icuSources/data/translit/pl_pl_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: pl_pl_FONIPA.txt
   8 # Generated from CLDR
   9 #
  10
  11 # Polish orthography to phonemic transcription.
  12 # http://en.wikipedia.org/wiki/Polish_phonology
  13 #
  14 # Transform input to normalized form NFC, and to lowercase.
  15 :: NFC () ;
  16 :: Lower () ;
  17 #
  18 #
  19 # Definitions.
  20 $voiceless = [cfhkpst];
  21 $vowel = [ aąeęioóuy ];
  22 #
  23 #
  24 # Digraphs and Trigraphs.
  25 #
  26 ch } i → ç ;
  27 ch → x ;
  28 ci } $vowel → t \u0361 ɕ ;
  29 ci → t \u0361 ɕ i ;
  30 cz → t \u0361 ʂ ;
  31 dzi } $vowel → d \u0361 ʑ ;
  32 dzi → d \u0361 ʑ i ;
  33 dz } $voiceless → t \u0361 s ;
  34 dz → d \u0361 z ;
  35 dź } $voiceless → t \u0361 ɕ ;
  36 dź → d \u0361 ʑ ;
  37 dż } $voiceless → t \u0361 ʂ ;
  38 dż → d \u0361 ʐ ;
  39 kw → kf ;
  40 krw → krf ;
  41 ni } $vowel → ɲ ;
  42 ni → ɲ i ;
  43 $voiceless { rz → ʂ ;
  44 rz } $voiceless → ʂ ;
  45 rz → ʐ ;
  46 sz → ʂ ;
  47 trw → trf ;
  48 tw → tf ;
  49 zi } $vowel → ʑ ;
  50 zi → ʑ i ;
  51 #
  52 #
  53 a → a ;
  54 ą } [bp] → ɔm ;
  55 ą } [kg] → ɔŋ ;
  56 ą } [cdt] → ɔn ;
  57 ą → ɔ\u0303 ;
  58 b } $voiceless → p ;
  59 b → b ;
  60 c → t \u0361 s ;
  61 ć → t \u0361 ɕ ;
  62 d } $voiceless → t ;
  63 d → d ;
  64 e → ɛ ;
  65 ę } [bp] → ɛm ;
  66 ę } [dt] → ɛn ;
  67 ę } [gk] → ɛŋ ;
  68 ę → ɛ\u0303;
  69 f → f ;
  70 g } $voiceless → k ;
  71 g → ɡ ;
  72 h } i → ç ;
  73 h → x ;
  74 i } $vowel → ʲ ;
  75 i → i ;
  76 j → j ;
  77 k } [bdzż] → ɡ ;
  78 k → k ;
  79 l } i → ʎ ;
  80 l → l ;
  81 ł → w ;
  82 m → m ;
  83 n → n ;
  84 ń → ɲ ;
  85 o → ɔ ;
  86 ó → u ;
  87 r → r ;
  88 si } $vowel → ɕ ;
  89 si → ɕi ;
  90 s → s ;
  91 ś → ɕ ;
  92 t → t ;
  93 u → u ;
  94 w } $voiceless → f ;
  95 w → v ;
  96 y → ɨ ;
  97 ź } $voiceless → ɕ ;
  98 ź → ʑ ;
  99 ż } $voiceless → ʂ ;
 100 ż → ʐ ;
 101 #
 102 #
 103 # Second pass: Phoneme-to-phone rules.
 104 :: Null ;
 105 #
 106 #
 107 tʐ → tʂ ;
 108 pʐ → pʂ ;
 109 xʐ → xʂ ;
 110 ʐt\u0361ʂ → ʂt\u0361ʂ ;
 111 #
 112 #
 113 # Final de-voicing.
 114 b } [$] → p ;
 115 d \u0361 z } [$] → t \u0361 s ;  # rydz → rɨt\u0361s
 116 d } [$] → t ;
 117 ɡ } [$] → k ;
 118 v } [$] → f ;
 119 ʐ } [$] → ʂ ;
 120 ʑ } [$] → ɕ ;
 121 z } [$] → s ;
 122