Sardines and Localization

Bagoum
Bagoum
Jun 28 · 17 min read

The mainline Disgaea games features stories broken up into several “chapters”. At the end of each chapter is a “Next time on…” segment (much like you see in your favorite anime), but, in the spirit of Disgaea, they’re all twisted in some absurdist manner.

In Disgaea 4, each segment begins normally, before quickly spiraling into Valvatorez ranting about sardines. On its face, this seems to have nothing to do with localization. But, in fact, every one of these nine segments suggests a problem for localization, and there’s… a lot to learn from comparing the original to the English release. And that’s exactly what we’re going to do here.

This article will be helpful no matter whether you know nothing about game localization (you’ll get an idea of some of the problems) or if you know everything about localization (you’ll learn about the most compressed Japanese-English case study ever written).

The power of sardines!

Japanese words are in italics when not translated. I leave some words untranslated for clarity. This will be especially important when discussing the word iwashi, which APPROXIMATELY translates to “sardine”. When I put something in brackets [], I’m recording what was written in the original.

1: Warm-Up: Kanji Jokes

Top half is translated Japanese (super-literal, maintaining even sentence structure where I can), bottom half is the English localization.

In kanji, iwashi [] is written as left-radical-fish [] on weak []!

…From today on, we should instead write it as left-radical-fish [] on strong []!

— — — —

The kanji for the word “sardine” is a combination of the characters “fish” and “weak”!

…From now on, let’s officially make the combination of “fish” and “strong” the correct kanji spelling for sardine!

The translation of the kanji joke here is as straightforward as possible. It’s also the most basic possible kanji joke, relying on observing that some character is composed of two other characters.

What makes kanji jokes difficult is that you can only translate them by crudely signposting that you’re actually talking about the linguistic structure of a different language. And, furthermore, it’s not obvious that people playing the localized versions will be able to follow. After all, even the term “kanji” is not particularly transparent, and people familiar with the Latin alphabet may have no conception of an ideographic language. Thus, you have two options: crudely signpost and hope that people figure it out, or rewrite it.

Incidentally, in Russian, iwashi is pronounced as “ivasii”.

— — — —

By the way, “sardina” is sardine in Russian.

On first glance, it looks like the localizers hacked their way into this translation. Is sardine written as “ivasii” or “sardina”? Actually, it’s both: “sardina” refers to sardines in general, whereas “ivasii” refers specifically to sardinops melanostictus, which is one of the fish to which iwashi refers. (Take this with a grain of salt, I don’t know Russian.) The localizers got incredibly lucky here, and chances are that an English -> Japanese localization would be screwed by such a line.

2: Translating Translation

This one is super hardcore, and probably a bit much for the second segment. Strap yourselves in…

Believe it or not, “sardine”, which is supposed to be a translation of iwashi, includes several small fish in the Clupeinae subfamily [nishin-aka] that aren’t iwashi! Like mamakari {sardinella zunasi}!

— — — —

The truth is, the word sardine refers to more than one kind of fish, including other small fish in the Clupeinae family! Like the Japanese shad!

Both versions discuss the English word “sardine” in this line. However, the English version misses the point, since the original Japanese was comparing “iwashi” and “sardine”. It’s kind of unsurprising to say that “this common word has multiple taxonomic referents”, but to point out differences in cross-lingual use is interesting. The variants also use different “other fish” examples, which is important in the next two lines.

There’s also a taxonomic error. “sardine” is generally taken to refers to several fish in the Clupeidae family [nishin-ka]. Clupeinae [nishin-aka] is a subfamily, and only some sardines are in it. The Japanese shad (konosirus punctatus) is in the family, but not the subfamily.

By the way, mamakari is written as 飯借, so the rule is that you must eat it with a meal [] borrowed [] from your neighbor!

— — — —

By the way, sardines are named after an island called Sardinia in the Mediterranean Sea!

The joke in the original is based on the kanji writing of mamakari, in a similar fashion to the “fish-weak” joke in the first segment. However, instead of translating it literally as they did in the first segment, the localizers rewrote it. In doing so, they lost the “silly rule” that Valvatorez invents (Valvatorez’s adherence to “silly” rules is an important character point in the story!). I assume this is because this kanji joke is a lot more roundabout and would require way more signposting to understand in English. This is a reasonable decision, but they should have included the character point somehow, maybe by adding a line like, “so if you visit Sardinia, you must eat sardines with every meal”!

