Tuesday, November 27, 2007

Language... (part 4) - Translation

The problem of multiple languages

There is language and then there are languages! We have been considering language as if we lived before Babel! The reality is not only that there are many hundreds of human language, but also that each of them have regional dialects numbering in thousands.

How then do we understand those who speak a different language to us? The answer has been to employ a highly specialised people called translators. We are all well aware of the perils of simultaneous translation of talks! But there was also another solution. One that we considered fanatical - everybody learn a common language. Esparanto was the proposed language. Today, we have a defacto standard - English. Not because of the English, but because of the Americans - and that too accelerated, thanks to the computer and the Internet. What we are at this moment doing is using this second solution!

In the world of networking, we have a plethora of protocols - languages used for computers within a network talking to each other. Since these were vendor dependent, they were different, and so needed to be translated between each other. If you think about it, we need one "gateway" to translate between two networks, but 3 to translate between 3, 6 to translate between 4 etc. Not a scalable solution!

Sometime in the mid 70's a proposal for an Esparanto for computers on networks was successfully made. It was called IP. Fundamentally it is a protocol for inter-network communication. Incidentally why it is better than English (in the human case) is that computers had no will to resist "learning" it! So, since 1984 in the UNIX world, and 1995 in the Windows world, all computers coming out of the production line came pre-installed with operating systems that were not only capable of speaking the proprietary vendor LAN protocol, but also a "second language", IP, that made it possible for them to talk to any other computer on the Internet - indeed any other device (cell phones, microwaves, hair dryers, lighting systems, cars…) on the Internet!

These two models for understanding languages of other "systems" (or people groups) can be explained thus:

  • The inter-lingua model - where a "higher level" meaning representation language is designed or adopted
  • The transfer model - where translation between any two languages requires a lot of hard linguistic expertise (fundamentally, how to replace the human translator)

No comments: