Sunday, March 31, 2013

An Amusing Game of Miscommunication: Google Translate Telephone



As many of your have probably figured out by now, I just have way too much time on my hands. I mean, think about it: I've actually written a blog entry in which I turned the 1973 animated version of Charlotte's Web into a horror movie involving mass insanity, devil worship, and allusions to Lord of the Flies and Animal Farm. If you should want to read that, well, my sympathies, first of all, but for what it's worth, you can find it by clicking here. 

Anyway, when I'm not blogging about things such as that, or how much I loathe the film Mr. Holland's Opus (yes, you can read that one by clicking on this sentence), I try to think of new and amusing things to do with the digital tools at my disposal. 

One such tool is Google Translate. I find Google translate way cool, and I used it in my last entry to reach out when I discovered that a couple of folks from Russia checked out my blog.

Of course, as I did this I kept my language as simple as possible. I knew that as good as the translation algorithm was, there still would be a lot that wouldn't quite make it to the other side.

I really saw this when I took what I had translated to Russian, and then translated it back to English. Inevitably, the cracks began to show, particularly if I used idioms.

Here's an example. Let's start with this bit of "The Oldest Established," from Guys and Dolls:

There are well-heeled shooters everywhere, everywhere
There are well-heeled shooters everywhere.
And an awful lot of lettuce
For the fella who can get us there.
If we only had a lousy little grand
We could be a millionaire!

Now, let's translate it to Russian:

Есть богатых стрелков везде, везде
Есть богатых стрелков во всем мире.
И очень много салата
Для парня, который может доставить нас туда.
Если у нас был только маленький паршивый великий
Мы могли бы быть миллионером!

...and now let's translate it back and see what happens:

There is plenty of shooters everywhere, everywhere
There is plenty of shooters worldwide.
And a lot of salad
For a guy who can get us there.
If we only had a lousy little grand
We could be a millionaire!

Okay, pretty good. It's actually kind of endearing, the way it turned "lettuce" into "salad".

But this got me thinking (which, as my friends can tell you is usually a bad thing): what would happen if I took a set of English sentences, and then translated them into another language, and then translated that translation into another language?

And what if I did that a few times?

What would it look like once it had completed its journey back to English?

And so, with that thought, I created a new game that I call Google Translate Telephone.

Telephone, you may remember, is the party game in which someone starts with a phrase, and then whispers it  to someone else. This person then whispers it to the next person, and so on, until the last person says what he or she heard. By this time, of course, the phrase has totally changed.

Let's give it a whirl. We will start with English:
I think that these people are lepers. I will ask them if they could please use a hand sanitizer before using the salad bar. I would not like them to contaminate the bacon bits.

and translate it to Russian:
Я думаю, что эти люди прокаженных. Я буду просить их, если они могли бы, пожалуйста, используйте дезинфицирующее средство для рук, прежде чем использовать салат-бар. Я бы не хотел, чтобы они загрязняют кусочки бекона.

then we'll take that Russian translation, and translate it to Czech:
Myslím si, že tito lidé jsou malomocní. Zeptám se jich, jestli by se prosím pomocí ruční sanitizer před použitím salátový bar. Nechtěl bych, aby znečištění slaninu.

then to Greek:
Νομίζω ότι αυτοί οι άνθρωποι είναι λεπροί. Θα τους ρωτήσω αν θα παρακαλούμε χρήση απολυμαντικό χεριών πριν χρησιμοποιήσετε το salad bar. Δεν θα ήθελα μπέικον ρύπανση.

then to Haitian Creole
Mwen panse ke moun sa yo se lepre. Yo pral mande si ou ta tanpri itilize dezenfektan pou men anvan ou sèvi ak ba a sòs salad. Mwen pa ta renmen polisyon bekonn.

