Abstract Social media plays an increasingly prominent role in the timeliness and sharing of information. While much research focuses on improving user stickiness, the impact of celebrities' views on user behavior remains under-explored. This study investigates the influence of celebrity language style on user communication and opinion dissemination. By analyzing the language style characteristics of over 20,000 celebrity posts on Sina Weibo and applying Speech Act Theory (SAT) alongside Latent Class Analysis (LCA), we identify key linguistic combinations that drive engagement. Results indicate that positive expressions—combining positive emotion, external focus, and analytical rigor—are the most effective for communication. Furthermore, celebrity emotional states significantly influence user sentiment, with a notable trend toward reduced positive and increased negative emotions in user responses.
Introduction
Celebrities wield immense power on social media, yet the mechanics of how their stylistic choices influence audience behavior are not fully understood. Most existing research focuses on brand-consumer interactions or general user behavior. This paper addresses this gap by examining "celebrity opinion dissemination." Specifically, it asks: Does a narrative style outperform an analytical one? Does focusing on external events versus internal feelings drive more engagement? And how does a celebrity’s emotional tone affect the sentiment of the resulting discourse?
Methodology and Theoretical Framework
The study utilizes a multi-step pipeline combining linguistic theory with statistical modeling to understand how information is presented and received.
Speech Act Theory (SAT)
The research frames social media communication as an "act" consisting of three components:
- Intraverbal act: The actual presentation of information.
- Idiosyncratic Acts: Grasping the intentions behind the post (e.g., narrative vs. analytical style).
- After words: The results of the information exchange, such as user communication behavior and emotional spread.
Data Processing and Analysis
The researchers crawled a dataset from Sina Weibo featuring 21,788 original posts from 9 high-profile accounts across three niches: Fashion Bloggers, Writers, and Celebrity Chefs. User engagement was measured through 129,214 comments and retweet counts. Posts were categorized across three dimensions: Narrative vs. Analytical, Internal vs. External Focus, and Emotional Tone. Latent Class Analysis (LCA) was then employed to identify "hidden" subgroups of posts sharing similar linguistic trait combinations.
Key Results and Discussion
Linguistic Style Dominance
The analysis revealed a clear preference for analytical styles across all celebrity categories. As shown in Figure 2, celebrities across different niches consistently favor an analytical approach over a narrative one, suggesting a trend toward more "official" and "rigorous" public personas.
Furthermore, celebrities focus on external events (social hot topics, products) significantly more than internal feelings, with external focus appearing in 59% to 71% of posts.
The Three Post Archetypes
The LCA model identified three distinct categories of celebrity communication:
- Positive Expression (71.3%): The most successful category, characterized by high positivity, analytical style, and external focus. These posts achieved the highest engagement levels.
- Negative Catharsis (15.5%): High negative sentiment or "venting." These posts typically saw the lowest user engagement.
- Official Elaboration (13.3%): High neutral sentiment and highly analytical. These dry, factual updates often failed to resonate emotionally with the audience.
Emotional Contagion
An interesting "emotional leakage" effect was observed. While celebrity positivity generally fosters user positivity, there is a slight "dampening" effect where user responses are often less positive than the original post. Conversely, negative emotions from celebrities tend to trigger an amplified negative response among users. This underscores the importance of positive guidance on public platforms to avoid the rapid dissemination of negative sentiment.
Conclusion and Future Work
This research demonstrates that a "winning combo" of positive emotion, external attention, and analytical rigor creates the most effective communication hook for social media audiences. However, the study is limited by its small sample of nine celebrities and its focus on text-only analysis. Future research should expand the sample size across more diverse industries and explore "Linguistic Style Matching" to see if user replies mirror the celebrity's writing style.
Summary updated