A filtered syntactical description is generated by removing the located elements.
This in mentioned way found syntactical and lexical information is used for creating a model of the speech.