_________________________ Papers _________________________

A Note on Plural Logic

Gustavo Fernández Díez

Abstract: A distinction is introduced between itemized and non-itemized

plural predication. It is argued that a full-fledged system of plural logic
is not necessary in order to account for the validity of inferences con-
cerning itemized collective predication. Instead, it is shown how this
type of inferences can be adequately dealt with in a first-order logic
system, after small modifications on the standard treatment. The pro-
posed system, unlike plural logic, has the advantage of preserving
completeness. And as a result, inferences such as ‘Dick and Tony
emptied the bottle, hence Tony and Dick emptied the bottle’ are shown
to be first-order.
Keywords: compound terms, itemized predication, singular logic, plu-
ral logic.

1 Introduction

(1) Dick and Tony emptied the bottle. Hence, Tony and Dick
emptied the bottle.
Formalize and prove first-order validity.
It sounds like a routine exercise of a Logic course. But:
(a) It does not correspond to:
E(d,b) & E(t,b) ∴ E(t,b) & E(d,b)
since Dick and Tony’s joint emptying of the bottle is not equivalent to
a pair of emptyings, one by Dick and one by Tony.

Organon F 17 (2010), No. 2, 150 – 162 © 2010 The Author. Journal compilation © 2010 Institute of Philosophy SAS
A Note on Plural Logic __________________________________________________ 151

(b) Using a 3-place predicate letter F for ‘x and y emptied z ’, we

get
F(d,t,b) ∴ F(t,d,b)
which is not a valid inference of first-order logic.1
(c) If we then put e.g. G1 for ‘x is a emptying event’, G2 for ‘y
was emptied in event x ’ and G3 for ‘z and w were the ones
who emptied in event x ’, we will obtain
∃x ( G1(x) & G2(b,x) & G3(d,t,x) ) ∴ ∃x ( G1(x) & G2(b,x) & G3(t,d,x) )
which is not a first-order inference either.
(d) If we use the set {d,t} we will come to
E({d,t},b) ∴ E({t,d},b)
which is a correct inference of set theory, but again, not of pure first-
oder logic.2
(e) And finally, using the so-called ‘plural logic’ we can render
(1) as:

∃xx ( E(xx,b) & d ‫ ط‬xx & t ‫ ط‬xx & ∀y ( y ‫ ط‬xx → y = d ∨ y = t ) )

∴ ∃xx ( E(xx,b) & t ‫ ط‬xx & d ‫ ط‬xx & ∀y ( y ‫ ط‬xx → y = t ∨ y = d ) )
i.e.: ‘There are some things xx which emptied the bottle, and only Dick
and Tony are among them. Hence, only Tony and Dick are among
them’. Plural logic is a recent system of non-classical logic which al-
lows quantification over plural variables (like ‘xx’), intended to stand
for various objects at the same time. The point of plural logic is to take
plural expressions of natural language at face value, instead of singu-
larizing them into sets, groups, or aggregates of any kind.
Within plural logic, inference (1) obtains as a consequence of the ‘in-
discernibility’ (or ‘extensionality’) axiom for pluralities.3 However, al-

1 Besides obscuring the fact that the activity denoted by the 2-place letter E and the 3-
place letter F is essentially the same.
2 Besides the dubious implication that the bottle would have been emptied by a set
(cf. Boolos 1984, 72, one of the pioneering works of plural logic).
152 _________________________________________________ Gustavo Fernandéz Díez

lowing quantification over plural variables is not without cost. Plural

logic is a much stronger theory than classical first-order logic, and in
particular, it was soon discovered to be not recursively axiomatizable.4
So we are left again with our initial question: is (1) not a first-order
inference, after all?

2 The Logical Analysis of Plural Predication

A plural predication is that in which a predicate is attributed to a
plurality of subjects. It is called ‘distributive’ when the predicate ap-
plies to each of the subjects of the plurality taken one by one. For
example, ‘Dick and Tony are fair-haired’, which is tantamount to say-
ing that Dick is fair-haired and Tony is fair-haired. And a plural pre-
dication is called ‘collective’ when the predicate only applies to the
subjects of the plurality taken collectively. For example, in the case of
inference (1), just seen.5

