Defining the Learning Problem The output of learning is complex Examples that t wanna contraction parasitic gaps reconstruction etc etc The output of learning is hard to observe Crucial input for learning is hard to observe ID: 932542
Download Presentation The PPT/PDF document "Linguistic Computation for Language Lear..." 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
Linguistic Computation for Language Learners
Slide2Defining the Learning Problem
The output of learning is complex
Examples: that-
t
,
wanna
contraction, parasitic gaps, reconstruction, etc. etc.
The output of learning is hard to observe
Crucial input for learning is hard to observe
It’s noisy (on both sides of the ear)
It’s dissimilar from what must be learned
It’s rare
Yet learning is robust
We should be able to describe the learning problem at multiple grains of analysis, just like the output
of learning
Slide3Obvious variation
English verbs precede their objects (
ate the pizza
)
Japanese verbs follow their objects (
piza-o
tabeta
)
English distinguishes the vowels in
sheep
and
ship
Spanish does not
All Russian verbs encode
aspect
(± completed action)
English verbs do not
etc. etc. etc.
Slide4Easy to Observe
English is an SVO language, Japanese is an SOV language
John ate the pizza.
John-
ga
piza-o
tabeta
.
English
wh
-questions involve
wh
-fronting, Chinese counterparts do not
Who did Sally meet __ ?
Sally met who?
[
Chinese]
English main verbs follow adverbs, French main verbs precede adverbs
Joe always drinks coffee in the morning.
Jean
boit
toujours
du café avec son petit
déjeuner
. [
CP’s
bad French]
J. drinks always coffee with his breakfast
Slide5Not-so-obvious variation
Example 1: Pronoun Interpretation
While
John
was reading the book,
he
ate an apple.
While
he
was reading the book, John ate an apple.John ate an apple while he was reading the book.He ate an apple while John was reading the book.Example 2: Constraints on questionsWhat do you think Sally ate ___?What do you think that Sally ate ___?Who do you think ___ ate the donut?Who do you think that ___ ate the donut?
English - ok
Russian - not ok
Impossible in
all
languages
English -
bad
Italian - ok
Slide6Language typology & learning
The Big Idea:
identifying constraints on language variation and explaining the success of language learning
are essentially the same problem
Universals:
properties that are common to all human languages do not need to be learned
Co-variation:
clusters of non-universal properties that consistently co-occur in a language reflect a single underlying trait (and so those properties do not need to be learned individually)
Ensuring learning success:
any non-obvious property that must be learned should be part of cluster that includes an ‘obvious’ property, thereby ensuring reliable learningCurrent status …
Slide7Possibly Universal
Principle C
While John was reading the book he ate an apple.
While he was reading the book John ate an apple.
John ate an apple while he was reading the book.
*He ate an apple while John was reading the book.
Sally thinks that she is the best dancer.
*She thinks that Sally is the best dancer.
Slide8A Constraint on Interpretation
S
NP
VP
V
NP
John
ate
the apple
S’
S
while
S
NP
VP
Comp
he
was reading the book
While he was reading the book, John ate the apple
Slide9A Constraint on Interpretation
S
NP
VP
V
NP
he
ate
the apple
S’
VP
while
S
NP
VP
Comp
John
was reading the book
He ate the apple while John was reading the book
Slide10Universals
Example 1: Pronoun Interpretation
While
John
was reading the book,
he
ate an apple.
While
he
was reading the book, John ate an apple.John ate an apple while he was reading the book.He ate an apple while John was reading the book.
English - ok
Russian - not ok
Impossible in
all
languages
3 year olds in
Russian & English
Kazanina & Phillips 2001
2.5
yr
olds:
Lukyanenko
et al. 2015
EARLY
LATE
Slide11Co-variation
Example 2: Constraints on questions
What do you think Sally ate ___?
What do you think that Sally ate ___?
Who do you think ___ ate the donut?
Who do you think that ___ ate the donut?
This variation is linked to the possibility of post-verbal subjects
e.g.,
telefono
Pavarotti (‘Pavarotti called’)Languages that allow post-verbal subjects also allow the red examplePost-verbal subjects are easy for learners to observe
English -
bad
Italian - ok
Slide12Indirect learning argument
A. Learning Happens
P1.
That-
t
effects are reliable, grammatical effects for individual speakers
P2. Members of a speech community agree [
convergence
] C1. Need to explain consensusP3. There is variation between communities C2a. Hard-coding the surface phenomenon is not viable C2b. Experience must explain the consensus [learning]B. Not DirectlyP4. Evidence for the surface pattern could come from P4a. Explicit instruction P4b. Distribution in input that reflects the community consensus
P4c. Input contains cues that are informative to more ‘selective’ learner
P5. None of the above works a. no relevant feedback. b
. relevant input is absent/misleading, …C. Epiphenomenon of ObservablesP6. Acceptable that-t sentences reflect alternative structural parse
P7. There are readily observable correlates of the alternative parse
Slide13Co-variation
Example 2: Constraints on questions
What do you think Sally ate ___?
What do you think that Sally ate ___?
Who do you think ___ ate the donut?
