# written by Bart Alewijnse
# Vieregge-Cuccuarini modified by W. Heeringa (X-SAMPA)
# configuration file written in Jan 2005
#
# A somewhat modified version of the feature system by Vieregge and 
# Cucchiarini. Most of the modifications (by Wilbert Heeringa) are in
# the addition of a number of features. 
#


#########################################################################
DEFINES
########

#default stuff.
VERSION 2    
TOP 65535
METHOD SUM      
TOKENSTRING RAW 
START 0

# The following is actually only half-arbitrary
# 
# 8 seems to be about the maximum usual difference between tokens
# in practice (Taking the distance to the indels vector into 
#              account; 5 is enough for most others)
# 
# However, because the output costs are clamped when they are
# converted into the features distance table (int range 0 to 65535,
#  which is equivalent to 0..1 for the leven executable) it's 
# possible to distort the costs: eg. when delete-and-insert becomes 
# relatively cheaper because the would-be-much-larger indel was clamped
# while the subst wasn't. 
# Therefore I'm taking a value a little higher, especially since further
# alterations could make the distance go to ten without too much trouble.
#  Note that the leven distance only really has comparison value 
# relative to other values under the same model anyhow.
SUBSTMAX  10.0

#the INDEL value is irrelevant here, as the use of the vector in the 
#INDELS section overrides it


#########################################################################
FEATURES
########

# All the 0's are default distances. 
#
# The point is that the characters that can be both vowel and consonant
# should be comparable to characters that are only vowel or consonant ONLY
# by the features of the type in which they match (eg. vowel features in 
# a vowel-vowel and vowel-both comparison).
#
# By default in VERSION 2, the distance between a feature that is undefined 
# in one token and defined in the other is its default difference times its
# weight. By making the default difference zero, this means if a feature
# is defined in one token only, it does not add anything to the distance.
# 
# This in turn means consonants and vowels can leave only their own
# features defined and 'both' can have both sections defined, and whatever
# isn't defined in one of the tokens in a token-token comparison will
# contribute 0 (0*weight) to the distance.
# 
# It's a bit convoluted, but it seems like the only way to allow the 'both'
# thing.
#

#  Type As in vowel/consonant/both, exists purely to disallow a vowel-consonant
#  subst by making it extremely costly, forcing leven to fall back on using an
#  insert and delete instead.
B 1.0 40.0 type

# While used as bitmaps, some features can take fractional values.
# It's possible to solve that with weights, but rather dirty.
N 0 1.0 adv1
N 0 1.0 adv2
N 0 1.0 adv3
N 0 1.0 adv4
N 0 1.0 high1
N 0 1.0 high2
N 0 1.0 high3
N 0 1.0 long1
N 0 1.0 long2
N 0 1.0 long3
N 0 1.0 round
N 0 1.0 nasal
N 0 1.0 diph

N 0 1.0 breath
N 0 1.0 creak
N 0 1.0 tone1
N 0 1.0 tome2
N 0 1.0 circ

N 0 1.0 place1
N 0 1.0 place2
N 0 1.0 place3
N 0 1.0 place4
N 0 1.0 voice
N 0 1.0 nas
N 0 1.0 stop
N 0 1.0 glide
N 0 1.0 lat
N 0 1.0 fric
N 0 1.0 trill
N 0 1.0 high
N 0 1.0 distr
N 0 1.0 syll

N 0 1.0 apic




#These two are purely for the indels. They don't have any other effect,
# because because they're either equal or swamped in value by 'type'
# because I'm lazy, these do work based on the nondefined-defined default
# distance, so that I only have to set them in the INDELS vector.
D 1 1 vInDel
D 1 1 cInDel
# In fact, their naming is almost redundant; the only detail is that it's
# in the indel vector, nowhere else, and has a default distance of 1.

