|
1
|
|
|
# Copyright 2014-2020 by Christopher C. Little. |
|
2
|
|
|
# This file is part of Abydos. |
|
3
|
|
|
# |
|
4
|
|
|
# Abydos is free software: you can redistribute it and/or modify |
|
5
|
|
|
# it under the terms of the GNU General Public License as published by |
|
6
|
|
|
# the Free Software Foundation, either version 3 of the License, or |
|
7
|
|
|
# (at your option) any later version. |
|
8
|
|
|
# |
|
9
|
|
|
# Abydos is distributed in the hope that it will be useful, |
|
10
|
|
|
# but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
11
|
|
|
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
12
|
|
|
# GNU General Public License for more details. |
|
13
|
|
|
# |
|
14
|
|
|
# You should have received a copy of the GNU General Public License |
|
15
|
|
|
# along with Abydos. If not, see <http://www.gnu.org/licenses/>. |
|
16
|
|
|
|
|
17
|
|
|
"""abydos.phones._phones. |
|
18
|
|
|
|
|
19
|
1 |
|
The phones module implements phonetic feature coding, decoding, and comparison |
|
20
|
|
|
functions. |
|
21
|
|
|
""" |
|
22
|
|
|
|
|
23
|
|
|
from typing import Dict, List, Optional, Sequence, Union |
|
24
|
|
|
from unicodedata import normalize |
|
25
|
1 |
|
|
|
26
|
|
|
__all__ = ['cmp_features', 'get_feature', 'ipa_to_features'] |
|
27
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
_PHONETIC_FEATURES = { |
|
30
|
|
|
't': 2783230754502126250, |
|
31
|
|
|
't͇': 2783230754502126250, |
|
32
|
1 |
|
't̪': 2783230479624219306, |
|
33
|
|
|
't̠': 2783231579135847082, |
|
34
|
1 |
|
'd': 2783230754501864106, |
|
35
|
1 |
|
'd͇': 2783230754501864106, |
|
36
|
|
|
'd̪': 2783230479623957162, |
|
37
|
1 |
|
'd̠': 2783231579135584938, |
|
38
|
|
|
's': 2783230754502125978, |
|
39
|
|
|
's͇': 2783230754502125978, |
|
40
|
1 |
|
's̪': 2783230479624219034, |
|
41
|
|
|
's̟': 2783230479624219034, |
|
42
|
|
|
'z': 2783230754501863834, |
|
43
|
|
|
'z͇': 2783230754501863834, |
|
44
|
|
|
'z̪': 2783230479623956890, |
|
45
|
|
|
'z̟': 2783230479623956890, |
|
46
|
|
|
'θ̱': 2783230754502125994, |
|
47
|
|
|
'θ͇': 2783230754502125994, |
|
48
|
|
|
'ɹ̝̊': 2783230754502125994, |
|
49
|
|
|
'ð̠': 2783230754501863850, |
|
50
|
|
|
'ð͇': 2783230754501863850, |
|
51
|
|
|
'ɹ̝': 2783230754501863850, |
|
52
|
|
|
'ɬ': 2783230754502060454, |
|
53
|
|
|
'ɮ': 2783230754501863846, |
|
54
|
|
|
'θ': 2783230479624219050, |
|
55
|
|
|
's̄': 2783230479624219034, |
|
56
|
|
|
'ð': 2783230479623956906, |
|
57
|
|
|
'ð̞': 2693158487076546986, |
|
58
|
|
|
'ʃ': 2783231556184615322, |
|
59
|
|
|
'ʃʲ': 2783231556184615322, |
|
60
|
|
|
'š': 2783231556184615322, |
|
61
|
|
|
'ɹ̠̊˔': 2783231579135846826, |
|
62
|
|
|
'ʒ': 2783231556184353178, |
|
63
|
|
|
'ʒʲ': 2783231556184353178, |
|
64
|
|
|
'ž': 2783231556184353178, |
|
65
|
|
|
'ɹ̠˔': 2783231579135584682, |
|
66
|
|
|
'c': 2783233463150095018, |
|
67
|
|
|
'ɟ': 2783233463149832874, |
|
68
|
|
|
'ç': 2783233463150094762, |
|
69
|
|
|
'ʝ': 2783233463149832618, |
|
70
|
|
|
'p': 2781720575281375914, |
|
71
|
|
|
'p̪': 2781702983095331498, |
|
72
|
|
|
'p͆': 2781702983095331498, |
|
73
|
|
|
'b': 2781720575281113770, |
|
74
|
|
|
'b̪': 2781702983095069354, |
|
75
|
|
|
'b͆': 2781702983095069354, |
|
76
|
|
|
'f': 2781702983095331242, |
|
77
|
|
|
'v': 2781702983095069098, |
|
78
|
|
|
'ɸ': 2781720575281375658, |
|
79
|
|
|
'ɸ̞': 2691648582733965738, |
|
80
|
|
|
'β': 2781720575281113514, |
|
81
|
|
|
'β̞': 2691648582733703594, |
|
82
|
|
|
'k': 2783233462881659562, |
|
83
|
|
|
'ɡ': 2783233462881397418, |
|
84
|
|
|
'g': 2783233462881397418, |
|
85
|
|
|
'x': 2783233462881659306, |
|
86
|
|
|
'x̞': 2981391846485961130, |
|
87
|
|
|
'ɣ': 2783233462881397162, |
|
88
|
|
|
'ɰ': 2981391846485698986, |
|
89
|
|
|
'ɣ̞': 2981391846485698986, |
|
90
|
|
|
'ɣ˕': 2981391846485698986, |
|
91
|
|
|
'q': 2783233480061528746, |
|
92
|
|
|
'ɢ': 2783233480061266602, |
|
93
|
|
|
'χ': 2783233480061528490, |
|
94
|
|
|
'x̣': 2783233480061528490, |
|
95
|
|
|
'χ̞': 2693161487514118570, |
|
96
|
|
|
'ʁ': 2783233480061266346, |
|
97
|
|
|
'ʁ̝': 2783233480061266346, |
|
98
|
|
|
'ʁ̞': 2693161487513856426, |
|
99
|
|
|
'ħ': 2783233503273855402, |
|
100
|
|
|
'ħ̞': 2693161510726445482, |
|
101
|
|
|
'ʕ': 2783233503273593258, |
|
102
|
|
|
'ʕ̝': 2783233503273593258, |
|
103
|
|
|
'ʕ̞': 2693161510726183338, |
|
104
|
|
|
'ɑ̯': 2693161510726183338, |
|
105
|
|
|
'h': 2783233503281129898, |
|
106
|
|
|
'h̃': 2693161510733719914, |
|
107
|
|
|
'ɦ': 2783233503280867754, |
|
108
|
|
|
'ʔ': 2783233503281179306, |
|
109
|
|
|
'm': 2709662981243185770, |
|
110
|
|
|
'm̥': 2709662981243447914, |
|
111
|
|
|
'ɱ': 2709645389057141354, |
|
112
|
|
|
'm̪': 2709645389057141354, |
|
113
|
|
|
'n': 2711173160463936106, |
|
114
|
|
|
'n̪': 2711172885586029162, |
|
115
|
|
|
'n̥': 2711173160464198250, |
|
116
|
|
|
