past participle is expected in the complement of an auxiliary such as hebben, an infinitive appears in the Dutch verbal complex. That phenomenon is known as the Infinitivus pro Participio effect (IPP). This phenomenon is absent in SF, where hawwe always takes a participial complement. (6) demonstrates the absence of IPP in SF. (7) illustrates its occurence in Dutch.
(6) a Sy hie Sinteklaas út 'e skoarstien kliuwen sjoenshe had s out the chimney climb seen-PP b * Sy hie Sinteklaas út 'e skoarstien kliuwen sjenshe had s out the chimney climb see-INF
(7) a * Zij had Sinterklaas uit de schoorsteen gezien klimmenshe had s out the chimney seen-PP climb b Zij had Sinterklaas uit de schoorsteen zien klimmenshe had s out the chimney see-INF climb
To my knowledge very little has been written on the presence or absence of IPP in IF. That is why I have adopted this phenomenon in the survey as well.
A distinction between the doelfoarm and the nammefoarm
Frisian has two types of infinitives, the doelfoarm, which ends in /-@n/ (written as -en) and the nammefoarm, which ends in /-@/ (written as -e). In SF those two infinitives are in complementary distribution. Doelfoarms occur, amongst others, after te and in the complement of perception verbs, such as sjen (see), and in the complement of bliuwe (remain). Nammefoarms occur in most other positions, e.g. in the complement of litte (let). (8) and (9) provide two examples.
(8) a Ik kin har dêr rinnen sjeni can her there walk-DF see b * Ik kin har dêr rinne sjeni can her there walk-NF see
(9) a * Ik sil har mar rêstich lêzen littei will her just calmly read-DF let b Ik sil har mar rêstich lêze littei will her just calmly read-NF let
In Dutch both infinitives are used as well, but here they can freely alternate. That is demonstrated in (11) and (12).
(11) a Ik kan haar daar zien lopen i can her there see walk b Ik kan haar daar zien lopei can her there see walk
(12) a Ik zal haar maar rustig laten lezen i will her just calmly let read b Ik zal haar maar rustig laten lezei will her just calmly let read
From now on I will describe the doelfoarm as an infinitive that bears the feature <+n> and the nammefoarm as an infinitive with the feature <-n>. For Dutch this feature is not relevant. Therefore I will describe the Dutch infinitive as unspecified for this feature, or <øn>.
Literature on the verbal complex in IF
There is some literature on inversion in IF, but in general few attempts have been made to list the properties of the verbal complex in IF. Important first steps were made by Eising, Koopmans & Wieberdink (1981) and Ytsma (1995). Both studies are mainly engaged with inversion. The number of studies dealing with a syntactical account of the properties of IF is also very small. IF is mentioned in Abraham (1994). The most interesting publication so far is an article by De Haan (1995), in which he proposes an account for a number of IF complexes.
Different visions on the origin of IF
Feitsma (1971) discusses a number of morphological properties of an IF variety in the early seventies. An important condition in her reasoning is the personal bilingualism that is characteristic of (practically) all Frisians. She claims that in Frisian there is an internal shift going on towards forms that are closer to Dutch. That does not constitute a problem for comprehension, as all Frisians also have a certain proficiency in Dutch. In this way changes in the formation of diminutives can be accounted for. Feitsma contends that bilingual speakers tend towards the creation of a 1-to-1 relationship between morphemes in both languages. Because of the relatively great resemblances between the two languages that is often easy to establish. In Sjölin (1976) personal bilingualism does also play a major role. He demonstrates that the resemblances between Frisian and Dutch are so prominent that Frisians can make use of so-called conversion rules. By applying these rules speakers of Frisian can `frisicize' Dutch morphemes and so form new Frisian words, such as oandacht, a direct frisification of Dutch aandacht (attention). Such a strategy would not only be applied when there is no Frisian word available, but also when the concerning Frisian word is considered `puristic' or `literary', which is the case for relatively many Frisian words, or when the word has strong emotional connotations. Similar strategies can be found amongst other groups of bilinguals too. An important remark made by Sjölin in connection with this is the following:
[...] een volledige competence in het Fries wordt [...] normaliter verkregen via het Nederlands. a full competence in Frisian is normally acquired through Dutch
So, what Sjölin says is that Frisian is formed partly on the base of Dutch language input. De Haan (1995) too brings forward the idea that IF is the result of `mixed input'. From this point of view the bilingual child in the early language acquisition period uses both Frisian and Dutch language input for the formation of a Frisian grammatical system, as a result of which certain properties of Dutch syntax are adopted in Frisian. That is not a very odd scenario if we take into account that many researchers have noted that young bilinguals pass a phase in which they cannot yet separate the two grammars. It has also been noted that the child's notion of the existence of two different languages is a factor in the determination of when the child can separate both grammatical systems. The relative proximity of Dutch and Frisian makes it likely that for the Frisian-Dutch bilingual this moment comes rather late, which leaves a long period in which `mixed input' is likely. A different point of view is taken in De Haan (1990, 1992), in which the phenomenon of inversion in IF is discussed. De Haan claims here that there is no such thing as Dutchification of Frisian. As the non-inverted (SF) verb order would be computationally more complex than the Dutch order, the bilingual child would switch to Dutch grammar for the formation of Frisian verbal complexes. This scenario could be called `mixed output'.
Rule extension or full change?
Breuker (1984a) claims that IF, at least as far as inversion is concerned, is not the result of rule extension, but of a change of rules. That is an important difference, for if we are dealing with rule extension, the existing syntactical system would remain intact, whereas in the case of a change the old system would be replaced by a new system. According to Breuker, many speakers systematically realize the inverted order. He states:
Yn har harsens sit net (of net mear) de Fryske mar Hollânske regel. In their brains there is no (or no longer) the Frisian, but the Dutch rule
The nature of IF: language or language acquisition stage?
As it is generally assumed that IF is used mainly by young speakers, it could be imagined that IF is not a fully developed language, but that the non-standard constructions that occur in it are language acquisition errors. That option is put forward by Hoekstra (1987), who says:
Dat de jongste groep minder de standertfolchoarder brûkt, kin derop wize dat dy learlingen wat it brûken fan 3 of mear tiidwurden yn ien sin oanbelanget, noch yn in learproses sitte. The youngest group using standard word order less often, might be an indication of these pupils being still in a learning proces, as far as the using of 3 or more verbs is concerned
If that is so, that would imply a whole different way of looking at IF, namely as a language acquisition stage. I will also take that possibility into account in this survey.
Objective of the survey
The main objective of the survey is to make an inventory of possible verbal complexes in IF. Subsequently I will try to compare the results of the survey to the ideas in the literature described above. Finally I will make a proposal for a syntactical model for the verbal complex in Continental West Germanic that can also account for the IF complexes encountered.
The survey
In the following sections I will give a description of the development of the tasks used for this survey en of the way they were used in the field work.
The types of verbal complexes tested
The four points of attention in section 1.1. through 1.4 have constituted the base for the tasks. I have only taken into account verbal complexes that exist of two uninflected verbs. Longer complexes tend to be unnatural in Frisian and would be difficult to elicit. In practice I have used only main clauses, in which the inflected verb does not make part of the final complex as a result of the V2 effect, and om ... te ... (for ... to ...) clauses. Those types of sentences are quite common and it is not necessary to test other, less common sentence types in this survey. Taking these conditions into account I have tested the IF translation equivalents of ten types of verbal complexes that contain one or more of the four points of attention and differ in SF and Dutch. Those are listed in section 5. Of course I will not claim to have covered the whole range of types of complexes, but the ones tested here will do for getting a good impression of the phenomena covered by the points of attention.
Method
Following the objectives listed in section 3, three aspects must be taken into account. First, sufficient data is needed for making an inventory. Secondly, indications must be found for deciding on the ideas in the literature on the diachronic development of IF. Third, the results should be fitted into a synchronic syntactical model. That tripartition is reflected in the tests used. The test material consists of five tasks, developed by Jehannes Ytsma and myself. In the following sections I will briefly discuss each task. The complete tasks are given in an appendix.
mF'94
In the first place the survey had to provide an inventory of the verbal complexes possible in IF. To reach that objective we have developed an elicitation task. In order to prevent that the respondents would have to ponder over the constructions being used and make mistakes as a result of that, I have tried to elicit utterances that are as close as possible to spontaneous speech. Transcripts of children's interaction have shown that completely sponateneous speech would yield too little verbal complexes. I found it an acceptable compromise to use a task existing of sentences that lack the sentence final verbal complex. The respondents had to complete the sentences, with the instruction that if possible they should use more than one word. To approach the ideal op spontaneous speech as closely as possible, this task had to be an oral one. An example of a sentence that was used is given in (13).
