Ask a German to describe a bridge, and they are likely to use words like beautiful and elegant. Ask a Spanish speaker, and they will use words like big and dangerous. Now, ask them to describe a key. The German will say hard and heavy while the Spanish speaker will say lovely and intricate. Why? According to work by Boroditsky and co-authors, that’s because in German the bridge takes a feminine article and the key takes the masculine. And, as you may have guessed, the reverse is true in Spanish.
Why this may matter for our economic lives is the topic of a fascinating new paper by Pamela Jakiela and Owen Ozier. Jakiela and Ozier assemble an epic dataset to look at how gendered grammar might influence economic outcomes. Pulling data from language textbooks, academic work by linguists, historical records, and even going to ask native speakers and translators, they are able to classify 4,336 living languages, covering 99 percent of the world population. What they are looking for is languages where the grammar is gendered – those that assign a gender to things, including inanimate objects. As they put it “this feature of language forces gender into every aspect of life: for a speaker of a gender language, gender distinctions are salient in every thought and utterance; every object is either masculine or feminine because it is intrinsically linked to a word that carries a grammatical gender.”
Jakiela and Ozier take this gendered language data to global labor, education and (for a smaller set of countries) gender norms data. Strikingly, speaking a gendered language is associated with an 11.89 percentage point decline in female labor force participation, and this is robust to controlling for a range of geographic controls (including Alesina et. al.’s suitability for the plow) as well as continent fixed effects. Lest you are worried that this is something about overall labor force participation, they find similar results for the gap between male and female labor force participation: gendered grammar is associated with a 14.64 percentage increase in the gap between men and women.
For education the results are much less clear (but wait for the within country estimates, below). The results bounce around a bit and the clearest relationship is in the gap between women and men in primary school completion, which rises with gendered grammar (but only significant at 10 percent, with all controls included).
For gender norms, they use the World Value Surveys data to construct an index. With all geographic and continent controls, there are significantly more regressive gender norms in countries with gendered grammar – on the magnitude of one standard deviation. As Jakiela and Ozier indicate, this is about the distance in gender attitudes between Ukraine (at the 55th percentile) and Trinidad and Tobago (80th percentile). Within this index, gendered grammar is a significant predictor of more regressive attitudes in 7 of the 8 questions, including “when jobs are scarce men should have more right to a job than women” as well as men making better businesses executives and political leaders. Finally, they find that these results hold equally for the attitudes held by men or women.
So, these are cross country regressions, which may make some of us nervous. Of course, there are many more languages than there are countries in the world. This allows Jakiela and Ozier to look within countries and here they focus on four countries (Kenya, Niger, Nigeria, and Uganda) where this substantial within country data on the indicators they are looking at as well as at least one gendered and non-gendered indigenous language.
Within country, the labor results hold, and are actually somewhat bigger (women who speak gendered languages are 18 percentage points less likely to be in the labor force). And now education is significant. When they control for country-round fixed effects, age, and religion, speaking a gendered language is associated with a 22 percentage point decline in the likelihood that a woman completed primary school, and a 16 percentage point decline for secondary school. The result for the gender gap in educational attainment is consistent with this result – so it’s not being driven by overall lower levels of schooling (although it’s interesting to see that men who speak a gendered language are also less likely to finish primary and secondary school). One reason there may be more action in the education results for these within country comparisons is that in this analysis they’re focusing on countries where not everyone is completing school.
This is a lot of cross country work, with serious potential endogenity problems. So Jakiela and Ozier do a heck of a lot to show us robustness. For the cross country work they throw in a host of controls, including “bad controls” like GDP per capita. They also plumb the structure of language and how it evolves to deal with possible non-independence within families. And they beat up on their measure of gendered language a bit too.
The looming question here is causality. Jakiela and Ozier follow Altonji et. al. (2005) and Oster (2007) and find that unobservables would have to be 1.44 times more correlated with treatment than observables to explain the link between gendered grammar and women’s labor force participation. For the gender gap in labor force participation this number is 3.23. These numbers are even higher for the within-country analysis. In the end however, Jakiela and Ozier conclude that causality can’t be ironclad here – we can’t measure everything that might matter.
What of policy? This is clearly tricky, and Jakiela and Ozier thoughtfully offer us this:
"Languages are a critical part of our cultural heritage, and it would be inappropriate to suggest that some languages are detrimental to development or women’s rights. However, languages do evolve over time; the direction of their evolution is shaped by both individual choices (for example, whether to use gendered pronouns like “he” or “she” or gender-neutral alternatives such as “they”) and conscious decisions by government agencies (e.g. the Académie Française) and other thought leaders (e.g. major newspapers and magazines). Our results suggest that individuals should reflect upon the social consequences of their linguistic choices, as the nature of the language we speak shapes the ways we think, and the ways our children will think in the future.”
