The translator, rewriter and editor have a few things to think about with regard to the sound effects before sending the script on to the letterer. Much of it depends on how the sound effects are being treated by the publisher, but with all formats, there are things to consider.
No SFX translation for the sound effects:
Even when there are is supposedly no translation or touch up for sound effects, there are still considerations. Some sound effects (and outside-the-bubble asides) can be considered dialog, others have linguistic properties, and others have a meaning that pertain directly to the flow of the story. I have never worked for a company that doesn't in some way translate the sound effects, so I can't advise on this case, but some consideration should be given to sound effects that move the story along. The click of a door when someone enters the room for example. I've seen that click appear on a white background with no other image or indication of what happened in the story. If the reader can't read the click, then the reader won't know how the new character suddenly appeared.
Glossary-style sound effects translation.
Time to own up to this. The glossary-style sound effects at Viz were partially (if not mostly) my fault. It was in response to certain pressures put on during the changeover from flipped to unflipped and from $16 graphic novels to $10 graphic novels. I figured a glossary was a way to both allow for not retouching the sound effects while still giving the reader the information. And once the contracts were signed, there was no going back -- all stories that had started that way had to be finished that way. I had thought that it would be used like Del Rey's translators notes are used today. One would read through it, and if there was something confusing, one could always flip to the back to figure out that particular problem, then go back reading and ignoring the sound effects. It was a way for the reader to get what they were getting anyway from companies like TokyoPop and CMX, and still have the translation if they wanted it. (It was an awful pain for the editors -- I know since I edited quite a few myself.) But many readers resented it. They felt they had to flip back and forth to the glossary. And they would rather not have the option of a glossary at all. As far as I can see, all of Viz's recent acquisitions have been full touch-up, so aside from grandfathered, contractually-bound series, the technique has been put to rest. It was an experiment that failed. Sorry to those who hated it, and to those few readers who liked it, I'm sorry that it didn't do as well as I hoped.
As a pretty-much dead style, I won't comment except to say that it was a situation where one could be as free with the effects' sounds as the translator wanted to be. But it was, as I noted above, an awful lot of extra work.
Subtitle-style sound effects:
To tell you the truth, I'm still getting used to this. I've been doing them for more than two years now, and I still find myself doing the sound effects as if they were full touch-up effects. But I have converted in a few areas. I've found that for subtitle-style SFX, one can use real words easier than one can for full touch-up. Words like "smile" and "chatter" work better in subtitle-style than they do in full touch-up SFX.
The translator for subtitle-style has a lot of leeway as to how to translate a sound since the subtitle can be placed on one side of a broken sound effect (a broken sound effect is an effect that has image information between letters of the effect. For example, on a dramatic entrance, the effect may start on the left of the entering character with a "do" sound and finish on the right of the character with the "n" sound), so the translator doesn't have to be creative on how to break up the single-syllable SFX. (For the above example, a sound effect of DOOM can be subtitled under the left-hand SFX portion with nothing by the right-hand side.)
But a possible difficulty in subtitle-style translation there is always the possibility that a bilingual is reading your translation, so it's best to keep the English sounds closer to the Japanese sounds than one would need to in a full touch-up situation.
Full Touch-Up effects:
One of the biggest concerns for translating in a full-touch-up-SFX situation is the background. You know that some poor letterer is going to have to take out the Japanese SFX and replace them with what you write. If the original Japanese sound effect is black letters on a white background, then the letterer will find it easy to white out the original effect and put in nearly any combination of letters in its place. Similarly white effects on a black background is easy. Effects on line drawings are more difficult for the letterer, but still doable. But effects on complicated images or (shudder) screen tone are the worst for a letterer. The translator should look at the background and make allowances for what is there.
If the background is screentone or complicated images, the translator (or rewriter and editor) should come up with a sound effect that will take up approximately the same space as the original effect. Also the English SFX should be letters that do a good job of covering the area. If it's on a screen tone, don't use dashes for example. An effect like ZLI---IP leaves an awful lot of background that needs to be touched up. If ZLOOP were used instead, then the letterer can get away with covering up most of the original sound effect with the letters, leaving only the edges to touch up.
If it's a broken sound effect (see the paragraph above), then one should match the broken effect. For a dramatic entrance example above, one would want the left hand portion to say DOO, and the right-hand side to finish it up with OM or OOM.
The other thing to be careful of is that unlike subtitle sound effects, real-world words don't look as good. It's usually best to add an extra letter as in HUGG for an embrace; change to other similar-sounding consonants such as CLICK being changed to KLIK; or remove vowels such as SLIP being changed to SLP, or SLLLP. (Of course, it's always better to be creative, but...)
In touch-up effects, many multiple letters help the letterer when he/she is working on an effect. For example, if one has an effect like DONNNNNNG, the letterer has the option to remove "N"s or add them to make the English SFX fit the space needed.
There are other considerations that I'm not thinking of right now, but I'll see if I can cover those in a future installment.