(13) Hasto in min sifer op dyn rippetysje? Dan hiest dyn rippetysje ek mar better ... have-you a bad mark on your exam? then had-you your exam PRT just better
In SF one would expect a completion like tariede moatten (prepare must-PP). Each sentence was constructed in such a way that it should elicit one of the ten types of complexes mentioned above. For practical reasons the number of sentences had to be limited. We have adopted between one and four sentences per complex type. To distract the respondents' attention from the verbal complexes we have also adopted six so_called distractors; sentences that could not be completed by using verbs. A number of preliminary versions of the task were tested in practice to check its usability. It showed that it was very difficult to elicit certain verbal complexes with to-infinitives. Therefore these types were not adopted in the task. To get at least an impression of these types, be it from a less spontaneous task, we have developed a (also oral) grammaticality judgement task. Because this task needed an inventory instead of supplying one, we had to resort to adopting those complexes that we expected to be possible on the ground of the existing literature. One type of complex occured in both tasks. It showed from the completion task that certain complexes were used that we had not anticipated. As we considered them too interesting to leave them out, they were added to the grammaticality judgement task after the second session. I will indicate where necessary that the number of respondents is smaller for these sentences. Four distractors were added to this task. To create some variation in the sentences other than in the verbal complex we have created some syntactically irrelevant lexical variation. As an example, three sentences that were used in this task are given in (14).
(14) c We kinne wol even by it kanaal te fiskjen gean we can PRT PRT at the canal to fish go b We kinne wol even by de sleat gean fiskjen we can PRT PRT at the ditch go fish c We kinne wol even by de see gean fiskje we can PRT PRT at the sea go fish
Henceforth I will be referring to the two oral tasks discussed here as mF'94. I will be referring to the completion task as mF'94-a and to the grammaticality judgement task as mF'94-b.
mN'94
We considered it interesting to see if Dutch would be affected by Frisian as well as the other way around. Also, if we find non-standard constructions in Dutch, that would provide evidence for the `mixed input' scenario. Therefore we have created a Dutch language task (hereafter mN'94), that is very similar to mF'94. The sentences in the two tasks are syntactically comparable. A few syntactically irrelevant lexical changes were made in order to prevent possible identification with the Frisian equivalent. The distractors are different from the ones used in mF'94. As in mF'94, the unexpected constructions with a to-infinitive have not been adopted until the third session.
sF'94
Another option from the literature concerns the possibility mentioned by Hoekstra (1987) of IF not being a fully developed language, but rather a stage in language acquisition, perhaps specific for bilingual children. In 1990 Jehannes Ytsma has used a written task with 102 school children in the eighth (and fifth) grade, corresponding to about age twelve. That task, hereafter sF'90, was a sentence completion task. The respondent had to choose two verbs from a given total of four, as demonstrated in (15).
(15) kinne - kommen - komme - kinnen Do soest ek wol op myn feestje can-NF come-DF come-NF can-DF you would also PRT on my party
I have used this task again for this survey (as sF'94). The respondents to this ('94/'95) survey were taken from the group that had also cooperated in Ytsma's 1990 survey, in order to make a real-time comparison. If Hoekstra's option is correct, we would expect the respondents to use significantly more SF constructions with sF'94, than they did with sF'90, since they were four years further in their language development.
Selection of the respondents
For practical reasons I have limited the number of respondents to 20. A good 20 addresses were selected from Ytsma's 1990 respondents. Ytsma distinguished three types of schools, dependent on the percentage of Frisian speaking pupils. For practical reasons I did not take any addresses from the schools with the smallest number of Frisian speaking pupils. Ytsma (1995) demonstrates that these three groups do not show any significant difference in language use, so that choice should not constitute a problem. Apart from this, the selection was made at random. The selected group was first requested to cooperate in a letter. A few days later I made appointments with 21 of them. One respondent had such great difficulties with the task, that I decided not to use his data. That brings the number of respondents to 20. The respondents were not paid for their cooperation.
The sessions
Apart from the one respondent who was excluded from the survey, there were no serious problems during the sessions. It did turn out that not all respondents knew the words farsk and de mar. Those problems were solved during the session and have been replaced later by fers and it kanaal. In all cases it was announced that there were five tasks. A brief description was given of every task. Before the start of each task, it was explained once more. The intention of the survey was not mentioned before the task was completed. The sentences in mF'94-a and mN'94-a were propounded in random order, but it was made sure that the distractors were used equably throughout the task. During the grammaticality judgement tasks, the order was fixed (see the appendix). The order in which the tasks were presented was also fixed: mF'94-a, mF'94-b, mN'94-a, mN'94-b, sF'94. The language used for the instructions was not fixed. It depended on the situation. In one case Dutch was used for the instruction, in a few cases the language of the task was used for the instruction with that task, but during most of the sessions, only Frisian was used. All sessions took between ten and 30 minutes. During all the sessions, I was the only interviewer.
The corpus
In this section I will first account for the use of the tasks. Thereafter I will present the outcomes of the survey.
Validity
I will assume the tasks used to be valid. An indication for the validity of mF'94 and mF'94 was provided by Ytsma (1995). These two tasks contained four test sentences that were syntactically comparable. Those are compared for inversion in table 1. The numbers in brackets correspond to the type numbers used in the appendix.
Table 1: Percentage of non-inverted complexes in sF'94 and mF'94 (n=20)
read: 65% of the respondents used non-inverted word order with sentence 1 in sF'94-a
sF'94-a mF'94
65 (1) 70 (#3ba) 65 (3) 65 (#9a) 70 (5) 65 (#2a) 74 (7) 80 (#8a)gem. 71 67
Table 1 shows that the tasks sF'94 and mF'94 predict comparable percentages for inversion. This gives us an indication for the validity of these two tasks for the phenomenon of inversion. The other points of attention were more difficult to compare.
Inventory
In the following subsections I will discuss the outcomes of the three Frisian language tasks, mF'94-a, mF'94-b and sF'94, as given in table 2, 3 and 4.
Table 2: Verbal complexes in mF'94-a (n=20)
read: 50% of the verbal complexes in 1a-d had the construction past participle-auxiliary
1 SF: [[ V<+pp> ] AUXV<-n> ] Du: [[ V<+pp> ] AUXV<øn> ] or [ AUXV<øn> [ V<+pp> ]] (optional inversion)
1 = V<+pp> AUXV<-n> 50 2 = AUXV<-n> V<-pp> 1 0 = other 49
a ... dan soed er no sechstich jier
then would he now sixty years
1 e.g. west ha (been-PP have-NF) 30
2 e.g. wêze wurden (be-NF become-PP) 5 0 65
b ... soe ik no fan skoalle ôf
would i now from school away
