Lang3 a set of sentences

Lang3 language means a set of produced sentences – all the actual or potential sentences that could be said or written. This sense has recurred throughout linguistics, starting with Bloomfield – a language is ‘the totality of utterances that can be made in a speech-community’ (Bloomfield 1927/1957: 26) down to Chomsky –  ‘a set (finite or infinite) of sentences, each finite in length and constructed out of a finite set of elements’ (Chomsky 1957: 13). A language is the sum of all the sentences its speakers have said or, in Chomskyan vein, could say. Language as a set of sentences is studied by corpus linguistics, whose task is to define the properties in a specific collections of texts, carefully chosen in advance. And it seems to be the sense in which language is internalised in usage-based learning and emergentism (Tomasello 1998): language emerges from the array of language data that the learner encounters, extracted from a corpus. In this sense language is not an abstraction but a concrete object, made up of physical sounds, gestures or written symbols. Patterns can be extracted from these primary data, both by the linguist and the learner. But they remain patterns of data rather than systems of knowledge or behaviour.

Lang4 shared possession of a community

The Lang4 sense of language treats it as a social phenomenon, a shared cultural product – ‘the English-speaking world’, ‘native speakers of Chinese’ etc. A language belongs to a particular human group and confers identity as a member of that group; according to Smolicz et al. (2003), one of the shared core values of a community is its language. It is tempting to equate the community with national boundaries – Japanese speakers tend to live in Japan –  but language communities pay little attention to political borders – Chinese is used all over the world. Nor is it necessary to have a country to have a language – versions of Romani are spoken in most European countries by tens of thousands to hundred thousands of people. A language in the Lang4 sense is possessed by everybody who can understand each other, setting aside the difficulties over language versus dialect.

Lang5 mental knowledge system

The final Lang5 sense of ‘language’ crucial to most linguistics is language as the mental possession of an individual – ‘a language is a state of the faculty of language, an I‑language, in technical usage’ (Chomsky 2005: 2). Language is not only out there in the world but inside the mind. A person who knows English in the Lang5 sense can connect the world outside to the concepts inside their minds in a particular way. The problem is how this sense corresponds to the abstract entity called English in the Lang2 sense or to the set of English sentences people encounter in the Lang3 sense. The mental knowledge of a grammatical rule is not the same as the rule in the Lang2 grammar book or as the same as the patterns in a Lang3 set of sentences.

The Chomskyan solution is to call what is in the mind a ‘grammar’, not a ‘language’: 'The grammar in a person's mind/brain is real; it is one of the real things in the world. The language (whatever that may be) is not' (Chomsky 1982: p.5). Hence the word ‘grammar’ is ‘systematically ambiguous’ between the Lang2 and Lang5 senses. Language in the mind is an epiphenomenon, a side-effect, rather than the real thing. What the speaker knows is a state of their mind and does not necessarily correspond to any of the actual languages of the world in a Lang3 sense, only to the possible schemata laid down in the Universal Grammar.

The other major quandary is the relationship between Lang5 mental knowledge and Lang3 set of sentences, usually phrased in terms of competence and performance. Discussions of linguistic competence usually point out: (a) studying the sentences produced in the past is looking at accidental creations rather than the potential sentences created by the mental language system; (b) many mental rules of grammar are not derivable from the properties of sets of sentence, as the vain hunt for discovery procedures showed (Harris 1951; Chomsky 1957), though similar to the path taken by emergentism (Tomasello 1998): the relationship between Lang3 and Lang5 is as murky as it has ever been. As is indeed the relationship between social Lang4 and mental Lang5, which continues to attract sniping from entrenched views on both sides, generative linguists insisting on the purity of their Lang5 accounts of competence, sociolinguists on the complex realities of Lang4 interaction between people. The Lang4 community and Lang5 mental senses are inextricably entwined, as de Saussure pointed out early in the last century, ‘le langage a un côté individuel et un côté social, et l’on ne peut concevoir l’un sans l’autre’ (de Saussure 1976: 24): language is two-sided as being both individual knowledge and collective possession.

