Mobile menu

CAT Tools with word for word pretranslations?
Thread poster: Angelo Burton
Angelo Burton
Local time: 23:42
English to German
Mar 25, 2009

Hello,

I wondered if there is a CAT Tool that includes the feature to pretranslate a file word by word and not by sentence.
I installed an evaluation version of Trados and found out that it brakes down your files into segments of several words each and requires a minimum rep. of 30% before translating. When it translates something automatically it also replaces the rest of the sentence.
My customer has a big glossary with mainly single words and it's really time consuming to compare the files with his glossary by hand. Also the chance to miss something when you have to check for 250+ words is quite high.

Can you recommend a CAT-Tool to me that could help me out with this problem?

Kind regards,
Angelo Burton


Direct link Reply with quote
 

Lucy Brooks  Identity Verified
United Kingdom
Local time: 22:42
German to English
+ ...
This isn't what translation memory is for Mar 25, 2009

I don't think you will find any CAT tool that does specifically what you want.
The idea of TM is that it is a database of past translations, not an automatic machine translator.
If I had the job that you currently have in hand, I would make a lot of use of the Concordance feature which shows up previous translation units containing the term you are translating.
You can get Trados to produce any concordance hits automatically so you don't have to do too many key-strokes.
Good luck
Lucy


Direct link Reply with quote
 

Giles Watson  Identity Verified
Italy
Local time: 23:42
Italian to English
Most CAT tools will help you with terminology... Mar 25, 2009

... including Trados, but not in the evaluation version, which doesn't include MultiTerm, the glossary management program in the suite.

You can import your customer's glossary into MultiTerm and then use it with Workbench and Word or TagEditor, which also features quality assurance functions to make sure that terminology use is consistent.

Best,

Giles


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 23:42
Member
French to English
+ ...
My problem exactly! Mar 25, 2009

This was exactly the problem I encountered when I first stared looking at CAT!

The whole process of segmentation, and the tiny likelihood of segments ever repeating in the kind of work I do, meant that what I needed was exactly this kind of glossary feature. I now work (ocasionally, and only under duress!) with Wordfast, where I am constantly frustrated by its 'stupidity — completely missing a 'fuzzy match' that is only one or two words removed from the source segment, and then on another occasion proposing a 76% match for a segment that contains even very few of the same letters, let alone actual words! I find the TM feature almost worse than useless.

However, now I have found out what it is called, and all the associated jargon (did I know what a 'placeable' was...?), the 'glossary' feature more or less does what you need it to: if you create (somehow) a Wordfast-formatted glossary (not too difficult to do from your customer's word list...), it will then highlight words it finds (well, sometimes, if it feels like it...) and offer them as 'placeables' — i.e. words that you can 'place' into your target segment translation space with just a simple keyboard shortcut — and Ctrl + Alt + [dwn arr] is certainly a lot easier than the hassle of retyping "non-steroidal anti-inflammatories" dozens of times!

Whilst waiting for these things to come along, I did find another solution of my own, which actually worked even more automatically and quickly, albeit very simplistically; originally, I used to do 'search-&-replace all' manually on a document before starting to translate it (these were documents of a particularly simplistic and repetitive format), and then once I'd got a long list of words I could reliable MT in this way, I got a friend of mine to write a dedicated program in VBA which would do the whole job for me in about 10 secs flat! This was really great for pre-translation! The only drawback was that glossary management had to be done in a very cumbersome and manual way, and the glossary had to be done in a very precise order, so it could deal with inflected forms correctly. However, the developer is interested in taking this idea further, should you feel that your work is going to be sufficiently recurrent to justify a dedicated, custom solution.

[Edited at 2009-03-25 18:51 GMT]

[Edited at 2009-03-25 18:53 GMT]


Direct link Reply with quote
 

Philippe Etienne  Identity Verified
Spain
Local time: 23:42
Member
English to French
TWB and Multiterm Mar 25, 2009

Giles Watson wrote:
...You can import your customer's glossary into MultiTerm and then use it with Workbench and Word or TagEditor, which also features quality assurance functions to make sure that terminology use is consistent.
...

I am not sure if this is what you meant, but after importing the glossary to Multiterm, you may also be able to pretranslate terms only, using the relevant "Translate terms" radio button in the Translate (Tools>Translate) window. I never did it but it must be there for a reason.
Philippe


Direct link Reply with quote
 

Vito Smolej
Germany
Local time: 23:42
Member (2004)
English to Slovenian
+ ...
TRADOS with MultiTerm, DEjaVu, Transit ... Mar 25, 2009

Tony M wrote:
...it will then highlight words it finds (well, sometimes, if it feels like it...) and offer them as 'placeables' — i.e. words that you can 'place' into your target segment translation space with just a simple keyboard shortcut — and Ctrl + Alt + [dwn arr] is certainly a lot easier than the hassle of retyping "non-steroidal anti-inflammatories" dozens of times!...

This is what MultiTerm does, when used in conjunction with TWB. Of course a fat Vocabulary is a precondition for this to work - pretty much the same situation as in the case of non-existing Translation memories.


