Institute for Applied Social Sciences (ITS), University of Nijmegen, PO Box 9048,
6500 KJ Nijmegen, The Netherlands
This paper explores the relations between standard-language and arithmetic test
performance and a range oflanguage-relatedfamilycharacteristicsin the Netherlands.
The sample consists of 7730 pupils from nearly 700 primary schools. The variables
analysed are: the language variety the child chooses in conversations with its father,
mother, siblings and friends;the language the parents communicate in; the importance
the parents attach to their home language; the parents’ command of standard-Dutch;
the parents’ educational level; the child’s gender; the province. The main question is
whether pupils who use standard-Dutch perform better than pupils who generally
speak another language variety, i.e. a Dutch dialect or Frisian. The results show that
only the parents’ educational level and the province are relevant in explaining differ- ences in standard-Dutch and arithmetic test results. Most remarkable are the results
from Limburg and Friesland. On average the children from both provinces speak
non-Dutch in 47% of the language domains. Yet the pupils from Limburg perform best
and the pupils from Friesland worst on the standard-Dutch and arithmetic test, even
after controlling for the family characteristics.Within these groups there are no differ- ences between the pupils who generally speak Dutch and the pupils who generally
speak a Limburg dialect or Frisian.
Recent Dutch studies into the effects of bilingualism on educational achievements have generally focused on non-indigenous pupils who are non-native
speakers of Dutch. Research into the educational achievements of children who
speak a Dutch dialect or vernacular at home, which Dutch sociolinguistics
initially concentratedon, has moved into the background.In this article we try to
fill the gap that exists with respect to the school success of non-standard Dutch
speaking pupils in primary education. The article is built up as follows. First, we
present a brief outline of the language situation in the Netherlands. Generally
speaking the Dutch language area covers both the Netherlands and parts of
Belgium. In this article we will, however, restrict ourselves to dialects and
vernaculars spoken within the boundaries of the Netherlands. Next, a number of
Dutch studies into the use of dialects and vernaculars and educational achievements are discussed. We present an outline ofresearch that was carried out into
this area in the 1970s and 1990s. After that, we look at the design of the Primary
Education cohort-study. We used the data from this cohort to answer the
research questions. The results of the analysis are dealt with next. This article is
rounded off with a summary and conclusions.
0790-8318/99/01 0001-22 $10.00/0 © 1999 G. Driessen & V. Withagen
Indigenous Language Varieties in the Netherlands
A large number of different dialects are spoken in the Netherlands in addition
to the standard language, i.e. standard-Dutch. Dialects which lie very close to
each other in geographical terms, are often fairly easily understood by people
living in a particular area. As the distance between the dialects increases, it
becomes more difficult for people to understand each other. It is therefore not
easy to make a clear classification of separate dialects. Dialectologists often
divide the Netherlands into a number of dialect groups (for an overview see
Daan & Blok, 1969).
In Figure 1 we present a topographicalmap of the Netherlands with its twelve
provinces. Within the Netherlands, the ‘Randstad’, which is situatedin the provinces of Noord-Holland, Zuid-Holland and Utrecht, is the economic, demographic, political and cultural centre. The Randstad is a conurbation in the
mid-west, which encompasses the four biggest cities of the Netherlands:
Amsterdam, Rotterdam, the Hague and Utrecht. Outside that area the population density (largely rural areas) is much lower, in particular in the north
(Friesland and Groningen) and the east (Drenthe) and south-west (Zeeland). In
the latter provinces about half of the working population is employed in agriculture. The two southern provinces of Noord-Brabant and Limburg and the
(eastern) provinces of Overijssel and Gelderland can be characterised as urban2 Language, Culture and CurriculumFigure 1 The provinces of the Netherlands.
ised rural areas. Flevoland is the newest Dutch province, in fact a polder which
has been reclaimed from an inner sea.
In the west of the Netherlands, the area between Amsterdam, Utrecht, the
Hague and Rotterdam, most people speak standard-Dutch. The dialects spoken
in this area, in the provinces of Noord-Holland and Zuid-Holland, are closest to
the standardlanguage. Most people therefore do not see themselves as speakers
of a dialect, even though their use of language has characteristicsof a dialect(Van
Hout, 1984; Hagen, 1989). Dialects have a relatively strong position in the north,
east andsouthofthe country.The general pattern is,the greaterthedistance fromthe west of the Netherlands, the greater the distance from the standardlanguage
(Hagen & Giesbers, 1988).
The Netherlands is one of the most urbanised areas in Europe. A lot of dialects
are therefore also city dialects, which largely have a low prestige. City dialects
are often associated with a lower social class. There are a few exceptions to this,
such as the city dialect of Maastricht, which is spoken by all social strata of the
Maastricht population and is more of an expression of regional or local identity
(Hagen, 1989). The city dialects of Noord-Holland and Zuid-Holland (e.g.
Amsterdam, Rotterdam, The Hague) are judged most negatively because the
link between dialect usage and socioeconomic status is most evident here. One
result of this stigmatisation is a reduction in the size of the dialect-speaking
group. We increasingly see this development taking place also in cities outside
the west of the Netherlands (Hagen & Giesbers, 1988).
Dialect-usage is more common in rural areas and in small towns than it is in
the cities. Rural dialects often have less of a stigma attached to them because they
are an expressionofregional identity ratherthan low social status.Here it is often
the case that people learn the standard language without giving up their dialect
(Hagen & Giesbers, 1988).
In Friesland, situated in the north of the country, a minority language is
spoken. Despite the strong influence of the standard language, the Frisian
language also has a strong position in particular in rural areas. In the cities of
Friesland a city dialect is spoken, which is a mixture of old Frisian dialects and
Dutch. Thanks to efforts of the Frisian movement and the Fryske Akademy,
Frisian has been recognised by a European Charter. One of the reasons forrecognising Frisian as a language is that there is a Frisian standard-language in addition to the Frisian dialects. This political recognition of Frisian as a language has
had important consequences for instruction in the language and educational
practice in the province of Friesland. Frisian has in the meantime become the
language spoken at all primary schools in Friesland, with the exception of some
exempted schools (Van Hout, 1984, Ytsma & De Jong, 1993). In addition to
Frisian, Low Saxon and Limburgs have recently been recognised as vernacular
by a European Charter.
Nationalfigures on theuse ofdialects in the Netherlands, such as the ones that
are available for Low Saxon in Germany (Stellmacher, 1994),do not exist. Careful
estimates have been made, however, for example, by Boves & Vousten (1996).
According to their analyses on average 12% of all parents speak a Dutch dialect
or Frisian with their child. They did however find major regional differences in
Language Varieties and Educational Achievement 3
this respect: in particular in the provinces of Overijssel, Drenthe, Limburg and
Friesland a dialect is often spoken in the home situation. There are, furthermore,
a number of sociolinguistic studies available carried out at a local or regional
level, such as the one by Van Hout (1989) for Nijmegen.
