The making of the English translation of “Eighteen O’Clock Music Bath”

(Update: I have since republished “Eighteen O’Clock Music Bath” in a larger compilation that you can find here.)

Recently I published my second eBook (Volume 2 of the “Science: Hopes & Fears” series) which contains my translation of the classic dystopian tale “Eighteen O’Clock Music Bath” (original title: 「十八時の音楽浴」）by Juza Unno (海野十三), who is acknowledged as a founding figure of Japanese science fiction.

Just like I did for the my previous book (Volume 1 of the same series), I thought I would write up some details on the process of making and publishing this book for those interested in translation or self publishing.

While I had a good deal of trouble with publishing, formatting, and other unexpected things for Vol. 1, I leveraged the knowledge I gained and was able to publish Vol. 2 in less than a month and a half. Volume 1 took closer to three months, and it had a little higher total word count, that was not enough to explain the 2x increase in speed for Volume 2.

While it took me a lot of time to do the story selection for Volume 1, I had pretty much already decided I wanted to (eventually) translate “Eighteen O’Clock Music Bath” even before I started Volume 1 itself. There was a host of reasons that made it an easy choice for Volume 2: experience with translating Juza Unno’s works, permission from his surviving relatives to translate and publish his works, as well as the fact  that「十八時の音楽浴」 was one of this author’s most famous works, not to mention that it was a dystopian story (which happens to be one of my favorite genres). Although some day I would like to translate Juza Unno’s longer works, the fact is that just doing this as a hobby means it would quite a long time to for those. Fortunately, “Eighteen O’Clock Music Bath” was just long enough to be published on its own, but not too long to make it a unreasonably long project for me.

As before, I started off working in Google Docs, first preparing the initial rough draft translation. While there was a few tricky spots in the stories of Volume 1, “Eighteen O’Clock Music Bath” was even more difficult, with many areas that were tricky to understand, let alone translate accurately. On my first run through I bolded all of the problem areas, then came back to those areas later and did more research as needed. I leveraged the help of several native Japanese speakers, and Yeti from Shosetsu Ninja also gave valuable feedback about some of the tougher spots. While a few places had text that was difficult because of the science-related terms used, most of the difficulty stemmed from the fact the work was from 1937, and hence there were a bunch of expressions and words that are no longer used in modern Japanese. So I had to often cross check several dictionaries and web searches, in addition to asking native speakers.

Whereas Volume 1 was made of five stories, this time it was one fairly long story (technically ‘novella’ length) which made it harder to work in parallel with separate files. To try and keep things efficient, I ended up gradually transitioning the chapters one by one to a stage marking “translation check finished” by removing the original Japanese text. Sherayuki, my other proofreader on this project, would then look at these while Yeti did some more checks and I continued refining text across several iterations. It was fun to work with three people at once on one project, and there was even a few discussions where all of us chimed in.

The longer length contributed to another challenge: my typical workflow where I would go over the entire story once in the evening (in a 1-3 hour window of time) was no longer feasible. For some of the iterations I would do a few chapters one day, then a few the next, and once I got to the end start over. But I began to feel that this somehow added a bias to the editing process. So for some of the middle iterations I actually generated a random number and focused on that one chapter in a sitting. The fact that the later chapters were much longer than the earlier ones also made things a little tricky. While I did several iterations over all the chapters, I spent significantly more time on the first and last chapter (especially the few first and last paragraphs) because I really wanted the prose there to be as perfect as possible. Also, while I consider Juza Unno’s style to be more functional than flowery and overly-descriptive, I felt he put extra effort into these parts to make them stand out in the original text.

For the translation process, I tried a little more to give an “aged feel” to both the dialog and the rest of the text. This was tricky since, to be honest, I haven’t read too much literature from the 1930s, but I skimmed some SF works from that era, as well as read Vladimir Nabokov’s Invitation to a Beheading (which I reviewed here) as another source of period English. (In retrospect, since it was a translation, it may not have been the best choice for reference). While I don’t claim the result is historically accurate, at minimum I tried to avoid use of modern-sounding expressions.

I spent a long time thinking about the proper translation of the title. While the story has been referred to in English as “The Music Bath at 1800 Hours”, personally I didn’t like that rendering (partially because of the “the” and “at” that seemed wordy), despite it being a pretty close literal translation. I wanted the “18 o’clock” part to be in the front for emphasis, but generally numbers are written out in the beginning of titles so I ended up choosing “Eighteen O’Clock Music Bath”. One of the other reasons I choose this title is the similarity to the famous dystopian novel 1984 (which was written roughly 12 years after this work), which is typically written out as “Nighteen Eighty-Four”.

The character’s names were also a little tricky, for example one of the main characters is named ミルキ, which can be written in romaji as “Miruki” or “Miluki”, but also could be expressed as “Mirki” or “Milki”. I knew that this work had some Western inspiration, and this name was clearly not a typical Japanese name (by the way, the letter “ru” is very rarely used in Japanese words except for loanwords from other languages, the end of some verbs, and a few other words). I ended up choosing “Miluki”, partially because I didn’t like the ring of “Miruki” and “Milki” sounded too much like “milk”. Another character’s name was コハク (“Kohaku”) which I rendered as “Kohak” since it seemed to fit a professor better. (Note that there is no way to directly express a constant at the end of a word in Japanese, except for “n”).

After the text was to a good level of quality, I converted to a Microsoft Word file and then imported into Draft2Digital’s tool to get a .MOBI file, using the same visual template as Volume 1 (I especially liked the drop caps, and I wanted to keep the overall look feel the same throughout all volumes). From there, I did the final editing and tweaking directly on the MOBI file using Sigil before submitting to Amazon. As before, I used Grammarly to catch a small number of issues that got missed in editing, and Amazon’s automatic spell checker found one or two issues as well (along with a bunch of false positives).

One thing that saved time on the formatting was that I had no translator’s notes to worry about, whose formatting proved to be troublesome in Volume 1. While there was a few places that I probably could have added a note for extra clarity, I realized that none of them were necessary so I omitted the translator’s note section altogether.

I decided to try out a new tool for the cover design: the open-source vector graphics editor Inkscape. Except for a few quirks, I found it pretty easy to use and feature-rich. While I think the cover may not be up to par of something a professional cover artist would do, given my limited experience in this area I think it came out pretty good.

If you got something out of this article (or enjoy my blog in general), please consider buying “Eighteen O’Clock Music Bath” as a token of your appreciation. (Amazon link)