Making the most of language acquisition of Syrian asylum permit holders in the Netherlands: the role of policy factors examined

Abstract

In this article, we examine the relationship between important types of policies for asylum permit holders in the Netherlands and the improvement in their command of Dutch. As far as asylum policy is concerned, we find that participation in activities in the asylum seekers reception centre – and in particular, following Dutch language classes – contribute to an improvement in Syrian asylum permit holders’ command of Dutch. On the other hand, a prolonged period of stay and frequent relocations between reception centres are not favourable. Asylum permit holders who have successfully completed the civic integration programme have a better command of the language than asylum permit holders who are still undergoing the programme. An important finding is that there seems to be a sort of double deficit in the area of civic integration: not only do the elderly and lower educated make less progress in learning Dutch, but they are also the ones more likely to receive a dispensation from the civic integration requirement, which places them at a further disadvantage. Third, we find that early participation in the labour market or as a volunteer is also beneficial for language proficiency.

Introduction

Mastering the language of the country of destination is essential for the successful integration of migrants. This constitutes an important form of post-migration human capital. Having a good command of the host country language increases the chances of obtaining paid employment (Aldashev et al., 2009; Dustmann & Fabbri, 2003; Dagevos & Odé, 2011; De Vroome & Van Tubergen, 2010; Chiswick & Miller, 2009; Martin et al., 2016; Zimmermann, 2005). When asked what integration means to them, asylum permit holders most often mention mastering of the Dutch language (Damen et al., 2019). According to them, language is the key to integration; without a command of Dutch, it is difficult to form social contacts and find work. Asylum permit holders have high expectations regarding the significance of acquiring Dutch language skills.

Previous research already showed that a set of individual factors are related to second language proficiency. Men, young people, highly educated, and healthy asylum permit holders have a better command of the host country language than women, older people, less educated asylum permit holders, and those suffering from poor health (Beiser & Hou, 2000; Cheung & Phillimore, 2017; Chiswick et al., 2008; Chiswick & Miller, 2001; Dagevos & Odé, 2011; Van Tubergen, 2010; Van Tubergen & Kalmijn, 2005; Pottie et al., 2008). Geurts and Lubbers (2017) already showed a positive association of the intention to stay in the host country on language proficiency. Besides these individual factors, we argue that policy factors may also affect the host country language proficiency of asylum permit holders. In this article we therefore focus on the significance of Dutch policy for the ability of Syrian asylum permit holders, who received a residence permit between 2014 and 2016, to improve their command of Dutch. Since 2014, Syrian asylum permit holders are by far the largest migrant group who have come to live in the Netherlands as refugees. This paper is unique since the impact of policy factors on language proficiency of asylum permit holders has not yet been studied in the Dutch context. In addition, this is the first paper in the Netherlands that addresses this issue using longitudinal data, which enables us to study not only correlation, but also causation.

This focus on policy factors is important and a contribution to the field because it enables governments and policymakers to improve evidence-based policymaking. This article also aligns to some extent with literature on the civic turn of integration policies (Borevi et al., 2017). Much of this literature is concerned with the normative aspects of civic integration programmes and naturalisation policies, but relatively little is known about how these kinds of polices influence the integration of migrants, something that is particularly relevant for the impact of civic integration programmes (Mouritsen et al., 2019).

In the Netherlands, three types of policy are relevant to the language acquisition of asylum permit holders: asylum policy, civic integration policy, and labour market policies. Asylum policy can influence the extent to which asylum permit holders learn Dutch because the way the asylum policy is implemented affects the length of stay in the reception centre and the number of times asylum seekers are forced to relocate to other reception locations as well as the opportunities available at reception centres to learn Dutch and engage in other activities.

Second, the civic integration policy is primarily intended to support and accelerate the language development process of asylum permit holders. The civic integration policy ¹ obliges asylum permit holders to attain at least A2-level ² mastery of Dutch within 3 years.

