https://orcid.org/0000-0002-3815-4705
Figures
Abstract
Previous studies have demonstrated that women in male-dominated fields face a challenging “double bind”—they must display traditionally masculine traits while maintaining feminine qualities. This study provides a corpus-assisted discourse analysis of gender performances by male and female entrepreneurs on Twitter (now X) at three levels: (1) issues, (2) personality traits, and (3) linguistic styles. The primary purpose is to examine how the “double bind” manifests in male and female entrepreneurs’ social media performances. The findings suggest that both male and female entrepreneurs strategically navigate the “double bind”, balancing gender role expectations while engaging their followers across diverse topics. Their performances feature a “double bind” with a twist, as both male and female entrepreneurs tend to adopt an interactive and engaging style rather than a combative and confrontational one to connect with their followers and maintain interactive intimacy.
Citation: Liu M, Wang G, Zhao R (2025) “Double Bind” with a twist: A corpus-assisted discourse study of gender performances of male and female entrepreneurs on Twitter (now X). PLoS One 20(8): e0331400. https://doi.org/10.1371/journal.pone.0331400
Editor: Rafael Galvão de Almeida, Universidade Federal do Tocantins, BRAZIL
Received: January 7, 2025; Accepted: August 14, 2025; Published: August 29, 2025
Copyright: © 2025 Liu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting information files.
Funding: This work was supported by The Hong Kong Polytechnic University (ID No.: 4-ZZSP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Previous studies have demonstrated that women in male-dominated fields face a challenging “double bind”—they must display traditionally masculine traits while maintaining feminine qualities [1]. In leadership roles, women who exhibit “toughness” may be criticized for lacking warmth, while those perceived as too feminine may be deemed incompetent to lead [2]. This dynamic is particularly evident in politics, where voters associate different competencies with gender. Male politicians are typically viewed as more capable in areas like economics, military affairs, and technology, while women are often relegated to “softer” domains such as education, healthcare, and social welfare [3]. Female politicians frequently adopt traditionally masculine characteristics to counter these stereotypes [4], while male politicians often emphasize their own softer qualities [5].
In entrepreneurship, success has traditionally been associated with masculine traits like risk-taking, assertiveness, and competitiveness [6]. This association can disadvantage women entrepreneurs, who may be expected to simultaneously demonstrate these qualities while maintaining traditionally feminine characteristics like warmth and relationality [7]. Studies have shown different expectations for male and female CEOs, with women facing additional pressure to demonstrate empathy and human resource management skills [8]. The disparity is particularly evident in how female CEOs are expected to be more caring and empathetic towards employee needs, while male CEOs are primarily evaluated on their business decisions and communication effectiveness. These gendered expectations create additional challenges for women in leadership positions, who must carefully navigate both professional competence and societal expectations of femininity [9].
The rise of social media has introduced new dimensions to this discourse [10]. While some researchers suggest that social media platforms could serve as equalizers for women entrepreneurs by providing new opportunities and reducing traditional barriers [11], others argue that these digital spaces may actually reinforce or amplify existing gender disparities [9,11,12]. The debate continues regarding whether social media truly empowers women in entrepreneurship or merely replicates offline gender biases in digital form [12]. Some studies highlight social media’s potential as a “safe space” where women can overcome traditional obstacles and access new economic opportunities [11,13], while others warn against overemphasizing the empowering effects of social media [9,12]. This complex relationship between gender, entrepreneurship, and social media platforms warrants further investigation. This study integrates corpus linguistics and critical discourse analysis to explore how the “double bind” emerges in the performances of male and female entrepreneurs on social media. The basic assumption is that gender performances of male and female entrepreneurs on social media feature the “double bind” with a twist: while both groups strategically adapt to traditional gender norms to foreground their credibility and competence, the interactive nature of social media also allows them to navigate the “double bind” strategically.
2. Cyberfeminism and gender performances on social media
This study is informed by cyberfeminism, which examines the Internet’s impact on gendered power dynamics through the lens of social construction [12,14]. It adopts the perspective that gender is a “social construct” rather than an inherent social category, emphasizing that gender is an active enactment—a matter of “doing” rather than a static state of “being” [15]. Accordingly, gender is perceived as both an outcome and an ongoing process shaped through various forms of expression and self-expression [16]. Within this socially constructed framework, gender identity is a pivotal concept that reflects the degree to which individuals embody masculine and/or feminine characteristics as delineated by conventional gender norms [13]. Thus, the act of constructing gender is inherently tied to power dynamics, a connection that is notably evident in the politics of gender portrayal within social media contexts [17].
The exploration of how women express their gender identities in the digital realm has been a recurring theme in cyberfeminist scholarship [12], reflecting the multifaceted ways in which women navigate the Internet, sometimes challenging but often perpetuating entrenched gender and racial hierarchies [14]. As understanding of the interplay between gender and digital culture evolves, a growing body of cyberfeminist work suggests that socio-economic disparities are often mirrored or even intensified in the online world [18]. The crux of the cyberfeminist debate centers on whether and how Internet technologies can disrupt prevailing gender disparities; this includes an examination of the degree to which inequalities evident in the offline world persist in the digital space [14].
