icuSources/data/translit/ja_Latn_ko.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: ja_Latn_ko.txt
   5 # Generated from CLDR
   6 #
   7
   8 # Japanese (Rōmaji) to Korean (Hangul) transliteration table for ICU.
   9 # Can be run in sequence after e.g. Katakana-Latin.
  10 #
  11 # Based on 문교부 고시 제85-11호 (1986. 1. 7.) 외래어 표기법
  12 # For background info, see http://ko.wikisource.org/wiki/문교부_고시_제85-11호
  13 # and http://ko.wikipedia.org/wiki/외래어_표기법 (外來語表記法)
  14 ::NFD(NFC);
  15 ::[:Latin:] Lower();
  16 $consonant = [ᄀᄁᄂᄃᄄᄅᄆᄇᄈᄉᄊᄋᄌᄍᄎᄏᄐᄑᄒ];
  17 $lengthMarker = [\u0302\u0304];
  18 #
  19 #
  20 # Drop hyphens and apostrophes.
  21 [\-\'] → ;
  22 #
  23 #
  24 # Turn long /e:/ into diphthong /ei/.
  25 e $lengthMarker → | e i ;
  26 #
  27 #
  28 # Ignore vowel length everywhere else.
  29 $lengthMarker → ;
  30 #
  31 #
  32 # Vowels.
  33 #
  34 [^$consonant] { ( [aiueoyw] ) → ᄋ | $1 ;  # Supply a required null initial.
  35 a → ᅡ ;
  36 i\~e → | ie ;  # イェ
  37 i → ᅵ ;
  38 u\~a → | wa ;  # クァ, グァ
  39 u\~i → ᅱ ;     # ウィ, クィ, etc.
  40 u\~e → ᅰ ;     # ウェ
  41 u\~o → ᅯ ;     # ウォ
  42 u → ᅮ ;
  43 e → ᅦ ;
  44 o → ᅩ ;
  45 #
  46 #
  47 # Geminates.
  48 #
  49 kk → ᆺ | k ;
  50 ss → ᆺ | s ;
  51 tt → ᆺ | t ;
  52 tc → ᆺ | c ;
  53 cc → ᆺ | c ;
  54 hh → ᆺ | h ;
  55 ff → ᆺ | f ;
  56 rr → ᆺ | r ;
  57 gg → ᆺ | g ;
  58 zz → ᆺ | z ;
  59 jj → ᆺ | j ;
  60 dd → ᆺ | d ;
  61 bb → ᆺ | b ;
  62 vv → ᆺ | v ;
  63 pp → ᆺ | p ;
  64 #
  65 #
  66 # Consonants.
  67 #
  68 ' ' { k → | g ;   # Beginning of a word (after space).
  69 ^k      → | g ;   # Beginning of the string.
  70 k → ᄏ ;
  71 #
  72 #
  73 sh → | sy ;
  74 su → 스 ;
  75 s → ᄉ ;
  76 #
  77 #
  78 te\~     → | t ;   # テュ
  79 to\~     → | t ;   # トゥ
  80 tsu\~    → | ch ;  # ツァ, ツィ, etc.
  81 tsu → 쓰 ;
  82 ts       → | ch ;
  83 ' ' { t  → | d ;
  84 ^t       → | d ;
  85 t → ᄐ ;
  86 ' ' { ch → | j ;
  87 ^ch      → | j ;
  88 ch → ᄎ ;
  89 #
  90 #
  91 n } [\ \'bcdfghjkmnprstwz] → ᆫ ;
  92 n$ → ᆫ ;
  93 n → ᄂ ;
  94 #
  95 #
  96 h → ᄒ ;
  97 fu\~ → | p ;   # フュ
  98 fu   → | hu ;
  99 f    → | p ;
 100 #
 101 #
 102 m } [bmp] → ᆫ ;
 103 m → ᄆ ;
 104 #
 105 #
 106 ya → ᅣ ;
 107 yi → ᅵ ;  # Added for convenience, after shi.
 108 yu → ᅲ ;
 109 ye → ᅨ ;
 110 yo → ᅭ ;
 111 #
 112 #
 113 r → ᄅ ;
 114 #
 115 #
 116 wa → ᅪ ;
 117 w → ;
 118 #
 119 #
 120 g → ᄀ ;
 121 #
 122 #
 123 zu → 즈 ;
 124 z → | j ;
 125 j → ᄌ ;
 126 #
 127 #
 128 de\~  → | d ;   # デュ
 129 dji\~ → | j ;   # ヂァ, ヂゥ, etc.
 130 dji   → | ji ;  # ヂ
 131 do\~  → | d ;   # ドゥ
 132 dzu\~ → | j ;   # ヅァ, ヅィ, etc.
 133 dzu   → | zu ;  # ヅ
 134 dz    → | j ;
 135 d → ᄃ ;
 136 #
 137 #
 138 b → ᄇ ;
 139 vu\~ → | b ;  # ヴァ, etc.
 140 v    → | b ;
 141 #
 142 #
 143 p → ᄑ ;
 144 #
 145 #
 146 ::NFC(NFD);
 147