Going beyond the native speaker in language teaching

Going beyond the Native Speaker

in Language Teaching
TESOL Quarterly, 33, 2, 185-209, 1999

It is often taken for granted that the only rightful speakers of a language are its native speakers. Linguists look at the intuitions of native speakers or collect quantities of their speech; language teachers encourage students to be like them. This paper argues that the prominence of the native speaker in language teaching has obscured the distinctive nature of the successful second language (L2) user. It puts forward some suggestions for how language teaching can recognise students as L2 users both in the classroom and in the world outside.

First the implications of the term native speaker need to be spelled out. The keynote is struck in what Davies (1991) claims to be its first recorded use: 'The first language a human being learns to speak is his native language; he is a native speaker of this language' (Bloomfield, 1933, p.43). In other words you are a native speaker of the first language (L1) that you learnt in childhood, called by Davies (1996, p. 156) the 'bio-developmental definition'. Being a native speaker in this sense is an unalterable historic fact; you cannot change your native language any more than you can change who brought you up. This definition is echoed in modern sources such as the Oxford Companion to the English Language (McArthur, 1992) and the corpus-based Collins COBUILD English Dictionary (1995).

This core meaning of native speaker is often supplemented by detailing the characteristics that native speakers share apart from their birth. Stern (1983) lists: (i) subconscious knowledge of rules, (ii) intuitive grasp of meanings, (iii) ability to communicate within social settings, (iv) range of language skills, and (v) creativity of language use. The Encyclopedic Dictionary of Applied Linguistics (Johnson & Johnson, 1998) adds (vi) identification with a language community. Davies (1996) adds (vii) ability to produce fluent discourse, (viii) knowledge of differences between their own speech and that of the 'standard' form of the language, and (ix) ability 'to interpret and translate into the L1 of which she or he is a native speaker'.

Some of these characteristics are in a sense obvious: native speakers are not necessarily aware of their knowledge in a formal sense (i and ii), but nor could they explain how they ride a bicycle. Some are debatable: many native speakers are unaware how their speech differs from the status form (viii), shown for example in the growing use of non-standard between you and I for between you and me even in professional speakers such as news-readers. Many native speakers are far from fluent in speech (vii), some having to communicate via alternative means, such as Stephen Hawking and Helen Keller. Some native speakers function poorly in social settings (iii). In the Chomskyan sense of creativity, any novel sentence uttered or comprehended is creative (v); a computer can create 'new' sentences, for instance the speech program that answers telephone directory enquiries with every possible telephone number. In a general literary sense, creativity belongs to a small percentage of native speakers, such as poets, rap singers and so on. The ability to interpret (ix) is only possessed by native speakers with a second language and not necessarily by all of them. Native speakers are free to disassociate themselves completely from their L1 community politically or socially (vi) without giving up their native speaker status, whether Karl Marx in London, James Joyce in Zurich or Albert Einstein in Princeton.

These characteristics are then not only variable but also in a sense accidental; lack of any of them would not disqualify a person from being a native speaker. A monk sworn to silence is still a native speaker. Many are also shared by non-native speakers almost regardless of their level of proficiency in the language: non-native speakers show a rapidly developing awareness of gender-linked pronunciation (Adamson & Regan, 1991) and of the status of regional accents (Dailey-O'Cain, 1998); what level of L2 English did it take for Marcel Duchamps to create 'surrealistic aphorisms' such as My niece is cold because my knees are cold (Sanquillet & Peterson, 1978, p.111)?

The indisputable element in the definition of native speaker is that a person is a native speaker of the language they learnt first; the other characteristics are incidental, describing how well they use the language. If you did not learn a language in childhood, you do not speak it as a native speaker. Later-learnt languages can never be native languages, by definition. Children who learn two languages simultaneously from birth have two first languages (Davies, 1991); we see later that this may not be the same as being monolingual native speakers of either language. L2 students cannot be turned into native speakers without altering the core meaning of native speaker in English. A view such as 'adults usually fail to become native speakers' (Felix, 1987, p.140) is like saying that ducks fail to become swans: adults could never become native speakers without being reborn. At best L2 learning produces an L2 user who is like a native speaker in possessing some of the nine aspects of proficiency detailed above to a high degree but who cannot meet the 'bio-developmental definition'. A central aspect of being a native speaker so far as this paper is concerned is that it is someone who speaks their first-acquired language. The variable aspects of 'proficiency' (Davies, 1996) or 'expertise' (Rampton, 1950) are a separate issue of quality rather than defining characteristics (Ballmer, 1981).

An additional assumption is often that the native speaker speaks only one language. Illich & Sanders (1988, p.52) point out that: 'From Saussure to Chomsky "homo monolinguis" is posited as the man who uses language—the man who speaks.' Ballmer (1981) and Paikeday (1985) include monolingualism in their extended definitions of native speaker. In Chomskyan linguistics monolingualism is part of the abstraction involved in obtaining the idealized native speaker, 'We exclude, for example, a speech community of uniform speakers, each of whom speaks a mixture of Russian and French (say, an idealised version of the nineteenth-century Russian aristocracy)' (Chomsky, 1986, p.17). Important as it is for other purposes to consider the different types of native speaker and the different abilities that native speakers possess, the distinction here is between people who speak the language they grew up with compared to those who speak another language as well—between monolingual native speakers and L2 users. The meaning of native speaker here will then be a monolingual person who still speaks the language they learnt in childhood.