... originally, I used to do 'search-&-replace all' manually on a document before starting to translate it (these were documents of a particularly simplistic and repetitive format), and then once I'd got a long list of words I could reliable MT in this way, I got a friend of mine to write a dedicated program in VBA which would do the whole job for me in about 10 secs flat!

... and this is again something that for instance Translator's Workbench together with MultiTerm can do - pretranslation, by using both TMs and term bases. But (again) no TM, no Termbase = no fun.

This "no fun" situation is actually one of the major causes of frustration with CAT tools - expensive, cheap or free for all -: they are more of a hassle than help when starting. Which is why the first night with the new wonder weapon of a CAT tool can end in a night of long noses.

Some additional remarks: term pretranslation makes sense if you work with target languages that do not know (anymore) the declension. Otherwise it is much too much hassle.

[Edited at 2009-03-25 19:08 GMT]


Direct link Reply with quote
 

Vito Smolej
Germany
Local time: 23:42
Member (2004)
English to Slovenian
+ ...
ignore a double Mar 25, 2009

pls ignore. a double.

[Edited at 2009-03-25 19:26 GMT]


Direct link Reply with quote
 

Vito Smolej
Germany
Local time: 23:42
Member (2004)
English to Slovenian
+ ...
I never did it but it must be there for a reason. Mar 25, 2009

Philippe Etienne wrote:
...you may also be able to pretranslate terms only, using the relevant "Translate terms" radio button in the Translate (Tools>Translate) window. I never did it but it must be there for a reason.
Philippe

The segments that do not get pretranslated, get scanned word by word, source terms with matches get replaced and the resulting mixture of words (source and suggested target items - there's variants regarding how the replacement is done) gets inserted into the target segment.

It's good or great when doing price lists, catalogs etc. But otherwise - my experience - it is too much of a bother.

[Edited at 2009-03-25 19:25 GMT]


Direct link Reply with quote
 
FarkasAndras
Local time: 23:42
English to Hungarian
+ ...
nothing to see, move on... Mar 25, 2009

250 terms is not a lot at all.
Read through the list once before you start, just to get a feel for what sort of stuff is in there.
Then import the term list to multiterm, set up WB to give you terminology hits and start translating. When you run into an inflected word form that you suspect could be in the glossary, look it up.
There is no need to create a horrid mess of an in-between document by automatically inserting target language terms into the source text, unless you are the sort of person who takes pleasure in fashioning sentences out of such gibberish.


Direct link Reply with quote
 

P Forgas  Identity Verified
Brazil
Local time: 18:42
Portuguese to Spanish
+ ...
wordfast Mar 25, 2009

You can use wordfast Classic with glossary and propagation.

You can download the program and the manual here: http://www.wordfast.net/

You can use the program full funcionality with the only limitation of the TM size (500 units, I think).

P.


Direct link Reply with quote
 

bohy  Identity Verified
France
Local time: 23:42
English to French
+ ...
I agree with FarkasAndras Mar 25, 2009

By far, the memory that serves us the best is in our head. I do hate pre-translation, and I would not encourage anybody to do so. It is most likely to end up in a document of poor style, I think.
250 words is not much; I have checked much bigger glossaries, mainly trying to spot words that I would have translated differently and where I have to remember what the customer wants. This does not amount to much, usually.
After that, of course, a glossary may be helpful.


Direct link Reply with quote
 

Tony M  Identity Verified
France
Local time: 23:42
Member
French to English
+ ...
Pretranslation — a good or bad idea? Mar 26, 2009

I always thought it was complete waste of time, and then I found one particular document format I was working on over a long period of time which consisted of glorified lists of words, with little or nor grammar or structure. In this instance, the automatic prior search-&-replace function did save me a huge amount of time; any 'mess' generated was readily spotted and easily corrected, and the post-editing mainly involved tweaking word orders and dealing with residual FR terms.

Since then I have found prior search-&-replace comes in handy for a number of clients; if I encounter a tricky-to-type technical term, the second time I encounter it, I automatically do a search and replace (adding a coloured highlight to help me spot it later) — and of course being wary of things like inflected forms, plurals etc. Experience has shown that doing it this way as I go through a doc does save me time, purely in terms of typing fiddly words (never my strongest point!) and then spell-checking them later!

Recently, I was persuaded to use CAT (Wordfast, as it happens) for a particularly large project, and specifically, because the agency wanted me to build up a glossary and TM for future use (no doubt by other, cheaper translators!). Being a mug, I agreed to — and spent many sleepless nights regretting it! Not only did the actual translation take a lot longer than it normally would have done... Now it's easy to blame that on my lack of experience with W/f, and I certainly do agree that it is highly intuitive to use — but the plain fact of the matter is that I can type faster than if I have to do all the fiddling around with opening and closing segments, waiting for it to do its stuff, scrolling through alternative TUs, deleting completely stupid offers of target segments, correcting poor segmentation (often caused by poor source text punctuation, but hey! how often does that happen?). Even with terms being proposed from the glossary, it does save a bit of time, but it's all too easy to get muddled up with the multiple keystrokes involved, and sometimes it's simply less hassle to type the flippin' word! Plus of course — ay, and there's the rub! — you do have to create that glossary first ... or in my case, as you go along! Now although it is faintly encouraging when you see your source segment all full of nicely-highlighted 'placeables' (that's "words recognized from the word list" for those who only speak English!) just begging you to go "Controlsy-Altsy li'l down arrowsy" to simply 'pop' them into place, you soon realize that the differences in FR and EN word order mean it just doesn't work like that, and sadly nice familiar 'Controlsy-Zedsy' doesn't work in this mode when you $§@*£ up, as you are invariably bound to do at some point!

the actual process of creating a glossary entry, although cleverly designed, ist still very time-consuming, comparing with just keeping on typing without breaking my concentration, as I would normally do.

