L9 - Skilled reading Flashcards
What are the 3 stages of reading development (Frith, 1986; Ehri, 2005)?
- Logographic:
- Words recognised by idiosyncratic global features. - Alphabetic:
- Letter-sound relationships learned; new words can be sounded out. - Orthographic:
- Words recognised automatically from orthographic features.
Why is automatic word identification important for reading comprehension?
i) Reading requires many simultaneous processes:
- access meaning of words.
- maintain meanings in memory.
- integrate meanings together.
- relate meanings to existing knowledge.
ii) Attentional resources are limited:
- attention must be divided between simultaneous tasks.
- if attention is overloaded, performance is poor.
- to limit attentional requirements we automate ‘low level tasks’.
iii) Automating a task requires:
- that the task is ‘invariant’.
- lots of practice.
Does context change lexical access?
- Yes, words are identified more easily in context.
- Prior context facilitates identification of brief stimuli.
- Lexical decision faster to ‘nurse’ than ‘butter’ when preceded by ‘doctor’.
What are the two ways which context could facilitate lexical access?
- Automatic spreading semantic activation in lexical/semantic memory.
- Attentional strategies, e.g. guessing, prediction.
What is the two process model of Semantic Priming (Posner & Snyder, 1975)?
i) Automatic process:
- fast, obligatory, parallel.
- pre-activates related items with no cost to unrelated.
- > facilitation without interference.
ii) Attentional process:
- slow, attentional mediation, serial.
- > increased benefit for related items, as well as cost to unrelated.
What did Stanovich (1986) demonstrate towards the relationship between contextual predictions and reading skill?
- Participants asked to read final word of a sentence aloud.
Predicable: “The banker locked the safe.”
Neutral: “They said it was in the hidden safe.”
Incongruent: “The train pulled into the hidden safe.”
- 4th & 6th graders show facilitation (predictable) as well as inhibition (incongruent).
- Adults show only small facilitation.
-> Reliance on contextual prediction reduces with reading skill.
What did Stanovich and West (1983) demonstrate in skilled readers between degraded and normal text?
Facilitation for predictable completions; no inhibition for incongruous.
-> Priming effects reflect automatic processes.
Inhibition for incongruent completions only when stimuli are degraded.
-> Use active prediction when identification is difficult.
Are both meanings of ambiguous words accessed even when context favours one?
Cross-modal priming (Swinney, 1979):
- Auditory: “… the man was not surprised when he saw several spiders, roaches and other bugs.”
- Visual Lexical decision: between ant, spy, sew.
Respond to both ‘ant’ and ‘spy’ -> context irrelevant meaning initially activated, then suppressed.
What are some factors which the degree of activation of context-irrelevant meanings depend on?
- Homograph dominance (same spelling, different meaning).
- Degree of context bias.
- Reading skill.
What is the difference between skill levels of comprehension with lexical ambiguities?
- Better comprehenders are quicker to use sentence context to suppress irrelevant meanings.
- Word identification is automatic and the effects of context come in later.
What is the Dual Route Model of word recognition?
- Basic premise: need two routes to explain 1) exception words (e.g. colonel, pint); 2) nonwords (e.g. slint).
- Grapheme/phoneme correspondence (GPC) ‘rules’ predict regular pronunciation of non-words.
- Systems operate in parallel to determine response.
What is some evidence towards the Dual Route Model?
Regularity x frequency interaction:
- Irregular, low frequency words take longer.
- > because grapheme-phoneme system conflict.
Acquired dyslexics:
i) Phonological dyslexics:
- words > non-words.
- lexical route intact, damaged GPC.
- can’t pronoun non-words, can pronounce irregular words.
ii) Surface dyslexics:
- non-words > words.
- GPC route intact, lexical route damaged.
- errors on irregular words.
- > Double dissociation.
- > Independent systems.
Outline the Interactive Activation Model:
- Connectionist model with ‘symbolic’ nodes for letters, words.
- Hierarchical layer arrangement of nodes.
- Parallel interactive activation and inhibition.
- Identification occurs when threshold is reached in a node.
- Lateral inhibition within levels (competition for best matching word).
What are the 3 ‘Benchmark Phenomena’ in IA Model?
1) Word frequency effects:
- identification threshold lower for common words.
2) Word superiority effects:
- letters in words receive top-down support from word nodes.
3) Semantic priming:
- active word nodes activate their semantic features at concept levels.
- top-down effects from the concept level -> activates likely upcoming words which pre-activates the representations at the word level.
Outline the Parallel Distributed Processing (PDP) Model aka Triangle Model?
- Connectionist model.
- Set of interconnected processing nodes.
- Learning algorithms (e.g. delta rule, back propagation).
- Multi-level architecture (“hidden units”).
- Extract statistical regularities between orthography, phonology, semantics.