Lasersohn, P. (2011) “Mass Nouns and Plurals,” in Claudia Maienborn, Klaus von
Heusinger and Paul Portner (eds.) Semantics: An International Handbook of Natural
Language Meaning, Volume 2, Berlin: de Gruyter, 1131-1153.
1131
46. Mass nouns and plurals
1. Introduction
2. Issues in the denotation of mass and plural NPs
3. Issues in the denotation of mass and plural DPs
4. Collective and distributive readings
5. References
Abstract
Mass and plural expressions exhibit interesting similarities in distribution and interpretation,
including cumulative reference, the ability to appear bare, and a parallel alternation
between existential and generic readings.They also exhibit important differences in agreement,
determiner choice, and in the types of quantification available. Major approaches to
plural denotation make conflicting claims whether plurality involves reference to collective
objects such as sets or mereological sums,or instead requires simultaneous saturation of an
argument place by multiple individuals. Theories of mass denotation differ as to whether
the count/mass distinction is a difference in discrete vs. continuous denotation, reference to
objects vs. the material they are composed of, or reference to mereological sums vs. classes
of individuals. Bare plurals and mass nouns sometimes denote “kinds”; there is disagreement
whether they also have an indefinite reading.Several kinds of plural and mass quantification
can be distinguished, depending on determiner choice, predicate modification, and
the use of a classifier or measure phrase. Plural quantifiers may interact to give a “cumulative”
reading, in which the quantifiers are scopally independent. Sentences containing
plurals sometimes exhibit an ambiguity between collective and distributive readings; the
number of readings and mechanisms for producing them is in dispute.
1. Introduction
Many—perhaps all—languages draw a distinction between mass nouns, prototypical
examples of which denote homogeneous substances such as water or gold, and count
nouns, prototypically denoting discrete, bounded objects such as people or chairs. Likewise,
many languages distinguish between singular nouns, which refer to single objects,
and plural nouns, which refer to multiple objects collectively. (Some languages distinguish
additional categories such as dual or paucal.) In this article, we survey a variety of
issues related to the count/mass and singular/plural distinctions.
1.1. Parallels between plural and mass expressions
We discuss mass and plural nouns together because they show interesting similarities.
Both exhibit cumulative reference (Quine 1960: 91); licensing inferences like those
in (1):
(1) a. A is water and B is water; therefore A and B together are water
b. A are apples and B are apples; therefore A and B together are apples
von Heusinger, Maienborn and Portner (eds.) 2011, Semantics (HSK 33.2), de Gruyter, 1131–1153
1132 IX. Noun phrase semantics
Singular count nouns do not license the same kind of inference; (2) is invalid:
(2) A is an apple and B is an apple; therefore A and B together are an apple
Singular count nouns instead exhibit divided reference; as Quine puts it,“To learn ‘apple’
it is not sufficient to learn how much of what goes on counts as apple; we must learn how
much counts as an apple, and how much as another.”
In addition, mass and plural nouns may appear (in English) with no overt determiner,
while a determiner is normally required for singular count nouns:
(3) I see water/horses/*horse
To the extent that *I see horse is acceptable, it involves either a conversion of horse
from a count noun into a mass noun, or a special “telegraphic” style of speech in which
determiners are omitted generally.
Determinerless (or“bare”) mass and plural noun phrases also show a parallel alternation
in interpretation, depending on the predicate with which they combine (cf. article
44 (Dayal) Bare noun phrases). If the predicate is stage-level (Carlson 1977a,b), the noun
phrase is understood as existentially quantified; (4a,b) are roughly equivalent to Some
water leaked into the floor and Some raccoons were stealing my corn:
(4) a. Water leaked into the floor
b. Raccoons were stealing my corn
If the predicate is individual-level,the sentence is understood as drawing a generalization
about objects of the kind picked out by the mass or plural noun:
(5) a. Water is wet
b. Raccoons are sneaky
If the predicate is kind-level, the mass or plural noun is understood as referring
to a “kind” of object, and the predicate is applied to this kind collectively, as a
whole:
(6) a. Water is common
b. Raccoons are extinct
Parallels such as these have led many semanticists to treat plural and mass expressions
together as “non-singular,” or even to identify mass nouns with lexical plurals
(Chierchia 1998a,b). But a completely unified analysis would seem to be impossible,
because mass nouns also show obvious differences from overtly plural nouns,
notably in their inability to combine directly with numerals and their selection by other
determiners:
(7) a. two horses/*water
b. many horses/*water
46. Mass nouns and plurals 1133
c. much *horses/water
d. few horses/*water
e. little *horses/water
1.2. Issues in what is meant by mass and plural
The use of the term mass in its technical sense in semantics appears to originate with
Jespersen (1913, 1924). Count is considerably more recent than mass; the earliest occurrence
I know of is in the anonymous (1952) Structural Notes and Corpus; the term was
popularized by Gleason (1955). However, earlier authors did employ comparable terms
such as thing-words (Jespersen 1913), bounded nouns (Bloomfield 1933), or individual
nouns (Whorf 1941). Jespersen characterized “mass-words” as “words which represent
‘uncountables’, i.e., which do not call up the idea of any definite thing, having a certain
shape or precise limits” (1913: 114), in contrast to thing-words, which represent countable
objects. He went on to note various syntactic differences between mass-words and
thing-words; but reference to countable or uncountable objects seems to have been the
defining distinction.
Jespersen was careful to note that the mass-word/thing-word distinction cross-cuts the
distinction between “material” and “immaterial” words, and cited this feature of his terminology
as providing an advantage over Sweet’s (1892) earlier classification into “class
nouns” and “material nouns.”Abstract nouns such as progress, admiration, or safety were
categorized as mass-words.
Bloomfield (1933: 205) partially continued Jespersen’s terminology, but distinguished
“mass nouns” from “abstract nouns,” placing both under a more general heading of
“unbounded nouns,” in opposition to “bounded nouns.” Many authors have continued
Bloomfield’s narrower use of the term mass, so that abstract nouns are excluded.
A related issue is whether to include words such as furniture and footwear as mass.
These pattern syntactically with ordinary mass nouns, combining with much rather than
many, failing to combine directly with numerals, etc.; but they hardly fit Jespersen’s
characterization as not calling up the idea of a “definite thing, having a certain shape
and precise limits.” An observation due to Roger Schwarzschild is that these nouns
admit modification with “stubbornly distributive” predicates of shape and size, unlike
prototypical mass nouns:
(8) a. This furniture is small
b. *This water is small
The use of the term mass was imported from linguistics into philosophy by Quine (1960),
and although Quine was careful to stress that the distinction between count and mass
terms was not in the “stuff” they denote, but only in whether they show cumulative or
divided reference, much of the subsequent philosophical literature has construed mass
so narrowly as to include only those words which serve as names for physical substances,
and not nouns like furniture or admiration.But many authors use mass in a broader sense
and distinguish substance nouns as a special subclass. This variation in what is meant
by mass leads some writers to eschew the term entirely, preferring non-count as more
clearly including a broader set of examples (Payne & Huddleston 2002, Laycock 2006).
1134 IX. Noun phrase semantics
Another point of variation is in whether mass should be understood to include some
morphologically plural examples. Jespersen argued that a wide range of plural nouns
were actually mass, including examples such as victuals, brains (as in blow out somebody’s
brains), dregs, proceeds, blues, creeps, and others. These impose plural agreement
on the verb, but combine with much rather than many:
(9) a. In this kind of work, brains are less important than guts
b. It doesn’t take much brains to figure this out
Here again Bloomfield (1933) introduced a shift in terminology, stipulating that mass
nouns“have no plural,”without discussing Jespersen’s examples;the idea that mass nouns
are always singular has been part of conventional wisdom ever since. Plural mass nouns
have been periodically rediscovered (McCawley 1975, Gillon 1992), and are treated in
detail in Ojeda (2005).
