PSY246 Lecture Notes - Lecture 9: Phoneme, Low Frequency, Semantic Dementia
2 Jun 2018
School
Department
Course
Professor
Word recognition and reading – Week 9 Lecture PSY246:
Lecture outline –
• Some facts about reading – why study reading
• Research methods used in reading research
o Lexical decision, naming, semantic categorization
o Stroop task, masked priming task
• Word recognition
o Automatic processing
▪ Retrieval of semantics (semantic stroop effect)
▪ Retrieval of phonology (homophone interference effect,
masked phonological priming effect)
o Interactive activation model
o Problem with the IA model: slot-coding
▪ Cambridge email and Transposed-letter (TL) priming effect
• Reading aloud
o Varieties of acquired dyslexias
▪ Surface dyslexia, phonological dyslexia, deep dyslexia
o Dual-route cascaded model
o Connectionist triangle model
• Eye movements in reading
o Saccades and fixations
o Application – does Spritz work?
Facts about reading:
• Reading is a culturally acquired skill
o Writing was invented only about 5400 yrs ago
o Unlike speech, children do not acquire reading without instruction
o About 10% of adults in Aust, the US and UK read below grade 4 level
and are defined functionally illiterate
BUT –
• Reading is largely automatic
o ‘reading is a form of telepathy…’ (McEwan, Atonement)
• Average adult readers
o Read (silently) at about 250 words per minute
o Can identify words in less than 100 ms
o Read aloud words in 500-600 ms
o Make judgements about word meaning in 600-900 ms
• Reading is an important skill
o Adults without effective reading skills are at a great disadvantage
• Reading is a highly skilled behaviour – largely automatic
• Reading is a complex skill
o Involves several perceptual, cognitive and linguistic processes
Reading involves several kinds of linguistic processing –
• Orthography: the spelling of words
• Phonology: the sound of words
find more resources at oneclass.com
find more resources at oneclass.com
• Semantics: the meaning of words
• Syntax: rules for combining words e.g. Jamie hit David (who did the hitting?)
• Discourse processing: making inferences
• Various tasks differ in the involvement of these kinds of processing
o Lexical decision task is largely based on orthography
o Semantic categorization requires semantics
o Reading aloud requires phonology
Research methods –
• RT (reaction time) tasks – speeded responses
o Lexical decision task – ‘is TRUCK a word? Is SLINT a word?’
o Semantic categorization task – is hawk an animal?
o Naming (pronunciation, read-aloud) task – saying a printed word out
loud as quickly as possible
• Automaticity
o Interference tasks e.g. Stroop colour naming
o Priming e.g. semantic priming, masked priming
• Record eye movements during reading
• Neuropsychology – study of patients
• Neuroimaging (ERP, EEG, MEG fMRI)
Automatic processing –
• In skilled readers, word identification is automatic
o Automatic processes are fast – use of speeded response tasks (RT as
DV) e.g. lexical decision task, naming task
o Automatic processes are obligatory/unavoidable
• Stroop colour naming task
o Name the font colour
o Stroop interference effect indicates word identification is automatic in
the sense that it is obligatory
Is semantic retrieval automatic?
• Semantic Stroop effect
o Greater interference with (incongruent) colour-associated word
distractor than colour-unrelated words e.g. lemon (green) > model
(green)
• Semantic retrieval is automatic
o Semantic Stroop effect is not reduced/eliminated by:
▪ Presenting the colour with a single letter
▪ Hypnotic word blindness suggestion (‘the characters you see
are gibberish’)
• Others argue otherwise
Unconscious word identification –
• Automatic processes are unconscious
o Masked priming → a prime is presented briefly and masked so that
participants cannot identify it
▪ Each trial, there is a forward mask (stream of hashes #)
find more resources at oneclass.com
find more resources at oneclass.com
▪ Compare effects of related prime vs. unrelated (control) prime
e.g. pudge-JUDGE vs. slint-JUDGE
▪ Relationship can be orthographic, phonological or semantic e.g.
hawk-EAGLE vs. sofa-EAGLE
Automatic phonological activation –
• Is phonology retrieved automatically in silent reading?
• Frost (1998): strong phonology hypothesis
o Phonological representation is a necessary product of processing
printed words
o Phonological processing is mandatory (obligatory) in reading
Phonological processes in reading (Van Orden 1987) –
• Homophones
o Words having the same pronunciation but spelt differently e.g.
ROWS/ROSE, THERE/THEIR
• Homophone interference effect:
o Participants make more errors when to homophones of category
member when asked to make semantic categorization:
▪ Is it a flower?; ROWS > ROBS
▪ Participants engaged in phonological processing, mistaking
‘ROWS’ for ‘ROSE’
Masked Phonological Priming –
• Rastle and Brysbaert (2006): meta-analysis Kinoshita & Norris 2012
o Word processing faster when preceded by phonologically identical
nonword primes than by primes similar in orthography but not
phonology
▪ Even when primes are masked and brief, so as not to be
consciously perceived
o Suggests phonological processing occurs rapidly and automatically
McClelland and Rumelhart’s (1981) Interactive Activation Model:
• Is the basis of almost all models of reading e.g. DRC (dual route cascaded
model)
• 3 levels of detectors = feature, letter, word
• Bottom-up excitation/inhibition and top-down excitation/inhibition
• Within-level inhibition = different features/letters/words are mutually
inhibitory
Word superiority effect –
• In a perceptual identification task, identifying a letter is easier if it occurs in a
word context (compared to a single letter context or a nonsense word)
• The IA model explains the word superiority effect in terms of the top-down
feedback from the word level to the letter level and also inhibition within the
word level
Slot coding –
• Used in the interactive activation model
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Word recognition and reading week 9 lecture psy246: In skilled readers, word identification is automatic: automatic processes are fast use of speeded response tasks (rt as. Dv) e. g. lexical decision task, naming task: automatic processes are obligatory/unavoidable, stroop colour naming task, name the font colour, stroop interference effect indicates word identification is automatic in the sense that it is obligatory. Is phonology retrieved automatically in silent reading: frost (1998): strong phonology hypothesis, phonological representation is a necessary product of processing printed words, phonological processing is mandatory (obligatory) in reading. ; rows > robs: participants engaged in phonological processing, mistaking. Is the basis of almost all models of reading e. g. drc (dual route cascaded model: 3 levels of detectors = feature, letter, word, bottom-up excitation/inhibition and top-down excitation/inhibition, within-level inhibition = different features/letters/words are mutually inhibitory. Slot coding : used in the interactive activation model. It cannot capture the similarity between words with letters in the wrong order.
