Pages in topic:   [1 2 3] >
CAT vs MT Tools??
Thread poster: JonAnderson

JonAnderson
Canada
Apr 13, 2015

Hi,

I am completely new to the world of translation. A friend of mine
is an executive of a multinational company and is helping me out
financially by offering me some translation work. The first
project is a large (120,000 word help file for CRM software) which
will be followed by other translation projects.

I researched the tools used by professional translators, found
this incredible website and read and watched tutorial videos on
all the major CAT software. Unfortunately, since I am not planning
to be a full-time translator (I have another career), I found the
CAT software model to be too laborious because of the large
quantity of initial translation needed to be done manually to
build the TMs from the ground up.

So I turned to MT based software as it offers instant translation.
I quickly zeroed-in on SYSTRAN which appears to be one of the most
established and robust option. I fully realize that I will need to
edit the initial translation and that having a TM tool inside the
MT software is essential. This lead me to considering buying the
SYSTRAN Premium (hybrid MT).

A few questions before I make that substantial investment in the next day or so as I need to get started this week:

- is SYSTRAN Premium the better MT option available?
- is it a mistake, considering my situation, to shy away from CAT software?
- what are the major trade offs in using MT instead of CAT?
- and most importantly, is SYSTRAN MT initial translation high quality?

Thanks in advance!
Jon


 

Tom in London
United Kingdom
Local time: 16:07
Member (2008)
Italian to English
Well put Apr 13, 2015

JonAnderson wrote:

.......... I found the
CAT software model to be too laborious because of the large
quantity of initial translation needed to be done manually to
build the TMs from the ground up......


Well put. That's why I can't be bothered with CAT tools. I've had a look at SYSTRAN but I think it s***s !

So far as CAT tools are concerned here's an amusing thread:

http://tinyurl.com/nyp66wh

[Edited at 2015-04-13 16:17 GMT]


 

Georgie Scott  Identity Verified
France
Local time: 17:07
Member (2009)
French to English
+ ...
A note on MT Apr 13, 2015

There are a couple of members on this site that are extremely against MT. Personally I find some of the animosity towards it is a little unjustified and would recommend taking it with a pinch of salt.

I can't give you any advice on that software though, I tend not to get involved with MT at the moment - because I have other things going on, rather than because I'm scared of the machines taking over.

Your comments about CAT tools are fairly justified, but if you are working on a 120,000 word project, once you are just a third of the way through, you'll probably already have quite an interesting TM already. It would also give you the option of creating a term base as you go along, so that the terminology you've decided on is suggested to you automatically - a bit like T9 on a phone.

In any case, it's up to you, but watch out for the trolls. And if you haven't already, look into term bases.


***That comment isn't directed at you by the way, Tom. I wouldn't be surprised if SYSTRAN is rubbish.

[Edited at 2015-04-13 16:43 GMT]


 

marcosky
English to Italian
+ ...
Omega T Apr 13, 2015

And what's about OmegaT?

 

autor  Identity Verified
Portugal
Local time: 16:07
Member
Portuguese to English
+ ...
In defence of CAT tools (sort of) Apr 13, 2015

Whilst I agree with some of the drawbacks and shortcomings highlighted here and in the associated thread, their usefulness really depends on the nature of translation work you get involved with.

For example, if you work for the same client for several years, and if there are routine reports that come up at least once a year, with only say less than 50% of the text changed from the previous year, then the client has an incentive to remain loyal, because by means of your TM, you will be quicker and cheaper than your competitors.

Also there are many clients out there that even if they know you use a CAT tool, don't insist on discounts, or that you hand over the TM or a bilingual version of the document.


 

Rolf Keller
Germany
Local time: 17:07
English to German
An old common saying: You can get it fast, cheap or good – pick any two. Apr 13, 2015

JonAnderson wrote:
120,000 word help file for CRM software


Such a file consists of thousands of short text snippets. In order to translate them into a well understandable and really helpful result the translator has to know and to take into account many references to the CRM software's details, dialog windows etc. But an MT software will not know anything about that CRM software; actually it will not even be able to discern which words are "special terms" (proper names, nouns that are converted into verbs, .... let alone sloppy language, shop language and wrong capitalization).

A simple example: "Put the customer into the number box" has to be translated like "Enter the customer's ID number into the Cust. No. box". How will MT software cope with that?

Depending on the conditions (your abilities as well as quality & style of the source text) the post-editing might take more time than a translation from scratch. My personal opinion: Post-editing is more difficult than translating.


 

Lori Cirefice  Identity Verified
France
Local time: 17:07
French to English
Depends on the type of document Apr 13, 2015

JonAnderson wrote:


- is it a mistake, considering my situation, to shy away from CAT software?



I'm not familiar with MT software at all, have only read a few things here and there. I do know that you can use MT within some CAT tools, kind of like a plug-in, but I have never tried these.

