Detecting China’s Use of Emotive Frames in Public Diplomacy: An NLP Approach

• In Event: China's Propaganda Apparatus

Abstract

Is China manipulating frames in its messaging to foreign audiences? If so, how would we measure this? This paper employs natural language processing (NLP) to detect when and how China uses strategic framing to target foreign audiences. It finds that Chinese state-backed media outlets strategically exploit emotive and value-resonant frames from Western outlets, and moreover, this strategic framing focusses on topics that discuss Western legitimacy or China’s own governance legitimacy.

Studies in public diplomacy show that countries can effectively target foreign audiences, as a means to shape the policies of foreign governments. Strategic framing of messages is one tool for this activity; however, questions remain as to which frames are most effective. Studies have suggested that frames that build on existing emotions, especially negative ones like fear or anxiety, are more likely to influence audiences. Successful frames also tend to be consistent with the audience’s pre-existing values or ideology. This paper builds on the strategic framing literature in testing whether China identifies and exploits pre-existing emotive and value-resonate frames in the target audience in maximising the efficacy of its public diplomacy campaign.

I compile a corpus of 1.8 million Facebook posts by Chinese English-language state-media outlets and mainstream Western news outlets over a ten-year period. I use two methodologies to gauge strategic framing: (1) a structural topic model detects the most prominent topics in the Chinese English-language state-media outlets and mainstream Western news outlets; and (2) a series of dynamic contextualised word embedding models trained at weekly intervals identify subtle frame shifts in the topics over time. This multi-method approach is novel in automating the identification of frame movements at specified time intervals.

I test the hypothesis that China is strategically exploiting emotive and value resonant frames in its target Western audience to maximise the efficacy of its public diplomacy messaging. Confirming evidence for this hypothesis is that frames used in Western media are then subsequently employed to frame Chinese state-backed media messaging. The findings broadly support the hypothesis, and in particular, they show that the most dramatic spikes in Chinese state-backed media topic prevalence occur in those topics that criticise Western governance, such as the Black Lives Matter movement. Chinese state outlets tend to adopt critical narratives from Western outlets to frame these topics. In contrast, topics that relate to China’s state legitimacy, such as the Hong Kong protests, are discussed more continuously but also shift frames more rapidly. These topics co-opt emotive and value-resonant frames that first appear in Western news outlets.

Beyond using NLP in a novel way to automatically detect frame shifts, more broadly, this paper advances the political communications literature by revealing strategic frame building techniques used by China in public diplomacy messaging.

Author

• Hannah Bailey, University of Oxford