Investigating Linguistic Variation in Chinese Podcasts: A Quantitative Stylistic Perspective

In recent years, the rise of internet media has fueled the development and prosperity of podcast audio programs. Podcasts, which use sound as their medium, are divided into two sub-registers: monologue and multi-person conversation. This article aims to explore the linguistic styles and distinctive characteristics of these two sub-registers within the podcast genre. Understanding these differences is essential for mastering the linguistic characteristics of the information age and has significant implications for applications in the field of natural language processing. Based on a self-constructed corpus of 323,994 words of spoken language from podcasts, a quantitative analysis was conducted from the perspective of stylometry to examine the linguistic features of Monologue Podcasts and Conversation Podcasts. The results reveal significant differences between Monologue Podcasts and Conversation Podcasts in terms of Word Classes usage across five categories, Word Length, Lexical Richness, Content Richness, and The Degree of Formality. Monologue Podcasts exhibit a higher level of formality, characterized by features typical of writ- ten language, whereas Conversation Podcasts are more colloquial and interactive in nature. The differences between Monologue Podcasts and Conversation Podcasts were analyzed from the perspectives of contextual features, communicative purposes, and cognitive demands. Additionally, accurately clustering the corpus into two categories using eleven specific indicators as text representation features, and representing texts through metric features of language structure, has significantly enhanced the interpretability of clustering and classification research.