As McAdam et al. [11] review, contemporary research in cyberfeminism pursues two main lines of inquiry. The first echoes the perspective of Plant [19], which posits that the Internet can alter traditional gender power dynamics. Researchers in this vein view the Internet as a “safe space” that provides women the comfort to thrive [11], bolstering women’s self-assurance, enhancing their communication abilities, fostering community engagement, and possibly unlocking economic possibilities. The second line of critique, however, questions the neutrality of the Internet as an entrepreneurial platform [20], and warns that the capacity of digital technology to empower economically marginalized women might be exaggerated, given that gender inequalities encountered offline tend to be replicated in the online environment [12].
While traditional studies used to conceptualize masculinity and femininity as bipolar ends of a single continuum and view women’s and men’s expected ways of thinking and behaving as either masculine or feminine [21], cyberfeminist research has underlined that a person can simultaneously possess and exhibit femininity and masculinity [14]. For example, a person scoring high in one dimension (i.e., femininity or masculinity) would exhibit the stereotypes of that gender (masculine or feminine) identity. The cyberfeminist theory expects that women engaged in digital space could exhibit traits that are both masculine (e.g., aggressive, assertive, and forceful) and feminine (e.g., affectionate, warm, and yielding), both instrumental (e.g., getting the thing accomplished) and expressive (e.g., sensitive to the needs of others) [22]. However, there remains a scarcity of research that analyzes the linguistic expressions of gender by both male and female entrepreneurs on social media.
3. From critical discourse analysis to corpus-assisted discourse studies
Critical discourse analysis (CDA) views discourse as a social practice and underlines the importance of examining discourse in its socio-political contexts [23]. It aims to expose the relations between language, power, and society [24]. CDA views gender as socially constructed and aims to explicate the “(re)production, negotiation, and contestation of gender ideologies and power relations in a particular context” [25]. It has been employed to uncover the construction of gender identities and dissemination of gender norms and roles [26]. However, CDA used to rely on a qualitative of small samples of texts, which has been criticized for a lack of methodological rigor and interpretation problems [27].
The recent development of corpus linguistics (CL) has contributed to growing studies in combining CL and CDA, but most of these studies fail to give a balanced combination of both disciplines [28]. Corpus-assisted discourse studies (CADS) is known for its emphasis on the balanced ‘synergy’ of quantitative corpus linguistic methods with detailed qualitative discourse analysis [29]. On the one hand, CL can benefit CDA by processing large samples of data efficiently and identifying language patterns that cannot be obtained through mere manual analysis. On the other hand, CDA can contribute to better interpretation and explanation of the findings generated by corpus analytic tools. CADS has been widely applied to the analysis of discursive contractions of different topics including war and politics [30–32], immigrants and minority groups [28], environment [33,34], social movement and protests [35].
It has also been used in the analysis of discursive constructions of gender [36]. However, these studies used to rely on traditional CL methods, such as keywords, collocates, and concordance [37], and more efforts are needed to take advantage of recent development in computational linguistics and text mining to realize the accurate processing of large samples of texts [38]. Newman et al. [38] compared the language of men and women across 14,000 samples of Twitter texts with the Linguistic Inquiry and Word Count tool. Windsor et al. [39] use a suite of computational linguistic approaches such as topic modelling, sentiment analysis and model the evolution of gendered language in the US senate. These studies have demonstrated the analytic potentials of quantitative analytic methods in revealing the linguistic differences of different genders.
4. Data collection and analytic methods
4.1. Data collection
Twitter, now known as X, stands as one of the world’s preeminent microblogging platforms, with around 600 million monthly users as of April 2025 [40]. The platform serves as a crucial channel for social connection and interaction, enabling users to communicate through concise messages of up to 280 characters [4]. Recent research indicates that social media platforms have emerged as supportive environments for female users, facilitating self-promotion and follower engagement strategies [41].
This study analyzes the 100 most recent posts from each of top 10 influential entrepreneurial male and female leaders, as ranked by Forbes in its “The Most Powerful Women List” (2021) and “The World’s Billionaire List” (2021). The inclusion of top 10 influential entrepreneurial leaders ensures that the sample is comprised of verified, credible figures in business and entrepreneurial contexts. Furthermore, the more influential these female entrepreneurs become, the more likely they are to experience the “double bind,” making their discourse particularly relevant for analysis.
Table 1 provides a detailed overview of their Twitter accounts, showcasing the names, positions, and number of followers at the time of data collection, which further validates the sample’s credibility. Data collection was conducted using the NCapture function within NVivo software, a tool specifically designed to streamline the process of gathering data from various sources. NVivo’s NCapture function utilizes Twitter’s API (Application Programming Interface) to access and collect data efficiently.
The Twitter (X) usernames of the top 10 male and female entrepreneurs were used as input search parameters, with the data collection configured to include only the 100 most recent posts from each account. To minimize data redundancy and ensure a diverse representation of daily topics, only the initial post of each day was selected [42]. This methodological decision allows for a focus on the entrepreneurs’ most deliberate and unprompted communications of the day, avoiding potential influence from interactions or reactions later in the day.
