Question
Asked 25 June 2013
Are machine translation and post-editing more efficient than ordinary translation? What about the quality?
Would anybody who has tried it please comment. Including which MT program has been used.
Most recent answer
Falcon Scientific Editing will be happy to edit your abstract for FREE so that you can determine what level of language editing service you require. Then, we can provide you with a quote for the remainder of your manuscript/paper. We also offer translation, illustration, and formatting services. Feel free to contact us at contact@falconediting.com.
All Answers (18)
South Westphalia University of Applied Sciences
Check Sharon O’Brien's (Dublin City University) work. She may not quite answer your question, but she's an expert.
Aarhus University
Thanks for your advice. O'Brien is certainly a capacity in this field. I will study her more intensely on this matter. Have you tried MT yourself or with students?
Soka University of America
It isn't exactly clear from your question what you are asking. If you are referring to some of the popular translation programs, such as Babylon, iTranslate, and Google, the results are mixed. I have used each of these three and found that performance depends on the target language. Google and iTranslate, for example, work fairly well with Spanish, but they produce gibberish when trying to translate Japanese and Greek. They do not work particularly well with Latin, but it is possible to sort out most of the errors--provided one knows a bit of Latin.
The issue of efficiency is problematic. I would argue that the most efficient method of translation is to get the help of someone who is fluent in both languages, but for many of us this option isn't feasible. MT using available software can be highly inefficient for languages like Latin, Greek, and Japanese because of the difficulties associated with the numerous errors produced.
Aarhus University
Thanks for your replies. They are useful. One of the things I have noticed is that students of translation as well as industry professionals use MT and post-editing. I'm curious about how that will change our work habits, so I'm glad that you have responded with those aspects included.
I have just read an exam paper where MT had definitely been used. Challenging in terms of tracking how students state sources, for instance.
I'm eager to hear more comments on this issue.
Thanks,
Birthe
University of the Witwatersrand
As an industry professional, I use Google Translate to translate documents of potential interest. It can translate faster than I can type, so I "only" have to fix up its mistakes.
If I have read a lot of the subject area, I need not have any knowledge of the source language, although this knowledge is needed to decode bits of inscrutable, bad machine translations.
I have read between your lines, and I share your concerns.
Aarhus University
Dear Ian,
I'm glad that you shared your work practices here. I know that many people use Google Translate, but they find it hard to 'admit'. Google Translate is here to stay, and although we as translation and language professionals hate the flaws of the program, it does help a lot of people gain information in their native language. Although flawed, the resulting text is detectable for most people, and if not, well then ..bad luck. Then ask somebody. In the midst of the criticism against Google Translate, I find that the -almost- global access to information for everybody is invaluable and part of a very important global democratic access and information sharing.
I'm planning to do some research on the value of MT (machine translation), so I would still welcome comments on the issue.
Thanks for your input.
Best regards,
Birhte
University of the Witwatersrand
You could start your research by reviewing the results of submitting "importance of machine translation" to Google.
Certainly, "statistical machine translation" is a new field for research, and you could do no harm by submitting it to Scholar, and reviewing the results.
Try also "statistiske maskinoversættelse" if GT is to be trusted, and "statistische maschinelle übersetzung uni düsseldorf" if Google Instant (German) is to be trusted.
Let me know if you need a review template.
Aarhus University
Hi Ian
Thanks so much for your reply. Sounds interesting, and I will certainly try some of your suggestions.
Have you got some kind of review template?
Best regards,
Birthe Mousten
Aarhus University
Thanks so much. I just saw it in my email. I will certainly take a look at it and try some of it.
1 Recommendation
University of the Witwatersrand
You also asked about quality. If one does an initial cheap (free) Machine Translation, then you have to spend less time typing and checking details such as spelling. So you can concentrate on the higher-level issues such as phraseology. The result should be a better quality translation.
Government Polytechnic college Nanded
Do you mean that Machine translation system with post editing? There are some translation tools which requires post editing (the tool will produce more than one translation from which user has to select correct according to context). The tools which use post editing facility will always provide better quality than normal translation.
University of the Witwatersrand
Machine translations that provide more than one possible translation, ordered according to probability, and ready for selection, are even more useful than those that pretend that one translation is right!
Aarhus University
Dear Goraksh and Ian,
Thanks for your comments.
Goraksh, what I am particularly interested in is the interface between MT and people's work processes in post-editing. That is where I think a lot of research can take place for the future - near future, that is! Professionals already use MT tools, both professional systems and freeware, yet almost the entire academic world shuns the field.
Ian, I am curious about the MT systems you mention that would suggest more than one translation. Which ones do you think of? I must admit that I am quite ignorant when it comes to knowing such systems.
University of the Witwatersrand
Google translate does the following:
Have you got some kind of review template? →
Har du fået en slags gennemgang skabelon?
If you turn on instant translation, then the moment
you type “Have”, the Danish word “Har” appears.
In addition, the following info useful to the translator appears:
have – verb
Frequent: have[har?] _ have, possess
Sometimes: vaere _ be, have, make
Seldom: eje _ own, possess, hold, have
Which give me some confidence that Har might indeed be a good choice.
So the translator's review template you are really looking for contains the useful back-translations and their probabilities!
Frequent: Danish _ English meanings list
Sometimes: Danish _ English meanings list
Seldom: Danish _ English meanings list
Aarhus University
Hi Ian
Thanks for your answer - I see what you mean now. I thought you knew systems that could produce different text versions, maybe based on complexity or something else. That would really be a novelty.
I do find the Google queries very useful - they can give the kinds of interactions that are useful.
Thanks again.
1 Recommendation
Falcon Scientific Editing will be happy to edit your abstract for FREE so that you can determine what level of language editing service you require. Then, we can provide you with a quote for the remainder of your manuscript/paper. We also offer translation, illustration, and formatting services. Feel free to contact us at contact@falconediting.com.
Similar questions and discussions
Recommendations
The standard distinction between poetic and referential language, the gulf between science and the humanities, and the distress many teachers of English feel when faced for the first time with the prospect of teaching technical writing are discussed. In the introduction of many technical writing textbooks. Technical communication is divorced from o...
The diversity of languages is one of the main barriers to international trade and a cause of relative isolation for parts of the population. Retaining and supporting this diversity is a determining factor for social and political cohesion within the European Union. European support in the field of language engineering will help to realise and explo...
Endangered languages may require more flexible language technologies than stable ones because they may not be standard- ized and they may be in a cycle of losing, replacing, and borrowing vocabulary and grammar. This paper argues that the cov- erage and content of language technologies should be in the hands of the speech com- munity, and that it n...