Linguistic creativity in virtual discourse as a test for artificial intelligence
Taymour, Maria Pavlovna PhD in Linguistics, Assistant Professor, Department of Linguistics and Professional Communication in Humanities and Applied Sciences, Moscow State Linguistic University, Moscow, Russia email@example.com
Abstract This work is devoted to the study of cognitive functions and characteristics of English-language Internet memes as polymodal units of online communication, as well as to the identification of potential difficulties in their analysis by artificial intelligence. Internet memes are considered as specific units of virtual communication, consisting of several different code components. The author studies such types of creativity in Internet memetics as a language game, violation of the language norm, as well as cases of using metaphors, metonyms and metaphtonyms as global mechanisms for the creative conceptualization of the world. The present study also focuses on cases of non-canonical use of mixed metaphors as several metaphors with different source domains combined in one phrase or sentence and malaphors as a hybrid of several proverbs or sayings in a single phrase. We show that a large number of Internet memes can cause changes in the formation of a fragment of the recipient's worldview under certain conditions, since they are aimed at influencing him. It is determined that the effectiveness of the inference of the meaning of such creative discursive elements by artificial intelligence depends on the number of involved associative links or frame structures in the mind of a recipient, where certain frames are formed on the basis of already existing associations and clichés. The results of the analysis of the creative potential of memes, which include mixed metaphors and malaphors, as well as the functioning of cognitive mechanisms of (de)compression of information, (de)focusing of attention, cognitive resonance and dissonance, are presented. It is determined that the cognitive dissonance that occurs in the addressee can be predicted by AI in the presence of the effect of deceived expectation in the discourse element where cognitively incompatible frames occur (for example, in the case of the presence of a malaphor in the verbal part of the meme).