Register  Login  Active Topics  Maps  

Who is Creating Parallel Texts?

  Tags: Bilingual texts
 Language Learning Forum : General discussion Post Reply
76 messages over 10 pages: << Previous 1 2 3 4 5 6 7 ... 9 10
kaptengröt
Tetraglot
Groupie
Sweden
Joined 4273 days ago

92 posts - 163 votes 
Speaks: English*, Swedish, Faroese, Icelandic
Studies: Japanese

 
 Message 73 of 76
18 January 2013 at 8:35am | IP Logged 
mrwarper wrote:

I don't agree. Software translation is notoriously bad because it's too literal, but it has become ubiquitous as well, so it is increasingly easier to assemble one on the spot (assuming it is necessary at all) from the original text and online translators / dictionaries. That's the magic of software, and you can't get any more literal than a bunch of single word look-ups strung together.


I suppose so, but when you don't have enough knowledge of the language (or even when you do) and you get:
»No, but I'm up there the glove on your return trip ticket. You have had to go home early, and had to knock the horse, which was a fishing trolley, and you Okuda on bad roads, to reach the station «.
She jumped and stared at my companion.
»It's no mystery, dear Miss« said Holmes smiling. »On the left arm, you have no fewer than seven forarslettur, and they are quite new. No vehicles splashing such person, unless hunting chariots, and they do it only when sitting on the left trolley loader «.

instead of "inside your left glove i see your return ticket... you had to take a dogcart on bad roads, you only get splashed when you sit to the left of the driver...". this was a minor, quick example but it can get a lot worse, especially in the case where it actually all SEEMS understandable in context and is in fact really wrong. i think being able to judge the quality of the translation can sometimes really depend on how good your language skills actually are, and so i wouldn't suggest it to a beginner. then again, someone who is good, doesn't need a more literal translation (although it is still useful). And the pull-down options for alternate translations don't always make any more sense, and it only gets worse when you start having stuff like minor prepositions that when paired with certain words make entirely different meanings.

I can't remember exactly what the sentence was, but I will always remember when I saw "She put the clothes into the washing machine" translated into English as something like "She cleaned up the bodies", and just the other day for some reason "Icelandic" (in Icelandic) was always being translated as "English" when I was messing around... I think machine translation can be just fine for really common languages (which are probably much better at translating), if you are not worried about mistakes like that or are good enough already to catch them.

Edited by kaptengröt on 18 January 2013 at 11:07am

1 person has voted this message useful





Iversen
Super Polyglot
Moderator
Denmark
berejst.dk
Joined 6638 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 74 of 76
18 January 2013 at 10:12am | IP Logged 
Google translate is notorious for changing common names, languages and things like that to other names etc. which play the same role ind the target language. For instance it recently changed Kuala Lumpur into Djakarta when I translated a text from Bahasa Malaysia into Bahasa Indonesia - but curiously enough not when I translated the same passage into English. It is strange that a company with Google's immense resources can't find a solution to this problem, possibly involving checklists of proper names and other things that either should be left alone or changed to the exact equivalent in the other languages.

Apart from that my formula for using Google Translate is that it is better than nothing if you simply don't understand anything in a foreign language - but only by a narrow margin. For instance my only chance of understanding Japanese or Korean texts is to let Google have a go, with all the risks which that implies.

I do however use Google to make bulk translations from a weak language into something I know well, like Danish or English, and then I use these translations when I make bilingual texts. The error rate is probably the same in both directions, but my chances of spotting the most grotesque errors are much higher in a language I know well. Furthermore I don't trust the translation blindly, but use it as a way to help myself to understand the foreign text - and even here there are many cases where it is obvious that Google's version can't be correct, and then I have to use my dictionary to find out what all the words in a passage mean in the hope that some kind of plausible interpretation then emerges. But by and large I can live with a fairly high number of translation errors, as long as I get just enough help to form my own opinion about difficult passages.

Besides I only make bilingual texts for languages which I study, which means that I have at least a minimal knowledge of their grammar and terminology - enough to spot for instance whether there is a negation in a sentence in the original text or not. And that's important because another wellknown error in Google Translate is its tendency to drop negations.

Concerning language combinations: The worst one by far seems to be Latin vs. Something-else, which generally produces rubbish in both directions. I am not surprised that Google has problems precisely with Latin, if they mainly base their analysis on Classical and Medieval texts and their translations into English. On top of that Latin has a fairly free word order, which means that you need a combination of grammatical analysis and simple common sense to make order in some sentences. And machine translation programs have severe problems in both these areas.

