Transcript Power Point presentation - University of Massachusetts Amherst

Diathesis Alternations and
NP Semantics
Barbara H. Partee
University of Massachusetts, Amherst
Acknowledgements




Thanks to the Fulbright Foundation for a Fellowship
to teach in Moscow in 2005.
Thanks to many students in classes at RGGU and
MGU for data and suggestions.
Thanks to Vladimir Borschev, Elena Paducheva,
Ekaterina Rakhilina, and Yakov Testelets for
ongoing discussion.
This material is based upon work supported in part
by the National Science Foundation under Grant No.
BCS-0418311 to B.H. Partee and V. Borschev.
2
Abstract



In this paper we examine the relationships among
diathesis alternations, the semantics of verbs, and
the referential status of NPs.
Some diathesis shifts are argued to involve changes
in the semantic type of NP arguments, including
possible alternations between ‘referential’ NPs and
‘property-type’ NPs.
We explore applications of this approach to
alternations of Genitive and Accusative, both with
intensional verbs like ždat’ and with the Genitive of
Negation.
3
1. Examples and issues.
For the purposes of this paper, I take the
notion of diathesis alternation in a broad
sense. I include not only familiar alternations
of syntactic patterns as in examples (1a-b):
(1) a. The farmers loaded the truck with (the) hay.
b. The farmers loaded (the/some) hay on the truck.


but also alternations of case assignment, as in
the next examples.
4

(2) a. On ždet
podrugu.
(Neidle 1988, p.31)
He waits girlfriend-acc
‘He’s waiting for his girlfriend.’
b. On ždet
otveta
na
vopros.
He waits answer-gen
to
question
‘He’s waiting for an answer to the question.’

(3) a.
On
ne
polučil pis’mo.
he
NEG received
‘He didn’t receive the letter.’
b.
On
ne
letter-acc
polučil pis’ma.
he
NEG received
‘He didn’t receive any letter.’
letter-gen
5
Examples and issues, continued.

Concern: the interaction between:



lexical semantic interpretation of the verb in its
various diathesis frames and
the semantic interpretation of noun phrase (NP)
arguments that appear in those frames.
Long-range goal: integration of lexical
semantics with compositional semantics,
(Borschev and Partee 1999, 2002, Partee and
Borschev 2003).
6
Diatheses and referential differences
(1a) The farmers loaded the truck with (*?the/*?some) hay.
(1b)
The farmers loaded (the/some) hay on the truck.
In (1a-b) we see a preference for the direct object to be
definite in either diathesis frame.
The difference is most pronounced for the role which is
‘Means’ in (1a) (sredstvo -- Apresjan) and ‘Patient’ in
(1b): hay is almost obligatorily non-specific in (1a) but
may be indefinite or definite in (1b).
7
Intensional verbs and referential status

In (2a-b) there is a clear difference in referential status of
the direct object argument in the two cases, and a
correlated difference in the sense of the verb.

(2) a. On ždet podrugu-acc
(Neidle 1988)
‘He’s waiting for his girlfriend.’
b. On ždet otveta-gen na vopros.
‘He’s waiting for an answer to the question.’

8
Genitive of Negation and Referential Status


Similar difference in referential status of the object in
(3a-b); in this case the verb itself does not seem
different, but negation together with the verb creates
a context somehow similar to the non-referential
complement of ždat’. This possibility was raised by
Neidle (1988), Partee and Borschev (2004), Kagan
(2005). (More on this later.)
(3) a. On ne polučil pis’mo-acc.
‘He didn’t receive the letter.’
b. On ne polučil pis’ma-gen.
‘He didn’t receive any letter.’
9
The Issues

The questions of concern in this paper are to
what extent the semantic interpretation of the
NP is part of, or is affected by, the semantics
connected with the diathesis alternations
seen in such pairs, and how the semantics of
the diatheses are connected with the lexical
semantics of the verb.
10
Not only diathesis alternations …

Correlation between different senses of a verb and
different interpretations of an NP argument:

(4) John is looking for a blue Volkswagen.
 a.
… He can’t remember where he parked it.
 b.
… If he finds one, he will buy it for his wife.

“Look for” in (4a) denotes a relation between two entities; in (4b)
it denotes a relation between an entity (the agent) and some
intensional object (Montague 1973), perhaps a property as
argued by Zimmermann (1993).

(4a-b) are not normally classed as diathesis alternation, but the
interaction of verb and NP interpretations is similar.
11
Filip: Verbal affixes, diathesis shift, and
quantificational effects

Filip (In press) has explored some cases that
were first discussed in Partee (1991, 1995),
in which verbal affixes together with diathesis
shifts can be used to express various kinds of
quantificational and closely related meanings
like measure, distributivity, totality, exclusivity
or exhaustiveness.
12
Filip: Verbal affixes, diathesis shift, and
quantificational effects, continued.

