18–20 Ara 2024
Istanbul University Faculty of Letters
Europe/Istanbul saat dilimi

From Natural (Human to Human) Discourse to Artificial Intelligence-Assisted (Human to Artificial Intelligence) Discourse: A Linguistic Approach

18 Ara 2024 16:40
30dk
A7 (Istanbul University Faculty of Letters)

A7

Istanbul University Faculty of Letters

Oral Presentation Theoretical Foundations Session 2.2 (Day 1)

Konuşmacılar

Azad Mammadov (Azerbaijan University of Languages - Azerbaijan University) Sabina Ismayilova

Açıklama

My research question is: can AI function as a communicator and replace human as an equal discourse participant? For this purpose, the paper focuses on the ways of reproduction of natural discourse production by AI. The most challenging issue from a linguistic perspective is how AI deals with various social and individual factors, which heavily influence the process of natural text construction and natural discourse production. Recent studies reflect general understanding of the problem suggesting the necessity to study artificial intelligence as a communicator within human-computer communication (Broussard 2018, Guzman & Lewis 2020). The most intriguing issue is the challenge AI supported non-human audience can face to identify implicit relations. It requires the study of not only explicit relations built in the process of text construction, but also implicit relations built during discourse production. In this connection, the paper focuses on the ways how AI supported non-human audience identifies explicit and implicit relations during production of AI assisted discourse.
The method of discourse analysis has been applied to study academic (Noam Chomskys Language and Mind), literary (Michael Ondaatjes The English Patient) and media (Reuters, BBC and The Sun) texts.

On the levels of text and discourse, grammatical and lexical markers of explicit and implicit relations perform key textual functions. Any AI assisted computer can`t participate in human- computer communication without identifying these explicit and implicit relations. For this purpose, AI needs more specific situational knowledge during reproduction of natural discourse.
Despite the fact that the texts under analysis are all in English, the study of the linguistic mechanisms of text construction demonstrates their universal nature. They are more or less common in the languages across genealogical and morphological classifications. This fact can give us a strong argument to consider this universal nature as very helpful for reproduction of natural discourse by AI. By contrast, culture-specific features of natural discourse are obvious obstacles, which could be resolved by alternative algorithms. In addition, AI must deal with genre specific features of each discourse.

Keywords: Natural discourse, human-computer communication, artificial intelligence, explicit relations, implicit relations

Institution / Affiliation / Kurum

Azerbaijan University of Languages / Azerbaijan University

Presentation language / Sunum Dili EN (English)
Disciplines / Disiplinler Linguistics / Dilbilim
E-mail / E-posta azad19622003@yahoo.com
ORCID ID 0000-0002-8357-1356
Country / Ülke Azerbaijan

Başlıca yazarlar:

Azad Mammadov (Azerbaijan University of Languages - Azerbaijan University) Sabina Ismayilova

Sunum Materyalleri