]> git.saurik.com Git - apple/icu.git/blob - icuSources/data/translit/und_FONIPA_und_FONXSAMP.txt
ICU-62107.0.1.tar.gz
[apple/icu.git] / icuSources / data / translit / und_FONIPA_und_FONXSAMP.txt
1 # © 2016 and later: Unicode, Inc. and others.
2 # License & terms of use: http://www.unicode.org/copyright.html#License
3 #
4 # File: und_FONIPA_und_FONXSAMP.txt
5 # Generated from CLDR
6 #
7
8 # Conversion between IPA and X-SAMPA phonetic transcription.
9 #
10 # See http://www.phon.ucl.ac.uk/home/sampa/x-sampa.htm for a description of
11 # X-SAMPA, an ASCII encoding of the International Phonetic Alphabet.
12 #
13 # The following obsolete or extended IPA symbols have no X-SAMPA equivalents
14 # and remain unaffected by this transform:
15 #
16 # ʞ LATIN SMALL LETTER TURNED K
17 # ʩ LATIN SMALL LETTER FENG DIGRAPH
18 # ʪ LATIN SMALL LETTER LS DIGRAPH
19 # ʫ LATIN SMALL LETTER LZ DIGRAPH
20 # ʬ LATIN LETTER BILABIAL PERCUSSIVE
21 # ʭ LATIN LETTER BIDENTAL PERCUSSIVE
22 #
23 # An IPA tie bar is transformed to an X-SAMPA underscore, per the official
24 # X-SAMPA guidelines. This can result in certain ambiguities: For example, the
25 # labial-velar nasal (http://en.wikipedia.org/wiki/Labial-velar_nasal) can be
26 # either written as [ŋ\u0361m] or [m\u0361ŋ] in IPA. However, neither version can be
27 # represented unambiguously in X-SAMPA: IPA [ŋ\u0361m] becomes X-SAMPA [N_m], which
28 # is also used to represent a hypothetical laminal (_m) velar nasal, IPA [ŋ\u033B];
29 # and IPA [m\u0361ŋ] becomes X-SAMPA [m_N], which can also represent a linguolabial
30 # (_N) nasal, IPA [m\u033C], which is more appropriately written [n\u033C]. To avoid
31 # unintended ambiguities, it may therefore be advisable to write affricates
32 # without tie bars.
33 #
34 $t = '_'; # X-SAMPA representation of IPA tie bar.
35 ::NFD(NFC);
36 # 5-character X-SAMPA representations
37 ʯ ↔ 'z`_w='; # LATIN SMALL LETTER TURNED H WITH FISHHOOK AND TAIL
38 # Alternative 5-character X-SAMPA representations
39 n\u031F\u030A ← 'n_+_0';
40 n\u0320\u030A ← 'n_-_0';
41 n\u032A\u030A ← 'n_d_0';
42 n\u033A\u030A ← 'n_a_0';
43 n\u033B\u030A ← 'n_m_0';
44 n\u033C\u030A ← 'n_N_0';
45 ɻ\u030A ← 'r\`_0';
46 # 4-character X-SAMPA representations
47 ǁ ↔ '|\|\'; # LATIN LETTER LATERAL CLICK
48 ʄ ↔ 'J\_<'; # LATIN SMALL LETTER DOTLESS J WITH STROKE AND HOOK
49 ʛ ↔ 'G\_<'; # LATIN LETTER SMALL CAPITAL G WITH HOOK
50 ʮ ↔ 'z_w='; # LATIN SMALL LETTER TURNED H WITH FISHHOOK
51 \u1DC4 ↔ '_H_T'; # COMBINING MACRON-ACUTE
52 \u1DC5 ↔ '_B_L'; # COMBINING GRAVE-MACRON
53 \u1DC8 ↔ '_R_F'; # COMBINING GRAVE-ACUTE-GRAVE
54 # Alternative 4-character X-SAMPA representations
55 ɭ\u030A ← 'l`_0';
