Language Learners' Imagined Communities: Model and Questionnaire Development in the Iranian Context

Introduction

Norton Peirce's (1995) study of second language (L2) learning experiences of five adult immigrant women in Canada laid the groundwork for the establishment of the theoretical relevance of identity in second language acquisition (SLA) research. She employed the term identity “to reference how a person understands his or her relationship to the world, how that relationship is constructed across time and space, and how the person understands possibilities for the future” (Norton, 2000, p. 5).

Two closely interrelated constructs which came into consideration as an extension of interest in identity research were the learners' investment in language learning and language learners' imagined communities/identities (Norton, 1997, 2001; Kanno & Norton, 2003). These three significant constructs reveal the ‘social turn’ (Block, 2003) in applied linguistics and can deepen our understanding of language learners’ experiences.

Known as a primarily sociological construct, investment was proposed by Norton Peirce (1995) in place of the psychological construct of motivation to complement the notions of motivation in the field of SLA (Norton & Gao, 2008). Investment can provide a more comprehensive analysis of language learning process in which the learner is reflected as a social agent who has an active role in the course of language learning. This construct signifies the commitment of learners to learn an L2, and considers their imagined identities and their hopes for the future (Norton, 2016).

As a construct closely connected with identity and investment, a language learner’s imagined communities/identities embodies any community of the imagination that language learners aspire to belong while they are learning a language (Norton, 2016). This term was initially used by Norton (2001) in SLA research by applying the concept of imagined community from Anderson (1991) and arguing how the desired community of two adult immigrant language learners as the participants in her study went beyond the book and the four walls of the classroom context both temporally and spatially. This notion inspired Norton to consider the non-participatory role of the learners in L2 classrooms (Norton, 2000, 2001).

As Wenger (1998) argues, direct involvement with community practices and investment in concrete relationships, what he recognizes as engagement, is not the only way individuals belong to a community. He considers imagination as another important source of community. In her work on identity and language learning, Norton (2000) draws on Wenger’s (1998) theory of imagination and learning, which states that imagination links current practices with broader activities and is a process of self-expansion by moving beyond time and place and picturing new images of individuals and the world. Hence, without any active participation, it can have a strong bearing on one's agency, motivation, investment, resistance to language learning and subsequent progress in language learning (Norton, 2000, 2001, 2016; Pavlenko & Norton, 2007). What is noteworthy is that imagined communities affect learners’ concomitant actions (and identity co-construction) and investment even more strongly than the ones with which they have daily engagement (Kanno & Norton, 2003).

The distinctive point about our imagination today is that it has been affected by two developments: communication technology and mass migration (Appadurai, 1996). The effect of the swift development of global communication systems and technological advances on the scope of imaginable communities is certain. These advances have had a significant impact on the extension of a sense of community (Almgren, 2000) and what is possible to imagine (Kanno & Norton, 2003).

In the 21st century, due to the existence of profound digital innovations, superdiversity, and mobility, learners can take part in unlimited spaces of learning and socialization, both face-to-face and virtual, which are developed with distinct and progressively invisible systemic forms of control and power structures (Darvin & Norton, 2015). At this time of heightened mobility that English has furnished many persons with an unparalleled amount of symbolic capital (Bourdieu, 1991), the issue of a present or future imagined self should be taken more deeply (Dawson, 2017).

Literature Review

The construct of imagined communitiesrefers to how people try to feel a sense of belonging to groups of people, not directly accessible, through the power of imagination (Norton & Kamal, 2003). Originally coined by Anderson (1991), it was used to reflect the construction of a nation state and was described as imagined because, ‘‘the members of even the smallest nation will never know most of their fellow-members, meet them, or even hear of them, yet in the minds of each lives the image of their communion’’ (p. 6).

Imagination is a beneficial force that provides hope for a bright future and can be the force for action (Appadurai, 1996). Emphasis on the future reflects a consideration of the significance of imagination in learning and teaching (Norton & Kamal, 2003). A fascinating aspect of imagination is that it should not be supposed to be the same as fantasy. Simon (1992) differentiates between “wishes,” in which there is no probability for action, and “hopeful imagination,” in which action becomes substantial in the fulfillment of desire and, subsequently, brings forth attempt for a better future.

Theories of imagination and learning were developed widely by Wenger (1998), who focused on the relationship between imagination and identity, assuming that imagination is the process of generating new images of the world and ourselves (p. 176). It is this conception of imagination that Norton (2001) extends to her work on L2 learning by focusing on the relationship between imagination and investment in communities of practice. Extrapolating the construct of imagined communities and relating it to the context of the classroom, Norton defines an imagined community as ‘‘a community of the imagination˗˗ a desired community that offers possibilities for an enhanced range of identity options in the future’’ (Norton, 2010, p. 355).

