Teachers ’ beliefs about multilingualism in a course on translanguaging

In recent years new ideas about multilingualism and translanguaging have been widely debated in applied linguistics and sociolinguistics. A growing number of studies have investigated different aspects of translanguaging. Thus far only a few studies have focused on teachers ’ beliefs, attitudes or ideologies. In an earlier study, we found monolingual assumptions to be strong among teachers in the Basque Country and in Friesland. In the current study in-service teachers ’ beliefs about multilingualism and translanguaging are discussed in relation to the key role that teachers can play in changing educational practices. During a course of continuing professional development in-service teachers received training on multilingual approaches. Before, during and after the course data were collected on their beliefs about multilingualism and translanguaging. The outcomes reveal some important changes in the teachers ’ beliefs about separating languages, mixing languages and languages supporting each other and the application of those beliefs in the classroom. The complex relationship between professional development and changes in teachers ’ beliefs and practices is placed in a broader context of multilingual approaches to teaching. © 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
This article reports on the outcomes of a study into the beliefs of three groups of in-service teachers who took a professional development training course in multilingualism, translanguaging and an integrated language curriculum. During the course the teachers were informed about various recent approaches, ideas and classroom practices and the aim of the course was that these teachers would accept and apply those ideas and practices in their own classrooms. As researchers involved in the training course, we wanted to find out if these teachers changed their views on multilingualism during or after they attended the course.
Our study deals with the professional development of practicing in-service teachers, mainly from primary and secondary education. We focus on teachers' beliefs, more specifically their beliefs about multilingualism and translanguaging. The central research question is: Do teachers' beliefs on multilingualism change after a training course on translanguaging in the classroom?
The article has the following structure. In 2.1 we discuss the concept of "multilingualism" in relation to pedagogical translanguaging and our earlier work on the "Focus on Multilingualism"  which was part of the training course. In 2.2 we reflect on the dimension of "teachers' beliefs", where we elaborate on beliefs in relationship to languages. This includes a discussion of some outcomes of earlier work and end with the research question that leads to this study. In section 3 we describe the design of the study. Section 4 presents the results of three scales that we have constructed on the basis of the questionnaire and in section 5 the article ends with a discussion and conclusion.

Theoretical framework
The study takes place in a context of continuing professional development. Looking at teachers in particular as the target group, the following definition can be used: "In education, the term professional development may be used in reference to a wide variety of specialized training, formal education, or advanced professional learning" (Edglossary.org, 2018). Or, to be somewhat more specific, the aim of the professional development of educationalists is to further "knowledge in a teacher's subject area -e.g., learning new scientific theories, … or learning how to teach subject-area content and concepts more effectively" (Edglossary.org, 2018). Formulated this way, it may sound rather commonsensical, but in practice to achieve such a goal of furthering knowledge is complex and full of obstacles. For example, training is related to issues about the relationship between theory and practice, and it is also about applying theoretical knowledge or new skills obtained in a training course in real-life teaching practice; it also often includes changing existing habits, some of which may have been ingrained for a long time.

Multilingualism
Over the past years applied linguists and sociolinguists have widely discussed new ideas about multilingualism and translanguaging. So much so that May (2014), titled his edited book "The Multilingual turn", and he mentions that multilingualism is a "topic du jour". He added an important caveat by remarking that "mainstream applied linguistics remains to this day largely untouched, uninterested" (May, 2014: 2). Conteh and Meier (2014) also published a book with the words "multilingual turn" in the title. They provide as a rationale for their book that, on the one hand, "in the globalised world, most societies are increasingly multilingual" while on the other hand, there are "tensions between diversity and inclusion in mainstream educational provision … minority languages are often seen as problem rather than a resource" (Conteh and Meier 2014: 1e2).
Since the publication of these two books an increasing number of researchers have turned their attention to multilingualism, although in society at large or in mainstream education monolingual ideas remain dominant.
The important role of multilingualism as a phenomenon in academia, in schools and in society is clear. Changes in the academic literature can be observed in an outburst of new terminology that has introduced, for example, concepts like "metrolingualism", "polylanguaging", "language meshing", or "translanguaging". In a critical review of those new terms, Pennycook (2016: 201) observed that "it is evident that something has been going on recently in sociolinguistics with a sudden upsurge … of new terminology." He wonders if there is really an ongoing paradigm shift or if some of the new jargon is "old wine in new bottles". His answer is that it is a bit of both. Some ideas and intentions are already present in earlier work, and those are reused and recirculated, but it is also obvious that something really is changing.