1 e.g. west ha (been-PP have-NF) 5 0 95
c ... soe ik beslist net sa folle fan dy pudding would i certainly not so much of that pudding
1 e.g. iten ha (eaten-PP have-NF) 70 0 30
d De smoarge kopkes en pantsjes moatte noch wol
the dirty cups and sauces must still PRT
1 e.g. ôfwosken wurde (washed-up-PP be-NF) 95 0 5
2 SF: [[ V<-n> ] MODV<+pp> ] Du: [ MODV<øn> [ V<øn> ]] (obligatory inversion and IPP)
1 = V<-n> MODV<+pp> 68 2 = MODV<-n> V<øn> 23 0 = other 8
a Dan hiest dyn rippetysje ek mar better
then had-you your test PRT just better
1 e.g. leare moatten (learn-NF must-PP) 70
2 e.g. moatte leare (must-NF learn-NF) 30
b ... dy hie sels in professor net
that had even a professor not
1 e.g. meitsje kind (make-NF could-PP) 70
2 e.g. kinne meitsje (can-NF make-NF) 20 0 10
c Einliks hie de toanielspiler altyd leaver in clown
really had the actor always rather a clown
1 e.g. wêze wold (be-NF wanted-PP) 65
2 e.g. wolle wêze (want-NF be-NF) 20 0 15
3ASF: [[ V<-n> ] MODV<-n> ] Du: [[ V<øn> ] MODV<øn> ] or [ MODV<øn> [ V<øn> ]] (optional inversion) 1 = V<-n> MODV<-n> 83 2 = MODV<-n> V<-n> 15 0 = other 3
a Ik soe dat puzeltsje bêst
i would that puzzle well
1 e.g. dwaan wolle (do-NF want-NF) 90
2 e.g. wolle dwaan (want-NF do-NF) 10
b ... neffens my moatst der toch ek wol mei tsjin in bal according-to me must-you it PRT also PRT with against a ball
1 e.g. skoppe kinne (kick-NF can-NF) 75
2 e.g. kinne skoppe (can-NF kick-NF) 20 0 5
3BSF: [[ V<-n> ] litte<-n> ] Du: [ laten<øn> [ V<øn> ]] (obligatory inversion)
1 = V<-n> litte<-n> 65 2 = litte<-n> V<-n> 33 3 = V<+n> litte<-n> 3
a Us heit soe ús hûs noait troch sa'n djoere skilder
our father would our house never by such-a expensive painter
1 e.g. fervje litte (paint-NF let-NF) 65
2 e.g. litte fervje (let-NF paint-NF) 35
b ... dat hy wol it net troch ien fan de bern that he wants in not by one of the children
1 e.g. dwaan litte (do-NF let-NF) 65
2 e.g. litte dwaan (let-NF do-NF) 30
3 e.g. fervjen litte (paint-DF let-NF) 5
7 SF: [[ V<-n> ] te litten<+n> ] Du: [ te laten<øn> [ V<øn> ]] (obligatory inversion)
1 = V<-n> te litten 28 2 = te litte V<-n> 25 6 = te V<-n> litten 3 7 = te V<+n> litten 3 9 = te V<+n> litte 3 10 = te litte V<+n> 35 11 = te V<+n> litten -> te litte V<-n> 3 0 = other 3
a dat hy gong nei de fytsmakker om syn bân
so he went to the bike-repairman for his tyre
1 e.g. plakke te litten (mend-NF to let-DF) 30
2 e.g. te litte plakke (to let-NF mend-NF) 25
7 e.g. te plakken litten (to mend-DF let-DF) 5
10 e.g. te litte plakken (to let-NF mend-DF) 35 0 5
b Nynke joech har treinkaartsje oan de kondukteur om it troch him
n gave her ticket to the guard for it by him
1 e.g. knippe te litten (clip-NF to let-DF) 25
2 e.g. te litte knippe (to let-NF clip-NF) 25
6 e.g. te knippe litten (to clip-NF let-DF) 5
9 e.g. te knippen litte (to clip-DF let-NF) 5
10 e.g. te litte knippen (to let-NF clip-DF) 35 11 (te litte knippe (self-correction)) 5
8 SF: [[ V<+n> ] PERCV<+pp> ] Du: [ PERCV<øn> [ V<øn> ]] (obligatory inversion and IPP) 1 V<+n> PERCV<+pp> 59 2 PERCV<-n> V<-n> 3 3 PERCV<-n> V<+n> 17 4 PERCV<+pp> V<+n> 12 6 V<-n> PERCV<+pp> 1 0 other 9
a Ik ha ús pake noch noait sûnder bril in krante
i have our grandfather yet never without spectacles a newspaper
1 e.g. lêzen sjoen (read-DF seen-PP) 74
2 e.g. sjen lêze (see-NF read-NF) 5
3 e.g. sjen lêzen (see-NF read-DF) 11
4 e.g. sjoen lêzen (seen-PP read-DF) 11
b ... mar ik haw him noch noait foar clown but i have him yet never as clown
1 e.g. spyljen sjoen (play-DF seen-PP) 60
3 e.g. sjen spyljen (see-NF play-DF) 10
4 e.g. sjoen spyljen (seen-PP play-DF) 10
6 e.g. spylje sjoen (play-NF seen-PP) 5 0 15
c Dy sangeres haw ik dat ferske al faker
that singer have i that song already more-often
1 e.g. sjongen heard (sing-DF heard-PP) 55
2 e.g. hearre sjonge (hear-NF sing-NF) 5
3 e.g. hearre sjongen (hear-NF sing-DF) 25 0 15
d ... mar ik haw him noch noait in sigaar but i have him yet never a cigar
1 e.g. roken sjoen (smoke-DF seen-PP) 47
3 e.g. sjen roken (see-NF smoke-DF) 21
4 e.g. sjoen roken (seen-PP smoke-DF) 26 0 5
9 SF: [[ V<+n> ] PERCV<-n> ] Du: [ PERCV<øn> [ V<øn> ]] (obligatory inversion)
1 = V<+n> PERCV<-n> 60 2 = PERCV<-n> V<-n> 25 3 = PERCV<-n> V<+n> 10 4 = V<-n> PERCV<-n> 5
a ... dat ik koe ús buorlju dúdlik oer polityk
that i could our neighbours clearly about politics
1 e.g. praten hearre (talk-DF hear-NF) 60 2 e.g. hearre prate 25 3 e.g. hearre praten 10 4 e.g. prate hearre 5
Table 3: Percentage of sentences judged as grammatical in mF'94-b (n=20)
read: 95% of the respondents judged the complex `te fiskjen gean' as grammatical in mF'94-b
4 SF: [[ te V<+n> ] V<-n> ] Du: [ V<øn> [ V<øn> ]] (obligatory inversion)
te fiskjen gean (to fish-DF go) 95 gean fiskjen (go fish-DF) 75 gean fiskje (go fish-NF) 60 gean te fiskjen (go to fish-DF) 50 fiskjen gean (fish-DF go) 70 fiskje gean (fish-NF go) 65
5 SF: [[ V<-n>] te learen<+n> ] Du: [ te leren<øn> [ V<øn> ]] (obligatory inversion)
skriuwe te learen (write-NF to learn-DF) 50 te leare skriuwe (to learn-NF write-NF) 65 te learen skriuwe (to learn-DF write-NF) 45
6 SF: [[ te V<+n> ] V<-n> ] Du: [[ te V<øn> ] V<øn> ]
te iten freegje (to eat-DF ask-NF) 95
7 SF: [[ V<-n> ] te litten<+n> ] Du: [ te laten<øn> [ V<øn> ]] (obligatory inversion)
te reparearje litte (to repair-NF let-NF) 39 (n=18) meitsje te litten (fix-NF to let-DF) 65 te litte meitsje (to let-NF fix-NF) 60 te litten meitsje (to let-DF fix-NF) 40 te reparearje litten (to fix-NF let-DF) 28 (n=18)
Table 4: Percentage of complexes used in sF'94 (n=20)
read: 65% of the respondents used the complex `falle litte' with sentence 1 in sF'94
1 Pas op! Do moast dy faas net
(watch out you should that vaze not)
1. falle litte 65 2. litte falle 25 3. falle litten 5 6. litte fallen 5 2 Do soest ek wol op myn feestje
(you would also PRT at my party)
1. komme kinne 50 2. kinne komme 45 3. komme kinnen 5 3 Ik wol Madonna graach ris yn it echt
(i want m much sometime in the real)
1. sjongen hearre 60 2. hearre sjongen 20 6. hearre sjonge 15 7. sjonge hearren 5 4 Dat boek kin dêr wol
(that book can there PRT)
1. lizzen bliuwe 55 2. bliuwe lizzen 15 3. lizzen bliuwen 10 5. lizze bliuwe 10 6. bliuwe lizze 5 7. lizze bliuwen 5 5 Klaas hie syn lytse suske
(k had his little sister)
1. helpe moatten 55 2. moatten helpe 60 3. helpe moatte 10 4. moatte helpe 10 6. moatten helpen 10 8. moatte helpen 10 6 Ho! Dêr hie ik myn bril hast
(oops there had i my spectacles almost)
1. falle litten 60 4. litte falle 10 5. fallen litten 5 6. litten fallen 5 7. fallen litte 5 8. litte fallen 15 7 Ik ha juf yn de stêd
(i have teacher in the town)
1. rinnen sjoen 80 2. sjoen rinnen 10 6. sjoen rinne 10 8 Jan is te lang op bêd
(j is too long in bed)
1. lizzen bleaun 50 2. bleaun lizzen 15 4. bliuwe lizzen 25 6. bleaun lizze 10
Inversion
Table 5 and 6 give the distribution of inverted word order in respectively mF'94-a and sF'94. The numbers in brackets correspond to the type numbers used in the appendix.
Table 5: Percentage of non-inverted and inverted complexes in mF'94-a (n=20)
read: 50% of the complexes of type 1 in mF'94-a had non-inverted word order
non-inverted inverted other
(#f1) 50 1 49 (#f2) 68 23 8 (#f3a) 83 15 3 (#f3b) 68 33 - (#f8) 75 19 9 (#f9) 65 35 - average 68 21 12
Table 6: Percentage of non-inverted and inverted complexes in sF'94 (n=20)
read: 70% of the complexes used with sentence 1 in sF'94 had non-inverted word order
non-inverted inverted
(1) 70 30 (2) 55 45 (3) 65 35 (4) 80 20 (5) 65 35 (6) 70 30 (7) 80 20 (8) 50 50 average 67 83
It shows from both table 5 and 6 that there is no type for which inverted word order is used more often than non-inverted word order. With type 1 in mF'94-a (past participle - auxiliary) there is almost no deviation from the SF norm; only 1% of the complexes has inversion.
