Language Comprehension Flashcards
Homophone
Words with same pronunciation but different meanings
Homograph
Words with same spelling but different meanings and pronunciations
Homonyms
Words with same spelling and pronunciation, but different meanings
Co-articulation
One phoneme is influenced by another when put together
Ambiguous speech streams
Being unable to tell when one word finishes and another one starts (I scream or ice cream)
Mental representations
A hypothetical cognitive signal that represents external reality
Kuhl & Miller (1975)
Chinchillas can discriminate between /t/ and /d/ sound
Trained to move to one side of cage for /t/ and the other for /d/
Point at which animals perceive the change of sound was the same as humans
Janet Werker
6-8 month old children can discriminate between /ba/ and /da/
They can also discriminate between Hindi sounds which adults cannot
This ability is lost at around 10 months (Werker & Tees, 1984)
How does frequency affect word recognition?
Short and frequent words are accessed faster than long, infrequent ones
How does neighbourhood density affect word recognition?
Lexical access is slower when words have many neighbours in the mental lexicon (words which differ by one phoneme)
Luce & Pistoni (1998)
Response to words like yacht is faster than to words like peach
This is because yacht has a low neighbourhood density (is a hermit word) and peach has a high neighbourhood density
How do we comprehend sentences?
Activation of lexical items - selection
Selection of lexical item via matching input to mental representation
Integration of lexical items into context
Cohort Model
Marslen-Wilson & Welsh (1979)
Access - lexical retrieval begins with first one or two speech sounds
Selection - words activated but not intended are kicked out until only one remains
Integration - semantic and syntactic properties of activated words are incorporated into high-level utterance representation
Revised Cohort Model
Marslen-Wilson & Welsh (1994)
Context influences integration of words into sentences, but not lexical selection
Initial activation is bottom up
Competitor effects leave one word more active than others (Hebbian)
Context can increase activation of cohort members
Activation of candidates who don’t fit context is not eliminated
Cohort members who fit bottom up input are activated regardless of context
Marslen-Wilson and Taylor (1980) - word monitoring
Participants asked to monitor speech for ‘motorway’
When the sentence made sense, participants could say the word 200ms after the start
Suggests that they recognise the word after a few sounds
When sentence doesn’t make sense, they are much slower
Context gives an advantage