One example is the use of the Czech prefix po- , discussed in
Partee (1991, 1995) and illustrated here by examples (5a-b) from
Filip (In press):

(5) a. Maloval
hesla
(na sténu). Czech
paint.past.impf.3sg slogan.acc
(on wall)
‘He painted (the/some) slogans (on the wall).’
b. PO-maloval
sténu
hesly.
tot-paint.past.
3sg wall.acc slogans.inst
‘He covered the wall with slogans.’
c. *PO-maloval hesla
na sténu.
tot-paint.past.pf.3sg slogan.pl.acc on wall.sg.acc
*‘He covered (the/some) slogans on the wall.’


13
(5)



a.
b.
Maloval hesla-acc (na sténu).
PO-maloval sténu-acc hesly-instr.
When po- is applied to an imperfective verb with a meaning like
‘write’, ‘draw’, etc., which takes an object of creation (5a), the
resulting perfective verb in (5b) takes as its direct object the
optional locative complement of the base verb, and the direct
object of the base verb is demoted to an optional instrumental
complement of the prefixed verb.
The meaning of the perfective verb pomaloval in (5b) is ‘he
painted all over X’ or ‘he covered X with painting’. Hence, the
prefix po- is in a certain sense quantificational but at a lexical
rather than a syntactic level (Partee 1995).
Note the different interpretations of the bare NP hesla ‘the/some
slogans’ as direct object of the imperfective verb in (5a), and in
instrumental form in (5b). In (5b) it can only be indefinite, and in a
sense is even ‘more indefinite’ than it can be in (5a).
14
Lines of investigation

Formal semanticists who have studyied connections
between verb semantics and the semantics of NP
arguments have emphasized quantificational properties,

relation between aspectual properties of verbs and the
count/mass distinction

(Krifka, Bach, Dowty, Filip, Partee).

Paducheva: correlations between decreased
assertiveness (snjataja utverditel’nost’) in sentences and
decreased referential status in NPs.

Yanko: diathesis alternations correlating with ThemeRheme structure (also in recent work of Levin and
Rappaport Hovav, below).
15
Debates about semantics of diathesis alternations



Formal semanticists have recently begun seeking arguments to
distinguish between diathesis alternations that carry genuine
semantic distinctions and those that may carry only Theme-Rheme
structure distinctions (Krifka 2004).
An example of the difficulty of sorting out such factors can be seen
in debates concerning semantics and pragmatics of the Dative
Alternation in English (6a-b).
(6)
a.
b.

Ann sold the car
NP0 V NP2
Ann sold Beth the car
NP0 V NP1 NP2
to
to
Beth
NP1
Rheme
Theme
The double object construction in (6b) is argued by Levin and
Rappaport Hovav and by Bresnan to be motivated largely to make
the “dative” argument the Theme.
16
(6)
a.
b.





Ann sold the car to
NP0 V NP2
to
Ann sold Beth the car
NP0 V NP1 NP2
Beth
NP1
On Krifka’s analysis, the lexical semantics of many of the verbs
that occur in both frames of (6a-b) is a manner of action.
The construction in (6a) contributes an entailment that NP0
caused NP2 to go to NP1.
The construction in (6b) contributes an entailment that NP0
caused NP1 to have NP2.
Some verbs, like give and sell, have so much information in their
lexical semantics that the constructions contribute nothing new,
and the sentences in the two frames end up truth-conditionally
equivalent; in this case, it is especially likely that differences in
Theme-Rheme structure will be the most salient differences.
But in other cases, as with different sorts of verbs of causation of
motion, like throw vs. push, the semantic difference between the
two frames may be quite noticeable.
17
(6)
a.
b.
Ann sold the car to
NP0 V NP2
to
Ann sold Beth the car
NP0 V NP1 NP2
Beth
NP1
Most relevant to this paper, we see the grounds for a prediction
of a difference in referential status to arise in certain cases.
 Consider the two subparts of the interpretations,
 “NP2 to go to NP1” for (6a) vs. “NP1 to have NP2” for (6b)
 In the first, NP2 must exist throughout the action; in the second,
NP2 may come into existence as a result of the action.
 Krifka shows cases where this results in only one of the two
diatheses being possible with some NPs.
(7)a. #The explanation gave a headache to Susan.
b. The explanation gave Susan a headache.
 The headache does not exist throughout the event, so only the
(7b) frame (‘cause to have’) is appropriate for ‘less referential’
headache.

