This paper argues that the starting-point for language teaching should be the recognition that the second language user is a particular kind of person in their own right with their own knowledge of the first language (L1) and the second language (L2), rather than a monolingual with an added L2. An L2 user is a person who uses another language for any purpose at whatever level (Cook, 2002a). They might be writers like Nabokov or Conrad creating novels, ethnic minority children acting as translators for their parents in medical consultations, tourists travelling on holiday, journalists plying their trade, businessmen doing deals on the internet, tennis-players giving television interviews. Some L2 users acquired their second language through practical living, others after long study in the classroom; some need it for survival in everyday existence, others for amusement, pleasure or education. In short L2 users are as diverse as the rest of humanity. Their needs and uses of language are as wide as monolinguals, if not wider.

The L2 user concept is based on the multi-competence view of second language acquisition, which has been developed as an overall approach to L2 learning since Cook (1991). Multi-competence means the knowledge of two or more languages in one mind. The term thus encompasses the concept of interlanguage, which has been used only for the L2 component, and the L1 component. It treats the mind of the L2 user as a whole rather than as having separate L1 and interlanguage components. It argues that studying second language acquisition means accepting this totality, not just the interlanguage component.

The main question for multi-competence research is how the two languages relate in the same mind. At some level, the two languages must obviously co-exist. The question is at which level they separate or, indeed, if they separate at all. One interpretation sees this as an 'integration continuum' (Cook, 2002a), going from total separation between the languages at one end to total integration at the other. This continuum may represent a person's development over time or it might be that particular individuals may have more or less integration depending on how long they have been learning a second language or it might depend on other individual factors. The continuum could also apply differently to various aspects of language within the same person's competence: pronunciation and vocabulary might be more likely to be integrated, grammar less likely. The position on the continuum may also vary for the individual from moment to moment according to Grosjean's concept of language mode, the integration of the two languages depending on the extent to which the speaker perceives they are in monolingual mode (using one language, whether L1 or L2) or bilingual mode (using both languages together) (Grosjean, 2001).

The term 'L2 user' is then crucial to the overall approach. Chomsky (e.g. 1986) insists that linguistics has to account for the linguistic knowledge of the adult native speaker; only after this has been described can linguistics go on to see how language is acquired and explain what knowledge of language is. The study of the first language starts from the mature L1 user not from children: the account of language acquisition depends on first describing the linguistic competence of the native speaker – what it is that is acquired.

Similarly second language acquisition (SLA) research is about the minds of people who have successfully reached a usable level of the second language, not just about how they learnt it. Some L2 users may also be L2 learners who are still acquiring language: an immigrant using the second language in the street becomes a student learning the language when they step through the classroom door. But we are no more justified in saying that an L2 user is a perpetual L2 learner than we are in saying an adult native speaker is an eternal L1 learner. When SLA research talks about everything to do with the L2 as 'acquisition' or talks about people who speak second languages as 'L2 learners' (or even the regrettable term 'L2er'), it implies that no person using a second language succeeds in getting to a state of using the language properly. A person of fifty who has used a language all their lives is not called an L1 learner; why should their use of a second language for, say, thirty years still be deemed learning? The only people who escape the label of 'L2 learner' are balanced bilinguals, equally native-like in both languages, like a double monolingual rather than an L2 user.

A crucial relationship that has been changed in the multi-competence perspective is then that between the native speaker and the person acquiring or using a second language. L2 research from the 60s onwards made use of the interlanguage concept to describe the independent language of the L2 learner. The aim was to describe learners in their own right, to look at their grammar, their phonology and their vocabulary as things of their own. Yet the research methods employed consistently involved measuring the L2 learner against the native speaker, whether in terms of Error Analysis (errors being things natives wouldn't say), obligatory occurrences (contexts where natives have to have particular forms), or grammaticality judgements (sentences natives reject). The model against which the learner was measured was how a native speaker performed.