Third, in response to the influx of asylum seekers in 2015, an influential policy brief in the Netherlands argued in favour of combining different forms of participation, such as work and learning the Dutch language (Engbersen et al., 2015). The underlying idea behind this is not only to encourage participation in the labour market from an early stage but also to combine various forms of social participation, which are assumed to promote the learning of the Dutch language. Therefore, in this article we also examine the relationship between social participation and learning the Dutch language.

This article focuses on answering the following research question: To what extent does policy contribute towards improving the language proficiency of Syrian asylum permit holders in the Netherlands?

Policy for improving asylum permit holders’ command of the Dutch language: what are the expected results?

Relationship between the asylum policy and improvement of Dutch language proficiency

As in most European countries, asylum seekers in the Netherlands are placed in reception centres whilst awaiting a decision on whether or not there asylum request will be granted. The way the asylum policy is implemented determines how long asylum seekers spend in a reception centre. It is known from earlier research that the length of stay influences the early integration processes. Less research has been carried out on the relationship between the number of times asylum seekers have had to relocate between reception centres and their subsequent integration. We also focus on the opportunities for participation during the reception period, and investigate the extent to which this influences their acquisition of Dutch language.

There is an extensive body of research literature on the relationship between the time spent in the reception centre and the socio-economic integration of asylum permit holders in the host country. This clearly shows that the length of the stay in the asylum seekers reception centre has a negative correlation with various integration indicators. A longer stay has negative consequences for the mental health of asylum permit holders (Hainmueller et al., 2016; Bakker et al., 2014; Ghorashi, 2005; Laban et al., 2004; Phillimore, 2011), which possibly has follow-on effects on various aspects of social participation, including language proficiency (Damen et al., 2022; Hvidtfeldt et al., 2018). Research among recently arrived asylum permit holders shows that it is not so much the length of time spent in the reception centre that is important for mental health, but rather the number of times they have moved from one to another reception centre prior to receiving a permit holder status (Weeda et al., 2019; Wijga et al., 2019; Van der Linden et al., 2022). Having to move several times from one reception centre to another can create uncertainty and mental health problems. All in all this can hamper the attainment of the Dutch language. Based on the above, asylum permit holders who spent a longer time in the reception centre and who have been relocated more often between different reception centres can therefore be expected to make slower progress in mastering the Dutch language. The reasons for moving to another reception centre can be diverse. The biggest factor was the high influx of refugees in the period around 2015. During the peak of the influx of refugees many ‘emergency locations’ where opened, with minimum facilities. The aim was to relocate people to regular reception centres as soon as possible.

Another development in recent years in the Dutch asylum policy is that there are more opportunities to engage in various activities during the period of reception, such as learning Dutch and doing volunteer work. These programmes (for learning the language and volunteer work) at asylum seekers reception centres are intended to prepare asylum seekers and asylum permit holders better for their integration in the Netherlands by giving them an opportunity to gainfully use the time spent in the reception centre. We argue that it is not a matter of personal choice (selection), but due to the availability of these programmes (policy) that makes participation in these activities possible.

There is a specific programme for asylum permit holders who are still living in the asylum seekers reception centre that aims to prepare them for their future life and civic integration within the municipality (Preparatory Civic Integration Programme (Voorinburgering)). This programme is conducted over a 14-week period and includes Dutch language classes, basic knowledge about Dutch society and the labour market, and personal counselling. The evaluation of Preparatory Civic Integration Programme (Bakker et al., 2020) shows that the rate of participation is around 70%. Just over half (55% in 2019) of the participants succeed in achieving language level A1 via this programme, which is the minimum targeted language level in the programme. However, this study also showed that only about 40 out of 160 reception centres offered the Voorinburgering programme. Consequently not all asylum permit holders had the opportunity to engage in this Voorinburgering programme. This depends on the reception centre that they stayed in. It is thus primarily differences in the Preparatory Civic Integration Programme which determine the probability of people in reception centres being able to follow Dutch lessons, and not so much their own preferences. The latter could indicate the presence of selection effects, making it more difficult to say anything about policy effects. Therefore we argue that it is interesting to see whether participation in language classes during the stay in the reception centre has an effect on second language proficiency later on.

Using the available longitudinal data, we investigate the expectation that asylum permit holders who participated in these activities (language classes, voluntary work, paid work, training) made more progress in mastering the Dutch language than those who did not participate.

