icuSources/data/translit/und_FONIPA_und_FONXSAMP.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: und_FONIPA_und_FONXSAMP.txt
   8 # Generated from CLDR
   9 #
  10
  11 # Conversion between IPA and X-SAMPA phonetic transcription.
  12 #
  13 # See http://www.phon.ucl.ac.uk/home/sampa/x-sampa.htm for a description of
  14 # X-SAMPA, an ASCII encoding of the International Phonetic Alphabet.
  15 #
  16 # The following obsolete or extended IPA symbols have no X-SAMPA equivalents
  17 # and remain unaffected by this transform:
  18 #
  19 #   ʞ  LATIN SMALL LETTER TURNED K
  20 #   ʩ  LATIN SMALL LETTER FENG DIGRAPH
  21 #   ʪ  LATIN SMALL LETTER LS DIGRAPH
  22 #   ʫ  LATIN SMALL LETTER LZ DIGRAPH
  23 #   ʬ  LATIN LETTER BILABIAL PERCUSSIVE
  24 #   ʭ  LATIN LETTER BIDENTAL PERCUSSIVE
  25 #
  26 # An IPA tie bar is transformed to an X-SAMPA underscore, per the official
  27 # X-SAMPA guidelines.  This can result in certain ambiguities: For example, the
  28 # labial-velar nasal (http://en.wikipedia.org/wiki/Labial-velar_nasal) can be
  29 # either written as [ŋ\u0361m] or [m\u0361ŋ] in IPA.  However, neither version can be
  30 # represented unambiguously in X-SAMPA: IPA [ŋ\u0361m] becomes X-SAMPA [N_m], which
  31 # is also used to represent a hypothetical laminal (_m) velar nasal, IPA [ŋ\u033B];
  32 # and IPA [m\u0361ŋ] becomes X-SAMPA [m_N], which can also represent a linguolabial
  33 # (_N) nasal, IPA [m\u033C], which is more appropriately written [n\u033C].  To avoid
  34 # unintended ambiguities, it may therefore be advisable to write affricates
  35 # without tie bars.
  36 #
  37 $t = '_';  # X-SAMPA representation of IPA tie bar.
  38 ::NFD(NFC);
  39 # 5-character X-SAMPA representations
  40 ʯ ↔ 'z`_w=';      # LATIN SMALL LETTER TURNED H WITH FISHHOOK AND TAIL
  41 # Alternative 5-character X-SAMPA representations
  42 n\u031F\u030A ← 'n_+_0';
  43 n\u0320\u030A ← 'n_-_0';
  44 n\u032A\u030A ← 'n_d_0';
  45 n\u033A\u030A ← 'n_a_0';
  46 n\u033B\u030A ← 'n_m_0';
  47 n\u033C\u030A ← 'n_N_0';
  48 ɻ\u030A ← 'r\`_0';
  49 # 4-character X-SAMPA representations
  50 ǁ ↔ '|\|\';       # LATIN LETTER LATERAL CLICK
  51 ʄ ↔ 'J\_<';    # LATIN SMALL LETTER DOTLESS J WITH STROKE AND HOOK
  52 ʛ ↔ 'G\_<';    # LATIN LETTER SMALL CAPITAL G WITH HOOK
  53 ʮ ↔ 'z_w=';       # LATIN SMALL LETTER TURNED H WITH FISHHOOK
