INTRODUCTION
　　In this chapter, I am going to discuss the significance of evaluating Error Analysis (EA) studies in 1970s and early 1980s and set the proper perspective toward the use of learner corpora in analysing learner language errors in order to better understand the process and sequence of acquisition of English as a second/foreign language
　　Before 1960s, when the behaviouristic viewpoint of language learning was prevailing, learner errors were considered something undesirable and to be avoided. It is because in behaviourists perspectives, people learn by responding to external stimuli and receiving proper reinforcement. A proper habit is being formed by reinforcement, hence learning takes place. Therefore, errors were considered to be a wrong response to the stimulus, which should be corrected immediately after they were made. Unless corrected properly, the error became a habit and a wrong behavioural pattern would stick in your mind.
　　This viewpoint of learning influenced greatly the language classroom, where teachers concentrated on the mimicry and memorisation of target forms and tried to instill the correct patterns of the form into learners' mind. If learners made any mistake while repeating words, phrases or sentences,? the teacher corrected their mistakes immediately. Errors were regarded as something you should avoid and making an error was considered to be fatal to proper language learning processes.
　　This belief of learning was eventually discarded by the well-known radically different perspective proposed by N. Chomsky (1957). He wrote in his paper against B.F. Skinner, that human learning, especially language acquisition, cannot be explained by simply starting off with a "tabula rasa" state of mind. He claimed that human beings must have a certain kind of innate capacity which can guide you through a vast number of sentence generation possibilities and have a child acquire a grammar of that language until the age of five or six with almost no exception. He called this capacity "Universal Grammar" and claimed that it is this very human faculty that linguistics aims to pursue.
　　This swing-back of pendulum toward a rationalistic view of language ability lead many language teachers to discredit the behaviouristic language learning style and emphasize cognitive-code learning approach. Hence, learners were encouraged to work on more conscious grammar exercises based on certain rules and deductive learning began to be focused again. This application of new linguistic insights, however, did not bear much fruit since Chomsky himself commented that a linguistic theory of the kind he pursued had little to offer for actual language learning or teaching (Chomksy 1966) .?
　　In the school of applied linguistics, however, this shift towards the innate human capacity raised a growing interest in the learner's powers of hypothesis formation as he moves towards the bilingual competence sufficient for his communicative needs. One major result of this shift of attention was an increasing concern in the monitoring and analysis of learner language. The concepts of 'interlanguage' and 'approximative system' presented challenging areas of descriptive enquiry.
　　In 1970s and early 80s, a large number of papers on error analysis were published throughout the world. However, it lost its attention and enthusiasm gradually as more and more criticism was made against the approach and method of error analysis. As the present writer makes an attempt to analyse learner language oncomputer, it is essential to review the previous work of error analysis and identify what it aimed to achieve and how it failed. Otherwise, it could be just a repetition of what was already done a decade ago and not very much meaningful. Error analysis using learner corpora must be significantly different from traditional error analysis, in quality and quantity. I would like to show the readers whether that is really the case.
I will first review some classic articles on error analysis by Corder, Selinker, Richards, among others and try to establish what the original purpose of error analysis was like or what it inteded to do. Then I will describe the criticisms against error analysis in 80s and early 90s and summarise what traditional error analysis failed to offer. Next, I will introduce recent research results on the role of negative evidence in language learning and data-driven learning in order to show the effectiveness of giving feedback to learners about their common errors in a new language learning perspective. Finally, I would like to make a systematic comparison between traditional error analysis and "learner corpus-based" error analysis so that hopefully I can convince the readers of the powers of learner corpora in systematic investigation of learner language.
　　Error Analysis (EA): its roots and development
　　Larsen-Freeman and Long (1991) claims that the study of SLA can be said to have passed through a series of phases defined by the modes of inquiry researchers have utilized in their work: contrastive analysis, error analysis, performance analysis and discourse analysis (p.81). As we look into the roots and development of error analysis, let us first overview contrastive analysis so as to gain better insight into how error analysis became more popular among SLA researchers.
