A Potentially Heteroglossic Policy Becomes Monoglossic in Context: An Ethnographic Analysis of Paraguayan Bilingual Education Policy

Profesora Noemí was one of three staff members at the Paraguayan Ministry of Education in 1993 who were charged with developing a bilingual education program design that would fulfill the new constitutional directive for mother tongue and bilingual education for all in a newly declared bilingual country. She and her two colleagues chose a strong, additive program design (Baker 2011), similar to many 90/10 enrichment dual language programs in the United States—this in a time and place where, despite most children's dominance in Guarani, instruction had always officially been in Spanish. They wanted the programs they designed to change that: they wanted to claim their language, to ensure that their language continued alive, while not laying aside Spanish. The program that emerged, along with its subsequent elaborations, is one that in many ways might be considered heteroglossic (García 2009)—or one that recognizes multiple, co-existing language norms and identities, one based on an assumption of multilingualism, rather than monolingualism, as the norm. Yet, when I spent a year sitting in classrooms created under this policy, instructional practices and the ideologies that framed them were decidedly monoglossic, oriented instead toward either a two-solitudes bilingualism (Cummins 2008) or a monolingual, standard Spanish-speaking norm.

The purpose of this paper is to describe, in part, how a policy with plenty of ideological and implementational space (Hornberger 2005) for multiple language norms was appropriated in ways that still did not adequately serve language minoritized children. While the counterhegemonic practice of teaching Guarani language in school was present in classrooms, differences in the varieties of Guarani used and how they were used implied uneven benefits for rural and urban children—thus limiting the real counterhegemonic effects of the policy. I argue that this inequitable pattern occurred in part because the national policy texts and local appropriations included assumptions of language separation as the norm, while norms of language use were characterized by much intralanguage variation and language mixing. Though the policy successfully disrupted the traditional exclusion of the minoritized language from education, it failed to disrupt hegemonic monoglossic ideologies that continue to exclude Guarani-dominant students from high-quality education. I discuss the importance of explicit articulations of heteroglossia in educational language policy for increasing the likelihood that policy can and will lead to greater educational equity and social justice.

Ethnographic and Discursive Approaches to Educational Language Policy

I frame this analysis within a larger effort to better understand the ways that educational language policies take shape through the actions of teachers, administrators, children, and community members, often in ways that are unpredictable from policy text alone (see for example, Hornberger 2005; Menken and García 2010). This work builds upon research framing policy as discourse (Bacchi, 2000; Ball, 1993), a focus that helps us to see policy as collections of linguistic signs that encode some set of representations that then must be made sense of in context by people who interpret them. It also builds upon a contemporary wave of language policy research that draws on anthropological approaches to policy as sociocultural practice (Levinson and Sutton 2001; Levinson et al. 2009; Shore and Wright 1997). Rather than assuming a simple, linear implementation in which official words have taken-for-granted referential meanings that do or do not become on-the-ground actions, these approaches characterize implementation as a process of appropriation: policy actors actively construct the meaning of policy, “interpreting and adapting such policy to the situated logic in their contexts of everyday practice” (Levinson and Sutton 2001, 17). This frame accounts for the roles of both actors’ agency and larger political and social structures in the implementation of policy, and it helps to reveal the simultaneous emergence and constraint of cultural patterns in the process. Much of this growing body of work uses ethnography to study language policy (Hornberger and Johnson 2007; Jaffe 1999; Johnson 2009a; McCarty 2011) in an effort to capture local actors’ meaning-making processes, and it attends in particular to the ways that educators act as policymakers from the bottom-up in otherwise top-down policy contexts (Johnson 2009b; Menken and García 2010; Ricento and Hornberger 1996). These ethnographic approaches dovetail with increasing attention to the role of local norms, practices, and epistemologies in various aspects of language policy activity (Canagarajah 2004; Hornberger 1996). Together this work demonstrates that when we look at language policy text, we have but a partial understanding of its meaning. For a fuller understanding of its meaning we must look to the contexts into which it is taken and that policy actors use to make sense of it.

