Today, I want to talk you about my occupation.
My company is localization company.
The localization (shorn term is l10n) is a business that translate computer manuals, user interfaces (menus, error messages, dialog boxes etc.), boxes, hardware (laptop PCs, cameras, cars etc.) manuals, publicity papers etc.
The difference between translation and l10n is localizers translate not only papers but also everything.
The range of our task is very wide and we have many businesses.
So we got to translate very much words in a short term.
And we got to translate them in a limited price than the usual translaters of literature.

Then we use computer to help the translation much more faster, cheaper and accurate at the same time.

I am a computer engineer to support the computing technologies to help such translaters.

One of the typical technique that the computer helps translaters is "translation memory".

Now, let's think we got to translate a manual of the word processer.
The manual has very huge volume.
Nowadays, it's not the printed book but the online help in DVD or on the Internet, but it contains thousand of pages.
And we got to translate them once in a year, with the limited time and cost.

Then you find that the manual is very similar to the previous version.
In other words, the new words are very limited.
Only you got to translate is changed terms and the topics of new functionality.

Then you need special software that provides "translation memory" functionality.

TM ("translation memory") is the database that includes the mapping of source language (let's say it is English) and its corresponding target language (let's say it is Japanese.)

You have the software to analyse the previous version to create the TM.
Then you have the software to analyse the new version with the TM.
Then you obtain the new sentences that don't exist in the previous version.
Then you got to translate only the new sentences.
So you can reduce your workload and your client would save the money.
And you can map the same translation to the same words/phrases.
It would improve the consistency of your translation with the help of computer.

The TM software can suggest you the "fuzzy" sentences --- the sentence that is new, but similar to the some of old material.
Then you can guess the meaning of the sentence and you can reuse the old terminology.

Let's think, if you got the brand new material (the material version one).
Can't you use TM because you don't have the mapping of source and target in the older translation?

Yes, you can!

There might be some (much) same or similar sentences in the single document.
They are called as reputations.
So if you translate the sentence once, the software would copy or suggest the translation for the rest of the reputations.
Very handy, and you can gain consistency.

The TM technology is very handy and helpful.
This technology is the crossover of the computer science and linguistics.

But nowadays, we are not satisfied with TM.
Then we take MT -- Machine Translation.
Let's discuss about it the next time.



Subscribe with livedoor Reader
Add to Google
RSS
このエントリーをはてなブックマークに追加