Highly gendered languages are linked to larger personality differences between men and women

The way a language encodes gender suggests a subtle link to how its speakers view their own personalities. A recent study published in the Journal of Personality and Social Psychology provides evidence that people who speak highly gendered languages tend to report larger differences in personality traits between men and women. These findings suggest that the grammatical and structural rules of our native tongues might shape our self-concept and the way we experience gender.

Language acts as a framework that people use to represent and categorize the world around them. Historically, thinkers have explored the idea of linguistic relativity. This concept proposes that the vocabulary and grammatical structures of a language provide an invisible scaffold that influences the way its speakers think and behave.

Previous evidence suggests that language can shape how people perceive concepts like time, space, and color. Recent research even hints that implicit gender stereotypes tend to align with the linguistic associations found in a person’s native language.

Different languages handle gender in very different ways. Some languages, like English, are considered natural-gender languages because they primarily use gendered pronouns like he or she but do not assign a gender to everyday objects. Other languages, like Spanish or German, feature grammatical gender.

In these languages, nouns are arbitrarily categorized as masculine or feminine, and this assignment changes the structure of surrounding words like adjectives and articles. Finally, there are genderless languages, like Estonian or Turkish, which do not use gendered pronouns or gendered nouns.

Study authors Roxana Hofmann and René Mõttus, affiliated with the Department of Psychology at the University of Edinburgh and the Institute of Psychology at the University of Tartu, wanted to explore how these linguistic categories map onto psychological self-perceptions. In doing so, they noticed vast linguistic diversity and sought to connect language and self-evaluation.

“We heard about these studies connecting language and perception, for example, of time, color, and objects,” Hofmann said. “They are tentative, but we were curious to see whether a similar association might exist for how people evaluate their own behaviors, thoughts, and feelings, in other words, personality traits.”

“Gender is such a distinguishing feature of languages, and gender differences in personality vary so much around the world, so we were wondering if there is a connection,” Hofmann explained.

“When assessing language genderedness, we were also surprised by how much linguistic genderedness varies around the world,” Hofmann added. “For example, we used a language typology commonly used when comparing language genderedness that classifies languages into ‘genderless’, ‘natural-gender’ and ‘gendered’.”

“Gendered languages are languages with grammatical gender, but even those vary in their degree of genderedness, such as the number of word classes that require gender agreement or the phonological transparency of gender cues,” Hofmann continued.

She noted her own background as an example of this variation. “For example, I grew up speaking German and Bulgarian. Both have grammatical gender (‘table’ is masculine in German and feminine in Bulgarian), but Bulgarian has much more pervasive and predictable gender agreement. In terms of word classes, articles and adjectives must agree with the noun’s gender in German, whereas in Bulgarian agreement extends to past participles too.”

“In terms of transparency, noun gender in Bulgarian is much more predictable from phonological cues compared to German,” Hofmann added. “It was interesting to learn about all the different ways languages encode gender, or don’t. My supervisor’s native tongue is Estonian, which doesn’t even distinguish gender in pronouns, making gender often ambiguous in everyday speech, even when referring to people.”

To explore this relationship across cultures, the researchers analyzed a massive existing dataset of personality assessments. This dataset included exactly 755,307 participants aged 18 to 99 from 122 different countries. These participants represented 48 dominant languages, ranging from globally widespread languages like Spanish and English to more regional ones like Luganda and Latvian.

Though they came from diverse backgrounds, all participants completed the personality test in English. Holding the assessment language constant helped the scientists avoid potential translation biases. If the questionnaire had been translated into 48 different languages, subtle shifts in wording could have skewed the answers. Instead, any personality differences observed among the groups could be attributed to the dominant native language spoken in the participants’ home countries.

The scientists used the IPIP-NEO-120, a widely recognized psychological questionnaire that measures five major personality traits. These traits include Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. Neuroticism refers to a person’s tendency to experience negative emotions like anxiety or sadness, while Openness describes a willingness to try new things and entertain new ideas. Extraversion involves outgoing and energetic behavior, and Conscientiousness measures organization and responsibility.

By analyzing the scores, the authors calculated the average standardized differences in these traits between men and women for each language group. They used a statistical measure that quantifies the exact gap between two averages. This allowed them to confidently compare the size of the gender divide across completely different cultures.

“We wanted to highlight how grateful we are for all the people who made the data public, which allowed us to test our research question,” Hofmann noted. “This included the anonymized personality trait data assessed across multiple countries, language typologies from the World Atlas of Language Structures, and also pretrained word embeddings that allowed us to estimate the gender association of words across multiple languages. It shows how valuable open access practices are for psychological research.”

