We are grateful to the audiences and organizers of the following events where the paper was presented: the Conference on (In)Definites and Weak Referentiality at the Federal University of Santa Catarina, Florianópolis (20 – 21 August 2012), the Colloquium of the Linguistics Department of the Hebrew University of Jerusalem (December 27, 2012), and the the Workshop on Aspect at Bar Ilan University (March 20, 2013). We are indebted to Claire Beyssade, Luka Crnič, Anita Mittwoch, and two anonymous reviewers, for their input. This work was supported by ISF grant 1157/10 to Edit Doron and Nora Boneh.
1The paper introduces a novel type of weak definites, which we call amount definites. The aim of the paper is to reformulate Poesio's original analysis of weak definites in terms of the choice-function analysis of (in)definiteness, and to explore its consequences by explaining the properties of amount definites.
2Weak definites are noun phrases which are definite in form, yet do not presuppose a unique referent, unlike regular definites. Following Poesio (1994) and Barker (2005), we view weak definite noun phrases as crucially involving relational nouns, and we extend this approach to the non-possessive weak definites discussed by Carlson and Sussman (2005). We draw attention to particular weak definites not yet found in the literature which are headed by amount nouns, including numerals and other measure nouns. Weak definites constructed from amount nouns will be called amount definites.
3The following are (attested) examples of amount definites in colloquial Modern Hebrew. In each case, the bolded noun phrase is interpreted like an indefinite, though it is definite in form and may be interpreted as definite in other environments: the three children, the five (or the hundred) Shekels, the glass of wine:
(1)
|
ha-mišpaxot
|
im
|
ha-šloša
|
yeladim
|
|
|
the-families
|
with
|
the-three
|
children
|
|
|
'the families with three children
'
|
|
(2)
|
ha-alut le-mišpaxa
|
lo
|
overet
|
et
|
ha-xamiša šqalim
|
|
the-cost per-family
|
neg
|
exceed
|
acc
|
the-five Shekels
|
|
'The cost per family does not exceed five Shekels.' (Shekel is a currency unit)
|
(3)
|
crixat ha-alkohol šelo hi bisvivot ha-kos yayin be-yom
|
|
|
consumption the-alcohol his is about the-glass wine in-day
|
|
|
'His alcohol intake is about a glass of wine a day.'
|
|
(4)
|
ha-be'ayot še nitqalim bahem
|
|
|
the-problems that one faces them
|
|
|
kše- xosxim et ha-me'a šqalim al ixsun
|
|
|
when one saves acc the-hundred Shekels on storage
|
|
|
'the problems one faces when one saves one hundred Shekels for storage.'
|
|
4Amount definites are related to amount relatives (Carlson 1977), illustrated in (5) below. In (5), one finds definiteness of the amount coupled with the indefiniteness of the substance, which characterizes (1) – (4).
(5)
|
It will take us the rest of our lives to drink the champagne that they spilled that evening. (Heim 1987: 40)
|
5The structure of the paper is as follows. We introduce weak definites in section 2 together with our analysis. It is crucial to our account that weak definites are headed by relational nouns. In Hebrew, relational nouns are idiomatically expressed in an inflectional form called the construct state. Section 3 introduces state inflection in general. Section 4 explores the connection of the construct state to relational nouns, and section 5 – its connection to the expression of definiteness and weak definiteness in Hebrew. In section 6 we turn to amount definites in colloquial Hebrew. We compare their properties to those of definite noun phrases where the amount noun is in the construct state form. We end with a semantic analysis of amount definites as weak definites.
6Weak definites are noun phrases (NPs) which are definite in form yet do not presuppose unique reference, unlike regular (strong) definites. Two subclasses of weak definites have been brought up so far in the literature. One involves possessive constructions, analysed by Poesio (1994) and Barker (2005). The other subclass is non-possessive, and was analysed by Carlson and Sussman (2005).
7The term weak definites and the original examples are due to Poesio (1994), who discussed a construction of possessive NPs with an indefinite possessor and a definite head (where the definiteness of the head is expressed either by the determiner the or by the Saxon genitive). Some examples are shown below:
(6)
|
a.
|
John got these data from the student of a famous linguist.
|
|
b.
|
My aunt got attacked by the parent of a student whom she had failed
|
|
c.
|
A bomb exploded outside the office of an American corporation.
|
|
d.
|
Bill found the wedding photo of a same-sex couple.
|
|
e.
|
He showed me the picture of a veiled woman holding a wounded relative in her arms.
|
(7)
|
a.
|
John got these data from a famous linguist's student.
|
|
b.
|
My aunt got attacked by a student's parent.
|
|
c.
|
A bomb exploded outside an American corporation's office.
|
|
d.
|
Bill found a same-sex couple's wedding photo.
|
|
e.
|
He showed me a veiled woman's picture.
|
- 1 We leave aside the strong reading, which these examples also have, where the definite NP presuppose (...)