Another complication is that a single form may sometimes be used as a mass noun, and
sometimes as a count noun. Beer is ordinarily mass, but may be used as a count noun to
refer to individual servings of beer or kinds of beer; many other mass nouns show a similar
alternation. Conversely, a count noun may also be used as a mass noun if one imagines the
objects it denotes being put through a “universal grinder” (Pelletier 1975); after putting a
steak (count) through the grinder,“there is steak all over the floor” (mass).
There is much less variation in what semanticists mean by plural than there is with
mass, but even here there are some complications. Plurality is associated with a variety of
morphosyntactic generalizations, which do not always coincide. A common observation
is that in some dialects of English, morphologically singular but semantically collective
nouns such as committee and government may impose plural agreement on verbs and
pronouns, as in (10):
(10) The government are failing to achieve their goals
These nouns do not combine with plural quantifiers or appear bare, however:
(11) *Many/*Five/*∅ government are failing to achieve their goals
Such nouns should be distinguished from lexical plurals, such as police or cattle, which do
appear bare and combine with some plural quantifiers:
(12) a. Cattle are slaughtered for their meat
b. This city has too many police
For many speakers, these nouns resist combining with numerals:
(13) ?Five police came walking down the road
Yet they are clearly plural rather than mass – so an inability to combine with numerals
should not be taken as the defining characteristic of mass nouns. The main patterns discussed
so far may be summarized in Tab. 46.1.
46. Mass nouns and plurals 1135
Tab. 46.1: Summary of patterns distinguishing subclasses of singular, plural and mass nouns
ordinary
singulars
cup
collective
singulars
government
lexical
plurals
police
ordinary
plurals
cups
heterogeneous
mass
nouns
furniture
homo-
geneous
mass nouns
water
plural
mass
nouns
dues
agreement sg sg/pl pl pl sg sg pl
many vs.
much
* * many many much much much
numerals * * ? ✓ * * *
bare * * ✓ ✓ ✓ ✓ ✓
cumulative
reference
no no yes yes yes yes yes
combine
with “stubbornly
dis-
tributive”
predicates
✓ ✓ ✓ ✓ ✓ * *
2. Issues in the denotation of mass and plural NPs
By an “NP” we here mean a phrase consisting of a common noun, possibly with complements
or modifi ers, but excluding any determiner; e.g. water, horse, books written by
Mark Twain, but not that water, a horse, or all books written by Mark Twain. We turn
to phrases including the determiner in section 3. For the sake of discussion, we assume
for most of this section that NPs are predicates, and hold or fail to hold of groups and/
or individuals; we turn to the idea that NPs may sometimes serve as something like the
name of a kind in section 3.1.
2.1. Approaches to plural denotation
Most analyses assume that plural predicates (including nouns) hold true of collective
objects of some sort, which I will call “groups.” (Readers are cautioned that group has
a more specifi c technical sense in some work, especially that derived from Link 1984,
Landman 1989a,b, 2000.) Thus, a plural noun such as horses will hold true of groups of
horses just as a singular noun like horse holds true of individual horses.
The issue then arises of what a “group” is. One option is to identify groups with sets.
However, some authors object to this identifi cation on the grounds that sets are abstract
mathematical objects, while the denotata of plural nouns may be concrete (Burge 1977,
Link 1983, 1984).As Link (1984: 247) puts it,“If my kids turn the living room into a mess
I fi nd it hard to believe that a set has been at work, and my reaction to it is not likely to
be that of a singleton set…” However, Black (1971) has argued that regarding the referents
of plural terms as sets actually clarifi es, rather than distorts, the notion of a set; and
in any case not everyone shares the intuition that sets of concrete objects are themselves
abstract (Cresswell 1985, Landman 1989a).
If groups are not identified with sets, they are usually taken to be concrete particulars
of some sort – often called “plural individuals,” though this is quite a departure from the
1136 IX. Noun phrase semantics
meaning of the word individual in ordinary, non-technical usage. The group of John and
Mary would be identified with a complex, spatially scattered individual with John and
Mary as parts; or, as it is usually termed, the sum of John and Mary, which we may notate
‘j+m’.
Typically it is assumed that the sum operation is associative, so that a+(b+c)=(a+b)+c.
Summing differs in this respect from set-theoretic pairing, since {a,{b,c}}≠{{a,b},c} when a,
b and c are distinct.This allows us a way of distinguishing the two approaches completely
independently from issues of abstractness and concreteness. This difference will play a
role in the analysis of distributivity (section 4).Another line of analysis denies that plural
predicates hold true of groups at all. Reference to groups is avoided by locating the plurality
in the denotation relation itself, rather than in the denoted object. This idea was
pioneered by Boolos (1984, 1985a,b) and developed in more detail by Schein (1993) and
subsequent literature; a related analysis of mass nouns is given in Nicolas (2008).
To illustrate, consider a revision to the standard notion of satisfaction. In the
usual semantics for a language with variables, interpretation is relative to a function
assigning exactly one value to each variable. In a system with plural variables, rather
than assigning each plural variable exactly one group as its value, we relativize interpretation
to relations rather than functions, so that an assignment may match a given
variable with more than one value. Then a formula containing a plural variable can
be satisfied by an assignment which gives multiple values a1
,…,an
,… to this variable,
without being satisfied by assignments which give the set of all these values {a1
,…,an
,…}
as the (sole) value for the same variable. The plurality is located in the assignment relation
itself, rather than in the assigned value. Predication in general can be treated as
satisfaction; adopting this technique in effect allows an argument place to be saturated
simultaneously by more than one individual, rather than by the group containing those
individuals. The primary advantage of such a technique is that it allows an analysis of
phrases like the sets which do not contain themselves which does not give rise to Russell’s
paradox; see the references above for details.
Whether one analyzes plural NPs as satisfied by sets, or sums, or simultaneously
by multiple individuals, certain more purely descriptive, theory-neutral issues must be
addressed. In what follows I will continue to phrase these issues in terms of the “groups”
denoted by a plural NP, but essentially the same questions arise in any approach;
readers who prefer a groups-free approach are invited to rephrase the discussion
accordingly.
Prominent among these issues is the question of how the denotation of a plural noun
relates to the denotation of the corresponding singular. A natural assumption to make
is that the plural noun holds true of all and only the groups of objects of which the corresponding
singular noun holds true; so that horses, e.g., will hold true of all and only the
groups whose members are individual horses. Note that this directly predicts that plural
nouns will have cumulative reference, on the plausible assumption that for any groups A
and B, there is a group whose members include all and only the members of A and the
members of B.
However, if we take seriously the idea that a group must contain more than one
member, this idea runs into immediate problems with examples using the determiner
no (Schwarzschild 1996: 5). A sentence of the form No A B is true iff there is nothing of
which both A and B are true. E.g. (14) is true only if there is nothing of which horses and
in the corral both hold true.
46. Mass nouns and plurals 1137
(14) No horses are in the corral
But suppose there is only one horse. Then there are no groups containing more than
one horse, so by our assumption that plural nouns hold only of groups, horses does not
hold true of anything.This renders (14) automatically true, even if the one horse is in the
corral – the wrong result.
This problem is easily solved if we allow plural NPs to hold of individuals and not
just groups. In particular, a plural NP should hold of all the same individuals as the corresponding
singular, as well as all groups of such individuals. Then if there is only one
horse, the plural noun horses will hold true of it and (14) is correctly predicted to be false
if the horse is in the corral.
Chierchia (1998b) defends the idea that plural nouns hold only of groups by assigning
a more complex denotation to no: Rather than taking no(A,B) as true iff A and B do not
overlap, he takes it as true iff π(A) and B do not overlap, where π(A) is the set of all subsets
and members of the union of all groups in A (and singletons of members of A). But
in the case just described, the plural noun denotation A is empty, so this more complex
procedure gains us nothing; incorrect truth conditions are still assigned. See Sauerland,
Anderson & Yatsushiro (2005) for additional considerations.
