Narratives of the Ukraine War on Chinese Social Media

• In Event: Mechanisms of Domination?: State Sponsored Information and Media Ecosystems

Abstract

As the Ukraine war unfolded, Western observers noted the presence of Russian narratives on Chinese social media and speculated that the Chinese government was flooding social media with such narratives to garner support for China’s alignment with Russia in the conflict. How prevalent were Russian narratives of the Ukraine war on Chinese social media? Between February 1, 2022 and April 18, 2022, we collected 475,231 Weibo posts published by 160,792 users related to the Ukraine War in real-time. We then hand-label a random sample of 3,200 posts to fine-tune a pretrained Chinese BERT with the Whole Word Masking model to identify Weibo content related to the Ukraine War. To describe narratives on Weibo toward the Ukraine war, we hand annotate posts to train similar supervised machine learning models. Specifically, we focus on the narrative adopted by Russia that the conflict is a result of Putin defending Russian sovereignty against Ukrainian aggression spurred on by the United States and NATO (Putin as victim narrative) as well as the opposing narrative, which captures the prevailing understanding of the Ukraine war in North America and Western Europe, that Putin's expansionist ambitions drove Russia to invade Ukraine (Putin as aggressor). We find that although the Putin as victim narrative is more prevalent (21% of posts) than the Putin as aggressor narrative (6% of posts), the majority of content on Weibo related to the war (73%) contained no opinion or opinions that do not fall into either of the two narratives. Contrary to speculations that the Chinese government flooded Chinese social media with Russian narratives of the Ukraine war, we find that Weibo accounts of unverified users, including those identifying as “self-media” (自媒体), were the accounts disproportionately more likely to post Putin as victim narratives (57% of posts) than Putin as aggressor narratives (50% of posts).

Authors

• Jennifer Pan, Stanford University

• Yingdan Lu, Northwestern University

• Anfan Chen, The Chinese University of Hong Kong