Descriptions, Presuppositional Indefinites, and Comparison Semantics of want Yu Izumi A PRIL 5, 2014 T OTAL : 5517 WORDS M AIN TEXT: 4857 N OTES : 660

Abstract One of the problems for Russell’s quantificational analysis of definite descriptions is that it generates unattested readings in the context of non-doxastic attitude verbs, such as want. The Fregean-Strawsonian presuppositional analysis of definite descriptions is designed to overcome such shortcomings of the quantificational analysis. Schoubye (2013), however, objects to the Fregean-Strawsonian solution to the problem of non-doxastic attitude verbs by generalizing the problem to indefinite descriptions. He claims that the generalized problem and presuppositional uses of indefinite descriptions demand a dynamic semantic revision of the standard view of definite and indefinite descriptions. In this paper, I defend the standard non-dynamic semantics of descriptions by refuting Schoubye’s objections to the Fregean-Strawsonian analysis. I argue that, once we take into account Villalta’s (2008) recent analysis of non-doxastic attitude verbs, we can solve Schoubye’s generalized problem. I also argue that presuppositional uses of indefinite descriptions are compatible with the standard quantificational account of indefinite descriptions. (150 words)

1

Introduction

Anders J. Schoubye (2013) poses a challenge to the standard view of the semantics of definite and indefinite descriptions. The standard view here includes both Russell’s quantificational analysis and FregeanStrawsonian presuppositional analysis of definite descriptions. Whichever analysis one would apply to definite descriptions, on the standard view, indefinite descriptions are analyzed as quantificational phrases. Schoubye attempts to motivate a dynamic semantic account of (in)definite descriptions by criticizing the standard view of descriptions. The goal of this paper is to refute his criticism and to defend the standard non-dynamic semantics of descriptions. The Fregean-Strawsonian presuppositional approach to definite descriptions overcomes many shortcomings of Russell’s quantificational approach (Heim, 1991; Heim and Kratzer, 1998; Elbourne, 2005, 2010, 2013; Glanzberg, 2009, among others). According to Schoubye, however, a broader consideration of both 1

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

definite and indefinite descriptions reveals the limitations of the Fregean-Strawsonian approach. He generalizes one major problem for Russell’s analysis of definite descriptions to some occurrences of indefinite descriptions. The problem involves the non-doxastic attitude verb want, and presuppositional uses of indefinite descriptions, in which the extensions of their descriptive contents are presupposed to be non-empty. He argues that the Fregean-Strawsonian approach is unable to address the generalized problem, and hence is inadequate. In what follows, first, I will sketch the major problem for Russell’s analysis of definite descriptions involving the desire verb want (Section 2). Second, I will present a Fregean-Strawsonian solution to the problem (Section 3). Third, I will discuss how Schoubye generalizes the problem in order to challenge the Fregean-Strawsonian approach (Section 4). Forth, I will argue that, once we adopt a recent account of nondoxastic attitude verbs proposed by Elizabeth Villalta (2008), we can adequately address the generalized problem without revising the standard view of descriptions (Section 5). Last, I will discuss presuppositions of indefinite descriptions outside of attitude ascriptions, which Schoubye regards as counterexamples to the standard view of descriptions. I will argue that Schoubye’s alleged counterexamples fall short of disproving the quantificational analysis of indefinite descriptions (Section 6).

2

The scope problem for Russell’s analysis of definite descriptions

One of the criticisms of Russell’s quantificational analysis of definite descriptions is that it makes wrong predictions when a description interacts with a non-doxastic attitude verb, such as want. The following examples were originally discussed by Irene Heim (1991), and later emphasized by Paul Elbourne (2005): (1)

a. Hans wants the banshee in his attic to be quiet tonight. b. Hans wonders whether the banshee in his attic will be quiet tonight. (Elbourne, 2005, p. 109)

According to Russell’s analysis, the descriptions in (1a-b) can, in principle, have both narrow and wide scope interpretations with respect to the attitude verbs. I schematically represent both scope possibilities as follows, where ‘ATTD’ stands for an attitude verb: (2)

a. ATTD [the x : F x] (Gx) b. [the x : F x] ATTD (Gx)

[email protected]

2

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

Consider the narrow scope structure (2a), in which the existence and uniqueness conditions are embedded inside the scope of the attitude verb. If (1a-b) could have the narrow scope logical form, then the following would have been acceptable paraphrases: (3)

a. Hans wants there to be exactly one banshee in his attic, and for it to be quiet tonight. b. Hans wonders whether the following is the case: there is exactly one banshee in his attic and that banshee will be quiet tonight.