To Japanese:
私はこれらの人々がハンセン病患者であると思います。あなたはサラダバーを使用する前に、手指消毒剤を使用してくださいならば、彼らは尋ねます。私はベーコン汚染を好きではないだろう。

to Arabic:
أعتقد أن هذا هو والأبرص هؤلاء الناس. قبل أن تتمكن من استخدام شريط صلاد، إذا كنت يرجى استخدام المطهر، وطلبوا منك. ولا أود التلوث لحم الخنزير المقدد.

to Korean:
이이 사람들의 나환자이라고 생각합니다. 당신이 샐러드 바를 사용하기 전에 세제를 사용하십시오, 그리고 경우에 그들은 당신을 부탁드립니다. 나는 오염 베이컨을 기원하지 않습니다.

to Hebrew:
חולי צרעת, האנשים האלה חושבים. השתמש בחומר ניקוי לפני שתשתמש בבר הסלטים, ואם הם שואלים. בייקון זיהום אני לא מתפלל.

to Esperanto:
Lepro pacientoj, ĉi tiuj homoj pensas. Uzu detergento antaŭ uzi la salato trinkejo, kaj se ili demandas. Lardo polucio mi ne preĝas.

to Traditional Chinese:
麻風患者,這些人認為。使用清潔劑之前使用的沙拉吧,如果你問。黃油污染,我不求。

back to English:
Leprosy patients, these people think. Use detergent before use salad bar, if you ask. Butter pollution, I do not seek.

What's awesome about this, of course, is that it has turned it into something that you could imagine Yoda saying while deep in the indignities of Alzheimer's.

But let's really give this game a run for it's money. Let's take a paragraph replete with idioms, and see how it comes out at the other end of a Google Translate torture session.

Hmmmm...how about this:

English:
I want to give Google Translate a run for its money, so I'm going to feed a lot of English idioms into it this time around. I think that it will strike people as funny just how much stuff gets lost in translation. In fact, not to put too fine a point on it or anything, I think that what's going to emerge at the other end of this is a nearly incomprehensible mishmash of linguistic fragments, utterly devoid of any coherence whatsoever. I really think the prospect of translating this into anything understandable is going to hit a wall pretty quickly. 

and let's really have fun with this by feeding it through nothing but languages far different from English.

Our first stop is Iceland. This, by the way, is a language in which its inhabitants are fiercely protective of it, which has resulted in the language being much the same as it was hundreds of years ago (there is actually a language board that approves of additions to official Icelandic; they had a tough time when personal computers came to the forefront, as it was necessary to have words for such things as "operating system").

This is pretty cool, really: if you're a kid in Iceland reading their epic legends, the stuff reads as if someone wrote it yesterday. At the same time, I'm sure that a number of the things I wrote are going to hit a wall when Google frantically tries to do its work:

Mig langar að gefa Google Þýða að hlaupa fyrir peningana sína, þannig að ég ætla að gefa mikið af ensku idioms í það að þessu sinni í kring. Ég held að það muni slá fólk sem fyndið hversu mikið efni týnist í þýðingunni. Í staðreynd, ekki að setja of fínn a benda á það eða eitthvað, ég held að það er að fara að koma fram á hinum enda þetta er nánast óskiljanlegt mishmash af tungumála brot algerlega gjörsneyddur hvaða samhengi alls. Ég held í raun möguleika á að þýða þetta í neitt skiljanlegt er að fara til högg a veggur ansi fljótt.

Now, from Icelandic, hmmmm....how about translating that Nordic speak into Swahili. Yeah. Swahili.

Mimi nataka kutoa Google Tafsiri kukimbia kwa pesa zao, hivyo mimi nina kwenda kutoa mengi ya Nahau Kiingereza katika hilo wakati huu. Nadhani itakuwa mgomo watu kama funny stuff kiasi gani anapata waliopotea katika tafsiri. Kwa kweli, si kwa kuweka pia faini uhakika juu yake au kitu chochote, nadhani ni kitakuja mwisho mwingine ni karibu incomprehensible mishmash ya lugha kuvunja kabisa bila ya yoyote muktadha wowote. Nadhani, kwa kweli, uwezo wa kutafsiri hii ndani ya kitu chochote inaeleweka inaenda kuitwanga ukuta pretty haraka.

