SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. A Field of Views
  5. 3. Against the Predicate Modifier Approach
  6. 4. In Defense of the Modal Account of the Progressive
  7. 5. Concluding Remarks
  8. References

When we talk about creation, we use the progressive and verbs of creation as in ‘Mary is building a house’. The modal account of the progressive says that a sentence such as ‘Mary is building a house’ is true just in case Mary eventually builds a house in all worlds in which her house-building proceeds normally. Recently, the modal account has come under fire from those who claim that it over-generates modal entailments and those who think the progressive should be treated as a predicate of events. By properly situating the modal account within a contemporary semantic framework for natural language modals, I argue that it gives the best semantic treatment of the progressive.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. A Field of Views
  5. 3. Against the Predicate Modifier Approach
  6. 4. In Defense of the Modal Account of the Progressive
  7. 5. Concluding Remarks
  8. References

Natural language provides many ways of expressing modal information. When talking about a favorite team that lost the previous day, it would be natural to say, ‘They could have won. If they had scored in the second half, then victory would have been ours!’ These sentences convey information about what could have been the case and not just about what actually is the case. Expressions that typically appear in sentences used to convey modal information include ‘possibly’, ‘necessarily’, ‘might’, ‘must’, ‘could’, ‘would’, and so forth. For the most part, this is uncontroversial.

What is controversial, on the other hand, is the extent to which natural language expressions beyond these usual suspects convey modal information. Although it might be natural to take a sentences such as ‘The Netherlands could have won the World Cup’ as expressing modal information, what about the sentence ‘John is building a house’, which is a typical example of an English sentence in the progressive? On the surface, this seems to be about what is actually the case and not about what might be the case. Terrance Parsons (1980), Zoltán Szabó (2008) and Graeme Forbes (2006) all argue that the English progressive is not a modal expression. Contrary to these philosophers, I argue that the progressive does convey modal information. Defending a modal account the progressive is important for our general understanding of the extent to which modality is intertwined with natural language.

To begin the case for the modal account of the progressive, consider the following two examples, both of which are well known from the literature. First, the sentence ‘Judy is crossing the street’ does not entail ‘Judy crossed the street’ since she might have changed her mind and turned back. Second, whereas ‘John baked a cake’ entails that there is a cake such that John baked it, the sentence ‘John is baking a cake’ does not entail that there is a cake such that John is baking it. If it is early enough in the process, there might only be a bowl of batter.

The modal account of the progressive explains these failures of entailment. For now, let the account state that a progressive sentence such as ‘Judy is crossing the street’ is true at a world w and time t just in case Judy eventually crosses the street in all worlds w′ that are like w up until t but in which things proceed normally after t.1 Since this account does not require that the actual world be in the set of worlds mentioned, the entailment to Judy eventually crossing the street as well as the entailment to there being a cake such that John is baking it are both blocked. In the first case, events in the actual world might not proceed normally and so Judy may never cross the street. In the latter case, since the truth of the progressive is dependent on what happens at later times in other worlds, no actual cake is needed before the baking process is completed. So it seems that the modal approach is treading on solid territory.

Nevertheless, there have been several recent attempts to dismiss the modal account of the progressive. In this article, I will argue that these attempts fail. Carefully examining a sophisticated formulation of the modal account will show why we should reject the current arguments against it. First, I will lay out the current field of views concerning the progressive. The chief difference in these views arises from a disagreement on whether the progressive itself should be assigned a modal semantics. After laying out these views, I will argue that the Predicate Modifier Account, one of the extensionalist views concerning the progressive, is not evidentially supported and carries unacceptable metaphysical commitments. I will then defend a contemporary modal account of the progressive from recent challenges against the progressive's modal entailment, as well as show that there is no problem from combinations with unless clauses. After arguing that the modal account ought to be preferred to its intensional rival the Adverbial Account on theoretical grounds, I will finish by explaining the significance for our understanding of the relationship between modality and natural language.2

2. A Field of Views

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. A Field of Views
  5. 3. Against the Predicate Modifier Approach
  6. 4. In Defense of the Modal Account of the Progressive
  7. 5. Concluding Remarks
  8. References

The field of views concerning the semantics of the progressive can first be divided into two camps: those that treat the progressive as a modal operator and those that treat the progressive extensionally. The first camp is primarily comprised of views that treat the progressive as a sentential modal operator.3 Among the views that treat the progressive extensionally, there is a further subdivision: those that introduce intensionality in the verb phrase that is scoped by the progressive and those that give an extensionalist treatment for both the progressive and the underlying verb phrase. In what follows, I will outline the main semantic proposals for each of these competing accounts.