Oh, and mamakari are also called sappa!

— — — —

Oh, and shads are also called river herrings!

While these lines follow directly from the “other fish” examples in the first line, the English version is weaker because the lines don’t flow. In Japanese, the topic progresses from (mamakari as sardine > mamakari’s kanji writing > mamakari’s name). In English, the progression is instead (shad as sardine > Sardinia??? > shad’s name). The execution could be better.

The problem is that while “sardine” is supposedly a translation of iwashi, it doesn’t include urume-iwashi or katakuchi-iwashi!

— — — —

In Japanese, sardine is translated as iwashi, but urume-iwashi and katakuchi-iwashi aren’t even included in the same family!

The Japanese line is straightforward and a further development of the preceding argument. In Japanese, there are three primary types of iwashi, which are ma, urume, and katakuchi. So, not only does “sardine” include non-iwashi fish, it also excludes two types of iwashi. The point being made is that the English word isn’t good.

Just like the last line, the logical progression is broken in English. Since the first line didn’t discuss the crosslingual usage, this line doesn’t follow from it. Also, it makes no sense. As an English reader, I have no idea what “urume-iwashi” or “katakuchi-iwashi” are supposed to be, or how they relate to “iwashi”, or why I should expect them to be in the same family. An additional “two of its subtypes…” would have helped immensely.

“Anchovy”, which is translated as katakuchi-iwashi, includes even etsu, which looks nothing like iwashi even if it is part of the Engraulidae family [katakuchi-iwashi-ka]!

— — — —

Katakuchi-iwashi, the anchovy, is in the Engraulidae family that includes the grenadier anchovy that doesn’t even look like a sardine! So why do they even call them iwashi!?

The logic in Japanese is not obvious when translated, so here’s an explanation. English categorizes katakuchi-iwashi and etsu both as “anchovy”, but etsu does not look like an iwashi, so the argument is that “anchovy” is a loose term. The fact that etsu is part of Engraulidae is given as a sort of concession to the English categorization. The reason that it is a concession is that the Japanese word for Engraulidae is just katakuchi-iwashi-ka, so stating “etsu is in katakuchi-iwashi-ka” implies that SCIENCE admits that etsu is similar to iwashi. However, Valvatorez asserts that etsu nonetheless looks nothing like iwashi.

Note that etsu is the Japanese grenadier anchovy.

The English logic takes the opposite position, and argues, from this concession on the taxonomy of katakuchi-iwashi and the grenadier anchovy, that katakuchi-iwashi shouldn’t be categorized as iwashi in Japanese. It argues that the Japanese categorization is loose.

So in these last two lines, we see the Japanese version arguing based on the appearance of iwashi that English categorizations are bad, while the English version argues based on the taxonomy that the Japanese categorizations are bad. The localization basically reverses the argument. I think this is a pretty solid decision, since it makes more sense to argue in favor of Japanese terminology in Japanese and in favor of English terminology in English. Since the side taken doesn’t have much consequence for the story, the localizers had the ability to switch at will. In game localization theory, one of the biggest questions is always: what are the boundaries in which we can modify things? I like that the localizers recognized here that they could switch around the argument, even if they butchered the execution. In most later cases, localizers don’t explore these boundaries.

3: Translating Words, Losing Structure

iwashi isn’t in iwashi-ka, but instead in Clupeidae [nishin-ka] Clupeinae [nishin-aka]!

— — — —

Sardines are not Clupeinae! They are Clupeiformes Clupeoidei, to be specific!

The Japanese joke here relies on an observation that you may have made in the previous section. Apparently, a lot of taxonomic names for orders and below in Japanese just take the name of a dominant member and stick a corresponding suffix on the end. (I think we do this in English taxonomy, but using Latin names, so nobody can tell.) For example, a certain species of herring is called nishin in Japanese. Its genus is nishin-zoku, its subfamily is nishin-aka, its family is nishin-ka, and its order is nishin-moku. It is therefore reasonable to expect that there might be an iwashi-ka (and someone who has as much faith in sardines as Valvatorez would believe as much), but in fact, no such thing exists, and (at least one) iwashi are subtyped under nishin’s family.

