Linguistics and second language research

Some relationshipS between linguistics and Second Language research	Vivian Cook Obscure Writings SLA Topics
Vivian Cook
In K. Sajavaara and C. Fairweather (eds.), Approaches to Second Language Acquisition, Jyväskylä Cross-Language Studies 17, 7-13, 1996
This paper first looks at some of the historical relationships between Second Language research and linguistics, then surveys briefly some of the current relationships, and finally develops a multi-competence position in terms of current linguistics.

Let us first see how Second Language (SL) research relates to the linguistics of its day. In the 1950s SL research was indeed taken seriously by linguists in a way it has not been since; perhaps no major linguist since Roman Jakobson (1953) has made such a ringing declaration as 'Bilingualism is for me the fundamental problem of linguistics'. The European schools of Contrastive Analysis, surveyed for example in Sajavaara (1981), form part of a long scholarly linguistic tradition; the American school of Contrastive Analysis exemplified by Lado (1957) has perfectly respectable linguistic origins.

The work of Lado (1957) and Weinreich (1953) with second languages matched the overall structuralist paradigm laid down for linguistics by Bloomfield (1933). On the one hand this could be seen in the learning model employed; to Lado, speakers control habits that they use automatically to produce speech without thinking, acquired through exposure and practice. The learning model thus is a form of behaviourism relying on frequency of occurrence, and on imitation. On the other hand there is the concept of phrase structure through which the relationship between the L1 and the L2 can be stated. Their analysis is within the broad American structuralist tradition, which sees grammatical relations as morphology and phrase structure, symbolised in syntactic trees and the boxes of substitution tables. There were thus strong links between the second language ideas and contemporary linguistics. The clear survivors of this period are phrase structure grammar, perhaps the dominant concept of syntax in SL research to this day, not to mention language teaching, and the notions of interference and transfer so crucial to Lado's and Weinreich's thinking.

Moving into the 1960s, in linguistics there came the rebellion against all that the Bloomfieldians held dear. Language was no longer acquirable through habit-formation, phrase structure was no longer an adequate way of describing syntax. This revolution generated ideas in first language acquisition that gradually percolated through to SL research. One such idea is the 'independent grammars assumption'. This arose generally out of Chomskyan thinking, McNeill (1965) perhaps having been the first to state it explicitly. The independent grammars assumption insists that children must be treated wholistically; their language is not just deficient fragments of the adult grammar but has a grammar in its own right. At each age the child has a perfectly usable system of language knowledge that is peculiar to that stage - three-year-old-ese or four-year-old-ese - however different it may be from an adult system. Children have importance in their own right, not just as defective adults: before you can begin to relate caterpillars to butterflies, you have to describe a caterpillar as a caterpillar, not as a butterfly that hasn't learnt to fly. The learners' sentences such as "Him go shop" are not just deviant versions of a native sentence: they reveal the grammar of the child at that stage. So linguists tried to describe the child's language in terms of its own independent grammar, Brown (1973) and Klima and Bellugi (1967) providing prime examples of grammars of children's speech.

The independent grammars assumption crept into SL research around 1970. Nemser (1971) called it the 'approximative system'; 'Learner speech at a given time is the patterned product of a linguistic system, La [approximative language], distinct from Ls [source language] and Lt [target language] and internally structured'. Pit Corder (1971) with 'transitional competence' and Selinker (1972) with 'interlanguage' expressed variations of the same idea with different emphases. The independent grammars assumption crossed into SL research from the mainstream of linguistics; it became a firm belief that the L2 learner's language should be treated as a system in its own right.

A second idea that filtered through into SL research from 1960s linguistics was the Language Acquisition Device (LAD). Chomsky (1964) expressed the problem of language acquisition as a metaphoric black box called LAD into which went sentences, and out of which came knowledge of language. LAD saw language acquisition as the active creation of language knowledge by the child's mind; it is a process in which a highly complex state of knowledge is created by the mind out of the sentences that are heard. LAD too became an integral part of our mental paraphernalia, partly through its stress on the creation of language knowledge by the mind, as emphasised by Selinker in the original formulation of interlanguage, or through the idea of creative construction espoused by Dulay and Burt (1973), partly surfacing explicitly within the hypotheses of Krashen (1985). Thus the first era of SL research in the 1970s was massively influenced by the linguistics of the previous decade.

Sadly linguistics and SL research have to a large extent parted company since those days. The ideas already described have continued to play their part; phrase structure grammar is the backbone of most SLA work on sequences of acquisition; lip-service is still paid to LAD and the independent grammars assumption. Habit-formation is perhaps missing except for work that comes within the bounds of psychology rather than linguistics. But later ideas from linguistics are few and far between.

