A Critical Review of Theoretical Perspectives : From Language Maintenance and Shift to Postmodern / Poststructuralist Bi / Multilingualism

This paper presents a critical review of three theoretical perspectives on language maintenance (LM) and language shift (LS) in minority language contexts. These three perspectives are (1) LS and subtractive bilingualism, (2) reversing LS and additive bilingualism, and (3) a critical perspective on bi/multilingualism. The review aims to demonstrate that much of the LM/LS literature as reflected in the first two perspectives (i.e., LS and subtractive bilingualism, and reversing LS and additive bilingualism) has been dominated by an essentialised view of language and its related concepts (i.e., identity and community) as whole, separate and autonomous entities within the bounds of nationstates. Such perspectives tend to reinforce a simplistic view of LM/LS as an all-or-nothing phenomenon and to advance its pessimistic outlook as language loss or language death. Hence, for a more fruitful framework, this paper presents a critical perspective on bi/ multilingualism that draws on postmodern and poststructuralist theories (Heller 2007a, 2012; Makoni and Pennycook 2007; Pennycook 2010). By seeking to investigate bi/ multilingual speakers’ local language ideologies and practices, this critical perspective enables not only a reconceptualisation of language, identity and community but also a more realistic and hopeful vision of bi/multilingualism in our pluralist, diverse, transnational and


Introduction
The topic of language maintenance (LM) and/or language shift (LS) has been an enduring concern for sociolinguists for decades.This linguistic phenomenon has typically been seen as resulting from a conflict between two ideological poles, the traditional and the modern, or in Saussure's (1983) terms, between "provincialism" and "intercourse".Edwards (2005) describes the former as "keeping communities faithful to their original habits" and the latter as "obliging people to move about and forcing them out of the shadow of the village belfry" (452).This Saussurean dichotomy reflects the tension in people's desires to retain something "smaller" and "traditional" against something "bigger", "predominant" and "impersonal".In the US, for instance, such a scenario is akin to a linguistic competition in which the dominant language (i.e., American English) always survives, while the minority languages are prone to marginalisation.
However, this zero-sum view of LM/LS that consists of winner and loser languages is naïve and overly simplistic.Rooted in modernist thought, LM/LS views language in totalistic terms and holds that language and its related concepts (i.e., identity and community) can only be maintained or shifted in their entirety.This explains why we have begun to witness this sociolinguistic topic moving away from the modernist ethos and towards postmodernist and poststructuralist theories.This movement is in response to the socio-political and socio-economic shifts that many nations are experiencing as a result of transnational migration and mobility, which, in turn, are driven by globalisation, the new economy and the rapid flow of information.
Along similar lines, this paper argues that LM/LS should no longer be seen as an allor-nothing phenomenon because this traditional view has been proven inadequate in its ability to capture a more complex and multifaceted sociolinguistic reality.This paper also shows how LM/LS research has progressed in new directions through a review of the literature published over the past three decades.

Critical Review of Theoretical Perspectives
This paper reviews three major theoretical perspectives that have been applied in much of the LM/LS-related research.It begins with a discussion of the first two perspectives rooted in the modernist, hence monolithic, ideology of monolingualism and bilingualism (Jaffe 2007).These perspectives include (1) LS and subtractive bilingualism and (2) reverse LS and additive bilingualism.The author discusses the socially contextualised significance of each perspective based on relevant empirical research and offers a critique drawing on prominent sociolinguists' insights on why they have been unsuccessful in capturing present-day multilingual realities.Finally, the author presents a current theoretical perspective characterised as a "critical perspective on bi/multilingualism" grounded in postmodern/ poststructuralist theories.This perspective emphasises examining bi/multilingual (i.e., minority language) speakers' language ideologies and practices (Heller 2007a;Makoni and Pennybook 2007), and it has been effectively used in minority language research over the past several years to depict the complexities of onthe-ground bi/multilingualism.The author concludes with a discussion of how the research findings obtained from this current perspective can elucidate minority speakers' LM/LS situations.
Before we proceed, it may be helpful to take note of the following.First, this paper by no means suggests the chronological succession of these theoretical perspectives; rather, they are presented in a logical succession without implying that the emergence of one perspective has necessarily caused the decline or demise of the other.Second, this paper bases its review on the language contact situations of language minority groups, according to Kachru (2001)'s three-concentric model, in inner-circle nations, including Australia, Canada, New Zealand, the USA and the UK, and some outer-circle countries, including Malaysia and Singapore.The former situations focus mainly on Asian immigrant communities, 1 i.e., East Asian, South Asian and Southeast Asian populations, whereas the latter study minority speakers of the Mandarin, Hakka and Teochew dialects, with Hakka and Teochew representing what Martin (2007, 495) calls "minorities within minorities". 2More notably, the Asian minority groups in both milieus are still largely underrepresented in applied linguistics research today.Third, this paper uses the juxtaposed term "bi/multilingualism" or "bi/multilingual" to encompass the coexistence and use of two or more languages by an individual speaker or a minority language community.However, when it appears as two separate terms (e.g., subtractive bilingualism, additive bilingualism), "bilingualism" is used to retain its original use, which is saturated with socio-political and socio-historical significance (Garcia and Flores 2012), whereas "multilingualism" embraces a new spirit of linguistic and cultural diversity (i.e., multilingual world, multilingual realities).