2.2. Approaches to mass denotation
A mass noun like water is frequently assumed to hold true of all and only the individual
portions of water – with no assumption that an individual “portion” must be physically
separated in any way. Thus, water will hold of the water in the top half of my glass, as
well as the water in the bottom half, the water in the top three quarters and the water
occupying the glass as a whole. Nor need portions be physically contiguous; the water in
two separate glasses may be considered together as a portion of water, of which the noun
water holds true. Assuming that for any two portions A and B, there is a portion A+B
consisting of them, we may stipulate that mass nouns are cumulative, holding of A+B
whenever they hold of A and of B.
Since plurals also show the cumulative reference property, this will not distinguish
mass nouns from plurals, or explain the differences between them, such as the ability
of plurals but not mass nouns to combine with numerals. We will consider four major
strategies for explaining the differences between mass and plural count NPs in semantic
terms.
One strategy is to assume that mass nouns, but not plurals, show distributive reference,
also sometimes known as divisive reference (not to be confused with Quine’s divided
reference) or Cheng’s condition (after Cheng 1973): If a mass noun holds of A, and B is
a part of A, then the mass noun holds of B as well. Some versions of this approach go
further and require that mass nouns be non-atomic; i.e., that for each A of which the mass
noun holds, there is some B which is a proper part of A, of which the mass noun also
holds. This implies that mass noun denotations have no minimal parts; one may divide
them without limit. If a noun’s denotation is cumulative, distributive and non-atomic, we
may call it continuous. Much of the attraction of analyzing mass nouns as denoting continuously
is that it offers an explanation why mass nouns do not combine with numerals:
One may divide their denotations in any arbitrary fashion into any number of parts, so
there is no basis for counting.
1138 IX. Noun phrase semantics
Unfortunately, a condition requiring continuous denotation does not achieve even
initial plausibility in the case of complex mass NPs like water covering the floor, since
some water could easily cover the floor without all its parts covering the floor. Yet such
complex mass NPs fail combine with numerals and other count determiners, just as
simple mass nouns do.
Moreover, it is quite debatable whether even lexical mass noun denotations are really
non-atomic; the individual hydrogen and oxygen atoms constituting an H2
O molecule
would not seem to be water. (It should be cautioned that the issue here is not whether
they would be water if separated from each other and released as gas, but whether they
are water when still part of the H2
O molecule – perhaps a trickier issue.) One may claim
that even if mass noun denotations are not actually continuous, the language portrays
them as if they were (Bunt 1985); but this would seem to imply that much of our ordinary
talk using mass nouns is literally false, a consequence many semanticists would want to
avoid.
A different approach to semantically distinguishing count and mass nouns is to
regard the mass nouns as holding of portions of material, while count nouns hold of
more abstract objects constituted of that material (Link 1983). This way of drawing the
distinction allows an easy solution to the “gold ring” paradox: It may be that a ring is
gold and the ring is new, but the gold is old. If we distinguish between the ring and
the gold which constitutes it, there is nothing to prevent one from being new and the
other old.
The philosophical merits of claiming that objects like rings are distinct from the portions
of material of which they are constituted may be debated. But in addition, the
proposal makes sense only under the very narrowest construal of the term mass noun,
in which it refers only to those nouns which function as names of physical substances.
Even though chair is a count noun and furniture is, by most definitions, a mass noun,
one hesitates to say that chairs are constituted of furniture in the way that rings are
constituted of gold, or that a chair can be new while the furniture it is constituted of may
be old.
A third approach to the semantics of the mass-count distinction, advanced especially
by Chierchia (1998a, 1998b), is to claim that mass nouns are essentially just lexical plurals,
so that the part/whole relation on the denotata of mass nouns coincides with the
subgroup relation on the denotata of plurals. Under Chierchia’s approach, a mass noun
like change is (nearly) identical in denotation to the plural noun coins; the mass noun
footwear is (nearly) identical in denotation to shoes, etc.
An analysis which drew no distinction at all between mass nouns and lexical plurals
would face several problems: First, there are clear examples of lexical plurals which are
not mass, such as police and cattle. Second, mass nouns and plurals combine with different
classes of determiners, and may not give equivalent truth conditions even when
they do combine with the same determiner. Most change is copper may be understood as
claiming that the copper coins exceed the other coins in some measure such as weight or
volume, while Most coins are copper requires specifically that the total number of copper
coins exceeds the number of other coins. Chierchia’s proposal addresses challenges like
these by allowing that mass nouns are not completely indistinguishable in denotation
from plurals: plural nouns hold only of groups and never of individuals, while mass nouns
may hold of both. But as pointed out in section 2.1 above, claiming that plural nouns
cannot hold of individuals makes the semantics of determiners like no problematic; until
46. Mass nouns and plurals 1139
a solution to this problem is offered, this strategy for representing the mass-count distinction
must be regarded as questionable.
A fourth approach to the semantics of mass NPs treats them not as predicates at all,
but as singular terms denoting sums. Water is not treated as a predicate holding true of
all individual portions of water, but instead as something like a name, denoting the sum
of all such portions. The inability of mass NPs to combine with numerals can then be
explained in the same way as the inability of proper names to combine with numerals:
it makes no sense to count a single object, as opposed to a set. The analogy to proper
names must not be pushed too far, since proper names normally do not combine with
quantifiers, while mass NPs do, including some quantifiers dedicated just to this purpose.
But as stressed by Roeper (1983), Lønning (1987), Higginbotham (1994), this approach
can explain a number of otherwise puzzling facts about mass quantification, if we assume
that the domain of possible mass NP denotations forms a Boolean algebra; see section
3.2 below.
To summarize, each of these strategies for identifying a semantic difference between
mass and count NPs faces significant challenges:There are direct counterexamples to the
claim that mass NPs denote continuously. Only a subset of mass NPs denote substances.
Treating mass NPs as holding of groups and individuals, but plurals only of groups, seems
incompatible with the semantics of no.And treating mass NPs as names of sums requires
an explanation why mass NPs but not names combine with quantifiers.
3. Issues in the denotation of mass and plural DPs
By a “DP” we mean a phrase consisting of an overt or covert determiner, together with
an NP, e.g. that water, a horse, or all books written by Mark Twain. Phrases of this category
may serve directly as arguments to a verb or other predicate. In some analyses, NPs may
also sometimes serve directly as arguments to predicates, so we include discussion of the
semantics of NPs in such analyses here as well.
We consider in turn bare plurals and mass nouns, plural and mass DPs with overt
quantificational determiners, and definite and conjoined DPs.
3.1. Bare plurals and mass nouns
As already mentioned in section 1.1, plural and mass nouns are distinguished from singular
count nouns in English by their ability to appear bare, and show a parallel alternation
in interpretation among existential, generalizing and kind-level readings, depending
on the type of predicate with which they combine (see (4) to (6) above).
The starting point for most modern literature on this pattern is Carlson (1977a,b),
which argued that bare plurals and mass nouns are interpreted unambiguously as something
like proper names of kinds. In this analysis, the existential interpretation exhibited
in examples like Raccoons were stealing my corn is not due to the internal semantics of
the bare NP, but is built into the meaning of the predicate with which it combines: We
relate each kind to the “stages” which realize it via a relation R, then represent steal, e.g.,
as λxλy᭚z[R(y,z) & steal(z,x)].This predicate can then apply to the kind “raccoons” collectively,
to yield truth conditions to the effect that there is at least one realization of this
kind that was stealing my corn.
1140 IX. Noun phrase semantics
Likewise, individual-level predicates like are sneaky are analyzed as containing a
hidden generic operator G, allowing them to take kinds as arguments while generalizing
about the individuals realizing those kinds. Raccoons are sneaky may be represented as
G(sneaky)(r),where G(P)(k) means that instantiations of kind k generally have property
P. (Carlson 1989 replaces G with a similar operator taking scope over entire sentences.)
Kind-level readings like those in Raccoons are extinct result from direct application of
the predicate to its argument, with no hidden quantification.