What areas of linguistics have been utilised in later SL research? One important area is the work of creolists; the comparison of Second Language Acquisition with the formation of creoles and pidgins by Schumann (1978), Andersen (1983), and Bickerton (1981), has provided stimulating accounts of SL development.

A second area is sociolinguistics. The example familiar from every student essay is communicative competence, taken from Hymes (1972). Also important was Labov's account of language variation from one task to another (Labov, 1966), which was the seed for the theories of Tarone (1983) and others developed in the 1980s. Other areas of the study of variation such as style and register and genre analysis have not however exerted much influence.

Thirdly, phonology for long was a neglected area in SL research; scanning the indexes in general introductions produces no page references for 'phonology', 'phonetics', or indeed 'pronunciation' in Larsen-Freeman and Long (1991) and two references to 'phonology' alone in Ellis (1985). While some interesting work took place in classic phoneme-based theory, such as Wieden & Nemser (1991), most L2 work was at a phonetic end concerned with the minutiae of Voice Onset Time rather than with contemporary phonological theory; some of this is described in Cook (1991b). Now, with the regular conference New Sounds in Amsterdam (Leather and James, 1990), more recent ideas are beginning to yield results, going beyond the phoneme to non-linear phonologies of various kinds (Durand, 1990).

But where is everything else? Morphology in SL research still seems to mean the grammatical morphemes of Brown (1973), rather than any systematic linguistic treatment on modern lines, using the kind of areas covered in Spencer (1991) or those currently being studied in L1 acquisition reported in Goodluck (1991) for instance. Vocabulary research still mostly relies on the theories of psychologists rather than those of linguists or indeed of those doing first language acquisition; it is again almost totally absent as a topic from Larsen-Freeman and Long (1991) and gets a mere three page references in Ellis (1985). It is encouraging to see the large-scale research on vocabulary acquisition by Viberg reported elsewhere in this volume. Syntax is not much better off. There are allusions to Halliday (1985) in the L2 literature, sometimes in the same breath as Hymes, but no actual L2 research. GPSG and LF grammar both have their advocates in linguistics, and LF has indeed been used for first language acquisition (Pinker, 1984): neither have been explored in SL research on any scale. Syntax in SL research still largely means the phrase structure grammar of an earlier era of linguistics. With certain exceptions, there is still little contact between SL research and the ordinary syntax studied in first year linguistics courses.

Obviously a list of gaps can go on for ever and is always open to the charge of ignorance of the relevant literature, which may be buried in journals outside the L2 area proper. Suffice to say, of the thousand odd references in Larsen-Freeman and Long (1991), about 32 belong to mainstream 1980s linguistics, 14 of those being due to Bickerton, Chomsky, and Givon. SL research is in danger of cutting itself off from other disciplines to do with language without even exploring whether their current manifestations are important to it.

The glaring omission from this list is Universal Grammar and its associated theory of syntax, now usually known as principles and parameters theory. Some SL research has indeed become tied in to UG; each development in the syntactic theory during the 1980s was followed by a veritable spate of research investigating it in SL learners - pro-drop in the early days (White, 1986), binding in the mid-period (Cook, 1990), more recently the structure of functional phrases (Vainikka and Young-Scholten, 1991). One can hardly complain that the mammoth 1980s industry on UG and Second Language Acquisition, reviewed for example in White (1989), is not linked to linguistics. But the complaint may still be made that UG has had to do too much by default; as a theory it is limited to claims about the core aspects of syntax - a handful of principles and parameters - rather than embracing the whole of SL research. However interesting within its own area, UG-related work is in danger of going into a world of its own, dealing with ever more esoteric syntax that changes faster and faster and dealing with a more and more limited set of research questions such as whether L2 learners have 'access' to UG. SL research needs a greater variety of input from linguistics than UG theory alone. This argument is developed more fully in Cook (to appear).

Let us now turn to a more general consideration of the relationship between SL research and linguistics. A starting point can be Chomsky's list of the three main questions that linguistics should try to answer (Chomsky, 1986):
1) What constitutes knowledge of language? The goal is to discover the reality of language in the mind: what does a human being know who knows English, say? The linguist aims at describing the language contents of the human mind in whatever terms are appropriate, describing the full computational system that relates the physical reality of the actual sounds of speech to the cognitive representation of their meaning. The study of language knowledge in this sense has led to a massive explosion of work in the past ten years.
2) How is knowledge of language acquired? A mind must be able to learn any possible human language, not just English or Finnish, say. The process of acquisition enables and constrains the knowledge that is acquired; the study of language knowledge cannot be separated from the study of acquisition. This goal has led on the one hand to syntactic descriptions that take in acquisition, on the other to accounts of first language acquisition that take in recent linguistic descriptions, for instance the work of Radford (1990) and Atkinson (1992).
3) How is knowledge of language put to use? This acquired knowledge is used for sociological and psychological ends, whether to communicate ideas to people, to propose marriage to someone, to keep one's diary, to try to save the blue whale, or to achieve the innumerable other functions that human language may have.

