Abstract: The ambiguity in syntactic structure is a hard stone in natural language parsing. This paper deals with the structure ambiguity problem by using a combined probability evaluation function, which employs the sentence structure features and context information during English parsing. Post parsing correction is implemented on the basis of rule based extensible frame, which possesses the feature of a self evaluation mechanism. These two evaluation methods have been provided. The combined probability evaluation function has provided an extensible statistic model for better parsing of English sentences, which is also applicable to other languages. The experiment shows a promising result of precision and recall.