**3. Theoretical framework**

A fairness argument in assessment suggests there needs to be comparability of interpretations and decision outcomes across groups. While the theoretical framework used for this chapter leans on a validity and fairness perspective in educational assessment, the empirical data are based on teachers' assessment beliefs and practices in multilingual education. Never the less, teachers' conceptions are situated in context which is determined by policy documents, teachers' disciplinary

**333**

logical stance (**Figure 1**).

*Linguistic Diversity and Comparability in Educational Assessment*

background and their experience of teaching language learners. Multilingual is here used as a generic term, including the use of two or more languages [37]. Policy is defined as a set of guidelines or organizational principles and operational practices that are informed by values [38]. The principles are expressed as educational standards and core objectives in various subject curricula at university level and the national curriculum in compulsory education. Therefore, three interdependent levels or perspectives need to be considered when looking at assessment, i.e. policy, teacher and pedagogy, where the latter represents how teaching is put into practice in the planning and sequencing of curricular content, in activities and assessment tasks. These three components, policy, teacher and pedagogy, are in turn influenced by the prevailing ontological and epistemological paradigm in society, which arguably are not always consistent. For the case of this chapter and study the ontological and epistemological perspectives of language are of particular interest. Ortega [39] defines two possible ontological perspectives on language: essentialist or non-essentialist. Whereas the essentialist perspective implies a monolingual worldview, based on traditional ontologies rooted in structuralism and named languages, using fixed developmental language dimensions to describe developmental stages, the nonessentialist perspective relates to what has been defined by Li Wei in 2019 as the post-multilingual era [40]. The multilingual turn means a shift away from named languages and represents an empowering ideology as can be found in the discourse surrounding translanguaging practices [41–44]. Whereas the essentialist view holds that there are named languages and traditional language categories cannot be ignored, the non-essentialist stance implies that language is rather seen as a process

than a product and there is a reluctance to speak of deficient language.

The figure below attempts to illustrate how the interdependent components must be considered in combination and interpreted in relation to a surrounding ontological and epistemological paradigm which is discussed below. The figure was first inspired by a model used by Nikula et al. [12] and was developed by Reierstam [15]. The current figure was further developed for the case of this chapter by Reierstam and Hellstén, by including the reference to the ontological and epistemo-

Assessment often comes at 'the end' of a module or of a course. In constructive alignment theory (cf [45]), a backwards design is advocated which means that the intended learning outcomes are taken as point of departure, "starting with the end", when planning instruction [46]. It starts with defining what the students must demonstrate they know and how well they must be able to do it according to the standards, and this decides what teaching, assessment and feedback methods to use during the course. If the language elements of a course are not specified in advance they cannot be taught nor assessed, for the assessment to be considered valid and fair. Messick [47] identified a threat to making valid interpretations of assessment outcomes which he labeled *construct-irrelevant variance*. It means that adequate and appropriate inferences cannot be made about a student's subject matter knowledge if an assessment contains for example, language elements that the student has not been given an opportunity to learn before being assessed. Therefore, the equitable and fair assessment requires that students be assessed in ways that are consistent with how they were taught. Kane [3] makes a distinction between procedural fairness and substantive fairness where the first stipulates that all students taking a test should be treated in essentially the same way, that is, by doing the same or equivalent assessments under the same conditions. This also requires that accomplishment ought to be evaluated by using the same rules and procedures. Substantive fairness on the other hand, implies that score interpretations and use of assessment are appropriate across groups, which albeit may require different types of assessments

and accommodations to better suit different needs of individual students.