The resulting corpora consist of 31,124 words from female entrepreneurs and 27,773 words from male entrepreneurs, providing a robust dataset for analyzing gendered discourse patterns in entrepreneurial contexts. The collection and analysis method complied with the terms and conditions for the source of the data.
4.2. Analytical framework and procedures
This study aims to analyze gender performances on Twitter at multi-levels of discourse: (1) (masculine vs. feminine) issues; (2) (masculine vs. feminine) personality traits; (3) masculine vs feminine linguistic style [38,41,43]. So-called masculine issues that are stereotypically assumed to be better handled by men include economy, military, foreign affairs, technology, science, crime, terrorism, and gun problems, while so-called feminine issues for which women are perceived to be more suitable include education, health, human right, women’s rights, animal right, child care, poverty, arts, environment, and social welfare [41,44]. Traditional social role theory posits that men and women are assumed to have different personality traits; men are expected to be strong, tough, assertive, competent, instrumental, and achievement-oriented (so-called masculine traits), whereas women are assumed to be warm, caring, understanding, compassionate, expressive, and family-oriented [41].
Contemporary computational literature often distinguishes men and women on pragmatic dimensions of “informativeness” and “involvement” based on earlier corpus-based contrasts of written and spoken genres [45]. The involvement dimension consists of linguistic resources that create interactions between speakers and their audiences; the informational dimension consists of resources that communicate propositional content. Early work compared frequencies of large word classes, such as parts-of-speech: the involvement dimension includes first and second person pronouns, present tense verbs, and contractions [45,46], while the informational dimension includes prepositions, attributive adjectives, and longer words. Informational word classes were used preferentially by men, while involvement and interaction are associated with women [47].
The online corpus-analytic tools Wmatrix 5.0 and KH Coder were used in this study to examine whether and to what extent male and female entrepreneurs vary in the choice of issues, personality traits and linguistic styles. Wmatrix 5.0 is a software for corpus analysis and comparison. It provides a web interface to natural language processing tools such as the USAS and CLAWS corpus annotation tools for English. USAS (UCREL Semantic Analysis System) is a framework for automatic semantic analysis of texts. It has a multi-tier structure with 21 major semantic fields and 232 sub-semantic categories. CLAWS (the Constituent Likelihood Automatic Word-tagging System) is a Part-of-Speech (POS) tagging tool for English texts.
With its corpus annotation tools, Wmatrix can extend keyword analysis to the analysis of key semantic categories (SMCs) and key parts of speech (POSs) [48]. Keyness is measured by log-likelihood (LL) values: the higher the keyness values, the more statistically significant the differences [48]. The two corpora were first compared with a general reference corpus, American English 2006 (AmE06), to generate two key SMC lists. The top 20 key SMCs were compared to identify those present in both corpora. Then, the two corpora were compared with each other to generate two additional key SMC lists which can suggest their preferential key SMCs. This study focuses on the top 20 preferential key SMCs of each corpus and identify their respective preferences for topics and personality traits. In order to guarantee that those key SMCs identified are statistically significant, the LL values of these key SMCs must be above 6.63 (p < 0.01). Additionally, the two corpora were compared to generate two key POS lists. The key POSs with LL values above 6.63 (p < 0.01) were analyzed to identify the distinctive POSs of the two corpora [49]. These findings suggest the characteristic linguistic styles of each corpus.
This study employs KH Coder, an open-source text mining software, to analyze variations in self-presentation between female and male entrepreneurs. Originally developed for quantitative content analysis, KH Coder integrates the Stanford POS Tagger for English word extraction, R for statistical computations, and MySQL for data management and retrieval [50]. The software offers a range of analytic tools, including frequency analysis, concordance examination, key term extraction, word association mapping, correspondence analysis, multi-dimensional scaling, hierarchical clustering, co-occurrence network visualization, and topic modeling [50,51].
For this analysis, we conducted word association analysis to examine terms frequently co-occurring with first-person pronouns (I, we, our) within a sentence. KH Coder visualizes these co-occurrence patterns through network diagrams, with line thickness indicating the strength of word associations. By default, the software displays the 60 strongest word pair connections, omitting unconnected terms from the visualization [50]. Within these networks, the software automatically identifies and color-codes distinct word clusters, revealing thematic patterns in how entrepreneurs represent themselves in their communications.
5. Data analysis and discussion
5.1. Analysis of shared key semantic categories
An analysis of the top 20 key semantic categories (SMCs) of both corpora (see Table 2) when they are compared with American English 2006 (AmE06) reveals 12 shared key SMCs. These include “Helping” (S8+), “Unmatched” (Z99), “Content” (E4.2+), “The universe” (W1), “Polite” (S1.2.4+), “Time: New and young” (T3-), “Time: Future” (T1.1.3), “Belonging to a group” (S5+), “Evaluation: Good” (A5.1+), “Money and pay” (I1.1), and “Time: Present; simultaneous” (T1.1.2). The I1.1 category, comprising terms such as stakeholders, investors, invest(ing), investment, and capital, reflects the expected financial focus of entrepreneurial discourse.
