"Errors", claims Olsson (1974) "constitute a way of learning." One problem is to define errors. George (1972) defines an error as "an unwanted form--specifically, a form which a particular course designer or teacher does not want." Johansson (1975) believes that "If native speakers hesitate about the acceptability of a word or construction it should not be considered an error." Tchalo (1987) makes a distinction between error and mistake. "A mistake is an inadvertant [sic] 'slip of the tongue or pen', whereas an error is a systematic (i.e. 'rule-governed') recurring 'mistake'. An error then is an extension and application of a rule beyond its domain" (p. 5).
How does one identify errors? In a paper presented at the American Language Institute, University of Southern California, Paul Schachter (1974) expressed his belief that, whereas previous syntactic contrastive analysis had emphasized differences, one should not overlook similarities between languages. In the interpretation of a study of Farsi-speaking and Arabic-speaking students, he found that the speakers felt comfortable using the fairly similar restrictive relative clauses construction and used it with about the same frequency as English speakers, but made errors because they had not learned to observe some minor ways in which the English construction differed from their own. Schachter was of the opinion that both Contrastive Analysis and Error Analysis could identify and explain certain data... but not all data.
A more militant stance had been adopted before by Newmark (1966) who had argued that CA was irrelevant and unnecessary; that errors which seemed to be caused by interference from the native language merely represented a gap in the learner's knowledge of the target language. The learner made up for this deficiency by substituting something he already knew (i.e. his native language), thereby producing incorrect structures in the target language.
A less extreme po...