A major argument for this approach is that it correctly predicts that the existential
quantifier associated with bare plurals and mass nouns always takes the narrowest possible
scope.Thus (15a) has only the reading which allows everyone to have read different
books about caterpillars, while (15b) is ambiguous, and admits a reading which requires
everyone to have read the same book about caterpillars – an unexpected difference if
the bare plural caterpillars expressed existential quantification as part of its internal
semantics:
(15) a. Everyone read books about caterpillars
b. Everyone read a book about caterpillars
A second argument comes from the fact that kind-level, individual-level and stage-level
predicates can be conjoined to take a single bare plural or mass argument:
(16) Raccoons are widespread, sneaky and have been stealing my corn
If bare plurals were ambiguous between existential, generalizing and collective
readings, examples like this would seem to impose conflicting requirements on how
to interpret the bare plural subject; but if bare plurals are unambiguously kind-denoting,
such examples are expected. The coordinate VP is straightforwardly analyzed as
in (17):
(17) λx[widespread(x) & G(sneaky)(x)& ᭚y[R(x,y) & stealing-my-corn(y)]]
A popular alternative analysis, developed in Wilkinson (1991), Krifka & Gerstner-Link
(1993) and Diesing (1992), claims that bare plurals and mass nouns are interpreted as
plural indefinites when they combine with stage- or individual-level predicates. Indefinites
are interpreted as contributing free variables to the semantic representation, with
no quantificational force as part of their internal semantics, as in Discourse Representation
Theory (Kamp 1981) or File Change Semantics (Heim 1982). The variable contributed
by an indefinite may be bound by a quantifier in the surrounding context, such as
the adverb usually in (18a), to yield truth conditions represented as in (18b):
(18) a. Bears usually have blue eyes
b. usually x (bear(x), x has blue eyes)
Or, the variable may be bound by a general operation of existential closure, as in Raccoons
were stealing my corn. To obtain a generic reading in examples like Raccoons are
sneaky, it is assumed that the variable is bound by a “generic operator” analogous to an
adverb of quantification:
46. Mass nouns and plurals 1141
(19) GEN x (raccoon(x), x is sneaky)
In Diesing’s version of this proposal, it is claimed that existential closure takes place at
the level of VP; bare plural or mass subjects of stage-level predicates are VP-internal,
hence existentially bound. Subjects of individual-level predicates are VP-external, hence
available for binding by the generic operator or other quantifiers. On the assumption
that quantificational determiners must scope higher than the existential closure operation
on VP, this correctly predicts that the existential quantification associated with bare
plurals in examples like (15a) always takes narrow scope.
This approach has an advantage over Carlson’s in that it predicts that bare plurals
are available for binding by adverbs of quantification, as in (18); such sentences require
extra stipulation if bare plurals are unambiguously kind-denoting. But Carlson’s analysis
has an advantage in predicting the conjoinability of kind-level with stage-level and
individual-level predicates, as in (16); if bare plurals combining with stage- and
individual-level predicates are indefinite rather than kind-denoting, extra stipulation
must be given for these examples.
A syntactic issue regarding bare plurals and mass nouns is whether they are DPs with
an implicit determiner, or simply NPs serving directly as arguments to the verb, with no
determiner at all, implicit or explicit. If the latter, it may be necessary to allow that NPs
may serve as something like names of kinds. This would force a revision to much of our
discussion in section 2, where it was assumed that NPs were predicates.
Chierchia (1998a,b) suggests that this is a point of parametric variation among languages:
In languages like Chinese or Japanese, NPs are unambiguously kind-denoting, so
that all NPs may appear bare. To combine such NPs with a determiner requires application
of a predicate-forming operation, whose output, Chierchia suggests, is mass; this predicts
that in such languages, NPs cannot be combined directly with numerals, but require
classifiers. In contrast, NPs in languages like French are unambiguously predicates and
never function as names of kinds; the prediction is that French NPs may not appear
without a determiner. Languages like English allow NPs to function both as predicates
and as names of kinds, according to whether they are count or mass. Mass nouns may
thus appear bare, while (singular) count nouns may not. Plural marking on a count noun
serves to form the name of a kind from a predicate, allowing plurals to appear bare as
well. See articles 96 (Doetjes) Count/mass distinctions, 44 (Dayal) Bare noun phrases,
and 47 (Carlson) Genericity for more discussion.
3.2. Quantified plurals and mass nouns
Plural DPs with quantificational determiners such as many, few, most, etc. differ from
singular DPs in allowing quantification over groups. But there appear to be several
different kinds of quantification over groups involved,and trying to give a unified account
of all of them is a challenge.
First, many plural quantifiers allow a reading which involves existential quantification
over groups of a size given by the determiner. With certain quantifiers, this reading is
most natural when the determiner heads a partitive construction as in (20a).
(20) a. Most/Many/All of the students gathered in the hallway
b. ?Most/?Many/?All students gathered in the hallway
1142 IX. Noun phrase semantics
(20a) may be paraphrased as “A group consisting of most/many/all of the students gathered
in the hallway.” Similar readings are available for non-partitive constructions, but
at least with some determiners, many speakers find these slightly degraded in comparison
to partitives as in (20b). Other determiners allow this reading naturally even in
non-partitives:
(21) Fifty/The students gathered in the hallway
Existential quantification over groups of the size given by the determiner gives the wrong
results for determiners like few, exactly fifty and other non-monotone-increasing quantifiers.
Sentence (22a) does not mean that at least one group consisting of few students
gathered in the hallway, but rather that the total number of students who gathered in
the hallway is few; (22b) does not mean that at least one group of exactly fifty students
gathered in the hallway, but rather that the total number of students that gathered in the
hallway was exactly fifty:
(22) a. Few of the students gathered in the hallway
b. Exactly fifty students gathered in the hallway
To obtain correct results in examples like these, the determiner should be analyzed as
placing a cardinality restriction on the maximal group satisfying both the NP and the
predicate, so that exactly fifty, e.g., denotes λXλY[|∪(X∩Y) | = 50], where X and Y range
over sets of groups.
An interesting observation due to Dowty (1986), made originally with respect to all
but equally applicable to many other plural determiners, is that they do not combine
naturally with predicates expressing pure cardinality:
(23) ??Most/??Many/??All of the students are numerous
Dowty suggests that although predicates like gather hold only of groups and not individuals,
they have “distributive subentailments” concerning the individual members of
those groups. If a group gathers in the hall, individual members of the group must come
into the hall and remain there long enough that they are all present at a common time.
In contrast, a predicate like be numerous carries no non-trivial entailments about the
individual members of the groups of which it holds. The determiners in examples like
(20a) serve to indicate that the subentailments of the predicate hold of some quantity
or proportion of individual members of the group; thus All of the students gathered in
the hallway requires that each individual student come into the hallway. Because be
numerous does not carry any distributive subentailments for the determiner to operate
on, the sentences in (23) are anomalous.
A different kind of quantification over groups is noted by Link (1987). (24) seems to
involve universal quantification over groups of competing companies:
(24) All competing companies have common interests
In this sort of example,the correct results may be obtained straightforwardly by assigning
the determiner its usual semantics in Generalized Quantifier Theory and letting plural
46. Mass nouns and plurals 1143
NPs and VPs denote sets containing groups.We let a group be in the denotation of competing
companies iff its members are companies in competition with each other, and
a group be in the denotation of have common interests iff its members have common
interests with each other; the determiner every indicates that the former set is a subset
of the latter.
However, it should be noted that this kind of reading is generally only available when
the NP contains a modifier such as competing which forces the NP to hold only of groups.
Indeed, if neither the NP nor the VP forces a collective reading, most quantifiers, even
if morphologically plural, are most naturally interpreted as quantifying simply over
individuals:
(25) Most/Few students wrote a good paper
The sentences in (25) mean that a majority/minority of individual students wrote a good
paper.
The definite determiner allows a collective reading even without such modification,
as do numerals:
(26) The/Three students wrote a good paper
One natural interpretation of the sentences in (26) is that the students collaborated in
writing a good paper.