  54 \u1DC4 ↔ '_H_T';  # COMBINING MACRON-ACUTE
  55 \u1DC5 ↔ '_B_L';  # COMBINING GRAVE-MACRON
  56 \u1DC8 ↔ '_R_F';  # COMBINING GRAVE-ACUTE-GRAVE
  57 # Alternative 4-character X-SAMPA representations
  58 ɭ\u030A ← 'l`_0';
  59 ɰ\u030A ← 'M\_0';
  60 ɳ\u030A ← 'n`_0';
  61 ɽ\u030A ← 'r`_0';
  62 # 3-character X-SAMPA representations
  63 ɓ ↔ 'b_<';     # LATIN SMALL LETTER B WITH HOOK
  64 ɗ ↔ 'd_<';     # LATIN SMALL LETTER D WITH HOOK
  65 ɠ ↔ 'g_<';     # LATIN SMALL LETTER G WITH HOOK
  66 ɻ ↔ 'r\`';        # LATIN SMALL LETTER TURNED R WITH HOOK
  67 ↗ ↔ '<R>';  # NORTH EAST ARROW
  68 ↘ ↔ '<F>';  # SOUTH EAST ARROW
  69 # Alternative 3-character X-SAMPA representations
  70 j\u030A ← 'j_0';
  71 ŋ\u030A ← 'N_0';
  72 ɥ\u030A ← 'H_0';
  73 ɱ\u030A ← 'F_0';
  74 ɲ\u030A ← 'J_0';
  75 # 2-character X-SAMPA representations
  76 ħ ↔ 'X\';      # LATIN SMALL LETTER H WITH STROKE
  77 ǀ ↔ '|\';      # LATIN LETTER DENTAL CLICK
  78 ǂ ↔ '=\';      # LATIN LETTER ALVEOLAR CLICK
  79 ǃ ↔ '!\';      # LATIN LETTER RETROFLEX CLICK
  80 ɕ ↔ 's\';      # LATIN SMALL LETTER C WITH CURL
  81 ɖ ↔ 'd`';      # LATIN SMALL LETTER D WITH TAIL
  82 ɘ ↔ '@\';      # LATIN SMALL LETTER REVERSED E
  83 ɚ ↔ '@`';      # LATIN SMALL LETTER SCHWA WITH HOOK
  84 ɝ ↔ '3`';      # LATIN SMALL LETTER REVERSED OPEN E WITH HOOK
  85 ɞ ↔ '3\';      # LATIN SMALL LETTER CLOSED REVERSED OPEN E
  86 ɟ ↔ 'J\';      # LATIN SMALL LETTER DOTLESS J WITH STROKE
  87 ɢ ↔ 'G\';      # LATIN LETTER SMALL CAPITAL G
  88 ɦ ↔ 'h\';      # LATIN SMALL LETTER H WITH HOOK
  89 ɧ ↔ 'x\';      # LATIN SMALL LETTER HENG WITH HOOK
  90 ɭ ↔ 'l`';      # LATIN SMALL LETTER L WITH RETROFLEX HOOK
  91 ɮ ↔ 'K\';      # LATIN SMALL LETTER LEZH
  92 ɰ ↔ 'M\';      # LATIN SMALL LETTER TURNED M WITH LONG LEG
  93 ɳ ↔ 'n`';      # LATIN SMALL LETTER N WITH RETROFLEX HOOK
  94 ɴ ↔ 'N\';      # LATIN LETTER SMALL CAPITAL N
  95 ɸ ↔ 'p\';      # LATIN SMALL LETTER PHI
  96 ɹ ↔ 'r\';      # LATIN SMALL LETTER TURNED R
  97 ɺ ↔ 'l\';      # LATIN SMALL LETTER TURNED R WITH LONG LEG
  98 ɽ ↔ 'r`';      # LATIN SMALL LETTER R WITH TAIL
  99 ʀ ↔ 'R\';      # LATIN LETTER SMALL CAPITAL R
 100 ʂ ↔ 's`';      # LATIN SMALL LETTER S WITH HOOK
 101 ʈ ↔ 't`';      # LATIN SMALL LETTER T WITH RETROFLEX HOOK
 102 ʐ ↔ 'z`';      # LATIN SMALL LETTER Z WITH RETROFLEX HOOK
