The article is devoted of problem definition of effective ways detection of lie, which are used for analysis in verbal components. Theoretical analysis described main ways of detection. They are verbal and nonverbal components. We have defined that nonverbal components were most criticism and are not enough reliable for lie detection, also authors of these method thought so. We have formulated the hypothesis that structure of lie massage can be change due to violation of textual cohesion during its making. During lie speaking a mental image will be different, because an author have to dream up text now, in contrast a true message. True image is in memory of individua. These factors less depend by emotional conditions, as nonverbal components, because they base on cognitive component of person. In the psycholinguistic researches we have highlighted main verbal features of lie, which were grouped in four classes. They consist of speech errors (reservation, word repetitions, word corrections), special speech constructions (contextual inserts, slang, nonpersonal constructions, description of other’s actions, unusual details), change of general structure of text (logical violence, thematical transmissions, avoid of theme), change of dualization (increase or decrease details, generalization, using direct speech. synonym). With the aim of check for hypothesis we have realized an experiment, which were analyzed true and lie messages of people. As result we found that these criteria can be used as diagnostical criteria, but they have different reliable level. In lie massages nonpersonal construction are used more often, make word correction, dualization (increase and decrease number of details). There are big violence of logic, tendential to decrease of text volume. After correlation we defined two types of liar. They are extraverted and introverted. Which have differences in communication skills, level of logic and detailed.