As noted by Scha (1981), if more than one plural quantifier is present in a clause, a
reading is available involving “cumulative quantification” (not to be confused with the
“cumulative reference” property discussed in section 1, above). (27) has a reading which
claims that the total number of Dutch firms that have an American computer is 600 and
the total number of American computers owned by a Dutch firm is 5000:
(27) 600 Dutch firms have 5000 American computers
Roberts (1987: 148ff), following unpublished work by Partee, suggests that the cumulative
reading is just a special case of an ordinary collective reading in which the predicate
takes two groups as arguments, so that (27) means simply a group consisting of 600
Dutch firms stands in the “have” relation to a group of 5000 American computers. But as
van der Does (1993: 545) and Schein (1993: 167) point out, this approach does not extend
easily to sentences containing monotone decreasing determiners. The correct truth
conditions for (28) are not obtained if we interpret the quantifiers according to their
standard semantics and assign them scope in the usual way:
(28) Fewer than 600 firms own fewer than 5000 computers
Scha’s analysis of this sort of example requires an unusual syntactic analysis in which the
two determiners combine to form a “compound numerical”: in (27), 600 and 5000 combine
to form an expression denoting λR[|proj1
(R)|=600 &|proj2
(R)|=5000], where projn
maps a relation onto the projection of its nth
argument place.The two NPs also combine to
form a “compound noun,” denoting the Cartesian product of the denotations of the NPs
which combine: DF×AC. The compound numerical combines with the compound noun
1144 IX. Noun phrase semantics
to form a complex DP or “noun phrase sequence” denoting λR[ | proj1
({<x,y>∈DF × AC
|R(x,y)})|=600 & |proj2
({<x,y>∈DF × AC|R(x,y)})|=5000]. This may then combine
with the 2-place predicate own to give the desired truth conditions. Many semanticists
have viewed this proposal as non-compositional, and a variety of subsequent proposals
have been made to interpret such sentences while retaining a more intuitive
constituency.
One family of analysis uses special mechanisms to pass information up the tree
which would be lost in ordinary semantic composition: Van der Does (1992) employs
product types to allow access to NP denotations above the level of the DP. Landman
(2000) proposes a complex system in which multiple semantic representations are
derived in parallel, then combined to form the asserted content of the sentence as a
whole; a related analysis is developed in Krifka (1999a). A different family of solutions
appeals to branching quantification (Westerståhl 1987, Sher 1990). Schein (1993) uses
to a neo-Davidsonian theory of thematic relations: each argument of the verb corresponds
to a separate clause in logical form, over which the corresponding quantifier may
take scope; the subject and object quantifiers thus remain scopally independent of one
another.The choice among these analyses is a major unresolved issue in the semantics of
plurality.
Quantified mass DPs generally fall into two patterns: In the first, a bare mass DP
combines with a measure phrase or classifier to form a complex count NP, which may
then combine with an ordinary count determiner, as in two liters of water, every loaf of
bread, etc. In the second, the mass NP combines directly with a determiner without a
measure phrase or classifier, in which case a mass determiner is required: much water,
all bread.
Measure expressions such as liter or loaf are most often analyzed in terms of measure
functions, i.e., functions from individuals to real numbers.As stressed by Lønning (1987),
Krifka (1989), Schwarzschild (2002), this kind of quantification requires additive measure
functions, so that whenever x and y do not overlap, f(x+y) = f(x)+f(y). (Hence *fifty
degrees Celsius of water.)
Where liter is the function mapping portions of material onto their volume in liters
and R relates kinds to their realizations as in section 3.1 above, we may analyze the
measure word liter as denoting λkλnλx[R(k,x) & liter(x) = n]. Two liters of water will
therefore denote λx[R(water,x) & liter(x) = 2], the set of individuals realizing the kind
“water” and measuring two liters. Note that the numeral two is not analyzed as a quantificational
determiner, but as something more like a proper name denoting the number 2,
and serving as an argument of liter.
Alternatively, we might treat liter as denoting λkλx[R(k,x) & liter(x) = 1], so that
liters of water simply denotes the set of 1-liter volumes of water. (This option must
probably be available anyway, for examples like every liter of water.) We might then
allow this to combine with the ordinary determiner two; but since every 2-liter volume
of water contains many more than two 1-liter volumes of water, this will not give the
right results unless we adopt a non-overlap condition, perhaps as part of the pragmatic
background.
This use of measure functions is extended to noun classifiers of the kind exemplified
in Chinese, Japanese and other East Asian languages in Krifka (1995). Sometimes it is
claimed that in these languages, all nouns are mass, since they all must combine with
classifiers before they may combine with numerals (Chierchia 1998a,b; Krifka 1999b).
46. Mass nouns and plurals 1145
However, even in classifier languages, some sort of mass/count distinction is often
detectable (Hundius & Kölver 1983, Cheng & Sybesma 1999).
Direct quantification of a mass NP, with no measure phrase or classifier, is possible
in English using quantifiers such as much, little, most, etc. As noted by Roeper (1983),
Lønning (1987), Higginbotham (1994) and others, we do not obtain correct results by
treating mass NPs as predicates holding of individual portions of “stuff” as in section 2.2
above, and treating these quantifiers as binding variables ranging over these portions.
(29) does not mean that for every x, if x is a portion of phosphorus, then either x is red
or x is black, since (29) may be true in the case where some portions are only partly red
and partly black.
(29) All phosphorus is either red or black
A related observation, first made by Bunt (1979), is that direct mass quantification normally
requires not only the NP, but also the scope of the DP to show cumulative and
distributive reference:
(30) Most water is wet/*heavy
Exceptions to this generalization have been noted and discussed by Higginbotham
(1994), but these may be regarded as special cases.
Assuming such a restriction, we define a sum operation on the extensions of cumulative,
distributive predicates: let σxP(x) denote the sum of all those objects x of which
P holds true, providing P refers cumulatively and distributively; undefined otherwise.We
apply this sum operation to both the NP and the verbal predicate before combining them
with the mass determiner; this treats the determiner as a relation between sums.
Assuming a Boolean part-whole structure on portions, we may now reconstruct the
theory of quantification in this Boolean algebra, rather than the power set algebra of the
universe of discourse (Roeper 1983, Lønning 1987, Higginbotham 1994). E.g. all may
be analyzed as holding between two portions x and y iff x is a material part of y, so that
All water is wet is true iff the sum of all water is a part of the sum of all wet material;
most may be treated as holding between x and y iff μ(x⋀y) > 1/2 μ(x), where μ is some
pragmatically salient measure function and ⋀ is the Boolean meet operation.
3.3. Plural and mass definites and conjunction
A related use of sum operations may be made in the analysis of plural and mass definite
DPs and in the analysis of conjoined DPs.An obvious limitation of Russell’s (1905)
theory of definite descriptions in terms of unique existential quantification is that it
does not apply to plural or mass definites: The horses are in the corral does not mean
that there is exactly one horse; The coffee is in the room does not mean that there is
exactly one portion of coffee. Yet the fact that the same word the is used both with singular
count NPs and with mass and plural NPs seems no accident; one would hope for a
unified semantics.
An idea suggested by Sharvy (1980) and popularized in the linguistics literature by
Link (1983), is to replace the Russellian representation of ‘The A is/are B’ in (31a) with
the representation in (31b), where ‘≤’ indicates the part-whole relation:
1146 IX. Noun phrase semantics
(31) a. ᭚x[A(x) & ∀y[A(y) → x=y] & B(x)]
b. ᭚x[A(x) & ∀y[A(y) → x≤y] & B(x)]
Now The coffee is in the room will be true iff there is a maximal portion of coffee, of
which all other portions are part, which is in the room.Assuming that the maximal group
of horses has its smaller subgroups and members as parts, The horses are in the corral
will require this maximal group of horses to be in the corral. But on the assumption that
no king of France contains another as part, The king of France is bald will require the
existence of a unique king of France: the Russellian truth conditions fall out as a special
case.
The maximality condition imposed in this analysis has the effect that the definite
description picks out the sum of the extension of the NP, on the assumption that the
NP refers cumulatively. (The sum operation here should not require that the NP have
distributive reference, unlike that used at the end of section 3.2.) If one prefers a presuppositional
analysis, the definite determiner may be treated as directly expressing the sum
operation, so that ‘The A is/are B’ is represented as in (32):
(32) B(σx(A(x))
Then the A will be undefined when A is not cumulative, e.g. if it is a singular count noun
with more than one element in its extension; the formula is therefore not assigned a truth
value, which we consider to be presupposition failure. See article 41 (Heim) Definiteness
and indefiniteness for more discussion.