8Poesio shows that despite the definite form of the possessive constructions in such examples, they do not presuppose uniqueness.1 It is not presupposed that the linguist in (6a) has a unique student or that the student in (6b) has a unique parent, etc. Indeed, these constructions appear in environments typical of indefinite NPs. Like indefinites, such as in (8), and unlike strong definites in (9), which are not felicitous discourse-initially, weak definites can introduce new participants into the discourse, as in (10):
(8)
|
a.
|
I met a student yesterday.
|
|
b.
|
Mary bought an office.
|
(9)
|
a.
|
#I met the student yesterday
|
|
b.
|
#Mary bought the office.
|
(10)
|
a.
|
I met the student of a famous linguist yesterday.
|
|
b.
|
Mary bought the office of an American corporation.
|
Like other weak NPs, they appear in the existential construction:
(11)
|
a.
|
There is a famous linguist's student waiting for you in the hall.
|
|
b.
|
There was a student's parent in the classroom.
|
|
c.
|
There is an American corporation's office just around the corner.
|
|
d.
|
There was the wedding photo of a same-sex couple on his desk.
|
|
e.
|
In today's paper there was the picture of a veiled woman holding a wounded relative in her arms.
|
9We may ask whether weak definites have any characteristics of definite NPs. (12a) below shows that, like definites, they allow the partitive construction, which is not the case for indefinites, cf. (12b). But notice that the examples in (12) are plural, and accordingly the definite in (12a) might actually be a strong definite:
(12)
|
a.
|
Yesterday I spoke to two of the students of a famous linguist.
|
|
b.
|
*Yesterday I spoke to two of several students of a famous linguist.
|
10Poesio (1994), and later also Barker (2005), note that there are also examples of weak definites with a definite possessor. There is no uniqueness presupposition in these examples either; it is not presupposed that the road has a single side or the kitchen a single corner etc:
(13)
|
It is safer to mount and dismount towards the side of the road, rather than in the middle of traffic.
|
(14)
|
It took him several minutes to reach the refrigerator nestled in the corner of the kitchen.
|
(15)
|
The baby’s fully-developed hand wrapped itself around the finger of the surgeon/ the surgeon's finger.
|
(16)
|
A secondary school student or the parent of the student may request that the student's name, address and telephone listing not be released.
|
(17)
|
Tell the employee of the store what you are looking for and she will provide you with options.
|
11A different class of weak definites, with no possessors whatsoever, was brought up by Carlson and Sussman (2005):
(18)
|
Mary went to the store.
|
(19)
|
I’ll read the newspaper when I get home.
|
(20)
|
Open the window, will you please?
|
(21)
|
Fred listened to the Red Sox on the radio.
|
(22)
|
Let's take the elevator.
|
12Carlson and Sussman demonstrate that these definites differ from strong definites. For example, whereas strong definites have reference that carries over in VP-ellipsis, i.e. strict identity, in the case of weak definites there is no requirement of referential identity in VP-ellipsis, and sloppy identity is possible. For instance, in (23), both people need to have heard about the same riot, but not on the same radio:
(23)
|
Mary heard about the riot on the radio, and Bob did, too.
|
|
(Carlson et al. 2006: 17)
|
13Sloppy identity in VP-ellipsis characterizes indefinites in general, and also the possessive weak definites discussed by Poesio and by Barker, such as those in (6-7) and (13-17) above.
14Possessive examples with indefinite possessors, like the ones in (6-7), are closest to indefinite NPs in interpretation. They appear in existential constructions, as was shown in (11) above, whereas the other examples do not:
(24)
|
a.
|
#There is the surgeon's finger in the picture.
|
|
b.
|
#There is the elevator.
|
- 2 Note that there are conditions on the particular objects denoted by the nouns, and thus it may not (...)
15The examples in (24) are infelicitous with an existential reading, and can only be given a "list" interpretation (Milsark 1977). On the other hand, the examples with indefinite possessors did give rise to an existential interpretation in (11). This is due to the lack of existential presuppositions of the possessors in (6-7), which contrasts with the existential presupposition of the definite possessors in (13-17) and the implicit context to which the weak indefinites are related in (18-22).2 This difference in existential presuppositions is at the basis of the difference in the felicity of the existential construction (de Jong 1987).
16Despite the difference in existential presuppositions of the possessors in the different types of examples, there is no difference in the existential presuppositions of the possessees. All these possessees equally presuppose existence. The existence presupposition is satisfied for each context of (13-22) on the basis of general knowledge of the lexicon and the world; e.g. it is normally presupposed that roads have sides, kitchens have corners, surgeons have fingers and that radios, newspapers, stores, windows etc are found within any discourse situation. As for (6-7), existence is easily accommodated (globally, in the terminology of Heim 1983).
17The puzzle of weak definites is that despite the existence presuppositions, there is no uniqueness presupposition. If there is no uniqueness in a weak definite, it is unclear how it can be expressed as a definite.
18We follow Poesio in analysing the head NP in weak definites such as (6-7) as a relation, whose denotation yields a set of values for each possessor. In (6a), for example, student is interpreted as a relation, which, when applied to some value of an indefinite linguist, returns a value which is also variable, and moreover varies with the value of the linguist. Hence, the reference of the possessee is not unique. But, according to Poesio, definiteness is satisfied by the dependence (anchoring) of the value of a relational noun on that of its possessor.
19We adopt the use of choice functions to determine the denotations both of indefinite and definite noun phrases, as proposed by von Heusinger (2004) and Schlenker (2004). Each indefinite noun phrase in the discourse is interpreted by a new choice function Fi, which accounts for the fact that each occurrence of a student in the discourse denotes a different individual. Definite noun phrases on the other hand are interpreted by a single choice function FC per context, which accounts for the fixed denotation of the student in a given discourse. The uniqueness in the latter case follows from the fact that FC always selects the most salient individual with the relevant property:
(25)
|
a.
|
a ~>
|
λP. Fi [λy. P (y)]]
|
i is a new index
|
|
b.
|
the ~>
|
λP. FC [λy. P (y)]]
|
FC selects the most salient P- individual in C
|
|
c.
|
theR ~>
|
λR. [λx. FC, x [λy. R (y, x)]]
|
FC, x selects a P-individual in C depending on x
|
20In the case of weak definites, we propose that the ordinary interpretation of the from (25b) is type-shifted so that it applies to relations rather than to properties. Thus theR in (25c) turns each relational noun into a functional noun, an idea already found in Dobrovie-Sorin (2001). Moreover, salience in C is replaced by dependence (anchoring) to the possessor x in C
21For example, the phrase the student of a famous linguist of (6a) is interpreted by combining the relation student with the functor theR whose role it is to shift relational interpretations to functional interpretations in the following manner:
(26)
|
a.
|
student ~> λx. [λy. student (y, x)]
|
|
|
theR ~> λR. [λx. FC, x [λy. R (y, x)]]
|
|
b.
|
theR-student ~> λx. FC, x [λy. student (y, x)]
|
22Student is a relation and it yields several individuals per possessor x. theR applies to this relation in (26b) and transforms it into a function. The function theR-student then applies to an individual x which is a famous linguist and yields the student of this individual which is selected by the contextual function FC, x.
