A large number of studies conducted in various countries have shown that teachers in both primary and secondary education often have negative attitudes towards multilingual approaches and that they are reluctant to include students’ native languages in the school environment (Flores & Smith, 2009; Pulinx et al., 2017; Putjata & Koster, 2021), even if they recognize the benefits of speaking more than one language (De Angelis, 2011; Mitits, 2018). Attitudes towards multilingualism have been found to be related to several factors, including the ethnic or linguistic composition of the student population (Bosch et al., 2024; Flores & Smith, 2009; Pulinx et al., 2017) and personal characteristics of the teacher such as language skills, ethnicity and gender (Flores & Smith, 2009; Pulinx et al., 2017; Rinker & Ekinci, under review). Another crucial factor appears to be teacher education, with teachers who have received training on multilingualism, L2 pedagogy or cultural diversity being consistently more positive about multilingual approaches (Alisaari et al., 2019; Bosch et al., 2024; Dursun et al., 2023; Flores & Smith, 2009; Mitits, 2018; Pohlmann-Rother et al., 2023).

Since multilingualism has been relatively high on the educational agenda in recent years, it is likely that an increasing number of teachers have obtained evidence-based knowledge and possibly developed more positive attitudes towards multilingualism. Some recent studies suggest that this is indeed the case. Focusing on Denmark, where multilingual education is a compulsory part of teacher training, Søndergaard Knudsen and colleagues (2021) found that teachers recognized the value of multilingualism and L1 maintenance, as well as their own responsibility for children’s bilingual development. Similarly, teachers in Sweden (Lundberg, 2019) and Finland (Alisaari et al., 2019) displayed relatively positive beliefs about multilingualism and awareness of recent concepts such as translanguaging. Yet, even in these studies, some participants were reluctant to apply multilingual approaches in their teaching.

Related to the current study, and partially based on the same dataset, Bosch and colleagues (2024) investigated the attitudes of primary school teachers in the Netherlands, Italy and Greece, using an adaptation of a survey developed by Pulinx et al. (2017). This study showed great variation in teachers’ attitudes in all three countries, although teachers in Greece were significantly more positive about multilingual approaches. For example, 26 % of the Dutch teachers believed that multilingual students should not be allowed to use their home language at school, compared to 18 % of the Italian teachers and 5 % of the Greek teachers. In contrast, 15 % of the Dutch teachers, 26 % of the Italian teachers and 69 % of the Greek teachers agreed that pupils should have the opportunity to study their home language at school. A substantial proportion of teachers expressed their concerns about the academic development of L2 learners, with 48 % of the Dutch teachers, 41 % of the Italian teachers and 38 % of the Greek teachers sustaining that the most important cause of academic failure is insufficient proficiency in the school language. In the Netherlands, but not in Italy and Greece, there was a positive correlation between teachers’ attitudes and the proportion of multilingual students, suggesting that experience with multilingual students may support openness to multilingual approaches. Moreover, in all three countries, teachers who had received training on multilingualism expressed on average significantly more positive attitudes towards multilingualism in education than teachers who had not received such training.

Similar studies conducted in the Netherlands have focused on primary teacher education students (Robinson-Jones et al., 2022), foreign language teachers in secondary schools (van Beuningen & Polišenská, 2019), and English and Dutch teachers in vocational education (van Batenburg et al., 2022). The results show that many Dutch teachers still believe in common misconceptions about multilingualism and fail to make use of students’ multilingual repertoires. In secondary and vocational education, a majority of language teachers prefer to take a monolingual approach and discourage the use of students’ home languages in class, while at the same time secondary school teachers indicated a need for guidance on language sensitization and functional multilingual language education (van Batenburg et al., 2022; van Beuningen & Polišenská, 2019). Primary education students showed slightly more positive attitudes towards multilingual approaches, but a large majority still believed that Dutch should be the only language used in class (Robinson-Jones et al., 2022).

In sum, teachers’ attitudes towards multilingualism appear to be related to several variables, including the sociopolitical context of the country in which they teach, the type of school, the composition of the student population, and their professional training, gender and personal language background. In the next section, we will discuss how teachers’ attitudes may also be influenced by the relative status of the specific languages involved, which may be determined by so-called ‘language hierarchies’.

While there is a large body of research available on teachers’ attitudes towards multilingualism, very little attention has been paid to how these might be influenced by language hierarchies and social prestige. The limited evidence that is available to date suggests that teachers’ beliefs may indeed be influenced by their attitudes towards specific languages and their relative social status (Putjata & Koster, 2021; Young, 2014).

