Japanese Romanization is the way that Japanese text gets transliterated into the Roman alphabet. The romanized text is referred to in Japanese as "Rōmaji", from Roman alphabet + "ji" meaning "characters" (much the way "Kanji" literally means "Chinese characters", hence it being the on'yomi reading of "hànzì"). The word Romanization can actually refer to using any Latin-based alphabet (French, German, Polish, ...) to write a words originally written with any non-Latin script, but in English-speaking fandom it almost universally refers to Japanese-to-English transliteration. See Romanization, Japanese Writing System.
Japanese to English
Japanese has a few quirks that don't exist in English. Although in general pronouncing kana is simple, there are challenges in representing it in Roman letters. There are several systems to do this:
- Hepburn romanization and its revised variants are the most widely used methods of transcription of Japanese., especially for formal and academic writing. The Hepburn system is intended for use by English speakers and is based on English phonology, so a native speaker of English with no knowledge of Japanese will be more likely to pronounce Hepburn-romanized words correctly than if a different system were used. Some linguists dislike the Hepburn method, as it can make the origins of Japanese phonetic structures unclear, but those in favor of it say that the Hepburn system isn't supposed to be used as a linguistic tool anyway; it was originally developed when the relationship between kana readings and pronunciation was looser than it is today.
- Kunrei-shiki Rōmaji (literally: Cabinet-ordered romanization system, romanized as "Kunrei-siki" in its own system) is based on the older Nihon-shiki system, and was modified for modern standard Japanese, essentially meaning words are romanized not as they appear, but how they sound in modern spoken Japanese.
- Nihon-shiki or Nippon-shiki Rōmaji ("Japan-style"; romanized as Nihon-siki or Nippon-siki in its own system) is the most regular out of all the major romanization systems for Japanese, and has a one-to-one relation to the kana writing systems. The intention of this system was to completely replace kanji and kana with a romanized system, which, its creator believed, would make it easier for Japanese people to compete with Western countries. Since the system was intended for Japanese people to use to write their own language, it is not designed to be easy to pronounce for English speakers (and isn't for the most part).
- There's also "word processor romanization" or "wāpuro" which is technically a workaround for inputting Japanese with a QWERTY keyboard but is also used for informal writing, especially on the web. It tends to ignore all the difficulties below and just give a direct transcription of the "standard" kana reading; as such, the spelling may not match the actual pronunciation of words.
R vs L
In the Japanese language, there is technically no "l" or "r" sound; instead, there is a single sound half way between both, kind of like a partly rolled "r". In natively Japanese words this is romanized as "r" in all systems. With loan words written in katakana, whether it is romanized as an "l" or "r" (or even "ll") depends on the source word.
Looking at the kana tables, you may notice that there is no "ye" sound, but the unit of currency of Japan is called the yen, which would be spelled "en" in any modern romanization scheme. This is a holdover from old Hepburn where え(e) was often romanized as "ye", as was the now-obsolete ゑ(we). This is usually limited to older things that want to keep it for old times' sake, such as Yebisu Beer from the Ebisu district of Tokyo; Tokyo's former name was romanized as "Yedo" in the first edition of Hepburn's dictionary.
Chi/Ti, Tsu/Tu, Shi/Si, Fu/Hu, Zu/Du/Dzu, Ji/Zi/Di
One difference between the major romanization systems has to do with how certain consonants are written. Certain consonant/vowel pairs sound more like what an English speaker would consider different consonants. Hepburn writes this as the sound (chi, tsu, shi, fu) and Nihon-shiki and Kunrei-shiki write this using the same consonant even if it doesn't match the English sound (ti, tu, si, hu). These romanizations are still taught in Japan, largely because beginning students of English in Japan have difficulties with the concept of letters as single sounds and consonant clusters are too much for them.
The inflected tsu, which sounds like zu (which has largely replaced it in modern Japanese), deserves its own mention. Kunrei-shiki joins modern Hepburn in using the phonetic zu. Nihon-shiki sticks with the same consonant for du. Old Hepburn broke its phonetic scheme to use dzu, which is where the "d" in "kudzu" and "adzuki" comes from.
The inflected versions of shi and chi sound like each other (the former character is more commonly used). Hepburn uses ji for both, whereas Kunrei-shiki uses zi for both. Nihon-shiki distinguishes them as zi and di.
There are three particles which in Kunrei-shiki are written differently from Nihon-shiki to correspond with their irregular pronunciations, which are a remnant of historical kana usage and can be easily confused with different characters:
- Ha (は), when used as the topical particle, is pronounced wa, and Hepburn and Kunrei-shiki follow the pronunciation.
- He (へ), when used as a directional particle, is pronounced e and written that way in Kunrei-shiki; Old Hepburn rendered it as ye.
