Reading Flashcards
Model 1 duel route cascaded model
Route 1 grapheme - phoneme conversion
Direct route from orthographic analysis system to grapheme- phoneme rule system
It is used to read non familiar words and non words
Mechanism - converts letters or groups of letters into sounds
Regularise the pronunciation of words
Route 2 lexicon plus semantic system
Words can only be read aloud correctly by first activating the relevant meaning in the semantic system.
Then using the meaning as the basis for selecting the correct pronunciation From the speech output lexicon
Route 3 ( lexicon only - reading without meaning)
Direct route from orthographic input lexicon to phonological output lexicon
Reading of single words in a rapid sequence do not need to involve accessing the meaning
The contribution of this direct route might help between words with similar meaning. Reducing the numbers of semantic reading errors.
Patients can read familiar and non familiar words but are unable to make semantic judgement
Model strengths
It accounts for
Surface dyslexia
Phonological dyslexia
Frequency effect on naming by healthy participants
Lexical decision performance on healthy individuals
Model limitations
Only deals with mono syllabic words (1 word)
not good at naming times (only accounts for 4.5% of the variance in word naming times)
Fails to treat words and non words differently
Semantic system of model isn’t implemented it only applies to alphabetic languages.
Model 2 distributed connectionist approach
2 crucial assumptions
- pronunciation of words and non words is effected by the pronunciation of similar words which is called neighbourhood effect
High frequency words have stronger influence then low frequency words
How the model works
The model learns to read (pronounced words) accurately
This model is based on back -propagation which means that the output of the model is compared to the expected correct response
The model was trained on 2,998 words
Performance of the model at the end of training
- irregular words took longer to name than regular words
- low frequency words took longer to name then high frequency words
The model read over 90% of the words correctly which is similar to proficient adult readers
Plaut et al (1996) made lesions to the network to stimulate surface dyslexia
Plaut et al (1996) model evaluation strengths
Some evidence supports notion that orthographic, semantic and phonological systems are used in parallel
Reading of non words is influenced by knowledge of reading words
Includes and explicit learning mechanism
Plaut et al (1996) model evaluation limitations
Only deals with monosyllabic words same criticism as duel route cascaded model coltheart et al (2001)
The models performance is not as accurate as coltheart et al (2001)
Did not simulate the behaviour of phonological dyslexic patients
