Predication:
why we (sometimes) need a
Bert Le Bruyn
SiN 2008
I am a linguist.
Outline
Outline
DUTCH & ENGLISH
● Facts
● Analysis of de Swart, Winter & Zwarts
● A problem
● A solution
● An extension
Facts
Standard observations
one set of nouns usually doesn’t take the indefinite article
= capacity nouns
Professions
Religions
Nationalities
lawyer
jew
Belgian
dictator
christian
American
…
…
…
ex. Hitler was dictator.
H was dictator
another set of nouns usually takes the indefinite article
= non-capacity nouns
The rest
ex. White Fang is een wolf.
WF is a wolf
Advanced observations
MARKED USES
capacity nouns can occur with the indefinite article
Marie is een dictator.
“non-capacity use”
M is a dictator
“Mary has characteristics that we
associate with dictators”
non-capacity nouns can occur without the indefinite article
Ik ben wolf.
I am wolve
“I play the part of wolve”
“capacity use”
NON-CAPACITY NOUNS
Hitler was dictator.
Ik ben wolf.
Hitler was dictator
I am wolf
INDEFINITE ART
NO INDEFINITE ART
CAPACITY NOUNS
Marie is een dictator.
White Fang is een wolf.
Marie is a dictator
WF is a wolf
de Swart et al. (2007)
I. In the lexicon there are two kinds of nouns
non-capacity nouns
capacity nouns
-> kind nouns
type e, subtype ‘kinds’
I. In the lexicon there are two kinds of nouns
kind nouns
type e, subtype ‘kinds’
capacity nouns
type e, subtype ‘capacities’
Capacities
capacity nouns
Kinds
kind nouns
e
II. When occurring in predicate position nouns
have to shift to type <e,t>
To do this both have a special type-shift:
non-capacity nouns REL (cf. Carlson 1980)
kind
set of instantiations of the kind
capacity nouns CAP
profession
set of people performing the profession
religion
set of followers
nationality
set of citizens
Sometimes this shift is made explicit:
Hij is advocaat van beroep.
He is lawyer of profession
Capacities
Kinds
capacity nouns
e
kind nouns
CAP
REL
<e,t>
III. NumP selects kinds
Assumption:
The indefinite article in predicate position sits in NumP.
-> The combination of the indefinite article and a capacity noun
leads to a sortal clash.
-> To solve this clash the capacity gets coerced into a kind.
-> The operator used to do this is called kind.
Capacities
Kinds
e
[presence of NumP]
capacity nouns
REL
<e,t>
III. NumP selects kinds
Assumption:
The indefinite article in predicate position sits in NumP.
-> The combination of the indefinite article and a capacity noun
leads to a sortal clash.
-> To solve this clash the capacity gets coerced into a kind.
-> The operator used to do this is called kind.
This coercion has a semantic effect:
KINDS
kinds are different from capacities in that they not only
group the individuals that perform a certain profession
but also those that have the characteristics associated
with the profession
Capacities
Kinds
e
[presence of NumP]
capacity nouns
REL
CAP
<e,t>
Capacities
Kinds
e
[presence of NumP]
capacity nouns
CAP
REL
<e,t>
Facts central in de Swart et al. (2007)
● Special status of capacity nouns.
● Unmarked reading of capacity nouns.
Hitler was dictator.
Hitler was dictator
(Application of CAP that maps professions
to the people that perform it)
● Marked reading of capacity nouns.
Marie is een dictator.
Mary is a dictator
(Coercion into kind + application of REL)
The problem
Observation
De Swart et al. (2007) treat the “kind” reading of capacity
nouns but don’t treat the “capacity” reading of kind
nouns.
Question
Can it be incorporated into their account ? NO !
indefinite article
+
capacity
absence of article +
kind
indefinite article
+
kind
cap
absence of article +
REL
indefinite article
+
instantiations
of the kind
kind
capacity
CAP
absence of article +
The problem
There is nothing for the capacity to clash with...
and therefore no reason for coercion of any kind.
individuals
having a
capacity
The solution: Part I
The problem
There is nothing for the capacity to clash with...
and therefore no reason for coercion of any kind.
strategy
get a meaning for the indefinite article
-> exploit presence
-> exploit absence
A meaning for the indefinite article (1)
Marking argumenthood
In languages that have articles they are obligatory in
argument position (in as far as they render the same
semantics as the bare form)
*I have dog.
*Woman came to see me.
Marking uniqueness
In languages that distinguish between a definite and an
indefinite article the definite article (in the singular) is marked
for uniqueness whereas the indefinite article is unmarked.
#I saw the priest.
I saw a priest.
A meaning for the indefinite article (2)
Indefinite article vs. bare form
both constructions are unmarked for uniqueness
wherever both are possible (i.e. in predicate position) the
construction with the indefinite article marks non-uniqueness
(marked form linked to marked meaning)
A meaning for the indefinite article (3)
What does non-uniqueness really mean ?
men
Marc
John
[[John is man.]]
