INTERNATIONAL JOURNAL OF APPLIED LINGUISTICSS, QVUoAlR. I1N3G, TNHoE. C2I,R 2C0L0E3S 115599 Squaring the circles: issues in modeling English worldwide PAUL BRUTHIAUX Texas A&M University The model originally promoted by Braj Kachru and representing English worldwide as Inner, Outer, and Expanding circles has helped valorize denigrated varieties by drawing attention to commonalities across old and new varieties and by altering perceptions of their com- municative potential and relative prestige. However, the model suf- fers from being based in a political/historical view of English worldwide and thus fails to capture transplantations of the language in locations not formally recorded by colonial history. Because it promotesspecific varieties, the model also ignores variation within locales, especially where the gap between those who know English and those who do not is vast. Overall, the model encourages broad-brush descriptions of manifestations of English across all three circles that do not stand up to sociolinguistic analysis. In response, it is suggested that the model can continue to serve as a shorthand for English worldwide but that it must adapt by (1) moving away from a focus on nation-states in favor of a sociolinguistic focus on English-speaking communities wherever they are found and (2) recognizing that fundamental differences across contexts for English worldwide cannot be glossed over in support of specific varieties if we are to arrive at descriptively adequate sociolin- guistics and socially relevant language policies. Introduction For the best part of the last two decades, commentators on English worldwide have taken as their theoretical premise the model consisting of three concentric circles originally proposed by Braj Kachru (1984, 1985, 1989). In this model,the “Inner Circle” comprises locations where English is the language of asubstantial, often monolingual majority (e.g. USA, UK, Ireland, Australia, etc.). A major characteristic of varieties spoken in these locations is that they are largely ©© BBllaacckkwweellll PPuubblliisshhiinngg LLttdd.. 22000033, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA 160 PAUL BRUTHIAUX endonormative, that is, they find within themselves the norms of correctnessand appropriateness to be propagated through language education and language planning. By contrast, the “Outer Circle” represents locations that typically came under British or American colonial administration before acceding to independence and where English continues to be used for interethnic communica- tion and as a dominant language by those at the top of the socioeconomicladder. These English-speaking – or at least English-knowing – communities range insize and geopolitical importance from India to Nauru through Nigeria, Kenya, the Philippines, Singapore, Fiji, and many more. Post-colonial “New Englishes”tend to generate ambivalence among commentators. While some stress the role played by these varieties in perpetuating socioeconomic divisions between those who have English and those who do not (e.g. Canagarajah 1999; Ramanathan 1999), others emphasize the way in which they encapsulate aspirations to modernity through participation in worldwide trade, access to technology, and the tying together of new and typically multilingual nations (e.g. Kanyoro 1991). Despite increasing linguistic self-reliance and a gradual shift from exonormative to endonormative attitudes (Banjo 2000), these Englishes continue to be affected by conflict between linguistic norms and linguistic behavior, with widespread perceptions among users that Anglo-American norms are somehow superior and that their own variants are therefore deficient. The “Expanding Circle”, meanwhile, represents societies where English is not passed on to infantsnatural- istically across generations but is taught in schools to an increasing number of learners and is used – by some, at least – in activities involving members of other linguistic local communities and in international trade or tourism. Given that English – it can be safely assumed – is now taught to someone somewhere in every nation on earth, the Expanding Circle presumably comprises everynation not included in the Inner or Outer circles. Randomly selected names include Brazil, Italy, Thailand, Morocco, and many more. In these locations, English tends to be exonormative in that speakers, educators, and policy-makers have traditionally looked to American or British models for linguistic norms. Judging by the number of scholarly sources in which reference is made to the “Three Circles” of English, the model has clearly had a major impact. Introduced at a time when the duopoly of American and British English was unquestioned and metropolitan attitudes to postcolonial variants often ranged from amused condescension to racist stereotyping (for reviews, see de Beaugrande 1999; Canagarajah 1999; Bhatt 2002), the model broke new ground in raising awareness of the very existence of dynamic varieties of English with growing populations of speakers and increasingly vibrant media, literatures, andpopular cultures. Startling though it may have seemed to many at the time, the very act of pluralizing “English” and encouraging serious debate regarding the nature and role of “New Englishes” denoted both imagination and courage. Indeed, the enterprise was far from innocent. In Kachru’s own words (reported in Prendergast 1998: 229), this terminological