I did this in early 2008, back when I had become linguistically interested in Japanese, but before I actually started learning the language. That was also the period of my life when I was the most hardcore weeaboo I'd ever been (and goddamnit I swear, there no relation between those two things!), so pretty much all the words I used as examples are either directly from animes or words I found in dictionaries to serve as examples of corner cases. You can probably tell exactly which terrible animus I was watching at the time I wrote this up, and judge me accordingly. I still think "Strawberry Power" is an awesome song

Precisely because I now actually know Japanese (to some degree anyway ), I have some ideas for improvements of the scheme, which I will mark with supplementary text in the main body of the text. These revisions are from May 2010.

The Quesa Japanese Romanization Scheme

I was bored in class one day, so I decided to create a new method of writing Japanese-language text using the Latin alphabet. My goals in this endeavor were threefold:

  1. 1. To make Quesa romanization unnecessarily complicated and unwieldy, with many irregularities. I want this to be the worst practical Japanese romanization ever.
  2. 2. To be unambiguous – i.e., a given Quesa-romanized string of text should represent exactly one string of Hepburn, and vice versa. It should still be usable, just annoying. Note that I do not require that a given string of Quesa-romanized text represent exactly one string of kana; in fact, the system does ignore certain distinctions in written kana when they have merged in sound, such as the distinction between じ and ぢ, ji, /dʑi /.
  3. 3. To make the Romanization look absolutely nothing like Japanese. Personally I think it looks kind of like Polish, what with the frequent use of <ł> and frequent graphical consonant clusters.

Quesa Romanization is an abugida – that is, most basic graphemes actually represent a single consonant followed by a specific vowel. Generally this vowel is /ɯ/, although the exact implied vowel depends on the consonant. To write a syllable containing a different vowel, the letter is followed by an explicit vowel letter, which suppresses the inherent vowel: t = /to/, ta = /ta/. x = /ha/, xo = /ho/. This permits long strings of consonant letters in the romanization, even though Japanese's phonotactics permit very few phonemic consonant clusters in the spoken language.

The Basics

(Note: Quesa-romanized letters and words will be written in italics. Equivalent Hepburn words will be written with bold italics. IPA or X-SAMPA is written /laik θɪs/)

Vowels

Japanese has a 5-vowel system distinguishing two lengths, common to many of the world's languages: /a e i o ɯ a: e: i: o: ɯ:/. The vowels are romanized simply as: a e i o y. Length is indicated by the addition of an acute accent ´: á é í ó ý. Tókjó - Tōkyō; However, if a lengthed vowel matches the inherent vowel of the preceeding consonant, the acute is omitted; the length is indicated only by the non-suppression of the vowel: ćisan – chīsana, but ćýŋok – chūgoku. Note that when the inherent vowel is /ɯ/ , the grapheme used to represent it is y, not <*i>u. y is also used instead of u when writing the sound /ɯ/ in isolation: yć – uchi. The grapheme u does exist, though, and its use is explained in the Nasalization section.

Plosives

p b t d k ŋ q represent, respectively /p b t d k g k/. ŋ is used to represent the phoneme /g/, reflecting the allophonic variation of /g/ as the velar nasal [ŋ], and facilitating Goal 3: alŋat – arigato. The inherent vowel for all non-coronals (that is, everything but /t/ and /d/) is u

q is a variant of k, differing only in the inherent vowel – q's inherent vowel is /a/, while k's is /ɯ/: qn – kana ; kł – kuru. *ka is never used; q is used in all situations to represent /ka/.

When the vowel of the syllable is neither /a/ nor /ɯ/, either k or q may be used: qoqoło and kokoło are both valid representations of Hepburn kokoro. Generally, k is preferred is the following vowel is /e/ or / i/, and q is preferred if the following vowel is /o/: kéićło – Keiichiro; qoqoło – kokoro.

Owing to the frequency of the syllables /te/ and /de/ because of their use as a variety of basic grammatical particles, these syllables are represented with the letters ʈ and ɖ, both with inherent vowel e: jõɖ - yonde; ʈŋmi - tegami

Fricatives

Japanese has several fricative and affricate sounds, displaying some interesting allophonic variation which affects their romanization. In this section we will look at the four sounds /ɸ s z h/. /ɸ/ is an allophone of /h/ occurring only before /ɯ/ in native vocabulary. In more recent loanwords from western langauges /ɸ/ is a phoneme in an of itself. In kana such words are written using the kana for fu + a single vowel kana. For instance, faia, a loan from English "fire", is written in katakana as ファイア

Quesa romanization uses two separate symbols for /ɸ/ and /h/: f for the former with the inherent vowel /ɯ/, and x for the latter, with the inherent vowel /a/: ftacme – futatsume ; fãtaźikh – fantajikku; xźmalmas – hajimarimasu. /s/ and /z/ are represented as s and z, respectively, both with the inherent vowel /ɯ/: ɖs – desu, qãnzki – kannazuki.