This led to a pervasive air of failure and gloom: the interesting thing about people acquiring second languages was why they were so bad at it, few if any achieving the levels in a second language any monolingual can attain in the first language. To take some representative quotations, which could be repeated from virtually every general book about second language acquisition, ‘failure to acquire the target language grammar is typical’ (Birdsong, 1992, p.706), ‘children generally achieve full competence (in any language they are exposed to) whereas adults usually fail to become native speakers’ (Felix, 1987, p.140), and ‘Unfortunately, language mastery is not often the outcome of SLA’ (Larsen-Freeman & Long, 1991, p.153). No criterion is proposed for L2 success other than being like a native speaker. Success means getting as close as possible to this target.

Is the native speaker target in fact attainable? A few people have been found who can pass for native speakers.. But their numbers are so small that they are as relevant to SLA research as Michael Schumacher's driving skills at racing a Formula 1 car are to my daily drive to work. The reasonable definition of a native speaker is a person speaking the language they learnt first in childhood. By this definition it is impossible for any L2 learner ever to become a native speaker without going back in time to their childhood; nothing learnt in later life could qualify you as a native speaker.

Using the native speaker target commits one to comparing the two groups of native speakers and L2 users as if one were intrinsically trying to become the other. The comparison is loaded because one group is defined in terms of the other. This does not occur in other areas of language study, following Labov's powerful arguments for linguistic differences between groups rather than deficits (Labov, 1969). Thus it is no longer felt to be proper for linguists to talk about the language of Black citizens of the USA as deficient with regard to that of whites, about the language of working-class children as deficient compared to that of middle-class children, or the language of women as a deficient version of men's language, though all of these were claimed at one time or another, and many are still held by non-linguists. Does measuring L2 users against L1 native speakers amount to falling into the same trap? Second language acquisition might be a unique case where we are justified in seeing one group of human beings in terms of another: while linguists don't treat women as failed men, SLA researchers may legitimately treat L2 users as failed native speakers. The reasons why L2 users should be treated as different rather than deficient will be elaborated below. L2 users have the right to speak English as L2 users rather than as imitation native speakers, are exemplified by the French wine-grower who said 'I speak English very badly but my French accent is perfect'. L2 users should be judged by what they are, L2 users, not what they can never be by definition, native speakers.

If L2 users are indeed unique users of language in their own right, not imitations of native speakers, what are their characteristics?

1) the L2 user's knowledge of the second language is typically not identical to that of a native speaker. Controversy has raged over whether a small proportion of native speakers can use language identically to monolingual native speakers. Some point to 'balanced' bilinguals whose second language still differs from native speakers in grammaticality judgment tests (Coppetiers, 1987), others to a small group of L2 students who cannot be distinguished from monolingual native speakers (Bongaerts et al, 1997). As we saw above, the fact that a few untypical people are able to run a hundred metres in less than ten seconds does not tell us much about the normal running abilities of the human race.

The knowledge of the second language of the vast majority of learners is different from that of native speakers. Much effort in SLA research has been devoted to seeing why this is so, whether through Error Analysis, access to Universal Grammar or L1 interference. Given that multi-competence means having two languages present in the same mind, it is hardly surprising that the knowledge of the second language is not like that of a monolingual. The L2 learner has had the first language always present while acquiring the language; the L2 user still has it somewhere in their mind whichever language they happen to be functioning in. The interlanguage component of multi-competence forms part of a system with another language and so is bound to be different from the L1 grammar of a monolingual who has only ever had one language.

But difference is not deficit. The language of L2 users may be a perfectly normal language system of its own type; why should the more complex state of the mind with two languages be measured against the comparatively simple single language system of a monolingual? If the target is not to imitate the native speaker, the question of whether the eventual state of language knowledge should be like that of a native speaker is a side issue, no more relevant than discussing how many men can pass for women – even if both topics provide stimulating discussion and amazing anecdotes.

As always with matters of language, the neutral linguists' view may differ from that of the man in the street. For instance, despite years of linguists asserting that all dialects of English are equal, they are still not perceived as such by many employers, parents and teachers. Because of the traditional power of the native speaker, there are sociolinguistic arguments about native speakers' expectations that non-native speakers should be like natives. These have to be borne in mind in that students have to fit in with a society in which native-speakerness is still highly-valued and so they may be disadvantaged by being taught that it is not a useful goal, however true this may be in a linguistic sense. Even here though, with languages like English, perhaps the majority of use takes place between non-native speakers with different L1s who do not necessarily have the same expectations as monolingual natives.

