On Conventions

I mentioned before that I wanted to write one of these posts about translation, and some people (i.e. two) expressed interest. So here goes:

Walls of text ahead, so I figured I'd give y'all something nice to look at first. Picture unrelated.
Source: http://www.pixiv.net/member.php?id=922199

Today I’d like to talk about conventions. No, not the kind that you spend sweating bullets in your heavy cosplay and eating nasty overpriced hot dogs at. I’m talking about translation conventions.

The way I use the word, a translational convention is a consistent choice that the translator makes that doesn’t affect the essential accuracy of a translation. This is a bit tricky because there is no fine line between what does and what does not affect accuracy. Rather, it’s very subjective, and depends on how you define “accuracy”. If a translation does not capture a character’s emotion, is it inaccurate? Does dropping honorifics affect the accuracy of a translation, or is it just a convention?

For the sake of this essay I’m going to define the difference between “essential accuracy” and "convention" like this:

1. If two translators both understand the sentence and agree on what it means in its original language, any differences in their translations are a matter of convention.

2. If two translators disagree on the meaning of the sentence in its original language, or one or both of them misunderstood the original sentence, then differences in their translations are a matter of accuracy.

By this definition, the choice to keep, translate, or ignore honorifics is a matter of convention. I doubt any translators of Japanese do not know what the honorifics mean, but different translators make different choices concerning how to deal with them. That is a convention. Choosing to consistently translate “masaka…” as “I can’t believe…” instead of “Don’t tell me that…” is a convention. Choosing to translate “otsukare-san” as “Mr. Otsukare” instead of “good work” is inaccuracy.

Now you may ask, “If conventions don’t affect the accuracy of a translation, what’s the big deal about them?” (sometimes even I ask myself this when I see people complaining about translations). Well, based on some of the examples I’ve provided, this hardly needs answering. I’m sure you’ve all seen someone get worked up over honorifics in translation. People clearly care about certain conventions that you choose.

Furthermore, consistency in translation is vital. One of the most common conventions that translators (especially fan translators) have to deal with is how you Romanize names. For Japanese names there are rules in place, but what about made-up names in fantasy or sci-fi genres? The specific choice you make isn’t terribly important so long as it makes enough sense; the most important part is sticking to it.

I own ADV’s release of Kaleido Star, and one thing that stuck out to me was the translation of one of the names: it seems they couldn’t decide whether Yuri Killian’s father was named “Aaron” or “Arlon.” Obviously this character was not Japanese, so either translation would have been fine based on the katakana spelling of his name. I think they settled on “Arlon” for most of the series, but “Aaron” definitely popped up a few times. Thankfully this Arlon/Aaron was a very minor character, but hopefully this discrepancy highlights the importance of sticking to translation conventions. After all, we can’t have the viewer losing track of who’s who while watching, can we?

Conventions are important even outside of translation. If you’ve ever read an article from the Associated Press you’ll notice that all instances of the word “adviser” will be spelled with an “e,” instead of being spelled “advisor” as you might expect. This is true of every single Associated Press article that mentions the word “adviser”. Why is this?

The Associated Press has a rule for its writers that goes like this: when two spellings of a word are valid, the spelling that comes alphabetically first should be used in all cases. You may argue that “advisor” is the more commonly used spelling, but that doesn’t matter. The AP’s writers cannot look up which spelling is more common for every single word they use. Most words would not have any statistics on them, and for the ones that do, it’s hard to make sure those statistics are reliable.

The AP wants no inconsistencies in its articles. Not only would it be unacceptable for two different AP writers to use different spellings, the same writer might accidentally switch spellings of a word on a long article, which is even worse. The AP cannot rely on something as unreliable as statistics. Therefore, it came up with a simple, arbitrary, but effective rule: use the one that comes alphabetically first. Use “octopi” instead of “octopuses.” Use “indexes” instead of “indices”. Does it fundamentally change the meaning of the article? No. But it’s a convention, and it’s important.


[Update: I realized I'd completely forgotten where I'd heard this tidbit on how the AP style book decides on spelling guidelines. I did a bit of Googling and I couldn't find any confirmation, though I did confirm that the AP recommends "adviser" over "advisor". Considering the lack of Google results, it could very well be a myth I heard from a teacher at some point. Take it with a grain of salt.]

