Entity-denoting amount relatives definitely exist!
1 Background
In his seminal paper Carlson (1977), Carlson proposed to recognize the existence of
a large and quite diverse class of non-appositive relative constructions, which while
sharing certain properties with restrictive relative constructions, must nonetheless be
carefully distinguished from the latter in morphological, syntactic, and semantic ways,
the distinctions having either language-specific or language-independent status. The
class proposed by Carlson subsumed data like (1a-d) (some of which were brought up
in subsequent literature), which, he argued, must be distinguished from data like (1e),
where the relative clause has restrictive status.
(1) a. [The three students {(that / *who} there are _ in the office] arrived an hour ago.
b. [The 250 pounds {that / *which} you weigh _ ] endanger your health.
c. It will take us a lifetime to drink [the wine {they spilled _ that night /
that there was _ in the swimming pool when the party was over}].
c'. Israel would need a century to recruit [the (500,000) soldiers that
the Chinese paraded _ last May Day].
d. He drank [the amount of wine that his brother drank _ (of beer)].
e. [The three students {who / that} _ are in the office] arrived an hour ago.
Of the various distinguishing properties proposed by Carlson, some were argued by
Grosu and Landman (2016) (henceforth: GL) to be incorrect, but three were argued to
be correct, with certain refinements. Two of these, if correctly analyzed by GL, need
to be viewed as language-independent, and the third, as English-specific. The former
two are are: [i] The complex DP immediately containing the relative is felicitous with
definite or universal, but not with existential import, and [ii] two relatives not
separated by comma intonation may neither ‘stack’ nor coordinate with proper
intersective import. The English-specific property (which is revealed by a comparison
of (1a-b) with (1e)) is: the relative clause may be introduced by that or , but not by
who/which.
The data in (1a-d) are not uniform in all respects, and arguably require somewhat
different analyses. For recently proposed detailed analyses of these four sub-cases, see
GL (section 3.3), Grosu (2009), GL (section 3.2) and GL (section 3.1) respectively.
From the perspective of these analyses, the property shared by these various
constructions is that amounts (degrees, numbers, etc.) are somehow involved in their
analysis, and in particular, in the abstraction process, and not that the complex DPs
denote ‘pure’ amounts, degrees, or numbers. In fact, data like [the number {that /
which} she mentioned] is prime, where the complex DP denotes a pure number, are
simply restrictive relatives in which the variable abstracted over ranges over numbers,
rather than over books or people (note the acceptability of the version with which).
In contrast, in data like (1a-d), the relevant degrees are always degrees that
measure something, and the complex DPs, when denoting a degree, always denotes a
degree of something, not a pure abstract degree. Thus, the DP in (1b) does not merely
denote the weight of 250 pounds, but 250 pounds as ‘your’ weight (for
argumentation that this is so, see (3) below and the preceding text); the DPs in (1c, c'),
under a pragmatically plausible interpretation, denote an amount of wine and a
number of soldiers respectively (for brief discussion of alternative, pragmatically less
plausible construals, see section 3); as for the DP in (1d), its only possible denotation
is an amount of wine.
The example in (1a) stands somewhat apart from those in (1b-d) in that the DP
does not denote just a number of students, the identity of the students being left open,
but rather a particular sum of students possessing a certain cardinality. In view of the
fact that this type of denotation is also found with complex DPs that include
incontrovertible restrictive relatives, as in (1e), a number of colleagues have
expressed skepticism that data like (1a) need to involve relativization/abstraction over
degrees within CP, some persons voicing their qualms in personal communications,
and others doing so in print, e.g., McNally (2008). The central goal of this paper is to
provide incontrovertible empirical evidence that abstraction over degrees within the
relative clause is compatible with an individual denotation for the complex DP.