choice constituted nothing short of “an insurgent linguistic weapon”. Though potentially a double-edged sword, the characterization of the enterprise as “liberation linguistics” (Kachru 1991), © Blackwell Publishing Ltd. 2003 SQUARING THE CIRCLES 161 a label that evolved in response to Quirk’s 1990 concern with over-politicization) was apt. Over time, the model has enabled a generation of analysts to explore changing sociolinguistic circumstances and to capture evolving relationships across varieties (for a historical survey of this process, see Leitner 1992) as well as between language and power. The model has played a significant role in valorizing non-metropolitan varieties of English and in encouraging perceptions of these varieties as the default code for increasing numbers of speakers, the object of serious scholarly investigation, and the uncontroversial vehicle for best-selling literature (Thumboo 2001; Talib 2002). One direct outcome of this perceptual shift has been the increasing number of recommendations that the teaching of English be made to reflect local identities and incorporate local as well as worldwide norms (among recent examples, see Pakir 1999; Eguiguren 2000; Kubota 2001; Bhatt 2001). Typical of this shift is the claim by Arua and Magocha (2000) that the variety of English now taking root among Botswana children meets all the criteria for a communicatively adequate code, is not perceived as inferior to other varieties, is comprehensible internationally, and is of a quality such that there is no reason why it should not be used for educational purposes. Yet, despite its evident merits and the contributions it has undoubtedly made to our appreciation of the modern sociolinguistic context of English world- wide, the Three Circles model is not without limitations. In this article, I review some of the major reasons why the model may no longer be appropriate for this evolving context. I argue that because it is descriptively and analytically incon- sistent as well as over-representative of a political agenda, the model has little explanatory power and makes only a minor contribution to making sense of the current configuration of English worldwide. In essence, the model suffers from the legacy of past successes. While the promotion of denigrated varieties was a just and timely objective, this concentration has left us with a primarily nation- based model which draws on specific historical events and which correlates poorly with current sociolinguistic data. Because it tries to account for varieties (in the Inner Circle), a multiplicity of speaker types (mainly in the Outer Circle), and geographical locations (in the Expanding Circle) all at once, this superficially appealing and convenient model conceals more than it reveals and runs the risk of being interpreted as license to dispense with analytical rigor. On balance, I suggest, the Three Circles model is a 20th century construct that has outlived its usefulness, and I will briefly sketch out at the end of this article a 21st century alternative that focuses instead on the specific sociolinguistic characteristics of English-speaking communities wherever they are found. Limitations of the model: the Inner Circle Curiously, perhaps, given the self-declared intention by promoters of the model to be subversive, the model reinforces perceptions of Inner Circle varieties of © Blackwell Publishing Ltd. 2003 162 PAUL BRUTHIAUX English as largely monolithic and standardized because it offers no account of dialectal variation within each of the varieties that it lists. This is especially true of African-American English, as the recent Ebonics controversy amply illustrates. As Lippi-Green (1997) and Collins (1999) show, the debate may have raged over whether or not Black English was to be regarded as a dialect or a language. However, the two sides were in agreement throughout in their view that Standard English and Black English do not simply constitute minorvariants of one variety. Though generally less controversial, the same debate has marked perceptions of what might constitute British English given that the variety con- tinues to be characterized by substantial dialectal divergence. As Millar puts it in Afendras et al. (1995: 299), “British English is not so much a cover term as a masking term: it hides major phonetic and phonologicalvariation and renders invisible very many speakers and several national identities”. By glossing over this variation, the model inadvertently contributes to perpetuating the notion that despite the variety traditionally referred to as Received Pronunciation (RP) being spoken by a tiny minority of British users, this single, supposedly homogenous and norm-giving variety should remain the preferred model for speakers who – for historical or personal reasons – tend to relate to British linguistic norms. In large measure this is due to the fact that discussions of standards – within the Three Circles model as elsewhere – rarely take into account the fundamental difference between spoken norms, spontan- eously shared by communities of speakers and hence not easily amenable to deliberate standardization, and written norms, which are relatively open to manipulation by institutional forces, especially through schooling. By ignoring the fact that written norms differ relatively little whereas spoken norms differ widely across Englishes old or new, the model misses an opportunity