Date post: | 16-Dec-2015 |
Category: |
Documents |
Upload: | iyana-corey |
View: | 214 times |
Download: | 0 times |
Recoordinating bare coordination
December 9th, 2010Going Romance
Bert Le Bruyn & Henriëtte de Swart
2
Spoon was
The phenomenon of bare coordination
I saw cats dogsand I saw
Context
We had to set the table for the queen. We arranged one crystal goblet, one silver spoon, two antique gold forks and two platinum knives.
Forks and knives were equally dirty
indefinite interpretation
definite interpretation
Plurals
Singulars
was set to the right of the plate* set to the right of the plate*Goblet spoon wereand only definite interpretation
Heycock & Zamparelli (2003)
??? There were goblet and spoon on the table.
3
The phenomenon of bare coordination
Heycock & Zamparelli (2003)
coordinatednot coordinated
bare singulars
bare plurals
indefinite definite indefinite definite
4
The phenomenon of bare coordination
Why is it bare singulars cannot occur bare whereas coordinated bare singulars can ?
When and why do bare coordinated nouns get a definite reading?
5
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
6
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
7
New facts
coordinatednot coordinated
bare singulars
bare plurals
indefinite definite indefinite definite
8
New facts
Et là on arrive dans un petit village où il y a école et point d’eau.And there we arrive in a small village where there is school and water point.
He had pad and pencil to picture the whole event.
There were goblet and spoon on the table.
> potentially bad because of the sequence Vpl Nsing
???
9
Recap
Basic dataCoordination lifts all semantic constraints on the absence of articles.
10
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
11
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
12
Roodenburg (2004)
The analysis in a nutshell
Premise 1: Bare Coordinated NPs are plural.
Conclusion: Bare coordinated NPs are allowed in argument position.
Premise 2: Bare Plural NPs are allowed in argument position.
> Cat and dog were fighting.
13
Roodenburg (2004)
The analysis in a nutshell
As for the definite readings: they’re akin to functional readings of bare plurals (Condoravdi 1994)
> Ghosts haunted the campus. Students were aware of the danger.
14
Roodenburg (2004)
Problem
Functional readings of bare plurals are limited to the subject position. Definite readings of coordinated bare nominals are not.
> Ghosts haunted the campus. We warned students about the danger.
> During the debate about the new law, the Prime Minister didn’t succeed in bringing proponents and opponents closer together.
15
Heycock & Zamparelli (2003)
The analysis in a nutshell
Focus on deriving the definite reading of bare coordinated nominals.
Proposal: allow for N-to-D raising of the coordinated phrase.
DP
CoordP
NP1 and NP2
16
Heycock & Zamparelli (2003)
Problems
Allowing for N-to-D raising for coordinated NPs begs the question why it wouldn’t be allowed for non-coordinated NPs.
N-to-D raising is often used for proper names but proper names arguably have a different semantics from definite DPs.
> Why is Goblet was set to the right of the plate bad ?
> Why can I saw Cat only mean that I saw someone by the name Cat?
17
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
18
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
19
Our analysis in a nutshell
> Blocked by I read a book and I read the book.
> Why is Goblet and spoon were set to the right of the plate good ?
> Not blocked!
20
Our analysis in a nutshell
coordinatednot coordinated
bare singulars
bare plurals
indefinite definite indefinite definite
> Classic blocking account:
indefinite bare singulars are blocked bydefinite bare singulars are blocked bydefinite bare plurals are blocked by the definite plural article
the definite singular articlethe indefinite singular article
21
Our analysis in a nutshell
coordinatednot coordinated
bare singulars
bare plurals
indefinite definite indefinite definite
> Not so classic blocking account:
A, thesing and theplural don’t apply at the coordination level.As a consequence they cannot be taken to block indefinite or definite readings of coordinated bare nominals.
22
Our analysis in a nutshell
A, thesing and theplural don’t apply at the coordination level.
