ECG Formalizing Cognitive Linguistics Community Grammar and Core Concepts Deep Grammatical Analysis Computational Implementation Test Grammars Applied Projects Question Answering Map to Connectionist Models Brain ID: 747159
Download Presentation The PPT/PDF document "Embodied Construction Grammar" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Embodied Construction GrammarECG(Formalizing Cognitive Linguistics)
Community Grammar and Core Concepts
Deep Grammatical Analysis
Computational
Implementation
Test Grammars
Applied Projects – Question Answering
Map to Connectionist Models, Brain
Models of Grammar AcquisitionSlide2
Simulation specification
The analysis process produces a
simulation specification
that includes image-schematic, motor control and conceptual structures provides parameters for a mental simulationSlide3
Summary: ECGLinguistic constructions are tied to a model of simulated action and perception
Embedded in a theory of language processing
Constrains theory to be usable
Basis for models of grammar learningPrecise, computationally usable formalismPractical computational applications, like MT and NLUTesting of functionality, e.g. language learning
A shared theory and formalism for different cognitive mechanisms
Constructions, metaphor, mental spaces, etc.
Reduction to Connectionist and Neural levelsSlide4
physics
lowest energy state
chemistry
molecular fit
biology
fitness, MEU
N
euroeconomicsvision threats, friendslanguage errors, NTL
Constrained Best Fit in Nature
inanimate animate
society, politics
framing, compromiseSlide5
Competition-based analyzer
An analysis is made up of:
A constructional tree
A semantic specificationA set of resolutions
Bill gave Mary the book
Mary
Bill
Ref-ExpRef-ExpRef-ExpGiveA-GIVE-B-Xsubjvobj1obj2book01@Man@WomanGive-Action@Bookgiverrecipient
theme
Johno BryantSlide6
Combined score determines best-fit
Syntactic Fit:
Constituency relations
Combine with preferences on non-local elementsConditioned on syntactic context
Antecedent Fit:
Ability to find referents in the context
Conditioned on syntax match, feature agreement
Semantic Fit:Semantic bindings for frame rolesFrame roles’ fillers are scoredSlide7
0Eve1
walked
2
into3the4
house
5
Constructs
-------------- NPVP[0] (0,5) Eve[3] (0,1) ActiveSelfMotionPath [2] (1,5) WalkedVerb[57] (1,2) SpatialPP[56] (2,5) Into[174] (2,3) DetNoun[173] (3,5) The[204] (3,4) House[205] (4,5)Schema Instances------------------- SelfMotionPathEvent[1] HouseSchema[66] WalkAction[60] Person[4] SPG[58] RD[177] ~ house RD[5]~ Eve Slide8
Unification chains and their fillers
SelfMotionPathEvent[1].mover
SPG[58].trajector
WalkAction[60].walker
RD[5].resolved-ref
RD[5].category
Filler: Person4
SpatialPP[56].mInto[174].mSelfMotionPathEvent[1].spg Filler: SPG58 SelfMotionPathEvent[1] .landmarkHouse[205].mRD[177].categorySPG[58].landmark Filler:HouseSchema66 WalkedVerb[57].mWalkAction[60].routineWalkAction[60].gaitSelfMotionPathEvent[1] .motion Filler:WalkAction60Slide9
Mother (I) give you this (a toy).
CHILDES Beijing Corpus (Tardiff, 1993; Tardiff, 1996)
ma1+ma
gei3
ni3
zhei4+ge
mother
give2PSthis+CLSYou give auntie [the peach].Oh (go on)! You give [auntie] [that].Productive Argument Omission (Mandarin)Johno Bryant & Eva Mok123ni3 gei3 yi22PSgiveauntie
ao
ni3
gei3
ya
EMP
2PS
give
EMP
4
gei3
give
[I]
give
[you] [some peach]
.Slide10
Arguments are omitted with different probabilitiesAll args omitted: 30.6% No args omitted: 6.1%Slide11
Analyzing ni3 gei3 yi2 (You give auntie)Syntactic Fit:
P(Theme omitted | ditransitive cxn) = 0.65
P(Recipient omitted | ditransitive cxn) = 0.42
Two of the competing analyses:
ni3
gei3
yi2
omitted↓↓↓↓GiverTransferRecipientThemeni3gei3omittedyi2↓↓
↓
↓
Giver
Transfer
Recipient
Theme
(1-0.78)*(1-0.42)*0.65 = 0.08
(1-0.78)*(1-0.65)*0.42 = 0.03Slide12
Using frame and lexical information to restrict type of reference
Lexical Unit
gei3
Giver
(DNI)
Recipient
(DNI)
Theme (DNI)The Transfer FrameGiverRecipientThemeMannerMeansPlacePurposeReasonTimeSlide13
Can the omitted argument be recovered from context?Antecedent Fit:
ni3
gei3
yi2
omitted
↓
↓
↓↓GiverTransferRecipientThemeni3gei3omittedyi2↓↓↓
↓
Giver
Transfer
Recipient
Theme
Discourse & Situational
Context
child mother
peach auntie
table
?Slide14
How good of a theme is a peach? How about an aunt?