Theoretical perspective I: Language shift and subtractive bilingualism
This theoretical perspective is represented by the LS and subtractive bilingualism discourses that result from the linguistic and cultural pressures experienced by minority language speakers, especially immigrants living in a country where its dominant language has more political, economic and educational value.This is a common scenario in countries that have enforced an unum national policy (Lo Bianco 2001) favouring linguistic and cultural assimilation predicated on a monolingual ideology.In the US, for instance, the Americanization Movement that began in the early 20th century sought to mould its linguistic diversity into one nation and define "good Americans" based on the English proficiency that all immigrants from Eastern and Southern Europe were required to attain (Grillo 1998).This national policy was introduced in the early 19th century when the European nationalist movement originated (Anderson 1991) and where the bounded concept of "one-language-one-nation" was propagated to strengthen the idea of uniting one people speaking one language in one territory.In other words, the nationalist ideology serves to construct language as bounded and autonomous, and as Pennycook (2007) notes, entails a set of biased beliefs about what counts as a "language" versus "dialect" (e.g., English vs. other ethnic languages), what is considered a standard language, what their nature and characteristics are and the people who speak it.At the same time, identity was also constructed as unified and stable to promote authentic Americanness by using language (i.e., American English) as a key determining factor.This ideology, in brief, constructs any demand for a place for non-English languages and identities as a threat to national security (May 2006).
Most essentially, this one-language-one nation-one-identity ideology laid the groundwork for "subtractive bilingualism" whereby minority languages were suppressed or at least not supported in the same way as the dominant language.This led to asymmetry in bilingual competences and thus a "language shift" among immigrant children.Generally, this imbalanced bilingual proficiency was measured by the degree to which it deviates from the idealised native-speaker norm, based on which a bilingual speaker is often equated with cultural and linguistic deficits, with bilingualism constructed as an obstacle to the development of the dominant language competence (Weinreich 1974).This facet of bilingualism was documented in research on school language practices where bilingual children were cast as a "problem" and were prohibited from using their minority languages if they wanted to succeed in the host country.Among such studies were Boggs's (1985) report of a school's failure to recognise and accommodate part-Hawaiian children's home socialisation of verbal discourse patterns (e.g., giving orders, endemic quarrelling, story-telling, etc.) grounded in a strict age-graded hierarchy, authority and their relationship with family members as well as Crago's (1992) discovery of the incongruities between the discourse patterns with which young Inuit (Eskimo) children were socialised at home and those who were taught by non-Inuit teachers (e.g., silence vs. talk, observing vs. doing), leading these teachers to misconstrue Inuit children as having learning disabilities.

Critique on the first perspective
The first perspective has been criticised for downplaying the human agency of minority language speakers (Block 2008;Jaffe 2007;Heller 2007a;Pennycook 2010;Makoni and Pennycook 2007;Martin-Jones, Blackledge and Creese 2012).The research has shown that attempts to homogenise language, people and culture under the same flag are delusions in any era.As Kumaravadivelu (2008) argued, even when the US was forging a "cultural assimilation" that also included linguistic assimilation, to make this nation an ideal melting pot, the pot was simply unmeltable.As reported in Novak's (1972) book "The Rise of the Unmeltable Ethnics", even the descendants of immigrants from some European countries, such as Poles, Italians, Greeks and Slavs, continued to maintain their ethnic heritage, ethnic consciousness and cultural practices.Such maintenance may be even more robust among immigrants from non-European countries, such as Latin America and Asia, who come from more distinctive racial, linguistic and religious backgrounds (Kumaravadivelu 2008).In addition, as Jaffe (2007) notes, the LS perspective is highly "essentialised" because it gives rise to language as bounded and autonomous codes that tie individuals to a single identity and implies that there is a wholesale transformation of a community's linguistic practices while, in fact, the matter is far more mixed and complex, especially when we consider the social, political and historical contexts of language domination, contact and change.