Both gender groups exhibit similar behavioral patterns in their communications, demonstrating caring (S8+), nurturing (A5.1+ and S1.1.2+), and emotional expressiveness (E4.2+ and E4.1+) [52]. The helping orientation (S8+) is particularly prominent across both groups. In the female entrepreneurs’ corpus, this manifests through terms like help (60), support (30), helping (25), inspiring (19), services (11), and supporting (11). The male entrepreneurs’ corpus shows similar patterns with help (46), support (31), helping (11), inspiring (10), protecting (10), and helped (10). These linguistic choices position both groups as helpers and supporters within their professional spheres.
Both male and female entrepreneurs express strong positive emotions in their communications, particularly through expressions of happiness (E4.1+) and satisfaction (E4.2+). The satisfaction category is dominated by words like proud, glad, and pleasure, while happiness is communicated through terms like happy, celebrating, and celebrate. This can be attributed to the curtesy on social media to celebrate and share good news. They also share a tendency toward positive evaluation (A5.1+), frequently using words such as great, progress, well, and good. These positive evaluative words suggest that both groups feel confident and satisfied with their achievements.
The analysis reveals that both female and male entrepreneurs foreground politeness (S1.2.4+). Female entrepreneurs use terms like thanks (32 occurrences) and grateful (14), while male entrepreneurs show comparable usage with thanks (41) and grateful (17). This suggests a shared recognition of the importance of maintaining professional courtesy and expressing appreciation.
Time-related themes emerge as another significant commonality across both groups, which can be seen from three distinct temporal categories. References to newness and youth (T3-) appear frequently, with new, young, and innovation being prominent terms. Future-oriented language (T1.1.3) is evident through frequent use of will, future, and tomorrow, while present-focused communication (T1.1.2) employs terms like today and now. This temporal focus suggests both groups are actively engaged in innovation and forward-thinking initiatives while maintaining awareness of current developments.
Despite these shared patterns, they also show some gender-specific preferences for some key SMCs. Female entrepreneurs demonstrate a stronger focus on business-related topics (I2.1), work and employment (I3.1), sales (I2.2), and science/technology (Y1). Male entrepreneurs, conversely, show greater emphasis on telecommunications (Q1.3), general money matters (I1), and weather-related topics (W4). However, in order to examine whether they show statistically significant differences in the choice of these SMCs, the following section compares the two corpora with each other to identify their preferential key SMCs.
5.2. Analysis of preferential key SMCs
Table 3 shows the top 20 preferential key SMCs when comparing male and female entrepreneurs. The analysis of topic preferences reveals distinct patterns between female and male entrepreneurs. Female entrepreneurs show preferences for traditionally feminine topics, as evidenced by several key semantic categories. Their discourse prominently features references to women and gender (S2.1) with terms like women (67 occurrences) and female (5), musical themes (K2) including music (7) and musical (6), family relationships (S4) such as family (37) and mother (4), and work-related topics (I3.1) with terms like work (76) and career (9). This pattern aligns with Hu and Kearney’s [53] observation that despite increasing female participation in the workforce, traditional social role divisions persist, with women maintaining stronger connections to family and domestic spheres.
However, female entrepreneurs also demonstrate significant engagement with traditionally masculine topics. This is evident in their focus on science and technology (Y1) using terms like technology (39) and engineering (5), business matters (I2.1) with references to business (39) and companies (32), and leadership (S7.1+) through words like leadership (28) and CEO (13).
Male entrepreneurs, for their part, emphasize distinctly masculine topics in their communication. Their discourse centers on telecommunications (Q1.3) with frequent mentions of telegram (61), financial matters (I1) including money (14) and finance (7), health-related issues like smoking (F3), and governmental affairs (G1.1) with references to government (10) and President (10). They also show particular attention to financial transactions (I1.2) through terms like pay (6) and spend (6).
Interestingly, male entrepreneurs also engage with traditionally feminine topics, particularly in two areas: education (P1), using terms like students (13) and education (10), and weather-related topics (W4), with frequent references to climate (41) and various weather phenomena. This suggests that while gender-based topic preferences persist, there is some degree of cross-over in both directions.
The analysis reveals a nuanced pattern of topic preferences that transcends traditional gender expectations. Female entrepreneurs demonstrate strong engagement with traditionally masculine domains like technology and business, while maintaining presence in feminine-associated areas like music and employment. Similarly, male entrepreneurs show significant activity in masculine-coded areas like finance while also engaging with traditionally feminine domains such as education and environmental topics. This mixed engagement challenges traditional gender-based competency perceptions where men are typically associated with economy, military, technology, science, and security issues, while women are typically associated with education, healthcare, human rights, social welfare, and environmental concerns [44]. The findings suggest that both male and female entrepreneurs strategically navigate the “double bind”, balancing gender role expectations while engaging their followers across diverse topics. Rather than strictly adhering to gender-typical domains, both male and female entrepreneurs demonstrate flexibility in topic selection, perhaps reflecting the evolving nature of entrepreneurial leadership in contemporary contexts.