It is noteworthy that more inverted complexes are used with type #f3b than with #f3a. The Dutch counterpart of #f3a has obligatory inversion, while non-inverted clusters with #f3b are at best marginal in Dutch. Following Ytsma (1995) and Wolf (1995a), this might be accounted for by the assumption that Frisian-Dutch bilinguals have a preference for inverted word order in Dutch when the selecting verb is a modal. This was also attested by Stroop (1970)
Another interesting observation is that the past participles sjoen (seen) and bleaun (remained) can occur to the left of their complement in both mF'94-a (type #f8) and sF'94 (type 7 and 8), whereas heard (heard) never does. The same is true when the selecting verb is realized as an infinitive (the IPP effect), it occurs systematically to the left of its complement.
Inversion with a to-infinitive
The picture looks different for to-infinitives. Whereas the majority of complexes without te in mF'94-a do not deviate from the SF norm, to-infinitives are used in non-inverted complexes in only 28% of the cases. Only 65% of the SF sentences with a to-infinitive are judged as grammatical in mF'94-b (see table 6). It is remarkable that a fair part of the respondents use constructions such as te reparearje litte(n), in which the complement verb is realized between te and the selecting verb. This word order is absent in Dutch and SF. In mF'94-b it is judged as grammatical in 39% and 28% with litte and litten respectively.
The IPP effect
Table 7 and 8 give the occurences of IPP complexes in the relevant types in mF'94-a and sF'94.
Table 7: Percentages of complexes with IPP and with a past participle in mF'94-a (n=20)
read: 68% of the complexes used with type #2 in mF'94-a used a past participle
PART IPP other
(#2) 68 23 8 (#8) 72 19 9 average 70 21 9
Table 8: Percentages of complexes with IPP and with a past participle in sF'94 (n=20)
read: 90% of the complexes used with type 7 in sF'94 used a past participle
PART IPP
(7) 90 10 (8) 75 25 average 83 18
The (SF) participial construction is clearly favoured in both tasks. Furthermore, whenever the IPP effect occurs, we also find inversion. This does not hold the other way around, as we have noticed that the participles sjoen and bleaun can also occur to the left of their complements.
A distinction between the doelfoarm and the nammefoarm
It shows from table 2 that the distribution of the two infinitives differs from the SF distribution. We could wonder if the free distribution (V<øn>) found in Dutch is adopted in IF. Table 2 shows that we never find a doelfoarm in clusters that do not have a doelfoarm somewhere in SF. That is quite remarkable, if you assume the two infinitives to be in free distribution. Yet, it is not always the same verb as in SF that is realized as a doelfoarm. There is one complex, te plakken litten, where we find two doelfoarms, although in SF you would expect only one. Secondly, it is striking that apparently nammefoarms can appear where you would expect a doelfoarm, as for example in te litte plakke. The picture is less clear in sF'94. We find here a small percentage of cases where a doelfoarm is realized where SF has a nammefoarm. As this is a writing task, it is not impossible that Dutch spelling conventions may have interfered.
Different sorts of Interference Frisian?
In this section I will give a brief overview of the coherence between a number of IF constructions that are found in this survey. I will begin with a discussion of the remarkable construction past participle - infinitive, that occurs with #f8 in mF'94-a and 7 and 8 in sF'94. The respondents who used this construction, did not use it consistently. As noticed before, only sjoen and bleaun can appear in this construction. All respondents do also use other constructions than this one, mostly the ones that equal the SF and Dutch possibilities. Some even use all three constructions. As an illustration, (16) shows a number of complexes used by the same respondent:
(16) a Ik ha ús pake noch noait sûnder bril in krante sjoen lêzen (mF'94-a:#f8a) b ... ik haw him noch noait foar clown sjoen spyljen (mF'94-a:#f8b) c ... ik haw him noch noait in sigaar sjoen roken (mF'94-a:#f8d) d Dy sangeres haw ik dat ferske al faker hearre sjongen (mF'94-a:#f8c) e Ik ha juf yn de stêd rinnen sjoen (sF'94:7) f Jan is te lang op bêd bliuwe lizzen (sF'94:8)
This respondent appears to prefer the construction past participle - infinitive in mF'94-a, but not in sF'94. There is no general picture, though. (17) gives a number of sentences used by one (other) respondent, who uses the inverted past participle only in sF'94.
(17) a Ik ha ús pake noch noait sûnder bril in krante lêzen sjoen (mF'94-a:#f8a) b ... ik haw him noch noait foar clown spyljen sjoen (mF'94-a:#f8b) c ... ik haw him noch noait in sigaar roken sjoen (mF'94-a:#f8d) d Dy sangeres haw ik dat ferske al faker sjongen heard (mF'94-a:#f8c) e Ik ha juf yn de stêd sjoen rinne (sF'94:7) f Jan is te lang op bêd bleaun lizze (sF'94:8)
This respondent uses the SF construction throughout mF'94-a, but not in sF'94. There, we find the construction past participle - infinitive. In this case the infinitive is a nammefoarm, and not a doelfoarm, as it would be in SF. There are nine respondents who use the construction past participle - infinitive at least once. 18 of the 20 respondents use the SF construction (not considering the form of the infinitive). Some respondents, who do use the construction with IPP and inversion, do not use the past participle - infinitive construction. Another remarkable construction is te - infinitive - litte(n), as in (18).
(18) Nynke joech har treinkaartsje oan de kondukteur om it troch him te knippe litten (mF'94-a:#f7b)
This construction is not very frequent. It was used three times in the survey. There was also one respondent who first used the complex te knippen litten and then corrected himself by saying te litte knippe. The use of the proper infinitive looks totally unmotivated in this construction. The four respondents who use this construction, do not use it consistently. They also use the Dutch or SF constructions. One respondent, who used the construction in (18), also used (19):
(19) ... hy gong nei de fytsmakker om syn fyts meitsje te litten (mF'94-a:#f7b)
The construction te litten meitsje was not found in mF'94-a, but 40% of the respondents judged it as correct in mF'94-b.
The Dutch language tasks
Table 9 and 10 provide the outcomes of the Dutch language tasks mN'94-a and mN'94-b. The numbers correspond to the numbers used in table 2.
Table 9: Verbal complexes in mN'94-a (n=20)
read: 22% of the verbal complexes in 1a-c had the construction past participle - auxiliary (n=20)
1 1 = [[ V<+pp> ] AUXV<øn> ] 22 2 = [ AUXV<øn> [ V<+pp> ]] 48 0 = other 34
a ... dan zou ze nu zestig jaar then would she now sixty year
1 e.g. geweest zijn (been-PP have-INF) 20
2 e.g. zijn geweest (have-INF been-PP) 50 0 30
b ... zou ik nu van school af
would i now from school away
1 e.g. geweest zijn (been-PP have-INF) 15
2 e.g. zijn geweest (have-INF been-PP) 30 0 55
c ... zou ik nooit van die limonade
would i never of that lemonade
1 e.g. gedronken hebben (drunk-PP have-INF) 21
2 e.g. hebben gedronken (have-INF drunk-PP) 63 0 16
2 1 = [ MODV<øn> [ V<øn> ]] 95
0 = other 5
a Dan had je je huiswerk maar beter
then had you your homework just better
1 e.g. moeten leren (must-INF learn-INF) 100
b Dat had zelfs de slimste jongen van de klas niet that had even the smartest boy of the class not
1 e.g. kunnen maken (can-INF make-INF) 95 0 5
c Eigenlijk had de clown altijd liever een akrobaat
really had the clown always rather an acrobat
1 e.g. willen zijn (want-INF be-INF) 90 0 10
3A 1 = [[ V<øn> ] MODV<øn> ] 3 2 = [ MODV<øn> [ V<øn> ]] 90 3 = as 2 (corrected after using 1) 3 0 = other 5
a Ik zou dat puzzeltje best
i would that puzzle well
2 e.g. willen doen (want-INF do-INF) 90
b ... volgens mij moet je er toch ook wel mee tegen een bal according-to me must you it PRT also PRT with against a ball
1 e.g. schoppen kunnen (kick-INF can-INF) 5
2 e.g. kunnen schoppen (can-INF kick-INF) 90 3 schoppen kunnen -> kunnen schoppen 5
3BNl:1 = [ laten [ V<øn> ]] 100
a Mijn oom zou zijn auto nooit door die monteur
my uncle would his car never by that mechanic
2 e.g. laten maken (fix-INF let-INF) 100
b ... ze zou het nooit door een van de kinderen
she would it never by one of the children
2 e.g. laten doen (let-INF do-INF) 100
7 -1 = [ te laten [ V<øn> ]] 98 2 = [ te [ V<øn> ] laten ] 3
a ... ging hij naar fietsenmaker om zijn band
went he to the bike-repairman for his tyre
1 e.g. te laten plakken (to let-INF mend-INF) 95
2 e.g. te plakken laten (to mend-INF let-INF) 5
b Fred gaf zijn treinkaartje aan de conducteur om het door hem f gave his ticket to the guard for it by him
1 e.g. te laten knippen (to let-INF clip-INF) 100
8 1 = [ PERCV<øn> [ V<øn> ]] 100
a Ik heb mijn oom nog nooit zonder bril een krant
i have my uncle yet never without spectacles a newspaper
1 e.g. zien lezen (see-INF read-INF) 100
b ... ik heb haar nog nooit de kamer
i have her yet never the room
1 e.g. zien stofzuigen (see-INF vacuum-clean-INF) 100
c Die zanger heb ik dat liedje al eens eerder
that singer have i that song already sometime before
1 e.g. horen zingen (hear-INF sing-INF) 100
d Ik heb hem nog nooit bier
i have him yet never beer
1 e.g. zien drinken (see-INF drink-INF) 100
9 1 = [ PERCV<øn> [ V<øn> ]] 100
a Ik kon de buren heel duidelijk over politiek i could the neighbours very clearly about politics
1 e.g. horen praten (hear-INF talk-INF) 100
Table 10: Percentage of sentences judges as grammatical in mN'94-b (n=20)
read: 45% of the respondents judged the complex `te vissen gaan' as grammatical in mN'94-b
4
te vissen gaan (to fish-INF go-INF) 45
gaan vissen (go-INF fish-INF) 90
gaan te vissen (go-INF to fish-INF) 15
vissen gaan (fish-INF go-INF) 10
5 schrijven te leren (write-INF to learn-INF) -
te leren schrijven (to learn-INF write-INF) 90
6 te eten vragen (to eat-INF ask-INF) 95
7 maken te laten (fix-INF to let-INF) -
te laten maken (to let-INF fix-INF) 90
te repareren laten (to repair-INF let-INF) -
Table 9 and 10 show that there are very few non-standard complexes used or accepted in Dutch. The only non-standard complex in mN'94-a is te plakken laten. Overall, the respondents show strong preference for inverted constructions.