3 Burgess (2004, 197); McKay (2006, Ch. 6); Linnebo (2009, §1.2).
4 Burgess (2004, 220); Oliver – Smiley (2006, 318); Yi (2006, 257).
5 We often find the collective/distributive distinction formulated as a differentiation
between the predicates themselves, as if each predicate carried within its meaning
an indication about whether it should be taken collectively or distributively, when
applied to a plural subject. According to such a view, a predicate P would be distri-
butive when it is an analytic truth that P applies to a plurality of things if and only
if P applies to each of them; and it would be collective otherwise (cf. Oliver – Smiley
2006, 322; and similarly in Linnebo – Nicolas 2008, 188; Linnebo 2009, §1.1). Howev-
er, we have to take into account that many predicates admit both readings. For ex-
ample, ‘lifted the table’ in the sentence ‘Dick and Tony lifted the table’, can be inter-
preted either collectively – they did it together – or distributively – they did it one
after the other. In cases like this the predicate in question would have to be re-
garded as a pair of different predicates, which is very artificial. And on top of it all,
such ‘mixed’ or ambiguous predicates turn out to be the most common ones in Eng-
lish (Winter 2002, 495; Yi 2005, 481).
Hence, it seems much more convenient to conceive the collective/distributive
distinction, as depending on the way in which the predicate applies to the plural
subject in order to form the statement in question. On this other view, the ambigui-
ty of a sentence such as ‘Dick and Tony lifted the table’ is regarded as a purely
structural ambiguity, just as a sentence of the form ‘p and q or r’ is ambiguous be-
tween its two possible readings, ‘(p & q)∨r’ and ‘p & (q∨r)’. The collec-
tive/distributive distinction should therefore be seen as dwelling on the structure of
the predication, rather than on the nature of the predicate used (a similar view is
A Note on Plural Logic __________________________________________________ 153

A much lesser known distinction is that between itemized and non-

itemized plural subjects. We shall say that the subject of a predication
is an itemized (or individualized) plural subject, when each of the ele-
ments which make up the plurality is individually referred to, or at
least the exact number of individuals which compose it is given. For
example: ‘Dick and Tony emptied the bottle’, ‘Dick and Tony are fair-
haired’, ‘Dick and somebody else emptied the bottle’, ‘two people
emptied the bottle’, or ‘two people are fair-haired’.
The opposite case is that of non-itemized (or non-individualized) plu-
ral subjects, in which there is no indication as to which or how many
individual members make up the plurality in question. Examples of
non-itemized plural subjects are: ‘various people emptied the bottle’,
‘various people are fair-haired’, ‘Dick and some other people emptied
the bottle’, and ‘Dick and some other people are fair-haired’.6
The difference between itemized and non-itemized plural predi-
cation suggests a more general distinction, between singular logics and
plural logics. A singular logic is that in which the individual (the singular
object) is the only vehicle to convey reference, predication and quantification.
A plural logic is one which admits genuine reference, predication and quanti-
fication over a plurality of objects, without reducing the plurality to a com-
posite unit of any kind. Of course, all plural logic proposals advanced so
far are ‘plural logics’ in this sense. And standard first-order logic is, in
this sense, a ‘singular logic’ beyond any doubt.
Moreover, it is clear that standard first-order logic does not enable
us to account for the logical validity of inferences regarding itemized
collective predication, as we have seen with the case of inference (1).
But in fact only a few modifications are needed to patch this up. The
resulting system will be a singular logic in the sense just given – in
fact, almost a variant of first-order logic as standardly formulated.
And it will preserve recursive axiomatizability.

defended in Moltmann 1997, 52; Cameron 1999, 129; Yi 2005, 481; McKay 2006, Ch.
1, Note 15; Fernández Díez 2006, 42).
6 This includes of course pluralities described by means of generalized quantifiers
(such as ‘most’, ‘nearly all’, ‘more than ten’, etc), unless the description happens to
be equivalent to stating an exact number of elements for the plurality in question
(as in ‘more than four and less than six’, for instance, which can only come out to
five).
154 _________________________________________________ Gustavo Fernandéz Díez

Therefore, such a system will be a considerable simpler and weaker

logical theory than plural logic. And consequently, if it does suffice for
the logical analysis of itemized plurals, it will serve as evidence for the
fact that the logical status of itemized plural predication is deep down
different – essentially simpler – from that of its non-itemized counter-
part.
This is by no means arguing against the pertinence of plural logic,
which has its natural field of action in the treatment of non-itemized
pluralities and non-itemized predication – for which it is the natural
option. The only thing I would like to suggest is that the introduction of
plural logic systems should not be motivated by examples of itemized collecti-
ve predication.7

3 The Logical Status of Predication over Itemized

Plurals
Inference (1) shows a typical feature of itemized plural predication:
in the absence of further indication, the order in which the subjects are
listed is immaterial. This occurs not only with collective predication,
but with distributive predication as well. For example:
(2) Dick and Tony are fair-haired. Hence, Tony and Dick are
fair-haired.
Another feature of plural predication, both collective and distributive,
is the immateriality of repetitions within the subjects listed:
(3) Dick and Tony emptied the bottle. Hence, Dick, Dick, and
Tony, emptied the bottle.

