Linguistic Characteristics of Social Media Messages Spreading across Geographic and Linguistic Boundaries

Abstract

Social media enable messages to be exchanged beyond geographic constraints. Some of the messages could be shared and forwarded by people with different cultural backgrounds across different geographical regions. Studying the content of messages that can reach diverse populations is important for practices such as movement propagation and global marketing. Existing studies mainly investigated the characteristics of messages that are popular, i.e., shared or forwarded by more users. As the diffusion of information is prone to be echoed inside certain geographical and linguistic boundaries, popular messages are not always to be shared and spread across geographical and linguistic boundaries. We investigated the linguistic characteristics of social media messages that can reach and be disseminated by people across nations, and across geographic and linguistic boundaries in the MeToo movement. Specifically, we analyze the diffusion paths of messages according to the geolocation of tweets and conducted statistical analysis to compare the linguistic characteristic of tweets that spread across geographical or linguistic boundaries with those that do not. We focus on the linguistic characteristics from three aspects: ‘emotions’, ‘social relations’, and ‘economics, politics, and religion’. Our findings reveal that popular messages tend to contain more negative emotions, however, messages with negative emotions are unlikely to be disseminated across geographical or linguistic boundaries. On the other hand, messages on economic topics or non-adults’ issues are more probable to be disseminated universally. The findings provide insights on the content that is more probable to be shared and disseminated by people with different cultural backgrounds across geographical regions.