The key SMC analysis also reveals distinct communication patterns among female entrepreneurs that align with traditional gender-based leadership expectations [3]. Their language choices reflect traditionally feminine personality traits, particularly in three key areas of interpersonal communication, community orientation, and emotional expression.
In terms of discourse markers (Z4), female entrepreneurs show a strong preference for gratitude and supportive expressions, with thank_you appearing 90 times, followed by encouraging phrases like please (5), bravo (5), and well_done (4). This emphasis on acknowledgment and appreciation suggests a conscious effort to build and maintain relationships with followers.
The prominence of reciprocity-related terms (S1.1.2+) demonstrates a clear community orientation, with frequent use of sharing-related vocabulary (share: 33, sharing: 15, shared: 15). Similarly, their use of positive affect terms (E2+) is characterized by strong emotional expression through words like love (41), loved (17), enjoyed (10), and appreciate (6).
These patterns align with previous research findings that women tend to reference emotions more frequently than men [54] and show higher use of positive emotional expression [55]. These linguistic choices reinforce the construction of female entrepreneurs as more emotionally expressive, relationship-focused, and community-oriented than their male counterparts [56], particularly evident in their frequent use of gratitude expressions and positive emotional language.
Female entrepreneurs demonstrate a distinctive approach to follower relationships, characterized by an emphasis on collaborative investigation and inclusive observation. This is evident in their language patterns across three key semantic categories: research-oriented terms (X2.4) such as research (19) and analysis (4); diversity-focused language (A6.3) including diversity (13) and diverse (5); and observational verbs (X3.4) like see (67) and watch (13).
This communication style presents a notable contrast to female politicians, who often adopt more masculine, combative approaches in their campaigns [5,57]. Instead, female entrepreneurs favor cooperative and relationship-focused communication strategies with their followers. They tend to show a reluctance to impose their views on others.
Male entrepreneurs exhibit communication patterns that align with traditional masculine personality traits [58]. Their language choices reflect several distinct characteristics: competitiveness and confrontation (evidenced in negative markers like n’t [89], not [68], no [18] in Z6), task orientation (shown through terms like principles [13] and objective [2] in G2.2), and security consciousness (reflected in words like privacy [11] and encrypted [2] in A10-).
A notably impersonal communication style emerges through their frequent use of existential verbs (A3+) such as is (283), are (104), and be (83). This is further reinforced by their tendency to reference general objects (O2), using terms like things (25), products (11), and tools (5). These patterns collectively construct an image of male entrepreneurs as emotionally controlled, assertive, and authoritative [56].
The contrast between male and female entrepreneurial communication styles is striking. While female entrepreneurs tend toward relational and people-oriented approaches [58], male entrepreneurs focus more on objects and impersonal topics [59]. Female entrepreneurs often emphasize work-related topics and consciously incorporate masculine traits while maintaining attention to women’s issues. In contrast, male entrepreneurs engage with a broader spectrum of challenging topics from education and climate to economics and politics, approaching these subjects from a notably critical perspective. Unlike their female counterparts, male entrepreneurs appear less concerned with consciously displaying gender-typical traits, instead maintaining their focus on objective and impersonal communication styles.
5.3. Analysis of key part-of-speech (POS) categories
The analysis of key Parts of Speech (POS) categories, evaluated at a statistical significance level of p < 0.01 (cutoff value 6.63), reveals distinct linguistic patterns between male and female entrepreneurs (see Table 4). A notable finding is the disparity in distinctive POS categories—female entrepreneurs exhibit only 7 categories compared to male entrepreneurs’ 15. This suggests that female entrepreneurs tend to adopt communication styles that more closely align with male patterns.
Female entrepreneurs demonstrate characteristic usage of specific linguistic elements: possessive pronouns (APPGE), third-person singular pronouns (PPHS1), prepositions (II and IF), verb participles (VVG), singular common nouns (NN1), and articles (AT1). This pattern, particularly the prominence of pronouns, aligns with previous research identifying pronouns as typical markers of female authorship [43]. While the use of possessive pronouns might suggest an involved communication style, this characteristic appears relatively modest in the analysis.
Male entrepreneurs, by contrast, demonstrate a more information-focused communication style. Their language is characterized by diverse verb forms (VVZ, VD0, VDI, VDZ), conjunctions (CS), numerical expressions (MC, NNU), general adjectives (JJ), adverbs (RR), and pronouns (NN, NP1). Notably, their use of negation words suggests a tendency toward critical engagement and counterarguments [54].
5.4. Word association analysis
Figs 1 and 2 (Generated by the KH Coder) show the words which strongly co-occur with first person pronouns (I, we, and our) in the two corpora. Although both corpora have identified 11 themes, the two corpora show their different preferences for constructing male and female entrepreneurs.
