Psychology 2134A/B Lecture Notes - Eye Tracking, Context-Dependent Memory
This preview shows half of the first page. to view the full 2 pages of the document.
LECTURE 2.6 – Language Processing Continued
Autonomous Theory predicts that top down effects will occur later. Contextual effects occur
post processing after our initial parse.
Connectionists propose that multiple syntactic structures are activated t once. The one with the
strongest activation wins out as our response. Constraint on activation is caused by sense biases
and contextual information.
Eye tracking is a more effective procedure than self paced reading. Self paced reading is very
slow and unnatural; it does not effectively examine people’s ability to parse a sentence.
If we recall from last lecture, minimal attachment is a form of online processing. As such,
simplicity is a key feature of minimal attachment and it prefers direct objects to sentences
For example “Mary suspected the man...”
o DO = “...from Calgary.”
o SO=”...was from Calgary.”
Frazier & Rainer showed these same sentences to people and tracked their eye movements.
They discovered that people looked longer at the “was” in the second sentence.
They interpreted this finding as the processing system taking longer because it expects minimal
attachment which prefers direct objects over sentence compliments.
Garnsey further studied the relationship between direct objects and sentence compliments. He
found that some verbs have a sense bias towards direct objects and others towards sentence
o DO VERB = “Realize”
o SC VERB = “Find”
With this knowledge in hand researchers presented people with ambiguous sentences. When
people reached the disambiguating part of the sentence it took them longer to process it if the
verb conflicted with its sense bias.
These results show that sense bias influences processing. This fining of top-down processing is
inconsistent with the Autonomous Theory but it is consistent with the Connectionist perspective
of multiple activations.
Minimal Attachment begins to fail however in the Salt Shaker Test. I cannot predict people’s eye
Instead we find that people are combing semantic and syntactic information very quickly. This
Salt Shaker Test demonstrated that we are using multiple parses at once and ruling out
erroneous ones with contextual information; not constructing little trees in our heads.
Children start to produce syntax at around 12 months. Infants tend to learn many more nouns
than verbs and use holophrases.
Holophrases are “one word” sentences that infants use to convey meaning. For example....
o Desire “Want!” “Gimme!”
o Emotion “Bad!” “Happy!”
o Naming “Kitty!” “Doggie!”
You're Reading a Preview
Unlock to view full version