Wikimedia Blog/Drafts/The design story behind the Universal Language Selector

[archived]

Title

The design story behind the Universal Language Selector

Body

Language is our vehicle to communicate knowledge. Allowing users to read and write in their own languages is essential for Wikimedia projects. Unfortunately many languages are not properly supported by most computers. This prevents many users to display content and/or type in their own languages. In order to overcome these language barriers, the Universal Language Selector was deployed to the Wikimedia projects recently.

Designing language tools is challenging for different reasons:

• Different users have different needs. Some users do not need language tools at all, while other users need them for the most basic activities.
• We need to support many languages. The solutions should work for more than 300 languages.
• We cannot speak all languages. We need to interact with different language communities to understand their different needs.

We faced the above challenges when designing the Universal Language selector and we are providing some insights on the design process we followed and the rationale for some of the design decisions.

Finding the right balance

Language is at the core of our experience as users. We use language for the most simple interactions such as reading, searching, typing or filling a form. For users that need the tools these tools should be easy to find when needed. For users that do not need the tools, they should not intrude in their way.

The Universal language selector supports different scenarios:

• Display content regardless of the availability of fonts on the user machine.
• Select and switch input methods. We found that input tools were easier to find wehn they were provided closer to the editing area since it avoided additional scrolling or navigation steps.
• Customise the User Interface language.

During our usability testing sessions we saw great interest from users of poorly-supported languages since they identified clearly the problem they were aimed to solve. For example, a Nepali user commented: "This looks very good and promising because in Nepali Wikipedia users don't find the input settings they want [...] and get away. "

Nevertheless, these scenarios are not only relevant for poorly-supported languages.

• Supporting mixed content.
• Accessibility.

Discoverability

Different wiki projects support multiple languages in different ways.

Wikis without interlanguage links such as Commons, show a language selector at the personal toolbar.

However, we found some problems when we tested this approach on wikis with interlanguage links such as Wikipedia. The main use case related to language for those wikis is to access information in a different language (as opposed to change the language of the menus), which is covered by interlanguage links. Introducing a new language-related area for those wikis for which already one exists, was confusing for users. Especially since the advanced tools were presented in a location that was more prominent than the main content selection scenario which is covered by the interlanguage links.

Recovery

We wanted our language tools to be safe for exploration. So that, if you make a mistake selecting a foreign language you are not left alone in a sea of uninteligible text.

• Once the User Interface language is changed, a tooltip informs you of the change with an option to go back to the previous language.
• Changes for language and font selection are previewed immediately so that the user does not have to accept in order to discover possible mistakes.
• Graphical aids are given so that you can navigate in a foreign language. After the initial use, most users were able to operate the Universal Language Selector even in a language they do not understand.

Fluency

• Anticipate user needs.
• Forgiving input.
• Avoid technical decisions. In general, users don't think in terms of content language and UI language as two separate things. Presenting an explicit selection of UI vs. content at the same level forces a decision on them which is based on technology terms instead of user terms. By making the common actions more prominent, and using terms such as "language of menus".

Repetition

For input methods, we spited the choice between input language and input method. There are many input methods for some languages, but users are likely to use one of those. In this way we can allow the user to switch between languages and provide the preferred input method based on previous selections.

Involving users from different language communities

One of the main problems when designing language tools is that the users that need them the most are uses whose only language is a poorly-supported language and are not technical-savvy. Those users cannot use an alternative language, cannot install fonts or input methods, and cannot easily communicate their problems to us.

32 users, 17 locations, 30 languages.

This is a barrier for users to consume and contribute their knowledge.

To help users to achieve this goal, new language tools have been added recently.

More information

If you are attending Wikimania 2013 in Hong Kong, you can get more information about the design process we follow when designing language tools at the talk Improving the user experience of language tools.

Notes

-->