Elbourne (2005, p. 110)

Uses of (1a-b), however, don’t seem to have the interpretations expressed by (3a-b). Thus, Russell’s quantificational analysis of definite descriptions generates unattested interpretations in non-doxastic attitude ascriptions. I will refer to this problem for Russell’s analysis ‘the scope problem’.1

3

A Fregean-Strawsonian solution to the scope problem

The scope problem for Russell’s analysis of definite descriptions is usually thought to provide support for the Fregean-Strawsonian presuppositional analysis of definite descriptions (Heim 1991; Heim and Kratzer 1998; Elbourne 2005, 2010, 2013; Glanzberg 2009). According to the Fregean-Strawsonian analysis, the existence and uniqueness conditions of a definite description are not asserted but presupposed by the speaker of a sentence containing the description. A definite description is a non-quantificational, referring expression, and it contributes an individual, not its descriptive content, to the asserted proposition. Definite descriptions don’t interact with other expressions in terms of scope. Apparent scopal properties of definite descriptions are explained away by a different mechanism, such as quantification over possible worlds or situations.2 Given the Fregean-Strawsonian account, for example, it is easy to see why (1a) isn’t synonymous with (3a). The desire report (3a) attributes to the subject the desire that there is exactly one banshee in his attic; the existence and uniqueness conditions are asserted to be part of the subject’s desire. According to the Fregean-Strawsonian account, by contrast, the existence and uniqueness conditions of the description in (1a) are merely presupposed by the speaker, not attributed to the subject, and (1a) doesn’t imply that Hans desires that there is exactly one banshee in his attic. The scope problem, therefore, doesn’t arise for the Fregean-Strawsonian account of definite descriptions.

[email protected]

3

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

4

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

Schoubye’s generalized scope problem

Schoubye (2013) rejects the sketched solution to the scope problem on the grounds that the FregeanStrawsonian approach fails to address a larger problem. He acknowledges that the sketched solution might describe the accurate truth conditions of uses of definite descriptions. He argues, however, that the scope problem is not restricted to definite descriptions, but applicable to indefinite descriptions.3 Since it addresses the scope problem only with respect to definite descriptions, the Fregean-Strawsonian solution can’t be applied to the problem involving indefinite descriptions. In other words, the scope problem generalizes to indefinite descriptions, whereas the Fregean-Strawsonian approach doesn’t. This result, according to Schoubye, gives us major motivation for adopting a dynamic semantic approach to (in)definite descriptions. In this paper, I will not evaluate his positive account of descriptions. Instead, I will resist his challenge on behalf of the Fregean-Strawsonian theorist by claiming that Schoubye fails to demonstrate the inadequacy of the standard view of descriptions.4 Let us first see how Schoubye generalizes the scope problem to indefinite descriptions. Consider (4), where an indefinite description is used together with a non-doxastic attitude verb. (4) Bertrand wants a murderer to be convicted tonight.

(Schoubye, 2013, p. 507, ex. 23)

Here is the scenario Schoubye presents for a use of (4). (CI)

Today three murderers are on trial. Bertrand has a desire that at least one of them be convicted, although he has no specific individual in mind. Bertrand is a normal person who prefers that murders were never committed. So he has no desire that there be a murderer. His desire that a murderer be convicted is conditional on the antecedent belief that the individuals on trial are in fact murderers. (Adapted from Schoubye, 2013, ‘Context I’)

According to the standard quantificational analysis of indefinite descriptions, (4) has two scope possibilities, as shown by (5a-b). (5)

a. DESb [∃x : murderer x] (convicted tonight x) b. [∃x : murderer x] DESb (convicted tonight x)

The narrow scope structure (5a) seems to state that Bertrand desires the following conjunction to hold: there is at least one murderer and he is convicted tonight. The scenario (CI), however, conflicts with such a reading because Bertrand, as a sensible citizen, has no desire that there be a murder. Furthermore, since Bertrand [email protected]

4

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

has no specific individual in mind, the wide scope interpretation (5b) is also incompatible with the scenario. Thus, the quantificational analysis cannot account for the interpretation we associate with a use of (4) in (CI). We can’t apply the Fregean-Strawsonian solution sketched in Section 3 to this example because it relies on the fact that a definite description can be analyzed as a referring expression of type ⟨e⟩, rather than a quantificational term. Schoubye claims that examples of this sort force us to abandon the standard view of descriptions.

5

How to solve the generalized scope problem

In this section, I will argue that the generalized scope problem can be handled without abandoning the standard view of descriptions. Schoubye anticipates and rejects four possible responses to the generalized scope problem from the perspective of the standard view. I want to focus on the one that is based on the semantics of propositional attitude verbs (Schoubye’s ‘Response 4’), which will be shown to be adequate to answer the scope problem.