toencourage rigorous analysis of commonalities and differences across varieties of English. In short, by oversimplifying in this manner, the model offers an incomplete and potentially misleading representation of one of its major components. A second limitation of the model is that it fails to account for varieties that meet conventional Inner Circle criteria except for the fact that they are spoken by a minority, often alongside evolving varieties taking root among neighboring communities in the country. Typical of this context is South Africa, a nation where English fulfills a nationwide range of functions including that of lingua franca, dominant medium of education, and symbol of political change and modernization. In a multicultural and multilingual country of such complexity, at least three major varieties of English operate with distinct social andfunctional distributions. One of these, which could be described as “White South African English”, is a recent transplant, in many ways comparable to Australian orNew Zealand English in their respective settings. Another, widely identified as “Black South African English”, represents an emerging variety not yet securely estab- lished but possibly evolving toward a competing standard of increasing prestige and power (van der Walt and van Rooy 2002). A third, labeled by Mesthrie (1996) “South African Indian English”, has its origins in the efforts of amigrant population speaking a range of North and South Indian languages tocommunicate © Blackwell Publishing Ltd. 2003 SQUARING THE CIRCLES 163 with English and Zulu speakers without having significant competence in either language, at least initially. The result is a variety that is distinct from bothWhite and Black South African Englishes as well as from mainland Indian varieties because it was influenced by a specific set of social conditions, namelyprolonged interactions with native and non-native speakers and teachers of English (often other Indians) in South Africa. Overall, as de Kadt (2000) argues, South Africa fits neither the “Inner” nor the “Outer” components of the model. One response to this conundrum has been to focus on the relatively stable White variety of South African English and to list it alongside less problematic Inner Circle varieties, an approach followed by Graddol (1997: 10). The other is simply to omit all reference to South African English in relation to Inner Circle varieties, as does Yano (2001). Clearly, listing White and Black South African Englishes as part of the Inner and Outer circles respectively would be politically divisive. Yet, even if we accept the characterization of Black South African English as rapidly evolving and therefore not amenable to classifica- tion, there is a case for identifying a White South African variety of English on the same basis as, say, an Australian or even an American variety in that these are spoken natively by most descendants of European immigrants and the descendants of other, more recent immigrants who adopted those speech norms. On this basis, especially once comparable populations in neighboring countries such as Zimbabwe or Namibia are taken into account, the total number of speakers of what might be labeled “White Southern African English” isprobably greater (approximately 4–5 million) than the entire English-speaking populations of Ireland or New Zealand. On demographic grounds alone, this makes leaving this population and the variety of English it sustains out of the Inner Circle untenable. On theoretical grounds, there is a case for questioning the validity of a model that stresses the common nature of varieties of English descended from a colonial power that exported its language and saw it gain additional speakers in at least five locations (Ireland, USA, Canada, Australia, and New Zealand) while leaving out an obvious candidate on the grounds of political sensibilities or simply poor fit with other varieties within the nation-state in question. In effect, the very reliance of the model on nation-states as its principal conceptual base is called into question. Limitations of the model: the Outer Circle THE OUTER CIRCLE AND (NON-)NATIVENESS While it is legitimate for a model of English worldwide to encourage a sense of increasing ownership of English among its many users, it must also address the complexities arising from multilingual settings in which an increasing number of these users operate. These complexities include the nature and scope of (non-)nativeness, whether and how these can be determined, and whether they matter. In practice, the Three Circles model appears to have muddied thewaters by idealizing the very distinction between native and non-native speakers that it © Blackwell Publishing Ltd. 2003 164 PAUL BRUTHIAUX set out to counteract. At a relatively trivial level, this is apparent in some of the odder attempts at grouping countries on the basis of the native or non-native competence of their populations. Graddol (1997), for example, lists two nations with largely bi-ethnic populations born of the earlier practice of importing Indian laborers: Fiji, and Trinidad and Tobago. Yet, while Trinidad andTobago is listed alongside India and others as having both native and non-native populations of English speakers, Fiji is not, despite the fact that its Indian population (still numbering close to 50% of the total) displays a similar con- tinuum from English-dominant bilinguals to