decent j to english pee cee appz?

[kotatsuneko]

got this recently:

http://pf.toshiba-sol.co.jp/prod/hon_yaku/seihin/mpr/index_j.htm

after seeing a brief piece on greggmans site, which is great for helping out with medical essays etc, but was wondering if there was anything in its league for day to day translating? i have pretty much every decent pc j to e dictionary going [up to last jan], but was wondering if anything decent had come out in the last year and would like to know peoples view on them, before splashing out, cheers!

[FG Lurker]

I've never seen any translation software that could do a decent job of "normal" conversations.

Maybe it's getting better, but I think we need a more CPU power and a lot more storage space before effective machine translation of non-technical documents can be done.

Just my opinion though...

[Socratesabroad]

[FG Lurker]

The high-end of machine translation has improved, but it still has a long way to go. The farther from technical documents and manuals you get, the more work that is needed in the machine translation field.

There has been a certain degree of success by taking known-to-be-accurate human translations and putting them into relational databases. The document is broken down and put into the database as phrases which are then related into paragraphs which are then built into the original document.

The translation software can then look for already-translated examples and reuse them.

This is much like how top chess computers (Deep Blue for example) work. They have massive databases of as many high-level chess games as they can get. The machine then uses database analysis to decide the next move based on the current board situation.

Translating like this still requires human proof-reading, but a lot of the "grunt work" for technical documents and manuals is done automatically. However we are still a long way from having such systems available at the consumer level.

[Socratesabroad]

[FG Lurker]

Automated translation and interpretation are good uses of logically thinking computers, if such machines ever progress beyond their current state. I don't see much hope within the next 10 years, but who knows...Breakthroughs tend to be very sudden.

[ttjereth]

CAT and Translation Support Software is usually referred to as "Translation Memory".

Basically a database of source-translation pairs. The software uses the database when you translate a new document by either comparing the document as a whole to the database and replacing selections with translations according to a (if you have a decent program anyway) user-defined match ratio/percentage.

Extremely useful and time saving for repetitive translations, like manuals etc. Can also be useful when the author of the original document has favorite "pet phrases" that are used over and over again, and can be accurately and reliably translated to a corresponding phrase in the target language.

Most translation memory software comes with an empty database(s) and you create your own (or swap with other translators). Not so useful the first time around, but once you have a decent database built things start to speed up.

Unfortunately, translation memory soft is really expensive (i.e. trados, cheapest version is what 7-800 $ u.s.?).

If you are interested

http://www.transassist.com

Site is all Japanese, but this is a relatively new translation memory that is about as cheap as they get. 39,900 yen I think. However, there is a free trial version available on the site if you are interested in just trying out translation memory software before plunking down several hundred dollars. You have to register to download the free trial and you will be sent a user survey by email afterwards asking what you thought of the software.

The software is actually designed by an American and the interface can be switched from Japanese to English, but as stated above the site is only Japanese at the moment.

Also, unfortunately the software is Windows only, no Mac version yet.

[ttjereth]

[Socratesabroad]

[ttjereth]