Taking a qualitative approach, Putjata and Koster (2021) used semi-structured interviews to investigate the language attitudes of teachers working in monolingual and German-Dutch bilingual kindergartens and primary and secondary schools in the German state of North Rhine-Westphalia. Their results showed that teachers in both monolingual and bilingual schools tended to adhere to a rigid monolingual standard, even though all of them believed it was important to speak multiple languages. While teachers working in bilingual schools expressed extremely positive views on multilingualism when speaking about the school’s bilingual curriculum, highlighting the benefits for language learning, they considered mostly negative aspects when discussing multilingualism involving minority languages. For example, some feared that children would use their L1 as a ‘secret language’ to exclude others. A similar double standard was reported by Portolés and Martí (2018), who focused on teachers working in Spanish-English bilingual schools in Spain.

Focusing on teachers in the Alsace area of France, Young (2014) also found evidence of language hierarchies. Using critical discourse analysis on interview data, she observed that the majority of her participants exclusively associated multilingualism with German/French bilingualism of the Alsace region or with English L2 learning, while they generally did not consider minority languages as legitimate languages in the school context. Moreover, many participants sustained that certain languages should be limited to specific times and spaces, emphasizing a strict dichotomy between the classroom and the schoolyard. Even though many teachers argued that children should be free to choose which language to use during play time, they typically believed that French should be the only language used in class, with 52 % mentioning implicit or explicit rules to discourage or prohibit the use of home languages.

Thus, teachers’ attitudes towards multilingualism appear to be influenced by the social status that is assigned to the specific languages involved, with some languages being delegitimized in certain formal contexts. Differences in the social status of languages generally reflect existing political and economic power relations (Appel & Muysken, 2005). The high social and economic evaluation of European languages contrasts with those of immigrant populations, which are often associated with poverty and/or a history of colonial domination (Helot & Young, 2002). The use of minority languages may also be perceived as a barrier to educational and occupational success (Agirdag, 2010; Agirdag et al., 2014) or to the use and development of the majority language (Van Avermaet & Sierens, 2010; Leseman, 2000).

This dual perspective on languages is not without implications for the educational context. For example, Pulinx and colleagues (2017) found a negative correlation between the degree of monolingual thinking among Flemish teachers and the trust they place in students. Low trust is associated with low expectations, which in turn impact the behavior of teachers and students, leading to lower academic outcomes (Rosenthal & Jacobson, 1968) – the so-called Pygmalion or Golem effect. Furthermore, comparing German-Dutch and Turkish-Dutch bilingual children in the Netherlands, Goriot and colleagues (2016) investigated children’s perception of their teachers’ appreciation of their home languages, and whether this perception was related to their executive functioning. Their results showed that German-Dutch bilinguals perceived more appreciation of their home language than Turkish-Dutch bilinguals. Moreover, the difference in perceived appreciation could partly explain group differences in the domain of working memory.

Teachers’ attitudes towards multilingualism and their adherence to language hierarchies may thus indirectly affect the educational outcomes of students who speak a minority language at home. However, more research on the influence of the social status of languages and its effect on teachers’ attitudes is needed, since so far only a few studies have investigated this issue, mostly taking a qualitative approach. By using a survey consisting of open and closed questions, we aim to contribute to our understanding of the role of language hierarchies in teachers’ attitudes towards multilingual practices in education.

This study investigated Dutch primary school teachers’ attitudes towards multilingualism and the use of other languages in the school environment, and whether such beliefs are influenced by the social prestige of the languages involved. Using an online questionnaire, participants were presented with a ‘vignette’, i.e. a hypothetical scenario about which they were asked some questions. The use of vignettes is a well-established method in qualitative research (Duarte & Günther-van der Meij, 2022; Finch, 1987; Steiner et al., 2016), which has certain advantages compared to traditional survey questions. Crucially, vignette questions offer a more realistic and tangible approach, since they are grounded in concrete, real-life scenarios, providing multi-dimensional representations of specific situations (Steiner et al., 2016).

