languages/Names.php

   1 <?php
   2 /**
   3   * These determine things like interwikis, language selectors, and so on.
   4   * Safe to change without running scripts on the respective sites.
   5   *
   6   * @ingroup Language
   7   */
   8 /* private */ $wgLanguageNames = array(
   9         'aa' => 'Afar',                 # Afar
  10         'ab' => 'Аҧсуа',   # Abkhaz, should possibly add ' бысжѡа'
  11         'af' => 'Afrikaans',    # Afrikaans
  12         'ak' => 'Akan',         # Akan
  13         'aln' => 'Gegë',       # Gheg Albanian
  14         'als' => 'Alemannisch', # Alemannic -- not a valid code, for compatibility. See gsw.
  15         'am' => 'አማርኛ', # Amharic
  16         'an' => 'Aragonés',    # Aragonese
  17         'ang' => 'Anglo-Saxon', # Old English
  18         'ar' => 'العربية',       # Arabic
  19         'arc' => 'ܐܪܡܝܐ',  # Aramaic
  20         'arn' => 'Mapudungun',  # Mapuche, Mapudungu, Araucanian (Araucano)
  21         'arz' => 'مصرى',    # Egyptian Spoken Arabic
  22         'as' => 'অসমীয়া',   # Assamese
  23         'ast' => 'Asturianu',   # Asturian
  24         'av' => 'Авар',     # Avar
  25         'avk' => 'Kotava', # Kotava
  26         'ay' => 'Aymar',                # Aymara, should possibly be Aymará
  27         'az' => 'Azərbaycan',  # Azerbaijani
  28         'ba' => 'Башҡорт',       # Bashkir
  29         'bar' => 'Boarisch',    # Bavarian (Austro-Bavarian and South Tyrolean)
  30         'bat-smg' => 'Žemaitėška', # Samogitian
  31         'bcl' => 'Bikol Central', # Bikol: Central Bicolano language
  32         'be' => 'Беларуская', #  Belarusian normative
  33         'be-tarask' => 'Беларуская (тарашкевіца)', # Belarusian in Taraskievica orthography
  34         'be-x-old' => 'Беларуская (тарашкевіца)',  # Belarusian in Taraskievica orthography; compat link
  35         'bg' => 'Български',   # Bulgarian
  36         'bh' => 'भोजपुरी',        # Bhojpuri
  37         'bi' => 'Bislama',              # Bislama
  38         'bm' => 'Bamanankan',   # Bambara
  39         'bn' => 'বাংলা',      # Bengali
  40         'bo' => 'བོད་ཡིག',        # Tibetan
  41         'bpy' => 'ইমার ঠার/বিষ্ণুপ্রিয়া মণিপুরী',        # Bishnupriya Manipuri
  42         'br' => 'Brezhoneg',    # Breton
  43         'bs' => 'Bosanski',             # Bosnian
  44         'bto' => 'Iriga Bicolano',              # Iriga Bicolano/Rinconada Bikol
  45         'bug' => 'ᨅᨔ ᨕᨘᨁᨗ', # Bugis
  46         'bxr' => 'Буряад',        # Buryat (Russia)
  47         'ca' => 'Català',      # Catalan
  48         'cbk-zam' => 'Chavacano de Zamboanga',  # Zamboanga Chavacano
  49         'cdo' => 'Mìng-dĕ̤ng-ngṳ̄',       # Min Dong
  50         'ce' => 'Нохчийн',       # Chechen
  51         'ceb' => 'Cebuano',     # Cebuano
  52         'ch' => 'Chamoru',              # Chamorro
  53         'cho' => 'Choctaw',             # Choctaw
  54         'chr' => 'ᏣᎳᎩ', # Cherokee
  55         'chy' => 'Tsetsêhestâhese',   # Cheyenne
  56         'co' => 'Corsu',                # Corsican
  57         'cr' => 'Nēhiyawēwin / ᓀᐦᐃᔭᐍᐏᐣ',                # Cree
  58         'crh' => 'Qırımtatarca',   # Crimean Tatar
  59         'crh-latn' => "\xE2\x80\xAAQırımtatarca (Latin)\xE2\x80\xAC",       # Crimean Tatar (Latin)
  60         'crh-cyrl' => "\xE2\x80\xAAКъырымтатарджа (Кирилл)\xE2\x80\xAC",       # Crimean Tatar (Cyrillic)
  61         'cs' => 'Česky',       # Czech
  62         'csb' => 'Kaszëbsczi', # Cassubian
  63         'cu' => 'Словѣньскъ',         # Old Church Slavonic (ancient language)
  64         'cv' => 'Чăвашла',       # Chuvash