those with little or no knowledge of English, and every shade in between (Siegel 1992). At the root of this lack of typological systematicity is a sense of ambiguity encouraged by the model. Clearly, there is a crucial sociolinguistic difference between what languages users know and what they do with what they know. Put another way, the language needs of adults do not necessarily require that they possess native or native-like competence across the entire system. However, this purely instrumental perspective bypasses the psycholinguistic underpinnings of multilingual competence. In practice, it should be uncontroversial to note that locations of the Outer type are characterized by a vast spread between those who know and use English in preference to any other language in their repertoire and those who know or use no English whatsoever. Unlike most speakers in Inner Circle locations, many Outer Circle residents cannot be said to share a reasonably stable linguistic system broadly recognizable as English with the rest of the population because they do not communicate with each other extensively in that language and thus do not expose their children primarily to input based on that shared system. This fundamental distinction and its consequences for the nature and scopeof Outer varieties of English cannot be glossed over without encouragingmisleading perceptions of each sociolinguistic setting, often based on oxymoronicreferences to “non-native varieties” (Singh, in Prendergast 1998), as well as unrealistic expectations of the social and educational potential of English in Outer Circle locations. Just as South African Indian English evolved in response to interac- tions between Indian migrants and speakers of local languages as well as native and non-native teachers of English (Mesthrie 1996), mainland Indian English features substantial code-mixing between English and at least one locallanguage even in supposedly English-medium education, with lower socioeconomic groups institutionally barred from access to English-medium schooling (Ramanathan 1999). At the syntactic level, it also exhibits substantial and systematic internal variation between standard and vernacular usage (Bhatt 2000). A further difficulty is that by grouping together nation-states on the basis of their shared colonial history at the expense of detailed sociolinguistic analysis, the model fails to discriminate between strongly multiethnic entities and strongly monolingual ones. For example, in strongly multilingual Nigeria, Mauritius, or Singapore, English is widely used in a variety of official and unofficial roles not only for education, administration and – for a few, at least – international communication but also for internal communication across © Blackwell Publishing Ltd. 2003 SQUARING THE CIRCLES 165 ethnic groups. By contrast, despite having come under British administrative control at approximately the same time, relatively monolingual Bangladesh or Hong Kong have little need for English as a tool for internal communication and tend to limit it to administrative and educational functions. As a result, the model makes few useful predictions regarding proficiency levels and commun- icative practices in each location, a deficiency that affects language educators and language policy-makers working in these communities. THE OUTER CIRCLE AND POSTCOLONIALISM A second limitation of the Outer Circle concept is that it places at its core nation-states born of the vagaries of (de)colonization and poorly reflects local sociolinguistic arrangements. Borrowing from Crystal (1997), for example, Graddol (1997: 11) lists a total of 63 countries with non-native populations of English speakers, a set of nations and territories of hugely different size and geopolitical importance, ranging from 1-billion strong regional superpower India to microscopic Nauru. Among the tinier nations, the Seychelles are listed but the Maldives are not, perhaps reflecting the fact that the former was a crown colony and the latter a mere protectorate. While this type of omission may seem relatively trivial and could easily be remedied, a greater flaw in the Outer Circle concept is that it encourages a correlation between new varieties of English and locations where (mostly)British colonial power left behind readily identifiable creations in the form of inde- pendent nation-states. This results in the model missing countries that became part of a colonial empire only in disguise. One example among many is Egypt, where British occupation began in 1882 and was reinforced at the close ofWorld War I when the country officially became a British protectorate, a situation that lasted until 1952, longer in fact than the more transparent American colonization of the Philippines. As Schaub (2000) shows, the British presence in Egypt has had a major impact on educational practices to the point where something like “Egyptian English” is common currency among professional and service-oriented groups working in engineering, business, medicine, and the tourist industry. One major difference is that, because the country is consider- ably less multilingual than, say, India, English in Egypt does not normally function as an interethnic lingua franca. However, its use among a professional class that could in theory make do with Arabic makes the omission of a country of such demographic and cultural importance from representations of the Outer Circle model