Thinking of relying on Google Translate for your translations? Think again!

There is a lot of buzz out there about Google Translate and other machine or software translations which are cheap and instant. As a translation company we routinely get clients who wonder why translation services cost what they do and take time when you can do everything instantaneously with computers these days.

Well… it’s not that simple. Machine translation can handle some very simple sentences just fine. When you start getting into idiomatic, marketing speak or highly technical, industry specific language, machine translation still falls very short.

Language is dynamic, complex and highly nuanced. No machine or software has been built that can mimic the human brain and how it produces language. On a daily basis individuals produce speech creatively and instantly using more knowledge of tone, style, context and meaning than any software can compute.

Moreover, no software has been created that can capture how a statement needs to be culturally adapted in order to make it relevant for your target audience. Maybe that financial product you’re selling doesn’t have a counterpart in Germany. Perhaps “the grass is always greener” has no value when taken word for word in Chinese. Not to mention the fact that some Asian languages are so different from Western languages, they don’t even have tenses! Google translate isn’t going to tell you that and most of the time it doesn’t handle the line in the right way.

To explain the point, let’s do a little test run. For a recent translation we had a quote from a customer:

It doesn’t make sense to go with anyone else.

The Chinese translation was:

看电影的伴侣非其莫属。

Let’s head over to Google Translate for a demo on how machine translation compares:

google_translation_1

OK… Hm. I’m thinking maybe there was something off in that line? Why don’t I click on the “Did you mean” link since I think that is giving me some sort of auto-correct suggestion.

google_translation_2

Well, that’s clearly not right. At least it got the language right. I once dropped in Japanese and it detected Chinese since there were a couple Kanji characters.

So maybe we have one too many characters here and the period has been misplaced. I wonder how it would change if I deleted one character at a time from the end.

I just want to see something that resembles “doesn’t make sense.” That’s the key phrase here.

google_translation_3

Deleting one character resulted in what? Let’s delete another.

google_translation_4

Wow. One character and two words disappeared? And why is Google Translate giving me a hyphen? What’s that all about?

google_translation_5

Hm. Nothing changed this time.

google_translation_6

I deleted one character and now it went from “Cinema partner non-” to “Couples watching movies!”

google_translation_7

Hmmm…

google_translation_8

To watch vs. to look. OK, those are similar. Fair enough.

google_translation_9

We’re still in sort of the same ball park here even though I deleted three characters. I wonder how “ball park” would translate in Chinese. Of course assuming Google Translate would pick up what I mean.

google_translation_10

Wow. Deleting one character now changed it to “See power!”

google_translation_11

This last image really is the best. It captures the exact reason why machine translation falls short, even today, with all our technological advances. See how one Chinese character has numerous possibilities for how it can be translated and how many different characters could be used for each of those terms.

This demonstration is not meant to say that Google Translate is wrong, but without a person to go through those terms and understand the context and the message you are trying to convey, how can you be sure that the right term will be chosen?

Additionally, when a sentence can change so much by deleting one character, do you really want to leave professional translations up to a machine? Asian languages can be particularly problematic because the structure, grammar and culture is vastly different and mistranslations, especially by a machine, are even more likely.