» Air Conditioner
» Translation
» Downloads
» Book Publishing
» Printing
» Health
» Cell Phone
» Weight Loss
Japanese Translation
The Complications of Japanese Localization
When Wild 9 was completed by Shiny in the fall of 1998, we worked on a few different skus of the title and then quickly moved the team onto other projects which were in need of help. Wild 9 was initially released in English, French, Italian, Spanish, and German including a low-violence version almost simultaneously, and by December of 1998 I figured that Wild 9 development was completely finished. At that time I was reminded that we still were scheduled to work on a Japanese version of Wild 9 to be released in Japan first-party by Sony. After already localizing Wild 9 into four other languages it seemed to be a slam dunk to localize into one more, but that assumption could not have been farther from the truth. With the exception of a low-violence version of a game, Japanese has proven to be the most demanding localization that I have experienced and certainly one which we were not adequately prepared for.
At the end on 1998 much of the original development team was no longer available to work on the title due to reallocation to other teams and people inevitably leaving to pursue other opportunities. To further complicate matters, our project servers, which were almost literally being held together by duct-tape, were failing fast. We assumed that even though most of the original team was unavailable that the former lead programmer and myself would easily accomplish the localization in about a month to a month and a half at most. The lead programmer by this time had left the company to pursue other opportunities but graciously agreed to help us with the localization on a contract basis for a month or so on weekends.
By the time the title was eventually approved by Sony Japan, we had worked on the localization with a skeleton team on weekends for not only the month we had originally allocated, but also an additional 11 months which we had not planned on. During the development of the Japanese version of Wild 9 which later was titled Wildroid 9 to appeal more to Japanese gamers, we not only worked with the original lead programmer but also three other programmers and a number of artists to finish the demanding localization.
Why were we so far off in our original assessment of the scope of the localization? Hopefully the following with highlight some of the pitfalls which should have been anticipated from the beginning.
The Basics – What To Keep In Mind When Localizing For Japan
When preparing to localize a game for Japan, there are really four basic things to keep in mind.
- Difficulty Level will have to be adjusted which is usually too high in American titles
which are being localized for Japan.
- Two-byte character support will have to be enabled to display Japanese characters on
screen.
* Lots of graphics will have to be modified to better suit the Japanese market.
* Sound assets will have to be modified to better suit the Japanese market.
The extent of how much the above will affect your title depends on what type of game you are localizing, but one should keep them all in mind as the above are generally applicable to all games.
One of the most important things to do early on before tackling any of the above is to find a Japanese contact that understands what Japanese gamers prefer in the way of character design, sound design, and difficulty. Without a solid understanding of what makes a localized game appealing to a Japanese gamer, it is impossible to achieve a successful translation.
Getting Started on the Specifics - Localization of On-Screen Text, Technical Hurdles
Since Wild 9 was being taken to the Japanese market as a Sony first-party title we were fortunate in that we at Shiny did not have to deal with the actual localization of the text itself. In fact, the localization of the text in the game was as simple as sending all of our text strings and printed manuals to Sony of Japan who promptly returned all of our assets translated into Japanese. If you are not fortunate enough to work with a large company like Sony to develop a first-party title, translation can be done either by your publisher or outsourced to a company who specializes in translation from English to Japanese.
So we were off the hook with the actual translation of the on-screen text, but we still had to work out how to display the Japanese characters on screen.
There are two very good ways to enable Japanese characters in your title, Double Byte Support (also referred to as DBCS) and Unicode. The method you choose to localize your title depends on what stage of development your game is in currently at when you think of localization.
The inherent problem with American titles is that all of the on-screen text is displayed 8-bit which allows for up to 256 different characters to be displayed on screen. Many Asian languages, Japanese included, utilize more than 256 characters making the 8-bit solution inappropriate. Single byte character sets like ANSI can’t handle more than 256 characters thus making the conversion to Double Byte support necessary. Double Byte support allows a mapping of the different characters to different values in which each value is 2 bytes wide. The first byte of the character is called with Lead Byte while the second byte is called the Trail Byte. To display a Double Byte character it is necessary to display both the lead byte as well as the trail byte. While not as powerful as Unicode, double byte is much easier to implement and will handle a conversion from English to Japanese quite easily. If your title has already been localized in Europe or if the localization to Japan is happening late in your title’s development then double byte support is the correct choice.
