The Contrastive Analysis Hypothesis derives from the school of psychology known as Behaviorism, which is based on the idea that learning is a matter of either positive or negative reinforcement contributing to the development of new habits. According to the Contrastive Analysis Hypothesis, learners of a second language have the most trouble when they come across some feature of the new language that's different from their native language, and have the least trouble with aspects of the new language that are relatively familiar. If the two languages are compared and analyzed in detail, it should be possible to accurately predict the specific errors the student is likely to make. This is called Error Analysis.
The Contrastive Analysis Hypothesis was initially very influential. During the height of behavioral psychology in the 1960s, studies seemed to support the hypothesis. However, supporters of Contrastive Analysis claim that it has predictive value, and this has proven not to be the case. The Error Analysis method used by supporters of the hypothesis often predicted that students would have difficulty with aspects of the second language that turned out not to be problematic for them. It also failed to accurately predict the aspects that did turn out to be difficult.
As behavioral psychology became less influential in the 1970s, linguists began to doubt that there was an easy and logical way to compare two languages, without which Contrastive Analysis isn't even possible. Because the underlying logic and structure of each language is different to at least some degree, it's unclear that some aspect of one language can be directly equated to a similar aspect of another language in order to compare and contrast them. Without a universally valid system for categorizing inguistic structures, Contrastive Analysis requires that too many differences be ignored.
The Contrastive Analysis Hypothesis is based on the assumption that most problems in language acquisition are caused by the unfamiliar. However, the familiar can cause just as many problems. French and English share many "cognates" or related words, and this similarity can cause the learner to think there is a connection between two words that actually have different meanings. For instance, an English speaker learning French might assume that the French word "travail" must mean either "travel" or "travail," but it actually means "work." Contrastive Analysis doesn't predict and can't account for this type of problem.