Now to finally get to my main point: today I’d like to talk about smart conventions.

Just because conventions don’t affect the essential accuracy of a translation does not mean that there’s no such thing as good conventions and bad conventions. Now, when a translator is faced with a choice of convention, he or she has a couple options. The first option is that the translator has a set of conventions that he or she uses for every single translation, and simply follows these conventions every single time. The second option is that the translator has no preexisting standards and thus makes an arbitrary choice of convention each time; in other words, whatever works is fine. Out of these two the first one is generally preferred, but these aren’t the only two options.

What I’d like to advocate is smart conventions. A smart convention is not a convention that is chosen arbitrarily, nor is it a convention that is followed just because it’s a rule. Instead, it’s chosen for real, specific reasons. Furthermore, a smart convention remembers why it was chosen. A smart convention can vary and adapt based on the situation; it knows why it exists, and knows when it can and should be broken.

Let’s change the direction briefly and bring up Crunchyroll. I can already hear the haters clamoring in the wings, but hear me out here (and for the record, yes I have a paid account at CR).

If you ask me, Crunchyroll could learn a bit from the AP. Whatever you have to say about the accuracy of their translations, what bothers me more is the lack of consistency. This is not so much the fault of the individual translators as the fault of a lack of consistent standards for all the translators: it seems to take the second option, i.e. arbitrary conventions. Some of the shows have honorifics in them, while others don’t. Some actually switched partway through the same series! I realize that for some older shows CR uses the DVD subtitles; that I have nothing to say about, since the translation will just be what the DVD licensors chose. However I have to wonder why shows simulcasting in the same season have such different conventions. Why not be like the AP, and define guidelines that apply to all the shows on CR?

That said, I actually wouldn’t end up advising CR to be like the AP and just define a catch-all convention. I’d rather CR use smart conventions.

Let’s provide a specific example. The CR translation of Gosick preserves honorifics. On the other hand, their translation for Hourou Musuko drops or translates honorifics (as Mr, Mrs, etc). I assume this is because the shows have different translators.

Now, let’s think about this for a second. Gosick is set in a fictional European country, and the characters in the series, while of course speaking Japanese, are “technically” speaking whatever they speak in Saubre (French?). On the other hand, Hourou Musuko is set in Japan, and furthermore a big focus of the show is the subtle relationships between the characters. The fact that most of her friends call Takatsuki “Takatsuki-kun” and that Nitori calls her “Takatsuki-san” is actually kind of a big deal.

Why translate Gosick with honorifics, when it’s actually set in Europe and would make more sense with no honorifics or translated honorifics like Mr and Mrs (or if you want to be even fancier, maybe Monsieur and Madame)? Why insistently drop honorifics in Hourou Musuko, when it’s set in Japan and the use of honorifics is actually relevant to the story?

Here is where the smart convention comes in. The option one convention might read: “Translate honorifics all the time for all shows.” The option two convention might read: “Do whatever you want with honorifics, but try to stick with that choice as long as you’re still working on that show.”

The smart convention might say: “If the show is set in Japan, favor leaving honorifics untranslated, especially if it might affect the plot. If the show is not set in Japan, favor localizing the honorifics to fit the country where it takes place. Any special cases are at your discretion, but be consistent.”

The vital takeaway here is to be aware of the importance of conventions but not to lose sight of why they exist—to choose your conventions for good reasons but be ready to break them when those reasons have changed. 

So I’ve been going on about “conventions, conventions” for over 1,400 words now. And hey, they really are what I consider one of the most important aspects of translation. A lot of how I judge a translation is based on how well the translators chose their conventions (by my definitions) and how well they stuck to them or changed them as needed. I’ve critiqued some professional translations in this essay but don’t think fan translations are exempt; in fact I have plenty more examples I left out of this essay for fear of making it even longer than it already is.


This post has been mainly about the importance of having conventions and introducing my little pet idea of a "smart convention," but I'd like to write another discussing how to choose these conventions, perhaps using some of the examples I left out here.

I am of course by no means perfect, and despite all this theorizing I do, I’m still working out how to practically apply these theories to my own translation style. All in all, hope somebody found this interesting, and I’ll see y’all on Thursday for more A-Channel.