If your title is in an early stage of development and you are planning to someday localize for both Europe and Japan, then Unicode is a possibility which may make life easier later in the project if planned on early. Microsoft defines Unicode as a worldwide character-encoding standard that uses 16-bit character values to represent all character used in modern computing. Unicode is an international character set that supports over 650 languages including Japanese, Chinese, French, and German. As stated, Unicode uses a 16-bit character set as opposed to the 8-bit character set commonly used in the U.S. This allows your title to display up to 65,000 characters, 200 times more than the capacity allowed by an 8-bit set. The Unicode standard assigns each character a unique 16-bit value and does not use complex modes or escape codes to specify modified characters or special cases.
There are two different ways in which Unicode can be encoded, UTF-16 and UTF-8. With UTF-16, all characters such as ASCII characters, Latin characters, and Asian ideographs are encoded as two-byte, 16-bit characters. It’s fixed width means that applications do not need to determine the boundaries of the data and is therefore best suited for processing character data in an application. The second type of Unicode encoding is UTF-8. UTF-8 is a 1-to-3 byte, variable width encoding. ASCII characters such as A-Z, a-z, and 0-9 are encoded as single-byte characters. Accented European characters are encoded as two-bytes and Asian ideographs are encoded as three-byte characters. Though this is a highly compact storage format this can be fairly dangerous to implement as one day you’ll find your memory footprint is three times larger than it was when you were developing your NTSC sku when the three-byte characters are enabled for your Japanese version.
UTF-16 and DBCS may seem somewhat similar but Unicode is a single, large character set which displays different languages from around the world while the Japanese DBCS character set only supports Japanese and English.
Unicode seems to be a magic solution, so why isn’t everyone using it? Unicode is difficult to implement if not planned for in the beginning of development and oftentimes would require a complete rewrite of most of your code if implementation were attempted late in the project. While Unicode is quite powerful, it is not a logical choice for implementation late in a project or if budget is a concern.
Localization of On-Screen Text, Practical Considerations
Inevitably the question will arise during your localization effort, “Do we have to localize all of the text?” Ideally, all of the text in the game would be localized but in the case of sections like the presentation system which is often created in Photoshop of Max rather than via text strings, this may not always be a practical solution.
In the case of Wildroid 9 we ended up not localizing all of the text at the suggestion of Sony. All of the in game dialog strings and on-screen help were localized but when it came to the menus we elected to go with a combination of Japanese and English text or English text only.
In regard to in game text, Japanese gamers are used to the basic English phrases such as “Press Start Button, Quit, Save, Load, and Options.” This being the case it is often not necessary to localize these common phrases but it is much more important to localize and complex critical messages such as ones which explain the status of memory cards, why a save is unsuccessful, and the like.
When not easily recognized, localized Japanese games primarily use Hiragana which is used to display Japanese words and Katakana which is used to display foreign words.
Sony Japan made our lives pretty easy as they ended up delivering a Japanese character set to us and then referring to the set with numbers which could be cut and pasted into the code. After calling the correct characters on screen we simply worked with Sony in Japan to break our lines in the correct places on screen which were inevitably longer in Japanese than they were in English.
Thus there is no need to translate all of your text to have a successful localization but of course the more the better to really do a first class job. I was told once but the head of European development at Interplay that given a really great game which has been partially localized and one which does not contain the same level of gameplay but has been fully localized, European gamers tend to purchase more copies of the title with complete localization over that with better gameplay. This point simply highlights the obvious, if the resources are available localize as much as you can.
Localization of Art
The conversion of art in a localized game for Japan is problematic as there is no magic formula for the overall style the art should be translated to and for the fact that so much of the art assets will have to be revised to better suit the Japanese market.
