#############################
# metaphonerules.d
#############################
# by Steve Southwell - BravePoint, Inc.
# Copyright 2003 - The FreeFrameWork Project, Inc.
# Use freely according to FFW License: http://www.freeframework.org/license.shtml
#############################
# How to use:
# 1.  Ensure that this file is somewhere in your propath
# 2.  Include metaphone.i in your program
# 3.  Use toMetaphone("yourtexthere").
#############################
# How to Edit
# 1. Lines beginning with # are comments
# 2. Anything after the # on a line is a comment
# 3. White space is fine, but will not be considered part of the substitution text
# 4. Each data line consists of From-text and to-text - delimited by a comma
# 5. The defaults provided should work pretty well for English.  
# 6. If you make improvements to this, please consider passing them along to us! 
#    webmaster@freeframework.org
# 7. The ^ character matches the beginnings of words
# 8. The $ character matches the ends of words
# 9. Substitution is case-insensitive.  All-caps are used for clarity
# 10. Substitutions are cumulative, so make sure you have them in the right order!
# 11. Use the special keyword <doubles> to indicate that doubled letters should be 
#     combined into a single letter.
# 12. To eliminate a certain pattern, just enter it on a line by itself, followed by 
#     a comma
#############################
# End results
# The text will end up consisting only of the following letters:
#   B,E,F,H,J,K,L,M,N,S,T,W,X,Y,Z
# Each of the letters has roughly its normal sound with the following exceptions:
#   X = sh or ch sound
#   Z = th sound
#   E and Y can only be at the beginning of words

# Y almost always sounds like I except at beginning of words:
^Y,%
Y,I
%,^Y

# Oddballs that mess up other rules
LAUGH,LAF   # LAUGH
ASCIS,ASHIS # FASCIST

# Beginning of word replacements:
^GN,^N      # GNOME
^GH,^G      # GHOST
^JUA,^WA    # JUAN
^KN,^N      # KNOW
^PN,^N      # PNEUMONIA
^PS,^S      # PSYCOLOGY
^PF,^F      # PFIZER
^RH,^R      # RHOMBUS
^TS,^S      # TSUNAMI
^WHO,^HO    # WHORE
^WH,^W      # WHALE,WHEN,WHY
^X,^S       # XEROX

# End of word replacements:
AJO$,AHO$   # NAVAJO
CHT$,T$     # YACHT
EJO$,EHO$   # VIEJO
FTH$,TH$    # FIFTH
GHT$,T$     # FLIGHT
GN$,N$      # FOREIGN
GNS$,NS$    # ALIGNS
GNED$,ND$   # ALIGNED
ILLA$,IA$	# TORTILLA, VILLA - Most likely Spanish pronunciation
ILLAS$,IAS$	
MB$,M$      # DUMB
MN$,M$      # CONDEMN
OGH$,O$     # VAN GOGH
OUGH$,OF$   # ENOUGH,COUGH,ROUGH
AUGH$,A$    # LIMBAUGH

# Other anywhere replacements
CCE,KSE     # SUCCESS
CCI,KSI     # SUCCINCT
PH,F        # PHARMACY,STAPH
SCE,SE      # SCENE
SCHO,SKO    # SCHOOL
SCHE,SKE    # SCHEMA,SCHEDULE
SCH,SH      # BUSCH
SCI,SI      # SCIENCE
GG,K        # EGG,BRAGGING


# First set of single letters
Z,S
X,KS
Q,K
V,F


# SH = X except when preceded by S (must be before doubled letter rule)
SS,%
SH,X
%,SS

# Before we check for doubles...
CK,K        # CLOCK,KICKER

# SPECIAL SYNTAX FOR REMOVING DOUBLED LETTERS:
<doubles>

# More combos:
CIA,XA      # OFFICIAL
DGE,JE      # JUDGE
DGI,JI      # ? ORIGINAL METAPHONE RULE ?
GI,JI       # GIN
GE,JE       # GENDER
SIA,XA      # PERSIAN
SIO,XA      # TENSION
TCH,X       # ITCH
TIA,XA      # MARTIAN
TIO,XO      # ACTION
TH,Z        # THIS,THAT,THE OTHER


# Still more combos:
CH,X        # SANDWICH
CI,SA       # CINDER
CE,SA       # CENTRAL
GH,K        # SPAGHETTI,AGHAST


# Second set of single letters:
C,K         
D,T       
G,K 

# All vowels now become A for a moment
E,A
I,A
O,A
U,A

# Now special rules that deal with proximity to vowels in general:

AH,%
%A,AHA
%,A         # H silent if after vowel and no vowel follows
W,%
%A,WA
%,          # W silent if NOT followed by a vowel like WRITE,WRAP,WRENCH

# Preserve beginning of word vowel, but delete the others.
^A,^E
A,

# ALL DONE!