by Haley Halcyon (Haruki Wakamatsu)
Africanist Korean Romanization was inspired by Latin-script variants for African languages, once you get over the surprise of these unfamiliar-looking letters, this system is surprisingly simple and easy to learn.
This idea was spawned by this thing I posted on Reddit to which I got an unexpectedly positive reaction!
Africanist Korean Romanization uses the following special characters.
Africanist Korean Romanization can convert to and from hangul with no loss of information. Using Wikipedia’s example phrase, meaning “a person who cannot do it”:
Hangul is written in a system called moasseugi, which assembles jamo (i.e. individual consonant or vowel letters) into blocks. There has been some historical discussion about switching to bureossugi, a system where jamo are written in a single line, when moving to adopt Western-style horizontal writing, dating back to December 1908. However, nothing really came of it.
Since ⟨ㅇ⟩ is silent when at the start of a syllable block and /ŋ/ at the end, it has been suggested to drop the silent ⟨ㅇ⟩ in bureosseugi. However, simply listing the jamo horizontally will be slow to handwrite, and cause ambiguity in sequences such as ⟨ㅐ⟩ vs. ⟨ㅏㅣ⟩ if the silent ⟨ㅇ⟩ is dropped.
Africanist Korean Romanization is a realization of that bureosseugi idea. It separates ⟨ㅇ⟩ into its two ancestors: the always silent ⟨ㅇ⟩ and the obsolete ⟨ㆁ⟩ letter standing for its /ŋ/ sound.
Revised Romanization is the current official standard for Korean romaja.
Revised Romanization is full of digraphs, such as eo ⟨ㅓ⟩, eu ⟨ㅡ⟩, ae ⟨ㅐ⟩, ch ⟨ㅊ⟩, ng ⟨ㅇ⟩, and ⟨ㅃ⟩ pp. This necessitates the use of hyphens to break up ambiguous sequences: for example, is aeo meant to be ⟨ㅏ어⟩ a-eo or ⟨ㅐ오⟩ ae-o? Is Gangwon meant to be Gang-won ⟨강원⟩ or Gan-gwon ⟨간권⟩?
By using more unique letters, Africanist Romaja reduces this ambiguity to a bare minimum. (aeo would be clearly disambiguated as aɔ ⟨ㅏ어⟩ and ɛo ⟨ㅐ오⟩.) The apostrophe is only ever used as the initial ⟨ㅇ⟩.
Unlike Revised Romanization, Africanist Romaja transcribes the spelling in hangul instead of pronunciation, potentially making Africanist Romaja a good learning aid in studying Korean word derivation rules and batchim sound changes.
Unlike in Revised Romanization, where “ㄹ” is ⟨r⟩ when initial and ⟨l⟩ when final, here “ㄹ” is always ⟨l⟩.
By directly representing the hangul spelling, Africanist Romaja does not directly represent the pronunciation of the word. The reader must apply the sound change rules in their minds, which are equal to that in hangul.
Its use of special characters requires the use of a specialized international keyboard. Shona no longer uses ⟨Ȿ ȿ⟩, making it currently unused by any natural language. Even without ⟨Ȿ ȿ⟩, no one natural language uses all the remaining letters of Africanist Romaja. Only specialist fonts, such as Andika used on this page, cover all the letters of Africanist Romaja.
It might be worth trying to make a variant of this system with more commonly used special characters, such as ⟨è⟩ for ⟨ɛ⟩.