>Indirect evidence
un homme et une femme (amale man and afemale woman) 1760000
un homme et femme (amale man and woman) 696
une femme et une fille (afemale woman and afemale girl) 885
une femme et fille (afemale woman and girl) 15
les hommes et les femmes (the men and the women) 3030000
les hommes et femmes (the men and women) 361000yahoo.fr 11/11/2010
Generalization:
Strong preference for repetition of the determiner; Suggests that the repetition of the determiner is the default; Suggests that the cases in which there is no repetition involve elided Ds.
23
Our analysis in a nutshell
A, thesing and theplural don’t apply at the coordination level.
>Direct evidence
Dog and cat were fighting. > bare coordination can trigger plural agreement
> there is a level of syntactic representation at which CoordPs have to have plurality specified (see also de Vries 1992)
> If Ds were to apply to CoordPs we would predict CoordPs to be able to take a plural article, even if both conjuncts are singular.
> This is however not the case.
*Dog and cat was fighting.
24
Our analysis in a nutshell
les hommes et les femmes the men and the women 3030000
les hommes et femmes the men and women 361000
les homme et femme the man and woman 99
les hommes et les garçons the men and the boys 2570
les hommes et garçons the men and boys 175
les homme et garçon the man and boy 1
les femmes et les filles the women and the girls 164000
les femmes et filles the women and girls 16000
les femme et fille the woman and girl 18
yahoo.fr 11/11/2010
25
Recap
Basic dataCoordination lifts all semantic constraints on the absence of articles.
Basic insightDeterminers don’t apply at the coordination level.
ImplementationClassic blocking ...
26
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
27
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
28
From ‘informal’ blocking to OTDP
NumP
CoordP
AND
NumP
NP
N
NumP
NP
N
DP DP
N-domain
CoordP-domain
N.B. Coordination can apply at the DP, NumP or NP-level.
N N
29
From ‘informal’ blocking to OT
a. FdrMark discourse referents
b. FplMark reference to a group
For each type of functional projection we have a faithfulness constraint.
DP
NumP
c. FdefMark definiteness
We add an extra one for D projections.
DP
For the two domains we add a markedness constraint.
d. *FunctNDon’t mark functional structure in the N-domain
e. *FunctCoordPDon’t mark functional structure in the CoordP-domain.
N-dom
CoordP-dom
30
From ‘informal’ blocking to OT
a. FdrMark discourse referents
b. FplMark reference to a group
For French and English the following ranking holds:
c. FdefMark definiteness
e. *FunctCoordPDon’t mark functional structure in the CoordP-domain.
d. *FunctNDon’t mark functional structure in the N-domain.
31
From ‘informal’ blocking to OT
Depending on the level at which coordination applies the ranking derives the following possibilities:
the cats and the dogsDP level coordination
cats and dogsNumP level coordination
cat and dogNP level coordination
Testable illegal structures:
I saw *(a) cat.Bare singular arguments
several cat and dogDs applying at CoordP
Untestable (?) illegal structures:
I saw cat and dogs (?)(meaning I saw cats and dogs)
Number at CoordP
32
Recap
Basic dataCoordination lifts all semantic constraints on the absence of articles.
Basic insightDeterminers don’t apply at the coordination level.
ImplementationClassic blocking ... and its formalization in OT.
33
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
34
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
35
The semantics of coordination
We assume the basic semantics of coordination at the level of sets is that of set intersection.
X Y
Bare coordination never has this basic semantics.
X and Y
> Bride and groom were extremely happy.
There was an extremely happy person who was both bride and groom.
36
The semantics of coordination
Two types of coordination:
> coordination with ‘joint’ readings
> coordination with ‘split’ readings
Bare coordination always concerns coordination with ‘split’ readings.
Our challenge will be to derive split readings without giving up the basic intuition of coordination being an instance of set intersection.
37
The semantics of coordination
How to go about this?
