Language and Cognition 2

The Science of Reading

Why is reading unique?

recent cultural invention

not universal

requires explicit teaching

Why is reading important globally?

illiteracy costs >$1 trillion annually (World Literacy Foundation, 2015)

What does recognising a word involve?

identifying letters

correct order

matching to mental lexicon

How fast is visual word recognition?

~500-600ms for:

naming

lexical decision

meaning judgment

What are key facts about eye movements?

fixations: 200-250 ms

saccades: jumps between words

fovea processes: ~4-5 letters

What variables affect reading speed?

frequency

word length

age of acquisition

semantic ambiguity

neighbourhood size

What is the MegaStudy approach?

Balota et al. (2014)

large datasets of words

regression analyses

Purpose: identify factors influencing word recognition

What is Forster’s model?

Forster (1976)

serial search model

words checked one-byone

What is the Logogen Model?

Morton (1969)

parallel processing

word detectors activated by input

What are connectionist models?

neural network inspired

parallel distributed processing

developed in 1980s

What is the Interactive Activation Model?

McClelland and Rumelhart (1981,1982)

structure:

3 levels: features, letters, words

mechanisms:

activation spreads, excitatory + inhibitory connections

What is the Word Superiority Effect?

letters recognised better in real words than non-worda

What are limitations of the IAC model?

poor letter position coding

cannot handle letter swaps well

ignores phonology

Why is sound important in reading?

alphabetic systems map letters - sounds

helps children “crack the code”

What are the two routes in reading?

indirect (phonological) - sounding out

direct (lexical) - whole word recognition

How do we understand ambiguous words?

multiple meanings activated

one selected rapidly

What is the Reordered Access Model?

meaning access influenced by: frequency (dominance), context

What did Duffy et al. (1988) show?

Method:

eye tracking during reading

ambiguous words (e.g. “port”)

Conditions:

early vs. late context

Results:

slower processing depending on: context timing, meaning dominance

Conclusions:

meaning selection is: dynamic, context-sensitive

What did Rodd et al. (2013) show?

Method:

priming - filler - test

Result:

recent exposure affects meaning choice

What did Rodd et al. (2016) show?

long-term experience shapes meaning preferences

example: rowers interpret “catch” differently

What are the core problems in language processing?

variability

timing

ambiguity

What are the key mechanisms?

parallel processing

top-down influence

learning and experience