Mobile menu

Advice on how to approach a large XML project
Thread poster: transslate
transslate
Local time: 11:02
German to English
+ ...
Sep 15, 2005

Hello,

I need some help on how to approach an XML job. The client wants to reduce cost of a 97,000 word XML document. There is only one big XML file (see attached) which has about 280k new words and 120k reps. We are currently working on this for a number of languages, but they are looking to add more. The problem is that since they are doing this the first time, it will be very expensive if they have to do everything
from scratch. He is considering machine translation (which I think is a bad idea). I would appreciate it if anyone could help figure out how to localize all of these words and keep the cost down by not having to translate every word.

[Subject edited by staff or moderator 2005-09-15 17:14]


Direct link Reply with quote
 
xxxBrandis
Local time: 20:02
English to German
+ ...
Tageditor is the best Sep 15, 2005

Hi! AFAIK Trados (6.5 onwards)suited for this kind of work. Best Brandis

[Edited at 2005-09-15 17:53]


Direct link Reply with quote
 

PCovs
Denmark
Local time: 20:02
Member (2003)
English to Danish
+ ...
Indeed TagEditor Sep 15, 2005

Brandis wrote:

Hi! AFAIK Trados (6.5 onwards)suited for this kind of work. Best Brandis

[Edited at 2005-09-15 17:53]


Trados will take care of the repetitions so that the client will only pay once for a segment which is e.g. repeated 1000 times or more (the translator will only have to translate this once - then Trados will recognize the segment).


Direct link Reply with quote
 
transslate
Local time: 11:02
German to English
+ ...
TOPIC STARTER
Help Sep 15, 2005

Thank you all for your replies. I realize the Trados issue but the client is looking at machine translation (bad!) and then editing the text,phrases,etc. Do you know any decent machine translation sw? The languages would be Dutch, Swedish and Portuguese. Thank you!

Direct link Reply with quote
 

Riccardo Schiaffino  Identity Verified
United States
Local time: 12:02
Member (2003)
English to Italian
+ ...
Sep 16, 2005



[Edited at 2005-09-16 04:43]


Direct link Reply with quote
 

PCovs
Denmark
Local time: 20:02
Member (2003)
English to Danish
+ ...
No machine translation tool is good enough yet. Sep 16, 2005

Riccardo Schiaffino wrote:



[Edited at 2005-09-16 04:43]


I rather liked this comment, why was it deleted?

Anyway, machine translation is really a bad solution, and I would not want to be the one proofreading and editing the outcome!

And I don't think I would help a client find a machine translation tool either, because I would feel that the quality of the outcome was somehow my responsibility, if you know what I mean.

But perhaps you could suggest that your client tries to translate something into his/her native language using a machine translation tool and then have a look at the outcome to see if he/she thinks it would be at all acceptable.

Just a suggestion.


Direct link Reply with quote
 

Roberta Anderson  Identity Verified
Italy
Local time: 20:02
Member (2001)
English to Italian
+ ...
Machine translation is really no option - false saving Sep 16, 2005

Machine translation is really no option.

Editing a machine-translated text is not the same as reviewing a translated text. Apart from the psychological aspect (not to be understimated - it's really frustrating, which leads to demotiviation on the reviewer side, which in turn is inevitably reflected in the final result), this kind of editing is usually paid on an hourly basis, and for such large volumes will most probably end up costing as much if not more than a "human" translation.
This latter is I think an argument that the client could appreciate.

Depending on the contents of this large project, the clients might be ablt to rip the benefits of a properly handled first translation in future updates.

Also, if it is a "first", all the more reasons to start off with a professional approach - whatever is done now will affect any future projects based on this first one.

cheers,
Roberta


Direct link Reply with quote
 

Heinrich Pesch  Identity Verified
Finland
Local time: 21:02
Member (2003)
Finnish to German
+ ...
First a specialist should look at the original text Sep 16, 2005

With xml one often encounters this: text is ripped off from a web site and delivered to translators. But first one should make sure that the original text is really ok and properly formatted. Each sentence should end with a full stop at least. And the spelling is good.
Only then translators should be envolved.
If the text is not worth to be translated by humans it is not worth to be translated at all (and should be deleted from the web).

Regards
Heinrich


Direct link Reply with quote
 

Cecilia Vela Segovia-Frund, CT  Identity Verified
Argentina
Local time: 15:02
Member (2003)
English to Spanish
Maybe Systran... Sep 16, 2005

I agree with the previous opinions. You'll never get a professional text from MT. In most texts, you will have lots of incomprehensible and even risible paragraphs.

Anyway, I understand you are trying to meet your client needs, and I respect the business aspects of our profession. All projects are different and they need different solutions.

I recommend you the following:

1. Check if source text was written by a Technical Writer. If internationalization was considered beforehand, he may have used the right words and structures to obtain better results from MT.

2. Get a good MT software. I've used Systran from www.systransoft.com and results are reasonable for non-human output (only paid versions).

3. Identify key terms in source text (placeables like product names that are not to be translated, specific industry terminology, etc.) and add manually to Systran glossaries of each target language.

4. Machine translate your text and have it reviewed by Linguistic Reviewers (you may ask for a very "shallow" review that will change only those sentences that the software will totally mistranslate, no more than 20% of sentences to make it cost effective).

