Want to learn how to avoid gender bias and sexism in translation? Pronouns can be tricky to navigate in some languages but hopefully, this blog will help.
Many languages use gender in a different way to English. If you studied French at school, you’d know that nouns are either masculine or feminine such as la chaussette (the sock) and le chapeau (the hat). This linguistic characteristic is known as ‘grammatical gender’.
In English, we don’t use grammatical gender, but traditionally we have used gender-specific pronouns such as ‘he’ and ‘she’. Although, in a bid to be more inclusive many brands are moving to the non-gendered ‘they’. But not all languages use gender-specific pronouns. Some languages use gender-neutral pronouns, with only gender-specific words like ‘mother’, ‘son’ and ‘man’ or ‘woman’.
For example, in Turkish, there is no ‘he’ or ‘she’, instead there is the gender-neutral ‘o’. Similarly, in Finnish ‘hän’ a third party gender-neutral pronoun is used. Several other languages including Armenian, Bengali, Hungarian, Japanese, Korean and Malay are also genderless languages.
This issue shouldn’t be a problem if your organisation is using an experienced native speaking linguist to translate your content, but be careful if you’re using machine translation and artificial intelligence (AI) tools. There are also security considerations when using free online translation tools.
Gender bias problems with machine translation
Historically, AI technology like Google Translate makes gender choices that reflect societal asymmetries. In plain English, this means that the neural algorithm searches for learning patterns in all the translation data that’s available online. From translated corporate websites to translated literature, it identifies the most commonly used pronoun associated with specific nouns or verbs.
Until recently, faced with the Turkish ‘o’, Google Translate translated ‘o bir doktor’ into English as ‘he is a doctor’ and translated ‘o bir hemşire’ as ‘she is a nurse’. In many ways, these translations are an unfortunate reflection of gender bias in society as a whole.
Google Translate announced last year the steps its taking to reduce gender bias in its translations. It now provides both feminine and masculine translations when translating phrases and sentences from Turkish to English, see below:
As Google acknowledges, gender bias is a problem for machine learning, and more work needs to be done to ensure that machine translation technology delivers better results. In fact, Google has also recently announced that its auto-suggestion tool for Google Mail will not suggest gender-specific pronouns. At this stage, the technology is not sophisticated enough to confidently recommend gender-specific pronouns without the risk of causing offence.
Steps to avoid gender bias and sexism in translation
With so many differences between languages, ensuring translations are free of gender bias can be a problem. In our opinion, it’s all about context. While a human translator knows whether to translate the Turkish ‘o’ or Finnish ‘hän’ into English as male, female or non-binary, machine translation tools like Google Translate aren’t advanced enough to put gender-neutral pronouns into context.
Human translators refer to the larger text to identify the gender of the subject, for example, a reference to ‘mother’ might clarify how a third party gender-neutral pronoun should be translated. And, if in doubt, a human translator can always find out for sure with a quick phone call or email to their client. This becomes even more important when producing gender-neutral language and the sensitivities surrounding this. It can also be useful if you need to localise your content or adapt the copy in a more creative way based on gender neutrality. Take a look at our blog ‘what is the difference between translation, transcreation and copywriting‘ to find out more.
That said, machine translation tools are a boom for any organisation that has large volumes of content to translate. They offer many benefits such as reducing translation times – they’re super fast – and translation costs are reduced as a result of the time savings.
However, we would always recommend that you use machine translation in conjunction with a human translator to proofread and check problematic elements such as gender.
My advice to you is to get an understanding of the linguistic characteristics of the languages your organisation handles, so you have an idea of any potential pitfalls. And if you need to translate content from a genderless language into English or another language, please don’t rely on machine translation to do the job. By all means, get the benefits of machine translation but also use a human translator to proofread and edit translations, so you don’t fall into the gender bias trap.