The design of a lexical difficulty filter for language learning on the Internet

The authors describe the design and implementation of a tool called the Lexical Difficulty Filter (LDF) intended to help learners or teachers in selecting sentences appropriate to the level of students for English vocabulary learning. The LDF serves as a bridge between learner and information resource bank. It selects suitable sentences from the output of standard English corpus concordancers. The core technology in the design of LDF is a fuzzy expert system. The results obtained are very encouraging in this pioneering work. We achieve 94.33% accuracy rate in imitating the judgments of a human expert in determining the degree of difficulty of a sentence for a given learner level.