#########################################################################
TEMPLATES
########
#
# vzero and czero set the respective section to non-undefined when it is clear
# that a token is a vowel (and/)or a consonant.
# (This avoids the default distance from being applicable for the same 
#  section - i.e. makes sure the distance comes purely from defined value
#  distances [note that in the TOKENS section templates are only called to
#  set to non-zero] )
#
T vzero
F adv1 = 0
F adv2 = 0
F adv3 = 0
F adv4 = 0
F high1 = 0
F high2 = 0
F high3 = 0
F long1 = 0
F long2 = 0
F long3 = 0
F round = 0
F nasal = 0
F diph = 0
F breath = 0
F creak = 0
F tone1 = 0
F tome2 = 0
F circ = 0

T czero
F place1 = 0
F place2 = 0
F place3 = 0
F place4 = 0
F voice = 0
F nas = 0
F stop = 0
F glide = 0
F lat = 0
F fric = 0
F trill = 0
F high = 0
F distr = 0
F syll = 0
F apic = 0

#note that type is a bitmap. Matching of 3 with 2, and 3 with 1 is inherent.
T vowel
F type = 1

T consonant
F type = 2

#This is from back when a character (i, j, u and w) could be seen as both a
# vowel and consonant, which had a problem. 
# It is still necessary for the comparison to the null vector in INDELS.
T both
F type = 3


T adv1
F adv1 = 1 

T adv2
F adv2 = 1

T adv3
F adv3 = 1

T adv4
F adv4 = 1

T high1
F high1 = 1
T high1half
F high1 = 0.5

T high2
F high2 = 1
T high2half
F high2 = 0.5

T high3
F high3 = 1
T high3half
F high3 = 0.5

T long1
F long1 = 1

T long2
F long2 = 1

T round
F round = 1
T roundhalf
F round = 0.5

T nasal
F nasal = 1

T diph 
F diph = 1

T breath
F breath = 1

T creak
F creak = 1

T tone1
F tone1 = 1

T tome2
F tome2 = 1

T circ 
F circ = 1


T place1
F place1 = 1
T place1half
F place1 = 0.5

T place2
F place2 = 1
T place2half
F place2 = 0.5

T place3
F place3 = 1

T place4
F place4 = 1
T place4half
F place4 = 0.5

T voice
F voice = 1

T nas 
F nas = 1

T stop
F stop = 1

T glide
F glide = 1

T lat 
F lat = 1

T fric 
F fric = 1

T trill
F trill = 1

T high 
F high = 1

T distr
F distr = 1

T syll
F syll = 1

T apic
F apic = 1


#########################################################################
INDELS
########
#
# Makes indels comparable with a neutral feature-vector that more or less
# represents silent and neutral.
# Basically works quite like a 'both' token in how comparison is done.
#
T both czero vzero adv1 adv2 high1 high2half roundhalf place1 place2 place3 place4
F vInDel = 1
F cInDel = 1
# These last two are here to give an inherent bast cost for being a vowel 
# or a consonant


#########################################################################
TOKENS
########

H i
T  vowel vzero high1 high2 high3
#under both also had: czero place1 place2 voice glide high