Perhaps one of the most difficult areas of Wild 9 to localize was the characters themselves. We knew that some changes would have to be made to the overall appearance of the characters but even then we were unprepared for the fact that most of the characters would have to be completely redesigned in the more Japanese market friendly anime style. Once again we were fortunate in that we were working directly with Sony Japan who were able to help us a great deal with the overall style of the characters as well as by having our designer Tom Tanaka on staff who already had a solid sense of anime character design.
Here is an image displaying the final version of Wex from the NTSC version of Wild 9, and Vex, as he was later renamed for easier pronunciation in Japanese, in his final sketch for the Japanese version of Wildroid 9.
Again, though the visual differences between the characters are obvious, it is difficult to come up with a standard style guide as to why Vex is more appealing to Japanese gamers than Wex. The Japanese gamer’s attraction to the anime style character is as much a cultural and historically conditioned affinity than anything else. When you compare the two characters you will notice that by nature the anime style Vex does not have the kind or darkness and hardness that the Wex character has on the left. Note that many times the American version of the heroic character contains this same hard, teeth-gritting quality as our final version of Wex does. As Wild 9 inherently has a violent game mechanic stressing the use of Wex’s weapon to torture his enemies, Sony’s Japanese producer Katsuhiko pushed hard for the softer anime style versions of our characters to soften the gameplay a bit. In the end Wex’s transformation into Vex not only took place in our presentation system and paper components but also in the game itself which required us to go back and retouch our in game models.
Localization of the Sound Effects
It goes without saying that the voice over samples were rerecorded by Sony Japan for the localization into Japanese but what about other sound effects? In the case of Wildroid 9 we not only changed the VO but also many of the incidental sound effects like the enemy grinders effects and the character screams. Again, as with the graphics, it was Sony’s recommendation with Wildroid 9 to tone down the overall violence of the samples in an attempt to soften the gameplay mechanic of using Wex’s weapon the torture his enemies. By making the screams a bit more comical and brief rather than blood curdling we were able to create a different atmosphere for the game worlds. It is also common for American games to have an overacted feel to the sound samples than those found in Japanese games. Localization of sound effects seemed to us to be a foregone conclusion but as with other aspects of the localization it was localized in much greater detail than we would have expected in our initial schedule predictions.
Expect the Unexpected
With the conversion of Wild 9 we were also constantly presented with obstacles to the localization which were unexpected around every corner, some of which should have been obvious as well as other which were not.
As Wildroid 9 is a being released as a first-party title in Japan by Sony we were subject to much more stringent submission and approval standards than we were in the U.S. and Europe with the NTSC and PAL versions of the title. The first oversight on our part was the differences between the Sony Japan Technical Requirements Checklist (TRC) and that of the one here in the U.S. and Europe. Make sure that when the localization is being scheduled that time is allotted for items like button reconfiguration, differing memory card management requirements, and presentation system formats. While changing the action button in Wild 9 from “X” here in the U.S. to “O” in the Japanese version is a small detail in itself, the amount of these little changes in the Japanese TRC took a bit more time than we expected.
When localizing the game the last thing we would have expected is months of back and forth with Sony’s internal QA department. After all, if the NTSC and PAL versions of the title passed QA already and the game was already in gamers hands in the U.S. and Europe, what could possibly be the hold up with the game in Japan? Lots of things we came to find out the hard way. As the title was to be released by Sony first-party it was scrutinized much more than the previous versions had. If it wasn’t the memory card system failing constantly for an impossibly long list of little problems than it was logic bombs or bugs in the game that were achieved by Sony’s internal testing teams performing detailed actions in the game which would produce bugs previously undiscovered. By the time the test reports came rolling in we no longer had the staff or the tools in place to deal with many of the complex issues found within the levels themselves making the fixes extremely difficult. The moral of the story is simply to allow a realistic amount of time and resources for QA even if the title has already been released in either the United States or Europe.
Another issue to be aware of which we encountered were the subtle cultural differences between the U.S. and Japanese gaming community.