Relationship between civic integration policy and improvement of Dutch language proficiency

Persons who came to the Netherlands as refugees and have been granted residency status are obliged to complete a civic integration programme within 3 years. They begin the programme once they have moved out of reception accommodation and are living in a municipality. The Dutch civic integration policy has been in place since 1998, and despite various changes in the system since then, learning the Dutch language has always been the primary objective. All asylum permit holders are required to pass the civic integration examination at language level A2 (of the Common European Framework of Reference for Languages) within a period of 3 years. ³ When they do not successfully pass the exam this has consequences for their stay in the Netherlands: without passing the civic integration exam they cannot request permanent stay in the Netherlands (after 5 years of stay). This shows the importance and impact of the Dutch civic integration policy. A social loan (from the government) is available to finance civic integration programmes (including language classes) which are provided by commercial enterprises. This enables all permit holders to follow such civic integration programmes.

Relationship between participation and improvement of Dutch language proficiency

Since the increased influx of asylum seekers in 2015, the Netherlands has outlined policies to increase the labour market participation of asylum permit holders. Participation, it is assumed, will not only lead to financial self-sufficiency, but will also strengthen integration in other domains. Working and other forms of participation (such as volunteering) are expected to help in learning the Dutch language (Van Niejenhuis et al., 2015; CPB/SCP, 2020). We expect that asylum permit holders who seek work, do paid or voluntary work will succeed in improving their language skills more so than those who do not.

Methods and analysis

The analysis was based on two waves of the survey ‘New Permitholders in the Netherlands’ (NSN2017 and NSN2019). ⁴ The survey was conducted at the request of four Dutch ministries, aiming to gain insight into early integration among refugees in the Netherlands. The first wave of the NSN was collected in 2017 ⁵ among Syrians aged 15 and older who received a (temporary) residence permit between January 1st, 2014 and July 1st, 2016. Family members who reunited in 2014/2015 also belong to the target population. A single random cluster sample was drawn from the target population by Statistics Netherlands. The clustering was based on municipalities. A cluster comprised five persons in small municipalities (fewer than 50,000 residents) and ten persons in larger municipalities (more than 50,000 residents). In the five largest cities in the Netherlands, a sample of Syrians was drawn which was in proportion to the number of Syrians living in those municipalities. Ultimately, all individuals from the target population have the same chance of being included in the sample.

The questionnaire was tested thoroughly and translated into Modern Standard Arabic. A sequential mixed-mode survey design was used; respondents were first invited to complete the survey online (CAWI) but if they didn’t, they were given the opportunity to complete the survey in person with an interviewer (CAPI). In case of no response, interviewers would visit respondents up to four times to make an appointment ⁶ . All interviewers spoke Arabic and were from the same origin country as the respondents. The second wave of the survey was collected in 2019, and a similar approach was taken.

In total, 3,209 Syrians completed the first survey in 2017, corresponding to a response rate of 81%. Statistics Netherlands was able to provide data for 2,944 people in 2019 and 2,544 participated in the second survey, resulting in a response rate of 86%. The high response rates can partly be attributed to the personal and repetitive approach, but it also shows people were eager to provide their input. The survey files were weighted by Statistics Netherlands to match the distribution in the sample with that in the population. The sampling, extensive fieldwork, bilingual interviewers, high response rate, and weighting of the data resulted in a unique and high-quality dataset. Moreover, the survey data was enriched with register data from Statistics Netherlands. For this study, we made use of a balanced panel, meaning that only respondents who participated in both waves were included in our analysis (N = 2,141). Additional comparison between all the respondents who took part in wave 1 to respondents who took part in wave 2 shows that there are no significant differences between the dependent variable regarding Dutch language acquisition. ⁷