Have you tried analyzing your 120,000 word project (with the Wordfast demo version for example) to see if it has any repetitions? If there are repetitions, you should consider using some form of CAT.

You would probably benefit from the glossary function of a CAT regardless of whether there are repetitions, especially for such a large project with more to come from the same company.


 

brg (X)
Netherlands
I think it is a mistake Apr 13, 2015

JonAnderson wrote:
- is it a mistake, considering my situation, to shy away from CAT software?


In fact, you have two options:
- use a CAT, and probably a part of its segments are reusable (to be analyzed before: 30%? 50%? 70%?)
- use MT (Systran or another) and you will have 120,000 words to revise entirely, maybe even retranslate, and to check consistency all over the files.

Good luck


 

Dan Lucas  Identity Verified
United Kingdom
Local time: 16:07
Member (2014)
Japanese to English
Termbases, definitely Apr 13, 2015

interpretwhisky wrote:
***That comment isn't directed at you by the way, Tom.

Well, just so that our new friend doesn't get the wrong idea, I think it is only fair to point out that while Tom is of course a wonderful chap, he is also a fervent anti-CAT person and in this he does not represent the mainstream. Full disclosure and all that.

Like interpretwhisky, I think the OP would get a fair bit of mileage from a good CAT tool with a 120k file that probably has a lot of repetition. And I find termbases far more useful than the TM.

Dan


 

Michael Beijer  Identity Verified
United Kingdom
Local time: 16:07
Member (2009)
Dutch to English
+ ...
CAT tool (with good terminology module), definitely. Apr 13, 2015

You'd have to be insane (or have a photographic memory) to translate a "120,000 word help file for CRM software" without a CAT tool. It'll probably contain nine gazillion repeated fragments, which I assume you will want to keep consistent somehow.

If you want to use MT, just use a CAT tool with an MT plugin (pretty much all of them nowadays).

[Edited at 2015-04-13 18:10 GMT]


 

FarkasAndras  Identity Verified
Local time: 17:07
English to Hungarian
+ ...
I would... Apr 13, 2015

Forget about MT and go with a plain CAT tool. MT is likely to introduce a level of unwanted complication. Building up a good TM may take time, but that's not the right way of looking at it. If a specific sentence is repeated 5 times in your text, you will only have to translate it once. The other 4 times you will just have to confirm the automatically retrieved translation (your translation, so you don't have to go through it with a fine-toothed comb again). Also, you can import termbases (term lists) and translation memories (previous translations) to jump start the database creation process. If the company had similar documents translated in the past, you should try and align them (=generate translation memories out of them).

Download the free trials of two or three main contenders and spend at least a day testing and researching each of them before you buy.


 

Philippe Etienne  Identity Verified
Spain
Local time: 17:07
Member
English to French
I'd use a CAT tool and not MT Apr 13, 2015

JonAnderson wrote:
The first project is a large (120,000 word help file for CRM software) which
will be followed by other translation projects.

While I understand that newcomers may find that translation takes too long to do from scratch and would rather have a pre-chewed MT version to work on, I am not sure that a raw MT program will save you much time, especially in its out-of-the-box form. Now if the translation user just needs a quick, dirty reference in-house, then MT may suffice.

As help files are usually quite repetitive, I'd choose a proper CAT tool and insert terms in a glossary as I go along. Depending on the amount of repetitions/similarities, I may be able to finish in 1.5 months working full-time on it (5 days/week). Or not.
For what it's worth, if the user interface (the actual program captions on the screen, such as Open File, Show Report, error messages, etc.) is meant to be translated, it should be done before (or in parallel with) the help files, as the latter constantly refer to the UI. You will also need to know the UI terms for the OS used with the CRM program.
Software projects can quickly get very messy if not properly managed.
JonAnderson wrote:
- what are the major trade offs in using MT instead of CAT?

It's like comparing the strengths/weaknesses of a saw vs. a hammer. Although they can both be used in carpentry, they don't serve the same purpose.
JonAnderson wrote:
- and most importantly, is SYSTRAN MT initial translation high quality?

From what I have seen, MT can save time (meaning: a "fit-for-purpose" outcome. Flow and language quality is another matter) when it is fed beforehand with human translations - usually from a CAT translation memory - and/or specific glossaries. If your MT program is not customised to your specific project, I don't think it will do any better than the free online tool Google Translate.

If similar projects come up afterwards, if your customer's quality requirements are "post-edited MT" and not "human-like quality", and if you have time to customise an MT program, maybe it could make sense to "teach" it the TM from the CAT tool and pre-translate subsequent batches with MT.

Philippe


 

Christine Andersen  Identity Verified
Denmark
Local time: 17:07
Member (2003)
Danish to English
+ ...
I would go for a CAT from day 1 Apr 13, 2015

Which CAT is a matter of taste.

Whether you like MT or not depends on a lot of factors, one being your language pair. For some the results are helpful. I generally find that it is not. I spend more time removing the inaccurate translations and correcting them than I save with anything helpful.

