Matlin, Chapter 9

1
Matlin, Chapter 9
2
History

• Phrase structure (as in diagramming sentences)
  constituentsunits of phrase structure Jarvella
  study
• Test sentences
• The confidence of Kofach was not unfounded. To
  stack the meeting for McDonald, the union had
  even brought in outsiders.

3
Jarvella study

• Other version
• Kofach had been persuaded by the international to
  stack the meeting for McDonald. The union had
  even brought in outsiders.
• Note middle part is the same (to stack the
  meeting for McDonald)
• Subjects remembered middle part better for first
  version. Constituent structure affects what is
  in working memory.

4

• Chomskytransformational grammar surface deep
  structure rules for transforming one to other
• Same deep structure can have more than one
  surface structure (i.e., theres more than one
  way to say the same thing)
• Ambiguityone surface structure has more than one
  deep structure

5
For fun ambiguous headlines

• http//rds.yahoo.com/S2766679/Khumorambiguoush
  eadlines/v2/SIDw/TIDQSSH_35/lWS1/R1/H0/IPCu
  s/SHE0/SIG1215cqok0/-http3A/www.monster-island
  .org/tinashumor/humor/headline.html

6
Comprehension

• Negatives are harder to understand If you dont
  think thats unlikely not to be the case, then
  you must not be less confused than you didnt
  think you werent.
• Passive voicerequires less familiar
  transformation
• Nested structurese.g., embedded clauses
  requires further transformation

7
Example of nested structure

• (from text) The plane that I want to take when
  I go to Denver after he returns from Washington
  leaves at 941.
• Punctuation helps indicate how to break up into
  phrases punct-2.htm

8
Neurolinguistics

• Left vs. right hemispheres some variation due to
  handedness gender some language processing
  more in left
• Aphasiaspecifics of disorder depend on location
  of brain damage (e.g., Wernickes for receptive,
  Brocas for expressive)
• Subtractive methods various tasks with
  different complexity levels subtract digital
  record of brain activation across tasks

9
Speech miscomprehension

• kissthisguy.com -- The Archive of Misheard
  Lyrics

10
Reading

• Eye movementssaccades guided by top-down
  processing
• Perceptual span 4 spaces to left of fixation,
  up to 15 to right of fixation
• Regressionsbad habit or driven by comprehension
  processes?
• Top-down processing context levels and
  comprehension process

11
Reading

• Individual differencesgood readers better at
  top-down processing (and may be faster at
  bottom-up as well)
• Working memorybetter for good readers than for
  poor readers

12
Reading theories

• Direct access hypothesisfrom visual stimulus to
  meaning (sight words)
• Indirect access hypothesisfrom visual to
  phonological to meaning phonologically-mediated
• Dual route hypothesiscan do direct or indirect
  direct more likely with high frequency words,
  indirect for low freq. also individual
  differences

13
Understanding discourse

• Forming a representation processing cycles
  (Kintsch)each sentence understood within context
  of previous text, continuously integrating with
  whole representation
• Inferences constructionist viewwe draw
  inferences (about causes, relationships) as we
  read (similar to precise elaboration)

14
Understanding discourse

• Conditions making inferences MORE likely large
  working memory, good metacomprehension skills,
  have good knowledge of topic
• Conditions making inferences LESS likely
  reading scientific texts, being depressed,
  reading series of articles on same topic

15
Artificial intelligence and reading

• FRUMPcan understand script-based info can
  make inferences, summarize, answer questions
• LSA (latent semantic analysis)compares degree of
  semantic similarity between two passages

16
Bilateral cooperative model of reading
speech
naming
understanding
Functional relationships
real world contexts
Phrase syntax
categories associations
Word linkages
associations, simple images
Phonetic recoding
holistic word recognition
feature analysis
feature analysis