]> git.saurik.com Git - apple/icu.git/blame - icuSources/data/translit/und_FONIPA_und_FONXSAMP.txt
ICU-66108.tar.gz
[apple/icu.git] / icuSources / data / translit / und_FONIPA_und_FONXSAMP.txt
CommitLineData
f3c0d7a5
A
1# © 2016 and later: Unicode, Inc. and others.
2# License & terms of use: http://www.unicode.org/copyright.html#License
3#
2ca993e8 4# File: und_FONIPA_und_FONXSAMP.txt
f3c0d7a5 5# Generated from CLDR
2ca993e8
A
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
37ʯ ↔ 'z`_w='; # LATIN SMALL LETTER TURNED H WITH FISHHOOK AND TAIL
38# Alternative 5-character X-SAMPA representations
39n\u031F\u030A ← 'n_+_0';
40n\u0320\u030A ← 'n_-_0';
41n\u032A\u030A ← 'n_d_0';
42n\u033A\u030A ← 'n_a_0';
43n\u033B\u030A ← 'n_m_0';
44n\u033C\u030A ← 'n_N_0';
45ɻ\u030A ← 'r\`_0';
46# 4-character X-SAMPA representations
47ǁ ↔ '|\|\'; # LATIN LETTER LATERAL CLICK
48ʄ ↔ 'J\_<'; # LATIN SMALL LETTER DOTLESS J WITH STROKE AND HOOK
49ʛ ↔ 'G\_<'; # LATIN LETTER SMALL CAPITAL G WITH HOOK
50ʮ ↔ '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
55ɭ\u030A ← 'l`_0';
56ɰ\u030A ← 'M\_0';
57ɳ\u030A ← 'n`_0';
58ɽ\u030A ← 'r`_0';
59# 3-character X-SAMPA representations
60ɓ ↔ 'b_<'; # LATIN SMALL LETTER B WITH HOOK
61ɗ ↔ 'd_<'; # LATIN SMALL LETTER D WITH HOOK
62ɠ ↔ 'g_<'; # LATIN SMALL LETTER G WITH HOOK
63ɻ ↔ '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
67j\u030A ← 'j_0';
68ŋ\u030A ← 'N_0';
69ɥ\u030A ← 'H_0';
70ɱ\u030A ← 'F_0';
71ɲ\u030A ← 'J_0';
72# 2-character X-SAMPA representations
73ħ ↔ 'X\'; # LATIN SMALL LETTER H WITH STROKE
74ǀ ↔ '|\'; # LATIN LETTER DENTAL CLICK
75ǂ ↔ '=\'; # LATIN LETTER ALVEOLAR CLICK
76ǃ ↔ '!\'; # LATIN LETTER RETROFLEX CLICK
77ɕ ↔ 's\'; # LATIN SMALL LETTER C WITH CURL
78ɖ ↔ 'd`'; # LATIN SMALL LETTER D WITH TAIL
79ɘ ↔ '@\'; # LATIN SMALL LETTER REVERSED E
80ɚ ↔ '@`'; # LATIN SMALL LETTER SCHWA WITH HOOK
81ɝ ↔ '3`'; # LATIN SMALL LETTER REVERSED OPEN E WITH HOOK
82ɞ ↔ '3\'; # LATIN SMALL LETTER CLOSED REVERSED OPEN E
83ɟ ↔ 'J\'; # LATIN SMALL LETTER DOTLESS J WITH STROKE
84ɢ ↔ 'G\'; # LATIN LETTER SMALL CAPITAL G
85ɦ ↔ 'h\'; # LATIN SMALL LETTER H WITH HOOK
86ɧ ↔ 'x\'; # LATIN SMALL LETTER HENG WITH HOOK
87ɭ ↔ 'l`'; # LATIN SMALL LETTER L WITH RETROFLEX HOOK
88ɮ ↔ 'K\'; # LATIN SMALL LETTER LEZH
89ɰ ↔ 'M\'; # LATIN SMALL LETTER TURNED M WITH LONG LEG
90ɳ ↔ 'n`'; # LATIN SMALL LETTER N WITH RETROFLEX HOOK
91ɴ ↔ 'N\'; # LATIN LETTER SMALL CAPITAL N