Measurement and analysis

We measure the extent of language improvement based on the difference in the language score assigned by the asylum permit holders themselves (on (Linden, 2022) a scale of 1–10) at the two measurement moments (2017 and 2019). The respondents are asked ‘How well do you speak the Dutch language? 1 means that I do not speak Dutch, 10 means that I speak Dutch very well’. This means that we are using a self-reported language score. Within the context of the survey, it was not possible to carry out an extensive language test. A possible disadvantage of self-reporting is that people may be unable to properly assess their own language skills and that the social desirability bias therefore influences the assessment. However, we have great confidence in the validity of this measurement due to several reasons. Firstly, self-reported measurements have also been used in other surveys involving migrants, with equally plausible results (see e.g. Bernhard & Bernhard, 2021; Van Tubergen, 2010). ⁸ In some of these surveys, it was possible to relate the interviewer’s assessment of the respondent’s command of Dutch to the respondent’s own estimation. There is quite a high correlation in such cases (0.66 in a study involving refugee groups, see Dagevos & Odé, 2011). In addition, the differences in the reported language proficiency in our sample between younger and older participants, the higher and lower educated, and asylum permit holders with different lengths of stay are in line with expectations (see Miltenburg & Dagevos, 2020). The interviews were conducted by persons with a Syrian background, rather than by Dutch interviewers or government officials, thus we do not expect social desirability to play a significant role. Another reason for using this variable is that we are interested in the individual progress in Dutch language proficiency, based on the difference in the language score between the two measurement moments. This involves an assessment by the same person at two measurement moments. We assume that within-subject reliability is fairly high here. Any subjective differences between respondents or groups of respondents in their estimated language proficiency are less important for the analyses. Therefore, the disadvantage of a subjective measurement of language proficiency seems to be limited. We also note that there is actually enough variation in the development of language proficiency over time between 2017 and 2019 to use hybrid models.

We examine whether the possible links between aforementioned factors related to policy and language improvement are unrelated to other factors or whether they are dependent on other determinants. We do this by carefully setting up the panel regression analysis model, step by step. Before we explain the set-up of this model, it is important to mention that, as in the case of the dependent variable, some of the independent variables also change over time (i.e. between the two measurement moments).

At the time of the survey, none or almost none of the respondents were living in the reception centre. The activities respondents undertook during their stay in the reception centre have been reported in the survey retrospectively, during the first wave of the survey in 2017. The measurements relate to length of stay in the asylum seekers reception centre ⁹ (in years), the number of relocations (the number of reception centres the asylum permit holder stayed in during the reception period: 0–9 centres), participation in language classes in the asylum seekers reception centre (dummy variable, yes = 1), and participation in another activity in the asylum seekers reception centre ¹⁰ (such as training, paid or voluntary work) (dummy variable, yes = 1). It is assumed that the participation characteristics may vary between the two measurement moments.

The ‘Civic integration status’ variable is measured based on four categories: (1) successfully completed the civic integration examination at some point in time (2017–2019) ¹¹ , (2) granted a dispensation at some point in time (2017–2019), (3) not subject to a civic integration requirement, and (4) subject to a civic integration requirement at both measurement moments and still undergoing the civic integration programme (= reference category). The category ‘dispensation’ (in Dutch: ontheffing) consists of asylum permit holders who have received a dispensation from the Dutch authorities. This type of dispensation is granted from the Dutch authorities to asylum permit holders who cannot pass the examination for example because of chronic bad health or have been granted a dispensation after they have already unsuccessfully tried multiple times to pass the examinations. The most likely to be granted dispensation are older and less-educated persons. Those not included in the target group of the civic integration policy (category 3) are often students and pupils in education.

We include ‘Active in the labour market’ as a dichotomous variable, where an asylum permit holder was active (either working or looking for work) at some point in time (2017–2019). The reference category consists of asylum permit holders who have never been active in the Dutch labour market. The ‘Active in voluntary work’ variable indicates whether asylum permit holders have volunteered at any point in time, compared to asylum permit holders who have never volunteered. Please note here that we build on previous research on the (change in) language proficiency of various migrant groups that has also included similar measurements such as work search intentions (see Asfar et al., 2019), language course participation (Bernhard & Bernhard, 2021), and having followed an integration course (Geurts & Lubbers, 2017).

See Table 1 for the descriptive statistics of the variables in the explanatory analyses.

