The paper presents the technique for analysis of text emotional nature which is a key characteristic of Mass media news text. Emotions inherent design its Emotional coloring and become a significant feature of mass media news texts. The technique proposed measures the degree of exposure of emotions and allocates them by rating. Emotional coloring is defined by emotional characteristics and by grammar categories, and a set of rules is applied to regulate wordforms interaction. Techniques for verbal units analysis are examined. The Heavy Natural Language Processing models and Machine learning techniques are considered. They are compared and the optimum one is defined to resolve the problem of Emotional coloring evaluation. A system prototype is developed on the basis of this technique. It allocates news by influence rating according to their key parameters. The examples of texts’ emotional nature recognition results by means of the prototype are presented. The visualization of emotional nature analysis results highlights additional features of the news text’s emotional nature and expresses them in numeric values. It is exposed both by sentences and by the whole news text, with tracking of news Emotional coloring dynamics. The results presented have application in analysis procedure intending to studying Mass media, particularly informational environment with concomitant factors, and their impact on political and social interrelation.