The discussion of native speaker introduced the term L2 user for someone who is using an L2, contrasted with L2 learner for someone still in the process of learning it. It is doubtless debatable when an L2 learner becomes a L2 user because of the difficulty in defining what the final state of L2 learning is; moreover some learners are already users whenever they step outside the classroom. While there are fuzzy aspects to this distinction, its rationale will emerge during the argument.

In recent years growing concern has been expressed over the role of the native speaker in language teaching and SLA research. Some have seen the issue in quasi-political terms as the exercise of power and status (Holliday, 1994); the native speaker concept has political and economic benefits for the countries from which particular languages originated (Phillipson, 1992). Others see it in cultural terms as the imposition of native speaker interaction norms contrary to the students' own preferred types of interaction (Kramsch & Sullivan, 1996). Others point out that 'one man in his time plays many parts': English-speaking people show they are men by using /' n/ in waiting (Trudgill, 1974); that they are American by having /r/ in corn; that they are British working class by 'dropping' the 'h' in hair (Milroy, 1983). Native speakers form only one of the social groups to which a speaker belongs (Rampton, 1950); the role of native speaker is no more basic than any other (Firth & Wagner, 1997). In practice, despite these objections, the native speaker model remains firmly entrenched in SLA research and language teaching.

Overt discussion of the native speaker is rare in language teaching. Indirect evidence for the importance of the native speaker in ELT is indeed the perennial issue of which kind of native speaker should be the model for language teaching (Quirk, 1950), which mostly assumes that the choice lies between different types or aspects of native speakers, not whether to use them at all. Stern (1983, p.341) puts it bluntly: 'The native speaker's 'competence' or 'proficiency' or 'knowledge of the language' is a necessary point of reference for the second language proficiency concept used in language teaching'. The Practice of English Language Teaching (Harmer, 1991) describes different areas of language in a chapter entitled 'What a native speaker knows' and goes on to say 'students need to get an idea of how the new language is used by native speakers' (Harmer, 1991, p.57), though the usage shifts into the combined expression 'native speakers (or competent users of the language)'. Kramsch (1998, p.28) sums it up pithily: 'Traditional methodologies based on the native speaker usually define language learners in terms of what they are not, or at least not yet'. Or, one might add, probably not ever.

One source of insight into language teaching is the coursebook, which provides a structure for many classes (Hutchinson & Hutchinson, 1994). The description of English underlying coursebooks seems implicitly native-based, even if it often reflects the idealised normative view of English of the teaching tradition rather than actual description. The Collins COBUILD English Course 1 (Willis & Willis, 1988), however, explicitly 'focuses on the real English students will encounter and need to use in today's world', based on a large database of native speaker usage. The model situations met in coursebooks are almost invariably native-to-native, apart from the typical opening lessons in which students introduce each other and exchange personal information, for example Units 1 in Headstart (Beavan, 1995) and True to Life (Collie & Slater, 1995).

1960s SLA research borrowed from L1 acquisition research the assumption that learners have language systems with distinctive features of their own (Corder, 1967; Cook, 1969). This assumption formed one aspect of the well-known interlanguage hypothesis (Selinker, 1972), implicit in the continuing aim of the SLA research field to describe and explain the L2 language system in its own right. In other words SLA research aims in principle to detach L2 learning from the native speaker.

In practice SLA research has often fallen into the 'comparative fallacy' (Bley-Vroman, 1983) of relating the L2 learner to the native speaker. This is reflected in the frequency with which the words succeed and fail are associated in a non-technical sense with the phrase native speaker, for example the view that fossilisation and errors in L2 users' speech add up 'to failure to achieve native-speaker competence, since in Chomsky's words, native speakers (NSs) are people who know their language perfectly' (James, 1998, p.2). Many SLA research methods involve comparison with the native speaker (Cook, 1997a; Firth & Wagner, 1997), whether grammaticality judgements, obligatory occurrences or Error Analysis.

A standard communication strategy is to describe an unknown object in terms of one that is already known (Poulisse, 1996); if you have never seen a tomato before, it can be described as a rather soft apple with a large number of pips. But this is no more than a temporary expedient until you have understood the unique properties of the object itself. SLA research can justifiably use native speakers as an entry-point to the language of L2 learners, provided it does not make them the measure of final L2 achievement. Klein and Perdue (1992, p.333) warn in particular of the danger of the 'closeness fallacy' of being deceived by learner utterances that have a false resemblance to those of the native speaker. The avowed aim of their large multi-language project was indeed to discover 'why ... adults attain the state they do' (Klein and Perdue, 1992, p 334). So, while much SLA research treats the L2 user as independent, the native speaker often maintains a ghost-like presence.