  65         'cy' => 'Cymraeg',              # Welsh
  66         'da' => 'Dansk',                # Danish
  67         'de' => 'Deutsch',              # German ("Du")
  68         'de-formal' => 'Deutsch (Sie-Form)',            # German - formal address ("Sie")
  69         'diq' => 'Zazaki',              # Zazaki
  70         'dk' => 'Dansk (deprecated:da)',                # Unused code currently falls back to Danish, 'da' is correct for the language
  71         'dsb' => 'Dolnoserbski', # Lower Sorbian
  72         'dv' => 'ދިވެހިބަސް',         # Dhivehi
  73         'dz' => 'ཇོང་ཁ',              # Bhutani
  74         'ee' => 'Eʋegbe',      # Éwé
  75         'el' => 'Ελληνικά',     # Greek
  76         'eml' => 'Emiliàn e rumagnòl',        # Emiliano-Romagnolo / Sammarinese
  77         'en' => 'English',              # English
  78         'en-gb' => 'British English',           # British English
  79         'eo' => 'Esperanto',    # Esperanto
  80         'es' => 'Español',     # Spanish
  81         'et' => 'Eesti',                # Estonian
  82         'eu' => 'Euskara',              # Basque
  83         'ext' => 'Estremeñu', # Extremaduran
  84         'fa' => 'فارسی',   # Persian
  85         'ff' => 'Fulfulde',             # Fulfulde, Maasina
  86         'fi' => 'Suomi',                # Finnish
  87         'fiu-vro' => 'Võro',    # Võro
  88         'fj' => 'Na Vosa Vakaviti',     # Fijian
  89         'fo' => 'Føroyskt',    # Faroese
  90         'fr' => 'Français',    # French
  91         'frc' => 'Français cadien', # Cajun French
  92         'frp' => 'Arpetan',     # Franco-Provençal/Arpitan
  93         'fur' => 'Furlan',              # Friulian
  94         'fy' => 'Frysk',                # Frisian
  95         'ga' => 'Gaeilge',              # Irish
  96         'gag' => 'Gagauz',              # Gagauz
  97         'gan' => '贛語',                      # Gan
  98         'gd' => 'Gàidhlig',    # Scots Gaelic
  99         'gl' => 'Galego',               # Galician
 100         'glk' => 'گیلکی',  # Gilaki
 101         'gn' => 'Avañe\'ẽ',  # Guaraní, Paraguayan
 102         'got' => '𐌲𐌿𐍄𐌹𐍃𐌺',    # Gothic
 103         'grc' => 'Ἀρχαία ἑλληνικὴ', # Ancient Greece
 104         'gsw' => 'Alemannisch', # Alemannic
 105         'gu' => 'ગુજરાતી',        # Gujarati
 106         'gv' => 'Gaelg',                # Manx
 107         'ha' => 'هَوُسَ', # Hausa
 108         'hak' => 'Hak-kâ-fa',  # Hakka
 109         'haw' => 'Hawai`i',             # Hawaiian
 110         'he' => 'עברית',   # Hebrew
 111         'hi' => 'हिन्दी',   # Hindi
 112         'hif' => 'Fiji Hindi/फ़ीजी हिन्दी',       # Fijian Hindi (falls back to hif-latn)
 113         'hif-deva' => 'फ़ीजी हिन्दी',     # Hindi
 114         'hif-latn' => 'Fiji Hindi',     # Hindi
 115         'hil' => 'Ilonggo',     # Hiligaynon
 116         'ho' => 'Hiri Motu',    # Hiri Motu
 117         'hr' => 'Hrvatski',             # Croatian
 118         'hsb' => 'Hornjoserbsce',       # Upper Sorbian
 119         'ht'  => 'Kreyòl ayisyen',             # Haitian Creole French
 120         'hu' => 'Magyar',               # Hungarian
 121         'hy' => 'Հայերեն',       # Armenian
 122         'hz' => 'Otsiherero',   # Herero
 123         'ia' => 'Interlingua',  # Interlingua (IALA)
 124         'id' => 'Bahasa Indonesia',     # Indonesian
 125         'ie' => 'Interlingue',  # Interlingue (Occidental)
 126         'ig' => 'Igbo',                 # Igbo
 127         'ii' => 'ꆇꉙ',       # Sichuan Yi
 128         'ik' => 'Iñupiak',     # Inupiak (Inupiatun, Northwest Alaska / Inupiatun, North Alaskan)