such as Crystal (1997) and Graddol (1997) somewhat puzzling. Similarly, nations such as Jordan, Iraq, or (then) Palestine, over which Britain had a brief mandate as they emerged from the disintegration of the Turkish empire after World War I, or those countries created on the edge of the Arabian peninsula in part to serve as administrative frameworks for oil production such as Kuwait or Abu Dhabi, came under British influence too late to be colored red on maps of the empire. Yet, despite not being listed as Outer Circle locations in sources such as Crystal (1997) or Graddol (1997), many of these nations are seeing English evolving into a language of widercommunication © Blackwell Publishing Ltd. 2003 166 PAUL BRUTHIAUX connecting a dominant local ethnic community to a vast multinational immigrant labor force, in addition to playing an increasing role in education at all levels. Another group of significantly English-speaking communities overlooked by the model includes portions of countries whose regional componentsexperienced diverging colonial influences but are now unitary nation-states. Cameroon, for example, is acknowledged by Graddol (1997) as having in excess of 6 million non-native speakers of English, mostly originating in the formerly German- controlled western portion of the country that came under British control at the end of World War I and was later joined to the larger French-controlled area to form post-independence Cameroon. Yet, no such recognition is accorded to Somaliland, the north-facing portion of Somalia that was a British protectorate from 1884 to 1960 – once again significantly longer than the more familiar American colonization of the Philippines – at which point it was joined to a greater Somalia. This ignores the fact that Somaliland has attempted since 1991 to have its de facto break with the formerly Italian-controlled southern portion of the country recognized internationally (with little success so far) and lists English along with Somali and Arabic as one of its three official languages. Another major country rarely discussed in relation to English worldwide and the Three Circles model is Ethiopia, a strongly multiethnic country never subjected to substantial interference by a dominant, English-speaking colonial power and yet where English has long played a role as a prestige language, as witnessed by Amharic-English code-switching patterns (Leyew 1998). Because it appears to offer a neutral tool for internal interethnic communication as well as – rightly or wrongly – a promise of modernization, English is currently being promoted throughout the country as an alternative to Amharic not only in a nationwide official role but also as the medium of education, a prospect of obvious interest to many non-Amharic-speaking Ethiopians (Ambatchew 1995; Bloor 1996; Hameso 1997; Boothe and Walker 1997). In this sense, despite substantially different patterns of European colonization, the current sociolin- guistic interplay of Amharic, other local languages, and English in Ethiopia recalls India, with its 50-year-old tussle between English and Hindi, the latter promoted by a dominant group as a national language of wider communication yet resisted by many non-Hindi speakers. In brief, a systematic focus on the largely accidental outcome of political history obscures major sociolinguistic similarities across former colonies and non-colonies alike and severely weakens the descriptive and explanatory scope of the model. Another case of English-speaking populations being treated differently by the model on primarily historical rather than sociolinguistic grounds relates to Central America. From Belize to Panama, the six Central American countries with access to a Caribbean coastline all have substantial populations of English- speaking descendents of relatively recent immigrants, mostly from Jamaica, originally contracted to work on plantations as well as on related projects such as railroad construction, and in Panama on the construction of the canal. Prior to these relatively recent developments, however, links among British traders in © Blackwell Publishing Ltd. 2003 SQUARING THE CIRCLES 167 Central American outposts and the key commercial hub of Jamaica were strong enough for repeated calls to be made for British colonial status to be granted to the entire Caribbean coast of Central America (Naylor 1989). Indeed, so strong was the link between these populations and de facto British colonial power that it was not until the 1948 revolution that the Jamaican residents of theCaribbean coast of Costa Rica were granted Costa Rican citizenship and thus freed from the need to turn to British consular authorities for protection andrepresentation (Harpelle 2001). Yet, only Belize (then British Honduras) succeeded in securing official British recognition as a crown colony, in part because relatively favorable geography made the threat of direct control being imposed from neighboring Guatemala more likely than in countries (especially along the Mosquito Coast of Honduras and Nicaragua) where British trading presence was confined toisolated estuary settlements largely inaccessible from the seat of Spanish-speaking power in remote highlands. Today, despite the inroads made by Spanish into these English-speaking communities in part as a result of the spread of primary education, a degree of cultural affinity continues to link the coastal people of these six