fs = code.split(".")[1][4]
pc = "3"
pcode = phila_system(i, phones, trans, fm, fp, fv, ps, fs, pc, phoneset)
if pcode == "33":
return True
elif pcode == "3":
return False
if __name__ == "__main__":
pron = {}
for line in open("dict", "r"):
if line.startswith(';'):
continue
(word, pron_string) = line.rstrip().split(' ', 1)
pron[word] = pron_string.split()
words = ['MADDER', 'BADNESS', 'BADDEST', 'RAN', 'SWAM', 'MATH', 'SAD',
'HAND', 'HAM', 'HALF', 'PATH', 'PASS', 'CASH', 'BANG', 'BAT',
'PAL', 'BAG', 'CAB', 'HAMMER', 'MANAGE', 'PLANET', 'PLANETS',
'MANNING', 'CLASSES', 'ASKING', 'CAN', 'BEGAN', 'ANNE',
'ASPECT', 'ASPIRIN', 'CASKET', 'ASKED', 'BASKETBALL', 'BANDSTAND',
'BACKSTROKE', 'TRANSFER', "GLADDEN", "TRANSMISSION",
'SANTA', 'GRANDMOTHER', 'RASCAL', "ASKIN'", "PASSIN'", 'PASSING',
'TRAFFIC', 'CATHOLIC', 'MARRY', 'SPANISH']
for word in words:
word_pron = pron[word]
old_coding = is_tense_old(word, word_pron)
new_coding = is_tense(word, word_pron)
#if old_coding != new_coding:
print word, old_coding, new_coding
def phila_system(i, phones, trans, fm, fp, fv, ps, fs, pc, phoneset):
"""redefines vowel classes for Philadelphia"""
orig_pc = pc
# Plotnik code returned by arpabet2plotnik
phones = split_stress_digit(phones)
# separate Arpabet coding from stress digit for vowels
# 1. New short-a coding using Kyle Gorman's syllabification &
# classification functions above
if pc == '3' and phones[i].label in {'AE0','AE1','AE2','AE'} and fm != '0':
tenseness = is_tense(trans, [ph.label for ph in phones[i:]])
# the second argument is a list of ARPABET phones for the word starting
# with the target vowel; this will not affect correctness of
# syllabification. is_tense returns True if tense, False if lax, and None
# if variable/"unclassifiable"
if tenseness == True:
pc = '33'
elif tenseness == False:
pc = '3'
else:
pc = '39'
# convert dictionary entries to LAX short-a for "-arry" words
if pc == '2' and 'ARRY' in trans.upper():
if len(phones) > i + 2 and phones[i + 1].arpa == 'R' and phoneset[phones[i + 2].arpa].cvox == '0':
pc = '3'
# random dictionary inaccuracies
if pc == '5' and trans.upper() == 'MARIO':
pc = '3'
# 2. /e/
if trans.upper() in ["CATCH", "KEPT"]:
pc = '2'
# 3. /oh/
if phones[
i].arpa == 'AA' and trans.upper() in ['LAW', 'LAWS', "LAW'S", 'LAWFUL',
'UNLAWFUL', 'DOG', 'DOGS', "DOG'S",
'DOGGED', 'ALL', "ALL'S", 'CALL',
'CALLS', "CALL'S", 'CALLING',
'CALLED', 'FALL', 'FALLS', "FALL'S",
'FALLING', 'AUDIENCE', 'AUDIENCES',
"AUDIENCE'S", 'ON', 'ONTO', 'GONNA',
'GONE', 'BOSTON', "BOSTON'S", 'AWFUL',
'AWFULLY', 'AWFULNESS', 'AWKWARD',
'AWKWARDLY', 'AWKWARDNESS', 'AWESOME', 'AUGUST',
'COUGH', 'COUGHS', 'COUGHED', 'COUGHING']:
pc = '53'
# 4. /o/
if phones[
i].arpa == 'AO' and trans.upper() in ['CHOCOLATE', 'CHOCOLATES', "CHOCOLATE'S",
'WALLET', 'WALLETS', 'WARRANT', 'WARRANTS',
'WATCH', 'WATCHES', 'WATCHED', 'WATCHING',
'WANDER', 'WANDERS', 'WANDERED', 'WANDERING',
'CONNIE', 'CATHOLICISM', 'WANT', 'WANTED',
'PONG', 'GONG', 'KONG', 'FLORIDA', 'ORANGE',
'HORRIBLE', 'MAJORITY']:
pc = '5'
if phones[i].arpa == 'AE' and trans.upper() in ['LANZA', "LANZA'S"]:
pc = '5'
# 4. /ah/
# if pc == '5' and phones[i].arpa == 'AA' and (i > 0 and phones[i-1].arpa != 'W') and (len(phones) != i+1 and phones[i+1].arpa != 'R'):
## x1 = max(0, i - 3)
## x2 = min(len(trans), i + 3)
# print "Checking a spelling in chunk %s of word %s (%s)." % (trans[x1:x2].upper(), trans, ' '.join([p.label for p in phones]))
# for t in ['AU', 'AW', 'AL']:
# if 'A' in trans[x1:x2].upper() and t not in trans[x1:x2].upper():
## pc = '43'
# 5. /iw/
if phones[i].label == "UW1":
# UW1 preceded by /y/
if i > 0 and phones[i - 1].arpa == 'Y':
pc = '82'
# words spelled with "-ew", e.g. "threw", "new", "brew"
if 'EW' in trans.upper():
pc = '82'
