##########################################################
#
#  This file contains rules for phonetic encoding of
#  Spanish words in an adapted SAMPA standard, to
#  meet the needs of the corrector.
#
##########################################################

# Rules change from a sound (one or more letters) or a category
# to another sound or category, in a certain context.
#
# Category names can have only one letter are usually uppercase.
#
# Contexts must contain a "_" symbol indicating where the replacement
# takes place; they may also contain letters, categories, and the special
# symbols ( ) (to enclose optional parts), ^ (word beginning), or $ (word end)
#
# Rules can only change letters/sounds to sounds, and categories to categories.
# If a category is to be changed to another category, they should contain the 
# same number of characteers. Otherwise the second category will have its last
# letter repeated until it has the same length as the first (if it is
# shorter), or characters in the second category that don't match
# characters in the first will be ignored.
#
# Don't use regex metacharacters (except for the parentheses, ^, and $) in
# context specifications, category names, or sound codes.

<Rules>

############
# Categories
############

# vowels, long, short
U=áéíóú
V=aeiou
F=aou
W=ei

##########
## Rules
##########

# remove accents
U/V/_

## remove duplicated letters
aa/a/_
ee/e/_
ii/i/_
oo/o/_
uu/u/_
dd/d/_
ff/f/_
gg/g/_
jj/x/_
mm/m/_
ññ/ñ/_
ss/s/_
yy/j/_
zz/z/_

## numbers soud as their names
1/uno/_
2/dos/_
3/tres/_
4/cuatro/_
5/cinco/_
6/seis/_
7/siete/_
8/ocho/_
9/nueve/_
0/cero/_

# strong and soft "r" in Spanish.
# NOTE: "4" is not SAMPA standard (should be "rr")
# This is adapted to get an encoding of one character
# per sound, which is convenient for the corrector.
rr/4/_
r/4/^_
r/4/_$

# make these first to avoid interactions with 
# rules handling "c" or producing "x".
# NOTE: "X" is not SAMPA standard (should be "tS").
# This is adapted to get an encoding of one character
# per sound, which is convenient for the corrector.
ch/X/_
x/X/_

# occlusive k sound for strong (F) and weak (W) vowels
# and /T/ sound of "c" with weak vowels (cero,cielo)
qu/k/_W
qu/ku/_F
qü/ku/_
q/k/_
cc/kT/_W
c/T/_W
c/k/_F
c/k/_

# gutural fricative (girafa,jarra) vs plosive (gato,guerra) g/j sound in Spanish
j/x/_
g/x/_W
gu/g/_W
gu/gu/_
gü/gu/_

# other sounds
ph/f/_
ll/L/_

z/T/_
h//_
v/b/_
i/j/_V
y/j/_
ñ/J/_

# rules for special characters
\-//_
\_//_
\&//_
\*//_
\+//_
\.//_
\$//_
\-//_

</Rules>


<Exceptions>
software sofwar
hardware xardwar
whisky wiski
shakespeare Xespir
</Exceptions>