Pen-based computing has not yet taken off, partly because of the lack of fast and easy text input methods. The situation is even worse for people using East Asian languages, where thousands of characters are used and handwriting recognition is extremely difficult. In this paper, we propose a new fast text input method for pen-based computers, where text is not composed by entering characters one by one, but by selecting words from a menu of candidates created by filtering the dictionary and predicting from context. Using our approach, users can enter Japanese text more than twice as fast as recognition-based and other existing text input methods. User studies and detailed analysis of the method are also given.