The Transfer Frame
Giver (usually animate)
Recipient (usually animate)
Theme (usually inanimate)
ni3
gei3
yi2omitted↓↓↓↓GiverTransferRecipientThemeni3gei3omittedyi2
↓
↓
↓
↓
Giver
Transfer
Recipient
Theme
Semantic Fit:
ni3
gei3
yi2
omitted
↓
↓
↓
↓
Giver
Transfer
Recipient
ThemeSlide15
The argument omission patterns shown earlier can be covered with just ONE constructionEach construction is annotated with probabilities of omission
Language-specific default probability can be set
Subj
Verb
Obj1
Obj2
↓
↓↓↓GiverTransferRecipientTheme0.780.420.65P(omitted|cxn):Slide16
Leverage process to simplify representationThe processing model is complementary to the theory of grammarBy using a competition-based analysis process, we can:
Find the best-fit analysis with respect to constituency structure, context, and semantics
Eliminate the need to enumerate allowable patterns of argument omission in grammar
This is currently being applied in models of language understanding and grammar learning. Slide17
Best-fit example with theme omitted
Subj
Verb
Obj1
Obj2
↓
↓
↓↓GiverTransferRecipientThemeYou give auntie [the peach].2Verb↓Transferlocal? omitted?local? omitted?local? omitted?locallocal
Subj
↓
Giver
omitted
local?
omitted?
local
Obj1
↓
Recipient
Obj2
↓
Theme
ni3
gei3
yi2
2PS
give
auntieSlide18
Lexical Unit
gei3
Giver
Recipient
Theme
How to recover the omitted argument, in this case the peach?
The Transfer Frame
GiverRecipientThemeMannerMeansPlacePurposeReasonTime(DNI)(DNI)(DNI)Discourse & Situational ContextchildmotherauntiepeachtableomittedObj2↓ThemeSlide19
Best-fit example with theme omitted
Oh (go on)! You give
[auntie] [that]
.
3
Verb
↓
Transferlocal? omitted?local? omitted?local? omitted?localomittedSubj↓Giveromittedlocal? omitted? local Obj1↓RecipientObj2↓
Theme
ao
ni3
gei3
ya
EMP
2PS
give
EMPSlide20
Lexical Unit
gei3
Giver
Recipient
Theme
How to recover the omitted argument, in this case the aunt and the peach?
The Transfer Frame
GiverRecipientThemeMannerMeansPlacePurposeReasonTime(DNI)(DNI)(DNI)Discourse & Situational ContextchildmotherauntiepeachtableomittedObj2↓Theme
omitted
Obj1
↓
RecipientSlide21
Modeling context for language understanding and learning
Linguistic structure reflects experiential structure
Discourse participants and entities
Embodied schemas:action, perception, emotion, attention, perspective Semantic and pragmatic relations: spatial, social, ontological, causal
‘Contextual bootstrapping’
for grammar learningSlide22
The context model tracks accessible entities, events, and utterances
Discourse & Situational Context
Discourse01
participants: Eve , Mother
objects: Hands, ...
discourse-history: DS01
situational-history: Wash-Action
Discourse:Slide23
Each of the items in the context model has rich internal structure
Situational History:
Discourse History:
Participants:
Objects:
Discourse:
Wash-Action
washer: Eve washee: HandsDS01 speaker: Mother addressee: Eve attentional-focus: Hands content: {"are they clean yet?"} speech-act: questionEve category: child gender: female name: Eve age: 2Mother category: parent gender: female name: Eve age: 33Hands category: BodyPart part-of: Eve number: plural accessibility: accessibleSlide24
Analysis produces a semantic specification
Linguistic Knowledge
Utterance
Discourse & Situational Context
Semantic Specification
World Knowledge
Analysis
“You washed them”WASH-ACTION washer: Eve washee: HandsSlide25
How Can Children Be So Good At Learning Language?
Gold’s Theorem:
No superfinite class of language is identifiable in the limit from positive data only
Principles & ParametersBabies are born as blank slates but acquire language quickly (with noisy input and little correction) → Language must be innate:
Universal Grammar + parameter setting
But babies aren’t born as blank slates!