- 3 Notice that body parts have to be visible and distinguishable in order to be included in the domain (...)
23The same account can be extended to examples with definite possessives such as examples (13-17). The only difference is that in these examples, the possessor is unique. Here too there are relational nouns such as side, corner, and body parts such as finger, leg, arm etc, where the object standing in the relation to the possessor is not unique. The use of the definite article is again justified by the contextual function FC, x which selects one of the various referents relative to the possessor x.3
24The same account can also be extended to non-possessive examples such as those in (18-22). The nouns here are relational nouns such as store, newspaper, radio, hospital, train etc, describing objects which have a particular conventional use per location x. These nouns are indeed interpreted as weak definites only in sentences involving this characteristic use (Carlson and Sussman 2005). The object fulfilling this characteristic use in a given context is not unique in a given location x, but a unique object is selected relative to x by FC, x. Therefore the use of a definite expression is justified here as well.
25Our account distinguishes between relational/functional nouns, where the anchoring to a possessor/user licenses a definite form, and nouns modified by adjuncts. Unlike the examples in (6-7) and (13-22) above, the bolded NPs in the examples (27-30) can only be interpreted as strongly definite, i.e. presupposed to be unique:
(27)
|
Call their attention to the book on the table.
|
(28)
|
I remember the walk with her on a clear day.
|
(29)
|
The truck was involved in the accident near a local intersection.
|
(30)
|
What is the small green leaf below a flower's petal?
|
26A relational denotation is required for a noun to be in the domain of the operator theR. We return to the characterization of relational nouns in section 4 below, but for now we illustrate the extensional nature of theR. Consider the contrast in example (31) below. This example involves the noun picture which can be interpreted either as an extensional or an intensional relation:
(31)
|
a.
|
He showed me the picture of a veiled woman holding a wounded relative in her arms.
|
|
b.
|
He showed me a picture of a veiled woman holding a wounded relative in her arms.
|
27In (31a), leaving aside the strong definite reading and concentrating on the weak reading, the noun picture is only interpreted as an extensional relation, i.e. as relating a particular individual to her picture. We claim that this is so since theR, which yields a weak definite, is only defined for extensional relations.
28On the other hand, in (31b), the noun picture is not necessarily extensional, i.e. it does not necessarily relate an actual woman to her picture. Instead it may be interpreted as intensional, with the noun phrase a veiled woman interpreted de-dicto relative to it. Under this reading, the picture depicts an imaginary woman-concept represented by the artist. Such an interpretation of a veiled woman in (31a) is not possible under the weak definite reading, and is only possible under the strong definite reading.
29Our analysis of weak definites relies on the determiner theR, realized in English as the or as 's, which turns a relational noun into a function. We now turn to the expression of relational nouns in Hebrew, where they typically appear in a particular morphological form, the construct state, a form which has special properties with respect to definiteness interpretation. In the next session, we describe the expression of definiteness in Hebrew and its relationship to the morphological inflectional category of state.
- 4 The morphological term emphatic is a Semiticists' term marking a particular value of the inflecti (...)
30The definite article ha- of Modern Hebrew (MH) historically originates as an inflectional prefix marking the emphatic state of the noun (es), in contrast to the absolute state, which lacks this prefix. Both differ from a third form of the noun, the indeterminate construct state (cs) which does not overtly mark the contrast between emphatic and absolute. These forms are part of the Central Semitic nominal inflection system described by traditional Hebraists as early as the Renaissance, e.g. Reuchlin (1506), Buxtorf (1651) and others, whereby all nouns are inflected for the category of state (in addition to familiar nominal categories such as gender and number).4
(32)
|
a.
|
absolute state
|
e.g. simla
|
'gown'
|
|
b.
|
emphatic state (es)
|
e.g. ha-simla
|
' es-gown'
|
|
c.
|
construct state (cs)
|
e.g. simla-t
|
'gown-cs'
|
- 5 We assume names and pronouns are inherently emphatic. Note also that the default form of the noun i (...)
31Both absolute state and es nouns have an inherent emphaticity value: [–emph] for the absolute state, and [+emph] for the es. cs is the form of the noun which is undetermined for emphaticity, and is marked neither as [+emph] nor as [–emph]. Thus, semantically speaking, the cs is unmarked:5
(33)
|
a.
|
absolute state
|
[–emph]
|
|
b.
|
emphatic state (es)
|
[+emph]
|
|
c.
|
construct state (cs)
|
|
32The cs noun is assigned an emphaticity value within the syntactic derivation through attachment to its annex (typically a possessor) in a construction called construct. The feature [±emph] of the annex serves to provide an emphaticity value to the construct head as well, i.e. to the cs noun.
- 6 We predict that adjacency plays a central role in the transmission of emphaticity to the cs-head. (...)
