[PPT] - Long-distance dependencies in continuation grammar Cara Su-Yi Leong PowerPoint Presentation

SLIDE 1

Long-distance dependencies in continuation grammar

Cara Su-Yi Leong & Michael Yoshitaka Erlewine cara@u.nus.edu mitcho@nus.edu.sg National University of Singapore PACLIC 33 Future University Hakodate, September 2019

SLIDE 2

Introduction

Barker & Shan (B&S) develop a Combinatory Categorial Grammar which uses the notion of continuations for semantic scope-taking. One hallmark

f B&S is their explanatory account of crossover efgects (Shan and

Barker 2006). Today: We critically evaluate the B&S framework, based on the behavior

f long-distance dependencies.
Data from quantifjer scope-taking and long-distance dependencies

motivate some refjnements to the B&S theory...

...but these necessary refjnements then result in undoing their

positive predictions for crossover efgects. Quantifjers, pronouns, and gaps all “take scope” in the same way for B&S, but their scope-taking behavior is empirically difgerent. We show that the B&S framework has fundamental diffjculties modeling such behavior. 2

SLIDE 3

Introduction

Barker & Shan (B&S) develop a Combinatory Categorial Grammar which uses the notion of continuations for semantic scope-taking. One hallmark

f B&S is their explanatory account of crossover efgects (Shan and

Barker 2006). Today: We critically evaluate the B&S framework, based on the behavior

f long-distance dependencies.
Data from quantifjer scope-taking and long-distance dependencies

motivate some refjnements to the B&S theory...

...but these necessary refjnements then result in undoing their

positive predictions for crossover efgects. Quantifjers, pronouns, and gaps all “take scope” in the same way for B&S, but their scope-taking behavior is empirically difgerent. We show that the B&S framework has fundamental diffjculties modeling such behavior. 2

SLIDE 4

Introduction

Barker & Shan (B&S) develop a Combinatory Categorial Grammar which uses the notion of continuations for semantic scope-taking. One hallmark

f B&S is their explanatory account of crossover efgects (Shan and

Barker 2006). Today: We critically evaluate the B&S framework, based on the behavior

f long-distance dependencies.
Data from quantifjer scope-taking and long-distance dependencies

motivate some refjnements to the B&S theory...

...but these necessary refjnements then result in undoing their

positive predictions for crossover efgects. Quantifjers, pronouns, and gaps all “take scope” in the same way for B&S, but their scope-taking behavior is empirically difgerent. We show that the B&S framework has fundamental diffjculties modeling such behavior. 2

SLIDE 5

§2 Background: Barker & Shan’s continuation- based grammar

3

SLIDE 6

B&S: CCG with continuations

Barker and Shan (Barker 2002; Shan 2004, 2007; Shan and Barker 2006; Barker and Shan 2006, 2008, 2014) develop a CCG using continuations. We refer to these works collectively as B&S. DP Mary m (DP\ S) / DP likes likes DP John j = S John likes Mary likes j m ← syntactic type ← surface form ← denotation 4

SLIDE 7

B&S: CCG with continuations

Barker and Shan (Barker 2002; Shan 2004, 2007; Shan and Barker 2006; Barker and Shan 2006, 2008, 2014) develop a CCG using continuations. We refer to these works collectively as B&S. DP Mary m (DP\ S) / DP likes likes DP John j = S John likes Mary likes j m ← syntactic type ← surface form ← denotation 4

SLIDE 8

Continuations

Continuations refer to the “computational future of an expression” (Shan and Barker 2006: 95), i.e. the procedures that will later apply to the

expression. B&S use continuation-passing to implement semantic scope.

In addition to common \ and / type constructors for left and right composition, B&S introduce and for continuation-passing. Informally, following B&S (2014: 6):

A B would be a B if we could add an A inside it;
C D would be a C if we could add a surrounding D.

5

SLIDE 9

Continuations

Continuations refer to the “computational future of an expression” (Shan and Barker 2006: 95), i.e. the procedures that will later apply to the

expression. B&S use continuation-passing to implement semantic scope.

In addition to common \ and / type constructors for left and right composition, B&S introduce and for continuation-passing. Informally, following B&S (2014: 6):

A B would be a B if we could add an A inside it;
C D would be a C if we could add a surrounding D.

