Microsoft Bing: Small Tsu = Porn?
I am on Facebook a lot now, and try to read some Japanese-language posts, especially ones written by my wife. They have the “See Translation” option, provided by Microsoft’s Bing, which supposedly translates the Japanese text into English.
The problem is, it often translates like it was written by a deranged screenwriter specializing in bad porn. Seriously, it’s like one of those Chinese dictionaries that resulted in obscene English labels in Chinese supermarkets.
Now I know that machine translation between European and East Asian languages is spotty at best, but one would think that certain words would simply not be in the translation matrix, or whatever it’s called.
However, it seems to be mostly related to a single Japanese character.
Take this sentence in Japanese:
A fair translation would be, “It was a wonderful performance in the men’s figure skating. Wow! Great!”
On Facebook, it was translated as:
Figure men’s amazing performance was … wow.! Nice boobs!
I tried going to Bing translation directly, pasted the sentence there—and it was even worse:
It was the figure men’s great acting. Amazing boobs! Nice boobs!
Turns out that the “boobs” comes whenever a small “tsu” (っ) appears out of place, used often in Japanese to create a sudden stop, acting kind of like an emphasis for the exclamation point. On Google translate, it comes out as “tsu” or (strangely) “LI.”
Here is a Bing translation of a single Facebook post:
In less than two hours March! (early!) fliped over my private calendar is out! (did buy a desk calendar “Hoshino Chan” thanks for accepted calendar for a super House by mistake, I have is and & my husband face big boobs a March to forgive.) was indeed warm day, so it was just a happy. Mood shop & cold hardens me you cum Sasha! I’ll do it! (what? for) of switch “chubby!”, I feel that it was. (Lol) weekend winter mode is, but another relapse is also no sense.
Seriously? “Big boobs,” and “cold hardens me you cum”? I’ve been getting questions from my family as to what exactly Sachi is writing in Japanese.
When I put the text into Bing’s official page, again it identified the small “tsu” (っ) as the part translated as “boobs”—but it also translated the exact same character into “cum” in another sentence! What the…?
This is what you get if the exact same message in Japanese is put into Google Translate:
March in less than 2 hours to go! (Ll soon) Tsu was turning a private calendar of my home!
(And me accepted me to buy a desk calendar calendar super home for you’ve had “Hoshi-chan” by mistake and … face Deka-tsuna March excuse of & husband thanks) Today is because it was a warm day, I was happy with it. For me, that hardens and cold moody & was the day that was popular, the feeling that “Pochi” was the switch “Yaruzo pretensions” of (what for) … (laughs) The weekend seems to winter mode, but there is no sense going back.
As you can see, the translation as a whole is better on Google. It’s still mangled, but much more clear, and no porn terminology.
The thing is, it’s not just the two strange “tsu” related hiccups I found—strange words find their way into the text fairly commonly. Here’s a collection of sentences that I have strung together from various sources, to give you an idea of what I’m talking about:
Gaping! I always do you have weed… Hand fetish with me! Will drink your father who ate and I’m sure. Walk to him cum ♪ suffice in the exercise of said. I was the time you pack. Also to go out, when combined in the dog ultra-most fortunately I’m a boobs. Requests off my husband cum!
So, was Microsoft’s software intentionally sabotaged, and after months or years nobody at Microsoft noticed? What the hell is going on there?