Psy1302 Psychology of Language

Psy1302 Psy1302 Psychology of LanguagePsychology of Language

Lecture 10Lecture 10

Ambiguity ResolutionAmbiguity Resolution

Sentence Processing ISentence Processing I

agendaagenda

Connecting word recognition with Connecting word recognition with sentence processing via ambiguity sentence processing via ambiguity resolution.resolution.

Lexical AmbiguityLexical Ambiguity Syntactic AmbiguitySyntactic Ambiguity

– & MORE MODELS!!!& MORE MODELS!!! Garden-Path ModelGarden-Path Model Constraint-Satisfaction ModelConstraint-Satisfaction Model

– & CLEVER but difficult to explain & CLEVER but difficult to explain experiments!experiments!

(so ask questions if you are lost!!!)(so ask questions if you are lost!!!)

AmbiguityAmbiguityTime flies like an arrowTime flies like an arrow

– Time proceeds as quickly as an arrow Time proceeds as quickly as an arrow proceeds.proceeds.

– Measure the speed of flies in the same way Measure the speed of flies in the same way that you measure the speed of an arrow.that you measure the speed of an arrow.

– Measure the speed of flies in the same way Measure the speed of flies in the same way that an arrow measures the speed of flies.that an arrow measures the speed of flies.

– Measure the speed of flies that resemble Measure the speed of flies that resemble an arrow.an arrow.

– Flies of a particular kind, time flies, are Flies of a particular kind, time flies, are fond of an arrow.fond of an arrow.

Qs about Online Ambiguity Qs about Online Ambiguity ResolutionResolution What alternatives are available at What alternatives are available at

different time points?different time points? What degree of commitment is What degree of commitment is

made to one or more made to one or more alternatives?alternatives?

What information is used to guide What information is used to guide these commitments?these commitments?

Lexical AmbiguityLexical Ambiguity(semantic, lexical)(semantic, lexical)

Cross-Modal PrimingCross-Modal Priming

Cross-Modal Priming Exp. Cross-Modal Priming Exp. 11(Swinney et al. 1978; Onifer & Swinney, (Swinney et al. 1978; Onifer & Swinney, 1981)1981)Rumour had it that for many years, the government Rumour had it that for many years, the government building had been plagued with problems. The man building had been plagued with problems. The man was not surprised when he found several (spiders, was not surprised when he found several (spiders, roaches, and other) bugs in the corner of his room.roaches, and other) bugs in the corner of his room.

ANT

SPY

SEW

ANT

SPY

SEW

{

0

10

20

30

40

50

60

70

80

immediate 3 syll

delay

Am

ou

nt

of

Pri

min

g(u

nre

late

d w

ord

RT m

inu

s

rela

ted

word

RT)

“ant”

“spy”

Cross-Modal Priming Exp. Cross-Modal Priming Exp. 11(Swinney et al. 1978; Onifer & Swinney, (Swinney et al. 1978; Onifer & Swinney, 1981)1981)

Cross-Modal Priming Exp. Cross-Modal Priming Exp. 11(Swinney et al. 1978; Onifer & Swinney, (Swinney et al. 1978; Onifer & Swinney, 1981)1981)Rumour had it that for many years, the government Rumour had it that for many years, the government building had been plagued with problems. The man was building had been plagued with problems. The man was not surprised when he found several (spiders, roaches, not surprised when he found several (spiders, roaches, and other) bugs (insects) in the corner of his room.and other) bugs (insects) in the corner of his room.

ANT

SPY

SEW

ANT

SPY

SEW

{

RiddleRiddle What has wheels and flies, but is not What has wheels and flies, but is not

an airplane?an airplane?

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

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

V

N

Cross-Modal Priming Exp. Cross-Modal Priming Exp. 22(Tanenhaus, Leiman, & Seidenberg, 1979; (Tanenhaus, Leiman, & Seidenberg, 1979; Seidenberg, Tanenhaus, Leiman, & Seidenberg, Tanenhaus, Leiman, & Bienkowski, 1982)Bienkowski, 1982) Noun reading: I bought a Noun reading: I bought a watchwatch..

Verb reading: I will Verb reading: I will watchwatch..

0 200 600

0 200 600

CLOCK

CLOCK

Cross-Modal Priming Exp. 2Cross-Modal Priming Exp. 2(Tanenhaus, Leiman, & Seidenberg, 1979; (Tanenhaus, Leiman, & Seidenberg, 1979; Seidenberg, Tanenhaus, Leiman, & Seidenberg, Tanenhaus, Leiman, & Bienkowski, 1982)Bienkowski, 1982)

Noun reading: I bought a Noun reading: I bought a watchwatch..