7 As it often occurs, e.g.: McKay (2006, Ch. 1): “Non-distributive predication. Standard
first-order logic does not provide adequate resources for properly representing
many ordinary things that we say: (1) Arnie, Bob and Carlos are shipmates”. Yi
(2005, 460): “The Fregean systems, Frege’s system and its descendants cannot deal
with the logic of the plural constructions (in short, plurals) of natural languages, such
as the following: [P1] Venus and Serena are tennis players, and they won a U. S. Open
doubles title”. Such examples are cases of itemized plural predication, and they are
easily treatable by means of the system which I will describe here.
A Note on Plural Logic __________________________________________________ 155

(4) Dick and Tony are fair-haired. Hence, Dick, Dick, and Tony,
are fair-haired.8
Another feature of plural predication, both collective and distribu-
tive, is substitutivity salva veritate:
(5) Dick and Tony emptied the bottle. Tony is Iron Man. Hence,
Dick and Iron Man emptied the bottle.
(6) Dick and Tony are fair-haired. Tony is Iron Man. Hence,
Dick and Iron Man are fair-haired.
And another is existential quantification over some of the subjects
of the plurality in question:
(7) Dick and Tony emptied the bottle. Hence, Dick and some-
body else emptied the bottle.
(8) Dick and Tony are fair-haired. Hence, Dick and somebody
else are fair-haired.
I claim all inferences (1) to (8) are logical inferences. As a matter of
fact, the even ones correspond to cases of distributive predication, and
can be easily accounted for in standard first-order logic by means of
the usual propositional disjunctions. Indeed (using H for ‘being fair-
haired’, and i for Iron Man):
(2a) H(d) & H(t) ∴ H(t) & H(d)
(4a) H(d) & H(t) ∴ H(d) & H(d) & H(t)
(6a) H(d) & H(t) ; t = i ∴ H(d) & H(i)
(8a) H(d) & H(t) ∴ ∃x ( H(d) & H(x) )
However, the odd inferences of this list ((1), (3), (5) and (7)) corre-
spond to cases of collective predication, and cannot be adequately
treated in standard first-order logic. To be sure, inferences (5) and (7)
could be validated by the introduction of the 3-place predicate letter F
mentioned in §1, but this does not work for (1) and (3).9

8 Assuming Dick being the same all along.

9 Besides the awkward implication that e.g. in ‘It was Dick and Tony who emptied
the bottle, and not Charles’ we would have to use two different predicate letters for
‘emptied’, thus: F(d,t,b) & ∼E(c,b) (cf. Note 1 again).
156 _________________________________________________ Gustavo Fernandéz Díez

I will now indicate how first-order logic can be adapted so as to

cope with these and similar inferences in a natural way. The resulting
system could be called ‘first-order logic for itemized collections’, or
‘first-order logic for itemized collective predication’.

4 First-order Logic for Itemized Collections: Syntax

To the usual individual terms of standard first-order logic, we add
compound terms. A compound term will be a discrete and linear list of
terms between square brackets, which behaving like a unit, acts itself
as a new term.10 Hence we will have:
(a) an individual variable is a term;
(b) an individual constant is a term;
(c) if r1 , …, rn are terms, then [r1, …, rn] is a term.11
Terms formed by clauses (a) and (b) are the usual individual terms;
terms formed by clause (c) will be called ‘compound terms’.
The inductive structure of the definition enables us to accommoda-
te nested structures of itemized collections. Like for example: ‘Dick
and Hannah, together with Tony and Grace, introduced John and Ali-
ce to swinging’:
(9) I ( [[d,h],[t,g]] , [j,a] )12
Next, the usual definition of formula does not need to be changed.
In particular, an atomic formula P(r1, …, rn) can be simply defined as

