Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Focus intervention effects and quantificationaldomains of focus-sensitive operators
Haoze Li 1 Hoi-Ki Jess Law 2
1The Chinese University of Hong Kong
2Rutgers University
GLOW 37April 2, 2014
1 / 46
A pretheoretical look
(1) German (Mayr 2013: 5; see also Beck 1996)
a. * Wenwho
hathas
nuronly
derthe
HansFHansF
wannwhen
angerufen?called
b. Wenwho
hathas
wannwhen
nuronly
derthe
HansFHansF
angerufen?called
’Who did only Hans call when?’
(2) Mandarin (Yang 2008: 69)
a. ?? Zhiyouonly
ZhangsanF
ZhangsanF
meinot
chieat
nawhich
daoCl
cai?dish
b. Nawhich
daoCl
caidish
zhiyouonly
ZhangsanF
ZhangsanF
meinot
chi?eat
’Which dish did only Zhangsan not eat?’
2 / 46
Roadmap
I The intervention hypothesisI Minimality (Beck 2006)I Non-additivity (Mayr 2013)
I The quantificational-domain hypothesis (our view)I Critical data in support of our view
I F-WH associationI Multiple-WH questionsI Alternative questions
I Focus interverntion beyond questionsI Contrastive topics
3 / 46
The intervention hypothesis
Some focus operator interferes with the interpretation ofwh-questions (Pesetsky 2000; Beck 2006; Cable 2012; Mayr 2013;cf. Tomioka 2007; Haida 2008)
(3) J[Q...Focus − sensitive operator ...WH...]Kg
= undefined⇒ Wh-questions cannot recieve a proper interpretation
4 / 46
Minimality (Beck 2006)
(4) ?*[Q ... [ ∼ [ ... WH ...]]]
I JWHKf = {a, b, c} (Hamblin denotation)
I JWHKg = undefined
I The role of Q is to elevate the focus semantic value of awh-containing constituent to the ordinary semantic value
I ∼ interferes with the association between Q and thewh-containing constituent
5 / 46
Non-additivity (Mayr 2013)
(5) ?*[Q ... [ non-additive operator [ ... WH ...]]]
I Additive operator Op:for any g, h, Op (g ∨ h) = Op (g) ∨ Op (h)
I an operator is a problematic intervener forwh-questions if it is non-additive
(6) Only John smokes or drinks. 6=
(7) Only John smokes or only John drinks.
I only is non-additive, hence a problematic intervener.
6 / 46
The quantificational-domain hypothesis
A warm-up: set membership relation
(8) α1 ∈ {α1, α2, α3}a set of α
(9) α1 /∈ { {α1, α2, α3} , {α4, α5, α6} }a set of sets of α
This very simple set membership relation is what lies in focusintervention effects.
7 / 46
Focus intervention is caused by the inappropriate quantificationaldomain of a focus-sensitive operator.
(10) ?*[ Q ... focus-sensitive operator [ XPF ... WH ...]]
(11)
operator
only/also
domain
a set of sets of α
scope
α
α does not belong to the quantificational domain
⇒ The quantificational structure induced by a focus-sensitiveoperator is illicit
8 / 46
Preliminary I: Focus semantics
Association with focus (Rooth 1985; Kratzer 1991)
(12) IP
He VP2
only VP1
met JohnF1
a. JJohnF 1Kg = John; JJohnF 1Kg ,h = h(1)
b. JVP1Kg = λy. y met John
c. JVP1Kg ,h = λy. y met h(1)
d. JVP1Kf = {λy. y met h(1) | h∈H}= {λy. y met John, λy. y met Peter, ...}
9 / 46
Only takes as its quantificational domain the focus semantic valueof VP1 ⇒ Association with focus
(13) Jonly VP1Kg
= JonlyKg (JVP1Kf ) (JVP1Kg )= λy.∀P∈ JVP1Kf [P(y)=1→P(y)=JVP1Kg (y)]
Notice P<e,t> ∈ JVP1Kf <<e,t>,t>
(14)
operator
only
domain
a set of properties
scope
property
10 / 46
Preliminary II: neo-Hamblin semantics
Semantics of wh-phrase
I Ordinary semantic value: a set of alternatives(Hamblin 1973)
I Focus semantic value: none(see also Eckardt 2007; contra Beck 2006)
Under Kratzer (1991)’s framework of focus interpretation
I JWHKg = {a, b, c} (ordinary semantic value)
I JWHKg ,h = {a, b, c} (secondary semantic value)
I no focus semantic value
11 / 46
(15) CP
Q IP
Libai VP
met who
a. JwhoKg = {John, Peter, ...}b. JmetKg = λx.λy. y met x
c. JVPKg = {JmetKg (x) | x ∈ JwhoKg}= {λy. y met John, λy. y met Peter, ...}(pointwise functional application, Yatsushiro 2009, seealso Hagstrom 1998)
12 / 46
(16) a. JIPKg = JLibai met whoKg
= {Libai met John, Libai met Peter, ...}b. JCPKg = JQ IPKg = JIPKg
(Kratzer and Shimoyama 2002)
13 / 46
Derivation of focus intervention effects
Focus intervention effects in Mandarin
(17) ?* Tahe
zhionly
rangallow
LibaiFLibaiF
jian-lemeet-Asp
shei?who
(see 20)
Intended ’Who was the person x such that he onlyallowed LibaiF to meet x?’