So in principle the proper goal for an L2 user is using the second language like an L2 user, not like an L1 user, with the exception, say, of those who want to be spies. This is easier said than done since there are no descriptions of what successful L2 use might be. Perdue (2002) describes the basic grammar that all L2 learners seem to acquire in the early stages, regardless of which L1 or L2 is involved. Jenkins (2000) has shown what a syllabus would look like that was based on the comprehensibility of English among L2 students. It may be that there is not a single successful end-point to L2 acquisition as there is for L1 learning and so many models of successful use are needed. Hence, for the time being, the native speaker model will have to do as a rough and ready approximation until there are the descriptions of L2 user grammar, L2 user frequency and L2 user phonology to put in its place. But this does not alter the fact that, apart from the sociological whims of the native speakers themselves and of those dominated by the monolingualist perspective, there is no good reason why the language of an L2 user should approximate that of a native speaker, Students should not be penalised for deviations from native speaker norms if they are functioning perfectly adequately in the second language, say by code-switching.

2) the L2 user has other uses for language than the monolingual. At one level there are uses of language which involve both languages more or less simultaneously such as translation and code-switching. Some see these as extensions to the monolingual's ability to paraphrase and change style (Paradis, 1997); others see the monolingual uses as limited versions of the full range available to L2 users (Cook, 2002a). But L2 users employ a wider range of language functions than a monolingual for the needs of their lives. An L2 user can be seen in terms of métissage – 'the mixing of two ethnic groups, forming a third ethnicity' (Canada Tree, 1996).

At another level, everything the speaker does is informed by the second language, whichever language they are using. Few L2 users can so compartmentalise their languages that they effectively switch one language off and function solely through the other. Their everyday use of language is subtly altered by their knowledge of other languages.

Furthermore the L2 user never gets to function in exactly the same situation as the L1 user; the very presence of an L2 user changes the perceptions of the participants. The language used by the L2 user may for example be more polite than that of natives, for instance 'Thank you very much indeed' rather than 'Thanks' (Cook, 1985) but this usage may mirror the native speaker's expectations: we don't expect L2 users to speak like us and regard near-nativenesss as suspicious, even spy-like. Native speakers do not talk in the same way to non-natives as they do to natives, partly in terms of syntax, partly in terms of how information is presented, so the extra formality of 'Thank you very much indeed' may be a correct response by the L2 user to the native speaker. Practical everyday situations such as shopping and going to the doctors are different when an L2 user is involved. So the typical dialogues in these situations depicted in language teaching coursebooks are misleading when they involve only natives and native-to-native speech. For the majority of learners non-native speech to non-native-speakers may be far more relevant and valuable.

3) the L2 user's knowledge of their first language is in some respects not the same as that of a monolingual. A recent volume (Cook, 2003) described a variety of effects of the second language on the first. The speaker's knowledge of their first language is undoubtedly influenced by the other languages they learn, whether in terms of:
– syntax: Japanese speakers of English are more prone to prefer plural subjects in Japanese sentences than Japanese who don't know English (Cook et al, 2003).
– the lexicon: experienced Russian speakers of Hebrew use a less rich vocabulary in Russian than comparative newcomers (Laufer, 2003).
– stylistic complexity: Hungarian children who have learnt English use stylistically more complex writing in the first language (Kecskes & Papp, 2000).
– pragmatics: Russian learners of English begin to rely on expressing emotions as states rather than as process (Pavlenko, 2003).
– phonology: French users of English pronounce the /t/ sound in French with a longer VOT than monolinguals (Flege, 1987).

It seems clear that the language processing of people who know another language is no longer the same as monolinguals, even in their first language, though the differences may be small and need complex techniques to establish. The relationship between the two languages in the mind of the L2 user goes in both directions, not just one.