'n̊': 2711173160464198250, |
|
117
|
|
|
'n̪̊': 2711172885586291306, |
|
118
|
|
|
'ɳ': 2711174259975563882, |
|
119
|
|
|
'ɳ̊': 2711174259975826026, |
|
120
|
|
|
'ɳ̥': 2711174259975826026, |
|
121
|
|
|
'ɲ': 2711175869111904874, |
|
122
|
|
|
'ñ': 2711175869111904874, |
|
123
|
|
|
'n̠ʲ': 2711173944966556266, |
|
124
|
|
|
'ɲ̟': 2711173944966556266, |
|
125
|
|
|
'ɲ̊': 2711175869112167018, |
|
126
|
|
|
'ɲ̥': 2711175869112167018, |
|
127
|
|
|
'ŋ': 2711175868843469418, |
|
128
|
|
|
'ŋ̊': 2711175868843731562, |
|
129
|
|
|
'ɴ': 2711175886023338602, |
|
130
|
|
|
'l': 2693158761954453926, |
|
131
|
|
|
'l̠': 2693159586588174758, |
|
132
|
|
|
'l̪': 2693158487076546982, |
|
133
|
|
|
'ɫ': 2693158721554917798, |
|
134
|
|
|
'lˠ': 2693158721554917798, |
|
135
|
|
|
'lˤ': 2693158721554917798, |
|
136
|
|
|
'l̴': 2693158721554917798, |
|
137
|
|
|
'ɫ̪': 2693158446677010854, |
|
138
|
|
|
'l̪ˠ': 2693158446677010854, |
|
139
|
|
|
'l̪ˤ': 2693158446677010854, |
|
140
|
|
|
'l̴̪': 2693158446677010854, |
|
141
|
|
|
'l̥': 2693158761954716070, |
|
142
|
|
|
'ʎ': 2693161470602422694, |
|
143
|
|
|
'l̠ʲ': 2693159546457074086, |
|
144
|
|
|
'ʎ̟': 2693159546457074086, |
|
145
|
|
|
'ʟ': 2693161470333987238, |
|
146
|
|
|
'r': 2711173160463936170, |
|
147
|
|
|
'r̥': 2711173160464198314, |
|
148
|
|
|
'ɹ': 2693158761954453930, |
|
149
|
|
|
'ð̠˕': 2693158761954453930, |
|
150
|
|
|
'ʀ': 2711175886023338666, |
|
151
|
|
|
'ʀ̥': 2711175886023600810, |
|
152
|
|
|
'j': 2981391846754134442, |
|
153
|
|
|
'i̯': 2981391846754134442, |
|
154
|
|
|
'j̊': 2981391846754396586, |
|
155
|
|
|
'w': 2978753018579036586, |
|
156
|
|
|
'u̯': 2978753018579036586, |
|
157
|
|
|
'ɰʷ': 2978753018579036586, |
|
158
|
|
|
'ʍ': 2978753018579298730, |
|
159
|
|
|
'w̥': 2978753018579298730, |
|
160
|
|
|
'ɰᵝ': 2978682649834858922, |
|
161
|
|
|
'wᵝ': 2978682649834858922, |
|
162
|
|
|
'ɥ': 2978753018847472042, |
|
163
|
|
|
'jʷ': 2978753018847472042, |
|
164
|
|
|
'ʡ': 2783233503274904234, |
|
165
|
|
|
'ʜ': 2783233503274903978, |
|
166
|
|
|
'ʢ': 2783233503274641834, |
|
167
|
|
|
'ʢ̝': 2783233503274641834, |
|
168
|
|
|
'ʢ̞': 2693161510727231914, |
|
169
|
|
|
'ᴙ': 2711175909236714154, |
|
170
|
|
|
'ʙ': 2709662981243185834, |
|
171
|
|
|
'ʋ': 2691630990547659178, |
|
172
|
|
|
'ɕ': 2783231539004746138, |
|
173
|
|
|
'ʑ': 2783231539004483994, |
|
174
|
|
|
'ɾ': 2711173160463935914, |
|
175
|
|
|
'ᴅ': 2711173160463935914, |
|
176
|
|
|
'ɾ̥': 2711173160464198058, |
|
177
|
|
|
'ɾ̊': 2711173160464198058, |
|
178
|
|
|
'ᴅ̥': 2711173160464198058, |
|
179
|
|
|
'ᴅ̊': 2711173160464198058, |
|
180
|
|
|
'ɾ̃': 2711173160463935850, |
|
181
|
|
|
'n̆': 2711173160463935850, |
|
182
|
|
|
'ⱱ': 2709645389057141162, |
|
183
|
|
|
'ⱱ̟': 2709662981243185578, |
|
184
|
|
|
'w̆': 2709662981243185578, |
|
185
|
|
|
'b̆': 2709662981243185578, |
|
186
|
|
|
'ɢ̆': 2711175886023338410, |
|
187
|
|
|
'ʀ̆': 2711175886023338410, |
|
188
|
|
|
'ʡ̯': 2711175909236713898, |
|
189
|
|
|
'ɺ': 2711173160463935910, |
|
190
|
|
|
'ʎ̯': 2711175869111904678, |
|
191
|
|
|
'ʟ̆': 2711175868843469222, |
|
192
|
|
|
'ʈ': 2783231854013754026, |
|
193
|
|
|
'ɖ': 2783231854013491882, |
|
194
|
|
|
'ʂ': 2783231854013753754, |
|
195
|
|
|
'ʐ': 2783231854013491610, |
|
196
|
|
|
'ɻ': 2693159861466081706, |
|
197
|
|
|
'ɽ': 2711174259975563690, |
|
198
|
|
|
'ɽ͡r': 2711174259975563946, |
|
199
|
|
|
'ɭ': 2693159861466081702, |
|
200
|
|
|
'ɭ̆': 2711174259975563686, |
|
201
|
|
|
'ɺ˞': 2711174259975563686, |
|
202
|
|
|
'ɺ̢': 2711174259975563686, |
|
203
|
|
|
'ꞎ': 2783231854013688230, |
|
204
|
|
|
'ʎ̝̊': 2783233463150094758, |
|
205
|
|
|
'ʟ̝̊': 2783233462881659302, |
|
206
|
|
|
'ʟ̝': 2783233462881397158, |
|
207
|
|
|
# affricates & co-articulated |
|
208
|
|
|
't͡ʃ': 2783231556184615833, |
|
209
|
|
|
'ʧ': 2783231556184615833, |
|
210
|
|
|
't͜ʃ': 2783231556184615833, |
|
211
|
|
|
't̠ʲʃ': 2783231556184615833, |
|
212
|
|
|
'č': 2783231556184615833, |
|
213
|
|
|
'd͡ʒ': 2783231556184353689, |
|
214
|
|
|
'ʤ': 2783231556184353689, |
|
215
|
|
|
'd͜ʒ': 2783231556184353689, |
|
216
|
|
|
'd̠ʲʒ': 2783231556184353689, |
|
217
|
|
|
'ǯ': 2783231556184353689, |
|
218
|
|
|
't͡s': 2783230754502126489, |
|
219
|
|
|
'ʦ': 2783230754502126489, |
|
220
|
|
|
't͜s': 2783230754502126489, |
|
221
|
|
|
't̪͡s̪': 2783230479624219545, |
|
222
|
|
|
't͡s̪': 2783230479624219545, |
|
223
|
|
|
't̟͡s̟': 2783230479624219545, |
|
224
|
|
|
't͡s̟': 2783230479624219545, |
|
225
|
|
|
'ʦ̪': 2783230479624219545, |
|
226
|
|
|
'ʦ̟': 2783230479624219545, |
|
227
|
|
|
't͡θ̠': 2783230754502126505, |
|
228
|
|
|
't͡θ͇': 2783230754502126505, |
|
229
|
|
|
't͡θ': 2783230479624219561, |
|
230
|
|
|
't͜θ': 2783230479624219561, |
|
231
|
|
|
't̪͡θ': 2783230479624219561, |
|
232
|
|
|
't̟͡θ': 2783230479624219561, |
|
233
|
|
|
'd͡z': 2783230754501864345, |