(18) ?* Zhiyouonly
LibaiFLibaiF
rangallow
tahim
jian-lemeet-Asp
shei?who
Intended ’Who was the person x such that only LibaiFallowed him to meet x?’
(19) ?* Tahe
haialso
rangallow
DufuF
DufuF
jian-lemeet-Asp
shei?who
Intended ’Who was the person x such that he alsoallowed DufuF to meet x?’
14 / 46
(20) ?*CP
Q IP
he VP2
only VP1
allow LibaiF1to meet who
a. JLibaiF 1Kg = Libai; JLibaiF 1Kg ,h = h(1)
b. JwhoKg = JwhoKg ,h = {John, Peter, ...}
15 / 46
(21) a. JVP1Kg = {λy. y allow Libai to meet x | x ∈ JwhoKg}
=
λy. y allow Libai to meet Johnλy. y allow Libai to meet Peter
...
b. JVP1Kg ,h = {λy. y allow h(1) to meet x | x∈JwhoKg ,h}
=
λy. y allow h(1) to meet Johnλy. y aloow h(1) to meet Peter
...
c. JVP1Kf = {JVP1Kg ,h | h∈H }
=
λy. y allow h(1) to meet Johnλy. y allow h(1) to meet Peter
...
| h∈H
⇒ a set of sets of alternatives
16 / 46
d. JVP1Kf = (a set of sets of alternatives)
λy. y allow Libai to meet Johnλy. y allow Libai to meet Peter
...
λy. y allow Dufu to meet Johnλy. y allow Dufu to meet Peter
...
...
17 / 46
Only takes as its quantificational domain the focus semantic valueof VP1:
(22) a. JonlyKg = λD.λF.λy. ∀P∈D [P(y)=1 → P(y)=F(y)]
b. Jonly VP1Kg = JonlyKg (JVP1Kf ) (JVP1Kg )
= JonlyKg (JVP1Kf ) (Jallow Libai to meet whoKg )
= JonlyKg (JVP1Kf )
λy. y allow Libai to meet Johnλy. y allow Libai to meet Peter
...
=
JonlyKg (JVP1Kf ) (λy. y allow Libai to meet John)JonlyKg (JVP1Kf ) (λy. y allow Libai to meet Peter)
...
18 / 46
Jonly VP1Kg = JonlyKg (JVP1Kf ) (JVP1Kg ) =λy.∀P∈ JVP1Kf [P(y)=1→P(y)=y allow Libai to meet John]λy.∀P∈ JVP1Kf [P(y)=1→P(y)=y allow Libai to meet Peter]
...
However, P <e ,t> /∈ JVP1Kf <<<e ,t>t>t>
⇒ Illicit quantification⇒ Focus intervention effects
19 / 46
Derivation of overt wh-fronting
(23) Shei1,who
tahe
zhionly
rangallow
LibaiFLibaiF
jian-lemeet-Asp
t1? (see 25)
Intended ’Who was the person x such that he onlyallowed LibaiF to meet x?’
(24) Na dao cai1,which-CL dish
zhiyouonly
ZhangsanF
ZhangsanF
meinot
chieat
t1?
Intended ’Which dish did only Zhangsan not eat?’