Among the new terms "translanguaging" has surfaced as the most widespread, or as Pennycook (2016: 202) signals, the "the term of choice". The concept of translanguaging has its origins in the context of bilingual education in Wales. There it was used to refer to "the process of making meaning, shaping experiences, gaining understanding and knowledge through the use of two languages" (Baker, 2011: 288). García (2009) added an important extension to the concept of translanguaging when she broadened it to mean "multiple discursive practices in which bilinguals engage in order to make sense of their bilingual worlds" (García, 2009: 45). This widening of the scope refers, among others, to the language practices of Hispanic students in New York and elsewhere, inside and outside the classroom. Acceptance of translanguaging as a legitimate practice is a key dimension of Garcia's work. In just a few years, the term translanguaging has spread widely in the field of multilingualism studies and at the same time, its meaning has evolved. Garcia in collaboration with Otheguy and Reid expanded its meaning to also include monolingual speakers. Thus, Otheguy, Garcia and Reid, (2015: 281) argued that translanguaging refers to the "deployment of a speaker's full linguistic repertoire without regard for watchful adherence to the socially and politically defined boundaries of named (and usually national and state) languages". Although these authors recognize that languages exist as important socio-political constructs, they defend a psycholinguistically "unitary view", i.e. a single undifferentiated cognitive competence. Li (2018) took the argument a step further when he proposed translanguaging as a general theory of applied linguistics. For him "translanguaging offers a practical theory of language that sees the latter [ ¼ language] as a multilingual, multisemiotic, multisensory, and multimodal resource that human beings use for thinking and for communicating thought" (Li, 2018: 26). He preferred not to use just "languaging" as a term because the added value of the concept "translanguaging" is highlighted in the "trans" prefix as it refers to: -"fluid practices that go beyond, i.e., transcend, socially constructed language systems and structures to engage diverse multiple meaning-making systems and subjectivities; -transformative capacity of the translanguaging process not only for language systems but also for individuals' cognition and social structures; -transdisciplinary consequences of re-conceptualizing language, language learning, and language use, and working across the divides between linguistics, psychology, sociology, and education" (Li, 2018: 27).
Jaspers (2018) observes two inspirations for translanguaging in his critique of the work of Garcia, Li Wei and colleagues. First, as a general theory of language and second, as a political project of transformation to recognize linguistic diversity at school. He discussed the different dimensions of the concept at length and he concluded that, in sum, translanguaging can have five meanings (p 3): "an innate instinct that includes monolinguals; the performance of fluid language use (mostly by bilinguals); a bilingual pedagogy; a theory or approach of language; a process of personal and social transformation". As Jaspers concluded, this is a lot of meaning to carry for one term.
In our own work on teachers we recognized the importance of this theoretical debate. Our model "Focus on Multilingualism"  looks at how languages are interrelated when conducting research, in teaching or when assessing different languages. The model takes three dimensions into account: (1) the multilingual speaker, (2) the whole linguistic repertoire and (3) the social context. Ideas about translanguaging can easily be integrated into this model (Cenoz & Gorter, 2013). Some teaching objectives were formulated as setting attainable goals, using plurilingual competence, developing integrated syllabi and creating resources with translanguaging activities.
