study, however, this alternative explanation is untestable. Nevertheless, it may be true.

Mason (1971) is also interpretable in these two ways. In this study, certain foreign students at the University of Hawaii were allowed to follow regular academic programs without extra ESL, despite the fact that their English placement scores indicated that they should be enrolled in English for foreign student classes. Post-tests given at the end of the semester showed no significant difference in increase in English proficiency between those excused from ESL and controls who took the required ESL classes. These data are again consistent with hypothesis I, but other explanations are available, including the "self-study" hypothesis described above, which is consistent with hypothesis II.

Carroll (1967) studied the second language proficiency of American college seniors majoring in foreign languages (French, German, Russian, Spanish). About 25 per cent of the total population of senior language majors that year (N = 2784) were given form A of the MLA Foreign Language Proficiency test in their language. Carroll's major finding was that, on the average, foreign language majors performed rather poorly: the median score on the MLA corresponded to a Foreign Service Institute rating of 2 plus (out of 5) (between "limited working proficiency" and "minimum professional proficiency"). Of more interest here is the relation found between attainment^* and measures of time spent in different linguistic environments. A strong relationship was found between time spent abroad (in the country where the target language was spoken) and test performance, with those who reported a year's study abroad doing best, followed by those who reported a summer abroad or a tour. Both of these groups outperformed those who had never been in the country where the target language was spoken.

A significant relationship was also found between test performance and the extent to which the target language was used in the students' home. (Native speakers of the language majored in were excluded from the study.) Those reporting frequent parental use of the target

42