> Enrichment of and
> First enrichment: turn and into a ‘matchmaker’
PQ ( )PQ x E E x( ) ( )
> Based on a proposal by Yoad Winter (p.c.)
38
The semantics of coordination
bride groom
( , )
( , )
( , )
( , )
( , )
( , )
39
The semantics of coordination
( , )
( , )
( , )
( , )
( , )
QxE ExP
40
The semantics of coordination
> Enrichment of and
> First enrichment: turn and into a ‘matchmaker’
PQ ( )PQ x E E x( ) ( )
> Second enrichment: add a function that turns (singular) couples into plural individuals.
> Based on a proposal by Yoad Winter (p.c.)
PQ ( )PQ x E E x( ) ( )RtoI
Relations to Individuals
RtoI(R) = {xy : R(x,y)}
How to go about this?
41
The semantics of coordination
bride and groom
42
The semantics of coordinationbride and groom
> Bride and groom were extremely happy.
> the unique plural individual consisting of a bride and groom was extremely happy
> extremely_happy( )
43
Recap
Basic dataCoordination lifts all semantic constraints on the absence of articles.
Basic insightDeterminers don’t apply at the coordination level.
ImplementationClassic blocking ... and its formalization in OT.
The semantics of bare coordinationEnriched version of an intersective semantics.
44
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
45
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
shortcut to conclusion
46
Why cat and dog is ‘definite’ by default
coordinatednot coordinated
bare singulars
bare plurals
indefinite definite indefinite definite
coordinatednot coordinated
bare singulars
bare plurals
indefinite definite indefinite definite
47
Cat and dog were fighting.
> Implicature of uniqueness
If there had been more cats and dogs, we could have told you so.
Given that we did not tell you, you can assume that there was only one cat and one dog.
> The effect of this implicature is almost indistinguishable from the contribution of the definite article.
Even though our semantic account predicts both a definite and an indefinite reading, pragmatically the indefinite reading is so close to the definite reading that one gets the impression there’s only a definite reading.
Why cat and dog is ‘definite’ by default
48
Predictions
... coordinated bare plurals should not have any preference for definite readings.
... the preference for definite interpretations should be cancelable.
Given that the implicature depends on the nouns being singular...
Given that we assume the default definite interpretation is an implicature...
> This is arguably what we find (see Heycock & Zamparelli).
> This is what we have demonstrated for existential contexts.
Why cat and dog is ‘definite’ by default
49
More predictions
... the definiteness effect should not only be found for coordinated nouns but also for uncoordinated singular nouns in languages that have a singular/plural distinction but no articles
Given that the implicature arises because of the competition between bare singulars and plurals...
> Languages like Hindi and Russian have indeed been argued to only allow for definite readings for bare singulars, despite their acceptability in existential environments (see Dayal 2004).
Why cat and dog is ‘definite’ by default
50
More predictions
... the definiteness effect should not only be found for coordinated nouns but also for uncoordinated singular nouns in languages that have a singular/plural distinction but no articles
... uncoordinated plural nouns in these languages should not show any preference for definite readings
Given that the implicature arises because of the competition between bare singulars and plurals...
> Languages like Hindi and Russian have indeed been argued to only allow for definite readings for bare singulars, despite their acceptability in existential environments (see Dayal 2004, Geist 2010).
> Uncoordinated bare plurals in Hindi and Russian have indeed been argued to allow both definite and indefinite readings (see Dayal 2004).
Why cat and dog is ‘definite’ by default
51
One more prediction
... there should be no definiteness effect in Chinese comparable to the one in Hindi and Russian
Given that the implicature arises because of the competition between bare singulars and plurals...
> Bare nominals in Chinese have indeed been argued to freely allow both for a definite and an indefinite reading (see Yang 2001).
N.B.
This implicature account can be formulated both under the analysis of the singular/plural contrast of Farkas & de Swart (2010) and the one in the tradition of Krifka (1989) (see a.o. Sauerland et al. 2005).