4) L2 users have different minds from monolinguals. The effects of the more complex system of multi-competence extend outside the area of language. Research over the past forty years has confirmed the effects of the L2 on the minds of the users, heralded by such traditional goals of language teaching as brain training. Children who have learnt a second language:
– have a sharper view of language if they speak a second language (Bialystok, 2001).
– learn to read more quickly in their first language (Yelland et al, 1993).
– have better 'conceptual development', 'creativity' and 'analogical reasoning' (Diaz, 1985).

Current research is exploring whether certain basic concepts are modified in those who know a second language. Athanasopoulos (2001) found Greek speakers who knew English had a different perception of the two Greek words covered by English blue, namely γαλαζιο (ghalazio, ‘light blue’) and μπλε (ble, ‘dark blue’), than monolingual Greek speakers. Bassetti et al (2002) found that Japanese people who had had longer exposure to English chose shape rather than substance more often in a categorisation experiment than those with less exposure. Some concepts in the L2 user's mind may then be influenced by those of the second language; others may perhaps take forms that are the same neither as the first language or the second. This does of course assume that people who speak different languages think to some extent in different ways, a revival of the idea of linguistics relativity that has been gaining ground in recent years (Levinson, 1996).

To sum up, L2 users have different language abilities and knowledge and different ways of thinking from monolingual native speakers. Rather than encouraging the students to approximate the native speaker as much as possible, teaching should in principle try to make them independent L2 users who can function across two languages, with mental abilities that the monolingual native speaker cannot emulate. The only caveat is that, given the bias in monolingual speakers toward the native speaker norm, some concession may have to be made for those students who want to take part integratively in native speaker communities rather than communicate with other non-native speakers, obviously less relevant for those learning an international language used across the globe such as English as opposed to a language spoken in a limited geographical area such as Japanese.

An implicit goal of language teaching has often been to get as close to the native speaker as possible, recognising the native speaker as having the only acceptable form of the language. If the arguments above are accepted and the sociolinguistic power of the native speaker can be set aside, a more achievable goal is to make students into successful L2 users. The native speaker target has been more a matter of exerting the power of the native speaker than a recognition of what students actually need or can achieve.

One perennial justification is that the students themselves want to be like native speakers. As Grosjean (1989) points out, L2 users are part of the same social climate as monolinguals and have come to accept that native speakers rule the roost. I couldn’t count how many times a perfectly fluent L2 user has apologised to me for their level of English; yet they were doing something I could not possibly do in their language: why wasn't I the one who was apologising for being a monolingual? Because, as a native speaker, I had the inherent power over others trying to claim membership of my group, even in a situation when multiple languages were in use. People who have spent their language learning lives trying to speak as much like native speakers as possible become upset when they are told such a target is meaningless: they want to hear praise that they could almost be mistaken for natives.

What is required is a proper description of L2 users to form the basis for teaching. Here comes the major problem for designing the language syllabus: the goals that can be defined in L2 user terms no longer constitute a single putatively unified target like the native speaker. What makes a successful level of L2 use for a particular individual or a particular country may not apply to others. An immigrant who wants to practice medicine in an English-speaking country needs very different L2 use from say a medical researcher who wants access to the medical literature and the web through English. A child in Shanghai who may never encounter a live native speaker of English needs different L2 use from a Chinese child in Vancouver. This is the dilemma that confronted ESP: as soon as you start looking at individual needs for a second language, you need to think of specialised goals.

Obviously many situations do generalise for large numbers of L2 users: travelling in English in non-English-speaking countries may be useful for large numbers of users who will seldom if ever converse with native speakers. The description of such contact situations may be a valuable part of the syllabus, rather than the typical native-to-native or native-to-non-native ones usually found in textbooks. Jenkins' (2000) account of the phonological needs of multilingual students in classrooms for talking to each other is an interesting example of this approach, though it does not reach outside the classroom environment. Eventually there may be descriptions of the language of successful L2 users on which to base our teaching. Meanwhile, because of their long tradition and their availability, descriptions of the native speaker such as grammar-books or corpora of native speaker texts may be what we have to fall back on out of necessity. But this is a temporary expedient; what is needed is proper descriptions of successful L2 users which can show their unique characteristics of grammar, vocabulary and pronunciation, rather than relegating them to deviation from the language of the native speaker

It is convenient to divide the goals of language teaching into external goals relating to the students' present or future use of the second language outside the classroom and internal goals relating to the students mental development as individuals (Cook, to appear).