Interlanguage refers to the knowledge of the second language in the speaker's mind. But this L2 interlanguage exists in the same mind as the L1. Since no word describes the knowledge of both L1 and L2, the term multi-competence was coined to refer to the compound state of a mind with two languages (Cook, 1991). Multi-competence covers the total state of language knowledge of a person who knows more than one language, including both the L1 competence and the L2 interlanguage.

Competence is a neutral term in linguistics for the native speaker's knowledge of language; it does not involve a value judgement about whether such competence is good or bad compared to some outside criterion. In a sense whatever the L1 native speaker does is right—subject of course to the vagaries of performance and the like. Multi-competence is intended to be a similarly neutral term for the knowledge of more than one language, free from evaluation against an outside standard. The difficulty is that, while all the speakers of a first language arguably have similar competences, L2 users notoriously end up with widely differing knowledge. Nevertheless, so far as any individual is concerned, there is a final state of L2 competence just as there is for an individual L1 learner, difficult as this may be to generalise across many L2 learners.

The term multi-competence itself does not commit the user to anything more than the need to refer to a person's total language knowledge. Multi-competence implies that at some level the sum of the language knowledge in the mind is relevant, not just the portions dedicated to the L1 or the L2. Language teaching is concerned with developing an L2 in a mind that already contains an L1; as Stern (1992, p.282) puts it 'whether we like it or not, the new language is learnt on the basis of a previous language'. This section looks at the nature of multi-competent minds that know two languages, contrasting this with the competence of the monolingual native speaker.

One component of multi-competence is the knowledge of the L2. Nobody is surprised that the second language of L2 users differs from that of L1 users. Very few L2 users could be mistaken for native speakers. Most L2 learners resign themselves to 'failing' to reach the native speaker target. Some research has looked at 'ultimate attainment' in L2 learning, starting by showing even fluent bilinguals can be told apart from monolinguals in grammaticality judgments (Coppieters, 1987; Davies, 1991), going on to demonstrate that some L2 users are nevertheless indistinguishable from native speakers in syntax (Birdsong, 1992) and phonology (Bongaerts, Planken & Schils, 1995). As put by White and Genesee (1996, 258), 'Ultimate attainment in an L2 can indeed be native-like in the UG domain'. But the comparison with the native speaker again creeps in; valid ultimate attainment seems to be phrased with reference to the native speaker's competence rather than in its own terms. The ultimate attainment of L2 learning should be defined in term of knowledge of the L2. There is no reason why the L2 component of multi-competence should be identical to the monolingual's L1 if for no other reason than that multi-competence is intrinsically more complex than monolingualism.

Whether or not one accepts that some L2 users can pass for native speakers, it is evident that these passers form an extremely small percentage of L2 users. Research with this group tells us about the achievements of a few unusual people such as those as typical of human beings as Olympic high jumpers or opera singers.

The other major component of multi-competence is the first language. An early definition held that transfer went in two directions, producing 'instances of deviation from the norms of either language which occur in the speech of bilinguals as a result of their familiarity with more than one language' (Weinreich, 1953, p.1). While the effects of the L1 on the L2 interlanguage are easy to see, the effects of the L2 on the L1 have been little discussed. Yet everyone who has been exposed to an L2 can tell anecdotes about its effects on their L1: for example my own speech has sentences such as What do you want for a book? and vocabulary such as pulli for pullover, probably showing the use of L2 Swiss-German as a child.

A body of research shows that this effect of the L2 on the L1 exists in most aspects of language. In terms of phonology, the timing of voicing at the beginning of plosive consonants (Voice Onset Time) in the L1 moves slightly towards that found in the L2, French speakers of English having a slightly longer VOT for /t/ in their first language, French, than monolinguals (Flege, 1987). In vocabulary, the meanings of L2 words affect the meanings of their twins in the L1: for example the meaning of the English word coin (piece of money) affects the understanding of the French word coin (corner) in French people who know English (Beauvillain & Grainger, 1987); loan-words have slightly different meaning in the L1 for people who know the L2 from which they are derived, for example Japanese bosu (gang-leader) is perceived as less related to crime by Japanese who know English boss (Tokumaru, in progress). In syntax too, L1 grammaticality judgments are affected by the L2: English speakers who know French judge English sentences with null subjects such as "Is raining" differently from monolinguals (, 1996); Francophones and Anglophones learning the respective L2s have different reactions to middle verb constructions in their L1 from monolinguals (Balcom, 1998). Several experiments show L2 users become slightly slower at processing the L1 as they gain proficiency in an L2 (Magiste, 1986). In reading also, Greeks who know English read Greek differently from monolinguals to some extent, for example being more affected by the order of presentation (Chitiri & Willows, 1997).

In short multicompetent L2 users do not have the same knowledge of the L1 as monolinguals, extending for some to the partial loss of their L1 (Seliger & Vago, 1991).

Though hard to disentangle from the last two points, L2 users have the L1 constantly available to them while processing the L2. For example L2 users compensate for gaps in their vocabulary with the same communication strategies as in their L1 (Poulisse, 1996). L2 users are faster and more accurate in an language-switching task than in a monolingual condition on an auditory version of the STROOP test which asks people to decide whether voices saying the words 'high' and 'low' in two languages are actually either high or low (Hamers & Lambert, 1972). L2 Spanish/English users understand sentences that are translations of Spanish idioms more quickly than monolinguals (Blair & Harris, 1981). L2 users tend to switch from one language to another for their own private purposes; 61% prefer the L1 rather than the L2 for working out sums, 60% for praying, while 61% use the L2 for keeping their diary, 44% for remembering phone numbers (Cook, 1998).