20 / 46
(25) CP
Q IP2
Who
λ2 IP1
he VP2
only VP1
allow LibaiF1to meet t2
a. JLibaiF 1Kg = Libai; JLibaiF 1Kg ,h = h(1)
b. Jt2Kg = Jt2Kg ,h = g(2)
c. JwhoKg = JwhoKg ,h = {John, Peter, ...}
21 / 46
(26) a. JVP1Kg = λy. y allow Libai to meet g(2)
b. JVP1Kg ,h = λy. y allow h(1) to meet g(2)
c. JVP1Kf = {JVP1Kg ,h | h∈H }
=
λy. y allow Libai to meet g(2)λy. y allow Dufu to meet g(2)
...
⇒ a set of alternatives
d. Jonly VP1Kg = JonlyKg (JVP1Kf ) (JVP1Kg ) =λy.∀P∈ JVP1Kf [P(y)=1→P(y)=y allow Libai to meetg(2)]Here, P<e,t> ∈ JVP1Kf<<e,t>,t>
⇒ Licit quantification
22 / 46
(27) a. JIP1Kg
= ∀P∈ JVP1Kf [P(he)=1→P(he)=he allow Libai tomeet g(2)] = he only allow Libai to meet g(2)
b. λ2.JIP1Kg [x/2] = λx. he only allow Libai to meet x
c. JIP2Kg = {he only allow Libai to meet x | x ∈ JwhoKg}
=
he only allow Libai to meet Johnhe only allow Libai to meet Peter
...
d. JCPKg = JIP2Kg
23 / 46
Interim summary
Configuration of focus intervention effects in wh-questions
(28) ?*[ ... focus-sensitive operator [ XPF ... WH ...]]
I XPF is as crucial as the focus-sensitive operator and the WH!
The quantificational-domain hypothesis of focus intervention
(29) Focus intervention effects arise iff what a focus-sensitiveoperator quantifies over is not a member of itsquantificational domain.
24 / 46
Critical data
When focus-sensitive operators do not interfere with questions
I F-WH association
I F-Alt association
Focus intervention in non-questions
I Contrastive topics
25 / 46
Association between focus-sensitive operators andwh-phrases (F-WH association)
Mandarin
(30) LibaiLibai
zhionly
jian-lemeet-Asp
shei?who
’Who was the person x such that Libai met x and nobodyelse?’
(31) Zhiyouonly
sheiwho
jian-lemeet-Asp
Libai?Libai
’Who was the person x such that x and nobody else metLibai?’
(32) LibaiLibai
haialso
jian-lemeet-Asp
shei?who
’Who was the person x such that Libai met x (andsomeone else)?
26 / 46
Turkish
(33) JohnJohn
sadeceonly
kim-iwho-Acc
gor-du?see-Past
’Who was the person x such that John met x and nobodyelse?’
(34) Sadeceonly
kimwho
John-iJohn-Acc
gor-du?see-Past
’Who was the person x such that x and nobody else metJohn?’
27 / 46
A contrast between F-WH association and focus intervention
(35) [Q ... focus-sensitive operator [ ... WH ... ]]
(36) *[Q ... focus-sensitive operator [XPF ... WH ... ]]
I In-situ wh-phrases can associate with focus-sensitive operators(Aoun and Li 1993; Li 2013).
I The presence of XPF makes a difference!
I The intervention hypothesis makes no reference to XPF ,hence fails to predict the contrast.
I The quantificational-domain hypothesis predicts that XPF
interacts with the WH-containing constituent and results inan inappropriate quantificational domain for thefocus-sensitive operator.
28 / 46
Derviation of F-WH association
(37) CP
Q IP
Libai VP2
only VP1
met who
a. JwhoKg = JwhoKg ,h = {John, Peter, ...}b. JVP1Kg = JVP1Kg ,h ={λy. y met John, λy. y met Peter, ...}⇒ a set of alternatives
29 / 46
JVP2Kg = Jonly VP1Kg = Jonly met whoKg
=
λy.∀P∈ JVP1Kg ,h[P(y)=1→P(y)=y met John]λy.∀P∈ JVP1Kg ,h[P(y)=1→P(y)=y met Peter]
...
Only takes as its quantificational domain the set of alternativesderived via the wh-phrase ⇒ F-WH association
30 / 46
(38) JCPKg
= JQ IPKg = JIPKg (Kratzer and Shimoyama 2002)= JLibai only met whoKg
=
∀P∈ JVP1Kg ,h[P(Libai)=1→P(Libai)=Libai met John]∀P∈ JVP1Kg ,h[P(Libai)=1→P(Libai)=Libai met Peter]
...