In our vignette questions, participants read a scenario about a child who migrated to the Netherlands three years ago, and who (despite having strong language skills in the L1) appears to have limited vocabulary knowledge in Dutch. To test the effect of language status, we manipulated the name, native language and country of origin of the child: around half of the participants read about Emile from France who spoke French (the high-prestige condition), while the other half read about Mohamed from Pakistan who spoke Urdu (the low-prestige condition). Comparing these two conditions, our aim was to investigate (1) teachers’ concerns regarding the development of Dutch in child L2 learners, (2) what they would advise parents of bilingual children, and (3) to what extent teachers allow, encourage or prohibit the use of home languages in the classroom versus in the schoolyard. We also examined whether teachers’ concerns about L2 development and attitudes towards home language use at school were related to their training on multilingualism and to the proportion of multilingual students in their class.

Hypothesizing that teachers’ attitudes towards multilingualism are influenced by language hierarchies, we expected teachers in the low-prestige condition to be more concerned about the child’s language development, more likely to advise the child’s parents to speak Dutch with their child, and more likely to prohibit the use of the L1 in class and in the schoolyard, compared to teachers in the high-prestige condition. Finally, given previous research partially based on the same dataset (Bosch et al., 2024), we expected teachers who have received training on multilingualism and teachers with more multilingual pupils in their class to be less concerned about the situation described in the scenario and more likely to have positive attitudes towards the use of the L1 at school.

A total of 104 Dutch primary school teachers were included in this study. Participants were recruited online through the networks of teacher education programs, through Facebook groups for teachers, and by approaching schools directly. The sample did not include teachers who worked in international schools.

The mean age of our participants was 44.6 (SD = 12, Range = 21-66) and they had on average 16.9 years of teaching experience (SD = 10.8, Range = 1-43). The large majority of them were women (90.4 %). Most teachers worked in the province of South-Holland (45.2 %), followed by North-Holland (21.2 %), North-Brabant (11.5 %) and Flevoland (9.6 %). Teachers worked in both urban and rural areas (48.1 % worked in a city with more than 100.000 inhabitants). A little over half of the participants (54.8 %) had previously attended a course or workshop on multilingualism, either during formal teacher education or as a form of in-service training.

There was great variation in the number of multilingual students present in the classrooms of the teachers. On average, 39 % of their students spoke another language at home (SD = 35.5 %, range = 0-100 %), and 80.8 % of the teachers indicated that they had experience teaching students who were still learning Dutch. Note that nine teachers (8.7 %) taught in reception classes for newly arrived migrant children (so-called ‘eerste opvangonderwijs’). The most common native languages that were reportedly spoken by the students of our respondents were (Moroccan) Arabic, Turkish, Polish, English, Bulgarian, Spanish, Chinese, Romanian, Somali, Russian and Syrian.

Participants were randomly divided over the two conditions: approximately half of the participants were in the high-prestige condition (N = 53), while the other half were in the low-prestige condition (N = 51). The distribution of variables across the two conditions is shown in Table 1. As can be seen from this table, the two sub-samples were comparable with respect to the relevant background variables. Although there were slightly more participants who had received training on multilingualism in the low prestige-condition, this association was not significant, X²(1, N = 104) = 1.009, p = .315.

Participants completed an online questionnaire on their beliefs about multilingualism and their experience teaching L2 learners, which was implemented in Qualtrics. Completing the full survey lasted approximately 15 to 20 minutes.

In the part of the survey that is relevant to this study, participants were asked to read a scenario about a bilingual child, imagine being the child’s teacher and answer some questions about how they would act in the presented situation. Participants were randomly assigned to either the high-prestige condition (in which the scenario was about a French boy called Emile who speaks French as his L1) or the low-prestige condition (in which the child in question was a Pakistani boy called Mohamed who speaks Urdu as his L1). The choice of these languages was based on their relative social status in the Netherlands: while Urdu is a non-European language that is likely to be associated with immigration and that may elicit negative attitudes in some participants, French is a prestigious, European foreign language taught in schools that is more likely to elicit positive attitudes. Yet, these languages are comparable in the sense that teachers in the Netherlands might realistically encounter them in class, even though neither of them are among the most common languages spoken by immigrant communities in the Netherlands.

Depending on the condition, participants were presented with the following scenario:

After reading the first part of the scenario, participants were asked whether they would be concerned about this situation. Responses were given on a four point Likert scale (not at all – not really – a bit – very much), after which participants were asked to motivate their answers. Using an open question, we then asked teachers what they would advise the parents of the child in this scenario to support their child’s language development. After reading the second part of the scenario, teachers were asked whether they would allow the children to translate to their L1 in class and whether they would allow them to use the L1 in the schoolyard. Again, responses were given on a four-point Likert scale (yes, definitely – more yes than no – more no than yes – absolutely not), followed by a question asking teachers to explain their answers.