|
234
|
|
|
'ʣ': 2783230754501864345, |
|
235
|
|
|
'd͜z': 2783230754501864345, |
|
236
|
|
|
'd̪͡z̪': 2783230479623957401, |
|
237
|
|
|
'd͡z̪': 2783230479623957401, |
|
238
|
|
|
'd̟͡z̟': 2783230479623957401, |
|
239
|
|
|
'd͡z̟': 2783230479623957401, |
|
240
|
|
|
'ʣ̪': 2783230479623957401, |
|
241
|
|
|
'ʣ̟': 2783230479623957401, |
|
242
|
|
|
'd͡ð̠': 2783230754501864361, |
|
243
|
|
|
'd͡ð̳': 2783230754501864361, |
|
244
|
|
|
'd͡ð': 2783230479623957417, |
|
245
|
|
|
'd͜ð': 2783230479623957417, |
|
246
|
|
|
'd̪͡ð': 2783230479623957417, |
|
247
|
|
|
'd̟͡ð': 2783230479623957417, |
|
248
|
|
|
'k͡x': 2783233462881659817, |
|
249
|
|
|
'ɡ͡ɣ': 2783233462881397673, |
|
250
|
|
|
'g͡ɣ': 2783233462881397673, |
|
251
|
|
|
'p͡f': 2781738167467420585, |
|
252
|
|
|
'p̪͡f': 2781702983095331753, |
|
253
|
|
|
'b͡v': 2781738167467158441, |
|
254
|
|
|
'b̪͡v': 2781702983095069609, |
|
255
|
|
|
'b̪͜v': 2781702983095069609, |
|
256
|
|
|
't͡ɕ': 2783231539004746649, |
|
257
|
|
|
't͜ɕ': 2783231539004746649, |
|
258
|
|
|
'ʨ': 2783231539004746649, |
|
259
|
|
|
'd͡ʑ': 2783231539004484505, |
|
260
|
|
|
'd͜ʑ': 2783231539004484505, |
|
261
|
|
|
'ʥ': 2783231539004484505, |
|
262
|
|
|
'ʈ͡ʂ': 2783231854013754265, |
|
263
|
|
|
't͡ʂ': 2783231854013754265, |
|
264
|
|
|
'ɖ͡ʐ': 2783231854013492121, |
|
265
|
|
|
'd͡ʐ': 2783231854013492121, |
|
266
|
|
|
't͡ɬ': 2783230754502060965, |
|
267
|
|
|
't͜ɬ': 2783230754502060965, |
|
268
|
|
|
'ƛ': 2783230754502060965, |
|
269
|
|
|
'd͡ɮ': 2783230754501864357, |
|
270
|
|
|
'c͡ç': 2783233463150095273, |
|
271
|
|
|
'c͜ç': 2783233463150095273, |
|
272
|
|
|
'ɟ͡ʝ': 2783233463149833129, |
|
273
|
|
|
'c͡ʎ̝̥': 2783233463150095269, |
|
274
|
|
|
'k͡ʟ̝̊': 2783233462881659813, |
|
275
|
|
|
'ɡ͡ʟ̝': 2783233462881397669, |
|
276
|
|
|
'q͡χ': 2783233480061529001, |
|
277
|
|
|
'ɢ͡ʁ': 2783233480061266857, |
|
278
|
|
|
'ɢ͜ʁ': 2783233480061266857, |
|
279
|
|
|
'ɧ': 2783231813614217626, |
|
280
|
|
|
'k͡p': 2781720534881839786, |
|
281
|
|
|
'k͜p': 2781720534881839786, |
|
282
|
|
|
'ɡ͡b': 2781720534881577642, |
|
283
|
|
|
'kʷ': 2780594634974997162, |
|
284
|
|
|
'k͡w': 2780594634974997162, |
|
285
|
|
|
'k͜w': 2780594634974997162, |
|
286
|
|
|
'kʷh': 2780594634974931626, |
|
287
|
|
|
'gʷ': 2780594634974735018, |
|
288
|
|
|
'g͡w': 2780594634974735018, |
|
289
|
|
|
'gʷh': 2780594634974669482, |
|
290
|
|
|
# implosives |
|
291
|
|
|
'ɓ̥': 2781720575281355434, |
|
292
|
|
|
'ƥ': 2781720575281355434, |
|
293
|
|
|
'pʼ↓': 2781720575281355434, |
|
294
|
|
|
'ɓ': 2781720575281093290, |
|
295
|
|
|
'ɗ̥': 2783230754502105770, |
|
296
|
|
|
'ƭ': 2783230754502105770, |
|
297
|
|
|
'tʼ↓': 2783230754502105770, |
|
298
|
|
|
'ɗ': 2783230754501843626, |
|
299
|
|
|
'ᶑ': 2783231854013471402, |
|
300
|
|
|
'ʄ̊': 2783233463150074538, |
|
301
|
|
|
'ƈ': 2783233463150074538, |
|
302
|
|
|
'cʼ↓': 2783233463150074538, |
|
303
|
|
|
'ʄ': 2783233463149812394, |
|
304
|
|
|
'ɠ': 2783233462881376938, |
|
305
|
|
|
'ʛ': 2783233480061246122, |
|
306
|
|
|
# clicks |
|
307
|
|
|
'ʘ': 2781720575281374890, |
|
308
|
|
|
'ʘʰ': 2781720575281309354, |
|
309
|
|
|
'ʘ̬': 2781720575281112746, |
|
310
|
|
|
'ᶢʘ': 2781720575281112746, |
|
311
|
|
|
'ʘ̃': 2781720575281112682, |
|
312
|
|
|
'ᵑʘ': 2781720575281112682, |
|
313
|
|
|
'ᵐʘ': 2781720575281112682, |
|
314
|
|
|
'ʘ̥̃ʰ': 2781720575281309290, |
|
315
|
|
|
'ᵑ̊ʘʰ': 2781720575281309290, |
|
316
|
|
|
'ʘ̃ˀ': 2781720575281358442, |
|
317
|
|
|
'ᵑʘˀ': 2781720575281358442, |
|
318
|
|
|
'ʘ̃͜ʔ': 2781720575281358442, |
|
319
|
|
|
'ᵑ̊ʘˀ': 2781720575281358442, |
|
320
|
|
|
'ǀ': 2783230479624218282, |
|
321
|
|
|
'ʇ': 2783230479624218282, |
|
322
|
|
|
'ǀ̬': 2783230479623956138, |
|
323
|
|
|
'ʇ̬': 2783230479623956138, |
|
324
|
|
|
'ᶢǀ': 2783230479623956138, |
|
325
|
|
|
'ᶢʇ': 2783230479623956138, |
|
326
|
|
|
'ǀ̃': 2783230479623956074, |
|
327
|
|
|
'ʇ̃': 2783230479623956074, |
|
328
|
|
|
'ᵑǀ': 2783230479623956074, |
|
329
|
|
|
'ⁿǀ': 2783230479623956074, |
|
330
|
|
|
'ᵑʇ': 2783230479623956074, |
|
331
|
|
|
'ǀ̥̃ʰ': 2783230479624152682, |
|
332
|
|
|
'ʇ̥̃ʰ': 2783230479624152682, |
|
333
|
|
|
'ᵑ̊ǀʰ': 2783230479624152682, |
|
334
|
|
|
'ᵑ̊ʇʰ': 2783230479624152682, |
|
335
|
|
|
'ǀ̃ˀ': 2783230479624201834, |
|
336
|
|
|
'ʇ̃ˀ': 2783230479624201834, |
|
337
|
|
|
'ᵑǀˀ': 2783230479624201834, |
|
338
|
|
|
'ᵑʇˀ': 2783230479624201834, |
|
339
|
|
|
'ᵑ̊ǀˀ': 2783230479624201834, |
|
340
|
|
|
'ᵑ̊ʇˀ': 2783230479624201834, |
|
341
|
|
|
'ǃ': 2783230754502125226, |
|
342
|
|
|
'ʗ': 2783230754502125226, |
|
343
|
|
|
'ǃʰ': 2783230754502059690, |
|
344
|
|
|
'ʗʰ': 2783230754502059690, |
|
345
|
|
|
'ǃ̬': 2783230754501863082, |
|
346
|
|
|
'ʗ̬': 2783230754501863082, |
|
347
|
|
|
'ᶢǃ': 2783230754501863082, |
|
348
|
|
|
'ᶢʗ': 2783230754501863082, |
|
349
|
|
|
'ǃ̃': 2783230754501863018, |
|
350
|
|
