Google Translate Is No Friend to Science or Business, Despite its Improvement
The telltale signs are there: the same word written three different ways, an acronym half the time and spelled out in full the other half, and comically literal terms.
This paper’s got Google Translate written all over it.
But I confess…
Google Translate’s gotten much better
Occasionally its translations need little or no correction. This can happen with stock phrases, simple grammar, and in highly formulaic writing such as common scientific methods.
Where it continues to fail is just about everywhere else, especially for (1) the conversational tone that modern-day marketing copy needs, and (2) the precision and clarity that scholarly writing needs.
If you only use Google Translate to point you in the right direction, to give you the gist, it can do the trick probably three-quarters of the time.
So, good job with that, Google. But you’re a long way from replacing human translation, and a much longer way from replacing source language written from scratch.
Meanwhile, every other translation tool I know is poor for giving the general idea of most text, and is unspeakable for creating publishable text.
I should also mention that the Japanese–English pair is my basis for this. There may be better progress in other pairs with more similarly structured languages.
Three takeaways:
- Google Translation is good for getting an idea of what’s going on, and as a start when doing direct, literal translations.
- All other translation tools are good for little more than looking up words and short phrases.
- No output from a translation tool should ever be used for published material.
Machine Translation Makes You Look Really Bad
These issues don’t stop academics and businesses alike from using translation tools for purposes when natural and accurate English is needed. The results are on the spectrum from humorous to head-scratching to maddening, depending on if you’ve had your coffee yet.
That’s not the fault of the tools. They’re amazing and evolving. Someday, probably in my lifetime, they’ll replace most human translation. But it’s not going to be next year and probably not next decade.
So when auto-translate embarrasses authors, especially those who can afford better, it’s the authors’ own fault. They’re being irresponsible and cheap, as well as disrespectful to the reader.
Try Bic Camera’s English page. It’s all piped through auto-translation and the results are pitiful. This is a company that does huge sales to tourists and resident non-Japanese. Maybe the Chinese and Korean pages are better?
Amazon.co.jp, to its credit, has excellent English e-mail and telephone support, which is a real rarity in Japan. Its phonetic translation, however, is absurd. Some of it could make a Trump blush: this one is NSFW (set your Amazon to English if you can’t see it).
So then why do a very well-off consumer electronics chain and one of the world’s richest companies use customer-facing auto-translation that produces gibberish and obscenities?
Other than cutting corners and sheer arrogance, there’s no clear rationale.
Google Translate is an Emerging Plague in Academic Writing
I’m also seeing greater dependence on Google Translate in academic writing.
Academics are smart people (duh). They’re hip to the fact Google can spit out language that may not immediately look like auto-translation. Also, by dropping an abstract or even a full manuscript in Google Translate, and then sending it for proofreading, they may save a few dollars. It’s understandable, it could be ignorance, and just may coerce an editor into hammering some sense into it.
Authors: please stop that. Editors: please decline the work offer. It’s not worth it for either of you.
When I’ve got my editor hat on, the Google Translate challenge makes my job even trickier. You may think it would make it easier for me to edit but it’s the opposite.
I used to be able to say, “This was auto-translated. Sorry, but I/we only edit English that authors write themselves. Please try to write it yourself or ask a colleague for assistance.” That’s a policy that editing companies with high standards use, for their own sake (to reduce complaints, needless labor, and borderline ghostwriting) and for author’s sake (to better ensure authenticity and original intent in the writing).
But now, I’m sometimes tricked, or I get a page into a job or two into a job and realize I was duped.
So if it tricked me, it must be pretty good, right? Well, no.
For one, the labor of editing takes on a snowball effect as I have to weed out the viable text from the mistranslated text, and as I have to align terms throughout. The author really should then confirm the accuracy as well. For another, I now don’t even know if the paper says what the author wants it to say; because the author didn’t write it, Google did, and to an extent, so did I.
Now even if I go through and make some sense out of the hodgepodge of words, it’s not authentic writing and there is a greater chance of misinterpretation. In science, that’s dangerous, fatal even!
It’s also a big fat waste of my time and the client’s money.
I’ll dissect an example to unpack the strengths and weaknesses of Google Translate and a couple of other popular platforms, and compare them with my human translation and transcreation (translating + adding context and tone suited to the audience).
How to (Not) Mangle an Auto-translation
Here’s a passage from a Yahoo! News Japan article about global warming. It’s written for a general audience, but it’s scientific content. It has some basic scientific terms. The tone is accessible but not at a conversational level. Ideally, someone with around a high school-level education should have no problem with it. It throws in an English phrase as well, presumably because there isn’t Japanese to quite capture the essence of that term.
The original is:
報告書には、異常気象、海面上昇、生態系、健康、食料、水資源といった各側面において、1.5℃温暖化すれば今よりリスクが大きくなり、2℃温暖化すればさらに大きくなることが包括的に述べられている。しかし、それだけならば「そりゃそうでしょ」と思うだけであり、問題はそのリスクをどの時点で受け入れられなくなるか（How dangerous is too dangerous?）だろう。
Reference: 「地球温暖化対策 なぜ1.5℃未満を目指すのか －IPCC特別報告書を読む」, Yahoo! Japan News. Retrieved from https://news.yahoo.co.jp/byline/emoriseita/20181104-00102886/
Me
My reasonably trustworthy translation is:
The report states that outcomes such as abnormal weather, sea level rise, and adverse effects on the ecosystem, health, and food and water will collectively expand with a 1.5°C temperature rise. A 2.0°C rise would be even more damaging. Such outcomes may seem obvious to people, but the real problem is: how dangerous is too dangerous? How much will we tolerate?