18

In the case of the Dative Alternation, we see
the situation that the semantics of the two
diatheses permits cases in which the
referential status of NP2 differs, but does not
require them to differ. A fully referential NP2
is possible in both frames, while nonreferential NP2 is possible only in
construction (6b).
19
Still to come:

In the second half of the paper, we look at the
interaction of indefiniteness and ‘decreased
referential status’ of NPs with diathesis
alternations and verbal semantics.
20
Referential Status in formal semantics

In formal semantics there has been a great deal of
work on NP interpretations, including much work on






quantification,
definiteness and indefiniteness
kind-denoting NPs
property-denoting NPs
type-shifting among different possible interpretations of
NPs
and recently on semantic typology and different kinds of
indefinite and quantificational NPs.
21
Generalized Quantifiers

Classic formal semantics adopted
Montague’s proposal (Montague 1973) for
the semantics of Noun Phrases (NPs). Every
NP was interpreted as denoting a
Generalized Quantifier, i.e. a set of sets, type
<<e,t>,t> (strictly, a set of properties, type
<<<s,<e,t>>,t>.)
22
NP interpretations: Montague
Some NP interpretations on Montague’s analysis:


John
λP[P(j)]
(the set of all of John’s
properties)
every student
λPx[student(x)  P(x)]
(the set of all of properties that every student has)

a student
λPx[student(x) & P(x)]
(the set of properties that at least one student has)

the king
λP [x[king(x) & y ( king(y)  y = x) & P(x))]
(the set of properties which the one and only king has)
23
Indefinites

What could “indefinite” mean in such a framework? The need for
such a distinction first arose in discussions of the prohibition of
“definite” NPs in English existential there-sentences: see the
contrast between the acceptable sentences in (9) and the
sentences in (10), which are anomalous without special contexts.
(9) a. There is a new problem.
b. There are three/many/several/few/no semantics textbooks.
(10) a. #There is every/neither/the linguistics student.

b. #There are most/both/the three democratic governments.
24

No intuitive notion of “definite” vs. “indefinite”
explains why some quantifiers, like three and
many, pattern with indefinites while others,
like every, most, and both, pattern with
definites. Milsark (1974, 1977) provided the
kernel of a semantic explanation, dividing
determiners into “weak” and “strong”, which
was further developed by Barwise and
Cooper (1981) and by Keenan (1987).
25
Semantic explanation – Milsark, Barwise and
Cooper, Keenan

Definition (Keenan 1987): A determiner D is
a basic existential determiner if for all models
M and all A,B  E,
D(A)(B) = D(AB)(E).

English test: “Det CN VP” is true iff “Det CN
which VP exist(s)” is true.
26
Examples

(i) Three is an existential determiner: Three
cats are in the tree iff three cats which are in
the tree exist.

(ii) Every is not existential:
Suppose there are 5 cats, and 3 are in the
tree. Then:
“Every cat is in the tree” is false but “Every
cat which is in the tree exists” is true.

27
Existential = Symmetric

Basic existential determiners = symmetric
determiners.


One can prove, given that all determiners are conservative
(Barwise and Cooper 1981), that Keenan’s basic
existential determiners are exactly the symmetric
determiners.
Symmetry: A determiner D is symmetric iff
for all A, B, D(A)(B) ≡ D(B)(A).
28
Weak and Strong Determiners



The determiners three, a, some, no, at least
three, exactly three, at most three are all
weak.
The determiners the, the three, every, both,
most, neither are all strong.
These semantic definitions laid a successful
groundwork for a great deal of further
research on the semantics of weak and
strong NPs and their distribution (Büring, de
Hoop, Partee, Rullmann).
29
Property-type NP interpretations

While some properties of ‘weak’ and ‘strong’ NPs can be
accounted for within the theory of generalized
quantifiers, as in the account above, it has been argued
that in some cases, weak NPs are really of “property
type” (Montague’s type <s,<e,t>>, functions from
possible situations to sets of entities), rather than
generalized quantifiers.

Property-type analyses of various “weak NPs” are
becoming increasingly common in Western formal
semantics, and they may have an important role to play
in accounting for non-referential readings of NPs in
various diatheses, possibly including the Russian
Genitive of Negation (Partee and Borschev, Kagan)
30
Property-type interpretations

Zimmermann 1993: argues against
Montague’s analysis of “intensional transitive
verbs” like seek

Montague: object is intensional generalized
quantifier, type <s,<s,<e,t>>,t>.

Zimmermann: object is property-type, type
<s,<e,t>>.
31
Fundamental properties of intensional contexts
(11)
Caroline found a unicorn.
(extensional, unambiguous)
(12) Caroline sought a unicorn.
(intensional, ambiguous)


Sentences with seek are ambiguous between a
specific and a non-specific reading (or transparent
vs. opaque reading). (11) is unambiguous, (12) is
ambiguous.
On the opaque reading of (12), the existence of a
unicorn is not entailed.
32
Fundamental properties of intensional
contexts, continued

Substitution of extensionally equivalent
expressions in an intensional context does
not always preserve truth-value.