# words spelled with "-u" after /t/, /d/, /n/, /l/, /s/, e.g.
# "Tuesday", "nude", "duty", "new"
if i > 0 and phones[i - 1].arpa in ['T', 'D', 'N', 'L', 'S']:
for t in ['TU', 'DU', 'NU', 'LU', 'SU']:
# make sure -u spelling is adjacent to consonant in orthography
if t in trans.upper():
pc = '82'
# 6. /Tuw/
if phones[i].label == "UW1" and trans.upper in ['THROUGH']:
pc = '73'
# if pc != orig_pc:
# print "\tPhila system reassignment: Changed class of vowel %s from %2s
# to %2s in word %s (%s)." % (phones[i].label, orig_pc, pc, trans, '
# '.join([p.label for p in phones]))
# 7. front vowels before r
if len(phones) > i + 1 and phones[i].arpa in ['EH', 'AE'] and phones[i + 1].arpa == 'R':
if len(phones) == i + 2: # word-final /r/
pc = '24'
if len(phones) > i + 2 and phoneset[phones[i + 2].arpa].cvox != '0':
# not word-final but also NOT intervocalic r
pc = '24'
return pc
pcode = phila_system(i, phones, trans, fm, fp, fv, ps, fs, pc, phoneset)
if pcode == "33":
return True
elif pcode == "3":
return False
if __name__ == "__main__":
pron = {}
for line in open("dict", "r"):
if line.startswith(';'):
continue
(word, pron_string) = line.rstrip().split(' ', 1)
pron[word] = pron_string.split()
words = [
'MADDER', 'BADNESS', 'BADDEST', 'RAN', 'SWAM', 'MATH', 'SAD', 'HAND',
'HAM', 'HALF', 'PATH', 'PASS', 'CASH', 'BANG', 'BAT', 'PAL', 'BAG',
'CAB', 'HAMMER', 'MANAGE', 'PLANET', 'PLANETS', 'MANNING', 'CLASSES',
'ASKING', 'CAN', 'BEGAN', 'ANNE', 'ASPECT', 'ASPIRIN', 'CASKET',
'ASKED', 'BASKETBALL', 'BANDSTAND', 'BACKSTROKE', 'TRANSFER',
"GLADDEN", "TRANSMISSION", 'SANTA', 'GRANDMOTHER', 'RASCAL', "ASKIN'",
"PASSIN'", 'PASSING', 'TRAFFIC', 'CATHOLIC', 'MARRY', 'SPANISH'
]
for word in words:
word_pron = pron[word]
old_coding = is_tense_old(word, word_pron)
new_coding = is_tense(word, word_pron)
#if old_coding != new_coding:
print word, old_coding, new_coding
def phila_system(i, phones, trans, fm, fp, fv, ps, fs, pc, phoneset):
"""redefines vowel classes for Philadelphia"""
orig_pc = pc
# Plotnik code returned by arpabet2plotnik
phones = split_stress_digit(phones)
# separate Arpabet coding from stress digit for vowels
# 1. New short-a coding using Kyle Gorman's syllabification &
# classification functions above
if pc == '3' and phones[i].label in {'AE0', 'AE1', 'AE2', 'AE'
} and fm != '0':
tenseness = is_tense(trans, [ph.label for ph in phones[i:]])
# the second argument is a list of ARPABET phones for the word starting
# with the target vowel; this will not affect correctness of
# syllabification. is_tense returns True if tense, False if lax, and None
# if variable/"unclassifiable"
if tenseness == True:
pc = '33'
elif tenseness == False:
pc = '3'
else:
pc = '39'
