Social media discourse is characterized by the post-truth phenomenon, in which feelings and personal beliefs appear to exert greater influence on shaping public opinions than truth. In such cases, the truth of news may be blurred and reported events are often reversed along revelations of the facts. Reverse news on social media is, in a sense, a typical instance of post-truth discourse. This study attempts to provide a corpus-based description of attitude appraisal and illocutionary acts in reverse news on social media, with the aim of investigating semantic and pragmatic properties of post-truth discourse. The corpora for this study consist of social media posts in Chinese and English about a defamation lawsuit, which were collected from Weibo and Facebook.

The results indicate that the English corpus emphasizes the use of appreciations and judgments with more complex co-occurring relations among three attitudinal domains while the Chinese corpus contains more judgments than appreciations and affects. Meanwhile, both corpora exhibit higher frequencies of assertives and expressives than the other acts, yet the Chinese corpus presents more complicated illocutionary act sequences than the English corpus. After the verdict, the frequencies of both attitude appraisal and illocutionary acts decrease in the Chinese corpus but increase in the English corpus. Based on this, we propose that the development of post-truth discourse may go through three functional stages: starting with expression, progressing to appeal, and ending with representation. The sequence of these stages may vary depending on sociocultural contexts.