The origin and nature of Interference Frisian
In this section I will go into the questions about the origin and nature of IF, as posed in section 2.
Rule extension or full change?
It shows from table 2 that there is no systematic realization of non-SF complexes, as also discussed in section 5.2.5. Nor can we say that IF is built by using the `rules of Dutch' as Breuker (1984a) proposes. A possible problem is the relatively low acceptance of complexes like meitsje te litten and skriuwe te learen in mF'94-b. Yet, several respondents have indicated that they had other motives for not accepting those complexes, which may explain at least part of the problem. For the majority of respondents, the data leave us to conclude that IF must be considered an extension of the SF grammatical system.
Language or language acquisition stage?
If IF is not a fully developed language, but a language acquisition stage, we would expect the respondents to have passed that stage by the time of the 1994 survey, as their ages were now about 16 years of age. This is clearly not the case. Furthermore, there is hardly any `progress' towards the SF forms. This is shown in table 11 through 13.
Table 11: Percentage of non-inverted complexes in sF
read: 75% of the respondents used a non-inverted complex with sentence 1 in sF'90
sF'90 sF'94 index
(n=20) (n=20) (sF'94 / sF'90) x 100 (1) 75 70 93 (2) 60 55 92 (3) 50 65 130 (4) 65 80 123 (5) 55 65 118 (6) 65 70 108 (7) 60 80 133 (8) 55 50 91 average 61 67 111
Table 12: Percentage of used complexes with a participle in sF
read: 65% of the respondents used a complex with a participle with sentence 5 in sF'90
sF'90 sF'94 index
(n=20) (n=20) (sF'94 / sF'90) x 100 (5) 55 70 127 (6) 75 70 93 (7) 85 90 106 (8) 65 75 115 average 70 76 110
Table 13: Percentage of complexes corresponding to the SF norm
read: 55% of the respondents used a complex that corresponds to the SF norm in both word order and word form with sentence 1 in sF'90
sF'90 sF'94 index
(n=20) (n=20) (sF'94 / sF'90) x 100 (1) 65 65 100 (2) 60 50 83 (3) 35 60 171 (4) 40 55 138 (5) 45 55 122 (6) 50 60 120 (7) 55 80 69 (8) 50 50 100 average 50 59 113
Although with most sentences there is a slight progress towards the SF norm, Ytsma (1995) shows that this is not a significant change. Also, at the age of 16 we expect people to have acquired full proficiency in their mother tongue. Furthermore, the outcomes deviate quite radically from the outcomes of a test used with a sample of the respondents' parents, as used by Jehannes Ytsma in 1992 (sF'92-parents). Compare table 11 to the outcomes of that task in table 14.
Table 14: Percentage of non-inverted complexes in sF'90 and sF'92-parents
read: 70% of the respondents used a non-inverted complex with sentence 1 in sF'90
sF'90 sF'92-parents
(n=102) (n=52) (1) 70 100 (2) 42 96 (3) 50 100 (4) 62 92 (5) 57 94 (6) 50 90 (7) 49 94 (8) 47 94 average 61 67 111
On the basis of these figures I conclude that it is very unlikely that IF is merely a language acquisition stage. That means that it is a fully developed variety. In the following sections I will investigate how this variety could best be described.
Mixed output?
As pointed out in section 2.1, De Haan (1990) mentions the `mixed output' scenario, in which speakers of IF switch to Dutch grammar for producing verbal complexes. This switch would be triggered by the relative perceptive complexity of SF word order as compared to the Dutch system. There are several reasons why this scenario is not desirable. The most direct reason comes forward from the actual data. There are various constructions in IF that cannot be the result of Dutch grammar output, such as sjoen rinnen and te knippe litten. Secondly, there are several differences between Dutch and Frisian verbal morphology. It is unclear how the distribution of cf. the two infinitives in IF could be accounted for in this scenario. Third, it is not clear what would trigger inversion in simple two-word complexes, that allow both inverted and non-inverted word order in Dutch, nor does it provide insight in the diffent inversion properties in the various types of complexes in IF. Finally and perhaps most elementary is the question whether we should assume two underlying syntactical systems for a certain variety at all. It is not clear to me how assuming such an aberrant model only for IF could be accounted for within the existing theory. Perhaps it is possible to apply a `mixed output' model to other varieties than IF. For now I will just state that from a theoretical point of view this model is undesirable.
Dutch
Few can be said about the complexes used in Dutch. Practically no non-standard complexes are found there. It is clear that the influence of Dutch on Frisian is one-way traffic. The strong preference for inverted complexes is noteworthy.
Towards a synchronic syntactical model for IF
Rejection of the above model implies of course that the IF should be described within one syntactical model. The question that remains now is if that is possible. In the following sections I will contend that that is indeed a possibility. I will also show how this model can be derived from the so-called `mixed input' scenario, presented in section 2.1. Furthermore, I will show how this model can serve to account for data from other Continental West Germanic languages.
Inversion
As noted before, the constructions found in IF are not simply Dutch glosses. Inversion occurs in IF constructions without IPP, where this is obligatory in Dutch. That pleads against the `mixed output' scenario, and for considering inversion a separate rule in IF that should be considered on its own. It shows from table 2 that the respondents use both inverted and non-inverted constructions with all types, except for complexes with a participle and an auxiliary. That appears to indicate that IF has an inversion rule that differs from the Dutch inversion rule. De Haan (1990) gives three rules for inversion in Dutch, as in (20).
(20) Inversion in Dutch (informally)
V" V"
===>
V$ V* V* V$
a. optional, if V* is a finite modal, and V$ and V* are non-branching;
b. optional, if V$ is a past participle; c. obligatory elsewhere
With (20), De Haan explains sentences as in (21).
(21) a dat hij het kan doen / doen kan that he it can-3SG do-INF b dat hij het gedaan heeft / heeft gedaanthat he it done-PP has-3SG c Hij heeft haar zien lopen / *lopen zien / lopen *gezienhe has her see-INF walk-INF seen-PP
Those rules cannot account for IF sentences as in (22) (taken from De Haan (1990).
(22) a Hij hat him sjen wold / wolle sjen he has him see-NF wanted-PP want-NF b om in doelpunt meitsje te kinnen / te kinne meitsjefor a goal make-NF to can-DF can-NF
According to De Haan, these sentences cannot be derived using (20), because there is no finite modal (22a) or past participle (22b) in the verbal complex. Therefore he concludes that there cannot be syntactical borrowing from Dutch. If we assume that the inversion rule in Dutch is not bound to the form of the verb, though, the sentences in (22) are no longer a problem. I propose (23) as a replacement of (20).