5.1

Heim’s conditional semantics and its problem

The basic strategy of Response 4 is to provide an account of non-doxastic attitude verbs that can derive the intuitive reading we associate with a use of (4) in (CI). As Schoubye puts it, according to Response 4, the scope problem is apparent only “because the full range of technical resources at our disposal (technical resources necessary for dealing with various intensional constructions) have not yet been utilized” (Schoubye, 2013, p. 510). He frames Response 4 in terms of Heim’s (1992) analysis of non-doxastic attitude verbs, which treats a desire report as a hidden conditional statement. For example, “John wants you to leave means that John thinks that if you leave he will be in a more desirable world than if you don’t leave” (Heim, 1992, p. 193). More specifically, Heim’s conditional semantics of want can be stated as follows, where Doxα (w) stands for the set of worlds that are compatible with α’s beliefs in w: (6)

α wants ϕ is true in w iff for every w′ ∈ Doxα (w): every ϕ-world maximally similar to w′ is more desirable to α in w than any non-ϕ-world similar to w′ .

[email protected]

(Heim, 1992, p. 193)

5

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

On this account, a desire report containing want quantifies over possible worlds that are compatible with the subject’s beliefs. For example, my utterance of (7) is true when, among the worlds that I believe to be plausible, I prefer the worlds in which I teach Tuesdays and Thursdays over the worlds in which I don’t. (7) I want to teach Tuesdays and Thursdays next semester. Those worlds that I consider utterly unattainable are irrelevant to evaluating the utterance. Perhaps, it would be more desirable if I could keep all the benefits, while teaching nothing; there are other preferred alternatives in which I don’t teach Tuesdays and Thursdays. Such peculiar circumstances are, however, implausible, and they seem irrelevant to evaluating (7). That’s why Heim’s analysis quantifies over the subject’s belief worlds. What matters is, among those worlds that are similar to the subject’s belief worlds, whether the ϕ-worlds are more preferable to the non-ϕ worlds for the subject. With this in mind, one might attempt to solve the generalized scope problem in the following way. The narrow scope logical form (5a) leads to the interpretation that, among those worlds similar to his belief worlds, Bertrand prefers those worlds in which a murderer is convicted over the worlds in which no murderer is. Since Bertrand assumes there to be murderers, (5a) only considers worlds in which murderers exist, and it doesn’t imply that Bertrand prefers a murderous world to a murder-free world. Thus, Heim’s analysis can attribute a natural interpretation to a use of (4) in (CI) without revising the standard view of descriptions. This attempt is problematic, however. Consider the following there-construction (8), which corresponds to our main example (4), repeated below. (4) Bertrand wants a murderer to be convicted tonight. (8) Bertrand wants there to be a murderer and for him to be convicted tonight. There is a felt contrast between (4) and (8): while we tend to judge (4) to be true in (CI), (8) seems to be false in the same context. Since Russell’s analysis takes the embedded clauses in (4) and (8) to express the same proposition, Heim’s analysis doesn’t distinguish them in terms of truth conditions, and it predicts (4) and (8) to be equally true or false in the same context. Thus, Schoubye concludes, Russell’s quantificational analysis of indefinite descriptions is unable to solve the generalized scope problem, even if we are given an adequate account of non-doxastic attitude verbs. Heim’s (1992) analysis, however, has been independently shown to be inadequate. In what follows, I will introduce Elizabeth Villalta’s analysis of want, which replaces Heim’s analysis and enables us to [email protected]

6

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

distinguish (4) and (8), even though Russell’s analysis predicts the embedded clauses to express the same proposition.

5.2

Villalta’s comparison semantics

Villalta (2008) points out an undesirable implication of Heim’s conditional analysis of non-doxastic verbs. Consider the following invalid inference from (9a) and (9b) to (9c). (9)

a. I want to teach Tuesdays and Thursdays next semester. b. I believe that I will teach Tuesdays and Thursdays next semester if and only if I work hard now. Therefore: c. I want to work hard now.

(Villalta, 2008, p. 478)

On Heim’s semantics, if (9a), then among those worlds similar to my belief worlds, all those worlds in which I teach Tuesdays and Thursdays next semester are more desirable to me than those worlds in which I don’t. But by (9b), among my belief worlds, I work hard now at every world in which I teach Tuesdays and Thursdays next semester. Thus, according to Heim’s analysis, (9c) is entailed by (9a) and (9b), which doesn’t reflect our judgment about the inference (9). Villalta’s diagnosis of the problem is that Heim unnecessarily uses non-ϕ-worlds to describe the subject’s desire that ϕ. A better way of describing the subject’s desire is to compare ϕ-worlds and contextually given alternatives. Roughly speaking, want expresses a ternary relation among the subject, the embedded proposition, and its contextually salient alternatives. More specifically, Villalta’s proposal goes as follows: (10)