In the Netherlands the use of dialect has declined considerably over recent
decades. The reduction in the use of the ‘old’ dialects is not only evident from the
smaller number of speakers, but also from the sociological and demographic
characteristics of dialect-usage. The number of domains in which dialect is
spoken has very much been reduced. There is a clear trend which shows that
dialect-speaking parents are increasingly starting to speak the standard
language with their children. It is becoming less common for people to speak
only dialect; they usually command the local dialect as well as standard-Dutch.
In addition, the linguistic structure of dialects is moving more closely towards
that of the standard language: new linguistic variants are developing including
varieties that lie somewhere between a dialect and the standard language. This
development of interim forms, ‘regiolects’, is taking place at all levels of the
language (Van Hout, 1984; Hoppenbrouwers, 1990).
Research into Language Varieties and Educational
Dutch research carried out in the 1970s into the educational achievements of
dialect or vernacular-speaking pupils largely focused on a particular region or
town in the Netherlands.Wijnstra (1976) wanted to make a contribution towards
the discussion on Frisian as a subject and as a language of instruction in primary
education. He did not find any differences in the written command of Dutch
among pupils at mono- and bilingual schools in Friesland who were being
instructed in Frisian under various conditions. He also did not discover any
differences between the Frisian children and a control group ofrural pupils fromthe central Netherlands. The Frisian-speaking pupils did however score considerably lower on reading and language tests at the end of primary school than
those in the rest of the Netherlands. The arithmetic scores of the Frisian group
were also lower, althoughWijnstra suggested in his comments that this might be
related to the amount of time spent on arithmetic in school. In the 1990s De Jong
& Riemersma (1994)took anotherlook at the educational achievements of pupils
in Friesland. They concluded that pupils at Frisian primary schools did not do
any worse than those in the rest of the Netherlands. According to their study,
instruction in Frisian therefore also does not have any consequences for the
fluency in Dutch of the Frisian children.
The so-called Kerkrade-project consisted of a sizeable study which was
carried out in Kerkrade, a medium-sized town in the south of the Netherlands, in
the 1970s and early 1980s (Stijnen & Vallen, 1981). Pupils who had grown up
speaking the Kerkrade dialect had a poorer command of standard-Dutch than
standard-languagespeakers, in particular when it came to their grammaticaland
communicative command of the standard language and their participation in
verbal interactions in class. Apart from this, the researchers came across hardly
4 Language, Culture and Curriculum
any differences in the educational achievements. The dialect-speakers were
however more frequently given lower recommendations for secondary education1 and more frequently had to repeat a year. The researchers therefore
concluded that the assessments of the teachers were also being influenced by
their attitudes towards the dialect. Another important finding from this study
was that the language factor and the social class factor each had an independent
effect on school achievement. This implies that the negative effect of dialect
speaking is felt in higher as well as in lower socioeconomic classes. In Kerkrade
research was not only carried out into the educational achievements of
dialect-speaking children, but also into how the discrepancies established could
be overcome. Once the dialect-speaking pupils were allowed to speak the dialect
at school,the general achievement levels of these pupils were not any lowerthan
those of their standard language-speaking fellow pupils.
Parallel to the Kerkrade-project a comparable study was carried out in
Gennep, a small city in the south-east of the Netherlands. The language background of the pupils appeared to have a great deal less influence in the tests
administered here than in the Kerkrade-project (Giesbers et al., 1978).
After very little attention had been paid to the educational achievements of
indigenous bilingual children for quite some time,the subject was once again put
on the agenda by some researchers in the 1990s. Results of a large-scale national
study carried outin secondary education became available forthe very firsttime.
Boves and Vousten (1996) noted that the educational achievements of pupils
who spoke a variety of Dutch or Frisian with their parents were lower than those
of children who generally spoke Dutch. For an explanation of the differences
established they refer to Jansen Heijtmajer and Cremers (1993) who also found a
difference in the educational achievement levels of these pupils. The indigenous
pupils in this study who generally spoke a dialect or Frisian at home, performed
worse even than their non-indigenous (mostly Turkish and Moroccan) peers.
The study concerned children whoseparents hadnot obtained any qualifications
after leaving primary school (low socioeconomic status parents). Boves and
Vousten (1996) agree with the theory presented by these researchers, that there
could be a link between the language spoken at home and the educationallevel of
theparents andthatthis is the causeofthedifference in the achievement levels. In
other words, the relation between language spoken at home and achievements
implies a statistically spurious effect.
A recent study by Van Reydt (1997) focused on the opinions of teachers with
respect to the achievements of primary school pupils with a dialect background
(compare the study of Stijnen & Vallen, 1981). The dialect-speakers were
assessed less favourably on various personality traits, including intelligence,
than the pupils who spoke the standard language. In the last year at primary
school they also less frequently received recommendations for higher forms of
secondary education. These results were in sharp contrast to a direct measurement of the opinions of the teachers. From this they generally appeared to have a
positive attitude towards dialect and dialect-speakers. The teachers, however,
did not approve of the use of dialect in the school situation.
As mentioned in the introduction,the general aim of this paper is to clarify the
Language Varieties and Educational Achievement 5
situation with respect to the school success of non-standard Dutch speaking children. To gain more insight into this situation we conducted an empirical study in
primary education. Centralto this study is the question as to whether the use of a
Dutch dialect or Frisian in the home situation by pupils in the fourth year of
primary education is related to their standard-Dutch and numeracy achievements. Using analysis of variance and correlational analysis as the main techniques we will focus on answering the following sub-questions:
· What kind of relationships are there between the language the children
speak at home, the language the parents use among themselves,the importance the parents attach to their children learning this language and the
parents’ command of the standard-Dutch language? · What kind ofrelationships are there between the language spoken at home,
the standard-language and numeracy achievements and a number of background characteristics, including province and the education of the
Sample, Instruments and Variables
The data for this article were derived from the national cohort-study Primary
Education(‘PRIMA’). This concerns a longitudinal study in which informationis
gathered once every two years in primary schools from the school management
team, the teachers, the pupils and their parents. The study was started in the
1994/95 school year at almost 700primaryschools andinvolves a totalof approximately 70,000 pupils from Years 2, 4, 6 and 8. For an extensive account and
description of PRIMA we refer to Jungbluth et al. (1996) and Van Langen et al.
In this article we make use of the data relating to pupils in the fourth year
(approximately 8 years of age),2 concerning the results of a standard-Dutch and
numeracy test that the pupils completed, and a number of data on the pupils and
their home situation collected with the aid of a written questionnaire completed
by the parents. The information provided by the parents is available for a total of
10,375 pupils (cf. Driessen & Haanstra, 1996).Within this group we made a selection of those children of which both parents were born in the Netherlands, i.e. the
indigenous pupils. The restriction to indigenous pupils meant that 7730 pupils
remained for the analyses.
Instruments and variables
The language test
The language test that was administered within PRIMA was specially developed for this study by CITO (National Institute for Educational Measurement).
