Designing forms for gender diversity and inclusion
Sabrina Fonseca

This is a great piece! Thank you for writing it; I’m glad people are thinking about the UX of gender selection.

I’ve got one bit to add from a horribly hard version of this I ran into at work a few years ago, for any UX designers who have to build products that work in multiple languages. (I was the person ultimately in charge of fixing this for Google)

If your product has to function not just in multiple languages, but across languages — e.g., if you have to form sentences like “[person] just sent you a picture” where that person has a different language setting than you — then the grammatical localization problem becomes tremendously harder.

If that person uses male or female pronouns, it’s pretty easy, but for anything else, it’s almost hopeless, because almost no languages have gender-neutral constructs for talking about people at all. English and Mandarin are really extreme outliers among languages; only pronouns have gender at all, and many (e.g. “you”) don’t even have that. Many more languages act like Romance languages, where pronouns, adjectives, and some verb forms (especially participles) inflect and have to agree in gender with nouns; at the extreme end are things like the Semitic languages, where nearly every form of anything inflects. In the Hebrew or Arabic version of that sentence, for example, “sent” has to have the same gender as [person] and “you” depends on gender — and both languages have exactly two grammatical genders.

That shouldn’t be too surprising; English-speakers have been at the forefront of the conversation about gender, and when combined with English’s very weak grammatical gender, that has made it the perfect place to start experimenting with pronouns and so on. The existing presence of a form like “singular they” (which was never uncommon! Consider how often you’ve said “they’re coming”) made it even easier. Experiments exist in a few other languages (e.g. the use of the suffix “-@” in Spanish), but they don’t map 1:1 onto analogous concepts in English, and they’re unfamiliar enough to most native speakers that they look like UI bugs, not language.

But if a user selected a pronoun like “xe” in English, or even just “they,” how would you translate this into other languages?

Fortunately, there’s a nice way out of this. Even though almost no language has a way to talk about “this person who is neither male nor female,” every language does have a way to talk about “that person on the other side of the room whose gender I can’t see clearly.” And this is something you can use in UI design.

This means that if your grammatical gender selection is limited to three states — male (he/him), female (she/her), and other (they/them) — it’s possible to both have visibility controls on gender (by defaulting to “other” whenever the user can’t see the other user’s gender) and make good UI sentences.

To make it work, you need some special instructions for your translators. Whenever sending off sentences with gender options in them (e.g., “[person/gender1] just sent [you/gender2] a picture,” which the translators need to turn into nine distinct sentences)

  • Tell them to interpret an “other” value of gender as “that person across the room whose gender you can’t see clearly.”
  • In the master example sentence, always give the examples using only male or female pronouns. The use of singular they in English is a nuanced idiom which many non-native speakers will miss, and if you give them an English master sentence “they just sent you a picture,” a good half of the translations you get back will be plural, not gender-neutral. Don’t confuse your translators.

And as the article above says, decouple “display gender” from “grammatical gender.” Grammatical gender needs to be one of those three options, or localization is hopeless; display gender can be just an abstract string. (In all of Google’s codebase, the display gender only shows up on your G+ profile, among a lot of other freeform text written in the person’s language of choice.) That means there’s no need to attempt to translate display genders, or create a list of “suggested genders,” which is a mug’s game if ever you’ve seen one — not only are ideas about gender rapidly evolving among a lot of different language speakers, but there are all sorts of very reasonable arguments over which terms are preferred or even offensive which there is no reason for your UI to get into.

(For example, while designing that interface, I came across a very thoughtful essay about why “transgender” was preferred and “transgendered” was offensive, and another very thoughtful essay arguing the exact opposite. My immediate response was that this was way outside of the space of things that I had any intelligent opinion about, and I couldn’t even imagine how I might try to localize either of those strings into German if they were actually enum selectors. Free-text strings are fantastic this way; there are lots of people for whom this is very important, and they can each choose exactly what they prefer, and it is no skin off anyone else’s back. As it should be.)

Since custom genders are still a pretty uncommon case, we simplified the UI a bit by merging things: at the top level, there’s a selector for “Male,” “Female,” “Decline to State,” and “Custom;” if you choose “Custom,” that opens up a pronoun selector (which gives example sentences in your own language) and a free-text field. Under the hood, those fill in a separate grammatical gender (a three-valued enum, default value OTHER) and an optional display gender; if the display gender isn’t explicitly set, it’s inferred from the grammatical gender and turned into a localized UI string.

The net result was a UI which lets anyone pick the gender they prefer, but doesn’t require everyone to fill in a lengthy form about it — something important because most people don’t have any interest in doing that.