Contexts of Appropriation as Scales

Ethnographic and discursive approaches to educational language policy (ELP) showing the primacy of contexts of appropriation are important for understanding how policies succeed and fail to meet the goal of improved education for marginalized students. My principal interest in language policy is in how we can better ensure that it does meet this goal. Following other ELP researchers (Blommaert 2007; Collins 2012; Hult 2010), I use the concept of spatiotemporal or sociolinguistic scales (Blommaert 2007, 2010; Canagarajah 2013; Collins, et al. 2009) as a way of understanding the contexts into which ELP is appropriated and the varying effects that appropriations have. Jan Blommaert (2007, 2010) shows how all language uses index sociolinguistic scales that have both spatial and temporal dimension and that are hierarchically ordered. This concept is useful in understanding how particular, local appropriations and instantiations (Johnson 2012) of ELP get linked to higher-scale policy texts, and it is also useful in understanding how these uses of linguistic resources serve or do not serve as means of empowerment for minoritized students in interactions.

Some uses of linguistic resources produce power in interaction by indexing higher sociolinguistic scales; they do this by upscaling or lifting a particular topic or moment to “a higher level of relevance, truth, validity or value” (Blommaert, 2007, 6). In such moves, sociolinguistic resources that are construed as decontextualized, translocal/widespread, more timeless, more homogenous—like a standard language, for example—tend to index higher scales, while those that are construed as more contextualized, locally situated, and heterogeneous—like a nonstandard or hybrid variety, for example—tend to index lower scales. Monoglossic ideologies such as Michael Silverstein's (1996) monoglot standard regiment these indexical relationships. Monoglossic ideologies assume “that legitimate linguistic practices are only those enacted by monolinguals” (García 2009, 115). While these ideologies permit the recognition of multiple languages, they erase (Irvine and Gal 2000) intralanguage variation and hybrid language use, as well as multiplicity and dynamism of identities. Concepts such as “mother tongue” and “second language” are monoglossic concepts (García 2009; Skutnabb-Kangas 1981).

Despite the power of these indexical tendencies, Suresh Canagarajah (2013) shows how people in multilingual communities also regularly disrupt them in interaction, drawing on ideologies that frame hybrid language practices as normative and indexical of higher, rather than lower, scales. For example, he analyzes how skilled African migrants working in English-dominant countries indexically link the use of diverse varieties of English to a translocal, cosmopolitan space: a higher scale rather than a lower, more local one. In those interactions, people scale, locate, contextualize heterogeneous practices up rather than down. These heteroglossic ideologies recognize “multiple co-existing norms which characterize bilingual speech” (García 2009, 117). This frame is derived from Bakhtin's (1981) heteroglossia—the multiplicity of social voices in the novel—and on understandings of languages as constructed (Makoni and Pennycook 2006). Heteroglossic ideologies recognize intralanguage variation and translanguaging (Creese and Blackledge 2010; García 2009; Hornberger and Link 2012) or translingual practices (Canagarajah 2013). They frame identities as multiple and shifting and languages as porous, interdependent, continuous, and as products of social process rather than as thing-like codes. In fact, a heteroglossic frame reveals not languages but mobile linguistic resources (Blommaert 2010; Canagarajah 2013) that are combined and recombined in ways that crisscross traditional linguistic and territorial boundaries. In what follows I describe aspects of Paraguayan policy for universal bilingual education that encourage its uptake as heteroglossic, and I also describe aspects that encourage monoglossic practices and frames. I then describe ways in which the policy was taken up in two elementary schools. Despite some important acknowledgment of and space for heteroglossia in policy text, monoglossic frames were those largely taken up in practice in the focal schools. Through those frames Guarani-dominant students’ languaging practices continued to index lower sociolinguistic scales, and the counterhegemonic potential of this policy was attenuated.

The Paraguayan Context and Study Methods

To understand how the national policy for universal bilingual education was appropriated in local contexts, I spent 11 months (January through November 2008) conducting fieldwork in two Paraguayan elementary schools—one rural and one urban. Data included nearly 500 hours of participant observation and field notes; audio recorded interviews with 91 individuals including teachers, administrators, parents, students (interviewed in groups), Ministry of Education supervisors and officials, language activists and language scholars; 14 hours of videotaped classroom interaction in each of two focal sixth-grade classrooms, one in each school; policy and school documents; and student work. Analysis of the data included open coding, initial memos, and focused coding (Emerson et al. 2011) and triangulation of patterns across multiple sources, time periods, and types of data (Hammersley and Atkinson 2007; Maxwell 1996), all facilitated by qualitative data analysis software. I compared these patterns across the two research sites.