Norton stated that when language learners are sitting in a classroom, they may be invested in communities that go beyond its four walls (Norton, 2001). In her studies (e.g. Norton, 2000, 2001), she demonstrated that the learners were capable of imagining the world as different from the predominant realities. Norton (2016) argues that learners' identities are shaped not only by material conditions, lived experiences and their investment in the real world, but also by their investment in the possibleworlds. Investment in these communities offers numerous possibilities for changing the social and educational aspects (Kanno & Norton, 2003).

Kanno and Norton (2003) argued that the notion of imagined communities endows us with deepening our understanding of learning both temporally and spatially. Temporally, it contributes to relating learners’ visions of the future to their principal actions and identities. It can be confirmed that what has not yet occurred in the future can be a source of motivation for the learners' present actions. Spatially, investigating the interaction between national ideologies and individual learners’ identities, and the impact of globalization and transnationalism on language learning and identity construction becomes possible.

The notion of imagined communities presents a theoretical framework for the consideration of desire, hope, and creativity in identity construction (Kanno & Norton, 2003). Identity, known as a multiple, fluid and a "site of struggle" construct (Norton, 2000, p. 127), is assumed as a central issue in examining language learners' imagined communities. Much research has pointed to the powerful force of the imagination in identity negotiations (e.g., Norton, 2001; Pavlenko & Norton, 2007). Norton (2001) argues that a learner's imagined community connotes an imagined identity, and the investment of learners in the target language is understandable in this context. Visualizing an imagined identity in an imagined community can be effective in a learner’s engagement with educational activities (Kanno & Norton, 2003). Pavlenko and Norton (2007) recognize five identity groups, i.e. postcolonial, global, ethnic, multilingual, and gendered identities that are related to English as an international language and may be affected by multiple memberships in imagined communities.

English as a foreign language (EFL)/English as a second language (ESL) learners and teachers should not focus just on one imagined community due to the unpredictability of life trajectories (Khrchenko, 2014). Similar imagined communities may bring about various forms of identities and investments for language learners. This point becomes significant when globalization, super-diversity, and mobility are taken into consideration (Dawson, 2017). Furthermore, EFL/ESL teachers should not determine imagined identities for language learners, nor should they consider the learners as tabula rasa in terms of their life experiences (Kharchenko, 2014).

Learners' participation in some imagined communities can be a strong power to revolutionize the available real communities of participation or to disregard the real community in favor of a future imagined one. The noticeable point is that a newly imagined community is not always a better one. In other words, imagination can have both damaging and consolidating functions (Kharchenko, 2014). What is clear at this point is that Norton’s (2000, 2001) works have indicated that students’ non-participation in particular language practices can be elucidated through their investment in specific imagined communities and the level of their access to these communities. L2 learners’ nonparticipation can be considered as a way of resisting their marginalization in the learning contexts which may be dissimilar to their imagined community (Song, 2018). If there is a clash between learners' projected identity in imagined communities and their perceived identity expectations enforced by others, the learners' participation in classroom practices is withdrawn (Norton, 2000). This can limit learner's access to full classroom participation, linguistic resources, L2 practice and academic success (Song, 2018). As Greene (1995) stated, “Imagining things being otherwise may be a first step toward acting on the belief that they can be changed” (p. 22). If the language learners' imagined communities are not approved by teachers, their non-participation may be worsened and this, in turn, can affect their learning trajectories negatively (Kharchenko, 2014).

Imagined community was the special issue of the Journal of Language, Identity, and Education in 2003. Numerous scholars did research on this construct. For example, Kanno and Norton (2003) pointed out two anecdotal pieces of evidence for the impacts of imagined community on identity and language learning. They showed how the inconsistency of the imagined communities with real ones damaged one participant’s desire to preserve his Japaneseness and for another participant to follow her ESL classes in Canada. In another study, Kanno (2003) focused on schools to investigate imagined community. Using an ethnographic approach, Kanno examined the policies and actions of four schools in Japan and how their visions were related to the concept of imagined identity/community. He observed that the schools' visions for the future of the students shaped the schools’ policies. The results indicated that since students at these four schools were planned to join different imagined communities, they were consequently trained for different types of bilingualism. Furthermore, Norton and Kamal (2003) investigated the experiences of Pakistani middle-school students who participated in a global community education project. In this project, the students aimed to improve the literacy and English skills of a group of Afghan refugee children. These students envisioned Pakistan in the future as an influential member of the international community, in which literacy, competence in English, and technological advances were essential to the development and serene coexistence.