Additionally, participants answered several questions regarding their personal background, teaching experience, training on multilingualism, and the characteristics of their school. These background questions were based on a questionnaire developed by Rinker and Ekinci (under review).

The analysis consisted of three parts. First, to examine to what extent teachers would be concerned, and whether teachers would allow the use of the L1 in the classroom and in the schoolyard, we conducted a descriptive statistical analysis by calculating the proportions of responses to the closed questions.

Second, taking a qualitative approach, we analyzed the open questions to examine why teachers would be concerned or not, why they would allow or prohibit the use of the L1 in a certain situation, and what they would advise the parents of the child in question to support their language development. Aiming to detect the most important patterns, we carried out a thematic analysis using a bottom-up inductive approach. The data were coded by the second author, in close collaboration with the first author. We identified several themes, which will be discussed in the next section.

Third, we ran two logistic regression models to test which factors were related to teachers’ concerns and their attitudes towards L1 use at school. For this analysis, ‘teachers’ concerns’ and ‘allowing L1 use’ were dummy-coded as binary dependent variables, by merging the responses ‘a bit’/‘very much’ and ‘more yes than no’/‘yes, definitely’ into the category ‘yes’ (coded as 1) and by merging the responses ‘not really’/‘not at all’ and ‘more no than yes’/‘absolutely not’ into the category ‘no’ (coded as 0). The motivation for using a binary dependent variable in the analyses was based on the fact that even though participants answered the questions on an ordinal scale to decrease the threshold for giving responses that might be socially undesirable and to gain more insight into how confident participants are about their responses, the answers to these questions remain essentially binary (i.e., ‘yes’ or ‘no’).

The first model tested whether the likelihood that teachers are concerned could be predicted by the main and interaction effects of condition (low-prestige, coded as +1/2, or high-prestige, coded as –1/2), training on multilingualism (yes, coded as +1/2, or no, coded as –1/2) and the proportion of multilingual students (number of multilingual students divided by total number of students, centered around the mean). The second model tested whether teachers’ attitudes towards L1 use at school could be predicted by the main and interaction effects of condition (low-prestige, coded as +1/2, or high-prestige, coded as –1/2), setting (in class, coded as +1/2, or in the schoolyard, coded as –1/2), training on multilingualism (yes, coded as +1/2, or no, coded as –1/2) and the proportion of multilingual students (centered around the mean). We conducted a top-down stepwise model comparison based on the Akaike Information Criterion to select the models with the best fit. The statistical analysis was conducted in R (R Core Team, 2023). Two participants were excluded from this part of the analysis, because of missing data for the questions that were used to calculate the proportion of multilingual students. The regression analysis was therefore based on 102 participants (53 in the high-prestige condition and 49 in the low-prestige condition).

The responses to the question ‘Would you be concerned?’ are summarized in Table 2. As can be seen from this table, approximately 2/3 of the teachers in our study would be concerned at least to some degree about the child described in the scenario (62.3 % in the high-prestige condition and 66.7 % in the low-prestige condition).

However, the difference between the two conditions was not significant (odds ratio = 1.14, 95 % CI = .498-2.63, p = .755). Since ‘training on multilingualism’ did not contribute significantly to the model fit, the best model was obtained by only including ‘proportion of multilingual students’ as a predictor. This model showed that teachers who have more multilingual students in their class were less likely to be concerned about the child in the scenario (odds ratio = .664, 95 % CI = .438–.997, p = .0498).

When asked why they would be concerned or not, 47 teachers (45.2 %) mentioned that the child in the scenario has only been in the Netherlands for a few years and that children need more time to learn a language and develop their vocabulary knowledge. For 24 of them (23.1 %) this was a reason not to worry, while the others were nevertheless a bit concerned. At the same time, 17 teachers (16.3 %) emphasized that three years of language exposure is a long time already and that, according to them, the child should have achieved a higher level of Dutch by now (e.g., “after three years he should speak Dutch fluently, in addition to Urdu”).

Moreover, 17 teachers (16.3 %) pointed out that the child in the scenario has strong language skills in the L1. For 8 of them (7.7 %) this was a reason not to be concerned. For example, one respondent mentioned that “when the mother tongue is well developed, the second language usually follows automatically – this may just take a lot of time”. In contrast, for 8 others (7.7 %) the pupil’s good L1 skills were all the more reason for concern. According to these teachers, the Dutch language knowledge of the child is expected to be at a good level after three years, especially considering his strong L1 skills.

