[本篇论文由上帝论文网为您收集整理，上帝论文网http://paper.5var.com将为您整理更多优秀的免费论文，谢谢您的支持]

Studies investigating the impact of non-manipulated overall instruction upon general proficiency may not provide a clear picture as to what acts as a causal factor in fostering learner's competence in second language. After all, classroom interaction involves a multiplicity of competing factors each affecting different aspects of learner's proficiency. In the midst of so many extraneous factors, one immediately finds himself in the difficult position of selecting the causal ones. For some, like Krashen, comprehensible input is the causal factor, while for others like Swain(1985), it is comprehensible output plus input. Swain claims that the concept of comprehensible input per se, is not enough to account for second language acquisition; comprehensible output should be included in any second language acquisition theory to better account for the acquisition process. According to Swain, language acquisition takes place when the learners realize how meaning is expressed accurately using their output as a means of hypothesis testing process. An L2 learner tests his hypotheses by trying them out in communicative situations. If his hypotheses prove to be successful in expressing his communicative intent then the hypotheses are confirmed; if not then they are revised and put into test again.

A strong version of Swain's output hypothesis cannot even account for first language acquisition. Brown and Hanlon 1970), for instance, have shown that parents do reinforce well-formed utterances of their children more than ill-formed ones. They found no significant correlation between parental approval and correctness of their children's utterances. Finally, it is concluded that output-based (dis)confirmations 'cannot be the forces causing the child to relinquish immature forms and adopt adult forms' and that a child revises his hypotheses not due to output disconfirmation but because of' the occasional mismatch between his theory of the structure of the language and the data he receives' (p.50), a finding which is quite in line with the predictions of the Input hypothesis. Similar results were reported by Hirsch-Pasek,Treiman and Schneiderman (1984), who replicated Brown and Hanlon's findings with a wider sample greater age range.