Thoughts on MTL for languages other than English?

[InnocentLurker]

I've been playing with MTL tools recently and I'm wondering whether it'd be worthwhile to add machine translations to my game.

 

[FromOtherSpace]

Few days ago, I saw one (indie) dev saying that 20% of world population speak english, so having more than 1 language is never a bad thing. He's not into adult gaming and don't say which language are "best".

Game market is huge in USA, China, Korea, Japan then in Europe (I don't remember exactly... Something like UK, Germany France, Netherland)

With Chinese, Indi, Japanese, Korean (which are "not" spoken out of their countries but have a huge population and/or market) English, Spanish, French and Arabian you can cover most of the world (those have influence widly out of their borders). Some language are spoke in many African countries, I don't remember their name.

 

[anne O'nymous]

Few days ago, I saw one (indie) dev saying that 20% of world population speak english, [...]

It's expected around 1.3 billions, therefore more like a roughly 15%, but the important point is that it's intended as fluent speakers. You can easily triple this percent if you include peoples able to understand it enough to follow a game story. But it doesn't mean that adding another language isn't a good idea, after all it still only make 45% of the population.
But in the same time, parts of the world where it would be unlikely to find someone having at least a basic level in English match with the parts of the world where it would be unlikely to find someone having a computer and internet connection good enough to get and play adult games. And in those countries, the ones that would have both the computer and internet connection would be parts of the middle class or upper one. Therefore there's high chance that they would have an English level good enough to understand the story ; and if it's not English, than it would probably be French or Spanish.

He's not into adult gaming and don't say which language are "best".

Because there isn't a best one beside English.

If you target South America, Spanish would be the best, but if you target Africa, it would be French. For Europe, it would be a tie between Spanish and French, but German would also be a good pick. Russian could also be a good choice there, but I guess that it's less learned in East Europe than it was thirty years ago.
Then in number of speakers, Mandarin and Indian should be what one pick. But in the same time I doubt that adult games are this welcomed in China, like I doubt that, both in China and India, there's really this many people with the needed computer and internet connection. In the same time, just 10% of one billion is already 100 millions possible players, what is half what you could get in Europe and a third for the USA.
Japanese could also be a good pick, it's the leading video gaming market, but they have a large video gaming scene, and seem less interested in Western production ; but it can be due to the fact that the percent of English speakers is relatively low.

In the end, would have I to chose, I would probably pick Spanish, French and German. Then on second pick Dutch, Italian and Russian.
Mandarin and Japanese would probably lead to the same number of possible players, but they are too difficult languages to rely on MTL for them. While an error can be corrected by the reader, at least half the time, in case of Mandarin/Japanese to English translation, the opposite would lead to too much confusion.
Arabian could also be a good pick, but most Arabian countries are sexually repressed, and some would even send you in jail if you're caught with such game. Therefore it would mostly apply to emigrants, so people who would also speak another language.
Apparently there's also a market for Turkish since a site specialized itself in translating (Ren'Py) adult games in that language. But I'm not really sure that it's an as big market as that.

Some language are spoke in many African countries, I don't remember their name.

In number of countries, Swahili is probably the first one, will Afrikaans should lead in number of speakers. But it fall back to the same issue than with Mandarin and Indian ; the part of the population that have the needed computer and internet connection is relatively low. And most of them would probably also speak either English or French, with a smaller parts for Spanish ; or at least understand it enough to follow the story in this language.

 

[FromOtherSpace]

I agree with anything you said.

Especially :

In the end, would have I to chose, I would probably pick Spanish, French and German.

It seem the better palette.

 

[InnocentLurker]

I doubt that adult games are this welcomed in China

I've been told that it's easier to get around the great firewall than you'd expect, with simply downloading a VPN (which are legal apparently). So it's more that the CCP likes to keep news out than to keep everyone in, especially capable people who would just find a workaround anyway.
There's also Taiwan, I know there's a good number of people with internet there.

I was hoping to get some opinions on the quality of MTL for languages like Chinese, but I can ask elsewhere if there aren't any speakers here.

In the end, would have I to chose, I would probably pick Spanish, French and German. Then on second pick Dutch, Italian and Russian.

Definitely. The Western European languages would probably have the best quality since they're such close cousins of English.

I'm less sure for any outside Europe, but the translation process seems pretty intuitive. I could include Chinese and Hindi just for the fun of it depending on how long it takes to generate.
Arabic I'd want to include too, but the right-to-left writing direction might break a lot of my UI.

Thanks for the reply!

 

[Tompte]

Speaking as a bilingual person, the quality of machine translation varies wildly between languages (and I argue it's still far from good enough to be used without a proof reader, regardless of language.)

English has a ton of words with identical spelling that can mean different things, making translation from English a harder problem than you'd might think. Most automated translators will make a weighted guess. So you'll read some text, and then suddenly there's a random word in the middle of a sentence, and you have no idea where it came from or what the original meaning was (unless you speak English and can infer the error and re-translate it yourself).

In my native language, which isn't a major spoken language, Google Translate is consistently wrong, to the point of being completely useless.

As far as I know, there is no machine translator that actually does a decent job. The results you get are too approximate to be used in a game or a story, if you ask me.

 

[InnocentLurker]

As far as I know, there is no machine translator that actually does a decent job. The results you get are too approximate to be used in a game or a story, if you ask me.

That's what I've been hearing yeah.
DeepL is supposedly pretty good, using some AI magic, but also pretty expensive, and not offering that many languages, etc.

I guess this tech is just going to have to get better before something like this is viable.

 

[Tompte]

DeepL is better than most but it still gives me weird results from time to time, like repeated sentences.
But where Google Translate gives you a word salad, DeepL appears to at least be trying. Tech is better but not yet where it needs to be, imo.