 103 ʑ ↔ 'z\';      # LATIN SMALL LETTER Z WITH CURL
 104 ʕ ↔ '?\';      # LATIN LETTER PHARYNGEAL VOICED FRICATIVE
 105 ʘ ↔ 'O\';      # LATIN LETTER BILABIAL CLICK
 106 ʙ ↔ 'B\';      # LATIN LETTER SMALL CAPITAL B
 107 ʜ ↔ 'H\';      # LATIN LETTER SMALL CAPITAL H
 108 ʝ ↔ 'j\';      # LATIN SMALL LETTER J WITH CROSSED-TAIL
 109 ʟ ↔ 'L\';      # LATIN LETTER SMALL CAPITAL L
 110 ʡ ↔ '>\';      # LATIN LETTER GLOTTAL STOP WITH STROKE
 111 ʢ ↔ '<\';      # LATIN LETTER REVERSED GLOTTAL STOP WITH STROKE
 112 ʰ ↔ '_h';      # MODIFIER LETTER SMALL H
 113 ʷ ↔ '_w';      # MODIFIER LETTER SMALL W
 114 ʼ ↔ '_>';      # MODIFIER LETTER APOSTROPHE
 115 ˆ ↔ '_\';      # MODIFIER LETTER CIRCUMFLEX ACCENT
 116 ˇ ↔ '_/';      # CARON
 117 ˑ ↔ ':\';      # MODIFIER LETTER HALF TRIANGULAR COLON
 118 ˠ ↔ '_G';      # MODIFIER LETTER SMALL GAMMA
 119 ˡ ↔ '_l';      # MODIFIER LETTER SMALL L
 120 ˤ ↔ '_?\';     # MODIFIER LETTER SMALL REVERSED GLOTTAL STOP
 121 \u0300 ↔ '_L';    # COMBINING GRAVE ACCENT
 122 \u0301 ↔ '_H';    # COMBINING ACUTE ACCENT
 123 \u0302 ↔ '_F';    # COMBINING CIRCUMFLEX ACCENT
 124 \u0304 ↔ '_M';    # COMBINING MACRON
 125 \u0306 ↔ '_X';    # COMBINING BREVE
 126 \u0308 ↔ '_"';    # COMBINING DIAERESIS
 127 \u030B ↔ '_T';    # COMBINING DOUBLE ACUTE ACCENT
 128 \u030C ↔ '_R';    # COMBINING CARON
 129 \u030F ↔ '_B';    # COMBINING DOUBLE GRAVE ACCENT
 130 \u0318 ↔ '_A';    # COMBINING LEFT TACK BELOW
 131 \u0319 ↔ '_q';    # COMBINING RIGHT TACK BELOW
 132 \u031A ↔ '_}';    # COMBINING LEFT ANGLE ABOVE
 133 \u031C ↔ '_c';    # COMBINING LEFT HALF RING BELOW
 134 \u031D ↔ '_r';    # COMBINING UP TACK BELOW
 135 \u031E ↔ '_o';    # COMBINING DOWN TACK BELOW
 136 \u031F ↔ '_+';    # COMBINING PLUS SIGN BELOW
 137 \u0320 ↔ '_-';    # COMBINING MINUS SIGN BELOW
 138 \u0324 ↔ '_t';    # COMBINING DIAERESIS BELOW
 139 \u0325 ↔ '_0';    # COMBINING RING BELOW
 140 \u032A ↔ '_d';    # COMBINING BRIDGE BELOW
 141 \u032C ↔ '_v';    # COMBINING CARON BELOW
 142 \u032F ↔ '_^';    # COMBINING INVERTED BREVE BELOW
 143 \u0330 ↔ '_k';    # COMBINING TILDE BELOW
 144 \u0334 ↔ '_e';    # COMBINING TILDE OVERLAY
 145 \u0339 ↔ '_O';    # COMBINING RIGHT HALF RING BELOW
 146 \u033A ↔ '_a';    # COMBINING INVERTED BRIDGE BELOW
 147 \u033B ↔ '_m';    # COMBINING SQUARE BELOW
 148 \u033C ↔ '_N';    # COMBINING SEAGULL BELOW
 149 \u033D ↔ '_x';    # COMBINING X ABOVE