A related idea is frequently invoked in analysis of conjoined DPs, as in (33):
(33) John and Mary are a happy couple
The conjunction in this sort of example cannot be reduced in any obvious way to sentential
conjunction; (33) does not mean “John is a happy couple and Mary is a happy
couple.” Instead, most analyses treat the coordinate subject John and Mary as referring
to the group of John and Mary, and let the predicate are a happy couple apply to this
group collectively.
Perhaps the simplest way to obtain this result is to treat and as ambiguous, between
the ordinary truth-functional and (or some generalization it across a type hierarchy) and
a “group-forming” and which maps any two individuals to the group consisting of them.
This idea dates to ancient times and is represented in the modern literature by Partee &
Rooth (1983) and many others; see Lasersohn (1995) for a historical overview.
A number of complications arise in such an analysis. First, group-forming readings of
conjunction are not limited to proper names and other individual-denoting DPs, but also
occur with indefinites and other quantificational DPs:
(34) a. A man and a woman own this house
b. Every student and every professor met to discuss their plans
Hoeksema (1983, 1988) discusses ways to adapt a group-forming conjunction operation
into Generalized Quantifier Theory and Discourse Representation Theory to deal with
such examples.
46. Mass nouns and plurals 1147
Another complication is that group-forming and must sometimes be done “in the
argument places” of NPs or other predicates, as in (35):
(35) This man and woman are in love
This can be accomplished by a suitable type-theoretical generalization of the groupforming
conjunction operation (Lasersohn 1995, Heycock & Zamparelli 2005).
But perhaps the most unsatisfying feature of an analysis which claims that conjunction
is ambiguous between truth-functional and group-forming and is the claim that and
is ambiguous at all. The putative ambiguity is too systematic and too common crosslinguistically
to be accidental; an analysis should at least make clear what these readings
have in common which leads them naturally to be expressed by the same lexical item,
and ideally should unify their semantics completely.
Lasersohn (1992, 1995) argues that examples like (36) require that the conjunction
be analyzed in terms of a group-forming operation on events, hence that verbal and
sentential conjunction in general can be assimilated to group-forming conjunction:
(36) This refrigerator runs alternately too hot and too cold
Winter (2001) argues for an assimilation in the opposite direction, noting that if one
treats proper names as generalized quantifiers in type <<e,t>,t> and allows them to conjoin
using the cross-categorial generalization of ordinary truth-functional conjunction in
the style of Partee & Rooth (1983),then John and Mary denotes the set of sets containing
John as a member and Mary as a member; the group of John and Mary is recoverable
from this set through a simple type-shifting operation. Conjunction itself is therefore
treated as unambiguous; the collective reading is obtained by applying this type-shifting
operation to the ordinary conjunction of John and Mary.
4. Collective and distributive readings
An important observation about sentences containing plural or conjoined DPs is that
they may be understood either collectively, as in (37a) and (38a), or distributively, as in
(37b) and (38b):
(37) a. Our problems are numerous
b. The children are asleep
(38) a. John and Mary are a happy couple
b. John and Mary are asleep
Sentence (37a) means that our problems, taken together as a group, are numerous – no
individual problem is numerous – and (38a) means that John and Mary together form a
happy couple, not that they each do. In contrast, (37b) entails that the individual children
are asleep, not that the group is somehow asleep independently of its members being
asleep, and (38b) is interpreted in the same way. The availability of these collective and
distributive interpretations depends in large part on the predicate. Certain predicates,
such as be asleep, cannot hold of a group without holding of its individual members;
others, such as be numerous, cannot sensibly apply to an individual.
1148 IX. Noun phrase semantics
A third class of predicates may apply both to groups (without necessarily applying to
their members) and to individuals: draw a picture. Sentences containing this third class of
predicates may be understood either collectively or distributively; (39) can mean either
that each child drew a picture, or that the children collaborated in drawing a picture
together:
(39) The children drew a picture
In examples with conjoined plural subjects, a distributive interpretation is possible even
with predicates which do not sensibly apply to individuals:
(40) The students and the professors met to discuss the issue
(40) may be understood as meaning either that the students met to discuss the issue, and
so did the professors; or that the students met with the professors to discuss the issue.
Examples like (40) suggest that distributive interpretations do not necessarily involve
application of a predicate to individuals as opposed to groups; but rather, application to
the members of the group denoted by the DP, whether these members are themselves
groups or individuals. Returning to an issue raised in section 2.1 above, this supports
the idea that group-formation is not associative, since an associative operation does not
permit the representation of higher-order groups: Where a and b are the students and c
and d are the professors, ((a+b)+(c+d)) = (a+b+c+d) if + is associative.
The idea that group-formation is associative has been defended in the face of such
examples by Schwarzschild (1992, 1996), who argues that the denotations of plural DPs
may be analyzed as always having a “flat” structure if interpretation is relativized to a
pragmatically established cover of the group denoted by the DP, following Gillon (1987).
(A cover of a set S is a set of subsets of S whose union equals S.) In this analysis, a predicate
applies to each cell in a pragmatically salient cover of the group denoted by its plural
argument. Shoes conventionally come in pairs, so we interpret (41) relative to a cover
which divides the set of shoes into matching pairs, yielding a reading that each pair of
shoes costs $50, rather than each individual shoe or the group of shoes as a whole:
(41) The shoes cost $50
Describing the group whose members are the individual students and the individual professors
using a coordinate DP like the students and the professors makes salient a cover
of this group which divides it into the group of the students and the group of the professors,
so that (40) may be interpreted as meaning that the students met and so did the
professors.
It should be noticed that even though a covers-based analysis allows the use of an
associative group-formation operation for the denotations of plural DPs, covers themselves
have a non-associative structure: {{a}, {b,c}} and {{a, b}, {c}} are both covers of {a,
b, c}, but must be distinguished from one another. The need for some technique for
representing non-associative groupings seems beyond dispute.
A covers-based analysis generates non-existent readings in some cases (Lasersohn
1989). If John, Mary and Bill are the teaching assistants and earned exactly $7000 each
last year, (37) is false, even though each cell in the cover {{John, Mary}, {John, Bill}}
earned exactly $14,000:
46. Mass nouns and plurals 1149
(42) The teaching assistants earned exactly $14,000 last year
Whether distributive interpretations make reference to covers,or simply involve applying
a predicate to each member of the group denoted by its plural argument, the issue arises
whether the collective/distributive alternation represents authentic ambiguity, or rather
a single reading which is general enough to cover both possibilities. Lasersohn (1995)
argues for an ambiguity, based on examples like (43):
(43) a. John and Mary earned exactly $10,000
b. John and Mary earned exactly $5000
Suppose John and Mary each earned exactly $5000; then both (43a) and (43b) are true.
This is easy to explain if there is an ambiguity, since then (43a) might be true relative
to one reading, while (43b) is true relative to the other. But if there is no ambiguity, we
face the paradox that there are two distinct amounts, both of which are the exact amount
which John and Mary earned .
As Roberts (1987) points out, an ambiguity is also helpful in explaining patterns
of anaphora. Sentence (44a) may be true in any of three types of situation: ones in
which John and Mary collectively lifted a piano, ones in which they each lifted the same
piano, and ones in which they each lifted a potentially different piano. But only the
first two cases may the sentence be continued as in (44b), where it is anaphoric to a
piano:
(44) a. John and Mary lifted a piano
b. It was heavy
If the three types of situation in which the sentence is true correspond to formally distinct
meanings of the sentence, one can attribute the difference in anaphoric potential to
differences in meaning. But if the sentence is assigned just one very general reading, true
in any of these three situation types, it is difficult to see how rules governing the distribution
of discourse anaphors could be coherently stated. Gillon (1987) provides additional
arguments for an ambiguity.
