The new improved Google Translate
Thread poster: Barnaby Capel-Dunn
I don’t know if any of you are familiar with an old Charles Addams cartoon showing a group of Christians preparing to meet their fate in the Colliseum. Just before the gates are thrown open, one of them sees the agenda giving details of the day’s "events" pinned on the wall. He glances through it and then turns to the others. "Holy smoke!", he says, "Have you guys seen this script?"
I felt much the same way when I tried out the new improved Google Translate engine, now using statistics rather than algorithms to generate its translations. (Previously it had been using Systran’s engine for all but a handful of languages.) I’ve been trying it out between French and English and am pretty impressed by the results it gives with what might loosely be termed information-based/technical texts. Perhaps not so much by the results themselves as by the potential for improvement in the future.
What does this mean for our profession?
Here is my spontaneous, unconsidered reaction :
- The PERCENTAGE of all translations handled by professionals is likely to decrease as people will rely more and more on the web for a wide variety of purposes. (This does not necessarily mean that the VOLUME treated by us will decline.)
- This is bad news for companies like Language Weaver currently offering statistical machine translation for a fee.
- Is it bad news for the likes of Trados in the long run? I don’t know, but given Google’s overall ambitions, I should have thought so.
- And what about us translators? My own view is that in the medium-term and for all but "editorial" translations, we are going to become proofreaders as much as if not more than translators. I can’t see any escaping that conclusion.
As I say, these remarks are not the fruit of great reflection, and there is an element of provocation about them.
What do you think ?
| || || |
| | Margreet Logmans
Local time: 17:46
English to Dutch
| Depends on a number of factors || Oct 25, 2007 |
I haven't tried the tool yet, but here are a few general observations.
I think the language combination makes a difference. Mainstream languages will sooner be fed into a computer than exotic ones. So those of us that translate in exotic combinations won't have much to fear any time soon.
Also, I think the type of documents matters. Legal and literature, for example, won't be left to computers easily.
On the whole, I'm not afraid yet. The profession will change, certainly. But I don't know if it will change in a big way.
After all, someone has got to run the texts through the Translate Engine, which takes time, and then they have to check and read the translation, or find someone to do this for them. I wouldn't be surprised if it turned out to be, not only safer in terms of quality, but also more economic in terms of time and fees to hire a human translator. Secretaries get paid by the hour too...
So don't give up hope yet! Remember Daniel survived the lion's den!
[Edited at 2007-10-25 10:12]
[Edited at 2007-10-25 10:13]
| || || |
| Medical/pharmaceutical should be fairly safe || Oct 25, 2007 |
At least, I hope so! But it's such a fast-moving field that it would be difficult for machine translation programs to keep up - and the importance is such that I doubt any respected pharma companies (if that's not an oxymoron) would want to rely on machine translation.
There's also to consider that a minor but significant part of this field involves the deciphering and translation of (extremely badly) handwritten medical records, and I can't see how that can ever be done without human input.
| | Tim Drayton
Local time: 18:46
Turkish to English
| Not so impressed || Oct 25, 2007 |
I agree that this is a development we need to watch. Out of curiosity, I entered the following brief German text, which I don't think is particularly demanding and involves no idiomatic language:
Mag. Georg Bauthen, Country Managing Partner von Ernst & Young Österreich:
„Dieser Award-Bewerb findet bereits in über 100 Städten in 40 Ländern statt und
weltweit nehmen mehr als 10.000 Unternehmerinnen und Unternehmer teil. Wir
sind sehr stolz, dass wir – Ernst & Young Österreich – diesen Bewerb nunmehr
auch in Österreich mit so großem Erfolg etablieren konnten. Selbstverständlich
findet der Entrepreneur Of The Year nunmehr alljährlich statt und wir beginnen
bereits morgen mit den Vorbereitungen und Einladungen für 2007.“
and obtained the following English translation:
Georg Bauthen, Country Managing Partner of Ernst & Young Austria: "This award competition will take place in more than 100 cities in 40 countries worldwide, and take more than 10,000 entrepreneurs. We are very proud that we, Ernst & Young Austria - this competition is now also in Austria with so much success to establish. Naturally, the Entrepreneur Of The Year every year, and now we begin tomorrow with the preparations and invitations for 2007. "
My first impression is that while all three sentences in the German text are full gramatical sentences, only the first translated sentence is even a proper sentence. The German tells us that the competition IS ALREADY TAKING PLACE (findet bereits ... statt ), whereas the translation tells us that it "will take place". It has totally failed to handle the separable verb "teilnehmen": while the original text talks of more than 10,000 entrepreneurs TAKING PART, the translation gives us some garbled nonsense about "taking entrepreneurs".
In the second sentence, we are told that E&Y are very proud to have been able to introduce the competition to Austria. The translation seems to ignore the existence of the past tense of the very common verb meaning "can/be able", i.e. "Konnten", and gives us the garbled "with so much success to establish".
In the third sentence, we get "Naturally, the Entrepreneur Of The Year every year, ..." and the words "findet ... nunmehr alljährlich statt" have been totally glossed over, failing to inform us that "will hanceforth take place every year".
Based on this result, I think MT still has a long way to go before it can provide anything more than the basic gist of the text, if even that.
| || || |
| | Isabel Fernandez
Local time: 11:46
English to Spanish
| Volume decrease || Oct 25, 2007 |
I am convinced that translations for editorial purposes will continue to be handed to translators. The issue is that now all of us are going to be competing for less work available (since many other texts will just be run through the automated tool). This may bring prices down, and I think we should all get prepared to receive less volume.
To report site rules violations or get help, contact a site moderator:
You can also contact site staff by submitting a support request »
The new improved Google Translate
|Déjà Vu X3 |
|Try it, Love it|
Find out why Déjà Vu is today the most flexible, customizable and user-friendly tool on the market.
See the brand new features in action:
*Completely redesigned user interface
*Inline spell checking
More info »
|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 »