 150 ᵻ ↔ 'I\';      # LATIN SMALL CAPITAL LETTER I WITH STROKE
 151 ᵿ ↔ 'U\';      # LATIN SMALL CAPITAL LETTER U WITH STROKE
 152 ⁿ ↔ '_n';      # MODIFIER LETTER LATIN SMALL LETTER N
 153 # Alternative 2-character X-SAMPA representations
 154 ʋ ← 'v\';         # LATIN SMALL LETTER V WITH HOOK
 155 ʲ ← '_j';         # MODIFIER LETTER SMALL J
 156 \u0303 ← '_~';    # COMBINING TILDE
 157 \u0329 ← '_=';    # COMBINING VERTICAL LINE BELOW
 158 # 1-character X-SAMPA representations
 159 c\u0327 ↔ C;      # LATIN SMALL LETTER C WITH CEDILLA (decomposed)
 160 æ ↔ '{';          # LATIN SMALL LETTER AE
 161 ð ↔ D;            # LATIN SMALL LETTER ETH
 162 ø ↔ 2;            # LATIN SMALL LETTER O WITH STROKE
 163 ŋ ↔ N;            # LATIN SMALL LETTER ENG
 164 œ ↔ 9;            # LATIN SMALL LIGATURE OE
 165 ɐ ↔ 6;            # LATIN SMALL LETTER TURNED A
 166 ɑ ↔ A;            # LATIN SMALL LETTER ALPHA
 167 ɒ ↔ Q;            # LATIN SMALL LETTER TURNED ALPHA
 168 ɔ ↔ O;            # LATIN SMALL LETTER OPEN O
 169 ə ↔ '@';          # LATIN SMALL LETTER SCHWA
 170 ɛ ↔ E;            # LATIN SMALL LETTER OPEN E
 171 ɜ ↔ 3;            # LATIN SMALL LETTER REVERSED OPEN E
 172 ɡ ↔ g;            # LATIN SMALL LETTER SCRIPT G
 173 ɣ ↔ G;            # LATIN SMALL LETTER GAMMA
 174 ɤ ↔ 7;            # LATIN SMALL LETTER RAMS HORN
 175 ɥ ↔ H;            # LATIN SMALL LETTER TURNED H
 176 ɨ ↔ 1;            # LATIN SMALL LETTER I WITH STROKE
 177 ɪ ↔ I;            # LATIN LETTER SMALL CAPITAL I
 178 ɫ ↔ 5;            # LATIN SMALL LETTER L WITH MIDDLE TILDE
 179 ɬ ↔ K;            # LATIN SMALL LETTER L WITH BELT
 180 ɯ ↔ M;            # LATIN SMALL LETTER TURNED M
 181 ɱ ↔ F;            # LATIN SMALL LETTER M WITH HOOK
 182 ɲ ↔ J;            # LATIN SMALL LETTER N WITH LEFT HOOK
 183 ɵ ↔ 8;            # LATIN SMALL LETTER BARRED O
 184 ɶ ↔ '&';          # LATIN LETTER SMALL CAPITAL OE
 185 ɾ ↔ 4;            # LATIN SMALL LETTER R WITH FISHHOOK
 186 ʁ ↔ R;            # LATIN LETTER SMALL CAPITAL INVERTED R
 187 ʃ ↔ S;            # LATIN SMALL LETTER ESH
 188 ʉ ↔ '}';          # LATIN SMALL LETTER U BAR
 189 ʊ ↔ U;            # LATIN SMALL LETTER UPSILON
 190 ʋ ↔ P;            # LATIN SMALL LETTER V WITH HOOK
 191 ʌ ↔ V;            # LATIN SMALL LETTER TURNED V
 192 ʍ ↔ W;            # LATIN SMALL LETTER TURNED W
 193 ʎ ↔ L;            # LATIN SMALL LETTER TURNED Y