The fact that constituted the initial impetus for Carlson’s hypothesis that data like
(1a) must involve abstraction over degrees is the well-known fact (noted as early as
Milsark 1974) that an individual variable in the existential context appears to be preempted by local Existential Closure, something that makes it unavailable for
abstraction, in contrast to variables in ‘ordinary’ argument positions, as, e.g., in (1e),
where abstraction seems to be unproblematic. Carlson also observed that degree
expressions which ‘modify’ (i.e., provide the measure of) individual-denoting
expressions escape Existential Closure, so that modifying degree variables are in
principle available for abstraction. However, as Grosu and Landman (1998) observed,
such an approach gives rise to a prima facie puzzle, which Carlson did not solve: if
abstraction targets degrees, how can the complex DP end up denoting entities? To
overcome this insufficiency, Grosu and Landman (1998) proposed the following kind
of solution, which was further refined in GL: Abstraction at the relative CP level does
not target a mere variable over degrees, but rather a variable over ordered pairs of
degrees and entities they measure, the resulting abstract being subsequently mapped
by an operation of Maximalization to a singleton that contains only the pair consisting
of the maximal entity and the maximal degree in the input abstract (if there is such a
pair, the operation being undefined otherwise), and a subsequent operation called
SUBSTANCE ensuring that the complex NP translates as a singleton containing the
maximal entity, with the maximal degree that measures it implicit in it. GL explain in
detail how the assumption of Maximalization accounts for the two languageindependent properties [i]-[ii] indicated in the second paragraph of this section; brief
explanations are provided below in appropriate places.
Subsequently to GL (1998), a number of authors proposed alternative analyses of
data like (1a), which were evaluated in GL (2016, section 5), where it was argued that
two of them, due to Herdan (2008) and McNally (2008), rely on incorrect empirical
assumptions, and that a third, due to von Fintel (1999), is close to their analysis, but
unnecessarily complex.
The central goal of this paper, stated three paragraphs above, is in effect
complementary to GL (section 5.3), where McNally’s (2008) counter-proposal was
critiqued on empirical grounds. Specifically, it addresses McNally’s conceptual
objection that it is puzzling, and presumably implausible, to assume that abstraction
over degrees operates in a construction that denotes entities. We will argue, on the
basis of data from Romanian, that the kind of construction that McNally doubted the
existence of is incontrovertibly found in at least one natural language, Romanian, and
must thus be allowed by UG. This task will be carried out in section 2. In section 3,
we point out certain parallelisms between English and Romanian entity-denoting
complex DP that involve relativization of degrees, and in section 4, we examine the
different division of labor that is found in English and Romanian with respect to the
relativizers used in non-appositive externally headed relative, paying special attention
to amount relatives. Section 5 is a brief summary of results.
2 Romanian constructions with degree relative pronouns
In arguing against GL, McNally noted that while who/which are typed in English as
relativizers of entities, null operators are un-typed, and may thus be used as
relativizers not only of degrees, but also of kinds, properties, and, of course, entities.
The inventory of Romanian relativizers is different, and includes, in addition to care
‘who/which’, an inflected set drawn from the interrogative paradigm and typed for
degrees: cât ‘how-much.MSG’, câtă ‘how-much.FSG’, câţi ‘how-many.MPL’, câte
‘how-many.FPL’. Degree relative pronouns are the only option in the counterparts of
English constructions like (1b), as illustrated in (2).
(2) [(Cele)
12 kilograme {cât
/ *pe care le} cântăreşte bagajul
The.F.Pl 12 kilos
how-much ACC which CL weighs
luggage-the
tău de mână] nu reprezintă o problemă.
your of hand not represent.3 a problem
‘[*(The) 12 kilos {that / *which} your hand-luggage weighs] do not
constitute a problem.’
For completeness, we note a language-specific property of the Romanian construction
in (2). While the corresponding English construction requires an overt definite article,
as can be seen in the translation of (2), the Romanian construction does not. The full
and reduced versions of (2) are synonymous, both being construed as definite
(unsurprisingly, since 12 kilos is a unique weight)1. Data like the reduced version of
(2) were discussed in some detail in Grosu (2009), where it was noted that such data,
when the relative is not flanked by pauses, are not construed appositively, as can be
appreciated by considering (3). Just like its fluent English translation, (3) has the
potentially odd meaning that the excessive weight of the king’s hand-luggage will
somehow prevent ‘you’ from boarding the plane, rather than the more reasonable
appositive meaning: 30 kilos, which is what the king’s hand-luggage weighed, would
prevent you from boarding the plane (if your hand-luggage weighed that much).