= {Libai only met John, Libai only met Peter, ...}
NB: Each proposition of this set encodes exhaustivity
⇒ An answer to this question must be exhaustive
31 / 46
Association with multiple WH
A focus-sensitive operator can be associated with multiplewh-phrases.
(39) Tahe
zhionly
[V P song-lesend-Asp
sheiwho
shenmewhat
shu]?book
’Who was the person x and what was the book y such thathe only sent x y?’
(40) Tahe
haialso
[V P song-lesend-Asp
sheiwho
shenmewhat
shu]?book
’Who was the person x and what was the book y such thathe also sent x y?’
32 / 46
(41) Tahe
zhionly
[V P song-lesend-Asp
sheiwho
shenmewhat
shu]?book
a. JVPKg = JVPKg ,h ={λy. y sent x z | x ∈ JwhoKg , z ∈ Jwhat bookKg } ={λy. y sent Peter a novel,λy. y sent John a journal, ...}⇒ a set of alternatives (see also Hagstrom 1998)
b. Jonly VPKg =λy.∀P∈ JVPKg ,h [P(y)→P(y)=y sent Peter a novel]λy.∀P∈ JVPKg ,h [P(y)→P(y)=y sent John a journal]
...
⇒ Licit quantification
33 / 46
F-Alt association
Hamblin semantics of alternative questions in English
Assume that the compositional analysis of alternative questionsfollows Hamblin semantics (von Stechow 1991; Biezma andRawlins 2012; see also Beck and Kim 2006).
(42) a. [CP Did John [DisjP dance or sing]]?
b. JDisjPKg = {λy. y danced, λy. y sang}c. JCPKg = {John danced, John sang}
34 / 46
In this framework, disjunctive phrases in alternative questions havethe same ordinary semantic value as wh-phrases in Mandarinwh-in-situ questions. Consequently, our analysis predicts thefollowing contrast:
(43) a. ?* [Q ... focus-sensitive operator [ XPF ... DisjP ... ]]
b. [Q ... focus-sensitive operator [ ... DisjP ... ]]⇒ Association with alternatives
F-Alt association
35 / 46
(44) Focus intervention effects (Beck and Kim 2006: 172)
a. ?* Did only MaryF introduce Sue [DisjP to Bill or (to)Tom]?
b. ?* Did only MaryF introduce [DisjP Sue or Molly] toBill?
c. ?* Did only JohnF drink [DisjP coffee or tea]?
(45) F-Alt association
a. Did Mary introduce Sue only [DisjP to Bill or (to)Tom]?
b. Did Mary only introduce [DisjP Sue or Molly] to Bill?
c. Did John only drink [DisjP coffee or tea]?
36 / 46
Beyond questions: Contrastive topic (CT) constructions
Constant (2010, 2012) argues that Mandarin has a CT operator-ne, which triggers topic movement of a focused phrase.
(46) DufuDufu
ailove
hedrink
hongcha.black.tea
’Dufu likes to drink black tea.’
(er)while
LibaiF1LibaiF
ne,CT.operator
t1 ailike
hedrink
kafeiF .coffeeF
’LibaiF likes to drink coffeeF .’
37 / 46
Constant (2012)’s analysis
(47) [IP LibaiFLibaiF
ne,CT.operator
he-ledrink-Asp
kafeiF ].coffeeF
a. JLibaiF Kg = Libai; JLibaiF Kf = {Libai, Dufu, ...}b. JkafeiF Kg = coffee; JkafeiF Kf = {coffee, tea, ...}c. JIPKg = Libai drank coffee
d.
JIPKf =
{Libai drank coffee, Libai drank tea}{Dufu drank coffee, Dufu drank tea}
...
⇒ a set of sets of alternatives
38 / 46
It is predicted that focus intervention effects occur when afocus-sensitive operator scopes over a CT construction.
(48) [zhiyouonly
[IP LibaiFLibaiF
ne,CT.operator
he-ledrink-Asp
kafeiF ]].coffeeF
a. Jonly IPKg =∀p∈ JIPKf [p = 1 → p = Libai drank coffee]
b. However, p /∈ JIPKf (see (47d))⇒ focus intervention effects
39 / 46
The prediction is borne out.
(49) * Zhiyouonly
[IP LibaiFLibaiF
ne,CT.operator
he-ledrink-Asp
kafeiF ].coffeeF
Intended ’Only LibaiF drank coffeeF .’