In a 60k word document, I created a glossary of some 3000 words (the majority single-word terms). Of those, quite a large proportion were probably accounted for by inflected versions — let's say there were actually around 2500 distinct terms. And of those, a surprisingly large number were words with more than one possible answer.

Take this as concrete example: in the particular context of my document, the FR word 'joint' occurred very frequently; leaving aside other far more entertaining possible meanings, it was likely to mean 'seal', 'gasket', or 'sealing washer'. So there it sits looking smugly down at me as an innocent turquoise 'placeable'; without another thought in the world, I gaily (don't know any other way to do it!) and dare I say slightly triumphantly hit Ctrl + Alt + → to select it, then Ctrl + Alt + ↓ to place it... I'm already half way out of my seat to go and make that well-deserved cuppa, when I see the glossary list pop up "which of the choices would you like to use?" it seems to sneer at me. I wouldn't mind a little sneer, if it at least had the intelligence to present them to me in order of likelihood, based on previous usage statistics; but no, in plain old alphabetical order, what else? My brain would have done all this on its own, without the need for smug highlighting or sneering insensitively contextual drop-downs, nor even nifty keyboard shortcuts Arthur Rubenstein would have been proud of, in less time than it takes to type 'gasket'.

I use this deliberately facile example just to illustrate some of the unexpected decisions one may have to take; I ended up wasting a vast amount of my precious translation time simply making earnest glossary entries, which I then for the most part ignored, as time got shorter and shorter, and turquoise highlights went whizzing past at a blurry speed as I manually retyped the same words, much faster (most times) — but with equal consistency! The only inconsistency I found was when I changed my translation decisions half way through; I then had to laboriously stop and manually edit the glossary entries (so it wouldn't keep offering me the old translations) — or do as I ended up doing and simply ignoring the glossary, going back afterwards and doing a correction so the glossary matched my translation. The end result is one not-very-useful (but lovingly-created!) glossary — and a translation that, far from being better and more consistent, is actually worse and less consistent than it would have been, since I ended up having to rush it so much!

What would be really wonderful would be if a glossary entry that was subsequently edited could suddenly be magically replaced everywhere throughout the text where it had ever been used — now that would be something really worthwhile! I've been told it ought to be something to do with 'propagate', but that whole subject area still remains a great mystery to me...

I am delighted though, that I have at last found out how to do regular 'slow' clean-ups, and delete TUs that I wish to revise — both of these (to me) key features proved far less intuitive to find than I would have expected from the rest of the software.

So in a way, I don't think you should worry too much about your little 250-word glossary; if the worst comes to the worst, use search-&-replace-all 250 times, or get a friend to write you a VBA routine like I did (there's no actual limit to the number of terms it can deal with, though I think from recollection mine was about 200)!


Direct link Reply with quote
 

Victor Dewsbery  Identity Verified
Germany
Local time: 23:42
German to English
+ ...
Deja Vu (Pretranslate, assemble) Mar 26, 2009

One of the advantages of DejaVuX is that it can pretranslate and assemble by using matches at the sub-segment level. It is of course rare that it will create a near-perfect translation by assembling from what it calls "portions" (although it occasionally happens), but it is an enormous help for terminology consistency.

As others have pointed out, 250 words is not a lot of material, so on its own this would not make a lot of impact on your translation, but if you already have a TM (translation memory, i.e. the sentence-by-sentence database) and a terminology database, it is then very helpful if the client gives you a project-specific glossary, which you can feed into the project-specific "lexicon".

It would be too much to describe all the bells and whistles here, but you can get a fully functional 30 day free trial from www.atril.com (to activate it, you need to get a demo licence key by e-mail).
If/when you buy, you need at least the Professional version of DVX to get the benefit of the major productivity features. But you can take your 30 days to find out whether this will be worthwhile for you.

I have happily worked with the program for almost 10 years, and most of my work does not include many exact or fuzzy matches at the sentence level, so I particularly appreciate the pretranslate and assemble functions.


Direct link Reply with quote
 


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

Moderator(s) of this forum
Maria Castro[Call to this topic]

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

CAT Tools with word for word pretranslations?

Advanced search


Translation news





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 »
memoQ translator pro
Kilgray's memoQ is the world's fastest developing integrated localization & translation environment rendering you more productive and efficient.

With our advanced file filters, unlimited language and advanced file support, memoQ translator pro has been designed for translators and reviewers who work on their own, with other translators or in team-based translation projects.

More info »



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