Opinion

Why Twitter Swahili Translation Sucks

As an African, there are many things we are for sure left out of every single time. Whether it is done by purpose or accidentally, I cannot tell. For example, when you read about companies like Elon’s Tesla plan for more markets and you find no single mention of an African country. Or you try to subscribe to NatGeo Magazine and discover African countries have the most expensive charges. Or like when Black Panther premiered and Buzzfeed wrote about the Kisumu launch as the ‘African Premiere’. One would think all Africans were gathered in Kisumu.

The continued sidelining of Africa is seen even on Twitter. Despite having a very active, and growing user base, Twitter still don’t have a single office on the world’s biggest continent. And worse off, they still translate Swahili tweets as Indonesian!

I decided to take a Swahili tweet, translate it to English, then to Indonesian. Maybe the languages are related, and the translation engine isn’t to blame. Here:

I don’t see any similarity. I don’t know Indonesian, but there’s no way one can confuse these two languages.

Indonesian is the official language of Indonesia. Swahili is a lingua franca in Africa and serves as a national language in Kenya, Tanzania, Uganda, and the Democratic Republic of Congo. It is spoken in many more countries including: Rwanda, Burundi, Malawi, Mozambique, Somalia, and South Africa. In the Middle East, Swahili is spoken in the Sultanate of Oman, and in the United Arab Emirates. Yet still Twitter chooses to translate Swahili as Indonesian.

So I decided to come up with some wild reasons why Twitter may think Swahili is Indonesian. Take to mind that I understand the translation is done by Microsoft.

Here are some reasons I could come up with:

Teaching bias to Machines:

This isn’t something new. There are machines that don’t recognise faces of black people. Because they were mainly trained using white faces. They thus become racist machines. I understand that it is a bot that does the translation. Could it be that the bots have learnt that bias against Africa. How? Maybe no one fed it enough Swahili. Why? Because no one cares about African languages. Also because: just ignore Africa.

Ignore Africans

Maybe it is a rule to ignore Africa and we still have no clue. Many companies ignore Africa so let’s go with it. You would think Africans don’t have electricity to drive electric cars. Or we don’t have Space centres to launch rockets. Or out of the close to 1 Billion Africans, we can’t find employable people so that’s why they don’t set up offices on our lands. This could be the biggest reason why Twitter chooses Indonesian over Swahili. Ignore them.

Kill African Languages

There are so many African languages. Try using any of these languages on Twitter to ask for support from the company. Or try translating even 5 of these languages using their translation features. Nothing. They will tell you ‘translate from Polish, Finnish, Indonesian, and other ‘advanced languages”, rarely African. No company has shown the littlest of concern over our languages. Google included. I much prefer Facebook’s approach to Swahili translation: using users to make the translations more accurate.

I am sure very many Africans speak over 3 languages. Yet most of the languages we speak are still not represented online. And still not considered as languages when you apply for a job. Yes, let’s kill African languages and remain with English, French, Spanish and Simple Chinese.


It is funny that a company like Microsoft who sometimes greet you in Swahili on their services is still poor at recognising Swahili when contracted to translate tweets.

I may not always use Swahili to tweet but I want my language to be respected. And not just Swahili, I want to understand South African jokes too.

2 Comments

  1. Hi Dickson, I hear your frustrations and understand where you are coming from. It is not just African languages that are suffering, but most other languages of the world. Unless you are an essential commercial language due to GDP numbers, very few recognize you in this global world. We recently published a study on GDP that shows the trend in the shrinking long tail of languages. Take a look and let me know what you think: https://globalvis.com/2018/03/english-chinese-language-oligopolies/

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Back to top button