lang-word-tokenizers / data /Chibchan.json
guipenedo's picture
guipenedo HF staff
do not propagate to the root
49dc1e7 unverified
{
"name": "Chibchan",
"depth": 0,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Chibchan A",
"depth": 1,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Boruca",
"depth": 2,
"iso_1_code": null,
"iso_3_code": "brn",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3374",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Teribe",
"depth": 2,
"iso_1_code": null,
"iso_3_code": "tfr",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3375",
"native_tokenizers": [],
"scripts": [
"Latn"
]
},
{
"name": "Guaymi\u00edc",
"depth": 2,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Ng\u00e4bere",
"depth": 3,
"iso_1_code": null,
"iso_3_code": "gym",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3377",
"native_tokenizers": [],
"scripts": [
"Latn"
]
},
{
"name": "Buglere",
"depth": 3,
"iso_1_code": null,
"iso_3_code": "sab",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3378",
"native_tokenizers": [],
"scripts": [
"Latn"
]
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3376",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Viceitic",
"depth": 2,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Bribri",
"depth": 3,
"iso_1_code": null,
"iso_3_code": "bzd",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3380",
"native_tokenizers": [],
"scripts": [
"Latn"
]
},
{
"name": "Cab\u00e9car",
"depth": 3,
"iso_1_code": null,
"iso_3_code": "cjp",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3381",
"native_tokenizers": [],
"scripts": [
"Latn"
]
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3379",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3373",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Chibchan B",
"depth": 1,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Pech",
"depth": 2,
"iso_1_code": null,
"iso_3_code": "pay",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3383",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Eastern Chibchan",
"depth": 2,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Colombian",
"depth": 3,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Northern Colombian",
"depth": 4,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Chimila",
"depth": 5,
"iso_1_code": null,
"iso_3_code": "cbg",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3387",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Arhuacan",
"depth": 5,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Kogi",
"depth": 6,
"iso_1_code": null,
"iso_3_code": "kog",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3389",
"native_tokenizers": [],
"scripts": [
"Latn"
]
},
{
"name": "Southern and Eastern Arhuacan",
"depth": 6,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Arhuaco",
"depth": 7,
"iso_1_code": null,
"iso_3_code": "arh",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3391",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Guamaca-Atanque",
"depth": 7,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Sanka",
"depth": 8,
"iso_1_code": null,
"iso_3_code": "mbp",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3393",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3392",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3390",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3388",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3386",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Southern Colombian",
"depth": 4,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Bar\u00ed",
"depth": 5,
"iso_1_code": null,
"iso_3_code": "mot",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3395",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Cundicocuyese",
"depth": 5,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Chibcha",
"depth": 6,
"iso_1_code": null,
"iso_3_code": "chb",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3397",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Tunebo, Barro Negro",
"depth": 6,
"iso_1_code": null,
"iso_3_code": "tbn",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3398",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Tunebo, Western",
"depth": 6,
"iso_1_code": null,
"iso_3_code": "tnb",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3399",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Tunebo, Angosturas",
"depth": 6,
"iso_1_code": null,
"iso_3_code": "tnd",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3400",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Tunebo, Central",
"depth": 6,
"iso_1_code": null,
"iso_3_code": "tuf",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3401",
"native_tokenizers": [],
"scripts": [
"Latn"
]
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3396",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3394",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3385",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Cuna",
"depth": 3,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Kuna, San Blas",
"depth": 4,
"iso_1_code": null,
"iso_3_code": "cuk",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3403",
"native_tokenizers": [],
"scripts": [
"Latn"
]
},
{
"name": "Kuna, Border",
"depth": 4,
"iso_1_code": null,
"iso_3_code": "kvn",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3404",
"native_tokenizers": [],
"scripts": [
"Latn"
]
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3402",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3384",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Votic",
"depth": 2,
"iso_1_code": null,
"iso_3_code": null,
"children": [
{
"name": "Mal\u00e9ku Ja\u00edka",
"depth": 3,
"iso_1_code": null,
"iso_3_code": "gut",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3406",
"native_tokenizers": [],
"scripts": []
},
{
"name": "Rama",
"depth": 3,
"iso_1_code": null,
"iso_3_code": "rma",
"children": [],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3407",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3405",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3382",
"native_tokenizers": [],
"scripts": []
}
],
"family": "Chibchan",
"tokenizers": {},
"node_i": "3372",
"native_tokenizers": [],
"scripts": []
}