The idea of linguistic multi-competence was first proposed in the early 1990s as the knowledge of more than one language in the same mind (Cook 1991). Recently it has been discussed in areas far outside its original remit – dynamic systems (De Bot, Lowie & Verspoor 2005), multilingualism (Herdina & Jessner 2002), macroacquisition of language by communities (Brutt-Griffler 2002), post-structuralist construction of identity (Golombek & Jordan 2005), lingua francas (Jenkins 2006), heritage languages (Valdés 2005) and cross-linguistic influence (Pavlenko & Jarvis 2006). This paper tries to accommodate these developments within a multi-competence framework. Since they raise fundamental issues and complex about the nature of language and of the language learner and user and range across many approaches to language, this can be far from a final account. Three questions will be tackled:
- who are the second language users?

- what is the ‘language’ they know?

- what is the community they belong to?

The classic view of second language acquisition (SLA) saw it as the learner creating an interlanguage by drawing on the first language (L1), second language (L2) and other factors (Selinker 1972). Initially the term multi-competence was devised as a convenient term for the knowledge of languages in one person’s mind (Cook 1991), i.e. the L1 plus the interlanguage. This had the consequence of separating someone who knows two languages from the native speaker as a person in their own right: the relationship between the L1 and the interlanguage within one mind is different from that between the interlanguage in one mind and the L2 in another mind (which is actually a first language for the person involved). Hence the term ‘L2 user’ became preferred over ‘L2 learner’ and its variants as it conferred separate identity rather than dependent status, implying the person is always learning, never achieving. The research that span off from this conceptualisation of multi-competence has concerned itself with the relationships between the two language systems in one mind, particularly reverse transfer from L2 to L1 (Cook 2003) and with the relationships between the language systems and the rest of the L2 user’s mind (Cook et al. 2006), visualised as an integration continuum between the two language systems (Cook 2003) rather than as the conventional division between compound and coordinate bilingualism (Weinreich 1953).

Question 1. Who are the L2 users?

People usually accept the idea of the monolingual native speaker without much quibbling; they feel they know what they’re talking about when they say ‘a native speaker of English’ or ‘belonging to the English-speaking community’. Despite warnings such as ‘a linguistic community is never homogenous and hardly ever self‑contained’ (Martinet 1953: vii), they tend to accept that native speakers form a uniform community. The arguments against this idealisation will not be developed here as it has been a well-worn path in applied linguistics to reject Chomsky’s definition of linguistic competence (Chomsky 1965) in favour of Hymesian communicative competence (Hymes 1972); nevertheless SLA research and language teaching have paid little attention to native speaker variation whether within or across individuals.

Defining the L2 user or the community of L2 users has proved far more problematic. The standard solution is to speak only of the community of monolingual native speakers, whether of the L1 or the L2, as a monolithic whole. Almost invariably this leads to successful L2 users being seen as those who can pass as members of the monolingual native speaker community rather than having membership of a community of their own, a condition that very few would pass, seen clearly in typical quotations like 'Relative to native speaker's linguistic competence, learners' interlanguage is deficient by definition' (Kasper & Kellerman 1997: 5).

The aspect focused on here is to remind researchers that both the languages that the L2 users know and the communities they belong to are rarely static. The classic SLA interlanguage model assumed clearly defined entities for L1 and L2; we all knew what these entities were. But, as Dynamic Systems Theory insists (DeBot et al. 2005) and attrition studies have shown (Schmid et al. 2004), language is rarely if ever still. Communities too are variable and flexible, adapting and changing continuously through macro-acquisition (Brutt-Griffler 2002). This section raises some issues about the changing languages and communities of L2 users.

First language change

- static first language

The first language may be static or changing. An individual may be an adult with a so‑called steady state of language knowledge, even if one accepts this is relative stasis rather than frozen. Only one’s vocabulary is believed to change appreciably during adult life. Similarly the language of a community may have the appearance of a static standard form. The English language is spoken of as if it has now achieved a fixed final form that brooks no change and to which everything else has been prologue. There is also the language frozen in time of some emigrant groups: an Italian-American actor-director who was interviewed on Sicilian Television used an Old Sicilian dialect that had to be translated for a modern Sicilian audience (http://www.italiansrus.com/articles/subs/hyphenated_italians_part2.htm). For some purposes in some cases a first language is static albeit in a highly idealised way. In most SLA research, stasis of the L1 is taken as the norm rather than seen as a moment when time stands still. The L1 is treated as fixed in the L2 user’s mind and in the community they belong to.

- developing first language

The individual may be a child developing their first language or, in the case of early bilinguals, first languages. Many L2 learners are not at an adult stage of development in their first language, particularly those in schools. The first language can also be developing in the L1 community, as in the case of creolisation where the group is inventing a new language from scratch, like the Nicaraguan sign language that sprang into being twenty years ago (Senghas et al. 2004). Like it or not, many, perhaps most, languages are developing in the individual and in the community. The L1 in many L2 users is not a constant static object that SLA research can take for granted and in some cases is being created in the L1 community.

- reducing first language

Alternatively a first language may be reducing, declining in some way. Individuals appear to lose some aspects of their first language, whether through lack of everyday use, brain injury or the effects of normal aging. In terms of the community, languages too may reduce. At one extreme there is the emotive issue of language death; languages may lose their last speakers, say Dyirbal speakers in Australia (Schmidt 1985). The languages may be temporarily suppressed like Min in Taiwan (Sandal et al. 2006) or Ulster Scots in Northern Ireland, now a recognised minority language in the European Union. It cannot be assumed that the L1 in the L2 user’s mind and L1 community is a constant.