Who do you think that ___ ate the donut?
This variation is linked to the possibility of post-verbal subjects
e.g.,
telefono
Pavarotti (‘Pavarotti called’)Languages that allow post-verbal subjects also allow the red examplePost-verbal subjects are easy for learners to observe
English -
bad
Italian - ok
/11308
159
2
13
0
Slide14Null Subject Parameter
Cluster of related properties vary together (
Rizzi
, Chomsky)
Null subjects
Lack of expletive subjects (‘It is raining’, ‘It is clear that it’s icy outside’)
Post-verbal subjects
Lack of that-trace effects
Suggestion: since the members of this cluster are not independent properties of language – they are reflections of a single underlying trait – a learner need only master one of them in order to know the status of all of them.
Slide15But does it work?
Parameters that work …
Parameter learning mechanisms …
Evidence of parameter-setting in learning …
Slide16So far
…
Language learning & language typology
Invariance & co-variation (~ principles & parameters)
Hard-to-observe variation must be linked to easy-to-observe variation
Universals don’t need to be learned
Clarification #1: could reflect domain-specific or domain-general properties of humans
Clarification #2: learning-as-experience/practice ≠ learning-as-choosing (cf. walking)
Co-varying properties reflect a shared trait
Clarification #1: we need to guard against accidental co-variationClarification #2: ‘shared trait’ usually understood as representational unit, but could involve different connections, e.g., property X makes it possible to learn Y
Slide17Next
…
Scope of language variation
Null subject parameter & that-trace effects
A valley in Sweden vs. The World
Some properties that are hard-to-observe, yet seem to vary
Verbs, scope
Consistent vs.
inconsistent variation
Understanding hard-to-observe variation: island constraintsReducing variation to other propertiesQuantitative measures of variability
Slide18Micro-variation
Greatly expanded database of language information
Worldwide typological surveys
Dense regional dialect projects
Reliable clusters are harder to find.
Not good news for learners.
Large-scale studies biased towards more easy-to-observe phenomena
Important challenge:
does variation in ‘non-obvious’ properties show micro-variation?
More rigid constraints in domains where learning is more difficult?Testing semantic variation.
Slide19Anders Holmberg
Newcastle, UK
Ian Roberts
Cambridge, UK
Theresa
Biberauer
Cambridge, UK
Fritz
Newmeyer
Vancouver, Canada
The Null Subject Parameter
(what’s left of it)
Winter Storm 2013
Slide20Slide21Null subjects
Gilligan 1987 (USC PhD): survey of 102 languages
Newmeyer
: “These results are not very heartening for […]
Rizzi’s
theory”
Slide22Roberts & Holmberg 2005
Slide23In a valley in Sweden …
Mainland Scandinavian: Danish, Swedish, Norwegian
Insular Scandinavian: Icelandic, Faroese, Medieval
MSc
.;
Älvdalen
dialect of Swedish
Cluster of properties distinguishes these two groups
Null
non-referential subjects in ISc. (‘Now have __ come many students’)Non-nominative subjects in ISc.‘Stylistic fronting’ in ISc. (‘Forth has come that fished has been illegally’)Verb-raising across negation in ISc.Richer subject-verb agreement in ISc.Like MSc
.: modern English, modern French (~)Like
ISc.: Old French, Middle English, Yiddish
Slide24Älvdalen
Slide25So far …
Structure of cross-linguistic variation is structure of the learning problem
Universals … Clusters/Parameters … Isolated facts.
Idea: clusters of surface properties reflect a shared underlying trait
Many questions about the prospects of identifying reliable clusters (
microvariation
).
Failure to find reliable clusters does not dissolve the question of hard-to-observe properties
Evidence for
microvariation & clusters may be misleading, due to: (i) data sampling bias towards obvious properties (ii) since clusters reflect abstract traits, their surface realizations may be ambiguous [different consequences for ±obvious properties]Status of clusters motivates contrasting learning approaches: (i) cue-based(ii) model-fitting … but either case requires detectible evidence that leads learner to change
Slide26Slide27what is the corpus?
questions, relative clauses, etc.
reliable or noisy input data?
hopefully parsed right
Slide28Slide29Slide30Slide31Slide32What PS model does well
Generalizes beyond input
Distinguishes non-occurring/difficult from non-occurring/impossible sentences
‘learns island constraints’
Slide33Does it do so well?
Distinguishing long/hard vs. impossible
Data Sparseness
too many categories
too little data
limits of trigrams
Cross-language variation
Generalizing across dependency types
Slide34Slide35Data Sparseness?
9
3
= 729
15
3
= 3,375
15
4
= 50,625Estimated corpus size = 175,000 wh
-questions (3-year period)
Approx. 5,000 per month, 160 per day
Slide36a. Who do you think that John met __? 2/20923
b
. Who do you think John met __? 236/20923
c
. *Who do you think that __ left? 0/20923
d
. Who do you think __ left? 24/20923
2011 Version
April 2012 Version
Slide37Co-variation
Example 2: Constraints on questions
What do you think Sally ate ___?
What do you think that Sally ate ___?
Who do you think ___ ate the donut?
