Some Novel Applications of Explanation-Based Learning to Parsing Lexicalized Tree-Adjoining Grammars