Reading models and dyslexia

1
Reading models and dyslexia

• Lorna Bourke

2
Aims

• To gain an awareness of acquired dyslexia in its
  different forms
• To gain understanding and knowledge of acquired
  dyslexia in terms of models of normal reading
• To be introduced to neuropsychological case
  studies that have advanced the study of acquired
  dyslexia
• To develop knowledge of the assessment battery
  typically used by Educational Psychologists to
  diagnose developmental phonological dyslexia

3
Strokes

• Most common type of brain injury
• Disruption of blood supply to the brain
• burst artery (haemorrhage)
• blocked artery (blood clot)
• Paul Broca (1860s)
• right handed people - language problems occur
  after damage to the left hemisphere of the brain
• left hemisphere responsible for language
  abilities (incl. reading and writing)
• Aphasias
• Disorders affecting the comprehension or
  production of language
• Different types depending on which aspects of
  speech processing have been impaired (Ellis
  Young, 1988)
• When reading problems are the predominant
  symptom Acquired Dyslexia

4
Acquired Dyslexia

• Carl Wernicke (late 18th Century)
• Neurologist
• Mid-1970s
• Cognitive Neuropsychology
• What do these patients tell us about
• The way in which we access the meaning of a word
  in reading?
• The way we read a word aloud
• Can we explain the data from these patients in
  terms of models of normal cognitive function?
• Ellis Young (1988) Dual-Route Models
• Should we abandon them?
• Single word reading
• Models of word recognition
• box and arrow flow diagrams
• Connectionist models

5
Shallice Warrington (1980)

• Patients differ in terms of severity of
  impairment but more importantly, the nature of
  the reading disability varies
• Peripheral Central Acquired Dyslexias
• Peripheral Dyslexia
• Damage to the visual analysis system
• Perception of letters in words is impaired
• Central Dyslexias
• Damage to processes beyond the visual analysis
  system
• Comprehension and/or pronunciation of written
  words

6
Peripheral Dyslexias Neglect Dyslexia

• Ellis, Flude Young (1987)
• Patient VB
• Stroke affected right hemisphere
• Speech unimpaired
• Problems reading
• Sometimes (12) the patient omitted the first
  letter
• CAGE gt age LEVER gt ever
• More commonly the first letter was substituted
• ELATE gt plate PEACH gt beach
• LIQUID gt squid a kind of sea creature
• Visual problem
• Identification of words spelled aloud to her
  excellent
• Rotation of passage of text 90 degrees
  improvement
• Errors involved replacements rather than
  deletions
• FABLE as table not able BEAT as heat
  rather than eat

7
Peripheral DyslexiasNeglect Dyslexia

• Riddoch, Humphreys, Cleton Fery (1991)
• Problem arises from failure to attend to left
  side of words
• JB
• GROSS as cross BOUGH as slough
• to left of words performance improved
• Supports attentional theory
• Different types of Neglect Dyslexia
• Reviewed by Ellis, Young Flude (1993)
• Attentional Dyslexia (Shallice Warrington,
  1977)
• Letter-by-Letter Reading (Patterson Kay, 1982)

8
Central DyslexiasNon-Semantic Reading

• WLP (62yr old female progressive dementia incl.
  memory loss)
• Schwartz, Marin Saffran (1979)
• Schwartz, Saffran Martin (1980)
• 7/20 correct matches of animal names to pictures
• 18/20 correct animal names read aloud
• Not reading via meaning
• Could read nonwords irregular words
• Comprehension of written words poor
• Impaired Semantic System

9
Central DyslexiasSurface Dyslexia

• Marshall Newcombe (1973)
• Reliance on sublexical route
• i.e. letter-sound conversion
• visual analysis system to the phoneme level
• Misread irregular words as regular words
• ISLAND as izland SUGAR as sudgar BROAD as
  brode
• Better reading of regular words
• Regularisation errors to irregular words
• Patterson, Marshall Coltheart (1985)
• Dual-route reading model
• Damage to the visual input lexicon
• Damage to the speech output lexicon
• Ellis Young (1988)