33The [+emph] feature of the es noun is typically interpreted as definite, as in (34a), and the [–emph] feature of the absolute noun as indefinite, as in (35a). In the construct, the same feature is shared by the cs-head noun as well. This results in a [+emph] interpretation of the cs-head in (34b), and a [–emph] interpretation in (35b):6
(34)
|
a.
|
ha-yalda
|
|
|
|
es-girl
|
|
|
|
[+emph]ha-yalda
|
'the girl'
|
|
b.
|
Simlat ha-yalda
|
|
|
|
gown-cs es-girl
|
|
|
|
simlat [+emph]ha-yalda
|
'the gown of the girl'
|
(35)
|
a.
|
yalda
|
|
|
|
girl
|
|
|
|
[–emph]yalda
|
'a girl'
|
|
b.
|
Simlat yalda
|
|
|
|
gown-cs girl
|
|
|
|
simlat [–emph]yalda
|
'a gown of a girl'
|
34Nouns in the absolute or emphatic state must lack an annex, as they do in (36a-b), assuming emphaticity can only be determined once. In contrast, nouns in the construct state must have an annex (usually a possessor), as shown by the contrast in grammaticality between (36c) which lacks an annex and (36d):
(36)
|
a.
|
soxaxnu
|
al
|
simla
|
|
|
|
we-spoke
|
of
|
gown
|
|
|
|
'We spoke of a gown.'
|
|
|
b.
|
soxaxnu
|
al
|
ha-simla
|
|
|
|
we-spoke
|
of
|
es-gown
|
|
|
|
'We spoke of the gown.'
|
|
|
c.
|
*soxaxnu
|
al
|
simlat
|
|
|
|
we-spoke
|
of
|
gown-cs
|
|
|
d.
|
soxaxnu
|
al simlat
|
|
ha-yalda
|
|
|
we-spoke
|
of
|
gown-cs
|
es-girl
|
|
|
'We spoke of the girl's gown'
|
|
- 7 We thus derive the well-known restriction that ha- is never prefixed to a cs noun.
Moreover, nouns in the construct state cannot be marked as emphatic, since this would yield double emphaticity marking:7
(37)
|
soxaxnu al (*ha-)simlat ha-yalda
|
|
we-spoke of (*es-)gown-cs es-girl
|
|
'We spoke of the girl's gown.'
|
35The correlation between the es form and definiteness is not a complete overlap, a problem which has generated a vast literature on the construct state aimed at theoretical accounts of the discrepancies between emphaticity (i.e. ha- inflection) and definiteness: Borer (1984, 1996, 1999), Ritter (1988), Englehardt (1998, 2000), Dobrovie-Sorin (2000, 2003), Danon (2001, 2008a, 2010), Heller (2002), Siloni (2003), Shlonsky (2004), and others.
(38)
|
ha-yalda
|
ha-gvoha
|
|
es-girl
|
es-tall
|
|
'the tall girl'
|
36A second illustration for the different domains of emphaticity and definiteness is illustrated by the es-noun in the following example, i.e. es-bride, which, though emphatic, is only definite in reading (a), but not in reading (b):
(39)
|
simlat
|
ha-kala
|
|
gown-cs
|
es-bride
|
|
a. 'the bride's gown'
|
es-bride is definite
|
|
b. 'the bridal gown'
|
es-bride is not definite
|
- 8 As shown in examples (34)-(35) above, a cs head shares the Emph feature of its annex. A NP with an (...)
37In (39), the difference in the definiteness of es-bride in (39a) vs (39b) stems from the fact that es-bride heads an NP in (39a) but does not head any NP in (39b). The structures are as follows:8
38In (40a), es-bride constitutes an NP which is the definite possessor. Accordingly, (40a) presupposes the existence and uniqueness of a bride in the context. But in (40b), es-bride is a noun modifier of the compound's head. It is not a NP, and thus, in (40b), the bride is not unique nor even presupposed to exist in the context.
39The example we now turn to has been given as an example of the seemingly indirect relation between emphaticity and definiteness (Borer 1988, Dobrovie-Sorin 2000). Yet it can be shown that this type of example does not differ from the previous example in (39). We call such examples in their (a) interpretation amount constructs.
(41)
|
kos
|
ha-yayin
|
|
glass-cs
|
es-wine
|
|
ambg: (Rosén 1957, Rothstein 2009)
|
|
a. 'the glass of wine'
|
|
b. 'the wine glass'
|
40Here it seems that in both readings, not just in the compound reading, the emphatic state noun es-wine might not be a definite NP. In reading (a), the measured noun es-wine is translated to English as the bare noun wine which is not a definite argument.
But in our view the structures for (41) exactly parallel the structures in (40) above:
41Here too, we claim, the simple relation between emphaticity and definiteness can be maintained. (42b), like (40b), consists of a compound, and it denotes a glass of a particular kind, a wine-glass. As in (40b), the emphaticity of es-wine serves to determine the emphaticity of the cs-head, which in turn determines the definiteness of the noun phrase headed by the compound.
42The interesting reading in this case is the one in (42a). Unlike the English translation in (41a), in Hebrew wine is a definite NP, headed by the emphatic-state measured noun es-wine. This definite NP denotes the maximal quantity of wine in the context (in accordance with e.g. Sharvy's 1980, Link's 1983 view of mass-term definiteness). As will be elaborated below, a cs noun combining with a definite argument NP is interpreted as a function, in this particular case a measure function restricting the amount (a glass) of the wine. The literal translation of (42a), though less idiomatic in English, would therefore be: the wine, of which the amount is a glass.