|
'ᵑǃ': 2783230754501863018, |
|
351
|
|
|
'ʗ̃': 2783230754501863018, |
|
352
|
|
|
'ᵑʗ': 2783230754501863018, |
|
353
|
|
|
'ǃ̥̃ʰ': 2783230754502059626, |
|
354
|
|
|
'ʗ̃̊ʰ': 2783230754502059626, |
|
355
|
|
|
'ᵑ̊ǃʰ': 2783230754502059626, |
|
356
|
|
|
'ᵑ̊ʗʰ': 2783230754502059626, |
|
357
|
|
|
'ǃ̃ˀ': 2783230754502108778, |
|
358
|
|
|
'ʗ̃ˀ': 2783230754502108778, |
|
359
|
|
|
'ᵑǃˀ': 2783230754502108778, |
|
360
|
|
|
'ᵑʗˀ': 2783230754502108778, |
|
361
|
|
|
'ǂ': 2783233463150093994, |
|
362
|
|
|
'ʄ̵': 2783233463150093994, |
|
363
|
|
|
'⨎': 2783233463150093994, |
|
364
|
|
|
'ǂʰ': 2783233463150028458, |
|
365
|
|
|
'ǂ̬': 2783233463149831850, |
|
366
|
|
|
'ᶢǂ': 2783233463149831850, |
|
367
|
|
|
'ǂ̃': 2783233463149831786, |
|
368
|
|
|
'ᵑǂ': 2783233463149831786, |
|
369
|
|
|
'ǂ̥̃ʰ': 2783233463150028394, |
|
370
|
|
|
'ᵑ̊ǂʰ': 2783233463150028394, |
|
371
|
|
|
'ǂ̃ˀ': 2783233463150077546, |
|
372
|
|
|
'ᵑǂˀ': 2783233463150077546, |
|
373
|
|
|
'ᵑǂ͡ʔ': 2783233463150077546, |
|
374
|
|
|
'ᵑ̊ǂˀ': 2783233463150077546, |
|
375
|
|
|
'ǃ͡s': 2783233463150093978, |
|
376
|
|
|
'ǂᶴ': 2783233463150093978, |
|
377
|
|
|
'ǁ': 2783230754502125222, |
|
378
|
|
|
'ʖ': 2783230754502125222, |
|
379
|
|
|
'ǁʰ': 2783230754502059686, |
|
380
|
|
|
'ʖʰ': 2783230754502059686, |
|
381
|
|
|
'ǁ̬': 2783230754501863078, |
|
382
|
|
|
'ʖ̬': 2783230754501863078, |
|
383
|
|
|
'ᶢǁ': 2783230754501863078, |
|
384
|
|
|
'ᶢʖ': 2783230754501863078, |
|
385
|
|
|
'ǁ̃': 2783230754501863014, |
|
386
|
|
|
'ʖ̃': 2783230754501863014, |
|
387
|
|
|
'ᵑǁ': 2783230754501863014, |
|
388
|
|
|
'ᵑʖ': 2783230754501863014, |
|
389
|
|
|
'ǁ̥̃ʰ': 2783230754502059622, |
|
390
|
|
|
'ʖ̥̃ʰ': 2783230754502059622, |
|
391
|
|
|
'ᵑ̊ǁʰ': 2783230754502059622, |
|
392
|
|
|
'ᵑ̊ʖʰ': 2783230754502059622, |
|
393
|
|
|
'ǁ̃ˀ': 2783230754502108774, |
|
394
|
|
|
'ʖ̃ˀ': 2783230754502108774, |
|
395
|
|
|
'ᵑǁˀ': 2783230754502108774, |
|
396
|
|
|
'ᵑʖˀ': 2783230754502108774, |
|
397
|
|
|
'ᵑǁ͡ʔ': 2783230754502108774, |
|
398
|
|
|
'ʖ̃͜ʔ': 2783230754502108774, |
|
399
|
|
|
'ǃ˞': 2783231854013753002, |
|
400
|
|
|
'ǃǁ': 2783231854013753002, |
|
401
|
|
|
'‼': 2783231854013753002, |
|
402
|
|
|
'!!': 2783231854013753002, |
|
403
|
|
|
'ǃ˞ʰ': 2783231854013687466, |
|
404
|
|
|
'‼ʰ': 2783231854013687466, |
|
405
|
|
|
'!!ʰ': 2783231854013687466, |
|
406
|
|
|
'ǃ̬˞': 2783231854013490858, |
|
407
|
|
|
'ᶢǃ˞': 2783231854013490858, |
|
408
|
|
|
'‼̬': 2783231854013490858, |
|
409
|
|
|
'!!̬': 2783231854013490858, |
|
410
|
|
|
'ᶢ‼': 2783231854013490858, |
|
411
|
|
|
'ᶢ!!': 2783231854013490858, |
|
412
|
|
|
'ǃ̃˞': 2783231854013490794, |
|
413
|
|
|
'ᵑǃ˞': 2783231854013490794, |
|
414
|
|
|
'‼̃': 2783231854013490794, |
|
415
|
|
|
'!!̃': 2783231854013490794, |
|
416
|
|
|
'ᵑ‼': 2783231854013490794, |
|
417
|
|
|
'ᵑ!!': 2783231854013490794, |
|
418
|
|
|
'ǃ̥̃˞ʰ': 2783231854013687402, |
|
419
|
|
|
'ᵑ̊ǃ˞ʰ': 2783231854013687402, |
|
420
|
|
|
'‼̥̃ʰ': 2783231854013687402, |
|
421
|
|
|
'!!̥̃ʰ': 2783231854013687402, |
|
422
|
|
|
'ᵑ̊‼ʰ': 2783231854013687402, |
|
423
|
|
|
'ᵑ̊!!ʰ': 2783231854013687402, |
|
424
|
|
|
'ǃ̃˞ˀ': 2783231854013736554, |
|
425
|
|
|
'ᵑǃ˞͜ʔ': 2783231854013736554, |
|
426
|
|
|
'ᵑ̊‼ˀ': 2783231854013736554, |
|
427
|
|
|
'ᵑ‼ˀ': 2783231854013736554, |
|
428
|
|
|
'‼̃ˀ': 2783231854013736554, |
|
429
|
|
|
'ᵑ̊!!ˀ': 2783231854013736554, |
|
430
|
|
|
'ᵑ!!ˀ': 2783231854013736554, |
|
431
|
|
|
'!!̃ˀ': 2783231854013736554, |
|
432
|
|
|
# vowels |
|
433
|
|
|
'i': 1826957412996131242, |
|
434
|
|
|
'ɪ': 1826957413067434410, |
|
435
|
|
|
'ɩ': 1826957413067434410, |
|
436
|
|
|
'u': 1825831513894594986, |
|
437
|
|
|
'ʊ': 1825831513965898154, |
|
438
|
|
|
'ɷ': 1825831513965898154, |
|
439
|
|
|
'ᴜ': 1825831513965898154, |
|
440
|
|
|
'ɯ̽': 1826957413872740778, |
|
441
|
|
|
'ʊ̜': 2115187790024452522, |
|
442
|
|
|
'e': 1826957430176000426, |
|
443
|
|
|
'ɛ': 1826957430247303594, |
|
444
|
|
|
'o': 1825831531074464170, |
|
445
|
|
|
'ɤ': 1826957430981306794, |
|
446
|
|
|
'ɔ': 1825831531145767338, |
|
447
|
|
|
'ʌ': 1826957431052609962, |
|
448
|
|
|
'a': 1826957425952336298, |
|
449
|
|
|
'a̟': 1826957425952336298, |
|
450
|
|
|
'æ̞': 1826957425952336298, |
|
451
|
|
|
'ɶ': 1825831526045493674, |
|
452
|
|
|
'æ': 1826957425885227434, |
|
453
|
|
|
'y': 1825831513089288618, |
|
454
|
|
|
'ʏ': 1825831513160591786, |
|
455
|
|
|
'ø': 1825831530269157802, |
|
456
|
|
|
'œ': 1825831530340460970, |
|
457
|
|
|
'ə': 1828083331160779178, |
|
458
|
|
|
'ɵ̞': 1825831531347093930, |
|
459
|
|
|
'ə̹': 1825831531347093930, |
|
460
|
|
|
'ɞ̝': 1825831531347093930, |
|
461
|
|
|
'ɘ̞': 1826957431253936554, |
|
462
|
|
|
'ɯ': 1826957413805631914, |
|
463
|
|
|
'ɒ': 1825831526850800042, |
|
464
|
|
|
'ɑ': 1826957426757642666, |
|
465
|
|
|
'ɨ': 1826957414069873066, |
|
466
|
|
|
'ï': 1826957414069873066, |