(3) [30 de kilograme cât
a cântărit bagajul
de mână al regelui]
30 of kilos
how-much has weighed luggage-the of hand GEN king-the.GEN
te
va
împiedica pe tine să urci
în avion.
CL.2SG will.3SG prevent ACC you SBJV climb.2SG on plane
‘The 30 kilos that the king’s hand-luggage weighed will prevent you
from boarding the plane.’
Now, the degree pronouns of the kind seen in (2) may also be used in entitydenoting DPs, thereby providing what we view as incontrovertible evidence for the
existence in natural languages of the kind of construction that was viewed as
implausible by McNally. Before illustrating this state of affairs, we point to a property
of Romanian grammar that makes it hard to illustrate exact Romanian counterparts of
English data like (1a). Thus, Romanian lacks an overt dummy subject in existential
constructions, so that the counterparts of the English constructions in (4a, b) are
distinguished only by the pre- versus post-copular position of the italicized nominal.
(4) a. Doi copii
(nu) sunt în cameră.
two children (not) are in room
‘Two children are not in the room.’
b. (Nu) sunt doi copii
în cameră.
(not) are two children in room
‘There are(n’t) two children in the room.’
A consequence of this state of affairs is that one cannot construct an unambiguous
Romanian counterpart of (1a). To see this, consider (5), and note that the gap can in
principle be either pre- or post-copular.
(5) [Cei
zece soldaţi câţi
( _ ) sunt ( _ ) pe baricadă]
the.MPL ten soldiers how-many
are
on barricade
au sosit
acum o oră.
have arrived now one hour
‘The ten soldiers that (there) are on the barricade arrived an hour ago.’
This situation does not, however, prevent us from demonstrating the existence in
Romanian of the kind of construction at issue. With respect to English, it was
necessary to resort to the existential context because the null operator is in principle
compatible with both a degree and an entity interpretation, and the existential context
eliminates the potential ambiguity by blocking the entity construal. To be sure,
Carlson (1977) noted that English constructions using the relativizers that or  and
having the gap in an ordinary argument position should in principle allow, in addition
to a restrictive analysis, the kind of analysis that is appropriate for data like (1a), but
was prevented from explicitly demonstrating this by the un-typed status of the null
operator in English. In Romanian, however, the overt degree pronoun is
unambiguous, and the existence of the relevant construction can be demonstrated
regardless of the position of the gap. In fact, the existence of such constructions can
be demonstrated with relatives that do not include a copular construction, as in (6).
Note that the fluent English translation of this example, which uses a null operator, is
analytically ambiguous in a way the Romanian sentence is not.
(6) [Cei
zece studenţi câţi
aşteaptă la uşă] îşi pierd răbdarea.
the.MPL ten students how-many wait
at door REFL lose patience-the
‘The ten students that are waiting outside are losing patience.’
For completeness, we note two types of consequences of the existence of overt
degree relative pronouns in Romanian. On the one hand, these pronouns permit the
formation of certain types of relative constructions that have no English counterparts,
in particular, free relatives like those in (7a-b), whose semantic properties are
comparable to those of (1a, c) respectively.
(7) a. [Câţi
(studenţi) s-au
prezentat] au luat examenul.
how-many students REFL-have presented have taken exam-the
‘All {those / the students} who showed up passed the exam.’
b. Voi
bea [cât
(vin) bea şi Ion].
will.1.Sg drink how-much wine drinks and Ion
‘I will drink as much (wine) as Ion does.’
On the other hand, being inherently more richly specified than the null operators of
English, they also restrict the range of externally-headed relative constructions in
which they may felicitously occur; we return to this point in section 4.
These considerations notwithstanding, the crucial importance of data like (5)-(7) is
that they clearly show that the involvement of degrees in the relative-internal
abstraction process is compatible with an entity-denotation for the complex DP, and
thus constitute, we submit, incontrovertible evidence for the existence of such
constructions; more colorfully put, they have the status of a ‘smoking gun.’ We
conclude on this basis that McNally’s conceptual objection to GL’s analysis is
without force.
For completeness, we wish to make one additional remark. While it is not possible
to construct an unambiguous exact counterpart of (1a), for reasons noted in
connection with (4)-(5), it is nonetheless possible to construct an entity-denoting
complex DP whose gap unambiguously lies in a different kind of existential context;
we return to this point in section 3.