10
Central DyslexiasPhonological Dyslexia

• Dont generally have the same speech (aphasic)
  problems as Deep Dyslexics
• Poor reading of nonwords relative to real words
• The problems with nonwords is not usually as
  severe as it is in Deep Dyslexia
• JD (Farah et al, 1996) 25-30 nonwords 75-80
  words
• WT (Coslett, 1991) 25 of nonwords 90 of words
• AM Patterson, 1982) 8 of nonwords 85 or words
• WB (Funnell, 1983) 0 of nonwords 90 of words
• No semantic errors in reading
• Some visual errors in reading (less than in Deep
  Dyslexia)

11
Central DyslexiasPhonological Dyslexia

• Several patients (though not all) have problems
  reading function words
• AM 90 content words 70 function words
• JD 75 content words 90 function words
• WB 90 content words 90 function words
• Several patients (though not all) have problems
  reading low imageable content words
• JD 88 HI image words 55 LO image words
• AM 90 HI image words 90 LO image words
• WB 90 HI image words 90 LO image words
• Comprehension of written words is poor in some
  patients and good in others
• WB poor at picture-word matching and deciding
  whether toast cake or bread
• AM virtually 100 on tasks of this kind
• WT was poor at comprehending abstract words

12
Central DyslexiasPhonological Dyslexia

• Explanations
• Impairment to the non-lexical reading route
• All patients, though variable in terms of
  severity
• Impairment to lexical routes
• Lexical-semantic route? (WB, WT poor
  comprehension of written words but not AM)
• Lexical non-semantic route? (AM function word
  deficit JD low imageable word deficit but not
  in WB or WT)
• At least one of the two lexical routes is
  preserved.
• It is therefore clear that the deficit in reading
  unfamiliar words can persist even when patients
  whose reading comprehension of familiar words
  is remarkably well preserved (AM).
• In surface dyslexia, however, one can observe the
  reverse of this.

13
Deep Dyslexia

• Marshall Newcombe (1966, 1973)
• Patients are unable to read unfamiliar words or
  nonwords correctly e.g. BLASP
• Demonstrate no ability to match words and
  psuedohomophomes (FOX PHOKS)
• Their reading of familiar words is much better.
  However, they are particularly poor at reading
  function words (e.g. because) and words of low
  imageability
• High Imageability Low Imageability
• mouth cause
• hospital purpose
• horse moment

14
Deep Dyslexia

• Visual Paralexias
• Participants make a preponderance of visual
  errors often referred to as visual paralexias
• An error in response to a written word
  paralexia
• An error in response to spoken word
  paraphasia
• When a participant makes a visual error, they
  respond with a word that is visually similar to
  the word they are trying to read
• i) DUGgtbug ii) PAMPERgtpaper iii) WASgtwash
  iv) UNIVERSALgtuniversity
• The error can occur at
• The start of the word (i)
• The middle (ii)
• The end (iii, iv)
• Response is of higher imageability than target
• Semantic Paralexias
• BOAT gt SHIP ILL gt SICK
• Sometimes both visual semantic
• SYMPATHY gt ORCHESTRA

The error can involve Substitution (i,
iv) Omission (ii) Addition (iii)
15
Video

• Reading for Meaning
• A Case Study of Deep Dyslexia
• Funnell, E., Humphreys (Eds)

16
Deep Dyslexia

• Level of impairment in Deep Dyslexia
• Right Hemisphere Reading (e.g. Coltheart, 1980,
  1987 see also Morton Patterson, 1987 Michel et
  al, 1996 Weekes et al, 1997 Price et al, 1998
  Coltheart, 2000)
• Severe Left Hemisphere Damage
• Relying on rudimentary right hemisphere reading
  system (can only deal with high imageability
  words)
• Why should such a system exist?
• Why the partial recovery in Klein et als
  patient?
• Why do some patients make semantic errors in
  picture naming as well?