UC | lc | Hangul | Notes |
---|---|---|---|
A | a | ㅏ | |
B | b | ㅂ | |
C | c | ㅊ | |
Ƈ | ƈ | ㅉ | Serer /ʄ̊/ |
D | d | ㄷ | |
E | e | ㅔ | |
Ɛ | ɛ | ㅐ | many African languages |
G | g | ㄱ | |
H | h | ㅎ | |
I | i | ㅣ | |
J | j | ㅈ | |
K | k | ㅋ | |
Ƙ | ƙ | ㄲ | Hausa /k’/, Pan-Nigerian Alphabet |
L | l | ㄹ | |
M | m | ㅁ | |
N | n | ㄴ | |
Ŋ | ŋ | ㅇ (batchim only) | many African languages |
O | o | ㅗ | |
Ɔ | ɔ | ㅓ | many African languages |
Ø | ø | ㅚ only | Based on Sámi /oe/ diphthong |
P | p | ㅍ | |
Ƥ | ƥ | ㅃ | Serer /ɓ̥/ |
S | s | ㅅ | |
Ȿ | ȿ | ㅆ | Formerly Shona for /sʷ/, now “sv” |
T | t | ㅌ | |
Ƭ | ƭ | ㄸ | Serer /ɗ̥/ (African Reference Alphabet) |
U | u | ㅜ | |
Ɥ | ɥ | ㅡ | Dan language /ɯ/ |
W | w | (ㅘ, ㅝ, etc.) | |
Y | y | (ㅑ, ㅕ, etc.) | |
’ | ’ | (ㅇ after batchim) |
한굴 | UC | lc |
---|---|---|
ㄱ | G | g |
ㄴ | N | n |
ㄷ | D | d |
ㄹ | L | l |
ㅁ | M | m |
ㅂ | B | b |
ㅅ | S | s |
ㅇ | (see “Ieung rules”) | |
ㅈ | J | j |
ㅊ | C | c |
ㅋ | K | k |
ㅌ | T | t |
ㅍ | P | p |
ㅎ | H | h |
ㄲ | Ƙ | ƙ |
ㄸ | Ƭ | ƭ |
ㅃ | Ƥ | ƥ |
ㅆ | Ȿ | ȿ |
ㅉ | Ƈ | ƈ |
한굴 | UC | lc |
---|---|---|
ㅏ | A | a |
ㅐ | Ɛ | ɛ |
ㅑ | YA | ya |
ㅒ | YƐ | yɛ |
ㅓ | Ɔ | ɔ |
ㅔ | E | e |
ㅕ | YƆ | yɔ |
ㅖ | YE | ye |
ㅗ | O | o |
ㅘ | WA | wa |
ㅙ | WƐ | wɛ |
ㅚ* | Ø | ø |
ㅛ | YO | yo |
ㅜ | U | u |
ㅝ | WƆ | wɔ |
ㅞ | WE | we |
ㅟ | WI | wi |
ㅠ | YU | yu |
ㅡ | Ɥ | ɥ |
ㅢ | ꞍI | ɥi |
ㅣ | I | i |
ㅡ이* | Ɥ’I | ɥ’i |
Situation | UC | lc |
---|---|---|
Final consonant /ŋ/ (batchim) | Ŋ | ŋ |
After a final consonant (batchim) that is not “ㅇ” /ŋ/ | ’ | ’ |
In the sequence “ㅡ이” | ’ | ’ |
After a vowel or at the start of a word | nothing |
This tool is powered by hangul-js
.
This tool will capitalize the first word of each sentence. If you need to capitalize other letters, please use the “Toggle case of selected” button.
This affects the output text box in the previous section, “Convert to/from hangul”.
*Please note that I don’t fully understand batcim assimilation rules either, so this guide may be inaccurate.
A batcim is a consonant placed at the end of a syllable. It means “support” or “prop” in Korean because it is written at the bottom of the syllable block in hangɥl.
The batcim consonant is also called joŋsɔŋ in/about Korean, and a “coda consonant” generally.
Some letters sound different when it becomes a batcim.
In hangɥl, ŋ is written using the dummy consonant ⟨ㅇ⟩. This is because, historically, they were two different letters with a different sound: ⟨ㅇ⟩ (silent) and ⟨ㆁ⟩ ŋ. Because you could just leave the ⟨ㅇ⟩ out at the end of a syllable block, and the sound ŋ did not appear at the start of a syllable, these two letters with similar shapes merged into one letter with two contextual sounds, one as a cosɔŋ (initial consonant) and another as a joŋsɔŋ (final consonant).
If the next syllable starts with a vowel (including y or w), you’re already there! If there is an apostrophe, simply remove it.
Certain Korean words with d’i and t’i change to sound like j’i and c’i.
The only tense consonants that can be batcim are ƙ and ȿ. They confer their tenseness to the following consonant, or just sound like themselves if it is before a vowel.
If the batcim forms a cluster containing h, it confers the aspiration to the consonant before or after it and then disappears.
As an exception, l and n do not have an aspirated form, so it just erases the h.
Some syllables have two consonants at the end.
In isolation or at the end of a word, the “higher-ranking” consonant is pronounced, while the “lower-ranking” consonant is ignored. The “rank” of a consonant is encoded in hangɥl consonant order: g > n > d > l > m > b ...
The list of valid double batcim are: lg, nj, nh, lb, lh, lm, bs, gs, ls, lt, and lp. The first 7 are the most common.
Before another consonant (i.e. there are three consonant letters in a row), discard the “lower-ranking” consonant, then apply regular sound changes. Exceptions apply, however:
Before an apostrophe, simply remove the apostrophe.
Some combinations between g/k and n/m cause one of them to turn into ŋ.
Some combinations between n/b/p and n/m/b cause one of them to turn into m.
All batcim pronounced like t turn into n before n or m.
l wins against n. The clusters -ln- and -nl- both turn into -ll-.
l draws against g. When they cluster, l turns into n and g turns into ŋ.
Surprisingly, l loses against m. The clusters -ml- and -lm- cause the l to turn into n.
When clustering against an unaspirated consonant, it causes that consonant to be tense.
When an unaspirated, aspirated, or tense consonant clusters with an unaspirated consonant, the consonant on the onset becomes tense.
Only when no other rules apply, clusters ending in n turn into nn.