We will conclude this section by showing that data like (5)-(6) exhibit the
maximalization properties of data like (1a). Maximalization was inferred by GL on
the basis of the two language-independent effects noted in the first paragraph of this
paper, i.e., [i] infelicity of existential force for the complex DP, and [ii] unavailability
of stacking or coordination with proper intersective import. Succinctly put, GL
attributed the effect in [i] to a pragmatic clash between the uniqueness of the
singleton’s membership and the fact that existential quantification, unlike
definiteness, leaves open the possibility of non-uniqueness, and the effect in [ii] to the
fact that the intersection of two singletons is either trivial or null.
[i] is illustrated with respect to both Romanian and English by the degraded status
of the reduced version of (8) and its English translation relative to the perfect
acceptability of the corresponding full versions.
(8) [?(Cei) zece soldaţi câţi
sunt acum pe baricadă ]
the ten soldiers how-many are now on barricade
n-au
mâncat de trei zile.
not-have eaten of three days
‘[#(The) ten soldiers there are now on the barricade] have not eaten for three
days.’
A point that needs to be mentioned, however, is that the reduced version of (8)
without flanking pauses around the relative seems to us less strikingly deviant than
the corresponding English translation. We surmise that this effect is due to a marginal
possibility of assigning to the Romanian relative an appositive construal even in the
absence of flanking pauses, something that is ruled out in English by the fact that an
appositive construal is not available in this type (with the relativizers Ø or that), even
with flanking pauses; rather, the English translation of the appositive construal of the
reduced version of (8) can only be something like ten soldiers, as many as there are
now on the barricade, have not eaten for three days.
The astute reader may have noticed that the unavailability of a non-appositive
construal in the reduced versions of (8) contrasts with the availability of such a
construal in the reduced version of (2) and in (3). The reason is that in, say, the
reduced version of (2), the expression 12 kilograme ‘12 kilos’ can function as the
proper name of a unique weight, and can combine with a maximalizing relative clause
to denote 12 kilos as the weight of a particular object (Grosu 2009 points out that this
kind of modification of a proper name bears an arguable family resemblance to a
situation where some individual, call him Bill, functions as both the judge and the
hangman of a town, so that one can form combinations like the boldfaced expression
in Bill as a judge has been on strike for a week, but as a hangman, he has never
stopped working). In contrast, zece soldaţi ‘tens soldiers’ cannot function as the
proper name of a plurality of entities, so that when combining with a maximalizing
relative clause, the definite article must be used. Therefore, the full version of (8)
contrasts in truth conditions with its reduced version, under an appositive construal of
CP. Whereas in the former case, the ten soldiers who fasted for three days are
necessarily those who are on the barricade at the moment of speech, this need not be
so in the latter case, which is true if the soldiers who fasted and those on the barricade
are distinct sums of individuals who share a common cardinality.
Concerning property [ii], consider (9) and (10).
(9) [Toţi turiştii
care se aflau pe vapor la 3 pm (şi) care se
aflau
all tourists-the who REFL found on boat at 3 pm (and) who REFL found
pe insulă la 2 pm] au ajuns târziu acasă.
on island at 2 pm have arrived late home
‘[All the tourists who were on the boat at 3 pm (and) who were
on the island at 2 pm] returned home late.’
(10) [Toţi turiştii
câţi
se
aflau pe vapor la 3 pm #(şi)
all tourists-the how-many REFL found on boat at 3 pm (and)
câţi
se aflau pe insulă la 2 pm] au
ajuns târziu acasă.
how-many REFL found on island at 2 pm have arrived late home
‘[All the tourists that there were on the boat at 3 pm #(and) that there were
on the island at 2 pm] returned home late.’
Assume for both of them the following context: The individuals a, b and c were on the
boat at 3 pm and the individuals b, c and d were on the island at 2 pm. In the reduced
version of (9), if there is no comma between the relatives, both clauses are restrictive,
and their construal is necessarily intersective, so that the complex DP denotes the sum
btc. In the full version of (9), this intersective construal is also available, along with
one obtained by the union of the two relatives, in which case the complex DP denotes
the sum atbtctd. In (10), on the other hand, where we have câţi/there clauses,
intersective construals are excluded, with the result that the full version
unambiguously denotes atbtctd, and the reduced version is infelicitous.