Considering the significance of imagined communities in the process of language learning, SLA researchers should explore the relationship between identity, imagination, and language learning to uncover the influence of the learners' imagined communities on their identity construction and participation (Dawson, 2017). Moreover, since globalization has heightened the range of possible worlds for individuals, extensive research on imagined communities seems to be worthwhile and urgent. The present study reflects such an issue.

Although the construct of imagined communities/identities is well integrated in the wide body of literature on language learning and teaching, most of the studies on this concept (e.g. Kanno & Norton, 2003; Norton, 2001; Pavlenko, 2003) have examined it in ESL ones and there is not enough research on this construct among EFL learners. Moreover, this research has been done through qualitative approaches˗ especially interview˗ and many researchers have not paid enough attention to quantitative or mixed-methods studies. This may be due to the fuzziness and malleability of the umbrella term of identity.

Since identity research was initially done by sociologists, anthropologists, and psychologists, a review of research in these disciplines shows that they generally prefer quantitative approaches (Khatib & Rezaei, 2013). However, in applied linguistics, quantitative approaches are normally ignored in identity research and its related constructs. Nevertheless, the disposition towards quantitative research in neighboring disciplines has been influential in applied linguistics and some studies have followed this movement (e.g. Ehala, 2012; Khatib & Rezaei, 2013; Soltanian, Ghapanchi, Rezaei, & Pishghadam, in press).

Although the literature indicates that studies conducted on imagined communities belong to the qualitative paradigm, now that many complex constructs in applied linguistics, e.g. language motivation, language anxiety, language competence, language identity, and recently investment in language learning, have been translated into quantifiable measures, imagined communities can similarly be explored with quantitative or mixed-methods approaches. To achieve this aim, this study followed three main steps. The first one was to develop a hypothesized model of language learners' imagined communities in Iran as an EFL context. In the second phase, a questionnaire was developed and validated to test the hypothesized model. Finally, the data collected through the questionnaire were fed into the model to see to what extent the model fit the data.

Theoretical Framework of the Study

To develop a model and questionnaire for language learner's imagined communities, the present researchers drew on some theories. The theories and studies on language and identity, especially Norton Peirce’s (1995) conceptualization of identity and language learning, were utilized to support the present study. Norton Peirce adopts a poststructuralist conception of social identity, and emphasizes “the role that social identity plays in L2 acquisition” (Ellis, 1997, p. 243). In Norton’s (1997) view, language learners negotiate their identity/sense of self while learning L2 and are influential in their own process of meaning construction in L2. Thus, “because learning transforms who we are and what we can do, it is an experience of identity” (p. 215). Moreover, as Song (2018) states, L2 learning changes learners' perception of themselves in a new sociocultural context.

This research is mainly informed by theories related to the role of imagination in learning, since the whole study highlights imagined communities and its relationship with language learning and considers the point that learning can be connected to learner participation in a wider world and with people beyond immediate social networks via imagination. The work of scholars, who focus on different but complementary functions of imagination, such as Vygotsky (1978), Anderson (1991), and Wenger (1998), as well as the works of Norton (2000, 2001), were used as strong frameworks in this study.

In Vygotsky's (1978) understanding of the links between imagination, consciousness, and control, imagination is extremely important as a new psychological function. He argues that imagination, linked to the development of learner’s consciousness, gains an important educational function (Pavlenko, 2003). Anderson's (1991) view of nation-states as imagined communities implies that imagination happens on a societal and not just an individual level, in the form of ideologies of nationhood. His view reflects the social context for imagination. Consequently, for Anderson (1991), imagination has ideological and identitary purposes (Pavlenko, 2003).Wenger’s (1998) view of imagination as a form of engagement with communities of practice, or situated learning theory, develops Anderson’s view of imagined communities to any community of practice an individual wishes to enter and presents imagination as both an individual and social process. He argues that imagination is a unique form of belonging to a particular community of practice via which individuals place themselves and others in the world and take in their identities “other meanings, other possibilities, other perspectives” (p. 178). So, for Wenger, imagination marks both an educational and an identitary role (Pavlenko, 2003).