(50) * ShiCleft.operator
[I P Dabufen de renF1
most DE personF
ne,CT.operator
douDOU
he-ledrink-Asp
kafeiF ].coffeeF
Intended ’It is most personsF who drank coffeeF .’
40 / 46
Conclusion
I The quantificational-domain hypothesis:Focus intervention is due to an inappropriate quantificationaldomain of a focus-sensitive operator.
I Attested predictionsI Focus-sensitive operators can take WH and DisjP in AltQs as
their associates without triggering focus intervention effects.I Focus intervention effects are not limited to questions.
41 / 46
Bibliography I
Aoun, J. and Li, Y.-H. A. (1993). Wh-elements in situ: syntax orLF. Linguistic Inquiry, 24:199–238.
Beck, S. (1996). Quantified structures as barriers for lf movement.Natural Language Semantics, 4:1–56.
Beck, S. (2006). Intervention effects in alternative questions.Natural Language Semantics, 14:1–56.
Beck, S. and Kim, S.-S. (2006). Intervention effects in alternativequestions. Journal of Comparative German Linguistics,9:165–208.
Biezma, M. and Rawlins, K. (2012). Responding to alternative andpolar questions. Linguistics and Philosophy, 35:361–406.
Cable, S. (2012). The grammar of Q. Oxford: Oxford UniversityPress.
42 / 46
Bibliography II
Constant, N. (2010). Mandarin ne as contrastive topic : the caseof ct questions. Presented at the 4th Workshop on Prosody,Syntax, and Information.
Constant, N. (2012). Topic abstraction as the source for nestedalternatives: A conservative semantics for contrastive topics. InArnett, N. and Bennet, R., editors, Proceedings of West CoastConference on Formal Linguistics, pages 120–130. Somerville,MA: Cascadilla Proceedings Project.
Eckardt, R. (2007). Inherent focus on wh-phrases. InPuig-Waldmueller, E., editor, Proceedings of Sinn andBedeutung 11.
Hagstrom, P. (1998). Decomposing questions. PhD thesis, MIT,Cambridge, MA.
Hamblin, C. L. (1973). Questions in montague english.Foundations of Language, 10:41–53.
43 / 46
Bibliography III
Kratzer, A. (1991). The representation of focus. In von Stechow,A. and Wunderlich, D., editors, Semantics: An internationalhandbook of contemporary research, pages 825–834. Berlin: deGruyter.
Kratzer, A. and Shimoyama, J. (2002). Indeterminate pronouns:The view from japanese. In Otsu, Y., editor, The Proceedings ofthe third Tokyo conference on psycholinguistics, pages 1–25.Hituzi Syobo, Tokyo.
Li, H. (2013). Association between focus particles and wh-phrases.In Goto, N., Otaki, K., Sato, A., and Takita, K., editors, TheProceedings of GLOW in Asia IX, pages 109–123. MieUniversity, Japan.
Mayr, C. (2013). Intervention effects and additivity. Journal ofSemantics, 0:1–42.
44 / 46
Bibliography IV
Pesetsky, D. (2000). Phrasal movement and its kin. Cambridge,MA: MIT Press.
Rooth, M. (1985). Association with focus. PhD thesis, UMass.
von Stechow, A. (1991). Focusing and background operators. InAbraham, W., editor, Discourse particles: Descriptive andtheoretical investigations on the logical, syntactic and pragmaticproperties of discourse particles in German, pages 37–81.Amsterdam, Philadelphia: Benjamins.
Yang, C.-Y. B. (2008). Intervention effects and the covertcomponent of the grammar. PhD thesis, National Tsing HuaUniversity, Taiwan.
Yatsushiro, K. (2009). The distribution of quantificational suffixesin japanese. Natural Language Semantics, 17:141–173.
45 / 46
Acknowledge
We are grateful to the Rutgers Semantics Research Group(SURGE) for their support in this presentation, in particular, MattBarrows, Yi-Hsun Chen, Veneeta Dayal, Jane Grimshaw, Ken Safir,Roger Schwarzchild, Kristen Syrett and Beibei Xu. We thankSimon Charlow, Ezra Keshet, Angelika Kratzer, Jo-Wang Lin andSatoshi Tomioka for their generous comments on various versionsof this work. We acknowledge the help of Sarah Hansen, YagmurSag, and Vartan Haghverdi with the English and the Turkish data.
46 / 46