|
467
|
|
|
'ʉ': 1825831514163030442, |
|
468
|
|
|
'ü': 1825831514163030442, |
|
469
|
|
|
'ɘ': 1826957431249742250, |
|
470
|
|
|
'ë': 1826957431249742250, |
|
471
|
|
|
'ɵ': 1825831531342899626, |
|
472
|
|
|
'ö': 1825831531342899626, |
|
473
|
|
|
'ɜ': 1826957431316851114, |
|
474
|
|
|
'ɛ̈': 1826957431316851114, |
|
475
|
|
|
'ɞ': 1825831531410008490, |
|
476
|
|
|
'ɔ̈': 1825831531410008490, |
|
477
|
|
|
'ä': 1826957427026078122, |
|
478
|
|
|
'a̠': 1826957427026078122, |
|
479
|
|
|
'ɑ̈': 1826957427026078122, |
|
480
|
|
|
'ɐ̞': 1826957427026078122, |
|
481
|
|
|
'ɐ': 1828083326865811882, |
|
482
|
|
|
'ɜ̞': 1826957426958969258, |
|
483
|
|
|
'ɞ̞': 1825831527052126634, |
|
484
|
|
|
'ɪ̟': 1826957413063240106, |
|
485
|
|
|
'ʏ̟': 1825831513156397482, |
|
486
|
|
|
'ʏ̫': 1825550038183881130, |
|
487
|
|
|
'ʏʷ': 1825550038183881130, |
|
488
|
|
|
'ɪʷ': 1825550038183881130, |
|
489
|
|
|
'y̫': 1825550038112577962, |
|
490
|
|
|
'yʷ': 1825550038112577962, |
|
491
|
|
|
'iʷ': 1825550038112577962, |
|
492
|
|
|
'u͍': 1825761145150417322, |
|
493
|
|
|
'ɯᵝ': 1825761145150417322, |
|
494
|
|
|
'ɯ͡β̞': 1825761145150417322, |
|
495
|
|
|
'ʉ͍': 1825761145418852778, |
|
496
|
|
|
'ɨᵝ': 1825761145418852778, |
|
497
|
|
|
'ɨ͡β̞': 1825761145418852778, |
|
498
|
|
|
'ɪ̈': 1826957414136981930, |
|
499
|
|
|
'ɨ̞': 1826957414136981930, |
|
500
|
|
|
'ɘ̝': 1826957414136981930, |
|
501
|
|
|
'ʊ̈': 1825831514230139306, |
|
502
|
|
|
'ʉ̞': 1825831514230139306, |
|
503
|
|
|
'ø̫': 1825550055292447146, |
|
504
|
|
|
'øʷ': 1825550055292447146, |
|
505
|
|
|
'eʷ': 1825550055292447146, |
|
506
|
|
|
'e̞': 1826957430180194730, |
|
507
|
|
|
'ɛ̝': 1826957430180194730, |
|
508
|
|
|
'ø̞': 1825831530273352106, |
|
509
|
|
|
'œ̝': 1825831530273352106, |
|
510
|
|
|
'o̞': 1825831531078658474, |
|
511
|
|
|
'ɔ̝': 1825831531078658474, |
|
512
|
|
|
'ɤ̞': 1826957430985501098, |
|
513
|
|
|
'ʌ̝': 1826957430985501098, |
|
514
|
|
|
# feature diacritics |
|
515
|
|
|
'ʰ': 65536, |
|
516
|
|
|
'ʱ': 327680, |
|
517
|
|
|
'̤': 327680, |
|
518
|
|
|
'ʼ': 16384, |
|
519
|
|
|
'̚': 131072, |
|
520
|
|
|
'̬': 262144, |
|
521
|
|
|
'̌': 262144, |
|
522
|
|
|
'̥': 524288, |
|
523
|
|
|
'̊': 524288, |
|
524
|
|
|
'ʷ': 6737807255011328, |
|
525
|
|
|
'ʲ': 103012106240, |
|
526
|
|
|
'̰': 278528, |
|
527
|
|
|
'̩': 1152921504606846976, |
|
528
|
|
|
'̯': 2305843009213693952, |
|
529
|
|
|
'̃': 64, |
|
530
|
|
|
'̨': 64, |
|
531
|
|
|
'͊': 128, |
|
532
|
|
|
'ˠ': 98180268032, |
|
533
|
|
|
'̴': 98180268032, |
|
534
|
|
|
'ˤ': 32505856, |
|
535
|
|
|
'̘': 23068672, |
|
536
|
|
|
'̙': 26214400, |
|
537
|
|
|
'˭': 688256, |
|
538
|
|
|
} |
|
539
|
|
|
|
|
540
|
|
|
_FEATURE_MASK = { |
|
541
|
|
|
'syllabic': 3458764513820540928, |
|
542
|
|
|
'consonantal': 864691128455135232, |
|
543
|
|
|
'sonorant': 216172782113783808, |
|
544
|
|
|
'approximant': 54043195528445952, |
|
545
|
|
|
'labial': 13510798882111488, |
|
546
|
|
|
'round': 3377699720527872, |
|
547
|
|
|
'protruded': 844424930131968, |
|
548
|
|
|
'compressed': 211106232532992, |
|
549
|
|
|
'labiodental': 52776558133248, |
|
550
|
|
|
'coronal': 13194139533312, |
|
551
|
1 |
|
'anterior': 3298534883328, |
|
552
|
|
|
'distributed': 824633720832, |
|
553
|
|
|
'dorsal': 206158430208, |
|
554
|
|
|
'high': 51539607552, |
|
555
|
|
|
'low': 12884901888, |
|
556
|
|
|
'front': 3221225472, |
|
557
|
|
|
'back': 805306368, |
|
558
|
|
|
'tense': 201326592, |
|
559
|
|
|
'pharyngeal': 50331648, |
|
560
|
|
|
'atr': 12582912, |
|
561
|
|
|
'rtr': 3145728, |
|
562
|
|
|
'voice': 786432, |
|
563
|
|
|
'spread_glottis': 196608, |
|
564
|
|
|
'constricted_glottis': 49152, |
|
565
|
|
|
'glottalic_suction': 12288, |
|
566
|
|
|
'velaric_suction': 3072, |
|
567
|
|
|
'continuant': 768, |
|
568
|
|
|
'nasal': 192, |
|
569
|
|
|
'strident': 48, |
|
570
|
|
|
'lateral': 12, |
|
571
|
|
|
'delayed_release': 3, |
|
572
|
|
|
} |
|
573
|
|
|
|
|
574
|
|
|
|
|
575
|
|
|
def ipa_to_features(ipa: str) -> List[int]: |
|
576
|
|
|
"""Convert IPA to features. |
|
577
|
|
|
|
|
578
|
|
|
This translates an IPA string of one or more phones to a list of ints |
|
579
|
|
|
representing the features of the string. |
|
580
|
|
|
|
|
581
|
|
|
Parameters |
|
582
|
|
|
---------- |
|
583
|
|
|
ipa : str |
|
584
|
|
|
The IPA representation of a phone or series of phones |
|
585
|
|
|
|
|
586
|
1 |
|
Returns |
|
587
|
|
|
------- |
|
588
|
|
|
list of ints |
|
589
|
|
|
A representation of the features of the input string |
|
590
|
|
|
|
|
591
|
|
|
Examples |
|
592
|
|
|
-------- |
|
593
|
|
|
>>> ipa_to_features('mut') |
|
594
|
|
|
[2709662981243185770, 1825831513894594986, 2783230754502126250] |
|
595
|
|
|
>>> ipa_to_features('fon') |
|
596
|
|
|
[2781702983095331242, 1825831531074464170, 2711173160463936106] |
|
597
|
|
|
