Você está na página 1de 45

Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Generic Plural (In)definites and Generic Bare Plurals : Sums or Groups?*

Abstract
The central claim of the paper will be that adverbial quantifiers cannot bind variables that
range over sums. The ban on quantifying over sums will be shown to account for a wide
empirical domain, covering in particular the generic readings of indefinite plurals and mass
nouns in French, bare plurals and bare mass nouns in Romance languages other than French,
and keine in German. Plural indefinites can be bound by adverbs of quantification only if
they can be represented as group-variables. Such ‘truly generic’ indefinites should be
distinguished from ‘pseudo-generic’ indefinites: because they are merely indirectly bound by
an adverb that quantifies over events, pseudo-generic indefinites can be modelled as sums of
individuals. The ban on quantifying over sums also bears on the analysis of generic sentences
built with names of kinds. It will be argued that in this case the adverb quantifies over the
locations at which the kind is realized.

Keywords: genericity, plurals, mass nouns, indefinites, sums, groups, adverbial


quantification, weak/strong indefinites

*
A previous version of part of the work presented here was published independently, in two
conference papers co-authored with Alda Mari (Dobrovie-Sorin & Mari (2007a,b)). In a
thoroughly revised form, sections 1-6 take up the contents of those papers; sections 7-8,
which are entirely new, extend the empirical domain to mass indefinites in French, German
keine, and to bare plurals in Romance languages on the one hand and in English on the other
hand. Also new is the account of generic sentences built with names of kinds. I would like to
thank Claire Beyssade, Patricia Cabredo-Hofherr and Olivier Bonami for enlightening
discussions bearing on some of the central claims made in the paper, and two reviewers of
Natural Language Semantics, for their comments, questions and scepticism, which have
forced me to prove that the proposed ban on quantifying over sums is not a mere stipulation,
but that indeed it makes correct predictions for a large domain of unrelated data.

1
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

1. Introducing the problem : from des-indefinites in French to bare plurals across


languages

Why is it that the generic reading of French des-indefinites is impossible in (1a-b), but
possible in (2a-b)?1
(1) a. *Des carrés ont quatre côtés.
des squares have four sides
b. *Des chats sont souvent intelligents.
des cats are often intelligent
(2) a. Des droites convergentes ont un point en commun.
des lines convergent have a point in common
‘Convergent lines have a point in common.’
b. Des amis intimes se critiquent toujours l’un l’autre.
des friends intimate always criticize each other
‘Intimate friends always criticize each other.’
Both types of examples are constructed in such a way that they are to be represented as
relying on generic quantification over (plural) individuals. The intuitive reading associated
with such examples is a generalization concerning (groups of) individuals, e.g., groups of
convergent lines or of intimate friends.
Compare the examples in (3),2 which arguably involve quantification over events
combined with the indirect binding of the plural indefinites :3
(3) a. Méfie-toi, des guêpes énervées sont un danger terrible.
watch out, des wasps excited are a danger terrible.
‘Watch out, excited wasps are a terrible danger.’
b. Des éléphants blancs se promenant dans la rue ont parfois/toujours suscité une

1
Examples (1) and (2) are versions of examples due to Corblin (1987).
2
The example in (3a) is borrowed from Dobrovie-Sorin & Laca (1998) and (3b) is adapted
from an Italian example due to Longobardi (2001).
3
The fact that the examples in (3) differ from those in (1)-(2) is also acknowledged in Farkas
& de Swart (2007), but these authors propose a quite different analysis (formalized within the
boxed representation of DRT) : examples of the type in (3) would rely on quantification over
events, with the indefinite being existentially closed, whereas examples of the type in (1)-(2)
would rely on quantification over events combined with the indirect binding of the indefinite.
Note that this notion of ‘indirect binding’ should be kept distinct from ours.

2
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

très vive curiosité.


des elephants white strolling in the street have sometimes/always aroused a
very vivid curiosity
'Always/sometimes if white elephants stroll in the street they arouse curiosity.'
The intuitive reading associated with examples such as (3a-b) is a generalization concerning
events, e.g., events of wasp-excitement or events of elephants-strolling-in-the-street. A brief
presentation of the difference between quantification over individuals and quantification over
events (combined with the indirect binding of indefinites) can be found in section 2.
Examples of the latter type, which are out of the main scope of the paper, will be briefly
discussed in the Appendix.
We will also leave aside plural indefinites headed by determiners such as plusieurs,
certains, 'several, some' etc. or by cardinals (for an insightful description of the basic data,
see Corblin (1987)). The generic readings of such plural indefinites (all of which supply
more lexical information than des-indefinites, which are merely ‘plural’) may also be argued
to rely on quantification over events rather than quantification over entities.
The central claim of the paper will be that adverbial quantifiers (and in particular the
GEN operator) cannot bind sum-variables, but only group-variables. The contrast between
(1a-b) and (2a-b) will be explained as being due to the fact that only certain nouns may –
when pluralized - provide a description for groups. The ban on quantifying over sums will be
shown to constrain not only adverbial quantification, but also adnominal quantifiers such as
keine in German.
On their existential reading, des-indefinites constitute the closest French counterparts of
bare plurals4 in English and Romance languages such as Spanish or Romanian. Our analysis
of French des-indefinites should therefore be able to account for the behavior of bare plurals
in the other Romance languages, as well as in English. The data seems puzzling at first sight:
(i) inside one and the same language, generic bare plurals exhibit no contrast comparable to
that shown in (1) vs (2) for French des-indefinites; thus, the examples corresponding to (1a-
b) and (2a-b) (where the des-indefinite is replaced by a bare plural) are all grammatical in
English but ungrammatical in Spanish, Romanian or Italian; (ii) in the latter languages,
generic bare plurals are only allowed in the type of examples shown in (3). The generic
readings of bare plurals in Romance languages (other than Brazilian Portuguese) thus appear

4
Although they go back to a partitive determiner, French des-indefinites do not have a
partitive meaning, unless they are used in contrastive contexts.

3
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

to be more constrained than the generic readings of des-indefinites, whereas the generic
readings of English bare plurals are less constrained, even less constrained than the generic
readings of singular indefinites. In a nutshell, the two sides of the puzzle will be solved as
follows: Romance bare plurals contrast with French des-indefinites insofar as they cannot
supply variables over groups; English bare plurals, on the other hand, can function as names
of kinds (Carlson (1977a,b)), or more precisely as intensional maximal sums of individuals
(Chierchia (1998)). For ease of reference we will use the term ‘name of kind’.
The paper is organized as follows. In section 2, the background assumptions regarding
the generic readings of indefinites are outlined, and in section 3 the core data and previous
accounts are introduced. Section 4 is devoted to a brief background on pluralities, where the
distinction between sums and groups is defined. In sections 5 and 6, it is respectively shown
that adverbs of quantification (Q-adverbs henceforth) cannot bind sum-variables, but only
group-variables. In section 7 the empirical basis is extended to keine in German and to bare
plurals in Romance languages and in English. I argue that English generic bare plurals cannot
be analyzed as plural indefinites, but only as names of kinds. Section 8 proposes an analysis
of generic sentences built with names of kinds and Q-adverbs. The Appendix is devoted to
those contexts in which plural indefinites are indirectly bound by an operator that quantifies
over events.
2. Background on generic indefinites : adverbial quantification over events or over
individuals?
According to the current view, stemming from Lewis (1975) and Heim (1982), the generic
reading of indefinites comes about when the variable supplied by an indefinite occurs in the
restriction of a Q-adverb. Two different types of representation can however be postulated,
depending on whether the variable supplied by the indefinite is directly or only indirectly
bound by a Q-adverb. Correlated with these two distinct representations, it is possible to
distinguish between two types of ‘generic’ (readings of) indefinites, which will be referred to
as ‘truly generic’ and ‘pseudo-generic’, respectively: the former occur in generalizations over
(singular or plural) individuals, and the latter in generalizations over events; in such cases,
the indefinite cannot be said to be generic, hence the label ‘pseudo-generic’.5

5
To my knowledge, this term was first used by de Swart (1991), but with a slightly different
meaning: because de Swart (1991) assumes that Q-adverbs can quantify only over events, no
indefinite is truly generic, all of them are pseudo-generic. With the meaning intended here,

4
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

2.1. Adverbial quantification over events and indirect binding of indefinites


Let us first consider the example in (4) and the corresponding Logical Form (LF)
representation in (4’), with the notation used in the current literature (see in particular Krifka
& alii (1995)):
(4) When John invites a friend, he always cooks dinner for her.
(4’) Alwayse,x (invite (e, John, x) ∧ friend (x)) [cook dinner for (e, John, x)]6
This LF is meant to notate the fact that individuals (participants to the event) covary with
events. Note however that the notation in (4’), where the Q-adverb binds two variables,
which was an adequate representation of unselective binding (Lewis (1975)), should be
revised if we want to assume that Q-adverbs selectively quantify over events (Rooth (1985,
1995), Schubert & Pelletier (1987, 1988), de Swart (1991)). Under this more restrictive
theory, the Q-adverb should bind an event-variable, but not the variable contributed by the
indefinite. What is then the analysis of the indefinite expressions occurring in examples
relying on quantification over events? A possible representation is given in (4”), based on the
neo-davidsonian view (Parsons (1990)) according to which th-roles can be viewed as
functions that apply to events and return individuals.7

the term ‘pseudo-generic’ was used in Dobrovie-Sorin (2003) and in Farkas & de Swart
(2007).
6
The LF in (4’) is imprecise insofar as it does not explicitly indicate that the events in the
restriction and in the nuclear scope are not identical, but rather overlapping sub-events. This
simplification does not affect the main argument.
7
Another possibility is to assume that the indefinite is bound by existential closure (see in
particular de Swart (1996) and Farkas & de Swart (2007)). Taking the liberty of translating
Farkas & de Swart’s boxed representations into tripartite representations, the example in (4)
would have the LF shown in (i):
(i) Alwayse (∃x invite(e, John, x))[ ∃x cook-for (e, John, x)]
The problem with this LF is that it does not ensure the identity between the invited person and
the person for whom John cooks. To solve this problem, the donkey pronoun her should be
treated as an E-type pronoun, yielding the LF in (ii):
(ii) Alwayse (∃x invite(e, John, x))[ cook-for (e, John, ιx. invite(e, John, x))]
The second problem is that we should also ensure that John cooks for the maximal entity that
is invited: e.g., if John invites Mary and Paul, he should cook for both of them, not just for

5
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(4”) Alwayse (invite (e, John, Theme (e)) [cook dinner for (e, John, Theme (e))]
According to this representation, the indefinite denotes an individual obtained by applying
the constant function ‘the Theme of’ to the event of invitation. The co-domain of the function
is restricted by the property denoted by the indefinite, in this case, the property of being (a)
friends (of John). The value of the function co-varies with the value of the event over which
the Q-adverb quantifies.

