Resumen de Employing text mining to investigate language attitudes and social identity on Chinese social media

Zhenzhen Zhang, Haomin Zhang

• Language attitudes are a central focus in sociolinguistics, with research methods steadily advancing. This study employed the text mining method, using keywords to retrieve 4.3k relevant Weibo posts to build a corpus. The BERTopic analysis reveals that discussions on Weibo primarily focus on three themes: cultural entertainment, language and regional issues, and education and exams. These topics demonstrate both short-term viral dissemination and long-term persistence. Manual coding indicates that discussions on language attitudes predominantly revolve around social contexts. Sentiment analysis shows that overall sentiment toward language attitudes is largely neutral (36.72%), with positive sentiment (33.36%) slightly exceeding negative sentiment (29.92%). Further analysis highlights that language attitudes are shaped by interpersonal emotional projection, social recognition and identity construction, and the functional role of language as a communication tool. Moreover, a societal expectation for ‘accent standardization’ is evident in discussions concerning both Mandarin and regional dialects.