Theoretical perspective II: Reversing language shift and additive bilingualism
The second perspective of LM and LS is characterised by a major discursive swing towards the minority language movement motivated by collective memories of the devalued bilingualism described above.This movement aims to "reverse language shift" (Fishman 1991), revive minority languages and protect them from a potentially irreversible loss, which typically occurs within three generations (Fishman 1991).It also supports dual language competencies, hence additive bilingualism, and posits that the acquisition of a mainstream language should not necessitate a loss of one's mother tongue.Consequently, a voluminous amount of research has emerged, employing the dominant discourses of "linguistic human rights" and "linguistic and cultural biodiversity".These discourses surfaced when the homogeneous ideology of the nation-state began to lose credibility, hence representing a shift away from a nationalist discourse of "one-languageone-nation" towards an international or supranational discourse (Muehlmann and Duchêne 2007) with a shared aim of thwarting the decline of "smaller" languages.
The linguistic human rights discourse recognises that everyone, regardless of ethno-linguistic and socioeconomic backgrounds, is entitled to bi/multilingual literacy and fluency because language rights are a prerequisite for human rights and a precondition for the promotion of global diversity (Skutnabb-Kangas 1999).In other words, this perspective acknowledges people's "natural rights" and "inalienable rights" and holds that everyone should live fully as human beings (Patrick 2007, 120).This discourse was widely promoted by several supranational agencies or NGOs, such as Linguapax, Terralingua and the Foundation for Endangered Languages (FEL) 3 (Muehlmann and Duchêne 2007).
The linguistic and cultural biodiversity discourse (Mühlhäusler 1996;Skutnabb-Kangas 2000) results from the heightened awareness among sociolinguists of language rights and was used to frame the discussion of bi/multilingualism in the 1990s.It consistently relies on the metaphor of language as a "species" while justifying that a language needs to interact with its environment and can "live," "survive" or "die".This "species" metaphor is often applied to endangered situations of smaller or minority languages as a result of environmental (i.e., cultural, economic and political) changes.More importantly, this discourse has prompted sociolinguists to adopt the metaphor of "loss" (Block 2008) while maintaining that the loss of language entails the loss of culture, which in turn is the loss of "another unique part of the mosaic that is humanity in all its diversity" (190).Effectively, this "loss" metaphor has shaped the LM/LS research in emotive and moralistic terms.For example, Wong-Fillmore (2000) documented how the loss of Cantonese led to a drastic decrease in household communication in a Chinese family in San Francisco, which in turn deteriorated their familial relations and deprived their children of an understanding of their ethnic values and cultural socialisation.Similarly, Downman (2006) has explored the impact of English language acquisition among Hmong families in Northern Queensland, Australia, how English undermined Hmong children's ethnic identity and how it destabilised the cultural traditions and family relationships by turning them against parental guidance and towards social life outside the Hmong community.

Critique on the second perspective
Despite the international stances advocated by supranational organisations, both discourses are still wedged in the essentialising logic about language and its associated concepts (i.e., identity and community), which many nation-states employ to marginalise minority languages.That is, they regard language as a whole and separate system associated with a single (ethnic) identity and as being located within a particular domain as a bounded territory.
With regard to the linguistic human rights discourse, when the minorities' language rights are advocated, such NGOs highlight the level of the threat to linguistic diversity by borrowing the nation-state's powerful discursive tactics of "enumeration" or counting the number of languages (Hill 1992).However, this tactic often backfires, resulting in the inclusion of only larger (minority) languages and excluding smaller languages such as dialects and other intra-language varieties.This discourse, therefore, paradoxically perpetuates the homogenising ideology of nationalism it always claimed to resist.According to Ricento (2005), this discourse leads to continued competition between majority and minority languages.In the case of the US, for example, while minorities reclaim their linguistic rights, the majority community also believes that English is their natural right, and since English carries higher instrumental (i.e., economic and patriotic) value, the non-English languages have to justify how they can maintain "divided loyalties" (Ricento 2005, 120).This linguistic tension explains in part why minority language speakers have not been successful in protecting their languages, which almost always end up being relegated to functionality in private domains (e.g., home, community, etc.) but rarely in public or institutional spaces (e.g., schools, public offices, etc.) Regarding the linguistic and cultural biodiversity discourse, by likening "language" to "species", it again recognises language as a discrete, bounded and well-defined object.It also overemphasises a gloomy notion of "loss" or "death" of a language species and refers to the LS towards dominant languages as a "crisis" (Krauss 1992).As stated earlier, it links the disappearance of linguistic varieties to that of endangered species and views the loss of biodiversity (i.e., the linguistic and cultural loss) caused by the encroaching modernisation into its ecological environment.However, according to Patrick (2007), language varieties and "endangered species" do not disappear in the same way because "although one language may fall out of use, new forms may be created that contain the discursive, lexical or pragmatic features of the old variety" (124).Therefore, a language variety, as seen from the perspective on-the-ground sociolinguistic realities, does not completely disappear or die but continues to develop and change.Patrick advises further that we should instead focus on actual language speakers as being located in certain political and economic contexts because it is the speakers of endangered languages that survive or die, not the languages.
Finally, the reverse LS perspective tends to romanticise the importance of traditional or smaller languages by situating the issues in the utopian past where the speakers are locked in a static notion of identity and place and by disregarding the fact that a shift towards more dominant languages is at times unavoidable due to the speakers' complex socio-political and socioeconomic situations (e.g., their geographical and social mobility) (May 2006).Moreover, by conceptualising language, culture, territory as essentialised, this perspective muddles the concept of "authenticity", especially in terms of who can legitimately represent "authentic" minority language speakers.This perspective consequently risks stereotyping and stigmatising minority language speakers themselves, especially those who are unable to conform to linguistic and cultural practices (e.g., speaking unmixed varieties, wearing non-traditional clothes, etc.) when, as is well known, such an idealised notion of authenticity is no longer easy to find in today's globalised world.

