OpenDoTT Studio Session 2: Localisation and Internationalisation

The second OpenDoTT Studio session on 31 August 2021 focused on localisation and internationalisation, and was co-hosted by Sarah Kiden (Marie Curie Research Fellow, Northumbria University) and Solana Larsen (Editor of the Internet Health Report, Mozilla). Our amazing panelists were Peiying Mo and Zibi Braniecki. At the start of the session, all attendees shared the number of languages they speak, noting that English was common among us, but collectively, we spoke more than 10 languages. Here are some highlights from the session.

What is the difference between internationalisation and localisation? 

From an end user perspective, internationalisation is the process that enables localisation. A metaphor would be if someone has a universal plug adaptor, they can travel without having to worry about electrical wall plugs or the wiring in buildings for countries they are visiting.

From a technical perspective, internationalisation encompasses two aspects: human and algorithmic/data driven. Internationalisation can be achieved simply by using an algorithm (e.g. convert a Jewish calendar to another format), while human translation and localisation can be likened to an iceberg (e.g. Firefox has over 10,000 unique strings used to create the interface but end users only see the final product). The litmus test for localisation is if something particular is being changed in a product for a particular group of users.

Globalisation is a broader term that covers issues from compliance, to language, legal, acumen, and marketing, among others.

 

 

To what extent do you need to build decisions into product design from the start?

Similar to planning for accessibility and security of projects, it should happen as early as possible. Building decisions into projects early helps to identify missing pieces of user interfaces and avoid coding assumptions into products. At Mozilla, and at many other large tech companies, designers and engineers are often based in North America, and may easily encode their ‘culture’ into projects even when they serve users worldwide. If they have not encountered other cultures personally, they may design products that only display 12 calendar months, assume weekends are on Saturdays and Sundays for everyone, or market products that are not relatable.

As designers/researchers, you are not required to know everything, said Braniecki. Instead, when you look at user interfaces, you grow the sensitivity to ask questions like: ‘Will it work in 300 [sic] countries?’, ‘Will this message translate?’ or ‘Is the emotional message I am sending consistent with the visual message?’

Are Mozilla products Universal-Acceptance ready?

Currently, Mozilla products support 80% of global language scripts (over 3,000 individual scripts) and over 200 locales, though not all of them are actively maintained. Mozilla is working with linguists and activists to revive and document more languages, as well as working with volunteer communities to localise projects and translate customer support and open source developer resources. This community driven approach has enabled Mozilla to maintain a high degree of global reach despite having a far smaller budget than big tech competitors.

Many language and anthropology projects are funded by companies like Microsoft, Apple, Google and IBM, who have the resources to grow language scripts themselves. This is one of the rare cases, where large corporations do something that does not immediately bring money.

Experience or advice on localisation efforts with connected devices/services like voice

Prediction: In about 10-15 years, most technology will be interacted with and directed by voice commands, which has a higher throughput than keyboards. There are some limitations as no voice systems currently have fully internationalised vocal capabilities. Producing (like computer speaking to a user) is much easier to achieve, and currently 50-80 languages are supported by various systems. The challenge for the industry is to train language models with different languages and accents to make them inclusive. The nuances are beautiful, messy and humane.


About the Panelists

Peiying-MoPeiying Mo is the Sr., Localisation Program Manager at Mozilla Corporation. She works on localisation for web projects, marketing, legal documentation and vendor management. Previously, she worked as the Localisation Program Manager for Media at Yahoo!.

 

Zibi-BranieckiZibi Braniecki is a Platform Internationalisation Tech Lead Manager at Mozilla working on Mozilla Platform (Gecko), product (Firefox) and Web Internationalisation and Localisation technologies. Zibi is one of the authors of the Fluent Localisation System and he represents Mozilla at TC39/ECMA-402 committee, MessageFormat 2.0 and ICU4X Work Group. When not in front of the keyboard, he’s captaining the Polish National Team in Ultimate Frisbee.