 129         'ike-cans' => 'ᐃᓄᒃᑎᑐᑦ',     # Inuktitut, Eastern Canadian/Eastern Canadian "Eskimo"/"Eastern Arctic Eskimo"/Inuit (Unified Canadian Aboriginal Syllabics)
 130         'ike-latn' => 'inuktitut',      # Inuktitut, Eastern Canadian (Latin script)
 131         'ilo' => 'Ilokano',     # Ilokano
 132         'inh' => 'ГІалгІай Ğalğaj',    # Ingush
 133         'io' => 'Ido',                  # Ido
 134         'is' => 'Íslenska',    # Icelandic
 135         'it' => 'Italiano',             # Italian
 136         'iu' => 'ᐃᓄᒃᑎᑐᑦ/inuktitut', # Inuktitut (macro language - do no localise, see ike/ikt - falls back to ike-cans)
 137         'ja' => '日本語',    # Japanese
 138         'jbo' => 'Lojban',              # Lojban
 139         'jut' => 'Jysk',        # Jutish / Jutlandic
 140         'jv' => 'Basa Jawa',    # Javanese
 141         'ka' => 'ქართული',        # Georgian
 142         'kaa' => 'Qaraqalpaqsha',       # Karakalpak
 143         'kab' => 'Taqbaylit',   # Kabyle
 144         'kg' => 'Kongo',                # Kongo, (FIXME!) should probaly be KiKongo or KiKoongo
 145         'ki' => 'Gĩkũyũ',    # Gikuyu
 146         'kj' => 'Kwanyama',             # Kwanyama
 147         'kk' => 'Қазақша',       # Kazakh
 148         'kk-arab' => "\xE2\x80\xABقازاقشا (تٴوتە)\xE2\x80\xAC",     # Kazakh Arabic
 149         'kk-cyrl' => "\xE2\x80\xAAҚазақша (кирил)\xE2\x80\xAC",     # Kazakh Cyrillic
 150         'kk-latn' => "\xE2\x80\xAAQazaqşa (latın)\xE2\x80\xAC",       # Kazakh Latin
 151         'kk-cn' => "\xE2\x80\xABقازاقشا (جۇنگو)\xE2\x80\xAC",       # Kazakh (China)
 152         'kk-kz' => "\xE2\x80\xAAҚазақша (Қазақстан)\xE2\x80\xAC",       # Kazakh (Kazakhstan)
 153         'kk-tr' => "\xE2\x80\xAAQazaqşa (Türkïya)\xE2\x80\xAC",      # Kazakh (Turkey)
 154         'kl' => 'Kalaallisut',  # Inuktitut, Greenlandic/Greenlandic/Kalaallisut (kal)
 155         'km' => 'ភាសាខ្មែរ',  # Khmer, Central
 156         'kn' => 'ಕನ್ನಡ',      # Kannada
 157         'ko' => '한국어',    # Korean
 158         'kr' => 'Kanuri',               # Kanuri, Central
 159         'kri' => 'Krio', # Krio
 160         'krj' => 'Kinaray-a', # Kinaray-a
 161         'ks' => 'कश्मीरी - (كشميري)',       # Kashmiri
 162         'ksh' => 'Ripoarisch',  # Ripuarian
 163         'ku'  => 'Kurdî / كوردی', # Kurdish
 164         'ku-latn' => "\xE2\x80\xAAKurdî (latînî)\xE2\x80\xAC",       # Northern Kurdish Latin script
 165         'ku-arab' => "\xE2\x80\xABكوردي (عەرەبی)\xE2\x80\xAC",       # Northern Kurdish Arabic script
 166         'kv' => 'Коми',     # Komi-Zyrian, cyrillic is common script but also written in latin script
 167         'kw' => 'Kernewek',             # Cornish
 168         'ky' => 'Кыргызча',     # Kirghiz
 169         'la' => 'Latina',               # Latin
 170         'lad' => 'Ladino',      # Ladino
 171         'lb' => 'Lëtzebuergesch',      # Luxemburguish
 172         'lbe' => 'Лакку',  # Lak
 173         'lez' => 'Лезги',  # Lezgi
 174         'lfn' => 'Lingua Franca Nova',  # Lingua Franca Nova
 175         'lg' => 'Luganda',              # Ganda
 176         'li' => 'Limburgs',     # Limburgian
 177         'lij' => 'Líguru',     # Ligurian
 178         'lld' => 'Ladin',       # Ladin
 179         'lmo' => 'Lumbaart',    # Lombard
 180         'ln' => 'Lingála',             # Lingala
 181         'lo' => 'ລາວ',# Laotian
 182         'loz' => 'Silozi', # Lozi
 183         'lt' => 'Lietuvių',    # Lithuanian
 184         'lv' => 'Latviešu',    # Latvian
 185         'lzz' => 'Lazuri Nena', #Laz
 186         'mai' => 'मैथिली', # Maithili
 187         'map-bms' => 'Basa Banyumasan', # Banyumasan