Coda
The review above shows that the two theoretical perspectives, "language shift and subtractive bilingualism" and "reversing language shift and additive bilingualism", share similar points of concern.First, both perspectives are linked to the essentialist notion of language, identity and community, leading to an understanding that the disappearance of a minority language causes a loss of identity and community membership.Second, these perspectives imply that language shift or loss should be deplored and revitalised in certain ways, a view that has deeply permeated the local ethnic and research communities for years.However, as Djité (2009) points out, such a view prevents us from considering that minority languages also need to use other languages for different purposes or adjust to the constantly changing and evolving patterns of language mixing in their linguistic repertoires, all of which represent a natural condition for bi/multilingualism.Next, from a methodological standpoint, much of the research often relies on the counting practices or data quantification of the number of speakers, especially in the forms of surveys.These numerical or statistical representations imply that researchers' view of languages that, again, are bounded, closed and geographically fixed entities (Moore, Pietikäinen and Blommaert 2010), which explain why the LM/LS issues have been conceptualised only from the perspective of either the maintenance or loss of a (whole) language.Relatedly, such findings often come from etic views of language experts (i.e., linguistic or sociolinguistic researchers), policy makers, and interviews with selected people in the focal ethnic communities (e.g., community leaders, parents, etc.) who tend to express a gloomy outlook about LS while virtually silencing the emic voices of minority language speakers and their lived experiences in their language contact situations.

Theoretical perspective III: Postmodern and post-structuralist bi/multilingualism
The last theoretical perspective for LM/LS is implicated in postmodern and poststructuralist theories (Gardner and Martin-Jones 2012;Heller 2007a;Jaffe 2007;Makoni and Pennycook 2007;Martin-Jones, Blackledge and Creese 2012).It has arisen out of sociolinguists' acute awareness of past research's continued lack of success in addressing language-related issues in the 20th century (Hornberger 2003).Unlike previous theoretical perspectives, the postmodern view of multilingualism holds that we will continue to face the same research dilemmas unless we focus on the deep roots of the problems, which requires examining these concepts (i.e., language, identity and community).

Postmodernism and post-structuralism
Postmodernism represents a reaction against modernisation, the era between the late 19th and early 20th centuries when the notion of truth predicated on "clarity, certitude, wholeness and continuity" was at its peak.Postmodernism refuses all these notions and espouses "ambiguity, relativity, fragmentation, particularity and discontinuity" (Crotty 2009, 185).It also rejects anything systematic in form, objective in thinking and scientific in reasoning but embraces play, irony and the absence of boundaries and even "mess" (ibid., 185).In this sense, postmodernism as a theoretical paradigm allows us to connect more intimately with language and its related issues in the real world of bi/multilingualism where we begin to witness the growing destabilisation of linguistic, identity and territorial boundaries.
Juxtaposed with postmodern bi/multilingualism is post-structuralism since both theories share similar intellectual orientations and are used virtually interchangeably in applied linguistics today.Post-structuralism represents a clear departure from the structuralist assumption that every language must have a "fundamental structure" or an "essence" that makes it a language (Sturrock 1993) in favour of the view that language and language practices are ambiguous and unstable and that one should question the straightforward relationship between the speakers (identity), the language they speak and the community (territory) to which they belong.In addition, post-structuralists see language as a vehicle through which differences between and within identity categories are socially and emergently constructed and consider the speaker's relation to language in terms of power relations implicated in discourses or other representational forms (e.g., ideology, norms, conventions, etc.).