When asked what they would advise the parents in the scenario to support their child’s language development, 22 respondents (21 %) mentioned advice regarding language use at home. Half of these respondents would advise parents to speak more Dutch with their child (e.g., “Try to speak Dutch at home, too”), while the other half would explicitly advise parents to keep using the home language (e.g., “Parents should above all continue to use the mother tongue with him”). In the high-prestige condition, 6 respondents would recommend to use Dutch and 3 would recommend to use the L1, whereas in the low-prestige condition, 5 respondents would recommend to use Dutch and 8 respondents would recommend to keep using the L1.

Additionally, a majority of teachers would advise the parents to stimulate reading in Dutch (59 %), and to have the child watch Dutch television or films (54.3 %). Another common advice, provided by 38.1 % of the teachers, was to increase the child’s exposure to Dutch by engaging in extra-curricular activities in Dutch or by establishing contacts with Dutch-speaking children outside of school. These recommendations were more or less equally common in the two conditions.

Table 3 and Table 4 summarize the responses to the questions regarding teachers’ attitudes towards L1 use during class and in the schoolyard, respectively. These results show a lot of disagreement: both in class and in the schoolyard, approximately half of our respondents would allow the use of the L1, while the other half would not.

The likelihood of teachers allowing L1 use was not significantly influenced by condition (high-prestige versus low-prestige, odds ratio = 1.18, 95 % CI = .641-2.16, p = .598), setting (classroom versus schoolyard, odds ratio = 1.32, 95 % CI = .728-2.43, p = .358) or the interaction between them (odds ratio = 1.76, 95 % CI = .530-5.88, p = .358). The best-fitting model included only the main effects of training on multilingualism and proportion of multilingual students. Teachers who had received training on multilingualism were estimated to be more than two times more likely to allow L1 use at school than teachers who had not received such training (odds ratio = 2.05, 95 % CI = 1.07-3.91, p = .029). The likelihood of a teacher accepting L1 use was also positively related to the proportion of multilingual students in a class, with teachers in more linguistically diverse environments being more likely to allow L1 use (odds ratio = 1.95, 95 % CI = 1.39-2.77, p = .0001).

While many teachers prefer to avoid the use of the L1 in class for translation purposes, because it is considered to be detrimental to the development of the Dutch language, most of them do recognize the practical advantages of it. For example, they argue that in some cases it may help to improve the pupil’s comprehension of the content of a lesson or to avoid misunderstandings. However, these considerations are often seen as a last resort. One teacher explains: “The child should extract as much information from the [Dutch] context as possible. If the assignment is not understood from this context, a classmate can translate the most important contents”. Another teacher proposes that L1 translations should always serve the purpose of Dutch language development: “If L1 use is functional and contributes to the development of Dutch yes – otherwise no”. Moreover, 21.2 % of the teachers mention that they would only allow L1 use after the student has tried using Dutch first, and they would always offer the Dutch translation together with the explanation in the L1.

With respect to teachers’ opinions about the use of the L1 in the schoolyard, our respondents gave various motivations. An important reason for allowing or encouraging L1 use, provided by 25 % of the teachers, was that it is beneficial for relaxation and well-being, or that it can give children a sense of safety, belonging and self-confidence. Although 11.5 % of the teachers believe it is important that Dutch is spoken in the schoolyard in order to promote social contacts with Dutch-speaking children (“It is desirable that he can use the language well enough to develop friendships with children from his class”), 3 teachers (2.9 %) argue that the use of the L1 is also conducive to social contact (“For establishing contact with peers it is nice if he develops communicative skills both in Urdu and in Dutch”). Both supporters and opponents of L1 use (20 teachers, 19.2 %) claimed to ensure that students do not isolate themselves from the group, and that others are not excluded. For some, that risk is a reason not to allow L1 use, while for others, L1 use is not a problem, provided no one is excluded. One respondent states that they would allow L1 use in the schoolyard, “depending on whether it helps to have contact with other children, or if it keeps them away from them”. In contrast, 12.5 % of the teachers think that the use of the home language in the schoolyard is desirable only occasionally, for example if a child doesn’t understand something or if they need to be comforted. Finally, 4 respondents (3.8 %) rely on an explicit or implicit school rule that prohibits the use of the L1 (e.g., “At a Dutch school we speak Dutch”).