Study 1 employed the National Corpus of Polish 2008-2012 (Pęzik, 2012; http://nkjp.pl/index.php?page=0&lang=1) as a representative sample of language data for the years 1999-2009 (see Formanowicz & Sczesny, 2016). In total, the national corpus includes over fifteen hundred million words. The sample consisted of linguistic data from the following sources: press (50%), books (29%), spoken language (10%), the internet (7%), and other written materials (4%). To search for feminine forms of job titles, we first obtained the official list of professions from the Polish Statistical Office, containing 2,592 professions (based on the most detailed specification of the Polish Ministry of Labor). We used the Polish dictionary contemporary to the corpus data (Dubisz, 2008) to classify all professions as having either only a masculine job title, both a masculine and a feminine form, or only a feminine form. From the initial list we excluded those professions that were listed as "other" (235 listings = 9.1% of the entire list) and 121 professions that were not included in the dictionary (4.7% of the total). We also excluded 318 professions for which no person was listed (12.3% of the total). Thus, the list consisted of 1,918 professions that were specific and in current use.

The search for job titles in the corpus was limited to single words, therefore several professions were collapsed into more comprehensive categories (e.g., 30 types of 'mechanics' and 89 types of 'teachers' were collapsed into the general categories of 'mechanic' and 'teacher', respectively). Moreover, we excluded professions that had a different meaning when shortened to a one-word term (e.g., 'insurance agent' - 'agent', 'master of production' - 'master'). Most importantly, we excluded professions whose feminine form could have a different meaning that would interfere with the corpus search (e.g., cukierniczka 'confectioner' (fem), which also means 'sugar bowl'). Interestingly, of 68 professions excluded due to such semantic asymmetries, 61 were working-class (45 with only a masculine job title and 16 with an available feminine form) and only 7 middle-class (6 with only a masculine job title and 1 with an available feminine form). This list comprised 354 professions, of which 131 (37%) had only the masculine form of the job title and 223 (63%) had both forms.

Each profession was categorized according to social class based on the distinction available on the website of the Polish Ministry of Labor, and specifically the classification of the professions based on the International Standard Classification of Occupations (ISCO) and the Polish indicator of the professional social standing in Poland (Domański et al., 2007). Specifically, the two first groups (Managers and Professionals) were classified as middle-class and the remaining groups as working-class (Technicians and Associate Professionals; Clerical Support Workers; Services and Sales Workers; Agricultural, Forestry and Fishery Workers; Craft and Related Trades Workers; Plant and Machine Operators and Assemblers and Elementary Occupations). To arrive at a clear distinction, we used only those professions that could be unequivocally categorized as either middle- or working-class. In those cases where multiple professions were collapsed into one, an educational level was assigned only when all professions shared the same status. Therefore, 20 professions were assigned no value, as they could belong to different categories: malarz 'painter', for instance, was categorized as both a high-status and a low-status profession in the original list, because the term can refer to both construction painters and artists. Thus, the final list consisted of 334 professions: 59% working-class professions (of which 47% had only the masculine form and 53% had both forms) and 41% middle-class professions (of which 62% had only the masculine form and 38% had both forms). The search in the Polish National Corpus included both singular and plural forms as well as different case forms of the feminine job titles. Feminine forms for the professions that do not have official feminine forms were coined according to the rules of the Polish language and based on similar existing professions (for instance, existing socjolożka 'sociologist' (fem) served as a model for non-existent but similar words such as geolożka 'geologist' (fem)). These feminine forms were then validated by means of an internet search to determine whether any Polish speakers use them. The use of feminine job titles was coded as follows: When no feminine title was used in the analyzed period, we assigned the value 0 (134 professions), and when at least one feminine form occurred, we assigned the value 1 (220 professions).

For professions with official feminine job titles, feminine titles were widely used in the text corpus; even for professions for which no official feminine titles exist, unofficial feminine titles were used (see Table 1).

