Nonverbal communication with emojis in social media dissociating hedonic intensity from frequency

Abstract

As a popular means of nonverbal communication in social media, emojis provide quick predictions about public sentiments towards social events. Previous analyses of emojis reported that people use positive emojis more frequently than negative emojis. However, psychological research reveals a negativity bias in sentiment, as seen in the phenomenon of loss-aversion, where negative sentiment due to a loss possesses a greater psychological valence than positive sentiment due to a gain of an equal amount. We propose that the frequency and intensity of emojis are dissociable. Whereas the frequency of emojis reflects social norms in public communication, the intensity reflects hedonic values and loss-aversion. We first developed a text-free emoji sentiment lexicon based on a survey with more than 900 users of Weibo (a Chinese version of Twitter). Using the sentiment lexicon, we then analyzed 8822 Weibo comments containing the indexed emojis in reaction to three controversial events (i.e., a murder case in which public opinion largely opposed the final verdict, a manslaughter case in which public opinion was supportive of the final verdict, and a public debate on an award-winning subject). The results showed that positive or negative emoji frequency was consistent with the majority sentiment (social norm) towards a controversial event. In contrast, the average intensity of negative emojis was stronger than positive emojis across all three samples, revealing a public sentiment version of loss-aversion. In all three samples, emoji polarity analysis served as a proxy for text sentiment analysis to capture public attitudes towards a social event.