And from there, let's take a trip to...hmmm...Azerbaijan. Yes. I like it:

Google onların pul üçün run, mən hazırda deyim English bir çox vermək gedirəm Tərcümə vermək istəyirəm. Mən bu stuff tərcümə itirildikdə necə çox məzəli insanlar tətil olacaq. Əslində, bu və ya bir şey çox gözəl nöqtə qoymaq üçün deyil, mən bu dildə demək olar ki, anlaşılmaz dolaşıq tamamilə heç bir kontekstində olmadan qırdı olan digər sonuna gedir edirəm. Mən əslində, qabiliyyəti olduqca tez divar edib gedən aydın bir şey bu tərcümə, düşünürəm.

From Azerbaijani, let's now take these words on a trip to....India. Ladies and Gentlemen, I give you the Hindi translation:

गूगल अपने पैसे के लिए चला जाएगा, मैं वाक्यांश मैं अंग्रेजी का एक बहुत कुछ करने के लिए जा रहा हूँ का अनुवाद करना चाहते हैं. मजेदार कितना सामान अनुवाद में खो जाता है, तो मैं लोगों को हड़ताल करेंगे. वास्तव में, यह एक बहुत अच्छी बात पर या कुछ डाल दिया है, लेकिन मैं कह सकता हूँ कि भाषा अस्पष्ट, उलझन में है और किसी भी संदर्भ के बिना पूरी तरह से तोड़ दिया है, मैं दूसरे छोर पर जाने के लिए करना चाहते हैं. मैं वास्तव में करने के लिए दीवार बहुत जल्दी से स्पष्ट कुछ अनुवाद करने की क्षमता है करने के लिए जा रहा हूँ, मुझे लगता है.

And...well, Vickey's dad is from Portugal, so:

Google terá a duração de seu dinheiro, eu tenho um monte de frases em inglês que eu vou deseja traduzir. Material quanto engraçado se perde na tradução, por isso vou atacar as pessoas. Na verdade, é uma coisa muito boa para colocar ou algo assim, mas eu posso dizer que a linguagem vaga, confusa e quebrou completamente sem qualquer contexto, eu gostaria de ir para o outro lado . Eu realmente gosto da parede muito rapidamente some'm claro vai ter a capacidade de traduzir, eu acho.

And from there, let's visit the Promised Land:

גוגל תפעל לכסף שלהם, יש לי הרבה משפטים באנגלית שאני רוצה לתרגם. חומר כמצחיק הולך לאיבוד בתרגום, אז אני אתקוף אנשים. למעשה, זה דבר טוב מאוד למקום או משהו כזה, אבל אני יכול לומר ששפה מעורפלת, מבולבלת ושברתה לחלוטין ללא כל קשר, אני רוצה ללכת לצד השני. אני ממש אוהב את הקיר במהירות רבה כמובן some'm יהיה היכולת לתרגם, אני מניח.

I never noticed how cool Hebrew looks in italics. Anyway, it's a small world. Next stop, Japan:

Googleは彼らのお金のために働くだろう、私が翻訳したい英語の多くを持っている。ユーモラスな材料は翻訳で失われているので、私は人々を攻撃します。実際には、それは非常に良い場所か何かですが、私は、その言語が曖昧と言うことができ、混乱し、完全にマーサにもかかわらず、私は反対側に行きたい。私は本当に、非常に迅速に壁が好きもちろんsome'm翻訳することができます、私は推測する。

(I'd like to point out how "Google" has managed to survive this entire trip intact. I also have no idea how "some'm" got in there, nor do I know why it struggles to remain untranslated.)

Anyway, let's keep moving. I give you Urdu, the national language of Pakistan:

گوگل ان کے پیسے کے لیے کام کریں گے، انگریزی کی ایک بہت ہے کہ میں ترجمہ کرنا چاہتے ہیں ہے. ونودی مواد کی وجہ سے یہ ترجمہ میں کھو گیا ہے، میں لوگوں پر حملہ کرے گا. سچ تو یہ ہے کہ اگرچہ یہ ایک بہت اچھی جگہ ہے یا کچھ اور ہے تو آپ کا کہنا ہے کہ اس زبان غیر واضح، مبہم ہے، مرتا کے باوجود مکمل طور پر، لیکن میں نے دوسری طرف جانے کے لئے کرنا چاہتے ہیں کر سکتے ہیں. تم سچ میں ترجمہ کر سکتے ہیں کورس کے some'm دیوار بہت تیزی سے محبت کرتا ہوں، مجھے لگتا ہے.

