1 Nandini Bhattacharya 16.01.2014 M.Phil in Linguistics. Sem –I P-2 Term Paper Semantics of Quantification in Bangla :- Abstract: Quantifications over individuals or objects are sometimes explicit or implicit in Natural languages. The question of how the quantification is encoded in various languages has been widely debated in formal semantics. The traditional view in Generalized Quantifier Theory is that the Noun Phrases in Natural Languages behave like quantified expressions and the quantifiers act like determiners. The existing literature mostly probes into the English data and there is a lot more need for cross-linguistic data to validate the Theory and examine some language specific phenomena concerning NP quantification. This paper, proposes a uniform survey of the existing literature highlighting the key points. The Paper also provides an adequate description of the quantifiers in Bangla. Together with that, this paper presents some analysis regarding the formalizations of some of the non-logical and partitive quantification in Bangla. Moreover, the last section of the paper examines the scopal interaction of quantificational NPs with Negation in Bangla. The analysis focuses on the semantics of the Noun Phrase quantification following Generalized Quantifier Theory, and sheds light on the language specific features of quantification in Bangla.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Nandini Bhattacharya 16.01.2014
M.Phil in Linguistics. Sem –I
P-2 Term Paper
Semantics of Quantification in Bangla :-
Abstract: Quantifications over individuals or objects are sometimes explicit or implicit
in Natural languages. The question of how the quantification is encoded in various
languages has been widely debated in formal semantics. The traditional view in Generalized
Quantifier Theory is that the Noun Phrases in Natural Languages behave like quantified
expressions and the quantifiers act like determiners. The existing literature mostly probes
into the English data and there is a lot more need for cross-linguistic data to validate the
Theory and examine some language specific phenomena concerning NP quantification. This
paper, proposes a uniform survey of the existing literature highlighting the key points. The
Paper also provides an adequate description of the quantifiers in Bangla. Together with that,
this paper presents some analysis regarding the formalizations of some of the non-logical
and partitive quantification in Bangla. Moreover, the last section of the paper examines the
scopal interaction of quantificational NPs with Negation in Bangla. The analysis focuses on
the semantics of the Noun Phrase quantification following Generalized Quantifier Theory,
and sheds light on the language specific features of quantification in Bangla.
2
1. Introduction :-
There are many semantic theories that propose methods for
the proper interpretation of quantified Noun Phrases in natural language. The Generalized
quantifier theory has been an influential sub-field of semantic theory of all. David Lewis
(1970) , first proposes that noun phrases should be interpreted as properties of properties
rather than as in first order predicate logic. His idea was based on the set theoretic notion of
set-subset and set membership and treated Noun Phrases as Generalized quatifiers. After
Lewis, the Generalized Quantifier theory (abbre. GQ Theory), has been developed by
Barwise and Cooper (1981), to the level of Universals. They suggest that all noun phrases
denote generalized quantifiers. They theorizes that the Quantifier Phrase (QP) is formed by
Quantifier determiners combined with Nominal arguments of type <e,t> . They also
proposed weak-strong distinctions of quantifiers. Carlson’s papers (1980) focus on the issue
of the treatment of the English bare plurals and how they interact with quantification.
Partee (2004) applies the semantic type theory and type-shifting principles to reconcile the
Generalized Quantifier Theory with the ‘non-uniform’ NP semantics proposed by Heim and
Kratzer (1998). This tradition has been further put forward by another notable semanticist
such as, Anna Szabolsci (2010).
There is a different between the accounts of the Syntax
and Semantics of quantification. The syntax of quantification has been proposed by kennedy
(1997) and Szabolcsi (1999/2001, 2010). The semantics of quantification is proposed by
Heim and Kratzer (1998) ,Chierchia Gennaro and Sally McConnell-Ginet ( 2000) and
Anastasia Giannakidou and Monika Rathert (2009). By emphasizing, the significance of the
GQ theory, Anastasia Giannakidou and Monika Rathert (2009) state that, the Generalized
Quantifier Theory has motivated an extensive and fruitful research agenda, since 1980s to
the current time. They point out that the framework of GQ Theory has featured in
“extensive studies of quantificational structures, with attention to the constituents of QPs,
and their scopal properties.” They have emphasized on the cross-linguistic study of
quantification in natural languages. There have also been notable researches in the syntax-
semantics interface of quantification as well.
3
2. Overview of Literature :-
Modern quantifiers were first introduced by Montague
(1973). Thereafter, the treatment of quantification has been a most discussed area of
Formal semantics. Following are the overviews of some of the eminent existing literature of
Generalized Quantifier theory in natural language semantics.