　　Contrastive Analysis
　　Before the SLA field as we know it today was establised, from the 1940s to the 1960s, contrastive analyses were conducted, in which two languages were systematically compared. Researchers at that time were motivated by the prospect of being able to identify points of similarity and difference between native languages (NLs) and target languages (TLs). There was a strong belief that a more effective pedagogy would result when these were taken into consideration. Charles Fries, one of the leading applied linguists of the day, said: "The most efficient materials are those that are based upon a scientific description of the language to be learned, carefully compared witha parallel description of the native language of the learner."(Fries 1945: 9)
　　Robert Lado, Fries' colleague at the University of Michigan, also expressed the importance of contrastive analysis in language teaching material design:
　　Individuals tend to transfer the forms and meanings and the distribution of forms and meanings of their native language and culture to the foreign language and culture - both productively when attempting to speak the language and to act in the culture and receptively? when attempting to grasp and understand the language and the culture as practised by natives. (Lado 1957, in Larsen-Freeman & Long 1991:52-53)
　　This claim is still quite appealing to anyone who has attempted to learn or teach a foreign language. We encounter so many examples of the interfering effects of our NLs.
　　Lado went on to say a more controversial position, however, when he claimed that "those elements that are similar to his native language will be simple for him, and those elements that are different will be difficult" (Lado 1957:2). This conviction that linguistic differences could be used to predict learning difficulty produced the notion of the contrastive analysis hypothesis (CAH): "Where two languages were similar, positive transfer would occur; where they were different, negative transfer, or interference, would result." (Larsen-Freeman & Long 1991: 53)
　　Corder (1967): Introduction of the Concept 'Error Analysis'
It was S.P. Corder who first advocated in ELT/applied linguistics community the importance of errors in language learning process. In Corder (1967), he mentions the paradigm shift in linguistics from a behaviouristic view of language to a more rationalistic view and claims that in language teaching one noticeable effect is to shift the emphasis away from teaching towards a study of learning. He emphasises great potential for applying new hypotheses about how languages are learned in L1 to the learning of a second language.? He says "Within this context the study of errors takes on a new importance and will I believe contribute to a verification or rejection of the new hypothesis." (in Richards 1974:.21)
　　Corder goes on to say that in L1 acquisition we interpret child's 'incorrect' utterances as being evidence that he is in the process of acquiring language and that for those who attempt to describe his knowledge of the language at any point in its development, it is the 'errors' which provide the important evidence.(ibid.: 23) In second language acquisition, Corder proposed as a working hypothesis that some of the strategies adopted by the learner of a second language are substantially the same as those by which a first language is acquired. (It does not mean, however, the course or sequence of learning is the same in L1 and L2.) By classifying the errors that learners made, researchers could learn a great deal about the SLA process by inferring the strategies that second language learners were adopting. It is in this Corder's seminal paper that he adds to our thinking by discussing the function of errors for the learners themselves. For learners themselves, errors are 'indispensable,' since the making of errors can be regarded as a divice the learner uses in order to learn. (Selinker 1992: 150)
　　Selinker (1992) pointed out the two highly significant contributions that Corder made: "that the errors of a learner, whether adult or child, are (a) not random, but are in fact systematic, and are (b) not 'negative' or 'interfering' in any way with learning a TL but are, on the contrary, a necessary positive factor, indicative of testing hypotheses. (ibid:151) Such contribution in Corder (1967) began to provide a framework for the study of adult learner lanugage. Along with the influence of studies in L1 acquisition and concepts provided by Contrastive Analysis (especially language transfer) and by the interlanguage hypothesis (e.g. fossilization, backsliding, langauge transfer, communication and learning strategies), this paper provided the impetus for many SLA empirical studies. EA in 1970s: popularity and criticism