6.2.1　lnterlanguage theory

At various periods second language theorists have offered contrasting and conflicting views of the concept of knowledge of language.In the early 1950s，most scholars would have assumed that knowing a language involved knowing a number of items and their potential arrangements；this item and arrangement grammar was just starting to be challenged by a notion of items and processes（rules）.The second language learner was seen as having an imperfect knowledge of the items and arrangements of the language he or she was learning；the gaps waited to be filled by learning.

The major innovation of the American school of contrastive analysis，following the work of Charles Fries and Robert Lado，was to propose that the gaps were generally filled，and some of the correct knowledge was replaced or confused，by the grammar of the learner's first language.Thus，a second language learner's knowledge of a second language was subject to interference from his or her knowledge of the first.

While this notion is still held to be important，the most widespread set of views at the present time has a somewhat different approach.Though，as will be seen，not constituting an organized and consistent body of theory，the most common approach is probably that loosely grouped under the rubric‘interlanguage’，a term that in its various uses can perhaps be best glossed as‘a second language learner's knowledge of target language’.Ellis（1985）regularly replaces it by the admittedly awkward but more accurate term‘language-learner language’.

The notion of interlanguage first appeared in the mid-1960s.One of its principal contributions was its underlying claim that the learner's knowledge is to be seen as a unified whole，in which new knowledge is integrated and systematically reorganized with previous knowledge of the native language.In this view，there are not gaps waiting to be filled by the first language，or even two competing language systems struggling for control，but rather the second language learner's knowledge is a complete whole，critically different from that of the first language learner.The principle may be summed up as a necessary condition：

（1）Conditions for second language learning

The earliest formulation of the notion was that proposed by S.Pit Corder（1967）（for example，McNeill，1966）which was starting to show how the study of children's developing linguistic competence could be enhanced by using as evidence the regular occurrence of errors that could not be the result of imitation and therefore must demonstrate the development of rules or generalizations.Following this lead and applying it to second language learning，Corder（1976）suggested that the study of errors made by second language learners could in the same way provide evidence of their developing systematization of the language they were acquiring.In his now classic paper，precisely titled‘The significance of learners'errors’，Corder suggested that errors of a systematic kind provided evidence of the nature of transitional competence，that is to say，the grammar or set of rules a learner used in producing sentences in the language he was learning.In a somewhat different model，but in general agreement，a paper by Nemser（1971）spoke of approximative systems—the term reflects the notion of successive approximations of target behavior used at the time in work in programmed instruction.

Corder's proposal was particularly important，for it shifted attention from the target language，or the target language and the native language，to the learner's own performance，and established this performance as a suitable object of research.The study of second language learning could thus move from the library to the classroom.

Selinker picked up Corder's idea in thinking about the learner's performance as an object of study，and tentatively tried out his own term：an“interlanguage”may be linguistically described using as data the observable output resulting from a speaker's attempt to produce a foreign norm，i.e.both his errors and his non-errors.It is assumed that such behavior is‘highly structured’（Selinker，1969：71）.The implication of this earliest formulation is that‘interlanguage’is a performance phenomenon，to be seen in the behavior of second language learners attempting to emulate the target language speaker's norm or competence：the underlying structure is to be derived from the‘observable output’.In Selinker's next article（1971），the term‘interlanguage’lost its inverted commas and moved into the title，since when it has been the focus of many papers and the central concept holding together the‘SLA’school.There has been criticism of the work in the field；a review by Kohn（1982）maintains that：‘Today，however，much of the enthusiasm of the early days has died down，and it is no exaggeration to claim that the main achievement so far has been in accumulating new speculations instead of turning old ones into hard knowledge’.A number of scholars have suggested explanations for this disappointment：

（a）the misuse of concepts related to the target language（the comparative fallacy）（Bialystok，1985）.

（b）the failure to deal with variability（Dickerson，1974）.

（c）the concentration on morpho-syntactic development and the failure to deal with semantic development（Ellis，1982）.

（d）the failure to recognize clearly the specific features that distinguish interlanguages from other natural languages（Adjemian 1976）.

（e）the failure to define the concept clearly（Bialystok 1985）.

For Selinker（1971），interlanguage has a different‘psychological infrastructure’from normal language.But it is far from being clear what the term interlanguage is meant to convey.Essentially，it would seem to make a claim that just as the structuralist explanation for the speech behavior of people in their native tongue is an abstraction called language，the structuralist explanation for the speech behavior of people in a second language is an abstraction called interlanguage.Thus，interlanguage is to be compared to language in the sense of‘langue’or grammar.The implication of this is clearest if one follows the Saussurean concept whereby language is an abstraction underlying the specific speech behavior of members of a community，or the Chomskyan axiom that a grammar represents the knowledge of the idealized monolingual in a homogeneous speech community.In this sense，interlanguage would represent the knowledge of the idealized speaker in a homogeneous community of second language speakers.In other words，it is a claim that second language speakers share a system that is different from the system of native speakers.

