Multimodal Neurotexts in Network Political Discourse: Manipulative Potential (Based on English-language Internet Memes about Scottish Regionalism)

The article is devoted to the study of multimodal neurotexts — texts generated by large language models and combining verbal and visual components. The material consisted of 45 English-language internet memes about Scottish regionalism, created by the Supermeme.ai neural network from OpenAI using three prompts with different degrees of evaluativeness. The aim of the research is to identify linguistic and visual means of manipulative influence in neurotexts and to compare them with the techniques characteristic of memes created by humans. It has been established that neurotexts reproduce the manipulative and identitarian characteristics of regionalist discourse: they employ metaphorization, linguocreative elements (alliteration, onomatopoeia, puns), internet slang, idioms, and cultural references. The cultural non-specificity of visual templates has been revealed: Scottish domestic specifics are illustrated by the visual imagery of American popular culture. It is shown that, contrary to the hypothesis, the memes generated from the three prompts retained the evaluativeness embedded in the first prompt, demonstrating a bias in favor of Scottish regionalism. The conclusion is made about the high manipulative potential of multimodal neurotexts and the risks of their dissemination during political campaigns and conflicts.