# convert dictionary entries to LAX short-a for "-arry" words
if pc == '2' and 'ARRY' in trans.upper():
if len(phones) > i + 2 and phones[i + 1].arpa == 'R' and phoneset[
phones[i + 2].arpa].cvox == '0':
pc = '3'
# random dictionary inaccuracies
if pc == '5' and trans.upper() == 'MARIO':
pc = '3'
# 2. /e/
if trans.upper() in ["CATCH", "KEPT"]:
pc = '2'
# 3. /oh/
if phones[i].arpa == 'AA' and trans.upper() in [
'LAW', 'LAWS', "LAW'S", 'LAWFUL', 'UNLAWFUL', 'DOG', 'DOGS',
"DOG'S", 'DOGGED', 'ALL', "ALL'S", 'CALL', 'CALLS', "CALL'S",
'CALLING', 'CALLED', 'FALL', 'FALLS', "FALL'S", 'FALLING',
'AUDIENCE', 'AUDIENCES', "AUDIENCE'S", 'ON', 'ONTO', 'GONNA',
'GONE', 'BOSTON', "BOSTON'S", 'AWFUL', 'AWFULLY', 'AWFULNESS',
'AWKWARD', 'AWKWARDLY', 'AWKWARDNESS', 'AWESOME', 'AUGUST',
'COUGH', 'COUGHS', 'COUGHED', 'COUGHING'
]:
pc = '53'
# 4. /o/
if phones[i].arpa == 'AO' and trans.upper() in [
'CHOCOLATE', 'CHOCOLATES', "CHOCOLATE'S", 'WALLET', 'WALLETS',
'WARRANT', 'WARRANTS', 'WATCH', 'WATCHES', 'WATCHED', 'WATCHING',
'WANDER', 'WANDERS', 'WANDERED', 'WANDERING', 'CONNIE',
'CATHOLICISM', 'WANT', 'WANTED', 'PONG', 'GONG', 'KONG', 'FLORIDA',
'ORANGE', 'HORRIBLE', 'MAJORITY'
]:
pc = '5'
if phones[i].arpa == 'AE' and trans.upper() in ['LANZA', "LANZA'S"]:
pc = '5'
# 4. /ah/
# if pc == '5' and phones[i].arpa == 'AA' and (i > 0 and phones[i-1].arpa != 'W') and (len(phones) != i+1 and phones[i+1].arpa != 'R'):
## x1 = max(0, i - 3)
## x2 = min(len(trans), i + 3)
# print "Checking a spelling in chunk %s of word %s (%s)." % (trans[x1:x2].upper(), trans, ' '.join([p.label for p in phones]))
# for t in ['AU', 'AW', 'AL']:
# if 'A' in trans[x1:x2].upper() and t not in trans[x1:x2].upper():
## pc = '43'
# 5. /iw/
if phones[i].label == "UW1":
# UW1 preceded by /y/
if i > 0 and phones[i - 1].arpa == 'Y':
pc = '82'
# words spelled with "-ew", e.g. "threw", "new", "brew"
if 'EW' in trans.upper():
pc = '82'
# words spelled with "-u" after /t/, /d/, /n/, /l/, /s/, e.g.
# "Tuesday", "nude", "duty", "new"
if i > 0 and phones[i - 1].arpa in ['T', 'D', 'N', 'L', 'S']:
for t in ['TU', 'DU', 'NU', 'LU', 'SU']:
# make sure -u spelling is adjacent to consonant in orthography
if t in trans.upper():
pc = '82'
# 6. /Tuw/
if phones[i].label == "UW1" and trans.upper in ['THROUGH']:
pc = '73'
# if pc != orig_pc:
# print "\tPhila system reassignment: Changed class of vowel %s from %2s
# to %2s in word %s (%s)." % (phones[i].label, orig_pc, pc, trans, '
# '.join([p.label for p in phones]))
# 7. front vowels before r
if len(phones) > i + 1 and phones[i].arpa in [
'EH', 'AE'
] and phones[i + 1].arpa == 'R':
if len(phones) == i + 2: # word-final /r/
pc = '24'
if len(phones) > i + 2 and phoneset[phones[i + 2].arpa].cvox != '0':
# not word-final but also NOT intervocalic r
pc = '24'
return pc