And they do not learn language in a vacuum!Slide26
Key ideas for a NT of language acquisitionNancy Chang and Eva Mok
Embodied Construction Grammar
Opulence of the Substrate
Prelinguistic children already have rich sensorimotor representations and sophisticated social knowledgeBasic Scenes
Simple clause constructions are associated directly with
scenes basic to human experience
(Goldberg 1995, Slobin 1985)
Verb Island Hypothesis Children learn their earliest constructions (arguments, syntactic marking) on a verb-specific basis(Verb Island Hypothesis, Tomasello 1992)Slide27
Embodiment and Grammar Learning
Paradigm problem for Nature vs. Nurture
The
poverty of the stimulusThe
opulence
of the
substrate
Intricate interplay of genetic and environmental, including social, factors.Slide28
Two perspectives on grammar learning
Computational models
Grammatical induction
language identificationcontext-free grammars, unification grammarsstatistical NLP (parsing, etc.)Word learning modelssemantic representations
logical forms
discrete representations
continuous representations
statistical modelsDevelopmental evidencePrior knowledgeprimitive conceptsevent-based knowledgesocial cognitionlexical itemsData-driven learningbasic sceneslexically specific patternsusage-based learningSlide29
Key assumptions for language acquisition
Significant prior
conceptual/embodied knowledge
rich sensorimotor/social substrateIncremental learning based on experienceLexically specific constructions are learned first.
Language learning tied to
language use
Acquisition interacts with comprehension, production;
reflects communication and experience in world.Statistical properties of data affect learningSlide30
Context
Eve
washer
Wash-Action
Hands
washee
Discourse Segment
addresseeattentional-focusAnalysis draws on constructions and contextbeforebeforeMeaningFormyouAddresseewasherWash-Actionwashedwashee
ContextElement
themSlide31
Learning updates linguistic knowledge based on input utterances
Learning
Discourse & Situational Context
Linguistic Knowledge
Analysis
Utterance
Partial
SemSpecWorld KnowledgeSlide32
Context
Eve
washer
Wash-Action
Hands
washee
Discourse Segment
addresseeattentional-focusContext aids understanding: Incomplete grammars yield partial SemSpecMeaningFormyouAddresseewasherWash-ActionwashedwasheeContextElement
themSlide33
Context
Eve
washer
Wash-Action
Hands
washee
Discourse Segment
addresseeattentional-focusContext bootstraps learning: new construction maps form to meaningMeaningFormyouAddresseeWash-ActionwashedContextElementthembeforebeforewasher
washeeSlide34
Context bootstraps learning: new construction maps form to meaning
Meaning
Form
you
Addressee
Wash-Action
washed
ContextElementthembeforebeforewasherwasheeYOU-WASHED-THEM constituents: YOU, WASHED, THEM form: YOU before WASHED WASHED before THEM meaning: WASH-ACTION washer: addressee washee: ContextElementSlide35
Grammar learning: suggesting new CxNs and reorganizing existing ones
reinforcement
reorganize
merge
join
split
Linguistic Knowledge
Discourse & Situational ContextAnalysisUtterancePartialSemSpecWorld Knowledgehypothesizemap form to meaninglearn contextual constraintsSlide36
Challenge: How far up to generalizeEat riceEat appleEat watermelon
Want rice
Want apple
Want chair
Inanimate Object
Manipulable
Objects
Unmovable ObjectsFoodFurnitureFruitSavoryChairSofaapplewatermelonriceSlide37
Challenge: Omissible constituentsIn Mandarin, almost anything available in context can be omitted – and often is in child-directed speech.Intuition:
Same context, two expressions that differ by one constituent
a general construction with the constituent being omissible
May require verbatim memory traces of utterances + “relevant” contextSlide38
When does the learning stop?Most likely grammar given utterances and contextThe grammar prior includes a preference for the “kind” of grammar
In practice, take the log and minimize cost
Minimum Description Length (MDL)
Bayesian Learning Framework
Schemas +
Constructions
SemSpec
Analysis + ResolutionContext FittingreorganizehypothesizereinforcementSlide39
Intuition for MDLS -> Give me NPNP -> the book
NP -> a book
S -> Give me NP
NP -> DET bookDET -> theDET -> a
39
Suppose that the prior is inversely proportional to the size
of the grammar (e.g. number of rules)
It’s not worthwhile to make this generalizationSlide40
Intuition for MDLS -> Give me NPNP -> the book
NP -> a book
NP -> the pen
NP -> a penNP -> the pencilNP -> a pencilNP -> the markerNP -> a marker
S -> Give me NP
NP -> DET N
DET -> the
DET -> aN -> bookN -> penN -> pencilN -> markerSlide41
Usage-based learning: comprehension and production
reinforcement
(usage)
reinformcent
(correction)
reinforcement
(usage)
hypothesize constructions& reorganizereinforcement(correction)constructiconworld knowledgediscourse & situational contextsimulationanalysisutterance
analyze
&
resolve
utterance
response
comm. intent
generate