Why this may matter for our economic lives is the topic of a fascinating new paper by Pamela Jakiela and Owen Ozier. Jakiela and Ozier assemble an epic dataset to look at how gendered grammar might influence economic outcomes. Pulling data from language textbooks, academic work by linguists, historical records, and even going to ask native speakers and translators, they are able to classify 4,336 living languages, covering 99 percent of the world population. What they are looking for is languages where the grammar is gendered – those that assign a gender to things, including inanimate objects. As they put it “this feature of language forces gender into every aspect of life: for a speaker of a gender language, gender distinctions are salient in every thought and utterance; every object is either masculine or feminine because it is intrinsically linked to a word that carries a grammatical gender.”
Jakiela and Ozier take this gendered language data to global labor, education and (for a smaller set of countries) gender norms data. Strikingly, speaking a gendered language is associated with an 11.89 percentage point decline in female labor force participation, and this is robust to controlling for a range of geographic controls (including Alesina et. al.’s suitability for the plow) as well as continent fixed effects. Lest you are worried that this is something about overall labor force participation, they find similar results for the gap between male and female labor force participation: gendered grammar is associated with a 14.64 percentage increase in the gap between men and women.
For education the results are much less clear (but wait for the within country estimates, below). The results bounce around a bit and the clearest relationship is in the gap between women and men in primary school completion, which rises with gendered grammar (but only significant at 10 percent, with all controls included).
For gender norms, they use the World Value Surveys data to construct an index. With all geographic and continent controls, there are significantly more regressive gender norms in countries with gendered grammar – on the magnitude of one standard deviation. As Jakiela and Ozier indicate, this is about the distance in gender attitudes between Ukraine (at the 55th percentile) and Trinidad and Tobago (80th percentile). Within this index, gendered grammar is a significant predictor of more regressive attitudes in 7 of the 8 questions, including “when jobs are scarce men should have more right to a job than women” as well as men making better businesses executives and political leaders. Finally, they find that these results hold equally for the attitudes held by men or women.
So, these are cross country regressions, which may make some of us nervous. Of course, there are many more languages than there are countries in the world. This allows Jakiela and Ozier to look within countries and here they focus on four countries (Kenya, Niger, Nigeria, and Uganda) where this substantial within country data on the indicators they are looking at as well as at least one gendered and non-gendered indigenous language.
Within country, the labor results hold, and are actually somewhat bigger (women who speak gendered languages are 18 percentage points less likely to be in the labor force). And now education is significant. When they control for country-round fixed effects, age, and religion, speaking a gendered language is associated with a 22 percentage point decline in the likelihood that a woman completed primary school, and a 16 percentage point decline for secondary school. The result for the gender gap in educational attainment is consistent with this result – so it’s not being driven by overall lower levels of schooling (although it’s interesting to see that men who speak a gendered language are also less likely to finish primary and secondary school). One reason there may be more action in the education results for these within country comparisons is that in this analysis they’re focusing on countries where not everyone is completing school.
This is a lot of cross country work, with serious potential endogenity problems. So Jakiela and Ozier do a heck of a lot to show us robustness. For the cross country work they throw in a host of controls, including “bad controls” like GDP per capita. They also plumb the structure of language and how it evolves to deal with possible non-independence within families. And they beat up on their measure of gendered language a bit too.
The looming question here is causality. Jakiela and Ozier follow Altonji et. al. (2005) and Oster (2007) and find that unobservables would have to be 1.44 times more correlated with treatment than observables to explain the link between gendered grammar and women’s labor force participation. For the gender gap in labor force participation this number is 3.23. These numbers are even higher for the within-country analysis. In the end however, Jakiela and Ozier conclude that causality can’t be ironclad here – we can’t measure everything that might matter.
What of policy? This is clearly tricky, and Jakiela and Ozier thoughtfully offer us this:
"Languages are a critical part of our cultural heritage, and it would be inappropriate to suggest that some languages are detrimental to development or women’s rights. However, languages do evolve over time; the direction of their evolution is shaped by both individual choices (for example, whether to use gendered pronouns like “he” or “she” or gender-neutral alternatives such as “they”) and conscious decisions by government agencies (e.g. the Académie Française) and other thought leaders (e.g. major newspapers and magazines). Our results suggest that individuals should reflect upon the social consequences of their linguistic choices, as the nature of the language we speak shapes the ways we think, and the ways our children will think in the future.”
Join the Conversation