2.2 Adverbial quantification over individuals


Let us next consider examples such as (5):
(5) A dog is usually intelligent.
Sentences of this type express generalizations over individuals (in this case, dogs) and the
most straightforward representation corresponding to this reading is the one in (5'), where
MOST is the LF translation of usually :
(5’) MOST x (dog (x)) [intelligent (x)]
Certain event-based approaches assume that Q-adverbs can quantify only over events (Rooth
1985, 1995, Schubert & Pelletier 1987, 1988, de Swart 1991, 1996).8 Thus, according to
Rooth (1995: 270-272), the LF in (5') is not directly assigned to (5), but is rather derived
from a more basic LF, in which the Q-adverb is analyzed as quantifying over events of
‘having some property P’ that is a prerequisite for being intelligent. I will assume without
further discussion that, at least for examples of the type in (5), built with i-level predicates,
we must allow Q-adverbs to directly quantify over individuals9 (Lewis 1975, Kamp 1981,

Mary. This restriction cannot be read off (ii). This problem can be fixed by representing
participants to events as functions from events to individuals, as proposed in the text.
8
These theorists assume, in line with Parsons (1990) and contra Kratzer (1988, 1995), that
any kind of sentential predicate can introduce an event-variable.
9
Empirical arguments against representing characterizing generic sentences of the type in (5)
in terms of quantification over events can be found in Dobrovie-Sorin (2002). Farkas & de
Swart (2007) suggest a relaxed version of de Swart’s (1991) event-based approach, according
to which in examples such as (5), the Q-adverb binds not only an event-variable, but also the
variable supplied by the indefinite. Compare examples of the type in (4), as well as those in
(3), which according to these authors would rely exclusively on quantification over events.
Quite clearly, such a differentiated analysis is an implicit acknowledgment of the fact that a
uniform analysis of Q-adverbs as exclusively quantifying over events cannot be maintained.

6
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Heim 1982, Kratzer (1988, 1995), Diesing 1992, Chierchia 1995b, Krifka & alii 1995).
In sum, we will assume that Q-adverbs are selective quantifiers (in the sense that they are
unable to bind more than one variable in a given context), but we will allow them to quantify
not only over events but also over individuals; this means that in a given context, a Q-adverb
binds either an individual-variable or an event-variable.

2.3. The syntax-semantics mapping rules


The question now arises as to the choice between these two possibilities: given a particular
example, how does a Q-adverb choose whether it will quantify over events or over
individuals? In other words, we need to make explicit the mapping rules between overt
syntax and LF. We will assume the following rules, which – under slightly different
formulations - can be found in the current literature:10
(6) If/when-clauses go to the restriction of Q-adverbs (Lewis (1975), Kratzer (1978,
1986, Heim (1982), Farkas & Sugioka (1983), etc.)
(7) Subjects of generic (i.e., lexically i-level or habitual) predicates go to the restriction of
Q-adverbs11.
Which of these rules applies in a given example depends on the lexical content of the main
predicate and on overt syntax: quite trivially, the rule in (6) can apply in (4), yielding (4'), or
a refinement thereof, but not in (5), which can instead be analyzed via the rule in (7).
The distinction between quantification over individuals and quantification over events
will be crucial for the analysis of the generic readings of plural indefinites to be developed in
this paper. The core of the paper is devoted to the analysis of truly generic plural indefinites,
which involve quantification over individuals. Pseudo-generic plural indefinites, which are
only indirectly bound by adverbs that quantify over events will be briefly examined in the
Appendix.

10
This is not an exhaustive list of the mapping rules that are needed for a complete account of
Q-adverbs. Because they are irrelevant for the examples under discussion here, we do not
introduce Rooth’s (1985, 1995) ‘association with focus’ algorithm, nor Schubert et Pelletier’s
(1987,1988)) rule, which is sensitive to presuppositions. Some other mapping rules will be
introduced in the Appendix, when they will become relevant.
11
This rule is a simplified version of Diesing's (1992) Mapping Principle, which is sufficient
for our present purposes.

7
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

3. Previous accounts of French plural indefinites headed by des


The contrast shown in (1) vs (2) was observed by Corblin (1987: 57-58), who proposed an
explanation relying on a pragmatic principle that basically says that the generic readings of
plural indefinites are blocked (or neutralized, in Corblin's terminology) in examples such as
(1a-b) because the intended meaning can be expressed in a more direct way by using the
corresponding examples built with a singular indefinite (see (8a-b)):12
(8) a. Un carré a quatre côtés.
'A square has four sides.'
b. Un chat noir est souvent intelligent.
‘A black cat is often intelligent.’
Corblin’s principle can also account for the acceptability of the examples in (2a-b), which
express generalizations over pluralities, and as such are not blocked by corresponding
examples with singular indefinites, which are ungrammatical:
(9) a. *Une droite convergente a un point en commun.
a line convergent has a point in common
b. *Un ami intime se critique toujours l’un l’autre.
a friend intimate always criticize each other
Note however that a blocking account cannot explain the unacceptability, viz.
ungrammaticality, of examples of the type in (10), built with mass-indefinites headed by de
la/du, which were not discussed in the previous literature:13

12
The same principle would explain – according to Corblin (1987) – the unacceptability of
the generic readings of cardinal plural indefinites, which will not be examined here. A similar
account is suggested in Krifka & alii (1995:35) for the unacceptability of examples of the
type ?Twelve cats are beautiful when they have white fur.
13
Some readers might be tempted to attribute the ungrammaticality of (10) to a constraint
similar to the one that prevents bare plurals (and bare mass nouns) from occupying the
preverbal subject position in Romance languages such as Italian, Romanian or Spanish
(Contreras (1986), Longobardi (1994, 2001)), e.g. :
(i) a. *Copii se jucau pe stradà.
children were playing in the street.
b. *Apà curgea din perete.
water leaked from the wall
This constraint does not hold for French indefinites headed by des or by de la/du :

8
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(10) a. *De l’eau est (généralement/toujours) liquide.


de l’ water is (generally/always) liquid.
b. *De l’eau gèle à 0° Celsius.
de l’ water freezes at 0° Celsius.
c. *Du vin est souvent cher ici.
du wine is often expensive here.
The intended meaning of these examples is a generalization over amounts of matter, which
makes the Blocking principle inapplicable, since no counterpart examples exist where the
mass indefinite would be replaced by another indefinite, referring to amounts of matter, for
which the generalization would hold more straightforwardly. Sure enough, the intended
meaning can be expressed by definite mass nouns:
(11) a. L’eau est (généralement/toujours) liquide.
the water is (generally/always) liquid.
‘Water is (generally/always) liquid.’
b. Le vin est souvent cher ici.
the wine is often expensive here.
‘Wine is often expensive here.’
c. L’eau gèle à 0° Celsius.
the water freezes at 0° Celsius.
‘Water freezes at 0° Celsius.’
Note however that no Blocking principle could be invoked to account for the contrast
between (10) and (11): under current theories of genericity, the examples in (11) rely on
adverbial quantification (expressed by overt adverbs, as in (11a-b) or by the covert Gen(eric)
operator, as in (11c), over the amounts of matter that instantiate kinds such as water or wine.
Since this meaning is presumably more directly expressed by the examples in (10) than by
those in (11), the ungrammaticality of (10) cannot be explained as being due to a blocking
effect.
An alternative line of investigation was proposed by Dobrovie-Sorin & Laca (1998)
and Dobrovie-Sorin (2003), according to whom the unacceptability of examples such as (1a-

(ii) a. Des enfants étaient en train de jouer dans la rue.


des children were playing in the street
b. Du beurre était en train de fondre sur la table.
du butter was melting on the table.

9
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

b) is not due to pragmatic principles, but rather to formal constraints: LF representations such
as (1') are ill-formed (as indicated by #) because the variables in the restriction and in the
nuclear scope range over different types of entities (pluralities14 and atomic individuals,
respectively), and as such cannot be bound by the same operator.
(1) a. *Des carrés ont quatre côtés
des squares have four sides
(1’) a. #GEN X (X is a plurality of squares) [x has four sides]
Whether the variables are plurality-variables (notated by capital letters) or atomic variables
depends on the lexico-syntactic properties of a given example: in (1a) the indefinite is plural,
and therefore it supplies a plurality-variable, whereas the main predicate selects atomic
individuals, hence it introduces variables over atomic individuals in the nuclear scope of the
tripartite configuration.
Like Corblin’s analysis, the analysis sketched above correctly accounts for examples
such as (2a-b), in which generic plural indefinites combine with i-level predicates that select
pluralities. Further examples are given in (12a-b):15
(12) a. Des pays limitrophes ont souvent des rapports difficiles.,
des countries neighboring often have des relations difficult
‘Neighboring countries often have difficult relations.’
b. De vrais jumeaux se ressemblent dans les plus petits détails.
de genuine twins SE resemble down to the smallest details
‘Genuine twins look alike down to the smallest details.’
Such examples are acceptable, because the variables in the restriction and in the nuclear
scope both range over pluralities. In (12’a), MOST is the translation of souvent ‘often’ and in
(12’b) the silent GEN operator is supplied by default:
(12') a. MOST X (X is a plurality of neighb. countries) [X have difficult relations]
b. GEN X (X is a plurality of true twins) [X look alike in the smallest details]
Note however that this proposal is no better than Corblin’s when it comes to accounting for

14
I use the term plurality to refer to both sums and groups. Since Dobrovie-Sorin & Laca
(1998) and Dobrovie-Sorin (2003) use the term ‘group’ in this unspecified way (and not in the
technical sense defined below, see § 4.2), I have replaced ‘group’ by ‘plurality’ in the
discussion of the work of these authors.
15
(12a-b) are based on examples by Corblin (1987). I have simply replaced the numerals
with des.

10
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

the ungrammaticality of adverbial quantification over amounts of matter : in examples of that


type (see (10)), no conflict arises between the variables in the restriction and in the scope,
which both range over amounts of matter. It seems fair to say that the two proposals are
basically alike insofar as they directly encode (albeit at different levels of representation) the
same basic intuition: generic plural indefinites cannot express generalizations over atomic
individuals, they can only express generalizations over pluralities.
The only genuine contribution due to Dobrovie-Sorin (2003) was the distinction,
which was not made in Dobrovie-Sorin & Laca (1998), between direct quantification over
pluralities (see the examples in (1)-(2)) and quantification over events combined with the
indirect binding of the indefinite (as in (3)). This distinction is crucial in explaining important
differences between the examples in (1)-(2) and (3), in particular the fact that in (1)-(2) the
generalization is necessarily about pluralities, whereas examples of the type in (3) allow for
number-neutralization:16 in (3a) danger may come from one single excited wasp and in (3b)
curiosity may be aroused by a single elephant strolling in the street.
The next step in the analysis of generic plural indefinites is due to Dobrovie-Sorin &
Mari (2007a,b), who claimed that adverbs of quantification can bind group-variables, but not
sum-variables. In this paper, I will provide further evidence in favor of this constraint by
extending the empirical basis to bare plurals, which are the closest counterparts of French
des-indefinites in Romance languages on the one hand and in English on the other hand. I
will furthermore show that the constraint also holds for adnominal quantifiers such as
German keine.
4. Basics about pluralities
Since the distinction between sums (Le_niewski (1916), Link (1983), Landman (1989a,b))
and groups (Link (1984), Landman (1989a,b)) is crucial for the proposal to be developed
here, let me briefly present each of these notions in turn.

16
Because he adhered avant la lettre to the view that the generic readings of indefinites
uniformly rely on quantification over events, this observation cannot be accounted for by
Corblin (1987), who noticed that number-neutralization was sometimes allowed, without
clearly circumscribing the type of examples in which that could happen (for further
discussion, see section 9.2. below).

