sition was taken by Celce-Muria (1974) who, discussing Contrastive Analysis and Error Analysis in perspective, believed that there were "a number of shortcomings and limitations in the CA hypothesis... First of all, CA is relevant only in those language learning situations where the students all speak the same native language. Second, its results are more reliable on the phonological level than on the syntactic or lexical level. Finally, certain kinds of difference between two languages seem to cause greater difficulty than others; this in turn has prompted the concept of 'hierarchy of difficulty" (p. 78).
The variety of opinions regarding errors may stem from a lack of clear definitions of "errors" as much as from the variety of sources of errors, of individual learning styles and metacognitive backgrounds, and of learning contexts.
Richards (1971) takes a non-contrastive approach to error analysis. He sees errors as traceable to transfer and interference. He classifies errors as being interlingual, intralingual, and developmental. Interlingual errors are those caused by interference of the learner's mother tongue. They are errors "which reflect the general characteristic of rule learning, such as faulty generalization, incomplete application of rules, and failures to learn conditions under which the rules apply" (p. 207). Intralingual errors are those originating within the structure of English itself. Such errors are developmental, usually temporary, and susceptible to correction. "They arise by a misinterpretation of the rules of the target language... The errors, as made by second language learners, are often similar to the errors made by children acquiring the language of L1... They are usually morphological or syntactical" (Elliott 1983, p. 136).
Earlier, Lado (1957) had come to the conclusion that interlingual errors were inevitable and resulted from interference. Richards (1975) pointed out that errors are not due just...