Some of these ideas were implemented in a research project on the "Development of learners' language awareness through Focus on Multillingualism", that became informally known as the "translanguaging project". The project was carried out at a primary school with pupils aged 3 to 12, which was located in a town of nearly 30,000 inhabitants in the Basque Country (Spain). Of this population 52.9% are Basque speakers, 19.7% of the inhabitants can understand, but not speak Basque and 27.4% are speakers of Spanish (Eustat, 2017). The school is set up according to the so-called D-model, which means that Basque is the basic language used for teaching all subjects. The exceptions are the teaching of Spanish and English as subjects on their own while English is also used a medium of instruction for one or two other subjects such as social science. The pedagogical intervention we carried out as part of the research project is related to the original Welsh approach to translanguaging but expands it from two to three languages. Our pedagogical translanguaging is applied to a range of activities with respect to vocabulary, reading comprehension, and writing. The overall aim is to strengthen and develop the communicative and academic competences of the pupils in the three languages Basque, Spanish, and English. The specific aims of the intervention are (1) to reinforce metalinguistic awareness, (2) to reinforce language awareness, (3) to develop literacy skills and (4) to link language and content (see also Leonet, Cenoz, & Gorter, 2017, 2020. Some teaching materials were developed specifically for the intervention, in to which different strategies were incorporated. Those included strategies and exercises that allow for multilingual practices that break through language separation and further those that emphasize reading and writing skills. The intervention took place over a period of 12 weeks in the three language classes of the 5th and 6th grades (the final two grades of primary education; average age 10.6 years). Data were collected by pre-and post-testing of language competences of all three languages, an attitude questionnaire, regular ethnographic classroom observations, focus groups with the students and in-depth interviews with the three teachers involved. Also the learning materials that included, for example, a language biography exercise, a task about the linguistic landscape of the town and exercises focusing on cognates, derivatives and compounds provided further research data and gave insight into the progress of the students. The teachers of Basque, Spanish and English all three provided positive feedback on their experience with the intervention. They observed that the level of Basque had not gone down and they found it an interesting idea to relate the three languages to each other in the same lesson, for example in exercises with cognates.
One of the main conclusions of the project was that an approach that is based on pedagogical translanguaging can be compatible with the support for and the development of a minority language such as Basque. At the same time, it became clear that pedagogical translanguaging need to be introduced gradually and adapted to the existing school pedagogy ; see also the Introduction to this Special Issue). We could apply our experiences with this research project in the development of the contents of the training course which is the subject of this study.

Teachers' beliefs
The concept of "teachers' beliefs" may at face value seem free from ambiguity and rather straightforward, however, what is meant by beliefs may not always be clear. In 1992 Pajares wrote what is now a classic and widely quoted article in which he recognized that it is challenging to define beliefs, to agree upon what we mean by beliefs and how they relate to behavior. The aim of his article was to "clean up a messy construct". Pajares (1992: 309) argued that many different terms are used in the literature to refer to beliefs or as he said "beliefs travel in disguise". He then mentions the following list of concepts related to beliefs:"attitudes, values, judgments, axioms, opinions, ideology, perceptions, conceptions, conceptual systems, preconceptions, dispositions, implicit theories, explicit theories, personal theories, internal mental processes, action strategies, rules of practice, practical principles, perspectives, repertories of understanding, and social strategy". Pajares' work was partly based on earlier studies by Rokeach (1968) who argued that all beliefs have three components: a cognitive component representing knowledge, an affective component capable of arousing emotion, and a behavioral component activated when action is required. This three-component model is part of a general theory of the organization of beliefs, values and attitudes (Rokeach, 1968). Pajares (1992), after an in-depth discussion of beliefs, provides the following definition of belief: "an individual's judgment of the truth or falsity of a proposition" (p 316). This definition has been widely used in the literature. It is relevant that he further writes "teachers' attitudes about educationdabout schooling, teaching, learning, and studentsdhave generally been referred to as teachers' beliefs" (p 316). Obviously, beliefs about education are just a part of all the beliefs a teacher has. Teachers, just like any other person, hold general beliefs about society, politics, the environment and other issues, but those more general beliefs are not usually taken into account or included in the concept of "teachers' beliefs". A further important assumption put forward by Pajares is that teachers' beliefs do not stand on their own, but that they influence pedagogical decisions, and that beliefs are typically resistant to change.
The ideas of Pajares were revisited twenty years later by Fives and Buehl (2012). They wanted to investigate if conceptual progress had been made and therefore carried out a meta-analysis based on some 300 articles about beliefs. In one long table they compare no less than 74 studies on seven dimensions. They observe that over a period of 20 years a lot had been investigated and published about beliefs and they dryly remark that it is not difficult to define beliefs because many definitions can be found in the literature. They themselves quote six different definitions, but as they conclude, it is far more difficult to reach agreement. In the end Fives and Buehl (2012: 471) argue "the pervasive conviction in the literature, schools, and teacher education programs is that teachers' beliefs matter. We also believe they do".