*DOI: http://dx.doi.org/10.5772/intechopen.97405*

#### *Linguistic Diversity and Comparability in Educational Assessment DOI: http://dx.doi.org/10.5772/intechopen.97405*

*Teacher Education in the 21st Century - Emerging Skills for a Changing World*

language learners where evidently language deficiency may be an important reason they are denied equal access to learning. Edyburn [32] notes that when supports are embedded already in the design of the curriculum, it reduces the likelihood of student failure and frustration as well as alleviating the stress for teachers to reactively create accommodations and modifications. When the UDL principles – to provide multiple means of representation, multiple means of action and expression and multiple means of engagement – are applied among language learners, the teachers will help the students to have an opportunity to build background knowledge, interact with information visually as well as auditorily. They will also provide rich scaffolding and support to help highlight the patterns of language and allow the students to have numerous opportunities to express their knowledge in meaningful ways (Novak, n.d.). The option to build in numerous, or different ways of communicating knowledge already at the initial planning of a course aligns with assessment validity theory which is further outlined below. UDL pedagogy has been identified as a way to provide fairer tests by using multiple ways of action and test items which take the students' diverse backgrounds into account, not relying on cultural information outside of what has been taught or presented in the test content [33, 34]. Fovet [34] notes that accommodations and inclusive pedagogy are often considered unnecessary for graduate students since they supposedly have already adopted academic strategies. However, as argued by Fovet [34], traditional forms of accommodations are ineffective with regards to the great diversity that can be found among students today. Haigh [35] further notes that teaching learners with different cultures, worldviews and aptitudes is an abiding problem in Higher Education which calls for a design for pluralism, especially in assessment, to reach educational inclusivity. UDL explicitly forwards a framework for the design of lesson plans and assessments based on three main principles: to provide multiple means of representation, multiple means of action and expression, and multiple means of engagement. This can also be identified in other models used for the instruction of language learners. Two decades ago, Gibbons [36] for instance, spoke of scaffolding learning by using a genre-based curriculum cycle, starting by building knowledge together of the field, modeling text types, working through joint construction of subject specific genre before reaching students' independent writing. While the underlying UDL methodology is already employed by many, whether identified specifically as universal design or not, this is challenging for teachers for many reasons, one being time constraints, or as mentioned here, unfamiliarity with language teaching strategies. It is not only a challenge, but a threat to fairness and comparability in education if this becomes a matter of individual choice and experience, rather than a shared responsibility. As mentioned here in the introduction inconsistencies in the alignment of policy and pedagogy involving difficulties in matching intended learning outcomes and standards with student cohorts which lack expected academic skills at the outset need to be regarded as a shared concern in quality education. In the next section the theoretical framework and underpinnings adopted in

A fairness argument in assessment suggests there needs to be comparability of interpretations and decision outcomes across groups. While the theoretical framework used for this chapter leans on a validity and fairness perspective in educational assessment, the empirical data are based on teachers' assessment beliefs and practices in multilingual education. Never the less, teachers' conceptions are situated in context which is determined by policy documents, teachers' disciplinary

**332**

this chapter are presented.

**3. Theoretical framework**

background and their experience of teaching language learners. Multilingual is here used as a generic term, including the use of two or more languages [37]. Policy is defined as a set of guidelines or organizational principles and operational practices that are informed by values [38]. The principles are expressed as educational standards and core objectives in various subject curricula at university level and the national curriculum in compulsory education. Therefore, three interdependent levels or perspectives need to be considered when looking at assessment, i.e. policy, teacher and pedagogy, where the latter represents how teaching is put into practice in the planning and sequencing of curricular content, in activities and assessment tasks. These three components, policy, teacher and pedagogy, are in turn influenced by the prevailing ontological and epistemological paradigm in society, which arguably are not always consistent. For the case of this chapter and study the ontological and epistemological perspectives of language are of particular interest. Ortega [39] defines two possible ontological perspectives on language: essentialist or non-essentialist. Whereas the essentialist perspective implies a monolingual worldview, based on traditional ontologies rooted in structuralism and named languages, using fixed developmental language dimensions to describe developmental stages, the nonessentialist perspective relates to what has been defined by Li Wei in 2019 as the post-multilingual era [40]. The multilingual turn means a shift away from named languages and represents an empowering ideology as can be found in the discourse surrounding translanguaging practices [41–44]. Whereas the essentialist view holds that there are named languages and traditional language categories cannot be ignored, the non-essentialist stance implies that language is rather seen as a process than a product and there is a reluctance to speak of deficient language.

The figure below attempts to illustrate how the interdependent components must be considered in combination and interpreted in relation to a surrounding ontological and epistemological paradigm which is discussed below. The figure was first inspired by a model used by Nikula et al. [12] and was developed by Reierstam [15]. The current figure was further developed for the case of this chapter by Reierstam and Hellstén, by including the reference to the ontological and epistemological stance (**Figure 1**).

