Psy1302 Psychology of Language

1
Psy1302 Psychology of Language

• Lecture 10
• Ambiguity Resolution
• Sentence Processing I

2
agenda

• Connecting word recognition with sentence
  processing via ambiguity resolution.
• Lexical Ambiguity
• Syntactic Ambiguity
• MORE MODELS!!!
• Garden-Path Model
• Constraint-Satisfaction Model
• CLEVER but difficult to explain experiments!
• (so ask questions if you are lost!!!)

3
Ambiguity

• Time flies like an arrow
• Time proceeds as quickly as an arrow proceeds.
• Measure the speed of flies in the same way that
  you measure the speed of an arrow.
• Measure the speed of flies in the same way that
  an arrow measures the speed of flies.
• Measure the speed of flies that resemble an
  arrow.
• Flies of a particular kind, time flies, are fond
  of an arrow.

4
Qs about Online Ambiguity Resolution

• What alternatives are available at different time
  points?
• What degree of commitment is made to one or more
  alternatives?
• What information is used to guide these
  commitments?

5
Lexical Ambiguity(semantic, lexical)
6
Cross-Modal Priming
7
Cross-Modal Priming Exp. 1(Swinney et al. 1978
Onifer Swinney, 1981)
Rumour had it that for many years, the government
building had been plagued with problems. The man
was not surprised when he found several (spiders,
roaches, and other) bugs in the corner of his
room.

ANT
ANT
SPY
SPY
SEW
SEW
8
Cross-Modal Priming Exp. 1(Swinney et al. 1978
Onifer Swinney, 1981)
80
70
ant
60
spy
50
Amount of Priming (unrelated word RT minus
related word RT)
40
30
20
10
0
immediate
3 syll
delay
9
Cross-Modal Priming Exp. 1(Swinney et al. 1978
Onifer Swinney, 1981)
Rumour had it that for many years, the government
building had been plagued with problems. The man
was not surprised when he found several (spiders,
roaches, and other) bugs (insects) in the corner
of his room.

ANT
ANT
SPY
SPY
SEW
SEW
10
Riddle

• What has wheels and flies, but is not an
  airplane?
• What has wheels and flies, but is not an
  airplane?
• What has wheels and flies, but is not an
  airplane?

V
N
11
Cross-Modal Priming Exp. 2(Tanenhaus, Leiman,
Seidenberg, 1979 Seidenberg, Tanenhaus, Leiman,
Bienkowski, 1982)

• Noun reading I bought a watch.
• Verb reading I will watch.

0
600
200
CLOCK
0
600
200
CLOCK
12
Cross-Modal Priming Exp. 2(Tanenhaus, Leiman,
Seidenberg, 1979 Seidenberg, Tanenhaus, Leiman,
Bienkowski, 1982)

• Noun reading I bought a watch.
• Verb reading I will watch.

0
600
200
clock
clock
clock
0
600
200
clock
clock
clock
13
Seidenberg, Tanenhaus, Leiman, Bienkowski, 1982
14
Effects of Frequency in Ambiguity Resolutions
Equibias Ambiguous Word
Non-Equibias Ambiguous Word
port
pitcher
15
Duffy, Morris, Rayner (1988)

• Varied frequency of homonyms
• Varied whether supportive context came before
  word or after word.

16
Older Eye-tracker

• low-level infrared light ? eye
• reflections from cornea and lens indicate
  position of eye fixation.

Head movements messes up calibration ? Bite bar
or head rest is needed
17
Duffy, Morris, Rayner (1988)
Supportive Context
No Supportive Context
Supportive Context
No Supportive Context
Control words in Parentheses
For Non-Equibiased, Context supports
non-dominant reading.
18
No Supportive Context
Non-Ambiguous Control
Equibias Ambiguous Word
pitcher
whiskey
Non-Ambiguous Control
Non-Equibias Ambiguous Word
soup
port
-- Thickness of the line indicates amount of
activation.
19
Adding Supportive Context
Non-Ambiguous Control
Equibias Ambiguous Word
pitcher
whiskey
Non-Ambiguous Control
Non-Equibias Ambiguous Word
soup
port
-- Thickness of line indicates amount of
activation.
20
Supportive Context
No Supportive Context
Equibias
supportive context
pitcher
Non-Equibias
port
supportive context
21
High reaction time
Supportive Context
No Supportive Context

• Equibiased
• Processing time lower when provided with prior
  disambiguating contextual support. (Reason
  because accessing both meanings)
• Non-equibiased
• Processing time high when provided with prior
  disambiguating contextual support supporting the
  less frequent meaning. (Reason made the less
  frequent more equal to the other meaning)
• Processing time low when not provided
  disambiguating contextual support for the less
  frequent meaning. (Reason not considering the
  less frequent meaning. In fact, time spent later
  in disambiguating region is higher due to a need
  to reanalyze).