In this article, we take the definition by Pajares as a starting point because it is in agreement with many other studies. For example, in the recent International Handbook on Research on Teachers' beliefs (Fives & Gill, 2015), Pajares' definition was quoted no less than five times. Since we are specifically interested in language beliefs, the same general handbook for the study of beliefs offered an opportunity to find out about the importance of "language" and related issues. It turned out that out of 27 chapters only one deals with language beliefs: "Teachers' Beliefs about English Language Learners" (Lucas, Villegas, & Martin, 2015). The chapter focuses entirely on English. Checking the same handbook we further find "bilingualism" as a key word three times (once as part of a reference) and for "multilingualism" we find no results. The concept "Second Language Acquisition" (or SLA) can be found two instances, but both in the same chapter by Lucas, Villegas and Martin. Of course, one handbook is insufficient as research data, but our tentative conclusion is that language beliefs are only a small issue in the wider field of teachers' beliefs in general.
At the same time it is also true that one can find numerous studies in the more specialized area of the language beliefs of teachers. Looking at such studies, these are usually not about any teachers. Most studies investigate language beliefs of language teachers, and in particular about teachers of second or foreign languages (e.g. Barcelos & Kalaja, 2011). More specifically, these studies are not about their language beliefs in general but focus more narrowly on beliefs about certain aspects of language, such as grammar, vocabulary or literacy (Borg, 2003(Borg, , 2009. For the purposes of this article, it is important to note that there are few publications concerning the beliefs about multilingualism of language teachers. Haukas (2016) carried out an exhaustive literature search and she only found four studies on teachers' beliefs on multilingualism. She summarizes and discusses the following studies: De Angelis (2011), Heyder and Sch€ adlich (2014), Jakisch (2014) and Otwinowska (2014). Haukas (2016) concluded that teachers in different countries (in her case Norway) have positive beliefs about multilingualism and teachers think that multilingualism should be promoted. She also noted that teachers do not support multilingualism in practice, that is, they do not make use of learners' previous linguistic knowledge in their own classrooms. This gap is not specific for language beliefs because, as Basturkmen (2012) showed in the majority of the studies that she reviewed, there is a limited correspondence between the teachers' stated beliefs and actual practices.
Due to the multilingual turn in applied linguistics and sociolinguistics, the topic of teachers' beliefs about multilingualism has received more attention in recent years. An additional literature search found five other studies. Young (2014) interviewed head teachers in the Alsace, France and she found that they continue to support language ideologies unfounded by research including monolingual myths, language hierarchies and separation of languages. She proposes to include a critical language awareness component in both initial teacher training and professional development. D egi (2016) examined the beliefs of English language teachers with regard to language use in the classroom. She found that while they have some basic knowledge of multilingualism, they fail to apply this knowledge in their classrooms. Otwinowska (2017) followed up her earlier study mentioned by Haukas by looking into attitudes towards the principles of a multilingual pedagogy among Polish teachers of English. She found a certain degree of multilingual awareness among these English teachers as well as a readiness to use multilingual approaches to teaching. Portol es and Martí (2018) investigated beliefs about the teaching of English as a third language among a sample of students in teacher training for preschool and primary school in the Valencian Community in Spain. The study investigated an undergraduate course on "English language teaching". Their results show that some beliefs become more favorable towards multilingualism, but they also confirm the persistence of a monolingual bias. Tarnanen and Palviainen (2018) carried out a meta-analysis of four studies of beliefs of Finnish teachers between 2016 and 2018. They showed that even though the teachers in the different studies acknowledge the language competencies of multilingual children, in the classroom they do not use them.
