The effectiveness of early foreign language learning in the Netherlands

This article reports on a number of projects on early English teaching in the Netherlands. The focus of these projects has been on the impact of English on the development of the mother tongue and the development of skills in the foreign language. Overall the results show that there is no negative effect on the mother tongue and that the gains in English proficiency are substantial. Given the specific situation in the Netherlands where English is very present, in particular in the media, a real comparison of the findings with those from other countries is problematic.


Introduction
In this contribution an overview of the research on early foreign language teaching in the Netherlands will be presented. Research on this topic started in the early 2000s with a number of small-scale projects and, more recently, a large, government-funded project was finished. The findings will be presented along 410 with some ideas and related research on future developments. First, a brief history of the provision of bilingual education in the Netherlands will be presented. Internationally there is a lot of research on early foreign language teaching, but we will not deal with that here since there is an excellent overview by Nikolov and Mihaljevi Djigunovi (2011).

Bilingual streams in secondary education
Bilingual education started in 1992, when a group of parents whose children attended secondary education in the Dutch section of international schools demanded that their children would have access to a form of bilingual education, too. International schools cater for children of ex-pats and embassy personnel, and they are not government-sponsored, which allows some freedom in the curriculum. A large part of the program in such schools is taught in English, and Dutch is taught as a second language. According to ministerial regulations, teaching in Dutch schools needs to be done in Dutch, unless the aim is to enhance levels of proficiency in a foreign language. This "loop" in the regulations has been used to set up bilingual streams in a small number of schools in the early 1990s. This "grassroots" movement of parents pushing for some form of bilingual education led to a program with an English stream in three schools in 1992. The number of bilingual schools grew rapidly to some 130 schools in 2014. While the intention was to provide bilingual education in various languages, in particular German in the border region in the eastern part of the Netherlands and French in the south, in effect almost all schools opted for English. At present there are only 23 schools that have a German-Dutch bilingual program; the rest is English-Dutch.
Over the years the European Platform for Internationalization of Education, a government-sponsored organization, has been active in developing bilingual education by forming networks of schools, providing training for teachers and supporting schools in developing a high quality program. There is now a well-developed quality control and accreditation system. There is a standard that schools aspiring to have a bilingual program are supposed to meet. It deals with the participation of native speakers, levels of English proficiency of the Dutch teachers, internationalization activities like school exchanges and the proportion of the program to be taught in English. It is now set on 50 % for the first four years and 30% for the last two years.
It could be argued that the development of bilingual education is the most important innovation in foreign language teaching in last 50 years, and research has now shown its effectiveness. An early but very important study was the one by Huibregtse (2001, see also Admiraal, Wetshoff, & de Bot, 2006. In this largescale project, 584 pupils in bilingual schools and 749 pupils in normal schools 411 have been tested twice a year over a period of 5 years. Four cohorts of pupils have been followed over 5 years. There were tests of receptive vocabulary, reading skills and oral proficiency. In addition, there was an extensive questionnaire on language use, attitude, motivation and socio-economic background. The main findings were that the bilingual group outscored the control group for lexical skills (when corrected for general aptitude and beginning level), reading comprehension, oral proficiency, pronunciation, and grades in the final national examination for the subjects taught in English (typically geography, history and mathematics). The most important finding, politically, was that here were no negative effect on Dutch proficiency according to grades in national examinations. The Huibregtse study has been extremely influential because it provided schools with strong arguments to set up a bilingual program and silenced doubts about its effectiveness.
The Huibregtse project looked at some of the early adopters and there is certainly the risk of a Hawthorne effect, as with all innovations: Schools, teachers and pupils feel "special," and may therefore be more motivated and willing to invest extra time and energy in the teaching program. These factors may be the real cause for the advantages mentioned. Therefore it was decided to set up a replication study in the late 2010s, the OTTO (Onderzoek Twee Talig Onderwijs research project on bilingual education) project (Verspoor, de Bot, & van Rein, 2011). The OTTO study used the same instruments as the Huibregtse study but also included a special group of schools, the Reformed schools, in which the pupils for religious reasons have no or very limited access to TV, movies and computer games. Altogether, there were four groups: Religious bilingual Religious mainstream Nonreligious bilingual Nonreligious mainstream The findings of this project are basically the same as those in the Huibregtse project, but in addition show that out-of-school contact plays a crucial role in the development of English: The best results are found in the nonreligious bilingual schools, followed by the religious bilingual, nonreligious mainstream and religious mainstream.