- Wo (を), which is obsolete in modern Japanese except when used as the objective particle and sounds like o in all cases, is usually rendered as o in modern Hepburn and Kunrei-shiki.
The Long Vowel Issue
In Japanese, vowels can be short or long. A long vowel (which just means that the syllable is held for slightly longer, not that the pronunciation is changed) is written in Japanese as two of the vowels in a row - except in the case of long o (which is usually written with a "u" character, as "ou", instead of "oo") and long e (which is usually written with an "i" character, as "ei", instead of "ee").
For example, the name of the city of Tokyo contains two long o vowels, and the Japanese kana (script) would be most directly transcribed as to-u-kyo-u.
There are several ways of presenting the long o:
- Hepburn technically requires a bar (macron) over the o (ō): Tōkyō. This can be hard to type, and may cause formatting issues when text is copied between different systems.
- Kunrei-shiki uses the circumflex (ô) to indicate long vowels. This is handled better on many computers, though many Japanese-language programs will still reject it as input.
- A double vowel (oo): Tookyoo. The problem with this is that in English this represents an entirely different sound - a long u, as in "spoon".
- The pair spelled the way they are in hiragana (ou): Toukyou. This is even worse than "oo", as in English this can be at least two different sounds: a diphthong as in the word "sound" or a long u as in "through".
- Rarely, an h after the vowel (oh): Tohkyoh. This can look unnatural, as no English words have this combination in the middle of a word.
- The long/short distinction omitted entirely, as is the case with Tokyo. Most English speakers wouldn't really know the difference between a short and long vowel unless it was pointed out to them, so this is probably the most common way to write it. The downside is that if you want to turn it back into Japanese, you would lose the extra information of long syllables.
Note: There are a few cases where the doubled spelling for long "o" actually is "oo". "Ooki" (big), "ooi" (many) and "ookami" (wolf) are three such words. There are a few rare cases of "ee" as well, like "neesan".
Similarly, there is such a thing as a "long consonant", which is usually written by a small "tsu" character (sokuon) before the syllable; this indicates that the consonant part of the syllable is held for longer. This is generally easier to deal with, as the English consonant is just doubled (e.g. "kappa"). It does get confusing when the character to be doubled is a "ch" or "sh" sound, though.
The main exception to the spelling rule is a double "m" or "n", which is written by an additional "n" character rather than a "tsu". R and H cannot be doubled in Japanese, but H can be doubled in katakana to represent the German "ch" sound (e.g. Heinrich or Ludwig would be spelled ha-i-n-ri-(small-tsu)-hi and ru-do-bi-(small tsu)-hi).
There are a few syllables that turn into combinations, like "ji-ya", "chi-yo", "ri-yu", etc., with the second syllable written smaller. (The smaller kana are a modern invention, and historical kana usage also included many confusing alternate spellings.) In modern Hepburn this is turned into "ja", "cho", "ryu", but you can also see "jya"; Nihon-shiki and Kunrei-shiki actually use "zya". Old Hepburn only did this consistently with sh(a/o/u), ch(a/o/u) and j(a/o/u), and wrote, for instance, the names of the city of Kyoto and the island of Kyushu as "Kiyoto" and "Kiushu." ("Kwa" and "gwa" used to occur in the on'yomi readings of many characters; language reforms have collapsed most cases to "ka" and "ga," though a few characters are still usually read as "kuwa.")
The "n" apostrophe
One more issue is how to treat "n" followed by a vowel. Since "n", unlike other consonants, does not have to have a vowel sound after it, it's ambiguous whether "ni", for instance, refers to a single syllable or to a "n" followed by a separate "i". The latter is usually the case in personal names containing "ichi" (e.g. Kenichi, Shinichi). Some systems use an apostrophe to indicate this. (Examples: ren'ai, "romantic love", vs. re'nai, "no re"; shin'en, "passion" vs. shinen, "thought".)
English to Japanese to English
Japanese is a language of syllables. Very few words can end in a consonant; most end in vowels. There are also fewer sounds in Japanese than in English. When an English word is presented in Japanese (generally in katakana, the script used for foreign characters), information is invariably lost. When it then gets translated back into English, the missing information often leads to mistranslations. This is a common malaise when Video Games get brought to English-speaking countries; many names and words are meant to be English, but the translators sometimes mess up on what they're actually saying.
Common transliteration problems from English to Japanese include:
- The lack of a differentiated "R" and "L" sound in Japanese. Japanese has only one sound, which is somewhere between the two. This is probably the most common challenge in romanization: figuring out whether a Japanese syllable is meant to be an R or an L. This is where the term "Engrish" comes from.
- Japanese don't really have a "f" sound; "f" is basically a somewhat stronger version of "h" (perhaps not coincidentally, "fuu" means "wind" is Japanese); the -u syllable is usually written in English as "fu" but the others are "ha, he, hi, ho". This makes for weird combinations like "fu-(small a)" for "fa". Sometimes the two are interchangable; for example, "hu" in Japanese would still be spelled with the "fu" syllable.