An additional example of an amount construct is given below, with a numeral head:
(43)
|
šlošet
|
ha-yeladim
|
|
three-cs
|
es-children
|
|
'the three children'
|
43Unlike the English translation in (43), in Hebrew children is a definite NP headed by the emphatic-state counted noun es-children. This definite NP denotes the maximal plurality of children in the context. The cs noun denotes the amount function which contributes a presupposition regarding the number of the children (we return to this below in section 6). The literal translation of (44) would therefore be: the children, of which the number is three.
- 9 Rather, it is the partitive interpretation of (45) which is special..
44Note that there is no way to block the definite reading of es-children in (44); it manifests itself clearly in an example such as (45) below. If we tried to generate an English-style analysis of (43), we would have to derive a partitive reading for (44) where it means three of the children, in parallel to the partitive (45). But (44) does not have a partitive reading.9
(45)
|
axad
|
ha-yeladim
|
|
one-cs
|
es-children
|
|
'one of the children'
|
- 10 Danon (2001, 2008a, 2010) maintains that annex es-nouns which are not definite are also found in co (...)
45We conclude that emphaticity corresponds to definiteness for argument/ predicate NP, but not for N which is the modifiee in an attributive relation such as adjectival modification or compound modification.10
46CS nouns in Hebrew are interpreted as relational nouns.
Cross linguistically, the most common relational (non-derived) nouns denote
-
inter-individual relations:
-
kinship (mother, uncle, cousin, grandfather, spouse)
-
socially defined (teacher, student, friend, lover, neighbour, stranger, expert, owner, colleague)
-
institutionally defined roles (captain (of a ship), capital (of a country), mayor (of a city), governor (of a province))
-
telic qualia (purpose and function)
-
use and control by owner (car, gown, pet...)
-
social institutions fulfilling particular use and purpose in given locations (hospital, school, newspaper, supermarket, public transportation)
-
abstract (behalf, sake)
-
agentive qualia
-
author, creator (picture, story)
-
individual-internal qualia relations:
-
part-whole relations:
-
body parts (hand, head, finger),
-
spatial parts (corner, side, coastline, periphery, vicinity, north, top, front, left)
-
temporal parts (beginning, end, middle)
-
membership (member, associate, inhabitant, citizen, employee)
-
intrinsic aspects of entities:
-
color, speed, weight, shape, temperature, price, size, amount...
- 11 In the periphrastic construction, both nouns appear in the absolute state or the es, and are theref (...)
47There is a cross linguistic tendency for more structural "cohesion" in relational constructions than in possessive constructions. In Hebrew, the construct state (cs) is the idiomatic form of relational nouns which allows them to appear in close association with their argument. The periphrastic construction, where the possessor is not an argument (Partee and Borschev 2001), does not seem suitable to express such relations:11
(46)
|
drom
|
ha-arec
|
? ha-darom
|
šel
|
ha-arec
|
|
south-cs
|
es-country
|
es-south
|
of
|
es-country
|
|
'the south of the country'
|
|
|
|
|
(47)
|
roš
|
ha-migdal
|
? ha-roš
|
šel
|
ha-migdal
|
|
head-cs
|
es-tower
|
es-head
|
of
|
es-tower
|
|
'the top of the tower'
|
|
|
|
|
(48)
|
txilat
|
ha-ši'ur
|
* ha-txila
|
šel
|
ha-ši'ur
|
|
start-cs
|
es-class
|
es-start
|
of
|
es-class
|
|
'the beginning of the class'
|
|
|
|
|
(49)
|
tovat
|
ha-mada'
|
* ha-tova
|
šel
|
ha-mada'
|
|
sake-cs
|
es-science
|
es-sake
|
of
|
es-science
|
|
'the sake of science'
|
|
|
|
|
- 12 There is an additional version of the construct which includes clitic doubling (Borer 1984); this v (...)
48The construct is only interpreted as relational, whereas the periphrastic possessive construction allows for contextual association between the possessor and the possessee (Rosén 1957, Doron & Meir, to appear):12
(50)
|
bnot ha-mora ha-banot
|
šel ha-mora
|
|
girls-cs es-teacher es-girls
|
of es-teacher
|
|
'the daughters of the teacher'
|
'the teacher's girls'
|
|
|
(not necessarily her daughters,
|
|
|
maybe her students, or girls
|
|
|
associated in any contextually
|
|
|
salient way)
|
(51)
|
ešet ha-cayar
|
ha-iša šel ha-cayar
|
|
woman-cs es-artist
|
es-woman of es-artist
|
|
'the wife of the artist'
|
'the artist's woman'
|
|
|
(not necessarily his wife, could be the woman he painted)
|
(52)
|
ceva ha-stav
|
ha-ceva šel ha-stav
|
|
colour-cs es-autumn
|
es-colour of es-autumn
|
|
'the colour of autumn'
|
'autumn's colour'
|
|
(the prevalent colour of nature
|
(the colour associated with autumn,
|
|
in that time of year)
|
e.g. the one in vogue in autumn fashion this year)
|
49The relation denoted by the cs noun can be constructed from a sortal noun by specifying a qualia relation. This type of relational interpretation was suggested by Heller (2002) following Vikner and Jansen (2002), as a means of coercing sortal nouns to a relational interpretation:
(53)
|
mexonit ha-šaxen
|
ha-mexonit šel ha-šaxen
|
|
car-cs es-neighbour.m
|
es-car ofes-neighbour.m
|
|
'the neighbour's car'
|
'the neighbour's car'
|
|
(the car he uses)
|
(could be the car he bet on)
|
(54)
|
glimat ha-melex
|
ha-glima šel ha-melex
|
|
gown-cs es-king
|
es-gown of es-king.