Next, the researchers needed to measure exactly how gendered each of the 48 languages was. Beyond expert classifications, they used natural language processing programs to analyze huge databases of text from movie subtitles and Wikipedia articles. These programs evaluated word embeddings, which are mathematical representations of how frequently words appear near one another in everyday sentences. The basic idea is that words sharing similar meanings will naturally show up in similar conversational contexts.

By calculating the mathematical distance between personality-related words and gendered anchor words like man or woman, the scientists estimated the implicit gender associations built into the everyday vocabulary of each language. If a language frequently placed nurturing words near feminine pronouns, the program captured that bias. This provided a quantitative measure of how much a culture’s written and spoken media links specific traits to specific genders.

The authors also prompted three advanced artificial intelligence models, specifically ChatGPT-5.2, Gemini 2.5 Pro, and DeepSeek-V3.2, to rate the 48 languages. The models were asked to score each language on a scale from zero to one based on how strongly its grammar and vocabulary encode feminine and masculine distinctions. All three models provided highly consistent ratings that aligned with the expert classifications and the text database analysis.

The data revealed a robust association between the genderedness of a language and the magnitude of personality differences between men and women. “We were surprised to find such a strong association between language genderedness and personality gender differences,” Hofmann shared. Speakers from countries with gendered languages tended to show significantly larger gender differences in personality traits compared to speakers from countries with genderless languages.

Specifically, the researchers found that these personality differences were most pronounced in social and emotional traits. The trait of Agreeableness, which reflects how cooperative, trusting, and compassionate a person is, showed the strongest link. In countries where the dominant language heavily encodes gender, the gap in Agreeableness scores between men and women tended to be much wider.

The researchers also checked to see if broader cultural norms could explain these variations instead of the language itself. “The take-away is quite simple: We showed that gender differences are on average larger in countries in which the dominant native language is gendered,” Hofmann told PsyPost.

“By gendered, we mean the extent to which each language’s grammar and lexicon encode binary (feminine/masculine) gender distinctions and associations,” Hofmann explained. “Importantly, language genderedness was associated with gender differences in personality, even after controlling for cultural differences in, for example, values toward individualism, masculinity, or attitudes toward gender equality.”

Providing context for the size of these gaps, Hofmann explained that psychological differences between men and women are generally subtle. “Gender differences are typically small and only detectable in averages,” she stated. “We previously estimated an association between gender and average personality domains ranging from r = .05 (Thailand) to .22 (Bulgaria), with U.S. adults somewhere in the middle at r = .11. For reference, the correlation between gender and height is estimated at r = 0.67 in the U.S.”

Despite the modest size of the psychological gaps themselves, their connection to language structure was highly consistent. “So men and women are generally similar around the world, but to the extent that differences do exist, we can predict them quite well from linguistic features related to gender, ranging from r = .51 to .59 for different measures of language genderedness,” Hofmann continued. “In other words, the size of the personality gap is small, but how well it lines up with a language’s genderedness is strong.”

While these patterns are noticeable, the study design has certain limitations. “There are two misinterpretations we would like to highlight,” Hofmann warned. “First, we cannot say anything on the individual level; the association we describe refers to group-level differences.”

“On the personality side, we quantified average differences between men and women who likely speak the same language (language was inferred from country identification),” Hofmann elaborated. “On the language side, we quantified the average gender association in words (for lexical gender, as grammatical gender features do not vary within languages). And both are associated with each other.”

The cross-sectional nature of the data also prevents researchers from drawing definitive conclusions about what causes what. “Second, we cannot say anything about the direction of causality, that is, whether language influences personality or the other way around,” Hofmann explained. It remains possible that gendered languages cause people to adopt more gender-typical personalities, but a society with historically strong gender differences might have also naturally developed a language that reflects those divisions over time.

Another limitation involves the way the researchers determined language use. They assumed that participants spoke the dominant language of the country they identified with. In highly multilingual countries, or in cases where the official language does not match the locally spoken dialect, this assumption could introduce some measurement errors. Additionally, because all participants completed the personality assessment in English, the sample only includes people with a relatively high level of English proficiency.

Future research aims to address these causal questions directly. “We would really like to figure out the direction of causality,” Hofmann said. “There is no straightforward way to do so, but we have some ideas to gain some clarity about language’s unique effects. For example, it would be interesting to test the consequences of gender-neutral language reform shifts (e.g., gender-neutral job titles replacing gendered ones, such as police officer instead of policeman).”

“We would also be curious whether individual differences in exposure to gendered language (e.g., media consumption, schooling) predict gender-personality differences,” Hofmann noted. “Another interesting avenue would be to assess bilinguals whose two native languages differ in genderedness, one gendered, one genderless, and compare their personality scores across the two languages to see whether gender differences are larger when assessed in the gendered one.”

The study, “Speaking of Gender: Language Genderedness and Its Association With Gender Differences in Personality Across 48 Languages,” was authored by Roxana Hofmann and René Mõttus.