Abstract
This research aims to analyse and compare the types of lexical cohesion used by Jack Ma and Elon Musk in their debate on artificial intelligence at the World Artificial Intelligence Conference (WAIC) 2019. It applies a descriptive quantitative method using Renkema (2004) framework, focusing on six types of lexical cohesion: repetition, synonymy, antonymy, hyponymy, meronymy, and collocation. The data were obtained from the debate transcript uploaded by “News China TV” on YouTube. The findings reveal that Jack Ma used more lexical cohesive devices (432 instances) than Elon Musk (343 instances), with repetition being the most dominant type. Jack Ma employed lexical cohesion to emphasize human values in responding to the advancement of AI, while Elon Musk used it to explain the technical and scientific aspects of the technology. Although both speakers used collocations at a similar frequency, their functions differed according to each speaker’s background and communication goals. These differences illustrate that lexical cohesion not only connects words in discourse but also reveals rhetorical strategies and speaking styles. This research contributes to a deeper understanding of how lexical cohesion operates in spontaneous argumentative discourse.