Summarizing the results of this section, we have shown that entity-denoting
complex DPs whose relatives make incontrovertible use of abstraction over degrees
exist in at least one natural language, and that such constructions share with English
constructions like (1a) the property that crucially distinguishes them from restrictive
relative constructions, namely, maximalization of entities and of their measuring
degrees within the relative CP.
3 Entity-denoting amount relatives with the gap in an existential context
As we saw in the preceding section, the existence of entity-denoting amount
relatives can be demonstrated in Romanian without placing the gap in an existential
context, as, e.g., in (6). Nonetheless, existential contexts give rise to some interesting
effects when combined with negation and/or modals, and we will briefly examine
such effects in this section, noting parallelisms between English and Romanian.
We begin by taking a closer look at the facts in (4), reproduced below for
convenience.
(4) a. Doi copii
(nu) sunt în cameră.
two children (not) are in room
‘Two children are not in the room.’
b. (Nu) sunt doi copii
în cameră.
(not) are two children in room
‘There are(n’t) two children in the room.’
As is well-known, the indefinite expression in (4b) has narrow scope, and is thus
targeted by negation, while the indefinite expression in (4a) may have wide scope,
thereby allowing negation to target the locative PP. Now, if we construct English data
with relatives that have gaps in the position of the italicized expressions in the full
versions of (4a-b), we obtain a coherent result in the former case and an incoherent
one in the latter, as can be seen in (11).
(11) a. [The five children who are not in the room] are having lunch.
b.#[The five children that there aren’t in the room] are having lunch.
The incoherence of (11b) is due, of course, to the fact that one cannot simultaneously
deny the existence of five children and assert that they are having lunch.
Now, there is a way to ‘salvage’ DPs like the bracketed one in (11b), for example,
by placing an existential modal in the matrix, as in (12a).
(12) a. [The fifty soldiers that there weren’t on the barricade when the enemy
attacked] would have sufficed to check its advance.
b. If there had been fifty soldiers on the barricade when the enemy attacked,
those fifty soldiers would have sufficed to check its advance.
c. [The fifty soldiers who weren’t on the barricade when the enemy
attacked] would have sufficed to check its advance.
(12) avoids incoherence because a sum of fifty soldiers whose existence is denied in
the world of evaluation exists in a possible world distinct from the latter, and the
sentence is construed as saying that if those fifty soldiers had existed in the world of
evaluation, the enemy’s advance would have been checked; an essential paraphrase of
(12a) is provided in (12b). For completeness, we note that this interpretation is also
available in (12c), where the relative is restrictive, but this example also allows an
alternative construal, on which fifty soldiers is interpreted in the matrix and negation
targets the locative PP; on this reading, the fifty soldiers exist in the world of
evaluation, they just were not present on the barricade when the enemy attacked.
The effects just noted may be reproduced in Romanian by using relatives with
degree pronouns and by placing the gap in different existential contexts, such as those
of the italicized items in (13a-b).
(13) a. Actriţa
nu are 5000 de admiratori.
actress-the not has 5,000 of admirers
‘The actress doesn’t have 5,000 admirers.’
b. Actriţei
îi
lipsesc 5000 de admiratori pentru a obţine un Oscar.
actress-the.DAT CL.DAT lack 5,000 of admirers for to obtain an Oscar
‘The actress lacks 5,000 (additional) admirers (needed) for her to obtain
an Oscar.’
Data partly comparable to (11b) and (12a) are exhibited in (14) and (15) respectively.
(14) #[Cei 5000 de admiratori {pe care nu i-a /
câţi n-a} avut actriţa]
the 5,000 of admirers Acc who not Cl-has how-many not-has had actress-the
au votat pentru alta.
have voted for another-one(F)
‘# The 5,000 admirers that the actress didn’t have voted for another one.’
(15) [Cei 5000 de admiratori {care / câţi
i-au
lipsit actriţei]
the 5,000 of admirers how-many CL.DAT-have lacked actress-the.DAT
ar
fi putut să-i
asigure un Oscar.
would be could Subj-CL sufficed an Oscar
‘The 5,000 admirers that the actress lacked could have got her an Oscar.’