>>> ipa_to_features('telz') |
|
598
|
|
|
[2783230754502126250, 1826957430176000426, 2693158761954453926, |
|
599
|
|
|
2783230754501863834] |
|
600
|
|
|
|
|
601
|
|
|
.. versionadded:: 0.1.0 |
|
602
|
|
|
|
|
603
|
|
|
""" |
|
604
|
|
|
features = [] |
|
605
|
|
|
pos = 0 |
|
606
|
|
|
ipa = normalize('NFD', ipa.lower()) |
|
607
|
|
|
|
|
608
|
|
|
maxsymlen = max(len(_) for _ in _PHONETIC_FEATURES) |
|
609
|
|
|
|
|
610
|
|
|
while pos < len(ipa): |
|
611
|
|
|
found_match = False |
|
612
|
|
|
for i in range(maxsymlen, 0, -1): |
|
613
|
|
|
if ( |
|
614
|
|
|
pos + i - 1 <= len(ipa) |
|
615
|
1 |
|
and ipa[pos : pos + i] in _PHONETIC_FEATURES |
|
616
|
1 |
|
): |
|
617
|
1 |
|
features.append(_PHONETIC_FEATURES[ipa[pos : pos + i]]) |
|
618
|
|
|
pos += i |
|
619
|
1 |
|
found_match = True |
|
620
|
|
|
|
|
621
|
1 |
|
if not found_match: |
|
622
|
1 |
|
features.append(-1) |
|
623
|
1 |
|
pos += 1 |
|
624
|
1 |
|
|
|
625
|
|
|
return features |
|
626
|
|
|
|
|
627
|
|
|
|
|
628
|
1 |
|
def ipa_to_feature_dicts(ipa: str) -> List[Dict[str, str]]: |
|
629
|
1 |
|
"""Convert IPA to a feature dict list. |
|
630
|
1 |
|
|
|
631
|
|
|
This translates an IPA string of one or more phones to a list of dicts |
|
632
|
1 |
|
representing the features of the string. |
|
633
|
1 |
|
|
|
634
|
1 |
|
Parameters |
|
635
|
|
|
---------- |
|
636
|
1 |
|
ipa : str |
|
637
|
|
|
The IPA representation of a phone or series of phones |
|
638
|
|
|
|
|
639
|
1 |
|
Returns |
|
640
|
|
|
------- |
|
641
|
|
|
list of dicts |
|
642
|
|
|
A representation of the features of the input string |
|
643
|
|
|
|
|
644
|
|
|
Examples |
|
645
|
|
|
-------- |
|
646
|
|
|
>>> ipa_to_feature_dicts('mut') |
|
647
|
|
|
[{'syllabic': '-', |
|
648
|
|
|
'consonantal': '+', |
|
649
|
|
|
'sonorant': '+', |
|
650
|
|
|
'approximant': '-', |
|
651
|
|
|
'labial': '+', |
|
652
|
|
|
'round': '-', |
|
653
|
|
|
'protruded': '-', |
|
654
|
|
|
'compressed': '-', |
|
655
|
|
|
'labiodental': '-', |
|
656
|
|
|
'coronal': '-', |
|
657
|
|
|
'anterior': '0', |
|
658
|
|
|
'distributed': '0', |
|
659
|
|
|
'dorsal': '-', |
|
660
|
|
|
'high': '0', |
|
661
|
|
|
'low': '0', |
|
662
|
|
|
'front': '0', |
|
663
|
|
|
'back': '0', |
|
664
|
|
|
'tense': '0', |
|
665
|
|
|
'pharyngeal': '-', |
|
666
|
|
|
'atr': '0', |
|
667
|
|
|
'rtr': '0', |
|
668
|
|
|
'voice': '+', |
|
669
|
|
|
'spread_glottis': '-', |
|
670
|
|
|
'constricted_glottis': '-', |
|
671
|
|
|
'glottalic_suction': '-', |
|
672
|
|
|
'velaric_suction': '-', |
|
673
|
|
|
'continuant': '-', |
|
674
|
|
|
'nasal': '+', |
|
675
|
|
|
'strident': '-', |
|
676
|
|
|
'lateral': '-', |
|
677
|
|
|
'delayed_release': '-'}, |
|
678
|
|
|
{'syllabic': '+', |
|
679
|
|
|
'consonantal': '-', |
|
680
|
|
|
'sonorant': '+', |
|
681
|
|
|
'approximant': '+', |
|
682
|
|
|
'labial': '+', |
|
683
|
|
|
'round': '+', |
|
684
|
|
|
'protruded': '-', |
|
685
|
|
|
'compressed': '-', |
|
686
|
|
|
'labiodental': '-', |
|
687
|
|
|
'coronal': '-', |
|
688
|
|
|
'anterior': '0', |
|
689
|
|
|
'distributed': '0', |
|
690
|
|
|
'dorsal': '+', |
|
691
|
|
|
'high': '+', |
|
692
|
|
|
'low': '-', |
|
693
|
|
|
'front': '-', |
|
694
|
|
|
'back': '+', |
|
695
|
|
|
'tense': '+', |
|
696
|
|
|
'pharyngeal': '+', |
|
697
|
|
|
'atr': '+', |
|
698
|
|
|
'rtr': '-', |
|
699
|
|
|
'voice': '+', |
|
700
|
|
|
'spread_glottis': '-', |
|
701
|
|
|
'constricted_glottis': '-', |
|
702
|
|
|
'glottalic_suction': '-', |
|
703
|
|
|
'velaric_suction': '-', |
|
704
|
|
|
'continuant': '+', |
|
705
|
|
|
'nasal': '-', |
|
706
|
|
|
'strident': '-', |
|
707
|
|
|
'lateral': '-', |
|
708
|
|
|
'delayed_release': '-'}, |
|
709
|
|
|
{'syllabic': '-', |
|
710
|
|
|
'consonantal': '+', |
|
711
|
|
|
'sonorant': '-', |
|
712
|
|
|
'approximant': '-', |
|
713
|
|
|
'labial': '-', |
|
714
|
|
|
'round': '0', |
|
715
|
|
|
'protruded': '0', |
|
716
|
|
|
'compressed': '0', |
|
717
|
|
|
'labiodental': '0', |
|
718
|
|
|
'coronal': '+', |
|
719
|
|
|
'anterior': '+', |
|
720
|
|
|
'distributed': '-', |
|
721
|
|
|
'dorsal': '-', |
|
722
|
|
|
'high': '0', |
|
723
|
|
|
'low': '0', |
|
724
|
|
|
'front': '0', |
|
725
|
|
|
'back': '0', |
|
726
|
|
|
'tense': '0', |
|
727
|
|
|
'pharyngeal': '-', |
|
728
|
|
|
'atr': '0', |
|
729
|
|
|
'rtr': '0', |
|
730
|
|
|
'voice': '-', |
|
731
|
|
|
'spread_glottis': '-', |
|
732
|
|
|
'constricted_glottis': '-', |
|
733
|
|
|
'glottalic_suction': '-', |
|
734
|
|
|
'velaric_suction': '-', |
|
735
|
|
|
'continuant': '-', |
|
736
|
|
|
'nasal': '-', |
|
737
|
|
|
'strident': '-', |
|
738
|
|
|
'lateral': '-', |
|
739
|
|
|
'delayed_release': '-'}] |
|
740
|
|
|
|
|
741
|
|
|
.. versionadded:: 0.4.1 |
|
742
|
|
|
|
|
743
|
|
|
""" |
|
744
|
|
|
features = [] |
|
745
|
|
|
pos = 0 |
|
746
|
|
|
ipa = normalize('NFD', ipa.lower()) |
|
747
