Word Recognition

Question 1

Moon et al (2013)

Answer 1

• 30 hour old babies exposed to words unique to their mothers language and words unique to another.
• Babies sucked dummies more when listening to foreign language
• Sound travels to womb

Question 2

Segmentation Problem

Answer 2

Word constraint Norris et al (1997) hard to segment word where it isn’t two separate words on their own.

Stress Culter and Butterfield (1992)
First syllable usually stressed

Question 3

Invariance Problem

Answer 3

• Speech is shaped by other factors such as accent or speed
• Co-articulation, different sound of the same letter depending what comes after it

Question 4

Context effects

Answer 4

Word more recognisable when context involved
Warren & Warren 1970
When missing phoneme replaced by silence it is easier to tell what the phoneme is rather than when replaced by cough as speech recognition system fills in gaps when silent

Warren & Warren 1970
Phoneme restoration, use context of sentence to figure out which letter is masked by noise

Question 5

TRACE theory (McClelland & Elman, 1986)

Answer 5

• Information flows bottom up and top down across three levels
• 3 levels: word units, phoneme units and feature units
• Explains context effects as higher levels of info affects lower levels of info (top down), allows missing phoneme to be activated
• Frauenfelder (1990) criticised trace for over predicting effects

Question 6

Cohort model (Marslen - Wilson, 1984)

Answer 6

Selection stage is influenced by:
- the auditory presentation of the word
- the semantic or syntactic context
- Can recognise word before uniqueness point because of the context of sentence
Two problems
- first phoneme is mispronounced word shouldn’t be recognised
- Semantic context doesn’t influence which words get to stay or leave cohort.

Question 7

Revised Cohort Model (Marslen - Wilson, 1990)

Answer 7

Solved the two problems
- context no longer influence early stages i.e. revised model is less interactive and more bottom up
- words not totally eliminated, activation level decreases instead which deals with mispronunciations

Question 8

Reading and Phonology

Answer 8

• reading is not an innate skill
• learn to produce speech before we learn to read

Question 9

Class example
Reading and Phonology

Answer 9

• Phonology made responding to stimuli difficult
  Although can sometimes make reading easier:
• Phonological neighbourhood, differ in one phoneme
• Phonological priming Klip primes clip

Question 10

Interactive Activation Model
(McClelland & Rumelhart, 1981)

Answer 10

• Visual version of trace model
• 3 levels; feature units, letter units, word units
• Recognise letters better when they are in words rather and isolated

Question 11

Dual-route cascaded model
(Coltheart et al, 2001)

Answer 11

Route 1:
Grapheme Phoneme conversation
Rules determined by the most common grapheme phoneme association of language

Route 2:
Lexicon and semantics

Route 3:
Lexicon only

Problems with it…
doesn’t explain why semantics are important for reading
doesn’t explain how reading is learnt

Question 12

Connectionist triangle model
(Harm & Seidenberg, 2004)

Answer 12

Two routes which take you from spelling to sound

1) direct pathway from orthography to phonology
2) Indirect pathway from orthography to phonology via semantics - support from semantics depends on task being completed

Semantics are important for this model less important for others
McKay et al (2008), words which had meaning read faster but only if consistent