Wednesday, September 19, 2007


We don't know exactly how it all got started, but each of us is born with the ability to learn languages. None of us is born with the knowledge of any particular language. We get that from having to interact with the people around us.

That works fine on a small scale, but what if you want to get the whole world talking? There are only a few basic possibilities:
  • One language: It's the only game in town. QED.
  • Lingua Franca: If there have to be multiple languages, at least have one that everyone speaks.
  • Multilingualism: If you want to understand someone, learn to speak their language or get them to speak yours.
  • Translate: If someone doesn't speak your language, get someone to translate what they're saying and likewise for you.
The world has never in recorded history been even close to monolingual, but it's not for lack of people making the suggestion, at some times more forcefully than at others.

Multilingualism is common in much of the world. Combined with Metcalfe's Law and/or its cousins, multilingualism tends to promote the emergence of at least a local lingua franca. It's not clear whether this discourages multilingualism and ultimately tends toward the one-language world. It might seem easier just to learn the lingua franca, but children have no trouble learning multiple languages and there is a strong tendency to maintain the local language when the lingua franca is not local.

Latin is an interesting case. At one time an educated person was expected to know Latin and publish in it. This did not, however, lead to the widespread revival of spoken Latin. One had, after all, to talk to the merchants in the square as well as one's learned colleagues.

Lingua francas (or linguae francae if you prefer) have so far been partial in any case, both in that different regions have different ones and that not everyone in a given region finds need to learn the lingua franca, or speak it as fluently as a native. It's probably too early to know exactly what effect the web will have. You'd want to bet on homogenization, but you wouldn't want to bet the farm.

And that leaves translation. It's expensive, awkward and not completely reliable, but it's still with us. If machines ever get as good as human translators it may even eliminate much of the need for the other approaches.

There are obvious analogies here to standards on the web. One significant difference is that enabling a service to understand a new format or writing a converter is much less expensive than teaching a person (even a child) a new language or hiring a translator.

1 comment:

David Hull said...

Note to self: wha t is the actual language distribution on the web (and by what measure)?