So people have ‘language’ in the sense they speak a Lang1 human language, which apes and dolphins do not; they have some relationship with a Lang2 abstract entity called the English language, etc, which speakers of other languages do not; they produce a Lang3 set of sentences labelled English rather than French; they are members of various Lang4 communities of English speakers that exclude, say, French speakers; and they have a mental Lang5 system of knowledge/processes, an English grammar, that differs from a mental grammar for French etc.

 ‘language’ and SLA research

Let us now try to link these five senses of the word ‘language’ to SLA research. Obviously once again this is a first attempt rather than the last word.

 Lang1 applied to the second language

In the Lang1 sense of a human representation system, adding a second language, say Chinese, to the repertoire of an individual who speaks English makes no difference: Chinese is another human language, another representation system. Occasionally SLA research claims that second languages are not learnt as languages at all but as some other type of knowledge (Clahsen & Muysken 1989). This would then put second languages in the same bracket as artificial computer languages like Prolog, which human beings can obviously learn and use but do not have the characteristics of human language. The Lang1 sense is not relevant to SLA research unless we deny second languages are human languages.

Lang2 applied to the second language

In the Lang2 sense of an institutional entity, an L2 user can be linked to two standard varieties of a language say, say British RP in the L1 and Northern Chinese Mandarin in the L2. But the knowledge of the second language in the L2 user’s mind in Lang5 is as remote from the abstract entity of Lang2 in the second language as in the first. If these two senses are not kept separate, language teaching and indeed SLA research may measure the L2 user against this Lang2 entity and find them wanting. Hardly surprising as these Lang2 entities do not represent the Lang5 linguistic competence of any individual. SLA research has tended to assume that the native speaker essentially has a perfect command of this abstract Lang2 entity and has compared this with the faltering steps of the L2 user (Cook 1979). If it is at all necessary to compare the languages of the monolingual native speaker and the L2 user, the same meaning of ‘language’ needs to be used for both, whether Lang5 mental knowledge with Lang5 or Lang3 set of sentences with Lang3.

Lang3 applied to the second language

In the Lang3 sense, the second language is another set of sentences. We will grant that it is possible to describe the input that the L2 user has received as a Lang3 set of L2 sentences. The problem is the L2 set that the L2 user produces. Selinker called interlanguage ‘the utterances which are produced when the learner attempts to say sentences of a TL [target language]’ (Selinker, 1972). It would be convenient if there were a set of L2 sentences: Weinreich (1953: 7) said that ‘A structuralist theory of communication which distinguishes between speech and language ... necessarily assumes that “every speech event belongs to a definite language”’. But code-switching research has shown that L2 users’ sentences can in effect belong to both languages simultaneously, say the matrix model of Myers-Scotton (2002). It is hard, if not impossible, to decide that some sentences of the L2 user belong to one language, some to another, without bringing in criteria from other senses of ‘language’; which sentences, say, use the word order of Lang2 English, which Lang2 Chinese?

The Lang3 starting point has to be the set of all sentences the L2 user produces, not just those assigned to a second language using a criterion from another sense. This demands an analysis of the whole, not an arbitrary division of sentences into languages A and B. There has been an increase in the use of learner corpora such as the Seidlhofer (2002) and Granger (2003) projects, which will hopefully start producing results. But corpora-based studies that concentrate solely on the second language may miss half the picture. One of the revelations of SLA research in the past few years has been the influence of the second language on the first (Cook 2003). A full account of the L2 user’s actual and potential sentences means looking at everything and not assuming that the first language can be taken for granted as if it were identical to that of a monolingual. Duncan (1989) argued that bilingual speech therapy should be based on the child’s first language as well as their second language; you can’t see what’s wrong with either if you don’t look at both (Stow & Dodd 2003). The same applies to SLA research. Studying the second language without the first is missing the unique feature of second language acquisition, namely the presence of a first language.