Assessment often comes at 'the end' of a module or of a course. In constructive alignment theory (cf [45]), a backwards design is advocated which means that the intended learning outcomes are taken as point of departure, "starting with the end", when planning instruction [46]. It starts with defining what the students must demonstrate they know and how well they must be able to do it according to the standards, and this decides what teaching, assessment and feedback methods to use during the course. If the language elements of a course are not specified in advance they cannot be taught nor assessed, for the assessment to be considered valid and fair. Messick [47] identified a threat to making valid interpretations of assessment outcomes which he labeled *construct-irrelevant variance*. It means that adequate and appropriate inferences cannot be made about a student's subject matter knowledge if an assessment contains for example, language elements that the student has not been given an opportunity to learn before being assessed. Therefore, the equitable and fair assessment requires that students be assessed in ways that are consistent with how they were taught. Kane [3] makes a distinction between procedural fairness and substantive fairness where the first stipulates that all students taking a test should be treated in essentially the same way, that is, by doing the same or equivalent assessments under the same conditions. This also requires that accomplishment ought to be evaluated by using the same rules and procedures. Substantive fairness on the other hand, implies that score interpretations and use of assessment are appropriate across groups, which albeit may require different types of assessments and accommodations to better suit different needs of individual students.

#### **Figure 1.**

*Three interdependent components in curriculum design and teaching, situated in a current ontological and epistemological paradigm [15].*

Effective teaching, learning and assessment practices are dependent on both the knowledge and the value systems adopted by educators ([4], p. 54). In linguistically diverse settings intercultural inclusive schools are more sensitive towards providing differentiated instruction and support to students in order to accommodate for the different needs and attributes of individual learners [4]. Teacher beliefs, particularly in relation to assessment is sometimes defined as "teachers' assessment literacy". To reflect on this, Abell and Siegel [48] developed a model to describe science teachers' assessment literacy in secondary education among language learners, a model which was used and modified by Reierstam in a recent study in Sweden [15]. The model which is presented below (**Figure 2**) includes four so called, knowledge elements: knowledge of the assessment purposes (Why?) knowledge of what to assess (What?), knowledge of assessment strategies (How?) and knowledge of assessment interpretation and action-taking (And?).

The assessment literacy framework above relies on teacher beliefs and an integrated view on teaching, learning and assessment. At the center lies the teacher's views of learning and assessment of the learner as these are believed to interact with the teachers' assessment practices. Biggs and Tang [49] note that "desirable student learning depends both on student-based factors – ability, appropriate prior knowledge, clearly accessible new knowledge – and on the teaching context, which includes teacher responsibility, informed decision making and good management." Teachers' previous experiences, knowledge and beliefs are shown to shape their views of how students learn and thus influence the teachers' priorities in practice. Entwistle et al. [50] describe beliefs as reflecting strongly felt ideas and implicit or tacit knowledge derived from experience and driven by emotions rather than from any predefined conceptual framework. Nespor [51] similarly states that beliefs can be argued to reflect personal truth and evaluative components that are guided by feelings rather than logic. Teacher beliefs are said to be associated with years of

**335**

statistics [55].

multilingual contexts.

**Figure 2.**

**4. Case in point from the Swedish context**

*Linguistic Diversity and Comparability in Educational Assessment*

*The assessment literacy model. Used in [15], adapted after Abell and Siegel [48].*

experience and professional education [52, 53]. However, Pajares [54] point out that there may be inconsistencies in teacher beliefs, which often depend on the role of contextual factors and the interconnectedness between the individual and the surrounding world. The possible inconsistencies in teacher beliefs and policy as described above, together with the argument that equitable and fair assessment hinges on equal access to knowledge, gives reason to believe that there is a lot to gain from finding empirically based descriptions of good assessment practices. The assessment literacy model as outlined above provides a structured way of describing some of the most critical key elements for more transparent assessment practices. There is no contradiction, as argued in this chapter, in identifying what the assessment requirements of the students are in order for providing essential and relevant learning opportunities and feedback along the way. At the same time it advocates for universal design for learning, positing flexibility in teaching and taking the individual students' backgrounds into account, as will be illustrated in a case in point from the Swedish context below. In the following, the assessment literacy model in **Figure 2** is used as a frame while presenting data from a study in Swedish

This part of the chapter presents data from a study in two parts investigating teachers' assessment beliefs and assessment practices in two multilingual contexts in Sweden. The first was conducted in content and language integrated, CLIL, schools where English is used as a medium of instruction in the content courses at upper secondary level, students aged 16–19. The second study covered teachers in contexts with newly arrived migrant students aged 13 to 19 [15]. Both studies included teachers in the social and the natural sciences. Between 2015 and 2017 there was an increase of older age migrant students with only a few years before they graduate, although Sweden has a fairly long history of welcoming immigrants. However, discouraging reports showing that among those who arrived in grades 6 to 9 only 28% qualified for upper secondary education. Questions of inclusion and how to help them succeed academically has become an urgent concern. At the time of the study every fourth student in Swedish compulsory schools had a foreign background and in the upper secondary the figure was 30% according to