I am a U.S. American who had lived and worked as a teacher in Capitán Antón for two and a half years and had conducted two summer research trips prior to fieldwork. I am a speaker of U.S. academic and Paraguayan Spanishes with limited oral and more literate competence in Guarani. My outsider identities afforded me some perceived neutrality in many questions of specific language policy, though I shared with participants that I was passionate about bilingual education and the inclusion of minoritized language speakers in school. The cultural and linguistic competences and social network I had developed previously, as well as daily living in town with my husband and our infant daughter, helped me to be seen at least as a knowledgeable foreigner with some grounding in the community. All interviews were conducted in Spanish, which for most adults was the language they expected to speak at school and with a foreigner. Nevertheless, I would undoubtedly have been able to access additional nuances and perspectives had my competence in Guarani been enough to truly allow interviewees to speak however they desired. Video recorded classroom interaction that included significant portions in Guarani was transcribed and translated into Spanish by my Guarani teacher.

The bulk of my time was spent at the two elementary schools in the two focal sixth grade classrooms. Being in both a rural school and an urban school was important because of differences in socioeconomic status and language dominance across these areas. Urban Escuela Coronel Insfrán (ECI) was the oldest and largest school (1,800 students pre-K through ninth grade) located in the center of Capitán Antón, a town of about 185,000 people in the Central Department of Paraguay. Profesora Carla, the teacher of the focal sixth grade classroom there, was a veteran teacher with 18 years of experience and a university degree in Guarani language, in addition to the profesorado degree required of all teachers. She lived in Capitán Antón and identified as Spanish dominant but spoke Guarani fluently and comfortably, and believed strongly in its inclusion in formal education. Her class included 23 students who participated in the study (of 27 total1), all 11 or 12 years old, and all identified Spanish as their first language except for one who listed Guarani, two who listed Jopara2 (mixed Guarani/Spanish), and two who listed both Spanish and Guarani. Their families tended to be middle class, with parents giving the following as their occupations: homemaker, worker, businessman, hairdresser, merchant, student, mechanic, jeweler, military employee, bus driver, bricklayer, auto body repairman, chauffeur, military captain and lawyer, guard, teacher, aviation mechanic, and butcher. Many spoke Guarani and identified as Spanish dominant.

Rural Escuela San Blás was located in a community I call Ypa Guazu, and although it was only 12 kilometers from the center of Capitán Antón, it took 40 minutes to reach by bus, and even those ran infrequently and unreliably during the time of my field work. In contrast to urban ECI, Escuela San Blás (ESB) was both smaller in number and size of buildings and classrooms and in number of students. Economic resources were also fewer. The school housed both the elementary level with about 100 students and a junior high and high school technical program focused on agriculture with between 40 and 80 students. Because of the small number of students, each teacher in the elementary school taught two grade levels—one in the morning and one in the afternoon. Profesora Elena taught fifth grade in the afternoon and sixth grade in the morning. She was also a veteran teacher with 14 years of experience at the time of the study, and in addition to the required profesorado, she was additionally working on her undergraduate degree at a university in the capital. Profesora Elena lived in a nearby community and described herself as more comfortable expressing herself in Guarani than in Spanish, and in general, she and the other teachers spoke Guarani among themselves. Her sixth grade class included 21 students who participated in the study (of 24 total), and they ranged in age from 11 to 15 years old. Some had experienced periods of illness or economic hardship that had kept them out of school for a time, and they were back to continue studying at an older age. All but three identified Guarani or Jopara as their first language. Their families tended to live on more limited means, and parents gave the following as their occupations: homemaker, vendor, ranch overseer, worker, seamstress/tailor, bricklayer, chauffeur, domestic employee, lottery ticket vendor, gardener, shoemaker, midwife, farmer, and several gave no answer. According to Profesora Elena and other teachers, many of the families experienced significant financial insecurity. Parents tended to speak Guarani with teachers and each other, though many spoke Spanish as well.