I have never tried SYSTRAN, but I think others I have worked with have. I have come round to having an open mind about MT, but I am not convinced.

Not all CATs are expensive. Some are free, or you can get free demo versions that work quite well.

I think in terms of productivity and getting paid, the latest Trados Studio pays for itself in spite of being one of the most expensive CATs on the market, but I have used Trados for many years now. I stay with it because I like the glossary, Multiterm, that goes with it, but others find that hard work. OK, it is, but once you get the hang of it, it has useful features that the simpler models don't have.

MemoQ is another popular option, and for a while I liked WordFast Classic, which is free until you build up quite a large TM, and then is reasonable value for money, but I got fed up with the glossary.
___________________________

If I wanted to work on a big project like yours, I would start feeding the glossary at once to keep track of the terminology.
(In Trados you can do it fast by highlighting the source and target term as you translate, and then clicking save, so it only takes a few seconds.)

Although it is true that the longer you use your CAT the more you benefit from it, you do actually benefit from day 1 with larger projects and around job 2 or 3 for the same client if you do smaller ones (under 1000 words each) as I do.

With a help file for CRM software I bet there are dozens of phrases, if not whole sentences, that come up again and again.

Feed them into your glossary, and Multiterm will insert whole strings, correctly spelled, at when you type the first four or five letters.

A few random strings from mine:

Ministerial Order on the Appointment of Academic Staff at Universities

This delightful string comes up now and then:

Bekendtgørelse nr. 806 af 12. juli 2004 om information og samtykke ved inddragelse af forsøgspersoner i biomedicinske forskningsprojekter

-- and Trados immediately finds:
Ministerial Order No 806 of 12 July 2004 On Information and Consent at Inclusion of Trial Subjects in Biomedical Research Projects

-- product names like Halvblonde -> Half Lace (one of Royal Copenhagen's patterns). I canot remember all of them offhand...

and lots more.
___________________________________

The big advantage of the CAT is that it is reasonably reliable. Of course you should never accept anything uncritically - the best translations may get out of date, and you have to be careful about updating the TM with the final version of your text after proofreading, or it will contain the typos and errors you have removed from the text.
However, with a little discipline, that is neither difficult nor time-consuming.

Machine translation is partly based on human translation, but it is still a lottery, even though sometimes the odds are more favourable than unfavourable.

I have exchanged TMs with colleagues or received them from agencies, and I can see who translated each segment and when as it comes up.

I can instantly correct or delete any errors I come across, and it is possible to update TMs more systematically, though it is still quite a job if they are large.

Anyway, when all is said and done, I find a CAT is well worth the money and time spent learning to use it properly.

Even our friend Tom who hates CATS has a system of macros ...

There is quite a good discussion of different CATs here:

http://www.proz.com/forum/getting_established/282653-cat_software_for_start_up_advice_needed.html

_____________________________

Whatever you decide, good luck with your project!


 

Tom in London
United Kingdom
Local time: 16:07
Member (2008)
Italian to English
me Apr 13, 2015

Can everyone please stop talking about me and my presumed notions, and focus on the topic?

TIA


 

Diana Obermeyer  Identity Verified
United Kingdom
Local time: 16:07
Member (2013)
German to English
+ ...
Play nice. Apr 13, 2015

Tom in London wrote:

Can everyone please stop talking about me and my presumed notions, and focus on the topic?

TIA


Anyway...

Personally, I haven't had favourable experiences with MT. However, I'm not very good at coping with linguistic torture, and sadly this is a common side effect of such tools. I couldn't comment on time savings, as I enter nervous breakdown mode when I have to read a substantial text of this kind, wherefore a comparison is rather tricky.

Yet, I fully understand that different people have different preferences. We all have different favourite subject areas, and some are more inclined towards editing tasks. The important thing is that you don't rely on the MT output too heavily and are confident that you will know when specific terminology is required. You may find it useful to have the framework in place and edit it yourself. I had a quick look at the product descriptions, and the premium version you mention has some of the functionality we (not speaking for Tom) enjoy with CAT tools - it remembers your terminology choices, you can build glossaries and it creates a TM in tmx format (which is compatible with most CAT tools).

The homepage also features a box where you can enter a random text to see the results. Why not enter a few phrases from your project to determine whether or not you consider the output helpful.

[Edited at 2015-04-13 19:35 GMT]


 
Pages in topic:   [1 2 3] >


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


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

CAT vs MT Tools??

Advanced search







SDL Trados Studio 2019 Freelance
The leading translation software used by over 250,000 translators.

SDL Trados Studio 2019 has evolved to bring translators a brand new experience. Designed with user experience at its core, Studio 2019 transforms how new users get up and running, helps experienced users make the most of the powerful features, ensures new

More info »
Protemos translation business management system
Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!

The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.

More info »



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