2.1 The Modal Account of the Progressive

2.1.1 Kratzer's Framework for Modals

The modal account of the progressive treats the progressive as one of a number of modal operators appearing in natural language. Angelika Kratzer (Kratzer, 1977, 1981, 1991) and others have developed a sophisticated framework that helps explain how modals work and how they are related to each other. Before applying this framework to the progressive, I will first lay out the essentials for understanding the semantics of natural language modal operators.4

To begin, consider the sentence:

  • 1.
    In view of what the law provides, Mary must not steal Bill's car.

(1) is interpreted with respect to a background set of propositions L(w) indicated by the phrase in view of what the law provides. The propositions in L(w) express the content of the laws in w. Each proposition can be thought of as the set of worlds in which the law it expresses is fully satisfied—e.g. the law against theft is represented by the set of worlds in which no theft occurs. These worlds are ideal from the point of view of the law since the law is fully satisfied at those worlds. We can represent the set of worlds in which all the laws of w are followed perfectly as ∩L(w). Assuming the laws in w are consistent, then ∩L(w) will be a non-empty set. (1) is true in w because in every world in which the laws in w are fully satisfied—i.e. all the worlds in ∩L(w)—Mary does not steal Bill's car. Following Paul Portner, call this the classical possible-worlds semantics for modality.

This story does not work for the following example. Suppose Mary has in fact stolen Bill's car. Then the following sentence seems to be true:

  • 2.
    In view of what the law provides, Mary must go to jail.

(2), however, cannot be analyzed in terms of ∩L(w). ∩L(w) gives all the worlds in which the laws in w are perfectly satisfied. In other words, no one goes to jail in any of the worlds in ∩L(w) since no laws are broken in these worlds. So it is not the case that Mary goes to jail in all the worlds in ∩L(w) and so (2) is false, contrary to intuition.

The problem seems to be that a fact at w that is relevant for the interpretation of (2) is incompatible with one or more of the propositions in L(w). To reconcile this, let M(w) be the set of facts in w relevant for the interpretation of (2). In this particular case, one element of M(w) will be the fact that Mary stole Bill's car. The semantics for (2) should care about worlds in which the facts in M(w) hold. Let ∩M(w) represent this set of worlds.

As stated above, L(w) is the set of propositions expressing the laws in w. The worlds that are ideal from the perspective of L(w) are those in which all the laws are satisfied. Among the non-ideal worlds, some are better than others. For instance, assume that there is a law against stealing cars, a law against stealing credit cards, and a law against stealing passports in L(w). Among the non-ideal worlds from the perspective of L(w), those in which one of these laws is broken are better than worlds in which that law and one or two of the others are broken. The set of laws can be seen as establishing an order on any set of worlds. Formally put, let w′ < L,w w′′ if, according to the law in w, w′ is a better world than w′′.5

(2) seems to be true in w because in the best worlds from the perspective of L(w) compatible with Mary stealing Bill's car, she goes to jail.6 We can now represent this. Let Best(M,L,w) be the set of worlds w′ in ∩M(w) such that there is no w′′ in ∩M(w) where w′′ < L,w w′. ‘In view of what the law provides, Mary must go to jail’ is true just in case for every world w′ in Best(M,L,w), Mary goes to jail in w′.

In this extension of the classical possible-worlds semantics for modality, M(w) is called the modal base (since it provides a baseline of facts relevant for the interpretation of the intended modal claim) and L(w) is called the ordering source (since it provides an ordering on the set of worlds in ∩M(w)). These are contextually sensitive parameters that depend on the context of utterance for determining the content of a modal claim.7,8 Different modals in natural language vary according to their relevant modal base and ordering source. The framework presented here is powerful enough to capture this variability while making explicit the underlying meaning unifying natural language modals. To extend this framework to the progressive in natural language, what is needed is the appropriate modal base and ordering source.

2.1.2 Extending Kratzer's Modal Framework to the Progressive

A modal theory of the progressive along these lines requires an appropriate modal base and ordering source. Consider the sentence ‘John is building a house’. The facts that are relevant for the truth of this sentence pertain to John's house building abilities, the availability of materials, and features of the land on which John is building.9 Since these facts pertain to the circumstances in which John finds himself building a house, let the modal base be a circumstantial one (not unlike the modal base used in interpreting the sentence ‘Mary must go to jail’). M(w) is the set of facts in w relevant for the interpretation of ‘John is building a house’. ∩M(w) represents the set of worlds in which the facts in M(w) hold.