A distinctive ability of L2 users is code-switching. When L2 users are talking to other people who know both languages, they may alternate between languages: Suami saya dulu slim and trim tapi sekarang plump like drum (Before my husband was slim and trim but now he is plump like a drum) was said by a Bahasa Malaysia teacher of English to fellow teachers in the staffroom. They can not only use either language separately but also both languages at the same time—what Grosjean (1989) calls the monolingual and bilingual modes. Code-switching has complex rules, partly at a pragmatic level of speaker and listener roles, partly at a discourse level for topic, partly at a syntactic level (see the range of papers in Milroy & Muyskens (1995)). Code-switching is the most obvious achievement of the L2 user that cannot be duplicated by monolingual native speakers, as they have no language to switch into. It shows intricate links between the two language systems in multi-competence: the L1 in the mind is not insulated from the L2.

Multicompetent speakers and monolingual native speakers also differ in certain thought processes. It may not be surprising that people who know two languages are slightly less effective at language-related cognitive tasks in the L2 than monolinguals (Cook, 1997b). Long term memory for information gathered in lectures is less efficient in an L2 (Long and Harding-Esch, 1977); working memory span in the L2 is usually slightly below the L1 level at all levels of L2 performance (Brown and Hulme, 1992; Service and Craik, 1993). But this also works in reverse: L2 users are slightly below the L1 performance of peer monolinguals at naming objects and following instructions to mark letters in words (Magiste, 1986); 'The very fact of having available more than one response to the same stimulus may lead to slower reaction times unless the two response systems are hermetically isolated from each other' (Magiste, 1986, p.118). In other words the minds of L2 users differ from monolinguals in several respects other than sheer knowledge of language.

Indeed this is one reason why, in many educational systems, second languages are taught in the first place. Learning a foreign language is seen as leading to 'an interest in language and culture' in Japan (Tokyo, 1950), to the ability 'to recognize cultural attitudes as expressed in language and learn the use of social conventions' in the UK (HMSO, 1995), and to 'courage, honesty, charity and unity' in the first form in Malaysia (Kementarian Pendidikan Malaysia, 1987). A particular benefit has often been claimed to be 'brain-training' – learning other mental skills. SLA research has indeed shown some truth to these claims, particularly the bilingual's keener awareness of language itself. Bilingual children are aware of grammatical properties of their L1 sooner than monolinguals (Galambos & Goldin-Meadow, 1950) and are better at judging how many words there are in a sentence. In particular bilingual children are more capable of separating meaning from form (Ben Zeev, 1977; Bialystok, 1986). Most remarkably, English-speaking children who learnt Italian for an hour a week in the first class of primary school showed advantages in learning to read over 'monolingual' children (Yelland, Pollard & Mercuri, 1993). Diaz (1985) lists many advantages for bilinguals such as 'measures of conceptual development', 'creativity', and 'analogical reasoning'.

Clearly multicompetent people differ from monolinguals in many ways. L2 users are different kinds of people, not just monolingual native speakers who happen to know another language. The native speaker goal of language teaching could be achieved only if the students remained unchanged by their new languages. But this is not the case, whether for better or for worse.

Most L2 users differ from L1 monolinguals in the way they know and use the L1 and the L2, but how are these differences relevant to questions abou the role of the native speaker as a model for L2 learners? Should such differences be seen as deficits from the native speaker standard?

Labov's classic argument held that one group should not be measured against the norm of another (Labov, 1969), whether whites against blacks or working-class against middle-class; this was in a sense a belated recognition of ethnocentrism (Sumner, 1506) in linguistics. People cannot be expected to conform to the norm of another group to which they do not belong, whether groups are defined by race, class, sex or any other feature. People who speak differently from some arbitrary group are not speaking better or worse, just differently. The comparison between groups yields differences, not deficits.

However, teachers, researchers and people in general have often taken it for granted that L2 learning is a special case in which one group can be properly judged by the standards of another. When the L2 users' grammar differs from native speakers, when their pronunciation betrays where they come from, when their vocabulary differs from native usage, these are treated as signs of their failure to become native speakers, not of their success in their own terms. Just as it was once claimed that women should speak like men to succeed in business, black children should learn to speak like white children, working class children should learn the elaborated language of the middle-class, so L2 users are commonly seen as failed native speakers.

According to the definition used above, L2 users are not monolingual native speakers and never will be, as incapable of changing places as are most women and men. L2 users have to be looked at in their own right as genuine L2 users, not as imitation native speakers. It is no more relevant for language teaching that a few L2 users can pass for native speakers than it is for the study of gender that the woman novelist James Tiptree Jr. wrote as a man or it is for the study of race that the clarinet player Mezz Mezzrow claimed to be a white negro. The study of L2 learning should not base itself on a handful of extraordinary people. There is therefore no special case for making L2 users an exception to the dictum that one group should not be measured against another. Comparing the characteristics of native speakers and of L2 users is like comparing tomatoes and apples, useful only at a gross level. Second language users should be treated as people in their own right, not as deficient native speakers. Halliday (1968, p.165) speaking of dialects said 'A speaker who is made ashamed of his own language habits suffers a basic injury as a human being: to make anyone, especially a child, feel so ashamed is as indefensible as to make him feel ashamed of the color of his skin'. Clearly up till now many people have had little compunction in treating L2 users in this way.