But Latin is also a problem in another way: with one exception all my Something --> Latin dictionaries are nearly useless when it comes to translating modern content into Latin. It seems that their authors simply took the words used to translate Latin into Something and turned the bucket upside down. But you can' do that - a Something -> Something-else directionary should primarily be concerned with things people of language X might want to express in that language because they almost certainly will also be the ones they want to express in language Y. And in the case of Latin this is not only a question of electronics and nuclear physics, but also of basic things in a modern society like 'teachers' or 'mass media'. I haven't tried to learn Ancient Greek or Hittite or Akkadian, but you might expect similar problems with dictionaries for these languages.


Edited by Iversen on 18 January 2013 at 10:30am

1 person has voted this message useful



kaptengröt
Tetraglot
Groupie
Sweden
Joined 4273 days ago

92 posts - 163 votes 
Speaks: English*, Swedish, Faroese, Icelandic
Studies: Japanese

 
 Message 75 of 76
18 January 2013 at 11:15am | IP Logged 
Iversen wrote:

Besides I only make bilingual texts for languages which I study, which means that I have at least a minimal knowledge of their grammar and terminology - enough to spot for instance whether there is a negation in a sentence in the original text or not. And that's important because another wellknown error in Google Translate is its tendency to drop negations.


I was actually wondering about something like this - with Google Translate you can move around the word order in the translation, but it is already connected to target-language words. If it thinks "cat and dog" are put together in a phrase or something, I can't split them. That's fine normally but when "cat" actually really needs to be separated and moved to another part of the sentence in the translation (since sometimes it really messes up in this recognition thing)... I was just wondering if there actually is a way to do that with Google Translate and I didn't notice. That doesn't have to do with "making parallel texts" though, but "when I want to spend time improving Google's translations for no reason".

I have, though, seen beginning learners use Google Translate all the time - I think it's alright with trying to get some kind of meaning from a language you don't know, but the problem is that I have seen a lot of learners use it to get meaning INTO the language they are learning, or to plain use it instead of a dictionary. For some reason I keep meeting Swedish learners who eventually reveal "no I have been looking up all your words in Google Translate!" when talking with me, err, why didn't you just use an actual online dictionary, like I do? Also I make a remark about how good/wrong their sentence was, and get "I just used Google Translate". I don't really get it, Google is usually my last resort for translations, or is for when I am too lazy to actually read something myself and want a quick translation to English so I can see what the topic of the message actually is.

So I guess, I really don't want to advocate Google Translate just because I have already seen so many people using it when they really shouldn't, and I don't want to make that number grow...

Edited by kaptengröt on 18 January 2013 at 11:22am

2 persons have voted this message useful



kaptengröt
Tetraglot
Groupie
Sweden
Joined 4273 days ago

92 posts - 163 votes 
Speaks: English*, Swedish, Faroese, Icelandic
Studies: Japanese

 
 Message 76 of 76
27 May 2013 at 7:35pm | IP Logged 
I magically stopped getting Email alerts for this website, so I forgot to post in the thread...

I put my more massive parallel text projects on hold, and started doing smaller ones here:
http://read-listen-study.tumblr.com/

Right now there is only one in Icelandic, and one in Japanese (lacking audio - and anyway that one is Kokoro, which has the first part of its audio findable online). But I plan to do more whenever I feel like it, and perhaps rope some friends into recording for Swedish and Russian and Japanese.

People are also welcome to submit or request there, but I ask that no one post copyrighted audio or entire copyrighted books (since people who like to sue people seem to easily find Tumblr posts). The Icelandic chapter was put online by the publisher, and Kokoro is copyright-free, and the Moomins translation was also put up by the translator of the official version, so those were okay to post.

Edited by kaptengröt on 27 May 2013 at 7:37pm



2 persons have voted this message useful



This discussion contains 76 messages over 10 pages: << Prev 1 2 3 4 5 6 7 8 9 10 

If you wish to post a reply to this topic you must first login. If you are not already registered you must first register


Post ReplyPost New Topic Printable version Printable version

You cannot post new topics in this forum - You cannot reply to topics in this forum - You cannot delete your posts in this forum
You cannot edit your posts in this forum - You cannot create polls in this forum - You cannot vote in polls in this forum


This page was generated in 0.3281 seconds.


DHTML Menu By Milonic JavaScript
Copyright 2024 FX Micheloud - All rights reserved
No part of this website may be copied by any means without my written authorization.