Caroline is looking for a unicorn
The set of unicorns = the set of 13-leaf
clovers
Not entailed: Caroline is looking for a 13leaf clover


33
Zimmermann’s account



(Montague’s account – see written text.)
Zimmermann: we can capture the relevant
generalizations if we treat definite and
indefinite arguments of intensional verbs, (but
not generalized quantifiers) as properties,
type <s,<e,t>>.
Zimmermann’s proposal is that a verb like
seek1 denotes a relation between an
individual and a property.
34
Zimmermann’s account, continued
Zimmermann: seek a unicorn:

seek’(^unicorn’)
( ^ is Montague’s ‘intension operator’)

This is a case of NP type-shifting by coercion: seek
demands a property-type argument.

We know that indefinite NPs easily shift into <s,<e,t>>
readings, as was shown for predicate nominals in
(Partee 1986).

transparent, or de re, reading: “quantify in” to e-type
argument position of seek2.
35
(2) a. On ždet podrugu-acc ‘He’s waiting for his girlfriend.’
b. On ždet otveta-gen. ‘He’s waiting for an answer.’



In the case of the potentially intensional verb
ždat’ in (2a-b), we see that its intensional
genitive-taking variant in (2b) has all the
properties of English seek.
Its extensional accusative-taking variant,
(2a), allows referential NPs and
quantificational NPs.
We predict that genitive should be disallowed
with essentially quantificational NPs such as
those formed with každyj.
36
What about Genitive of Negation?


Hypothesis (Neidle, Partee & Borschev,
Kagan)
Wherever we see Nom/Gen and Acc/Gen
alternation (under negation, just as under
intensional verbs):


Nom or Acc represents an ordinary e-type
argument position (‘referential’; and may be
quantified)
Gen NP is always interpreted as property-type:
<e,t>, or <s,<e,t>>.
37
Russian Genitive of Negation, continued.


In the case of Genitive of Negation, the
construction is not intensional.
But Russian linguists from Jakobson to
Paducheva have argued that Genitivemarked NPs have reduced “referential
status”, and Western linguists have generally
claimed that they must be “indefinite”.
38
Kinds of reduced referentiality

Negation is not really intensional; there seem to be
different kinds of ‘reduced referentiality’.





Intensional NPs: existence in possible worlds
Abstract NPs: peace, justice, trouble: may exist in actual
world, but their existence is less “concrete”.
Quantificational NPs: many different kinds, some just as
‘real-world existent’ as a definite NP, some not (Paducheva,
others)
Indefinites under negation: if non-presuppo-sitional, then
non-existence may be implied.
Predicate NPs, Property-type NPs: less referential just as
adjectives are less referential.
39
Genitive as a mark of reduced referentiality


A Genitive NP as an argument of a verb may always
be “less referential” in some sense; competing
theories may ‘all be right’:
There may well be more than one way for an NP to
be ‘less referential’ (much as there are several
different kinds of ‘imperfective’ meaning), including
being ‘quantificational/partitive’, being
‘modalized/intensional (not necessarily actual),
being property-type or kind-type or ‘abstract’
40
Are Gen Neg NPs property-denoting?

Evidence in favor: parallels to ждать.
(a) Петя нашел ответ.
(b) Петя не нашел ответ.
(c) Петя не нашел ответа.
Accusative implies actual-world existence,
Genitive does not.
41
Parallels between Gen Neg and Subjunctive




From Kagan 2005
25 a. Ivan ne počuvstvoval, čto bylo
xolodno
Ivan NEG felt
that be(past) cold
Ivan didn’t feel that it was cold.
b. Ivan ne počuvstvoval, čtoby
bylo
xolodno
Ivan NEG felt
that-subj be(past) cold
26 a. Ivan ne počuvstvoval xolod.
Ivan NEG felt
cold(acc)
Ivan didn’t feel the cold.
b. Ivan ne počuvstvoval xoloda.
Ivan NEG felt
cold(gen)
Such parallels support a property-type analysis
42
Problems for property-type analysis

Evidence casting doubt on property
analysis:
(a) Я не видела Машу.
(b) Я не видела Маши.
The (b) case causes problems for all “quantificational”
approaches to the Genitive of Negation, unless we
suggest a meaning like “any trace of Masha”.
(c) Ваня не решил все задачи.
(d) Ваня не решил всех задач.
Exs. (c-d) may differ in scope, but not in intensionality.
43
Possible non-uniform analysis


There may be more than one way that the
‘reduced referentiality’ of Gen Neg NPs
comes about, licensed by different classes of
verbs. (Cf. multiple kinds of ‘Imperfective’
meanings.) The property-type idea may be
correct for a number of cases, but other
‘quantity’-based ideas may be better for other
cases.
Work in progress!
44