|
|
'the king's gown'
|
'the king's gown'
|
|
(he wears it)
|
(he may have ordered it for his wife)
|
50The relational association within the construct is true for derived nouns as well. In the case of event-nominalization, the construct most often takes its annex as internal rather than external argument (Rosén 1957):
(55)
|
giluy ha-meragel
|
ha-giluy šel ha-meragel
|
|
discovery-cs es-spy
|
es-discovery of es-spy
|
|
'the discovery of the spy'
|
'the spy's discovery'
|
|
(spy is internal arg only)
|
(spy may be external arg)
|
In the case of agentive-nominalization, the construct takes its annex only as internal argument:
(56)
|
roceax roš ha-memšala
|
|
murderer-cs head-cs es-government
|
|
'the murderer of the prime-minister' (prime-minister internal arg only)
|
|
ha-roceax šel roš ha-memšala
|
|
es-murderer of head-cs es-government
|
|
'the prime-minister's murderer' (could be a murderer hired by the PM)
|
51For the purposes of this article, we only discuss the interpretation of cs nouns where they head a noun phrase rather than a compound:
52Dobrovie-Sorin (2000) and Heller (2002) have suggested an interpretation of NCS as a function of type <e,e>. cs is viewed as an operation which shifts the denotation P of the noun N to the denotation PCS of NCS in the following way:
(58)
|
a.
|
PCS ~> λx. ιy P(y, x)
|
P is a relational
|
|
b.
|
PCS ~> λx. ιy [P(y) & RP (y, x)]
|
P is sortal
|
53RP is a context-independent relation determined by P which is a restricted possessive relation (unlike the contextual non-restricted possessive relation found in periphrastic possessives) which is the coerced qualia relation suggested by Heller (2002) following Vikner and Jensen (2002).
- 13 The amount constructs in (42a) and (44) above are additional examples of constructs with functional (...)
This approach is appropriate where P (or RP) is not just relational but functional, as in:13
(59)
|
a.
|
avi
|
ha-kala
|
|
|
|
father-cs
|
es-bride
|
'the father of the bride'
|
|
b.
|
beyt
|
avixay
|
|
|
|
house-cs
|
Avihai
|
'the home of Avihai'
|
54It relies on the additional presupposition that for each individual x in the domain, there is a unique y related to it by P (or RP). Yet there are cases where this presupposition fails, i.e. cases when P (or RP) is not functional but relational. These are the cases where we get weak rather than strong definiteness, as was observed by Danon (2001) (though Danon himself treats these examples as indefinites):
(60)
|
a.
|
regel
|
ha-šulxan
|
|
|
|
leg-cs
|
es-table
|
'the leg of the table'
|
|
b.
|
xalon
|
ha-mexonit
|
|
|
|
window-cs
|
es-car
|
'the window of the car'
|
|
c.
|
dod
|
ha-kala
|
|
|
|
window-cs
|
es-car
|
'the window of the car
|
|
d.
|
ovedet
|
ha-šagrirut
|
|
|
|
uncle-cs
|
es-bride
|
'the uncle of the bride'
|
|
e.
|
tošav
|
ha-ezor
|
|
|
|
inhabitant-cs
|
es-area
|
'the inhabitant of the area'
|
|
f.
|
talmid
|
ha-xug
|
|
|
|
student-cs
|
es-department
|
'the student of the department'
|
The following is an attested example:
(61)
|
ha-pinuy le-xadarmiyunye'asebe-livuy
|
|
es-evacuationto-room-cs emergencywill-be-donein-company-cs
|
|
mevugar –horeha-talmidoxaverha-cevetha-xinuxi
|
|
adult –parent-cses-studentormember-cses-teames-educational
|
|
'Evacuation to the emergency-room will be done in the company of an adult
|
|
- the parent of the student or the member of the educational team.' (internet)
|
55The cs-nouns in (60), like leg or window etc, are not semantically unique, and we attribute the definiteness of these examples to the contextual function FC,x discussed in section 2 above. For this type of relational nouns, we define cs as a relation which relates the interpretation P of NP to a property:
(62)
|
a.
|
PCS ~> λx λy P(y, x)
|
P is a relational
|
|
b.
|
PCS ~> λx λy[P(y) & RP (y, x)]
|
P is sortal
|
The weak definite reading is a shift of the interpretations in (62) to the one in (63):
(63)
|
PCS → λx. FC, x [λy. PCS (y, x)]
|
- 14 The bolded NP in (64) is indeed a weak definite, as it does not presuppose uniqueness of a dress pe (...)
56In the examples in (60) above, this shift is triggered by the presence of the feature [+emph] which originates in the annex and is interpreted as definiteness. But as in Poesio's examples, the shift in (63), yielding the weak definite reading, is also found where the possessor is indefinite. In such case, the shift is triggered by the definiteness of the clitic (-a '3fs' in (64)), which doubles the indefinite possessor and formally serves as the [+emph] annex:14
(64)
|
b-a-mexira ha-pumbit nimkera simlat-a šel saxkanit mefursemet
|
|
in-es-auction es-public was-sold dress-cs-3fs of actress famous
|
|
'The dress of a famous actress was sold in the auction.'
|
57We thus agree with the received view in the literature on the construct state (other than Danon) that a definite annex necessarily generates a definite (either weak or strong) construct.
58As shown in section 3 above, cs-nouns denoting amounts combine with referring definite NPs. Examples are shown again below, where the literal translation emphasizes the fact that these examples are constructed by merging the amount noun externally to the definite NP (rather than as part of the definite NP, as in English):
(65)
|
a.
|
šlošet
|
ha-targilim
|
b.
|
kos
|
ha-yayin
|
|
three-cs
|
es-exercises
|
|
|
glass-cs
|
es-wine
|
|
'the three exercises'
|
|
'the glass of wine'
|
literally:
|
|
'the exercises, of which there are three' 'the wine, of which there is a glass'
|
- 15 There is syncretism of the absolute and construct forms of the noun kos 'glass', as seen in (65b) a (...)