The extent and scope of CLIL has been more difficult to define, since there is considerable variation in how the English medium instruction is implemented, but according to a recent survey around 4% of the upper secondary schools have

*DOI: http://dx.doi.org/10.5772/intechopen.97405*

*Linguistic Diversity and Comparability in Educational Assessment DOI: http://dx.doi.org/10.5772/intechopen.97405*

**Figure 2.**

*Teacher Education in the 21st Century - Emerging Skills for a Changing World*

Effective teaching, learning and assessment practices are dependent on both the knowledge and the value systems adopted by educators ([4], p. 54). In linguistically diverse settings intercultural inclusive schools are more sensitive towards providing differentiated instruction and support to students in order to accommodate for the different needs and attributes of individual learners [4]. Teacher beliefs, particularly in relation to assessment is sometimes defined as "teachers' assessment literacy". To reflect on this, Abell and Siegel [48] developed a model to describe science teachers' assessment literacy in secondary education among language learners, a model which was used and modified by Reierstam in a recent study in Sweden [15]. The model which is presented below (**Figure 2**) includes four so called, knowledge elements: knowledge of the assessment purposes (Why?) knowledge of what to assess (What?), knowledge of assessment strategies (How?) and knowledge

*Three interdependent components in curriculum design and teaching, situated in a current ontological and* 

The assessment literacy framework above relies on teacher beliefs and an integrated view on teaching, learning and assessment. At the center lies the teacher's views of learning and assessment of the learner as these are believed to interact with the teachers' assessment practices. Biggs and Tang [49] note that "desirable student learning depends both on student-based factors – ability, appropriate prior knowledge, clearly accessible new knowledge – and on the teaching context, which includes teacher responsibility, informed decision making and good management." Teachers' previous experiences, knowledge and beliefs are shown to shape their views of how students learn and thus influence the teachers' priorities in practice. Entwistle et al. [50] describe beliefs as reflecting strongly felt ideas and implicit or tacit knowledge derived from experience and driven by emotions rather than from any predefined conceptual framework. Nespor [51] similarly states that beliefs can be argued to reflect personal truth and evaluative components that are guided by feelings rather than logic. Teacher beliefs are said to be associated with years of

of assessment interpretation and action-taking (And?).

**334**

**Figure 1.**

*epistemological paradigm [15].*

*The assessment literacy model. Used in [15], adapted after Abell and Siegel [48].*

experience and professional education [52, 53]. However, Pajares [54] point out that there may be inconsistencies in teacher beliefs, which often depend on the role of contextual factors and the interconnectedness between the individual and the surrounding world. The possible inconsistencies in teacher beliefs and policy as described above, together with the argument that equitable and fair assessment hinges on equal access to knowledge, gives reason to believe that there is a lot to gain from finding empirically based descriptions of good assessment practices. The assessment literacy model as outlined above provides a structured way of describing some of the most critical key elements for more transparent assessment practices. There is no contradiction, as argued in this chapter, in identifying what the assessment requirements of the students are in order for providing essential and relevant learning opportunities and feedback along the way. At the same time it advocates for universal design for learning, positing flexibility in teaching and taking the individual students' backgrounds into account, as will be illustrated in a case in point from the Swedish context below. In the following, the assessment literacy model in **Figure 2** is used as a frame while presenting data from a study in Swedish multilingual contexts.

## **4. Case in point from the Swedish context**

This part of the chapter presents data from a study in two parts investigating teachers' assessment beliefs and assessment practices in two multilingual contexts in Sweden. The first was conducted in content and language integrated, CLIL, schools where English is used as a medium of instruction in the content courses at upper secondary level, students aged 16–19. The second study covered teachers in contexts with newly arrived migrant students aged 13 to 19 [15]. Both studies included teachers in the social and the natural sciences. Between 2015 and 2017 there was an increase of older age migrant students with only a few years before they graduate, although Sweden has a fairly long history of welcoming immigrants. However, discouraging reports showing that among those who arrived in grades 6 to 9 only 28% qualified for upper secondary education. Questions of inclusion and how to help them succeed academically has become an urgent concern. At the time of the study every fourth student in Swedish compulsory schools had a foreign background and in the upper secondary the figure was 30% according to statistics [55].