...and while we're at it, let's hit the Arabian Peninsula:

جوجل ستعمل لأموالهم، والكثير من اللغة الإنجليزية التي يتم ترجمتها إلى يريد. انه مضحك لأن يتم فقدان المحتوى في الترجمة، وسوف تهاجم الناس.سچ تو یہ ہے کہ اگرچہ یہ ایک بہت اچھی جگہ ہے یا کچھ اور ہے تو آپ کا کہنا ہے کہ اس زبان غیر واضح، مبہم ہے، مرتا کے باوجود مکمل طور پر، لیکن میں نے دوسری طرف جانے کے لئے کرنا چاہتے ہیں کر سکتے و. يمكنك حقا ترجمة سريعة جدا some'm بالطبع أحب الجدار، على ما أعتقد.

...and we're clearly starting to hit a wall here, because when we try to translate it from Arabic, some characters just remain untranslated, as if Google Translate is just saying "no more, please...no more." But there will be more. Oh, yes, there will be more.

Let's go to Thailand:

Google จะทำงานเพื่อเงินของพวกเขาและจำนวนมากของอังกฤษซึ่งเป็นที่ต้องการแปลเป็น มันตลกเพราะพวกเขาจะสูญเสียเนื้อหาในการแปลและจะโจมตีคน. S چ Tu یہہےکہگ r چہیہیک B ہทีچھی J گہہے O کچھ ur ہے Tu پککہ na ہےکہ S Zaban ไม่ชัดเจน Mb ہ ม. ہےมาร์ธาکے Baugod M ک มล. พัฒนาپ r, เพื่อیک n M یںยังไม่มีข้อความที่ے Dousari พรรค ม.ค. ےکےลี่ےک RNA چہทีےہیںک R S T کے จริงๆคุณสามารถแปลได้อย่างรวดเร็วมาก some'm แน่นอนฉันรักผนังผมคิดว่า

to the Philippines:

Tatakbo ang Google para sa kanilang pera at ng maraming ng Ingles, na isinalin sa. Nakakatawa dahil sila ay mawala ang mga nilalaman na i-interpret at pag-atake ng mga tao. S چ Tu یہہےکہگ r چہیہیک B ہ t چھی J گہہے O کچھ ur ہے Tu پککہ na ہےکہ S Zaban malinaw Mb ہ m ہے Martha کے Baugod M ک-unlad ML پ r, sa یک n M یں N. ng partido ے Dousari Jan ےکے Li ےک RNA چہ ےہیںک RST کے ako ay napakabilis, maaari mo ba talagang some'm kurso na gustung-gusto ko ang wall, tingin ko.

(and still, "Google" survives.)

to Malaysia:

Google bertanding untuk wang mereka dan banyak bahasa Inggeris, diterjemahkan. Lucu kerana mereka akan kehilangan kandungan untuk mentafsir dan menyerang orang. S چ Tu یہہےکہگ چہیہیک B r ہ t چھی J گہہے O ur ہے Tu کچھ پککہ untuk ہےکہ S Zaban jelas Mb ہ m ہے Martha کے Baugod M ک pembangunan ML پ r, n M یں یک N. ے Dousari Jan ےکے Li parti ےک RNA چہ ےہیںک RST I کے pantas, anda benar-benar boleh some'm tentu saya memuja dinding, saya fikir.

to China (traditional Chinese, mind you...none of this simplified stuff):

谷歌爭奪他們的金錢和大量的英語翻譯。有趣,因為他們將失去內容進行解釋和攻擊。小號چ塗یہہےکہگچہیہیکB Rہ噸چھیĴگہہےØ烏爾ہے塗کچھپککہ到ہےکہ小號咋辦明確MBہ米瑪莎کےہےBaugod中號ک發展MLپR,N MیںیکN.一月Dousariےکے李ےےک黨的RNAچہےہیںکRST我کے快,你才能真正some'm當然,我很喜歡在牆上,我想。