5

SLIDE 10

Multi-level towers

B&S introduce multi-level towers with the interpretation in (??), where higher levels of the towers represent continuation-passing. (1) C B A expression f[ ] a := C (A B) expression λκ . f (κ (a)) 6

SLIDE 11

Composition of towers

Composing two expressions: (2) C D A left-exp g[ ] x D E A\ B right-exp h[ ] f = C E B left-exp right-exp g(h[ ]) f(x) Notice that adjacent types on the higher levels have to match. 7

SLIDE 12

Composition of towers

Composing two expressions: (2) C D A left-exp g[ ] x D E A\ B right-exp h[ ] f = C E B left-exp right-exp g(h[ ]) f(x) Notice that adjacent types on the higher levels have to match. 7

SLIDE 13

Scope-taking and type-shifters

Scope-taking expressions like quantifjers have two-level denotations.

S S DP someone ∃x . [ ] x (DP\ S) / DP likes likes S S DP everyone ∀y . [ ] y

8

SLIDE 14

Scope-taking and type-shifters

Scope-taking expressions like quantifjers have two-level denotations.

S S DP someone ∃x . [ ] x S S (DP\ S) / DP likes [ ] likes S S DP everyone ∀y . [ ] y

We use lift (??) to match non-scope-taking expressions for composition. 8

SLIDE 15

Scope-taking and type-shifters

Scope-taking expressions like quantifjers have two-level denotations.

S S DP someone ∃x . [ ] x S S (DP\ S) / DP likes [ ] likes S S DP everyone ∀y . [ ] y = S S S someone likes everyone ∃x . ∀y . [ ] likes y x

8

SLIDE 16

Scope-taking and type-shifters

Scope-taking expressions like quantifjers have two-level denotations.

S S DP someone ∃x . [ ] x S S (DP\ S) / DP likes [ ] likes S S DP everyone ∀y . [ ] y = S S S someone likes everyone ∃x . ∀y . [ ] likes y x

↓

= ⇒ S someone likes everyone ∃x . ∀y . likes y x

We then lower (??) the expression at the end. 8

SLIDE 17

Scope-taking and type-shifters

We can also derive inverse scope using multi-level towers.

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / DP likes [ ] [ ] likes S S S S DP everyone ∀y . [ ] [ ] y

We use “internal lift” to raise ∀ to a higher level. 9

SLIDE 18

Scope-taking and type-shifters

We can also derive inverse scope using multi-level towers.

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / DP likes [ ] [ ] likes S S S S DP everyone ∀y . [ ] [ ] y = S S S S S someone likes everyone ∀y . [ ] ∃x . [ ] likes y x

9

SLIDE 19

Scope-taking and type-shifters

We can also derive inverse scope using multi-level towers.

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / DP likes [ ] [ ] likes S S S S DP everyone ∀y . [ ] [ ] y = S S S S S someone likes everyone ∀y . [ ] ∃x . [ ] likes y x

↓↓

= = ⇒ S someone likes everyone ∀y . ∃x . likes y x

9

SLIDE 20

Pronouns

The syntactic category A ▷ B represents a B that contains an unbound pronoun of category A, for example: DP ▷ S S DP he λx . [ ] x Pronouns are represented as inherently multi-level towers, meaning that they are also scope-taking expressions. 10

SLIDE 21

Pronouns

The DP ▷ S type propagates to the left, denoting an open pronoun exists in the expression until it is bound (hypothetically). DP ▷ S DP ▷ S DP John [ ] j DP ▷ S DP ▷ S (DP\ S) / S said [ ] said DP ▷ S S DP he λx . [ ] x S S DP / S cried [ ] cried We apply bind (??) to John for it to bind the pronoun to its right. 11

SLIDE 22

Pronouns

The DP ▷ S type propagates to the left, denoting an open pronoun exists in the expression until it is bound (hypothetically). S DP ▷ S DP John [ ] j j DP ▷ S DP ▷ S (DP\ S) / S said [ ] said DP ▷ S S DP he λx . [ ] x S S DP / S cried [ ] cried We apply bind (??) to John for it to bind the pronoun to its right. 11

SLIDE 23

Movement

Continuation-passing provides an in-situ account of movement dependencies using gaps. Gaps introduce a variable and λ binder like pronouns: DP S S DP λx . [ ] x Similarly, DP S propagates to the left to get bound. 12

SLIDE 24

Crossover

One advertised feature of B&S’s proposal is its explanation for crossover efgects (Postal 1971) using linear evaluation. (??) a. Which girli did John introduce to heri second cousin?

b. ??Which girli did John introduce heri second cousin to

? (??)

a. ✓whi ...

i ... proi

b. *whi ... proi ...

i

13

SLIDE 25

Crossover

One advertised feature of B&S’s proposal is its explanation for crossover efgects (Postal 1971) using linear evaluation. (??) a. Which girli did John introduce to heri second cousin?

b. ??Which girli did John introduce heri second cousin to

? (??)

a. ✓whi ...

i ... proi

b. *whi ... proi ...

i

13

SLIDE 26

Explaining crossover efgects

For B&S, in grammatical wh-pro-binding confjgurations (??/??a), the gap binds the pronoun to its right.