The first language component of the multi-competence in the L2 user’s mind is not then necessarily static but developing in the case of children or reducing as in attrition. The individual’s first language, taken for granted in SLA research, is complex and shifting. The ‘L1’ construct is an abstraction, a snapshot of a moving target. The same is true of the language as the possession of a community. Notionally there may be a synchronic moment that isolates a state of a language from previous and future states – Modern English as spoken in 2006 – fleeting as this may be. The dynamic nature of the L1 community needs to be taken into account in SLA research.

Second language change

- static second language

The static L2 user is an individual with a putatively constant L2 knowledge, someone using the second language as part of their repertoire for their own purposes, say Roger Federer being interviewed in English or a doctor in Spain treating a Japanese patient in English, that is to say ordinary people anywhere happening to use another language than their first. The L2 user community may also be notionally static in that it involves a long-standing use of the two languages, such as the Polish/English community living in West London. For children it may be the micro-community of the bilingual family. This is distinct from the notion of ‘fossilisation’, with its negative connotations; people and communities reach a stable level of language for their purposes.

- developing second language

The developing L2 user is the classic figure studied by second language acquisition research – the L2 learner, who forms the subject matter of the vast majority of SLA research and hardly needs enlarging on here. Communities also develop second languages in many ways. One case is the Italian learnt by Spanish-speaking migrant workers in German-speaking Switzerland, a logical if surprising solution to working together (Schmid 1994). While the change in the individual is taken for granted, the change in the community may also be relevant to SLA research.

- reducing second language

L2 attrition is also prent in the individual. Many school learners retain rather little of their second language ten years later. Expats’ children returning to their home country may rapidly lose their other language (Kanno 2000). While ‘attrition’ is the usual term for this phenomenon, this involves a negative metaphor of invasion by the first language, which may well characterise some cases, not others, which are more driven by lack of use or other factors. In terms of L2 user communities, this reduction is most well-known as the familiar three-generation shift from first language to second language seen in many immigrant populations (Fishman 1991).

So the individual’s language knowledge may be notionally static, developing or reducing. The L2 community similarly stays the same or changes in various ways. The second language in SLA research models is as much a label for a mass of varying attributes as the first language. Till now the specialist area of SLA research that has concerned itself with language change in general is attrition. But attrition as language change is not so much an extra area of study as an integral part of any SLA model. The ideal situation of the unvarying L1 and L2 is seldom found in individuals or communities and should perhaps form an exceptional situation in SLA research rather than the norm.

To sum up this section, L2 users have a varied set of first languages and a varied set of second languages, whether static, developing or reducing. The habit of identifying the first language and second language as solid entities in SLA research belies their inherent variability and diversity. At some level it may indeed be necessary to reify the L1 and L2 into these highly abstract entities for our own research objectives but, like the Chomskyan definition of linguistic competence, we have to be careful not to throw the baby out with the bathwater. The classic interlanguage triad of L1, L2 and interlanguage ignored the variation within the constructs of L1 and L2.

Question 2. What is the language the L2 user knows?

The other construct that forms part of the classic SLA model is language. The issue of what language the L2 user knows depends on the meaning given to the word ‘language’, barely debated in SLA research, apart perhaps from discussions about language versus dialect, for instance Li Wei (2000), which we will not develop here (but see Cook (2006) for some discussion). Five meanings of the word ‘language’ will be distinguished here, not intended as final definitions, but reflecting some of the broad meanings that ‘language’ has within linguistics that are relevant to SLA research. Figure 1 summarises these five meanings for convenience.

 Lang1 human representation system

At some level human beings are different from other creatures because they possess a systematic representation system that allows an indefinite number of sentences to be spontaneously created within a shared context. Exactly where the differences between human language and animal communication systems lie is as controversial as ever,  currently over whether the distinctively human aspect of the Narrow Language Faculty is recursion, as asserted by Hauser et al. (2002), or whether this can be used by other species such as starlings (Gentner et al. 2006). Whether human language is primarily for communication, for taking part in a group (Malinowski 1926) or for organising the contents of the mind is similarly a bone of contention. Nevertheless in one way another a central meaning of ‘language’ is as a defining property of human beings.

Lang2 abstract external entity

Language is also a countable noun in English, as in ‘the English language’ or ‘the Chinese language’. There are discrete entities called ‘English’ and ‘Chinese’, codified in the rules of a grammar book and the entries of a dictionary and sometimes controlled through an institution such as the French Academy. Often this sense refers to a prestige ‘standard’ variety of the language spoken by a minority of people and jealously guarded against dialectal forms and historical shift, as witness the perpetual defence of standard English against the barbarians in books such as Truss (2003). Chiefly this standard is seen as the written language rather than the spoken; the exception is the question of accent, usually defined in terms of the status speakers of a class and regional variety of the language such as British Received Pronunciation (RP) rather than Geordie or Parisian French rather than Geneva French. English in this sense is no concern of the person in the street but belongs to the cultivated elite living in the capital city of an ex-colonial power. Yet no single person actually knows a language in this sense – the Oxford English Dictionary has some 650 thousand entries of which no speaker of English knows more than a fraction. While the institutional object of language bears some relationship to what people know it is more like that between the ideal model of driving laid down in the UK Highway Code and an individual’s behaviour driving to work in the morning. In some ways Lang2 represents the maximum that a speaker of a standard variety of a language could know, rather than the small amount that any actual individual knows or the variations in any individual’s speech due to age, region, class and all the other sociolinguistic variables.