RE: Translation Memory System

["Glenn Maxey" <glenn -dot- maxey -at- voyanttech -dot- com>]

> -----Original Message-----
> From: bounce-techwr-l-58477 -at- lists -dot- raycomm -dot- com
> [mailto:bounce-techwr-l-58477 -at- lists -dot- raycomm -dot- com]On Behalf Of susan
> larsson
> Sent: Thursday, May 24, 2001 2:37 AM
> To: TECHWR-L
> Subject: RE: Translation Memory System
>
>
> On 04:06 PM 5/23/2001 -0600, Glenn Maxey wrote the following:
> >If you can reduce the amount of text that the translators have to look at
> >even with a tool, you'll bring down the costs of subsequent update
> >translations. Typically, translators will charge 20% of the full
> >translation
> >just to open a file and run translation memory against a file that they
> >previously translated. In other words, they aren't doing anything except
> >running a tool that could very well "100% match" everything, yet
> >you'll be billed 20% of the full translation cost.
>
> Just curious - as a tech writer, if there is existing
> documentation and you
> are asked to update it - do you charge nothing to review what already
> exists? Do you only look at the part that has changed, without
> seeing it in the context of what went before, and what follows?

A former manager of mine used to say: "There's not a sentence in the manuals
that I couldn't or wouldn't re-write if given the opportunity."

He wasn't saying that what was written was wrong. He was saying that
depending upon the star alignments and which side of the bed he got up on,
he'd change a sentence one way and on the next day change it back given the
opportunity. We all have our stylistic whims.

The hard part is to look at a sentence and say, "I don't need to change this
even though I could." When looked at as part of a whole and amidst other
more pressing projects, it is wise to prioritize what has to be changed from
what could be changed. This was his real message.

That being said, I'll risk lessening my stature in the eyes of my peers by
answering your questions truthfully.

Depending upon:

- where we are in the release cycle,
- what my relative knowledge of the subject matter is,
- how much material is involved,
- what other projects are on my plate, and
- my level of involvement in the project from the beginning,

I admit that I do NOT always review the entire suite of what already exists
when I am charged with updating it. I must distinguish between what HAS TO
BE changed and what CAN BE changed, and prioritize accordingly. I must have
some trust that what a co-worker or I did at a previous date was and still
is correct. Hence, my efforts need to focus on what is missing (e.g., new
stuff) and wrong (as per feedback from engineering, QA, etc.)

Hence, I do make changes without seeing the entire context.


> Translators do not work with words in a vacuum either;

I have worked with many translators who did work in a vacuum. They were not
subject matter experts. They had no training in the field that the
documentation dealt with. They were working not just on the edge of their
understanding, but beyond it. (This does not mean that they did a bad job;
it just means that they were indeed working in a vacuum compared to those of
us in-house.)

Hence, standard operating procedures (that even they recommend and impose
upon us) dictate that their translation of, say, the GUI be sent to an
in-country SME like a Sales Engineer for review before they move on.
Corrections to those terms gets put in the translation memory for use later
on.

Moreover, if the scope of the project was big enough and the time-line short
enough, they divided the work up among several translators. They often got
individual chapters out of context (yet another vacuum).

Let's be honest with ourselves here. If they received out of context a
highly technical chapter in some obscure subject area that had been
translated previously by a colleague and could be translated to 90% directly
from translation memory, do you really think that they'll re-visit any of
the 90%? Or will they focus on the 10% that hasn't been translated?

Time is money. If they have little understanding of the subject matter, they
will make the same assumptions that I do. They have to trust that what their
co-worker did was and is correct unless they have reason to believe that it
isn't, as would be the case with negative feedback from in-country subject
matter experts who reviewed the translated material.

However, even with such negative feedback, if you know your CAT, you can
optimize it. You and/or your in-country SME can change the translation
memory accordingly. Thereafter, there will be little reason to re-visit the
90% except to see what borders new/changed material.

The most I would expect from a translator is to view the preceding and
following sentences/paragraphs for any new or changed sentence/paragraph
that appears. I do not expect them to read the entire chapter/manual. Even
if they did, I've seen first-hand how the subject matter even in small
quantities made their eyes glaze over as the technical details whushed by
over their heads.

Hence, it is perfectly acceptable to cut corners by reducing what gets sent
to translators. (Just enough context to give an untranslated out-of-context
sentence/paragraph meaning.)

I repeat that I would not go to the effort to do this unless the number of
translation languages and/or the size of the material made sending projects
to translation in their entirety cost-prohibitive (and unless I was the
full-time translation coordinator and had time for such fun).

> if we do not examine the context, we cannot make the style consistent;
> we cannot guarantee that the changes will fit.

True. However, being the Devil's advocate, even the tools of the translation
trade do not always put things into context.

