Pages in topic:   < [1 2]
How reliable are Machine translations in general nowadays?
Thread poster: Elke Fehling

Daniel Frisano
Switzerland
Local time: 11:00
English to Italian
+ ...
MT Jul 19, 2018

Hans Lenting wrote:

Better see the 'MT' as Mining Terminology than as Machine Translation. Don't use MT systems to translate segments, don't use it to translate complete texts. Pick what's useful for you.



I see it more as Misleading Tinsel. Its utility is dubious, it is and will always be neck-deep full of mistakes, and the risk of doing more harm than good is constantly there. Good luck finding your way with a GPS device that keeps sending you in wrong directions.

That said, please use it, and use it relentlessly. Use nothing but MT. After all, the more bad translations are there around, the more the good ones will stick out. Soon you will also trigger a self-feeding loop: the results of MT will become part of the base from which MT takes its input, until the system will eventually spin off into... I'm curious to see where!

[Edited at 2018-07-19 17:56 GMT]


 

Hans Lenting  Identity Verified
Netherlands
Member (2006)
German to Dutch
As a pro you stay in charge Jul 19, 2018

Daniel:

I see it more as Misleading Tinsel. Its utility is dubious, it is and will always be neck-deep full of mistakes, and the risk of doing more harm than good is constantly there. Good luck finding your way with a GPS device that once in a while sends you in the wrong direction.



Yes, full of mistakes, but that’s where you come in. And about the comparison with the navigation software: that is the actual end product, whereas DeepL is just a tool that you can use to create your end product. As a professional translator you stay in charge during the whole translation process.


Jorge Payan
Giovanni Guarnieri MITI, MIL
Angie Garbarino
 

Daniel Frisano
Switzerland
Local time: 11:00
English to Italian
+ ...
All-in Jul 19, 2018

Hans Lenting wrote:

Yes, full of mistakes, but etc.



Now you're starting to sound like a used car salesman trying to talk me into buying an old wreck. I hope the cut you get from DeepL is worth the effort!

Anyway, as I said, I am for using MT. It will make a lot of people happy, myself included.

[Edited at 2018-07-19 19:56 GMT]


 

Hans Lenting  Identity Verified
Netherlands
Member (2006)
German to Dutch
Not very respectful Jul 19, 2018

Daniel Frisano wrote:

Hans Lenting wrote:

Yes, full of mistakes, but etc.



Now you're starting to sound like a used car salesman trying to talk me into buying and old wreck. I hope the cut you get from DeepL is worth the effort!

Anyway, as I said, I am for using MT. It will make a lot of people happy, myself included.


Dear Mr Frisano,

First you cut away a crucial part of my statement then you make a strange insinuation.


Jorge Payan
Angie Garbarino
 

Daniel Frisano
Switzerland
Local time: 11:00
English to Italian
+ ...
Right Jul 19, 2018

Giovanni Guarnieri wrote:

Putting single words in the engine? Of course the result is going to be comical.


Good point. Let's talk about consistency over longer sentences then. I hope that we all agree that any kind of tool, to be really helpful, should at the very least maintain consistency within the same sentence. Voilà:

Untitled-1

Again, this is the very first example I tried. Took 2 minutes. It's just a gift that keeps giving.

I urge everybody to use MT though. After all, clients don't care much about consistency - especially in patents.

As for terminology mining, good thing we have Google Translator, DeepL, etc. They are certainly more accurate and comprehensive than this:

Untitled-1

[Edited at 2018-07-20 01:49 GMT]


 

Hans Lenting  Identity Verified
Netherlands
Member (2006)
German to Dutch
MT suggestions are not the end product: you still have to work Jul 20, 2018

Daniel Frisano wrote:

Giovanni Guarnieri wrote:

Putting single words in the engine? Of course the result is going to be comical.


Good point. Let's talk about consistency over longer sentences then. I hope that we all agree that any kind of tool, to be really helpful, should at the very least maintain consistency within the same sentence. Voilà:

Untitled-1



You are in charge of the consistency. Software like CafeTran will assist you by forcing DeepL to consistently use the same target term. Besides that, you can easily change any suggested target term to the one you prefer, with 2 mouse clicks.



Again, this is the very first example I tried. Took 2 minutes. It's just a gift that keeps giving.

I urge everybody to use MT though. After all, clients don't care much about consistency - especially in patents.

As for terminology mining, good thing we have Google Translator, DeepL, etc. They are certainly more accurate and comprehensive than this:

Untitled-1

[Edited at 2018-07-20 01:49 GMT]


Mining will give you the rough diamonds, which you will have to polish yourself.

Resources like LEO are excellent for many subjects, but they won’t go into detail as much as MT systems do. What’s in LEO should already be present in the glossaries of an experienced technical translator. These words are well-defined and not the challenge. For the challenges you can get good suggestions by MT. You’ll have to verify their quality yourself. That still is your work.


Giovanni Guarnieri MITI, MIL
Matthias Brombach
 

Daniel Frisano
Switzerland
Local time: 11:00
English to Italian
+ ...
Got it Jul 20, 2018

Hans Lenting wrote:

You are in charge of the consistency ...

... you can easily change any suggested target term to the one you prefer ...

... which you will have to polish yourself.

... You’ll have to verify their quality yourself. That still is your work.



Got it. I still do all the work anyway, since Artificial Dumbness can't incorporate in its famed algorithms a hyper-complicated rule like "translate like technical words/terms into like technical words/terms". Over and out.

[Edited at 2018-07-20 11:11 GMT]


 

Daniel Frisano
Switzerland
Local time: 11:00
English to Italian
+ ...
Ultimate proof of utter stupidity Jul 20, 2018

Just to end on a high note:

Untitled-1

This moron can't even recognize itself. That's artificial intelligence for you.

I encourage everyone to use this worthless piece of junk, especially in EN>IT, DE>IT, FR>IT. You will make a lot of us happy.


 

Elke Fehling  Identity Verified
Local time: 11:00
English to German
+ ...
TOPIC STARTER
Funny Jul 20, 2018

Daniel Frisano wrote:

This moron can't even recognize itself. That's artificial intelligence for you.

I encourage everyone to use this worthless piece of junk, especially in EN>IT, DE>IT, FR>IT. You will


I don't agree with everything you say, because I still find this tool quite useful. But your example is very entertaining


 

Jean Dimitriadis  Identity Verified
France
Local time: 11:00
Member
English to French
+ ...
DDDL Jul 20, 2018

Don't dig too DeepL…

 

Hans Lenting  Identity Verified
Netherlands
Member (2006)
German to Dutch
Don't dig too deep Jul 21, 2018

Jean Dimitriadis wrote:

Don't dig too DeepL…


DeepL has been mainly fed with instructive texts. It performs well ('above average') when translating such texts. Don't expect DeepL to perform well on patents and the like.


Giovanni Guarnieri MITI, MIL
Angie Garbarino
 
Pages in topic:   < [1 2]


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


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

How reliable are Machine translations in general nowadays?

Advanced search






SDL MultiTerm 2019
Guarantee a unified, consistent and high-quality translation with terminology software by the industry leaders.

SDL MultiTerm 2019 allows translators to create one central location to store and manage multilingual terminology, and with SDL MultiTerm Extract 2019 you can automatically create term lists from your existing documentation to save time.

More info »
TM-Town
Manage your TMs and Terms ... and boost your translation business

Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.

More info »



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