An illustration can be found in the way that the measure of success in L2 learning is often held to be the amount of foreign accent—the extent to which people's pronunciation conforms to native standards. Joseph Conrad is taken as a L2 learning failure because Virginia Woolf among others claimed he was 'a foreigner. Speaking only broken English', despite the excellence of of his written English or indeed of his second language, French (Page, 1986). Apart from a few die-hard writers of letters to the newspapers, nobody would claim that speakers of Brummy and Glaswegian fail to acquire native speaker language because they were born in Birmingham or Glasgow. Consciously or unconsciously, people proclaim their membership of particular groups through the language they use. However, L2 learners are not supposed to reveal which part of the world they come from; they are failures if they have foreign accents, as much research into age differences in language learning assumes (Cook, 1986). Why should English people who sound as if they come from Houston be accepted as L1 successes when Polish people speaking English are deemed L2 failures for sounding as if they come from Warsaw? A French winegrower once said, perfectly sensibly, 'My English is not good but my French accent is perfect'. L2 users belong to the general group of L2 users, to smaller groups of L2 users with particular L1s, and to many other language groupings in the languages they know. The one group they can not belong to is native speakers by up-bringing. Only if the native speaker is the sole arbiter of language can L2 learners be seen as failures for revealing the social groups to which they belong.

An objection that is sometime raised to this argument is that it is the L2 users themselves who want to be native speakers. Even bilinguals, according to Grosjean (1989), 'often assume and amplify the monolingual view and hence criticize their own language competence'. Their attitudes are the product of the many pressures on them to regard L2 users as failed natives. Bilinguals have accepted the role assigned to them in a society dominated by monolinguals where bilingualism is a problem, monolingualism is not, just as psychologists once used to talk of 'African precocity' in children's development, not 'Euroamerican retardation' (Berry, Poortinga, Segall & Dasen, 1992). But this does not mean these attitudes are right. Members of various groups have indeed wanted to change the color of their skin, the straightness of their hair, or the shape of their eyes to conform to other groups but this highlights the status of various groups in society not their intrinsic deficits. The only occasion on which L2 users can justifiably be measured against native speakers is when they are passing for natives, for example when making translations to be read as native rather than non-native texts.

Monolingual bias is also reflected in the prevalent use of the term L2 learner for anybody who knows a second language while L1 learner is not applied to an adult native speaker. People who learn a second language are implied to be in a permanently unfinished state, never reaching a final form (Firth & Wagner, 1997, p.292). Hence L2 user is used here, for the person who uses a second language and L2 learner for the person in the process of learning it. While complete consistency is impossible, it seems preferable at least to attempt to credit successful L2 learners with the status of users. It does, incidentally, seem even more condescending to reduce L2 acquirer to L2er (Schwartz and Sprouse, 1996, p.42).

The logical consequence of the above arguments is that language teaching should be more aware of the student as a potential and actual L2 user and less concerned with the monolingual native speaker. Abandoning the native speaker totally may be unrealistic since the stereotype is so entrenched in teachers' and students' minds. Yet teaching can at least take some steps in the right direction. The following suggestions are written within an EFL background; some may apply rather differently to the teaching of English to students residing or intending to reside in an English-speaking country; indeed some of them have already been assimilated, for example the use of students' L1s in Special Alternative Instructional Programs in the USA (Lucas & Katz, 1994). They are more concerned with the language side of L2 teaching, that is to say the teaching of the components of competence such as syntax, vocabulary and phonology, than with the pragmatic side where the independence of some L2 use has already been accepted.

Language teaching has traditionally balanced the educational gains for the student's mind, attitudes and personality from learning the second language against the social and communicative gains from being able to use a second language for practical purposes. The aims of language teaching can be divided into 'internal' classroom goals that relate to the students' life within the classroom, such as communicating their backgrounds and feelings to each other, and 'external' target goals that relate to the students' use of English outside the classroom, such as travelling or living in an English-speaking environment (Cook, 1983).

The classroom internal goals are not therefore directly related to the actual use of the L2 in the world outside, whether by native speakers or by L2 users, and so are less affected by any change in the status of the native speaker. The process syllabus in which students negotiate continuously over what they want to do and achieve (Breen, 1984), relates neither to the native speaker nor to the L2 user, only to the students' own wishes. Community Language Learning allows the students themselves to shape the processes and goals in the classroom without reference to anything outside (Curran, 1976). Though the students are still doubtless influenced in their choices by target-based perceptions of what they will need as L2 users and of the status of native speakers, in principle they can decide what they like.