S/(DP S) which girl λκ . wh(girl y ∧ κ(y)) ... DP S S DP λx . [ ] x ... DP ▷ S S PP to her second cousin λc . [ ] second cousin c

bind (??) cannot apply to the fronted wh-phrase itself. Therefore, the binding is grammatical in (??a), but in (??b) there is no preceding gap that can bind the pronoun, leading to ungrammaticality. This explains the crossover asymmetry. 14

SLIDE 27

Explaining crossover efgects

For B&S, in grammatical wh-pro-binding confjgurations (??/??a), the gap binds the pronoun to its right.

S/(DP S) which girl λκ . wh(girl y ∧ κ(y)) ... DP S DP ▷ S DP λx . [ ] x x ... DP ▷ S S PP to her second cousin λc . [ ] second cousin c

bind (??) cannot apply to the fronted wh-phrase itself. Therefore, the binding is grammatical in (??a), but in (??b) there is no preceding gap that can bind the pronoun, leading to ungrammaticality. This explains the crossover asymmetry. 14

SLIDE 28

Explaining crossover efgects

For B&S, in grammatical wh-pro-binding confjgurations (??/??a), the gap binds the pronoun to its right.

S/(DP S) which girl λκ . wh(girl y ∧ κ(y)) ... DP S DP ▷ S DP λx . [ ] x x ... DP ▷ S S PP to her second cousin λc . [ ] second cousin c

bind (??) cannot apply to the fronted wh-phrase itself. Therefore, the binding is grammatical in (??a), but in (??b) there is no preceding gap that can bind the pronoun, leading to ungrammaticality. This explains the crossover asymmetry. 14

SLIDE 29

Explaining crossover efgects

For B&S, in grammatical wh-pro-binding confjgurations (??/??a), the gap binds the pronoun to its right.

S/(DP S) which girl λκ . wh(girl y ∧ κ(y)) ... DP S DP ▷ S DP λx . [ ] x x ... DP ▷ S S PP to her second cousin λc . [ ] second cousin c

bind (??) cannot apply to the fronted wh-phrase itself. Therefore, the binding is grammatical in (??a), but in (??b) there is no preceding gap that can bind the pronoun, leading to ungrammaticality. This explains the crossover asymmetry. 14

SLIDE 30

Preview

§3 B&S does not restrict quantifjer scope-taking... We formalize a suggested fjx

A . A complicates the binding of embedded gaps and pronouns...

We propose minimal modifjcations

B so that we can model

such examples together with

A .

§4

A and B together lead to incorrect predictions for crossover,

undoing a key advantage of the B&S framework. §5 Discussion 15

SLIDE 31

Preview

§3 B&S does not restrict quantifjer scope-taking... We formalize a suggested fjx

A . A complicates the binding of embedded gaps and pronouns...

We propose minimal modifjcations

B so that we can model

such examples together with

A .

§4

A and B together lead to incorrect predictions for crossover,

undoing a key advantage of the B&S framework. §5 Discussion 15

SLIDE 32

Preview

§3 B&S does not restrict quantifjer scope-taking... We formalize a suggested fjx

A . A complicates the binding of embedded gaps and pronouns...

We propose minimal modifjcations

B so that we can model

such examples together with

A .

§4

A and B together lead to incorrect predictions for crossover,

undoing a key advantage of the B&S framework. §5 Discussion 15

SLIDE 33

Preview

§3 B&S does not restrict quantifjer scope-taking... We formalize a suggested fjx

A . A complicates the binding of embedded gaps and pronouns...

We propose minimal modifjcations

B so that we can model

such examples together with

A .

§4

A and B together lead to incorrect predictions for crossover,

undoing a key advantage of the B&S framework. §5 Discussion 15

SLIDE 34

§3 Scope-taking across clause boundaries

16

SLIDE 35

Scope

(??) # Someone said [everyone is married to Sue].