(23) Inversion in Dutch (informally)
a. always optional
b. in some complexes non-inverted complexes are ruled out by independent factors What these independent factors look like I do not know. Answering that question is beyond the scope of this article. I only want to claim that it is possible to derive optional and obligatory inversion from successive application of one optional inversion rule and a set of other rules, instead of from a more fundamentally different set of non-successively applicable rules. (23a) could be borrowed into Frisian. In that way it is possible to derive the sentences in (22). It looks as if another mechanism rules out inversion of a participle and an auxiliary. Whether those constructions are always ungrammatical (one respondent does use it) and if this extend to other participles as well, is something that needs further investigation.
Inversion with a to-infinitive
There are several complexes in IF with a to-infinitive. In all cases, the verb selected by te is litte (let). (24) lists the complexes encountered in table 2.
(24) a om it troch him knippe te litten b om it troch him te litte knippec om it troch him te litte knippend om it troch him te kontrolearen littee om syn bân te plakken litten
De Haan (1995) has found that (24a-b) occur in IF. He proposes the following model to account for these data.
(25) a
I'
V I
V V I Vi
<-n> <+n>
knippe ei te litten
b
I'
V I
ei
I Vi
te V V <-n>
litte knippe
(25) represents the PF structures of (24a-b) respectively. Now his argumentation is as follows: Verbal complexes are analyzed as V-clusters that are the result of V-raising, as originally proposed by Evers (1975). Subsequently, V is adjoined to I. Morphological features are assigned at PF, as proposed by Den Dikken (1989). Now litten in (25a) receives its doelfoarm morphology from te. The verb knippe receives its nammefoarm morphology from the selecting verb litten. In (25b) te selects a doelfoarm. The <+n> feature is assigned to the complex V. Now the feature goes down (`percolates') into the structure to a node where it can be spelled out. According to the Right-Hand Head Rule (RHHR) it can only percolate to the rightmost node of every branching, as morphological features are spelled out on the head. That means that litte is no longer a possible landing site for the doelfoarm feature. knippe would be, but this verb has already received a <-n> feature from the selecting verb litte. It is assumed that this feature assignment has precedence above feature assignment by the RHHR. This means that the doelfoarm feature is not spelled out. However, this is not compatible with complexes as in (24c), where we find both inversion and doelfoarm morphology on litten. I do not think that that is a problem for the model. The doelfoarm morphology must be accounted for. In this case, te is the only possible assigner of doelfoarm morphology. If we assume equal precedence for both ways of feature assignment, perhaps with personal variation, the data in (24a-c) can be accounted for within this model. The sentences in (9d-e) are not frequent, but since they occur, I will discuss them here. In the present model complexes like these are described as an inverted complex V that is adjoined to I. It is rather easy to account for doelfoarm morphology on the rightmost verb, knippen. As I will argue in section 7.4, nammefoarm morphology on either verb is not problematic to the model. However, there are two occurences of doelfoarm morphology on the left verb in this survey, both uttered by the same respondent. In mF'94-b those are judged as grammatical by respectively 45% and 40% of the respondents. I do not want to exclude the possibility that the pronunciation of final /n/ in mF'94-b is so unconspicuous that it is perceived as a personal peculiarity, rather than an ungrammatical utterance. It is also possible that it is due to the somewhat deviant morphological properties of litte, which, according to De Haan (1992a), behaves as a `syntactically bound morpheme' in Frisian. I have to leave these sentences open as a problem.
The IPP effect
It shows from table 2 that there are three complex types with complexes that have an IPP counterpart in Dutch. Those are given in (26).
(26) a ik haw him noch noait in sigaar roken sjoen b ik haw him noch noait in sigaar sjen rokenc ik haw him noch noait in sigaar sjoen roken
(26a) gives the SF construction, without inversion and without IPP. (26b) corresponds to the Dutch construction with both inversion and IPP. (26c) then, does have inversion, but not IPP.
De Haan (1995) observes the two constructions in (26a) and (26b). De Haan tries to account for both within one model. Again his argumentation is based on the RHHR. His proposal is based on PF structures such as the ones in (27).
(27) a V <+pp>
V V <+n> <+pp>
roken sjoen
b V <+pp>
V V <+n>
sjen roken
In the case of (13a) the verb rinnen receives the <+n> feature for doelfoarm morphology from the selecting verb sjoen. Participial morphology is assigned to the complex V by the selecting verb haw. Because of the RHHR it can percolate to the rightmost node sjoen, where it is spelled out. In (13b) rinnen receives its doelfoarm morphology from sjen. As a result, rinnen is no longer a possible landing site for the <+pp> feature, which is not spelled out. The leftmost node sjen is not assigned a morphological feature. Now De Haan claims that a verb that does not have a morphological feature, is assigned infinitival morphology by default. If we formalize this, we get a set of rules that derives the IPP effect. See (28).
(28) The IPP effect
a. Default-morphological rule: a verb that is not assigned a morphological feature, receives a default-morphological feature F
b. F-form: the feature F is spelled out as a nammefoarm <-n>
With (27) and (28) it is possible to account for (26a) and (26b), but not for (26c). In the following I will make a proposal to include (26c) in the theory, but first I want to point out again the distinction discussed in 5.2.1 between the distribution of the participles sjoen and heard. Whereas we do find (26c), (29) does not occur.
(29) * Dy sangeres haw ik dat ferske al faker heard sjongen
A similar distinction is found in another survey I have conducted amongst a group of 23 primary school children in 1995. Here a grammaticality judgement task was used.
(30) a Ik ha Peter yn de stêd sjoen rinnen i have p in the town seen-PP walk-DF b * Ik ha Lee Towers noch noait heard sjongeni have l.t. yet never heard-PP sing-DF
The IPP equivalent is acceptable in IF, as shown in (31).
(31) a Dy sangeres haw ik dat ferske al faker hearre sjongen b Ik ha Lee Towers noch noait hearre sjongeni have l.t. yet never hear-NF sing-DF
Apparently the participles sjoen and also bleaun may be inverted for many speakers, whereas heard may not. I think this difference is essential for the explanation of (26c). It is noted that heard is a regularly formed participle, that carries the participial suffix -t/-d. This is not the case for sjoen and bleaun that are irregular in that they lack this suffix. My claim is that regularly formed participles are analyzed as morphologically marked, whereas sjoen and bleaun are not. That idea is supported by the fact that a rising number of Frisians uses sjoend and bleaund. This is likely to be a process of regularization of these exceptions to make them morphologically marked as well. It is clearly not simply an extension of the regular participial suffix, as prefixed derivations of sjoen and bleaun, such as besjoen (`looked at') never take the suffix. This indicates that participles can be marked by both suffixes and prefixes. Now if we replace (28) by (32), we can account for all the IF data presented here.
(32) The FPP effect
a. Default-morphological rule: a verb that is not assigned a morphological feature, receives a default-morphological feature F
b. F-form: the feature F is spelled out as a not-morphologically marked verbal form
Both the nammefoarms sjen, bliuwe and hearre, and the participles sjoen and bleaun are F-forms. The regularly formed participle heard is not. Because the default-morphological rule in (32a) allows only F-forms as the leftmost daughter node in (27b), we can account for the judgements given in table 2.
Now how does the FPP effect work with regard to other Continental West Germanic languages? Would a universal claim for this effect hold? In (33-36) I present some data from SF, IF, Low German and Dutch.
SF: (33) a Ik ha juf sjongen heard b * Ik ha juf heard sjongen c * Ik ha juf sjongen hearre d * Ik ha juf hearre sjongen
IF: (34) a Ik ha juf sjongen heard i have teacher sing-DF heard-PP b * Ik ha juf heard sjongen c * Ik ha juf sjongen hearre i have teacher sing-DF hear-NF d Ik ha juf hearre sjongen
LG (35) a Ik hev Clara singen hoyrd i have c sing-INF heard-PP b * Ik hev Clara hoyrd singen c * Ik hev Clara singen hoyren i have c sing-INF hear-INF d * Ik hev Clara hoyren singen
Dutch (36) a * Ik heb juf zingen gehoord i have teacher sing-INF heard-PP b * Ik heb juf gehoord zingen c * Ik heb juf zingen horen i have teacher sing-INF hear-INF d Ik heb juf horen zingen
SF and Low German do not have inversion, so here the FPP effect does not apply. These languages only allow complexes with the order infinitive-participle. Dutch has obligatory inversion. Past participles always have a prefix in Dutch, what makes them morphologically marked. The infinitive is a default-morphological form, just like the nammefoarm in IF. In other words, the IPP effect is really a sub-effect to the FPP effect. That accounts for the data in (36).It is also likely that German haben is strictly left-directional in its assignment of the <+pp> feature. If we assume that in V2 clauses, haben and its complex complement can invert before the V2-movement takes place, haben can no longer assign a feature, as its complement is now to the right. This would also evoke the FPP effect.
