|
 |
GoogleTranslator is killing my future | ||
 Language Learning Forum : Languages & Work |  | |
| laconic Newbie United States ourmadworld.com Joined 4906 days ago 1 posts - 1 votes Speaks: English* Studies: Spanish, Latin | Message 33 of 63 22 June 2011 at 7:37pm | IP Logged | ||
Careful with using "never." If you can learn idioms, I can program a machine to translate them given enough time and resources. Our brains are not magic. They are just optimized to function well with the communication we utilize. Computers are optimized for different purposes. This may not always be the case. 1 person has voted this message useful | |||
| hjordis Senior Member United States snapshotsoftheworld. Joined 5188 days ago 209 posts - 264 votes  Speaks: English* Studies: French, German, Spanish, Japanese | Message 34 of 63 22 June 2011 at 7:43pm | IP Logged | ||
I do think that eventually machine translation will take over. I don't know when that will be, but I think a complete take over is far in the future. I'm not too worried about it. Even though I want to be a translator in the future, there are other things I can do. 1 person has voted this message useful | |||
| tornus Diglot GroupieRegistered users can see my Skype Name Joined 5145 days ago 82 posts - 113 votes Speaks: French*, English Studies: Spanish, Swedish, Danish | Message 35 of 63 22 June 2011 at 9:19pm | IP Logged | ||
|
When there is not enough information and context, that is to say most of the time, an online translator can't give a good translation.
A human translator can analyze sentences and figure out what's missing or come up with a good proposition. 1 person has voted this message useful | |||
 Iversen Super Polyglot Moderator Denmark berejst.dk Joined 6705 days ago 9078 posts - 16473 votes Speaks: Danish*, French, English, German, Italian, Spanish, Portuguese, Dutch, Swedish, Esperanto, Romanian, Catalan Studies: Afrikaans, Greek, Norwegian, Russian, Serbian, Icelandic, Latin, Irish, Lowland Scots, Indonesian, Polish, Croatian Personal Language Map | Message 36 of 63 23 June 2011 at 2:31am | IP Logged | ||
No, but that is not the point here - we all know that translation programs sometimes make gross errors, and often the same types of errors again and again. The point is that legions of proofreaders lost their job when spell checkers became reasonably, but not totally trustworthy. And now translation programs are moving in that direction. Right now the translation programs have problems on certain points (and 'forgetting' negations is also a problem for Google T with other language combinations), but I am fairly sure that many of these problems can and will be solved. And then many instutions and home page owners will swallow a couple of camels and leave the translation of their stuff to the machines, leaving just some literary and juridical and highly technical texts for the human translators. Interpreters will probably survive 5-10 years longer than translators, thanks to the incurable mumblers, babblers, snufflers, spoonerizers, lovers of rare words and surviving perfectionists of this world. But their fate is probably also sealed. Edited by Iversen on 23 June 2011 at 2:37am 3 persons have voted this message useful | |||
| hrhenry Octoglot Senior Member United States languagehopper.blogs Joined 5132 days ago 1871 posts - 3642 votes Speaks: English*, SpanishC2, ItalianC2, Norwegian, Catalan, Galician, Turkish, Portuguese Studies: Polish, Indonesian, Ojibwe | Message 37 of 63 23 June 2011 at 3:13am | IP Logged | ||
There are still proofreaders out there doing their job, just not as many. I don't believe interpreters will become extinct either. The numbers may decrease, but they'll still be around. I have a very hard time picturing any bureaucratic organization such as the UN or anything having to do with the EU going without any interpreters in my lifetime, at least. The same goes for legal and certain medical proceedings. The outcome of something mistranslated or misinterpreted causing irreversible harm is too great. And really, I think that a human face needs to be put on all things political, legal and medical, due to their inherent personal nature. R. == 3 persons have voted this message useful | |||
| translator2 Senior Member United States Joined 6921 days ago 848 posts - 1862 votes Speaks: English* | Message 38 of 63 23 June 2011 at 3:36am | IP Logged | ||
|
I would not worry about it. The new translation systems (like Google Translate) use a statistical method to make guesses as to the translation using bilingual texts translated by human beings. Without examples of human translation, they cannot work. In fact, many times Google omits words or phrases it does not "understand" and sometimes the "translation" is the complete opposite of the source text. This is not always obvious when the text is somewhat plausible.