✓∃ > ∀, *∀ > ∃

17

SLIDE 36

B&S overgenerates

B&S overgenerates the unattested inverse-scope reading of Someone said everyone is married to Sue (??):

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S S S S DP everyone ∀y . [ ] [ ] y S S S S DP\ S is married to Sue [ ] [ ] married to s = S S S S S

sm. said ev. is married to Sue

∀y . [ ] ∃x . [ ] said (married to s y) x

↓↓

= = ⇒ S

sm. said ev. is married to Sue

∀y . ∃x . said (married to s y) x

18

SLIDE 37

B&S overgenerates

B&S overgenerates the unattested inverse-scope reading of Someone said everyone is married to Sue (??):

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S S S S DP everyone ∀y . [ ] [ ] y S S S S DP\ S is married to Sue [ ] [ ] married to s = S S S S S

sm. said ev. is married to Sue

∀y . [ ] ∃x . [ ] said (married to s y) x

↓↓

= = ⇒ S

sm. said ev. is married to Sue

∀y . ∃x . said (married to s y) x

18

SLIDE 38

B&S overgenerates

B&S overgenerates the unattested inverse-scope reading of Someone said everyone is married to Sue (??):

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S S S S DP everyone ∀y . [ ] [ ] y S S S S DP\ S is married to Sue [ ] [ ] married to s = S S S S S

sm. said ev. is married to Sue

∀y . [ ] ∃x . [ ] said (married to s y) x

↓↓

= = ⇒ S

sm. said ev. is married to Sue

∀y . ∃x . said (married to s y) x

18

SLIDE 39

Scope Island Evaluation

B&S note this problem. Charlow (2014: 65) suggests that scope islands must be evaluated by “collapsing it into a single level.” We codify this requirement as follows: (??) Scope Island Evaluation If the expression is a scope island, apply lower as many times as possible (↓*). Embedded tensed clauses are scope islands. 19

SLIDE 40

Scope Island Evaluation

B&S note this problem. Charlow (2014: 65) suggests that scope islands must be evaluated by “collapsing it into a single level.” We codify this requirement as follows: (??) Scope Island Evaluation If the expression is a scope island, apply lower as many times as possible (↓*). Embedded tensed clauses are scope islands. 19

SLIDE 41

Scope Island Evaluation

B&S note this problem. Charlow (2014: 65) suggests that scope islands must be evaluated by “collapsing it into a single level.” We codify this requirement as follows: (??) Scope Island Evaluation If the expression is a scope island, apply lower as many times as possible (↓*). Embedded tensed clauses are scope islands. 19

SLIDE 42

Scope Island Evaluation

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S S S S DP everyone ∀y . [ ] [ ] y S S S S DP\ S is married to Sue [ ] [ ] married to s 20

SLIDE 43

Scope Island Evaluation

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S S S S S everyone is married to Sue ∀y . [ ] [ ] married to s y 20

SLIDE 44

Scope Island Evaluation

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S everyone is married to Sue ∀y . married to s y 20

SLIDE 45

Scope Island Evaluation

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S S S S S everyone is married to Sue [ ] [ ] ∀y . married to s y 20

SLIDE 46

Scope Island Evaluation

S S S S DP someone [ ] ∃x . [ ] x S S S S (DP\ S) / S said [ ] [ ] said S S S S S everyone is married to Sue [ ] [ ] ∀y . married to s y

↓↓

= = ⇒ S

sm. said ev. is married to Sue

∃x . ∀y . said (married to s y) x 20

SLIDE 47

Problems with embedded gaps and pronouns

Recall that pronouns and gaps are “scope-taking” in B&S: they posit a λ binder on a higher level, to be bound from the left. (??) a.

✓ Which girli did you say [Mary saw i]?

b.

✓ Every girli said [Mary saw heri].

Scope Island Evaluation blocks the binding of gaps (10a) and pronouns (10b) in embedded tensed clauses, contrary to fact. 21

SLIDE 48

Problems with embedded gaps and pronouns

Recall that pronouns and gaps are “scope-taking” in B&S: they posit a λ binder on a higher level, to be bound from the left. (??) a.

✓ Which girli did you say [Mary saw i]?

b.

✓ Every girli said [Mary saw heri].

Scope Island Evaluation blocks the binding of gaps (10a) and pronouns (10b) in embedded tensed clauses, contrary to fact. 21

SLIDE 49

Problems with embedded gaps and pronouns

Recall that pronouns and gaps are “scope-taking” in B&S: they posit a λ binder on a higher level, to be bound from the left. (??) a.