Early English teaching
So the system of bilingual secondary schools was well established and is now moving from the highest school types (preuniversity) to lower types and has shown to be effective there as well. The success of bilingual education in secondary education led to demands by both parents and schools to introduce some form of bilingual education in primary education as well. There has been and still is substantial resistance against the introduction of English teaching in the early grades since there is the common fear that time spent on English goes at the expense of the development of Dutch. Though a wealth of research findings in other countries has shown that the introduction of bilingual education is not harmful for the development of the mother tongue, there appeared to be a need to also show this for the Dutch setting.
One of the big debates with respect to early foreign language teaching has to do with the optional age to start (Muñoz, 2006). While in the layman's view the assumption of the earlier the better still lingers on, there is now no final conclusion on this issue. Also, the terms used in the debate are not always clear. While in European policy documents an early start is at the beginning of primary education, which in many countries is at age 4 or 5, other research, in particular from Spain, seems to take 8 years as an early start. Still, many proponents of a really early start, which in the Netherlands includes both schools and parents, have chosen the first grade to start teaching English. The motives of schools to start with early English are not always purely educational. There is competition between schools and early English is used as a unique selling point to attract more pupils. But up until recently, the empirical support for an early start was lacking. As Nikolov and Mihaljevi Djigunovi (2011) mention in their review of the literature: "No comparative study was found on how weekly exposure impacts on outcomes" (p. 97).

Empirical evidence: Some early projects
Research on the effects of an early start began with a number of small-scale projects reported on in Goorhuis-Brouwer and de Bot (2010). The studies followed three groups of children longitudinally from their entrance in the school till the end of second grade. The Reynel test for language comprehension in Dutch and English was used, and after one year of English the Dutch pupils reached an age equivalent for English monolingual children of 2 years and 5 months and a mean score for language production of 2 years and 1 month. There was normal development of Dutch proficiency: The large majority of the children had scores within the normal standardized range. An interesting finding was that there were no negative findings for non-Dutch children. If anything, they showed similar development in Dutch and were equivalent or better than the Dutch children in English. Numbers are small for this group, so no hard conclusions can be drawn for this particular group. The same holds for children who showed a low score at the beginning and who might have problems in their language development. These children improved somewhat more (but started lower) than the average children in the studies.

The Foreign Languages in Primary Schools project (FLIPP)
As with bilingual education in secondary education, the Dutch ministry of education was keen to monitor at an early stage what the effects of an early start might be. The main worry was that starting early with English might go at the expense of the mother tongue development. The main questions in the FLIPP study (Unsworth, Persson, Prins, & de Bot, 2014) were: 1. How do the English language skills of the children develop over time? 2. What is the relationship between the development of the children's English language skills and the development of Dutch as a first and a second language?
The project was a 2-year longitudinal study with three measurement times: at the beginning of schooling, at the beginning of second grade and the end of second grade (2010/2011/2012). The first factor looked at was the total number of minutes of English per week. The assumption was that the development of children's lexical and grammatical knowledge would be related to amount of exposure. Earlier findings for the Dutch setting showed an improvement after 1 year of 3 hours per week, though there was a stagnation after 2 years (Goorhuis-Brouwer & de Bot, 2010). With two years of 1 hour per week, there was only rudimentary understanding and no productive skills development (Aarts & Ronde, 2006). The second factor was the teachers' proficiency in English. Is a native speaker needed for significant development, or is a fairly high level of proficiency of regular teachers sufficient? Research by Larson-Hall (2008) suggests that differences between native speaker teachers and nonnative speaker teachers become relevant after hundreds of hours of education only.
There were three factors in the design: Early/late start (age 4 or 8/9) English proficiency of teachers (A2-C1) Number of minutes/week of English lessons Many factors are likely to play a role in language development, and in order to measure language development as "pure" as possible, working memory, nonverbal memory, IQ and contact with English were controlled for. The language tests for English included receptive vocabulary, phonology, morphology and syntax, while for Dutch, only vocabulary data have been gathered. The design was as follows: 1  Bishop, 2003) Score: total number of items correctly answered Dutch receptive vocabulary

Main results of FLIPP
The EFLT children improved significantly for both vocabulary and grammar in English. There is a significant effect for the number of minutes of English lessons per week: 60 min or less per week leads to significantly lower scores for English, compared to children with more than 60 min but less than 120 and the children with 120 min or more. The language proficiency of the teacher is a good predictor of the results for vocabulary after 1 year and of the results for grammar after 2 years: Children with a nonnative-speaker teacher of English at level B (CEFR) score significantly lower than children with a native-speaker teacher only or a native-speaker and a nonnative at C level.
The Dutch vocabulary of pupils develops according to age norms. Only a few students' scores fell below age-appropriate norms, and that number becomes even smaller in the course of time. The Pupil Monitoring System data are consistent with this analysis as are data from teachers' perceptions.
EFLT children with a non-Dutch (and non-English) language spoken at home score similarly to age-and Socio-Economic Status (SES)-matched monolingual Dutch-speaking children. The numbers are too small to draw far-reaching conclusions from these results, but they are consistent with earlier findings for the same population by Goorhuis and De Bot (2010).
In terms of age of onset of English instruction, the data show that in the course of 2 years, the later starters make more progress than the early starters, which also corresponds with previous research (e.g., Nikolov, 2009). A real comparison between an early and late start can only be made when children who start with English in Grade 7 can be compared with EFLT children in Grade 8. The control variables (age, SES, contact with English outside school, working memory, IQ) explain comparatively little variance in the results.