Who do you think that ___ ate the donut?
This variation is linked to the possibility of post-verbal subjects
e.g.,
telefono
Pavarotti (‘Pavarotti called’)Languages that allow post-verbal subjects also allow the red examplePost-verbal subjects are easy for learners to observe
English - bad
Italian - ok
Slide38Null Subject Parameter
Cluster of related properties vary together (
Rizzi
, Chomsky)
Null subjects
Lack of expletive subjects (‘It is raining’, ‘It is clear that it’s icy outside’)
Post-verbal subjects
Lack of that-trace effects
Suggestion: since the members of this cluster are not independent properties of language, a learner need only master one of them in order to know the status of all of them.
Slide39Data Noisiness
The PS corpus of
wh
-questions is
very
clean. This is surprising.
… and it is valuable for their learner’s success
But does this really help the learner?
Input to learners contains many errors, e.g., in agreement, verb complementation
*The plate next to your toys need to be put away. *Can you fill the milk into that cup. [many utterances from non-native speaking parents]So doesn’t the learner need to assume that all input could be noisy?
Slide40Variation in Island Constraints
Some constraints are universal (mostly), some clearly vary
Some variation seems reducible to other properties
Operationalizing “island effects”
Different types of amelioration
Many UMD contributions: Masaya Yoshida, Jon
Sprouse
, Lisa Pearl, Akira
Omaki
, Dave Kush, Dustin Chacón, Nick Huang
Slide41Island Constraints
Complement clause …
John thinks [that Mary gave a book to which boy]?
Which boy does John think that Mary gave a book to __?
Relative clause …
John likes the book [that Mary gave to which boy]?
*Which boy does John like the book that Mary gave to __?
Adjunct (conditional) clause …
John will cry [if Mary gives a book to which boy]?
*Which boy will John cry if Mary gives a book to __?
Slide42That-t Effects
Example 2: Constraints on questions
What do you think Sally ate ___?
What do you think that Sally ate ___?
Who do you think ___ ate the donut?
Who do you think that ___ ate the donut?
This variation is linked to the possibility of post-verbal subjects
e.g.,
telefono
Pavarotti (‘Pavarotti called’)Languages that allow post-verbal subjects also allow the red examplePost-verbal subjects are easy for learners to observe
English - bad
Italian - ok
/11308
159
2
13
0
Slide43Escapable Relative Clauses
English
*The man [
who
i
[the suit [
RC
that __
i
is wearing]] is dirty] arrived late.Japanesekiteiru yoohuku-ga yogoreteiru sinsiis.wearing suit.nom dirty.is gentlemanMajor Subject Construction (Japanese, Korean, Chinese)[
IP sono
sinsii-ga
[NP [CP proi __j
kiteiru] [yoohukuj]]-ga yogoreteiru]
that gentleman-nom pro wearing-is suit-nom dirty-is‘That gentleman is such that the suit that he is wearing is dirty.’
[CP
Opi [
IP __i [NP [CP proi __
j kiteiru] yoohukuj]-ga
yogoreteiru] [sinsii]] Op pro wearing-is suit-
nom dirty-is gentleman ‘The gentleman who the suit that he is wearing is dirty.’
Slide44Escapable Relative clauses
Relative Clauses
John knows a man [who believes in aliens]
John knows a man [who believes in what]
*What does John know a man [who believes in ___]
Swedish
Den
teorin
känner jag ingen [som tror på __]that theory know I noone [who believes in __]EnglishThis is a theorem that I need to find somebody who understands __.
*I studied the theorem that John met the mathematician who proved __.
What did John go to the store to buy __?
Slide45English
Japanese
Which boy
does John like [
NP
the book [
RC
that Mary
gave to
]] ?
*
どの男の子に
太郎は
[[
花子が
あげた RC
]本
NP]]が好きなの?
Wh-Question Formation
which boy
*
Scrambling
Scrambling cannot escape relative clauses.
(Saito 1985)
Wh-movement cannot escape relative clauses.
(Ross 1967)
OK
Which
boy
does John think [
CP
that Mary
gave a book to __ ]
OK
どの男の子に
太郎は
[
CP
花子が本を__あげたと
]
思っているの?
Relative Clauses are island
s
Cross-Language Uniformity
Slide46Cross-language Variation
English
*Which boy will John cry if Mary gives a present to __?
Japanese
Dono-gakusee-ni
Taroo-wa
[
Hanako-ga
__ present-
o
which-student-
dat
T-top H-nom present-acc
ageta-ra
]
nakidasu
-no?
give-cond cry-Q?
“Which student will
Taroo
cry if
Hanako gives a present to”
Adjunct Clauses are islands in English, but not in Japanese.
Slide47Languages
Head Final
Null Arguments
Scrambling
Wh-in-situ
(Overt) Det
Indeterminate
system
Japanese
Japanese
Type
Korean
Japanese
Type
Malayalam
Japanese
Type
Russian
Japanese
Type
Romance
Japanese
Type
Chinese
Chinese
Type
Basque
English
Type
Turkish
English
Type
English
English
Type
Cuzco Quechua
English
Type
Typological variation
(
Yoshida 2006: summary of previous studies and field work
)
We cannot attribute adjunct (non)islandhood to just one of these features
What is the combination of features that contains sufficient features?
