Timeline for What gets lost in translation?
Current License: CC BY-SA 3.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 25, 2017 at 10:19 | comment | added | ninjalj | @AlexP: expanding on gender, there are different gender systems, e.g. while modern IndoEuropean languages have a system of male/female/neuter gender, other languages have animate/inanimate gender, while others have no gender systems, yet others have really complex gender systems... | |
| May 25, 2017 at 10:14 | comment | added | ninjalj | @EricLippert: Google Translate has transitioned away from Markov models to neural networks translating sentences for lots of language pairs. | |
| May 24, 2017 at 17:00 | history | edited | AlexP | CC BY-SA 3.0 |
Brevity
|
| May 24, 2017 at 16:53 | comment | added | Mike L. | @EricLippert Google's claim is closer to reality than Microsoft's (they do translate words, or attempt to anyhow). What it says on Microsoft website is probably some marketing person's interpretation of the tool which bears little resemblance to how either a computer scientist or a linguist would describe it. | |
| May 24, 2017 at 8:54 | history | edited | AlexP | CC BY-SA 3.0 |
On
|
| May 24, 2017 at 8:47 | history | edited | AlexP | CC BY-SA 3.0 |
On
|
| May 24, 2017 at 0:30 | comment | added | AlexP | @EricLippert: So they are not lying, they are just describing the function of their services; it doesn't matter that what the machine does is not actually translating the input text if the humans who read the output perceive it as a translation. Neither Google nor Bing will take any responsibility for the equivalence of the input and output texts: the services are provided "as is". They are infinitely better than nothing and quite a lot better than previous efforts at machine translation. | |
| May 24, 2017 at 0:26 | comment | added | AlexP | @EricLippert: They provide free services transforming input text in output text, which, in most situations, can be interpreted by a human reader as translation. It is a very valuable service because, for example, without it I cannot read Chinese at all; however, since the computers which do the text transformation do not actually extract meaning from the input text the results are unrealiable. For example, there was a recent article discussing how Bing made the French phrase "mes chers compatriotes" translate to "my fellow Americans". | |
| May 24, 2017 at 0:17 | comment | added | Eric Lippert | @AlexP: So just to be clear, your claim is that when Google Translate says "Google's free service instantly translates words..." and when Microsoft Translator says "Microsoft Translator enables you to translate text or speech,..." and so on, that they are lying? or are they perhaps merely mistaken about the service they are providing, and in fact they provide some service other than translation? I find this claim... interesting. | |
| May 24, 2017 at 0:13 | comment | added | AlexP | @EricLippert: Whatever Google Translate does is not translation, it is a form of text transformation which humans can usually interpret as translation. Language Log has a recurrent topic discussing bizarre Google Translate output, with occasional surrealist poetic excursions; see for example Your gigantic crocodile and the locus classicus, The sphere of the sphere is the sphere of the sphere. | |
| May 23, 2017 at 23:23 | comment | added | Eric Lippert | Translation by humans is not done word-by-word. Google Translate makes no attempt to do a grammatical or semantic analysis of its input; it just applies a Markov model to sequences of individual words in one language and tries to make the best correspondence to sequences of words in other languages. Which is why if you type in bits of United Nations public documents into Google Translate you get great translations. Everything else, not so much. | |
| May 23, 2017 at 10:26 | comment | added | WGroleau | ¡Lo que realmente quiero es un perro caliente! | |
| May 23, 2017 at 8:50 | history | edited | AlexP | CC BY-SA 3.0 |
Wording
|
| May 23, 2017 at 8:45 | history | answered | AlexP | CC BY-SA 3.0 |