Postmodern and post-structuralist perspectives on language and language practice
The postmodern/post-structural view of language has gained currency in socially oriented bi/multilingual research in the past several years.This perspective was particularly encouraged by the two pioneering monographs, Disinventing and Reconstituting Languages, edited by Sinfree Makoni and Alastair Pennycook (2007), and Bilingualism: A Social Approach, edited by Monica Heller (2007a).Their research seeks to problematise the concept of language in relation to other social constructs (e.g., identity, community, culture, etc.).Makoni and Pennycook (2007) suggest that in order to move away from a view of language as being linked to a single identity and geographical space, we need to understand how language was historically constructed.The authors recall the colonial period when many languages were "invented" or "called into being" (10) by European colonial administrators and Christian missionaries through the metalinguistic ideologies of naming, categorising and counting (e.g., language census, family trees, etc.).To elaborate, this invention process entails the categorisation of languages into major groups of families, and to make counting possible, these languages were named and catalogued anew, necessitating boundaries among languages, the blurring of boundaries between languages and smaller varieties (e.g., dialects) or even the exclusion of small languages from the lists.This "invention" process, which "all too often start[s] with the enumerative strategy", explains why most bi/multilingualism policies have not been successfully implemented in many countries because such policies tend to "miss the qualitative question of where diversity lies" (16).As a solution, the authors urged a new way of "disinventing" and "reconstituting" languages that no longer rest on the notion of uniformity and homogeneity (e.g., language as system or entity) but on the belief that language should exist in and of itself.
Likewise, Heller (2007a) would certainly agree that language and its related notions (i.e., identity and community) are social constructs, although she adopts a somewhat different approach to examining bi/multilingualism.While Makoni and Pennycook focus on the history of language construction, Heller moves the issue further to present contexts characterised by the intensity of language contact as a result of social and political changes in many nations inspired by the globalisation and transnational trends in various public spheres (e.g., politics, economics, technologies and migration).These new trends have created new sites for new globalised and localised lifestyles, ideologies and discourses that transcend and problematise state boundaries.Hence, while recognising the fluid mobility of social actors around the globe, they also rupture the direct link between language (national or cultural) and identity and muddle our idea about what counts as a minority language and who counts as an authentic (minority) speaker in a (minority) community.It is therefore no longer uncommon to experience such a destabilised connection of language, culture, individual identity and territory as illustrated by Heller (2007a, 343) below: A Chinese couple on a downtown Toronto thoroughfare stop a Chinese girl and ask for directions.But they ask in Chinese, and she doesn't speak the language of the ancient who once came to Canada from China.
In the late 1980s, I go to visit a Serbo-Croatian (sic) heritage language class in Zurich.Many of the students are Albanian speakers from Kosovo.In Toronto, third generation speakers of southern Italian dialects are sent to learn standard Italian as their "heritage language".So what is next?At this juncture, once we realise that linguistic boundaries are blurred, identity is not unitary but multiple and fluid, and community boundaries are deterritorialised, sociolinguists such as Makoni and Pennycook, and Heller propose two research directions to enable us to rethink these concepts in bi/multilingual research: language ideology and language practice.The first direction investigates people's beliefs and ideas about language(s) in connection with ideologies, society, culture and cognition.The other direction aims to discover the situated forms of talk or (bi/multilingual) practices themselves (Heller 2007a;Makoni and Pennycook 2007).