Can we find any features that uniquely distinguish
these languages from others?
Slide48Languages
Head Final
Null Arguments
Scrambling
Wh-in-situ
(Overt) Det
Indeterminate
system
Japanese
Japanese
Type
Korean
Japanese
Type
Malayalam
Japanese
Type
Russian
Japanese
Type
Romance
Japanese
Type
Chinese
Chinese
Type
Basque
English
Type
Turkish
English
Type
English
English
Type
Cuzco Quechua
English
Type
Typological variation
(
Yoshida 2006: summary of previous studies and field work
)
We cannot attribute adjunct (non)islandhood to just one of these features
What is the combination of features that contains sufficient features?
Slide49Parasitic Gaps
Coordinate Structures
Slide50Assignment #2
Read the following:
Noam Chomsky. 1975.
Reflections on Language
. NY:
Praeger
. (chapter 1)
Steven Pinker. 1989.
Learnability
and Cognition. MIT Press. (chapter 1)Takuya Goro. 2007. Language-specific constraints on scope interpretation in first language acquisition. PhD dissertation, U of Maryland. (selections)Each of these works describes a learning problem in a different domain of grammar. To what extent do these problems present the same or different challenges for a learner? To what extent might the challenges be addressed by assuming that the child has the benefit of substantial innate knowledge, or a very powerful distributional learning mechanism?Due Weds 2/28
Slide51Phenomenon #1
Subject-auxiliary inversion & structure-dependence
Wallace has always liked cheese.
Has Wallace always liked cheese?
Gromit
is afraid of penguins.
Is
Gromit
afraid of penguins?
The dog that is afraid of penguins has always liked cheese.…?
Slide52Phenomenon #2
Dative alternation
John gave a book to Mary.
John gave Mary a book.
John sent a book to Mary.
John sent Mary a book.
John bought a book for Mary.
John bought Mary a book.
*John donated the museum a painting.
*John delivered Mary a book.*John purchased Mary a book.
Slide53Phenomenon #2
Locative alternation
John poured the water into the glass.
*John poured the glass with water.
*John filled the water into the glass.
John filled the glass with water.
John sprayed the water onto the wall.
John sprayed the wall with water.
Slide54Scope Variation
Scope Flexibility:
Some animal ate every piece of food.
Takuya
Goro
,
UMd
2002-7, Assoc. Prof.
Tsuda
College, Japan
Slide55Some animal ate every piece of food.
Ambiguous between surface and inverse scope.
Slide5611/04/2007
GORO et al./BUCLD32
56
Scope rigidity in Japanese
Someone criticized every professor
>> / >>
Dareka-ga dono kyoujyu mo hihan-sita
someone-nom every professor criticize-did
“Someone criticized every professor”
>> /
* >>
Japanese is a “scopally rigid” language. (e.g., Hoji 1985; Marsden 2004)
Slide57Backwards Anaphora
Principle C
While he was reading the book John ate an apple.
*He ate an apple while John was reading the book.
Russian counterparts
(Kazanina & Phillips 2001 et seq.,
Avrutin
&
Reuland
2002)*While he was reading the book John ate an apple.*He ate an apple while John was reading the book.The story that she read upset the girl.While his mother was reading the book John ate an apple.Before he read the book John ate an apple.
Slide58Aspectual Interpretation
Simple Clauses
Dutch PST – completion entailment
Russian IMP – no completion entailment
With overt frame of reference
Dutch PST – no completion entailment
Russian IMP – no completion entailment
Kazanina & Phillips, 2007,
Cognition
Slide59Long-distance Reflexives
Chinese
ziji
(‘self’)
Zhangsan
renwei
Lisi
zhidao Wangwu xihuan zijiZ thinks L knows W. Likes selfVarying discourse conditions on long-distance antecedents (in addition to syntactic conditions) (Cole, Hermon, & Lee (2001)Discourse conditions on logophors (Sells, 1987)SOURCE (source of communication)SELF (one whose mental state the sentence describes)PIVOT (perspective of the sentence)
Singapore Mandarin: PIVOTSingapore
Teochew: PIVOT + SOURCE/SELF
Slide60State of play so far
Defining
endstate
Identifing
what has to be learned, and how (adding/subtracting)
Identify what evidence is needed to accomplish that learning
So what can children tell us?
Various things …
Identify moments in development: learn about over/
undergenerationIdentify approximate timing of change, constrains learning modelsLearn about input, and possibly about uptake(And how do we figure out what children know?)
Slide61Verb Alternations
Slide62Baker (1979)
Alternating Verbs
John gave a cookie to the boy.
John gave the boy a cookie.
Mary showed some photos to her family.
Mary showed her family some photos.
Non-Alternating Verbs
John donated a painting to the museum.
*John donated the museum a painting.
Mary displayed her art collection to the visitors.*Mary displayed the visitors her art collectionLearnability problem: how to avoid overgeneralization
Slide63Verb Argument Structure
“Locative Verbs”
Sally poured the water into the glass.
