Project “Sexy Dynamite”

Okay. So this piece is crazy. Let's get that out of the way first.

When Edward Keyes first asked me to find a suitable project for him, I had no idea that this was how it was going to end. My record, in that regard, is pretty suspect: in 2005 I made him translate this odd fairy tale about a boy, his bird, and golden fruit; then in 2006 I steered him smack-dab into this even odder non-fairy tale about dead girls and cantakerous fish; and now in 2008 we've got the oddest three-minute-long novel game we've ever seen.

So, uh, yeah, I enjoy torturing him very much.

And, uh, yeah, the Virgin Mary Osaka Bancho is watching.

The script of this piece is simple and monosyllabic — as you can see for yourself, a good deal of it is spent in incoherent screaming, wet thudding sounds, and occasional large carnivorous reptiles being sent into the stratosphere, in roughly that order. The entire piece is otherwise written at about a fifth or sixth grade comprehension level, and as such the resultant translation was nice, competent, and done in a very small amount of time. The rest of this peer review will therefore focus on the technical aspect of the project. My reason for doing this is twofold — first, every other peer review presented here focuses on the translational aspect and neglects the technical aspect of projects; and second, there were many technical issues that surfaced during the localization of this piece that may be of interest to translation teams.

When Edward first brought me a complete build for peer review, the translation was complete, but many fit-and-finish issues remained:

  • Not only did the piece require Japanese as the default ACP, it also required unit formatting to be set to Japanese as well.
  • Graphical onomatopoeia was still all Japanese.
  • There were multiple margin overruns and button formatting issues.
  • Font kerning and spacing were severely deranged in the Iteko monologue.
  • Not enough cowbell.

Some of these issues were easily resolved; for instance, the margin overruns were simply avoided by manually linebreaking a little more intelligently. However, there were multiple other issues that were quite a bit more perplexing: particularly, why was font display so messed up in the Iteko monologue, and what were we going to do about the graphical onomatopoeia? The first was interesting because we did not know what the mechanism of the bug was until later on; the second was interesting because raised an interesting localization issue that translation teams — professional and amateur — often find themselves wrestling with.

Edward discusses the specifics of the Japanese ACP and the Iteko monologue issues far better than I ever could in his notes, so I'll leave those topics well alone. However, I'd like to take a moment to talk about the graphical onomatopoeia issue. As you already know, onomatopoeia in any language involves a mishmash of syllables that, when put together, hold no intrinsic meaning. Their sole purpose is to imitate sounds; as such, they are generally either single clipped syllables (bang, boom, crash, et cetera) or are multisyllabic but with multiple repeated letters (kablooey comes to mind). In Japanese it is no different: many common onomatopoeitic sounds, such as ドーン and ガーン, are single clipped syllables, and many of them share phonetic elements. With the character set {ド, ガ, ー, ン} you can construct both ドーン and ガーン, because the ーン is preserved in both words. Porn took just that approach to visual onomatopoeia — he more or less created multiple graphic files, each of which had a single syllabic glyph on it, and then mixed and matched them in-engine to arrive at the desired sound effect.

So what is the translator to do with this? Clearly, we can't simply force a 1-to-1 conversion; onomatopoeitic words don't tend to cross cultures very well. We can't try to finagle combinations by spreading letters across graphic files the way that Porn did, because the graphic files that Porn is displaying at any given time don't match up with the graphic files we might want. On the other hand, if you made each graphic file whole onomatopoeitic word by itself, now we have three words where there used to be one and we haven't made our situation any better vis-a-vis the combinations issue.

The solution Edward came up with — and I think it works very well — was to figure out exactly which onomatopoeitic sounds were being triggered when, and then to match appropriately neutral English onomatopoeitic words to match. The usage of this neutrality was decisive, as it allowed us to take a simple solution — making each graphic file its own onomatopoeitic word — and use it without having to fall back to anything fancy. Edward then hand-drew his own replacement graphics, keeping with the MSPAINT FOREVER spirit of the original. This problem solving process strikes me as particularly elegant, and certainly helps explain why I find him one of the best technical people in the community.

One other interesting issue. The "rage gauge" in the game allows for four 1-byte characters, and that's it. No possibility of expansion of the space, and in any case the words being displayed in that area would take up way too much room when translated into English anyhow:

  • 熱 translates to "hot" (rage levels 0-9)
  • 高熱 translates to "really hot" (rage levels 10-19)
  • 爆熱 translates to "bursting hot" (rage levels 20-29)
  • 極熱 translates to "burning hot" (rage levels 30-74)
  • 紅蓮 translates to "crimson" (rage levels 75-149)
  • 自熱 translates to "white hot" (rage levels 150-254)
  • 灼熱 translates to "red (as in surface-of-the-sun) hot" (rage level 255)

Clearly, there was no way this was all going to fit. And also clearly, there was no way we were going to be able to hack the game given the urgent time frame to make more spaces possible. We therefore decided to restrict ourselves to replacements that were only four characters long or less:

  • 熱 → RAGE
  • 高熱 → PAIN
  • 爆熱 → HURT
  • 極熱 → MAIM
  • 紅蓮 → KILL
  • 自熱 → DIE!
  • 灼熱 → RED!

While not at all a direct 1-to-1 translation, we felt this certainly preserved the spirit of the scale. We conferred with Porn and explained the rationale for the change, and he not only approved, he indicated that he liked this as much as, if not better (!) than, the original. And that was how the rest of the project went: hand-in-hand with the original creator.

All in all, this was a refreshing project to supervise and peer-review. Edward and I certainly learned a lot: we reaffirmed the lessons we'd learned from True Remembrance, particularly in regard to communicating with the original creator; we learned to even better bounce ideas and tactics off each other before we proceeded; and we learned to split the earth ... um ... yeah.

Best I stop here, eh?

28 November 2008
Seung Park
Festival Coordinator, al|together 2008