Local language ideologies
According to Kroskrity (2000), language ideology concerns understanding how people locally construct language through contextual sets of beliefs and attitudes.Pennycook (2007) further suggested that to deconstruct people's widespread misconceptions about language as homogeneous and stable, we need to understand "what the speakers believe about their language and/or other people's languages" and "how they analyze their talk" (22).For Heller (2007a), accessing such local knowledge enables us to discover the following: Who remains wedded to formerly dominant ideas about bounded systems, and who is trying to dislodge them?In favor of what?Why? … Whose interest it is to construct language(s) and their relationships in certain ways, and in whose interest it is to attribute what value to various ways of managing ideologically organized linguistic forms and practices -or rather of managing the variability that has always been at the heart of questions about bilingualism (341).
Research investigating local knowledge of bi/multilingual speakers is still sparse.This is especially true of studies (Blackledge and Creese 2010;Manosuthikit 2013) that focus on Asian immigrant communities and their heritage language experiences in English-speaking countries.Often, such research uncovers how two different generations, the older and younger generations, tend to hold discrepant perceptions about language and language practices, as presented in the following four studies.
A study by Blackledge and Creese (2010) reported the teachers' and parents' beliefs about Bengali and Sylheti (the standard and dialectal varieties spoken in Bangladesh, respectively) in community language schools in Birmingham, London.It was observed that both groups associated Bengali with the national "heritage" and as a prestige language while describing Sylheti as a low-status variety spoken by the poor and less educated.The researchers also noted that this local belief affected the classroom practices where only Bengali was emphasised.
Likewise, Manosuthikit (2013) studied how five sets of Burmese parents living in a US metropolis locally constructed the Burmese and English languages.The data revealed that the Burmese parents generally associated Burmese with their ethnic and cultural identity (i.e., obedience, politeness and respect) and the practice of separate bilingualism (i.e., unmixed Burmese).At the same time, they viewed English as their children's bilingual asset that offered a better educational and professional future while rejecting the American culture outright.
Shifting to the younger generation' local language ideologies, Preece (2006) examined how three British-born South Asian university students in London talked about language and discovered that their language choices was not tied to one identity or community but to various facets of their relationship with the people around them.One student associated her language choices (i.e., Urdu and Punjabi) with a desire to build solidarity with her Pakistani parents; another student positioned his parents' Pakistani culture as old-fashioned and saw himself struggling with the use of Urdu and Punjabi in family gatherings; the last student, an Asian man from Kenya, positioned himself as an English language deficit user due to his non-native accent despite his high English proficiency.Nevertheless, this last student felt confident as a native speaker of Gujarati, which he considered his "mother tongue".
In the community context, Manosuthikit (2013) investigated how a group of generation 1.5 teenagers in a US metropolitan city positioned English and their parents' mother tongue, Burmese.The researcher found that, unlike their parents who correlated a language with one's cultural identity, these teenagers viewed the languages far more pragmatically.They considered English as a language of egalitarianism and networks of friends and saw Burmese as a difficult language infused with social hierarchy.They, however, adopted the essential role of Burmese in their familial relationships, a sense of belonging to this community, and a deeper emotional effect it could have on them when verbally expressed.Documenting similar findings predicated on young speakers' linguistic pragmatics, Leng (2016) reported the language ideologies of 53 Teochew speakers in multilingual Singapore, most of whom were in their twenties.The researcher found that despite living under a powerful state policy that promoted the use of English and Mandarin, most continued to use Teochew, which they believed could serve not only as a communicative function, especially when conversing with their Teochew-speaking grandparents at home, but also its emotional function since the dialect also symbolised their core cultural identity.In essence, they held the realistic view that multilingualism served its own specific purposes.That is, while English helped them gain access to Western science and technology and Mandarin serves to engrain Confucian ethics and values, as supported by the government, Teochew, their mother tongue, was an indispensable language that helped them to remember their roots and more deeply understand "what they are" and "where they came from".
In another study, Wang (2016) explored the attitudes of 10 second-generation families in a Hakka Catholic community in Malaysia regarding language use for religious activities at home.Wang found that children tended to have different views from their Chinese-Malaysian parents in terms of the competency, future prospects and use of Hakka.They viewed that having an aural proficiency is sufficient for anyone to be considered a Hakka and that Mandarin, rather than Hakka and even Malay, was a more important language for practical purposes (e.g., education, travel).Although some held that the Hakka dialect is part of their culture and must be passed on to the next generation, they also thought that Mandarin, due to its greater popularity and practicality, can also serve as an identity marker that unified the Chinese population in Malaysia.In particular, Mandarin was seen as a better choice for them to participate in religious rituals at home because most in the younger generation used Mandarin in daily life.Again, the LS in this community was believed to take place due to the speakers' pragmatic considerations.
The studies above appear to confirm a divergence in the local perceptions about language based on the age of individuals (e.g., younger vs. older generation).That is, the older generations tend to view language in symbolic terms by mapping it onto identity, community and tradition, whereas the younger groups saw language more in instrumental or pragmatic terms by using it as a tool for partaking in community activities as well as establishing and negotiating personal relationships and group memberships.