Semantics of want based on comparison of alternatives α wantsc ϕ is true with respect to variable assignment g and world w iff for every proposition ψ such that ψ ̸= ϕ and ψ ∈ g(c), for every ψ-world, there is a ϕ-world that is more desirable to α at w (where g(c) gives a set of propositions that are alternative to ϕ).5

(based on Villalta 2008, 479)

On this account, every desire report is considered with a contextually salient set of alternatives. For example, in order to evaluate my utterance of (9a), we might consider the set of alternatives {I teach Mondays, Wednesdays, and Friday next semester, I teach weekends next semester}, and we would judge it to be true when I prefer those worlds in which I teach Tuesdays and Thursdays over those worlds specified by the set of [email protected]

7

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

alternatives. It is important to realize that we normally invoke different sets of alternatives when evaluating different reports, even if the embedded propositions are equivalent. As for (9c), we might introduce a different set of alternatives, such as {I work a little but effectively now, I work moderately but productively now}. Given these alternatives, I wouldn’t in fact want to work hard now because I prefer the alternative possibilities. Thus, (9c) can be false, even if (9a) and (9b) are true. According to Villalta’s comparison semantics of want, the inference (9) is invalid because of the equivocation of want, which is a contextsensitive expression. Let us now apply this semantics to Schoubye’s generalized scope problem. First consider the main example (4) again, repeated here as (11). In the context of (CI), its narrow scope logical form is understood as expressing something like (11a). (11)

Bertrand wants a murderer to be convicted tonight. a. Bertrand desires the following to be the case: there is at least one murderer and he is convicted tonight (rather than any of the following alternatives: {There is at least one murderer and he is not convicted tonight, There is at least one murderer and he will be acquitted tomorrow, ...})

The alternative set in (11a) depicts what can happen to some of the murderers in the scenario, and it excludes the possibility that there is no murderer because Bertrand doesn’t believe in such a possibility. Next consider the corresponding there-construction (8), also repeated here as (12), whose embedded clause is truth-conditionally equivalent to that of (11). Since (8) and (12) are different desire reports, we can introduce a different set of alternatives, as in (12a). (12)

Bertrand wants there to be a murderer and for him to be convicted tonight. a. Bertrand desires the following to be the case: there is at least one murderer and he is convicted tonight (rather than the alternative: {There is no murderer})

According to the context (CI), (12a) is false. Thus, Villalta’s analysis truth-conditionally distinguishes (11) and (12), even if we assume their embedded clauses to be equivalent, as the standard view of descriptions suggests. Once Villalta’s semantics of want is accepted, Schoubye’s objection to the standard view of

[email protected]

8

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

descriptions loses its force. Russell’s analysis of indefinite descriptions can accommodate our main example (11) used in (CI).6

6

Presuppositional indefinites

In this section, I will examine another set of sentences that Schoubye presents to undermine Russell’s quantificational analysis of indefinite descriptions. I will argue that he fails to establish the incorrectness of the quantificational analysis. Let us consider his example (13), which is uttered by someone who wants to bake an apple pie. (13)

I know we have flour, sugar, cinnamon, and nutmeg. If some apples in the pantry are ripe, we should be alright.

(Schoubye, 2013, p. 508, ex. 24)

The speaker seems to take it for granted that there are apples in the pantry; (13) isn’t a hypothetical claim that, if there are apples in the pantry, and they are ripe, they should be alright. The following contrast confirms this observation. (14)

a. I know we have flour, sugar, cinnamon, and nutmeg, but I’m not sure we have any apples. #If some apples in the pantry are ripe, we should be alright. b. I know we have flour, sugar, cinnamon, and nutmeg, but I’m not sure we have any apples. If there are some apples in the pantry and they are ripe, we should be alright. (Schoubye, 2013, p. 508, ex. 25-6)

When the speaker explicitly denies the assumption that there are some apples, (13) becomes infelicitous as in (14a), whereas its there-construction counterpart (14b) is just fine. In both (13) and (14a), the indefinite description some apples is used with the presupposition that there are apples in the pantry. Following the relevant literature, I will call such a use of indefinite phrase a ‘presuppositional indefinite’ (von Fintel, 1998; Asarina, 2012; Magri, 2012). Schoubye claims that, if a presuppositional indefinite, such as some apples, is merely analyzed as expressing existential quantification, then the associated presupposition would be inexplicable. Thus, he suggests, the standard view of descriptions must be abandoned. He then posits an ambiguity in indefinite descriptions to compute out the desired truth conditions. A presuppositional indefinite is analyzed as having a presupposition because it contains the quantifier someps that, by definition, introduces a presupposition.7

