Talk:Universal Language Selector Mobile

two character to indicate a language

You mention a two letter code to indicate a language. Many if not most of our Wikipedia language have three letters. Thanks, Gmeijssen (talk) 10:53, 17 February 2012 (UTC)

ah, then two/three letters it is depending on the language --Planemad (talk) 12:43, 18 February 2012 (UTC)

what language is spoken in what country

Are you aware how many languages are spoken in the United States? It is not only English and the native American languages. There are several languages not originating in the USA who have a substantial number of speakers.. think Chinese, Arabic, Korean but also Hmong ...

The notion that flags help indicate languages spoken is a long standing fallacy. Thanks, Gmeijssen (talk) 10:53, 17 February 2012 (UTC)

Choosing the US should obviously give languages like Spanish, Polish etc that are common there. The languages you would get on selecting a country should include common ones used by the population. You would obviously not have Zulu if you selct the US. The idea is to eliminate irrelevant languages from a selection list by using the country as a filter. If you were a Korean person in US and you had no idea to read English, then you would need to select Korea as the country to get the Korean language, it just does not make sense that he would choose the American flag rather than the Korean one to get his language. Would you argue that an Arab immigrant in the US would select the US flag rather than his own home country? --Planemad (talk) 11:52, 17 February 2012 (UTC)

The argument is with flags as a metaphor. They stink. Gmeijssen (talk) 10:50, 18 February 2012 (UTC)

Could you be a little more descriptive of why flags would be difficult/controversial for a user? A scenario would really help me understand why you would think that this is such an extreme problem and to improve upon the proposal--Planemad (talk) 12:41, 18 February 2012 (UTC)

As far as I can see flags would here be used to indicate countries, which I suppose is fine, and not languages, which would indeed be bad. SPQRobin (talk) 12:57, 18 February 2012 (UTC)

According to WolphramAlpha's statistics, Polish is less spoken in the United States than Chinese, Arabic, or Korean. --Yair rand (talk) 15:34, 19 February 2012 (UTC)

Countries

Using individual countries as units for language divisions could be problematic for a number of reasons:

• What is or isn't a "country" is really disputed, and we have no real way of deciding one way or another whether something is a country. (Do we use the flag of South Ossetia/any other disputed area for people in that region?)
• Languages divides don't really work well for countries. Inuktitut is spoken by 69.5% of w:Nunavut, but it probably wouldn't even be considered a major language for the entire country of Canada, for example. Similar situations occur in very many countries.
• Do we even have a reliable way of finding out what country the user is located in? I think we can get a decent idea of where someone is based on their IP address, and are usually able to find out what country that is, but why would we make the understanding of what languages the user speaks more ambiguous than necessary by first matching IP area to country, and then country to language, rather than just directly figuring out as well as we can what languages are spoken by people with that set of IP addresses?

--Yair rand (talk) 16:46, 19 February 2012 (UTC)

• There certainly is no clear solution for that, maybe it can be called select region instead of country and not showing flags in those cases.
• If a particular language is spoken by native people of a particular country it should feature in the list, irrespective of it being a major/minor language. The question that must be asked is with what country would the Inuktitut speakers of w:Nunavut associate themselves with. If its Canada, then the language must feature in the Canada list.
• The problem arises when a user would like to select languages that are not mapped to his ip. how do you navigate a list of 300 languages? you can then group them by ip numbers, but thats not usable by humans --Planemad (talk) 19:07, 19 February 2012 (UTC)

Universal Language Picker

What about this idea from Guillaume Paumier which I still really like. Simple and powerful autocompletion, augmented with browser and location detection would beat any GUI solution hands down I suspect. TheDJ (talk) 20:08, 19 February 2012 (UTC)

His idea is really good, except that it fails if the input language is one which the user is not familiar with. To take an example, when i was in Seoul, major websites like google would load the korean page by default, and the keyboard input and os interface was korean. In such a situation how would i go about selecting a minor language from India that I am comfortable with? --Planemad (talk) 06:12, 20 February 2012 (UTC)

I made a really quick mockup autocomplete-style language selector. This lets you autocomplete based on the code, or the local language name, or the English language name, but doesn't currently show you a "complete" list.

What we'd expect to see for the minor-Indian-language case is probably something like this:

1. phone locale is en_IN or something, probably?
2. app defaults to English
3. go into the language picker
4. start typing "telugu" or "kannada" or "marathi"
5. it comes up, probably within the first couple of characters typed

If more fully integrated with multiple local names and transliteration, this should work pretty well. But, you have to guess that it might actually be an option. If the initial list is realllly long you might be more inclined to say 'hey i'm probably in there somewhere, let's start typing'. --brion (talk) 01:03, 22 February 2012 (UTC)

I like the autocompletion UI. Instead of utilizing large lists, we could test a bunch of messages and see if they increase people's likelihood to select their language, e.g. "Wikipedia is available in 250 languages, including [random 3 language names]. Find Wikipedia in your language by typing its name above." Eloquence (talk) 02:43, 22 February 2012 (UTC)

It works well as I expected. But lazyman like me would not like to type but prefer to choose from a list or a table. For touch devices, the language search input box would be fine if it would be at the top and I want to see a selection table at the below which covers a number of language codes, a table consists of 4x5 or 5x7 cells. In the table, language code such as 'en', 'ko', or 'koi' would be listed by an order such as alphabetically or by the number of articles of project. --Ryuch (talk) 05:16, 24 February 2012 (UTC)