In an earlier study we investigated teachers' beliefs about languages (Arocena et al., 2015). Teachers in Friesland, the Netherlands and in the Basque Country, Spain were interviewed about their beliefs on multilingualism and the use of different languages. The study showed that teachers in both regions hold positive beliefs about the value of multilingualism but many of their beliefs are not congruent with their actual classroom practices. At the same time, the teachers also believe in the positive effect of language separation for teaching and most teachers' beliefs are based on monolingual assumptions about language teaching. This result is rather similar to what Haukas (2016) reported, as well as some of the other studies above. In a follow-up study (Arocena, 2017) used the data on beliefs from the earlier project, but now she looked exclusively at the teachers in the Basque Country. She combined the data on beliefs about multilingualism from interviews with observation data on code-switching in the classroom. It stands out among her findings that teachers hold strong beliefs about the exclusive use of the target language to ensure maximal input and exposure. Moreover, teachers are hesitant to allow for the use of the L1 or the L2 in the L3 classroom. Notwithstanding a verbalized belief in language separation, several of the same teachers use Basque or Spanish through frequent code-switching in the English language classroom. One of the interesting things Arocena extracted from her observations is the identification of some obvious "missed opportunities", where teachers could have used an intentional code-switching strategy, for example, "to enhance reading comprehension in English" (Arocena, 2017).
We have seen that strong beliefs continue to exist about the benefits of language separation. It is an ideology well rooted in teaching practices, where teachers try to avoid translation or interactions between languages. This has been referred to as "two solitudes" (Cummins, 2005: 588) or "separate bilingualism" (Blackledge & Creese, 2010). It is a preference for what Li (2011: 374) calls "One Language Only (OLON)" which is different from the practices of multilingual speakers in real life (Cenoz & Gorter, 2014). Multilingualism and translanguaging are complex concepts which are not so easy to measure by means of a questionnaire among a group of teachers who participate in a course (see section 3.2 on data-collection below). Several interpretations of translanguaging prefer to avoid terminology like 'code-switching' and 'language-mixing' because these suggest that languages are bounded entities with fixed codes, whereas translanguaging prefers to emphasize the fluidity of boundaries or even only acknowledge the existence of so-called named languages as socio-political constructs. Yet, it is clear that many teachers continue to think and talk in terms of alternating between given languages and they may believe that even if code-switching is something multilinguals engage in, it should be kept outside of the language classroom. It made sense to include "code-switching" and "language-mixing" as concepts in the data collection instrument. Another idea we wanted to promote in the course was the notion that languages are not competing and should not be isolated, but on the contrary, knowledge of one language can support the learning of another language. The idea that languages can mutually support each other was included in several items.
After this discussion of theoretical concepts, in particular multilingualism and teachers' beliefs, summarizing research studies and considering the aim of our study, we have formulated the following research question: To what extent do teachers' beliefs about multilingualism change during and after a training course on translanguaging in the classroom? We will answer the research question based on the answers we obtained by means of the questionnaire.
3. Design of the in-service training course and of the research study 3.1. Aims, structure and content of the training course At the request of the regional Ministry of Education, the DREAM research group at the University of the Basque Country (UPV/EHU) in Donostia-San Sebastian organized a professional development training course for in-service teachers on new ideas about multilingualism. The course was titled "Focus on Multilingualism: Language Integration in the Curriculum" and its aim was to strengthen the competencies of language and content teachers in order to deal with multilingualism. The course used a holistic approach adopting a multilingual focus in which pupils are considered as emergent multilinguals who can use their linguistic repertoire when learning and using languages. The specific aim was to develop competencies to implement an integrated language curriculum in primary or secondary school. The course combined theoretical perspectives with the application of strategies to integrate different languages in the same classroom, across language classes and also languages and content.
The course consisted of three parts. First, there was an intensive onsite seminar of four full days and then a second online section which lasted seven weeks. The third and final part was a one day session as a follow-up and closure of the course. The intensive part consisted of lectures by local and invited teachers, short assignments and debates. Several academic researchers, colleagues in the field, who came from different parts of the world, were invited to share their ideas on multilingualism and translanguaging. Colleagues from Mondragon University in the Basque Country and from our own University of the Basque Country also taught some sessions. Our organizing team of four members took care of short daily sessions at the start of each day and a concluding half hour at the end of the day, as well as a substantial introduction and a longer final session. The online part of the course was distributed over seven weeks. Each week the participants would find background information and assignments divided into the sections "Blueprint" and "Implementation". The Blueprint section contained reading materials about theory and teaching practices as well as ideas and statements for online discussions in which all participants were supposed to take part. The participants received readings that they discussed in a forum, for example, about the use of cognates or about our principles of sustainable translanguaging. The "Implementation" section contained instructions to apply the ideas from the first part of the course in their own classrooms. Each participant had to develop a lesson plan for implementation in their own class (see Cenoz & Santos in this Special Issue for outcomes of the implementation).