Table 1. Descriptive statistics of the variables in the explanatory analyses, 2017–2019 (weighted, balanced panel, N = 2141)

Source: The Netherlands Institute for Social Research (SCP)/Statistics Netherlands (CBS) (NSN ‘17 and NSN ‘19 enriched with register data), weighted data.

^rd= variable from register data.

Analysis: hybrid model and additional random effects model

We use a hybrid model so that we can make optimal use of the panel nature of the data (Allison, 2009). A hybrid model is a combination of a fixed effects model (FE model) and a random effects model (RE model). Unlike in an FE model, a hybrid model has the advantage of consistently estimating both the effects of time-variant factors (such as completion of the civic integration programme) and time-invariant factors (Schunck & Perales, 2017). The hybrid model estimates the within-subject effects for the time-variant factors and the random effects (a combination of within-subject and between-subject effects) for the time-invariant factors. The within-subject effects display the effects of change over a period of time. The random effects for the time-invariant factors can be interpreted as differences between individuals in the outcome measure. ¹² Table 2 shows the results.

The within-subject effects based on the FE approach allow for the best causal interpretations. At the same time, the FE approach is a ‘rigorous’ way of establishing associations of time-variant factors with the outcome measure. This approach obviously depends on the moments in time when the respondents were studied. The FE approach only takes into account changes that occur between Wave 1 and Wave 2. Therefore, if the changes occurred before Wave 1 (2017) or after Wave 2 (2019), they are not taken into consideration. The FE approach does not take into account respondents who show no changes with regard to the dependent variable between the observed moments in time. Given this ‘rigorous’ approach, the absence of a significant fixed effect does not necessarily mean that there are no significant associations between the time-variant variables and the outcome measure. That is why we have estimated an additional RE model (please see Table 3). Although an RE model makes it less easy to make causal inference, it still provides information on whether significant differences exist between individuals. We use the additional RE model mainly to supplement the interpretation of time-variant variables in cases where there is no significant effect of time-variant factors in the hybrid model. A comparison with the findings of the additional RE model indicates whether there are any significant differences between individuals, in which case we can speak of significant associations, rather than causal inference (or causal relationships).

Results

Table 2 shows the results of the hybrid model to explain this language improvement. ¹³ In Table 2, Model 1 only takes into account factors related to policy. Table 2, Model 2 includes individual characteristics as well: gender, age, age at arrival ¹⁴ , education in country of origin or other foreign country ¹⁵ , length of stay in the Netherlands ¹⁶ and intention of taking up residence in the Netherlands ¹⁷ and ‘Perceived health status’ (as an indicator of both physical and mental health).

In Step 3, we add the variables relating to social contacts in the Netherlands and whether asylum permit holders feel at home here. The assumption is that both variables have a positive association with improved Dutch language proficiency. ‘Contacts with Dutch neighbours’ and ‘Contacts with Dutch friends’ are both included as four dummy variables. In cases where the associations between factors related to policy and language improvement (Table 2, Model 3) remain significant, we can conclude that these associations exist even after checking for the influence of other characteristics.

Table 2. Explanatory analyses of the language improvement of Syrian asylum permit holders in the Netherlands, 2017–2019 (hybrid model, balanced panel, standard errors in brackets)

Source: The Netherlands Institute for Social Research (SCP)/Statistics Netherlands (CBS) (NSN ‘17 and NSN ‘19 enriched with register data), unweighted data.

^r= RE estimator, ^w= FE estimator.

***p < 0.01, **p < 0.05, *p < 0.1.

Longer stay in reception centres and more frequent relocations interfere with the ability to master the Dutch language

In line with our expectations, we found that asylum permit holders with longer stay in the asylum seekers reception centres had a lower language score than those with shorter stay. The number of relocations between reception centres is also negatively associated to the language score of asylum permit holders.

Participation in activities in the asylum seekers reception centre contributes to language proficiency

The hybrid model shows that participation in language classes in the asylum seekers reception centre helps improve asylum permits holders’ command of Dutch. Participation in other activities in the asylum seekers reception centre, such as voluntary work, paid work or a training course, also contributes to this, although this association is not as strong. ¹⁸ The impact of the activities during the time spent in the reception centre stands on its own and is therefore not dependent on the impact of other characteristics. These findings confirm the importance of meaningful activities during the stay in the reception centre, particularly the following of language classes.