More importantly, computers must be programmed and as a cursory glance through the various translator help sites and glossaries will show, many times even professional translators cannot agree on the correct translation of a given term. Will we give the computer all possible translations? How will it decide which one to use? Take the following sentence - He has a part in his hair. Native English speakers will immediately recognize that this refers to the way this particular gentleman combs his hair. A computer (without experience as a living breathing human) would not be able to rule out the unlikely possibility that this person is walking around with a part or piece of some object in his hair. Indeed, translating this sentence with Altavista’s Babel Fish results in: German: Er hat ein Teil in seinem Haar. French: Il a une pièce dans ses cheveux. Spanish: Él tiene una pieza en su pelo. A more famous example is “Time flies like an arrow.” (from Pinker’s “The Language Instinct” I think) Our human intuition tells us that this refers to how quickly time seems to pass. A computer, however, would have to somehow eliminate the following possible translations: “Take a group of flies (insects) and time them in the same manner as one would time (the speed of) an arrow.” “Time flies (a type of fly like a fruit fly, horse fly, etc.) are particularly fond of a certain arrow.” “Take a group of flies (insects) and time them in the same manner as an arrow would time them.” “Time the speed of flies that look like an arrow” Of course you can program the computer to "understand" these specific examples and translate them, but there are always new sentences. This just concerns meaning. Computers would also have to contend with cultural adaptations, emotions, localization, recognition of source language errors, etc. etc. etc. Here is a video about Machine Translation in 1954 and predictions for the future: Machine translation in 1954 What I have noticed with the recent incarnation of translation software (based on frequency statistics rather than a simple dictionary) is that many times the program seems to come up with a "translation" that, although ungrammatical, may seem to make sense to the monolingual person, but is completely wrong. It also sometimes just leaves a lot of words out for no reason. 1) Here is a simple example from Italian to English: Papà parlava di lavoro e mamma si occupava dell'ospitalità. [literally: Dad talked about business and Mom took care of the hospitality] Google's translation: Mom and Dad talking business took care of hospitality. By what rule did Google combine both subjects together (especially since both verbs are singular)? Why did Google interpret this sentence in this manner? 2) Another example: Tutte queste percezioni hanno contribuito a formarmi e me ne rendo conto nelle fasi decisionali. Google's translation: All these perceptions have contributed to and I realize in the early stages of decision making. For some reason, Google omits the word "formarmi" ([contributed to] training/molding/shaping me). 3) È un’azienda innovativa e flessibile. Google's translation: It is an innovative and flexible. [omits the word "company"] Why? 4) Sometimes a particular nuance is omitted: Purtroppo non è più possibile un contatto diretto con tutti. [literally: Unfortunately, direct contact with everyone "is no longer" possible.] Google's translation: Unfortunately you can not direct contact with everyone. Finally, there is this article from Microsoft: The Microsoft Translator team has given up and concluded that “no matter how many machines you throw at translation, it is still impossible to get the correct, error-free, contextually accurate translation every time.” Microsoft’s solution to this problem is the Collaborative Translations Framework, which supposedly combines the scale and speed of automatic machine translation with the accuracy and context awareness of human translation: Link to complete article Edited by translator2 on 23 June 2011 at 3:43am 5 persons have voted this message useful | |||
| Iversen Super Polyglot Moderator Denmark berejst.dk Joined 6705 days ago 9078 posts - 16473 votes Speaks: Danish*, French, English, German, Italian, Spanish, Portuguese, Dutch, Swedish, Esperanto, Romanian, Catalan Studies: Afrikaans, Greek, Norwegian, Russian, Serbian, Icelandic, Latin, Irish, Lowland Scots, Indonesian, Polish, Croatian Personal Language Map | Message 40 of 63 23 June 2011 at 10:38am | IP Logged | ||
|
Originally developers thought that machine translation should be solved by combining a grammar and a dictionary ... this sounds like a caricature of black school language teaching.
Now they have tried for a while only to use statistical methods ... this sounds like a caricature of grammarless 'natural' learning, except that the comprehensible texts have been dropped in favor of bilingual texts. Somewhere in between it must be possible to use the statistical methods, but within a framework where information from grammars and from dictionaries with morphological information is taken into account. This would for instance reduce the number of lost negations, errors of congruence and botched-up pronouns. It would also be a good idea to let the program check proper names so that for instance country names and languages aren't translated with names of other countries and languages. Google should be able to deliver that information to Google Translate. Finally the treatment of 'unknown' words could be improved a lot. It is stupid just to drop any word that the computer can't translate, and if it stays untranslated in the text it should be marked so that it isn't seen as a loanword in the target language. But most of the 'missing' words could be found in any decent dictionary, so this is just one more case where the pure statistical methods aren't sufficient. However the cure is there: just look all such single words up in a dictionary and evaluate the relevance of the possible translations - that's much better than just dropping them. The most problematic case is that where one language simply doesn't care about information that is vital in another (such as the lack of information about numerus or tempus in Bahasa). But what is the risk? Even a random choice between two possible constructions should give 50% correct guesses, and if you use some heuristic rules of thumb to make the choice (for instance the frequency of present tense verbs against that of imperatives, or that of singular against plural) this success rate can be augmented a lot. This is one area where heuristic rules made by AI could be relevant. Human translators can of course draw upon their knowledge of the world and simple logic to clear up cases of colourless green ideas and timing flying flies. But the example with the flies is based on not one, but TWO ambiguous words in a language with few morphological markers - most sentences aren't that ambiguous. Poets and other people with a keen interest in wordgames will of course always delight in constructing clever traps, but the important thing is that there are ways to make machine translations better, and this will happen. This assumption is not based on a prediction made by Fred IBM Flintstone 57 years ago, but on what has happened within the last five years or so, plus some analysis of typical errors made by Google T and other translation systems of 2011. I have not said that translators will die completely out, but at some point people won't pay for their services when the alternative is a passable machine translation. Edited by Iversen on 23 June 2011 at 11:10am 2 persons have voted this message useful |
|  | 
Printable version
|
| Copyright 2024 FX Micheloud - All rights reserved No part of this website may be copied by any means without my written authorization. |