11
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

4.1 Sums as derived plural individuals


Link’s (1983) basic claim is that sums of individuals are also individuals, hence the term
‘plural individual’. More precisely, sums are derived entities, obtained from other entities by
the join-operation. A sum requires no particular principle of composition other than set
formation or set union, nor any coherence relation (Higginbotham 1983). This observation is
known under the name of principle of universal existence of sums, which says that given any
two elements in a domain, there is a sum of these elements : there is for instance a sum of my
computer and me, a sum of my computer and my cup ; consequently, there is also a sum of
my computer, my cup and myself.
The crucial difference between singular individuals and sum-individuals is that the
former are primitive, whereas the latter are derived individuals. Correlatively, the algebraic
structures of their respective domains are different.
Sums are currently represented as elements of an i-join semi-lattice (Scha (1981), Link
(1983), Landman (1989a,b)) :

(13) a+b+c

a+b a+c b+c

a+ b+ c+

Let us assume that a, b, and c are individual boys, e.g., Paul, John and Saul, which can also
be represented as minimal sums of boys,17 notated a+, b+ and c+ in (13). All the other nodes
in this lattice are non-minimal sums of boys.
Assuming a DRT-type of analysis (Heim (1982)), plural indefinites can be
represented as variables ranging over sum-individuals.

17
In order for our domain to be closed under join, we need to assume that all the elements are
of the same type : atomic individuals are therefore to be represented as minimal sums. Under
a set-theoretic representation of plural individuals, atomic individuals that are part of an i-join
semi-lattice would be represented as sets containing just one element, e.g., {a}, {b}, {c}.

12
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

4.2 Groups as primitive plural individuals


Link (1984) observed that certain pluralities cannot be modelized as sums, which led him to
assume a more complex ontology, which contains group individuals, in addition to singular
individuals, among its primitive entities. Insofar as groups cannot be equated with the
collection of their members, they are ‘atoms’ in the ontology. Nevertheless, group-atoms
differ from singular entities : they contain members, unlike singular entities, which explains
why in certain languages, e.g., English, certain collective nouns, even if morphologically
singular, may trigger plural agreement on the verb, e.g., The committee were impressed by
the candidates. For this reason, group entities are classified under plural entities in (14). This
expositional choice is further motivated by the fact – which will be crucial for our analysis of
generic plural indefinites - that not only collective singular DPs, but also certain plural DPs
denote groups.
(14) Types of entities:

Singular Plural
ENTITIES
Primitive singular individuals group individuals

Derived sum individuals

This ontology will be assumed in this paper.


Groups crucially differ from sums in that their domain of reference is not an i-join
semi-lattice ordered by the part-whole relation, but rather an unordered set, just like that of
singular individuals.
Groups are clearly needed for the analysis of collective nouns such as committee,
press, mafia, etc. :
(15) a. The committee gathers at 6:00 PM
b. The orchestra played this concerto many times this year.
Groups exist over and above the existence of their members : although the members of a
given committee may vary from one meeting to another, may change in number or may be
replaced in the course of time, the committee as an entity remains the same. Similarly for an
orchestra. Not so with sums : if the members of a sum change, the sum itself changes.

13
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

There can be sums of groups, in the same way as there are sums of singular
individuals. Just as for singular individuals, sum-formation corresponds to pluralization, e.g.,
the committees, the mafias, etc.
Precisely because sums are derived entities, which exist only as the collection of their
members, the members of a sum are accessible, which means that sum-individuals are
compatible with distributive predication, as in (16a).18 Groups differ from sums insofar as
they form indivisible wholes (Link (1984), Simons (1987), Moltmann (1997), Landman
(2000), among others), which means that their members are not accessible. This is illustrated
in (16b), where the property of being dangerous does not distribute down to the individual
members of the mafia:
(16) a. These children are dangerous.
b. This mafia is dangerous.

In order to facilitate the reading of this paper, let me say that although the distinction
between groups and sums will be central, the distinction between distributive and collective
readings will be only marginally touched upon.
5. Sum-variables cannot be bound by an adverbial quantifier
In this section, I will present the analysis proposed in Dobrovie-Sorin & Mari (2007a,b),
according to which the restricted distribution of generic plural indefinites is due to the
impossibility of quantifying over sums. This analysis will be further developed in sections 6-
7 below.
Let me first make it clear that in this paper we are interested only in the constraints
that are specific to plural indefinites. Those constraints that bear on both singular and plural
indefinites will not be investigated, but merely taken care of in the choice of examples. One
such constraint is the ‘nomicity’ of the generalization (Lawler (1973) and Burton-Roberts
(1977), followed by many others). If the predicate expresses an essential/nomic property of
the individual denoted by the indefinite, an overt Q-adverb is not needed, as shown in (17a-
b), for singular and plural indefinites, respectively ; compare the ungrammatical examples in
(18a-b), which express descriptive generalizations :
(17) a. An opera is polyphonic.

18
For a short remark on why collective readings are possible for sum-denoting indefinites see
§ 6.3. below.

14
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

b. Parallel lines meet at the infinite.


(18) a. *An opera is popular.
b. *Intimate friends have different tastes.
It is also known that indefinites can be used in generic sentences that express descriptive
generalizations provided that a Q-adverb is inserted. Hence the grammaticality of (19a-b),
which differ from (18a-b) only by the presence of a Q-adverb :
(19) a. An opera is usually popular.
b. Intimate friends sometimes have different tastes.
The contrast between (18) and (19) indicates that the nomicity constraint does not hold of Q-
adverbs in general, but only of the GEN operator (a recent account can be found in
Greenberg (2007)). Since the analysis of the GEN operator is not under investigation here,
and in order to keep things as simple as possible, we will rely on examples with overt Q-
adverbs whenever the absence of such an element would lead to ungrammaticality.
Let us now consider the examples in (20a-b), represented as in (20’a-b), where MOST
and FEW respectively translate souvent ‘often’ and rarement ‘rarely’:
(20) a. *Souvent, des bébés se ressemblent dans les plus petits détails jusqu’à l’âge de trois
mois.
often, des babies look alike down to the smallest details until the age of 3months.
b. *Des droites sont rarement parallèles.
des lines are rarely parallel
(20’) a. MOST X (X is a sum of babies) [X resemble each other]
b. FEW X (X is a sum of lines) [X are parallels]

Note that in the representations in (20’), no conflict arises between the variable in the
restriction, which ranges over sums (because it corresponds to a plural indefinite) and the
variable in the scope, which also ranges over sums (because predicates such as resemble or
be parallel select pluralities, i.e., can only apply to groups or sums). LFs such as (20’) are
therefore ruled in by the principle invoked by Dobrovie-Sorin & Laca (1998) and Dobrovie-
Sorin (2003), which leaves us with no explanation for the ungrammaticality of the
corresponding examples.19

19
Dobrovie-Sorin (2003 : 64-65) notices this problem and proposes a further constraint on the
generic readings of plural indefinites: the nominal predicate must be relational, which allows
its plural form to denote a ‘set of (stable) pairs of individuals’. This constraint, which was not

15
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

A similar problem arises with respect to the examples in (10), repeated in (21), built
with mass indefinites. These examples are ungrammatical, although no clash arises between
the variables in the restriction and in the scope, which both range over amounts of matter
(notated with capital letters):
(21) a. *De l’eau est (généralement/toujours) liquide.
de l’ water is (generally/always) liquid.
b. *De l’eau gèle à 0° Celsius.
de l’ water freezes at 0° Celsius.
c. *Du vin est souvent cher ici.
du wine is often expensive here.
(21’) a. ALL X (X is a sum of water) [X is liquid]
In order to solve the problem raised by the examples in (20), Dobrovie-Sorin & Mari
(2007a,b) formulated the constraint in (22) :
(22) Sums cannot be quantified over.
This constraint rules out as illegitimate not only LFs such as (20’), but also LFs such as (21’),
since mass indefinites resemble plural indefinites insofar as they translate as sum-variables.20
Let us now examine again the example in (1a), repeated under (23), and the
corresponding LF, in which we have replaced ‘plurality’ with ‘sum’:
(23) *Des carrés ont quatre côtés.
des squares have four sides
(23’) # GEN X (X is a sum of squares) [x has four sides]
The LF in (23’) can be straightforwardly ruled out as a violation of the principle in (22), and
therefore the clash between the type of variables in the restriction and in the scope (which
was crucial for Dobrovie-Sorin & Laca (1998) and Dobrovie-Sorin (2003)) need not be
invoked at all.
It is easy to observe that des-indefinites can be represented as sum-variables bound by
existential closure:
(24) Dans la rue, des étudiants étaient en train de danser.
in the street, des students were dancing.
‘In the street, sm (unstressed some) students were dancing’

fully exploited by Dobrovie-Sorin (2003), became central in Dobrovie-Sorin & Mari’s


(2007a,b) account.
20
Dobrovie-Sorin & Mari (2007a,b) did not consider mass indefinites.

16
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(24’) ∃X (X is a sum of students ∧ X is dancing)


We must thus conclude that the constraint on quantification stated in (22) does not concern
existential closure. This is not surprising, given the widely accepted assumption that the
syntactic domains of existential closure and of adverbial quantification are separated :
indefinites occurring in a VP-external position are mapped into the restriction of an adverbial
quantifier (and as a consequence get bound by that quantifier), whereas indefinites occupying
a VP-internal position are mapped into the scope, where they are existentially closed
(Diesing (1992)). If the constraint proposed in (22) is correct, existential closure and
adverbial quantification differ not only regarding their respective syntactic domains, but also
regarding the type of variables that they can bind (Dobrovie-Sorin (2007)).
The prohibition against quantifying over sums proposed in this section may be judged
‘stipulative’ insofar as it does not follow from some other, more general constraint, but it
cannot be discarded as a mere restatement of the data: it has a genuine predictive force, being
able to account not only for our initial examples, built with plural indefinites, but also for the
unacceptability of generic mass indefinites. Further empirical support will be provided in
section 7 below, where the constraint in (22) will be shown to make correct predictions for a
wide range of data, coming from German, Italian and other Romance languages.
Quite obviously, the ban on quantifying over sums is not a logical constraint, but may
instead constitute a fundamental limitation of the computational system of natural languages,
which is arguably more constrained (less powerful) than logical systems. According to the
characterization given in section 4.1 above, sums are not primitive entities in the ontology,
but rather derived entities, which are related to one another by the part-whole relation.
Compare primitive entities, which can be said to be ‘strongly individualizable’ in the sense
that every entity is disjoint from all the other entities of the domain. The constraint in (22)
can thus be viewed as the consequence of an individuability condition that says that
quantification can only apply to strongly individualizable entities, i.e., to elements of
unordered sets:
(25) Only elements of unordered sets can be quantified over.21

21
A similar, viz. identical, constraint on quantification can be found in Kratzer (1995 :169) :
‘Quite generally, any sort of quantification seems to require that the domain of quantification
is set up in such a way that its elements are truly distinct’, where ‘truly distinct’ means ‘not
related to each other by part-whole relations’.

17
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

6. Generic plural indefinites as group-variables bound by a Q-adverb


Given the ontology assumed in section 4 above, groups differ from sums in being primitive,
rather than derived, entities. This means that the domain of groups is an unordered set of
entities, just like the domain of singular individuals. The individuability constraint on
quantification stated in (25) therefore predicts that group-variables – unlike sum-variables -
can be quantified over. In this section, I show that this prediction is correct.