2.1. Barwise and Cooper (1981/2002):-
According to Barwise and Cooper, the quantifiers of first order
logic are not adequate to properly treat the quantification in Natural languages. There are
natural language sentences that can’t be symbolized using Restricted quantification.
Barwise and Cooper argues that the syntactic structure of quantified sentences in predicate
calculus is different from the syntactic structure of the quantified sentences in natural
language. In the article, Barwise and Cooper, discusses the notion of Generalized
Quantification and formalizes it. They propose a detailed analysis of the possible
implications of Generalized Quantifier theory of natural language. They attest their theory
with appropriate examples from English language data. In the article, Barwise and Cooper,
cites some English quantified NPs, such as ‘more than half’, ‘most of’, ‘no one’, ‘only one’
etc, and reasons why these quantifiers can’t be treated appropriately using traditional first
order logic. They argues that, if ‘E’ is an arbitrary non-empty set of things , first order logic
only allows quantification over objects in E, it doesn’t permit the quantification over the
arbitrary sets of things. So, Generalized quantifier theory in Model-theoretic Semantics,
provides a way to treat and formalize the determiners like ‘most’, ‘many’, ‘few’ etc.
Barwise and Cooper suggests, that in a sentence like “More
than half *emphasis mine+ of John’s arrows hit the target”, ‘more than half’ doesn’t behave
like a quantifier but like a determiner. This determiner combines with a set expression (i.e.
set of John’s arrows) to produce a quantifier. Barwise and Cooper argues that this structure
of the logical quantifiers, such as ‘more than half’, ‘most’ etc, corresponds in a similar way
to the structure of English NPs. The determiner combines with the set of things in a
quantified NP and the whole NP is the quantifier, for example, “most people”, “more than
4
half of John’s arrows’ etc. Barwise and Cooper terms these kinds of quantifiers as “non-
logical quantifiers”. They also argues that the truth value of these quantified sentences will
not depend upon ‘a priori’ logic but will depend on the ‘underlying measure of infinite sets
that is one is using’, which at the same time needs to be included in the Model.
Barwise and Cooper argues that there should be a fixed set of
contexts that determines the meaning of the basic expressions in the quantified sentences.
They propose this assumption as “fixed context assumption”. So the interpretation of the
non logical quantifiers, like ‘most’, ‘may’, ‘more than half’, ‘few’ etc will depend on the
Model and will vary from model to model. According to Barwise and Cooper that the
interpretation of the logical quantifier ‘Every’ (∀ ) remains same in every model (M).
According to Barwise and Cooper, the quantifiers are used to denote the property of a set,
for example, the Existential quantifier (∃) asserts that the set of individuals (x) have the
property which contains at least one member. The Universal quantifier (∀ ) asserts that the
set contains all the individual. Barwise and Cooper argues that a quantifier divides up the
family of sets provided by the model (M). When the quantifiers are combined with some
sets it will produce the truth value (T) and with combing with other sets will produce the
truth value (F). Barwise and Cooper argues that the denotation II Q II, of a quantifier symbol
Q, can be formalized as follows,
II ∃ II = { X ⊆ E I X ≠ ∅ } [ E = set of entities provided by M]
II ∀ II = { ∃ }
II Finite II = { X ⊆ E I X is finite }
II More than half of N II = { X ⊆ E I X contains more than half of the Ns }
II Most N II = { X ⊆ E I X contains most Ns }
Therefore, the quantifiers functions from set of things/individuals to the property of a non-
empty set.
5
According to Barwise and Cooper, the predicate calculus
can’t adequately translate the subject NPs with non logical quantifiers. So they propose a
simple formalization. For example, they cites the sentence, “Most babies sneeze”, and
formalizes it as , (Most babies)(sneeze). The sentence will have the truth value T, if the set
of sneezers contains most babies. Barwise and Cooper argues that proper names within a
NP also acts as a quantifier. The NP, with a proper name represents a family of sets,
containing that particular individual denoted by the proper name. Barwise and Cooper
proposes that the semantics of L(GQ) or Generalized quantifier logic, will be defined so that
‘thing’ always denotes the set E of things in the Model (i.e. the set of individuals or objects).
They also proposes the syntactic formation rules for L(GQ) using the logical symbols.