These dates are only partly comparable to the corresponding English data, because
care is also a possible relativizer out of existential contexts, in contrast to English
who/which, which are excluded in such cases; we return to this point in section 4.
The facts presented in this section, while orthogonal to those presented in section 2,
strengthen the general view that appeal to degrees within the relative clause is not
inconsistent with an individual denotation of the complex DP. The specific point
made here is that the deviance of (11b) and (14) is traceable to the fact that the
cardinality properties of the external nominal are necessarily interpreted within the
relative, more precisely, in the scope of CP-internal negation.
4 The division of labour between relativizers in English and Romanian
As noted in section 1, the English relativizers who/which are appropriate in
externally-headed non-appositive constructions only when interpreted as abstractors
over a variable of the type of entities; this makes them inappropriate for amount
relatives of all sub-types. In contrast, the null operator is appropriate, as far as we can
tell, in all cases of the kind under consideration, and thus in all amount relatives.
In Romanian, the situation is somewhat different, probably due to the fact that the
degree pronouns are inherently specified in certain ways (null operators are
presumably unspecified). By and large, such pronouns are appropriate only when the
CP-external material that belongs to the complex DP provides some characterization,
whether precise or approximate, of the cardinality/size of the denotatum of the
complex DP. This point was illustrated with respect to data comparable to (1b) in
Grosu (2009), and we will provide illustrations concerning data comparable to
(1a,c,d) in this section. We note that the inability to use degree pronouns in some
constructions parallel to (1a,c,d) is compensated in most cases by an ability to use
care, which thus has a much wider range of use than the English elements who/which,
being excluded just in constructions like (1b) (see (2)).
With respect to data like (1a), where the DP denotes an individual (sum) identical
to the one defined within the relative, note that in (5)-(6), the cardinality indicator is
the numeral zece ‘ten’, and in (10), it is the universal quantifier toţi ‘all’. In all these
cases, the relativizer may also be care, with no difference in truth conditions, the
difference, if any, being pragmatic, in the sense that the speaker wishes to focus on
cardinality/amount when using degree pronouns2.
If the numerals in (5)-(6) are suppressed, the CP-external part of the DP being
replaced with soldaţii ‘the soldiers’ and studenţii ‘the students’ respectively, and if the
CP-external material in (10) is simply turiştii ‘the tourists’, these examples become
significantly degraded. Such data may, however, be salvaged by using care.
Concerning data comparable to (1c-d), reproduced below for convenience, we have
provided no examples so far. Such data contrast with data like (1a) in that there is no
necessary identity between the individuals/substance defined within the relative and
the individuals/substance denoted by the complex DP, identity being limited to
cardinalities/amounts of individuals/stuff.
[1] c. It will take us a lifetime to drink [the wine {they spilled _ that night /
that there was _ in the swimming pool when the party was over}].
c'. Israel would need a century to recruit [the (500,000) soldiers that
the Chinese paraded _ last May Day].
d. He drank [the amount of wine that his brother drank _ (of beer)].
Note that (1c) allows a pragmatically reasonable reading, which concerns drinking an
amount of wine identical to the one that was spilt last night/at the party, as well as a
pragmatically implausible one, which concerns drinking the very spilt wine.
Similarly, (1c') has a reading that involves recruiting a number of soldiers identical to
the one paraded by the Chinese, as well as a reading that involves recruiting those
very individuals. As for (1d), allows only the former type of reading. We provide in
(16) identity-of-amount/cardinality data in which degree pronouns are acceptable, as
the restrictions on their use are respected, and in (17), data in which the use of degree
pronouns is unacceptable, due to violations of those restrictions. In virtually all these
cases, the use of care is permitted3.
(16) a. Ne
va trebui o veşnicie ca să bem
cei 80 de litri de vin
us.DAT will need an eternity that SBJV drink.1PL the 80 of liters of wine
{pe care i-au
/ cât
au}
băut ei aseară.