[email protected]

9

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

Before discussing why presuppositional indefinites don’t threaten the standard view of descriptions, I will briefly explain why this ambiguity proposal is unattractive. Schoubye suggests that there is a distinct form of some (i.e., someps ) that has a presuppositional meaning. But this option is also available to the Russellian theorist; if she is allowed to produce the desired truth conditions by brute force of ambiguity, then she could also introduce a homophonous expression, someps , that is Russellian, but requires the extension of its nominal predicate to be non-empty.8 That would accommodate the previous examples (13) and (14a). The introduction of a new expression, however, doesn’t explain why (13) does not have a mere existential, non-presuppositional interpretation, unlike its there-counterpart; something like someps can produce the correct truth conditions, i.e., the readings available to (13) and (14a), whereas it fails to explain the relevant negative facts. If we can associate both presuppositional and non-presuppositional meanings with an occurrence of some, then why can’t we use (13) or (14a) without presupposing the existence of apples? Neither the standard quantificational analysis nor Schoubye’s alternative proposal gives us an answer to this question. Schoubye (2013, 531, note 31) fully acknowledges that he doesn’t address the question about when indefinite descriptions are associated with presuppositions. The standard quantificational analysis also makes no predictions about when presuppositional indefinites are available. Schoubye’s ambiguity proposal for presuppositional indefinites fares no better than the standard analysis. I will now explain why examples like (13) don’t pose a threat to the standard quantificational analysis of indefinite descriptions. I will argue that the observed presuppositional properties don’t originate in indefinite descriptions themselves. In other words, revising the semantics of some or a(n) alone can’t be the right solution to the problem of presuppositional indefinites. Consider (15), which is just like (14a) except that it contains different predicates. If the semantics of some/a(n) were solely responsible for the presuppositional indefinites in (13) and (14a), then (15) should be as unacceptable as (14a). (15) I know we have flour, sugar, cinnamon, and nutmeg, but I’m not sure we have any apples. If some ripe apples are found in the pantry, we should be alright. (15) can be felicitously used in the same context we have been using for (13) and (14a). Thus, neither the context nor the use of an indefinite description alone leads to a presuppositional interpretation. In (13) and (14a), it is mainly the linguistic environment constituted by the adjective ripe and the prepositional phrase in the pantry that triggers a presupposition.

[email protected]

10

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

Furthermore, compare (14-15) with the following sentences, which contain the bare plural apples in the place of some apples. (16)

a. I know we have flour, sugar, cinnamon, and nutmeg, but I’m not sure we have any apples. #If apples in the pantry are ripe, we should be alright. b. I know we have flour, sugar, cinnamon, and nutmeg, but I’m not sure we have any apples. If there are apples in the pantry and they are ripe, we should be alright. c. I know we have flour, sugar, cinnamon, and nutmeg, but I’m not sure we have any apples. If ripe apples are found in the pantry, we should be alright.

(16a) seems as infelicitous as (14a). The second sentence of (16a) carries a presupposition that there are some apples in the pantry, although it doesn’t contain an indefinite description. (16b-c) also establish that bare plurals and indefinite descriptions behave similarly in these constructions. Thus, we can’t attribute the cause of presuppositional indefinites to the semantics of some/a(n) alone. Schoubye’s claim that presuppositional indefinites demand a revision of the semantics of descriptions isn’t justified. Merely abandoning Russell’s analysis wouldn’t help us explain the relevant facts. We need to figure out how the adjective ripe (and perhaps in the pantry, too) interacts with indefinite and bare plural phrases in order to account for the above examples. The ultimate account may turn out to be compatible with Russell’s analysis. Schoubye provides no reason to believe that it must be developed in a way that excludes Russell’s analysis. Schoubye’s non-attitudinal examples, therefore, don’t pose a threat to the standard analysis of indefinite descriptions. The same point can be made using a different set of examples. The pattern we have observed is similar to a contrast between individual-level predicates (e.g., tall) and stage-level predicates (e.g., available) (Milsark, 1974; Carlson, 1977a). Molly Diesing (1992) makes a generalization that the indefinite subject of an individual-level predicate is interpreted presuppositionally. A clear example can be found in (von Fintel, 1998). (17)

a. #I’m not sure yet whether there are any mistakes at all in this book manuscript, but we can definitely not publish it if some mistakes are major.

