Language Comprehension And Production Flashcards
What is the difference between spoken vs written word recognition?
Spoken: 1st mode of communication - developmentally and historically; involves the analysis of a speech signal that is only available for a short amount of time
Written: literacy relatively recent development - not all those who know a language are literate; written words are available for as long as the reader need them
How do we recognise words?
Lack of invariance - co-articulation of phonemes can make it hard to separate them
Speaker variability - no two speakers sound the same
Segmentation - speakers do not always leave clear breaks between the words
Why is the study of word recognition important?
Teaching children to read
Remediation of illiteracy
Rehabilitation (e.g. Brain injury)
What is a common task for studying word recognition?
The lexical decision task
What is the lexical decision task?
Participant has to press a key stating yes or no if a word is real or not
Response time and accuracy are recorded
What does the lexical decision task tell us about word recognition?
The frequency effect: quicker response time for more frequently occurring words
The lexical status effect: a non-word takes longer to respond to than a word
Non-word legality effect: quicker response time with non-plausible letter strings
What are additional phenomena that can occur in the lexical decision task?
Word similarity effect: slower and more inaccurate rejection of non words that are similar to words
Repetition priming effect: quicker response if the same word is presented twice
Semantic priming effect: quicker response if the word is preceded by a semantically related word
What are the two models of lexical processing?
Serial access models
Direct access models
What is the search model?
There are access files divided into bins
Approximately according to a first letter of word
Words within bins are ordered according to frequency
How would you use serial processing to recognise a word in the search model?
Select the correct bin
Check each entry until a match is found
Retrieve corresponding semantic (meaning) information from the master file
What does this model explain?
Some experimental data- frequent words found quicker as they are at the top of the bin (faster RT)
Non-words are rejected when all the words in a bin have been checked (slower RT)
What is the word superiority effect?
Letters are easier to identify in a real written word
Bias to identify partially obscured letters so that the letter string forms a word
The serial search model doesn’t account for this
What is the interactive activation model?
One of the first computational models
Model combines bottom up and top down information (interactive)
Readers/hearers can use their (top down) knowledge of words to help identify letter sequences from (bottom up) visual/auditory
What are Artificial Neural Networks?
Built by grouping together individual units (neurons) to form layers
neutrons are units and connections are weights
Knowledge is stored in weights and is acquired through learning
Knowledge is also distributed (held within multiple connections)
What is the visual activation model?
One process;
The visual feature layer (input) - the letter level - the word level (output)
Neural activity flows in both directions
The model shows that there is a perceptual aspect to visual word recognition