A logistic regression with social class as the focal predictor and use of feminine job titles as the criterion, while controlling for the availability feminine job titles, tested whether the log odds of occurrences of feminine titles were related to the social class of professions. The regression was significant, χ²(2) = 168.34, p < .001, Nagelkerke R² = .54. As predicted, the social class of professions predicted the use of feminine job titles, B = 1.04, Wald(1) = 10.79, p = .001, while controlling for the availability of feminine forms, B = 3.86, Wald(1) = 91.40, p < .001. The odds ratio of 2.82 indicated that feminine job titles were used about three times more often for middle-class professions than for working-class professions in the Polish National Corpus.

To rule out the possibility that differences in GIL in lower- or middle-class job roles are due to gender segregation, we mapped each profession onto the percentage of women in each category, using data from the Polish Statistical Office, covering 2004-2008 (Wage Structure by Profession; Polish Statistical Office, 2004-2008). Since many job titles in the Polish National Corpus are quite detailed, we aggregated several of them into broader categories for the purposes of this analysis. For instance, we grouped specific medical titles such as cardiologist, endocrinologist, and general practitioner under the broader category of Health Professionals (excluding nurses and midwives), which has a percentage of women of 63.9%. Similarly, titles like editor or economist were coded under the category Social Sciences and Related Professionals, where the percentage of women is 80.3%. In total, we successfully coded 343 professions, which accounts for approximately 87.1% of the total professions (see Supplementary Materials for further details, Hodel et al., 2025).

Subsequently, we ran a logistic regression with social class as the focal predictor and the use of feminine job titles as the criterion, while controlling for the availability of feminine forms and the percentage of women in each profession. This analysis tested whether the log odds of occurrences of feminine titles were related to the social class of the profession. The regression was significant, χ²(2) = 140.57, p < .001, Nagelkerke R² = .53. Again as predicted, social class of profession was a significant predictor of the use of feminine job titles, B = 0.93, Wald(1) = 7.135, p = .008, while controlling for the availability of feminine forms and the percentage of women in each profession, B = 0.00, Wald(1) = 0.00, p = .997. The odds ratio of 2.54 indicated that feminine job titles were used about two and a half times more often for middle-class professions than for working-class professions in the Polish National Corpus.

Study 1 found that the social class of professions predicted the use of feminine job titles in the national corpus of Polish, at a time when GIL had hardly been promoted and masculine generics represented the norm (Koniuszaniec & Blaszkowska, 2003). This difference was evident regardless of whether the forms concerned were novel (no official feminine form existing) or well-established (official feminine form existing) in the Polish language. The difference found may contribute to a higher visibility of middle-class women than working-class women in a variety of professions.

Study 2 examined the dataset of German job titles in Swiss job advertisements, already used in a cross-cultural study published by Hodel et al. (2017). This dataset contained a considerable amount of masculine and gender-inclusive job titles (the dataset for Polish could not be used as it only contained 20 (6%) gender-inclusive job titles). The German dataset included job advertisements taken from the online job search engine careerjet.ch (downloaded in June 2012). This engine provided a categorization of branches of which three branches had been selected, namely health care, restaurants and food services, and constructional steel and metal work. All three branches comprised working- and middle-class professions. For each of the three branches more than 1,000 job advertisements were listed (ranging between 1,440 for constructional steel and metal work and 7,056 for health care). The dataset included 120 randomly selected job advertisements per branch. Two independent raters coded the linguistic form of the job titles as either masculine (e.g., Koch ‘[male] cook’) or gender-inclusive (e.g., Koch/ Köchin ‘[male] cook/ [female] cook’; for more information on the coding of the linguistic forms, see Hodel et al., 2017).

Social class was coded by two new independent coders, based on the European Socio-economic Classification scheme (ESeC; Wirth & Fischer, 2008), which enables the classification of each profession regarding its socio-economic status into one of nine categories (see Table 2).

The ESeC categories correspond with the ISCO classification applied in Study 1. In line with Study 1, categories 1 and 2 were coded as middle-class (e.g., managers, higher technicians) and categories 3 to 9 were coded as working-class professions (e.g., lower technicians, routine workers).