There isn’t really a joke in the English version, because nobody really has any reason to believe that sardines should be in Clupinae or not, except for the one comment from the previous segment (which nobody could possibly remember). This said, there’s not much they could have done: there’s no way to translate the term iwashi-ka; since English taxonomy uses Latin names, a fictional taxonomic structure can’t be recognized by a common reader. In general, translating taxonomy words out of Japanese loses the simple structure that allows it to make jokes like iwashi-ka.

But, to be spiteful: according to my research on fishbase.de, there are actually sardines in different orders. (Most sardines are in the order Clupeiformes, but some are Characiformes or Cypriniformes. Though, to my knowledge, all Clupeiformes sardines are in the suborder of Clupeoidei.) This occurs because the word “sardine” was borrowed from French and then spread globally via American “diplomacy”, so it ends up applied to many similar-looking fish across the world, whereas “iwashi” has only ever been used in Japanese to describe fish in Japanese waters.

ma-iwashi and urume-iwashi are in Clupeidae [nishin-ka]! But! katakuchi-iwashi is — how could this be?! — in Engraulidae [katakuchi-iwashi-ka]!

— — — —

Although, Japanese sardines and round herrings are Clupeidae… and Japanese anchovies belong to the Engraulidae gang!

The Japanese engages in a logical continuation of the first line, discussing the taxonomic placement of the subtypes of iwashi. The English makes no sense, because as an English reader, I have no idea why round herrings or anchovies are relevant to the discussion. This disparity occurs because they translated the line literally (ma-iwashi = Japanese sardine, urume-iwashi = round herring, katakuchi-iwashi = Japanese anchovy), but without the knowledge that all three fish are subtypes of iwashi, there’s nothing that links them together. In Japanese, this knowledge is a given, since the names of the three fish are an obvious pointer. But in English, the naming structure is completely different. Thus, simply by translating the names from Japanese to English, the sentence loses its logical structure. Translation is magical!

Also, if you’re really excited by taxonomy, here’s a fun fact. urume-iwashi, the round herring, is technically no longer in Clupeidae. Instead, it’s part of the Dussumieriidae family. The story is that Dussumieriidae used to be part of Clupeidae as a subfamily, but was recently (?) separated into its own family. My source for Japanese taxonomy doesn’t list a Japanese name for this family. However, when it was a subfamily of Clupeidae, it was called — wait for it — urume-iwashi-aka. So we can assume the family is called urume-iwashi-ka. This means that all three iwashi are in different families, and ma-iwashi (the “true” Japanese sardine) is the only one without a family named after itself.

4: Proverbs

“Rinse iwashi seven times and it will taste like tai”! Wash it well to take away the fishy smell, and reveal its hidden power that can rival even tai!

— — — —

Just rinse them really well, and you will taste just how great these fish are!

Translating proverbs is a classical problem in localization. This is because a literal translation often makes no sense without cultural background knowledge, and a “similar proverb” may not be available or may derail the rest of the writing.

tai is another type of fish well-known in Japan. Given just this dictionary knowledge, we have no way of understanding the proverb. We require some cultural background knowledge, which is that iwashi is a low-tier and smelly fish, whereas tai is a high-tier and not-smelly fish. Then, the proverb makes sense: it claims that, simply by washing this low-tier fish, we can get a taste comparable to that of a more prized fish.

The localizers took the easy way out on this: they simply dropped the proverb and approximated the point being made. In this case, a “similar proverb” would have derailed the dialogue, since the original dialogue necessitates that the proverb be about iwashi. When there are multiple “sources of strength” for a specific linguistic phrase — in this case, the cultural status of the proverb, the proverbial meaning, and the literal reference to iwashi — translation is more difficult. However, it’s usually better to settle for a weaker phrase than to drop it altogether.

5: Language Jokes

Some fish that in Japan have “iwashi” in their names are not considered to be iwashi overseas!

But what lies at the bottom of this decision to name large fish swimming foreign seas as “oki-iwashi” and deep-sea fish as “itohiki-iwashi” is the relentless Japanese iwashi spirit!

The day that all fish are in Japanese named “xxx-iwashi” may be close!

— — — —

Even if some fish have “iwashi” in their names, they’re not always considered to be sardines outside of Japan!

Big fish that swim in foreign seas are called “oki-iwashi” and “itohiki-iwashi”, even though they are deep-sea fish.

There’s no end to the sardine spirit in Japanese culture! Someday, every fish will be called “something-iwashi”!

This segment is a commentary on Japanese language and culture. While there would be nothing strange about this for a Japanese reader, I expect it would be an awkward experience for an American playing through an English game to suddenly be accosted by facts about Japanese.