As a larger context, Paraguay has been known in the sociolinguistic literature for widespread, stable bilingualism in Paraguayan Spanish and Paraguayan Guarani (Fishman 1967; Rubin 1968), a language of indigenous origin now spoken largely by people who do not identify as indigenous.3 Census data from 1950 through 2002 indicate that for some time the proportion of Paraguayans who identify as speaking both languages has been around half (Gynan 2001; Paraguay DGEEC 2002; Zajícová 2009). Census data and popular perception find Guarani use concentrated in rural areas and bilingualism in urban areas (Gynan 2001). A number of researchers have observed some functional distribution of the two languages such that Guarani is preferred for domestic, informal, and oral contexts, and Spanish is preferred for public, formal, and written4 contexts (Choi 1998; Corvalán 1982; Pic-Gillard 2004; Rubin 1968).

Despite the importance of these patterns, the production of census data and the enumeration of languages and speakers are, by nature, monoglossic processes that erase sociolinguistic diversity and are deeply connected to the production of unequal power relations (Gal and Irvine 1995; Hill 2002; Mühlhäusler 1996). When the 2002 Census asked, “What is the language most commonly used in the household?” and heads-of-household were permitted only one answer, a vast proportion of everyday languaging activity in Paraguay was rendered invisible. Similarly, influence of Spanish on Guarani and of Guarani on Spanish in Paraguay is well documented (Choi 1998; De Granda 1995; Gynan 2001; Morínigo 1931), as is the existence of a continuum of hybrid varieties including Jopara or Jehe'a (both terms for mixed language) (Boidin 2000; Krivoshein de Canese and Corvalán 1987; Lustig 1996; Meliá 1974) and guaraní paraguayo (Paraguayan Guarani). When respondents, census takers, and researchers ask about or talk about Spanish or Guarani, a wide variety of languaging practices may be subsumed or elided. Fluidity across language boundaries and variation in speakers’ practices across the contexts, periods, and movements of their lives may be erased. As more and more scholarship shows, describing how languages and varieties are spread across people and domains of use, enumerating and naming languages, speakers, and varieties all inherently involve some ideological and discursive fixing of language use that does not necessarily represent all the languaging practices of a community. These are problems common to all sociolinguistic descriptions, and they are particularly salient when describing Paraguay, where it must be emphasized that varieties of Guarani and Spanish are not discrete and neither are the social identities associated with them, even though census takers, teachers, parents, journalists, linguists often represent them as such. These problems of representation are the very patterns that the concepts of heteroglossia and translanguaging target for disruption. They are also problems of representation key to understanding how a strong, additive bilingual program design—the kind doing the most to improve education for language minoritized children in some contexts (Thomas and Collier 2002)—has not yet fulfilled its promise.

A Strong, Additive Bilingual Program Design in Policy Text

As described above, the new Paraguayan constitution of 1992 included a mandate that all students would receive initial literacy instruction in their mother tongue (either Spanish or Guarani) and instruction in both official languages thereafter. The design of policy and programs that would fulfill this constitutional mandate fell primarily to a team of educator consultants at the Ministry of Education (MEC hereafter), including Profesora Noemí (Interview, 10/13/2008) and Profesora Teresa (Interview, 10/15/2008). According to both, they studied bilingual program designs in the research literature and consulted with international bilingual education researchers. They produced a plan that included two program designs, or modalidades (modalities): one for students described as hispanohablantes—Spanish speakers—and one for students described as guaranihablantes—or Guarani speakers. Both specified exclusive use of the students’ mother tongue in the first year (grade 1). Each year thereafter the proportion of L2 would increase until roughly equal proportions (Spanish 55%, Guarani 45%) after sixth grade. Importantly, within those proportions both languages were to be used as mediums of instruction as well as subjects of instruction.