Barwise and Cooper, proposes the semantic analysis of
L(GQ) by using the set theoretic notions. They formalizes some of the English quantifiers,
such as, ‘some’, ‘every’, ‘no’, ‘both’, etc. According to Barwise and Cooper,‘ II some II is the
function which assigns to each A ⊆ E the family, II some II (A) = { X ⊆ E I X ∩ A ≠ 0 - ‘. They
explains their formalism as, for each of the determiner D, II D II (A) is a family of sets Q with
the property that X ∈ Q iff (X ∩ A) ∈ Q. They describe this property of II D II by proposing
that the quantifier II D II (A) lives on A. They proposes the universal semantic feature of the
determiners that they have the ability to assign to any set A a quantifier, that is the family of
sets, which ‘lives on A.’ To give evidence of this, Barwise and Cooper provides examples of
the applications of the logic of the GQ Theory to the fragments of English syntax.
Barwise and Cooper proposes some Semantic
Universals, relating to the Generalized Quantifier Theory. They proposes that, “ Every
natural language has syntactic constituents (called noun-phrases) whose semantic function
is to express generalized quantifiers over the domain of discourse.” They argues that, apart
from that it is the whole NP who participate into scope relationships in a proposition. They
proposes another Semantic Universal, “If a language allows phrases to occur in a dislocated
position associated with a rule of variable binding, then at least NPs (i.e. the syntactic
category corresponding to quantifiers over the domain of discourse) will occur in this
position.” After that, Barwise and Cooper proposes the Determiner Universal and states
that ,”Every natural language contains basic expressions, (called determiners) whose
6
semantic function is to assign to common count noun denotations (i.e. , sets) A quantifier
that lives on A. They cites the example, ‘Many men run ↔ Many men are men who run.’
Here, the quantifier ‘many’ lives on the set of ‘men’.
Barwise and Cooper draws the distinction between the
‘weak’ and ‘strong’ quantifiers. According to them, the weak quantifiers are the followings:-
‘a’, ‘some’ , ‘one/two/three etc’, ‘many’, ‘a few’, ‘few’, ‘no’ , and the strong quantifiers are
the followings :- ‘both’, ‘all’, ‘every’, ’each’, ‘most’, ‘neither’ etc. They also proposes the
‘Monotone increasing’ and ‘Monotone decreasing’ quantifiers and formalizes the
‘Monotonicity correspondence Universal’ and ‘Monotonicity Constraints’. They attests their
arguments with English data. Barwise and Cooper concludes by critiquing Montague’s
(1974) Quantifier Theory. So, therefore, Barwise and Cooper paved the pathways for a
greater grounding of The Generalized Quantifier Theory.
2.2. Carlson (1980) :-
According to Carlson, the different meanings of the
English bare plural arise because of the manner in which the context of the sentence
interacts with the bare plural NPs. Similarly we can argue that, the set readings of the
quantifiers such as, ‘some’, ‘many’, ‘most’, ‘few’, ‘all’ etc in English, differs. These
differences arise because of the way in which the context or presupposition set of the
sentence interacts with the quantified expressions and the set NPs. Carlson argues that the
interactions of quantified NPs with negation and bare plurals depends on their relative
scope properties. Carlson proposes that the NPs, such as, ‘Any dog’, ‘All dogs’, ‘Every Dog’
and ‘Each dog’ , all can be formalized as , (∀x) (Dog (x) ) , using the first order logic. He
argues that though these quantifiers are quite distinct from each other, it is plausible to
represent all of them in a unified manner. He suggests that a bare plural is also nothing but
the singular form addition to the quantifier ‘every’, which can also be represented as a
quantified NP. According to Carlson a ‘generic’ and ‘existential’ quantifier can be used to
formalize a bare plural. Later in his thesis, Carlson examines the interaction between the
7
determiner with the bare plurals and also studies some set-theoretic approaches to analyse
the bare plurals in English.
2.3. Partee (2004):-
In her article, ‘Many Quatifiers’, Partee analyses the treatment
of ‘many’ and ‘few’ in Formal semantics. According to Barbara Partee, the quantifier ‘ many’
and ‘few’, are ambiguous. She argues that these quantifiers are vague between their
‘cardinal’ and ‘proportional’ readings. Partee critiques Montague’s treatment of ‘many’ and
‘few’ as context-dependent quantifiers only and also highlights other proposals in the
literature in relation to the quantifiers ‘many’ and ‘few’. Partee argues that ‘many’ is like
‘every’ and ‘most’ on the proportional readings. She suggests that the quantificational cases
are almost paraphrasable by partitives. The only difference between them can be that the
restrictor clause of the quantifiers is open-ended set and the partitives involve a definite set.
Partee analyses the behaviour of ‘many’ and ‘few’ by formalizing their cardinal, proportional
and generic readings and sheds lights on some potential research issues in this area.