10 Just as in ‘Dick and Tony emptied the bottle’, the phrase ‘Dick and Tony’ takes the
argument place of the one who empties, with respect to the 2-place relation ‘x emp-
tied y’.
11 I take a language without descriptions and function symbols, for simplicity.
12 Where predicate letter I stands for the 2-place relation ‘x introduced to swinging to
y’. In this example both places of the relation are filled up by compound terms, the
first of which is made up itself of two compound terms. Such nested pluralities are
not normally accepted in plural logic treatments, although some authors have ad-
vocated for them (Rayo 2006; Linnebo – Nicolas 2008). Our treatment covers them,
but only in the context of itemized collections.
A Note on Plural Logic __________________________________________________ 157

an n-place predicate letter P followed by n terms. And quantification

will be allowed – as usual – over individual variables only.13
As for the deductive system, the usual quantification rules (or axi-
oms) will enable us to validate inference (7): ‘Dick and Tony emptied
the bottle. Hence, Dick and someone else emptied the bottle’:
E ( [d,t] , b ) ∴ ∃x E ( [d,x] , b )
And in order to cover the other odd inferences of our list (that is,
inferences (1), (3) and (5)), we introduce a Rule for the Indiscernibility of
Compound Terms (RICT), according to which two pluralities are equal
if and only if they have exactly the same elements:

[ r1, …, rn ] = [ s1, …, sm ]


(r1 = s1∨…∨r1 = sm) & … & (rn = s1∨…∨rn = sm) & (s1 = r1∨…∨s1 = rn)
& … & (sm = r1∨…∨sm = rn)

the double horizontal line meaning that the rule can be used in both
directions, downwards and upwards. In other words: that [r1, …, rn]
equals [s1, …, sm] if and only if every one of r1, …, rn equals at least one
of s1, …, sm, and vice versa.14,15
Indeed, applying RICT (in the upward direction) we obtain:

d=d t=t

( d = t ∨ d = d) & ( t = t ∨ t = d) & ( t = d ∨ t = t) & ( d = d ∨ d = t)

[d , t ] = [ t , d ]

13 Recall that our aim is that the proposed system constitutes a singular logic in the
sense given in §2. Hence the basic vehicle to convey reference, predication and
quantification must only be the individual object.
14 In the form of an axiom:
[r1, …, rn] = [s1, …, sm] ↔ ((r1 = s1∨…∨r1 = sm) & … & (rn = s1∨…∨rn = sm) &
(s1 = r1∨…∨s1 = rn) & … & (sm = r1∨…∨sm = rn))
15 Rule RICT is a sort of weaker counterpart of the indiscernibility axiom for full plural
logic systems (which we mentioned in §1), adapted to the case of itemized pluralities.
158 _________________________________________________ Gustavo Fernandéz Díez

from which inference (1) follows: E ( [d,t] , b ) ∴ E ( [t,d] , b ). 16

By a similar application of RICT we get:

d=d t=t

(d = d ∨ d = d ∨ d = t) & (t = d ∨ t = d ∨ t = t) & (d = d ∨ d = t)
& (d = d ∨ d = t) & (t = d ∨ t = t)

[d,t]=[d,d,t]

from which inference (3) follows: E ( [d,t] , b ) ∴ E ( [d,d,t] , b ).

And finally, again by means of upwards RICT we obtain:

d=d t=i i=t


( d = d ∨ d = i) & ( t = d ∨ t = i) & ( d = d ∨ d = t) & ( i = d ∨ i = t)

[d , t ] = [ d , i ]

from which inference (5) follows: E ( [d,t] , b ); t = i ∴ E ( [d,i] , b ).

5 First-order Logic for Itemized Collections: Semantics

Let ℑ be any interpretation of the formal language, including an
assignment of values to the variables. Then we give the denotation of
individual terms as usual. And next, we simply take the denotation of
a compound term [r1, …, rn] under ℑ to be the set formed by the deno-
tation under ℑ of each of its constituents. That is to say, we give the
interpretation of compound terms by means of the inductive clause
[r1, …, rn]ℑ = { r1ℑ, …, rnℑ }
Thus, itemized collections are represented as sets.17

16 So we have finally shown (1) to be a first-order inference, as we set out to do in §1.

17 Strictly speaking, the clause has to be spelt out by induction on the degree of com-
plexity of the term. Thus [r1, …, rn] has complexity 1 when all r1, …, rn are individual
terms (i.e., terms of complexity 0), and in that case the values r1ℑ, …, rnℑ are already
A Note on Plural Logic __________________________________________________ 159