This program design can be seen as heteroglossic in some respects and monoglossic in others. It was monoglossic, first, in that the program design was originally built upon an assumption that all children would be either monolingual Guarani speakers or monolingual Spanish speakers. This monolingualism was assumed to belong not just to individual children but also to entire school communities, as program designs were assigned by school based on regional MEC supervisors’ perceptions of the dominant language in a school community (Profesora Noemí, Interview, 10/13/2008). Second, the program designs included very specific percentages of time that assumed the possibility of distinct separation of the languages. Third, the variety of Guarani originally selected and assumed by many to be necessary for instruction and materials was a purified register of Guarani, often called guaraní escolar or School Guarani. It differed from the everyday and hybrid varieties spoken by most people and its use conveyed an assumption that “language” means standard language.5

However, the policy might be seen as heteroglossic in other important respects. First, it was based on an assumption of biliteracy as an educational goal for all and not solely for speakers of the minority language. It was designed to maintain and develop both minoritized and majority languages with the goal of all students—Guarani dominant and Spanish dominant alike—becoming “coordinate bilinguals” (Paraguay MEC 2000). Second, MEC policy texts released after the first few years of implementation (Paraguay MEC 2000; Paraguay MEC 2004; Paraguay MEC 2006) articulated details that were significantly more heteroglossic than earlier materials and descriptions of program designs.6 These further policy articulations announced the development of a third modality of 50/50 design for students who came to school already bilingual to some extent. It acknowledged the possibility that schools might implement more than one program design and indicated that the educational community (not the MEC) should select which design(s) would be implemented in a given school. While the initial instructional materials developed between 1994 and 1997 for grades 1–6 used School Guarani, the MEC began recommending a different approach in 1997 for materials developed for grades 7–9 (Paraguay MEC 2000),7 one that “takes as a reference point the real language, the language of functional communication of Guarani speakers” and which, for example, would include loan words from Spanish. This approach contextualized the use of multiple, including mixed, varieties at higher sociolinguistic scales.

Acknowledging the existence of multiple varieties of Spanish and Guarani, this policy included “reflections on the concept of language” (18) itself and on language as inherently dynamic and subject to change through the situated actions of speakers across time and space. In an upscaling move, policy specified that the variety of Guarani that should be used in schools should be Paraguayan Guarani, “the language used by a majority of the Paraguayan population in their daily communication and capable of being adapted easily to current needs of oral and written language because it neither rejects its roots nor closes itself off from evolution and change” (Paraguay MEC 2000, 17). It called for flexible use of the languages based on their use in daily life, recommending the creation of terminology from Guarani morphemes as much as possible but also permitting the use of Spanish loan words in classroom materials under certain circumstances. This official upscaling of flexible language practices might have contributed to empowerment for speakers who used mixed varieties. For example, a teacher might frame her use of Jopara in instruction as evidence of adaptation to students’ and her own daily lives and in alignment with official national policy. Instead of the common framing of Jopara as “bad Guarani and bad Spanish,” she would be locating it at a higher sociolinguistic scale and potentially producing interactional power for herself and her students, in part by drawing on policy as a resource for this scaling. Actual appropriation in 2008, however, showed that very few of these heteroglossic perspectives had been taken up in the focal schools.

Monoglossic Appropriations in Rural and Urban Contexts

In both the urban and the rural focal schools, Guarani was indeed instructed as a language in approximately the required percentage of time. This, itself, constituted a counterhegemonic practice in that all students—majority and minority alike—were required to learn the minoritized language. But the ways in which Guarani was taken up as a medium of instruction differed significantly, both from policy text and across the two schools. These different mobilizations of Guarani also differed in the degree to which they were hetero- or monoglossic, and concomitantly in the sociolinguistic scales that they indexed. In turn, this produced uneven benefits for Guarani dominant and Spanish dominant children. Policy was taken up in such a way that, while the hegemony of Spanish in education was indeed disrupted, larger educational inequities remained intact.

Table 1 summarizes uses of Guarani in the schools and how these practices compared/contrasted with policy text and across the two schools. I describe how these practices were heteroglossic or monoglossic and, relatedly, the extent to which they scaled Guarani-dominant children's language practices up or down. These differences had important implications for students that I will discuss below. Descriptions of policy text are paraphrased from several documents (Paraguay 1992a, 1992b, 1992c; Paraguay MEC 2000, 2004, 2006).

