Submitted via IRC for Bytram
[...] When I first got interested in the subject, in the mid-1970s, I ran across a letter written in 1947 by the mathematician Warren Weaver, an early machine-translation advocate, to Norbert Wiener, a key figure in cybernetics, in which Weaver made this curious claim, today quite famous:
When I look at an article in Russian, I say, "This is really written in English, but it has been coded in some strange symbols. I will now proceed to decode."
[...] The practical utility of Google Translate and similar technologies is undeniable, and probably it's a good thing overall, but there is still something deeply lacking in the approach, which is conveyed by a single word: understanding. Machine translation has never focused on understanding language. Instead, the field has always tried to "decode"—to get away without worrying about what understanding and meaning are. Could it in fact be that understanding isn't needed in order to translate well? Could an entity, human or machine, do high-quality translation without paying attention to what language is all about? To shed some light on this question, I turn now to the experiments I made.
It is a bit on the long side but Douglas Hofstadter very clearly exposes what language translation is and that Google Translate does not do it that way
Source: The Shallowness of Google Translate
(Score: 0) by Anonymous Coward on Saturday February 03 2018, @02:56PM (2 children)
It used to be really bad that it couldn't even translate a single common word correctly:
http://i.imgur.com/mzq1TLm.png [imgur.com]
http://i.imgur.com/ZlRvgBs.png [imgur.com]
Context: https://arstechnica.com/information-technology/2015/03/german-pro-basketball-team-relegated-to-lower-division-due-to-windows-update/?comments=1&post=28743959 [arstechnica.com]
(Score: 0) by Anonymous Coward on Saturday February 03 2018, @04:09PM (1 child)
Translating single words is far harder than translating simple sentences. Take a look at common words in the dictionary and they'll sometimes have a dozen or more definitions attached. And those definitions are just the ones that are common enough to appear in the dictionary. There's usually ones that are more slangy and not yet common enough for inclusion.
Then, you have to figure out which one of those definitions in the original language gets paired up with what word or phrase in the target language.
Compare that with a sentence where there's more context included and where they can just give it a known translation if need be that will usually work out reasonably well.
(Score: 0) by Anonymous Coward on Sunday February 04 2018, @03:17AM
You should see the context too. The failure was also in a context of other text. The images were from some testing after the failure was noticed.