icuSources/data/translit/ja_Latn_ko.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: ja_Latn_ko.txt
   8 # Generated from CLDR
   9 #
  10
  11 # Japanese (Rōmaji) to Korean (Hangul) transliteration table for ICU.
  12 # Can be run in sequence after e.g. Katakana-Latin.
  13 #
  14 # Based on 문교부 고시 제85-11호 (1986. 1. 7.) 외래어 표기법
  15 # For background info, see http://ko.wikisource.org/wiki/문교부_고시_제85-11호
  16 # and http://ko.wikipedia.org/wiki/외래어_표기법 (外來語表記法)
  17 ::NFD(NFC);
  18 ::[:Latin:] Lower();
  19 $consonant = [ᄀᄁᄂᄃᄄᄅᄆᄇᄈᄉᄊᄋᄌᄍᄎᄏᄐᄑᄒ];
  20 $lengthMarker = [\u0302\u0304];
  21 #
  22 #
  23 # Drop hyphens and apostrophes.
  24 [\-\'] → ;
  25 #
  26 #
  27 # Turn long /e:/ into diphthong /ei/.
  28 e $lengthMarker → | e i ;
  29 #
  30 #
  31 # Ignore vowel length everywhere else.
  32 $lengthMarker → ;
  33 #
  34 #
  35 # Vowels.
  36 #
  37 [^$consonant] { ( [aiueoyw] ) → ᄋ | $1 ;  # Supply a required null initial.
  38 a → ᅡ ;
  39 i\~e → | ie ;  # イェ
  40 i → ᅵ ;
  41 u\~a → | wa ;  # クァ, グァ
  42 u\~i → ᅱ ;     # ウィ, クィ, etc.
  43 u\~e → ᅰ ;     # ウェ
  44 u\~o → ᅯ ;     # ウォ
  45 u → ᅮ ;
  46 e → ᅦ ;
  47 o → ᅩ ;
  48 #
  49 #
  50 # Geminates.
  51 #
  52 kk → ᆺ | k ;
  53 ss → ᆺ | s ;
  54 tt → ᆺ | t ;
  55 tc → ᆺ | c ;
  56 cc → ᆺ | c ;
  57 hh → ᆺ | h ;
  58 ff → ᆺ | f ;
  59 rr → ᆺ | r ;
  60 gg → ᆺ | g ;
  61 zz → ᆺ | z ;
  62 jj → ᆺ | j ;
  63 dd → ᆺ | d ;
  64 bb → ᆺ | b ;
  65 vv → ᆺ | v ;
  66 pp → ᆺ | p ;
  67 #
  68 #
  69 # Consonants.
  70 #
  71 ' ' { k → | g ;   # Beginning of a word (after space).
  72 ^k      → | g ;   # Beginning of the string.
  73 k → ᄏ ;
  74 #
  75 #
  76 sh → | sy ;
  77 su → 스 ;
  78 s → ᄉ ;
  79 #
  80 #
  81 te\~     → | t ;   # テュ
  82 to\~     → | t ;   # トゥ
  83 tsu\~    → | ch ;  # ツァ, ツィ, etc.
  84 tsu → 쓰 ;
  85 ts       → | ch ;
  86 ' ' { t  → | d ;
  87 ^t       → | d ;
  88 t → ᄐ ;
  89 ' ' { ch → | j ;
  90 ^ch      → | j ;
  91 ch → ᄎ ;
  92 #
  93 #
  94 n } [\ \'bcdfghjkmnprstwz] → ᆫ ;
  95 n$ → ᆫ ;
  96 n → ᄂ ;
  97 #
  98 #
  99 h → ᄒ ;
 100 fu\~ → | p ;   # フュ
 101 fu   → | hu ;
 102 f    → | p ;
 103 #
 104 #
 105 m } [bmp] → ᆫ ;
 106 m → ᄆ ;
 107 #
 108 #
 109 ya → ᅣ ;
 110 yi → ᅵ ;  # Added for convenience, after shi.
 111 yu → ᅲ ;
 112 ye → ᅨ ;
 113 yo → ᅭ ;
 114 #
 115 #
 116 r → ᄅ ;
 117 #
 118 #
 119 wa → ᅪ ;
 120 w → ;
 121 #
 122 #
 123 g → ᄀ ;
 124 #
 125 #
 126 zu → 즈 ;
 127 z → | j ;
 128 j → ᄌ ;
 129 #
 130 #
 131 de\~  → | d ;   # デュ
 132 dji\~ → | j ;   # ヂァ, ヂゥ, etc.
 133 dji   → | ji ;  # ヂ
 134 do\~  → | d ;   # ドゥ
 135 dzu\~ → | j ;   # ヅァ, ヅィ, etc.
 136 dzu   → | zu ;  # ヅ
 137 dz    → | j ;
 138 d → ᄃ ;
 139 #
 140 #
 141 b → ᄇ ;
 142 vu\~ → | b ;  # ヴァ, etc.
 143 v    → | b ;
 144 #
 145 #
 146 p → ᄑ ;
 147 #
 148 #
 149 ::NFC(NFD);
 150