Upcoming Platform-Wide Automatic Translation Experiment

how does ‘poor moderation’ correlate to localization? lmao

When you do translate a game to not official supported languages by roblox. The filter has a high chance of not allowing you to apply the translation, even if the translation has nothing bad in it. I know about this from previous experience i had with translating my games on other languages.

1 Like

The filter applys also to the website (example: titles, descriptions, group posts and also translations.). Because translations is something that everyone can access, roblox moderates translations based on their current filter( this applies to all languages). The problem is that since the filter is not the most accurate think, it can limit the ways you can translate your game if not completly blocking you because “your translation has been moderated”. Also i didn’t metion that the localisation team has anything to do with the filter, I said than the filter affects the translation of games.

The main think that i try to achieve with my feedback is that Automatic Translation is not a good and a not perfect idea. We are still very back in the automatic translation technology. It is not a good idea to run an experiment like this Especialy in the most popular games on the platform.

I’m curious on how it would work. Would it detect the language they are speaking then the language you are speaking and then translate? Otherwise the translator would have no source.

1 Like

It works like Roblox’s system works. You specify the games source language, and then (assuming you manually add keys/values or use automatic capture) it’ll generate a blank localisation table.

That table can then be edited, be it by translators or by AI.


I feel like this is the end of the line for most translators.


I am very interested in this feature and would see it as regrettable if people try to ruin its existence before even using/seeing it.

Clearly translators will not lose their job, but this has the potential to benefit smaller developers that can not afford expensive, yet great, translations services. It’s important to remember that Roblox has intelligent engineers and they understand the importance of collecting data and attempting to provide context while training the AI that will auto translate games.

dues this mean we could also understand what other people are saying in the chat? its a freaking pain talking to a guy while having to run everything they say, and what you want to say threw google translate…
there are a few times where i was the only english speaking person on a server, so i got a good idea knowing what its like for them on roblox when only english speakers are playing, i felt a bit out of place…

Wouldn’t this affect text that has textScaled false by turning it into a mess to read and fit on the gui because the string is suddenly twice the length as before?

1 Like

As a translator, I don’t think this will be very useful.

If we look at the current automatic translation… it’s really a disaster
Some words can have a lot of meanings and how can we be sure that it will find the appropriate meaning? More attention can be given to translators than automatic translation
For example, things like the role of “Trusted Translator” can be created for translators.

Some translators aren’t really reliable, for example …

He is advertising his own. It’s really ridiculous to do this in the game
And there are many translators like that

Automatic translation can be useful to get rid of such people. But I’m not sure how accurate the translations will be, I’m worried.


I’m a native Spanish speaker that has worked 1 or 2 times as a translator and trust me, this is not a good idea at all. Automatic translations are almost in every case absolutely TERRIBLE, and I believe 99% of non-English Roblox users would rather have the game in English than having it poorly translated.

Please make this experiment opt-in instead of opt-out. Automatic translations miss a LOT of context and can be extremely distracting lowering user-retention.

I can understand that many English speakers seem to take translations for granted, but honestly, if you really want to see the effect this change can have in your game, translate a bunch of strings in your game to Spanish/Portuguese/Russian/Chinese, translate them back, and imagine you’re a poor user playing your game for the first time with nothing other than flashy buttons and those poorly translated strings.


I’m a translator myself, but I prefer seeing this from a game developer standpoint. This will have a very negative impact on user retention, and there’s no denying that. A lot of English speaker developers seem to believe that automatic translations will magically make non-English players “understand” how to play their game, when in it reality just makes it worse.


So you think this will actually increase demand for developers due to their lack of trust within Roblox’s experiment?

If this update is implemented the whole niche of translation work will be gone. I recommend to reconsider.

I am worried about the implications of this. Back when the Spanish category in the devforum still used google translate, it started to lose sense, or sometimes even change the meaning. I was frustrated enough back then.

Thankfully, now the devrel manually translate the announcements, which improves the quality by orders of magnitude. Google translate tends to butcher everything up, sometimes creating incoherent sentences.

