Language modelling word FST

1
Language modelling (word FST)

• Operational model for categorizing
  mispronunciations

prompted image circus
step 1 decode visual image
(circus)
(cursus)
(circus)
step 2 convert graphemes to phonemes
(/k y r s y s/)
(/s i r k y s/)
(k i r k y s)
step 3 articulate phonemes
(/s i r k y s/)
(/k y - k y r s y s /) (/ka - i - Er - k y s/)
spoken utterance correct, miscue (step3)
or error (steps 1, 2)
2
Language modelling (word FST)

• Prevalence of errors of different types (Chorec
  data)

Children with RD tend to guess more often
Important to model steps 1 and 3 step 2 not so
important
3
Creation of word FST model step 1

• correct pronunciation
• predictable errors
• (prediction model needed)

4
Creation of word FST model step 3

• Per branch in previous FST
• Correctly articulated
• Restarts (fixed probabilities for now)
• Spelling (phonemic) (fixed probabilities for now)

5
Modelling image decoding errors

• Model 1 memory model
• adopted in listen project
• per target word
• create list of errors found in database
• keep those with P(list entry error TW) gt TH
• advantages
• very simple strategy
• can model real words non-real-word errors
• disadvantages
• cannot model unseen errors
• probably low precision

6
Modelling image decoding errors

• Model 2 extrapolation model (idea from ..)
• look for existing words that
• expected to belong to vocabulary of child (
  mental lexicon)
• bare good resemblance with target word
• select lexicon entries from that vocabulary
• feature based expose (dis)similarities with TW
• features length differences, alignment
  agreement, word categories, graphemes in common,
  
• decision tree ? P(entry decoding error
  features)
• keep those with P gt TH
• advantage can model not previously seen errors
• disadvantage can only model real word errors

7
Modelling image decoding errors

• Model 3 rule based model (under dev.)
• look for frequently observed transformations at
  subword level
• grapheme deletions, insertions, substitutions
  (e.g. d ? b)
• grapheme inversions (e.g. leed ? deel)
• combinations
• learn decision tree per transformation
• advantages
• more generic ? better recall/precision compromise
  
• can model real word non-real word errors
• disadvantage
• more complex time consuming to train

8
Modelling results so far

• Measures (over target words with error)
• recall nr of predicted errors / total nr of
  errors
• precision nr of predicted errors / nr of
  predictions
• F-rate 2.R.P/(RP)
• branch average nr of predictions per word
• Data test set from Chorec database