 188         'mdf' => 'мокшень кяль',             # Moksha
 189         'mg' => 'Malagasy',             # Malagasy
 190         'mh' => 'Ebon',                 # Marshallese
 191         'mi' => 'Māori',       # Maori
 192         'mk' => 'Македонски', # Macedonian
 193         'ml' => 'മലയാളം',   # Malayalam
 194         'mn' => 'Монгол', # Halh Mongolian (Cyrillic) (ISO 639-3: khk)
 195         'mo' => 'Молдовеняскэ',     # Moldovan
 196         'mr' => 'मराठी',      # Marathi
 197         'ms' => 'Bahasa Melayu',        # Malay
 198         'mt' => 'Malti',        # Maltese
 199         'mus' => 'Mvskoke',     # Muskogee/Creek
 200         'mwl' => 'Mirandés',   # Mirandese
 201         'my' => 'Myanmasa',             # Burmese
 202         'myv' => 'эрзянь кель',         # Erzya
 203         'mzn' => 'مَزِروني',            # Mazanderani
 204         'na' => 'Ekakairũ Naoero',             # Nauruan
 205         'nah' => 'Nahuatl',             # Nahuatl, en:Wikipedia writes Nahuatlahtolli, while another form is Náhuatl
 206         'nan' => 'Bân-lâm-gú', # Min-nan -- (bug 8217) nan instead of zh-min-nan, http://www.sil.org/iso639-3/codes.asp?order=639_3&letter=n
 207         'nap' => 'Nnapulitano', # Neapolitan
 208         'nb' => "\xE2\x80\xAANorsk (bokmål)\xE2\x80\xAC",              # Norwegian (Bokmal)
 209         'nds' => 'Plattdüütsch',      # Low German ''or'' Low Saxon
 210         'nds-nl' => 'Nedersaksisch',    # Dutch Low Saxon
 211         'ne' => 'नेपाली',   # Nepali
 212         'new' => 'नेपाल भाषा',                # Newar / Nepal Bhasa
 213         'ng' => 'Oshiwambo',            # Ndonga
 214         'niu' => 'Niuē',       # Niuean
 215         'nl' => 'Nederlands',   # Dutch
 216         'nn' => "\xE2\x80\xAANorsk (nynorsk)\xE2\x80\xAC",      # Norwegian (Nynorsk)
 217         'no' => "\xE2\x80\xAANorsk (bokmål)\xE2\x80\xAC",              # Norwegian
 218         'nov' => 'Novial',              # Novial
 219         'nrm' => 'Nouormand',   # Norman
 220         'nso' => 'Sesotho sa Leboa',    # Northern Sotho
 221         'nv' => 'Diné bizaad', # Navajo
 222         'ny' => 'Chi-Chewa',    # Chichewa
 223         'oc' => 'Occitan',              # Occitan
 224         'om' => 'Oromoo',               # Oromo
 225         'or' => 'ଓଡ଼ିଆ',         # Oriya
 226         'os' => 'Иронау', # Ossetic
 227         'pa' => 'ਪੰਜਾਬੀ', # Punjabi
 228         'pag' => 'Pangasinan',  # Pangasinan
 229         'pam' => 'Kapampangan',   # Pampanga
 230         'pap' => 'Papiamentu',  # Papiamentu
 231         'pdc' => 'Deitsch',     # Pennsylvania German
 232         'pdt' => 'Plautdietsch',        # Plautdietsch/Mennonite Low German
 233         'pfl' => 'Pfälzisch',  # Palatinate German
 234         'pi' => 'पािऴ', # Pali
 235         'pih' => 'Norfuk / Pitkern', # Norfuk/Pitcairn/Norfolk
 236         'pl' => 'Polski',               # Polish
 237         'plm' => 'Palembang',           # Palembang
 238         'pms' => 'Piemontèis',         # Piedmontese
 239         'pnt' => 'Ποντιακά',    # Pontic/Pontic Greek
 240         'ps' => 'پښتو',     # Pashto, Northern/Paktu/Pakhtu/Pakhtoo/Afghan/Pakhto/Pashtu/Pushto/Yusufzai Pashto
 241         'pt' => 'Português',   # Portuguese
 242         'pt-br' => 'Português do Brasil',      # Brazilian Portuguese
 243         'qu' => 'Runa Simi',    # Quechua
 244         'rm' => 'Rumantsch',    # Raeto-Romance
 245         'rmy' => 'Romani',      # Vlax Romany
 246         'rn' => 'Kirundi',              # Rundi/Kirundi/Urundi
 247         'ro' => 'Română',     # Romanian