6.1 Plural nominal predicates that denote sets of groups

Given that sum-denotation is freely available (recall that any random collection of elements,
e.g., two tables, several houses, etc., can form a sum), it goes without saying that any
pluralized nominal predicate denotes a set of sums. More interestingly, since group-
denotation is severely constrained (strong coherence relations are required for group
formation (Landman 1989b, Simons 1987, Moltmann 1997), we expect only a sub-class of
plural nominal predicates to be able to denote sets of groups. In this section, I will
circumscribe a restricted class of group-predicates. Plural indefinites built with nominal
predicates of this type will correlatively be analyzed as supplying group-variables (rather
than sum-variables).
Sums satisfy the type of inference shown in (26a):
(26) a. If A are children and
B are children
A and B are children
This type of inference does not hold for a restricted class of pluralized relational predicates :
(26) b. If A are twins and
B are twins
# A and B are twins
Because any sum of sums is a sum itself, the inference relation in (26a) is satisfied. The
inference relation in (26b) is not satisfied because the subject refers to a group in the
antecedent proposition and to a sum of groups in the consequent proposition: twins denotes a
property of groups, and therefore, in (26b), A and B refer to groups; however, the
conjunction [A and B] that appears in the consequent does not itself denote a group, but
rather the sum of the two groups. But a sum of groups does not yield a group.
We may conclude that, when pluralized, sortal predicates denote properties/sets of sums,
whereas (a subclass of) relational nominal predicates provide descriptions for primitive plural
entities, i.e., groups:

18
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(27) a. Plural sortal (object-referring) nouns denote properties/sets of sums.


b. Certain plural relational nouns may denote properties/sets of groups.
The possibility stated in (27b) is due to the fact that relations provide the coherence criterion
that is necessary for group-formation.

6.2 Group-variables can be bound by an adverbial quantifier


Our theory correctly predicts that examples such as (28) are grammatical, since pluralized
relational predicates provide descriptions for groups and quantification over groups is
allowed by the constraint in (25) (because the domain of groups is a set, and as such it is a
legitimate domain for quantification):
(28) a. Des droites convergentes ont un point en commun.
des lines convergent have a point in common
‘Convergent lines have a point in common.’
b. Des pays limitrophes ont souvent des rapports difficiles.
des countries neighboring often have des relations difficult
'Neighboring countries often have difficult relations.'
(28') a. GEN X (X is a group of convergent lines) [X has a point in common]
b. MOST X (X is a group of neighboring countries) [X have difficult rel.]
Compare the example in (20b), repeated here :
(20) b. *Des droites sont rarement parallèles.
‘des lines are rarely parallel.’
Examples of this type are ungrammatical because plural sortal nouns such as ‘lines’ do not
provide a description for groups, but only for sums. The corresponding LFs are illegitimate
because quantifying over sums is banned:
(20’) b. # FEW X (X is a sum of lines) [X are parallel]
The contrast between (29a) and (29b) can be explained in the same way:
(29) a. Des jumeaux se ressemblent dans les plus petits détails.
des twins look alike down to the smallest details.
‘Twins look alike down to the smallest details.’
b. *Des bébés se ressemblent dans les plus petits détails jusqu’à 3 mois.
des babies look alike down to the smallest details until 3 months
(29’) a. Gen X (X is a group of twins) [X look alike down to the smallest details]
b. # GEN X (X is a sum of babies) [X look alike .. ..]

19
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

6.3 Summarizing
Plural indefinites supply variables over either sums or groups. The ontological distinction
between sums and groups correlates with a different behavior regarding quantification. It is
possible to quantify over primitive entities (atomic individuals or group individuals) that
belong to unordered domains, but not over derived entities (sums), which are related to each
other by the part-whole relation. As a consequence, the generic reading of plural indefinites –
which crucially relies on adverbial quantification – can arise only if the plural indefinite can
supply a variable over groups. Only a sub-class of pluralized relational nouns can provide a
description for a group.
Two reviewers have questioned the proposal made here that group-denotation
depends on the indefinite being built with a relational noun by invoking examples of the type
in (30), which show that in existential/episodic contexts, collective readings are possible with
plural DPs built with sortal nouns, and conversely, distributive readings are possible with
plural DPs built with relational nouns:
(30) a. Two men lifted this table (together).
b. Two twins lifted this table yesterday (one after the other).
Observations of this kind do not go against my proposal provided that we assume, contra
Landman (1989a,b) that the difference between collective and distributive readings of plural
DPs cannot be analyzed as being triggered by the plural DP being represented as either a sum
or a group (by virtue of type-shifting operations). Instead, plural DPs in existential contexts
can be analyzed as sums, which as such allow both distributive and collective readings,
depending on the ‘cover’ that is associated with that sum, see (Gillon (1992) and
Schwarzschild (1992, 1996)). Collective readings can be forced by using disambiguating
predicate-modifiers such as together (Lasersohn (1992)).
Note also that two twins in (30b) may refer to a sum, despite the fact that the nominal
predicate is relational. This indicates that plural indefinites built with relational nouns can be
represented both as sums and as groups. Both representations are possible in episodic
contexts; in generic contexts, sum-denotation is ruled out.
7. Extending the empirical basis : from French to English via German and Italian
In this section we will test the predictive force of our account by first proposing a solution for
an intriguing puzzle regarding keine in German. We will then examine bare plurals, which in
many languages are the closest counterparts of French des-indefinites. We will explain why

20
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Romance bare plurals are more constrained than des-indefinites, whereas English bare plurals
are less constrained.
7.1 Keine
Kratzer (1995 : 144-147) observes that - but does not explain why - the negative plural
indefinite keine (unlike the singular indefinite kein) cannot function as a quantifier, but can
only be bound by existential closure.22 This generalization would explain the
ungrammaticality of (31)-(32) (corresponding to Kratzer’s (51)-(52)):
(31) *… weil keine Ärzte altruistisch sind.
since no physicians altruistic are
(32) *… weil das keine Kandidaten wissen.
since this no candidates know.
Given Diesing’s (1992) assumptions, predicates such as altruistic and know do not allow
their subject to appear inside the VP, which in turn prevents the subject, in this case keine N,
from being bound by existential closure. The only possibility would be for keine to be
analyzed as a quantifier. The ungrammaticality shown in (31)-(32) leads Kratzer to conclude
that this option is not available for the plural form keine, although it is unproblematic for the
singular form kein. The examples in (33)-(34) correspond to Kratzer’s (55)-(56)):
(33) … weil kein Arzt altruistisch ist.
since no physician altruistic is
‘…since no physician is altruistic.’
(34) … weil das kein Kandidat weiss.
since this no candidate knows.
‘... since no candidate knows this.’
Stated as it is, the constraint on keine is completely mysterious: why is it that plurality
prevents a negative determiner from being represented as a negative quantifier?
This puzzle can be easily solved by extending our account of French des-indefinites
to keine: examples of the type in (31) are correctly ruled out, because the corresponding LF
representation violates the ban on quantifying over sums. No problem arises for the singular
form kein, which introduces atomic individuals in the restriction:
(31’) # NO X [X is a sum of doctors] [X is altruistic]

22
The presumed inability of keine to function as a quantifier was out of the scope of Kratzer
(1995), who focused on the s-level vs i-level distinction, and related to it, on the constraints
on existential closure.

21
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(33’) no x [x is a doctor] [x is altruistic]


Our analysis correctly predicts the grammaticality of examples such as (35), which escaped
the attention of Kratzer. Since they are built with i-level predicates (which moreover are not
unaccusatives), these examples show that keine can translate as a negative quantifier:

(35) a. weil keine paralellen Linien einen Punkt gemeinsam haben.


since no parallel lines a point in common have
‘since no parallel lines have a point in common’
b. weil keine benachbarten Länder sehr gut miteinder auskommen.
since no neighbouring countries very well together get along
‘since no neighbouring countries get along together well’
Note that the nominal predicates in (35a-b) provide descriptions of groups, and therefore they
are ruled in by the present account, since groups can be quantified over:
(35‘) a. no X [X is a group of parallel lines] [X has a point in common]
To conclude, the distribution of keine can be accounted for by our analysis of des-indefinites:
quantification over groups is allowed, whereas quantification over sums is illegitimate. This
is a non-trivial extension of our account, which is shown to cover not only Q-adverbs, but
quantifiers in general, regardless of their particular syntactic status, e.g., DP-internal
quantifiers such as keine.23
This does not mean, however, that we expect all quantificational determiners to be
able to quantify over groups, as keine does. With possible interesting exceptions, DP-internal
quantification seems to be designed for quantification over atomic individuals : note, for
instance, that altough some English determiners subcategorize for plural NPs, e.g., most/all
students vs *most/all student, they do not quantify over groups, but over atomic individuals,
just like those determiners that subcategorize for singular NPs , e.g. every/each student vs
*every/each students.

23
The difference between kein and keine can be reproduced in English. Although no is
invariable, it may be argued that it quantifies over either atomic individuals or groups,
depending on whether its sister NP is singular or plural:
(i) No doctor is altruistic.
(ii) ??No doctors are altruistic.
(iii) *No parallel line has a point in common.
(iv) No parallel lines have a point in common.

22
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

This difference between adverbial quantification and DP-internal quantification can


be reasonably attributed to the syntax-semantics rules that determine the restrictive term of
quantifiers: indefinite DPs supply variables (which can be modelled as discourse referents in
DRT implementations) over atoms/groups/sums depending on the nominal predicate ;
compare quantificational DPs : in this case there is no syntactic constituent that can be said to
translate as a variable (in DRT-terms, no discourse referent is introduced); instead, the NP
that is a sister of the quantificational Determiner supplies a set of entities over which the
Determiner quantifies. The observation is that in the unmarked case, DP-internal constituents
that are sisters of quantificational determiners supply sets of atomic individuals, even if the
noun is relational and pluralized.24 Further investigation of this issue will be left for another
occasion.
7.2 A note on quantifying over amounts of matter
Our ban on quantifying over sums makes the following prediction: no DP-internal quantifier
should be able to quantify over amounts of matter (recall that no adverbial quantifier is
allowed to bind a mass-variable). This prediction seems to be invalidated by examples such
as (36), borrowed from Gillon (1992:631-633), who explicitely assumes that quantification
over sums/amounts of matter25 is legitimate:
(36) a. All water is wet.
b. All information is valuable.
c. All regular mail in Canada is 38 cents.
d. No furniture on this floor has four legs.
A complete analysis of these examples (which is not provided by Gillon himself) involves at
least three factors: (i) Doetjes’s (1996, 1997) distinction between ‘count mass nouns’ such as
information, mail or furniture, and ‘mass mass nouns’ such as water, gold, etc. We should
check whether these two classes of mass nouns behave alike with respect to quantification;
(ii) We should check whether the constituents that follow the quantifiers in (36a-d) are to be
analyzed as NPs or rather as DPs headed by an empty Det: in other words, we should check

24
This observation allows us to answer the following question raised by a reviewer : why is it
that the ban on quantifying over sums does not rule out examples such as ‘Some boys fell
asleep’ ? The answer is that in this example, no sum-variable is introduced ; the binary
quantifier some quantifies over the set of atomic individuals introduced by the NP boys.
25
In Gillon’s (1992 : 632) own words, ‘[the quantifiers] range over elements in […] the
greatest aggregate in the domain of discourse of which the mass noun is true.’

