effective use of translation memory (was Re: Use of American/British English (was Re: request for string addition))

[Tim Foster <Tim Foster Sun COM>]

hey Folks,

On Mon, 2004-02-23 at 00:09, Jeff Waugh wrote:
> Consider translation memory tools. If I wrote my software assuming 
> Aussie English, you wrote yours for British English, and Jody wrote
> his using Canadian/Quebecian Frenglish, anyone who used translation
> memory tools would very unhappy with us all.

Sorry to chime in with what's effectively just a "me too" post, but
Jeff's point is an excellent one. Getting a good translation memory in
one commonly used source language has been an important thing inside
Sun, for this very reason.

Taking the idea further, having documentation authors writing in a
consistent style has also done wonders for the amount of reuse we can
get from a translation memory system.


<blatent-advertising type="vaguely on-topic">

The other day, just for kicks[1], I ran a few some StarOffice/OpenOffice
sentences through our tools (we don't yet have OO translations in our
TM, but are working on it[2]) just to see what sort of matches I would
get.

Interestingly, I got quite a few matches from the Evolution Users Guide
and a good few from the GNOME Admin guide, due to the similar nature of
the StarOffice translations and the content in the GNOME docs. ( of
course, most hits were from other projects, like CDE, Netscape and
Solaris, since we have more of those translations in the TM, but that
will change over time I hope)

>From a 9600 word file where none of the previous translated versions of
that file had been included in our TM, we got about 500 words per
language translated completely, 60 words 95% translated, 250 words 85%
translated and 150 words 75% translated. I thought this was pretty
respectable : lots of "free" translations for 9 languages in 30 minutes,
and I got to drink coffee and read the days news while the server was
munching away on the translation work :-)

We're working on a paper for GUADEC this year to tell more about our
experiences with the use of translation memory tools on GNOME, and will
hopefully have more news about the availability of some of our tools, if
you guys want to play with them. Would this be of interest ?

</blatent-advertising>


	cheers,
			tim


[1] no, really

[2] - since StarOffice source was German up till a while ago, we were
facing this very problem ourselves : only this time it was en-US vs.
de-DE instead of en-US vs. en-CA, but it's the same principle.

-- 
Tim Foster - Translation Technology Engineer, Software Globalization
http://sunweb.ireland/~timf
http://www.netsoc.ucd.ie/~timf