This test gives an indication of the general proficiency level in Dutch. The test
consists of 60 multiple-choice items and is made up of three types of exercises:
morphological, syntactic and semantic. The pupils have to decide whether the
words presented have been correctly formed or formed according to conven6 Language, Culture and Curriculum
tions, whether the build-up of the sentences presented is correct in terms of
grammar and whether the meaning of the words, word groups and sentences
presented is correct (Van Bergen, 1987). The reliability of this test (measured
according to KR20) is high and amounts to 0.85 (cf. also Driessen et al., 1994).
Some (translated) examples of items from this test:
Is this how we say the following, YES or NO? ‘My father brought himself a
new car’.
Is this sentence correct, YES or NO? ‘There goes he on his bicycle.’
Does this sentence make sense, YES or NO? ‘Pleasure is another word for
Does the part that is in italics make sense, YES or NO? ‘It is very importantfor
a dog to get the right kind of food. A dog can get healthy as a result of eating the
wrong kind of food.‘ The score on this test is determined as the total number of correctly answered
items. The average score is 47.8 with a standarddeviation of 6.9 and an empirical
range of 7–60.
The numeracy test
The numeracy test was also developed by CITO and covers a number of
numeracy skills: counting, ordering, structuring (numbers up to 100); automation (numbers up to 10), division, addition and subtraction; calculations (up to
100), additions and subtractions that do not go beyond ten; measurements (time,
length, surface, roads, money). The test consists of 40 multiple-choice items; the
reliability (established according to Cronbach’s Alpha) is high at 0.87.
Some examples (the exercises are read out loud by the invigilator; the pupils
have the test book containing the exercises in front of them):
9 can be divided into 6 and …
Ann has 20 guilders. She buys a pen for 7 guilders. How much money has
she got left?
70 minus 20 is?
Here too,the score on the test is determined as the number of correctly answered
items.The average is 31.9 and the standarddeviation 6.0,withan empiricalrange
of 5–40.
The questionnaire
The parents of the pupils completed an extensive questionnaire. The questions, among other things, related to socio-ethnic background and cultural, religious and educational aspects. In addition, several questions were asked about
language choice patterns and language proficiency. A number of these questions
are of importance to this article. From the child’s perspective, four of the questions were related to the language they choose to use in conversations with four
groups of people (domains):
Language Varieties and Educational Achievement 7
‘Which language does the child generally use? (a) with the mother/carer;
(b) with the father/carer; (c) with siblings; (d) with friends.’ The corresponding answer categories are: (1) Dutch; (2) a Dutch dialect or Frisian; (3)
another language (e.g. Turkish or Spanish).
From the parents’ perspective, the following question relating to choice of
language was asked:
‘Which language do you generally speak with your partner?’ Here 18
possible answer categories are presented, the first three of these are: (1)
Dutch; (2) a Dutch dialect; (3) Frisian. The other categories involve foreign
Also from the parents’ point of view, an attitudinal question was asked:
‘Do you feel it is important for your child to have a good command of the
language you speak with your partner?’ The answer categories are: (1) no,
not important; (2) yes, somewhat important; (3) yes, very important.
Finally each of the parents were asked about their command of the Dutch
language. In this four language modalities were distinguished.
‘To what extent do you/does your partner have a command of the Dutch
language? Notincluding dialects and Frisian.’(a) understand; (b) speak;(c)
read; (d) write. The corresponding answer categories are: (1) poor;(2) fairly
good; (3) good.
In additionto the answers to the questions relating to the choice oflanguage used
and command of the Dutch language, we also know which province the pupils live
in (for a breakdown of the distribution see the last column of Table 1), the pupils’
gender (50.7% boys) and the educational level of their parents. The latter is established as the highest level of education completed within the family (i.e. father and
mother). The corresponding categories are: (1) no more than primary education; (2)
junior secondary vocational education; (3) junior general secondary education or
years 1–3 of senior general secondary education/pre-university education; (4)
senior secondary vocational education; (5) years 4–5/6 of senior general secondary
education/pre-university education; (6) higher professional education; (7) university education. The distribution of the pupils over the education categories is as
follows: 5.5, 16.3, 15.5, 21.3, 14.0, 20.6 and 6.8%.
Choice of language and language proficiency
The informal spoken language used by the pupils
In the question related to the informal spoken language generally used in the
four domains (with father, mother, siblings, friends) three answer categories
were originally used. As a result of the selection of only indigenous pupils the
number of parents that opted for the third option (‘another language’) was so
small (maximum 0.2%), that we decided not to take this category into consider8 Language, Culture and Curriculum
ation. Next we looked at the relationship between Dutch versus dialect/Frisian
in the four domains; the percentages for Frisian and dialect have been added
together for this purpose. It appears that within the family this relationship is
always roughly the same. Of the children, 86.8% speak Dutch with their father,
86.5% speak Dutch with their mother and 87.0% speak Dutch with siblings. Aslightly bigger proportion of the children speak Dutch with friends, i.e. 88.6%.In
order to arrive at one score for these choices of language we carried out a principal component analysis. One general factor stands out here with 91.4%explained variance. The reliability (KR20) of this factor is 0.97. Subsequently a
new variable was constructed by summarising the relative number of times that
the children speak Dutch in the four domains. The score on this variable
‘informal spoken language used by child is Dutch’ ranges from 0% (Dutch is not
spoken in any of the domains)to 100% (Dutch is spoken in all of the domains).In
the first column of Table 1 we present the distribution of this variable, in which
the scores have been broken down according to province, and the mean score as
well as the standarddeviation for the Netherlands as a whole are given. In addition the Eta2 coefficient is given at the bottom of the table; this represents the
proportion of explained variance.3 To give a topographical impression we
present the mean scores broken down by province in Figure 1 as well.
On average the children speak Dutch in more than 87% of the four domains.
This percentage varies a great deal, however, depending on the province the
child lives in. The Eta2 value gives an indication of this; from this it is possible to
deduce that 28% of the variance in the informal spoken language used is related
to differences between provinces. We can distinguish roughly three categories
Language Varieties and Educational Achievement 9
Province Language
Groningen 87.9 58.1 2.38 2.94 237
Friesland 53.3 42.2 2.49 2.94 423
Drenthe 80.6 44.4 2.09 2.91 272
Overijssel 92.2 64.3 2.35 2.93 890
Flevoland 99.1 91.1 2.80 2.98 112
Gelderland 97.0 77.5 2.59 2.94 885
Utrecht 99.6 96.5 2.90 2.96 341
Noord-Holland 96.7 95.9 2.87 2.96 872
Zuid-Holland 99.2 98.0 2.91 2.96 1235
Zeeland 85.1 69.1 2.55 2.96 171
Noord-Brabant 95.7 81.3 2.64 2.92 1242
Limburg 53.4 39.4 2.51 2.93 1049
Eta2 0.28 0.23 0.10 0.01
Total 87.2 73.9 2.63 2.94
SD 31.9 43.9 0.68 0.16
Table 1 Informal spoken language used by the child Dutch, informal spoken language used by the parents Dutch, importance of learning informal spoken language
used by the parents and parents’ command of Dutch, per province (means)
here. In Friesland and Limburg the children speak Dutch in 53% of the domains.