56 ɰ\u030A ← 'M\_0';
57 ɳ\u030A ← 'n`_0';
58 ɽ\u030A ← 'r`_0';
59 # 3-character X-SAMPA representations
60 ɓ ↔ 'b_<'; # LATIN SMALL LETTER B WITH HOOK
61 ɗ ↔ 'd_<'; # LATIN SMALL LETTER D WITH HOOK
62 ɠ ↔ 'g_<'; # LATIN SMALL LETTER G WITH HOOK
63 ɻ ↔ 'r\`'; # LATIN SMALL LETTER TURNED R WITH HOOK
64 ↗ ↔ '<R>'; # NORTH EAST ARROW
65 ↘ ↔ '<F>'; # SOUTH EAST ARROW
66 # Alternative 3-character X-SAMPA representations
67 j\u030A ← 'j_0';
68 ŋ\u030A ← 'N_0';
69 ɥ\u030A ← 'H_0';
70 ɱ\u030A ← 'F_0';
71 ɲ\u030A ← 'J_0';
72 # 2-character X-SAMPA representations
73 ħ ↔ 'X\'; # LATIN SMALL LETTER H WITH STROKE
74 ǀ ↔ '|\'; # LATIN LETTER DENTAL CLICK
75 ǂ ↔ '=\'; # LATIN LETTER ALVEOLAR CLICK
76 ǃ ↔ '!\'; # LATIN LETTER RETROFLEX CLICK
77 ɕ ↔ 's\'; # LATIN SMALL LETTER C WITH CURL
78 ɖ ↔ 'd`'; # LATIN SMALL LETTER D WITH TAIL
79 ɘ ↔ '@\'; # LATIN SMALL LETTER REVERSED E
80 ɚ ↔ '@`'; # LATIN SMALL LETTER SCHWA WITH HOOK
81 ɝ ↔ '3`'; # LATIN SMALL LETTER REVERSED OPEN E WITH HOOK
82 ɞ ↔ '3\'; # LATIN SMALL LETTER CLOSED REVERSED OPEN E
83 ɟ ↔ 'J\'; # LATIN SMALL LETTER DOTLESS J WITH STROKE
84 ɢ ↔ 'G\'; # LATIN LETTER SMALL CAPITAL G
85 ɦ ↔ 'h\'; # LATIN SMALL LETTER H WITH HOOK
86 ɧ ↔ 'x\'; # LATIN SMALL LETTER HENG WITH HOOK
87 ɭ ↔ 'l`'; # LATIN SMALL LETTER L WITH RETROFLEX HOOK
88 ɮ ↔ 'K\'; # LATIN SMALL LETTER LEZH
89 ɰ ↔ 'M\'; # LATIN SMALL LETTER TURNED M WITH LONG LEG
90 ɳ ↔ 'n`'; # LATIN SMALL LETTER N WITH RETROFLEX HOOK
91 ɴ ↔ 'N\'; # LATIN LETTER SMALL CAPITAL N
92 ɸ ↔ 'p\'; # LATIN SMALL LETTER PHI
93 ɹ ↔ 'r\'; # LATIN SMALL LETTER TURNED R
94 ɺ ↔ 'l\'; # LATIN SMALL LETTER TURNED R WITH LONG LEG
95 ɽ ↔ 'r`'; # LATIN SMALL LETTER R WITH TAIL
96 ʀ ↔ 'R\'; # LATIN LETTER SMALL CAPITAL R
97 ʂ ↔ 's`'; # LATIN SMALL LETTER S WITH HOOK
98 ʈ ↔ 't`'; # LATIN SMALL LETTER T WITH RETROFLEX HOOK
99 ʐ ↔ 'z`'; # LATIN SMALL LETTER Z WITH RETROFLEX HOOK
100 ʑ ↔ 'z\'; # LATIN SMALL LETTER Z WITH CURL
101 ʕ ↔ '?\'; # LATIN LETTER PHARYNGEAL VOICED FRICATIVE
102 ʘ ↔ 'O\'; # LATIN LETTER BILABIAL CLICK
103 ʙ ↔ 'B\'; # LATIN LETTER SMALL CAPITAL B
104 ʜ ↔ 'H\'; # LATIN LETTER SMALL CAPITAL H
105 ʝ ↔ 'j\'; # LATIN SMALL LETTER J WITH CROSSED-TAIL
106 ʟ ↔ 'L\'; # LATIN LETTER SMALL CAPITAL L
107 ʡ ↔ '>\'; # LATIN LETTER GLOTTAL STOP WITH STROKE
108 ʢ ↔ '<\'; # LATIN LETTER REVERSED GLOTTAL STOP WITH STROKE
109 ʰ ↔ '_h'; # MODIFIER LETTER SMALL H
110 ʷ ↔ '_w'; # MODIFIER LETTER SMALL W
111 ʼ ↔ '_>'; # MODIFIER LETTER APOSTROPHE
112 ˆ ↔ '_\'; # MODIFIER LETTER CIRCUMFLEX ACCENT
113 ˇ ↔ '_/'; # CARON
114 ˑ ↔ ':\'; # MODIFIER LETTER HALF TRIANGULAR COLON
115 ˠ ↔ '_G'; # MODIFIER LETTER SMALL GAMMA
116 ˡ ↔ '_l'; # MODIFIER LETTER SMALL L