Similar stress on the classroom can be found in task-based learning, a movement that now brings together areas ranging from the procedural syllabus (Prabhu, 1987) to the psychology of attention (Skehan, 1998). Writers on task-based learning seem divided over the extent to which tasks should be related to what happens outside the classroom. Nunan (1995, p.62) divides tasks into 'real-world' tasks that is to say 'the sorts of tasks required of them in the world beyond the classroom', and pedagogic tasks 'things which it is extremely unlikely they would be called upon to do outside the classroom'; Willis (1996) however, does not make external relevance one of the categories of task. Skehan (1998, p.96) considers it desirable for tasks to have real-world relevance 'but difficult to obtain in practice'. Task-based teaching has not been concerned with external goals because of its primary concern with the classroom itself. Issues about native speakers and L2 users are relevant only to the extent that tasks refer to 'the world beyond the classroom'.

These approaches value language teaching as an educational activity benefiting the students in many ways, not for utilitarian ends outside the class. The native speaker model is unnecessary: students get many things out of learning the language other than sounding like native speakers. The alternative aims of proficiency or expertise could be applied to these classroom-based goals. Skehan (1998) for instance sets goals of fluency, accuracy and complexity, without explicit mention of either the native speaker or L2 user. These are L2 student goals rather than L2 user goals—abilities that students acquire through L2 learning unrelated to target use.

At the other extreme, target-based external goals were emphasized in the heydays of the audiolingual and communicative methods of teaching. Audiolingualism stressed the situations and language used by natives (Rivers, 1964). Communicative teaching analysed the students' needs in terms of notions, functions, topics and so on (Van Ek, 1975), leading to the familiar lists of vocabulary and structures in coursebooks such as Reward (Greenall, 1994) to this day. As communicative needs were seldom established by empirical research into what happens in L2 user situations, the native speaker role is all pervasive. External target-based teaching is also sometimes found in English for Specific Purposes where detailed analyses are made of the English needed in specific situations—restaurants (Bung, 1973), medical research papers (Nwogu, 1997) or science lectures (Jackson and Bilton, 1994). Again in so far as such descriptions reflect what native speakers do, not skilled L2 users, they only have indirect links to the L2 user target.

A practical way of moving towards an L2 user model is to present students with examples of the language of L2 users and of language addressed to L2 users; the 'pedagogic corpus' of language the students encounter (Willis 1993) should be expanded to include specimens of the language that L2 users need rather than native speakers. This is not the same as saying that the students should listen more to each other but that they should encounter skilled L2 use. Willis (1996, p.12) points out that an 'internationally acceptable version of the target language' rather than a native speaker variety could be used. At least some of the authentic recordings used in the classroom could show skilled L2 use; at present they are authentic for native speakers not for L2 users. Many examples of L2 English are available from the media; most Continental politicians manage to give fluent television interviews in English, even if English politicians rarely manage the reverse; English language newspapers from many parts of the world can easily be accessed over the Web; for example the Straits Times from Malaysia (http//www.straitstimes.asia1.com) or the Santiago Times from Chile (http//santiagotimes.cl) provide good L2 user English as well as native-produced articles.

Teaching can also reflect the language used by L2 users to other L2 users, the most extreme perhaps being codeswitching. For example the New Crown English course in Japan uses some code-switching in dialogues (Morizume et al, 1997). Some of the language students encounter could reflect the modifications made by L1 users in their speech to L2 users, say providing information more explicitly (Arthur, Weiner, Culver, Young & Thomas, 1980); if the students have heard only native-to-native speech, they will be unaccustomed to such features when they eventually encounter them.

A related point concerns the situations from the outside world that are modelled in teaching. Taking the situations in coursebooks as representative, these fall into two broad types. One is the exclusively native situations in which all the roles are taken by native speakers, as seen on virtually every page of any coursebook, particularly the 'authentic' conversations in the COBUILD course (Willis & Willis, 1988) which rely on recordings of English native speakers talking about themselves and carrying out tasks with each other such as giving directions and identifying photos. While this may well cover relevant vocabulary said by 'real' native speakers, which is indeed the main aim of the course, the conversations are between native speaker friends and acquaintances, with hardly an L2 user in sight. To take another direction, the 'communicative aims' in the beginners' course Flying Colours (Garton-Springer & Greenall, 1950) include 'asking who people are', 'greeting people', 'talking about people's homes' and so on; the word 'people' is not explained but the actual text shows that, with few exceptions, the people in the situations are native speakers of English, even if some are multi-ethnic.

The other main type is mixed situations in which one or more roles are played by an L2 learner or a low-level L2 user, a typical example being the foreigner asking the way from the native speaker, seen in virtually all communicative or audiolingual materials. Situations involving low-level L2 users may be relevant, provided they do not fall into the funny foreigner stereotype of Manuel in Fawlty Towers, the comic Spanish waiter perpetually misunderstanding everything addressed to him in English. One possibility is indeed to reverse the roles so that the native speaker is ignorant, the L2 learner omniscient, as in some English courses where a native shows an English person the sights and customs of the home country; the course Angol Nyelv Alapfoken (Edina and Ivanne, 1987) for example features English used by travel agents and tour guides in Hungary. It is, to say the least, unhelpful and unmotivating if the only L2 users the students see in the classroom are incompetent and ignorant.