In Drenthe, Zeeland and Groningen the percentage lies between 80 and 90, while
in the other provinces this is more than 90. In Utrecht, Zuid-Holland and
Flevoland the children speak Dutch in more than 99% of the domains.
In order to check whether there is any correlation between the pupil’s gender
and the educational level of the parents, and the informal spoken language used,
we have calculated the correlation coefficients. For gender this is 0.00 (p = 0.994)
which shows that there is absolutely no correlation with the choice of language
used; for the educational level of the parents the coefficient is 0.11 (p =0.000),
which points to a very slight correlation: children with parents with a higher
level of education speak Dutch slightly more often.
The informal spoken language used by the parents
With respect to the question relating to the informal spoken language used by
the parents among themselves we decided against taking the foreign languages
(0.4%)into consideration.The result is thedichotomyDutch versus Dutch dialect
or Frisian.The final score indicates what percentage of parents speak Dutch with
each other. The distribution on this characteristic per province can be seen fromthe second column of Table 1 and also in Figure 1.
Across the Netherlands as a whole, almost 74% of the parents speak Dutch
among themselves. In Limburg, parents speak the least amount of Dutch among
themselves, i.e. 39.4%,followed by the parents from Friesland and Drenthe with
respectively 42.2 and 44.4%. In four provinces more than 90% of the parents
speak Dutch: Zuid-Holland, Utrecht, Noord-Holland and Flevoland.
Here, too, we checked to see if there was any correlation between the informal
spoken language used by the parents and their educational level. The correlation
between both of these factors amounts to 0.14 (p = 0.000): parents with a higher
level of education therefore speak Dutch slightly more frequently than parents
with a lower level of education.
If we compare the percentages with regard to the informal spoken language
used by the child with those related to the informal spoken language used by the
parents (see Figure 1 or first and second column Table 1), it is noticeable that the
percentages vary considerably. In all cases it is true to say that the parents more
frequently speak dialect or Frisian among themselves than the children with
parents, siblings and friends. The biggest difference occurs among the people
from Drenthe: only 44% of these parents speak Dutch, as opposed to 81% of their
children. In provinces such as Groningen, Overijssel, Gelderland and
Noord-Brabant there appears to be a similar generation effect. In provinces such
as Utrecht, Noord-Holland and Zuid-Holland this is however hardly the case,
although it must be mentioned that hardly any dialect is used here.
The importance of learning the informal spoken language used by the
Slightly more than 74% of the parents feel that it is also important for their
child to have a good command of the language the parents speak among themselves, almost 15% feel this is reasonablyimportant and more than 11% feel thatit
10 Language, Culture and Curriculum
is not important. The mean score for the whole of the Netherlands is 2.63, but
there are considerable differences between the provinces. It is considered least
important in Drenthe (2.09) and most important in Noord-Holland and
Zuid-Holland, Utrecht and Flevoland (scores of 2.80 and higher).
The correlationwith genderis -0.01 (p = 0.223),which points to the factthatthe
parents do not differentiate between boys and girls when it comes to the amount
of importance they attach to their children learning their informal spoken
language. There is however a slight correlation with the educational level of the
parents of 0.18 (p = 0.000):more highly educated parents attach more importance
to it than parents with a lower level of education. There also appears to be a very
strong correlation between the informal spoken language used by the parents
and the importance they attachto their children learning this language:r = 0.66 (p = 0.000). The parents who speak a dialect or Frisian among themselves have a
score of 1.89 on the importance they attach to their children learning their
informal spoken language, while the parents who speak Dutch have a score of
2.90. This difference amounts to approximately one and a half standard deviations. As a result it looks as though parents who speak a dialect or Frisian attach
relatively little value to passing that dialect or Frisian on to future generations.
Closer analysis of these data makes it clear however, that it is possible to distinguish two sub-categories within the category of parents who speak a dialect or
Frisian. The first is made up of the parents from Friesland and Limburg, and the
second of parents from the other provinces who speak a dialect. The mean scores
for both groups are respectively 2.25 and 1.62, a difference of almost one standard deviation. Within the category of Dutch-speaking parents there are virtually no differences between the provinces as far as the importance they attach to
their children learning the informal spoken language is concerned. The conclusion therefore is that it is largely the dialect-speaking parents who do not come
from Limburg or Friesland who do not attach a great deal of importance to
passing on their dialect to future generations; the parents from Limburg and
Friesland on the other hand do feel that this is important.
The parents’ command of Dutch
With regard to the command of Dutch eight sub-questions were asked. It
should be remembered that this concerns self-reported language proficiency.
Using principal component analyses we checked to see whether it would be
possible to reduce these questions to a more limited number of factors.It appears
to be possible to distinguish one factor that explains 43.2% of the variance. The
reliability of this factor (Cronbach’s Alpha) is 0.81. Here, too, we subsequently
constructeda new variable by summarising the scores on the eight questions and
dividing them by the valid number of values. This gives the mean of the scores of
the father and mother with respect to understanding, speaking, reading and
writing the Dutch language. The final score on this variable ‘parents’ command
of Dutch’ranges from 1 ‘poor’, via 2 ‘fair’ to 3 ‘good’. In the third column of Table
1 we present a summary of this variable, in which the scores have been broken
down according to province.
We can be quite brief when it comes to an explanation. The total mean score of
Language Varieties and Educational Achievement 11
2.94 indicates that the parents are in general of the opinion that they have a good
command of the Dutch language on each of the modalities. The scores per province furthermore make it clearthat there are hardly any differences in thatrespect;
the Eta2 coefficient confirms this conclusion with only 1% explained variance.
There is a reasonably strong correlationof 0.26 (p = 0.000)between the parents’
command of Dutch and their level of education: parents with a higher level of
education have a better command of Dutch than parents with a lower level of
education. There is a considerably weaker correlation of 0.14 (p = 0.000) between
the command of Dutch and the informal spoken language used by the parents
when speaking among themselves.
Choice of language, language proficiency, and language and
numeracy skills
In Table 2 we present a summary of the language and numeracy skills of the
children, as measured with the aid of the CITO tests, broken down according to a
number of the variables outlined above. We have condensed some of these variables for the benefit of this presentation. We reduced informal spoken language
used by the child to three categories: (1) Dutch in none of the four domains; (2)
Dutch in one to three of the four domains;(3) Dutch in allfour of the domains. The
parents’ command of Dutch has been condensed to: (1) poor/fair; (2) fairly good;
(3) good. In the case of the provinces a fourfold division was used based on a
combination of, on the one hand, the unique character of the language/the dialect
and, on the other hand, the relative number of times the children speak Dutch: (1)
Friesland (because of Frisian, a Frisian vernacular or dialect);(2) Limburg (because
of the Limburg dialect, i.e. language); (3) Groningen, Drenthe and Zeeland
(because Dutch is spoken in less than 90% of the domains here); (4) Overijssel,
Flevoland, Gelderland, Utrecht, Noord-Holland, Zuid-Holland and
Noord-Brabant (because Dutch is spoken in more than 90% of the domains here).
