Objective: This article introduces the Russian Constructicon—an open-access electronic database documenting over 2,200 fixed syntactic constructions of the Russian language—as a novel linguistic resource for the analysis of causative semantics in human-computer dialogue. The study presents the theoretical foundations, structure, and functionality of the Constructicon, demonstrating how it systematically catalogues stable syntactic patterns with their semantic, syntactic, and pragmatic properties. The aim is to show that fixed syntactic constructions recorded in the Russian Constructicon can serve as an effective tool for identifying hidden causative meanings in this new type of discourse.
Method: The study examines the Russian Constructicon's architecture, semantic classification system (53 semantic tags including Assessment, Attitude, and Intensity), and search functionality. The empirical material consists of an author's corpus of 1,000 Russian-language dialogues extracted from public online sources (VK, 4pda.ru, ixbt.com). The study employs contextual analysis, functional-semantic analysis, and comparison of empirical data with the semantic construction types identified in the Russian Constructicon.
Results: The analysis reveals that users interacting with AI regularly reproduce three types of fixed constructions documented in the Constructicon. First, constructions with negative evaluation and fatigue semantics ("хватит с меня") account for 15.7% of all causative utterances, where users demand the assistant stop repetitive or irrelevant behavior. Second, constructions expressing prohibition through contrast ("не умничай, а ответь") constitute 12.7%, where users attempt to regulate the manner of the assistant's response by rejecting anthropomorphic imitation. Third, constructions denoting abrupt change of state ("меня прорвало") represent 8.0%, where users describe loss of emotional control caused by assistant failures. The combined share of these three construction types constitutes 36.4% of all causative utterances in the corpus. Notably, a systematic search of the Russian Constructicon for terms related to human-computer interaction reveals that the word "диалог" ("dialogue") yields no results, and related terms yield only isolated constructions (e.g., "компьютер"—7 constructions, "общение"—1 construction, "помощник"—1 construction). This gap confirms that the Constructicon, while comprehensive in documenting constructions of the standard language, has not yet systematically captured constructions specific to digital discourse.
Conclusions: The article argues for the expansion of the Constructicon to include constructions specific to human-computer interaction. The constructicographic approach enables the transition from analyzing individual utterances to systematically describing typical user reactions through the "inventory" of stable constructions. The findings demonstrate that communication with AI is not chaotic but relies on codified linguistic patterns transferred from interpersonal communication. The concept of "covert grammar" [Katsnel'son 1972] explains the conflict between anthropomorphic grammatical form (imperative, 2nd person) and machine ontology. The term "quasi-causativity" is proposed to capture this tension, demonstrating that the phenomenon is not a marginal failure but a significant marker of technological anthropomorphism requiring consideration in dialogue interface design.