Local language practices
The second research direction looks at how linguistic resources are locally deployed by bi/multilingual speakers.Extending the notion of language use, Heller (2007a) and Pennycook (2010) call it practice.According to Pennycook (2010), to think about language as practice is to make a "social activity" central to our analysis while considering language as a necessary "resource" for engaging in a social activity.Heller (2007a, 341) elaborates on this aspect of practice as follows: The observable ways in which people draw on linguistic resources in situ, and how this connects up to the circulation of resources over time and space as well as to the circulation of people through activities where resources and discourses are produced and distributed.How to capture what exactly it is that people are doing with languages?How to explain what it means?How does it fit in with other social practices?
The first study by Blackledge and Creese (2010) that was described earlier effectively illustrates Heller's point.It explored the teachers' and students' routinised language practices in Bengali heritage language classrooms 4 in London and Manchester and found that their practices primarily rested on two ideologies: separate and flexible bilingualism.That is, while the students fluidly used multiple linguistic resources, such as Bengali, Sylheti, English or a mix of these varieties (flexible bilingualism), most teachers insisted on enforcing the use of only Bengali in class by forbidding students from using English or mixed linguistic repertoires (i.e., separate bilingualism).
In another study, Curdt-Christiansen (2016) sought to understand how Singaporean family members negotiated a foreign language policy (FLP) in everyday face-toface interactions.Although the government strongly advocates separate or balanced bilingualism and views the use of mixed codes, such as Singlish, as equivalent to a poor command of English, the researcher discovered that Singaporean children (and some parents) in a HDB housing community centre negotiated this policy by resorting to translanguaging strategies or the meshing of languages and/or dialects, i.e., Mandarin, Teochew and English or Singlish.This vibrant linguistic practice, though not embraced by either the government or many parents in public and private domains, simply represents the family members' routine use of multiple codes to negotiate meanings and achieve their communicative aims.
The practice-oriented research also encompasses an identity dimension besides the linguistic one.Research (Canagarajah 2006;Manosuthikit and De Costa 2016) has shown that language practices also create space for bi/multilingual speakers' identities because language practices concern how speakers selectively activate their linguistic resources to have an identity or dynamic positioning in various ways.
For instance, Canagarajah (2006) ethnographically explored the linguistic life of Tamil youths in Lancaster, London and Toronto, and observed that they displayed multiple identities, i.e., Tamil and American, according to the activities in which they were participating (e.g., family's religious gathering, meeting with friends, etc.) and strategically shuttled between communities of practice through the use of various linguistic forms (e.g., English, Tamil, mixture of both) and discourses to serve different purposes, including signalling and negotiating memberships in different communities.
In another study, Manosuthikit and De Costa (2016) examined how bilingual teens in a US urban city used the Burmese address terms such as aunty (aunt), U (uncle) and ko/ako (older brother).The study revealed the participants' tactical recourse to bilingual choices of the address terms within and beyond the constraints of the Burmese and American norms.These teens engaged in the fluid negotiation of self-positioning and other-positioning motivated by factors such as ethnicity, solidarity, familiarity and in-group/out-group memberships.Their social practices were found to "reproduce" (i.e., comply with) ideologies in line with the Burmese or American systems (i.e., use and non-use of kin titles) and "reconstruct" (i.e., transform) the practices that are markedly deviant from the structural cores (i.e., use of mixed or hybrid address terms and avoidance of all addressing choices) of both cultures.
In short, the studies above demonstrate the tendency of young social actors' local practices to disrupt the smooth correspondence of the one-language-one-identityone-nation (community) ideology.Their pragmatics-based practices involve the deployment of bi/multilingual resources with or without clear boundaries (e.g., mixed codes) to fulfil the goals of constructing and negotiating multiple identities and shuttling in and out of different group memberships instead of belonging to one particular community.These studies also demonstrate that language, when viewed as a resource, neither has clear boundaries nor a priori functions because it is the speakers and social meanings they accommodate, not distinct codes or language in use, that lie at the core of interactions.
Notably, the amount of research on local bi/multilingual practices has been growing over the past several years.This research trend, like the studies above, signals a shift from seeing language in terms of structure, system and categorical boundaries towards the view that focuses on bi/multilingual speakers and their agentive use of whatever linguistic resources are available to them to achieve their communicative aims.Such everyday practices have emerged under different labels, namely code-meshing (Canagarajah 2011)

