UDC 004.032.26+81.23
student of the 3rd year of the day-time form of education of the Foreign Languages Faculty of the EE «BrSU named after A. Pushkin», Brest
Research supervisor –M. P. Kontsevoy, senior lecturer of the Department of German Philology and Linguodidactics BrSU named after A. Pushkin
Linguistic foundations of the problemsof subjectivity of virtual interlocutors
The mass production of LLM texts in various literary genres is the most important problem facing modern language education [2]. This poses a threat to diminish the value of human authors and creators, as well as the innovation of language and literature. The purpose of the article is to provide grammatical tools for their appropriate discussion in the context of scientific discourse based on the psycholinguistic foundations of LLM personification.
The Large Language Model (LLM) is a powered by artificial intelligence model that is capable of generating natural language texts from large amounts of data. LLM creates books’ texts [3], articles, diplomas, textual illustrations, and word and expression visualizations.
Fundamental psycholinguistic mechanisms serve as the foundation for how people perceive and evaluate of LLM as a true language subject. The human psyche is projective, which psychologically causes the loss of the ability to tell a real person apart from a virtual interlocutor in a dialogue partner. The perception of the internal as coming from the outside is specifically caused by this. The psyche’s contents are projected onto communication and other exterior reality. The person is likely to think that his or her interlocutors are who he or she thinks they are. The thoughts, emotions, and experiences that individuals have are spontaneously and unconsciously transferred to speech and communication agents.
The characteristics of nominative languages, which fundamentally oppose the active subject and the passive object, are at the core of the mechanisms of personification, projection, and transference. This opposition produces the complete verbal representation of the world. This is conveyed linguistically in the metaphoricity of language as its universal. Although metaphor is the most significant heuristic tool of cognition (including scientific one), it can be misleading when used carelessly. There is a chance that the LLM scientific discourse will slip into a similar nominative metaphoricity trap [2].
The idea that the LLM is a brand-new subject of social reality cannot be regarded as sufficient or unique. Thinking about nominative metaphoricity gives rise to the potential of seeing that this linguistic metamodel makes a semantic error by transferring the object of consideration from one category to another. In this instance, the LLM has been moved from the category of instrument to that of subject (author, personality, individual, identity), leading to the assignment of traits that it neither possesses nor is capable of possessing. After that, even the most subtle arguments can be presented for or against its characteristics, but it doesn’t matter because the LLM was previously classified in the same category as humans. Additionally, by being categorized alongside humans, it gains new characteristics and is therefore made a part of a whole other set. A thorough examination of the fragment of the actual reality as well as in the space of potentialities is made possible by the system into which the element is incorporated providing it with new traits and features. The LLM also fits this description.
The means of solving the problems that arise are to be found in the same area where they arise. In this instance, the language itself has the tools needed to alter the psycholinguistic metamodel. It is suggested to use the toolkit of ergative languages to avoid the logical categorization problem. The toolkit of ergative languages, whose grammar is characterized by the opposition of agent (producer of action) and patient (bearer of action) rather than the subject-object opposition seen in nominative languages. When a verb is transitive, the ergative case encodes agence, revealing the source of the directed action. The instrumental case, also known as casus instrumentalis in Latin, is found in nominative languages and indicates a tool or instrument that acts on other objects or produces certain actions.
This results in two different ways of expressing the generation of literary content: 1) The author (person, subject, personality) creates (writes, composes) the text; 2) The text is created (generated, written) by the LLM. But in no way does the LLM create it. Simply because it is not the author, subject, person, personality.
What kind of subject then creates the text that is produced by the LLM? Here we should consider that in ergative languages there is no subject (and no object opposed to it), there is an agent and a patient. The LLM acts as the agent, the text acts as the patient. The question of the active subject in this language metamodel (frame of reference) does not make complete sense. The activity of the LLM can be described in the phenomenon, suggested and described by Roland Barthes, of the «scriptor» who «is born simultaneously with the text» [1] and has no being outside writing and no time, outside the speech (written or spoken) act. The LLM, as Barth’s scriptor, is not a subject in relation to which the text created (with its help, but not created by it) would be a predicate. LLM is not a subject (a person, an author) and is not valid in relation to his scientific discourse of will, intellect, understanding, ability, opinion, knowledge and other personal qualities. LLM is to be understood, not as a new subject that confronts the individual, but only as a personification of the individual. Such an understanding of LLM allows us to distinguish between author and scriptor, creativity and technical production, which, in turn, actualizes the problematic of the meaning and purpose of literary activity.
LLM embodies a new tool (digital agent) of the contemporary communicative environment: the personification of the linguistic subject. The language and literary culture face a value issue in the proper psycholinguistic comprehension and assimilation of LLM, which calls for an appropriate reaction based on logical reflection on the situation. Understanding the inevitable creation of a new linguistic phenomena, as conditioned by the entire history of language and speech development, seems important (on the basis of the ergative linguistic metamodel). This reflection requires asking the appropriate questions, whose discovery is a significant issue in and of itself (scientific and educational). The way we respond to the unavoidable, where we can discover (or overcome) both the good and the bad, will determine how our future turns out.
Bibliographic list
- Барт, Р. Смерть автора / Р. Барт // Избранные работы: Семиотика. Поэтика : пер. с фр. / сост., общ. ред. Г. К. Косикова. – М. : Прогресс, 1989. – С. 384–391.
- Концевая, Г. М. Феномен «Медиаробота» в логической ловушке номинативных метафор /Г. М.Концевая, М. П. Концевой // Медиа и власть: власть медиа? : материалы сежд. н.-об. конф., Казань, 10.11.2020 г. – Казань: Изд. КУ 2020. – С. 154–159.
- ChatGPT launches boom in AI-written e-books on Amazon [Electronic resource]. – Mode of access: https://ir-handouts.com/2023/02/23/chatgpt-launches-boom-in-ai-written-e-books-on-amazon. – Date of access: 10.03.2023.
