Abstract
According to White (1989), since the Subset Principle may not effectively work in L2 acquisition, a parameter with multiple values with three or more values, which meets the Subset Condition, may need negative evidence, while a binary parameter with two values may be reset through positive evidence. This paper attempts to see if White's theory would work in an experimental-classroom teaching. I used two language properties: the first is the reflexive which has a multiple parameter with three values and meets the Subset Condition, and the other is the relative clause which has a binary parameter with two values. The subjects were 32 students from a private junior high school. They were divided into two groups; a positive-input and a negative-input group. Three sessions were held and three grammatical judgment tests were undertaken: pre-test, post-test 1, and post-test 2. The mean score of the reflexive for the positive-input group was higher than that of the relative clause, which does not seem to support White's theory. On the other hand, the result of the negative-input group was significantly better than that of the positive-input group in both language properties, which renders a support for her theory.
