Whilst at StatusCampMontreal today, a few of us in the Internationalization session were discussing interfaces for switching to another language. I hit upon a possibly rather silly idea where you can use something visual to help predict someone’s language. This is only a rough sketch, so it’s probably not much use as it is to anyone right now, but perhaps this might come in handy for a better idea or become something that someone else can build upon. Note: this looks at a general problem and does not necessarily solves issues for status.net.
So, generally, there are two problems to overcome when a user encounters a multilingual site (imagine many languages) that may or may not be serving the correct language to them by default:
- a user needs to know how to switch to a different language
- a user needs to choose the correct language (their preferred language)
Bear in mind they are likely to be looking at the site in a language they don’t understand at this point.
My idea is more inclined towards solving the 2nd step. What if you could serve an image of a common object, and let the user type in the word they have for that object? The “common object” example I have here is an apple, but you may conceivably use images of the sun, the moon, a tree, etc.
The interface could look something like this:
(In the top right corner I’m using a language selector that I’ve used in past designs, using the presence of different languages — in this case, Chinese, French and English — to convey that there are further language settings, an attempt at addressing problem #1).
Basically, we provide a way for the user to type into a textbox the name of the object they see. The idea is that they would do this in whatever language they are already comfortable in. (In my example sketch, I’m writing “epal”, the word for “apple” in Malay.) There are languages where some nouns are similar, so this may need to be repeated a few times for an accurate detection.
There are obvious flaws with this visual language detection method:
- if you can’t see the image — if you’re not a sighted user or if you’re using a device that cannot render an image — this cannot be used
- choosing culturally-neutral images might not be trivial
- there needs to be some way of indicating (with little or no text) that you’re supposed to type in the name of the object you see, and that’s a hard concept to convey
- from a user flow perspective, this could be very confusing to be suddenly served an image of an apple…
Some good things about this:
- you can conceivably make a decent language guess if you have a few images, and you only need a fairly limited corpus to search through
- a user won’t have to wander through a big list of languages they don’t care about in order to find the one that matters to them