Suggestion: Once translated, you can suggest your client to place a hidden access counter in each of the htm pages of his web site in the different languages, and to agree to manually translate with your agency the pages that reach a certain number of clicks. I've heard of some one-million-word technical catalogs where the "best-clicked" pages (those bringing a higher profit to the company) are finally translated by humans to boost sales of those preferred products, while those with only a few clicks are not reviewed. I don’t know if it is wise, but it may be a “customized” answer to a client with a tight budget.

Regards,

Cecilia Vela-Segovia
ATA Certified Eng>Spa Translator


Direct link Reply with quote
 

PCovs
Denmark
Local time: 20:02
Member (2003)
English to Danish
+ ...
A quick review of Systran Sep 16, 2005

Cecilia Vela Segovia-Frund wrote:

I agree with the previous opinions. You'll never get a professional text from MT. In most texts, you will have lots of incomprehensible and even risible paragraphs.

...

2. Get a good MT software. I've used Systran from www.systransoft.com and results are reasonable for non-human output (only paid versions).

3. Identify key terms in source text (placeables like product names that are not to be translated, specific industry terminology, etc.) and add manually to Systran glossaries of each target language.

...

Regards,

Cecilia Vela-Segovia
ATA Certified Eng>Spa Translator




There's obviously a lot of work to be done beforehand when using this translation tool, and I still doubt the efficiency.

Using the link provided by Cecilia, I entered "Thank you all for your answers" (English to Swedish) and the translation was: "tacka all dig för dina svar".
When entering this sentence for translation from Swedish to English, the English translation reads: "ewe all you pursues your replies".

Another test: I entered "Depress the clutch pedal" and the translation was "Tryck ned den pedal- kopplingen". When turning this sentence around the English version now reads "Depresses the pedal- kopplingen".

Thess two examples do not make me sure that this tool should be able to translate a text into something that could be understood by anyone.

Anyone familiar with Swedish will know that theese translations would not be good and would need quite a bit of rewriting.

Just a simple test, mind you, but this is what Systran has to offer as a sales argument on their web-site, so...


Direct link Reply with quote
 

Stefano Papaleo  Identity Verified
Italy
Local time: 20:02
Member (2005)
English to Italian
+ ...
Be very careful Sep 16, 2005

I agree with Roberta: MT is no option, especially for XML. You or whoever will have to edit the text will end up cursing the client and have to spend hours & hours changing big chunks of text... or even translating from scratch all over again. Even if sometimes even Google automatic translator gets it right for easy things and some language pair there's nothing like a human native.
Also, like someone said, check that the original text is good and localization ready, sometimes variables and word order may be an issue... and here MT won't help you at all.
Trados or other good CAT tools can help you reuse the already translated segments and will make yr. client save some money.
Have you thought about an overall prince instead of a per word rate? Calculate a good ammount of hours, unpredictable things etc. and propose that to your client. Or some discounted price for the updates... although here it depends on how frequent they are and how many words they involve.
If it's the first time then it's important to get it right from the very start, and if the client simply has not enough money to play big games and comes up with any kind of excuses to save save save, then save your own time and go find someone else... these guys are only good at wasting your time and patience. If, on the contrary, they underestand the issues and money isn't their only concern then try to find a comprise that keep you both satisfied.

Best of luck

Stefano


Direct link Reply with quote
 

Patricia Posadas  Identity Verified
Spain
Local time: 19:02
Member (2002)
English to Spanish
+ ...
TagEditor Sep 17, 2005

No doubt this or other recognized software are the only way to really 'save' money.

I agee with the above views on MT, on source text quality and with Stefano:

"If it's the first time then it's important to get it right from the very start, and if the client simply has not enough money to play big games and comes up with any kind of excuses to save save save, then save your own time and go find someone else... these guys are only good at wasting your time and patience".

My 10 p.

Patricia


Direct link Reply with quote
 
HowardS
English
Systran is not as good as its own sales hype Sep 26, 2005

Cecilia Vela Segovia-Frund wrote:

2. Get a good MT software. I've used Systran from www.systransoft.com and results are reasonable for non-human output (only paid versions).
Cecilia Vela-Segovia
ATA Certified Eng>Spa Translator




I purchased Systran Pro v5 Global only to find it wasn't a global solution at all.

I then purchased the Arabic Translator.

Total expenditure about $1200 US.

I recently found that the translations were unreadable.

Translated web pages and incoming emails were also impossible to understand.

Despite showing the translations to Systran they did not give me a refund as I had had the software for more than 30 days.

The response contradicted the advertisement on their web page in relation to the quality I could expect and even their own "small print" contradicts it.

The software is totally overated and has many design faults and bugs in my opinion.

Dont waste your money.

Howard
Business User


Direct link Reply with quote
 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Advice on how to approach a large XML project

Advanced search






SDL MultiTerm 2017
Guarantee a unified, consistent and high-quality translation with terminology software by the industry leaders.

SDL MultiTerm 2017 allows translators to create one central location to store and manage multilingual terminology, and with SDL MultiTerm Extract 2017 you can automatically create term lists from your existing documentation to save time.

More info »
Wordfast Pro
Translation Memory Software for Any Platform

Exclusive discount for ProZ.com users! Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value

More info »



All of ProZ.com
  • All of ProZ.com
  • Term search
  • Jobs