(von Fintel, 1998, ex. 9c)

b. I’m not sure yet whether there are any mistakes at all in this book manuscript, but we can definitely not publish it if some major mistakes are found. (von Fintel, 1998, ex. 9b) [email protected]

11

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

(17a) is pragmatically odd because the context explicitly excludes the presupposition that there are mistakes in the manuscript, whereas the stage-level predicate be found out permits a non-presuppositional reading, as in (17b). Likewise, a bare plural, such as dinosaurs, when combined with an individual-level predicate, ceases to have a mere existential interpretation, as in (18a). (18)

a. Dinosaurs were warm-blooded. (✓Gen, *∃) b. Dinosaurs ate kelp. (✓Gen, ✓∃)

(Carlson, 1977b, ex. 74)

While (18b) is ambiguous between a report of a kelp-eating event and a generic assertion, (18a) has only a generic interpretation. When an individual-level predicate is the main predicate, an existential interpretation is excluded. How should we account for this contrast? If we merely revise the semantics of some/a(n), then we would fail to explain a broader pattern among different nominal expressions. We must examine the semantics and pragmatics of individual-level predicates, and understand how they interact with indefinite descriptions and bare plurals. An adequate account will also need to capture a number of other peculiarities exhibited by individual-level predicates. Unless the ultimate account is shown to be incompatible with Russell’s analysis of indefinite descriptions, examples like (17a) don’t count as evidence against Russell’s analysis. In the remainder of this section, I will discuss two possible sources of the presuppositions that we associate with indefinite descriptions. Although I don’t mean to provide a definite answer on this issue, it is not unreasonable to suppose that the standard view of descriptions is compatible with presuppositional indefinites. A first possibility is the pragmatics of discourse. Daniel B¨uring (1996) presents a purely pragmatic account of why some discourse-initial uses of quantified noun phrases, such as some cowboys in (19a), have a stronger meaning than existential quantification. The stronger meaning of (19a) is represented by the partitive construction (19b), which suggests that there is a contextually salient group of cowboys identifiable by the hearer of the utterance. (19)

a. SOME cowboys decided to go HOME

(Capitals indicate stress.)

b. ≈ Some of the cowboys decided to go home. (B¨uring, 1996, ex. 1) According to B¨uring, there is no special syntactic or semantic rule operating on some, but a general pragmatic procedure blindly computes out a question-answer structure on the basis of the alternative meanings [email protected]

12

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

available to (19a). As a result, even an ‘out-of-the-blue’ use of (19a) is understood as an answer to the question (20). (20)

What did the cowboys decide on?

(B¨uring, 1996, ex. 13)

Since (20) can be seen as an antecedent of (19a), there is a group of cowboys ‘mentioned’, in some sense, in the conversation. The speaker of (19a) accordingly assumes there to be some cowboys in the context. If a pragmatic account can be extended to presuppositional indefinites discussed above, then there would be no need to abandon Russell’s quantificational analysis of indefinite descriptions. A second possible source of presuppositional indefinites is a general difference between complex noun phrases in argument position and those in existential there-sentences. Various noun phrases, including some F , are known to have interpretations that are independent from the interpretations of the rest of the sentences (Enc¸, 1981; Musan, 1997, among many others). For example, consider the following sentences. (21)

The professors in this department are quite young. In fact, ... a. many professors were in kindergarten in the ’80s. b. # there were many professors in kindergarten in the ’80s.

(Keshet, 2008, 42)

The complex noun phrase many professors in (21a) is evaluated with respect to the speech time, which is different from the time referred to in the rest of the sentence. The speaker is talking about professors today, not those in the ’80s. By contrast, (21b) somehow excludes such an interpretation; the individuals denoted by many professors in (21b) have to be in kindergarten in the ’80s, while being professors. We can make the same observation with respect to an indefinite description some F. There is a clear contrast between a sentence containing a quantified noun phrase as an argument and its there-construction counterpart, and we have been given no reason to believe that an adequate account of this contrast needs to exclude Russell’s analysis of indefinite descriptions. Although I don’t have to be committed to any particular account of the above contrast, let me mention one such account for the sake of concreteness. Ezra Keshet (2008, 2010) attempts to account for the contrast in (21a-b) and other related facts in terms of general distributional constraints on ‘situation pronouns’, which are associated with all predicative elements in sentences, and provide domain restriction. Here is a brief sketch of his proposal. In (21a), many professors is associated with a situation pronoun that can be distinct from the situation pronoun associated with be in kindergarten in the ’80s. On the other hand, in (21b),

[email protected]

13

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

many professors and be in kindergarten in the ’80s must be associated with the same situation pronoun because of their structural relation. The independent situation pronoun attached to many professors in (21a) contextually provides a local domain with respect to which the predicate professors is evaluated, regardless of the interpretation of the main predicate. As a result, many professors in (21a) can be understood as ‘many of the professors in this department today’. On the other hand, many professors in (21b) is associated with the same situation pronoun as that for be in kindergarten in the ’80s. As a result, when evaluating (21b), we can’t appeal to a local domain that contains those individuals at the department today. (21b) is about those who were professors and in kindergarten at the same time (or situation). If we can extend this proposal to indefinite descriptions, then an occurrence of some apples in subject position is compatible with an existential presupposition because it can be understood as ‘some apples in the contextually salient domain of things’ due to an independent situation pronoun. Thus, we can account for presuppositional indefinites without revising the semantics of indefinite descriptions. No matter how we end up explaining the range of facts we have seen in this section, what is called for is not a revision of the semantics of indefinite descriptions. Therefore, contrary to Schoubye’s suggestion, his examples don’t establish the limitations of the standard quantificational view of indefinite descriptions.

