A natural language processing apparatus includes a result acquisition unit that acquires a plurality of analysis results indicating parts of speech of morphemes contained in one or more common sentences from a plurality of types of morphological analyzers, a pattern acquisition unit that detects a common segmentation point in the plurality of analysis results, extracts one or more parts of speech corresponding to a character string segmented at the common segmentation point from each of the analysis results, and acquires a set of the parts of speech as a part-of-speech differing pattern, and a candidate specifying unit that extracts the part-of-speech differing pattern with the number of appearances being equal to or less than a predetermined threshold and specifies the character string corresponding to the extracted part-of-speech differing pattern as a character string containing a candidate for an unknown word.