（2）Second language learner language approximates native speaker language

This is stated as a graded condition because the closeness of approximation will vary；it is a typical condition because there are cases where the learner may not in fact aim at native speaker language（Coppieters，1987，raises the interesting possibility that native-speaker performance may be based on different underlying competence.）.This may be because of the kind of speech to which the learner is exposed；the teacher，for instance，may not be a native speaker，or may be distorting the language he is using into the kind of foreigner talk that leads to the development of pidgins.Another reason may be connected to the learner's lack of integrative motivation and consequent unwillingness to adapt all the features of native speech，particularly a native speaker accent.At the same time，the condition does not deny the possibility of finding evidence for systematic developments that do not follow native speaker goals；a study such as Huebner（1985）shows the kind of evidence that is revealing.

Selinker，Swain，and Dumas seem to say that to count as interlanguage all one has to find is evidence of similar strategies.But this misses a point that has been central to linguistics at least since Saussure：the social nature of a language.This was recognized by Corder earlier when he called the approximative system an‘idiosyncratic dialect’；until these strategies evolve into a socially recognized variety，they have no more claim to the status of language than any other set of personal characteristics of speech.From this point of view，there is no such thing as an interlanguage.There are approximative systems，transitional competences，evolving or almost static results of attempts by a second language learner to understand the system of the target language.Groups of learners with similar language backgrounds and similar language learning experience have temporarily similar systems；similarly，learners isolated from native speakers of the language tend to develop‘fossilized’systems or pidgins.But there is some danger in using a term for these idiosyncratic transitory systems that suggests the kind of stability implied by calling them a language.

When the variety of a language used by people speaking it as a second or foreign language achieves some kind of social status，it seems appropriate to give it its own label.‘Second Language Acquisition’or like‘language-learner language’used by Ellis（1985）is probably to be preferred.

The full implication of the issue of variability for interlanguage was first noticed by Tarone：the interlanguage is governed by the rules of any other natural language，then it should turn out to have the same characteristics of variability as in other natural languages.Thus，five axioms were proposed by Labov（1969）.Deletion，and inherent variability will apply in general to studies of second language learning.There will be style-shifting，but in the case of a language learner，not all the socially appropriate shifts.These shifts will reflect the amount of attention paid to speech.As a result，there is special value in gathering evidence of the vernacular in informal usage.The observer's paradox here is that it is easiest to collect evidence in a situation where speech，because it is being observed，is not natural.

Littlewood（1981）has discussed the several kinds of variation that one might expect to find in a second language learner's speech.Variation within an individual learner is likely to result from changes in communicative function（redundant features will be omitted），changes in linguistic environment（as Dickerson shows，the learning of a phonetic rule involves slowly learning new environments to apply it），and changes in socio-situational factors（i.e.the changes between classroom and out of classroom performance）.Littlewood speculates that the mechanisms involving first and second language acquisition are similar.He proposes，however，that second language learners have not only social norms，but also pedagogic norms，most evident in school situations.In separating the pedagogic norms from the social，he is drawing attention to the social artificiality of the normal classroom.He applies his theory to types of learners：the natural learner with low integrative motivation，the student with low integrative motivation to whom formal instruction is given，and the student aiming at social acceptance.

The variationists like Tarone are also clearly much concerned with discovery procedures.According to Tarone，who assumes that the goal of research is to describe the‘grammatical and phonological system which underlies learner performance’，（1982：70）the best data to use are intuitional data supplemented by data gathered in actual meaningful communicative use：the former will give only a limited picture of the learner's system，for it will be a limited view of the target language.Tarone does not believe that these varying kinds of evidence show the existence of the same underlying linguistic norm，but rather that they show the existence of several linguistic norms，a continuum of styles ranging from a‘superordinate’style produced when the speaker pays the most attention to form，to a vernacular style produced when the least attention is paid.As far as one can see from available evidence，the vernacular style is the most impermeable to other influences.

So，in these few studies where evidence is available on both the vernacular and the careful styles of second language learners，the evidence seems to show that the superordinate style is characterized by the presence of more target language variants，more socially marked native language variants，and（in some case）aberrant forms...The evidence of style-shifting of this kind is quite sparse，but the evidence now available is consistent with the hypothesis that the IL（interlanguage）superordinate norm is more permeable to the target language and native language rule systems...（Tarone，1982：77）

Tarone points out the significant difference between her view of language knowledge and that implied by Krashen's Monitor Model.The Monitor Model assumes two distinct systems，one derived from learning and the other from acquisition；the variability model postulates a continuum，and so one system.The variability model，to the extent that it is concerned with competence rather than performance，is neutral as to conscious versus unconscious processes.And variability assumes influence from the native as well as the target system when the attention level is high.