Lucas
Matthew
-> John belongs to the set of men. TRUE
[[John is a man.]] -> John belongs to the set of men TRUE
and there is at least one other man.
A meaning for the indefinite article (3)
What does non-uniqueness really mean ?
men
John
[[John is man.]]
-> John belongs to the set of men. TRUE
[[John is a man.]] -> John belongs to the set of men FALSE
and there is at least one other man.
Interludium
Supervisor:
What does it mean to be unique ?
What does it mean to be non-unique ?
Me (to myself):
What does she mean ?
I already gave an explicit semantics...
p
a
s
t
men
To decide whether John is a man...
Marc
John
Lucas
Matthew
it is completely irrelevant to know there
are other men...
Marking non-uniqueness is therefore
completely irrelevant and should be
proscribed.
unless...
Mary saw cats.
Mary saw tall cats.
Mary saw taller cats.
* Mary saw tallest cats.
-> Mary saw the tallest cats.
even though the superlative guarantees uniqueness /
maximality by itself the definite article has to be used
Mary saw cats.
Mary saw tall cats.
Mary saw taller cats.
* Mary saw tallest cats.
-> Mary saw the tallest cats.
despite the fact that marking definiteness is irrelevant we
cannot but mark it
What is it that distinguishes DPs containing superlatives
from all other DPs ?
they guarantee uniqueness / maximality independently of
the model
Suggestion
If uniqueness / maximality is guaranteed modelindependently it has to be marked.
-> this constraint overwrites the relevance criterion
Extension
If non-uniqueness is guaranteed model-independently it
has to be marked.
-> this constraint overwrites the relevance criterion
The solution: Part II
Background on kinds (1)
Kinds are regularities that occur in nature.
(Chierchia 1998)
One important corollary for me:
-> they should – in potential – have more than one member
Background on kinds (2)
What does it mean to have “– in potential – more than one member” ?
+
#
KIND
>1
A kind can only be a kind if it has at least two instantiations.
Background on kinds (2)
What does it mean to have “– in potential – more than one member” ?
+
#
KIND
>1
It is model-independently guaranteed that kinds have at
least two members.
Capacities
Kinds
capacity nouns
e
kind nouns
If we assume REL takes a kind
and returns a set containing all
members of the kind (including
those of other worlds)...
... it is model-independently
guaranteed that the obtained set
contains at least two members.
CAP
REL
<e,t>
# members > 1
Pulling things together (1)
Whenever REL applies nonuniqueness has to be marked.
This has to be marked (for the
singular) with the indefinite
article.
Pulling things together (2)
Whenever REL applies nonuniqueness has to be marked.
This has to be marked (for the
singular) with the indefinite
article.
In as far as CAP doesn’t
necessarily give rise to sets with
at least 2 members...
...the relevance criterion tells us
that non-uniqueness should
never be used with capacity
nouns.
The indefinite article ends up being unambiguously
connected to REL and its absence to CAP.
Capacities
Kinds
capacity nouns
e
kind nouns
Ø
IND. ARTICLE
CAP
REL
<e,t>
An extension
Nitty-gritty facts about Dutch
Paul was journalist toen Parijs werd
aangevallen.
Paul was journalist
when Paris
was being attacked
“Paul was a journalist when Paris was under attack.” capacity noun / bare
?? Paul
was een journalist toen Parijs werd
aangevallen.
Paul was a
journalist
when Paris was being attacked
“Paul was a journalist when Paris was under attack” capacity noun / non-bare
?? Marie
is een meisje in de eerste jaren van haar leven.
Marie is a
girl
in the first
years
“Marie is a girl during the first years of her life.”
of
her
life
non-capacity noun / non-bare
Paul was meisje in het spel dat we gisteren speelden.
Paul was girl
in the game that we yesterday played
“Paul played the role of girl in our game yesterday.”
non-capacity noun / bare
Feeling of “what?! / why?! / how?!” when adding
adverbial modification to predication with the ind. art.
Capacities
Kinds
capacity nouns
e
kind nouns
contextual restriction
CAP
no contextual restriction
REL
<e,t>
Nitty-gritty facts about English
Henry is treasurer.
Mary is deputy leader of the party.
Ann is head of the department.
They refer to ‘unique professions’.
Two assumptions
1. English has a general ban on bare singulars. I take this to be a
syntactic ban.
2. The indefinite article in English is the default way of avoiding bare
singulars.
An account
The indefinite singular can only be used as a way to avoid bare nominals...
... insofar as its use does not clash with its semantics.
Given that sets originating in capacities always come with a contextual
restriction...
we expect a possible clash between the indefinite singular as the default
way to avoid bare nominals...
and its semantics.
In those cases we expect bare nominals to be possible in English.
-> predicts the behaviour of “unique professions” in English !
-> strong confirmation of the fact that the indefinite article marks
non-uniqueness !
Descargar

Dia 1