The basic need is to present situations in which L2 users take part. The unequal gender roles in EFL textbooks have been pointed out by, for example, Sunderland (1992), with women being fewer in number, lower in status and age, and less active conversational participants. The status of L2 users is in even more need of redress, since their positive representation is virtually non-existent. At one level there is the simple need to demonstrate that L2 users exist in the world as role models that students can try to emulate. Psychology books have lists of famous bilinguals including for instance Gandhi, Picasso, Marie Curie and Samuel Beckett (Grosjean, 1982, p.285); the famous people in coursebooks tend to be Ronald Reagan, Queen Elizabeth II and the Beatles (Greenall, 1994, p.83), hardly known for their L2 skills. Making some parts of language teaching reflect an L2 user target would at least show the students that L2 users exist in their own right and are not just pale shadows of native speakers.

A possible technique for introducing L2 user situations into teaching can be found in the cross-cultural training in Cushner and Brislin (1996). This presents a series of key intercultural problems; students discuss the alternative interpretations suggested and then see which of them is most likely. For example one case study features an American student in Germany who is perplexed by her apparent rejection by her German colleagues; the students discuss the possible causes and discover that the most likely reason is her lack of interest in politics. While the selection of such situations or alternatives would be difficult, at least this brings the figure of the L2 user into the classroom as a person in between two cultures.

An interesting type of L2 user is in fact the non-native speaker teacher. Often it is assumed that native speakers intrinsically make better teachers than non-natives; 'Learn French from the French' is an advertising slogan for a language school in London. Medgyes (1992) comes to a more balanced conclusion about possible advantages and disadvantages of being a native speaker. However, students may feel overwhelmed by native speaker teachers who have achieved a perfection that is out of their reach; as Kramsch (1993, p.9) puts it, 'non-native teachers and students alike are intimidated by the native-speaker norm'. Students may prefer the fallible non-native speaker teacher who presents a more achievable model.

Most orthodox EFL teaching methods minimise the role of the L1 (Howatt, 1984, p.212), called by Stern (1992) the 'intralingual strategy'. Apart from the never-dying but usually decried grammar/translation method, virtually all language teaching methods since the Reform Movement of the 1880s have insisted that teaching techniques should not rely on the L1, whether the audio-lingual and audio-visual methods, the communicative methods or the Silent Way; 'inventories of classroom techniques exist of which only a handful are not intralingual' (Stern, 1992, p.289). The insistence on the L2 by methodologists does not mean that the L1 has not in practice been used in most classrooms, but that this goes against the official doctrine. The UK National Curriculum for Modern Languages is typical in saying 'The natural use of the target language for virtually all communication is a sure sign of a good modern language course' (DES, 1950, p.58). Partly the following arguments are an attempt to reassure teachers and students that they are right to use the L1 in the classroom, despite the prevailing climate of opinion.

Exceptions to this orthodoxy are Community Language Learning with its reliance on translation (Curran, 1976) and a small group of teaching methods that employ alternating languages. These include: the New Concurrent Method (Jacobson & Faltis, 1950), which advocates controlled codeswitching; reciprocal language teaching, in which matching pairs or groups of students who want to learn each other's language alternate languages as they choose (Hawkins, 1981; Cook, 1989); and the Tandem computer network, which gets pairs of students learning different languages to send each other e-mails in their respective L2s . Apart from these more radical alternatives, at best in coursebooks the L1 supplies meanings for words or an occasional discussion topic; The Beginners' Choice (Mohamed & Acklam 1992) for example asks students to decide whether adjectives go before or after nouns in their L1s.

At least two ways of using the L1 in the classroom should be distinguished. One is for presenting meaning: when students need the meaning of a new word or grammatical structure, they can access it through translation into their L1, whether by the teacher or by looking it up in a dictionary, or through explanation in the L1, again from the teacher or from a grammar-book. Clearly multi-competence theory looks for such links between the languages that can be developed, such as translation, rather than trying to see the languages as being in two separate compartments. One reason for the lack of reliance on the L1 has undoubtedly been practical convenience for the teacher. Given that much EFL methodology arose from multilingual adult classes, teachers could not use the L1s of their pupils for conveying meaning as they might know at most one or two of them.

The other main use of the L1 is during the classroom activities. Teachers are encouraged to use the L2 throughout the class, as we have seen; students are expected to use the L2 even in activities where there would be natural codeswitching with fellow students who share the same L1. A typical remark is 'If they are talking in small groups it can be quite difficult to get some classes – particularly the less disciplined or motivated ones – to keep to the target language' (Ur, 1996, p.121). While the practical rationale of the teacher's ignorance of the students' languages again excuses this in multilingual classes, this restriction should not apply to those classes where the students share a common L1. L2 users have the L1 permanently present in their minds. Every activity the student carries out visibly in the L2 also involves the invisible L1. The superficial L2 nature of the classroom disguises the presence of the L1 in the minds of the students. From a multi-competence perspective, all teaching activities are 'crosslingual' in the sense of Stern (1992); the difference is whether the first language is visible or invisible, not whether it is altogether absent.