Verb reading: I will Verb reading: I will watchwatch..

0 200 600

0 200 600

clock

clock

clock

clock

clock

clock

Effects of Frequency in Effects of Frequency in Ambiguity ResolutionsAmbiguity Resolutions

pitcher port

Equibias Ambiguous Word

Non-Equibias Ambiguous Word

Duffy, Morris, & Rayner Duffy, Morris, & Rayner (1988)(1988)

Varied frequency of homonymsVaried frequency of homonyms Varied whether supportive Varied whether supportive

context came before word or after context came before word or after word.word.

low-level low-level infrared light infrared light eye eye

reflections reflections from cornea from cornea and lens and lens indicate indicate position of position of eye fixation.eye fixation.

Older Eye-trackerOlder Eye-tracker

Head movements messes up calibration Bite bar or head rest is needed

*Control words in Parentheses

*For Non-Equibiased, Context supports non-dominant reading.

Duffy, Morris, & Rayner Duffy, Morris, & Rayner (1988)(1988)

Supportive Context

No Supportive Context

Supportive Context

No Supportive Context

pitcher

port



whiskeyNon-Ambiguous Control

soupNon-Ambiguous Control

No Supportive ContextNo Supportive Context

-- Thickness of the line indicates amount of activation.

pitcher

port



whiskeyNon-Ambiguous Control

soupNon-Ambiguous Control

Adding Supportive Adding Supportive ContextContext

-- Thickness of line indicates amount of activation.

+ supportive context


pitcher + supportive context

port

Equibias

Non-Equibias


Supportive Context No Supportive Context

= High reaction time

Equibiased: Equibiased: – Processing time lower when provided with prior disambiguating Processing time lower when provided with prior disambiguating

contextual support. (Reason: because accessing both contextual support. (Reason: because accessing both meanings)meanings)

Non-equibiased:Non-equibiased:– Processing time high when provided with prior disambiguating Processing time high when provided with prior disambiguating

contextual support supporting the less frequent meaning. contextual support supporting the less frequent meaning. (Reason: made the less frequent more “equal” to the other (Reason: made the less frequent more “equal” to the other meaning)meaning)

– Processing time low when not provided disambiguating Processing time low when not provided disambiguating contextual support for the less frequent meaning. (Reason: not contextual support for the less frequent meaning. (Reason: not considering the less frequent meaning. In fact, time spent later considering the less frequent meaning. In fact, time spent later in disambiguating region is higher due to a need to reanalyze).in disambiguating region is higher due to a need to reanalyze).

Supportive Context No Supportive Context

Lexical AmbiguityLexical Ambiguity

Current conclusionsCurrent conclusions Parallel, rather than serial Parallel, rather than serial

activation activation Relative strength of activation Relative strength of activation

depends on: depends on: – Degree of contextual constraint Degree of contextual constraint

available available – Frequency of use of each meaning Frequency of use of each meaning

SyntaxSyntax

Another level up!Another level up! ParsingParsing:: figuring how the words figuring how the words

in a phrase or sentence combine, in a phrase or sentence combine, using the rules in a grammar.using the rules in a grammar.

ParserParser

Syntactic AmbiguitySyntactic Ambiguity

Is the woman insured?Is the woman insured?

Woman drives off with what she Woman drives off with what she thought was her date’s car (but wasn’t) thought was her date’s car (but wasn’t) and then totaled it. Can she get money and then totaled it. Can she get money from her insurance company:from her insurance company:

Contract says:Contract says:– Such insurance as is provided by this policy Such insurance as is provided by this policy

applies to the use of a non-owned vehicle applies to the use of a non-owned vehicle by the named insured and any person by the named insured and any person responsible for use by the named insured responsible for use by the named insured provided such use is with the permission of provided such use is with the permission of the owner.the owner.

Does he deserve jail Does he deserve jail time?time?

Drug dealer tried to swindle an (unbeknownst to him) Drug dealer tried to swindle an (unbeknownst to him) undercover cop by selling a bag of powder that had undercover cop by selling a bag of powder that had only a minuscule trace of methamhetamine. The only a minuscule trace of methamhetamine. The quantity was not harmful.quantity was not harmful.