- The lack of ending consonants. "n" is the only consonant that Japanese allows to end a syllable, so when foreign words are transliterated into the Japanese syllabaries they end up with extra vowels. "U" is commonly used since it's often elided in speech when it comes between unvoiced consonants; for this reason, Old Hepburn usually wrote it as an apostrophe. When "r" comes at the end of a syllable or is followed by a consonant, it's typical to double the preceding vowel (represented by a long dash in katakana), so for instance "number" becomes "nanbaa". The trailing "s" of plural nouns is often omitted, because the Japanese language lacks plurals.
- Japanese is not written with spaces or capitals. Translators have to figure out where the spaces go, which can be challenging. (Although there is a special dot symbol which can be used to separate words when necessary, e.g. to separate personal name from surname.)
- Missing sounds. Japanese has fewer sounds than English. Examples include:
- "th" is turned into "s" when it's not voiced, like in the name "Smith" (su-mi-su). When it's voiced, the "s" is, too: "the" becomes "za" (although "z" is pronounced more like "dz" in most cases). Since the Japanese language doesn't have definite articles, "the" is sometimes omitted, contributing to The "The" Title Confusion.
- "v" can be written as "u" with a dakuten on it, followed by a vowel, but more often is just rendered with a "b" (e.g. "violin" would be "ba-i-o-ri-n"). This has caused the weapon name "Vulcan cannon" to be mistranslated as "Balkan cannon" in such games as Magical Chase and Forgotten Worlds.
- The "tee" sound doesn't exist in Japanese. It can also be written using "te-(small i)", but it's often replaced by "chi". So "steal" gets turned into "su-chi-ru". Similarly, the "dee" sound in words like "melody" has to be written as "de-(small i)".
- The lengthened "a" and "o" are used to stand in for English diphthongs like "ar," "er" and "ur"; together with the R/L collapse noted above, this is how "claw" is often mistranslated as "crow" and vice versa, and why so many Japanese people will accidentally say "ice cream corn" instead of "cone."
- Japanese treats "n" before any labial consonant ("m", "p", and "b") as sounding like "m" rather than "n", which gets confusing when it is written as n when romanized. This leads to words like "combo" and "computer" being turned into "ko-n-bo" and "ko-n-pyuu-ta", rather than "ko-mu-bo". The same thing happens with the English prefix "in-" when added to a word beginning with "p" (i.e., in+possible=impossible). This assimilation sometimes also happens in other languages, whether the speakers are aware of it or not.
- The "w" sound exists in Japanese only in the syllable "wa"; when "wa" is the wrong sound, the consonant gets replaced by the vowel "u", and is either followed by a large or small vowel. "Wind" becomes either "u-i-n-do" or "u-(small i)-n-do"; in the modern Hepburn system the latter is parsed as "wi-n-do". Sometimes this "u" absorbs the following vowel: "wolf" turns into "u-ru-fu", "wood" is "u-(small tsu)-do", "woman" is "uu-ma-n".
- The five vowels "a", "e", "i", "o" and "u" are each pronounced one way in Japanese, but in context these vowels are rarely pronounced the same way in English. The katakana used to represent loanwords may either attempt to approximate the English vowel sounds or represent the vowels as written (which in words not native to English is often closer to the original pronunciation). One old controversy among Japanese speakers was whether Ultima should really begin with "a" instead of "u".
Sounds that don't fit nicely into English or are unusual can be even more confusing.
Some fun examples of missed Romanization:
- Star Ocean: The Second Story: Scylla → su-ku-ra → Scewer
- Final Fantasy VIII: Thamasa Soul → sa-ma-sa → Samantha Soul
- Tales of Phantasia: Stirge → su-te-i-ji → Stage
- Shadow Hearts: From The New World: Shub Niggurath → she-bu-ni-gu-ra-su → Jeb Niglas
- Final Fantasy Tactics: Breath → bu-re-su → Bracelet
- Wild ARMs 1: Jack Vambrace (a vambrace is an arm guard) → va-n-bu-re-i-su → Jack Van Burace
- Mega Man Xtreme 2: Iris → a-i-ri-su → Aillis
- Final Fantasy V: Wyvern → wa-i-baa-n → Y Burn
- Lufia II: Rise of the Sinistrals: Rafflesia → ra-fu-re-shi-a → La Fleshia
Note that some names were originally in Japanese but meant to ''sound'' English. These names have no "real" translationnote , and can result in all kinds of arguments. A good example is the town called "ri-ze-n-bu-r" from Fullmetal Alchemist, which has been variously translated as Resembool, Risembul, Riesenburgh, and Liesenburgh.