The difference between the goals of SL research and the Chomskyan goals seems then quite simple: SL research is concerned with people who know two or more languages, not just one. Let us explore the consequences of making the word "language" plural:
i) What constitutes knowledge of languages?
If you know two languages, you know two grammars; one mind possesses two systems of language knowledge. The goal of SL research is therefore to describe grammars of more than one language existing in the same person.
ii) How is knowledge of languages acquired?
If you know two languages, you have been through the acquisition process twice. SL research needs to explain how the mind can acquire two grammars and discover whether first language acquisition differs from second language acquisition, or whether they are different aspects of a single process.
iii) How is knowledge of languages put to use?
If you know two languages, you can decide when to use one or the other according to the situation you find yourself in and the people you are talking to. To achieve this goal, SL research must show how the same speaker utilises knowledge of two languages for a range of functions.

Thus, making the three goals of linguistics multilingual forces SL research to answer the knowledge question by describing the knowledge of languages of the second language speaker and its differences and similarities from that of a monolingual speaker, to answer the acquisition question by seeing how this complex state of knowledge originates, and to answer the use question by examining how knowledge of both languages is put to use. In a sense these three goals could not be denied by L2 researchers: the subject matter of the discipline is indeed how minds know more than one language, how they acquire more than one language, and how they can use more than one language. Some version of these goals should not be controversial for L2 research; there is a clear relationship between the goals of linguistics and the goals of SL research.

But accepting the first goal as the description of the L2 learner's grammar means starting from a state of the learner's mind. The knowledge of the second language has to be looked at as an independent grammar in its own right before going on to the monolingual L2 grammar. In some ways this goal reiterates the independent grammars assumption - the priority of the learner's own system, his or her interlanguage. Interpreted in UG terms, the question is a matter of how the L2 grammar exploits the possibilities inherent in human language, or ignores them. Then, as a second step, the learner's interlanguage may be compared with the target L2 as spoken by monolinguals. Bringing in the target L2 too soon may warp our analysis of the learner's own grammar towards the idiosyncrasies of the L2 rather than seeing it as a possible human language that has to be discussed as a thing of its own. Much UG research has fallen into this trap by measuring access to UG against the L2 rather than against the possibilities available in human language itself.

However, this rephrasing of the knowledge question can be taken further. The first question defines the goal as the knowledge of languages. But there isn't a word for knowledge of languages; a monolingual has linguistic competence; an L2 learner has interlanguage; no term encompasses the composite knowledge of both an L1 and an L2 in the same mind. Hence there is a a need for the term "multi-competence" to refer to one mind's knowledge of two languages (Cook, 1991a). The knowledge question must treat the L2 as a part of a more complex system that includes the L1 rather than in isolation; interlanguage is part of this double-facetted knowledge. Research may go on to find out what distinguishes this multicompetent state from the monolingual one, whether there are differences in knowledge of the L1 or knowledge of the L2, and how these two grammars get along with each other. We could look for example on the one hand at the problems that Italian learners have with putting subjects into all English sentences, on the other at the problems that Italians that know English have with leaving them out in Italian. Interlanguage does not stand alone by definition but is always part of a mind where another language coexists, a part of multi-competence.

The acquisition question is how the mind acquires knowledge of two languages: how does this multi-competent state of mind come into being? As Sridhar & Sridhar (1986) point out, 'Paradoxical as it may seem, SL researchers seem to have neglected the fact that the goal of SLA is bilingualism'. It is not enough to look just at the interlanguage; it is not how a second grammar pops out of the acquisition device that is interesting - that is a version of imitation monolingual acquisition - but how this double system evolves. The concept of interlanguage has treated the L2 as separate, as if it was just another L1. Hence we have heard time and again how bad second language learners are: they are seen as pale and ineffectual shadows of those marvellous monolinguals. But they are not; they are something different, the proud possessors of multi-competence, a means by which one person can know and use more than one language. Astronomers can look at stars in isolation; a double star system is a different system rather than being two single star systems. Multi-competence extends the learners' knowledge beyond that of a single language rather than simply being a degraded version of monolingual knowledge.

In conclusion, at one level linguistics can provide tools of analysis and ways of thinking about language that SL research can and should take advantage of. If nothing more, it is clearly beneficial to both disciplines that a channel of communication be kept open so that current research in both disciplines remain in touch. At another level linguistics can provide models and theories of language and language acquisition that can be adapted or tested in the SL arena. Here the relationship to SL research becomes closer; SL research takes part in the same endeavour as first language acquisition, sociolinguistics, phonology, etc, as one of the sciences of language. Whether the adaptation of Chomsky's goals to SL research is accepted or not, the goals of SL research are different from those of linguistics and should be different.