As a translator, I have a love / hate relationship with Google Translate. On the one hand, I share some trepidation about "the computer replacing the human" translator (while realizing that really won't happen), and at the same time, I find machine translation to be sometimes a time-saving tool (particularly when directly integrated into a CAT tool like WordFast or OmegaT). It is a mixed bag, but on the whole, I think that our world is a bit richer, because of Google Translate.
We translators know about Google Translate, but what didn't you know ? I have put together a short list of "things you probably didn't know about Google Translate":
- The head of Google Translate is a German. His name is Franz Ochs, and he is an expert in computers and language. He now works in Mountain View, Google's home base in the Bay Area of California, but he is originally from Germany. He studied in Germany and California and was later asked by the U.S. government for help after 9/11, but he went to work at Google.
- One main goal of Google Translate is to empower non-native speakers of English. Let's face it, 70% of the Internet sites in the world are written in English, with American English being particularly dominant. If you are a teenager in China or Mali or Brazil, maybe you have not mastered English yet, but want to read certain websites. Google Translate is designed to help you figure out, in an instant, what a website is about.
- Google uses as its source text four main things: Biblical texts (the Bible has been translated into every language known to man), texts from the United Nations (UN), and texts from The European Union (EU). This might be one reason that GT does a better job with European languages, than with non-European langauges. Another main source is - surprisingly - mystery novels. For this reason, Google Translate produces relatively decent legal and diplomatic texts. And if you want a chunk of a John Grisham novel translated, it can probably do that too.
- Google translate does not "think", it uses a statistical approach. From that point of view, it really - in my view - is not that different from a CAT (computer-assisted) tool. In fact, Google Translate, as I mentioned, is often used directly with a CAT tool, and it is a decent tool.
- Google Translate is amazingly bad at simple German syntax. It really is quite awful (I am speaking from personal observation here).
- Google uses English as a "pivot". A pivot is a node through which everything else flows. For instance, if you use Google Translate to translate from Greek to Norwegian, Google Translate will not match Greek source text with Norwegian target text. Instead, it matches Greek with English and then English with Norwegian. English is used as a kind of "lingua franca" or intermediate language for the tool.
- Franz Ochs, the head of Google Translate, has admitted on more than one occasion, that he does not use the tool much at all ! But he did use it on a trip to Japan, to translate menus.
- Google has human translators do its own translations. It does not use its own tool (thanks to David Bellos for that insight).
- Google Translate is helping preserve some endangered languages. Dialects of Maori that are no longer spoken, etc.
- There are confidentiality issues - for translators and translation companies - in using Google Translate ! Think about it: you are feeding your client's confidential source text into a machine that holds it "forever" and is available then to "the world", i.e. everyone. Few think about this, I am sure.
- Ochs thinks that improvements will continue in the tool, but admits he does not know where the tool's limits are.
- Google is fighting a battle - whether it knows it or not - against prescriptivism, the idea that there is a "right" way to speak, defined by pre-set rules (a view for instance that is very strong in France and in other countries). Rather than rules, it looks for how language is actually being used, "on the street", so to say (closer to the German way of lexicography).
- You can set your website so that it won't be translated by Google Translate.
- Good translators often Google Translate instead of a dictionary. One experienced patent translator said he does it "When I get lazy".

Because my relatives in Czech Republic read my blog posts through Google Translate, they often tell me about the strange things that I say in my posts when you run them through Google Translate.
ReplyDeleteFor example, my brother asked whether my dog Lucy does or does not like to eat other dogs?
What I was actually saying in English was that Lucy would like to eat my food and that she does not like to eat dog food.
Lucy definitely does not like to eat other dogs!
«There are confidentiality issues - for translators and translation companies - in using Google Translate ! Think about it: you are feeding your client's confidential source text into a machine that holds it "forever" and is available then to "the world", i.e. everyone. Few think about this, I am sure.»
ReplyDeleteIt's alarming that the majority of our colleagues fail to understand that we should NEVER use GT for our professional translations. I could flood you with links including exchanges in more than one list of fora, but judging from your "few think about this", I guess you already know. Unbelievable as it may sound, our colleagues do no see the danger for the profession (teaching the machine to replace us) and/or the legal consequences of breach of confidentiality.
THANK YOU!
To clarify, the confidentiality issue only applies to Google Translator Toolkit, not to the simple Google Translate website.
ReplyDeletePlease show me exactly how you can get someone else's confidential information out of Google Translate?
ReplyDeleteAs for "teaching the machine to replace us", think about it. Where does GT get its aligned texts? Not from what you put into it, because that would be useless: a source text paired with a less-than-perfect translation.
Sometimes translators need to use a little logic, too. :-)
"When asked whether they use Google Translate in their workflow, almost one half of the participants raised their hands. However, when I showed a slide with the following excerpts from Google’s TOS, hands clapped over mouths all around the room:
ReplyDelete“ …By submitting, posting or displaying the content you give Google a perpetual, irrevocable, worldwide, royalty-free, and non-exclusive license to reproduce, adapt, modify, translate, publish, publicly perform, publicly display and distribute any Content which you submit, post or display on or through, the Services.”
“… You agree that this license includes a right for Google to make such Content available to other companies, organizations or individuals with whom Google has relationships for the provision of syndicated services, and to use such Content in connection with the provision of those services.”
In human language that means that by uploading any content to any one of Google’s services, including Gmail, Docs, and Translate, we allow Google unlimited use of our content and give up all rights to the privacy of information. (I would love to be proven wrong on this, but after searching through all of Google’s licensing and privacy materials, I have not been able to find anything to the contrary).
Google already makes extensive use of search word analysis for displaying ads that match the content of Gmail messages. As Orwellian as that may sound, the greatest privacy issues are posed by Google Translate.
Google has also been actively crippling smaller languages by imposing "translations" that fit the English language rules. In my opinion as an educated native speaker of a smaller language, Google's human translators know crap about their native languages into which they translate. Not that it matters - the important thing is apparently that they know enough English...
ReplyDeleteDear John,
ReplyDeleteThought you might find this very interesting. "We were stunned when court interpreter, Ting Chin Kin, admitted to using the free online service of “Google Translate” to translate Teoh Beng Hock’s alleged suicide note. Was she unaware of the gravity of her work in the court proceedings?"
If our colleagues do not understand how confidentiality is compromised when online MT is used, we will be soon reading about legal actions against fellow translators. http://tinyurl.com/6tsqdu8
Best,
Oh yeah how about the hilarious translation of the Italian slogan "Dalla pecora al pecorino" (which means: "From sheep to Pecorino cheese"), translated by GT into the quite obscene "From sheep to Doggy-Style"!!!
ReplyDeleteAs a translator, I depend heavily on Google translate tool, but I've noticed that it gives some awkward translations sometimes, especially for culture-specific expressions. Yet, it still surprises me with many accurate technical terms.
ReplyDeleteThanks for the information
Ruba Khadamaljamei
Not sure why it would matter that the head of Google MT is from Germany, but if you mention it, please use the correct name: Franz Josef Och.
ReplyDelete