In Fig 1, the most prominent theme is Theme 1. It consists of such words as I, we, be, that, have, that, want, help, share, principle, ask, so, many, celebrate, community, deeply, grateful, and gift. It suggests that male entrepreneurs tend to foreground themselves in sharing principles, celebrating achievement and showing gratitude. However, in Fig 2, I and we only appear in Theme 10, which consists of I, we, have, be, team, and proud. They foreground female entrepreneurs’ pride to be a team.
The different themes also suggest their different preferences for topics/issues. Male entrepreneurs tend to foreground the feminine topics of family support (Theme 2), equity (Theme 6), safety improvement (Theme 8), team and partnership (Theme 11). Besides, they also foreground the masculine topics of politics (Theme 10) and science and technology (Theme 3). Male entrepreneurs tend to underline the action taken to achieve the goals, as can be seen from Theme 4 (continue and call), Theme 5 (meet and US), Theme 7 (come and fight), and Theme 9 (spend and time).
By contrast, female entrepreneurs show their preferences for the masculine topics of science and technology (Theme 1), market (Theme 3), and leadership (Theme 4). However, they also address some feminine topics, including community development and stakeholder value (Theme 2), family (Theme 5), and employees (Theme 6). Nevertheless, instead of foregrounding action, female entrepreneurs tend to foreground emotion and belief, as can be seen from such words as want, allow, commit, commitment, excite, welcome, honor, proud, believe, announce, discussion, inspire and hope.
6. Discussion and conclusion
Therefore, the findings reflect a “double bind” with a twist in the gender performances of male and female entrepreneurs on social media, where both groups strategically adapt to traditional gender norms while leveraging social media’s interactive nature. Both male and female entrepreneurs emphasize helping, emotional expressiveness, gratitude, and future-oriented language. These shared performances demonstrate how social media encourages interactive and collaborate communication styles, which partially dissolves traditional gendered expectations of assertiveness for men and emotional expressiveness for women. This adaptive approach reflects the “twist”, where social media allows for blending traditionally masculine and feminine traits. Unlike female politicians who are supposed to adopt a more combative masculine style to foreground their competence and leadership [60], female and male entrepreneurs do not rely on confrontation but interaction and engagement to build connections and solidarity with their followers.
This can be seen from their topic preferences. Female entrepreneurs show strong preferences for traditionally feminine topics like family and community while also engaging with masculine-coded domains such as technology, business, and leadership. This reflects their need to balance feminine and masculine traits to gain credibility while maintaining likability. Male entrepreneurs, conversely, emphasize masculine topics like finance, politics, and science while also engaging with feminine themes such as education, climate, and equity. This suggests that male entrepreneurs tend to adopt traits associated with femininity to enhance follower engagement and relatability.
Their performances in personality traits and styles show even more complicated pictures. Despite both genders’ shared performances, Twitter, as a social media platform, continues to reproduce traditional gender stereotypes about gender-related personality traits [56]. Female entrepreneurs tend to foreground emotions (e.g., pride, commitment, and inspiration) and relational values, reinforcing their alignment with traditional feminine traits. Male entrepreneurs, on the other hand, emphasize actions and achievements, reflecting traditional masculine traits of goal orientation and assertiveness. Despite this, male entrepreneurs also integrate emotional elements like gratitude and community-building, showcasing a nuanced adaptation of feminine communication styles.
Therefore, both male and female entrepreneurs navigate the digital double bind strategically. Female entrepreneurs face the challenge of balancing professional credibility with societal expectations of femininity. They engage in “soft self-promotion” by emphasizing community and emotional values while subtly asserting their leadership and expertise. Male entrepreneurs must also navigate a version of the double bind, as adopting excessively feminine traits could undermine their credibility. Both groups strategically adapt their communication to build professional authority while maintaining relatability.
To sum up, the above findings suggest that while traditional gender norms persist in shaping entrepreneurial communication [56], social media provides a platform for more flexible and nuanced performances. By revealing the “double bind” with a twist, this study highlights how male and female entrepreneurs strategically blend masculine and feminine traits to meet the complex demands of digital entrepreneurship. This suggests a gradual evolution in gendered communication, where traditional norms are negotiated rather than entirely overturned [9].
This analysis ultimately challenges the traditional binary treatment of gender in computational models, suggesting that gender categories are normative rather than merely descriptive [43]. While gender remains a powerful force in structuring social interactions, its expression is more nuanced and complex than simple binary categories suggest, particularly in the digital entrepreneurial space. As Hu and Kearney [53] argue, gender differences shouldn’t be viewed purely negatively, but rather as potential vehicles for women to represent their distinct values and increase their visibility within the entrepreneurial space. This perspective suggests that the goal should not necessarily be complete homogeneity in entrepreneurial discourse, as such homogeneity might simply reflect the forced accommodation of marginalized groups to dominant norms. Instead, the preservation of certain differences might serve as a means for women to maintain their authentic voices while navigating the complex demands of digital entrepreneurship. Therefore, gender performances in other professions are worthy of a close examination in future studies [60].
References
- 1. Gherardi S, Poggio B. Creating and recreating gender order in organizations. J World Bus. 2001;36(3):245–59.
- 2. Jones JJ. Talk “like a man”: The linguistic styles of Hillary Clinton, 1992–2013. Perspect Polit. 2016;14(3):625–42.