The present study was also built on Norton’s (2001) conceptualization of imagined communities which definitely connected the notions of imagination and imagined communities with L2 learning and use and with classroom practice. Through this lens, it can be argued that learners' behaviors and choices are linked to their investment in particular imagined communities. For Norton (2000, 2001), imagination has both educational and identitary functions (Pavlenko, 2003). Imagined communities/identities are related to this study as it revolves around how English language causes Iranian EFL learners to have different imaginations expanding their range of identities and reaching out to wider worlds and how their imagined communities/identities affect their investment in language learning and offer them possibilities for the future.

Markus and Nurius’s (1986) theory of possible selves as the link between motivation and behavior was another framework employed in this study. Wenger’s (1998) views are consistent with the psychological theory of possible selves which characterizes humans’ thoughts of what they might become, what they would like to become, and what they are afraid of becoming, hence connecting behavior, cognition, and motivation (Pavlenko & Norton, 2007). The notion of imagined communities can expand one's range of possible selves. Not only does it provide identity prospects in the future, but it is also an image of how they identify themselves and who they want to become (Kanno & Norton, 2003).

The concept of investment (Norton Peirce, 1995) also merits attention due to its significant function in conceptualizing the identity/imagined community relationship. Moreover, the concepts of investment and imagined communities are directly related to agency and identity construction. So, Norton Peirce’s (1995) investment hypothesis in L2 learning and Darvin and Norton's (2015) investment model also made the cornerstone of the present research and provided the theoretical lens through which the present researchers would address the complex relationship between Iranian EFL students, their L2 interactions in the classroom, and their constantly shifting identities and investment in learning English. Norton (2000) believes that learners' investment in language learning happens probably because of their recognition that they might have different future possibilities. On the other hand, if learners can imagine themselves participating in different imagined communities, they can show more investment in language learning.

Globalization and language learning and teaching subjects (e.g. Coupland, 2010) were also utilized in this study since the notion of imagined communities at the present time is closely connected with globalization and technological advances. The theory of bilingualism and bilingual education (e.g. Baker, 2011) was also one of the theories which helped in composing some items of the questionnaire. Moreover, sociolinguistics of identity (e.g. Omoniyi & White, 2006), sociology of language (e.g. Bourdieu, 1991; Spolsky, 2011) and sociocultural theory and its extension to SLA studies (Lantolf & Appel, 1994; Vygotsky, 1978) were helpful to develop the model since this study deals with the sociolinguistic domain of language studies. These theories have encouraged SLA researchers to consider the socio-historical contexts informing language acquisition and understand the relationships between individual and society, learning and becoming, and the role of human agency and interaction in SLA. In addition, language policy issues (e.g. Spolsky, 2003) were also used because in Iran the main language policy is to value Persian language approximately in all centers. Eventually, some local works on identity and investment were effective in developing the model (e.g. Khatib & Rezaei, 2013; Soltanian, Ghapanchi, Rezaei, & Pishghadam, in press).

Method

Participants

This research was conducted in one year from February 2017 to February 2018. The respondents were English language learners in four universities and five language institutes in three cities in Iran, from different ages, genders, language proficiency levels, and educational backgrounds. For the initial piloting phase of the study, there were 58 respondents and for the reliability and validation phases, 389 participants, including 233 female (59.9%), and 156 male (40.1%) took part. The comments of six experts and non-experts on the diction of the items, and content of the questionnaire were taken into consideration and some revisions were made in the items of the questionnaire based on them. To seek experts' opinions, the researchers drew on Delphi technique. The steps taken in the collection of the experts' opinions via several rounds of interviewing or survey and applying the received feedback into a unified whole as a point of agreement among the experts shapes the modified version of a multiphase research approach for tracing experts’ opinions called Delphi technique (Timmerman, Strickland, Johnson, & Payne, 2010). In the current study, to reach a consensus in the experts' opinions on the components of the model and questionnaire items, the researchers asked the experts to reassess their initial judgments regarding specific components and items provided in previous rounds and modify their opinions in later iterations by reviewing and assessing the comments and feedback provided by other experts. In fact, a summary of other experts' comments was given to each expert. In this way, they became aware of the range of opinions and the reasons underlying those opinions.