Discussion
The critical review above inevitably returns to our crucial question: How does this postmodern/post-structuralist perspective on bi/multilingualism illuminate our understanding of the language maintenance and shift phenomenon?As discussed earlier, the older generations (e.g., parents and teachers) influenced by public language policies tended to embrace balanced bilingualism as two distinct, separate and bounded languages, especially by expecting younger generations to speak unmixed languages and to adopt neatly separated cultural identities.Such a stance is reminiscent of the language revitalisation movement aimed at protecting minority languages from a potentially irreversible shift (Fishman 1991) and indicates a fear and anxiety of the potential loss of language, culture and identity (Blackledge and Creese 2010).At the same time, this stance also suggests an expression of resistance to the powerful English ideology, as Blackledge and Creese (2010, 122)

elaborate below:
Separate bilingualism offers an opportunity to counter the structures and systems of mainstream discourses by insisting that a particular ideology of bilingualism and a standard version of the [heritage] language continue to be valid in the face of the powerful English monolingual structures.
Nevertheless, as we have seen, the younger generations (Manosuthikit and De Costa 2016) tend to have more situated and fluid views about bilingualism based mainly on linguistic pragmatism.For them, language is less of a concern for preservation but more as a tool for participating in the social world, signalling particular types of identities and belonging to more than one group.Such stances about language and identity among the youth are highly evident in their strategic, conscious and purposeful employment of semiotic resources to fulfil communicative and ideological functions in various social contexts.Through such resource use, the younger generation is also able to expand their identity repertoires, hence refusing to be boxed into one subject position.
Oftentimes, the younger generation is branded as culprits for language shift or loss.However, this paper argues that this depends on the perspective we take.If we see language as separate, bounded and countable entities while romanticising the notion of "maintenance" as locked in time and space, as echoed in the parents' and teachers' views, we then are led to believe that linguistic changes and transformation is but a gloomy sign of LS.By contrast, if we accept social and cultural changes as part of the natural process, then we acknowledge that maintenance should also involve constant changes through the acquisition of newness and through difference (Pennycook 2010, 105).Such a position allows language not only to adjust for survival but also to paradoxically "remain the same" (ibid.105).These changes "from below" recognise the capacity of human agency to rise above the social structures and to creatively construct a new social (bilingual) reality.From this perspective, we can argue that we have not lost the language or its accompanying culture; rather, we gain them both.These gains are not just the new linguistic and identity resources but also create a new capacity to negotiate multiple affiliations and accommodate linguistic and cultural differences.These differences are part and parcel of bi/multilingual speakers' everyday sociocultural engagements as well as requisite for surviving and thriving in this postmodern multilingual world.
Finally, the myriad findings on local language ideologies and the practice discussed above can be best encapsulated in Pauwels' (2016) recent remarks about language maintenance: The term 'language maintenance' evokes both a sense of 'stability' and a level of abstraction that may seem increasingly at odds with the linguistic realities characterizing many communities and societies around the world.The new realities are best described as highly dynamic, with constantly and rapidly changing language constellations.
To put it simply, language maintenance 5 can no longer sufficiently depict societies where multilingualism is becoming a normal form of communication.These multilingual realities have become more apparent not only in Western urban settings but also in parts of Asia (and Africa), where there is a growing presence of plurilingual speakers who concurrently use not just two languages but multiple linguistic varieties and codes.Such a linguistic phenomenon can be particularly observed in the more dynamic contexts of transnational mobility and migration that result in different family members living in different countries.This dynamic element of the phenomenon may present additional challenges to the study of LS/LM currently (Pauwels 2016), but it certainly holds promise as a new research direction that richly deserves further investigation.

Conclusion
To conclude, the investigation of "bottom-up" language ideologies and practices has, in fact, been increasingly taken up by language policy and planning researchers who aim to scrutinise the scholarship at the micro level outside of the educational context (Kirkpatrick and Bui 2016).It marks a move away from the traditional research model that centres on the national, official and top-down approach.This research direction, as shown above, draws on socially inspired theories and ethnographic approaches while emphasising the local agency of language minority speakers.It also offers space for community members to voice and negotiate community needs via emic views and language-identity practices.All of these goals have been accomplished to advance our understanding of grassroots linguistic realities, to generate equity in bilingual practices, to defy unequal official language policies and to bring about sociocultural transformation (McCarty 2011).Notes 1. Except for two studies focusing the indigenous populations in Hawaii, the US, and Quebec, Canada (see subhead "Theoretical perspective I: Language shift and subtractive bilingualism").2. "Minorities within minorities" (Martin 2007, 495) generally have to encounter another layer of power relations, which makes their linguistic and cultural experiences even more multifaceted.3. Linguapax (founded in 1987), Terralingua (with its wider networks than Linguapax) and FEL (based in the UK) aim to protect, maintain, restore and promote the diversity of languages and cultures, including endangered languages in all contexts.4. Bengali is one of the four heritage languages in eight complementary schools in England that Blackledge and Creese (2010) investigated.5. Like Pauwels (2016), Jaffe (2007) also referred to the term "language shift" as a monolithic notion that can no longer be used to understand the LM/LS situations in today's multilingual contexts.