23
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

whether the LF representation of all water is wet really involves quantification over random
quantities of matter (it is this LF that is predicted to be illegitimate by our proposal), or rather
examples of the type in (36a) are to be analyzed as ‘water is all/always wet’, where the mass
noun does not translate as a set of sums of matter but instead functions as a name of kind26
(for a brief analysis of generic sentences built with names of kinds, see section 8 below). (iii)
Finally, we should check whether quantification is over quantities of matter or rather over
kinds of matter, e.g., does (36a) mean ‘all the random quantities of water are wet’ or rather
‘all the kinds/any kind of water are/is wet’. This investigation must also be left open for
further research.
7. 3 Bare plurals in Romance languages
In Romance languages such as Spanish, Romanian or Italian,27 bare plurals constitute the
closest counterparts of French des-indefinites:
(37) a. Jean lisait des poèmes. (French)
John readimparfait des poems.
‘John was reading poems.’
a'. Ion citea poezii. (Romanian)
John readimparfait poems.
‘John was reading poems.’
b. Dans la cour jouaient des enfants. (French)
in the yard playedimparfait des children.
‘In the yard were playing children.’
b'. In curte se jucau copii. (Romanian)
in the yard playedimparfait children.
‘In the yard were playing children.’
We might therefore expect that bare plurals allow generic readings in those contexts in which
des-indefinites do so. This expectation is not fulfilled. Instead, what we observe is that

26
It might turn out that this type of analysis is allowed only in languages such as English,
where bare mass nouns can be analyzed as names of kind.
27
Most of the examples of bare plurals in this section are from Romanian. The reader is
referred to Dobrovie-Sorin & Laca (2003) for examples from Italian and Spanish. Bare nouns
in Brazilian Portuguese, which show very different properties (Schmitt & Munn (1999)), will
be left aside here.

24
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Romance bare plurals are more constrained than French des-indefinites :28
(38) a. *Drepte paralele se încruci_eaz_ la infinit. (Romanian)
lines parallel SE cross at the infinite.
‘Parallel lines cross each other at the infinite.’
b. *Drepte convergente au un punct în comun.
lines convergent have a point in common.
‘Convergent lines have a point in common.’
c. *T_ri învecinate au adesea interese opuse.
countries neighboring have frequently interests opposed.
‘Neighboring countries frequently have conflicting interests.’
The contrast between Romanian examples such as (38) and their grammatical French
counterparts built with des (see examples of the type in (2)) is however not problematic for
our theory. To explain the ungrammaticality of (38) all we need is to recall that although
Romance bare plurals are indefinite-like expressions, they cannot be assimilated to
indefinites.29 For the problem at hand, the minimal assumption that is required is that bare
plurals differ from indefinite plurals insofar as they do not supply group-variables:
(39) Romance bare plurals cannot supply group-variables.
This negative characterization leaves open the choice between property-denotation (cf. van
Geenhoven (1996) as well as some of the theorists quoted in footnote 29) and sum-
denotation. For arguments in favor of analyzing existential bare plurals as sum-variables the
reader is referred to Dobrovie-Sorin (2007).
In sum, Romance bare plurals differ from French des-indefinites in so far as they cannot
supply group-variables. This impossibility can be accounted for by assuming that in
languages with articles, overt articles are needed in order for DPs to be able to refer to
individualizable entities (i.e., to entities that denote in an unordered domain, e.g., singular
entities or groups). This generalization captures the fact that many languages have an article

28
Note that the bare plurals in (38a-c) are modified by adjectives, and as such their
ungrammaticality cannot be attributed to the syntactic constraint regarding the preverbal
subject position (Contreras (1986), Longobardi (2001), among others).
29
Carlson’s (1977a,b) generalizations regarding the necessary narrow scope of English bare
plurals were shown to hold of Romance bare plurals (and mass nouns) by Laca (1990),
McNally (1995, 2004), Dobrovie-Sorin (1997a,b), Dobrovie-Sorin & Laca (1996, 2003),
Longobardi (2001), Delfitto (2002) and Bleam (2006), among others.

25
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

for singular indefinites but not for plural indefinites: in the unmarked cases, singular
indefinites refer to atomic entities, whereas plural indefinites refer to sums.
7. 4 Italian degli/dei-indefinites vs. Italian bare plurals
Italian is interesting insofar as it has both existential bare plurals and degli/dei-indefinites,
which are comparable to des-indefinites. Our analysis correctly predicts that the
crosslinguistic contrast between French and the other Romance languages can be observed
inside Italian itself (the glosses for (40)-(41) are the same as those of (38)):
(40) a. *Rette parallele s’incrociano all’infinito.
b. *Rette convergente hanno un ponto in commune.
c. *Paesi limitrofi hanno spesso rapporti difficili.
(41) a. Degli/dei rette parallele s’incrociano all’infinito.
b. Degli/dei rette convergente hanno un ponto in commune.
c. Degli/dei paesi limitrofi hanno spesso rapporti difficili.
The contrast illustrated in (40)-(41), which has escaped the attention of Italian scholars,
shows that Longobardi (2001) is wrong in assuming that the generic readings of bare plurals
are parallel to the generic readings of (singular) indefinites. Under the account proposed here,
the ‘truly generic’ readings (to be distinguished from ‘pseudo-generic’ readings illustrated in
(42) below) of bare plurals are ruled out, because bare plurals cannot supply variables over
primitive entities (and more precisely groups), in contrast with both singular indefinites and
degli/dei indefinites, which supply variables over singular individuals and groups,
respectively.
We have so far considered examples in which bare plurals are arguably directly bound
by adverbs of quantification. Compare examples such as (42), borrowed from Longobardi
(2001):
(42) Elefanti di colore bianco hanno creato sempre/spesso in passato grande curiosità.
elephants of color white have created always/often in the past big curiosity
‘White colored elephants have created always/often in the past a lot of curiosity.’
Note that this example, as well as all of Longobardi’s (2001) examples are of the type shown
in (3), which we have set aside as involving adverbial quantification over events combined
with the indirect binding of plural indefinites (or of bare plurals). Because in examples of this
type, the plural indefinite is not directly bound by the Q-adverb, it is out of the reach of the
ban against quantifying over sums. This explains why bare plurals, which can be represented
only as sum-variables, can appear in examples such as (42), but not in examples such as (40),
where they must be directly bound by the Q-adverb.

26
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

In sum, Longobardi is right to say that the generic readings of Italian (and more
generally, Romance) bare plurals necessarily rely on quantification over events, but he is
wrong to believe that this is also true for DPs headed by overt indefinite (singular or plural)
determiners.30 The latter can also be directly bound by Q-adverbs, as shown in (41) for
degli/dei, and by numerous examples of generic singular indefinites, e.g., A student is usually
intelligent.
Insofar as the distinction between directly and indirectly bound indefinites is crucial, the
account proposed here argues against the uniform analysis of adverbs of quantification as
quantifiers over events and in favor of more flexible approaches according to which they may
quantify over either individuals or events (Lewis 1975, Kamp 1981, Heim 1982, Chierchia
1995b, Krifka & alii 1995).

7.4 English generic bare plurals are not indefinites


The generalizations reached above regarding the generic readings of French des-indefinites
and Romance bare plurals shed serious doubt on the predominant analysis of English,
according to which generic bare plurals are plural indefinites (Burton-Roberts 1977, Diesing
(1992), Kratzer (1988, 1995)) bound by a Q-adverb, in particular by the GEN operator. If
English generic bare plurals were to be analyzed as plural indefinite-like expressions of
unspecified cardinality, they should behave on a par with French des-indefinites (and
Romance bare plurals), and therefore examples such as (43), corresponding to (44), should be
unacceptable (as indeed are their Romance counterparts). This is not what we find:31
(43) a. Cats are rarely intelligent.

30
This assumption, which is not necessary for Longobardi’s line of argumentation, was
borrowed from the predominant semantic analysis, according to which adverbs of
quantification uniformly quantify over events (Rooth 1985, 1995, Schubert & Pelletier 1987,
1988, de Swart 1991, 1996).
31
Because she endorses the current analysis of the English examples in (43), de Swart (1993,
1996) suggests that the limitations shown by French des-indefinites are due to supplementary
restrictions linked to the fact that des/du goes back to a partitivity marker, which would be
incompatible with the quasi-universal, generic interpretation. Partitivity is often invoked in
connection with the generic reading of des. But insofar as this notion is indeed relevant, it
holds in exactly the same way for indefinite bare plurals (see Longobardi (2001) on the
generic reading of Italian bare plurals). In other words, des-indefinites are partitive compared
to definite plurals, but this is a general property of indefinites, not of des in particular.

27
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

b. Indians usually die young.


c. Romanian students speak French.
(44) a. *Des chats sont rarement intelligents.
b. *Des Indiens meurent en général jeunes.
c. *Des étudiants roumains parlent français.
(the glosses of (44a-c) are the English examples in (43a-c) prefixed with des)
Given the unacceptability of the plural indefinite subjects in the French examples in (44), the
English bare plurals in (43) cannot be analyzed as indefinite-like expressions. We may
instead assume that they function as names of kinds (as first proposed by Carlson (1977a,b)),
and more precisely as intensional maximal sums of individuals, which can be obtained by
applying a nominalizing operator labelled ‘Down’ to the set of pluralities denoted by a plural
noun (Chierchia (1998)).32 This means that the French counterparts of (43) are not the
examples in (44), but rather those in (45), with definite plural subjects (see Dobrovie-Sorin &
Laca (1996, 1998)):33
(45) a. Les chats sont rarement intelligents
b. Les Indiens meurent en général jeunes.
c. Les étudiants roumains parlent français.
(the glosses of (45a-c) are the English examples in (43a-c) prefixed with ‘the’)
What we need to explain now is why examples of the type in (43) and (45), built with kind-
referring expressions, allow for (overt or covert) adverbial quantification, whereas examples
built with plural indefinites do not.

8. Names of kinds and quantification over spatiotemporal locations


According to the current view, generic sentences built with names of kinds and Q-adverbs

32
Chierchia’s analysis of English bare plurals in terms of intensional maximal sums (which
was extended to Romance definite plurals by Dobrovie-Sorin & Laca (1998)) makes it
possible to distinguish these expressions from kind-referring singular definites, which can be
analyzed as relying on an iota operator that applies to a property of kinds (Dayal (2004)).
33
According to Dobrovie-Sorin & Laca (1996, 1998), Chierchia’s (1998) ‘down’ operator,
which can be null in English, is necessarily realized as a definite article in Romance
languages. A similar proposal can be found in Dayal (2004). For a different analysis of the
crosslinguistic variation between Romance languages (and Hungarian) and English, see de
Swart & Farkas (2005), Farkas & de Swart (2007).

