Gender Differences in Social Media Language and Identity Construction from a Big Data Perspective

Abstract

With the rapid development of social media and big data natural language processing (NLP) technology, online language has become a critical carrier of individual gender identity presentation and social identity construction. This study adopts a big data quantitative analysis framework, crawling 1,286,420 valid social media posts from Twitter (X), Instagram and Weibo, constructing a balanced gender corpus, and applying linguistic feature quantification formulas, independent sample t-test and logistic regression model to explore gender differences in social media lexical features, syntactic styles and emotional expression, as well as the internal mechanism of language-based gender identity construction. The results show that female users are significantly inclined to use relational, emotional and interactive linguistic markers, while male users prefer agentic, task-oriented and concise expression styles; gender linguistic differences have a significant predictive effect on online identity construction, with an optimal gender language classification accuracy of 59.09%. This study verifies the applicability of social role theory in online discourse scenarios, constructs a quantitative evaluation model of social media gender language, and provides empirical support for revealing the interactive relationship between digital language behavior and social identity shaping.