The extent and scope of CLIL has been more difficult to define, since there is considerable variation in how the English medium instruction is implemented, but according to a recent survey around 4% of the upper secondary schools have some kind of CLIL provision [56]. Although the CLIL programs attract motivated students with high language proficiency levels, teachers note that it is hard to know if lower achievements are due to the fact that the productive language skills are not there, or if it depends on limited content knowledge [15, 23]. In both contexts there are thus concerns with regard to fairness and accuracy in assessment since teachers acknowledge that the students risk being at a disadvantage, when using a nonnative language as the medium of instruction. The main focus in the two studies were therefore how teachers perceive of language in relation to subject content, how it is dealt with before, after and in the actual assessment situation. The teachers' views of the students and their needs, the reported assessment principles and stated local policy were found to vary depending on school and the individual characteristics of the teachers, such as education and subject discipline. Whereas document analyses of assessment samples and teacher interviews were used in both studies (N = 12 in Study I, N = 13 in Study II), a survey (N = 196) was conducted in Study II. For the analyses of the survey data both descriptive and inferential statistics were employed to describe and compare teachers' beliefs. Qualitative content analyses were used for the interviews and assessment samples. For more see Reierstam [15].

#### **4.1 Assessment values and principles**

Without going into detail about the education system in Sweden, both of the above studies confirmed that there exists an accountability culture in Swedish schools [57]. Even though many teachers, especially among the migrant students, express a concern for the learners, the concern about reliability and legal justice in assessment was more pronounced. National guidelines from the National Agency for Education as well as recent research and literature advocate the use of the students' strongest language in the initial phase. *Translanguaging* practices, allowing for students to use all of their language resources are recommended. Although it is stipulated by law that newly arrived students should be provided study counselors who speak their language, many schools fail to do so. When asked who they collaborate with, teachers in the present study confirmed that there was little support from study counselors. Very few of the teachers reported to implement translanguaging pedagogy, most teachers agreed that the students need to learn Swedish. A couple of the interviewed teachers were a bit uncertain if the students are allowed to use any language to show proof of content knowledge, if so, they must they argued, but also agreed that this would be difficult for practical reasons. In the CLIL context some teachers were very determined to only use English, whereas others said that they allowed students to use either English or Swedish but they tried to stick with English. All of this relates to what was mentioned in the introduction about inconsistencies in language policy and vacillation between alternative ontological paradigms. On the one hand Swedish is considered the norm, or English for the CLIL incentive, on the other the communicative aspects with a focus on process rather than product, could be distinguished in some cases, as advocated in advice for the newly arrived migrants. Here it becomes a matter of the teachers' own interpretation about how language, and what type of language, is representative of the intended learning outcomes and how it is to be assessed.

When using oral follow-up for letting students elaborate on their individual answers as a didactic method of allowing for accommodation for their deficient language, the teachers expressed concern about their contributing to invalidating the student's performance: *"How do I know what the student's actual level is?"* This aligns well with the claims that in language testing, oral assessment is a field of its own, raising questions in terms of the validity and reliability of the dialogic coconstruction between the student and the teacher [58, 59]. In the migrant student

**337**

*Linguistic Diversity and Comparability in Educational Assessment*

context, teachers stated that they preferred for students to do the assessment in Swedish, as opposed to their native language, or with a translator since this would be out of their reach and control. In the CLIL context, the teachers were more flexible in letting students mix Swedish and English, since the teachers understand both languages and owing to the fact that there is no explicit language policy at upper secondary level which limits the use of English. (The English language has a unique position in Swedish society and it is stipulated in the Swedish School Ordinance that up to 50% of the instruction in compulsory schooling can be provided in English [60]. A pedagogy-driven, or learner-centered assessment approach was more common among those teachers teaching younger newly arrived students. Teachers with a double teaching profile, that is, a language (typically Swedish, Swedish as a second language or English as a foreign language) and a content subject, also showed statistically significant higher tendency towards this trend, as did the data showing a providing of language learning opportunities in the classrooms for such students. Teachers in both contexts expressed that they were not specifically interested in collaborating with language teacher colleagues or mother tongue instructors, if they were to collaborate with a colleague they preferred someone from the same subject discipline. Collaboration with colleagues in the same discipline has been noted to build assessment literacy in teachers [61]. A general comment noted from the teachers in the study, was that time constraints prevented them from having exchange

