Re: translation software (a bit of an advert in there)

[Sarah Carroll <sarahc -at- INDIGO -dot- IE>]

Hi Damien,

You wrote:
> I know there is translation software 
> which is used by translation agencies
> but have been asked to check if there is 
> any which can be used before text
> is sent to the translation agencies.  

This really depends on what you want to do before
you send the text to the agencies, and on what you
expect the agencies themselves to do.

I think when you refer to "translation software" you are
referring to translation memory (TM) systems. There are
other types of translation software available and various
ways to use a TM system.

What is a TM system? Basically it is a piece of software
that allows the user to translate a segment (may be a
sentence, a heading, an index entry), and then store that
segment and its translation in a special kind of database,
called a translation memory. If the same, or very similar,
segment appears later on in the translation, it is pulled
from the database by the system, and presented to the
translator to use. Very useful tool when there is a lot
of repetition in a document.

TM systems are generally of greatest use when there are
a) several translators working on the same material -- improves
consistency
b) frequent updates to a project during the translation life-cycle --
saves a lot of cutting and pasting
c) future updates to the same the document set are planned --
when the documents will be substantially similar, you can
use the TM
d) your doc and help are single sourced -- you can use
a lot of the segments in the TM

Most of that refers to how the translators/agencies
use TM systems. They will also generally use a terminology
management system along with the TM system.

Good TM systems also provide various other tools,
for example word counting and statistics on how much
repetition there is in the document. You might carry 
out that sort of activity before you send the files to the
translator/agency.

When you are in the situation where you have had
your documentation set translated before, and, (this is
important), you are happy with the quality of the translation,
but you don't have a translation memory from it, you may
want to use the previous translation in an update to the doc.
Then you would use an Alignment tool. This allows you to
create a translation memory from the previous material. 
Alignment needs to be done, generally, by a native speaker.
This is another activity you might carry out before sending
the files to be translated.

You might also "pre-translate" the files. That is, use an
existing translation memory to automatically insert the
translated segments into the new files. Translators will
generally insist on checking these segments, but it can
speed things up.

You might also set up the translation memory. In some
TM systems, you would have to go through the files
and set up "segmentation rules", that is, tell the system
not to create a new segment when it comes across
a period at the end of a particular abbreviation you are
using in your documentation.

If you are using FrameMaker or Interleaf files for your
documentation, you might use The S-Tagger (this is
the advertisement bit, but if you're reading this far, you
should probably get to know about it <g>) to convert
the FrameMaker or Interleaf files to STF, which is the
tagged text format used by The S-Tagger. Translating
STF files is a lot easier than translating Interleaf or
FrameMaker files. E-mail me mailto:sarahc -at- indigo -dot- ie
for more information, or look at http://www.trados.com/stagger

There are also Machine Translation systems. These
systems are very complex and expensive, mostly. Baby
MT systems (like the Systran one that AltaVista uses)
are really basic. The translations they provide are generally
not usable without substantial editing. MT systems take
a lot of work, and are really only useful (IMO) when you
have millions of pages of material which is all very
similar (in machine terms) and you have put a lot of
work into training the system. Some users are now
implementing MT systems with a feedback into a
TM system which feeds back to the MT system, confused?
Well, ask me for more info.

Translation memory software I recommend, but I
am biased as I work with TRADOS and develop
The S-Tagger are:
TRADOS WinAlign - Alignment software
TRADOS MultiTerm - Terminology management system
TRADOS Translator's Workbench - TM system
The S-Tagger from TRADOS - conversion solution for
FrameMaker and Interleaf files

Ah go on, I'll mention some others.
Disclaimer about the URL's -- if they don't work, please
don't innundate me with castigations! This is the info.
I have available to me at this time and I have checked 
them all.

Commercially available:
Alpnet Joust * AlpNet * http://www.alpnet.com/frame.htm
Corel Catalyst * Corel * http://www.corel.ie/catalysthome.htm
Déjà vu * Atril * http://www.atril.com
IBM Translation Manager * IBM * http://www.software.ibm.com/ad/translat/eqfn0mst.htm
Lant Eurolang Optimizer * Lant * http://www.lant.be
Star Transit * Star AG * http://www.star-ag.ch

Some Service Providers who have their own systems:
Foreign Desk 5.0 * International Communications International Communications * http://www.intl.com/default1.htm
Amptran * SDL Ltd. * http://www.sdl.co.uk/
ILE's LXEdit, OpenTag, Borneo * ILE * http://www.ile.com/

There is also the LISA OSCAR TMX format that you might be
interested in. This is in development, but should be adopted
by the LISA (Localisation Industry Standard Association) as
a standard format for the exchange of data between translation
memory systems in August. Again, e-mail me if you want 
more info. on that.

Advertisment and lecture over, usual disclaimers, 
Best
Sarah
sarahc -at- indigo -dot- ie
(Dublin, Ireland)