117 ˤ ↔ '_?\'; # MODIFIER LETTER SMALL REVERSED GLOTTAL STOP
118 \u0300 ↔ '_L'; # COMBINING GRAVE ACCENT
119 \u0301 ↔ '_H'; # COMBINING ACUTE ACCENT
120 \u0302 ↔ '_F'; # COMBINING CIRCUMFLEX ACCENT
121 \u0304 ↔ '_M'; # COMBINING MACRON
122 \u0306 ↔ '_X'; # COMBINING BREVE
123 \u0308 ↔ '_"'; # COMBINING DIAERESIS
124 \u030B ↔ '_T'; # COMBINING DOUBLE ACUTE ACCENT
125 \u030C ↔ '_R'; # COMBINING CARON
126 \u030F ↔ '_B'; # COMBINING DOUBLE GRAVE ACCENT
127 \u0318 ↔ '_A'; # COMBINING LEFT TACK BELOW
128 \u0319 ↔ '_q'; # COMBINING RIGHT TACK BELOW
129 \u031A ↔ '_}'; # COMBINING LEFT ANGLE ABOVE
130 \u031C ↔ '_c'; # COMBINING LEFT HALF RING BELOW
131 \u031D ↔ '_r'; # COMBINING UP TACK BELOW
132 \u031E ↔ '_o'; # COMBINING DOWN TACK BELOW
133 \u031F ↔ '_+'; # COMBINING PLUS SIGN BELOW
134 \u0320 ↔ '_-'; # COMBINING MINUS SIGN BELOW
135 \u0324 ↔ '_t'; # COMBINING DIAERESIS BELOW
136 \u0325 ↔ '_0'; # COMBINING RING BELOW
137 \u032A ↔ '_d'; # COMBINING BRIDGE BELOW
138 \u032C ↔ '_v'; # COMBINING CARON BELOW
139 \u032F ↔ '_^'; # COMBINING INVERTED BREVE BELOW
140 \u0330 ↔ '_k'; # COMBINING TILDE BELOW
141 \u0334 ↔ '_e'; # COMBINING TILDE OVERLAY
142 \u0339 ↔ '_O'; # COMBINING RIGHT HALF RING BELOW
143 \u033A ↔ '_a'; # COMBINING INVERTED BRIDGE BELOW
144 \u033B ↔ '_m'; # COMBINING SQUARE BELOW
145 \u033C ↔ '_N'; # COMBINING SEAGULL BELOW
146 \u033D ↔ '_x'; # COMBINING X ABOVE
147 ᵻ ↔ 'I\'; # LATIN SMALL CAPITAL LETTER I WITH STROKE
148 ᵿ ↔ 'U\'; # LATIN SMALL CAPITAL LETTER U WITH STROKE
149 ⁿ ↔ '_n'; # MODIFIER LETTER LATIN SMALL LETTER N
150 # Alternative 2-character X-SAMPA representations
151 ʋ ← 'v\'; # LATIN SMALL LETTER V WITH HOOK
152 ʲ ← '_j'; # MODIFIER LETTER SMALL J
153 \u0303 ← '_~'; # COMBINING TILDE
154 \u0329 ← '_='; # COMBINING VERTICAL LINE BELOW
155 # 1-character X-SAMPA representations
156 c\u0327 ↔ C; # LATIN SMALL LETTER C WITH CEDILLA (decomposed)
157 æ ↔ '{'; # LATIN SMALL LETTER AE
158 ð ↔ D; # LATIN SMALL LETTER ETH
159 ø ↔ 2; # LATIN SMALL LETTER O WITH STROKE
160 ŋ ↔ N; # LATIN SMALL LETTER ENG
161 œ ↔ 9; # LATIN SMALL LIGATURE OE
162 ɐ ↔ 6; # LATIN SMALL LETTER TURNED A
163 ɑ ↔ A; # LATIN SMALL LETTER ALPHA
164 ɒ ↔ Q; # LATIN SMALL LETTER TURNED ALPHA
165 ɔ ↔ O; # LATIN SMALL LETTER OPEN O
166 ə ↔ '@'; # LATIN SMALL LETTER SCHWA
167 ɛ ↔ E; # LATIN SMALL LETTER OPEN E
168 ɜ ↔ 3; # LATIN SMALL LETTER REVERSED OPEN E
169 ɡ ↔ g; # LATIN SMALL LETTER SCRIPT G
170 ɣ ↔ G; # LATIN SMALL LETTER GAMMA
171 ɤ ↔ 7; # LATIN SMALL LETTER RAMS HORN
172 ɥ ↔ H; # LATIN SMALL LETTER TURNED H
173 ɨ ↔ 1; # LATIN SMALL LETTER I WITH STROKE
174 ɪ ↔ I; # LATIN LETTER SMALL CAPITAL I