The main purpose of assessment was seen in collecting evidence of attainment of knowledge and to have a basis on which to ascertain grading. Teachers also mentioned the formative assessment aspect and especially among the newly arrived migrant students, to communicate to students what is missing and how they are progressing. Teacher also showed an awareness of their responsibility to equip students for the next level of learning, required for gaining competence for writing in higher education levels. Several teachers noted that their students are quite satisfied with presenting their knowledge in the oral format, after getting used to not having to write, but the teachers stressed that they must learn how to write to qualify into upper secondary school levels. However, a majority of the teachers interviewed said that they prefer the written assessment formats (see below) since this shows more reliable results. There is a consensus among the teachers that students may prefer written examinations due to being used to this form of assessment in previous schooling. As noted above, the purpose of accountability seemed the strongest motive in assessment and teachers referred to the need to be able to meet the stated

The 'what' of assessment is closely linked to instructional goals. In Swedish education, the standards and knowledge requirements of the national curriculum together with national tests administered in many subjects, are explicitly claimed to safe-guard the equity in education regardless of the location of the schools in different parts of the country. As noted above, the importance of national curricula seems to be undisputed by teachers in Sweden. However, what is interesting from a language perspective is the relative uncertainty and the inconsistencies that prevail between what language is taken to mean in relation to subject content, and which language can or should be used by the students, as mentioned above. First of all, the *subject specific concepts* seemed to be perceived as 'subject content' by the subject

knowledge requirements in the subject curricula.

*DOI: http://dx.doi.org/10.5772/intechopen.97405*

with colleagues.

**4.2 Why assess**

**4.3 What to assess**

#### *Linguistic Diversity and Comparability in Educational Assessment DOI: http://dx.doi.org/10.5772/intechopen.97405*

context, teachers stated that they preferred for students to do the assessment in Swedish, as opposed to their native language, or with a translator since this would be out of their reach and control. In the CLIL context, the teachers were more flexible in letting students mix Swedish and English, since the teachers understand both languages and owing to the fact that there is no explicit language policy at upper secondary level which limits the use of English. (The English language has a unique position in Swedish society and it is stipulated in the Swedish School Ordinance that up to 50% of the instruction in compulsory schooling can be provided in English [60]. A pedagogy-driven, or learner-centered assessment approach was more common among those teachers teaching younger newly arrived students. Teachers with a double teaching profile, that is, a language (typically Swedish, Swedish as a second language or English as a foreign language) and a content subject, also showed statistically significant higher tendency towards this trend, as did the data showing a providing of language learning opportunities in the classrooms for such students. Teachers in both contexts expressed that they were not specifically interested in collaborating with language teacher colleagues or mother tongue instructors, if they were to collaborate with a colleague they preferred someone from the same subject discipline. Collaboration with colleagues in the same discipline has been noted to build assessment literacy in teachers [61]. A general comment noted from the teachers in the study, was that time constraints prevented them from having exchange with colleagues.

### **4.2 Why assess**

*Teacher Education in the 21st Century - Emerging Skills for a Changing World*

**4.1 Assessment values and principles**

the intended learning outcomes and how it is to be assessed.

When using oral follow-up for letting students elaborate on their individual answers as a didactic method of allowing for accommodation for their deficient language, the teachers expressed concern about their contributing to invalidating the student's performance: *"How do I know what the student's actual level is?"* This aligns well with the claims that in language testing, oral assessment is a field of its own, raising questions in terms of the validity and reliability of the dialogic coconstruction between the student and the teacher [58, 59]. In the migrant student

some kind of CLIL provision [56]. Although the CLIL programs attract motivated students with high language proficiency levels, teachers note that it is hard to know if lower achievements are due to the fact that the productive language skills are not there, or if it depends on limited content knowledge [15, 23]. In both contexts there are thus concerns with regard to fairness and accuracy in assessment since teachers acknowledge that the students risk being at a disadvantage, when using a nonnative language as the medium of instruction. The main focus in the two studies were therefore how teachers perceive of language in relation to subject content, how it is dealt with before, after and in the actual assessment situation. The teachers' views of the students and their needs, the reported assessment principles and stated local policy were found to vary depending on school and the individual characteristics of the teachers, such as education and subject discipline. Whereas document analyses of assessment samples and teacher interviews were used in both studies (N = 12 in Study I, N = 13 in Study II), a survey (N = 196) was conducted in Study II. For the analyses of the survey data both descriptive and inferential statistics were employed to describe and compare teachers' beliefs. Qualitative content analyses were used for the interviews and assessment samples. For more see Reierstam [15].