The tablenot only shows the meanscoresper categorybut alsotheEta2 coefficients.
First, some comments with regard to language proficiency. The table shows that
as far as language proficiency is concerned, the informal spoken language used,
by the child orthe parents, makes very little difference. It explains only 1% of the
variance in language proficiency. The difference between the children who
speak Dutch in none ofthe four domains and the children who speak Dutch in all
of the four domains is almost 2 points, i.e. two test items. When it comes to the
effect of the informal spoken language used by the parents on language proficiency, it appears that this also does not amount to more than 1% explained variance. The effect of the parents’ commandof Dutch on the language proficiency of
their children is somewhat greater, i.e. 2%. The difference between parents who
have a poor to fair command of Dutch and parents who have a good command,
however, is more than 3 points. The latter amounts to almost half a standard
deviation, which can be seen as a reasonably strong effect. The difference
between boys and girls is not relevant with less than half a point. The biggest
differences occur with respect to the educational level of the parents; this characteristic explains 6% of the variance. Major differences exist between the two
extreme educational categories: children of parents with at most primary educa12 Language, Culture and Curriculum
tion do less well by more than seven items than children ofparents with a university education; this difference represents more than one standard deviation. The
differences according to the amount of importance attached to learning the
informal spoken language are not relevant overall; but it is true to say that the
children of parents who attach most importance to this have the highest
language scores. Finally, language proficiency is broken down according to
various parts of the country. The Frisian children with a score of 45.2 are least
proficient in Dutch. The language proficiency of children from Groningen,
Drenthe and Zeeland, where Dutch is spoken in less than 90% of the domains, is
slightly better with 46.7 points. The language proficiency of the children fromLimburg and the children from the other provinces, where Dutch is spoken in
more than 90% of the domains, is the same. The latteris striking because the children from Limburg speak a dialect in almost half of the domains.
When it comes to numeracy skills we can be fairly brief. Even though there is a
slight tendency for Dutch as the informal spoken language used by the child and
informal spoken language used by the parents to go together with better
numeracy skills,the differences are notrelevant according to the Eta2 values. The
relevance of the amount of importance parents attach to their children learning
Language Varieties and Educational Achievement 13
Language Numeracy Language Numeracy
Language child Dutch Command Dutch parents
0 domain 46.3 31.5 poor/fair 45.1 30.7
1-3 domains 47.0 31.5 fairly good 46.7 30.8
4 domains 48.1 32.0 good 48.4 32.3
Eta2 0.01 0.00 Eta2 0.02 0.01
Language parents Dutch Gender
Dialect/Frisian 47.0 31.6 boy 47.6 32.4
Dutch 48.2 32.1 girl 48.0 31.4
Eta2 0.01 0.00 Eta2 0.00 0.01
Education parents Part of country
Primary 43.9 29.2 Friesland 45.2 29.9
Junior vocational 46.1 30.3 Limburg 48.1 33.0
Junior general 47.0 31.6 Gr/Dr/Zl 46.7 31.5
Senior vocational 47.6 31.6 Ov/Fl/Gl/Ut/
48.1 31.9
Senior general 48.8 32.8 Eta2 0.01 0.01
Higher professional 49.5 33.1
University 51.2 34.4
Eta2 0.06 0.05
Importance language parents
Not important 47.2 31.6
Somewhat important 46.7 31.1
Very important 48.2 32.1
Eta2 0.01 0.00
Table 2 Language and numeracy skills by choice of language, language proficiency
and background characteristics (means)
their informal spoken language, the parents’ command of Dutch, and gender
also appears to be very limited. Educationallevel appears to be the mostrelevant
factor here, too. The difference between the two extreme educational categories
amounts to more than 5 points, or 0.87 of a standard deviation. As far as part of
the country is concerned, we see that the children from Limburg score highest
and those from Friesland score lowest.
If we compare the findings forlanguage with those for numeracy, it is striking
that with one exception the factors examined are hardly relevant when it comes
to explaining the differences in language proficiency and to an even lesser extent
thedifferences in numeracy skills. The only thing thatreally matters is the educational level of the parents. It is furthermore striking that the children fromLimburg do best in both language and numeracy, while they, together with the
Frisian children, do not speak Dutch in much more than half of the domains.This
stands out all the more, because the Frisian children do the least well in language
and numeracy (cf. also Van Langen & Vierke, 1992). For that matter, it is true to
say that within the group of children from Limburg and the group of children
from Friesland there are no differences in language proficiency and numeracy
skills according to the extent to which the parents or the children use a dialect.
The respective achievements of the children from Limburg and Friesland, therefore, do not appear to be affected by the extent to which standard-Dutch is
The above findings show that the relationship between the various factors is
clearly not as unambiguous as it would appearto be at first glance.In orderto get
a better insight into this, we looked at the interrelationship between the various
characteristics. These have been presented in Table 3 in the form of correlation
In general in research methodology the criterion forthe relevance of a correlation is put at a coefficient of approximately 0.15; this comes down to about 2%explained variance. From this point of view it becomes clearin our study that the
number of relevant correlations is very limited. Children of highly-educated
14 Language, Culture and CurriculumLanguage
skills child
skills child
Numeracy child 0.41
Gender child 0.03 -0.09
Language child 0.08 0.03 0.00
Education parents 0.24 0.21 -0.02 0.11
Language parents 0.08 0.04 0.00 0.61 0.14
0.07 0.05 -0.01 0.19 0.18 0.66
Command Dutch
0.12 0.08 -0.01 0.11 0.26 0.14 0.17
Table 3 Relationships between choice of language, language proficiency and numeracy skills and background characteristics (correlations)
parents with a better command of Dutch have a relatively high proficiency level,
as well as children of parents who have a better command of Dutch. Children
who speak relatively much Dutch often also have parents who speak Dutch
among themselves. More highly educated parents speak Dutch slightly more
often and alsohave a better commandofDutch. They also attachsomewhatmore
importance to their children learning to speak the same language they speak
among themselves. Parents of children who relatively often speak Dutch also
attach slightly more importance to their children learning that language. Finally,
parents who speak Dutch relatively often and attachimportance to their children
speaking the language they speak among themselves have a somewhat better
command of Dutch. As far as numeracy is concerned there is only one relevant
correlation with the educational level of parents: children of more highly
educated parents do slightly better on the numeracy testthan children of parents
with a lower level of education.
In order to get more insight in the interrelationship between the various characteristics and the language proficiency level and numeracy skills of the pupils,
we finally carried out a number of variance and regression analyses.
First of all the results of the analyses in which the language proficiency level is
the variable to be explained. In an initial analysis we checked to see if the differences between the various parts of the country continue to exist when one first
takes the differences in the education of the parents into account. We therefore
wanted to check whether differences in language achievements between various
parts of the country could possibly be the result of differences in the level of
education of the parents between different parts of the country. From this analysis it became clear that educational level explains 6.1% of the variance and that
subsequently the part of the country explains a further 1.1%.This is just as much
as in the bivariate analyses; compareTable 2.The analysis also shows that the Eta
and Beta coefficients remain exactly the same (0.25 for education, respectively
0.11 for part of the country); controlling for the other variable in the comparison
therefore has no effect.
