Language Reflects Societal Views: Men First, Women Last

As language users, we have a lot of flexibility in the way we use words. Only the rules of grammar might constrain the order of words used and consequently change the meaning of a sentence. The sentence “Romeo kissed Juliet” has a different meaning than “Juliet kissed Romeo” (and if you disagree because both kiss each other, try replacing “kissing” with the less friendly “killing,” and you get my point). But other than that, language users have a lot of flexibility. As long as language users obey the rules of grammar, they can express any word in any sentence they like in any order. Shakespeare could have written about “Juliet and Romeo” as easily as he could have written about “Romeo and Juliet.” It is therefore not surprising that researchers have sometimes argued that language is arbitrary. Language users are free to use any word in any combination at any time.

However, even though this may seem to make a lot of sense, when we look at the actual language that is produced, the presumed flexibility is far less obvious than what might have been expected theoretically. Take for instance the order in which male and female words are expressed in language—assuming for the time being that there are only two types of gender, masculine and feminine. We could take a male word and a female word and count the number of times with which the masculine word precedes or succeeds the feminine word. In fact, let’s take any two words, one being male, the other being female, and count the number of times these two words appear in a particular order in a five-word combination. And let’s look at all five-word combinations in a collection of one trillion words from the English language, all the language that Google has available.

If a language user has all the flexibility in the world to use any word combination, barring ungrammatical combinations, it is as likely that the order “ladies” precede “gentlemen” as that “gentlemen” precede “ladies.” And that does not quite turn out to be the case. One order is far more common than the other. Now, “ladies and gentlemen” are the polite conventional exception to the rule. When you look at a series of words, you actually find a very different and rather sexist pattern. In the majority of the cases, male words precede female words. That’s right, men go first and women go last, despite that there is nothing that stops a language user from reversing that pattern. If we compute the number of a series of men and women words, the pattern is obvious.

• "men" – "women" is 2.1 times more likely than "women" – "men"
• "husband" – "wife” is 3.4 times more likely than "wife" – "husband"
• "boy" – "girl" is 2.7 times more likely than "girl – boy"
• "father" – "mother" is 1.2 times more likely than "mother" – "father"
• "male" – "female" is 2.3 times more likely than "female" – "male"
• "brother" – "sister" is 3.8 times more likely than "sister" – "brother"
• "gays" – "lesbians" is 3.6 times more likely than "lesbians" – "gays"
• "king" – "queen" is 3.1 times more likely than "queen" – "king"
• "prince" – "princess" is 4 times more likely than "princess" – "prince"
• "actor" – "actress" is 10 times more likely than "actress" – "actor"

The question of course is why we find this sexist pattern? One answer has to do with frequency. Male words are more frequent than female words and it is often the case that what is frequent in language precedes what is infrequent. Common words precede uncommon words. But that leaves the question: Why are the male words more common than the female words?

Here, things get interesting. The order of words may say something about the way language is shaped, which in turn tells us something: How the perceptual world is shaped. Language users perceive the world and code it in such a way that frequent information precedes less frequent information. And that means that male concepts in the perceptual world are more frequent than female concepts.

But it may also work the other way around. When we hear or read male words preceding female words, we may (accidentally) map the linguistic information onto their perceptual views of the world. It is as if thought is reflected in language and language biases thought. This has been a controversial topic in psychology, but it is an interesting idea nevertheless. Perceptual views in society are encoded in language, views that may get reinforced by language.

It is the case that society is reflected in language, then we would expect that over the last few decades male dominance has been changing. That is, the difference between male-female word order versus female-male word order is then expected to become smaller over time because of societal changes. And that is exactly what we find. Perhaps not fast enough as we would like to see, but the proportion of male dominance in word order has been declining over the years. Take for instance the order of “men and women” versus “women and men” since 1960 and 2019 (2019 is the last year for which we can get this data). An interesting pattern emerges.

"Men and Women" versus "women and men" frequencies over time

Source: Max Louwerse

Now one should not draw conclusions from one item, in this case, one-word pair. It may be the case that “Women and Men” was launched as a fashionable clothing brand in 1975 and that is the reason for these patterns. A detailed and exhaustive comparison would require more than a post entry, but to show that this finding is not incidental, take the comparison between “male and female” versus “female and male.”

"Male and female" and "female and male" frequencies over time

Source: Max Louwerse

Essentially, the same pattern emerges. Over time, slowly but surely male dominance in word order reduces, but it has also plateaued. What does this tell us? First, that things are changing in language and in society. We know that male dominance in society has changed over the last six decades, and we also see that male dominance in word order has changed. Secondly, it says something about the arbitrariness of language. It is easy to say that language users can choose any word order they like, but in fact consistent patterns emerge with men going first and women going last. There is nothing to stop a language user to use a different pattern, but we don’t. And finally, it tells us something about the psychology of language. It tells something exciting: Societal views seem to be reflected in language.

And this raises exciting questions. Exciting questions for search engines, machine translation tools, and AI algorithms. But also for the psychology of language, as it sheds light on the question: How language reflects thought, and how thought may be biased by language.