adult performer: we have interpreted the presence of the "natural order" (the child's order of acquisition or difficulty order) in the adult performer as a manifestation of the acquired system without substantial interruption or contribution from the conscious grammar, or Monitor.

As we shall see, out current results dovetail nicely not only with out intuitions about the use of conscious rules but also with speculations and results from other areas of second language acquisition research.

Grammatical morpheme studies with adults began with our (Bailey, Madden, and Krashen, 1974) finding that adult second language acquirers (ESL students at Queens College) showed a "natural order" for eight grammatical morphemes, a difficulty order similar to that found in child second language acquirers by Dulay and Burt, with oral language elicited by the Bilingual Syntax Measure (Burt, Dulay, and Hernandez, 1975). We also reported no difference in rank order between Spanish-speakers and non-Spanish-speakers, which was consistent with Dulay and Burt's finding of no first language influence in their child second language study (Chinese-speakers and Spanish-speakers; Dulay and Burt, 1974a). We also noted that our difficulty order was similar to the order found by de Villiers (1974) for adult agrammatics.

Following our study, Larsen-Freeman (1975) reported a "natural order", again cross-sectionally, for both BSM elicited speech and an imitation task. Her own tests, which included "reading", "writing", and "listening", did not show a natural order, and I speculated (Krashen, 1976a, 1977a) that this was due to the intrusion of the conscious grammar, which caused an elevation in the accuracy of those items that were "easiest" to learn (third person singular morpheme, regular past morpheme). When performance is "Monitor-free" (little time, focus on communication and not form) we thus see the "natural order" for grammatical morphemes, a difficulty order similar to that seen in the child. When performance is "monitored", the natural order is disturbed. The BSM, it was suggested, is such a Monitor-free test, while Larsen-Freeman's pencil and paper "grammar-type" tests invite the use of the Monitor. We return to Larsen-Freeman's results below.

Our study using the SLOPE test, an oral production test designed by

52