This is OK, but it isn’t a great translation because I’m not a great translator and because I’ve read a lot into the author’s intent. I’ve had to infer, opening a wider gap for misinterpretation. Japanese has fewer words than English.
As seen in the list of adverse outcomes, the author never actually says “adverse outcomes”, and instead relies on the supposedly common understanding that effects of global warming are negative. The author also uses the word for “resources,” but I think it’s implied that food and water are resources in this context, so I chopped it. A more literal translation would be wordy and clumsy. The first sentence is challenging because of the fundamental Japanese grammar that sends the verb to the end of the sentence and because of the writer’s verbosity. The second is challenging because it slips into conversation for a short phrase, but mixes it with common formal grammar.
Google Translate
Google Trans gives me:
In the report, it is comprehensive that the risk becomes larger if it gets warmed at 1.5 ° C, and if it becomes warmer at 2 ° C, it will become even bigger in each aspect such as abnormal weather, sea level rise, ecology, health, food and water resources It is stated. However, it only seems to be “that ‘s so,” and the problem would be how dangerous is too dangerous? At which point the risk could not be accepted.
We can see Google has no trouble with the shopping list of issues related to global warming but struggles with the writer’s wordplay, such as the embedded self-reflection, which I paraphrased. It needs a lot of cleaning up to make it fit for print, but it’s enough to give you an idea of what the writer is on about.
👨🏫Gist Grade: B-
👨🏫Publishable Grade: D
Weblio
Weblio is a popular Japanese dictionary and translation tool. It gives me:
In abnormal weather, a rise in sea level, ecosystem, health, food, each side such as aquatic resources, a risk grows big from now if I make 1.5 degrees Celsius warming, and, in a report, it is spoken comprehensively that I grow bigger if I do 2 degrees Celsius warming. However, I only think “so, and the risk is not accepted at which point in time, or, only as for it, the problem will be it” (How dangerous is too dangerous?).
Weblio agrees with my choice of “ecosystem” over “ecology.” That’s more pertinent and related to global warming. Nice work. “Water resources” becomes “aquatic resources,” which is perfectly correct, but is a bit of an industry/academic term and less accessible to a lay reader. Weblio also slips into the first person so that the writer ends up becoming the very cause of global warming. The second sentence is a word salad.
👨🏫Gist Grade: D+
👨🏫Publishable Grade: F
Excite Japan
Yes, Excite still exists and it thrives in Japan. It’s like an ’80s metal band. Excite Japan’s still-popular translation tool gives me:
When warming globally 1.5 ℃ in abnormal weather, sea level rise, the ecosystem and each flank such as the health, the food and the water resources, risk will be a report big more than now, and when warming globally 2 ℃, it’s stated comprehensively to become bigger. But when it’s only that, you just think “If shaving, it’s so.”, and as of which wouldn’t a problem accept the risk any more?
Excite’s problem is it both goes literal and seems to have no ability to apply context. It also can’t reposition the grammar to an English subject-verb-object sequence and seems to translate word-by-word. The obscure “flank” is an accurate translation of the word, but an obscure one and incorrect for the context. It also dumbly transliterates the self-reflection, 「そりゃそうでしょ」. “Sorya sou desho”, if literally translated might be, “That’s the way it is, right?” but actually translates more like “Well, duh” or “No s**t Sherlock.” Excite translates the sorya as the verb for shaving. The end result is that Excite lets us know there’s some sort of issue here about global warming, and shaving.
👨🏫Gist Grade: D
👨🏫Publishable Grade: F
Advice for Editors
Follow a no-auto-translations policy with your clients unless you really need the money. Doing so will result in more natural English, it’s the ethical thing to do, and frankly, editing auto-translated English is insulting to a real editor.
We’re not machines; we’re skilled in natural English communication, and we advocate it. We want to produce really good work for our client authors, not give them something that just barely makes sense, or even worse, is inaccurate.
I learned Rule #1 in journalism school copy editing: Do no harm.
Low-grade budget proofreading companies and Fiverr freelancers, you should do the same. But I know you scrap for jobs and lofty standards for editing and ethics may be secondary. I get it.
Keep in mind you’re doing science a disservice when you proofread auto-translated text. Unless you have the ability to read the source language alongside the translation and you’re well-versed in the scientific topic, you may be misrepresenting valuable information, which can hurt your client’s reputation and, at its worse, result in publishing inaccurate results.
My fellow editors and proofreaders, I hope you can refrain from editing what is clearly auto-translated text.
And those who don’t speak English fluently, please do your best in your own English or ask a colleague for help. Use translation only as a guideline for your work, never use auto-translation for anything more than basic understanding, and find a qualified translator and editor so you get your intent across effectively.
The original version of this article is on the Scize blog at scize.com.