Similar analyses were also carried out to see whether after controlling for the
other characteristics (informal language child, parents’ command of standard-Dutch, informal language parents, importance language parents) there
would still be some effect for part of the country. In all instances the Eta and Beta
values did not diminish by more than 0.01 after controlling, which means that
also after controlling for the various characteristics the effect of part of the
country remains the same.
In a finalregression analysis we entered all of the above-mentioned characteristics:gender and informal language child, education, informal spoken language
and importance language parents, parents’ command of Dutch, and part of the
country. As far as the latteris concerned we have formed a dichotomy: the provinces in which Dutch is spoken in about half of the domains (Friesland and
Limburg) and the other provinces, where Dutch is spoken in more than 80% of
the domains.We did this to check to see which of these characteristics carry most
weight when it comes to explaining the differences in the language proficiency
levels of children. The results can easily be summarised:the level of education of
Language Varieties and Educational Achievement 15
theparents explains 5.6%of the variance, and the otherfive characteristics collectively less than 0.8%. The Beta of education parents amounts to 0.21; the highest
Beta for the five remaining characteristics is the one relating to the parents’
command of Dutch: 0.06.
The results withregard to numeracy skills hardlydeviate from those relatedto
language proficiency. In general the percentages of explained variance are somewhat lower and the Eta and Beta coefficients are slightly less strong. The correlations are however about the same as those for language.
Summary and Conclusions
In short it is possible to draw the following conclusions:
· Nearly 13% of the Dutch indigenous children in the fourth year of primary
education occasionally speak a dialect or indigenous minority language
(i.e. Frisian). There are considerable differences here, however, between
provinces. The use ofdialects andminority languages is concentratedin the
provinces of Friesland and Limburg. In additiondialectis spoken relatively
more frequently, i.e. in more than 10%ofthe fourdistinguished domains,in
Groningen, Drenthe and Zeeland. · About 74% of the parents speak Dutch among themselves; a quarter of
them therefore speak dialect or Frisian. This means that parents far more
frequently speak a dialect or Frisian than their children. The difference
within the family is particularly great (more than 10% difference) in
Groningen,Drenthe, Overijssel,Gelderland, Noord-Brabant and Limburg. · More than 74%ofthe parents feel thatit is importantfortheir child to have a
good command of the language the parents speak among themselves.
There are however major differences here. In particularthose parents who
speak a dialect among themselves and do not come from Friesland or
Limburg attach the least importance to this. · More highly educated parents have a better command of Dutch than
parents with a lowerlevel of education. They also more often speak Dutch.
There is only a weak correlation between command of Dutch and the
informal spoken language the parents use among themselves. · The number of relevant correlations between language proficiency and
numeracy skills on the one hand, and, the other variables on the other, is
very limited. It is in particular the education of the parents that plays an
important role. In addition the parents’ command of Dutch is of slight
importance. Differences in language proficiency according to the informal
spoken language used by the child, informal spoken language used by the
parents and gender are very small.These differences are even smallerwhen
it comes to numeracy skills. · The part of the country in which the child lives appears to be a relevant
factor in the relationship between choice of language and language proficiency and numeracy skills. It is clearly possible to distinguish parts of the
16 Language, Culture and Curriculum
country in which a dialect or Frisian is spoken to a greater or lesser extent.
At the same time these parts of the country also differ according to the
language proficiency and numeracy skills of the children, which furthermore cannot be explained by other characteristics such as the level of
education of the parents.
Some methodological remarks
Before we continue with an elaboration and discussion of the conclusions, we
would like to draw attention to some methodological points.
When looking for an explanation of the differences between Frisian and
Limburg pupils, the nature of the tests used in PRIMA might come into the
picture. These tests are of a multiple-choice character. While administratively
convenient, interpreting the tests’ results in some instances might pose a
problem. Regarding the validity, such tests have in general proved to be fairly
good predictors of secondary education levels. Still, an interesting pointremains
as to whether syntactically and semantically differences between Frisian and
Standard-Dutch on the one hand and between Limburgs and Standard-Dutch on
the other hand are reflected equitably in the test items. What is at stake here is
whether it is easier for the speakers of Limburgs to pick correct items by making
analogies with their first language than for the Frisian pupils. Or, from another
perspective, whether the Frisians are possibly more troubled with interference
errors than Limburgers. As to the question of the linguistic distance, there is
evidence that the distance between Limburgs and Standard-Dutch is greater
than that between Frisian and Standard-Dutch (cf. Hoppenbrouwers, 1990; De
Jong & Riemersma, 1994; Ytsma, 1995). This actually means that the Limburgs
speaking pupils are at a disadvantagecompared with the Frisians, which implies
that if the dialect were a relevant factorthe Limburg pupils would perform even
less well than the Frisian pupils. Apart from this, studying item-bias (when
pupils from different subgroups, but of a similar skill level, stand an unequal
chance of giving the proper answer to the item concerned) is extremely difficult
and surrounded with a lot of uncertainty and ambiguity (cf. Hagen, 1989;
Uiterwijk, 1994; Vallen & Stijnen, 1996).
Some of the variables analysed in this article pertain to self-reported language
behaviour and language proficiency. Little is known as regards the validityofthe
answers. As already stated in the introduction to this article, many people living
in the west of the Netherlands tend to see themselves as speakers of Standard-Dutch, although their language use shows characteristics of a dialect (Van
Hout, 1984; Hagen, 1989). This would mean that the percentages in Table 1
present an underestimation of the real dialect use in the west of the Netherlands.
At the same time in some regions and cities dialect use is associated with lowsocioeconomic status (but certainly not everywhere; in the Limburg capital of
Maastricht,forinstance,dialect use carries a lot of prestige; cf. Hagen & Giesbers,
1988). As a consequence, this might lead to socially desirable responses, and
again to an underestimation of the real dialect use. Another point of discussion
pertains to the parents’ assessment of their command of standard-Dutch. We do
Language Varieties and Educational Achievement 17
not know whether their self-rated fluency in Dutch is accurate, and — more
important — whether this accuracy is similar for all the groups studied.
Although recent review studies (cf. Driessen, 1995) show that in explaining
differences in achievement levels the parents’ educational level is a far more reliable and stronger predictor than other socioeconomic status characteristics, we
performed some additional analyses with occupational status and family
income. First we checked whether the Frisian and Limburg families differ in
status and income. This proved not to be the case. Next we carried out a series of
analyses of variance to check whether after controlling for status and income the
differences in achievement levels between Frisian and Limburg pupils remain
the same. The results showed that occupational status and family income do not
influence the correlations between achievement levels and province. Furthermore, this finding not only applies to the Frisian and Limburg pupils, but to the
pupils from the other provinces as well. In other words: status and income do not
add anything in explaining the existing achievement differences.