2.4. Chierchia Gennaro and Sally McConnell-Ginet ( 2000) :-
According to Chierchia and McConnell-Ginet, the
quantificational expressions introduce the power to convey generalizations into natural
languages. The quantifiers express the quantity of the individuals in a given domain (F1)
have a given property. They analyses the quantifier logic in the truth-conditional semantic
theory framework. Chierchia and McConnell-Ginet cites an example, “Everyone likes Loren.”
and states that the sentence is uttered many times, each time pointing at a different
individual until each individual in that domain has been pointed at. They argues that,
relative to that “pointing”, each of the sentences can be assigned the truth value of whether
8
they are true or false. If any single individual in that particular domain doesn’t like ‘Loren’
then the proposition yields the truth value F.
According to Chierchia and McConnell-Ginet, the
quantified expressions denote how many different values of the set of entities we have to
consider. They argue that the quantified expressions are the generalizing component. They
propose that the quantifier sentences are built out of sentences that contain different
variable set. Chierchia and McConnell-Ginet, argues that quantification has two
components, one contains the ordinary attribution of properties to referents/entities, and
another contains an instructions of how many such referents should have those properties.
They proposes a formulae, ∃x3Q ( x3 ), in which, the variable x3 is bound. Chierchia and
McConnell-Ginet, proposes that “an occurrence of a variable xn is syntactically bound iff it is
c-commanded by a quantifier coindexed with it.”, the ‘coindexed’ quantifier here refers to
∃xn and ∀xn etc.
Chierchia and McConnell-Ginet hypothesize a syntactic
account of the quantification in relation to c-command and scope interaction. They propose
that an occurrence of xn is syntactically bound by a quantifier, such as, Qn iff Qn is the lowest
quantifier which c-commands xn . They also provide with a semantic account of the
quantification. They suggests that it is part of the semantics of the Pronouns that they can
refer to any individual at all time in a given set , and also can be used with quantifiers to
denote something general about such a set. Chierchia and McConnell-Ginet also presents an
interpretation of the predicate calculus by using independent ‘value’ assignment functions.
They argues that the Models for the predicate calculus are made up of two things, first is a
specification of the sets or the domain of discourse and second a specifications of the
extensions of the language constants. They propose the structure of the Model for the
predicate calculus in semantics.
According to Chierchia and McConnell-Ginet, the
quantification in predicate calculus and the quantification in natural language are
connected. They applies the predicate calculus model to English quantificational NPs. They
argues that in care of English quantificational NPs, there is a presupposition set that
determines the truth-conditionality of the proposition. They also argues that in English, the
9
quantifying expressions such as , ‘every’, ‘some’ are always accompanied by some nominal
expressions, that restricts the universe of discourse to a specific set of individuals. Chierchia
and McConnell-Ginet proposes a grammar of English F2 and illustrates its association with a
certain class of English quantified sentences, for example ‘every man is hungry’ etc. They
analyses the scope taking and binding phenomena in the syntactic structure of the QP
which parallels the compositionality of the semantic representation of the natural language
Quantification. Chierchia and McConnell-Ginet concludes by formalizing the interaction
between the predicate calculus and the logical operator ‘if’ in the LF representation. .
Chierchia and McConnell-Ginet also highlight the significance of the Generalized Quantifier
approach (sets of sets). They argues that GQ Theory provides a compositional semantics for
NPs, it allows to bring out the cross-categorical nature of the logical words, such as, ‘and’,
‘or’ etc, it provides a precise criteria for NPs that characterize the distribution of Negative
Polarity Items and it gives an explanation for a substantial universal characteristics of
natural language determiners.
2.5. Szabolsci (2010) :-
According to Szabolcsi, the quantified expressions of the logical
language are different from the quantification in natural language. The syntax of logical
language specifies how the quantifier operator combines with expressions to yield new
expressions, and the semantics specifies their effects. Szabolcsi argues that the scope of an
operator in logic results from the constituent that it is attached to. But in natural language
one has to distinguish between semantic scope and syntactic domain. Szabolcsi argues that
the scope of a quantifier A is the property that is asserted to be an element of A on a given
derivation of sentences. If the property A incorporates another property B, such as,
quantifier, negation, model, then A automatically takes scope over B. Szabolcsi argues that
natural language quantifies over times and worlds in a syntactically explicit manner.
