The paper studies analytical tools and procedures which can supplement the psycholinguistic expertise of polycode and polymodal conflictogenic texts. The author introduces the procedure of “syncrisis”, new for this type of expertise (specialty: “investigation of speech products”), and describes its conduct algorithm in detail. The paper focuses on the problems of speech as a specific form of mental activity and “translation” of the psycho - state into discourse. Today, emotional computing and visual information processing constitute a fairly large segment of the academic space, and the systems of detecting and recognizing states are rapidly developing. And it becomes possible to analyze behavioral models, physiological indicators, as well as changes in person's mood. Programs used to identify the manifestations of the data from sphere can be divided into classes based on the type of response: facial, motor, and verbal. The mimic service interprets facial microexpressions, classifying them by basic emotions: joy, anger, sadness, surprise, fear, disgust, and neutral state. The ranking and classification of basic emotions occur through neural networks trained on a large experimental corpus of photos. The author also considers subconscious formal grammatical indicators that are not controlled by consciousness and the quantitative characteristics of which may have stable correlations with personal human behavior.