We excluded 20 job titles as they could not be clearly classified as masculine or gender-inclusive and 11 job titles as they could not be clearly classified regarding social class (e.g., internship in healthcare, security and environment). The final dataset consisted of 329 job titles of which 66% were working-class professions (for which 39% used a masculine form and 61% a GIL form) and 34% were middle-class professions (for which 6% used a masculine form and 94% a GIL form).

A logistic regression with social class as the predictor and language use (GIL vs. masculine) as the criterion that tested whether gender-inclusive job titles are related to social class of the profession was significant, χ²(1) = 47.41 p < .001, Nagelkerke R² = .19. As expected, the social class of a profession was a significant predictor of the use of feminine job titles, B = 2.27, Wald(1) = 30.02, p < .001. The odds ratio of 9.68 indicated that gender-inclusive compared to masculine job titles were used ten times more often for middle-class professions than for working-class professions.

Subsequently, to rule out the possibility that differences in GIL in lower- or middle-class job roles are due to gender segregation, we used the ISCO-19 classification (Swiss National Statistical Office, 2010-2012) and calculated the percentage of women in each profession (see Supplementary Materials, Hodel et al., 2025). A logistic regression with social class as the predictor, percentage of women in each profession as the control variable, and language use (GIL vs. masculine) as the criterion was conducted to test whether gender-inclusive job titles are related to the social class of the profession. The model was significant, χ²(2) = 70.21, p < .001, Nagelkerke R² = .286. Again as predicted, social class of profession was a significant predictor of the use of feminine job titles, B = 1.63, Wald(1) = 13.43, p < .001, while controlling for the percentage of women in each profession, B = 0.02, Wald(1) = 18.83, p < .001. The odds ratio of 5.08 indicated that the odds of using gender-inclusive job titles, compared to masculine ones, were five times higher for middle-class professions than for working-class professions.

Study 2 documented the social class difference for gender-inclusive job titles in job advertisements in Switzerland, where the use of feminine forms has been well established (Bußmann & Hellinger, 2003) and official GIL regulations have been adopted in many contexts, for example, in job advertisements (Hellinger, 2004). The fact that GIL is used less often for working-class professions in job advertisements may deter women from applying for these positions.

As the social class difference of GIL use has been documented for two grammatical-gender languages, it can most likely be generalized to other grammatical-gender languages. Generalization to natural gender or genderless languages, however, may be limited, as the introduction of gender-inclusive forms in these languages can easily be made through neutralization (in natural gender languages) or by simply avoiding gender references at all (in genderless languages; Sczesny et al., 2016). Future research should therefore investigate whether social class differences can also be found when using other language cues in natural and genderless languages, such as grammatical prominence (e.g., male role nouns more often perceived as agents than female one) or word order (e.g., male role nouns first; for an overview on linguistic gender cues see Formanowicz & Hansen, 2022).

The data used for this research were collected between 1999 and 2012, raising the question of whether the social class difference we found is still observable today. In recent decades, developments in GIL, including new forms of job titles, have taken place in both languages under investigation (e.g., Cieszkowski, 2015; Dargiewicz, 2021). For instance, in German job advertisements, the masculine generic form (e.g., Ingenieur ‘male engineer’) can nowadays be expanded by the abbreviation m/f/d (male/female/diverse) and is recommended by the European Parliament to indicate that the job offer is addressing people of all genders (European Commission, 2008; although this form of job title is classified as gender-inclusive, it fails to reduce male bias (Fatfouta & Sczesny, 2023).

Do such developments challenge the validity of the social class difference in GIL use we identified? This would be the case if GIL nowadays represents the standard in a speech community, that is, if it were widely accepted and commonly used. Under such conditions, the difference between middle-class and working-class professions should be less pronounced or even eliminated. However, the social acceptance of GIL (and consequently its use) is still rather low in both the countries under investigation. For instance, in current surveys, only 14% of 160 Polish respondents (Ivanová & Kyselová, 2022) and 24% of 30,754 Swiss respondents (Tamedia, 2023) were in favor of GIL. Thus, in both countries, we would still expect to find the social class difference in more recent datasets. Future research, nevertheless, needs to confirm the social class difference in GFL use, and may also check for cross-cultural variation in levels of acceptance of GIL.