17
Arguments against the Right-Hemisphere Theory

• Plaut Shallice (1993)
• Seems unlikely that a system would only be
  available when there is brain injury
• Klein et al (1994) spontaneous recovery of
  semantic impairments suggests 2 systems
• Roeltgen (1987)
• Doesnt tell us much about reading processes

18
Theoretical Interpretations of Difficulties
Schematic of the Dual Route Model of Reading
Acquisition  
PRINT
Letter Recognition
Graphemic Parsing
Visual Input Lexicon
Grapheme to Phoneme
Semantic System
Phoneme Level
Speech Output Lexicon
Phonemic Blending
Spoken Response
Route 1 Route
2 Route 3
Source Ellis Young (1995)
19
Deep Dyslexia

• Patients use the normal damaged reading system
• Reading lexicon problem? (visual errors)
• Might the visual errors reflect the loss of
  representations in the visual input lexicon?
• NO
• If so, the visual errors would occur equally for
  low high imageability words (Shallice, 1988).
  They dont!
• If so, there would be a poor performance on
  lexical decision (Patterson, 1978)

20
Seidenberg McClelland (1989)Triangle Model
Context
Meaning
Orthography
Phonology
A general framework for orthographic,
phonological and semantic processing of words.
See also Plaut et al (1996)
21
Seidenberg McClelland (1989)Triangle Model

• There are two routes instead of three
• Both nonwords and words can be read aloud by a
  single route
• Connectionist/neural network model of this route
• An additional route for pronouncing words aloud
  via semantics exists (Plaut Shallice, 1993)
• Appealing model in the way that it encodes
  relationships between phonemes and letters (the
  alphabetic principle) without the need for a
  non-lexical route
• English is quasi-regular (mint, lint, splint.
  pint)
• The model can capture the int gt short i sound
  rule
• But still manage to store the exception pint gt
  long i sound rule
• It is a model that learns and represents the
  statistical relationships between letters and
  their associated phonemes in English words (e.g.
  80 chance of being short i in int

22
Seidenberg McClelland (1989)Triangle Model

• In the model there are
• Input units representing orthographic information
  (400)
• Output units representing phonemic information
  (460)
• Hidden units that are vital for encoding
  exceptions (200)
• Inhibitory (suppresses another unit) excitatory
  (activates another unit) links
• Thresholds and weights (associative strengths)
  (will vary from one unit to another)
• Distributed representations (no lexicon)

23
Seidenberg McClelland (1989)Triangle Model

• The rule governing int could be captured by
  assuming that a hidden unit becomes activated
  when the last 3 letters are int.
• This hidden unit will activate a short i rather
  than a long i in the output units
• However, the rule must be abandoned when a p
  is the first letter.
• There will be a different hidden unit that
  becomes activated when p is the first letter.
  This will inhibit the unit that activates the
  short i in the output units
• The rule will be employed otherwise will be
  employed for nonwords that end in int
• By using a complex set of associations involving
  a large number of hidden units, complex
  relationships between written and spoken English
  can be encoded.
• Within a single system
• Both words and nonwords can be assigned
  pronunciation
• Both regular and irregular words can be read
  successfully

24
Deep Dyslexia

• Problem exists in the SEMANTIC SYSTEM?
• Imageability effect Semantic errors in reading
  and picture naming
• YES
• Plaut Shallice (1993)
• Semantic features have been lost from semantic
  memory. High imageability words contain more
  features and therefore are more resistant to
  feature loss.
• Visual errors occur because the meaning of the
  target word cannot be accessed in the semantic
  system. So a visually similar words meaning is
  selected instead.
• A different view meanings of low imageability
  words are stored in a separate semantic system,
  or a separate part of the semantic system, which
  has been damaged.

25
Deep Dyslexia

• NO
• Marshall Newcombe (1973)
• The semantic errors might arise because normal
  semantic representations are imprecise anyway.
  They are an inevitable consequence of reading via
  semantics. If so, why dont normal participants
  make semantic errors when relying on semantics?
• Impairment of the system that enables us to read
  unfamiliar words
• YES!
• Might this system prevent semantic errors in
  normal subjects?

26
Research in Practice

• http//www.dyslexia-inst.org.uk/assess.htm