No causal effect between completion of the civic integration exam and language proficiency

Contrary to expectations, the hybrid model does not allow us to conclude that successful completion of the civic integration examination contributes to improving the language proficiency of asylum permit holders. After the addition of the background characteristics of asylum permit holders (in Step 2), the initial significant effect of completing the civic integration disappears. This concerns the within-subject coefficient, which means that persons who have passed their civic integration exam during this period do not show a greater improvement in their language proficiency level than persons who did not pass the exam.

As mentioned earlier, the analysis conducted here is a rigorous test for establishing causal relationships between time-variant factors (such as completing the civic integration examination) and improvement of language acquisition over time. That is why we also look at the results of the additional RE model (see Table 3). This shows that there is a positive association between completing the civic integration examination and command of the Dutch language: asylum permit holders who have successfully completed the civic integration programme have a better grasp of the language than asylum permit holders who are still undergoing the civic integration programme. Note that the RE model is not about change, therefore we speak of language proficiency instead of language improvement here.

Table 3. Additional explanatory analysis of the language proficiency of Syrian asylum permit holders in the Netherlands, 2017–2019 (random effects model (RE), unweighted balanced panel, standard errors in brackets)

Source: The Netherlands Institute for Social Research (SCP)/Statistics Netherlands (CBS) (NSN ‘17 and NSN ‘19 enriched with register data), unweighted data.

***p < 0.01, **p < 0.05, *p < 0.1.

Lag in language improvement among asylum permit holders who receive a dispensation from the civic integration requirement

As already mentioned in the introduction, participants of civic integration programmes are required to pass the civic integration examination within a period of 3 years. Some participants of the civic integration programme receive a dispensation due to poor health or if candidates have repeatedly taken the civic integration examination, but were unable to successfully pass the examination. The analysis shows that the language skills of asylum permit holders who have such a dispensation (often older people and less educated people) improved less than that of the group participating in the civic integration programme. This analysis therefore shows that there is a significant negative association between the dispensation and gaining a command of the language. Not just the dispensation itself, but also the personal characteristics of the asylum permit holders who receive such a dispensation, have a negative association with the improvement of Dutch language skills.

Asylum permit holders who actively participate make more progress in mastering the Dutch language

The analyses in Table 2 show that asylum permit holders who were active in the labour market (either had a paid job or were looking for work) at one or both measurement moments (2017–2019) improved their command of the language more than asylum permit holders who were not similarly active. This is a within-subject effect: those who became active in the labour market between 2017 and 2019 improved their Dutch language skills more (by 0.37 points) than those who did not. We did a robustness analysis by using also a pooled OLS model and FE model (in addition to the hybrid model), in which ‘look for work’ and ‘actually work’ are included as separate dummy variables (Table A1, see the Appendix). The model has shown itself to be stable. We also find a positive association in the case of volunteer work. Asylum permit holders who became active in volunteer work between 2017 and 2019 make more progress in terms of language than those who were not involved in any volunteering activity at either measurement moment.

These findings stress the importance of early participation in the labour market or in other activities. Previous research has shown that almost all those who were active in the labour market in 2017 were combining this with learning Dutch (see Miltenburg et al., 2019). This finding confirms the assumption that participation and learning Dutch reinforce each other.

Apart from the contribution of policy-related factors to the language improvement of asylum permit holders the findings illustrate the importance of human capital: asylum permit holders who are young, highly educated and healthy are capable of making more progress in learning the language. The length of stay in the Netherlands is also a key factor.

Conclusion and discussion

In this article, we have examined the relationship between important types of policies for asylum permit holders and the improvement in their command of Dutch. As far as asylum policy is concerned, we find that participation in activities in the asylum seekers reception centre – and in particular, following Dutch language classes – contribute to an improvement in Syrian asylum permit holders’ command of Dutch. On the other hand, a prolonged period of stay and frequent relocations between reception centres are not favourable.