Obviously, legal frameworks and policies that aimed to promote gender-inclusive language across Europe (e.g., Gender-sensitive communication: A guide to non-discriminatory language; European Institute for Gender Equality, 2020) have not been resounding. Nevertheless, there are ongoing political debates about the representation of sexual and gender minorities in language that have already resulted in adjustments of legal frameworks in some European countries (e.g., the German Civil Status Law granted intersex individuals in Germany the ability to select "diverse" instead of "male" or "female" when indicating their sex in civil status records at registry offices, Federal Anti-Discrimination Agency, 2018), which might inspire further advancement towards GIL across Europe. Therefore, our findings serve as a starting point for the observation of social class differences in GIL use over time and across languages. They thereby serve as an indicator of the success of language reforms that aim to promote gender equality for both social classes in grammatical-gender languages.

A more complex explanation of the effects of social status on GIL arises from the observation that working-class jobs often display more pronounced gender segregation compared to middle-class occupations (England et al., 2020). In the present research it was, however, not appropriate to analyze the impact of gender segregation because the datasets used comprised of detailed professions (e.g., psychiatrist for children, physician for internal medicine, physician for anesthesia) and data on gender segregation was only available on an aggregated level (e.g., medical doctors). Future research therefore needs to disentangle the influences of social class and gender segregation on GIL use and explore how the interplay of social class and gender segregation impacts GIL use.

Finally, we proposed that the social class difference in GIL use may be caused by middle-class women who are more engaged in collective action towards gender equality on behalf of their group’s interest than working-class women (Freeman, 1995; Scarborough & Risman, 2020). Thus, middle-class women’s collective actions may have greatly impacted the more prevalent use of GIL for their professions. As the aim of the present research was to identify the social class difference but did not test whether this difference is an outcome of collective action, future research should directly test whether women’s collective action for GIL depends on their social class. Moreover, the underlying psychological mechanisms of collective action on GIL await further investigation. Besides social class differences in egalitarian values (Carvacho et al., 2013; Kalmijn & Kraaykamp, 2007) and belief that social change is possible (Fernández-Ballesteros et al., 2002), system-based anger (Jost et al., 2017) or identification with women as a group (Radke et al., 2016) might also contribute to the difference in the promotion of GIL by middle- and working-class professions.

The successful implementation of GIL plays an important role in achieving gender diversity as such reforms aim to reduce androcentrism (Bailey et al., 2019). By creating a psychologically safe workplace (Rosenberger & Claypool, 2023; Stout & Dasgupta, 2011), GIL may thereby contribute to improving organizational effectiveness (e.g., Ely & Thomas, 2020). However, the fact that GIL is less common for working-class than middle-class professions indicates that this reform is not universal across all speech-communities, potentially limiting working-class women’s professional opportunities. Importantly, although working-class women may prefer the status quo, they would nevertheless profit from a successful implementation of GIL in various ways, for instance, by becoming more visible (Stahlberg et al., 2001) and by increasing their job opportunities (Bem & Bem, 1973; Yavorsky, 2019).

Given that GIL use is more common for middle-class professions than for working-class professions, organizations and policymakers need to focus on promoting GIL within working-class communities and professions. This could involve revising language used in official documents, educational materials, and workplace communications to ensure that GIL is normalized and actively encouraged across all social strata. For instance, human resource departments who recognize the lack of GIL use in working-class professions may tailor job advertisements for these professions accordingly to attract a more diverse pool of applicants. GIL will thereby contribute to the reduction of masculine defaults (Cheryan & Markus, 2020) and contribute to workplace cultures towards greater equality and inclusivity.

Social class has received limited attention from linguists and social psychologists in past research and turned out to be an intriguing yet underexplored factor in the present research. This highlights the importance of taking a closer look at the promotion of GIL, as its implementation has proven to be more successful for middle- than for working-class professions. To increase the successful implementation of GIL, gender equality must be perceived as important and achievable in working-class communities. Fostering egalitarian values and beliefs in successful social change in working-class communities might thus provide a more impactful means of implementing GIL and other gender equality initiatives. In doing so, working-class women would also profit from GIL language reforms and have greater access to more diverse job opportunities in the future.