Data Collection Procedure

The Hypothesized Model of Language Learners' Imagined Communities in Iran

Developing a model for language learner's imagined communities, which is followed by developing a reliable and valid questionnaire to test the model, is beneficial to alter the common qualitative approach to imagined communities research and make it possible to do large-scale surveys. The initial aim of this study was specifying the main components of language learner's imagined communities in Iran to develop the language learner's imagined communities model. That is, the researchers wanted to know what components exactly form the Iranian language learner's imagined communities. The hypothesized model was developed after the researchers reviewed the related literature in a detailed way and went through interviews with a group of language learners and consultations with experts in the fields of sociology and sociolinguistics. Twenty language learners were interviewed, in 20 to 30 minutes, to draw themes from what they said. The interviews with the language learners were held in Persian. The content of the interviews emphasized language learners' imaginations while learning English especially in the language classroom. To ensure the comprehensiveness and accuracy of these components, the specified components were shown to three experts in the fields of linguistics, sociolinguistics and sociology to reflect on them. Their constructive comments were taken into consideration and some modifications were made in the components based on them.

To test the model, a questionnaire was developed and validated in the next stage through a number of meticulous stages. The data from the questionnaire was then fed into the model to test the model fitness. All these stages with their complex statistical procedures are explained henceforth.

Questionnaire Development

In the present study, Brown (2001), Dörnyei (2010) and Khatib and Rezaei (2013) were used as guidelines for the questionnaire development and validation. To develop a reliable and valid questionnaire, the researchers followed several rigorous and systematic steps. In the following sections, these steps are explained.

Item generation. To accomplish the first step of questionnaire development, the related literature must be reviewed to be informed about the existing instruments and establish a robust theoretical framework for the instrument. Since the researchers had already reviewed the related literature, they bypassed it and moved to the self-initiative item generation step directly.They employed multi-item scale in generating the items.

The researchers were careful to generate items tapping the components of the model. They tried to generate simple and short items using natural language away from any ambiguous and loaded words. Both positively and negatively worded items were included to avoid the problem of responding haphazardly. Moreover, the sensitive items were not placed at the beginning of the questionnaire (Dörnyei, 2010).

Finally, a pool of items was generated and after frequent reviews, some of these items were discarded. The researchers aimed to develop a short questionnaire but not to the point of excluding the fundamental points. This goal was achieved by having the questionnaire not take more than 25 minutes to be filled out.

Item checking with experts. After generating the items, six experts and non-experts were requested to check the accuracy, intelligibility, content representativeness and bias in the items. Since the non-experts were also the final respondents to the questionnaire, their feedbacks were informative for the researchers to know which items must be removed due to their unnecessary jargon and loaded words. The panel of experts included professionals in the fields of applied linguistics, sociology, sociolinguistics, and survey design.

The decision to keep an item or omit it was finally made based on the panel of experts' opinions. This panel was requested to rate the items of the questionnaire based on a Likert-type scale from one to four. According to this scale, they commented on the items as 'Not important to be included in the survey', 'Somehow important to be included', 'Important to be included', and finally 'Extremely important to be included in the survey'. They were further asked to express their final decision on each item by telling either 'delete' or 'preserve' the item. If the majority agreed on the acceptability of the item, the item was subsequently kept. As a general rule in this study, items which received more than 70% of acceptability were kept for the next step. The results of the responses were helpful in decreasing the items from 72 to 61 items. Hence, 11 items were omitted because of a number of reasons mentioned by the panel including the redundancy, length, irrelevance, and ambiguity of the items.

Item translation and revision.One of the researchers translated the English version into Persian to make it easy for all respondents, who were from different language proficiency levels, to complete the questionnaire. Back-translation was employed to ensure the accuracy of translation. Then the English back-translation and the original English items were examined, and the Persian translation of some items was revised. It was double-checked again by another expert for translation accuracy. Finally, two Persian editors of Iranian newspapers edited the Persian version and made it standard Persian. The Persian version was administered upon the participants' requests to increase the return rate.

Initial piloting and item analysis. In this step, the questionnaire was administered by hand to 58 students similar to the target population, i.e., Iranian EFL learners, for which the questionnaire was intended to be developed. Before administering the questionnaire, the respondents were ensured that the information obtained would be kept anonymous to make them relaxed while answering the sensitive items. The title of the questionnaire, imagined communities among Iranian EFL learners, was deleted during its administration since it might have been influential in the responses.

Respondents showed their degree of agreement/disagreement to each statement on a six-point Likert-type scale including strongly agree, agree, slightly agree, slightly disagree, disagree and strongly disagree. To score the items, ‘strongly agree’ got six points, ‘agree’ five points, ‘slightly agree’ four points and so on. For the negatively worded items, reverse coding was done and the scoring was reversed.

The respondents' feedbacks were useful in modifying some items and removing four. Therefore, the remaining items in the questionnaire were 57.