CITO 2012 project
The national testing agency (CITO) carried out a longitudinal study with a representative sample of 1,400 pupils in the 8th grade to test English proficiency (CITO, 2012). In order to test the impact of an early start, six bilingual schools with pupils who had been taught English from Grade 1 to 8 were added to sample. The main findings of the study were that pupils in bilingual schools reach higher levels of speaking proficiency, but not of other aspects of proficiency. An interesting additional finding was that attitudes of pupils decline over time: While English is something new and exciting in the first few years, it becomes an ordinary school subject in later years.

Conclusion
The findings so far indicate that an early start with English leads to higher levels of proficiency compared to controls. There are no indications that more English goes at the expense of the development of the mother tongue. There may be additional cognitive advantages, but these have not been tested in this study. One of the problems in the FLIPP study was that schools change rapidly: Some schools that had been labeled as providing 1 hour of English per week by nonnative teachers may the next year decide to have 3 hours with a native speaker teacher. Such changes are normal but problematic for longitudinal research such as this one. But it means that a real evaluation can only be done once the system has settled and stabilized. That may take a couple of years. The positive results found will be important for the development of early English teaching in the same way the Huibregtse study was important for the growth of bilingual education in secondary education.
One important aspect has not been taken into consideration in the studies: What happens in the classroom? It is possible that 1 hour of really good teaching is more effective than 3 hours of mediocre teaching. However, what constitutes good teaching in this context is not so easy to define.
In the near future a number of issues will come up: Will it be possible to find schools that do other languages, like Spanish, Turkish or Chinese?
Will it be possible to have more than one foreign language in primary education, or will English remain the most important and only one? How can we achieve optimal articulation between levels, that is, what happens to the pupils that enter secondary education with a fairly high level of proficiency? Will they be mainstreamed until the other pupils who came with less English have caught up, or will there be accommodation and differentiation in the English lessons? There is a need for improved teacher education. Several teachers felt they were not really equipped to teach English because in their preservice training, English, though officially part of the curriculum, had effectively been neglected. Multilingualism in preschool and day care: A fairly recent development in the Netherlands is the call for more multilingualism, which actually means more English in preschool education and daycare. The official stance is that daycare can only be provided in Dutch, but as with the development of bilingual streams in secondary education and primary education as described above, there is pressure from (generally higher educated) parents to have the option to provide daycare that is done partly in English. One of the problems that will come up is that there is again a problem of articulation between levels in the educational system, this time between preschool and primary education: If some children come into primary education with a relatively high level of proficiency in English, the need of differentiation according to level already starts in Grade 1. On the other hand, as far as there are any data on the very early foreign language learning in an educational setting, they do not seem to point to large gains in proficiency. In a project on teaching English as a foreign language to 3-4year-olds in commercial institutes in China, first indications are that the children appreciate the lessons, but gain very little from them (Sun, Steinkraus, & de Bot, 2014). To what extent an early basis for English will be beneficial in the long term is unclear. The focus of this special issue is on the age-factor. This debate has evolved from the question whether language learners after the critical period are able to acquire a language at native level to the realization that there may be some gifted and very dedicated learners who manage to become fluent enough to pass for natives. For the majority of learners this is neither a goal nor a necessity. In this respect the English as a lingua franca movement is relevant: Most communication in English will increasingly be between speakers of nonnative or nonstandard varieties of British/American English and in that situation the native norm becomes basically irrelevant.
Still the question remains whether an early start with foreign languages in school is a good idea. The findings from the projects reported on here suggest that an early start does not go at the expense of the development of the mother tongue and that there are substantial gains in English proficiency. What the best starting age is cannot be deduced from our data: As mentioned, a real comparison between an early and a late start is only possible when sufficient numbers of learners have gone through the full 8 years of primary education. Even if in that comparison the early starters outperform the late starters it will have to be shown that the investment in time and energy was worth it. If there is only a slight difference, a late start may be preferably in terms of efficiency.
A final point is to what extent our findings are generalizable to other situations/settings/countries. Similarly to the Nordic countries, the language setting in the Netherlands is such that English is becoming a second rather than a foreign language: It is very present in the media and in the linguistic landscape. Contact with English is almost unavoidable. The data from the FLIPP project show that children at the age of 4 already enter primary education with some English they have picked up in various ways. The omnipresence of English and the generally positive attitudes towards the language make the situation in the Netherlands different from countries like Germany, France and Spain, in particular with respect to dubbing or subtitling of foreign movies and TV programs. It is quite likely that there will be an interaction over time between learners' developing proficiency and the use of resources from the environment: Higher proficiency allows for more efficient use of language in the ambiance, and that will enhance proficiency again. The dynamics of that process are still to be explored.