ACC which CL-have.3PL how-much have.3PL drunk they last-night
‘We will need an eternity to drink the 80 liters of wine they drank last night.’
b. Ne
va trebui o veşnicie ca să recrutăm milioanele de soldaţi
us.DAT will need an eternity that SBJV recruit.1PL millions-the of soldiers
{pe care le-au
/ câţi
au}
recrutat chinezii
într-un
ACC who CL-have.3PL / how-many have.3PL recruited Chinese-the in-a
singur an.
single year
‘We will need an eternity to recruit the millions of soldiers that the Chinese
recruited in a single year.’
(17) a. Ne
va trebui o veşnicie ca să bem
vinul
us.DAT will need an eternity that SBJV drink.1PL wine-the
{pe care l-au
/ *cât
au}
băut ei aseară.
ACC which CL-have.3PL how-much have.3PL drunk they last-night
‘We will need an eternity to drink the 80 liters of wine they drank last night.’
b. Ne va trebui o veşnicie ca să recrutăm soldaţii {?pe care
us will need an eternity that SBJV recruit.1PL soldiers-the ACC which
i-au
/
*câţi
au}
recrutat chinezii
într-un singur an.
CL-have.3PL / how-many have.3PL recruited Chinese-the in-a
single year
‘We will need an eternity to recruit the millions of soldiers that the Chinese
recruited in a single year.’
c. A arătat aproape curajul
{pe care l-a
/*cât
a}
has shown almost courage-the ACC which CL-has how-much has
arătat tatăl
lui.
shown father-the his
‘He showed almost the (amount of) courage that his father did.’
5 Summary and results
In this paper, we have presented evidence from Romanian that complex DPs
containing a relative clause in which the relativization/abstraction process
incontrovertibly targets degrees may have an individual denotation. We have also
shown that such DPs exhibit precisely the maximalization effects and the cardinalityreconstruction effects that are found in English data like (1a). If so, the analysis
proposed in GL (section 3.3) for English data like (1a) seems highly appropriate for
Romanian data like (5)-(6), and is in any event conceptually unobjectionable for the
former, contra McNally (2008).
References
Carlson, Gregory, 1977a. Amount relatives. In: Language 53, pp. 520-542.
Grosu, Alexander, 2009. Two kinds of degree-denoting relatives: Hebrew versus
Romanian. Brill’s Annual of Afroasiatic Languages and Linguistics, vol. 1
Grosu, Alexander and Fred Landman, 1998. Strange relatives of the third kind. In:
Natural Language Semantics 6, pp. 125-170.
Grosu, Alexander and Fred Landman, 2016. Amount relatives. The Blackwell
Companion to Syntax II, Chapter 7. Downloadable from Alexander Grosu’s
website.
Herdan, Simona, 2008. A superlative theory of amount relatives. In: Chang, Charles and
Hannah Haynie (eds.), Proceedings of the West Coast Conference on Formal
Linguistics WCCFL 26. Cascadilla Proceedings Project, Somerville.
McNally, Louise, 2008. DP-internally only, amount relatives, and relatives out of
existentials. In: Linguistic Inquiry 39, pp. 161-169.
Milsark, Gary, 1974. Existential Sentences in English. Ph.D. Dissertation, MIT.
von Fintel, Kai, 1999. Amount relatives and the meaning of chains. Handout of a talk
given at the University of Edinburgh.
1
We are aware of only one significant difference between data like (2) with and without the definite
article, and it is morpho-syntactic in nature. Thus, in a complex DP that serves as subject of its matrix,
if the definite article is present, the matrix verb necessarily agrees in number with the DP, and if the
definite article is absent, agreement is optional. To illustrate, note that in (3), where the head noun
kilograme is morphologically plural, the matrix auxiliary va 'will' is singular. It is possible to replace
this form with vor 'will.Pl', but if we prefix cele 'the.F.Pl' to the DP, the plural form of the auxiliary is
obligatory.
Presumably, the expression 30 de kilograme may be viewed as denoting either a group of units of
one kilo each, and thus, an atom (of weight), or a sum of such units, and thus, a plurality. The latter
option is, we assume, what licenses the use of a plural definite article in the full version of (2), which,
as head of the complex DP subject, coerces plural agreement on the matrix verb.
2
In (10), as indicated in relation to (9), the use of care allows an additional intersective construal,
which is unavailable with a degree pronoun.
3
We do not know why (17b) is slightly degraded with care, but whatever the reason, the version with
cât seems to us distinctly worse.