Further to the conclusions presented earlier we would like to make a fewcomments. From the above it emerges that social class (or one of its indicators,
parental education) plays a relatively dominant role in explaining achievement
differences. This is in line with findings as presented in the Dutch and international literature (cf. Rossi & Montgomery, 1994; Evans, 1995; Day, Van Veen &Walraven, 1997). Some of the studies that we reviewed in the first part of this
paper have explicitly taken social class into account when trying to explain
language differences between standard-Dutch and dialect speakers. The findings of these studies, however, were not unambiguous. Our study shows that
social class actually is the only relevant factor; the language factor is of no relevance. Differences with earlierfindings might be related to the scope of our study
(national instead of local or regional) and to the fact that we have used recently
collected data.
The findings show that it is possible to speak of a development whereby less
dialect or Frisian is being spoken from generation to generation. This language
shift and erosion of dialects and Frisian are an almost universal phenomenon (cf.
Hagen, 1987;Fishman, 1991).There is a clear parallel here with the language situation of non-indigenous people; compare for example Driessen (1992, 1997). It
furthermore appears that parents feel that it is less importantfortheir children to
learn to speak the dialect correctly. This in particular applies to the parents who
do not originally come from Friesland or Limburg; parents of children fromFriesland and Limburg attach relatively more importance to it. These findings
suggest that the dialects and indigenous minority languages are slowly dying
out (cf. Ytsma, 1995; Ytsma et al., 1994). Just exactly at what kind of pace this is
likely to happen is still unclear (Van Hout, 1984). Moreover, the language shift
does not necessarily mean that these will in all cases be replaced by Dutch. There
are signs that developments are taking place in the shape of ‘regiolects’
(Hoppenbrouwers, 1990).
18 Language, Culture and Curriculum
Another point concerns the effect of the part of the country. From our data it is
noticeable that the children from Limburg perform best in both language and
arithmetic, while their Frisian peers perform worst in both tests. This is striking
because this is coupled to the fact that children from Limburg and Friesland
speak about the same amount of Dutch. It has furthermore been shown that this
is not due to possible differences in the educational level of the parents. With
respect to non-indigenous children in this context, the interdependence and the
threshold hypotheses by Jim Cummins may be significant (Cummins, 1991). Arough interpretation of these two hypotheses is that the language proficiency
level in Dutch is dependent on the proficiency level in the language spoken at
home (i.e. dialect or Frisian) and that a certain level needs to be achieved in both
languages before bilingualism can be expected to have positive effects on cognitive development. There are two important conditions for success, namely that
the children have to be exposed to Dutch to a sufficient degree (at school orin the
home environment) and that they have to be sufficiently motivated to learn
Dutch (see also De Jong & Riemersma, 1994). Quite apart from the fact that these
hypotheses are extremely hard to prove empirically (cf. Driessen, 1996), they do
not provide any clues as to why the children from Limburg perform much better
than the children from Friesland. This finding in fact does not match the expectations at all. After all, in Frisian primary schoolsFrisian is a subject and a language
of instruction (although the extent of this strongly varies from school to school
and in general is very limited; cf. Inspectie van het Onderwijs, 1989), Frisian
furthermore has a great deal more status as a language (although the Limburg
dialect has recently also been officially recognised as a European indigenous
minority language) and there is a written standard variant of Frisian.
One possible explanation for the differences between Friesland and Limburg
which is occasionallyput forwardis thatit could depend on the differences in the
types of schools available. In Limburg there are a lot of schools for special education (for children with learning difficulties and behavioural problems), while
there are only very few of these schools in Friesland. This means that Frisian children are more frequently forced to stay in regular primary education, while their
peers from Limburg are referred to special primary education (cf. Petersen &Oudenhoven, 1994).The better performance of the children from Limburg could,
therefore, be because the very worst performing pupils are no longer in primary
education there and can, therefore, also no longer bring down the average test
scores, which is the case in Friesland. This explanation, however, is not
completely satisfactory. In 1997 a national newspaper started to publish
so-called league tables of secondary school attainments. From these tables it
becomes clear that the differences that exist in primary education are present in
secondary education as well: the secondary schools in Limburg perform best.
Until now no explanation has been found for this phenomenon.
What are the consequences of this forthe didactic approach in primary education? In the Kerkrade-project it was noted that dialect-speaking children fromLimburg in primary education performed less well in terms of language than
pupils who did not speak a dialect (Stijnen & Vallen, 1981). Our study does not
confirm this conclusion (some 15 years on), when one takes the educational level
Language Varieties and Educational Achievement 19
of the parents into account. While children from Limburg do extremely well,
children from Friesland do extremely poorly. Our study does not confirm that
this is due to the fact thatthey speak relatively little Dutch. Closer analysis shows
that on average all Frisian children perform badly,regardless of how much standard-Dutch they speak. As a result of this, there appears to be little reason for not
using the native language in class (cf. Orstein-Galicia, 1994), Although it should
be noted that research in this area has not as yet led to uniform outcomes.
The authors would like to thank Ton Goeman of the Meertens Institute
(Amsterdam) for his help with the topographical map of the Netherlands. The
Foundation for Behavioural and Educational Sciences of the Netherlands Organization for Scientific Research (NWO, the Hague) is gratefully acknowledged
for providing the data of the cohort-study Primary Education on which the
empirical part of this article is based.
Any correspondence should be directed to Dr Geert Driessen, Institute for
Applied Social Sciences (ITS), University of Nijmegen, PO Box 9048, 6500 KJ
Nijmegen, The Netherlands ([email protected]).

  1. When the children are in the last year at primary school, they have to choose a secondary school. The parents will get advice in this from the head teacher orthe child’s class
    teacher. This recommendation largely depends on how well the child is doing at
  2. Dutch primary education is intended for 4 to 12-year-olds and consists of eight years.
    In the first two years (the infant school years) play takes up a central place; in the third
    year formal instruction in reading, numeracy and writing commences. 3. As far as the p-values are concerned we would like to point out that they are less informative here; when analysing such large groups of pupils diferences very soon are
    Boves, T. and Vousten, R. (1996) Thuistaal en schoolresultaten. In R. van Hout and J.
    Kruijsen (eds) Taalvariaties.Toonzettingenen Modulatiesop eenThema. Dordrecht: Foris.
    Cummins, J. (1991) Interdependence of first- and second-language proficiency in
    bilingual children. In E. Bialystok (ed.) Language Processing in Bilingual Children. Cambridge: Cambridge University Press.
    Daan, J. and Blok, D. (1969) Van Randstad tot Landrand. Amsterdam: Noordhollandse
    Day, C., Van Veen, D. and Walraven, G. (eds) (1997) Children and Youth at Risk and Urban
    Education. Research, Policy and Practice. Leuven/Apeldoorn: Garant.
    De Jong, S. and Riemersma, A. (1994) Taalpeiling yn Fryslan. Onderzoek naar de beheersing
    van het Fries en het Nederlands aan het Einde van de Basisschool. Ljouwert: Fryske
    Driessen, G. (1992) Etnische herkomst, verblijfsduur, thuistaal en taalvaardigheid
    Nederlands. Toegepaste Taalwetenschap in Artikelen 44, 7–19, 111.