This is a good time to introduce some terms from game localization theory. There are two terms that describe general approaches to localization, which are domestication and exoticization.

In domestication, you take all the references to the source culture and replace them with references to the target culture. In high-level terms, this is what Nintendo does with Fire Emblem, and it’s the reason there’s so much content modification in Fire Emblem localization (see my post on localization and censorship). Because Nintendo is obsessed with its brand image, it has to make sure that each of its localizations present a similar image, based on the cultural standards of the locale. On a lower level, you might see the replacement of pop culture references, mythological references, character dialog, and — most importantly for age ratings — references to sex and drugs. Several examples are given in Di Marco (2007).

In exoticization, you take all the references to the source culture and highlight them, so the localized version gets a distinct “foreign” feel. A recent example of this is Sekiro, which is one of the most exoticized games ever created. This isn’t only because it’s steeped in the cultural and artistic motifs of Japan and its mythology, but also because FROM maintains the absolute ubiquity of these motifs in the localization, and even soaks the localization in “exotic” features like kanji. When enemies perform perilous attacks, you’ll see a red kanji 危 (hint: it means danger) appear over your head, localization or not. Whenever you do something game-changing (killing a boss, lighting a bonfire), a massive vertical kanji message will appear on screen, with a tiny English translation underneath. All of this reads “exotic” to a foreign gamer, but is quite familiar to Japanese gamers.

[If Sekiro is exoticization, what is Dark Souls? Dark Souls is strange in that it starts off as a “foreign”-feeling game in Japan, with English-only voiceovers and massive English messages, but then becomes a fairly familiar European fantasy when localized. Kind of like angels in Evangelion. I can write on this later, perhaps.]

If we take a commentary on Japanese language/culture and keep it in the localization, we are engaging in exoticization. In itself, this is fine, but there is one serious side-effect of the exoticization strategy: it highlights the real-life culture behind the references. In Sekiro, this is beneficial, since the game is set within an explicitly Japanese context. But Disgaea is not about Japan — it’s about demons who live in the Netherworlds. While there is a human world, it’s by no means our world (as shown in the time-loop episode from Disgaea 4). References to foreign cultures thus tend to be distracting, and fail to integrate into the rest of the game. Imagine if I, in the next paragraph, were to randomly make a reference to Bulgarian folk music; wouldn’t the strangeness of this reference be slightly awkward, or at least more awkward than a reference to a culturally familiar trope like “American diplomacy”?

6: Proverbs, Redux

Speaking of heads, have you heard of the proverb “better to be the head of an iwashi than the tail of a tai”?

— — — —

Ever hear, “Better to be the head of a sardine, than the tail of a sea bream”?

tai can be translated as “sea bream”. I’ve never heard of this fish, personally.

The Japanese line is straightforward given our previous discussion. Given that tai is a high-tier fish and iwashi is a low-tier fish, this is clearly comparable to “it’s better to be a big fish in a small pond…”

This time around, the localizers were forced to translate the proverb literally, since the rest of the dialogue depends on the exact logical structure of the proverb and the relationship between tai and iwashi. And it’s awkward, because sea breams are not at all familiar to non-Japanese audiences. It’s in cases like this where you should scrap the entire dialogue and rewrite it with a different joke. (Or, if you’re really daring, you can merge the original proverb and a similar proverb: “better to be a big sardine in a small pond than a small sea bream in a big pond”. This, though, is an advanced localization technique; don’t try it at home.)

An exoticization strategy fundamentally requires that an American audience can appreciate the “Japanese-ness” of the image. However, in this case, an American audience would probably not even recognize the usage of “sea bream”, let alone the Japanese nature of the proverb. Exoticization often seems like the easy way out (since you only have to make a literal translation), but it also tends to lead to nonsense like this, where the English simply doesn’t read at all.

7: Soramimi (mishearing, lit. sky-ear)

Valvatorez: That’s right, the true enemy! Purines [purin-tai]!

Fuka: Pudding [purin]? But I love pudding.

Valvatorez: … Not that purin! purin may be purin but it’s a terrifying sort of purin!

— — — —

Valvatorez: That’s right, our new enemy is the “purine”!

Fuka: Prunes? Ugh, I hate prunes…

Valvatorez: … It’s “purines”! They are definitely more horrific than prunes!