|
|
|
|
|
748
|
|
|
maxsymlen = max(len(_) for _ in _PHONETIC_FEATURES) |
|
749
|
|
|
|
|
750
|
|
|
while pos < len(ipa): |
|
751
|
|
|
found_match = False |
|
752
|
|
|
for i in range(maxsymlen, 0, -1): |
|
753
|
|
|
if ( |
|
754
|
|
|
pos + i - 1 <= len(ipa) |
|
755
|
1 |
|
and ipa[pos : pos + i] in _PHONETIC_FEATURES |
|
756
|
1 |
|
): |
|
757
|
1 |
|
feature_int = _PHONETIC_FEATURES[ipa[pos : pos + i]] |
|
758
|
|
|
feature_dict = {} |
|
759
|
1 |
|
for feature in _FEATURE_MASK.keys(): |
|
760
|
|
|
# each feature mask contains two bits, one each for - and + |
|
761
|
1 |
|
mask = _FEATURE_MASK[feature] |
|
762
|
1 |
|
# the lower bit represents + |
|
763
|
1 |
|
pos_mask = mask >> 1 |
|
764
|
1 |
|
|
|
765
|
|
|
masked = feature_int & mask |
|
766
|
|
|
if masked == 0: |
|
767
|
|
|
feature_dict[feature] = '0' # 0 |
|
768
|
1 |
|
elif masked == mask: |
|
769
|
1 |
|
feature_dict[feature] = '+/-' # +/- |
|
770
|
1 |
|
elif masked & pos_mask: |
|
771
|
|
|
feature_dict[feature] = '+' # + |
|
772
|
1 |
|
else: |
|
773
|
|
|
feature_dict[feature] = '-' # - |
|
774
|
1 |
|
features.append(feature_dict) |
|
775
|
|
|
pos += i |
|
776
|
1 |
|
found_match = True |
|
777
|
1 |
|
|
|
778
|
1 |
|
if not found_match: |
|
779
|
1 |
|
features.append({}) |
|
780
|
1 |
|
pos += 1 |
|
781
|
1 |
|
|
|
782
|
1 |
|
return features |
|
783
|
|
|
|
|
784
|
1 |
|
|
|
785
|
1 |
|
def get_feature(vector: List[int], feature: str) -> List[Union[int, float]]: |
|
786
|
1 |
|
"""Get a feature vector. |
|
787
|
1 |
|
|
|
788
|
|
|
This returns a list of ints, equal in length to the vector input, |
|
789
|
1 |
|
representing presence/absence/neutrality with respect to a particular |
|
790
|
1 |
|
phonetic feature. |
|
791
|
1 |
|
|
|
792
|
|
|
Parameters |
|
793
|
1 |
|
---------- |
|
794
|
|
|
vector : list |
|
795
|
|
|
A tuple or list of ints representing the phonetic features of a phone |
|
796
|
1 |
|
or series of phones (such as is returned by the ipa_to_features |
|
797
|
|
|
function) |
|
798
|
|
|
feature : str |
|
799
|
|
|
A feature name from the set: |
|
800
|
|
|
|
|
801
|
|
|
- ``syllabic`` |
|
802
|
|
|
- ``consonantal`` |
|
803
|
|
|
- ``sonorant`` |
|
804
|
|
|
- ``approximant`` |
|
805
|
|
|
- ``labial`` |
|
806
|
|
|
- ``round`` |
|
807
|
|
|
- ``protruded`` |
|
808
|
|
|
- ``compressed`` |
|
809
|
|
|
- ``labiodental`` |
|
810
|
|
|
- ``coronal`` |
|
811
|
|
|
- ``anterior`` |
|
812
|
|
|
- ``distributed`` |
|
813
|
|
|
- ``dorsal`` |
|
814
|
|
|
- ``high`` |
|
815
|
|
|
- ``low`` |
|
816
|
|
|
- ``front`` |
|
817
|
|
|
- ``back`` |
|
818
|
|
|
- ``tense`` |
|
819
|
|
|
- ``pharyngeal`` |
|
820
|
|
|
- ``atr`` |
|
821
|
|
|
- ``rtr`` |
|
822
|
|
|
- ``voice`` |
|
823
|
|
|
- ``spread_glottis`` |
|
824
|
|
|
- ``constricted_glottis`` |
|
825
|
|
|
- ``glottalic_suction`` |
|
826
|
|
|
- ``velaric_suction`` |
|
827
|
|
|
- ``continuant`` |
|
828
|
|
|
- ``nasal`` |
|
829
|
|
|
- ``strident`` |
|
830
|
|
|
- ``lateral`` |
|
831
|
|
|
- ``delayed_release`` |
|
832
|
|
|
|
|
833
|
|
|
Returns |
|
834
|
|
|
------- |
|
835
|
|
|
list of ints |
|
836
|
|
|
A list indicating presence/absence/neutrality with respect to the |
|
837
|
|
|
feature |
|
838
|
|
|
|
|
839
|
|
|
Raises |
|
840
|
|
|
------ |
|
841
|
|
|
AttributeError |
|
842
|
|
|
feature must be one of ... |
|
843
|
|
|
|
|
844
|
|
|
Examples |
|
845
|
|
|
-------- |
|
846
|
|
|
>>> tails = ipa_to_features('telz') |
|
847
|
|
|
>>> get_feature(tails, 'consonantal') |
|
848
|
|
|
[1, -1, 1, 1] |
|
849
|
|
|
>>> get_feature(tails, 'sonorant') |
|
850
|
|
|
[-1, 1, 1, -1] |
|
851
|
|
|
>>> get_feature(tails, 'nasal') |
|
852
|
|
|
[-1, -1, -1, -1] |
|
853
|
|
|
>>> get_feature(tails, 'coronal') |
|
854
|
|
|
[1, -1, 1, 1] |
|
855
|
|
|
|
|
856
|
|
|
.. versionadded:: 0.1.0 |
|
857
|
|
|
|
|
858
|
|
|
""" |
|
859
|
|
|
# :param bool binary: if False, -1, 0, & 1 represent -, 0, & + |
|
860
|
|
|
# if True, only binary oppositions are allowed: |
|
861
|
|
|
# 0 & 1 represent - & + and 0s are mapped to - |
|
862
|
|
|
|
|
863
|
|
|
if feature not in _FEATURE_MASK: |
|
864
|
|
|
raise AttributeError( |
|
865
|
|
|
"feature must be one of: '{}'".format( |
|
866
|
|
|
"', '".join(_FEATURE_MASK.keys()) |
|
867
|
|
|
) |
|
868
|
|
|
) |
|
869
|
|
|
|
|
870
|
|
|
# each feature mask contains two bits, one each for - and + |
|
871
|
|
|
mask = _FEATURE_MASK[feature] |
|
872
|
|
|
# the lower bit represents + |
|
873
|
|
|
pos_mask = mask >> 1 |
|
874
|
1 |
|
retvec = [] |
|
875
|
1 |
|
for char in vector: |
|
876
|
|
|
if char < 0: |
|
877
|
|
|
retvec.append(float('NaN')) |
|
878
|
|
|
else: |
|
879
|
|
|
masked = char & mask |
|
880
|
|
|
if masked == 0: |
|
881
|
|
|
retvec.append(0) # 0 |
|
882
|
1 |
|
elif masked == mask: |
|
883
|
|
|
retvec.append(2) # +/- |
|
884
|
1 |
|