    Driessen, G. (1995) Het relatieve belang van sociaal milieu en etnische herkomst voor de
    20 Language, Culture and Curriculum
    verklaring van onderwijsachterstanden. Een overzicht van kwantitatief-empirische
    studies; bevindingen en problemen. Tijdschrift voor Onderwijsresearch 20, 341–362.
    Driessen, G. (1996) Minority language and culture teaching in the Netherlands: Policies,
    arguments, evaluation and prospects. Compare 26, 315–332.
    Driessen, G. (1997) From mother tongue to foreign language. Prospects for minority-language education in the Netherlands. In Th. Bongaerts and K. de Bot (eds)
    Perspectives on Foreign-language Policy. Studies in Honour of Theo van Els. Amsterdam/Philadelphia: Benjamins.
    Driessen, G. and Haanstra, F. (1996) De Oudervragenlijst Basisonderwijs en Speciaal
    Onderwijs. Te chnische Rappo rtage PRIMA-Coho rtonderz oe k 1994/95. Amsterdam/Nijmegen: SCO/ITS.
    Driessen, G., Van Langen, A. and Oudenhoven, D.(1994)De ToetsenvoorhetCohortPrimair
    Onderwijs. Verantwoording. Nijmegen: ITS.
    Evans, P.(1995)Children and youth‘atrisk’.InOECD Our ChildrenatRisk. Paris: OECD.
    Fishman, J. (1991) Reversing Language Shift. Theoretical and Empirical Foundations of
    Assistance to Threatened Languages. Clevedon: Multilingual Matters.
    Giesbers, H., Kroon, J. and Liebrand, R. (1978) Dialect en School. Een Empirisch
    Sociodialectisch Onderzoek naar de Invloed van Dialectgebruik op Schoolprestaties in het
    Basisonderwijs in Gennep. Nijmegen: KUN.
    Hagen, A. (1987) Dialect speaking and school education in Western Europe. In U.
    Ammon, K. Mattheier and P. Nelde (eds) Sociolinguistica. Volume 1. Tübingen: Max
    Niemeyer Verlag.
    Hagen, A. (1989) The Netherlands. In U. Ammon, K. Mattheier and P. Nelde (eds)
    Sociolinguistica. Volume 3. Tübingen: Max Niemeyer Verlag.
    Hagen, A. and Giesbers, H. (1988)Dutch sociolinguistic studies. InternationalJournal of the
    Sociology of Language 73, 29–44.
    Hoppenbrouwers, C. (1990) Het Regiolect. Van Dialect tot Algemeen Nederlands. Muiderberg: Coutinho.
    Inspectie van het Onderwijs (1989) Het Onderwijs in het Fries op de Basisschool. Stand van Zaken 1988–1989. Een Kwantitatief Inspectieonderzoeknaar het Onderwijs in het Fries op de
    Basisschool in Friesland. Zoetermeer: Inspectie van het Onderwijs.
    Jansen Heijtmajer, W. and Cremers, P. (1993) Nederlands als tweede taal. Moeilijk voor allochtone èn autochtone tweetaligen. Samenwijs 14, 104–8.
    Jungbluth, P., Van Langen, A., Peetsma, T. and Vierke, H. (1996) Leerlinggegevens Basisonderwijs en Speciaal Onderwijs. Technische Rapportage PRIMA-Cohortonderzoek
    1994/95. Amsterdam/Nijmegen: SCO/ITS.
    Ornstein-Galacia, J. (1994) The dilemma of dialect in the classroom. Pro and contra.
    ITL/Review of Applied Linguistics 103/104, 89–112.
    Petersen, B. and Oudenhoven, D. (1994)Verwijzing naar het SpeciaalOnderwijs.Een Analyse van Regionale Verschillen. Nijmegen: ITS.
    Rossi, R. and Montgomery, A. (1994) Educational Reforms and Students at Risk. A Review of
    the Current State of the Art. Washington DC: US Department of Education.
    Stellmacher, D. (1994) NiedersachsischerDialektzensus. Statistisches zum Sprachgebrauch imBundesland Niedersachsen. Stuttgart: Steiner.
    Stijnen, P. and Vallen, A. (1981) Dialec t als Onderwijsprobleem. Een
    Sociolingustisch-Onderwijskundig Onderzoek naar Problemen van Dialectsprekende
    Kinderen in het Basisonderwijs. ‘s-Gravenhage: Staatsuitgeverij.
    Uiterwijk, H. (1994) De Bruikbaarheid van de Eindtoets Basisonderwijs voor Allochtone
    Leerlingen. Arnhem: Cito.
    Vallen, T. and Stijnen, S. (1996) Terug naar Kerkrade. Een zoektocht naar sporen van het
    Kerkradeproject. In R. van Hout and J. Kruijsen (eds) Taalvariaties: Toonzettingen en modulaties op een thema. Dordrecht: Foris.
    Van Bergen, J. (1987) Verantwoording Constructie Toetsen voor de Evaluatie van het
    Onderwijsvoorrangsbeleid. Arnhem: CITO.
    Language Varieties and Educational Achievement 21
    Van Hout, R. (1984)Sociolinguistics in the Netherlandic language area. In K. Deprez (ed.)
    Sociolinguistics in the Low Countries. Amsterdam/Philadelphia: Benjamins.
    Van Hout, R. (1989) De Structuur van Taalvariatie. Een Sociolinguistisch Onderzoek naar het
    Stadsdialect van Nijmegen. [S.l.: s.n.]
    Van Langen, A. and Vierke, H. (1992) Het Onderwijs in de Noordelijke Plattelandsgebieden.
    Samenstelling en Prestaties van de Leerlingpopulatie. Nijmegen: ITS.
    Van Langen, A., Vierke, H. and Robijns, M. (1996) Veldwerkverslag Basisonderwijs en
    Speciaal Onderwijs. Technische Rapportage PRIMA-Cohortonderzoek 1994/95. Nijmegen/Amsterdam: ITS/SCO.
    Van Reydt, A. (1997) Dialect en Onderwijs in Emmen. Amsterdam: Thesis Publishers.
    Wijnstra, J. (1976) Onderwijs aan van Huis Friestalige Kinderen. Verslag van een Evaluatie-Onderzoek in een Meertalige Regio. ‘s-Gravenhage: Staatsuitgeverij.
    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. Ljouwert: Fryske Akademy.
    Ytsma, J. and De Jong, S. (1993) Frisian. In G. Extra and L. Verhoeven (eds) Community
    Languages in the Netherlands. Amsterdam/Lisse/Berwyn PA: Swets & Zeitlinger.
    Ytsma, J., Angels Vilador, M. and Giles, H. (1994) Ethnolinguistic vitality and ethnic
    identity: Some Catalan and Frisian data.InternationalJournal of the Sociologyof Language
    108, 63–78.
    22 Language, Culture and Curriculum