Why cat and dog is ‘definite’ by default
52
• New facts• Previous analyses• Our analysis in a nutshell• Our analysis in OT• The semantics of and• Surprise Bonus
Roadmap
to conclusion
53
The phenomenon of bare coordination
Why is it bare singulars cannot occur bare whereas coordinated bare singulars can ?
When and why do bare coordinated nouns get a definite reading?
> Articles don’t apply at the coordination level
> No blocking of bare coordinated forms
> Semantically, definite/indefinite readings are available through type-shifting > Pragmatically, bare singulars prefer ‘definite’ readings
54
References
Dayal, 2004, ‘Number marking and (in)definiteness in kind terms’, Linguistics and Philosophy 27, 393-450.
Farkas & de Swart, 2010, “The semantics and pragmatics of plurals”, Semantics and Pragmatics 3.
Geist, 2010, “Indefinite NPs without indefinite articles”, presentation at SUB 2010.
Heycock & Zamparelli, 2003, “Coordinated bare definites”, Linguistic Inquiry 34, 443-469.
Heycock & Zamparelli, 2005, “Friends and colleagues”, Natural Language Semantics 13, 201-270.
Krifka, 1989, “Nominal reference, temporal constitution and quantification in event semantics”, in: Bartsch, van Benthem & van Emde Boas (eds.), Semantics and contextual expression, Foris.
Roodenburg, 2004, Pour une approche scalaire de la déficience nominale, Ph.D. Dissertation, Universiteit van Amsterdam.
Sauerland, Anderssen & Yatsushiro, 2005, “The plural is semantically unmarked”, in: Kepser & Reis (eds.), Linguistic evidence, de Gruyter.
Yang, 2001, Common nouns, classifiers, and quantification in Chinese, Ph.D. Dissertation, Rutgers University.
Zwarts, 2009, Bare constructions in Dutch, Ms., Utrecht University.
55
This presentation builds on work that we carried out with Vera Mulder and Paulien Hesselink. We hereby gratefully acknowledge their contribution.
http://www.hum.uu.nl/medewerkers/b.s.w.lebruyn/weakreferentiality/index.htm
On the elided D analysis of single determiner
coordinations
57
Problems for the elided D analysis
This incompatible man and woman.
??This incompatible man and this incompatible woman.
Heycock & Zamparelli (2005)
My five maternal aunts and uncles were all Methodists.
My five maternal aunts and my five maternal uncles were all Methodists.
default: 5 individuals
default: 10 individuals
58
Problems for the elided D analysis
> All disambiguating cases involve some kind of modification.
> Modification overrules the restriction on determiners at the coordination level.
> Can be linked to a general constraint on modification of bare nominals (see Zwarts 2009).
59
Problems for the elided D analysis
This soldier and sailor agreed.
*Ce soldat et marin étaient d’accord.
The soldiers and sailors agreed.
Les soldats et marins étaient d’accord.
If we assume French doesn’t allow the non-repetition of determiners, why would there be a difference between singular and plural determiners ?
Heycock & Zamparelli (2005)
60
un homme et une femme (amale man and afemale woman) 1760000
un homme et femme (amale man and woman) 696
une femme et une fille (afemale woman and afemale girl) 885
une femme et fille (afemale woman and girl) 15
les hommes et les femmes (the men and the women) 3030000
les hommes et femmes (the men and women) 361000yahoo.fr 11/11/2010
338000
uk.yahoo.uk 05/12/2010restricted to UK pages
801000
276
1250
1580000
26100
Problems for the elided D analysis
FRENCH ENGLISH
This is a strong argument iff:> French shows no preference for repetition of plural determiners. > A language like English shows no difference in the repetition of singular and plural determiners.
61
Conclusion
There is still some work to be done to make a waterproof case for the elided D analysis.
Crucially however there are no knock-down arguments against it and we do have a strong argument against an analysis that takes determiners to apply to the coordination of two bare singulars.