In the top north east corner of France, the locals speak a language that the French call a patois, but what in fact is just a dialect of Dutch. "Dunkirk" just means "Dune Church".
A local there could speak a sentence to a neighbour, and ask the neighbour to pass it on, and being passed from neighbour to neighbour, every neighbour repeating only what they heard, that sentence could travel from northern France, into Belgium, through into the Netherlands, into northern Germany and end up in Denmark. The last person in the chain wouldn't be able to converse with the first person in the chain, but the message would have passed untranslated all that way.
The sentence could just as easily have taken another route, through Belgium, into Luxembourg, into Western France, through Switzerland, northern Italy, into Austria, and end up in the east Czech Republic.
In the same way, a sentence spoken in Portugal could pass through Spain to Catalonia, into southern France, along into Italy. Again the first person wouldn't be able to converse with the last person in the chain, but they are joined with what is called a "dialect continuum": there are no sharp language borders.
So if people at the edges of what we call languages can converse with each other, where does one language end and the other begin? How do we define a language? How many languages are there? In any case, linguists typically give a figure of 4000-5000 languages, and that's good enough for us now.
There is an unfortunate property of some countries that they have the same name as the majority language spoken there. France and Germany for instance, but not Austria or the United States.
I say "unfortunate" because there is a regrettable tendency for people to link countries with languages. A good example of this is that many web sites that offer versions of their pages in different languages use clickable flags to lead you to the different versions. But there are less than 200 countries, so it is clear that flags are a bad way to represent languages.
Take France, for instance. I count eight native languages there. Are people who live there but speak another language to click on the flag of another country to get to their own language?
I see this a lot as an English speaker. Typically to select an English version of a site, I am required to either click on the US flag or on the Union Jack, (which is the flag of the United Kingdom). However the British flag represents several different groups of people who speak different languages (such as the Welsh and Scots). There is an English flag, but most people wouldn't recognise it as such if they saw it.
By using flags to represent languages, you run the risk of injuring people's cultural and political sensibilities. There are countries that speak essentially the same language, but each give it a different name. By assigning a flag to the language you may be requiring someone to associate themselves with a country they don't wish to be associated with (and this includes the US and UK flags...) If you take the other view, of offering the flag icon as an entrance for inhabitants of that country (rather than speakers of the language that happens to bear the name of that country) then you have to know of all the languages spoken there, and avoid giving any precedence.
And after all, there is no real need for icons anyway: it is after all about language, and using the name of the language in that language is surely the best way to represent them.
By the way, people make the same mistake when designing the web address for a language version of a site: for instance you see web addresses with ".../gb/" or ".../us/" in them, using a country code to distinguish them. Don't do this! There is a perfectly good list of two-letter codes for languages (ISO 639). English is "en" for instance. Use those.
© Copyright Steven Pemberton, Amsterdam, 1998. All rights reserved.
First published in the SIGCHI Bulletin, January 1998