*Sally poured the glass with water.
*Sally filled the water into the glass.
Sally filled the glass with water.
Sally piled the books on the table.
Sally piled the table with books.
Slide64Verb Argument Structure
“Locative Verbs”
Sally poured the water into the glass.
*Sally poured the glass with water.
*Sally filled the water into the glass.
Sally filled the glass with water.
Sally piled the books on the table.
Sally piled the table with books.
Figure-verbs -- manner of motionpour, spill, drip, shake, etc.
Ground-verbs
-- change of state
fill, cover, decorate, soak, etc.
Alternator-verbs
-- manner & changepile, scatter, load, etc.
Slide65Verb Classes
Assumptions
1. Linking rules are consistent across languages
2. Linking rules need not be learned
Slide66Verb Argument Structure
Demonstrations of productivity (Gropen et al., 1991)
Children (aged 3;5 upwards) are taught only the meaning of new verbs
Children infer appropriate syntactic frames from the meanings that they’re taught
Slide67Slide68Slide69Slide70Experiment 2 - ‘purer endstate verb’
Slide71Languages Vary
English
*John decorated the flowers in the room.
John decorated the room with flowers.
Korean
Yumi-ka ccoch-ul pang-ey cangsikha-yess-ta
Nom flowers-Acc room-Loc decorate-Past-Dec
‘John decorated the flowers in the room.’
Yumi-ka pang-ul ccoch-ulo cangsikha-yess-ta
Nom room-Acc flowers-with decorate-Past-Dec‘John decorated the room with flowers.’Change-of-state--> Ground Frame
Korean is more liberal than English
Slide72English
John piled the books on the table.
John piled the table with books.
Korean
Yumi-ka chaek-lul chaeksang-ey ssa-ass-ta.
Nom book-Acc table-Loc pile-Past-Dec
‘Yumi piled books on the table.’
*Yumi-ka chaeksang-lul chaek-elo ssa-ass-ta.
Nom table-Acc books-with pile-Past-Dec
‘Yumi piled the table with books.’Languages VaryKorean is more restrictive than English - conflates pile-class and pour
-class.
Slide73English
Korean
Turkish
Chinese
Japanese
Yoruba
Hebrew
French
Spanish (Castilian)
Spanish (Argentinian)
Arabic
Thai
Luganda
Malay
Hindi
Ewe
Italian
Brazilian Port.Russian
Polish
Cross-Language Survey
Typological Survey
Slide74Cross-Language Survey
Survey I
English Turkish
Korean
Luganda
French Hindi
Japanese Hebrew
Chinese Malay
Thai Arabic
Survey IIItalian YorubaPolish EweJapanese RussianFrench EnglishBrazilian Portuguese Spanish (Argentinian) Spanish (Castilian)
Less detailed
classification used
(~15 verbs)
Slide75Cross-Language Survey
Survey I
English Turkish
Korean
Luganda
French Hindi
Japanese Hebrew
Chinese Malay
Thai Arabic
Survey IIItalian YorubaPolish EweJapanese RussianFrench EnglishBrazilian Portuguese Spanish (Argentinian) Spanish (Castilian)
More detailed
classification used
(~30 verbs)
Slide76Verb Argument Structure
Simple VP Structures
She filled the water into the glass.
She stuffed feathers into the pillow.
Simple VP Structures
She poured the glass with water.
She piled the shelf with books.
Adjectival Passives
The filled water.
The stuffed feathers. (*)Verb SerializationShe pour-put the glass.She pile-put the shelf.
English
Korean
diff.
same
Korean
diff.
Korean
diff.
English
Korean
diff.
same
(Kim, 1999; Kim, Landau, & Phillips, 1999)
Verb class contrasts that seem to disappear in simple/frequent
structures reemerge in constructions that are less frequent.
Slide77Verb Argument Structure
Languages vary, but not in systematic ways … relatively
Learners must use the input – but how?
How do they identify what is possible and impossible?
How could (built-in) knowledge of cross-language regularities help them?
Slide78Phenomenon #1
Subject-auxiliary inversion & structure-dependence
Wallace has always liked cheese.
Has Wallace always liked cheese?
Gromit
is afraid of penguins.
Is
Gromit
afraid of penguins?
The dog that is afraid of penguins has always liked cheese.…?
Slide79Phenomenon #2
Dative alternation
John gave a book to Mary.
John gave Mary a book.
John sent a book to Mary.
John sent Mary a book.
John bought a book for Mary.
John bought Mary a book.
*John donated the museum a painting.
*John delivered Mary a book.*John purchased Mary a book.
Slide80Phenomenon #2
Locative alternation
John poured the water into the glass.
*John poured the glass with water.
*John filled the water into the glass.
John filled the glass with water.
John sprayed the water onto the wall.
John sprayed the wall with water.
Slide81Scope Variation
Scope Flexibility:
Some animal ate every piece of food.
Takuya
Goro
,
UMd
2002-7, Assoc. Prof.
Tsuda
College, Japan
Slide82Some animal ate every piece of food.
Ambiguous between surface and inverse scope.
