PURPOSE: To discriminate even such words that are not easily discriminated by collection of character strings in a simple procedure and with high precision and to improve the correcting efficiency of documents in an interactive form by discriminating a desired word based on the morphological analysis result and generating the identification information.

CONSTITUTION: A natural language processor consists of a morphological analysis means 101 which performs the morphological analysis processing to the supplied document information, and a document correcting means 103 which applies the interactive document correction processing to the document information 102 that undergone the morphological analysis. A specific expression identification information generating means 105 discriminates the expression area related to the word of a specific type in a document and generates the specific expression identification information 104 to show the expression area. The means 103 includes a pending expression area specifying means 106 which specifies a pending expression area out of the discriminated expression area, an input means 108 which fetches the correction instruction input 107, and a correcting means 109 which corrects the pending expression area based on the input 107. Thus the treating of such a word that has large difference of expressions can be improved in the natural language processing.