Szabolcsi addresses some issues in Generalized Quantifier Theory that poses problem of
analysis. However, Szabolcsi, argues that those problems arise due to the absence of fully
articulated compositional analysis. Szabocsi states that, “ GQ theory can accommodate so-
10
called referential indefinites, non distributive readings of plurals and conjunctions, and
type multiplicity, and it could adopt the stipulation that the “topics sets” of all GQs are
presupposed to be non-empty.” Szabolcsi analyses the scope interaction and the behaviours
of quantified NPs cross-linguistically.
2.6. Syntax of Quantification :-
Anna Szabolcsi (2001), analyses the Theory of Quantifier
Scope in this article. Szabolcsi suggests that the scope of an operator such as, a quantifier, is
the domain within which the operator has the potentiality to affect the interpretation of
other expressions. For example, ‘every boy’ affects the interpretation of the predicte by
inducing referential variation. She argues that the notion of scope is similar in both logic and
logical syntax. Szabolcsi argues that the surface structure S, directly determines the scope
interactions between different operators, such as, quantifiers, pronouns, negative polarity
items etc. She presents an analysis of the scope interactions of the quantifier phrases in
natural languages. According to Szabolcsi, “if QE/1 is in the domain of QE/2 but not vise
versa, QE/1 must take wide scope. If both are in the domain of the other, the structure is
potentially ambiguous. If neither QE is in the domain of the other, they must be interpreted
independently.” (here, QE stands for Quantified Expressions).
Szabolcsi reviews Montague (1974) in order to analyse the
rules of quantification in abstract syntax. Szabolcsi also reviews the ‘Quantifier Raising
‘phenomena in natural languages and tries to draw an affinity between these two
approaches. She reviews the syntactic theories of scope reading from 1980s to the 1990s
Minimalism approach. She critiques those theories by stating that those theories only deal
with the syntax of quantifiers, such as ‘every’ and ‘some’. Szabolcsi proposes that the
varying scope of indefinite quantifiers need to be attributed to unselective binding. She
11
argues that the indefinites should be treated as variables. Szabolcsi concludes by critiquing
whether the spell-out syntax is sufficient to capture the scope relationship of indefinite
quantifiers and whether the cross linguistic data uniformly supports such interpretations.
Christopher Kennedy (1997), proposes an account of
Antecedent –contained deletion principle and analyses its relation to the syntactic theory of
Quantifier Raising. Kennedy argues that the matrix reading of the embedded ACD is present
in the Quantifier Raising account only if we assume that QR can move quantified DPs out of
non-finite clauses. He states that the wide scope interpretations of embedded quantifiers
are parallel to the matrix readings of embedded ACD. Kennedy suggests that the principles
that force the LF movements of lexical materials also force the PF movement, that is
interpreted at the interface level. According to him, quantifiers impose two requirements on
the sentence structure, firstly, a quantifier must bind a variable and secondly, nominal
quantification in natural language is restricted. Kennedy argues that the ACD, which show
that the Quantifier Raising selects both a quantificational determiner and its restriction,
gives evidence of the presence of this relationship at LF, as a relation between a head and
its complement. Kennedy argues that both the ACD analysis and Quantifier Raising Theory is
compatible with each other and presents the syntactic representation of the Quantifier in
the framework of the Quantifier Raising Theory..The syntactic representation of the
Quantification, that Kennedy proposes, is restricted quantification unlike the claims of
unrestricted quantification in natural language.
12
3. Survey of Bangla Quantifiers :-
Following is a description of the quantified expressions in Bangla, and
the Structure and behaviour of the Quantificational Phrases in Bangla. The various nominal
quantifiers and some adjectival and adverbial quantifiers are discussed.
3.1. Nominal Quantifiers :-
There are many quantifiers in Bangla. The Universal quantifier in
Bangla are the followings:- /sɔb/ (every), /sɔbai/ ( everyone), /sɔb kɪchu/ (everything). In
Bangla sentences, /sɔb/ quantifies over both animate and inanimate, (± count) NPs. In
Bangla, the meaning of / kɪchu / is ‘some’, which when occurs after / sɔb / denotes
inanimate object N. Followings example sentences show the occurrence of /sɔb/ in various
contexts:-
i. sɔb chele -ra boi porche
every boy -plu.ani. Book read.Pres.con.
“Every boy is reading”.
ii. sɔb pʰul -gulo lal
every flower -plu.inani. red.
“ Every flower is red.”
In the above sentences the set of entities/objects quantified by the Universal quantifier, /
sɔb / is restricted by the contexts.
The quantifiers in Bangla can be classified into + count
quantifiers, - count quantifiers, and (± count) quantifiers. The + count quantifiers in Bangla,