- 16 The fact that cs-nouns denoting amounts do not combine with indefinite NPs can be explained as foll (...)
On the other hand, indefinite nouns are combined with a specifier NP where the amount noun is in the absolute state:15, 16
(66)
|
a.
|
šloša targilim
|
b.
|
kos yayin
|
|
|
Three exercises
|
|
glass wine
|
|
|
'three exercises'
|
|
'a glass of wine'
|
- 17 The process of emphaticity sharing in the construct was described in (34b-35b) above. This process (...)
59In Modern Hebrew, there is a second, colloquial, construction for definite NPs with numeral/ amount nouns, in addition to the formal one shown in (65); we have called the colloquial construction amount definites. In (67a) below we repeat the formal construction (65a), and we illustrate the parallel amount definites in (67b). In the formal construction, the counted noun is in the es form; in the amount definites, it is the numeral which is in the es form, while the counted noun is in the absolute form. On principle, this difference of attachment of the definite article (the es inflectional morpheme) is simply a difference of register and is not reflected in the interpretation. It seems to correlate with the tendency to avoid the process of "emphaticity sharing" in the colloquial language. In the formal (a) constructions in (67-69) below, the construct state numeral shares the emphatic feature of its emphatic annex. In the colloquial construction in (b), the numeral is independently marked as emphatic:17
(67)
|
a.
|
šlošet
ha-targilim
|
b.
|
colloquial:
|
ha-šloša targilim
|
|
|
three-cs es-exercises
|
|
|
es-three exercises
|
|
both: 'the three exercises
|
|
|
|
(68)
|
a.
|
Xamešet ha-šqalim
|
b.
|
colloquial:
|
ha-xamiša šqalim
|
|
|
five-cs es-Shekels
|
|
|
es-five Shekels
|
|
both: 'the five Shekels'
|
|
|
|
(69)
|
a.
|
Kos ha-yayin
|
b.
|
colloquial:
|
ha-kos yayin
|
|
|
glass-cs es-wine
|
|
|
es-glass wine
|
|
both: 'the glass of wine'
|
|
|
|
60Semantically, when an indefinite with a numeral or a measure phrase such as in (66) appears within an intensional construction, as in (70) below, the numeral is always interpreted de-re, though the noun itself may be interpreted de-re or de-dicto (Heycock 1995). Thus, (70) is ambiguous, and can be interpreted such that the requirement is to solve either particular exercises, the number of which is three, or a particular number of exercises, which is three:
(70)
|
kedey
|
la'avor, carix
|
li-ftor
|
šloša
|
targilim
|
|
in-order
|
to-pass, required
|
to-solve
|
three
|
exercises
|
|
'In order to pass, one is required to solve three exercises.'
|
|
a. 'the requirement is to solve specific exercises, the number of which is three.'
|
|
b. 'the requirement is to solve a number of exercises, which happens to be three.'
|
61Based on the literal interpretation we derived for the corresponding definite in (65a), we predict that (71) unambiguously only has the first reading, which is indeed the case:
(71)
|
kedey
|
la'avor, carix
|
li-ftor
|
et
|
šlošet
|
ha-targilim
|
|
in-order
|
to-pass, required
|
to-solve
|
acc
|
three-cs
|
es-exercises
|
|
'In order to pass, one is required to solve the three exercises.'
|
|
only (a): 'the requirement is to solve the exercises, the number of which is three.'
|
62To express the (b) reading, we need the colloquial amount definite construction where the numeral is not in the cs from, but constitutes a NP which can be LF-raised on its own outside the scope of the intensional predicate (i.e. separated from its annex), so that the numeral alone is de-re. Thus the same sentence with the amount definite regains the ambiguity of the indefinite:
(72)
|
kedey
|
la'avor, carix
|
li-ftor
|
et
|
ha-šloša
|
targilim
|
|
in-order
|
to-pass, required
|
to-solve acc
|
es-three
|
exercises
|
|
|
'In order to pass, .....'
|
|
a. 'the requirement is to solve the exercises, the number of which is three.'
|
|
b. 'one must solve the required number of exercises, which is three.'
|
63Notice that the (b) readings in (70) and (72) are not exactly identical. In (72) the number of exercises is an explicit part of the requirement. We will account for that by postulating that the emphatic numeral is interpreted as salient in the context.
64Similarly, when there is a relative clause, only the amount definite gives rise to an amount relative. The relative clause in (73a), with the formal cs numeral, only denotes a property of objects, whereas the one in (73b), with the colloquial es numeral, can also be interpreted as an amount relative:
(73)
|
a.
|
patarti
|
yoter mi-
|
šlošet
|
ha-targilim
|
še-
|
at
|
patart
|
|
|
I-solved
|
more than
|
three-cs
|
es-exercises
|
that
|
you-fs
|
solved
|
|
|
'I solved more than the three exercises that you solved.'
|
|
b.
|
patarti
|
yoter mi-
|
ha-šloša
|
targilim
|
še-
|
at
|
patart
|
|
|
I-solved
|
more than
|
es-three
|
exercises
|
that
|
you-fs
|
solved
|
|
|
'I solved a higher number of exercises than the number of the three exercises that you solved.'
|
65While (73a) means that you solved three exercises and I solved the same ones and more, (73b) also means that you solved three exercises and I solved at least four.
66We claim that both the formal and colloquial numeral constructions in (67-69) are definite in form, yet the colloquial construction, the amount definite, may be interpreted as indefinite. Thus, amount definites constitute a novel type of weak definites.