A sentence such as ‘John is building a house’ seems to be true because if he is not interrupted, he will build a house.10 The ordering source for the progressive will focus on those worlds in which John is not interrupted. Let O(w) be the set of outside factors that need to go right for John, if the proposition John builds a house is to be true. These outside factors include, but are not limited to, John not getting struck by lightning, not losing all his building funds, and not accidentally cutting off his limbs or breaking his legs while working.11

From the perspective of O(w), the ideal worlds are those in which none of these outside factors occur and John completes his house. But among the non-ideal worlds, some are better than others. John's losing a finger with no other outside factors occurring is better from the perspective of O(w) than John's losing a finger and breaking his legs. The set of outside factors O(w) can be seen as establishing an order on a set of worlds. Formally put, let w′ < O,w w′′ if, according to the set of outside factors O(w) that would interrupt John's building if they occurred, w′ is a better world than w′′.

‘John is building a house’ seems to be true because in the best worlds from the perspective of O(w) compatible with John's abilities and the circumstances in which his building occurs, John eventually builds a house. We can now represent this. Let Best(M,O,w) be the set of worlds w′ in ∩M(w) such that there is no w′′ in ∩M(w) where w′′ < O,w w′. ‘John is building a house’ is true just in case for every world w′ in Best(M,O,w), John eventually builds a house in w′.

2.2 Extensionalist Accounts of the Progressive

Both extensionalist accounts assume that the progressive is a predicate that takes events as arguments. A chief difference between them is in how they handle a central problem that arises as a result of this common extensionalist claim. In the following sections, I present the event semantic framework for the extensionalist views, a problem for this framework, and two distinct ways to resolve this problem called the Adverbial Approach and the Predicate Modifier Approach.

2.2.1 Event Semantics Background to Extensionalist Accounts of the Progressive

Both extensionalist accounts of the progressive assume a Neo-Davidsonian event semantics framework. It is useful to see how the progressive is embedded within this framework and the initial problem it licenses in order to understand the motivations for the competing extensionalist views.

Since Donald Davidson's seminal work on events and the logic of adverb dropping inferences, it is common to treat verbs and adverbs along the lines of more familiar predicates such as ‘is red’.12 Where ‘is red’ is standardly treated as saying something about an object, Davidson's proposal is to treat verbs and adverbs as saying something about an event. Furthermore, instead of treating a verb phrase such as ‘stab in the back’ as a single unstructured predicate of events, Davidson introduces structure: ‘stab’ and ‘in the back’ are treated as separate predicates of events, with a sentence such as ‘Brutus stabbed Caesar in the back’ amounting to there being an event that is a stabbing and that is in the back.

Extending Davidson's proposal, Parsons (1990) treats the progressive as a predicate of events. More precisely, he treats the progressive as a relation between an event and a time, saying that the event holds at some time. On this proposal, a sentence such as ‘Brutus was stabbing Caesar in the back’ will amount to there being an event that is a stabbing, that is performed by Brutus, that is directed at Caesar, that is in the back, and that holds at some time.13

Parsons also extends Davidson's account to include perfective aspect. More precisely, he treats perfective aspect as a relation between an event and a time, saying that the event culminates at some time. On this proposal, a sentence such as ‘Brutus stabbed Caesar in the back’ will amount to there being an event that is a stabbing, that is performed by Brutus, that is directed at Caesar, that is in the back, and that culminated at some time.

Tying these intuitive proposals together into a formal proposal, we get the following formulas for the example sentences ‘Agatha was crossing the street’ and ‘Agatha crossed the street’:

  • 3.
    1. Agatha is crossing the street.
    2. et(crossing(e) ∧ Agent(e,Agatha) ∧ Theme(e,the street) ∧ Hold(e,t))
  • 4.
    1. Agatha crossed the street.
    2. et(crossing(e) ∧ Agent(e,Agatha) ∧ Theme(e,the street) ∧ Cul(e,t))

As I explained above, the predicate ‘Hold’ is provided by the progressive and the predicate ‘Cul’ is provided by the perfective.14 Additional predicates such as ‘Agent’ and ‘Theme’ are referred to as thematic roles and are meant to assign relevant participants to events, such as the person doing the crossing and the thing that is crossed.15