elif masked & pos_mask: |
|
885
|
1 |
|
retvec.append(1) # + |
|
886
|
1 |
|
else: |
|
887
|
1 |
|
retvec.append(-1) # - |
|
888
|
1 |
|
|
|
889
|
|
|
return retvec |
|
890
|
1 |
|
|
|
891
|
1 |
|
|
|
892
|
1 |
|
def cmp_features( |
|
893
|
1 |
|
feat1: int, |
|
894
|
1 |
|
feat2: int, |
|
895
|
1 |
|
weights: Optional[ |
|
896
|
1 |
|
Union[Sequence[Union[int, float]], Dict[str, Union[int, float]]] |
|
897
|
|
|
] = None, |
|
898
|
1 |
|
) -> float: |
|
899
|
|
|
"""Compare features. |
|
900
|
1 |
|
|
|
901
|
|
|
This returns a number in the range [0, 1] representing a comparison of two |
|
902
|
|
|
feature bundles. |
|
903
|
1 |
|
|
|
904
|
|
|
If one of the bundles is negative, -1 is returned (for unknown values) |
|
905
|
|
|
|
|
906
|
|
|
If the bundles are identical, 1 is returned. |
|
907
|
|
|
|
|
908
|
|
|
If they are inverses of one another, 0 is returned. |
|
909
|
|
|
|
|
910
|
|
|
Otherwise, a float representing their similarity is returned. |
|
911
|
|
|
|
|
912
|
|
|
Parameters |
|
913
|
|
|
---------- |
|
914
|
|
|
feat1 : int |
|
915
|
|
|
A feature bundle |
|
916
|
|
|
feat2 : int |
|
917
|
|
|
A feature bundle |
|
918
|
|
|
weights : None or list or tuple or dict |
|
919
|
|
|
If None, all features are of equal significance and a simple normalized |
|
920
|
|
|
hamming distance of the features is calculated. If a list or tuple |
|
921
|
|
|
of numeric values is supplied, the values are inferred as the weights |
|
922
|
|
|
for each feature, in order of the features listed in _FEATURE_MASK. |
|
923
|
|
|
If a dict is supplied, its key values should match keys in |
|
924
|
|
|
_FEATURE_MASK to which each weight (value) should be assigned. Missing |
|
925
|
|
|
values in all cases are assigned a weight of 0 and will be omitted from |
|
926
|
|
|
the comparison. |
|
927
|
|
|
|
|
928
|
|
|
Returns |
|
929
|
|
|
------- |
|
930
|
|
|
float |
|
931
|
|
|
A comparison of the feature bundles |
|
932
|
|
|
|
|
933
|
|
|
Examples |
|
934
|
|
|
-------- |
|
935
|
|
|
>>> cmp_features(ipa_to_features('l')[0], ipa_to_features('l')[0]) |
|
936
|
|
|
1.0 |
|
937
|
|
|
>>> cmp_features(ipa_to_features('l')[0], ipa_to_features('n')[0]) |
|
938
|
|
|
0.8709677419354839 |
|
939
|
|
|
>>> cmp_features(ipa_to_features('l')[0], ipa_to_features('z')[0]) |
|
940
|
|
|
0.8709677419354839 |
|
941
|
|
|
>>> cmp_features(ipa_to_features('l')[0], ipa_to_features('i')[0]) |
|
942
|
|
|
0.564516129032258 |
|
943
|
|
|
|
|
944
|
|
|
.. versionadded:: 0.1.0 |
|
945
|
|
|
.. versionchanged:: 0.4.1 |
|
946
|
|
|
Added weights parameter for modifiable feature weighting |
|
947
|
|
|
|
|
948
|
|
|
""" |
|
949
|
|
|
if feat1 < 0 or feat2 < 0: |
|
950
|
|
|
return 0.0 |
|
951
|
|
|
if feat1 == feat2: |
|
952
|
|
|
return 1.0 |
|
953
|
|
|
|
|
954
|
1 |
|
# This should be handled some other way since this will take a long time |
|
955
|
1 |
|
# when done repeatedly. Maybe convert to a class & save the weights list. |
|
956
|
1 |
|
if weights is not None: |
|
957
|
1 |
|
if isinstance(weights, dict): |
|
958
|
|
|
weights = [ |
|
959
|
|
|
weights[feature] if feature in weights else 0 |
|
960
|
|
|
for feature in sorted( |
|
961
|
1 |
|
_FEATURE_MASK, key=_FEATURE_MASK.get, reverse=True |
|
962
|
1 |
|
) |
|
963
|
1 |
|
] |
|
964
|
|
|
elif isinstance(weights, (list, tuple)): |
|
965
|
|
|
weights = list(weights) + [0] * (len(_FEATURE_MASK) - len(weights)) |
|
966
|
|
|
else: |
|
967
|
|
|
raise TypeError('weights must be a dist, list, or tuple.') |
|
968
|
|
|
|
|
969
|
1 |
|
magnitude = sum(weights) if weights else len(_FEATURE_MASK) |
|
970
|
1 |
|
|
|
971
|
|
|
""" |
|
972
|
1 |
|
# Alternate implementation |
|
973
|
|
|
diff_feats = 0 |
|
974
|
1 |
|
i = 0 |
|
975
|
|
|
while feat1 or feat2: |
|
976
|
1 |
|
f1 = feat1 & 0b11 |
|
977
|
1 |
|
f2 = feat2 & 0b11 |
|
978
|
1 |
|
if (not (0b11 in {f1, f2} and (f1 in {0b01, 0b10} or |
|
979
|
1 |
|
f2 in {0b01, 0b10}))) and (f1 != f2): |
|
980
|
1 |
|
diff_feats += weights[i] if weights else 1 |
|
981
|
1 |
|
|
|
982
|
1 |
|
feat1 >>= 2 |
|
983
|
1 |
|
feat2 >>= 2 |
|
984
|
1 |
|
i += 1 |
|
985
|
1 |
|
|
|
986
|
1 |
|
return 1 - (diff_feats / magnitude) |
|
987
|
1 |
|
""" |
|
988
|
|
|
|
|
989
|
|
|
featxor = feat1 ^ feat2 |
|
990
|
|
|
diffbits = 0.0 |
|
991
|
|
|
i = 0 |
|
992
|
|
|
while featxor: |
|
993
|
|
|
if featxor & 0b1: |
|
994
|
|
|
diffbits += weights[i] if weights else 1 |
|
995
|
|
|
featxor >>= 1 |
|
996
|
|
|
if featxor & 0b1: |
|
997
|
|
|
diffbits += weights[i] if weights else 1 |
|
998
|
|
|
featxor >>= 1 |
|
999
|
|
|
i += 1 |
|
1000
|
|
|
return 1 - (0 if not diffbits else (diffbits / (2 * magnitude))) |
|
1001
|
|
|
|
|
1002
|
|
|
|
|
1003
|
|
|
if __name__ == '__main__': |
|
1004
|
|
|
import doctest |
|
1005
|
|
|
|
|
1006
|
|
|
doctest.testmod() |
|
1007
|
|
|
|
chrislit /
abydos