Slide8311/04/2007
GORO et al./BUCLD32
83
Scope rigidity in Japanese
Someone criticized every professor
>> / >>
Dareka-ga dono kyoujyu mo hihan-sita
someone-nom every professor criticize-did
“Someone criticized every professor”
>> /
* >>
Japanese is a “scopally rigid” language. (e.g., Hoji 1985; Marsden 2004)
Slide8411/04/2007
GORO et al./BUCLD32
84
Experiment – basic logic
Dareka-ga
dono
tabemono
-mo
tabeta
someone-nom every food ate
“Someone ate every food”Pig1 Pig2 Pig3
Cream puff Banana CarrotThe
>> interpretation (surface scope
) is false;
no single individual ate every foodThe
>> interpretation (inverse scope
) is true;
every food was eaten by somebody
Slide8511/04/2007
GORO et al./BUCLD32
85
Experiment – basic logic
Dareka-ga
dono
tabemono
-mo
tabeta
someone-nom every food ate
“Someone ate every food”Pig1 Pig2 Pig3
Cream puff Banana CarrotJapanese adults should consistently reject the test sentence.
Conservative learners of Japanese should consistently reject the test sentence.
Slide8611/04/2007
GORO et al./BUCLD32
Felicity conditions for using indefinites
The “speaker-unknown” context (e.g., Haspelmath 2000)
Someone stole my bike!
The speaker doesn’t know the identity of the individual who stole his bike.
Slide8711/04/2007
GORO et al./BUCLD32
The “eating-game” story
12 groups of animals: each consists of 3 animals of the same kind
Each group was invited to eat three pieces of food
A gold medal was awarded to the teams where each member ate a different food
Slide8811/04/2007
GORO et al./BUCLD32
“Team Pigs”
Slide8911/04/2007
GORO et al./BUCLD32
“This one ate the banana…”
Slide9011/04/2007
GORO et al./BUCLD32
“And this big pig ate the cream puff…”
Slide9111/04/2007
GORO et al./BUCLD32
“The small pig ate the carrot…”
Slide9211/04/2007
GORO et al./BUCLD32
“They got a gold medal!”
Slide9311/04/2007
GORO et al./BUCLD32
The “eating game” phase goes on until all the 12 teams finish their trials.
Among the 12 teams, only 4 teams get a gold medal
After the “game” phase finishes, we move back to the first team, the pigs.
Kermit the Frog guesses how well each team did in the game.
Slide9411/04/2007
GORO et al./BUCLD32
Experimenter:
Kermit! Do you know what this team did in the game?
Slide9511/04/2007
GORO et al./BUCLD32
Kermit: Well, I don’t really remember what they did…but they have a gold medal!, which means,
“TEST SENTENCE”
Slide9611/04/2007
GORO et al./BUCLD32
Results: % acceptance of inverse scope
Slide9711/04/2007
GORO et al./BUCLD32
97
Japanese children = English children/adults
Japanese children accepted the inverse scope interpretation just as much as English children/adults did.
The degree of acceptance of inverse scope interpretations in English adults is quite similar to results from other studies (e.g.,
Kurtzman
& MacDonald 1993)
Slide9811/04/2007
GORO et al./BUCLD32
98
Scrambling and Scope Reconstruction
Slide9911/04/2007
GORO et al./BUCLD32
99
Scrambling and scope-reconstruction
Dareka-ga dono tabemono mo tabeta
Someone-nom every food ate
“Someone ate every food”
>> / * >>
[Dareka-o]
i
dono doubutsu mo t
i
tataita
someone-ACC every animal hit
Lit. “Someone, every animal hit” >> / OK
>>
Slide10011/04/2007
GORO et al./BUCLD32
100
Not all QNPs scope-reconstruct
[Piza mo pasuta mo]
i
Taroo-dake-ga t
i
tabeta
both pizza and pasta Taroo-only-NOM ate
Lit. “Both pizza and pasta, only Taroo ate”
>> ONLY /
*ONLY >> A scrambled X mo Y mo “both X and Y” does not scope-reconstruct.
Slide10111/04/2007
GORO et al./BUCLD32
101
Experiment – basic logic
[
Aoi
hako
mo
kuroi
hako
mo]i
Pikachu-dake-ga
t
i
aketaboth blue box and black box Pikachu-only-nom openedLit. “Both blue box and black box, only Pikachu opened”
The
>>ONLY interpretation (surface scope):
“Only Pikachu opened the blue box, and
only Pikachu opened the black box”
The ONLY>> interpretation (inverse scope
): “Pikachu was the only one who opened
both of the boxes”
nobody
else opened
any
of the boxes
nobody else opened
both
of the boxes
Slide10211/04/2007
GORO et al./BUCLD32
102
Experiment – basic logic
Situation
blue box black box
Pikachu
Doraemon
* *
Anpan-man *
The >>ONLY interpretation (
surface scope
):
nobody else opened any of the boxesThe ONLY>>
interpretation (inverse scope):
nobody else opened
both of the boxes
Slide10311/04/2007
GORO et al./BUCLD32
103
Experiment – basic logic
Situation
blue box black box
Pikachu
Doraemon
* *
Anpan-man *
The >>ONLY interpretation (
surface scope
) is false
;Anpan-man opened the black box
The ONLY>> interpretation (inverse scope
): nobody else opened
both of the boxes
Slide10411/04/2007
GORO et al./BUCLD32
104
Experiment – basic logic
Situation
blue box black box
Pikachu
Doraemon
* *
Anpan-man *
The >>ONLY interpretation (
surface scope
) is false
:Anpan-man opened the black box
The ONLY>> interpretation (inverse scope
) is true;
nobody else opened
both of the boxes
Slide10511/04/2007
GORO et al./BUCLD32
105
Experiment – basic logic
Situation
blue box black box
Pikachu
Doraemon
* *
Anpan-man *
Japanese adults should consistently reject the test sentence.