 248         'roa-rup' => 'Armãneashce', # Aromanian
 249         'roa-tara' => 'Tarandíne',     # Tarantino
 250         'ru' => 'Русский',       # Russian
 251         'ruq' => 'Vlăheşte',  # Megleno-Romanian (falls back to ruq-latn)
 252         'ruq-cyrl' => 'Влахесте',       # Megleno-Romanian (Cyrillic script)
 253         'ruq-grek' => 'Βλαεστε', # Megleno-Romanian (Greek script)
 254         'ruq-latn' => 'Vlăheşte',     # Megleno-Romanian (Latin script)
 255         'rw' => 'Kinyarwanda',  # Kinyarwanda, should possibly be Kinyarwandi
 256         'sa' => 'संस्कृत',        # Sanskrit
 257         'sah' => 'Саха тыла', # Sakha
 258         'sc' => 'Sardu',                # Sardinian
 259         'scn' => 'Sicilianu',   # Sicilian
 260         'sco' => 'Scots',       # Scots
 261         'sd' => 'سنڌي',     # Sindhi
 262         'sdc' => 'Sassaresu',   # Sassarese
 263         'se' => 'Sámegiella',  # Northern Sami
 264         'sei' => 'Cmique Itom', # Seri
 265         'sg' => 'Sängö',              # Sango/Sangho
 266         'sh' => 'Srpskohrvatski / Српскохрватски', # Serbocroatian
 267         'shi' => 'Tašlḥiyt',    # Tachelhit
 268         'si' => 'සිංහල',      # Sinhalese
 269         'simple' => 'Simple English',   # Simple English
 270         'sk' => 'Slovenčina',  # Slovak
 271         'sl' => 'Slovenščina',        # Slovenian
 272         'sm' => 'Gagana Samoa', # Samoan
 273         'sma' => 'Åarjelsaemien gïele',       # Southern Sami
 274         'sn' => 'chiShona',             # Shona
 275         'so' => 'Soomaaliga',   # Somali
 276         'sq' => 'Shqip',                # Albanian
 277         'sr' => 'Српски / Srpski',        # Serbian
 278         'sr-ec' => 'ћирилица',  # Serbian cyrillic ekavian
 279         'sr-el' => 'latinica',  # Serbian latin ekavian
 280         'srn' => 'Sranantongo',         # Sranan Tongo
 281         'ss' => 'SiSwati',              # Swati
 282         'st' => 'Sesotho',              # Southern Sotho
 283         'stq' => 'Seeltersk',           # Saterland Frisian
 284         'su' => 'Basa Sunda',   # Sundanese
 285         'sv' => 'Svenska',              # Swedish
 286         'sw' => 'Kiswahili',    # Swahili
 287         'szl' => 'Ślůnski',   # Silesian
 288         'ta' => 'தமிழ்',      # Tamil
 289         'te' => 'తెలుగు',   # Telugu
 290         'tet' => 'Tetun',       # Tetun
 291         'tg' => 'Тоҷикӣ/tojikī', # Tajiki (falls back to tg-cyrl)
 292         'tg-cyrl' => 'Тоҷикӣ',    # Tajiki (Cyrllic script) (default)
 293         'tg-latn' => 'tojikī', # Tajiki (Latin script)
 294         'th' => 'ไทย',    # Thai
 295         'ti' => 'ትግርኛ',         # Tigrinya
 296         'tk' => 'Türkmen',     # Turkmen
 297         'tl' => 'Tagalog',              # Tagalog (Filipino)
 298         #'tlh' => 'tlhIngan-Hol',       # Klingon - no interlanguage links allowed
 299         'tn' => 'Setswana',             # Setswana
 300         'to' => 'faka-Tonga',           # Tonga (Tonga Islands)
 301         'tokipona' => 'Toki Pona',      # Toki Pona
 302         'tp' => 'Toki Pona (deprecated:tokipona)',      # Toki Pona - non-standard language code
 303         'tpi' => 'Tok Pisin',   # Tok Pisin
 304         'tr' => 'Türkçe',     # Turkish
 305         'ts' => 'Xitsonga',             # Tsonga
 306         'tt' => 'Tatarça/Татарча',      # Tatar (multiple scripts - defaults to Latin)
 307         'tt-cyrl' => 'Татарча',  # Tatar (Cyrillic script)
 308         'tt-latn' => 'Tatarça',        # Tatar (Latin script)
 309         'tum' => 'chiTumbuka',  # Tumbuka
 310         'tw' => 'Twi',                  # Twi, (FIXME!)
