Ann Fathman, confirmed our earlier results. Again, adult performers showed a difficulty order similar to that seen in children acquiring ESL. Again, there was no strong first language influence. Also, there was no difference in rank order between formal learners and informal acquirers, the former groups being those who reported having had a great deal of instruction in English and little real use, and the latter being those who reported a great deal of real world exposure to English but little or no classroom instruction. Our interpretation of this result was that both groups were dependent on the acquired system, since the test did not encourage conscious monitoring (Krashen, Sferlazza, Feldman, and Fathman, 1976).

At about this time, Roger Andersen (1976) reported a "natural order" for adult ESL students in Puerto Rico using compositions. We also undertook a composition study soon after (Krashen, Butler, Birnbaum, and Robertson, 1978): in our study, we asked ESL students to write under two conditions, "fast", in which they were told to write as much as possible in a short time (in the style of Brière's 1966 "Quantity before Quality" study), and "edited", in which they were encouraged to go over their work carefully. We found a natural order in both conditions, with only small evidence of a contribution from the conscious grammar in the edited condition. Our interpretation of these results was that students were concerned with communication when writing rather than with form; apparently the "focus on form" condition for Monitor use discussed earlier (see Introduction) is more crucial for bringing out the Monitor than is the "time" condition: our subjects did indeed have time, but they did not use it for the conscious grammar, for conscious monitoring. This is not to say they did not edit--it does imply that when they do edit, they do not use their conscious knowledge to any great extent when communication is the issue.

A more recent study confirms this, and suggests that it may take a very special kind of task to encourage subjects to use the Monitor. (This is not to say that we feel Monitor use in inherently good. We are interested here in the theoretical question of when people use it and when they do not.) Noel Houck, Judith Robertson, and I have just completed a study in which we asked USC ESL students to transcribe their own speech and then to correct their own transcripts. Both activities show a natural order for grammatical morphemes, despite

53