✓ Which girli did you say [Mary saw i]?

b.

✓ Every girli said [Mary saw heri].

Scope Island Evaluation blocks the binding of gaps (10a) and pronouns (10b) in embedded tensed clauses, contrary to fact. 21

SLIDE 50

Embedded gaps

Clauses with DP gaps that undergo Scope Island Evaluation will be of category DP\ S, the category of a clause missing a DP to its left. DP S DP S DP Mary [ ] m DP S DP S (DP\ S) / DP saw [ ] saw DP S S DP λx . [ ] x = DP S S S Mary saw λx . [ ] saw x m

The syntactic category of the expression cannot compose as a S.
The λ binder for the gap is on the lowest level and ceases to

propagate as a scope-taking expression. 22

SLIDE 51

Embedded gaps

Clauses with DP gaps that undergo Scope Island Evaluation will be of category DP\ S, the category of a clause missing a DP to its left. DP S DP S DP Mary [ ] m DP S DP S (DP\ S) / DP saw [ ] saw DP S S DP λx . [ ] x = DP\ S Mary saw λx . saw x m

The syntactic category of the expression cannot compose as a S.
The λ binder for the gap is on the lowest level and ceases to

propagate as a scope-taking expression. 22

SLIDE 52

Embedded gaps: Intermediate gaps

We resolve this problem by positing intermediate gaps at the left edge of gapped clauses after they undergo Scope Island Evaluation. (??) DP S S DP λy . [ ] y DP\ S Mary saw λx . saw x m = DP S S S Mary saw λy . [ ] saw y m Adding the intermediate gap results in the same meaning as the embedded clause prior to Scope Island Evaluation. 23

SLIDE 53

Embedded gaps: Intermediate gaps

We resolve this problem by positing intermediate gaps at the left edge of gapped clauses after they undergo Scope Island Evaluation. (??) DP S S DP λy . [ ] y DP\ S Mary saw λx . saw x m = DP S S S Mary saw λy . [ ] saw y m Adding the intermediate gap results in the same meaning as the embedded clause prior to Scope Island Evaluation. 23

SLIDE 54

Embedded gaps: Intermediate gaps

We resolve this problem by positing intermediate gaps at the left edge of gapped clauses after they undergo Scope Island Evaluation. (??) DP S S DP λy . [ ] y S S DP\ S Mary saw [ ] λx . saw x m = DP S S S Mary saw λy . [ ] saw y m Adding the intermediate gap results in the same meaning as the embedded clause prior to Scope Island Evaluation. 23

SLIDE 55

Embedded gaps: Intermediate gaps

We resolve this problem by positing intermediate gaps at the left edge of gapped clauses after they undergo Scope Island Evaluation. (??) DP S S DP λy . [ ] y S S DP\ S Mary saw [ ] λx . saw x m = DP S S S Mary saw λy . [ ] saw y m Adding the intermediate gap results in the same meaning as the embedded clause prior to Scope Island Evaluation. 23

SLIDE 56

Embedded gaps: Intermediate gaps

We resolve this problem by positing intermediate gaps at the left edge of gapped clauses after they undergo Scope Island Evaluation. (??) DP S S DP λy . [ ] y S S DP\ S Mary saw [ ] λx . saw x m = DP S S S Mary saw λy . [ ] saw y m Adding the intermediate gap results in the same meaning as the embedded clause prior to Scope Island Evaluation. 23

SLIDE 57

Embedded pronouns

We encounter a similar problem with embedded bound pronouns: DP ▷ S DP ▷ S DP Mary [ ] m DP ▷ S DP ▷ S (DP\ S) / DP saw [ ] saw DP ▷ S S DP her λx . [ ] x = DP ▷ S S S Mary saw her λx . [ ] saw x m

The resulting expression after Scope Island Evaluation is of category

DP ▷ S and cannot combine with an S-selecting verb.

An additional problem: no existing expressions combine with

expressions of category DP ▷ S on the lowest level. 24

SLIDE 58

Embedded pronouns

We encounter a similar problem with embedded bound pronouns: DP ▷ S DP ▷ S DP Mary [ ] m DP ▷ S DP ▷ S (DP\ S) / DP saw [ ] saw DP ▷ S S DP her λx . [ ] x = DP ▷ S Mary saw her λx . saw x m

The resulting expression after Scope Island Evaluation is of category

DP ▷ S and cannot combine with an S-selecting verb.

