【正文】 ive to language teachers in that teachers can consciously apply the predictive power to the making of teaching plan, and hence, the solving of language difficulties.. Linguistic Aspect of Contrastive Analysis and Its LimitationsThe origins of CA were pedagogic. It includes a linguistic aspect and a psychological aspect. The linguistic aspect of CA, in the first place at least, is based on structuralist linguistics. Most of the contrastive studies carried out have been based on surface structure characteristics, such as those described by the structuralists. By paring language items, contrastive analysts soon realized that there were degrees of similarity and difference. Here are some of the possibilities that a parative study might reveal:1). No difference between a feature of the first and secondlanguage. For example, the contracted form J`ai in French is mirrored by the contracted form I39。ve in English.2). Convergent phenomena(. two items in the first languagebee coalesced into one in the L2).3). An item in the first language is absent in the target language.For example, in German, subordinate clauses require a different word order from main clauses, whereas in English the word order is the same in both clause types.4). An item in the first language has a different distribution from the equivalent item in the target language. For example, in many African languages [? ] occurs word initially, but in English it only occurs word medially or finally (. singer or thing).5). No similarity between first language feature and target language feature. For example, in Spanish, negation is preverbal (no se)。 whereas in English it is postverbal (I don39。t know). In addition, English negation involves the use of the auxiliary system, whereas Spanish negation does not.6). Divergent phenomena (. one item in the first language bees two items in the target language). For example, where the L2is French, English the diverges into le and la.It is one thing to develop categories, such as those mentioned above, for classifying the ways in which two languages differ. It is quiteanother, however, to relate these linguistic differences to learning difficulty. Stockwell, Bowen and Martin (1965), and Prator (1967) have proposed that linguistic differences can be arranged in a hierarchy of difficulty. Prator, for example, suggests that the above categories be ordered from zero to greatest difficulty. However, this claim is based neither on a psycholinguistic theory which explains why some differences create more learning difficulty than others do, nor on empirical research. It is based only on the conviction that the degree of linguistic difference corresponds to the degree of learning difficulty. Evidences prove that it is problematic. In practice, there will be some items in a second language which present greater difficulty than others do, even though they belong theoretically to the same level of difference. Whitman and Jackson (1972:40) conclude from their study of errors produced by Japanese learners of English, …relative similarity, rather than difference, is directly related to levels of difficulty. Lee (1968) reported that when he started to learn what was for him an exotic language (Chinese), he experienced very little L1 interference: his explanation was that his L1 and Chinese were so far apart that he was lifted into a new orbit of noninterference. That is, in this new orbit, there would have been no positive transfer, nor did negative transfer potential seem to exist. Therefore, we may conclude that difference and difficulty are not identical concepts. The former derives from linguistic description and the latter from psychological processes, and there is no reason to believe that they should correlate with each other in a reliable way.. Psychological Aspect of Contrastive Analysis and Its LimitationsThe psychological aspect of CA is based on behaviorist learning theory. The behaviorist approach claims that we can predict difficultiesand errors by means of contrastive analysis, on the basis of an analysis of two related linguistic systems.In his preface, Lado (1957) states that, The plan of the book rests on the assumption that we can predict and describe the patterns [of L2] that will cause difficulty in learning and those that will not cause difficulty. Oller (1971:79) again speaks of CA as ...a device for predicting points of difficulty and some of the errors that learners will make. Closest to Lado39。s view is the psychological one that CAs identify the conditions conductive to two kinds of transfer, positive and negative. Going one step forward, since negative transfer is normally manifest in errors, it is claimed that CAs predict error. And finally, since errors signal inadequate learning, there is the conclusion that CAs predict difficulty.In practice, the claim has not been strongly supported by the evidence. Teachers have found that errors predicted by contrastiveanalysis have often not occurred, whereas many actual errors would not have been predicted. Therefore, it concludes that whichever analysis is used, CA is of little use in predicting the items which are proved difficult. This CA Hypothesis whose primary aim is to reliably predict language difficulty and errors is termed strong version by Ronald Wardhaugh (1970). In accordance with the strong version, Wardhaugh proposed the weak version of CA Hypothesis, which cla ims merely to have the power to identify or diagnose errors that have been mitted, that is to say, after we have observed what errors learners actually make, CA can help to explain some of these errors, namely, those which are due to transfer. Wardhaugh (1970) suggested that CA Hypothesis is only tenable in its weak or diagnostic function, and not tenable as a predictor of error, …the weak version requires of the linguist only that he use the best linguistic knowledge available