Law saysLaw says– Every person who sells any controlled substance Every person who sells any controlled substance

which is specified in subdivision (d) shall be which is specified in subdivision (d) shall be punished.punished.

– (d) Any material, compound, mixture, or (d) Any material, compound, mixture, or preparation which contains any quantity of the preparation which contains any quantity of the following substance having a potential abuse following substance having a potential abuse associated with a stimulant effect on the central associated with a stimulant effect on the central nervous system: Amphetamine, nervous system: Amphetamine, Methamphetamine…Methamphetamine…

The bully hit the girl The bully hit the girl with thewith the......

...stick....stick.

...wart....wart. (*garden-pathed)(*garden-pathed)

The woman felt The woman felt the furthe fur......

...and then left....and then left. ...was very expensive. ...was very expensive. (*garden-pathed)(*garden-pathed)

Local Ambiguities Local Ambiguities (Being led down the “garden-(Being led down the “garden-path”)path”)

Local AmbiguitiesLocal Ambiguities

The bully hit the girl with the wart and then…The bully hit the girl with the wart and then…

The bully hit the girl with the stick and then…The bully hit the girl with the stick and then…

Ambiguous SentencesAmbiguous Sentences

time

yesterday today

Last night, the car crashed.

time

yesterday today

The car crashed.

The reporter said the car crashed last night.

Homework sentence


time

…car...

time

…car....

The reporter said the car crashed last night.

S

NP VP

S AdvP

NP

V

VP

The reportersaid

last night

the carcrashed

V

The reporter

NP VP

S

NP

V

VPsaid

the carcrashed

S

V AdvP

last night


Jamie saw the man with the telescope.

Homework sentence


Jamie saw the man with the telescope.

S

NP VP

PN NP PPV

Det N P NPJamie saw

the manwith

the telescope

S

NP VP

PN NP

PPV

Det N P NP

Jamie saw

the manwiththe telescope

NP

Traditional ViewsTraditional Views(contrasting lexical and syntactic (contrasting lexical and syntactic ambiguity)ambiguity)

Table from MacDonald, Pearlmutter, & Seidenberg Paper

Garden-Path ModelGarden-Path Model(Frazier & Fodor, 1978)(Frazier & Fodor, 1978)

Serial:Serial: the processor initially the processor initially identifies only one analysisidentifies only one analysis– selected based on structural simplicityselected based on structural simplicity

Modular:Modular: Initial structure built on Initial structure built on the basis of syntactic category the basis of syntactic category labels.labels.– revision process incorporates other revision process incorporates other

information.information.

Garden Path ModelGarden Path ModelSelecting the initial analysisSelecting the initial analysis

When word is identified, its syntactic When word is identified, its syntactic category is retrievedcategory is retrieved

Parser identifies which rules of the grammar Parser identifies which rules of the grammar contain that categorycontain that category

Analysis selected on the basis of structural Analysis selected on the basis of structural simplicitysimplicity– Late ClosureLate Closure– Minimal AttachmentMinimal Attachment

Garden Path ModelGarden Path ModelHeuristics for SimplicityHeuristics for Simplicity

Late ClosureLate Closure– When possible, attach incoming lexical When possible, attach incoming lexical

items into the clause or phrase currently items into the clause or phrase currently being processed (i.e., the lowest possible being processed (i.e., the lowest possible nonterminal node dominating the last item nonterminal node dominating the last item analyzed).analyzed).

Minimal AttachmentMinimal Attachment– Attach incoming lexical items into the Attach incoming lexical items into the

phrase-marker being constructed with the phrase-marker being constructed with the fewest nodes consistent with well-fewest nodes consistent with well-formedness rules of language.formedness rules of language.

Late ClosureLate Closure

time

yesterday today

..car…

time

yesterday today

..car…

S

NP VP

S AdvPNP

V

VP

The reportersaid

last night

the carcrashed

V

The reporter

NP VP

S

NP

V

VP

said

the carcrashed

S

V AdvP

last night

Late ClosureLate Closure

The reporter said the car crashed last night

S

VP

Vsaid

The reporter

NP

NP

the car

S

VP

crashedV

AdvP

last night

1 or 2?1

2

Late ClosureLate Closure Choice #1Choice #1 Choice #2Choice #2

…car...

time

Last night…

S

NP VP

S AdvP

NP

V

VP

The reportersaid

last night

the carcrashed

V

The reporter

NP VP

S

NP

V

VPsaid

the carcrashed

S

V AdvP

last night

Minimal AttachmentMinimal AttachmentS

NP VP

PN NP PPV

Det N P NPJamie saw

the manwith

the telescopeS

NP VP

PN NP

PPV

Det N P NP

Jamie saw

the man withthe telescope

NP

Minimal AttachmentMinimal Attachment

NP

PNJamie

S

VP

Vsaw

Jamie saw the man with

PP

Pwith

1 or 2?