An additional problem: no existing expressions combine with

expressions of category DP ▷ S on the lowest level. 24

SLIDE 59

Embedded pronouns: prolift

We propose a new type-shifter: (??) B C DP ▷ A expression f [ ] λx . g(x)

prolift

= = = = = = ⇒ DP ▷ B C A expression λx . f [ ] g(x) prolift returns the pronoun’s λ binder and corresponding DP▷-category to a higher level from which it can propagate leftward. 25

SLIDE 60

Embedded pronouns: prolift

We propose a new type-shifter: (??) B C DP ▷ A expression f [ ] λx . g(x)

prolift

= = = = = = ⇒ DP ▷ B C A expression λx . f [ ] g(x) prolift returns the pronoun’s λ binder and corresponding DP▷-category to a higher level from which it can propagate leftward. DP ▷ S Mary saw her λx . saw x m 25

SLIDE 61

Embedded pronouns: prolift

We propose a new type-shifter: (??) B C DP ▷ A expression f [ ] λx . g(x)

prolift

= = = = = = ⇒ DP ▷ B C A expression λx . f [ ] g(x) prolift returns the pronoun’s λ binder and corresponding DP▷-category to a higher level from which it can propagate leftward. S S DP ▷ S Mary saw her [ ] λx . saw x m 25

SLIDE 62

Embedded pronouns: prolift

We propose a new type-shifter: (??) B C DP ▷ A expression f [ ] λx . g(x)

prolift

= = = = = = ⇒ DP ▷ B C A expression λx . f [ ] g(x) prolift returns the pronoun’s λ binder and corresponding DP▷-category to a higher level from which it can propagate leftward. DP ▷ S S S Mary saw her λx . [ ] saw x m 25

SLIDE 63

Summary

To accurately model restrictions on quantifjer scope-taking, we

codifjed a suggestion by Charlow (2014) as Scope Island Evaluation.

To model the grammatical binding of embedded gaps and pronouns

and maintain Scope Island Evaluation, we proposed intermediate gaps and prolift. 26

SLIDE 64

Summary

To accurately model restrictions on quantifjer scope-taking, we

codifjed a suggestion by Charlow (2014) as Scope Island Evaluation.

To model the grammatical binding of embedded gaps and pronouns

and maintain Scope Island Evaluation, we proposed intermediate gaps and prolift. 26

SLIDE 65

§4 Crossover in long-distance confjgurations

27

SLIDE 66

Consequences for crossover

These three amendments to the theory — Scope Island Evaluation, intermediate gaps, and prolift, all necessary to account for the behavior of quantifjer scope-taking as well as long-distance dependencies — together lead to incorrect predictions for crossover efgects. 28

SLIDE 67

Crossover with embedded clauses

Recall that the derivation and explanation of crossover asymmetries (??) were claimed to be an advantage of the B&S framework. (??)

a. ✓whi ...

i ... proi

b. *whi ... proi ...

i

The revised B&S framework predicts the crossover violation in (??b) to be grammatical, contrary to fact. (??) a. Which girli do you think [ loves heri mother]? b.

?? Which girli do you think [heri mother loves

]? For grammatical embedded gap binding, we hypothesize an intermediate

gap. This gap precedes the pronoun and can bind it!

29

SLIDE 68

Crossover with embedded clauses

Recall that the derivation and explanation of crossover asymmetries (??) were claimed to be an advantage of the B&S framework. (??)

a. ✓whi ...

i ... proi

b. *whi ... proi ...

i

The revised B&S framework predicts the crossover violation in (??b) to be grammatical, contrary to fact. (??) a. Which girli do you think [ loves heri mother]? b.

?? Which girli do you think [heri mother loves

]? For grammatical embedded gap binding, we hypothesize an intermediate

gap. This gap precedes the pronoun and can bind it!

29

SLIDE 69

Crossover with embedded clauses

Recall that the derivation and explanation of crossover asymmetries (??) were claimed to be an advantage of the B&S framework. (??)

a. ✓whi ...

i ... proi

b. *whi ... proi ...

i

The revised B&S framework predicts the crossover violation in (??b) to be grammatical, contrary to fact. (??) a. Which girli do you think [ loves heri mother]? b.

?? Which girli do you think i [heri mother loves

]? For grammatical embedded gap binding, we hypothesize an intermediate

gap. This gap precedes the pronoun and can bind it!