28
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

are to be represented as shown in (45’), in which the name of kind has been shifted to the set
of atomic individuals that realize the kind (Chierchia (1998) among others) and the Q-adverb
quantifies over those individuals; FEW is the translation of rarement ‘rarely’:
(45’) a. FEWx (x is a cat) [x is intelligent]
In what follows I will show that this account cannot be extended to names of kinds built with
mass nouns and I will propose that at least in this type of generic sentences, the Q-adverb
quantifies over the locations at which the kind is instantiated.
Consider the contrast shown in (46) vs (47) between definite and indefinite mass
nouns in French, which is parallel to the contrast between definite and indefinite plurals
shown in (45) vs (44) above:
(46) a. L’eau est parfois solide.
the water is sometimes solid
‘Water is sometimes solid.’
b. Le pain est souvent bon.
the bread is often good
‘Bread is often good’
(47) a. *de l’eau est parfois solide.
de l’ water is sometimes solid
b. *du pain est souvent bon.
du bread is often good
The contrast is reproduced in Romance languages such as Romanian or Spanish, where – in
the relevant contexts - definite mass nouns contrast with bare mass nouns:
(48) a. Apa este cîteodat_ solid_. (Romanian)
water-the is sometimes solid.
‘Water is sometimes solid.’
b. *Ap_ este cîteodat_ solid_.
water is sometimes solid.
A possible representation of examples of the type in (46) is the one given in (46’), built on
the model of (45’). In (46’), the name of kind has been shifted to the set of the amounts of
matter that realize it and parfois ‘sometimes’ translates as SOME:
(46’) a. SOMEX (X is a sum of water) [X is solid]
The problem is that the LF shown in (46’) is ruled out by the ban on quantification over
sums, and therefore we incorrectly predict the corresponding examples to be ungrammatical.
Note, on the other hand, that if we give up our constraint, we can no longer account for the

29
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

ungrammaticality of the examples in (47) and (48b), built with French indefinite mass nouns
and Romanian bare mass nouns, respectively.
We are thus led to conclude that the contrast between definite mass nouns and
indefinite/bare mass nouns in the context of Q-adverbs cannot be explained if we assume that
in the examples built with names of kinds, e.g., (46), the Q-adverbs bind variables over
instances of kinds.
In what follows I will propose that generic sentences built with names of kinds
(definite expressions in Romance languages and bare nouns in English) rely on kind-
predication combined with adverbial quantification over the spatiotemporal locations
(notated l) at which kinds are realized:

(46”) SOMEl (the kind-water is realized at l) [the-kind-water is solid at l)

The notion of ‘spatiotemporal location’ is borrowed from Kratzer (1988, 1995) and regarding
its definition I will remain as vague as Kratzer herself, who assumes that the Davidsonian
argument is an argument for spatiotemporal location (as first proposed by Lemmon (1967)),
but does not commit herself to the assumption that the Davidsonian argument is an event
argument (p. 128). For my present purposes, (arguments for) spatiotemporal locations cannot
be identified with events (or event-arguments), which means that l-variables may co-occur
with e-variables (see Section 9).
Going back to the representation in (46”), no event-variable is present here, because
the example is built in such a way (categorical statement built with a name of kind and an i-
level predicate) that no event-predicate can be mapped onto the restriction. But why should
we be allowed to insert a location-variable? The presence of such a variable cannot be
induced by the main predicate, which is i-level, and as such – according to Kratzer (1995) -
does not have a location argument. It seems reasonable to assume that spatiotemporal
locations may come in due to the fact that kinds can be viewed as entities that are scattered
over spatiotemporal locations. The fact that the location argument may relate to the subject
rather than to the predicate is observed by Kratzer (1995 : 155-156) herself. Thus, in an
example such as (49), corresponding to Kratzer’s (73), the tense predicate before-now,
corresponding to the past tense morpheme, can apply either to the predicate French (in which
case this predicate becomes an s-level predicate, and the corresponding interpretation is that
Henry has lost his French nationality) or to the subject Henry (in which case the main
predicate stays i-level, and before-now applies only to Henry, which means that he is dead):

30
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(49) Henry was French.

The resemblance between (49) under its second reading and (46) seems clear : in both cases,
the main predicate is i-level and in both cases the location-argument bears on the subject of
the predication.
Unlike events, which need descriptive predicates (dancing, singing, inviting, just as
being intelligent or sad), locations lack descriptions (other than the trivial sortal property of
being a spatiotemporal location). How can we then isolate a set of locations over which a Q-
adverb may quantify? The LF in (46”) gives us the answer for the particular example
considered here: we collect all the spatiotemporal locations at which a kind is realized. The
set of spatiotemporal locations can constitute an appropriate domain of quantification only if
we consider minimal locations (see Kratzer (1995 :169)). Otherwise, quantifying over
locations violates the prohibition against quantifying over sums.
The reading associated to (46”) would be: ‘at some of the minimal spatiotemporal
locations at which it is realized, the kind water is solid’. Given this analysis, examples of this
type can be assigned a truth-value by checking, for every spatiotemporal location at which
water is present, whether all the water present at that location is solid or not; the sentence in
(46a) will be said to be true if there are at least some locations at which all the water is solid.
The LF representation in (46”) is truth-conditionally equivalent to that in (46’), since
quantifying over kind-realizations is equivalent to quantifying over the spatiotemporal
locations at which the kind is realized. Crucially, however, no variables over realizations of
kinds are introduced and consequently representations such as (46”), unlike those in (46’),
are out of the reach of the ban on quantifying over sums.
The analysis proposed here for mass nouns can also account for the examples in
(43/45), built with bare plurals and definite plurals, respectively, which refer to maximal
sums of atomic individuals. Since atomic individuals are the instances of kinds at minimal
spatiotemporal instances, quantifying over locations comes out equivalent to quantifying
over the atomic individuals that compose the kind. Analyzed in this way, examples of the
type in (43/45) are however not quantificational statements about individual
children/Indians/cats, but instead they attribute properties to the corresponding kinds and
they specify the percentage of minimal locations at which those predications come out true.
Our proposal allows us to answer an interesting question raised by a reviewer
regarding examples of the type shown in (50), built with a name of kind and a collective

31
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

predicate:
(50) Students often meet in cafes.
Relevant for our present discussion is the interpretation that attributes the habitual property
of meeting in cafes to the kind ‘students’,34 which is formalized as the intensional maximal
sum of students. Under the assumption that Q-adverbs quantify over kind-realizations, the
problem is that the Q-adverb seems to be unable to find some variable that it can legitimately
bind: (i) it cannot quantify over atomic individuals, because the predicate meet is collective;
(ii) it cannot quantify over sums, because of the ban against quantifying over sums; (iii) it
cannot quantify over groups, because students is not a relational predicate. This problem does
not arise if we assume quantification over locations: as in the simpler cases examined above,
the Q-adverb quantifies over minimal locations, but because meet is a collective predicate, at
least two students must be present at each minimal location.
Summarizing, generic sentences built with kind-referring mass nouns (e.g., Romance
definite mass nouns and English bare mass nouns) can only be analyzed as relying on kind-
predication combined with quantification over the minimal spatiotemporal locations at which
the kind is realized. The same analysis can be adopted for kind-referring plurals. Depending
on whether the main predicate is collective or not, the minimal locations will contain at least
two individuals or only one individual that instantiates the kind.
9. Appendix: Pseudo-generic plural indefinites
The main body of this article was concerned with plural indefinites that are directly bound by
a Q-adverb. This Appendix is dedicated to those plural indefinites that are indirectly bound
by a Q-adverb that quantifies over events. We do not aim at an in-depth analysis of such
pseudo-generic indefinites, but only at a brief presentation of the empirical criteria that allow
us to distinguish them from the truly generic indefinites examined above.

9.1 Adverbial quantification over events and the indirect binding of plural indefinites
Consider the examples in (51):
(51) a. Des pipelettes ne se supportent pas longtemps.

34
An alternative analysis, according to which (50) relies on quantification over events, is
irrelevant here. Moreover, such an analysis is not supported by the interpretation that is
intuitively associated to (50) : this example does not say that most student-meetings occur in
cafés (as it should, if it relied on quantification over events) but rather that meeting in cafes is
a habit of students.

32
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

des chatterboxes not SE stand PAS for a long time.'


'Chatterboxes don’t stand each other for a long time.'
b. Des petites filles sont souvent en train de te préparer une surprise.
des young girls are often preparing a surprise.'
'Young girls are often preparing a surprise.'
Let us first see whether these examples can be analyzed as involving direct quantification
over (groups of) individuals. In other words, can they be assigned representations such as
those in (51')?
(51') a. GEN X (chatterboxes (X)) [X don't stand each other ..]
b. MOST X (small girls (X)) [X are preparing a surprise ..]
These LFs are derivable from the corresponding examples by applying Diesing’s Mapping
Principle (provided that the main predicates are analyzed as habitual).35 But such
representations are ruled out as illegitimate by the proposal made in this paper: (i) X cannot
be a sum-variable because sums cannot be quantified over; (ii) X cannot be a group-variable
because chatterboxes and small girls are not relational predicates.
Let us then consider an alternative analysis, according to which examples of the type in
(51) rely on quantification over events:
(51”) a. GENe (be together (e, Theme (e), Loc (e))
[do not stand each other for a long while (e, Theme (e), Loc (e))
b. MOST e (be together (e, Theme (e), Loc (e))
[are preparing a surprise (e, Theme (e), Loc (e)]
As in section 2 above, event-participants are represented as individuals obtained by applying
a function to an event; the co-domain of the function is a set of sums of chatterboxes in
(51”a) and a set of sums of small girls in (51”b).
The representation in (51”) relies on the assumption stated in (52):
(52) Plural indefinites may supply an event predicate paraphrasable by ‘act as a group/be
together at a certain Location (in space or time)’.
This proposal is inspired by Lasersohn’s (1992) analysis of together as inducing one of three
possible readings, described as ‘collectivizing’ (paraphrasable by ‘act as a group’), spatial
proximity and temporal proximity.

35
According to Dobrovie-Sorin (2002) such habitual/dispositional predicates rely on a unary
HAB operator functioning as a frequency adverb (de Swart 1991), which binds the location-
variable of s-level predicates.

33
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Note that the notion of group relevant here, which would be more adequately referred to
as ‘sum participating to a group action’, should be distinguished from the notion of group
used in previous sections, where ‘group’ refers to a primitive entity in the domain of
reference. Reference to primitive groups depends on either collective nouns or (a sub-class
of) pluralized relational nouns, whereas collectivizing and proximity-based ‘groups’ merely
depend on pluralization.
It should also be observed that ‘act as a group’ or ‘be together at a certain Location’ are
not trivial properties: (i) unlike trivial properties, they cannot be supplied by default in any
syntactic context, but instead depend on the presence of a plural DP (and can be reinforced
by the presence of together and other comparable markers, see Lasersohn (1992)) ; (ii) they
are s-level properties (which have a Location-argument, as indicated in (51”)), and as such
they are incompatible with an i-level predicate being mapped onto the scope of the Q-adverb.
Coming back to the examples in (51), they can be analyzed as relying on adverbial
quantification over events of acting together or simply being together (in space or time).
The LF's in (51”a-b) correctly capture the intuitive readings that speakers associate with
(51a)-(51b): ‘whenever chatterboxes happen to be together, they don't stand each other for a
long time’ ; ‘most of the time, when small girls happen to be together, they are up to
mischief’. Compare readings such as ‘chatterboxes have the habit of not standing each other
for a long time’ or ‘most (groups of) small girls have the habit of being up to mischief’,
which would be read off the representations in (51’), but which do not seem to correspond to
the intuitive readings.
One may still wonder whether LFs of the type shown in (51”a-b) are not ruled out by our
constraint against quantifying over sums (of events). What we need is to construe events as
primitive entities belonging to an unordered domain, which can be ensured by considering
only the minimal locations at which an event takes place.
Let us next consider the examples in (53), due to Heyd (2002), who observed that the
generic reading of des-indefinites is facilitated by the presence of a nominal modifier. She
furthermore observes the contrast between (53) and (54):
(53) a. Des lions blessés sont vulnérables.
des lions injured are vulnerable
'Injured lions are vulnerable.'
b. Des enfants malades sont grincheux.
des children ill are irritable
'Ill children are irritable.'