And let's hit hit Iran, and make it Persian. Just because:

گوگل برای رقابت برای پول خود را و بسیاری از ترجمه انگلیسی است. جالب توجه است، چرا که آنها را از محتویات را به تفسیر و حمله را از دست بدهند. صور چ پوشش داده شده یہہےکہگچہیہیک B R ہ تن چھی J گہہے Ø تو ہے پوشش داده شده کچھپککہ به ہےکہ در بوق و کرنا قرار به روشن MB ہ سکته قلبی ماشا کےہے Baugod متوسط ​​ک توسعه ML پ R، N M یںیک N. ژانویه Dousari ےکے لی ےےک حزب RNA چہےہیںک RST I کے سریع، شما واقعا می تواند برخی از 'متر البته، دوست داشتم به دیوار، من فکر کردم.

(at this point, you'll notice Google translate is just sort of throwing up random letters, as if it's having a seizure or something)

On to Norway. We've been sadly neglecting Scandinavia:

Google å konkurrere for pengene, og mange av de engelske oversettelsene. Interessant, fordi de mister innholdet til å tolke og angripe. Hva slags ہ BR Y ہہے K ہ GCH ہ Y ہ Yk dekket Ch ھ Y J G ہہے Ø ہے du dekket ہے K ہ KCH ھ PKK ہ å sette trompeten til MB ہ Mi K ےہے Masha K D gjennomsnittlig Baugod ML c R, NM Y ں Yk N. Januar Dousari ے K ے Lee ےے K partiet RST I K ے RNA Ch ہےہ Y ں K rask, kan du virkelig ha noen 'm selvfølgelig, jeg elsket veggen, tenkte jeg.

...did you know that Google Translate also has the ability to translate into Basque? Neither did I:

Google dirua lortzeko lehiatuko dira, eta English itzulpen asko. Interesgarria, bere edukiak interpretatzeko eta eraso galtzen dutelako. Zer ہ BR Y ہہے K ہ GCH ہ ہ Y yk estalitako Ch ھ YJG ہہے Ø ہے ہے K ہ ہ KCH ھ PKK estali tronpeta jarri MB ہ Mi K ےہے Masha KD batez bestekoa Baugod ML c R, NM Y ں yk N. urtarrilaren Dousari ے F ے Lee ےے K party RST RNA IK ے Ch ہےہ Y ں K azkar, benetan ahal izango duzu 'jakina m, horma maite dut, pentsatu nuen.

And you know what? I'm getting homesick. Let's translate this back into English. Now, just to remind you, this was what I wrote before we went on our journey:

I want to give Google Translate a run for its money, so I'm going to feed a lot of English idioms into it this time around. I think that it will strike people as funny just how much stuff gets lost in translation. In fact, not to put too fine a point on it or anything, I think that what's going to emerge at the other end of this is a nearly incomprehensible mishmash of linguistic fragments, utterly devoid of any coherence whatsoever. I really think the prospect of translating this into anything understandable is going to hit a wall pretty quickly. 

....and this is what I ended up with:

Google will compete for money, and a lot of Inglesa translation. Interestingly, because the loss of its contents to interpret the attack. What ہ BR GCH ہ ہ ہ Y Y Y ہہے K Ch covered ھ YJG ہہے Ø ہے ہے K ہ ہ ہ MB KCH ھ PKK to cover the trumpet ےہے Mash Mi K KD Baugod average ML c R, NM AND ں yk N. January Dousari ے F ے ے IK Lee ےے K Ch party RST RNA ہےہ Y ں K fast, you can really 'm of course, I love the wall, I thought.

There's no place like home. There's no place like home.

So okay, let me get rid of all the stuff that just completely got lost on the way there and the way back, and see what I'm left with. To tell you the truth, there's actually a sort of haunting, poetic beauty to it (I also strained out RNA and RST, along with "Dousari," which has no meaning that I know of, but nonetheless sounds like the name of a character in Dune or something):

Google will compete for money, and a lot of Inglesa translation. Interestingly, because the loss of its contents to interpret the attack. What covered to cover the trumpet Mash average AND January Lee party  fast, you can really of course, I love the wall, I thought.