Conservative learners of Japanese should consistently reject the test sentence.
Slide10611/04/2007
GORO et al./BUCLD32
106
The experiments
A Truth Value Judgment Task with computer-generated animation
Participants:
Japanese children
(N=16, 4;11-5;10, Mean 5;6)
Japanese adults
(N=16)
4 crucial test trials + 6 fillers
Participants who made more than two errors in filler trials were excluded from the final data analysis.
Slide10711/04/2007
GORO et al./BUCLD32
107
Pikachu is trying to open those boxes using his psychic-power…
Slide10811/04/2007
GORO et al./BUCLD32
108
“Pika-pika-pii!”
Slide10911/04/2007
GORO et al./BUCLD32
109
“There are beautiful rings in the box!”
Slide11011/04/2007
GORO et al./BUCLD32
110
“Take one of those rings, Pikachu.”
Slide11111/04/2007
GORO et al./BUCLD32
111
The final outcome
Slide11211/04/2007
GORO et al./BUCLD32
112
Kermit: I know what happened! “
TEST SENTENCE
”
Slide11311/04/2007
GORO et al./BUCLD32
113
Results: % acceptances of the crucial test sentences
Slide11411/04/2007
GORO et al./BUCLD32
114
Japanese children
Japanese adults
Japanese adults consistently rejected the inverse scope interpretation with scrambled
X mo Y mo.
In contrast, Japanese children overwhelmingly accepted the inverse scope interpretation of the crucial test sentences.
Slide115Takuya
Goro
,
UMd
2002-7, Assoc. Prof.
Tsuda
Coll., Japan
Tests of interpretations that involve uncertainty
Japanese disjunction
Scope flexibility (we just saw this)
Slide116English vs. Japanese (1)
John speaks Icelandic
or
Swahili.
(
but I’m not sure which language he can actually speak…
)
John-
wa
Icelandic ka Swahili-wo hanas-u. John-TOP or -ACC speak-pres.
(but I’m not sure which language he can actually speak…
)
The interpretations of disjunctions are more or less same in both languages.
Slide117English vs. Japanese (2)
John does
n’t
speak Icelandic
or
Swahili.
John doesn’t speak Icelandic
AND he doesn’t speak Swahili.
John-
wa
Icelandic ka Swahili-wo
hanasa-na-i
John-TOP or -ACC speak-neg-pres.
John
doesn’t speak Icelandic OR he doesn’t speak Swahili.
(I know it is either one of those languages that John cannot speak, but I’m not sure which one…)
Slide118‘Neither’ interpretation in Japanese
John-
wa
Icelandic
mo
Swahili
mo
hanas-u
. John-TOP also also speak-pres. “John speaks both Icelandic and Swahili”John-wa Icelandic mo Swahili mo
hanase-na-i
John-TOP also also speak-
neg-pres.
John speaks neither Icelandic nor Swahili.
Slide119Disjunction and parameter
Let’s say that UG provides the universal disjunction operator OR, associated with a parameter={+PPI, -PPI}
OR(+PPI)
disjunctions in Japanese / Hungarian / Russian / Italian…
OR(-PPI)
disjunctions in English / German / Korean…
(cf.
Szabolcsi 2002)
Slide120Question about children
Can Japanese children accept the wide-scope reading of
ka
in (4)?
John-
wa
Icelandic
ka
Swahili-
wo hanasa-na-i John-TOP or -ACC speak-
neg-pres.
Can they accept (4) in the situation where John cannot speak Icelandic but he can speak Swahili?
If they have the –PPI setting, they should say “No”
Slide121Experimental conditions and the felicity of test sentences
John-
wa
Icelandic
ka
Swahili-
wo
hanasa-na-i
John-TOP or -ACC speak-neg-pres.
Situation: John cannot speak Icelandic but he can speak Swahili
Experimental context should make the sentence perfectly felicitous under AB (adult) interpretation; otherwise, children’s negative responses may not be counted as evidence for children’s conjunctive interpretation of
ka.
Slide122Felicity conditions for
AB
The speaker knows that something with affirmative expectation turned out to be false.
otherwise
, he wouldn’t use negation.
The speaker knows that it is either A or B (but not both) that is false.
otherwise
, he would say AB.The speaker doesn’t know which one is false.otherwise, he would simply say
A, or B.