Just for some fun and useful examples to illustrate some cultural behavior differences between the Japanese and Americans for the next time you are talking to a Japanese developer in the next few days, try these on:
- Though we take a casual stance here when standing and talking to a peer, the
Japanese prefer more of an “at attention” stance and will find it uncomfortable if you stand and talk to them with your hands in you pockets or with you arms crossed in front of you rather than standing and talking with your arms at your sides or with your hands clasped in front of you. If you cross your arms in front of you while talking it conveys disapproval or sternness toward your Japanese peer.
- If you have a contract or other important document which you are giving to a Japanese
superior or elder it is considered impolite to pass the item with one hand rather than passing it with both hands.
- If you receive a business card from a Japanese peer, you can not only cause serious
offense but also potentially ruin a business deal if you quickly put it into your pocket. Always take a moment to read the card before putting it away.
- Some of you, like myself, may carry their business cards in their rear pants pocket…this
is considered to be improper to many Japanese businessmen. Never hand them your card out of your back pocket.
The relevance of all this is just to be careful and keep an eye open for possible cultural faux pas when localizing your game for Japan. If you have a consultant in Japan as we did with Wildroid 9, all the better.
In Wild 9 all of the characters were designed with a very cartoony style which meant that many of the characters in the game and even some of the directional signs in the game were drawn with only four fingers on each hand. Seems like it wouldn’t be a problem, right? Actually, for cultural and historical reasons the representation of characters with less than five fingers is a major problem with a Japanese game. Characters with less than five fingers are representative of organized crime in Japan and must be avoided at all costs. This small detail with our character design was in the end extremely problematic as levels had to be re-rendered to add new directional signs and may characters including our end boss had to have their models revised to include the additional digit.
A Brief Japanese Market Analysis
Though a full market analysis could take up an entire hour lecture in itself, I must at least mention that even with a full localization for an American or European title headed to Japan one must set realistic sales expectations. With Wild 9 we have spent almost a year of evenings and weekends and plenty of cash with external contractors getting the title wrapped up for release in Japan. When all is said and done however, the average localized American title will sell only five thousand units in the mature PlayStation market where one will encounter heavy competition from new, cutting-edge Japanese titles. Unless a deal is struck which is profitable for the developer, the royalties alone might not make a full conversion realistic.
There are of course exceptions to the rule. Crash Bandicoot: Warped sold over one million units in Japan before discount and other titles like the Tomb Raider series sold well due to the fact that the overall premise of the game is like a Hollywood adventure movie which is popular in Japan, but created abroad.
There are also games in which localization will not be an issue. Typically sports titles based on the NFL, MLB, NHL, and X-Games are the type of titles that Japanese gamers would be interested in since the U.S. is the home of the sport itself. Formula One is another example of a title which originates in Europe yet is popular in Japan even with minimal localization. In the end however, the actual commercial success of course is dependent on how popular the actual event is in Japan, thus NFL, MLB, and NHL will not sell through well in Japan even though they only need minimal localization.
Games which are heavily dependent on their characters or the environment in their design are very difficult to localize. Any titles which belong in the action/adventure genre are likewise unlikely to achieve critical and commercial success in Japan.
Conclusion
In the end I believe we did a good job localizing Wild 9 for Japan even though the timeframe for such localization was much greater than our initial expectations. The title has just been released in Japan and thus time will tell if the concept of the game will catch on with Japanese gamers.
Though the localization of Wild 9 I have learned some important lessons about what qualities a successful title in Japan must possess. To reiterate some points about successful titles in Japan which have been localized from the U.S. and Europe.
- Character based games must be based on such a character that will be appealing to a
Japanese gamer. The Crash series is a good example.
- Games with themes or concepts which are originally from the U.S. or Europe which are
popular in Japan will generally sell well. Formula One or Extreme Games are good examples.
- Games which have a new and innovative concept will do well such as the Tomb Raider
series or Destruction Derby when it was first released.
- Games which do not necessarily have a new concept but have one which Japanese
developers would not think to develop like Themepark.
These types of games when localized properly stand a good chance of success and make your efforts worthwhile.
by Stuart Roch : stuart_roch@shiny.com