92ɸ ↔ 'p\'; # LATIN SMALL LETTER PHI
93ɹ ↔ 'r\'; # LATIN SMALL LETTER TURNED R
94ɺ ↔ 'l\'; # LATIN SMALL LETTER TURNED R WITH LONG LEG
95ɽ ↔ 'r`'; # LATIN SMALL LETTER R WITH TAIL
96ʀ ↔ 'R\'; # LATIN LETTER SMALL CAPITAL R
97ʂ ↔ 's`'; # LATIN SMALL LETTER S WITH HOOK
98ʈ ↔ 't`'; # LATIN SMALL LETTER T WITH RETROFLEX HOOK
99ʐ ↔ 'z`'; # LATIN SMALL LETTER Z WITH RETROFLEX HOOK
100ʑ ↔ 'z\'; # LATIN SMALL LETTER Z WITH CURL
101ʕ ↔ '?\'; # LATIN LETTER PHARYNGEAL VOICED FRICATIVE
102ʘ ↔ 'O\'; # LATIN LETTER BILABIAL CLICK
103ʙ ↔ 'B\'; # LATIN LETTER SMALL CAPITAL B
104ʜ ↔ 'H\'; # LATIN LETTER SMALL CAPITAL H
105ʝ ↔ 'j\'; # LATIN SMALL LETTER J WITH CROSSED-TAIL
106ʟ ↔ 'L\'; # LATIN LETTER SMALL CAPITAL L
107ʡ ↔ '>\'; # LATIN LETTER GLOTTAL STOP WITH STROKE
108ʢ ↔ '<\'; # LATIN LETTER REVERSED GLOTTAL STOP WITH STROKE
109ʰ ↔ '_h'; # MODIFIER LETTER SMALL H
110ʷ ↔ '_w'; # MODIFIER LETTER SMALL W
111ʼ ↔ '_>'; # MODIFIER LETTER APOSTROPHE
112ˆ ↔ '_\'; # MODIFIER LETTER CIRCUMFLEX ACCENT
113ˇ ↔ '_/'; # CARON
114ˑ ↔ ':\'; # MODIFIER LETTER HALF TRIANGULAR COLON
115ˠ ↔ '_G'; # MODIFIER LETTER SMALL GAMMA
116ˡ ↔ '_l'; # MODIFIER LETTER SMALL L
117ˤ ↔ '_?\'; # 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
147ᵻ ↔ 'I\'; # LATIN SMALL CAPITAL LETTER I WITH STROKE
148ᵿ ↔ 'U\'; # LATIN SMALL CAPITAL LETTER U WITH STROKE
149ⁿ ↔ '_n'; # MODIFIER LETTER LATIN SMALL LETTER N
150# Alternative 2-character X-SAMPA representations
151ʋ ← 'v\'; # LATIN SMALL LETTER V WITH HOOK
152ʲ ← '_j'; # MODIFIER LETTER SMALL J
153\u0303 ← '_~'; # COMBINING TILDE
154\u0329 ← '_='; # COMBINING VERTICAL LINE BELOW
155# 1-character X-SAMPA representations
156c\u0327 ↔ C; # LATIN SMALL LETTER C WITH CEDILLA (decomposed)
157æ ↔ '{'; # LATIN SMALL LETTER AE
158ð ↔ D; # LATIN SMALL LETTER ETH
159ø ↔ 2; # LATIN SMALL LETTER O WITH STROKE
160ŋ ↔ N; # LATIN SMALL LETTER ENG
161œ ↔ 9; # LATIN SMALL LIGATURE OE
162ɐ ↔ 6; # LATIN SMALL LETTER TURNED A
163ɑ ↔ A; # LATIN SMALL LETTER ALPHA
164ɒ ↔ Q; # LATIN SMALL LETTER TURNED ALPHA
165ɔ ↔ O; # LATIN SMALL LETTER OPEN O
166ə ↔ '@'; # LATIN SMALL LETTER SCHWA
167ɛ ↔ E; # LATIN SMALL LETTER OPEN E
168ɜ ↔ 3; # LATIN SMALL LETTER REVERSED OPEN E
169ɡ ↔ g; # LATIN SMALL LETTER SCRIPT G
170ɣ ↔ G; # LATIN SMALL LETTER GAMMA
171ɤ ↔ 7; # LATIN SMALL LETTER RAMS HORN
172ɥ ↔ H; # LATIN SMALL LETTER TURNED H
173ɨ ↔ 1; # LATIN SMALL LETTER I WITH STROKE