This leads us to deal with two kinds of ‘entities’ made up from the
individual members of the interpretation ℑ: on the one hand, simple
entities, which will just be the individuals of U (where U is the univer-
se of ℑ); and on the other hand, complex entities, which will be finite
sets of individuals of U, or finite sets of finite sets of individuals of U,
or combinations of these, etc. etc. In more precise terms (by inducti-
on): every individual of U will be a simple entity (entity of complexity
0) of U; and if u1, …, un are entities of U, the more complex of which
has complexity k, then {u1, …, un} will be a complex entity of U of com-
plexity k+1. No need to say that ℑ will assign simple entities to indivi-
dual terms, and complex entities to compound terms.18
Likewise, we shall consider that a property defined on U is a set of
entities of U, and more in general, that an n-place relation defined on U
is a set of n-tuples of entities of U. Then the interpretation of an n-place
predicate letter P under ℑ (say Pℑ) will be an n-place relation on U,
i.e., a set of n-tuples of entities of U. So that an n-tuple of entities of U
will be said to satisfy P under ℑ, if and only if it belongs to the relation
(set) Pℑ. Thus, if P is for example a 2-place predicate letter, its interpre-
tation Pℑ will be a set of ordered pairs, pairs whose components do
not necessarily have to be plain individuals of the universe U (as it
occurs in the standard first-order logic treatment), but that can be
complex entities, as long as they are built up starting from the indivi-
duals of U only.19

given. And [r1, …, rn] has complexity k+1 when the most complex term among r1, …,
rn has complexity k, and in that case the values r1ℑ, …, rnℑ are taken for granted by
the inductive clause of the definition.
18 Complex entities are bound to be finite sets (of individuals of U, or of finite sets of
individuals of U, etc.), in fair accord with the syntactic structure of compound terms
in our language, which can only consist of finite lists of terms (cf. §4–(c)). It could
not be otherwise in a singular logic system, in which the only vehicle to convey ref-
erence is the individual object (cf. §2); and this is so, of course, given the impossibil-
ity of denoting an infinite plurality by making individual reference to each of its
elements. Restriction to finite sets, incidentally, eliminates any possibility of Rus-
sell’s paradox style arguments affecting our system.
19 Sets which have among its elements single individuals as well as other sets of dif-
ferent complexities, are indeed familiar from set theory. The same can be said of n-
tuples whose components are single individuals as well as sets and other n-tuples
of different complexities. This involves no risk of ambiguity: letting u, v and w be
160 _________________________________________________ Gustavo Fernandéz Díez

Then we can simply say that an atomic formula such P(r1, …, rn) is
true under ℑ if and only if the n-tuple (r1ℑ, …, rnℑ) belongs to the re-
lation Pℑ. And this will hold irrespectively of whether some of the
terms r1, …, rn turn out to be compound terms or not, or what amounts
to the same thing, irrespectively of whether some of the entities r1ℑ, …,
rnℑ turn out to be complex entities or not.
In our logical system a 2-place predicate such as ‘x emptied y’ can
be satisfied by an ordered pair whose two components are single in-
dividuals (as it happens for instance in ‘Charles emptied the bottle’),
but also by an ordered pair whose first component is not a single in-
dividual of the universe, but a complex entity composed of two of
them (as it happens in ‘Dick and Tony emptied the bottle’). Something
which is in perfect match with our intuition that the 2-place predicate
‘x emptied y’ (or, in other words, the relation corresponding to it) is
essentially the same in the case where it was a single person who
emptied the bottle, that in the case where it was two people together,
hand in hand, who carried out the emptying.
Similarly, in our system, a 2-place predicate such as ‘x introduced to
swinging to y’ can be satisfied by an ordered pair whose first and se-
cond components are complex entities of different levels (as it happens
for instance in ‘Dick and Hannah, together with Tony and Grace, intro-
duced John and Alice to swinging’). And a similar remark applies, mu-
tatis mutandis, to all other cases of plural collective predication.
This settings make it very easy to obtain proofs of soundness and
completeness for our proposed system. Indeed, soundness of rule
RICT follows at once from the extensionality of sets.
And as for completeness, only a few minor modifications are nee-
ded on the customary Henkin-style proof.20 Indeed, the usual proof
proceeds by taking any consistent set of formulas Φ, and enlarging it
into a maximal consistent and existentially saturated Ψ. Then an equi-

