by Carme Mangiron
Over four decades, the video game industry has become a world-wide phenomenon which generated 152.1 billion dollars in 2019 and is expected to continue growing (Newzoo, 2019). In fact, the video game industry has even thrived during the global crisis caused by the COVID-19 pandemic (The Washington Post, 2020), as people around the globe were in lockdown and games became an excellent option to pass the time and have contact with others when playing online. To a large extent, the global success of the video game industry is due to game localization, as 50% of revenues
come from international markets (Chandler and Deming, 2012). The main skopos or purpose of game localization is to produce a target market version with the same functionality as the original that also provides a similar gameplay experience (O’Hagan & Mangiron, 2013). It involves complex technical, linguistic, cultural, legal and marketing processes and is considered a hybrid type of translation, as it shares features in common with audiovisual translation and software localization, and, to a lesser extent technical translation and literary translation.
Most games are developed in English or Japanese, as Japan has always been one of the main players in the video game industry. However, many Japanese games are translated into English first and then into other languages using English as a pivot language. Localization demand has been growing over the last decade, as has the number of the target languages. While in the early days games were usually translated into French, Italian, German and Spanish (FIGS), nowadays they are also being translated into Korean, Chinese, Russian, Polish, Brazilian Portuguese, Turkish and Thai.
Translation volume is also increasing, as many games have expansions, downloadable content (DCL), patches and regular updates. Often game translators are also responsible for associated materials, such as websites, marketing texts, strategy guides, health and safety, and legal texts.
When localizing a game, it is important to use easy-to-understand, and idiomatic language as players are multitasking and to enjoy the game they need to easily understand what they must do in order to progress. In addition, in games that are set in fantasy worlds there is a tendency towards adaptation of cultural references and humor, which requires translators to be creative. In the case of games set in historical periods or based on events or literary or audiovisual works, translators have less creative freedom and will have to use their documentation skills to ensure that their translation is consistent with historical facts or established translations.
There are also several constraints a game translator faces, such as lack of context or limitations of space. Nowadays, most games are being localized in an agile localization process following the simultaneous-shipment (sim-ship) model—they are being translated as they are being developed—, so translators have to work with a source text that is constantly changing and without access to the game’s visual context. Translating games, which are multimedia, multimodal and audiovisual texts, without access to the visual content is known in the industry as blind translation (Dietz, 2006) and is one of the main hurdles game localizers must overcome. The fact that video games are nonlinear texts, because they unfold depending on the decisions and actions taken by the player, also poses challenges to the translator. Sometimes they may even translate all the dialogue by a certain character without knowing who the character is talking to or their responses. Also, original text may be populated by tags and variables. Tags or control codes are instructions to the game engine and if they are translated or modified the game code will be broken.
Variables are placeholders that will be replaced by different information depending on the progress of the player. Translators must figure out how to replace the variable and formulate the sentence in which they appear in a way that will be correct in all cases. This is particularly challenging when localizing from Japanese or English into Romance languages because in those languages adjectives must agree in gender and number with the nouns they modify. For example, when translating a string like “You obtained a <color> <object>.” into Spanish, the variable <color> must be moved to after <object> to reproduce the most common syntactic order, which is noun + adjective. Also, the fact that nouns can be feminine or masculine and the article and the adjective must agree with them poses a problem, as we could end up with a sentence such as: “Obtuviste un espada antiguo”, which is grammatically incorrect. In this case, a possible solution is to translate the sentence as “Obtuviste: <object> de color <color>, so there would be no issues with the indeterminate article “un/una” or with the adjective. Another major constraint in game localization is space limitations, especially in the user interface and in subtitles, which means that translators have to think of creative ways to produce a translation that appeals to players and fits in the available space.
Additionally, there are different levels of localization, such as box and docs, partial localization and full localization (Chandler and O’Malley Deming, 2012). The term box and docs refers to the translation of the box and supporting materials, such as the manual and associated texts. Partial localization refers to the translation of all text-only assets into the target language, such as the user interface, system messages, tutorials, narrative and descriptive passages and dialogues without audio. Audio and cinematic assets are kept in the original language and subtitled in the target language. Full localization is used when all textual, audio and cinematic assets are translated or dubbed into the target language.
To localize a game, there are generally three stages: a) pre-localization, b) localization, and c) post-localization. Pre-localization includes all the work carried out before the translation starts. For translators, this consists of familiarizing themselves with the game, by playing it, if it happens to be available, or examining all the available information, such as walkthroughs and screenshots, if it’s not. If more than one translator is involved, they also create glossaries and style guides in order to ensure consistency. This is followed by the localization process, which includes translating, reviewing and editing the text. If the game is fully localized, once the translation is finished the script is recorded in a studio with a dubbing director and dubbing actors. After the localization process is finished, post-localization consists of game implementation, carried out by a localization engineer, and followed by quality assurance (QA), also known as testing or debugging. Because
game localizers often work without access to the visual context, the QA process is of paramount importance. It is the first time that the translated text can be seen in context so errors due to the
lack of context, as well as any other errors, known as bugs, such as functionality, linguistic, textual, cosmetic and translation errors, to name but a few, are amended at this stage.
The most common tools used in game localization are word processing and spreadsheet software and computer-assisted translation tools (CAT tools). In fact, Excel is the favored file format for developers because it makes importing and exporting text into the code very easy. However, Excel is not a tool developed for translation purposes, so often localization vendors import Excel files into CAT tools in order to optimize the translation process and provide consistency across translators. Customers often indicate their required tools, but memoq and Memsource are widely used in the game localization industry. Other tools that are used are SDL Trados, Smartling, XTM, POEditor, Transifex, and Wordbee. Alternatively, developers may use a proprietary tool that incorporates content management functionality as well as CAT tools, and provides contextual information, such as who is talking, who they are talking to, their age and gender, etc. They may even provide dialogue trees, in which a full conversation can be viewed, with all possible replies, which makes it easier to understand and translate properly.
Additionally, a number of game companies that develop triple A games—blockbusters in the game industry—have started to use artificial intelligence for dubbing games. The software manipulates graphics so that facial muscles move in sync with target text phonetic content. Another emerging technology in the game localization landscape is machine translation (MT), which some companies, such as Electronic Arts, have started to use in combination with postediting for some content, such as tutorials, user guides and websites. However, mistranslations, terminology inconsistencies, problems translating tags and variables and style and spacing issues have been detected in postedited machine translation (Anselmi and Rubio, 2020).
This means that more research is needed before MT can be applied to some game content and it seems unlikely that it will be used for text requiring creativity or idiomatic language.
Undoubtedly, game localization is a challenging type of audiovisual translation. Game translators deal with different text types, often work without access to the original game and encounter challenges such as tags, variables and space restrictions. However, when compared with different types of translation, game translators are granted more freedom and have a more authorial role, leaving their imprint on the target text. As demand for game localization is growing, it is undoubtedly an interesting option for audiovisual translators wishing to embark on new ventures.