174ɪ ↔ I; # LATIN LETTER SMALL CAPITAL I
175ɫ ↔ 5; # LATIN SMALL LETTER L WITH MIDDLE TILDE
176ɬ ↔ K; # LATIN SMALL LETTER L WITH BELT
177ɯ ↔ M; # LATIN SMALL LETTER TURNED M
178ɱ ↔ F; # LATIN SMALL LETTER M WITH HOOK
179ɲ ↔ J; # LATIN SMALL LETTER N WITH LEFT HOOK
180ɵ ↔ 8; # LATIN SMALL LETTER BARRED O
181ɶ ↔ '&'; # LATIN LETTER SMALL CAPITAL OE
182ɾ ↔ 4; # LATIN SMALL LETTER R WITH FISHHOOK
183ʁ ↔ R; # LATIN LETTER SMALL CAPITAL INVERTED R
184ʃ ↔ S; # LATIN SMALL LETTER ESH
185ʉ ↔ '}'; # LATIN SMALL LETTER U BAR
186ʊ ↔ U; # LATIN SMALL LETTER UPSILON
187ʋ ↔ P; # LATIN SMALL LETTER V WITH HOOK
188ʌ ↔ V; # LATIN SMALL LETTER TURNED V
189ʍ ↔ W; # LATIN SMALL LETTER TURNED W
190ʎ ↔ L; # LATIN SMALL LETTER TURNED Y
191ʏ ↔ Y; # LATIN LETTER SMALL CAPITAL Y
192ʒ ↔ Z; # LATIN SMALL LETTER EZH
193ʔ ↔ '?'; # LATIN LETTER GLOTTAL STOP
194ʲ ↔ \'; # MODIFIER LETTER SMALL J
195ˈ ↔ '"'; # MODIFIER LETTER VERTICAL LINE
196ˌ ↔ '%'; # MODIFIER LETTER LOW VERTICAL LINE
197ː ↔ ':'; # 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
202β ↔ B; # GREEK SMALL LETTER BETA
203θ ↔ T; # GREEK SMALL LETTER THETA
204χ ↔ X; # GREEK SMALL LETTER CHI
205↑ ↔ '^'; # UPWARDS ARROW
206↓ ↔ '!'; # DOWNWARDS ARROW
207# Compatibility rules for variant or obsolete IPA symbols
208g → g; # LATIN SMALL LETTER G (redundant, for additional clarity)
209ȵ → J; # LATIN SMALL LETTER N WITH CURL
210ɩ → I; # LATIN SMALL LETTER IOTA
211ɷ → U; # LATIN SMALL LETTER CLOSED OMEGA
212ɼ → 'r_r'; # LATIN SMALL LETTER R WITH LONG LEG
213ɿ → 'z='; # LATIN SMALL LETTER REVERSED R WITH FISHHOOK
214ʅ → 'z`='; # LATIN SMALL LETTER SQUAT REVERSED ESH
215ʆ → S\'; # LATIN SMALL LETTER ESH WITH CURL
216ʇ → '|\' ; # LATIN SMALL LETTER TURNED T
217ʓ → Z\'; # LATIN SMALL LETTER EZH WITH CURL
218ʖ → '|\|\'; # LATIN LETTER INVERTED GLOTTAL STOP
219ʗ → '!\'; # LATIN LETTER STRETCHED C
220ʚ → '3\'; # LATIN SMALL LETTER CLOSED OPEN E
221ʠ → 'G\_<_0'; # LATIN SMALL LETTER Q WITH HOOK
222ʣ → d $t z; # LATIN SMALL LETTER DZ DIGRAPH
223ʤ → d $t Z; # LATIN SMALL LETTER DEZH DIGRAPH
224ʥ → d $t 'z\'; # LATIN SMALL LETTER DZ DIGRAPH WITH CURL
225ʦ → t $t s; # LATIN SMALL LETTER TS DIGRAPH
226ʧ → t $t S; # LATIN SMALL LETTER TESH DIGRAPH
227ʨ → 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
231φ → 'p\'; # GREEK SMALL LETTER PHI
232ꞎ → 'K`'; # LATIN SMALL LETTER L WITH RETROFLEX HOOK AND BELT
233::NFC(NFD);
234