AUTOCAUSATIVITY AS A TYPE OF CAUSATIVE RELATIONS IN HUMAN–AI DIALOGUES

Abstract

The article offers a theoretical substantiation of autocausativity as an independent type of causative relations that emerge in human–computer assistant interaction (chatbots, voice assistants). In the classical model of causativity, the source of causation is an animate agent who, through his or her volition, initiates a change in the state of another person. In dialogues with artificial intelligence, however, a different configuration arises: a computer assistant’s utterance can modify the user’s emotional, cognitive, or behavioural state. To denote this phenomenon, the notion of CausatorCA (computer assistant as a causator) is introduced. Drawing on classical works on causativity and on contemporary research in functional grammar and emotive causativity, the mechanisms of this phenomenon are analysed. To explain why users address a machine as an animate being, A. V. Bondarko’s category of personhood is invoked: as an actualisation category, personhood does not distinguish whether the first- or second-person position is occupied by a human or by a program. Special attention is paid to the Russian Constructicon – a linguistic resource that allows filtering constructions by the semantic role of “Causator”. It is concluded that existing classifications of semantic roles need to be expanded to account for the CausatorCA phenomenon, and that the Constructicon’s tools are suitable for a systematic study of this phenomenon.