Getting rid of the native speaker target changes the external goals of language teaching. The students' goal is to be able to use both the L2 and the L1 in the appropriate situations and for the appropriate uses. It is not necessarily to be like a native speaker or to mingle with native speakers, even if this may be an appropriate goal for some people. One of the unfortunate side effects of the communicative revolution in language teaching was its almost behaviourist emphasis on external goals than on the internal development of the learners' minds. Hence for many years textbooks and syllabuses were concerned with how students should use the language in conversation and how they could convey ideas to other people (Wilkins, 1976) not with internal goals

Yet a traditional benefit of language teaching was the internal goal of improving the students' mind within a humanist education. As we saw earlier, there are indeed cognitive changes in L2 users' minds compared to monolinguals, mostly to their benefit. At the level of national curricula, the UK Modern Language Curriculum (DfEE, 1999) expects pupils to 'understand and appreciate different countries', to 'learn about the basic structures of language' and how it 'can be manipulated'. The curriculum for Israel (1998) divides language teaching into domains: the domain of appreciation of literature and culture 'addresses the importance of fostering understanding and developing sensitivity to people of various cultural backgrounds' and the domain of language helps 'pupils develop their language use as well as gain further insight into the nature of their mother tongue'. At the level of the students themselves, Coleman (1996) found that popular reasons for learning a modern language among UK university students were 'because I like the language' and 'to have a better understanding of the way of life in the country or countries where it is spoken' – internal goals.

So the consequence of an L2 user approach for the goals of language teaching means on the one hand basing the target on the external needs of L2 users not native speakers, on the other focussing on desirable internal changes in the student's mind. L2 users can add the ability to use a second language to their existing abilities so that they can behave as no monolingual can do. They change the contents and processes of their minds in a way no monolingual can match. Education has as always to balance the external value of a subject on the curriculum for the future social and career needs of the students against the internal value of the changing ways in which the students' minds can function.

If the native speaker is no longer the standard against which L2 users are measured, what does this do to the position of native and non-native speaker teachers? Successful students will never become like the native speaker teacher since the vast majority of them will not get remotely approximate native speaker speech and will not think in monolingual ways but L2 user ways. They are far more likely to become like the non-native speaker teachers, who are using a second language efficiently for a particular purpose. The only advantage of native speaker teachers is precisely that they are native speakers: if this is now immaterial to the goals of language teaching, it is no longer an asset. And of course in EFL settings nativeness is a waning asset since the first language is progressively changed in a native speaker in a non-native environment (Porte, 2003).

This approach ties in with the debate over the merits of native speaker and non-native speaker teachers in language teaching (Medgyes, 1992; Cook, 1999). In many parts of the world it is simply taken for granted that native speakers are best. Language teaching institutions stress this in their advertisements and in their employment policies. A trawl of the web immediately finds a school in Brazil that wants 'Native English speaker, bilingual, university degree', one in Italy that wants 'experienced, qualified professional native speaking English Language Teachers', one in Indonesia that needs 'Native EFL teachers' and one in China looking for 'Enthusiastic NATIVE English Teachers'. It is obviously felt to be a major selling point for an institution to have native speaker teachers, almost everywhere in the world.

Students are not necessarily as impressed by native speaker teachers as one might suppose. In a survey of attitudes to monolingualism I slipped in some questions about other topics, such as whether they agreed or disagreed with the statement 'Native speakers make the best language teachers'. The students in several countries were either adult learners of English or child learners aged on average 14. As seen in Figure 1, the approval rating for native speaker teachers expressed out of a hundred ranged from 72% for children in England down to 33% for children in Belgium and from 82% for adults in England to 51% for adults in Taiwan. While this indeed confirms a general preference for native speakers, this is far from overwhelming, perhaps with the exception of English adults. L2 students do not feel very strongly about the advantages of native-speaker teachers. Given that students are partly reflecting the knee-jerk reactions of the societies in which they dwell and the beliefs of their teachers and parents, it is surprising that they are comparatively luke-warm about native speaker teachers.