NP

theDet

man

N

1

2

Minimal AttachmentMinimal Attachment Choice #1Choice #1

PP

Pwith

NP

PNJamie

S

VP

Vsaw

NP

theDet

man

NPP

Pwith

NP

PNJamie

S

VP

Vsaw

NP

theDet NP

Choice #2Choice #2

NP Det + NNP NP + PP

manN

1 extra node

Garden Path ModelGarden Path ModelHeuristics for SimplicityHeuristics for Simplicity

Late ClosureLate Closure– When possible, attach incoming lexical When possible, attach incoming lexical

items into the clause or phrase currently items into the clause or phrase currently being processed (i.e., the lowest possible being processed (i.e., the lowest possible nonterminal node dominating the last item nonterminal node dominating the last item analyzed).analyzed).

Minimal AttachmentMinimal Attachment– Attach incoming lexical items into the Attach incoming lexical items into the

phrase-marker being constructed with the phrase-marker being constructed with the fewest nodes consistent with well-fewest nodes consistent with well-formedness rules of language.formedness rules of language.

Ambiguities: Ambiguities: Late Closure and Minimal Late Closure and Minimal AttachmentAttachment NP/VP Attachment Ambiguity:NP/VP Attachment Ambiguity:

– The cop [saw [the burglar] [with the binoculars]]The cop [saw [the burglar] [with the binoculars]]

– The cop saw [the burglar [with the gun]]The cop saw [the burglar [with the gun]]

In-Class Exercise (see also homework)

Ambiguities: Ambiguities: Late Closure and Minimal Late Closure and Minimal AttachmentAttachment

NP/S Complement Attachment Ambiguity:NP/S Complement Attachment Ambiguity:

– The athlete [realized [his goal]] last weekThe athlete [realized [his goal]] last week

– The athlete realized [[his shoes] were across the room]The athlete realized [[his shoes] were across the room]



Clause-boundary Ambiguity:Clause-boundary Ambiguity:

– Since Jay always [jogs [a mile]] the race doesn’t Since Jay always [jogs [a mile]] the race doesn’t seem very longseem very long

– Since Jay always jogs [[a mile] doesn’t seem Since Jay always jogs [[a mile] doesn’t seem very long]very long]



Reduced Relative-Main Clause Ambiguity:Reduced Relative-Main Clause Ambiguity:

– [The woman [delivered the junkmail on Thursdays]][The woman [delivered the junkmail on Thursdays]]

– [[The woman [delivered the junkmail]] threw it away][[The woman [delivered the junkmail]] threw it away]



Relative/Complement Clause Ambiguity:Relative/Complement Clause Ambiguity:

– The doctor [told [the woman [that he was in love with]] [to The doctor [told [the woman [that he was in love with]] [to leave]]leave]]

– The doctor [told [the woman] [that he was in love with her]]The doctor [told [the woman] [that he was in love with her]]


Garden-Path ModelGarden-Path Model

Strengths:Strengths: Considers our working memory Considers our working memory

capacitycapacity Speed achieved by considering one Speed achieved by considering one

interpretationinterpretation Explains broad range of phenomenaExplains broad range of phenomena

Models of Sentence Models of Sentence ProcessingProcessing Garden-Path ModelGarden-Path Model

– AutonomousAutonomous Late closureLate closure Minimal attachmentMinimal attachment

Constraint-Based ModelConstraint-Based Model– InteractiveInteractive

Lexical BiasesLexical Biases Referential ContextsReferential Contexts Structural BiasesStructural Biases }

Cues from multiple sourcesconstrain interpretation

Traditional ViewsTraditional Views(contrasting lexical and syntactic (contrasting lexical and syntactic ambiguity)ambiguity)

Table from MacDonald, Pearlmutter, & Seidenberg Paper

Constraint-Satisfaction Model SAYS it’s not the right characterization!

Psy1302 Psychology of Language

Documents

speed of flies

ambiguitytime flies

onifer swinney

vncrossmodal priming

clockclockcrossmodal

arrow proceeds

dominant reading

government building