29

SLIDE 70

Crossover with embedded clauses

Recall that the derivation and explanation of crossover asymmetries (??) were claimed to be an advantage of the B&S framework. (??)

a. ✓whi ...

i ... proi

b. *whi ... proi ...

i

The revised B&S framework predicts the crossover violation in (??b) to be grammatical, contrary to fact. (??) a. Which girli do you think [ loves heri mother]? b.

?? Which girli do you think i [heri mother loves

]? For grammatical embedded gap binding, we hypothesize an intermediate

gap. This gap precedes the pronoun and can bind it!

29

SLIDE 71

??Which girli do you think [heri mother loves

]?

DP ▷ (DP S) DP S DP her mother λd . [ ] mother d DP S DP S (DP\ S) / DP loves [ ] loves DP S S DP λx . [ ] x = DP ▷ (DP S) S S her mother loves λd . λx . [ ] loves x (mother d)

↓∗

= = ⇒ DP ▷ (DP S) her mother loves λd . λx . loves x (mother d)

lift, prolift

= = = = = = = = = = = ⇒

30

SLIDE 72

??Which girli do you think [heri mother loves

]?

DP ▷ (DP S) DP S DP her mother λd . [ ] mother d DP S DP S (DP\ S) / DP loves [ ] loves DP S S DP λx . [ ] x = DP ▷ (DP S) S S her mother loves λd . λx . [ ] loves x (mother d)

↓∗

= = ⇒ DP ▷ (DP S) her mother loves λd . λx . loves x (mother d)

lift, prolift

= = = = = = = = = = = ⇒

30

SLIDE 73

??Which girli do you think [heri mother loves

]?

DP S DP S DP you [ ] you DP S DP S (DP\ S) / S think [ ] think DP S DP ▷ S DP λy . [ ] y y DP ▷ S S DP\ S her mother loves λd . [ ] λx . loves x (mother d)

=

DP S S S you think her mother loves λy . [ ] loves y (mother y)

↓∗

= = ⇒ DP S you think her mother loves λy . loves y (mother y)

31

SLIDE 74

??Which girli do you think [heri mother loves

]?

DP S DP S DP you [ ] you DP S DP S (DP\ S) / S think [ ] think DP S DP ▷ S DP λy . [ ] y y DP ▷ S S DP\ S her mother loves λd . [ ] λx . loves x (mother d)

=

DP S S S you think her mother loves λy . [ ] loves y (mother y)

↓∗

= = ⇒ DP S you think her mother loves λy . loves y (mother y)

31

SLIDE 75

??Which girli do you think [heri mother loves

]?

S/(DP S) which girl λκ . wh(λg . girl g ∧ κ (g)) DP S you think her mother loves λy . loves y (mother y) = S which girl do you think her mother loves wh(λg . girl g ∧ loves g (mother g)) 32

SLIDE 76

Summary

Scope Island Evaluation, intermediate gaps, and prolift — all necessary to model both quantifjer scope-taking and long-distance binding — together overgenerates long-distance crossover confjgurations such as (??b) as acceptable. 33

SLIDE 77

§5 Discussion

34

SLIDE 78

Summary

B&S develop a CCG where continuations are passed linearly to model scope-taking and binding. A claimed advantage of this framework (Shan and Barker 2006) is its explanation for crossover efgects (Postal 1971). 35

SLIDE 79

Summary

B&S develop a CCG where continuations are passed linearly to model scope-taking and binding. A claimed advantage of this framework (Shan and Barker 2006) is its explanation for crossover efgects (Postal 1971). 35

SLIDE 80

Summary

Today, we discussed the behavior of quantifjers, pronouns, and gaps in embedded clauses, which have not been seriously discussed previously in this literature.

Limitations on quantifjer scope-taking motivate Scope Island

Evaluation, a requirement that all scope islands (including embedded tensed clauses) be fully lower-ed.

The availability of long-distance movement and binding

dependencies motivated further refjnements to the theory.

The revised B&S theory correctly accounts for limitations on

quantifjer scope, while allowing for long-distance movement and binding, but makes incorrect predictions for crossover efgects. 36

SLIDE 81

Summary

Today, we discussed the behavior of quantifjers, pronouns, and gaps in embedded clauses, which have not been seriously discussed previously in this literature.

Limitations on quantifjer scope-taking motivate Scope Island

Evaluation, a requirement that all scope islands (including embedded tensed clauses) be fully lower-ed.