The pros and cons of native and non-native speaker teachers from the L2 user perspective are then:

- non-native speaker teachers provide models of proficient L2 users in action in the classroom. Here is a person who knows two languages using the second language effectively, showing that it is possible to do this in a language that is not one's own. The native speaker teacher who does not know the first language of the students is only a model of something alien that the students can never emulate. Here also is a person who can speak from personal experience of the differences that L2 learning has made to their mental lives, say, their appreciation of other cultures and their feeling for language, something that the monolingual native speaker cannot do and probably does not even appreciate that it exists.

- non-native speaker teachers present examples of people who have become successful L2 users. The non-native teacher has been through the same route as the students and has acquired another language, a living demonstration that this is possible for non-native learners. They have shared the student's own experience at some time in their lives and have learnt the language by the same route that the students are taking. The native speaker teacher cannot appreciate their experiences and problems except at second hand.

- non-native speaker teachers often have more appropriate training and background. Expatriate native speakers come in from outside the country and do not necessarily see the culture of the classroom and the values of the educational system in the same way; they usually do not have as deep a knowledge of the educational system as the indigenous teacher. Many native speaker teachers have not had the same level of teacher training as the native or have not been trained in the educational system and beliefs of that country. Though these are accidental by-products – not all native speakers are handicapped by ignorance of the local situation and by lack of training – this certainly applies to a fair proportion of the expats teaching English as a Foreign Language in different parts of the world.

- non-native speaker teachers may have the disadvantage of lesser fluency etc. The above is predicated on the non-native teacher indeed being an efficient L2 user who can speak fluently and communicate within the classroom. It may be, for one reason or another, that some teachers do not attain this level; in many countries teachers do not feel their command of English is adequate for the demands of their task. But this is only relevant if it reflects shortcomings in being L2 users, not shortcomings in being like a native speaker: the type of English needed for successful use as a language teacher may differ considerably from that of the native speaker.

If the first language is always part of the L2 user's multi-competence, we have to re-examine its role in language teaching. Current language teaching has mostly tried to minimise the use of the first language in the classroom (Cook, 2001). A typical view is expressed in the UK National Curriculum (DfEE, 1999):'The natural use of the target language for virtually all communication is a sure sign of a good modern language course'. This view has prevailed for the past hundred years: 'It is assumed throughout that the teacher’s success is judged by the rarity of his lapses into the foreign tongue' (Thorley, 1918). If the L2 user has two languages available in the same mind, teaching should make systematic, deliberate use of the first language, partly by developing methods that incorporate both languages, partly by evaluating when the L1 can be used effectively within the L2 classroom, both as part of a true L2 user situation and to help the students' learning. This means recognising the classroom as a true L2 use situation, not an imitation L1 situation, so that the first language can be used:

• as a way of conveying L2 meaning. In many circumstances the first language is as effective a means of conveying new meanings or words or constructions as any other, provided it does not lead to treating the meanings of the second language as translation equivalents of the first language.

• as a short-cut for explaining tasks, tests etc. If the point of a teaching activity is the task itself, not the way of getting people to do the task, the teacher has to find the quickest and most effective way of getting the task underway, which may well be to use the the students' first language. Rather one minute of instructions in the first language and 9 minutes in the second language doing the task than 9 minutes of instructions in the L2 and 1 minute in the L2 doing the task.

• as a way of explaining grammar. If the goal is for students to understand the grammar itself rather than to benefit from the incidental language involved, again the teacher has to choose the best vehicle to achieve this, which may be the first language, as indeed is already the practise by many teachers (Franklin, 1990). This has the incidental advantage of building on grammatical models and terminology familiar to the students rather than on the Latin-based concepts and terms of the English-speaking world's tradition of school grammar.

• for practising L2 uses such as code-switching. Mostly frowned on within communicative teaching, code-switching in the classroom is nevertheless almost inevitable when the students know each other's first language. Rather than being shunned, code-switching can be deliberately controlled by the teacher as a gain for the students, as recommended in the New Concurrent Method (Jacobson and Faltis, 1990).