7

Conclusion

In this paper, I have closely examined presuppositional indefinites, i.e., uses of indefinite descriptions that come with presuppositions that the extensions of their descriptive contents are non-empty. The standard view of descriptions, which includes the Fregean-Strawsonian presuppositional approach to definite descriptions, semantically attributes no presupposition to a use of an indefinite description. Contrary to Schoubye’s suggestion, however, such cases don’t constitute a counterexample to the standard view. First, I have discussed the cases that involve the non-doxastic attitude verb want, and shown that Villalta’s comparison semantics of want adequately explains the relevant data without revising Russell’s analysis of indefinite descriptions. Second, I have discussed the presuppositional indefinites outside of attitude ascriptions. It turned out that existential presuppositions don’t originate in the semantics of indefinite descriptions. I have mentioned two possible sources of the presuppositions, which are compatible with the standard view of descriptions. Unless the ultimate account of presuppositional indefinites excludes the standard view, it remains as a viable theoretical option.

[email protected]

14

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

Notes

1 See

(Kaplan, 2005; Neale, 2007) for an attempt to defuse the presented worry about desire verbs. For a

rejoinder to Kaplan and Neale, see (Elbourne, 2010). 2 For

example, the de dicto reading of Necessarily, the number of planets is eight can be obtained by

assuming the modal operator to bind a world or situation index associated with the description: for every world w, the number of planets in w is eight. See (Percus, 2000; Elbourne, 2005). 3 Schoubye

calls the generalized problem ‘the existence problem’ in his paper, which he also claims to

apply to pronominal expressions. 4 Elbourne

(2013, pp. 163-7) defends the Fregean-Strawsonian account on different grounds. His main

point is that Schoubye’s objection to an account of indefinite descriptions cannot constitute an objection to an account of definite descriptions. 5 The

index c associated with want might be syntactically represented, just as many argue so with respect

to domain restriction indices. Alternatively, it might belong to the domain of pragmatics. What is important is that want is now treated as a context-sensitive expression. How we should think about context-sensitivity in general is not important for our purposes. 6 One

might object that this solution is problematic because there seems to be nothing preventing us

from assigning the proposition (11a) to (12), which doesn’t have such a reading, and therefore the solution overgenerates interpretations. But this does’t undermine my conclusion above. First of all, the alternative propositions with which a desire report is evaluated are pragmatically selected, and any random set of propositions can, in principle, be used to evaluate it, just as any random individual can be the referent of a pronominal expression, as long as it is compatible with the context of use and agreement features (e.g., gender). To say that a certain occurrence of a pronoun can’t mean every possible referent is hardly an objection to the view that a pronoun is context-sensitive, and can be used to refer to various things. Likewise, to say that (12) could mean a different thing in a different context is by no means a criticism of the proposed solution.

[email protected]

15

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

Second, existential there-sentences have a number of peculiar features (as we will see some of them below). It is unreasonable to suppose that some such feature excludes a range of interpretations, even if they are compatible with the semantics of want and some/a(n) themselves. One possibility is a constraint on the distribution of noun phrases in there-sentences. English existential there-sentences often, but not always, disallow a lexically definite postverbal noun phrase (e.g., There is the cat). Many linguists have tried to account for the distribution pattern (see Francez 2007, and references therein). For example, Ward and Birner (1995) claim that ‘the postverbal NP of an existential there-sentence consistently represents a hearer-new discourse entity (p. 729, my emphasis). According to their proposal, what the noun phrase represents has to be new to the hearer in the conversation (or at least treated as if it is new). If something like this is correct, then it is no wonder that (12) can’t be used to talk about the previously mentioned murderers in the context. I want to emphasize that there is no need for me to be committed to any particular account of any particular feature of there-sentences. We just have to acknowledge that (11) and (12) are different in a variety of ways, and the semantics of indefinite descriptions cannot be held responsible for every difference. 7 See

his example (40) (Schoubye, 2013, p. 521), where the presuppositional use of an indefinite de-

scription is analyzed as having the form of ‘Aps /Someps ’. ‘Aps /Someps ’ is then defined as introducing the presupposition that the extension of the descriptive content is non-empty. 8 It

(a)

is easy to present a logically coherent lexical meaning as follows: Jsomeps K = λF⟨e,t⟩ : |F | ≥ 1 . λG⟨e,t⟩ . ∃x(F (x) ∧ G(x))

That is, someps is a partial function that takes two properties and returns Truth if and only if there is an object that has both properties, and the extension of the first property must be non-empty. If nothing has the property represented by ‘F ’, someps returns nothing because it is undefined for empty properties.

