first word 0.0032741949999999997
current word 0.003265588
last word 0.003232208
pos model 0.003185353
word perplexity 0.003166182
next word 0.003142314
word recognition 0.0031359499999999998
word cor 0.0031202129999999997
strong word 0.003108423
language model 0.0030948549999999997
word correspondence 0.0030906659999999997
word correspondences 0.003067729
word classification 0.003064827
perfect word 0.003056451
word replacement 0.0030517799999999996
atis word 0.0030462339999999997
word matching 0.003046118
word aligner 0.00304611
word transcriptions 0.003045872
word identities 0.003045146
word fragment 0.003045146
word classifica 0.003043741
word matches 0.003043741
word replacements 0.003043741
word match 0.003043741
model probability 0.002937757
acoustic model 0.002531192
base model 0.002523012
guage model 0.0025120439999999997
model tones 0.0025077759999999998
anguage model 0.002496226
backoff model 0.0024810609999999997
other words 0.002255002
model 0.0022264
pos tag 0.0021079230000000003
previous words 0.001885307
repair tag 0.001837633
intended words 0.0017827449999999999
actual words 0.001760525
speech repair 0.0017601029999999998
consecutive words 0.001745275
clue words 0.001745275
pos tags 0.001614724
term tag 0.001587978
other features 0.00157554
language models 0.001571416
same pos 0.00156126
pos tagging 0.001549926
tag probabilities 0.001536216
ing speech 0.0015268319999999999
new speech 0.0015267899999999999
current speech 0.001522688
second tag 0.001517603
words 0.00151506
tone tag 0.001489608
speech repairs 0.001472899
speech processing 0.0014464599999999999
tag values 0.001420547
training data 0.001409747
modeling speech 0.0013939509999999998
speech recognition 0.0013930499999999998
wrong tag 0.001378373
correct pos 0.001370085
repair tags 0.0013444339999999998
speech rec 0.0013212529999999999
speech recognizer 0.0013093999999999998
fluent speech 0.001305044
spontaneous speech 0.0013047509999999998
tect speech 0.00130431
neous speech 0.001301189
speech recognizers 0.0013011499999999998
corporate speech 0.001300592
pos tagset 0.00128913
test set 0.001279463
statistical language 0.001264359
overall pos 0.001251898
tagging performance 0.001248125
ing data 0.0012352370000000001
trains corpus 0.001219129
unique pos 0.00121392
pos classification 0.00120944
silences pos 0.001205824
language modeling 0.001190966
recognition language 0.001190065
intonational features 0.0011824119999999999
basic language 0.001180109
tree algorithm 0.001151967
first repair 0.0011485179999999998
other repairs 0.001141401
language mode 0.001126724
other work 0.0011224350000000002
atis corpus 0.001119254
taains corpus 0.001117849
tree boundary 0.0011044140000000002
tistical language 0.001097804
tree approach 0.00109641
term tags 0.001094779
tagging error 0.001085018
speech 0.00107144
textual features 0.001066192