Without going into detail about the education system in Sweden, both of the above studies confirmed that there exists an accountability culture in Swedish schools [57]. Even though many teachers, especially among the migrant students, express a concern for the learners, the concern about reliability and legal justice in assessment was more pronounced. National guidelines from the National Agency for Education as well as recent research and literature advocate the use of the students' strongest language in the initial phase. *Translanguaging* practices, allowing for students to use all of their language resources are recommended. Although it is stipulated by law that newly arrived students should be provided study counselors who speak their language, many schools fail to do so. When asked who they collaborate with, teachers in the present study confirmed that there was little support from study counselors. Very few of the teachers reported to implement translanguaging pedagogy, most teachers agreed that the students need to learn Swedish. A couple of the interviewed teachers were a bit uncertain if the students are allowed to use any language to show proof of content knowledge, if so, they must they argued, but also agreed that this would be difficult for practical reasons. In the CLIL context some teachers were very determined to only use English, whereas others said that they allowed students to use either English or Swedish but they tried to stick with English. All of this relates to what was mentioned in the introduction about inconsistencies in language policy and vacillation between alternative ontological paradigms. On the one hand Swedish is considered the norm, or English for the CLIL incentive, on the other the communicative aspects with a focus on process rather than product, could be distinguished in some cases, as advocated in advice for the newly arrived migrants. Here it becomes a matter of the teachers' own interpretation about how language, and what type of language, is representative of

**336**

The main purpose of assessment was seen in collecting evidence of attainment of knowledge and to have a basis on which to ascertain grading. Teachers also mentioned the formative assessment aspect and especially among the newly arrived migrant students, to communicate to students what is missing and how they are progressing. Teacher also showed an awareness of their responsibility to equip students for the next level of learning, required for gaining competence for writing in higher education levels. Several teachers noted that their students are quite satisfied with presenting their knowledge in the oral format, after getting used to not having to write, but the teachers stressed that they must learn how to write to qualify into upper secondary school levels. However, a majority of the teachers interviewed said that they prefer the written assessment formats (see below) since this shows more reliable results. There is a consensus among the teachers that students may prefer written examinations due to being used to this form of assessment in previous schooling. As noted above, the purpose of accountability seemed the strongest motive in assessment and teachers referred to the need to be able to meet the stated knowledge requirements in the subject curricula.

## **4.3 What to assess**

The 'what' of assessment is closely linked to instructional goals. In Swedish education, the standards and knowledge requirements of the national curriculum together with national tests administered in many subjects, are explicitly claimed to safe-guard the equity in education regardless of the location of the schools in different parts of the country. As noted above, the importance of national curricula seems to be undisputed by teachers in Sweden. However, what is interesting from a language perspective is the relative uncertainty and the inconsistencies that prevail between what language is taken to mean in relation to subject content, and which language can or should be used by the students, as mentioned above. First of all, the *subject specific concepts* seemed to be perceived as 'subject content' by the subject matter teachers and was reported as very important in assessment. However, in study II among migrant students, subject concepts were perceived as difficult for the students. *Subject-specific skills* (e.g. to argue, to reason and be able to draw conclusions) that are stated in the knowledge requirements were considered the most important for assessment. The *general academic language* as well as *subject-specific written or oral genre* were not reported to be as important for assessment, and yet many teachers stated that lack of language represented a problem when interpreting assessment results. Teachers also expressed challenges due to insufficient language in relation to higher-order skills which meant that attainment of the highest grades was considered difficult or even unrealistic. Most teachers in both contexts stated that they do not assess language.

The confusion as to what represents content and language becomes evident in contexts with language learners. Since standards are used which also include expected outcomes in relation to communication skills, the solution is to identify and map the language skills that are embedded in the curriculum. The solution then is to integrate language development within disciplinary teaching and learning [10, 29]. Some may claim that the solution is in finding assessment formats that avoid the use of language, and construct-irrelevant variance, but this seems unrealistic as can be seen below.