For example, Trados displays a sentence at a time. If an index entry or
figure caption are in the text, I've experienced being prompted to translate
them as soon as Trados stumbled upon them. It broke the flow of the
sentences and paragraphs.

GUI translations are completely out-of-context.

Using the CAT like a translator would to search on fuzzy-matches (sentences
that haven't been translated or that were changed) is like looking at a
bunch of sentences out-of-context. In all honesty, in most cases you don't
need the entire context -- just the sentence itself -- in order to be able
to translate it, although you can easily get to the context.

> The fact is that even with 100% matches, the match
> may not be correct - and work will be required to correct it.

That's true. However, if you or someone you respected did the initial
translation, how much incentive do you have to re-visit the 100% matches to
verify that they are correct?

Let's be honest, if you aren't being paid to re-translate the 100% matches
and if no in-country SME flagged anything wrong, you'll hit "next fuzzy
match" and never see it. Yet you'll charge at least 20% of the translation
costs for all words, sentences, paragraphs, and chapters that 100% matched.

Depending upon the amount of material and number of languages, this 20%
charge can quickly add up. It could be enough to hire an in-house
translation coordinator who would take charge and then send out what needs
to be translated.

> ><snip>
> >Frank, you're in Germany and your posting was English. I suspect you're
> >bilingual and probably do small English/German translations. You'll find
> >Trados a big help even for small projects not related to the big French
> >translation.
>
> Being bilingual is far from the main requirement for being a good
> translator. The translator needs every one of the skills of a technical
> writer - plus a thorough understanding of the source language
> combined with the ability to adapt concepts so they work in the
> target language as well.

Right. Frank's native language is German and his posting was in near
flawless English. He's a technical writer. His experience at his company has
given him near SME status well beyond what an external translator would
have. Hence, if he received something in English relating to what his
company does, I think that we could expect something wonderful in German.

If his boss approaches him with a small translation assignment for some web
page, white paper, letter, or advertising brochure, I expect that he would
do a fine job English->German and something adequate German->English (except
for the advertising brochure where adequate isn't good enough).

My point was that when these assignments come his way -- and in the real
world we know that they do --, Frank will be able to get a lot of milage out
of having a CAT at his disposal.

> The goal is to produce a document that looks like it was written by a
> native technical writer from the start, without ever having been
> in another language.

No argument there.

> If you were to apply your ideas of cutting corners using CAT to your own
> profession of tech writing, I'm sure you'd see what I mean.

Cutting corners is the nature of business. It is done in technical writing,
and I have presented ideas (fully supported and recommended by Trados) in
how to cut corners in the translation of subsequent versions.

I'm not saying that I agree with it. I'm just saying that it is done.

At my last employer and while I was still new and heavily involved with
getting a release out the door, the translation coordinator sent out the
English online help to be translated into several languages BEFORE I could
review it. A significant portion of it was written and/or maintained by
non-native English speakers. Talk about irking me off!

When I did have the chance to review it, I was explicitly told NOT to. The
subsequent cost of re-translation was deemed too high (even though the CAT
would have fuzzy matched pretty accurately my tweaks.) Misinformation and
their lack of understanding of the CAT allowed them to feel no sorrow about
eating the cost of garbage-in-garbage-out translations and our time spent
with the translators clarifying poorly written source text. I was accused of
wanting to "gold plate" the documentation.

Having the CAT, subsequently attending training on the CAT, and
understanding the CAT allowed me to push successfully for a much-needed
online help re-organization and re-design, anyway, which was reflected in
all languages. I made major changes... although not to the sentence level
unless it was obviously really bad. The sentence level would have been my
targets for subsequent releases had I opted to stay.

If I would not have understood the CAT, the cost-cutting measures of the
company would have had me sailing out the door much sooner. It would have
been too frustrating being prevented from improving what I could and what
screamed for it.


> Translators use CAT - but not for the purpose of saving money
> for end clients, but to
> ensure optimal quality and consistency - for the end client.

This is not an "either-or" argument. It's "all of the above" and points not
yet mentioned, like how easy it makes certain aspects of translation. Fuzzy
matches in the course of translation go a long way into suggesting the right
terminology even if the exact context isn't right.



^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

*** Deva(tm) Tools for Dreamweaver and Deva(tm) Search ***
Build Contents, Indexes, and Search for Web Sites and Help Systems
Available now at http://www.devahelp.com or info -at- devahelp -dot- com

Sponsored by Information Mapping, Inc., a professional services firm
specializing in Knowledge Management and e-content solutions. See
http://www.infomap.com or 800-463-6627 for more about our solutions.

---
You are currently subscribed to techwr-l as: archive -at- raycomm -dot- com
To unsubscribe send a blank email to leave-techwr-l-obscured -at- lists -dot- raycomm -dot- com
Send administrative questions to ejray -at- raycomm -dot- com -dot-  Visit 
http://www.raycomm.com/techwhirl/ for more resources and info.