Interesting evidence for the FPP effect can also be found in the following Flemish dialect data, presented by Vanden Wyngaerd (1994).
(37) Jan is weest voetballen j is weest play-soccer-INF
In this dialect, the past participle is formed with the ge- prefix, just like in Dutch. The past participle of `to be' is geweest. For independent reasons, this dialect does not have an infinitive that can be used in the verbal complex. Now what we see is that the verbal form weest appears, which is the past participle with the prefix stripped off. This form is no longer morphologically marked, so it is a F-form that can appear in the left position in the complex.
A distinction between the doelfoarm and the nammefoarm
So far I have argued that all differences between IF and SF are triggered by the adoption of optional inversion from Dutch into Frisian. In this section I will argue that one more rule is adopted from Dutch. We have seen in table 2 that, with only a single exception, a doelfoarm does never occur in IF complexes, where the SF counterpart does not also have a doelfoarm. Where SF has a doelfoarm, we frequently find a nammefoarm in IF. My argumentation is based on that generalization, as formulated in (38).
(38) Infinitive generalization for IF a. a doelfoarm can also be realized as a nammefoarm b. a nammefoarm cannot be realized as a doelfoarm
This implies that a surface doelfoarm must always be accounted for from <+n> assignment by another element, as I have done in section 7.2. My claim is that the bilingual child analyses the infinitival situation in Dutch as underlyingly <+n>, with free deletion of the final /n/ in the phonology. That rule could look as in (39).
(39) N-deletion in Dutch: final /n/ can be freely deleted
This is a rule the child could adopt into IF. This implies that the Dutch situation of <øn> is not adopted into IF, i.e. that the underlying distinction <±n> is still present in IF, as expressed by the infinitive generalization in (38). As I have also said in section 7.2, it may not necessarily be the case that a `misplaced' doelfoarm is filtered out in perception, as it is an unconspicuous phonological element. That would account for people accepting complexes such as te learen skriuwe in mF'94-b.
Summary
In this article I have provided a first step towards an inventory of the verbal complex in IF. I have shown that IF grammar is an extension of SF grammar and not a replacement of the SF system. IF is not a language acquisition stage, so I take it to be a fully developed variety. This variety is not the result of `mixed output'. It has its own grammar that can be described with a single syntactical model, if we assume that the bilingual child has used Dutch language input in the formation of its Frisian grammar (the `mixed input' scenario). When we assume that the rules of optional inversion and (phonological) n-deletion are adopted from Dutch, it is possible to account for nearly all of the IF data encountered in this survey by using a set of universal rules.
References
Abraham, W. (1994), `Kaynes Asymmetriehypothese und die Abfolge im V-Komplex', Groninger Arbeiten zur Germanistischen Linguistik 37, p. 19-46. Aitchinson, J. (1991), Language change: progress or decay, Cambridge: Press Syndicate of the University of Cambridge. Appel, R. & P. Muysken (1987), Language contact and bilingualism, London: Edward Arnold. Arnberg, L. & P. Arnberg (1992), `Language awareness and language separation in the young child', in: Harris, R. (ed.), Cognitive processing in bilinguals, Advances in psychology 83, Amsterdam: Elsevier. Besten, H. den & H. Broekhuis (1989), `Woordvolgorde in de werkwoordelijke eindreeks', GLOT 12, p. 79-137. Boelens, K. & J. Ytsma (1989), `Bern en brekking; In ûndersyk nei taalwinning en taalferoaring', It Beaken 51-2, p. 103-114. Booij, C., J. Meinardi & K. Spoelman (1980), Een onderzoek naar veranderingen in het Gronings; De alternantie in het presens enkelvoud en de groene volgorde, Ms., Groningen. Breuker, P. (1984a), `Oer it lienen fan bûne morfemen út it Hollânsk yn it Westerlauwerske Frysk', in: N. Århammer et al (eds.), Miscelannea Frisica; In nije bondel Fryske stúdzjes, Assen: Van Gorcum. Breuker, P. (1984b), Foar de taalspegel; Koart oersjoch fan Hollânske ynslûpsels yn it Frysk, Ljouwert: AFUK. Dikken, M. den (1989), `Verb Projection Raising en de analyse van het IPP-effect', TABU 19-2, p. 59-75. Eising, G., W. Koopmans & M. Wieberdink (1981), Een onderzoek naar interferentie in de vervoeging van het zwakke werkwoord en de volgorde in de bijzin in het Fries van Abbega, Ms., Groningen. Eisma, D. (1989), Tiidwurden, Ljouwert: AFUK. Evers, A. (1975), The transformational cycle in Dutch and German, diss. Utrecht, RUU. Feitsma, A. (1971), `Onderzoek naar tweetaligheid in Friesland', in: A. Feitsma & M. van Overbeke, Tweetaligheidsproblemen, Bijdragen en mededelingen der dialectencommissie van de KNAW, Amsterdam: Noord-Hollandsche Uitgevers Maatschappij. Gorter, D. et al (1984), Taal yn Fryslân, Ljouwert: Fryske Akademy. Grosjean, F. (1982), Life with two languages, Cambridge: Harvard University Press. Haan, G. de (1987), `De syntacticus als frisist', in: S. Dyk. & J. Hoekstra (eds.), Ta de Fryske syntaksis, Ljouwert: Fryske Akademy. Haan, G. de (1990), `Grammatical borrowing and language change: the Dutchification of Frisian', in: D. Gorter et al (eds.), Fourth International Conference on Minority Languages. Vol I: General Papers, Clevedon: Multilingual Matters. Haan, G. de (1992a), `The verbal complex in Frisian', Us Wurk 1992, p. 58-92. Haan, G. de (1992b), Meertaligheid in Friesland, inaugural address Groningen, Dordrecht: ICG-Printing. Haan, G. de (1995), `Recent changes in the verbal complex of Frisian', to appear in NOWELE, Festschrift for N. Århammer. Hoeksema, J. (1988), `A constraint on governors in the West-Germanic verb cluster', in: M. Everaert et al (eds.), Morphology and modularity, Publications in language sciences 29, Dordrecht: Foris. Hoekstra, E. (1994a), `Over de distributie van infinitieven op -E en -EN in het Fries en het Westfries', lecture TABU-dag, 24-6-1994, Groningen. Hoekstra, E. (1994b), Analysing linear asymmetries in the verb clusters of Dutch and Frisian and their dialects, Working Papers in the syntax of West-Germanic dialects, Amsterdam: P.J. Meertens Instituut. Hoekstra, J. (1987), Ik hoop dat se de spigel net sille brekke; In ûndersyk nei taalgebrûk by frysktalige mafû-learlingen, Ms., Harns. Hoekstra, J. (1989), `A note on the typology of te-infinitives in Frisian', NOWELE 14, p. 19-51. Hoekstra, J. (1990), `Trijetiidwurdkonstruksjes', It Beaken 52, p. 59-95. Hoekstra, J. (1992), `Fering tu-infinitives, North Sea Germanic syntax and Universal Grammar', NOWELE suppl. vol. 8, p. 99-142. Jong, S. de & A. Riemersma (1994), Taalpeiling yn Fryslân; Onderzoek naar de beheersing van het Fries en het Nederlands aan het einde van de basisschool, Ljouwert: Fryske Akademy. Kalma, D. (1950), `Sizze, to sizzen, sizzen. (Mienefoarm, ferhâldingsfoarm, tiidwurdfoarm)', De Pompeblêden XXI, p. 83-89. Kayne, R. (1993), The antisymmetry of syntax, Ms., New York: CUNY. Parodi, T. (1990), `Word order and case morphology', in: J. Meisel (ed.), Two first languages; Early grammatical development in bilingual children, Dordrecht: Foris. Popkema, J. (1987), `Inkelde opmerkings oer Fryske en Nederlânske syntaksis', in: S. Dyk & J. Hoekstra (eds.), Ta de Fryske syntaksis, Ljouwert: Fryske Akademy. Reuland, E. (1990), `Infinitieven in het Fries en de aard van functionele categorieën', TTT 9-4, p. 287-309. Sjölin, B. (1976), Min Frysk; een onderzoek naar het ontstaan van transfer en "code-switching" in het gesproken Fries, Bijdragen en mededelingen der dialectencommissie van de KNAW, Amsterdam: Noord-Hollandsche Uitgevers Maatschappij. Stroop, J. (1970) Tiersma, P. (1985), Frisian reference grammar, Dordrecht: Foris. Tiersma, P. (1986), Language change in a bilingual community: some Frisian data, paper presented at Dutch Linguistics Colloquium, 9-11-1985, University of California. Wolf, H. (1995a), Feroarings yn de Fryske tiidwurdlike einrige; In ûndersyk nei in tal resinte syntaktyske feroarings by Frysktalige jongerein, Ms., Ljouwert. Wolf, H. (1995b), `Ynvertearre tiidwurdkloften yn it Ynterferinsjefrysk', Tydskrift foar Fryske Taalkunde 10-1, p. 1-11. Wolf, H. (1995c), `De distributie van de twee infinitieven in het Interferentiefries; Taalveranderingsonderzoek op het kruispunt van syntaxis en sociolinguïstiek', Link 7-3/4, p.30-36. Wyngaerd, G. vanden (1994), `IPP and the structure of participles', Groninger Arbeiten zur Germanistischen Linguistik 37, p. 265-276. Ytsma, J. (1995), Frisian as first and second language; Sociolinguistic and socio-psychological aspects of the acquisition of Frisian among Frisian and Dutch primary school children, diss. Tilburg, Ljouwert: Fryske Akademy. Zwart, J.-W. (1994), Verb clusters in Continental West Germanic dialects, paper presented at 18th Annual Meeting of the Atlantic Provinces Linguistic Association, 29-10-1994, Saint John.