The professional development course was held three times. The first edition ran between October 2016 and January 2017, the second edition took place from February to May of the same year and the third and final edition between October 2017 and January 2018. The first edition was aimed primarily, but not exclusively, at English teachers in public primary and secondary schools. The second edition also included school advisors and the third edition was open to public and private schools and included teachers from a range of subject areas. The language used in the intensive part of the training course was English, but questions and comments could be in Basque or Spanish. During the online part the base language of the materials, instructions, etc. was also English, but the participants could use Basque or Spanish for debates or assignments.

Data-collection
The main data-collection instrument was a paper and pencil questionnaire with mainly closed questions about basic background data and a series of statements about multilingualism and translanguaging. In the section on background data questions were asked, among others, about age, first language and number of years of teaching experience. Teachers were also asked if they had taught Basque, Spanish and/or English as a language and through which medium of instruction. The respondents were also required to self-evaluate their language proficiency in Basque, Spanish and English on a scale from 1 to 10. The second part of the questionnaire consisted of 43 Likert-type statements related to beliefs about aspects of multilingualism or translanguaging, the use of languages in the classroom, separating languages, monolingual norms and other issues there were going to be dealt with during the course. They items were scored on scales from 1 "completely disagree" to 7 "completely agree". For example, "It is good to use translation in language classes" or "Being bilingual is beneficial when learning another language".
The questionnaire was administered at the start of the course and the second part with the 43 statements was repeated after the intensive part at the end of the fourth day. After the participating teachers had completed the seven week online part, they were asked to fill out the same list of items for a third time at the end of the follow-up day. In this way beliefs about multilingualism were thus measured at three different moments because we wanted to find out about the development of their beliefs over time. The questionnaires were presented in Basque and Spanish, but all participants chose to fill in the Basque version. Not all the teachers could attend every session and not every participant completed all the online tasks, forums or debates. After entering the quantitative data in SPSS, we discarded incomplete cases of participants for dataanalysis we had to discard. In the end we obtained 124 questionnaires that were filled in three times by the same participants.
Based on our theoretical ideas we grouped the items into three scales that we called "separating languages", "mixing languages" and "languages supporting each other". We selected the items in these three subscales based on their content related to the same theme and we tested the reliability of those scales for internal coherence (Cronbach Alpha). The reliability tests for each of these three scales turned out to be satisfactory. We discarded 10 of the 43 items from the analysis. We analyzed the items through averages for each item and for all items of each scale combined.

Background characteristics of the teachers
A short profile can be presented of the in-service teachers participating in the course. We only include those teachers who filled in the questionnaire three times (N ¼ 124). The participants were 80% female and 20% male, which seems a typical distribution in the educational system today and their age ranged from 25 to 63 years old (mean 45.3 years). All but four participants were born in the Basque Country, Spain. These in-service teachers are active at different levels of education. Most participants teach in secondary education (46%) or in primary education (41%). The remaining 13% teach at various levels: nine at a teacher training institution, three at primary and secondary levels combined, two at pre-primary level, one in vocational training, and one in an adult education institution. On average the participants had been in the teaching profession for 20.2 years (SD ¼ 9.6), ranging from 1 year for recent starters to 38 years for the most experienced.
It is obvious that this is not an average group of teachers because 78% are teaching or have taught English as a subject (which was the specific target group of the first edition of the course). At the same time 66% teach or have taught Basque as a subject and 76% through the medium of Basque. These percentages are not remarkable because most of them took part in the transition of the education system in which Basque has become predominant language of instruction (see . Among the participants, 75% were language teachers. While some taught either Basque, Spanish, English or another language (French, German or Latin), others taught either a combination of two or three languages or a combination of a language and a content subject. The other 25% included content teachers, as well as a few coordinators and advisors who do not teach.
In terms of their first language, 45% reported Spanish, 44% Basque, and the other 11% both Basque and Spanish. The selfreported language proficiency for the three languages was as follows (on a scale from 1 to 10) (Table 1).
