Until very recently machine translation (MT), or “computer translation” as it is sometimes referred to, was a dirty word in the translation industry synonymous with atrocious quality. Indeed, it used to be that clients threatened to pull an account, or worse, if machine translation was used on their projects. And, in turn, predictably, translation service companies threatened divine retribution against any translator tempted to forego the hard work of professional translation for the short-cut afforded by machine translation.
How times change!
This article proudly declares in the future “Human and Machine Translation Cloud Helps Overcome Language Boundaries”1. Folks, that future is at hand -- Consumers of professional translation services should be left in no doubt that machine translation is already an integral part of the translation cloud!
I know, it's a storm cloud
Origins of the translation cloud
Let’s backup here a moment and take a look at translation clouds. The cloud is a recent development in the translation industry, which:
- Is a technique enabled by advances in cloud computing technology that allows project participants to simultaneously collaborate through shared resources, the purpose of which is to enhance efficiency and communication while eliminating costly managerial tasks.
Quality translation is NOT, nor ever was, a stated objective of translation clouds.
- Was born out of the business necessity to compete with cut-throat competitors originating in China and India etc., who employed a simple translation model:
Cheap Translation: Machine Translation + Post-editing by Novice Translators
Professional translators shunned this formula due to the sweat-shop rates of US$0.01~0.03 per word; as a result, these Asian companies were forced to rely on novice translators, in other words unqualified and inexperienced bilinguals. (In fact, many translation directories where these jobs are advertised banned the posting of jobs below a certain rate, commonly set at US$0.05 per word.)
Separating the wheat from the chaff: Clarity in the Cloud
A few translation cloud vendors commendably advertise their products and services as what you see is what you get.
However, many more couch use of machine translation in vague terms and slick marketing jargon that confuses customers, while essentially adopting the same translation model as their Asian competitors. Here’s a quintessential example, where the article’s author acknowledges the limitations of computer translation (machine translation & computer assisted translation)2, then proceeds to artfully champion it’s role in translation clouds:
“Think of computer translation as a child learning to write.”
By drawing on the analogy of a child learning, the author implies computer translation, be it machine translation or computer assisted translation, is capable of intelligent translation through learning -- Let’s be crystal clear here, IT IS NOT. Computer translation is software that can only do what it’s programmed to do. Fact is, science doesn’t even understand yet how children learn languages!
(Further, the most optimistic estimates by artificial intelligence experts put the kind of intelligence capable of human quality translation in computers at least a generation away.)
Translation cloud chokes on Baskin-Robbins
The article goes on to claim an accuracy rate of 65% for computer translation, with only 35% of the translation requiring “the help of professional and novice translators”.
This claim is beyong ridiculous! Here’s the essence of the problem:
- Simply matching up text segments from previous translations stored in database, which is how computer translation works, without regard for context knocks that accuracy rate right down to 30%, if not lower.
Let me introduce a real world example to illustrate this point. The other day my pre-school daughter barfed on a Baskin-Robbins Mint Chocolate Chip cone, declaring it was so sweet “it made me sick”. Now, this is a phrase she bandies around whenever she experiences discomfort, like the time she related how a bunch of kids trashed her favourite playground. Explaining her feelings she fumed “It made me sick!” I’m sure you can see the phrase (text segment) is the same; however, the context is completely different thus producing the completely different meanings of physical discomfort and mental discomfort.
That's merely the kids stuff; however, it brings into sharp relief the fatal flaws of the popular implementation of the translation cloud.
And, we’re only just getting to the 35% part -- This is where the cloud really starts looking more and more like that simple translation model out of Asia. I just love the way the author saddles “professional and novice translators” in the same sentence: the acceptance through association sales technique, like in commercials! The real question, though, is why are novice translators being introduced into what translation cloud operators are billing as professional translation services? There’s actually a logical explanation for this: Crowdsourcing of novice translators.
Machine translation is still a dirty word
The Wall Street Journal noted way back in 2002 "I definitely would not use machine translation (MT) for business or anything remotely requiring accuracy." The same holds true today! That is, there haven’t been any breakthroughs, or even noteworthy advances, in computer translation technology that supports the quality claims of translation cloud companies. Further, there won’t be for at least another 30 years or so…
The efficiencies and costs benefits you think you’re reaping with the translation cloud are simply NOT there. Your actual costs will be lower, and you may be able to get your hands on that project a bit sooner; however, in terms of quality you can expect, the true translation cost is in the clouds -- Sky high! An influential industry survey shows 80% of international businesses lose money to poor quality translation…Machines translation is still a dirty word where it matters most -- Amongst the professional translators that ensure you don’t lose money to translation errors.
About the Author
Ivan Vandermerwe is the CEO of SAECULII YK, owner of Japan Translation Services Tokyo Visit SAECULII for the latest professional articles and news on Japanese Translation Services
Copyright (C) SAECULII YK. All rights reserved. Reproduction of this article is permitted with inclusion of the "About the Author" reference as is (including text links, japan-translators.saeculii.com/english/services/japanese-translation-services.cfm), and this copyright information.