In this paper I examined the theory of feature percolation in detail. The feature percolation I examined is reminiscent of the notion of feature instantiation found in the theory of GPSG. Little attention has been given to the development of a general theory of how features combine and percolate. The purpose of this paper is to show that the feature percolation principles, when incorporated into the context of government binding theory, allow a unified treatment of a number of unrelated syntactic phenomena. I have discussed (a) sub-command facts. (b) differences in blocking effects of long distance reflexives, and (c) that-trace effects in English and other languages on the basis of the feature percolation principles. As a result, this general principle will eliminate a number of separate and arbitrary principles. The question that remains for further examination is to see whether this feature percolation principle can be extended to other syntactic features.

• 박만도