Appendix: the tasks used in the survey
mF'94-a
1a. At Elvis Presley net stoarn wie, dan soed er no sechstich jier ... 1b. At ik tsien jier earder geboaren wêze soe, soe ik no fan skoalle ôf ... 1c. At ik wist dat it my sa swier op 'e mage falle soe, soe ik beslist net sa folle fan dy pudding ... 1d. Even wachtsje jonges! De smoarge kopkes en pantsjes moatte noch wol ... 2a. Hasto in min sifer op dyn rippetysje? Dan hiest dyn rippetysje ek mar better ... 2b. Dy som is sa moeilik, sei master, dy hie sels in professor net ... 2c. Djip yn syn hert hie de toanielspiler einliks altyd leaver in clown ... 3aa. Ik soe dat puzeltsje bêst ... 3ab. Al is sa'n tennisskoech der net foar bedoeld, neffens my moatst der toch ek wol mei tsjin in bal ... 3ba. Us heit soe ús hûs noait troch sa'n djoere skilder ... 3bb. Heit fynt autowaskjen sok moai wurk, dat hy wol it net troch ien fan de bern ... 4. Om't ik sin oan farske fisk hie, sei ús heit tsjin my: Ast nocht haste, kinne we wol even by de mar ... 5. Lytse bern kinne noch net skriuwe, dêrom binne der ek skoaljuffen om de bern dy't dat noch net kinne ... 6. Dy minsken hawwe al sa lang gjin fatsoenlik iten hân, sille we har ... 7a. Jan hie in lekke bân, dat hy gong nei de fytsmakker om syn bân ... 7b. Nynke joech har treinkaartsje oan de kondukteur om it troch him ... 8a. Ik ha ús pake noch noait sûnder bril in krante ... 8b. Se sizze dat hy hiel goed foar clown spilet, mar ik haw him noch noait foar clown ... 8c. Dy sangeres haw ik dat ferske al faker ... 8d. It kin bêst wêze dat dy man sigaren rookt, mar ik haw him noch noait in sigaar ... 9. De muorren wienen sa tin, dat ik koe ús buorlju dúdlik oer polityk ... -Us mem hat in hiele grutte trui breide, mar doe't dy út de woskmasine kaam, wie it noch mar in hiel lyts ... -In swier skilderij moat oan in grouwe spiker, mar foar in lytsenien kinst wol ta mei in lyts ... -In strúsfûgel leit in hiel grou aai. In hintsje net, dy leit mar in ... -De bestjoerder fan de sportauto stjoerde mei in smel ... -Dy jonge hat twa ferskillende earen: in grut ear en in folle lytser ... -Dy tsjerke fersakket hielendal. Yn 'e toer sit in brede skuor. In pear jier lyn wie dat noch net sa. Doe siet der yn dy muorre noch mar in lyts ... mF'94-b
Jan gong nei de klokmakker om syn horloazje te reparearje litte We kinne wol even by it kanaal te fiskjen gean Dy kursus is der om bûtenlanners in brief skriuwe te learen We kinne wol even by de sleat gean fiskjen Dat soe ik no noait net dwaan, sei Piter Us heit gong nei de garaazje om syn auto meitsje te litten We kinne wol even by de see gean fiskjen Us mem gong nei de fytsmakker om har fyts te litte meitsje Pyt wol graach witte wat de mooglikheden binne We kinne wol even by de see gean te fiskjen Leararen binne der om bern in opstel te leare skriuwe We kinne dy minsken wol te iten freegje In oantal minsken sil dat TV-programma net wurdearje We kinne wol even by it kanaal fiskjen gean Us pake gong nei de garaazje om syn auto te litten meitsje It hûs wat dêr stiet, hat in skean dak Dizze kursus is bedoeld om de minsken in brief te learen skriuwe We kinne wol even by de see fiskje gean Jan gong nei de klokmakker om syn horloazje te reparearje litten mN'94-a
1a. Als Marilyn Monroe niet overleden was, dan zou ze nu zestig jaar ... 1b. Als ik twintig jaar eerder geboren zou zijn, zou ik nu van school af ... 1c. Als ik wist dat het zo vies zou smaken, zou ik nooit van die limonade ... 2a. Heb je een slecht cijfer voor je rapport? Dan had je je huiswerk maar beter ... 2b. Dat raadseltje is zo moeilijk, dat had zelfs de slimste jongen van de klas niet ... 2c. Eigenlijk had de clown altijd liever een akrobaat ... 3aa. Ik zou dat puzzeltje best ... 3ab. Al zijn klompen er dan niet voor gemaakt, volgens mij moet je er toch ook wel mee tegen een bal ... 3ba. Mijn oom zou zijn auto nooit door die monteur ... 3bb. Moeder vindt het schilderen van de voordeur zulk leuk werk, ze zou het nooit door een van de kinderen ... 4. Omdat ik zin had in verse vis, zei vader tegen mij: Als je zin hebt, kunnen we wel even bij de sloot ... 5. Kleine kinderen kunnen nog niet klokkijken, daarom zijn er leraren om de kinderen die dat nog niet kunnen ... 6. Die mensen hebben al zo lang geen fatsoenlijk eten gehad, zullen we ze ... 7a. Vader had een lekke band, daarom ging hij naar de fietsenmaker om zijn band ... 7b. Fred gaf zijn treinkaartje aan de conducteur om het door hem ... 8a. Ik heb mijn oom nog nooit zonder bril een krant ... 8b. Peter zegt dat zijn moeder heel vaak de kamer stofzuigt, maar ik heb haar nog nooit de kamer ... 8c. Die zanger heb ik dat liedje al eens eerder ... 8d. Misschien drinkt die man best wel eens bier, maar ik heb hem nog nooit bier ... 9. De muren waren zo dun, ik kon de buren heel duidelijk over politiek ... -Onze nieuwe buurman zei laatst dat ... -Is dat een blauwe auto of denk je -Johan zag laatst een nieuwe ... -Ik hou niet van zulke grote ... mN'94-b
We kunnen wel even bij het meer te vissen gaan Peter wist daar niet van waar het over ging Die cursus is er om kinderen een brief schrijven te leren We kunnen wel even bij de sloot gaan vissen Dat zou ik nou nooit niet willen Jan ging naar de garage om zijn auto maken te laten Nynke ging naar de fietsenmaker om haar fiets te laten maken Ze waren het er wel mee eens dat het zo moest We kunnen wel even bij de zee gaan te vissen Deze opdrachten zijn er om kinderen een opstel te leren schrijven Degene die dat wil, diegene mag dat wel doen We kunnen die mensen wel te eten vragen We kunnen wel even bij het meer vissen gaan Jan ging naar de klokkemaker om zijn horloge te repareren laten sF'94
1. falle - litten - litte - fallen Pas op! Do moast dy faas net
2. kinne - kommen - komme - kinnen Do soest ek wol op myn feestje
3. sjonge - hearren - hearre - sjongen Ik wol Madonna graach ris yn it echt
4. bliuwe - lizzen - lizze - bliuwen Dat boek kin dêr wol
5. helpe - moatte - moatten - helpen Klaas hie syn lytse suske
6. litten - falle - fallen - litte Ho! Dêr hie ik myn bril hast
7. rinnen - sjen - sjoen - rinne Ik ha juf yn de stêd
8. bleaun - lizze - lizzen - bliuwe Jan is te lang op bêd
Notes |