Lang4 applied to the second language

In the Lang4 sense ‘shared possession of a community’, much SLA research is crucially concerned with how people gain membership of a community along with the identity that comes with it, whether this community is in their present situation or their future plans, an abstract imagined community or a concrete reality. Language community and identity are basic to second language acquisition in terms both of the community the learners start out from and of the wider community they end up in.

But is there an L2 user community different from the monolingual native speaker community? According to Chomsky ‘A community with more than one language, or indeed more than one dialect, would not be homogenous: the language of a mixed community: 'would not be "pure" in the relevant sense, because it would not represent a single set of choices among the options permitted by UG but rather would include "contradictory" choices for certain of these options’ (Chomsky 1986: 17). Classically SLA research tacitly adopted this Chomskyan view of the homogenous monolingual community. L2 learners were assumed to want to belong to the community of native speakers; passing for a native speaker was a crucial test issue, which virtually all of them failed. Passing or a native speaker became a shibboleth for L2 research whether in discussions of the availability of Universal Grammar (Cook & Newson 2007) or of the age factor: ‘Those studies cited for phonology have shown that some learners can achieve very high levels of native-like pronunciation in mostly constrained tasks but have yet to show that later learners can achieve the same level of phonology as native speakers in production’ (de Keyser & Larson Hall 2005: 96): the only thing that proves L2 users have UG or that they are affected by age is whether or not they speak like natives. The concept of community will be returned to in the next section.

Lang5 applied to the second language

The Lang5 mental sense of ‘language’ for second language acquisition research means that the two languages are present in the same mind, i.e. multi-competence. In this broad sense all second language acquisition involves multi-competence. However, the two languages seem to be regarded by researchers as sharing the same mind more or less by accident. We study the first language or we study the second; we search for different locations for languages in the brain as if there were separate pigeonholes; we compare how good people are at one language and how bad they are at the other, usually to the detriment of the second language. At some level, however, the mind is a whole; the question for SLA research is where, if at all, it divides into different languages, according to the possibilities in the integration continuum (Cook 2003), and how it keeps them separate when necessary (Lambert 1990). As with Lang3, looking at the mental system of the second language and excluding the first ignores the basic premise of L2 acquisition that two languages are involved.

Second language acquisition research has again to take on board that the Lang5 language in the L2 user’s mind has as much and as little connection to the abstract entity of Lang2 as the information in the Highway Code has to someone’s driving. A Lang5 mental system within the speaker’s mind is not an external Lang2 institutional entity. Nor is the Lang5 linguistic competence of individuals the same as their Lang3 performance even if once again there is a relationship of some type. Extrapolating from sentences that people have said to what they know is as problematic as it has always been.

So what does ‘an L2 learner of English’ mean or ‘acquiring L2 English syntax’ or ‘speaking L2 English’? In the SLA literature these have frequently collocated with words like ‘fail’ and ‘lack of success’. The first three senses seem to have little connection with success: L2 users do not succeed at Lang1 human language because they have it already; they do not succeed in learning a Lang2 entity because nobody has done or ever could; they do not produce a ‘better’ or ‘worse’ Lang3 set of sentences, but neutral patterns. Success may be a legitimate question to ask in terms of Lang4 and Lang5. For Lang4, L2 users are gaining membership of an L2‑using community and we can ask whether they are succeeding or not; it is uninteresting to ask whether they succeed in passing as members of a monolingual native speaker community as this denies their distinct status; it could only be valid for those who wish to deny or disguise their origins, say spies and terrorists. In the Lang5 sense, L2 users are gaining another language system in the mind, a grammar in Chomsky’s sense, not a language. We have to ask at what level one mind has a single grammar with two subsystems say or two distinct grammars. Once we acknowledge the arbitrariness of calling the systems of the mind in the Lang5 mental sense by the name of Lang2 objects like English or Chinese, the language system in the L2 user’s mind can be explored as a whole, no longer counting languages.