 194 ʏ ↔ Y;            # LATIN LETTER SMALL CAPITAL Y
 195 ʒ ↔ Z;            # LATIN SMALL LETTER EZH
 196 ʔ ↔ '?';          # LATIN LETTER GLOTTAL STOP
 197 ʲ ↔ \';           # MODIFIER LETTER SMALL J
 198 ˈ ↔ '"';          # MODIFIER LETTER VERTICAL LINE
 199 ˌ ↔ '%';          # MODIFIER LETTER LOW VERTICAL LINE
 200 ː ↔ ':';          # MODIFIER LETTER TRIANGULAR COLON
 201 ˞ ↔ '`';          # MODIFIER LETTER RHOTIC HOOK
 202 \u0303 ↔ '~';     # COMBINING TILDE
 203 \u0329 ↔ '=';     # COMBINING VERTICAL LINE BELOW
 204 \u0361 ↔ $t;      # COMBINING DOUBLE INVERTED BREVE
 205 β ↔ B;            # GREEK SMALL LETTER BETA
 206 θ ↔ T;            # GREEK SMALL LETTER THETA
 207 χ ↔ X;            # GREEK SMALL LETTER CHI
 208 ↑ ↔ '^';          # UPWARDS ARROW
 209 ↓ ↔ '!';          # DOWNWARDS ARROW
 210 # Compatibility rules for variant or obsolete IPA symbols
 211 g → g;            # LATIN SMALL LETTER G (redundant, for additional clarity)
 212 ȵ → J;            # LATIN SMALL LETTER N WITH CURL
 213 ɩ → I;            # LATIN SMALL LETTER IOTA
 214 ɷ → U;            # LATIN SMALL LETTER CLOSED OMEGA
 215 ɼ → 'r_r';        # LATIN SMALL LETTER R WITH LONG LEG
 216 ɿ → 'z=';         # LATIN SMALL LETTER REVERSED R WITH FISHHOOK
 217 ʅ → 'z`=';        # LATIN SMALL LETTER SQUAT REVERSED ESH
 218 ʆ → S\';          # LATIN SMALL LETTER ESH WITH CURL
 219 ʇ → '|\' ;        # LATIN SMALL LETTER TURNED T
 220 ʓ → Z\';          # LATIN SMALL LETTER EZH WITH CURL
 221 ʖ → '|\|\';       # LATIN LETTER INVERTED GLOTTAL STOP
 222 ʗ → '!\';         # LATIN LETTER STRETCHED C
 223 ʚ → '3\';         # LATIN SMALL LETTER CLOSED OPEN E
 224 ʠ → 'G\_<_0';  # LATIN SMALL LETTER Q WITH HOOK
 225 ʣ → d $t z;       # LATIN SMALL LETTER DZ DIGRAPH
 226 ʤ → d $t Z;       # LATIN SMALL LETTER DEZH DIGRAPH
 227 ʥ → d $t 'z\';    # LATIN SMALL LETTER DZ DIGRAPH WITH CURL
 228 ʦ → t $t s;       # LATIN SMALL LETTER TS DIGRAPH
 229 ʧ → t $t S;       # LATIN SMALL LETTER TESH DIGRAPH
 230 ʨ → t $t 's\';    # LATIN SMALL LETTER TC DIGRAPH WITH CURL
 231 ˔ → '_r';         # MODIFIER LETTER UP TACK
 232 ˕ → '_o';         # MODIFIER LETTER DOWN TACK
 233 \u030A → '_0';    # COMBINING RING ABOVE
 234 φ → 'p\';         # GREEK SMALL LETTER PHI
 235 ꞎ → 'K`';         # LATIN SMALL LETTER L WITH RETROFLEX HOOK AND BELT
 236 ::NFC(NFD);
 237