We propose a semantic tagger that provides high level concept information for phrases based on several kinds of low level information about words in clinical narrative texts. It delineates such information from the statements written by doctors in patient records. The tagging, based on Hidden Markov Model (HMM), is performed on the text that have been tagged with Unified Medical Language System (UMLS), Part-of-Speech (POS), abbreviation tags, and numeric tags. It reuses UMLS, POS, abbreviation, clue words, and numerical information to produce higher level concept information. Our unknown phrase guessing method for a robust tagger also uses the existing information calculated in the training data. In short, the semantic tagger gives more meaningful and abstract information by integrating different kinds of low-level information. The result can be used to extract clinical knowledge that can support decision making or quality assurance of medical treatment.