Nasals

Japanese has two nasal phonemes, /m/ and /n/, as well as the moraic syllable-final (or coda) nasal. The first two are simply m and n, with inherent vowel /a/. The coda nasal is indicated by writing the tilde diacritic, ~, over the vowel of that syllable: dãzẽ - danzen; -sã - the honorific -san. For a coda nasal folloiwng a /ɯ/, instead of *ỹ, u is written, without a tilde: umé - unmei, udo - undou. Japanese permits syllables with both long vowels and a nasal coda in loaned vocabulary (ex. スプーン, Hepburn supuun). In Quesa romanization, such a combination of length and nasal coda is represented with the diacritic ^: spû - supuun.

Liquids

The approximants /ɰ/ and /j/ exist in Japanese. /ɰ/ is romanized as , with the inherent vowel /a/: vtaš – watashi. /ɰ/ cannot occur before any other consonant (with one exception to be dealt with later), so there is no need to ever write another vowel symbol after it. /j/ is romanized as j, also with the inherent vowel /a/: jsaxiło – yasahiro. The lateral /ɺ̠/, Hepburn r, is represented as ł, with the inherent vowel /ɯ/ : miʈł – miteru. There exist two variants of this grapheme, all with the base sound /ɺ̠/ but with different inherent vowels. l has the inherent stem vowel /i/: lbõ – ribon, while r has the inherent stem vowel /a/: daqr - dakara. Just as with the k/q pair, it is mandatory to use the consonant grapheme when its own inherent vowel needs to be expressed, but either of the three may be used when the vowel is /o/ or /e/: le, łe and re are all valid representations of re. However, ł is preferred for these two vowels.

Palatalized Consonants

Palatalized consonants, such as the ky in Kyōto, are written with a following j: kjót - Kyōto. The inherent vowel of all such consonants is /a/, following the inherent /a/ of the j grapheme: xphjk – happyaku.

Geminate Consonants

A geminate consonant (i.e. a doubled consonant, like the tt in zutto) is indicated by writing the single consonant, then following it with ,: zth – zutto ; xphjk – happyaku. The inherent vowel is unchanged when geminated – in zth the final o is omitted because it is the inherent vowel of an un-geminated t.

Inherent Vowel Separator

In some cases, it is useful to indicate that the inherent vowel of a consonant letter is not supressed by a following orthographic vowel. Hayaku, for instance, cannot be Quesa-romanized as *xjk, because the the j would suppress the inherent /a/ of the x, resulting in *hyaku. To avoid this, an apostrophe ' is added after the consonant letter, to indicate that the inherent vowel is not suppressed. Hayaku in Quesa romanization is x'jk.

Allophonic Variants

Several Japanese consonants have allophonic variants when they form a syllable with the vowels /ɯ/, /i/ or both:

Special Notation for Grammatical Particles

The grammatical particles no and wo have special status in Quesa romanization. Ordinarily, the Quesa romanized forms of those syllables are identical or similar to their Heburn romanization, namely *no, *vo. To achieve greater obfuscation, they may be romanized as n' and v' in all circumstances, even when the syllables in question are not used as particles: anta n' yć v'– anata no uchi wo; nan'x – Nanoha. Unlike most of the rules of Quesa romanization, however, the decision to use no/n' and vo/v' is arbitrary, provided that the syllables are not in fact used as grammatical particles. Nn'x and nnox, for instance, are both perfectly valid ways of representing Nanoha. *anta no, however, is not permitted.

Example Text

Hepburn:
anata ga sabishii toki itsudemo watashi tonde yuku kara deaete shiawase na no nani mo kowakunai kitto tatakaeru zutto mamoritsuzukitai anata no egao dakara isshou ni

and corresponding Quesa:
anta ŋa sabiśi tki icɖmo vtaś tõɖ jyk qr ɖ'aeʈ ś'avse n n' nni mo kovkn'i kith tataq'eł zth mmolczkitai anta n' eŋao daqr iśhó ni

A longer example text (and one that is not tremendous weeabooery, although arguably literature weeabooery ):

Hepburn

Wagahai wa neko de aru. Namae wa mada nai. Doko de umareta ka tonton kentou ga tsukanu. Nandemo usugurai jimejime shita tokoro de nyaanyaa naiteita koto dake wa nisshi shite ita.

and Quesa

Vŋx'i v neqo ɖ ał. nm'e v mda n'i. doqo ɖ ymłeta q tõtõ qẽto ŋ cqnu. Nã ɖ mo ysŋur'i źmeźme śta tqoło ɖ njanja n'iʈ'ita qot dake v nish śʈ'ita.