H y
T  vowel vzero high1 high2 high3 round

H 1
T  vowel vzero adv1 adv2 adv3 high2 high3 long1

H }
T  vowel vzero adv1 adv2 high1 high2 high3 round

H M
T  vowel vzero adv1 adv2 adv3 adv4 high1 high2 high3

H u
T  vowel vzero adv1 adv2 adv3 adv4 high1 high2 high3 round voice glide distr

H I
T  vowel vzero high1 high2 high3half

H Y
T  vowel vzero high1 high2 high3half round

H U
T  vowel vzero adv1 adv2 adv3 adv4 high1 high2 high3half round

H e
T  vowel vzero high1 high2

H 2
T  vowel vzero high1 high2 round

H @\
T  vowel vzero adv1 adv2 high1 high2

H 8
T  vowel vzero adv1 adv2 high1 high2 round

H 7
T  vowel vzero adv1 adv2 adv3 adv4 high1 high2

H o
T  vowel vzero adv1 adv2 adv3 adv4 high1 high2 round

H @
T  vowel vzero adv1 adv2 high1 high2half roundhalf

H E
T  vowel vzero high1

H 9
T  vowel vzero high1 round

H 3
T  vowel vzero adv1 adv2 high1

H 3\
T  vowel vzero adv1 adv2 high1 round

H V
T  vowel vzero adv1 adv2 adv3 adv4 high1

H O
T  vowel vzero adv1 adv2 adv3 adv4 high1 round

H {
T  vowel vzero high1half

H 6
T  vowel vzero adv1 adv2 high1half

H a
T  vowel vzero

H &
T  vowel vzero round

H A
T  vowel vzero adv1 adv2 adv3 adv4

H Q
T  vowel vzero adv1 adv2 adv3 adv4 round

H p
T  consonant czero stop distr

H b
T  consonant czero voice stop distr

H t
T  consonant czero place1 stop

H d
T  consonant czero place1 voice stop

H t`
T  consonant czero place1 place2half stop

H d`
T  consonant czero place1 place2half voice stop

H c
T  consonant czero place1 place2 stop high distr

H J\
T  consonant czero place1 place2 voice stop high distr

H k
T  consonant czero place1 place2 place3 stop high

H g
T  consonant czero place1 place2 place3 voice stop high

H q
T  consonant czero place1 place2 place3 stop

H G\
T  consonant czero place1 place2 place3 voice stop

H ?
T  consonant czero place1 place2 place3 place4 stop

H m
T  consonant czero voice nas distr

H F
T  consonant czero voice nas

H n
T  consonant czero place1 voice nas

H n`
T  consonant czero place1 place2half voice nas

H J
T  consonant czero place1 place2 voice nas high distr

H N
T  consonant czero place1 place2 place3 voice nas high

H N\
T  consonant czero place1 place2 place3 voice nas

H B\
T  consonant czero voice trill distr

H r
T  consonant czero place1 voice trill

H R\
T  consonant czero place1 place2 place3 voice trill

H 4
T  consonant czero place1 voice

H r`
T  consonant czero place1 place2half voice

H p\
T  consonant czero fric distr

H B
T  consonant czero voice fric distr

H f
T  consonant czero fric

H v
T  consonant czero voice fric

H T
T  consonant czero place1half fric

H D
T  consonant czero place1half voice fric

H s
T  consonant czero place1 fric

H z
T  consonant czero place1 voice fric

H S
T  consonant czero place1 fric high distr

H Z
T  consonant czero place1 voice fric high distr

H s`
T  consonant czero place1 place2half fric

H z`
T  consonant czero place1 place2half voice fric

H C
T  consonant czero place1 place2 fric high distr

H j\
T  consonant czero place1 place2 voice fric high distr

H x
T  consonant czero place1 place2 place3 fric high

H G
T  consonant czero place1 place2 place3 voice fric high

H X
T  consonant czero place1 place2 place3 fric

H R
T  consonant czero place1 place2 place3 voice fric

H X\
T  consonant czero place1 place2 place3 place4half fric

H ?\
T  consonant czero place1 place2 place3 place4half voice fric

H h
T  consonant czero place1 place2 place3 place4 fric

H h\
T  consonant czero place1 place2 place3 place4 voice fric

H K
T  consonant czero place1 lat fric

H K\
T  consonant czero place1 voice lat fric

H w
T  consonant czero round voice glide distr
# under both alzo vzero adv1 adv2 adv3 adv4 high1 high2 high3

H P
T  consonant czero voice glide

H v\
T  consonant czero voice glide

H r\
T  consonant czero place1 voice glide

H r\`
T  consonant czero place1 place2half voice glide

H j
T  consonant czero place1 place2 voice glide high
#under both, also had czero high1 high2 high3 place1 place2 voice glide high

H M\
T  consonant czero place1 place2 place3 voice glide high

H l
T  consonant czero place1 voice lat

H l`
T  consonant czero place1 place2half voice lat

H L
T  consonant czero place1 place2 voice lat high distr

H L\
T  consonant czero place1 place2 place3 voice lat high



#This is a hack to introduce a certain 65535 indel, so that
# neither features or leven will rescale. This is simply
# a string that will never exist as real input and has a
# large distance with respect to everything.

H #**
F type = 4




# Length modifiers. This is a slightly odd distance system,
# summarized in this distance matrix: 
#           es   s  hl
#       s    1    
#       hl   1   1    
#       l    3   2   1

#extra short
M _X
F long1 = 1

#normal length, aka 'short': stays zero

#half-long
M :\
F long2 = 1

#long
M :
F long2 = 1
F long3 = 1


M ~
F nasal = 1





MI _L
MI _H
MI _=
MI _~
MI _t
MI _q
HI H