It certainly doesn't seem like a difficult task from a conceptual standpoint -- even just running a block of text through secondhand solutions such as BabelFish can produce something that looks somewhat translated. But as Vance explained, simply changing the text from one language to another is an issue, and linguistics only scratches the barest element of what's needed for adapting a game. "Localization should be as much a science as an art," as he put it during the panel. While the focus remained on localizing Guild Wars from a business standpoint, players will still find interesting facts within the discussion.
As Vance explains, GW was targeted for 10 different languages within the same client. This tied into the simple business costs of adapting a game to different languages, a cost that rises as you factor in customer service, voice work, text, QA for patches, and so forth. There's a certain base number of sales required for even starting localization, simply because there comes a point at which the cost of translation exceeds the expected profits.
Of course, GW has always been world-ranging not just in terms of built-in languages. Each of the major campaigns has had a very specific regional feel, with Prophecies being very heavily based on a northern European feel, Factions centered primarily around an East Asian setting, and Nightfall taking place in an African and Middle Eastern environment. Rather than trying to segregate the expansions via real-world milieu, however, the devs devised each campaign around a set of values that would be resonant for a given setting -- Factions was a storyline that felt very Asian and thus was set in an appropriate environment.
Nightfall also presented a hurdle because of its setting -- because of a great deal of cultural baggage in certain regions of the world based on the dark-skinned cast dominating the setting. There were also a great deal of tensions between European nations and northern African nations at the time of the game's release, further exacerbating the issue. The writers were under a strict guideline not to portray any of the dark-skinned NPCs of the storyline as being villainous or evil, to avoid buying into a stereotype or causing longer-term issues for the game's international community.
But the localization issues weren't limited to the cultural issues. Eye of the North was originally slated to have a viral promotional campaign centering around Gwen, a campaign dubbed "Gwen is Missing." Unfortunately, the marketing team pulled the plug due to an unusually high number of missing children in the U.S. and the UK when the campaign was originally going to be running.
Cultural issues informed a great deal of the localization troubles the game faced early on, even disregarding the issues that the Chinese government had with certain imagery surrounding the Necromancer. (It's a well-known fact that images of death and decay are frowned on by the Chinese authorities, which is a bit of an issue for a class solely focused upon exacerbating rot and animating corpses.) Language servers, for instance, were a fix implemented after the regional server system was implemented in Europe. While all European players were able to play, there were particular frictions between German and French players, which prompted ArenaNet to move to limiting servers based on language within the region.
The cultural switches also led to the creation of the party search feature in the game, as Korean players were disappointed by the game's options for forming a party. Unlike U.S. players, who tend to use chat and abbreviations to search, Korean players prefer built-in UI tools, so the development team needed to add in a feature to help keep the Korean community engaged rather than isolated.
As Vance stressed, knowing that the game would be broadly localized led to a number of procedures from the start of development that helped keep the process simple. For example, rather than binding hotkeys to values, the devs bound the game's commands to physical keys, avoiding a keybinding that makes no sense across cultural lines (using "Y" for "yes" makes sense for English, but no sense for German). There were also a set of unified and consistent terms used by the localization teams so that characters, skills, items, and so forth could all be identified consistently, rather than muddying the waters with inconsistent translations or references.
Vance closed by discussing the difficulty in handling the voice work across multiple languages. Simply translating and recording different lines isn't always sufficient, assuming that the cinematics and storyline will make sense across multiple cultures in the first place. (Vance cites Prince Rurik and King Adelbern as a dynamic that made perfect sense in America -- but in Asian cultures, Rurik is a disobedient young man who should rightly be punished for failing to respect his father's wisdom.)
Especially for cinematics, many of the recorded lines of dialogue had to match up with pre-existing screen changes, as the scene would shift after a certain amount of time for more cinematic angles or due to something happening on screen. In addition, some languages required special consideration. The line "You have become a great warrior" is the same whether the subject is male or female in English but changes based on gender in French -- and in order to support the localization, both lines had to be the same length and had to be referenced dynamically based on language settings and the situation.
While Vance's discussion didn't lead to important content reveals, it discussed an important and yet oft-overlooked aspect of development. Despite all of the issues it presented, Vance concluded his discussion by saying that the GW story proves supporting a broader family of languages can be profitable and successful in the long term.