34
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(54) a. Des maladies cardiaques sont dangereuses (OK taxonomic; ≠ générique)


des illnesses cardiac are dangerous
'Cardiac illnesses are dangerous.'
b. Des éléphants d'Afrique ont de grandes oreilles (OK taxonomic; ≠ générique)
des elephants from Africa have big ears
'African elephants have big ears.’
According to Heyd, adjectives modifying the subject of a characterizing sentence can be
represented as event-predicates occurring in the restriction only if they can function as
sentential predicates. Because modifiers such as blessé ‘injured’ and malade ‘ill’ can
function as sentential predicates, the examples in (53) can be represented as in (53’a)-(53’b):
(53’) a. GENe (wounded (e,Theme (e))) [vulnerable (e,Theme (e))]
b. GENe (ill (e, (e,Theme (e) )) [irritable (e, (e,Theme (e))]
Modifiers such as cardiaques ‘cardiac’ and d'Afrique ‘from Africa’ cannot function as
sentential predicates (*Ces maladies sont cardiaques ‘these illnesses are cardiac’; *Ces
éléphants sont d'Afrique ‘these elephants are from Africa’), and therefore they cannot
provide the descriptive content of an event-variable)36, so that the examples in (54a)-(54b)
can only be represented as in (54’a)-(54’b), the restrictions of which contain a complex
nominal predicate that provides the descriptive content of a sum-variable (predicates such as
cardiac illnesses and elephants from Africa cannot provide a restriction for (stable) groups of
individuals):
(54') a. # GENX (X is a sum of cardiac illnesses) [X is dangerous]
b. # GENX (X is a sum of elephants from Africa) [X has big ears].
These LFs are illegitimate (because sums cannot be quantified over), which accounts for the
unacceptability of the corresponding examples.37

36
It so happens that the modifiers in (54) are s-level predicates, whereas those in (55) are i-
level. Although this contrast may be relevant for a fully developed analysis of these examples,
we will not take it into account here. In other words, we assume, as does Heyd, that i-level
modifiers might function as event-predicates (Parsons (1990)). This is clearly the case in
donkey-sentences built with relatives and i-level predicates, e.g., A woman who owns a dog is
usually intelligent.
37
The (im)possibility of functioning as a sentential predicate should not be understood as
some kind of lexical feature that characterizes certain nominal modifiers as opposed to others.
Most modifiers, e.g., black or premature, can function as sentential predicates, but the nouns

35
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

One may now wonder whether examples such as (2), repeated under (55), are to be
analyzed as relying on quantification over groups of individuals, as proposed in section 5.1
above (see the LF in (55'a)) or as relying on quantification over events, as Heyd would
assume (see (55’’a)):
(55) a. Des droites convergentes ont un point en commun.
des lines convergent have a point in common
‘Convergent lines have a point in common.’
b. Des amis intimes se critiquent toujours l’un l’autre.
des friends intimate SE criticize always each other
‘Intimate friends always criticize each other.’
(55') a. GEN X (X is a group of convergent lines) [X have a point in common]
(55”) a. GEN e (be convergent (e, Theme (e)) [have a point in common (e, Theme (e))]
Both analyses might be adequate for (55a-b). However, we may find acceptable examples
with unmodified plural indefinites allowing the generic reading:
(56) a. Des soeurs rivalisent souvent.
des sisters compete often.
‘Sisters compete with each other often.’
b. Des jumeaux ont souvent des affinités.
des twins have often affinities
‘Twins often have affinities.’
(Examples borrowed from Carlier (2000:184))
Since these examples do not contain a nominal modifier (nor any other element that might
supply an event description), they can only be analyzed as relying on quantification over

they are constructed with can be chosen in such a way that the modifier acts as a classifying
(relational) predicate, rather than as a sentential (intersective) predicate:
(i) *Des chats noirs sont intelligents.
des black cats are intelligent
(ii) *Des enfants prématurés marchent rarement avant 10 mois.
des premature babies rarely walk before the age of 10 months
Because of the classifying interpretation of black and premature when they modify cats and
children respectively, these adjectives cannot supply event-predicates, which forces the Q-
adverb to illegitimately quantify over sums of black cats and premature children.

36
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

(groups of) individuals. Such a representation corresponds to the intuitive reading: ‘most
groups of sisters are groups of rivals, most groups of twins are groups of people showing
affinities among each other’.
The analysis sketched in this section and the one proposed by Heyd (2002) are alike
insofar as nominal modifiers that can function as sentential predicates are analyzed as
supplying the restrictive term of a generalization over events. The two proposals differ,
however, regarding the existence of truly generic indefinites. Heyd (2002) follows the view
(see in particular Rooth 1985, 1995 and de Swart 1991) that Q-adverbs can quantify only
over events (adverbial quantification over individuals is always indirect), whereas we assume
that Q-adverbs may also directly quantify over individuals. The problem with the current
view adopted by Heyd is that it cannot explain why examples of the type in (56) are
grammatical.
Carlier (1989) observed that modalized contexts, in particular deontic/prescriptive ones,
facilitate the generic reading of plural indefinites:
(57) a. Des jeunes filles doivent se montrer discrètes.
des young ladies must SE show discrete.'
'Young ladies must show a discrete behavior.'
b. Des hommes forts peuvent soulever une voiture.
des men strong can lift a car
'Strong men can lift a car.'
c. Des élèves ne peuvent pas travailler dans ce genre de salle
des students not can not work in this kind of room
'Students cannot work in this kind of room.'
Within the proposal made here, the grammaticality of plural indefinites in modalized
contexts can be accounted for by assuming that the modal operator induces quantification
over events (the plural indefinite being only indirectly bound), which may be related to the
fact that modals introduce quantifiers over possible worlds (see in particular Kratzer (1981)).
Modal operators do not characterize individuals per se, but rather express possibilities or
necessities given a law that characterizes the behavior of individuals in particular situations.
Against this background, the analysis that we may suggest for the examples in (57) can
be paraphrased as in (57’) below, where the if-clause contains an event-predicate that
determines the modal basis, i.e., restricts the set of possible worlds that one must consider:
(57’) a. If they are in a social situation, young ladies must be discrete.
b. If they try / whenever they try to lift a car, strong men are able to lift a car.

37
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

c. If they try / whenever they try to work in this kind of room, students cannot do
so.
A fully developed analysis of examples of this type, which is out of the scope of the present
paper, would need to make explicit the way in which modal operators trigger quantification
over events.
Summing up, plural indefinites can be indirectly bound by a Q-adverb that quantifies
over events whenever an event-predicate can be supplied in the restriction. This constraint is
quite weak, since event-predicates can be easily supplied in various ways: (i) be together/act
together at location l is a default interpretation for plural indefinites; (ii) nominal modifiers
that can function as sentential predicates may function as event-predicates; (iii) modal
operators induce quantification over events; in other contexts, not discussed here, event-
predicates may be supplied due to the presuppositions of the predicate in the nuclear scope
(Schubert & Pelletier (1987, 1988)) or as a result of Rooth’s (1985, 1995) association-with-
focus algorithm. We can thus explain why pseudo-generic plural indefinites have a larger
distribution than truly generic plural indefinites, which are subject to quite severe constraints
(see sections 4-6 above): (i) the main predicate must be an i-level predicate that selects
pluralities and (ii) the nominal predicate must provide a description for groups, a requirement
that is met by certain pluralized relational nouns, but not by object-referring nouns.

9.2 Indirect binding and number neutralization


Corblin (1987) observed that ‘strictly distributive’ readings (which can be more adequately
described as involving number-neutralization) are marginally possible with des-indefinites in
examples of the type shown in (57): in (57a) the requirement of discrete behavior applies not
only to groups of young ladies, but also to individual ladies; (57b) does not exclude the
possibility that only one very strong man is able to lift a car, and finally the unmarked
reading of (57c) attributes inability to work to any individual student in the room. Number-
neutralization also holds for the examples in (3), repeated under (58), and more generally for
all the examples examined in this Appendix (with the sole exception of those sentences in
which the predicate mapped to the scope selects pluralities):
(58) a. Méfie-toi, des guêpes énervées sont un danger terrible.
watch out, des wasps excited are a danger terrible.
‘Watch out, excited wasps are a terrible danger.’
b. Des éléphants blancs se promenant dans la rue ont toujours/parfois suscité une
très vive curiosité.

38
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

des elephants white strolling in the street have always/sometimes arousen a


very vivid curiosity.
'Always/sometimes if white elephants stroll in the street they arouse curiosity.'
The sentence in (58a) may indeed be interpreted not only as a warning against groups of
excited wasps, but also against a single wasp. In (58b), the curiosity may have been induced
not only by groups of elephants strolling in the street, but also by a single elephant.
How can we distinguish between those generic sentences that allow and those that do not
allow number neutralized readings for des-indefinites? Corblin (1987: 75-76) suggests that in
the unmarked case, des-indefinites are number-neutral. Number neutralization would be
blocked in a quite circumscribed environment: when the main verb denotes a property that
‘notoriously characterizes’ each member of a given class of individuals38. This suggestion
cannot help us understand the unacceptability of many examples, e.g., (44), which are built
with predicates that cannot be viewed as defining properties of the individuals they are
predicated of, and yet block number neutral readings.
The puzzle regarding the number-neutral readings of generic bare plurals can be solved
under the analysis proposed in this paper, which distinguishes between truly generic plural
indefinites (i.e., plural indefinites that are directly bound by a Q-adverb) and pseudo-generic
plural indefinites (i.e., plural indefinites that are indirectly bound by a Q-adverb that
quantifies over events). As explained in previous sections, the former can be represented only
as group-variables, which is incompatible with number-neutralization. Pseudo-generic
indefinites, on the other hand, are not (directly) bound by the Q-adverb (which quantifies
over events), and therefore they can be represented as sums of unspecified cardinality,
including sums made up of just one element. Hence the effect of number-neutralization.
In conclusion, by exploiting the difference between quantification over individuals and
quantification over events, our proposal is able to rule out the unacceptable examples
considered in sections 4-6 while at the same time accounting for the acceptable status of the
examples examined in this Appendix:

10. Conclusions

38
Corblin (1987) does not define the notion of ‘property that notoriously characterizes each
member of a class’; he suggests (personal communication (2002)) that the relevant properties
are those that should be listed among the defining properties of the individuals of which they
are predicated: having four (equal) sides can be viewed as an essential property of squares.

39
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

The paper showed that an adequate analysis of the generic readings of plural indefinites and
bare plurals can be given only if we assume a flexible analysis of Q-adverbs, according to
which they can quantify over either (i) (atomic or plural) individuals or (ii) minimal locations
at which kinds are realized or (iii) events, the choice between these three options depending
on the syntactic context. It was proposed that sum-denoting indefinites are ruled out if they
are directly bound by a Q-adverb, but allowed if they are indirectly bound by a Q-adverb that
quantifies over events. Plural indefinites can be directly bound by a Q-adverb only if they can
be represented as group-variables, which is possible only if the nominal predicate is
relational. An important consequence of the ban on quantifying over sums is that Q-adverbs
cannot be assumed to quantify over amounts of matter, which explains why indefinite mass
nouns cannot be truly generic. As to generic sentences built with mass-kinds, they are to be
analyzed as relying on adverbial quantification over the locations at which the kind is
realized.

References
Bleam, Tonia 2006. The Role of Semantic Type in Differential Object Marking. In L.
Tasmowski and S. Vogeleer (eds.), Proceedings of the Indefiniteness Conference,
Brussels 2005, Belgian Journal of Linguistics.
Burton-Roberts, Noel 1977. Generic sentences and analyticity. Studies in Language 1, 155-
196.
Carlier, Anne 1989. Généricité du syntagme nominal sujet et modalités, In M. Wilmet, (ed.),
Généricité, spécificité et aspect. Travaux de linguistique 19, 33-56.
Carlier, Anne 2000. Les articles du et des en synchronie et en diachronie. Revue Romane 35-
2, 177-206.
Carlson, Gregory N. 1977a. Reference to kinds in English, Diss., University of
Massachusetts.
Carlson, Gregory N. 1977b. A unified analysis of the English bare plural. Linguistics and
Philosophy 1, 413-457.
Carlson, Gregory N. and Pelletier Francis J. (eds) 1995. The generic book. Chicago:The
University of Chicago Press.
Chierchia, Gennaro 1995a. Dynamics of Meaning. Anaphora, Presuppositions and the
Theory of Grammar. Chicago: The University of Chicago Press.
Chierchia, Gennaro 1995b. Individual-level predicates as inherent generics. In G.-N. Carlson
and F.J. Pelletier (eds), 176-223.

