Translating fun: how to make games the world will enjoy

Utopos

Back in the early ‘90s, when Atari ruled the earth, I was one of the legions of bedroom coders fighting for the affections (and pocket money) of the world’s gamers. I was developing Utopos, my first game for the legendary Atari ST, when I received a letter from Sebastien Larnac, a French gamer who had got hold of a demo copy. Sebastien offered to give me a full French translation in return for a free copy of the full game.

I agreed, and as a result around half of my sales ended up coming from France. This lesson in the power of translation stuck with me, and I have always insisted on multiple languages in any product I work on. Now, nearly 20 years later, Sebastien is still working with me.

Since the 20th anniversary of the release of my first game is approaching, I spoke to Sebastien about Utopos, and the possibilities for today’s generation of budding game translators.

What made you decide to write and offer your services as a translator?

- Well it all started back in 1994 or so. At that time I was a heavy ATARI STe fan, and still am to be honest. As a consequence, I came to know about and fall in love with demos, being artistic demonstrations using the limited hardware as far as possible (and very often beyond that). At that time I strongly wanted to become a game translator, something I actually had to forget about later. So I contacted you to offer a French translation against a free copy. You know, French people have never been really good at languages, though as these were my main studies I thought it could only help to sell that great game.

Utopos seems like a lifetime ago. Do you even remember what it was about?

- Of course! It was a great 50 fps gravity shooter using STe expanded hardware, opening upper and lower borders for instance, adding sampled FX and soundtrack and it was terribly addictive! I played it for years with a good friend of mine and believe me if at first it’s a bit tricky to fight against gravity to avoid crashes, once you get into it, games are simply insanely fun! In the summer of 1995 when I travelled to Helsinki to achieve a compulsory training period for my studies, I clearly remember you and the other guys working on Utopos 2 and I even seem to remember there was a PC version started too. So, as you can see, I haven’t forgotten about that game. Thanks to it, I made a friend, it allowed me to stay in Finland for 2 months and to work again with Transfluent about 20 years later.

Are there any translation challenges that are unique to working in games?

- To be honest, I don’t really remember the process of translating I used back then. Did I only have text file or could I directly test translations in the game? I really cannot tell but I know that as a first translation, it showed me something very useful: every sentence gets longer in French, and you have to make things as short as possible. Having gaming experience helps a lot: even today you can play games with such a poor and inadequate translation that it almost kills all the fun. Knowledge is not enough, you have to love what you are doing.

Do you think the recent boom in low-budget indie game development means there’s room for today’s multilingual gamers to do what you did?

- Well I hope and think that there is still room for beginners out there who want to offer their services to translate independent games but getting hired for long is another story, something that made me change my plans. Even though I still work on translations here and there, and it is something that I really love and think about as a mental challenge, my main job for the last 12 years has been teaching computer studies to kids in schools and also to adults in an open university and I love the job!

Thanks Sebastien! Let’s check back in 20 years to see where we are!

Photo credit: pennstatelive

People love a good language myth. The idea that Eskimos have 50 words for snow has circulated for years, despite being completely untrue. Myths like this seem to be based on the idea that language comes from the differences in our cultures, rather than in what they share. One language myth that comes up again and again is the idea of untranslatability.

The net is filled with lists of untranslatable words and concepts, but the truth is that while there may be words that don’t translate readily – especially using a machine translator – a skilled human translator can always get the meaning across. If a word were truly untranslatable, there would be no way to even explain it in another language. As the (slightly deranged) philosopher Ludwig Wittgenstein once wrote: “If a lion could speak to us, we could not understand him.” My point is that while language may be a core part of our identity and even affect the way we think, our ideas are still fundamentally human and translatable.

Hyggelig: nice word, but hard to translate

One particularly pleasant example of a supposedly untranslatable word is the Danish word “hyggelig,” which can be heard in many parts of Scandinavia. Machine translators tend to render it in English as “nice,” but while that is technically accurate in the most basic sense, there’s a whole lot more to hyggelig than that.

Have you ever sat around a campfire with your closest friends and felt a wave of peaceful coziness wash over you? That’s hyggelig. But as well as describing an emotion, there is also an aspirational sense to the word: hyggelig is something to strive for. Anything that may lead to a feeling of hyggelig is itself hyggelig.

The easiest way to explain this is to look at the word’s origin: hyggelig comes from the Old Norse “hyggja,” meaning to believe, to wish or to intend, and “huggan,” meaning care, consolation and comfort. Taken together, these Old Norse concepts signify a warmth towards the whole of humankind, which might be rare in today’s dog-eat-dog world (especially when stuck in traffic) but which is worth striving for.

Goodwill to all of humankind is a little more than just “nice.” While there is no one English word that reflects every facet of hyggelig, it can still be translated: A sensitive translator can pick from the many English words that correspond to hyggelig feelings. By using a phrase made from words like ‘cozy,’ ‘warm’ or ‘welcoming,’ a translator can capture the sentiment of hyggelig.