Given that an authentic ambiguity exists, the issue arises where in the sentence it
is located. Early analyses often took for granted that DPs were ambiguous between
collective and distributive readings, but many analyses now attribute the ambiguity to
the predicate. A standard argument for this approach (e.g. Dowty 1986) comes from
examples like (45):
(45) John and Mary met in a bar and had a beer
The natural interpretation is that John and Mary met collectively in the bar, but each had
a separate beer; if we locate the collective/distributive alternation in the subject DP, this
example would seem to impose conflicting requirements on the interpretation of John
and Mary. But the correct interpretation may be obtained by locating it in the predicates:
under its distributive reading, had a beer holds of a group iff each of its members had
a beer; this predicate may be sensibly conjoined with met in a bar to yield a complex
predicate applying to the group of John and Mary.
1150 IX. Noun phrase semantics
Frequently, distributive readings are attributed to a hidden operator attached to the
predicate, following Link (1991) and Roberts (1987); predicates may be ambiguous
because this operator may be present or absent.Notated‘D
’,this operator may be defined
as in (41), where ‘yΠx’ means that y is a member of group x:
(46) D
P = λx∀y[yΠx → P(y)]
See Schwarzschild (1996) for an analogous operator making reference to covers.
Lasersohn (1998a) generalizes a similar operator type-theoretically to account for
distributivity in non-subject argument places.
A collective reading may be forced by modifying a predicate with an adverbial expression
such as together or as a group.As pointed out by Lasersohn (1990, 1995, 1998b), this
presents a problem for analyses in which the extensions of distributive predicates are
not distinguishable in principle from the extensions of collective predicates. If John and
Mary lifted the piano distributively but not collectively, (47) is false; if they each lifted
the piano individually and also lifted it collectively, (47) is true. But in either case, the
extension of lifted the piano would seem to be the set containing John, Mary and the
group of John and Mary – and if the extensions are identical, there is no way for together
to operate on them differently to provide distinct truth values in the two cases:
(47) John and Mary lifted the piano together
Lasersohn suggests that collective and distributive readings may be extensionally distinguished
using a hidden event argument, as in Davidson (1967). An event of John and
Mary lifting the piano distributively will be composed of smaller events of John lifting
the piano and Mary lifting the piano; an event of John and Mary collectively lifting the
piano will not. This allows a definition of together as λPλgλe[P(g)(e) & ~᭚e’᭚x[e’≤e &
x≠g & P(x)(e’)]. For alternative analyses, see Schwarzschild (1994), Moltmann (1997,
2004).
5. References
Anonymous 1952. Structural Notes and Corpus. Washington, DC: American Council of Learned
Societies.
Black, Max 1971.The elusiveness of sets. Review of Metaphysics 24, 614–636.
Bloomfield, Leonard 1933. Language. New York: Holt, Rinehart & Winston.
Boolos, George 1984.To be is to be the value of a variable (or to be some values of some variables).
Journal of Philosophy 81, 430–449.
Boolos, George 1985a. Nominalist Platonism. Philosophical Review 94, 327–344.
Boolos, George 1985b. Reading the “Begriffsschrift”. Mind 94, 331–344.
Bunt, Harry C. 1979. Ensembles and the formal semantic properties of mass terms. In: F. J. Pelletier
(ed.). Mass Terms: Some Philosophical Problems. Dordrecht: Reidel, 249–277.
Bunt, Harry C. 1985. Mass Terms and Model-Theoretic Semantics. Cambridge: Cambridge University
Press.
Burge,Tyler 1977.A theory of aggregates. Noûs 11, 97–117.
Carlson,Greg 1977a.Reference to Kinds in English. Ph.D.dissertation,University of Massachusetts,
Amherst, MA.
Carlson, Greg 1977b. A unified analysis of the English bare plural. Linguistics & Philosophy 1,
413–457.
46. Mass nouns and plurals 1151
Carlson, Greg 1989. On the semantic composition of English generic sentences. In: G. Chierchia,
B. Partee & R.Turner (eds.). Properties,Types and Meaning, vol. II: Semantic Issues. Dordrecht:
Kluwer, 167–192.
Cheng,Chung-Ying 1973.Comments on Moravcsik’s paper.In:J.Hintikka,J.Moravcsik & P.Suppes
(eds.). Approaches to Natural Language. Dordrecht: Reidel, 215–220.
Cheng, Lisa Lai-Shen & Rint Sybesma 1999. Bare nouns and not-so-bare nouns and the structure
of NP. Linguistic Inquiry 30, 509–542.
Chierchia, Gennaro 1998a. Reference to kinds across languages. Natural Language Semantics 6,
339–405.
Chierchia, Gennaro 1998b. Plurality of mass nouns and the notion of “semantic parameter”. In:
S. Rothstein (ed.). Events and Grammar. Dordrecht: Kluwer, 53–103.
Cresswell, Max 1985. Review of F. Landman & F. Veltman (eds.). Varieties of Formal Semantics
(Dordrecht, 1984). Linguistics 23, 626–634.
Davidson,Donald 1967.The logical form of action sentences.In:N.Resher (ed.).The Logic of Decision
and Action. Pittsburgh, PA: University of Pittsburgh Press, 81–95.
Diesing, Molly 1992. Indefinites. Cambridge, MA:The MIT Press.
van der Does, Jaap 1992. Applied Quantifier Logics: Collectives, Naked Infinitives. Ph.D. dissertation.
University of Amsterdam.
van der Does, Jaap 1993. Sums and quantifiers. Linguistics & Philosophy 16, 509–550.
Dowty, David 1986. Collective predicates, distributive predicates, and all. In: F. Marshall (ed.). Proceedings
of the Third Eastern States Conference in Linguistics (= ESCOL ’86). Columbus, OH:
Ohio State University, 97–115.
Gillon, Brendan 1987. The readings of plural noun phrases in English. Linguistics & Philosophy
10, 199–219.
Gillon, Brendan 1992. Toward a common semantics for English count and mass nouns. Linguistics
& Philosophy 15, 597–639.
Gleason, Henry Allen 1955. An Introduction to Descriptive Linguistics. New York: Holt, Rinehart
& Winston.
Heim, Irene 1982. The Semantics of Definite and Indefinite Noun Phrases. Ph.D. dissertation.
University of Massachusetts,Amherst, MA. Reprinted:Ann Arbor, MI: University Microfilms.
Heycock, Caroline & Roberto Zamparelli 2005. Friends and colleagues: Plurality, coordination, and
the structure of DP. Natural Language Semantics 13, 201–270.
Higginbotham, James 1994. Mass and count quantifiers. Linguistics & Philosophy 17, 447–480.
Hoeksema, Jack 1983. Plurality and conjunction. In: A. ter Meulen (ed.). Studies in Modeltheoretic
Semantics. Dordrecht: Foris, 63–84.
Hoeksema, Jack 1988.The semantics of non-Boolean ‘and’. Journal of Semantics 6, 19–40.
Hundius, Harald & Ulrike Kölver 1983. Syntax and semantics of numeral classifiers in Thai. Studies
in Language 7, 165–214.
Jespersen, Otto 1913. A Modern English Grammar on Historical Principles, Part II: Syntax (1st
vol.). London:Allen & Unwin.
Jespersen, Otto 1924. The Philosophy of Grammar. New York:W.W. Norton & Co.
Kamp, Hans 1981.A theory of truth and semantic representation. In: J. Groenendijk,T. Janssen &
M. Stokhof (eds.). Formal Methods in the Study of Language.Amsterdam: Mathematical Centre,
277–322.
Krifka, Manfred 1989. Nominal reference, temporal constitution, and quantification in event
semantics. In: R. Bartsch, J. van Benthem & P. van Emde Boas (eds.). Semantics and Contextual
Expression. Dordrecht: Foris, 75–115.
Krifka,Manfred 1995.Common nouns:A contrastive analysis of Chinese and English.In:G.Carlson
& F. Pelletier (eds.). The Generic Book. Chicago, IL:The University of Chicago Press, 398–411.
