icuSources/data/translit/ro_ro_FONIPA.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: ro_ro_FONIPA.txt
   8 # Generated from CLDR
   9 #
  10
  11 # Romanian orthography to phonemic transcription.
  12 # http://en.wikipedia.org/wiki/Romanian_phonology
  13 #
  14 # TODO: Currently this transform does not palatalize consonants.
  15 $VowelEI = [e i î];
  16 $VowelAOU = [a â ă o u];
  17 $Vowel = [$VowelEI $VowelAOU];
  18 $Boundary =  [^[:L:][:M:][:N:]];
  19 :: NFC () ;
  20 :: Lower () ;
  21 # Special cases.
  22 eoai → eo\u032Faj ;  # eg. leoaică → /leo\u032Fajkə/, not /le\u032Fo\u032Faikə/
  23 # Triphthongs.
  24 eai → e\u032Faj ;
  25 eau → e\u032Faw ;
  26 eoa → e\u032Fo\u032Fa ;
  27 i\-ai → jaj ;
  28 ia\-i → jaj ;
  29 iau → jaw ;
  30 iei → jej ;
  31 ieu → jew ;
  32 $Boundary {eu} → jew ;
  33 ioa → jo\u032Fa ;
  34 ioi → joj ;
  35 i\-oi → joj ;
  36 iou → jow ;
  37 oai → o\u032Faj ;
  38 uai → waj ;
  39 uau → waw ;
  40 uăi → wəj ;
  41 # Diphthongs.
  42 ai → aj ;
  43 âi → ɨj ;
  44 ăi → əj ;
  45 au} r → au ;
  46 au → aw ;
  47 âu → ɨw ;
  48 ău → əw ;
  49 ea → e\u032Fa ;
  50 ei → ej ;
  51 eo → e\u032Fo ;
  52 e\-o → e\u032Fo ;
  53 eu → ew ;
  54 e\-u → e\u032Fu ;
  55 ia → ja ;
  56 i\-a → ja ;
  57 ie → je ;
  58 ii → ij ;
  59 io → jo ;
  60 iu} [$Vowel $Boundary] → iw ;
  61 iu → ju ;
  62 oa → o\u032Fa ;
  63 oi → oj ;
  64 ou → ow ;
  65 ua → wa ;
  66 uâ → wɨ ;
  67 uă → wə ;
  68 ue → we ;
  69 ui → uj ;
  70 uu → uw ;
  71 a → a ;
  72 â → ɨ ;
  73 ă → ə ;
  74 b → b ;
  75 ch → k ;
  76 {c [ei]} $VowelAOU → t\u0361ʃ ;
  77 {c} [ei] → t\u0361ʃ ;
  78 c → k ;
  79 d → d ;
  80 e → e ;
  81 f → f ;
  82 gh → ɡ ;
  83 {g [ei]} $VowelAOU → d\u0361ʒ ;
  84 {g} [ei] → d\u0361ʒ ;
  85 g → ɡ ;
  86 h → h ;
  87 i → i ;
  88 î → ɨ ;
  89 j → ʒ ;
  90 k → k ;
  91 l → l ;
  92 m → m ;
  93 ng → ŋ ;
  94 n → n ;
  95 o → o ;
  96 p → p ;
  97 q → k ;
  98 r → r ;
  99 s → s ;
 100 ş → ʃ ;
 101 ș → ʃ ;
 102 t → t ;
 103 ţ → t\u0361s ;
 104 ț → t\u0361s ;
 105 u → u ;
 106 v → v ;
 107 x → ks ;
 108 y → i ;
 109 z → z ;
 110 [:P:]+ → ' ';
 111 # Romanian does not have any gemination.
 112 # https://en.wikipedia.org/wiki/Gemination#Latin_and_Romance_languages
 113 ::null;
 114 pp+ → p;
 115 bb+ → b;
 116 tt+ → t;
 117 dd+ → d;
 118 kk+ → k;
 119 dd+ → d;
 120 ɡɡ+ → ɡ;
 121 ff+ → f;
 122 vv+ → v;
 123 hh+ → h;
 124 ss+ → s;
 125 zz+ → z;
 126 ʃʃ+ → ʃ;
 127 ʒʒ+ → ʒ;
 128 rr+ → r;
 129 ll+ → l;
 130 jj+ → j;
 131 ww+ → w;
 132