Additionally, grammatical gender may be lost between translations. Spanish example:
“The cat” can be translated as either “El gato” (m) or “La gata” (f). Which means that automatic translation will more often than not, default to male grammatical gender.

Another implication would be the loss of connotation. Connotation is important when transmitting an idea. For example, both skinny and slender have the same meaning, but skinny has a more negative connotation than slender. Connotation transmits emotion and intention. Translators are only able to understand the denotative component of language, not the connotative.

Another example, this time with French. The word gamer, when translated to french by google translate, comes up as joueur or joueuse (yes, more grammatical gender problems). But, when either of those words get translated back into English, it comes out as player. Showing how information will inevitably be lost during translation.

Until humanity creates an AI able to comprehend language like we do, automatic translation will have the aforementioned issues, and many more. A human translator is able to keep the emotions, intent, and intentions of the original text when translating.


I second this. English => Spanish automatic translation grinds my gears. This is why I always use services in English instead. Most software that use automatic translation to Spanish end up messing all the text up to the point that it makes no sense.

More on this here.


I’m a multilingual and it’s really funny to see when a player says something ridiculous in a different language(that I understand) behind someones back.
If this turns out well, maybe no one EVER has the chance to say something rude at a person without nobody else understanding because of the translations.
This is a major update and I hope it goes as planned.

Hello again!

Thank you for all of your feedback. I wanted to answer some of the FAQs I’ve seen here in the comments to help provide more color to this experiment.

First and foremost, why are we running this experiment?

In our research, we have gathered that bridging the language barrier gap will exponentially help us reach our goal of joining communities globally and we wanted to ensure that all of you were able to do so with the click of a button. Therefore, we want to test our automatic translation services at scale. We have already run a series of small scale experiments that have given us positive directional signals and believe that this large one will do the same. We want to assure you all that throughout the course of this experiment, our team will be monitoring metrics closely. We have your best interests in mind and want to make sure you all succeed so if we get negative signals, we will pull back our experiment!

For those of you asking, why is this experiment opt-out? For all of us to have the greatest understanding of the impact of automatic translations in game, we need this test to be run on a large percentage of playtime for international players. With more players and playtime coverage that we get in this test, we will be able to best accurately judge our translation features and worldwide impact.

How does this experiment impact existing localization systems and settings?

If your game has custom localization systems in place that use embedded tables OR you have already been managing a cloud localization table with existing translations, those will ALWAYS override translations in our experiment. Any translation work already done will not be lost and you can continue to add translations to your game which will override translations set by our experiment. Our experiment is meant to fill in any blanks and translate text that hasn’t already been handled by each developer. That being said, at this time, we will be overriding developer settings for Automatic Translations that may have been set in the Configure Localization page.

Also, if you have the AutoLocalize property on GUIs set to be false, we will be respecting that setting. If your game has custom chat capabilities, we ask that you set the AutoLocalize property to be false so that it will not get translated. More info available here.

What content will be translated?

At this time, this experiment will only extend to in-game text. So, it will not include images or content like game info/products. Additionally, we will not be translating system in-game chat although that is something we are exploring for the future.

What languages will this experiment run in?

Spanish, French, German, Portuguese, Japanese, Korean, Simplified Chinese and Traditional Chinese.

How does this impact translators?

Translators are, and will always be, a valuable part of our Roblox ecosystem. Their ability to provide contextual translations often yields higher quality results than an automated system. That being said, we know that not every developer has the luxury of outsourcing and paying translators for their services. To that end, we want to provide built-in product functionality so all developers can reach global players. If you are a developer that has the ability to hire a translator for your game, we encourage you to continue to do so. We previously built features specifically for human translator workflows (Web Translator Portal, translator contribution reports, etc) and will continue to keep them in mind as we develop our translation services going forward.

Thank you again for being a part of our international efforts. As Roblox looks to expand our global reach, we want to make sure you all are part of this journey.


@RoxyBloxyy When will this roll out for all games? I’m looking to translate my game for as many languages as possible, but don’t have the time to do it all myself.

1 Like