At this point, of course, I set a challenge for myself: how do I write a narrative that actually uses this, consecutively, so that it makes some sort of sense? Hmmm....let's see, let's see (okay, I'll insert a couple of words for the sake of syntax, and I think you'll see, cyberpunk sci-fi is pretty much the only avenue open to me):

"Google will compete for money," I said, as the clean coal smokestacks belched thick sulfuric acid clouds into the night sky.

"And a lot of Inglesa translation, if you know what I mean, huh?" Snix said, elbowing me as one of the plutonium mine workers lining up for their weekly pay envelope, suddenly fell on the ground, his body wracked with seizures. Poor guy. Radiation protection suits ran you at least 35 Kelzigs, and that was with the new currency. 


All around me, people spoke in two or three dozen slangs, tongues that had sprouted from language track brain implant mods that allowed kids to have their own teen speak, sort of the way they spoke gibberish back in the 21st century so as to confuse their parents, and keep their secret plans between them. 


You could actually get a euphoric rush from particular translations of great works. All across the world, programmers frantically tweaked languages, creating new dialects every day. One of the more popular ones was a grafting of the ancient romance languages into a Spanish/French/Italian/Greek/Latin/Hebrew/Aramaic hybrid called The Inglesa Translation.


But these implants led to more hacks inside a person's head than you could imagine, and that's where Snix and I came in. All across the wireless network, trolls prowled, attacking someone's mind with words as sweet as an "I love you" on the lips of a child bringing breakfast in bed to Mom on her birthday. 


Those seemingly sweet, innocent words were part of an innocent language of childhood, a popular implant. The user, with the implant, processed language in such a way that every word felt fresh and new. 


It made any user vulnerable to a neural hack attack, though. It was a simple concrete language, and in giving it the joy of childhood, the programmers realized they would have to lose something, and that something was the more sophisticated syntactical constructs, the parts that interpreted abstract ideas, which were, so many times, the foundation of duplicity. 

Because of the loss of its contents to interpret the attack, this simplistic language, with its innocence and wonder, gave its user a decreased awareness of any guile or deceit in the thoughts of the trolls that now invaded the mind.

When a joy stealer entered the mind, he first pilfered the simple joys of each day. The first cigarette. The morning cup of coffee. From there, it was a simple matter to reach back into the person's memories of single-digit-age birthdays, and holiday mornings when the family exchanged gifts.


They called it The Trumpet Mash, because it actually gave its victim a moment of pure, timeless bliss before all the joy left their mind via the neural language implant, traveling through the network to the subdural processor chip of the troll who stored those joys, and sold them on the black market at a premium. 


So Snix and I were following those signals, and every so often, we got lucky and busted some punks, and a few people actually kept their joy. For another day, at least.


But this took time, and while we covered the network, who watched the station, where the forces of good, already stretched thin, kept criminals overnight before they were arraigned the next morning?


What covered the station while we were on our daily beat, which was to cover the Trumpet Mash with a net of law and order? Did we even have four cyberdogs in front of the station house tonight? Or were they out stopping drunken fights, they way they always did?


On one of the holographic billboards, the trailer for the newest January Lee film played and repeated, played and repeated. 


Two kids, out of their minds on about a hundred different languages at the same time, were so overcome with the understanding of the world from so many viewpoints that they spoke in monosyllabic, euphoric bursts:


"Oh, man, this isn't even CLOSE to the way I felt yesterday! This is, this like, AVERAGE, AND..."


"JANUARY LEE!" the other screamed, seeing the holographic trailer.


"JANUARY LEE!" the first kid screamed back, "PARTY! FAST!"


"You can really party fast, of course," the second kid said, before crumpling up in hysterical laughter. 


I just watched them, scanning the network with my digital optics, seeing the signals as they shot here and there, looking for anything out of the ordinary as the signals went in out of people's heads, in and out. 


I leaned against the wall, sipping my coffee.


I love the wall, thought. 


...and when you think about it, don't we all love the wall?

But I digress. I'm sure you can come up with ways to mess around with Google Translate far more amusing and creative than I could have come up with. Use it. Enjoy it. Amuse yourself.



No comments: