PET - Process Embedded Translation Thread poster: Alexander Ulichnowski (X)
|
|---|
Alexander Ulichnowski (X)
Local time: 06:05
What is your opinion about this hot topic? Is this the future? As a translator I work with a unique workflow and interface developed by Xplanation (www.xplanation.com). First of all the text is downloaded. The interface not only shows TM (Translation Memory) but also MT (Machine Translation). The translator polishes the text and uploads it. 3 important steps are followed : pre-edition, post-edition and revision... See more What is your opinion about this hot topic? Is this the future? As a translator I work with a unique workflow and interface developed by Xplanation (www.xplanation.com). First of all the text is downloaded. The interface not only shows TM (Translation Memory) but also MT (Machine Translation). The translator polishes the text and uploads it. 3 important steps are followed : pre-edition, post-edition and revision. Finally the text returns to the person who asked for the translation. Billing is done automatically.
[ This Message was edited by: on 2002-01-15 09:57 ] ▲ Collapse
| | | |
Jeff Allen 
France
Local time: 06:05
Multiplelanguages
+ ...
| reply on process embedded translation | Dec 25, 2004 |
Alexander Ulichnowski wrote: What is your opinion about this hot topic? Is this the future? As a translator I work with a unique workflow and interface developed by Xplanation ( www.xplanation.com). First of all the text is downloaded. The interface not only shows TM (Translation Memory) but also MT (Machine Translation). The translator polishes the text and uploads it. 3 important steps are followed : pre-edition, post-edition and revision.
There have been many large translation projects moving in this type of direction over the past 10-15 years. I believe that good hybrid workflows that make the best of all the strong points of each technology to get the job done can improve the overall translation process. The obstacles are:
* budget
* system compatibility
* training the translation system users with a useful and productive methodology
* resistance to using technologies for fear of being replaced
See my related threads on Proz at:
http://www.proz.com/post/190436#190436
also:
http://www.geocities.com/mtpostediting/
Jeff
http://www.geocities.com/jeffallenpubs/
| | | |
I believe that apart from it being a pretty good way to ease your working day this type of translation poses the danger of mistaking connotations of words.
The final product may thus lack meaning and essence.
But it all depends on register. So, with technical texts, it's O.K., but with fiction, that's something completely different.
| | |
|
|
|
Isn't it nice to see CATs and PETs running around the translators courtyards! 
But I personally, although highly interested in these tools, feel almost terrified when I am to use them.
Is this PET a "kind soul"?
I mean, is it easy to work with? How much knowledge should one have in order to use it ? Does it support Cyrillic?
And how is i... See more
| | | |
Jeff Allen
France
Local time: 06:05
Multiplelanguages
+ ...
| reply on PET | Jan 30, 2005 |
cameliafrunza wrote:
I believe that apart from it being a pretty good way to ease your working day this type of translation poses the danger of mistaking connotations of words.
The final product may thus lack meaning and essence.
But it all depends on register. So, with technical texts, it's O.K., but with fiction, that's something completely different.
No problem with multiple connotations with words when you know how to use the system well by working with them.
The final product is always verified be a human translator.
Different levels of register can also be dealt with.
As for types of texts, it is only limited as you want it to be. I've used MT on many marketing brochures, Commercial replies to bids, legal documents, religious texts (very little repetition there), etc.
Jeff
| | | |
Jeff Allen
France
Local time: 06:05
Multiplelanguages
+ ...
| types of PET tools | Jan 30, 2005 |
Rossitsa Iordanova wrote: Isn't it nice to see CATs and PETs running around the translators courtyards! But I personally, although highly interested in these tools, feel almost terrified when I am to use them. Is this PET a "kind soul"? I mean, is it easy to work with? How much knowledge should one have in order to use it ? Does it support Cyrillic? And how is it available? Thanks in advance! Ross
Ross,
Yes, I implemented a full technical writing and translation workflow system at Caterpillar using SGML encoding. That included Russian. That was 10 years ago.
I know work with various commercial tools. Russian and other Eastern language scripts are supported.
The key issue to working with any of these kinds of language technology systems is making sure that the person who implements them understands real-world translation workflows, and preferably those that include corporate workflows and freelance work cycles.
These should not be implemented as push-button magic-wand systems, because for them to be productivity-enhancements systems, they must be carefully adapted to the existing workflow, and teaching the translators, revisers and others in the workflow on how to use the tools.
Jeff
| | | |
Pablo Roufogalis (X)
Colombia
Local time: 23:05
English to Spanish
| The future... | Feb 27, 2005 |
I do believe the future will be as this:
1) A powerful, centralized software that will provide pre-processed files for all the humans involved. They would be closed, proprietary systems that would take care of all aspects of the business.
2) The first translations would be done by a machine and a huge TM and term database that, in non-trivial phrases, will provide background and options for the first human translation and will be right about one of them most of the time... See more I do believe the future will be as this:
1) A powerful, centralized software that will provide pre-processed files for all the humans involved. They would be closed, proprietary systems that would take care of all aspects of the business.
2) The first translations would be done by a machine and a huge TM and term database that, in non-trivial phrases, will provide background and options for the first human translation and will be right about one of them most of the time.
At this stage you have one or several products, that can be sold directly to end-customers, cheaply or more cheaply. Skip step 3 and go directly to 4. Translators do not collect 200 dollars.
3) A first human translation done by a poor fellow in a poor country getting just a few bucks. He would use a proprietary, closed software that is connected at all times (or connects periodically, to the huge server above via the internet.
4) A final review/approval process that could be done by a capable translator or... the end customer with an easy to use tool, proprietary tool. His corrections will become part of the server's TM.
The business as we know it would be seen then as hand-broidery vs. a huge factory in India producing millions of cheap garments per year. It would have its place but it won't be the rule.
Do you believe Microsoft is a partner in Trados for the couple of millions in revenues that it provides to their cashflow?
Are we all luddites here? ▲ Collapse
| | |
|
|
|
| I have some doubts | Mar 7, 2005 |
2) The first translations would be done by a machine and a huge TM and term database that, in non-trivial phrases, will provide background and options for the first human translation and will be right about one of them most of the time.
May be it is a great thing from scientific point if view.
However my experience of working with pretranslated texts I have been receiving from my customers is very negative.
It takes so much time to edit given texts that I simply delete all pretranslated sentences and translate them from scratch now.
It sounds strange but typing of a new sentence takes less time than editing of a ready construction by dragging words to the right places. And very often a pretranslated text is an example of "do not speak in such a way!"
| | | |
| What about the knowledge base of the translator | Mar 5, 2009 |
I"m facing a choice: working or not with such tools (that be imposed soon)
but it seems to me that with these tools, the greatest danger would be that none of the terminology or translations memories done would be kept on my computer, and as we all know, we tend to forget things so what if I do a real good job in this field + terminology job, goes on to translate smth else, and comes back a year later to this specific field for another client, and all the knowledge I gathered has b... See more I"m facing a choice: working or not with such tools (that be imposed soon)
but it seems to me that with these tools, the greatest danger would be that none of the terminology or translations memories done would be kept on my computer, and as we all know, we tend to forget things so what if I do a real good job in this field + terminology job, goes on to translate smth else, and comes back a year later to this specific field for another client, and all the knowledge I gathered has been updloaded to this Translation agency's server, and nothing is left on my computer (maybe not even the result, ie. the translation) ?
How are young translators ever going to gain some knowledge if all they are asked is to serve as a "typing operator", filling in segments...
TIA for your answer (i know you started this post long ago, but I hope smbdy can pick up the conversation).
mac2forum ▲ Collapse
| | | |
JapanLegal
Japan
Local time: 13:05
Member (2009)
Japanese to English
+ ...
| I could not agree more | Mar 9, 2009 |
Sergei Tumanov wrote:
2) The first translations would be done by a machine and a huge TM and term database that, in non-trivial phrases, will provide background and options for the first human translation and will be right about one of them most of the time.
May be it is a great thing from scientific point if view. However my experience of working with pretranslated texts I have been receiving from my customers is very negative. It takes so much time to edit given texts that I simply delete all pretranslated sentences and translate them from scratch now. It sounds strange but typing of a new sentence takes less time than editing of a ready construction by dragging words to the right places. And very often a pretranslated text is an example of "do not speak in such a way!"
I can't tell you how many times I've had to work with a totally unusable machine-translated text. Yes, I could *vaguely* understand what was happening *most* of the time, but trying to massage such error-ridden word barf into tight and lucid text was much more difficult than it would have been to write the whole thing from scratch. Suggestion is powerful. In the vast majority of cases, it's much easier to work from one language into another than it is to edit low-quality "writing" into something publishable.
| | | |
Login to reply/comment 
Translation news