175 ɫ ↔ 5; # LATIN SMALL LETTER L WITH MIDDLE TILDE
176 ɬ ↔ K; # LATIN SMALL LETTER L WITH BELT
177 ɯ ↔ M; # LATIN SMALL LETTER TURNED M
178 ɱ ↔ F; # LATIN SMALL LETTER M WITH HOOK
179 ɲ ↔ J; # LATIN SMALL LETTER N WITH LEFT HOOK
180 ɵ ↔ 8; # LATIN SMALL LETTER BARRED O
181 ɶ ↔ '&'; # LATIN LETTER SMALL CAPITAL OE
182 ɾ ↔ 4; # LATIN SMALL LETTER R WITH FISHHOOK
183 ʁ ↔ R; # LATIN LETTER SMALL CAPITAL INVERTED R
184 ʃ ↔ S; # LATIN SMALL LETTER ESH
185 ʉ ↔ '}'; # LATIN SMALL LETTER U BAR
186 ʊ ↔ U; # LATIN SMALL LETTER UPSILON
187 ʋ ↔ P; # LATIN SMALL LETTER V WITH HOOK
188 ʌ ↔ V; # LATIN SMALL LETTER TURNED V
189 ʍ ↔ W; # LATIN SMALL LETTER TURNED W
190 ʎ ↔ L; # LATIN SMALL LETTER TURNED Y
191 ʏ ↔ Y; # LATIN LETTER SMALL CAPITAL Y
192 ʒ ↔ Z; # LATIN SMALL LETTER EZH
193 ʔ ↔ '?'; # LATIN LETTER GLOTTAL STOP
194 ʲ ↔ \'; # MODIFIER LETTER SMALL J
195 ˈ ↔ '"'; # MODIFIER LETTER VERTICAL LINE
196 ˌ ↔ '%'; # MODIFIER LETTER LOW VERTICAL LINE
197 ː ↔ ':'; # MODIFIER LETTER TRIANGULAR COLON
198 ˞ ↔ '`'; # MODIFIER LETTER RHOTIC HOOK
199 \u0303 ↔ '~'; # COMBINING TILDE
200 \u0329 ↔ '='; # COMBINING VERTICAL LINE BELOW
201 \u0361 ↔ $t; # COMBINING DOUBLE INVERTED BREVE
202 β ↔ B; # GREEK SMALL LETTER BETA
203 θ ↔ T; # GREEK SMALL LETTER THETA
204 χ ↔ X; # GREEK SMALL LETTER CHI
205 ↑ ↔ '^'; # UPWARDS ARROW
206 ↓ ↔ '!'; # DOWNWARDS ARROW
207 # Compatibility rules for variant or obsolete IPA symbols
208 g → g; # LATIN SMALL LETTER G (redundant, for additional clarity)
209 ȵ → J; # LATIN SMALL LETTER N WITH CURL
210 ɩ → I; # LATIN SMALL LETTER IOTA
211 ɷ → U; # LATIN SMALL LETTER CLOSED OMEGA
212 ɼ → 'r_r'; # LATIN SMALL LETTER R WITH LONG LEG
213 ɿ → 'z='; # LATIN SMALL LETTER REVERSED R WITH FISHHOOK
214 ʅ → 'z`='; # LATIN SMALL LETTER SQUAT REVERSED ESH
215 ʆ → S\'; # LATIN SMALL LETTER ESH WITH CURL
216 ʇ → '|\' ; # LATIN SMALL LETTER TURNED T
217 ʓ → Z\'; # LATIN SMALL LETTER EZH WITH CURL
218 ʖ → '|\|\'; # LATIN LETTER INVERTED GLOTTAL STOP
219 ʗ → '!\'; # LATIN LETTER STRETCHED C
220 ʚ → '3\'; # LATIN SMALL LETTER CLOSED OPEN E
221 ʠ → 'G\_<_0'; # LATIN SMALL LETTER Q WITH HOOK
222 ʣ → d $t z; # LATIN SMALL LETTER DZ DIGRAPH
223 ʤ → d $t Z; # LATIN SMALL LETTER DEZH DIGRAPH
224 ʥ → d $t 'z\'; # LATIN SMALL LETTER DZ DIGRAPH WITH CURL
225 ʦ → t $t s; # LATIN SMALL LETTER TS DIGRAPH
226 ʧ → t $t S; # LATIN SMALL LETTER TESH DIGRAPH
227 ʨ → t $t 's\'; # LATIN SMALL LETTER TC DIGRAPH WITH CURL
228 ˔ → '_r'; # MODIFIER LETTER UP TACK
229 ˕ → '_o'; # MODIFIER LETTER DOWN TACK
230 \u030A → '_0'; # COMBINING RING ABOVE
231 φ → 'p\'; # GREEK SMALL LETTER PHI
232 ꞎ → 'K`'; # LATIN SMALL LETTER L WITH RETROFLEX HOOK AND BELT
233 ::NFC(NFD);
234