Why hasn't MT Mastered (yet) video games Localization? a summary of my contribution to W.L UK event PART 2!!

Last week in part 1 of this post covered the reasons why MT and videogames are not a perfect symbiosis at the moment.

But I understand that many of you would like to try MT for games, you would like to explore the viability of such a localization strategy.

In my previous article, I explained the 3 main reasons why I think MT has not been successful (yet) in localizing videogames, but maybe despite what I explained, you still want to try and explore the potential of MT in games! so if you find yourself in that scenario, you should consider the following 2 ideas to successfully explore the binomial MT and videogames.

There are 2 fundamental areas you should consider when implementing MT as part of your game localization strategy.

1.- The lower the sensitivity priority of the content to be localized, the better the MT output will be

The first considerations you must asses are that each type of content deserves its own translation approach

Some content may be perfectly suitable for translation with MT.

Examples of that content might be User Forums or FAQs; while MT may not be the best recommendation for all your game content, there may still be some circumstances where MT can be your ally.

The following Nimdzi's infographics can give you a good idea of which path from a content perspective is the right one to follow if you decide to explore what MT can do for you.

View fullsize

Source: Nimdzi Insights

Forums, FAQs, or blog posts are types of content where you can be successful in applying MT.

This type of content is less creative, and the English applied tends to be more simplified. In fact, in many cases, it is what is called "controlled English". Controlled English (also known as simplified English) was originally developed for the aerospace industry (mistakes in that industry pay dearly! so clarity in the language is a vital requirement). Since it is a controlled language, it’s used a limited vocabulary where each word has a unique and clearly defined meaning.

By keeping the use of words and linguistic construction simple and consistent, this type of English minimizes the potential for misunderstanding because there is consistency….and this is where TM shines!

Consequently, the chances of obtaining a good result by applying machine translation in a FAQ are much higher than in a marketing campaign. FAQs will use a simple grammatical structure to explain technical problems/troubleshooting

If we look again at the image of the content of the pyramid, material that is very insensitive to localization will probably meet the needs of its potential users just by applying MT and publishing a raw MT; while in cases where the sensitivity priority is low/medium you may need a combination of MT + post-editing. That will increase the chances of producing a good localized copy of your content!

If you consider the previous pyramid, the content from bottom to top is easier to process through an MT solution.

Focus on that particular content and in that order if you want to implement MT as part of your game localization strategy. When you are producing content that is more informative and less intended to appeal to the emotions or passion of the players, then MT could do a better job.

Players' experience with content that is not yellow is easier to achieve with MT, so if you ever start an MT program for video games, start with the red content, then the blue, then the purple.

2.-How will MT be integrated with your workflow and tools?

The second aspect you have to stop and think about is how you are going to integrate MT into the publisher's technological framework.

In the following graphic, I create a workflow with what I think we can consider as a "standard" configuration in a game company with localization as part of its continuous delivery framework

View fullsize

Whichever MT solution you decide to adopt (Amazon Translate, Watson, Google Translate, Microsoft Translator...) you must ensure that the chosen solution is integrated into the current development workflow of the game.

All affected stakeholders by localization activities (game developers, QA testers, Localization Project Managers, and LSPs) should be involved in the conversation to understand how MT will be implemented into the technology framework and keep in mind that the framework should be prepared to adapt/integrate to this new technology because:

1. MT is not a stand-alone solution

2. It is not a plug-and-play technology

3. and it should be integrated (and thoroughly tested) with your current game development environment

When implementing MT you need to ensure you test the workflow between the code repositories used by your developers vs how those files might be processed by your MT engine

If you use a connector you must ensure that the bridge you are creating between the strings in English, the TMS, and the MT engine is handling efficiently all the localization logistics and managing properly the state of translation delivery

Not all content is equally complex, for this reason, you might want to start applying MT to that less creative content to make sure you take the time to analyze how to integrate the technology chosen by TM into the current development workflow of the game.

Globalizing a videogame is a multi-step process. As we've seen, there are so many things to consider when translating a game that there are more than a few places where MT could fail.

You have to consider how to integrate creative content, with more technical and repetitive content, for the first type, MT won't work for you, but maybe for the second..., it will work for you!

I hope this post has been useful, and if you have experience implementing MT in videogames I'd love you to leave your comments below, or better yet! Send me an email and we can have a virtual coffee in Zoom(?) that would be awesome as I'd like to chat with you and see how you approached the complexity of integrating MT technology as part of your daily work

Have a great week!

@yolocalizo