Obviously, these teachers have on average a very high competence in the three languages and they can all be regarded as highly multilingual. The scores for Spanish are close to the maximum possible. The slightly lower scores for speaking and writing in Basque may be related to learning Basque as a second language. The average scores for English are somewhat lower compared to Basque and Spanish because English is the teachers' third language. The self-reported score for English can be offset against the information about their English qualifications. Ninety-four percent reported having a qualification for English, 4% at B1-level, 30% at B2, 42% at C1 and 23% at C2 level.

Beliefs about multilingualism: separating languages, mixing languages and language supporting each other
We created three different scales based on three constructs: separating languages, mixing languages and languages supporting each other. These were related to a classification of the questionnaire items (as was explained in the methods section 3.2.). Here we can report the results for each of the three scales.
The first scale was labeled "separating languages": this refers mainly to monolingual ideas about the use of one language at a time and the exclusive use of the target language. This scale contained 8 items and these items were formulated negatively in respect to the ideas that were transmitted during the course (Table 2).
We can see from this table that beliefs about separating language in the classroom were not strong at the start of this course in which these teachers had chosen to enroll voluntarily. In that sense, the group was probably already predisposed to having less favorable beliefs about keeping languages separate. There are substantial differences between the items, ranging from almost complete disagreement with the statement on the need to teach languages in isolation (item 1) to more moderate views on monolingual textbooks (item 7) or using only the target language while teaching (item 8). The items form a reasonably reliable scale (Cronbach a in all three cases 0.70 or more). On average the participants have a score on this language separation scale of 2.62 which is on the disagreement side of the scale (1.00 would be "fully disagree" and 4.00 is the middle of the scale).
After the intensive part of the course, the average scores for these eight items went in the direction intended by the content of the course: the participants became less convinced that separating languages is a good idea. After the period during which the participants carried out the tasks of the online course and implemented a lesson plan in their own class, they came back for a follow-up day and filled in the same questionnaire again. The average beliefs measured by this scale remained unfavorable toward separating languages. It is remarkable that the beliefs of the teachers do not seem to change much further during the online part of the course or on the follow-up day. With minor variations the items have similar scores compared to the post-test.
The second scale is on "mixing languages" which refers to the idea that languages can be used by alternating them and that mixing languages is not bad in itself. This scale consisted of 12 Likert items and gave the following results (Table 3).
A substantial change in beliefs about language mixing occurred after the intensive part of the course. Overall, the beliefs change in the direction intended by the course. We see a decrease in the beliefs (items 1 to 6) that are negative about mixing languages and we see an increase towards more positive beliefs for items 7 to 12 which are more favorable toward using more than one language. This change in beliefs persists on the follow-up day. The online tasks and the teachers' own experiences in working with translanguaging for at least one lesson, do not seem to change their beliefs about mixing much further. For each of the three times we measured, we obtained a solid Cronbach a for the scale.
The third scale is about "languages supporting each other". The basic idea is that one language can be helpful in learning another or that comparing languages is useful. The scale contained 13 items (Table 4).
In this third scale we observe again the same change in the direction of beliefs: the participants become more inclined to support beliefs about comparing languages, knowing one language can be helpful in learning another language, and the transfer of knowledge of one language to a second or a third language. In all cases the change is in the desired direction. There was much less change between the post-test and the delayed post-test, given that there was already a lot of agreement among the participants, a change was always less likely. Overall, the participants turn out be convinced that languages can support each other in different ways. For each of the three measurements (before, during and after the course) we obtained a high Cronbach a for the scale.
In the next section we will confirm the results for this quantitative part by presenting some answers to three open questions we asked from the participants on the follow-up day.

Qualitative results
The teachers on the training course were also asked to give individual feedback about their experiences on the follow-up day. They had to write answers to three open questions which will be discussed one by one.
The first question was "Do students learn more or less with these multilingual activities?" and one of the participants wrote as an answer: "I think they reinforce their previous knowledge and they acquire new ones." It was a typical answer to this question given in similar words by various participants. The participants were convinced that the multilingual approach had a positive effect on the learning outcomes of their students.