- 18 Another characteristic of the colloquial construction is the approximative flavour of some exampl (...)
67The indefinite interpretation of amount definites is revealed by their acceptability in environments where only NPs with indefinite interpretation are appropriate. In such environments, e.g. (74) – (77) below, the formal numeral constructions in (a) are judged by speakers as unacceptable, even within the formal register. Only the colloquial amount definites in (b) are judged as acceptable:18
(74)
|
|
ha-mišpaxot
|
im
|
|
|
|
|
es-families
|
with
|
|
|
|
a.
|
# šlošet
|
ha-yeladim
|
|
|
|
|
three-cs
|
es-children
|
|
|
|
|
'the families with the three children'
|
|
|
b.
|
ha-šloša
|
yeladim
|
|
|
|
|
es-three
|
children
|
|
|
|
|
'the families with three children'
|
|
(75)
|
ha-alut
|
le-mišpaxa
|
lo
|
overet
|
et
|
|
es-cost
|
per-family
|
neg
|
exceeds
|
acc
|
|
a.
|
#xamešet
|
ha-šqalim
|
|
|
|
|
five-cs
|
es-Shekels
|
|
|
|
|
'The cost per family does not exceed the 5 Shekels.'
|
|
b.
|
ha-xamiša
|
šqalim
|
|
|
|
|
es-five
|
Shekels
|
|
|
|
|
'The cost per family does not exceed 5 Shekels.'
|
(76)
|
crixat ha-alkohol šelo hi bisvivot
|
|
|
Consumption es-alcohol his is about
|
|
|
a.
|
#kos ha-yayin be-yom
|
|
|
|
glass-cs es-wine in-day
|
|
|
|
'His alcohol intake is about the glass of wine a day.'
|
|
|
b.
|
ha-kos yayin be-yom
|
|
|
|
es-glass wine in-day
|
|
|
|
'His alcohol intake is about a glass of wine a day.'
|
|
(77)
|
habe'ayot še- nitqalim bahem kše- xosxim et
|
|
|
es-problems that one faces them when one saves acc
|
|
|
a.
|
#me'at ha-šqalim al ixsun
|
|
|
|
hundred-cs es-Shekels on storage
|
|
|
'the problems one faces when one saves the one hundred Shekels for storage'
|
|
|
b.
|
ha-me'a šqalim al ixsun
|
|
|
|
es-hundred Shekels on storage
|
|
|
'the problems one faces when one saves one hundred Shekels for storage.'
|
|
68We will not discuss each one of these minimal pairs separately. As an illustration, we formally present the contrast in (74) repeated here in (78).
(78)
|
a.
|
# ha-mišpaxot im šlošet ha-yeladim
|
|
|
es-families with three-cs es-children
|
|
|
'The families with the three children'
|
|
b.
|
ha-mišpaxot im ha-šloša yeladim
|
|
|
es-families with es-three children
|
|
|
'The families with three children'
|
69We propose the following semantic translations for the absolute, emphatic and construct state forms of the numeral respectively:
(79)
|
a.
|
šloša 'three' ~> lP. Fi [ly. P(y) & |y| = 3] i is new
|
|
|
where |y| denotes the number of atoms that the sum individual y consists of
|
|
b.
|
ha-šloša 'es-three' ~> λR. [lx. FC, x [ly. R (y, x) & |y| = 3]
|
|
c.
|
šlošet 'three-cs' ~> λx. FC
|
[λy. y=x & |y| = 3]
|
|
|
|
= λx. x if |x| = 3
|
70Thus, šloša in (79a) is an indefinite determiner interpreted with a new choice function; ha-šloša in (79b) is defined as the corresponding weak definite; šlošet in (79c) is a functional cs noun which serves as a test: it maps an individual to itself if this individual consists of 3 atoms. We can thus rewrite (79c) as the identity function λx. x, in case |x| = 3.
71We construct the two different interpretations of the sentences in (78), and account for their difference in acceptability:
72The oddity of (80) stems from the fact that the families in its denotation all include the same unique group of three children, i.e. the most salient plurality (typically the maximal one) of three children. But knowledge of the world tells us that different families normally have different groups of children. In (81) on the other hand, each family u has a different group of children yielded by FC, u. The weak definite nature of the numeral satisfies the requirement of definiteness agreement between the head and the modifier (Danon 2008b):
73Relational nouns R are nouns of type <e,<e,t>>, and we have proposed a particular rule which shifts their denotations to functions, i.e. type <e,e>, yielding weak definites:
(82)
|
λx. [λy. R(y, x)] → λx. FC, x [λy. R(y, x)]
|
74In English, this type-shift is triggered by the presence of the definite article the in weak definite possessive constructions such as the one in (83):
(83)
|
the leg of the table
|
75The definite article in this construction is attached to the relational head and not to the phrase as a whole, and shifts the relational noun's interpretation to a function, as in (82). This function maps the possessor to a single leg, without the presupposition that the leg is unique to begin with.
76We have argued that in Hebrew, the type-shift in (82) is triggered by the cs form of the head in combination with the es form of the annex, e.g.
(84)
|
regel ha-šulxan
|
|
leg-cs es-table
|
|
'the leg of the table'
|
77We have also discussed an additional type of weak definites, the colloquial Hebrew amount definites, where the relevant type-shift is triggered by an es numeral. For example, whe have shown that the es numeral ha-šloša 'es-three' triggers the type shift in (85) when it combines with a relational noun such as yeladim 'children'.
(85)
|
λx. [λy. R(y, x)] → λx. FC, x [λy. R(y, x) & |y| = 3]
|