Toward More Heteroglossic Policy Appropriations

In the analysis above, I have shown that, while policy text offered opportunity for heteroglossic interpretation and appropriation, it was appropriated primarily in monoglossic ways in the two schools in this study. This is to say that the counterhegemonic potential in the Paraguayan national policy for universal bilingual education was significantly attenuated. The national policy, while based on assumptions of bilingualism as the norm and including some heteroglossic frames, was appropriated into contexts where monoglossic frames dominated and language separation was ideologically normative (despite being empirically exceptional). As practiced in the two schools in this study, policy offered urban, middle-class, Spanish dominant students an opportunity for additive, enrichment bilingual education, while it offered the rural, more economically disadvantaged, Guarani dominant students a more transitional model (García 2009). In both cases, monoglossic ideologies and practices were used to locate students’, teachers’, and communities’ translanguaging practices at lower scales where they were more invisible and less powerful.

It is important to recognize that, unlike many other contexts and especially in the United States, bilingualism was taken for granted. In general, no one argued that Guarani should not be a part of education. Profesora Carla at urban ECI was unequivocal in her support for Guarani in school and embraced her identity as a speaker of Jopara and School Guarani in additional to Spanish. Profesora Elena at rural ESB expressed pain from how urbanites sometimes saw her when she spoke Guarani, and her feelings about how to use it in teaching were complex, but she never suggested it should not be part of instruction. Guarani language activists and others, including Profesora Teresa and Profesora Noemí who had formulated the original program designs, strongly defended the importance of Guarani and the rights and interests (as they saw them) of Guarani speakers, and for them, the development and promotion of purified School Guarani was a key part of the defense. Indeed, I would agree that the instruction and use of School Guarani has been a positive development. It was probably a contributor to the passage of the Ley de Lenguas (Languages Law) in 2010 (Paraguay 2010), which guarantees rights of Guarani speakers and communities, protects space for Guarani in mass media, and may lead to expanded roles for and status of literacy in Guarani. The instruction and promotion of School Guarani at the urban school was almost certainly an important contributor to students’ positive ideologies about Guarani and Jopara and to their self-identification as Guarani speakers (Mortimer 2016; Mortimer and Wortham 2015). However, without also being accompanied by positive valorization and use of more fluid translanguaging practices, the policy has aided majoritized children more than it has minoritized children.

If what we want are educational language policy and practices that increase educational equity, that always add to and never subtract from students’ communicative repertoires, and that locate minoritized students and communities and their languaging resources at the center of instructional practices, then we will have to find ways to move bilingual education policy beyond monoglossic frames for languages and speakers, beyond only language separation, and toward more flexible language practices. The assumption of bilingualism as the norm for all, as in the Paraguayan policy and in other instances of what Ofelia García (2009) terms recursive and dynamic types of bilingual education, is one important step. But the assumption of translanguaging practices is another step, however, that while receiving much attention and support in research is still rarely represented in policy. Scholars call for more attention to conceptualizations of language in our analyses of language policy (Collins 2011; Jaffe 2011; Makoni and Pennycook 2006; Ricento 2006) and to language hybridity, fluidity, and mobility in pedagogy (Blommaert 2010; Canagarajah 2013; Cummins 2008; García 2009; Lin 2013). A rapidly growing body of work documents productive translanguaging practices in classrooms (Creese and Blackledge 2010; Gort and Pontier 2012; Hornberger and Link 2012; Martínez et al. 2015; Mortimer 2014; Mortimer et al. 2014; Palmer et al. 2014; Sayer 2008, 2013).

Yet, such conceptualizations are not often present in or explicitly supported by policy, and even multilingual education policy reproduces colonial relations (Flores 2013; Mortimer et al. 2014). Even in a case like this one, where policy did include more fluid understandings of language, monoglossic ideologies are those used to interpret policy, and translanguaging practices occurred under the radar. Movement toward more heteroglossic policy as practiced in educational contexts will require both a more explicitly heteroglossic conceptualization in policy itself of the multiple varieties subsumed under language and the promotion and circulation of heteroglossic ideologies that can be used to interpret policy. Both will be required to further ensure that educational language policy serves goals of greater equity and inclusion for minoritized students.