Validation and reliability estimation. In the last two stages, the researchers investigated face validity, content validity, and construct validity of the questionnaire as the main types of validity for questionnaire validation in this study. The internal consistency of the questionnaire was also measured through Cronbach's Alpha coefficient. In the next section, the results for these two stages are given.

Results

The Hypothesized Model of Language Learners' Imagined Communities in Iran

After reviewing the literature and relevant theories iteratively, conducting interviews with 20 language learners and consultations with a cadre of experts in the fields of applied linguistics, sociology, and sociolinguistics, the researchers found a number of components which encapsulated Iranian language learner's imagined communities.

The eight components identified for language learners' imagined communities in Iran were as follows:imagination and the learners' desires for belonging and recognition, expanding one's range of possible selves by imagined communities, marginalization, non-participation and resistance in language classroom or outside, trying to attain a legitimate membership (moving from peripherality to legitimacy), gender, power, and material inequalities, access to different capitals (economic, cultural, social, and symbolic), identity construction and promotion, and finally language learner's agency. Table 1 provides the definition for each of these components.

Table 1. Components of Language Learners' Imagined Communities in Iran along with Their Definitions

Instrument Validation

Three types of validity were examined in this study: face validity, content validity and construct validity. As response, predictive and concurrent validities were not applicable in this study, they were not investigated.

To establish the face validity of the questionnaire, the questionnaire was distributed in a good shape enjoying reader-friendly font type, margin, color, and paper. So, the researchers did their best to meet the criteria for face validity.

To check the content validity of the questionnaire, as mentioned before, three experts commented on the representativeness and appropriateness of the questionnaire items and its instructions. Moreover, six English language learners in language institutes and university were asked to read the items and give their comments on their intelligibility. Some minor modifications were made in the wording of the items based on their comments.

Two techniques were utilized by the researchers to establish the construct validity of questionnaire. They first searched for the relevant theories of language learners' imagined communities in the literature and ensured themselves of the congruency of the questionnaire items with those theories. Second, to statistically check the construct validity, the researchers intended to run both exploratory and confirmatory factor analyses. However, after consultations with some statisticians and researchers in the field, they ignored exploratory factor analysis because the factors had already been specified iteratively by detailed review of the literature and the well-developed underlying theories while proposing the hypothesized model in the previous steps. In the following section, the procedure for running confirmatory factor analysis (CFA) is described.

Confirmatory Factor Analysis and Testing the Model Fitness

The purpose of running CFA in this study was to check if the questionnaire data fit the hypothesized model. To do so, the questionnaire was administered to 389 English language learners whose ages ranged from 14 to 36 years with a mean age of 20 years. They were from different universities and language institutes in Mashhad, Neyshabur, and Semnan, Iran. The number of years the participants had studied English language was different. Generally, it ranged from 5 to 12 years. The questionnaire was sent via Telegram or email and in some cases it was given by hand. The response rate of the questionnaire was 98.2%.

Table 2 demonstrates the descriptive statistics (age, gender, language proficiency level, and education level) for the participants in this phase of the study. The Statistical Package for Social Sciences (SPSS version 22) was used for inputting data and computing descriptive statistics.

Table 2. Demographic Information of the Participants Taking Part in CFA Phase

Analysis of Moment Structures (AMOS 21) software was employed to test the hypothesized model. AMOS software was run to perform CFA and Structural equation modeling (SEM). SEM is a powerful multivariate analysis technique used to explore causality in models and the causal relations among variables and confirm the proposed structural theory. SEM indicates the relationship between latent variables, that is, the components of language learners' imagined communities, and the observable variables, i.e. the questionnaire items generated for each of the components.

Before testing a structural model, all latent variables should be validated using CFA (Hair, Anderson, Tatham, & Black, 1998). Based on the CFA analysis, the association between each sub-factor of the proposed model was analyzed and the results can be seen in Figure 1. The model with all factor loadings can be seen in this figure. This model included eight sub-constructs (sub-scales): A. Imagination and the learners' desires for belonging and recognition (IM1), B. Expanding one's range of possible selves by imagined communities (IM2), C. Marginalization, non-participation and resistance in language classroom or outside (IM3), D. Trying to attain a legitimate membership (moving from peripherality to legitimacy) (IM4), E. Gender, power, and material inequalities (IM5), F. Access to different capitals (economic, cultural, social, and symbolic) (IM6), G. Identity construction and promotion (IM7), and H. Language learner's agency (IM8).