40
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Chierchia, Gennaro 1998. Reference to Kinds across Languages. Natural Language


Semantics 6-4, 339-405.
Contreras, Heles 1986. Spanish bare NPs and the ECP. In Ivonne Bordelois, Heles Contreras,
Karen Zagona (eds.) Generative Studies in Spanish Syntax. Dordrecht: Foris.
Corblin, Francis 1987. Indéfini, défini et démonstratif. Constructions linguistiques de la
référence. Genève: Droz.
Corblin, Francis and Henriëtte de Swart, eds. 2003. A Handbook of French Semantics.
Stanford: CSLI.
Dayal, Veneeta 2004, Number marking and (in)definiteness in kind terms. Linguistics and
Philosophy 27. p. 393-450
Delfitto, Denis 2002. Genericity in Language. Issues of Syntax, Logical Form and
Interpretation. Alessandria: Dell'Orso.
Diesing, Molly 1992. Indefinites. MIT Press, Cambridge, Mass.
Dobrovie-Sorin, Carmen 1997a. Types of Predicates and the Representation of Existential
Readings. In A. Lawson (ed), Proceedings of SALT VII. Ithaca, N.Y.: Cornell
University Press, 117-134.
Dobrovie-Sorin, Carmen 1997b. Classes de Prédicats, Distribution des Indéfinis et la
Distinction Thétique-Catégorique. Le Gré des Langues XX: 58-97.
Dobrovie-Sorin, Carmen 2002. Adverbs of quantification and genericity. Proceedings of
CSSP, Paris, 2001; electronic version in Yehuda N. Falk (ed.), Proceedings of Israel
Association of Theoretical Linguistics 17, Jesuralem: The Hebrew University of
Jerusalem, Hebrew University of Jerusalem English Department.
<http://atar.mscc.huji.ac.il/~english/IATL/17/
Dobrovie-Sorin 2003. Generic Plural Indefinites and (Un)selective Binding, In F. Corblin,
Francis and H. de Swart, (eds.). A Handbook of French Semantics. Stanford: CSLI.
Dobrovie-Sorin, Carmen 2007. Existential Bare Plurals : from Properties back to Entities,
submitted to Lingua.
Dobrovie-Sorin, Carmen and Laca, Brenda 1996. Generic Bare NPs, manuscript, University
of Paris 7/Strasbourg 2.
Dobrovie-Sorin, Carmen and Laca, Brenda 1998. La génericité entre la référence à l'espèce et
la quantification générique, Actes de Langues et Grammaires III, Université de Paris 8,
165-179.

41
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Dobrovie-Sorin, Carmen and Laca, Brenda 2003. Les noms sans déterminant dans les
langues romanes. In D. Godard (ed.), Syntaxe des langues romanes, CNRS Editions:
Paris.
Dobrovie-Sorin, Carmen and Beyssade, Claire 2004. Définir les indéfinis. CNRS Editions:
Paris.
Dobrovie-Sorin, Carmen and Mari, Alda 2007a. Generic Plural Indefinites: Sums or Groups?
Proceedings of Nels 2006.
Dobrovie-Sorin, Carmen and Mari, Alda 2007b. Constraints on Quantificational Domains :
Generic Plural des-indefinites in French. Proceedings of Sinn und Bedeutung 2006.
Farkas, Donka and Sugioka, Yoko 1983. Restrictive If/When clauses. Linguistics and
Philosophy 6: 225-258.
Farkas, Donka and de Swart, Henriëtte 2007. Article Choice in Plural Generics, Lingua.
Gillon, Brendan 1992. Towards a common semantics fro English count and mass nouns.
Linguistics and Philosophy, 15, 597-640.
Godard, Danièle (ed). 2002. Syntaxe des langues romanes. CNRS Editions: Paris.
Greenberg, Yael 2007. Exceptions to Generics : Where Vagueness, Context Dependence and
Modality Interact. Journal of Semantics, 1-37.
Heim, Irene 1982. The semantics of definite and indefinite noun phrases, Diss., University of
Massachusetts.
Heyd, Sophie 2002. Prédication et interprétation générique des syntagmes nominaux en des
en position sujet, abstract of a communication. Indéfinis et prédications en français
(Paris, octobre 2002).
Higginbotham, Jim 1983. Logical Form, Binding and Nominals, Linguistics Inquiry 14, 395-
420.
Kamp, Hans 1981. A Theory of Truth and Semantic Representation, In J. Groenendijk, T.
Janssen, and M. Stokhof, (eds.), Formal Methods in the Study of Language, Amsterdam
(Mathematisch Centrum), 277-322.
Kanazawa, Makuto and Piñon, Christopher J. (eds.). Quantifiers, Deduction and Context.
Stanford: CSLI.
Keenan, Ed (ed.) 1975. Formal Semantics of Natural Language. Cambridge: Cambridge
University Press.
Kratzer, Angelika 1978. Semantik der Rede. Kontexttheorie-Modalwörter, Konditionalsätze.
Königstein: Scriptor.

42
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Kratzer, Angelika 1981. The notional category of modality. In H-J Eikmeyer and H. Reiser
(eds.), Words, Worlds, and Contexts: New Approaches to Word Semantics, 38-74.
Berlin: de Gruyter.
Kratzer, Angelika 1986. Conditionals. In A. M. Farley, P. farley and K.E. McCullough (eds.),
Papers from the Parasession on Pragmatics and Grammatical Theory, 1-15. Chicago:
Chicago Linguistic Society.
Kratzer, Angelika 1988. Stage-level and individual-level predicates. In M. Krifka (ed.), 247-
284.
Kratzer, Angelika 1995. Stage-level and individual-level predicates. In G.N. Carlson and F.
Pelletier (eds), 125-175.
Krifka, Manfred (ed.) 1988. Genericity in Natural Language. Proceedings of the 1988
Tübingen Conference, Tübingen (SNS-Bericht 88-42).
Krifka, Manfred, F.J. Pelletier, G.N. Carlson, A. ter Meulen, G. Chierchia, G. Link 1995.
Genericity: an introduction. In G.N. Carlson and F. Pelletier (eds), 1-124.
Krifka, Manfred 1995. Focus and the Interpretation of Generic Sentences. In G.N. Carlson
and F. Pelletier (eds), 238-264.
Laca, Brenda 1990. Generic objects: Some more pieces of the puzzle. Lingua 81, 25-46.
Landman, Fred 1989a. Groups I. Linguistics and Philosophy 12.6:559-605.
Landman, Fred 1989b. Groups II. Linguistics and Philosophy 12.6: 723-744.
Landman, Fred 2000. Events and Plurality. Dordrecht: Kluwer Academic Publishers.
Lasersohn, Peter 1992. Group Action and Spatio-Temporal Proximity, Linguistics and
Philosophy
Lawler, John 1973. Tracking the Generic Toad. CLS 9, 320-331. Chicago : Chicago
Linguistic Society.
Lemmon, E. J. 1967. Comments on D. Davidson’s ‘The Logical Form of Action Sentences’.
In N. Rescher, ed., The Logic of Decision and Action, 96-103. Pittsburgh : University of
Pittsburgh Press.
LePore, Ernest (ed.). New Directions in Semantics, London: Academic Press.
Le_niewski, S. 1916. Foundatations of the general theory of Sets. I. In S.L. Surma et al.
1992. Collected Works, vol. I, Dordrecht: Kluwer Academic Publishers.
Lewis, David K. 1975. Adverbs of Quantification. in Ed. Keenan (ed.), Formal Semantics of
Natural Language. Cambridge: Cambridge University Press, 3-15.

43
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

Link, Goderhard 1983. The logical analysis of plural and mass terms: a lattice theoretic
approach. In R. Bauerle, Ch. Schwarze and A. von Stechow (eds.). Meaning, Use and
Interpretation of Language. Berlin: de Gruyter.
Link, Goderhard 1984. Hydras. On the logic of relative clause constructions with multiple
heads. In F. Landman and F. Veltmann (eds.). Varieties of Formal Semantics.
Dordrecht: Foris.
Longobardi, Giuseppe 1994. Reference and Proper Names. Linguistic Inquiry 25: 609-665.
Longobardi, G. 2001. How Comparative is Semantics? A Unified Parametric Theory of
Bare Nouns and Proper Names. Natural Language Semantics 9, 335-369.
McNally, Louise 1995. Bare plurals in Spanish are Interpreted as Properties. In G. Morrill
and D. Oehrle (eds), Proceedings of ESSLI Workshop on Formal Grammar. Barcelona:
Universitat Politecnica de Catalunya, 197222. See also Catalan Journal of Linguistics
3, The Semantics of Nominals, 115-133, 2004.
Milsark, Gary 1977. Toward an Explanation of Certain Peculiarities in the Existential
Construction in English. Linguistic Analysis 3, 1-30.
Moltmann, Friedrike. 1997. Parts and Wholes in Semantics. Oxford: Oxford University
Press.
Parsons, Terry 1990. Events in the Semantics of English. Cambridge, Mass: MIT Press.
Rooth, Mats 1985. Association with Focus, Ph.D. diss., University of Amherst,
Massachusetts.
Rooth, Mats 1995. Indefinites, adverbs of quantification, and Focus Semantics. In G.N.
Carlson and F. Pelletier (eds), 265-299.
Scha, Remko 1981. Distributive, Collective and Cumulative quantification. In J.
Groenendijk, T. Janssen and M. Stokhof (eds.), Formal Methods in the Study of
Language. Mathematical Center Tracts: Amsterdam.
Schmitt, Cristina and Munn, Alan 1999. Against the Nominal Mapping Parameter: Bare
nouns in Brazilian Portuguese. Proceedings of NELS 29
Schubert, Lenhart K. and Pelletier, Francis J. 1987. Problems in the Representation of the
Logical Form of Generics, Plurals, and Masis Nouns. in: LePore (ed.), 385-451.
Schubert, Lenhart K. and Pelletier, Francis J. 1988. An Outlook on Generic Statements.: M.
Krifka (ed.), 357-372.
Schwarzschild, Roger 1992. Types of plural individuals. Linguistics and Philosophy 15, 641-
675.
Schwarzschild, Roger 1996. Pluralities, Kluwer, Dorcrecht.

44
Generic Plural (In)definites and Generic Bare Plurals: Sums or Groups?

De Swart, Henriëtte 1991. Adverbs of Quantification: A Generalized Quantifiers Approach,


Ph.D. Dissertation, Groningen.
De Swart, Henriëtte 1993. Généricité, phrases conditionnelles et l´interprétation de des N.
Faits de Langues 4.
De Swart, Henriëtte 1996. (In)definites and genericity. in M. Kanazawa and C.J. Piñon
(eds.), 171-194.
De Swart, Henriëtte & Donka Farkas 2005. Généricité et (in)définitude. Une analyse dans la
théorie de l'optimalité. In C. Dobrovie-Sorin (ed.). Noms nus et généricité, Paris:
Presses Universitaires de Vincennes, 97-126.
Simons, P. 1987. Parts: A Study in Ontology. Oxford: Oxford University Press.
Van Geenhoven, V. 1996. Semantic Incorporation and Indefinite Descriptions: Semantic and
Syntactic Aspects of West Greenlandic Noun Incorporation, Ph.D. dissertation Universität
Tübingen. Stanford: CSLI Publications, 1998.

45

Você também pode gostar