X-Git-Url: https://git.saurik.com/apple/icu.git/blobdiff_plain/4388f060552cc537e71e957d32f35e9d75a61233..0f5d89e82340278ed3d7d50029f37cab2c41a57e:/icuSources/data/translit/es_es_FONIPA.txt

diff --git a/icuSources/data/translit/es_es_FONIPA.txt b/icuSources/data/translit/es_es_FONIPA.txt
index b464b6fa..89196cf3 100644
--- a/icuSources/data/translit/es_es_FONIPA.txt
+++ b/icuSources/data/translit/es_es_FONIPA.txt
@@ -1,42 +1,75 @@
-# ***************************************************************************
-# *
-# *  Copyright (C) 2004-2012, International Business Machines
-# *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
-# *
-# ***************************************************************************
+# © 2016 and later: Unicode, Inc. and others.
+# License & terms of use: http://www.unicode.org/copyright.html#License
+#
 # File: es_es_FONIPA.txt
-# Generated from CLDR 
+# Generated from CLDR
+#
+
+# Tranformation from Spanish to Spanish in IPA transcription (es_FONIPA).
+# Not fully phonemic, since we mark up allophonic variants of voiced stops,
+# e.g. we break down /b/ into [b] and [β].
+#
+# See e.g. "Ortografía de la lengua española" by Real Academia Española,
+# available in PDF format on the web (non-stable URL).
 #
-$bow = [-\ $] ; # Beginning of word.
+# Definitions.
+$bow = [-\ $] ;  # Beginning of word.
 $consonant = [bβdðfgɣʝklʎmnŋɲθprɾstʧx] ;
-$syll = '.' ; # Syllable boundary.
+$syll = '.' ;    # Syllable boundary.
+#
+#
+#
+#
 ::NFC;
 ::Lower;
-$bow { ct → | t ; # ctónico (?)
-$bow { cz → | z ; # czar
-$bow { gn → | n ; # gnomo
-$bow { mn → | n ; # mnemotécnico
-$bow { ps → | s ; # psicología
-$bow { pt → | t ; # pterodáctilo
-$bow { x → | s ; # xilófono
-$bow { i → i ;
-$consonant { i } [ aáeé oóuú] → j ;
-[aeo] { i } [^aáeé oóuú] → i\u032F ;
-i } [ aáeé oóuú] → ʝ ;
-i → i ;
-[aeo] { y } [^aáeéiíoóuú] → i\u032F ;
+#
+#
+#
+#
+#
+# Word-initial cluster simplification.
+#
+$bow { ct → | t ;  # ctónico (?)
+$bow { cz → | z ;  # czar
+$bow { gn → | n ;  # gnomo
+$bow { mn → | n ;  # mnemotécnico
+$bow { ps → | s ;  # psicología
+$bow { pt → | t ;  # pterodáctilo
+$bow { x  → | s ;  # xilófono
+#
+#
+#
+# Vowels and glides.
+#
+$bow       { i                 → i ;
+$consonant { i } [ aáeé  oóuú] → j ;
+[aeo]      { i } [^aáeé  oóuú] → i\u032F ;
+i } [ aáeé  oóuú] → ʝ ;
+i                 → i ;
+#
+#
+[aeo]      { y } [^aáeéiíoóuú] → i\u032F ;
 y } [ aáeéiíoóuú] → ʝ ;
-y → i ;
-[aeo] { u } [^aáeéiíoó ] → u\u032F ;
-u } [ aáeéiíoó ] → w ;
-ü } [ eéií ] → w ;
-u → u ;
-ü → u ; # Should not be needed, but just in case.
+y                 → i ;
+#
+#
+[aeo]      { u } [^aáeéiíoó  ] → u\u032F ;
+u } [ aáeéiíoó  ] → w ;
+ü } [   eéií    ] → w ;
+u                 → u ;
+ü                 → u ;  # Should not be needed, but just in case.
+#
+#
 [aá] → a ;
 [eé] → e ;
-í → i ;
+í    → i ;
 [oó] → o ;
-ú → u ;
+ú    → u ;
+#
+#
+#
+# Consonants.
+#
 b → β ;
 cch → ʧ ;
 ch → ʧ ;
@@ -65,7 +98,7 @@ rr → r ;
 r → ɾ ;
 ss → s ;
 s → s ;
-tx → ʧ ; # for loanwords from Basque, Catalan
+tx → ʧ ;  # for loanwords from Basque, Catalan
 t → t ;
 v → β ;
 w → $syll w ;
@@ -73,10 +106,40 @@ x } h?[aáeéiíoóuú$] → ks ;
 x } [^aáeéiíoóuú$] → s ;
 x → ks ;
 z → θ ;
+#
+#
+#
+# Second pass: phoneme-to-phone rules. Differentiation of /β/ into [b] and [β],
+# place assimilation of [n], etc.
+#
 ::Null;
+#
+#
 [-\ ] → ;
-[mnɲŋ $] { β → b ;
+#
+#
+[mnɲŋ  $] { β → b ;
 [mnɲŋlʎ$] { ð → d ;
-[mnɲŋ $] { ɣ → g ;
+[mnɲŋ  $] { ɣ → g ;
+#
+#
 n } [gɣk] → ŋ ;
+#
+#
+# Optional: Place assimilation of n before labial consonants.
+#
+# n } [bβpfm] → m ;
+#
+# Optional: Voicing of [s].
+#
+# s } [bβdð] → z ;
+# s } [gɣ][^ei] → z ;
+# s } [mnɲŋlʎrɾ] → z ;
+#
+# Optional: Lenition of [k] before [θ].
+#
+# k } θ → ɣ ;
+#
+#
 ::NFC;
+