- 3. Heilman ME. Description and Prescription: How Gender Stereotypes Prevent Women’s Ascent Up the Organizational Ladder. J Soc Issues. 2001;57(4):657–74.
- 4. Evans HK, Clark JH. “You Tweet Like a Girl!”. Am Polit Res. 2015;44(2):326–52.
- 5. Huddy L, Terkildsen N. Gender Stereotypes and the Perception of Male and Female Candidates. Am J Poli Sci. 1993;37(1):119.
- 6. Wang YF, Lee SK, Ye Q. Opinion leaders in eco-innovation diffusion: Analysis of information networks for waste separation in Shanghai. Res Conserv Recycling. 2021;174:105822.
- 7. Wright KAM, Holland J. Leadership and the media: Gendered framings of Julia Gillard’s ‘sexism and misogyny’ speech. Australian J Polit Sci. 2014;49(3):455–68.
- 8. Baxter J. Resolving a gender and language problem in women’s leadership: Consultancy research in workplace discourse. Discourse Commun. 2017;11(2):141–59.
- 9. Duffy BE, Pruchniewska U. Gender and self-enterprise in the social media age: a digital double bind. Inf Commun Soc. 2017;20(6):843–59.
- 10. Bachmann I, Harp D, Loke J. Covering Clinton (2010–2015): meaning-making strategies in US magazine covers. Feminist Media Stud. 2017;18(5):793–809.
- 11. McAdam M, Crowley C, Harrison RT. Digital girl: cyberfeminism and the emancipatory potential of digital entrepreneurship in emerging economies. Small Bus Econ. 2020;55(2):349–62.
- 12. Martinez Dy A. Levelling the playing field? Towards a critical-social perspective on digital entrepreneurship. Futures. 2022;135:102438.
- 13. Wang Y, Li Y, Wu J, Ling L, Long D. Does digitalization sufficiently empower female entrepreneurs? Evidence from their online gender identities and crowdfunding performance. Small Bus Econ. 2022;61(1):325–48.
- 14. Daniels J. Rethinking Cyberfeminism(s): Race, Gender, and Embodiment. wsq. 2009;37(1–2):101–24.
- 15.
Coates J. Women, men and language: A sociolinguistic account of gender differences in language. 3rd ed. London and New York: Routledge; 2015.
- 16.
De Lauretis T. The technology of gender. In: De Lauretis T, editor. Technologies of Gender: Essays on Theory, Film, and Fiction. Indianapolis: Indiana University Press; 1987. p. 1–30.
- 17. Caldeira SP, De Ridder S, Van Bauwel S. Exploring the Politics of Gender Representation on Instagram: Self-representations of Femininity. J Divers Gender Stud. 2018;5(1):23–42.
- 18. Wajcman J. Feminist theories of technology. Cambridge Journal of Economics. 2009;34(1):143–52.
- 19.
Plant S. Zeros and Ones: Digital Women and the New Technoculture. Doubleday; 1997.
- 20. Dy AM, Marlow S, Martin L. A Web of opportunity or the same old story? Women digital entrepreneurs and intersectionality theory. Human Relat. 2017;70(3):286–311.
- 21. Liñán F, Jaén I, Martín D. Does entrepreneurship fit her? Women entrepreneurs, gender-role orientation, and entrepreneurial culture. Small Bus Econ. 2020;58(2):1051–71.
- 22. Martiarena A. How gender stereotypes shape venture growth expectations. Small Bus Econ. 2020;58(2):1015–34.
- 23.
Fairclough N. Language and Power. London: Longman; 1989.
- 24.
Wodak R, Meyer M, editors. Methods of Critical Discourse Studies. 3rd ed. London: Sage; 2016.
- 25. Liu M, Zhao R, Feng J. Gender performances on social media: A comparative study of three top key opinion leaders in China. Front Psychol. 2022;13:1046887. pmid:36438362
- 26. Lazar MM. Feminist Critical Discourse Analysis: Articulating a Feminist Discourse Praxis1. Crit Discourse Stud. 2007;4(2):141–64.
- 27.
Baker P. Using corpora in discourse analysis. London; New York: Continuum; 2006.
- 28. Baker P, Gabrielatos C, KhosraviNik M, Krzyżanowski M, McEnery T, Wodak R. A useful methodological synergy? Combining critical discourse analysis and corpus linguistics to examine discourses of refugees and asylum seekers in the UK press. Discourse Soc. 2008;19(3):273–306.
- 29.
Partington A. Corpora and discourse, a most congruous beast. In: Partington A, Morley J, Harrman L, editors. Corpora and Discourse. Bern: Peter Lang; 2004.
- 30. Chen F, Wang G. A war or merely friction? Examining news reports on the current Sino-U.S. trade dispute in The New York Times and China Daily. Crit Discourse Stud. 2020;19(1):1–18.
- 31.
Morley J, Bayley P. Corpus-Assisted Discourse Studies on the Iraq Conflict: Wording the War. London; New York: Routledge; 2009.
