Putting Trans into Translation
Gender-fair language makes people angry. Manuel Lardelli has experienced it more than once. “You always get the question: What's the point of your work?”, he says. Manuel is doing research on the translation of non-binary language like the singular “they” from English into German and is currently completing his PhD at the University of Graz. His research involves looking at both the translation and post-editing process of texts written with non-binary language. He also co-organised a participatory workshop for machine translation experts, translation practitioners and non-binary and intersex community members, to establish strategies in which people who live outside the categories of male and female can be represented in translated texts. The adversity he faces when presenting his work is shocking to me.
“At my university I held a talk about how to integrate gender-fair language and a gender-fair attitude in your classroom.”, he continues “And there were some lecturers who were very resistant and made fun of me and my work. For example they wrote an email to the whole institute saying ‘We are so fortunate that we have very intelligent people who tell us we can't use male and female any more’. [...] They have this attitude that you are stupid because you are trying to promote some change or trying to make it visible. You explain that you can use for yourself whatever form you want. It is a basic matter of respect, that if a person tells you ‘I use the pronouns ‘them/them’ you use them. [...] It is difficult to get the point across. Many feel threatened, although I am not threatening them.”
Social hierarchy has always been expressed in human language, and the expression of gender can be seen as one aspect of this. On one end of the spectrum, there are languages that don’t have gendered nouns and pronouns: Finnish, for example, doesn’t use “he” or “she” but the pronoun “hän” to refer to people regardless of their gender. English has some gendered nouns (e.g. “sister” or “father”) and lots of gender-neutral ones (like “sibling” and “parent”) and both gendered and gender-neutral pronouns, not to mention a wide range of more or less established neo-pronouns (e.g. ‘ze’, ‘xir’ and ‘eir’).
While I was living in the UK, using singular “they” and gender-neutral terms for myself was simple enough. My two native languages, German and Russian, don’t make the task of avoiding binary gender as easy as that. In German, most nouns that describe humans are gendered either male or female and no established gender-neutral pronouns exist yet. Additionally, in Russian, all adjectives and past tense verbs carry binary gender markers. So in Russian, I can't say the sentence “I went to the shop” without declaring a binary gender for myself. Moving back to Germany I basically gave up on asking people not to gender me. Especially in my somewhat conservative workplace introducing neo-pronouns seems a much too daunting task. Sometimes changing the language means stepping back into the closet.
This is the conundrum that Manuel dives into with his research. During his master thesis he examined gender bias in machine translation, but when he started following feminist activists online he became doubtful about his prior work. “Some things that I’ve written in my master’s thesis about gender are not so true, because gender is not binary and we have many identities.”, he says “You look at gender bias papers and you always see a footnote: ‘gender is actually non-binary but we don’t know how to approach this thing’.” And so Manuel set out to approach it.
One of his experiments looks at various strategies of translating texts about the TV series “Sex Education” from English into German. “Non-binary people are presented in positive terms and it is not about the fact that they are non-binary but that they are featured in a TV series.”, says Manuel. While gender-neutral nouns and the singular “they” are used in English, several different strategies could be used in German: rewording the sentence to avoid gender, using special characters to indicate non-binary gender (e.g. “*” instead of a gendered word ending) or using a new declination system where a fourth gender is added. Manuel met with translators online, watched them translate and measured the time it took them to translate the sentences. As a comparison, he also watched translators post-edit already translated text to make it gender-fair. Afterwards he talked to them about their experience.
The results were surprising: He expected the usage of a new declination system to be very demanding in terms of time. After all, it requires learning a new set of pronouns and noun endings. But looking at the time measurements of translation and post-editing, there was only a small difference between this and the other strategies. “This can mean in terms of neo-pronouns people have to get used to it, but after that they could be not very demanding.”, says Manuel. On the other hand, the attitude of the participating translators towards gender-fair translation strategies varied widely. While some translators said they enjoyed the task, others said they hated it. One person even likened the usage of a new declination system to a destruction of language. Generally, professional ethos overrode personal concerns: If a client requested gender-fair translation, they would not refuse.
Machine translation, on the other hand, presents a different set of challenges. “We have to shift the paradigm," says Manuel. “We can not train machine translation to use a specific strategy, because we would need very large corpora where we implement the strategy. Instead, we should allow people to select a preferred strategy and identify their own pronouns. And this is tricky because it is not so easy to implement.”
Additionally, one of the main findings of Manuel’s research is the importance of context. Who is talking and who is the intended audience? A human translator can adapt to changing context in ways that a machine translation system can’t. “If I was translating for a general audience, I would maybe reword a sentence to make it gender-neutral unless it was really necessary to state that a person is, for example, gender-queer. If I was writing for the queer community I would use (as I would call them) ‘disruptive’ strategies. We see that someone is gendered in a particular way and it is on purpose because gender is very important in this case.”
For the future Manuel wishes for one thing in particular: Better translations. “I watch many TV shows and I see that nowadays there are many non-binary characters. But the translations are so awful. You don't understand what is going on because they use literal translation, they misgender all the time. So as a translator it is very important to me to find good products. I hope that through my work maybe some translators are now starting to introduce gender-fair language.” Better translations and a higher awareness of non-binary genders could be mutually reinforcing. “When I try to explain to my parents what I am doing I have to really go to the basics, saying: ‘there are not only men and women but there are also other people’. We need the basics explained. Maybe then we can find more suitable strategies to represent everyone in language.”
Manuel feels especially passionate about participatory and intersectional research, which is still largely missing from translation studies. Mostly, he sees his work as only the beginning of a much needed engagement with the topic. “I thought about it for a long time: Should I as a white cis-gender man carry out research in this field? I am not the best person, because I do face discrimination [as a bisexual man] but not this kind of discrimination. But at the moment there is so little research and someone should at least try and start [...]. It is important for me to start a discussion knowing that I am not the best expert because it is important to work with people from other disciplines, from other backgrounds, other genders and so on. I think this is the key.”
You can find more of Manuel’s research and writing here
This post was written by Sabine Weber. Sabine is a queer person who just finished their PhD at the University of Edinburgh. They are interested in multilingual NLP, AI ethics, science communication and art. They organized Queer in AI socials and were one of the Social Chairs at NAACL 2021. You can find them on twitter as @multilingual_s