#### **4.4 How to assess**

While the intended learning outcomes and aim of the subject come first, the intended 'how' of assessment needs to be considered already at the outset. In a backwards design, where all students are to be given access and opportunities to learn the necessary skills and language, the targeted forms of presentation need to be included in the planning of a course. "[I]t is not the elements per se, but rather their organization and sequencing, expression, relative emphasis and degree of alignment that differentiate effective from ineffective subject design" ([46], p. 95). In a universal design there needs to be flexibility to cater for students' various needs, but it has also been stated that what is good for disadvantaged students is good for all. In this study several teachers said that they had noted that all students, regardless of first language, need help with their language development, even native Swedish speakers. Using an alignment and fairness perspective the how of assessment can be said to include both preparation for assessment, the actual assessment design and format, including accommodations and support used for language learners in this case as well as the grading process that follows. As noted previously the teachers in both contexts expressed a preference for written assessment formats, but especially teachers of newly arrived migrants said that they use "anything that works" to get some information about the students' content knowledge.

Teachers in both contexts mentioned oral assessment serving as an accommodation for poor written expression, which brings up concerns as to whether the oral assessment results might indeed equal written results, and whether students might have been given the opportunity to practice their oral expression during the course of study leading into the testing phase. Other accommodations that are typically suggested as reducing the negative impact of insufficient language proficiency include, for instance, allowance for the use of dictionaries, the use of simplified language or extended time (cf [62]). However, Abedi [63] raises concerns over the validity of certain accommodations, since studies show that they may affect the end results of second-language speakers and native speakers alike, thus representing a threat to the validity of test-results. Stobart [64] mentions three areas where equity should be considered in relation to fairness in assessment across groups: questions of access (differences in resources), curriculum (what is taught, why and how);

**339**

**Figure 3.**

*use. ([15], p. 224).*

robust practices to emerge.

*Linguistic Diversity and Comparability in Educational Assessment*

equity in assessment, which will be discussed below.

**4.5 And? The consequences of assessment**

and assessment (appropriateness of form, content and mode for different groups). Regarding access and assessment, some of the teachers expressed that it is up to the students, or the language teachers, to learn/teach the necessary language, whereas others were doing their utmost to support the students. Especially teachers with a language teaching background, or who had collaborated with a language teacher colleague, reported to try to model language and used a universal design for learning. The variation between teachers raises concerns with regard to fairness and

It is well known, that language proficiency impacts the results of contentarea assessments. Research shows that multilinguals, who hold the same ability levels with native speakers show lower probability of correctly solving test items in mathematics [65–67]. Kane's [3] distinction *substantive fairness,* which was described previously, means that different assessment procedures are required for different students in order to achieve comparable results. In a similar vein Gee [68] and Stobart [64] state that equity in assessment hinges on students' equal opportunities for learning. Wherever there are standards and intended learning outcomes, teachers need to consider what is required of the students in terms of language proficiency. It has been argued that the alignment between specific forms of learning, assessment format and how the assessment data/results are used has been disconnected [69]. In the Swedish case in point a variation in how teachers perceive of their role as language educators can be seen, how and if they prepare students with required language skills for succeeding in the assessment. There are also inconsistencies in language policy. Such can be found, both in what appears to be a confusion or unawareness about the different underlying ontological and epistemological foundations. The discrepancy exists between on the one hand, a non-essentialist disaggregated translingual view, apart from the named languages, and a traditional essentialist view which allows for predefined language standards and learning outcomes. No matter which foundations, consequences need to be considered in order to ensure equitable and comparable assessment practices, that is, to define intended learning outcomes (ILOs) in the target language (TL). This needs to be adopted both for the disciplinary language and the language of instruction, and needs to align the assessment format and procedures with the instructions (see **Figure 3** below). This amounts to a adoption of a non-essentialist ontology which allows any language to be used, as is prevalent e.g. in translingual practices. The translingual pedagogy therefore, requires other measures in order to achieve comparability of results, including analyses of its consequences for society and students' opportunities in life. These aspects present a gap in current research which need more input in order for empirically based

*Alignment of languages in education with intended learning outcomes, target language (TL) and future* 

*DOI: http://dx.doi.org/10.5772/intechopen.97405*

and assessment (appropriateness of form, content and mode for different groups). Regarding access and assessment, some of the teachers expressed that it is up to the students, or the language teachers, to learn/teach the necessary language, whereas others were doing their utmost to support the students. Especially teachers with a language teaching background, or who had collaborated with a language teacher colleague, reported to try to model language and used a universal design for learning. The variation between teachers raises concerns with regard to fairness and equity in assessment, which will be discussed below.
