The "self-study" reinterpretation of Upshur's and Mason's results, as well as Carroll's "year abroad" and "home use" data, remains a plausible, but difficult to test, explanation.

In the following section, it is proposed that modifications of hypotheses I and II are correct: formal and informal environments contribute to second language competence in different ways or, rather, to different aspects of second language competence. To support this hypothesis, we will reinterpret the literature surveyed here in terms of the acquisition-learning distinction. In addition, we will consider some cross-sectional data that eliminate certain alternative explanations presented in the literature survey and confirm several hypotheses presented above.

Contributions of Formal and Informal Environments

It is not simply the case that informal environments provide the necessary input for acquisition while the classroom aids in increasing learned competence. The reinterpretation of the Krashen et al. series as well as the Friedlander et al. data described above suggests, first of all, that informal environments must be intensive and involve the learner directly in order to be effective. One might then distinguish "exposure-type" informal environments and "intake-type" environments. Only the latter provide true input to the language acquisition device. Second, it seems plausible that the classroom can accomplish both learning and acquisition simultaneously. While classwork is directly aimed at increasing conscious linguistic knowledge of the target language, to the extent that the target language is used realistically, to that extent will acquisition occur. In other words, the classroom may serve as an "intake" informal environment as well as a formal linguistic environment.

Both of these points are illustrated and confirmed by new data on proficiency and linguistic environment using the SLOPE test with adult learners of English. The subject pool was the same as used in Krashen, Sferlazza, Feldman, and Fathman (1976): sixty-six subjects were tested, with thirteen first language groups being represented. Some had studied English intensively while others had encountered English only in informal environments. Table 2 shows the relationship

47