Traditional phylogenetic tree are build based on the alignment of fragment of annotated sequences of the interested genomes. In this paper, we will present an alternative algorithm of building phylogenetic tree based on the characteristic 'signature' word usage in genomes analogous to the natural language processing. Here, we apply statistical N-gram analysis on whole genome sequences of several organisms. We calculated the occurrences of difference N-grams and the contrast value and departure values, which are the deviations of N-grams occurrences from their expectations, for every chromosome of 28 organisms. It could be shown that a few particular genome N-grams are found in abundance in one organism but occurring very rarely in other organisms, there by serving as genome signatures. Later, we consolidate the signature information on each organism to a features vector that consists of the average of the contrast value and departure values for 2-gram, 3-gram, 4-gram and 5-gram. From the features vector, we build the phylogenetic tree using correlations as the similarity measures, we could reproduce the taxonomy tree of 28 organisms. [ABSTRACT FROM AUTHOR]