The message that much teaching seems to convey is that the students should aim at L2 use unrelated to the L1, something that is virtually impossible to achieve and that denies their status as L2 users. Though teaching manuals such as Willis (1996) or Scrivenor (1994) now countenance some L1 use, the implication is that ideally the students would not be using their L1; 'As an ideal I would like a classroom where learners were free to use their own tongue but in fact mostly chose to use English' (Scrivenor, 1994, p.192). L2 use is not seen as desirable but as a necessary evil. One practical suggestion is that teachers should see the L1 as a positive factor in the class rather than as a negative factor that has to be endured. This may simply be putting more positive light on what already happens in many classrooms. Such a change has already taken place in some second language classrooms (Lucas & Katz, 1994); teachers can come to accept mixed languages in the classroom, however reluctant they are at first (Giauque & Ely, 1950).

A second suggestion is to introduce activities that deliberately involve both languages. The Institute of Linguists (1988) for instance asks elementary students to listen to messages in the L2 and to relay them in any language they like, whether L1 or L2; they test advanced students by getting them to write a report in either language based on a series of interviews and texts in the L2. The classic dual-language task was translation, which might be used as a vehicle for more communicative exercises: write down your favourite recipe in your L1 and then decide how you would explain it in English to a fellow student with a different L1. These activities above all see the student as an 'intercultural speaker' (Byram & Zarate, 1994), not an imitation L1 user, doing something the monolingual can never achieve. The use of such activities in teaching can go some way to develop the student as an L2 user not an imitation native speaker.

If the aim of teaching is to create L2 users, the description of English that is logically required is a description of L2 English. One of the claims of applied linguistics has always been that language teaching can make use of descriptions supplied by linguists (Corder, 1973); much applied linguistics today is indeed description-oriented rather than problem-oriented.

A trend in recent years is the use of language corpora for descriptive purposes. The COBUILD project for example produced a large database of English from which it could derive grammars, dictionaries and teaching materials (see for example the list in Payne (1995)). Such descriptions would be far more useful if L2 users were represented in the corpora. We do not at present have a clear idea of what typical success-ful L2 users know except through the distorting mirror of descriptions of native speakers. Furthermore corpus-based description is relevant to teaching only in so far as it is linked to a testable theory of language learning: it needs to attain the level of explanatory adequacy of showing how language is learnt, not just the level of observational adequacy of listing thousands of occurrences said by hundreds of people (Cook, 1985).

In the absence of descriptions of L2 users on which to base language teaching, one possibility is to see what can be gleaned from accounts of L2 learning. Collections of learner English such as the Longman Corpus of Student English potentially could act as stepping stones. Syllabuses and teaching materials could suggest intermediate goals for the students on their way to becoming successful L2 users. For example the ESF project (Klein & Perdue, 1997) discovered that L2 learners of European languages acquired a basic grammar consisting of three rules: a sentence may be Subject-Verb-Object Jane drinks beer, Subject-Copula-Adjective Beer is good, or Verb-Object Drinking beer. This L2 grammar is valid not just for L2 English but also for L2 German, Dutch, French and Spanish, almost regardless of the learner's L1. While this is an interim stage of L2 learning, it nevertheless provides a useful description of an L2 target for the beginner stage. An additional claim made in much contemporary work with syntax is that the initial stages of L2 acquisition depend upon word order rather than inflection (Klein & Perdue, 1997; Pienemann, 1985), a finding of major importance for the teaching of English, which traditionally spends considerable effort on the plural –s, past tense –ed and so on.

The suggestion of using L2 user language should not be over-stressed in that the differences between L2 users and native speakers described above could be marginal; L2 user goals could be hard to define because of the great variation between L2 users. Nevertheless taking the description of the native speaker as the basis of language teaching is in a sense a temporary short-cut that avoids describing what L2 users are like rather than the more satisfactory solution of tackling the L2 users themselves.

The vital point of this argument is not so much the specific suggestions as the attitudes which underly language teaching. If students and teachers see L2 learning as a battle that they are fated never to win, little wonder they become dispirited and give up. If they are convinced of the benefits of learning a second language and recognise their unique status as standing between two worlds and two cultures, more students may go on higher levels of L2 use; those who do give up may feel more satisfied with the level of L2 use they achieve. The graded objectives movement in language teaching tried to set interim targets (Harding, Page & Rowell, 1981) so that at whatever level students stop learning a language, they take away something of benefit. A beginners first year EFL course took a world-wide external goal to be travelling travelling abroad using English (Cook, 1980); the students who stopped after one year still gained a useful skill, based on the L2 user, not the monolingual native.

The two main suggestions in this paper are to place more emphasis on the successful L2 user and to use the L1 more in teaching. Together with the attitude change, these can bring language teaching back to the reality that it is helping people to use second languages, not to imitate native speakers. While it is utopian to imagine that students, teachers, or indeed L2 researchers would overnight give up their reliance on the native speaker, judicious changes such as these can at least begin to acknowledge that L2 users have strengths and rights of their own by giving the students role models of L2 users in action and by requiring the use of both languages by one person: in short convincing students that they are successful multicompetent speakers, not failed native speakers.