This is where the fundamental difference between human and machine translators lies: while machines might be able to provide a literal translation that expresses one part of a complex, emotionally nuanced concept, it takes a real human – with real feelings – to see and accurately translate the whole picture.

In February Google Translate finally added Esperanto to its roster of languages, a full century and a quarter after it was invented by L.L. Zamenhoff. Does this mean that we should start cramming our brains full of Esperanto and forget about our mother tongues?

The answer, of course, is no. 125 years after it was created, Esperanto continues to be spoken by only a small fraction of the world’s population. If there is a global language today it is surely English (or ‘Globish’), although given only around half of internet users today understand English, it is not much of a global language. Yet the failure of Esperanto to take over the world may hold lessons for the future of the world’s languages, machine translation and Globish.

The importance of culture

So why has Esperanto failed to gain traction around the world? One problem critics point to is its Eurocentrism. The Esperanto vocabulary seems to be disproportionately based on the romance languages. If it had incorporated a wider spectrum of languages it may have been more popular. However, even in Europe, the population of Esperanto speakers remains very small.

Perhaps a bigger problem is that without an associated culture or nation there is no one to promote Esperanto, no one making films in it, no one conducting trade or foreign affairs using it. This is in stark contrast to English: No country has more cultural exports or global influence than the US (or the British before them). This explains the prevalence of English as the world’s second language.

But it is culture that will probably help prevent Globish taking over the world. Language is more than just a way to communicate. For example, although I enjoy speaking English, it will never replace my mother tongue, Finnish. Unlike my adopted languages, Finnish is so ingrained in my culture that it forms part of my identity.

Google it?

It is culture that also causes problems for machine translators. Culture produces idiomatic phrases and slang terms unique to that culture’s language. A human translator needs to be familiar enough with both languages to understand the context of the term and think of a phrase that carries the same sentiment.

Even software such as Google Translate, which relies on statistical analysis rather than grammatical rules, has difficulty with slang and idioms. The internet is full of articles documenting its troubles.

The irony is that the problems machines have with translating cultural references make Esperanto the ideal language for machine translators. Without an established culture generating slang and idioms, Esperanto should create fewer difficulties for machine translators. (Of course, if everyone spoke Esperanto there wouldn’t be much need for machine translation!)

So what does the future hold for global languages, culture and machine translation? Of course, no one really knows. If the story of Esperanto has taught us anything, perhaps it is that the evolution of language and culture is difficult to predict.

Nevertheless, here goes: I think because the internet now makes it so easy to communicate in multiple languages, English, Chinese and Spanish will all continue to be very prominent, with local languages growing in popularity. If anything, I can see cultural differences diminishing due to globalization and the open internet, but not to the detriment of local languages. I think those who embrace local languages will prosper and those who attempt to stick to a global language (eg. English) will ultimately fail. For at least the medium term, therefore, I see there being plenty of work for translators.

When I was at school, as far as I can recall, my language teachers could all speak the language they were supposed to be teaching me.How the world has changed. Today Senja Larsen is helping thousands of people learn Swedish, despite the fact she can’t really speak it herself.

Senja Teaches you Swedish is an online language school with a difference. As Senja herself learns Swedish, a few words at a time, she shares the knowledge on Facebook.

Each mini-lesson focuses on a few useful words, with extra information supplied by a team of around 20 volunteers chosen by Senja. Though the lessons are posted by Senja and her volunteers, anyone can contribute. The page’s lively – and often hilarious – comments make each lesson more memorable, and help to give context to the words being learned. It seems to be working: the Senja Teaches You Swedish page now has over 13,000 members, all learning and discussing the Swedish language together.

Senja was inspired by her friend Samu, who offers Estonian lessons online, and the idea is proving infectious. As well as Senja and Samu’s language lessons, there is even an electronics school.

Senja now hopes to take her school into the offline world, and is using KickStarter to raise funding for a book. As I write this, Senja has raised $4,800 of her $12,000 goal, with two weeks left to go. I spoke to Senja about how she came up with the idea for her unique language school, and her hopes for the future.

Senja Teaches You Swedish is a totally original idea. What gave you the idea of sharing your learning process in this way?

- I wanted to learn Swedish but there was no place one could learn a word or two every day. In effect I established the site because I wanted to join it.

And what led from there to the KickStarter page?

- Publishers were interested in our book but having one would have doubled its costs for the end user. Since the content of the book is crowdsourced, we thought crowd funding would suit it perfectly.

How did you recruit your group of volunteers to help with the daily postings?

- They found the school! One is in Australia, another in California. As we have an online school, it is possible to participate regardless of time, place and level of knowledge.

What makes this such a good way to learn a language?

- Learning languages is a slow process. Words are the building blocks. In effect if you learn 1000 words in a language, you will understand most of what is going on.

What do you hope to achieve in the long term?

- More fabulous fun. I get to work with a wonderful group of people and continuously meet new ones. I learn new skills, which make me better at my work, which is PR and communications. As a bonus I learn new words in a language I did not speak one year ago. What more could one want?