nations across regional borders as well as with neighboring English- speaking Caribbean islands (Harpelle 2001). Yet little of this intricate sociolin- guistic pattern is captured by a model that maps varieties narrowly onto national boundaries. Once again, only a focus on accidents of political history, not sociolinguistic observation, explains why of all speakers of English in Central America, only those in Belize are mentioned in most discussions of regional varieties of the language and accorded Outer Circle status. As for nearby Panama, its absence from classifications such as Crystal (1997) or Graddol (1997) is doubly puzzling in that, in addition to being home to many English-speaking descendents of former plantation and canal laborers, this American-inspired creation and the construction and administration of thecanal that justified it led to the growth of a bilingual administrative cadre of Spanish speakers with its own localized variety of English as part of its communicative repertoire. This should have predicted that the country would be accounted for in much the same manner as Puerto Rico, the American Virgin Islands, or American Samoa, all of which receive recognition in Crystal’s and Graddol’s taxonomies. In brief, because it bases its coverage on political history (and inconsistently, at that) as opposed to sociolinguistics, the concept of an OuterCircle of new varieties of English is severely deficient in the explanatory andpredictive power needed to account for the complex ecology of English worldwide. Limitations of the model: the Expanding Circle The conceptual inconsistency that weakens the concepts of Inner and Outer Circles of English worldwide is also apparent in discussions of the Expanding Circle because it is not always clear whether the concept is meant to cover countries, country-based varieties, speakers, or non- (or barely-) speaking © Blackwell Publishing Ltd. 2003 168 PAUL BRUTHIAUX learners. One consequence of this lack of clarity is that “me-too” calls are heard periodically for additional varieties to be admitted to the ever-expanding family of new Englishes. One such recent call comes from Shim (1999), who argues that a long tradition of largely standardized English language teaching and testing in Korea has resulted in an increasingly endonormative localstandard shared by all members of the teaching profession. On closer examination, such periodic sightings of emerging varieties appear to have more in common with corn circles than with sociolinguistic ones. The question of what constitutes a variety of a language is a thorny one. The key issue is whether there exists in a particular location a core of speakers who not only know some English (e.g. Shim’s Korean teachers) but also use it to a reasonable level of proficiency for a substantial part of their daily activities, whether for internal communication in a multiethnic society, for international communication with other native or non-native speakers of the language, or for academic purposes in a location where English plays a significant role as a medium of instruction despite not having a substantial presence locally. This is fundamental because for a variety to emerge, local practices must surely gain norm value through recurring, spontaneous use across a range ofcommunicative functions as well as in emblematic domains such as the media, artistic creation, and popular culture. In other words, idiolects will converge as speakers accom- modate to each other and gradually evolve a set of norms that most implicitly recognize as a common bond. However, in countries where English is taught widely but not used internally, the conditions for the emergence of such norms are simply not in place. As a result, the kind of English spoken locally among a narrow professional circle – here, teachers – no more constitutes the basis for a variety of English than do restricted profession-based codes such as Airspeak, the worldwide medium of air traffic control. One example of a restricted speech form of this type is “Thai English”, which Smalley (1994) characterizes as difficult for foreigners to understand and inadequate for communicative needs beyond classroom practices. Although the increase in transnational communica- tion across Europe has led to a well-documented claim for variety status to be accorded to English used as a lingua franca (or “ELF”) by second language speakers among themselves (Jenkins 2000; Seidlhofer 2001), the domain of such language use remains restricted to specialized transactions (businessnegotiations, industrial cooperation, tourism, etc.) by a relatively small number of speakers, and broader variety-creating conditions remain largely absent. Admittedly, allowance could be made for nations such as the Netherlands or the Scandinavian countries where English is widely used in higher education. In addition, the relatively small populations of these countries and theirsubstantial involvement in international trade mean that at any given time a relatively high number of people will be involved in transnational communication in English, and this may provide part of the necessary social platform for norms todevelop. In this sense, there may be a marginal case for speaking of “Dutch English” or “Norwegian English”, though even this scenario is denied by Preisler with respect to Denmark (in Afendras et al. 1995). If this is the case as regards a © Blackwell Publishing Ltd. 2003