The second question was oriented to the future: "Do you see yourself including more multilingual activities in your classes?" It turned out that almost all participants answered in the affirmative. One illustrative example is: "Yes, and we are working on it; anything we hang on the wall; posters, signs, etc. We try to make it trilingual and with pictures. We think these visual aids are a very good support to help students learn and use the three languages we try to teach in the school". Based on the answers to this NB Likert items from "completely disagree" 1 to "completely agree" 7. Items are ordered from lowest to highest average agreement on the pre-test score. The mean scores on all items are significantly different and also the averages for all 12 items. question about their future orientation, one would assume that many participants intended to continue with the multilingual activities they had learned about during the training course. The third question tried to probe further and see if there was a wider dissemination effect of the course. We asked the participants "Do you think other teachers in your school will be ready to include multilingual activities"? Not all participants could or were bold enough to answer this question positively, but one answer was "I have already started to "sell" them translanguaging because I believe in it." This teacher obviously had been convinced and was trying to overcome possible resistance from colleagues, a factor that was mentioned by some of the other participants.

Discussion and conclusions
The study reported here shows that an in-service professional development course can change the beliefs of in-service teachers about insights into multilingualism and principles of translanguaging. The teachers attended four days of lectures by international specialists, held debates and carried out practical activities. Thereafter they worked on theoretical ideas and practical implementation for seven weeks and finally they came back for one day of additional information, reflection and feedback. Three times, they filled in a questionnaire on beliefs about multilingualism and translanguaging. These were mostly teachers from primary and secondary education with an average of over 20 years teaching experience. Most of them were language teachers, either of English, on which the first edition of the course was focused, of Spanish or of Basque, the main language of instruction in the education system.
In the questionnaire we distinguished three groups of items regarding the beliefs of teachers. We called these scales (1) "separating languages", (2) "mixing languages", and (3) "languages supporting each other". On each of these three sets of items we saw a change in the direction intended by the course, albeit not always to the same extend. The largest effects were between the beginning of the course and the fourth day. There was no significant further change after the online part of the course, but the effect did not wear off either. We cannot answer with our data whether the changes towards more positive beliefs about multilingualism and translanguaging is a lasting one. One limitation is that we were not able to do a follow-up after for example six months or one year. Of course, the size of the sample, although adequate in comparison to other studies, is another limitation as well as the fact it was that the teachers who decided themselves to participate. The answers to our open questions confirm the positive experiences of the participants and those qualitative results offer some reason for optimism in the future.
The outcomes contribute to our understanding of the complexity of the relationships between teachers' beliefs about languages, multilingualism and how teachers' classroom practices can be changed. During the training course the teachers told us that they were often relieved when they were finally allowed, and even encouraged, to do what in many cases they already practice but were afraid to admit to others. Now they could use different languages in the same lesson, especially in an intended way. Several of the teachers participating in the training course have progressed towards implementing a form of pedagogical translanguaging (see Cenoz & Santos this Special Issue). It remains to be seen if they are able to maintain their more positive beliefs and elaborate their translanguaging approaches over a longer time. That could be an excellent topic for a future research study. Haukas (2016) made clear that teachers often have positive beliefs about multilingualism, but they do not act according to these beliefs in practice. The group we studied here demonstrates that positive beliefs can be put into practice, but we also are aware of the obstacles and resistance that many teachers have to overcome in mainstream education. We have reached a small number of educators with our course, but many of their colleagues are probably not yet convinced of the new ideas on multilingualism and translanguaging that this course advocated. Borg (2017: 86) already observed the general issue that "situational constraints often prevent teachers from putting their beliefs into practice". This was confirmed by Basturkmen (2012) who reviewed several studies on the relationship between beliefs and practices.
In conclusion, four recommendations can be given. First of all, recent ideas about multilingualism and translanguaging that through the "multilingual turn" have spread widely in applied linguistics and sociolinguistics deserve a place in initial teacher training programs. Second, pedagogical translanguaging needs to be introduced gradually and adapted to existing school curricula. Third, implementation of pedagogical translanguaging has a greater chance of successful implementation when it is tailor-made for a specific education context, taking into account the languages used in the wider social context and the aims regarding the multilingual competences of the school. Finally, pedagogical translanguaging can be compatible with revitalization efforts on behalf of minority languages given that certain principles are adhered to . The multilingual turn, the acceptance of new multilingual realities and the introduction of translanguaging pedagogies are part of a wider renewal of education, which involves the steady adaptation of teachers to a world that is constantly changing.