To examine the validity of the measurement model and check the model fitness, goodness-of-fit indices were used (Kline, 2011). There are several fit indices that show the adequacy of the measurement models. In this study, Chi-square/degree of freedom (χ²/df), goodness-of-fit index (GFI), comparative fit index (CFI), Normed Fit Index (NFI) and root mean squared error of approximation (RMSEA) were used. To have a fit model, χ²/df should be a value of less than 3 (Tseng & Schmitt, 2008), GFI CFI, and NFI should be above .90 (Hu & Bentler, 1999), and RMSEA should be less than .08 (MacCallum, Browne, & Sugawara, 1996).

Because some measurement models did not show adequacy to the data, the researchers made some modifications on the model. These modifications included the removal of one item, (q.34) in IM5 sub-construct, due to low loadings. Error terms of four IM6 items (items 37 and 38 and 41 and 43) were correlated because each pair of these items refers to the same content. After doing these revisions, the model indicated acceptable fit to the data. Goodness of fit indices for SEM before and after modification can be seen in Table 3. This table shows that all the goodness-of-fit indices are within the acceptable range. Therefore, the scale enjoyed perfect validity and the model seemed to be a fit one. The final model showed a very good fit to the data. In other words, the data gathered in this study seemed to support the model.

Table 3. Goodness-of-fit Indices for the Model

Figure 1 illustrates the schematic representation of the final model of language learners' imagined communities in Iran. Path coefficients are also put on the pathways from each latent variable to other latent or observable variables to demonstrate the strength of relation or correlation among the variables. The results indicated that no modification was needed.

Figure 1. Final model of language learners' imagined communities in Iran

Note: IM1, IM2, IM3, IM4, IM5, IM6, IM7, and IM8 are the components identified in this study

Reliability Index

To measure the internal consistency of the questionnaire, Cronbach’s Alpha coefficient was employed. The researchers followed Dörnyei (2010) to determine the acceptable measure for the reliability index of the questionnaire and chose indexes below 0.60 as weak and above that as acceptable. To calculate the reliability, it had already been administered to 389 Iranian English language learners in the CFA phase. The results for the Cronbach’s Alpha analyses showed that the internal consistency of the questionnaire was 0.95, and for the eight subscales of the questionnaire, it was estimated to be 0.96, 0.94, 0.93, 0.94, 0.92, 0.97, 0.84, and 0.89 respectively. Table 4 shows the reliability indices obtained from Cronbach's alpha analyses for the subscales of the questionnaire along with the items. As this table demonstrates, the questionnaire gained high indices of Cronbach's alpha as a whole as well as in its subscales.

Table 4. Subscales of the Questionnaire, the Related Items, and Reliability Indices

Discussion

This study made an endeavor to develop and validate a questionnaire for understanding Iranian EFL learners' imagined communities. To this end, initially an eight-factor model of language learners' imagined communities was developed based on the comprehensive review of the related literature and the established theories, researchers' consultations with experts and interviews with language learners.

The results of the first phase of this research, which shows the main components of Iranian EFL learners' imagined communities, are especially informative. The first component, i.e. imagination and the learners' desires for belonging and recognition, specifies the desired community of language learners. The second component which forms the expansion of learner's range of possible selves by imagined communities reflects the different selves that language learners may imagine. The third component, i.e. classroom nonparticipation, indicates the reasons and forms of learners' disengagement in the classroom practices. The fourth component, recognizing language learners as legitimate individuals, stems from the point that language learners must be recognized as legitimate members who can claim the right to speak (Norton, 2013). The fifth component, gender, power, and material inequalities, was included to show how different types of inequality, which may exist in the language classroom, affect learning conditions. The sixth component is indicative of the numerous capitals gained through language learning. The seventh component, i.e. identity construction, is the core element and reflects the identities shaped by the learner during the process of language learning. The final component specifies the learners' agency and responsibility in the language class.

The hypothesized model was examined on a sample of 389 EFL learners using CFA. The data collected through the 57-item questionnaire were fed into the model. Actually, as the third objective of this study, the fitness of the model was tested via SEM. During CFA phase, it was shown that some measurement models did not show adequacy to the data. So, some modifications were made on the model. These included the removal of one item, (q.34) in IM5 sub-construct, due to low loadings. It did not load on gender, power, and material inequalities. Furthermore, error terms of four IM6 items (items 37 and 38 and 41 and 43) were correlated because there was content overlap in each pair of these items. To achieve satisfactory fit, it was essential to include a correlated error term between them. After making these revisions, the model indicated acceptable fit to the data. Therefore, the CFA confirmed the factor structure of the model, and all the eight initially proposed components in the instrument were verified by the collected data. The majority of the questionnaire items remained intact and only one item was deleted. The calculated model-fit approximations verified the CFA model as valid for recognizing Iranian EFL learners' imagined communities. Moreover, the Cronbach's Alpha estimated the reliability of all the items as 0.95. All of the eight factors yielded good reliability estimates ranging from 0.84 to 0.97. On such grounds, it can be claimed that the scale developed in this study can be considered as an efficient scale for understanding the status of language learners' imagined communities in the Iranian context.