 311         'ty' => 'Reo Mā`ohi',  # Tahitian
 312         'tyv' => 'Тыва дыл',     # Tyvan
 313         'tzm' => 'ⵜⴰⵎⴰⵣⵉⵖⵜ',    # (Central Morocco) Tamazight
 314         'udm' => 'Удмурт',        # Udmurt
 315         'ug' => 'Uyghurche‎ / ئۇيغۇرچە',      # Uyghur
 316         'uk' => 'Українська', # Ukrainian
 317         'ur' => 'اردو',     # Urdu
 318         'uz' => 'O\'zbek',      # Uzbek
 319         've' => 'Tshivenda',            # Venda
 320         'vec' => 'Vèneto',     # Venetian
 321         'vi' => 'Tiếng Việt',       # Vietnamese
 322         'vls' => 'West-Vlams', # West Flemish
 323         'vo' => 'Volapük',     # Volapük
 324         'wa' => 'Walon',                # Walloon
 325         'war' => 'Winaray', # Waray-Waray
 326         'wo' => 'Wolof',                # Wolof
 327         'wuu' => '吴语',              # Wu Chinese
 328         'xal' => 'Хальмг',                # Kalmyk-Oirat (Kalmuk, Kalmuck, Kalmack, Qalmaq, Kalmytskii Jazyk, Khal:mag, Oirat, Volga Oirat, European Oirat, Western Mongolian)
 329         'xh' => 'isiXhosa',             # Xhosan
 330         'xmf' => 'მარგალური', # Mingrelian
 331         'ydd' => 'מיזרח־ייִדיש', # Eastern Yiddish
 332         'yi' => 'ייִדיש', # Yiddish
 333         'yo' => 'Yorùbá',     # Yoruba
 334         'yue' => '粵語',      # Cantonese -- (bug 8217) yue instead of zh-yue, http://www.sil.org/iso639-3/codes.asp?order=639_3&letter=y
 335         'za' => '(Cuengh)',             # Zhuang
 336         'zea' => 'Zeêuws',     # Zeeuws/Zeaws
 337         'zh' => '中文',                                               # (Zhōng Wén) - Chinese
 338         'zh-classical' => '古文 / 文言文',                 # Classical Chinese/Literary Chinese
 339         'zh-cn' => "\xE2\x80\xAA中文(中国大陆)\xE2\x80\xAC",      # Chinese (PRC)
 340         'zh-hans' => "\xE2\x80\xAA中文(简体)\xE2\x80\xAC",  # Chinese written using the Simplified Chinese script
 341         'zh-hant' => "\xE2\x80\xAA中文(繁體)\xE2\x80\xAC",  # Chinese written using the Traditional Chinese script
 342         'zh-hk' => "\xE2\x80\xAA中文(香港)\xE2\x80\xAC",            # Chinese (Hong Kong)
 343         'zh-min-nan' => 'Bân-lâm-gú',                                # Min-nan -- (see bug 8217)
 344         'zh-sg' => "\xE2\x80\xAA中文(新加坡)\xE2\x80\xAC",         # Chinese (Singapore)
 345         'zh-tw' => "\xE2\x80\xAA中文(台灣)\xE2\x80\xAC",            # Chinese (Taiwan)
 346         'zh-yue' => '粵語',                                   # Cantonese -- (see bug 8217)
 347         'zu' => 'isiZulu'               # Zulu
 348 );