With regard to civic integration policy we find that there is no causal relationship with language improvement, but there is a positive association. Asylum permit holders who have successfully completed the civic integration programme have a better command of the language than asylum permit holders who are still undergoing the programme. An important finding is that there seems to be a sort of double deficit in the area of civic integration: not only do the elderly and lower educated make less progress in learning Dutch, but they are also the ones more likely to receive a dispensation from the civic integration requirement, which places them at a further disadvantage.

Third, we find that early participation in the labour market or as a volunteer is also beneficial for language proficiency.

Relevance of these findings for civic integration and asylum policies in other countries

What lessons can be drawn from these Dutch findings for integration and asylum policies in other countries? Various European countries have some form of civic integration policy in place, although these differ in terms of design and implementation. Our study in the Netherlands aligns in certain aspects with that of Mulvey (2018) and Calo et al. (2022) who studied to what extent policies implemented in the UK in the period 2000 and 2014 fostered or hindered the integration of refugees. We believe that our findings contain relevant lessons for civic integration policies in other countries. While our research shows that we must be cautious about making causal statements about the completion of the civic integration programme, it also reveals that there are distinct consequences if some people are excluded from the policy at any given time. This is certainly true for the Netherlands, because this exclusion often involves groups that have difficulty learning the language even under normal circumstances; hence, it is particularly important to involve this particular group in the civic integration policy, as is the intention under the new Dutch civic integration policy. Another lesson relates to opportunities for further research. Comparative studies on differences in outcomes could shed more light on the effective mechanisms of different forms of civic integration policy. The methodology used in this study, i.e. panel analyses, could possibly serve as an example. This kind of research is important in enabling good policy choices to be made. It is evident from the literature on the civic turn in integration policy that little is known about the effectiveness of this policy and that the choices made by policymakers are based principally on ideological motivations and expectations with regard to turning immigrants into ‘good citizens’. The most important factor, however, is whether such policy matters in terms of language proficiency, work, social contacts, and feeling at home in the host country.

We also argue that our findings regarding reception facilities could be relevant for other countries. Reception facilities exist in many European countries, and although there are differences in the organisation and regulations, it is plausible that mechanisms of length of stay, relocations, and activities also play a role in other countries (see Bakker et al., 2016; CPB/SCP, 2020).

Past research in various countries has shown that an extended period of stay in the reception centre gets asylum permit holders off to a poor start. This has consequences for their mental health and further integration opportunities, such as mastering the Dutch language. To our knowledge, far less research has been conducted on the relationship between participation in activities at reception centres and early integration processes. There are good reasons to assume that the mechanisms we have identified here are also applicable to the reception of asylum seekers in other countries. A short and active reception period with few relocations helps asylum permit holders get off to a good start in their country of destination.

Finally, our results highlight the large variation within the group of asylum permit holders and the corresponding difference in the pace at which they learn Dutch. These differences in composition will also exist in other countries and this indicates the importance of differentiated policy based on the capacities and aspirations of asylum permit holders. There is no one-size-fits-all approach. The conclusion that we have drawn for the Netherlands – a differentiated group requires a differentiated policy – also seems valid for policies in other countries.

Successful policy focuses on more than just the asylum permit holders

Partly due to its causal nature, this article sheds new light on the determinants behind asylum permit holders’ acquisition of the Dutch language, thereby offering important focus areas for policy formulation both in the Netherlands and abroad. The importance of a proper command of Dutch for the successful integration of asylum permit holders cannot be underestimated. At the same time, Ghorashi and Van Tilburg (2006) point out that participation opportunities for asylum permit holders in the Netherlands are limited due to exclusionary tendencies. A one-sided emphasis in the integration policy on language and education ignores the negative discourse concerning migrants and the barriers they face (see also Ponzoni et al., 2017). Integration policy must therefore encompass a broader spectrum of policies than only those aimed at encouraging Dutch language proficiency. It is also important to eliminate forms of exclusion and discrimination. The onus of becoming a valuable member of society cannot be laid on the asylum permit holder alone – they also must be provided with the necessary opportunities for doing so.