This paper discusses linguistic and psychological aspects of the problem of automatic modusdictum analysis of texts published in social networks and other electronic media. Thereupon, theoretical questions are raised anew on the linguistic nature of modus, on the means to express “ego-meanings” in speech, on the differentiation of proper modus (autoreferential signs) and modal-evaluating predicates in dictum position, on the implicit methods of communicating modus information, and the resources to read this information based on discursive speech practices (conventional meanings). The applied goal of the paper is to provide “humanitarian” (psychological and linguistic) support for development of machine “mining” programs, i. e. automatic monitoring of network content and text identification with a certain subjective modality. To achieve this goal, we describe, in particular, such lexical-grammatical features of the texts that can be significant for determining psychological state of an individual or a professional group to identify certain public opinions. Conceptually, this research is connected with the idea of speech system which is manifested both at the level of styles and genres and within independent communicative units, as well as with one of the most important trends in the field of artificial intelligence — the method of relational-situational analysis of texts in natural language. Thematic groups of words (TGW) were compiled including “evaluation collocations” typical of those texts. The templates created on the basis of psychological and linguistic description model suggested in this paper can be used hereafter to develop algorithms for automatic monitoring of the network texts of a given theme (professional stability or mobility, professional crisis, etc.) and evaluation.

* Работа выполнена при частичной поддержке РФФИ (гранты № 17-07-00651 «Разработка моделей и методов конструирования сценариев поведения на основе анализа текстов» и № 18-29-22027 «Персональные когнитивные ассистенты, сопровождающие деятельность человека в информационном пространстве»).

The goal of this paper is to convey the meaning of the category of speech system for computer cognitive modeling. By implementing it, the authors prove the importance of supplementing relational-situational analysis, based linguistically on the concept of functional syntax, with the technology of creating templates which are the formalization of those sections of stylisticspeech system that mark the mental processes under study. A review of works on speech system since the 1920s is given. It is shown that in recent decades, when studying speech system, the primary interest of linguists lies in the processes of verbal communication and in its main unit — a text. It is noted that the quantitative aspects of the language functioning in various areas and communicative situations are being examined not only in quantitative linguistics but also in the works on artificial intelligence. We prove the hypothesis that the given approach to the automatic analysis of texts is realized in all cases of social communicative practice, when to express the certain affective states or cognitive actions, we can find relatively stable ways of choosing and using linguistic means. From psychological point of view, this is the area of compressed internal actions, which, if necessary, can be expanded and expressed in the form of a verbal reaction or a verbal report. It is a broad research field covering the typologies underlying the known psychological inventories or recorded in descriptions of human behavior in various spheres of activity. The paper demonstrates the resources to analyze stylistic-speech system for the automatic search of professional crisis signs in social networks. The extensive material presented by the fragments of 86 network discussions describes the patterns of choice and use of multi-level language units to express the typical emotional states of currently experienced professional crisis. The linguistic markers of the mentioned states are considered in detail.

* Работа выполнена при поддержке РФФИ (грант № 18-00-00233 комфи «Методы комплексного интеллектуального анализа информации различных типов для социогуманитарных исследований в социальных медиа»).

The authors of this article present the results of the psycho-linguistic stage of studies which aim to develop instruments for automated identification of online communication texts according to a set dictum and modus parameters. The intentional, volitional, and emotional content of the texts was analyzed which determines their genre and linguo-semiotic peculiarities. As a result of an introspective analysis of the content of texts on the topic of a professional crisis, a number of communicative-textual categories were identified that are significant from the point of view of diagnosing the psychological state of the subject of speech. A description is given of both segment markers of subjective meanings — predicates that directly reflect the intentional-volitional state of the speaker and text fragments that, not being direct modus indicators, with a high degree of probability refer in this context to a certain internal state of the author text. The well-known discursive conventions are taken into account. Methodologically, the authors are guided by the theoretical guidelines of the concept of stylistic-speech consistency (M. N. Kozhina) and the model of relational-situational analysis developed within the framework of the theory of artificial intelligence by G. S. Osipov. When analyzing the modus-dictum content of texts, a scenario approach and methods of “internal and external reconstruction” of subjective meanings were used, interpreted with reference to both segment markers within the analyzed text unit (utterance), and their connections with such adjacent (external) text fragments, which, in the context of a network discussion, refer to certain intentional-volitional states of the subject of speech and are, in fact, their semantic anaphors. It is shown that the content of texts on the subject of a professional crisis is characterized by a high degree of emotiveness, the automatic “reading” of which is an important task of computer linguistics and artificial intelligence.