distinct objects, for instance, the set-theoretical definition of the ordered pair
([u,v],w) makes it clearly different from the pair (u,[v,w]):
([u,v],w) = ({u,v},w) = { {{u,v}} , {{u,v},w} }
≠ (u,[v,w]) = (u,{v,w}) = { {u} , {u,{v,w}} }
And a similar remark applies to n-tuples in general (see e.g. Roitman 1990, 29 – 30).
20 As exposed e.g. in Chang – Keisler (1990, 61 – 66).
A Note on Plural Logic __________________________________________________ 161

valence relation between terms is introduced, in such a way that r∼s if

and only if the formula r = s belongs toΨ. A model is thereby defined
that assigns to each term its own equivalence class, which makes it re-
latively easy to prove that such a model satisfies the whole ofΨ, hence
rendering Ψ (and Φ ) satisfiable. And from this satisfaction theorem (if
Φ consistent, Φ satisfiable) the completeness theorem follows at once.
All this can be done just the same in our first-order logic for itemi-
zed collections. In particular, the equivalence relation will be esta-
blished not only for individual terms, but for compound terms too.
And the fact that the set Ψ is maximal consistent, immediately guaran-
tees that it contains a formula of the form [r1, …, rn] = [s1, …, sm] if and
only if, for every ri among r1, …, rn it contains at least a formula ri = sj
with sj among s1, …, sm, and vice versa.21 From which satisfaction by
the constructed model will be ensured.
Hence we can conclude: (a) that our proposed system is a singular
logic in the sense given in §2, in that the individual is the essential
vehicle to convey reference, predication and quantification; (b) that,
unlike full-fledged plural logic formal systems, it preserves complete-
ness; and (c) that it suffices to account for the logical validity of infe-
rence (1) and related inferences, showing them to be first-order.22

Department of Philosophy
University of Murcia
P.O. Box 4021
E–30071 Murcia
Spain
[email protected]

References
BURGESS, J. P. (2004): E Pluribus Unum: Plural Logic and Set Theory. Philosophia
Mathematica 12, No. 3, 193 – 221.

21 This is of course in virtue of RICT.

22 Research supported by the Spanish Government Grants FIS2007–63830 and
FFI2009–13687–C02–01. This paper has greatly benefited from the comments of two
anonymous reviewers of Organon F, to which I express my sincere gratitude.
162 _________________________________________________ Gustavo Fernandéz Díez

BOOLOS, G. (1984): To Be is to Be the Value of a Variable (or to Be Some Values

of Some Variables). Journal of Philosophy 81, No. 8, 430 – 449. Quoted here
by the reprint in Boolos, G. Logic, Logic and Logic. Cambridge (MA): Har-
vard University Press, 54 – 72.
CAMERON, J. R. (1999): Plural Reference. Ratio (new series) 12, No. 2, 128 – 147.
FERNÁNDEZ DÍEZ, G. (2006): El análisis lógico de la predicación colectiva. Críti-
ca, Revista Hispanoamericana de Filosofía 38, No. 114, 21 – 44.
CHANG, C. C. – KEISLER, H. J. (1990): Model Theory (3rd edition). Amsterdam:
Elsevier.
LINNEBO, Ø. (2009): Plural Quantification. In: Zalta, E. N. (ed.): The Stanford
Encyclopedia of Philosophy (Spring 2009 Edition), URL =
<http://plato.stanford.edu/archives/spr2009/entries/plural-quant/>.
LINNEBO, Ø. – NICOLAS, D. (2008): Superplurals in English. Analysis 68, No. 3,
186 – 197.
MCKAY, T. J. (2006): Plural Predication. Oxford: Clarendon Press.
MOLTMANN, F. (1997): Parts and Wholes in Semantics. Oxford: Oxford Universi-
ty Press.
OLIVER, A. – SMILEY, T. (2006): A Modest Logic for Plurals. Journal of Philosophi-
cal Logic 35, No. 3, 317 – 348.
RAYO, A. (2006): Beyond Plurals. In: Rayo, A. – Uzquiano, G. (eds.): Absolute
Generality. Oxford: Clarendon Press, 220 – 253.
ROITMAN, J. (1990): Introduction to Modern Set Theory. New York: John Wiley &
Sons.
WINTER, Y. (2002): Atoms and Sets: A Characterization of Semantic Number.
Linguistic Inquiry 33, 493 – 505.
YI, B-U. (2005): The Logic and Meaning of Plurals. Part I. Journal of Philosophi-
cal Logic 34, No. 5 - 6, 459 – 506.
YI, B-U. (2006): The Logic and Meaning of Plurals. Part II. Journal of Philosophi-
cal Logic 35, No. 3, 239 – 288.