diff options
| author | Matthieu Pignolet <m@mpgn.dev> | 2025-05-05 19:06:18 +0400 | 
|---|---|---|
| committer | Matthieu Pignolet <m@mpgn.dev> | 2025-05-05 19:06:18 +0400 | 
| commit | c4d22da11b779d92faca44ac2a6b11cefc0e3d77 (patch) | |
| tree | ad7240b525c6ef7c03cef9c00242bffdfa3448ac /src/extract.json | |
| parent | fe305988e05f8a59a80459577712b84da00a6e69 (diff) | |
feat: add extract script for constants
Diffstat (limited to 'src/extract.json')
| -rw-r--r-- | src/extract.json | 1009 | 
1 files changed, 1009 insertions, 0 deletions
diff --git a/src/extract.json b/src/extract.json new file mode 100644 index 0000000..4e952e1 --- /dev/null +++ b/src/extract.json @@ -0,0 +1,1009 @@ +{ +  "C_skip": 10, +  "C_sub": 35, +  "C_exp": 45, +  "C_vwl": 5, +  "consonants": [ +    "B", +    "N", +    "R", +    "b", +    "c", +    "d", +    "f", +    "g", +    "h", +    "j", +    "k", +    "l", +    "m", +    "n", +    "p", +    "q", +    "r", +    "s", +    "t", +    "v", +    "x", +    "z", +    "\u00e7", +    "\u00f0", +    "\u0127", +    "\u014b", +    "\u0256", +    "\u025f", +    "\u0262", +    "\u0263", +    "\u0266", +    "\u026c", +    "\u026e", +    "\u0270", +    "\u0271", +    "\u0272", +    "\u0273", +    "\u0274", +    "\u0278", +    "\u0279", +    "\u027b", +    "\u027d", +    "\u027e", +    "\u0280", +    "\u0281", +    "\u0282", +    "\u0283", +    "\u0288", +    "\u028b", +    "\u0290 ", +    "\u0292", +    "\u0294", +    "\u0295", +    "\u0299", +    "\u029d", +    "\u03b2", +    "\u03b8", +    "\u03c7", +    "\u0290", +    "w" +  ], +  "R_c": [ +    "aspirated", +    "lateral", +    "manner", +    "nasal", +    "place", +    "retroflex", +    "syllabic", +    "voice" +  ], +  "R_v": [ +    "back", +    "lateral", +    "long", +    "manner", +    "nasal", +    "place", +    "retroflex", +    "round", +    "syllabic", +    "voice" +  ], +  "similarity_matrix": { +    "bilabial": 1.0, +    "labiodental": 0.95, +    "dental": 0.9, +    "alveolar": 0.85, +    "retroflex": 0.8, +    "palato-alveolar": 0.75, +    "palatal": 0.7, +    "velar": 0.6, +    "uvular": 0.5, +    "pharyngeal": 0.3, +    "glottal": 0.1, +    "labiovelar": 1.0, +    "vowel": -1.0, +    "stop": 1.0, +    "affricate": 0.9, +    "fricative": 0.85, +    "trill": 0.7, +    "tap": 0.65, +    "approximant": 0.6, +    "high vowel": 0.4, +    "mid vowel": 0.2, +    "low vowel": 0.0, +    "vowel2": 0.5, +    "high": 1.0, +    "mid": 0.5, +    "low": 0.0, +    "front": 1.0, +    "central": 0.5, +    "back": 0.0, +    "plus": 1.0, +    "minus": 0.0 +  }, +  "salience": { +    "syllabic": 5, +    "place": 40, +    "manner": 50, +    "voice": 5, +    "nasal": 20, +    "retroflex": 10, +    "lateral": 10, +    "aspirated": 5, +    "long": 0, +    "high": 3, +    "back": 2, +    "round": 2 +  }, +  "feature_matrix": { +    "p": { +      "place": "bilabial", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "b": { +      "place": "bilabial", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "t": { +      "place": "alveolar", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "d": { +      "place": "alveolar", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0288": { +      "place": "retroflex", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0256": { +      "place": "retroflex", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "c": { +      "place": "palatal", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u025f": { +      "place": "palatal", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "k": { +      "place": "velar", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "g": { +      "place": "velar", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "q": { +      "place": "uvular", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0262": { +      "place": "uvular", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0294": { +      "place": "glottal", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "m": { +      "place": "bilabial", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0271": { +      "place": "labiodental", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "n": { +      "place": "alveolar", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0273": { +      "place": "retroflex", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0272": { +      "place": "palatal", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u014b": { +      "place": "velar", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0274": { +      "place": "uvular", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "N": { +      "place": "uvular", +      "manner": "stop", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "plus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0299": { +      "place": "bilabial", +      "manner": "trill", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "B": { +      "place": "bilabial", +      "manner": "trill", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "r": { +      "place": "alveolar", +      "manner": "trill", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0280": { +      "place": "uvular", +      "manner": "trill", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "R": { +      "place": "uvular", +      "manner": "trill", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u027e": { +      "place": "alveolar", +      "manner": "tap", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u027d": { +      "place": "retroflex", +      "manner": "tap", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0278": { +      "place": "bilabial", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u03b2": { +      "place": "bilabial", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "f": { +      "place": "labiodental", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "v": { +      "place": "labiodental", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u03b8": { +      "place": "dental", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u00f0": { +      "place": "dental", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "s": { +      "place": "alveolar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "z": { +      "place": "alveolar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0283": { +      "place": "palato-alveolar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0292": { +      "place": "palato-alveolar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0282": { +      "place": "retroflex", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0290": { +      "place": "retroflex", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u00e7": { +      "place": "palatal", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u029d": { +      "place": "palatal", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "x": { +      "place": "velar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0263": { +      "place": "velar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u03c7": { +      "place": "uvular", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0281": { +      "place": "uvular", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0127": { +      "place": "pharyngeal", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0295": { +      "place": "pharyngeal", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "h": { +      "place": "glottal", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0266": { +      "place": "glottal", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u026c": { +      "place": "alveolar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "minus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "plus", +      "aspirated": "minus" +    }, +    "\u026e": { +      "place": "alveolar", +      "manner": "fricative", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "plus", +      "aspirated": "minus" +    }, +    "\u028b": { +      "place": "labiodental", +      "manner": "approximant", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0279": { +      "place": "alveolar", +      "manner": "approximant", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u027b": { +      "place": "retroflex", +      "manner": "approximant", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "plus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "j": { +      "place": "palatal", +      "manner": "approximant", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "\u0270": { +      "place": "velar", +      "manner": "approximant", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "l": { +      "place": "alveolar", +      "manner": "approximant", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "plus", +      "aspirated": "minus" +    }, +    "w": { +      "place": "labiovelar", +      "manner": "approximant", +      "syllabic": "minus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "aspirated": "minus" +    }, +    "i": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "high", +      "back": "front", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "y": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "high", +      "back": "front", +      "round": "plus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "e": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "front", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "E": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "front", +      "round": "minus", +      "long": "plus", +      "aspirated": "minus" +    }, +    "\u00f8": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "front", +      "round": "plus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "\u025b": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "front", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "\u0153": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "front", +      "round": "plus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "\u00e6": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "low", +      "back": "front", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "a": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "low", +      "back": "front", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "A": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "low", +      "back": "front", +      "round": "minus", +      "long": "plus", +      "aspirated": "minus" +    }, +    "\u0268": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "high", +      "back": "central", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "\u0289": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "high", +      "back": "central", +      "round": "plus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "\u0259": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "central", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "u": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "high", +      "back": "back", +      "round": "plus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "U": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "high", +      "back": "back", +      "round": "plus", +      "long": "plus", +      "aspirated": "minus" +    }, +    "o": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "back", +      "round": "plus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "O": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "back", +      "round": "plus", +      "long": "plus", +      "aspirated": "minus" +    }, +    "\u0254": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "mid", +      "back": "back", +      "round": "plus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "\u0252": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "low", +      "back": "back", +      "round": "minus", +      "long": "minus", +      "aspirated": "minus" +    }, +    "I": { +      "place": "vowel", +      "manner": "vowel2", +      "syllabic": "plus", +      "voice": "plus", +      "nasal": "minus", +      "retroflex": "minus", +      "lateral": "minus", +      "high": "high", +      "back": "front", +      "round": "minus", +      "long": "plus", +      "aspirated": "minus" +    } +  } +}  | 