- 32. Liu M. New trend, but old story: A corpus-assisted discourse study of Tibet imaginations in Anglo-American newspapers. Critical Arts. 2019;33(1):1–14.
- 33. Liu M, Li C. Competing discursive constructions of China’s smog in Chinese and Anglo-American English-language newspapers: A corpus-assisted discourse study. Discourse Commun. 2017;11(4):386–403.
- 34. Liu M, Zhang Y. Discursive constructions of scientific (Un)certainty about the health risks of China’s air pollution: A corpus-assisted discourse study. Language Commun. 2018;60:1–10.
- 35. Ming L, Ma J. The politics of fear in Hong Kong protest representations. J Lang Polit. 2021;21(1):37–59.
- 36. Wang G, Ma X. Representations of LGBTQ+ issues in China in its official English-language media: a corpus-assisted critical discourse study. Criti Discourse Stud. 2020;18(2):188–206.
- 37.
Poole R. Corpus-assisted ecolinguistics. London: Bloomsbury Publishing; 2022.
- 38. Newman ML, Groom CJ, Handelman LD, Pennebaker JW. Gender differences in language use: an analysis of 14,000 text samples. Discourse Process. 2008;45(3):211–36.
- 39. Windsor L, Mitchell SM, Osborn T, Dietrich B, Hampton AJ. Gender, language, and representation in the United States Senate. J Lang Polit. 2022;21(6):919–43.
- 40.
Duarte F. X (Formerly Twitter) User Age, Gender, & Demographic Stats (2025). 2025 [cited 2025 12 August]. Available from: https://explodingtopics.com/blog/x-user-stats
- 41. Lee J, Lim Y. Gendered campaign tweets: The cases of Hillary Clinton and Donald Trump. Public Relat Rev. 2016;42(5):849–55.
- 42. Joshi A, Sparks R, Karimi S, Yan S-LJ, Chughtai AA, Paris C, et al. Automated monitoring of tweets for early detection of the 2014 Ebola epidemic. PLoS One. 2020;15(3):e0230322. pmid:32182277
- 43. Bamman D, Eisenstein J, Schnoebelen T. Gender identity and lexical variation in social media. J Sociolinguist. 2014;18(2):135–60.
- 44. Sanbonmatsu K. Gender Stereotypes and Vote Choice. Am J Polit Sci. 2002;46(1):20.
- 45.
Biber D. Dimensions of register variation: A cross-linguistic comparison. Cambridge: Cambridge University Press; 1995.
- 46.
Biber D. Variation across Speech and Writing. Cambridge: Cambridge University Press; 1988.
- 47. Herring SC, Paolillo JC. Gender and genre variation in weblogs. J Sociolinguist. 2006;10(4):439–59.
- 48. Rayson P. From key words to key semantic domains. Int J Corpus Linguist. 2008;13(4):519–49.
- 49. L’Hôte E. New Labour and globalization: globalist discourse with a twist? Discourse Soc. 2010;21(4):355–76.
- 50. Higuchi K. A two-step approach to quantitative content analysis: KH-Coder tutorial using Anne of Green Gables (Part I). Ritsumeikan Soci Sci Rev. 2016;52(3):77–91.
- 51. Higuchi K. A two-step approach to quantitative content analysis: KH Coder tutorial using Anne of Green Gables (Part II). Ritsumeikan Soc Sci Rev. 2017;53.
- 52.
Alvesson M, Billing YD. Understanding Gender and Organizations. London: Sage; 2009.
- 53. Hu L, Kearney MW. Gendered Tweets: Computational Text Analysis of Gender Differences in Political Discussion on Twitter. J Lang Soc Psychol. 2020;40(4):482–503.
- 54. Mulac A, Studley LB, Blau S. The gender-linked language effect in primary and secondary students’ impromptu essays. Sex Roles. 1990;23(9–10):439–70.
- 55. Mehl MR, Pennebaker JW. The sounds of social life: a psychometric analysis of students’ daily social environments and natural conversations. J Pers Soc Psychol. 2003;84(4):857–70. pmid:12703653
- 56. Katila S, Eriksson P. He is a Firm, Strong‐Minded and Empowering Leader, but is She? Gendered Positioning of Female and Male CEOs. Gender Work Organization. 2011;20(1):71–84.
- 57. Lee J. ‘You know how tough I am?’ Discourse analysis of US Midwestern congresswomen’s self-presentation. Discourse Commun. 2013;7(3):299–317.
- 58.
Holmes J. Gendered Talk at Work. Oxford: Blackwell; 2008.
- 59. Park G, Yaden DB, Schwartz HA, Kern ML, Eichstaedt JC, Kosinski M, et al. Women are Warmer but No Less Assertive than Men: Gender and Language on Facebook. PLoS One. 2016;11(5):e0155885. pmid:27223607
- 60.
Liu M, Zhao R. Gender performances of male and female politicians on social media: A corpus-assisted discourse study. In: Zhu H, Debing F, Xinren C, editors. Social Identity and Discourses in China’s Digital Communication. London: Routledge; 2024. p. 69–90.