The proper way to translate a language joke is more in line with domestication: you have to modify the joke so that it fits the target language. In Japanese, “purine” sounds like “pudding”, so Fuka comments that she likes pudding. In fact, except for the suffix -tai, they are homophones. This allows Valvatorez to make a joke where he tries to distinguish the terrifying purine-purin from the delectable pudding-purin.

In English, there’s no homophone of “purine”, and “prune” is the closest food word, so Fuka comments that she doesn’t like prunes. This prevents Valvatorez from making the silly comment, and he instead has to settle for a fairly basic line. I would have liked to see a prune joke here to make up for the lost homophonic joke.

There is one problem: the on-screen image is still one of pudding.

8: Translating Words, Losing Structure, Redux

Remember how we discussed at length the difference between the categories of iwashi and sardine? Well, here it strikes again.

Wow! Young iwashi fish are used in the preparation of chirimenjako!

… That’s right! While shirasu is a general term referring to many types of young fish, the one that’s most commonly eaten is young katakuchi-iwashi!

— — — —

The truth is…! Did you know that various kinds of fish larvae are used in chirimen-jako?

… That’s right! Shirasu usually refers to any kind of fish larva, but the larvae of Japanese anchovies are the most commonly eaten shirasu!

The problem for the English localization is straightforward: why is Valvatorez talking about Japanese anchovies? His shtick is sardines — what’s going on here?

As long as the English word “sardine” doesn’t encompass katakuchi-iwashi, there’s no way to translate this section while still making it relevant to sardines. In other words, a literal translation has made this section completely meaningless in English.

As in 6, the best way to fix this would have been to scrap the dialog and rewrite it with a different joke.

9: Translating Words, Losing Structure, Reredux

In this final segment, Valvatorez doesn’t actually get to talk much, so he makes one interjection at the end:

Just remember this one thing! The name of sokonokogiri-iwashi-tsubu-iwashi! Because it sounds cool!

— — — —

Just don’t forget the name of this fish, Bathyprion danae Xenodermichtyhys nodulosus! Because it just sounds cool!

In the original, it’s absolutely obvious why Valvatorez brings up this name: it’s a fish which has iwashi in its name twice! But the English localization runs into the same problem as the previous segment: there’s no reason for Valvatorez to be talking about this fish. It has zero sources of strength. It’s just a random term out of nowhere.

Why does this occur? Because “bathyprion danae” is a literal taxonomic translation of sokonokogiri-iwashi. Did you see that coming?

The solution to this would be fairly easy. Just make up a fish name that sounds like “sardine”. For example, how about “sardina sardinops sardinella pilchardus”? It’s not a real fish, but all the words are from actual taxonomic categorizations for certain types of sardines. Hire me NISA, I know you need a localizer for the D4 remake.

Lessons and Conclusion

If there are a few things you should take away from this:

  • Literal translations can easily end up being completely nonsensical. See 3,6,8,9.
  • Literal translations, if done in the exoticization strategy, can highlight features of the source culture. This may be good or bad depending on the game. I don’t think it helps a game about demonology.
  • Nonliteral translations can drastically increase readability without significant consequence for narrative or personality. See 2 for a good idea that had botched execution. See 7 for a decent execution that could have been more thorough. See 9 for an example where this was not done to a great cost.
  • Sometimes you just have to throw out the original and write something new. See 3,6,8 for examples where insistence on fidelity to the original structure leads to garbage. See mamakari in 2 where this is done on a minor level.
  • Jokes that rely on language-specific features are difficult to translate without rewriting. See 1,3,5 where kanji, taxonomy, and naming are untranslatable. See 4 where the localizers don’t even bother, and 6 where a literally translated proverb doesn’t follow well. In these cases, you may be able to get away by simply talking about the source language instead, as in 1 and 5. But be careful about talking to Americans about kanji.

That’s all for this article. When the Disgaea 4 remake comes out in October, I might pick it up to see if they’ve fixed the localization at all.

References

Di Marco, Francesca. 2007. “Cultural Localization: Orientation and Disorientation in Japanese Video Games.” Revista Tradumàtica — Traducció i Tecnologies de la Informació i la Comunicació 05 : Localització de videojocs. http://www.fti.uab.cat/tradumatica/revista/num5/articles/06/06art.htm.

Bagoum

Written by

Bagoum

Software engineer, epic gamer, and student of Barthes and Skinner.