Further research is required to examine why item 34 was omitted in the confirmatory analysis. It can be argued that the removal of this item was probably caused as a result of the educational system of Iran in which students do not usually experience co-education until they enter the university. Hence, they cannot be sure of better learning in a co-educated classroom.

Generally, this study increases our awareness of EFL learners' imagined communities both theoretically and practically. It can prompt language teachers to delve deeply into their learners' imaginations while learning English and employ this power to intensify the sense of devotion to learn this language, since the learner’s investment in learning the target language has a direct connection to the imagined community to which the learner desires to belong (Norton & Kamal, 2003).

Since learning another language, probably more than any other educational activity, demonstrates learners' desire to multiply their range of identities and to imagine wider worlds (Norton, 2001), language teachers should admit learners’ imagined communities and the connections to investment and identity construction. What makes attention to learners' imagined communities significant is that these communities have the potentiality to be even more powerful than face-to-face communities in shaping the language learners' investment (Kanno & Norton, 2003) and that language teachers are influential in generating, encouraging, or preventing such imagined communities (Murphey, Chen, & Chen, 2005).

Conclusion and Implications

Following the point that sociolinguistics research in language learning has been confined to exclusively qualitative approaches, the current researchers tried to do a different work and move towards quantitative approaches in imagined communities research. In fact, the absence of an instrument to specify language learners' imagined communities motivated this study. The results of the study indicated that though the model was the first attempt for the Iranian context, it enjoyed a reasonable degree of reliability and validity as verified by the statistical indices from SEM. Furthermore, the questionnaire showed a respectable degree of reliability and validity.

Although qualitative approaches are the top priorities in sociolinguistics research, due to giving thorough accounts, their latent problems such as being time-consuming, costly for administration and scoring (Khatib & Rezaei, 2013), and being less generalizable (Rezaei, 2017) should not be overlooked. The model and questionnaire development can overcome these limitations and provide the ground for ongoing research (Khatib & Rezaei, 2013). The model developed and validated in this study can be a framework for research on imagined communities in Iran. Teachers can usefully employ the questionnaire to recognize EFL learners' imagined communities and teach accordingly. Moreover, the procedures utilized can provide guidelines for future researchers to develop and validate a model.

Future studies on language learners' imagined communities may reconfigure or expand the model for the possible inclusion of other components or items to improve both the model and questionnaire. Studies are required to test if this model is truly applicable in Iran across time and place. These studies may also reduce probable confounding variables and improve the reliability and validity of the model and establish it as a robust one.

Considering the contextual nature of constructs like imagined communities, the researchers should be careful about the generalizability of the model proposed here which is limited to the Iranian context. Studies should be done to test if this model is applicable in Iran across time and place and also if it can be used as a valid model in other contexts. Future research can examine the effect of different variables, e.g. demographic data, on imagined communities. Researchers in non-Iranian contexts can also use the model and questionnaire if they are rechecked for their reliability and validity and altered based on the contextual variances. Given the unavoidable effects of technological advances on the scope of individuals' imaginable communities, it can be inferred that imagined communities, similar to identity and investment, is a fluid, fuzzy, multiple, malleable, and dynamic construct with a contextualized nature. Hence, developing a more fluid model is something which bears further investigation. That model can encompass the constructs and visually depict the interrelatedness of the subcomponents in a spherical way.

The criticism which might be made about the current model is the effort to quantify imagined communities which is qualitative, dynamic, and fuzzy in nature. As the literature shows, models have also been developed for other fuzzy concepts in language research such as language proficiency, anxiety, critical thinking, language competence and language identity. The last point which is worth mentioning is that although questionnaires have various advantages, such as being invaluable instruments for large-scale surveys in a short time, meeting generalizability in results, yielding rich data, and making it possible to extrapolate data easily and score objectively, in doing research studies it is recommended to mix quantitative approaches with qualitative ones to overcome the defects in the data collected through each approach (Rezaei, 2017).