Photo credit: DrJohnBullas

Imagine you’re a rock star, sunning yourself on the deck of your megayacht in the South China Sea. Margarita in one hand, smartphone in the other, the world at your feet.

Your agent has been nagging you to put in some work promoting your new album, so you fire off a quick tweet to your army of loyal followers. As the retweets roll in you congratulate yourself for mastering the art of self-promotion. You look up at the lights of Hong Kong on the horizon and it hits you. Why aren’t you talking to them as well?

As we’ve mentioned before, multilingual tweeting can really help get your message across. Whether you’re a celebrity or in business, or you just want to be heard, the key to the new global economy lies in mastering the art of global communication.

But before you order another margarita and set to work running your tweets through an automatic web translator, you might want to check out just how confusing – and potentially disastrous – the results could be.

The big problem with machine translation is its lack of sensitivity. For example, machine translators don’t understand jokes, puns or idioms very well. This means that your carefully-crafted tweets may not make sense, or worse, come to mean something completely different.

Preserving meaning in translations between English and Chinese is notoriously difficult, even for humans. It’s no surprise the machines can’t keep up. Unfortunate translations have inspired everything from memes to academic discussion. So how can you make sure a mistranslated tweet doesn’t make you famous for the wrong reason?

For example, say you wanted to translate the English phrase ‘like father, like son.’ After feeding the phrase into a popular online translator I was given 父亲一样，喜欢儿子, which has the literal meaning ‘His father, like a son.’ Not only is the translation confusing, it is also unnecessarily long. This could be a pain if you’re trying to keep to Twitter’s 140 character limit. The most literal translation of ‘like father, like son’ in Chinese is 有其父必有其子, which simply means ‘the father is like the son.’ But if you really want to be clear and concise, the phrase you need is 虎父无犬子, or ‘a tiger does not father a dog.’ This version, which a human translator would use, retains the original meaning, uses fewer characters, and has a far more poetic feel than the cold, clunky machine translation.

When it comes to fast, accurate, culturally sensitive translation, nothing beats the power of the human brain.

Now for a quick commercial break: At Transfluent, we use over 15,000 human brains to translate messages into more than 60 languages. By linking your Twitter feed with our automated, human-powered system, you can combine the speed of a machine translation with the accuracy only a human can provide.

And what’s more, you can do it all from your yacht.

Transfluent for Apps

August 1, 2012 Helsinki – Transfluent (www.transfluent.com) today announced an API (Application Programming Interface) for easy translation of mobile applications developed for Apple’s iOS devices, including the iPhone and iPad, and Android devices, into multiple languages.

The Transfluent API for Apps is the world’s first professional language translation service that can be embedded directly into the application build process. This means that every time the developer makes changes to the application, all of the different language versions also get updated at the same time.

Mobile applications are a truly global market, and localizing an application typically increases sales significantly. Despite the growing need for app localization, there has not been an efficient method of translating the apps into multiple languages until now.

The Transfluent API makes it possible to develop applications directly in multiple languages (as opposed to translating once the development has been completed). This shortens the time-to-market, increases the quality of the localized versions and lowers the cost of maintaining the multiple language versions.

Transfluent’s backend is powered by a network of more than 15,000 professional translators in over 60 languages. Each translation task is assigned to two independent translators – one does the translation and another proofreads the text. The API helps the developer specify context for the translation task, including screenshots taken from the application.

“For the best results, we recommend starting the translation process early on in the development of the app. Developing the app and testing it in multiple languages throughout the process guarantees that the localized versions work and feel as native as the original. Loyal users who are native speakers of various languages can be a great resource in making sure all the terminology is spot-on!” comments Jani Penttinen, founder and CEO of Transfluent.

Transfluent keeps track of the previous translations so that only changes and additions are translated. If required, the text can be sent to translation dozens of times a day, but translation will only performed where needed.

Transfluent has quickly become a trusted partner for prominent application developers, such as DocScanner developer Haave Inc., Recoil, NonStop Games and Kapu Toys.

Tommy Incrosnatu, CEO of Kapu Toys commented on Transfluent: “Kapu Toys is developing new kinds of mobile games for children. Transfluent has supplied us with comprehensive translation solutions which have helped us in production and marketing. Our co-operation with Transfluent has worked well; the service is quick and knowledgeable. Now, with the new API, updating new translation versions is even easier and more trouble-free.”

About Transfluent

Transfluent is a provider of rapid, professional translations for social media, websites and applications. Transfluent works in Facebook, Twitter and WordPress and the Transfluent API can be used to embed the near-realtime translations in applications. Transfluent is powered by over 15 000 professional translators, available 24/7, in over 60 languages.

Transfluent is one of the pioneers of professional crowdsourcing. Using crowdsourcing techniques with professional resources gives Transfluent the ability to deliver high quality translations as fast as humanly possible.

Transfluent for Apps can be found at http://www.transfluent.com/apps/