Slide123Creating Uncertainty
Two sub-sessions
(1) The “eating-game”
12 animals try to eat 3 kinds of food. Depending of how good they did, they get a particular kind of medal as a prize.
(2) Truth Value Judgment
Kermit guesses how good each animal did on the basis of the medal the animal has.
Slide124Participants
Japanese monolingual children in
Sumire
kindergarden
, Totsuka, Yokohama.
N=30, Age: 3;7-6;3, Mean: 5;3
Slide125Experimenter: Look at this! There are animals going to play an “eating-game”!!
Slide126Experimenter: Here’s a piece of cake, a green pepper, and a carrot. All animals love cakes, but they don’t like vegetables. So here’s the rule: if one eats not only the cake but also the vegetables, he’ll get a better prize.
Slide127Experimenter: For example, if one eats the cake, and the pepper, and also the carrot…then he’ll get a shining gold medal!
Slide128Experimenter: If one eats the cake, and either one of the vegetables, but not both…then he’ll get a blue medal.
Slide129Experimenter: If one eats only the cake, but none of the vegetables, then he’ll get a cross…
Slide130Experimenter: Now, here comes a pig. He will play the game.
Slide131Experimenter: The pig first picked up the cake. Yes, he loves cakes and of course he ate it!
Slide132Experimenter: Then he picked up the pepper. He doesn’t like peppers…but he managed to eat it up!
Slide133Experimenter: Then he picked up the carrot…Oh no, he couldn’t eat the carrot!
Slide134Experimenter: So, the pig ate the cake, and he ate the pepper, but he didn’t eat the carrot. Which prize will he get?
Slide135Experimenter: Yes, a blue medal!
Slide136Experimenter: Now here comes another animal…
(the “eating-game” goes on until all the 12 animals finish their trials. Every animal eats the cake. 4 of them eat both vegetables, other 4 eat either one of them, and other 4 eat neither)
Slide137(After the “game” phase, we move back to the first animal, the pig)
Kermit: Ok, now I’m going to guess how well those animals did with this game. Umm, the pig … I don’t remember what he ate … oh, but, he has a blue medal!
Slide138Kermit: Now I know what happened.
The pig ate the cake, but, he didn’t eat the pepper
ka
the carrot!
(the test sentence)
Slide139Experimenter: Was Kermit correct?
(And the truth-value judgments go on…)
Slide140Felicity of the test sentence
Kermit knows that something with affirmative expectation turned out to be false,
because it is not a gold medal that the pig has
.
Kermit knows that it is either A or B (but not both) that is false,
because it is not a cross that the pig has
.
Kermit doesn’t know which one is false,
because he cannot see which food is left. Adult group (Age 29-32, N=10) accepted the sentence 100% of the time (20/20).
Slide141Result (1): the wide-scope reading of “A
ka B”
“he didn’t eat the carrot
ka
the pepper”
for an animal with a blue medal
The sentence is true under adult Japanese interpretation, but false under the narrow-scope, conjunctive interpretation of
ka. The acceptance rate is 25% (15/60)4 kids were adultlike: 4;11, 5;5, 5;10, 6;2.If we exclude them from the count, then the acceptance rate is 13% (7/52)
Slide142Further support: narrow-scope
ka
“he didn’t eat the carrot
ka
the pepper”
for an animal with a cross
The sentence is true under the narrow-scope, conjunctive interpretation of
ka
. The acceptance rate is 78% (47/60)The result makes a lot of sense given that children accepted the wide-scope ka 25% of the time.
Slide143Result (2): children’s performance on “A
mo
B
mo
”
“he didn’t eat the carrot
mo
the pepper
mo
” He didn’t eat the carrot or the pepperfor an animal with a cross (true under adult interpretation) 95% acceptance (57/60)for an animal with a blue medal (false under adult interpretation)
95% rejection (57/60)
Children did very well with A mo
B mo.
Slide144The ideal control item:
nanika
nani - ka
“something”
what
nani - mo
“anything”
John-wa nanika tabe-nakat-ta
John-TOP something eat-neg-past There is something that John didn’t eat
John-wa nanimo tabe-nakat-ta
John-TOP anything eat-neg-past
John didn’t eat anything
Slide145The control experiment
Subjects: N=30, Age: 3;7-6;3, Mean: 5;4
A ka B
is replaced with
nanika
;
A mo B mo
is replaced with
nanimo
Food: 3 different vegetables, and 4 animals don’t eat anythingget a crossAll the other details are the same with the previous experiment.
Slide146Result (1): the wide-scope reading of
nanika
“he didn’t eat
nanika
”
for an animal with a blue medal
The sentence is true under adult Japanese interpretation, but false under the narrow-scope interpretation of
nanika
. The acceptance rate is 88% (53/60)They can access the wide-scope interpretation!
Slide147Result (2): children’s performance with
nanimo
“he didn’t eat
nanimo
”
He didn’t eat anything
for an animal with a cross (true under adult interpretation) 100% acceptance (60/60)for an animal with a blue medal (false under adult interpretation) 85% rejection (51/60)
Children did fairly good with nanimo.