22
Lexical Ambiguity

• Current conclusions
• Parallel, rather than serial activation
• Relative strength of activation depends on
• Degree of contextual constraint available
• Frequency of use of each meaning

23
Syntax

• Another level up!
• Parsing figuring how the words in a phrase or
  sentence combine, using the rules in a grammar.
• Parser

24
Syntactic Ambiguity
25
Is the woman insured?

• Woman drives off with what she thought was her
  dates car (but wasnt) and then totaled it. Can
  she get money from her insurance company
• Contract says
• Such insurance as is provided by this policy
  applies to the use of a non-owned vehicle by the
  named insured and any person responsible for use
  by the named insured provided such use is with
  the permission of the owner.

26
Does he deserve jail time?

• Drug dealer tried to swindle an (unbeknownst to
  him) undercover cop by selling a bag of powder
  that had only a minuscule trace of
  methamhetamine. The quantity was not harmful.
• Law says
• Every person who sells any controlled substance
  which is specified in subdivision (d) shall be
  punished.
• (d) Any material, compound, mixture, or
  preparation which contains any quantity of the
  following substance having a potential abuse
  associated with a stimulant effect on the central
  nervous system Amphetamine, Methamphetamine

27
Local Ambiguities (Being led down the
garden-path)

• The bully hit the girl with the...
• ...stick.
• ...wart. (garden-pathed)
• The woman felt the fur...
• ...and then left.
• ...was very expensive.
  (garden-pathed)

28
Local Ambiguities

• The bully hit the girl with the wart and then
• The bully hit the girl with the stick and then

29
Ambiguous Sentences
Homework sentence
Last night, the car crashed.
The car crashed.
yesterday
today
yesterday
today
time
time
30
Ambiguous Sentences
S
The reporter
31
Ambiguous Sentences
Homework sentence
32
Ambiguous Sentences
33
Traditional Views(contrasting lexical and
syntactic ambiguity)
Table from MacDonald, Pearlmutter, Seidenberg
Paper
34
Garden-Path Model(Frazier Fodor, 1978)

• Serial the processor initially identifies only
  one analysis
• selected based on structural simplicity
• Modular Initial structure built on the basis of
  syntactic category labels.
• revision process incorporates other information.

35
Garden Path ModelSelecting the initial analysis

• When word is identified, its syntactic category
  is retrieved
• Parser identifies which rules of the grammar
  contain that category
• Analysis selected on the basis of structural
  simplicity
• Late Closure
• Minimal Attachment

36
Garden Path ModelHeuristics for Simplicity

• Late Closure
• When possible, attach incoming lexical items into
  the clause or phrase currently being processed
  (i.e., the lowest possible nonterminal node
  dominating the last item analyzed).
• Minimal Attachment
• Attach incoming lexical items into the
  phrase-marker being constructed with the fewest
  nodes consistent with well-formedness rules of
  language.

37
Late Closure
38
Late Closure
crashed
The reporter
said
the car
last night
1 or 2?
1
2
S
39
Late Closure

• Choice 1

• Choice 2

S
The reporter
car...
40
Minimal Attachment
41
Minimal Attachment
Jamie
saw
man
with
the
1 or 2?
1
2
42
Minimal Attachment

• Choice 1

• Choice 2

NP ? Det N NP ? NP PP
NP
1 extra node
NP
Det
the
PP
P
with
43
Garden Path ModelHeuristics for Simplicity

• Late Closure
• When possible, attach incoming lexical items into
  the clause or phrase currently being processed
  (i.e., the lowest possible nonterminal node
  dominating the last item analyzed).
• Minimal Attachment
• Attach incoming lexical items into the
  phrase-marker being constructed with the fewest
  nodes consistent with well-formedness rules of
  language.

44
Ambiguities Late Closure and Minimal Attachment
In-Class Exercise (see also homework)

• NP/VP Attachment Ambiguity
• The cop saw the burglar with the binoculars
• The cop saw the burglar with the gun

45
Ambiguities Late Closure and Minimal Attachment
In-Class Exercise (see also homework)

• NP/S Complement Attachment Ambiguity
• The athlete realized his goal last week
• The athlete realized his shoes were across the
  room

46
Ambiguities Late Closure and Minimal Attachment
In-Class Exercise (see also homework)

• Clause-boundary Ambiguity
• Since Jay always jogs a mile the race doesnt
  seem very long
• Since Jay always jogs a mile doesnt seem very
  long

47
Ambiguities Late Closure and Minimal Attachment
In-Class Exercise (see also homework)

• Reduced Relative-Main Clause Ambiguity
• The woman delivered the junkmail on Thursdays
• The woman delivered the junkmail threw it
  away

48
Ambiguities Late Closure and Minimal Attachment
In-Class Exercise (see also homework)

• Relative/Complement Clause Ambiguity
• The doctor told the woman that he was in love
  with to leave
• The doctor told the woman that he was in love
  with her

49
Garden-Path Model

• Strengths
• Considers our working memory capacity
• Speed achieved by considering one interpretation
• Explains broad range of phenomena

50
Models of Sentence Processing

• Garden-Path Model
• Autonomous
• Late closure
• Minimal attachment
• Constraint-Based Model
• Interactive
• Lexical Biases
• Referential Contexts
• Structural Biases

Cues from multiple sources constrain
interpretation
51
Traditional Views(contrasting lexical and
syntactic ambiguity)
Constraint-Satisfaction Model SAYS its not the
right characterization!
Table from MacDonald, Pearlmutter, Seidenberg
Paper