References Asarina, A. (2012). Presuppositional indefinites are positive poliarity items. In Choi, J., Hogue, E. A., Punske, J., Tat, D., Schertz, J., and Trueman, A., editors, Proceedings of the 29th West Coast Conference on Formal Linguistics, pages 11–19, Somerville, Massachusetts. Cascadilla Proceedings Project.

[email protected]

16

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

B¨uring, D. (1996). A weak theory of strong readings. In Galloway, T. and Spence, J., editors, Proceedings of the Semantics and Linguistic Theory 6, pages 17–34, Ithaca, New York. Cornell University. Carlson, G. N. (1977a). Reference to Kinds in English. PhD thesis, University of Massachusetts, Amherst, Massachusetts. Published in 1980. New York: Garland Publishing. Carlson, G. N. (1977b). A unified analysis of the English bare plural. Linguistics and Philosophy, 1(3):413– 456. Diesing, M. (1992). Indefinites. The MIT Press, Cambridge, Massachusetts. Elbourne, P. D. (2005). Situations and Individuals. The MIT Press, Cambridge, Massachusetts. Elbourne, P. D. (2010). The existence entailments of definite descriptions. Linguistic and Philosophy, 33(1):1–10. Elbourne, P. D. (2013). Definite Descriptions. Oxford University Press, Oxford. Enc¸, M. (1981). Tense without Scope: An Analysis of Nouns as Indexicals. PhD thesis, University of Wisconsin, Madison, Wisconsin. Francez, I. (2007). Existential Propositions. PhD thesis, Stanford University, Stanford, California. Glanzberg, M. (2009). Descriptions, negation, and focus. In Stainton, R. J. and Viger, C., editors, Compositionality, Context and Semantic Values: Essays in Honour of Ernie Lepore, Studies in Linguistics and Philosophy 85, pages 193–220. Springer. Heim, I. (1991). Articles and definiteness. In Stechow, A. V. and Wunderlich, D., editors, Semantics: An International Handbook of Contemporary Research. Mouton de Gruyter, Berlin. Published in German as “Artikel und Definitheit”. Heim, I. (1992). Presupposition projection and the semantics of attitude verbs. Journal of Semantics, 9(3):183–221. Heim, I. and Kratzer, A. (1998). Semantics in Generative Grammar. Blackwell, Malden, Massachusetts. Kaplan, D. (2005). Reading ‘On Denoting’ on its cenetary. Mind, 114(456):933–1003.

[email protected]

17

MANUSCRIPT

A PRIL 5, 2014

Yu Izumi

D ESCRIPTIONS , P RESUPPOSITIONAL I NDEFINITES , want

Keshet, E. (2008). Good Intensions: Paving Two Roads to a Theory of the De re/De dicto Distinction. PhD thesis, MIT, Cambridge, Massachusetts. Keshet, E. (2010). Situation economy. Natural Language Semantics, 18:385–434. Magri, G. (2012). No need for a dedicated theory of the distribution of readings of english bare plurals. In Proceedings of the Semantics and Linguistic Theory 22, pages 383–402. Milsark, G. (1974). Existential Sentencs in English. PhD thesis, Massachusetts Institute of Technology, Cambridge, Massachusetts. Musan, R. (1997). Tense, predicates, and lifetime effects. Natural Language Semantics, 5(3):271–301. Neale, S. (2007). Heavy hands, magic, and scene-reading traps. European Journal of Analytic Philosophy, 3(2):77–132. Percus, O. (2000). Constraints on some other variables in syntax. Natural Language Semantics, 8(3):173– 229. Schoubye, A. J. (2013). Ghosts, murderers, and the semantics of descriptions. No´us, 47(3):496–533. Villalta, E. (2008). Mood and gradability: an investigation of the subjenctive mood in Spanish. Linguistic and Philosophy, 31:467–522. von Fintel, K. (1998). Evidence for presuppositional indefinites. Manuscript. Ward, G. and Birner, B. (1995). Definiteness and the English Existential. Language, 71(4):722–742.

[email protected]

18

MANUSCRIPT

A PRIL 5, 2014