Everyone wants work of good quality, but what that looks like can vary from person to person and project to project. This is especially true for game localization. A typical game localization workflow will pass through the hands of many stakeholders, covering such aspects as project management, translation, review, content management and production, to name just a few. Some of these concerns are objective in nature: cost of translation, turnaround time, volume of text, etc. Some are more subjective, like player satisfaction and title-specific preferences.
In general, ultimately it is the client’s standards that must be maintained. “In recent years quality has been defined as a moving target, as it is intrinsically linked with client expectation,” says Laura Casanellas, Machine Translation and Localization Consultant. “At the end of the day, the best translation is one that provides a seamless experience to the player. When they never think that what they are reading or hearing is nothing less than original content from their own language.”
To make sure that all bases are covered, it’s important to know at the outset of a project what quality means, what gets in the way of its execution, and how to ensure it is maintained.
It’s natural to consider context when defining quality. For example, a couple of spelling mistakes may not seem severe on a product website but will stand out on a certified translator's website. Similarly, accuracy errors are more egregious in a medical translation than in marketing materials. Textual style is vital for a publishing house, but not so for a company that writes user guides.
“The translation needs to be devoid of linguistic errors, stylistically pleasing, and successful at adapting dialogue to a specific locale,” relates Laura. “For that reason, in game localization, errors related to style or context can be heavily penalized. Also, since games are usually localized to be played on many platforms, translating the terms of each platform (called first-party terminology) is essential. If this is not achieved, the translator will be presented with a critical error that will fail the whole translation from a quality point of view.”
In the games industry, while quality managers mostly focus on linguistic perfection, cost and time are equally important, if not critical. After all, at the end of the day all entities handling projects must deliver translations by the release date to obtain their return on investment.
Many respected companies within the industry have well-structured localization quality assurance teams. They consist of developers, testers, project managers, and linguists. Well-defined procedures and roles allow them to make sure the appropriate scrutiny is always present.
Frequently, these companies measure quality in localization by means of client satisfaction surveys. This requires an open and effortless communication channel with a client and a frank, well-grounded business relationship with them.
“Quality at PTW is assessed by a linguistic quality measuring system that allows for recognition of error types and evaluation of their severity,” explains our Global Quality Manager, Olga Ewa Mielczarek. “It is used as a basic matrix for measurements performed on linguist outputs in our evaluation tool. Thanks to this, we can evaluate linguists’ performance on a monthly, quarterly, or yearly basis and track quality data from various angles, such as per source, target language pairs, projects, or client groups.”
There are many good quality references in the industry, like the Multidimensional Quality Metrics (MQM) framework, used by localization teams to structure quality programs. The MQM framework is comprehensive, as it contains a very large list of potential errors that cover most possibilities, but it’s also flexible as each company incorporates the error types that are important for their own content types.
Most CAT tools nowadays offer linguistic quality assurance and automatic QA settings that can be defined following well-known frameworks like the MQM.
If we try to define quality specifically from a gamer’s perspective, high quality standards are discernible when translated in-game dialogue and text does not sound awkward, and the language that the characters in a game use has a natural flow.
When the quality of translation falls short of gamers’ expectations, it is almost always because linguists are faced with a lack of context during the translation phase.
There are times when context from reference material is limited or is not provided at all. “Localization QA teams testing these games at a later stage work on iterative builds and thus can easily tell what situation certain characters are in and what characters’ moods are,” says Olga. “But linguists, who must make the dialogue sound natural, often work with plain text divided into segments, which are frequently divorced from the context of the overall narrative. Thus, any form of familiarization with the game a linguist can have is beneficial for all.”
All those involved need to be educated on the consequences of asking for sudden changes that might have a huge cascading effect on the translation. They can affect key terminology that might need to be fully updated in existing translation memories, with subsequent cost to the delivery time and budget.
Often, guidelines are too vague, giving translators a large range of options that they might not need. When something is important, it should become a directive when writing a project guideline, rather than an option. Sometimes queries have not been addressed on time or have not been passed on to translators and reviewers. Any such omission can have an impact on milestones and the eventual deadline for submission. In rare cases, translation memories are not up to date and contain inconsistencies or errors. This can cause trouble farther down the pipeline because they might not be caught in time, given that translation memories by definition need to be trusted sources.
Another factor that can influence quality is the game-specific preferences that clients have for a title. “These must be fully communicated from Account Managers to Project Managers and from Project Managers to Translators, Post-editors, and Reviewers,” says Laura. “They are unique to each title and include onomatopoeia (shouts and exclamations), capitalization, gender forms, addressing forms, and untranslatable terms. It is crucial for translation teams to define needs and requirements as early as possible, typically at the kick-off meeting.”
Finally, an all-too-common occurrence (especially in live games) is when in-game content has been updated and needs to be translated again, but the deadline cannot be extended as the release date has already been announced.
Olga discusses the quality program at PTW. “We’ve implemented quality improvements and procedures like Source Analysis, Quality Assurance Checks, Sanity Checks, RCA (Root and Cause Analysis) procedures, and Style Guide creation projects, to name a few. We also use the linguistic evaluation platform known as ContentQuo, allowing for further quality checks and quality data analysis.”
Widespread usage of ContentQuo can help boost quality levels for all Localization teams. Naturally, to streamline and maximize the use of this tool, additional help from extended teams will be necessary. Extra in-house or external linguistic quality auditors are needed to ensure that translated content reflects the source, that the appropriate terminology is used, and that consistency is guaranteed throughout all projects and files. Linguists and linguistic auditors play the role of certified subject matter experts assessing both our clients’ and our own content.
“The most important thing in a quality program is that everybody in the localization team knows it well and they know the part they need to take in it,” explains Laura. There are different responsibilities for the different players:
The Quality Manager defines the strategy, creates the quality model the LQA testing will be based on, designs the workflow, selects the tools that are going to be needed to put the program in place, trains the staff involved (PMs, vendor manager, linguists), analyzes and interprets quality results, and sets up the quality strategy.
The Vendor Manager works closely with the Quality Manager, checking quality trends, designing ways to improve the quality, and training linguistic resources in better practices.
The Language Coordinators coordinate LQA teams, liaise with linguists, and arbitrate disputes.
The Project Managers support Language Coordinators and the Quality Manager, informing them of any changes in the project and sharing project resources with them.
The Reviewers perform LQA testing, which will be a snapshot of the translation quality of the project. They need to be disciplined and fair as their role is observational in terms of scoring the quality of the piece, but also educational, in terms of transmitting instructions and suggestions to translators when necessary.
Translators need to understand how LQA works and be ready to learn from the issues reported by the Reviewers to deliver better quality in the future. It's also important to note that smaller teams might not have a Language Coordinator and thus that role might be performed by the Project Managers.
Enhanced quality standards ensure a win-win situation for all sides involved in game localization projects. Seamless game experiences create satisfied players who spread positive feedback that boosts everyone’s reputation. Happy clients offer increased volumes of work, higher revenue, and form long-lasting business relationships. A satisfied localization team works in an environment where each person knows their responsibility and their place in the workflow, and thus teams grow stronger as members become more experienced and confident.