Krifka, Manfred 1999a. At least some determiners aren’t determiners. In: K. Turner (ed.). The
Semantics/Pragmatics Interface from Different Points of View.Amsterdam: Elsevier, 257–291.
Krifka, Manfred 1999b. Mass expressions. In: K. Brown & J. Miller (eds.). Concise Encyclopedia of
Grammatical Categories.Amsterdam: Elsevier, 221–223.
1152 IX. Noun phrase semantics
Krifka, Manfred & Claudia Gerstner-Link 1993. Genericity. In: J. Jacobs et al. (eds.). Syntax: Ein
internationales Handbuch zeitgenössischer Forschung—An International Handbook of Contemporary
Research (HSK 9.1) Berlin: de Gruyter, 966–978.
Landman, Fred 1989a. Groups I. Linguistics & Philosophy 12, 559–605.
Landman, Fred 1989b. Groups II. Linguistics & Philosophy 12, 723–744.
Landman, Fred 2000. Events and Plurality. Dordrecht: Kluwer.
Lasersohn, Peter 1989. On the readings of plural noun phrases. Linguistic Inquiry 20, 130–134.
Lasersohn, Peter 1990. Group action and spatio-temporal proximity. Linguistics & Philosophy 13,
179–206.
Lasersohn, Peter 1992. Generalized conjunction and temporal modification. Linguistics & Philosophy
13, 381–410.
Lasersohn, Peter 1995. Plurality, Conjunction and Events. Dordrecht: Kluwer.
Lasersohn, Peter 1998a. Generalized distributivity operators. Linguistics & Philosophy 21,
83–93.
Lasersohn, Peter 1998b. Events in the semantics of collectivizing adverbials. In: S. Rothstein (ed.).
Events and Grammar. Dordrecht: Kluwer, 273–292.
Laycock, Henry 2006. Words without Objects: Semantics, Ontology, and Logic for Non-Singularity.
Oxford: Clarendon Press.
Link, Godehard 1983.The logical analysis of plural and mass terms:A lattice-theoretical approach.
In: R. Bäuerle, Ch. Schwarze & A. von Stechow (eds.). Meaning, Use and Interpretation of Language.
Berlin: de Gruyter, 302–323.
Link, Godehard 1984. Hydras. In: F. Landman & F. Veltman (eds.). Varieties of Formal Semantics.
Dordrecht: Foris, 245–257.
Link, Godehard 1987. Generalized quantifiers and plurals. In: P. Gärdenfors (ed.). Generalized
Quantifiers: Linguistic and Logical Approaches. Dordrecht: Reidel, 151–180.
Link, Godehard 1991. Plural. In: A. von Stechow & D. Wunderlich (eds.). Semantik—Semantics.
Ein internationales Handbuch der zeitgenössischen Forschung—An International Handbook of
Contemporary Research (HSK 6). Berlin: de Gruyter, 418–440.
Lønning, Jan Tore 1987. Mass terms and quantification. Linguistics & Philosophy 10, 1–52.
McCawley, James 1975. Lexicography and the count-mass distinction. In: Proceedings of the Annual
Meeting of the Berkeley Linguistics Society (=BLS) 1. Berkeley, CA: Berkeley Linguistics
Society, 314–321.
Moltmann, Friederike 1997. Parts and Wholes in Semantics. Oxford: Oxford University Press.
Moltmann, Friederike 2004.The semantics of together. Natural Language Semantics 12, 289–318.
Nicolas, David 2008. Mass nouns and plural logic. Linguistics & Philosophy 31, 211–244.
Ojeda, Almerindo 2005. The paradox of mass plurals. In: S. S. Mufwene, E. J. Francis & R. S.
Wheeler (eds.). Polymorphous Linguistics: Jim McCawley’s Legacy. Cambridge, MA: The MIT
Press, 389–410.
Partee, Barbara & Mats Rooth 1983. Generalized conjunction and type ambiguity. In: R. Bäuerle,
Ch. Schwarze & A. von Stechow (eds.). Meaning, Use and Interpretation of Language. Berlin: de
Gruyter, 361–383.
Payne, John & Rodney Huddleston 2002. Nouns and noun phrases. In: R. Huddleston & G. Pullum
(eds.). The Cambridge Grammar of the English Language. Cambridge: Cambridge University
Press, 323–523.
Pelletier, Francis Jeffry 1975. Non-singular reference: Some preliminaries. Philosophia 5, 451–465.
Quine,Willard van Orman 1960. Word and Object. Cambridge, MA:The MIT Press.
Roberts, Craige 1987. Modal Subordination and Pronominal Anaphora in Discourse. Ph.D. dissertation.
University of Massachusetts,Amherst, MA.
Roeper, Peter 1983. Semantics for mass terms with quantifiers. Noûs 17, 251–265.
Russell, Bertrand 1905. On denoting. Mind 14, 479–493.
Sauerland, Uli, Jan Andersen & Kazuko Yatsushiro 2005. The plural is semantically unmarked. In:
S. Kepser & M. Reis (eds.). Linguistic Evidence: Empirical, Theoretical and Computational Perspectives.
Berlin: Mouton de Gruyter, 413–434.
1153
Scha, Remko 1981. Distributive, collective and cumulative quantification. In: J. Groenendijk,
T. Janssen & M. Stokhof (eds.). Formal Methods in the Study of Language 2.Amsterdam: Mathematical
Centre, 483–512.
Schein, Barry 1993. Plurals and Events. Cambridge, MA:The MIT Press.
Schwarzschild, Roger 1992.Types of plural individuals. Linguistics & Philosophy 15, 641–675.
Schwarzschild, Roger 1994. Plurals, presuppositions and the sources of distributivity. Natural Language
Semantics 2, 201–248.
Schwarzschild, Roger 1996. Pluralities. Dordrecht: Kluwer.
Schwarzschild, Roger 2002. The grammar of measurement. In: B. Jackson (ed.). Proceedings of
Semantics and Linguistic Theory (= SALT) XII. Ithaca, NY: Cornell University, 225–245.
Sharvy, Richard 1980. A more general theory of definite descriptions. Philosophical Review 89,
607–624.
Sher, Gila 1990.Ways of branching quantifiers. Linguistics & Philosophy 13, 393–422.
Sweet, Henry 1892. A New English Grammar, Logical and Historical. Oxford: Clarendon Press.
Westerståhl, Dag 1987. Branching generalized quantifiers and natural language. In: P. Gärdenfors
(ed.). Generalized Quantifiers: Linguistic and Logical Approaches. Dordrecht: Kluwer, 269–298.
Whorf, Benjamin Lee 1941.The relation of habitual thought and behavior to language. In: L. Spier
(ed.). Language, Culture and Personality: Essays in Memory of Edward Sapir. Menasha, WI:
Sapir Memorial Publication Fund, 75–93.
Wilkinson, Karina 1991. Studies in the Semantics of Generic Noun Phrases. Ph.D. dissertation.
University of Massachusetts,Amherst, MA.
Winter, Yoad 2001. Flexibility Principles in Boolean Semantics: The Interpretation of Coordination,
Plurality, and Scope in Natural Language. Cambridge, MA:The MIT Press.
Peter Lasersohn, Urbana, IL (United States)
47. Genericity
1. Preliminaries
2. Sentential genericity
3. Generic reference
4. Rationale for generic reference
5. What types of English DP’s can have generic interpretations?
6. Generic quantification
7. What types of DP’s can express generic reference across languages?
8. Indefinite singulars
9. If there are genera, what are they?
10. References
Abstract
Generic and habitual sentences are how natural language expresses regularities, laws,
generalizations, habits, dispositions, etc. One example would be “Bears eat honey.” They
are opposed in concept to episodic sentences, whose truth conditions concern whether or
not an event of a given type occurs or fails to occur in a world of evaluation, whether as
singular events or quantified over. An example would be “Some bears are eating some
honey”. Generic sentences often include as a part a generic noun phrase such as “bears”
whose denotation is argued to be a kind of thing, rather than being some quantification
von Heusinger, Maienborn and Portner (eds.) 2011, Semantics (HSK 33.2), de Gruyter, 1153–1185