The availability of long-distance movement and binding

dependencies motivated further refjnements to the theory.

The revised B&S theory correctly accounts for limitations on

quantifjer scope, while allowing for long-distance movement and binding, but makes incorrect predictions for crossover efgects. 36

SLIDE 82

Summary

Today, we discussed the behavior of quantifjers, pronouns, and gaps in embedded clauses, which have not been seriously discussed previously in this literature.

Limitations on quantifjer scope-taking motivate Scope Island

Evaluation, a requirement that all scope islands (including embedded tensed clauses) be fully lower-ed.

The availability of long-distance movement and binding

dependencies motivated further refjnements to the theory.

The revised B&S theory correctly accounts for limitations on

quantifjer scope, while allowing for long-distance movement and binding, but makes incorrect predictions for crossover efgects. 36

SLIDE 83

Choose two:

37

SLIDE 84

Choose two:

37

SLIDE 85

The problem

At issue is B&S’s uniform treatment of quantifjer scope-taking, pronominal binding, and fjller-gap (movement) dependencies. But these dependencies are sensitive to difgerent locality restrictions:

Quantifjers generally resist scoping out of tensed clauses, although

there is some speaker variation (Wurmbrand 2018).

Movement dependencies can cross tensed clauses, but are sensitive

to syntactic islands (Ross 1967).

Pronominal binding is insensitive to both tensed clause boundaries

and syntactic islands. Our demonstration here challenges a unifjed approach to these phenomena, in turn challenging the B&S program itself. 38

SLIDE 86

The problem

At issue is B&S’s uniform treatment of quantifjer scope-taking, pronominal binding, and fjller-gap (movement) dependencies. But these dependencies are sensitive to difgerent locality restrictions:

Quantifjers generally resist scoping out of tensed clauses, although

there is some speaker variation (Wurmbrand 2018).

Movement dependencies can cross tensed clauses, but are sensitive

to syntactic islands (Ross 1967).

Pronominal binding is insensitive to both tensed clause boundaries

and syntactic islands. Our demonstration here challenges a unifjed approach to these phenomena, in turn challenging the B&S program itself. 38

SLIDE 87

The problem

At issue is B&S’s uniform treatment of quantifjer scope-taking, pronominal binding, and fjller-gap (movement) dependencies. But these dependencies are sensitive to difgerent locality restrictions:

Quantifjers generally resist scoping out of tensed clauses, although

there is some speaker variation (Wurmbrand 2018).

Movement dependencies can cross tensed clauses, but are sensitive

to syntactic islands (Ross 1967).

Pronominal binding is insensitive to both tensed clause boundaries

and syntactic islands. Our demonstration here challenges a unifjed approach to these phenomena, in turn challenging the B&S program itself. 38

SLIDE 88

Thank you!

We thank Chris Barker for helpful correspondence and encouraging discussion and Kenyon Branan for comments on this presentation.

Questions?

39

SLIDE 89

References I

Barker, Chris. 2002. Continuations and the Nature of Quantifjcation. Natural Language Semantics 10:211–242. Barker, Chris, and Chung-Chieh Shan. 2006. Types as Graphs: Continuations in Type Logical Grammar. Journal of Logic, Language and Information 15:331–370. Barker, Chris, and Chung-Chieh Shan. 2008. Donkey anaphora is in-scope

binding. Semantics and Pragmatics 1:1–46.

Barker, Chris, and Chung-Chieh Shan. 2014. Continuations and Natural

Language. Oxford University Press.

Charlow, Simon. 2014. On the semantics of exceptional scope. Doctoral Dissertation, New York University. Postal, Paul M. 1971. Cross-over Phenomena. Holt, Rinehart and Winston. Ross, John Robert. 1967. Constraints on variables in syntax. Doctoral Dissertation, Massachusetts Institute of Technology, Cambridge, MA.

40

SLIDE 90

References II

Shan, Chung-Chieh. 2004. Delimited continuations in natural language. In Proceedings of the Fourth ACM SIGPLAN Continuations Workshop. Shan, Chung-Chieh. 2007. Linguistic side efgects. In Direct compositionality,

ed. Chris Barker and Pauline Jacobson, 132–163. Oxford: Oxford University

Press. Shan, Chung-Chieh, and Chris Barker. 2006. Explaining Crossover and Superiority as Left-to-right Evaluation. Linguistics and Philosophy 29:91–134. Wurmbrand, Susanne. 2018. The cost of raising quantifjers. Glossa 3:1–40.

41