Parsons' proposal quickly generates an unwanted entailment. Consider our recurring example of John and his incomplete attempt to build a house. Since the sentence that is relevant is ‘John is building a house’ we will have at least the following in its formal representation: (i) a predicate ‘building’ representing the content of the verb; (ii) a relation ‘Agent’ between the event denoted by the verb and the agent of the event, in this case John; (iii) a relation ‘Theme’ between the event denoted by the verb and the denotation of ‘a house’, which is given in the formal representation by an existentially quantified formula; and (iv) a relation ‘Holds’ between the event denoted by the verb and a time, which represents the progressive. The result is as follows:

  • 5.
    1. John is building a house.
    2. et(building(e) ∧ Agent(e,John) ∧ ∃x(house(x) ∧ Theme(e,x)) ∧ Hold(e,t))

Now (5b), which is the translation of (5a) according to Parsons' semantic theory, entails ∃x(house(x)). This is the entailment we do not want to license. So (5b) must be revised.

The problem with (5b) as a representation of (5a) arises from the introduction of the predicate ‘Theme’, the representation of ‘a house’ with an existentially quantified formula, and the entailment they license. This motivates the extensionalists to look for a solution in revisions to these parts of the representation. The problematic entailment seems to result from ‘Theme’ relating the event of John's building to a house that is the denotation of ‘house’ and that is being affected over the course of John's building. Both extensionalist accounts focus on this feature in their revision to (5b).

2.2.2 Extensionalist Account of the Progressive I: Adverbial Approach

In its general form, the adverbial approach revises (5b) by introducing a new thematic relation ‘Theme*’ that relates events to properties.16,17 The property in question is denoted by the indefinite noun phrase ‘a house’ in ‘John is building a house’. Letting ‘P’ name the property denoted by ‘a house’, we get the following revision to (5b) according to the adverbial approach:

  • 6.
    et(building(e) ∧ Agent(e,John) ∧ Theme*(e,P)) ∧ Hold(e,t))

This account avoids the counterexample to (5b) since the revised thematic relation relates John's building to P (a property) and not to any particular house.18

2.2.3 Extensionalist Account of the Progressive II: Predicate Modifier Approach

Building on Parsons' event semantics, the predicate modifier approach introduces a predicate modifier ‘IP’ that takes predicates such as ‘house’ and returns a complex predicate ‘IP(house)’. An object satisfies the formula ‘IP(house)(x)’ just in case it is a house-in-progress. As a further constraint on ‘IP’, the predicate it modifies (e.g., ‘house’ in ‘IP(house)’) cannot be exported. In other words, it cannot be used to describe the object it applies to.19

‘IP’ is introduced as part of the meaning of indefinite noun phrases when in the object position of progressive sentences with verbs of creation. When ‘a house’ occurs in ‘John is building a house’ it will be assigned ‘IP’ in the following way:

  • 7.
    et(building(e) ∧ Agent(e,John) ∧ ∃x(IP(house)(x) ∧ Theme(e,x)) ∧ Hold(e,t))20

Given the constraints on ‘IP’, (7) entails that John is building some thing while blocking the entailment to the existence of a house that he is building.

3. Against the Predicate Modifier Approach

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. A Field of Views
  5. 3. Against the Predicate Modifier Approach
  6. 4. In Defense of the Modal Account of the Progressive
  7. 5. Concluding Remarks
  8. References

I have two criticisms of the predicate modifier approach. First, the evidence in favor of it is questionable at best. Second, the predicate modifier view carries unintuitive metaphysical commitments. These metaphysical commitments are not acceptable given the lack of evidence for the approach. Hence, an alternative semantics for the progressive, one that is supported by evidence and that does not carry questionable metaphysical commitments, should be accepted instead. As I argue in the next section, this alternative semantics is the modal account of the progressive.

Although the predicate modifier approach does not entail the existence of a house, it does entail the existence of some object. But does this entailment comport with the intuitive evidence concerning progressive sentences with verbs of creation? Furthermore, does the intuitive evidence favor an extensionalist view as opposed to an intensionalist view like the modal account?

Consider Ponce de León's search for the non-existent fountain of youth. As Szabó correctly points out, it is correct to say that if Ponce de León was searching for the fountain of youth, then he was searching for something. The reason Szabó gives is that single world quantifiers in English (such as ‘something’, ‘somewhere’, ‘somehow’, but not ‘someone’) are insensitive to intensional contexts. So the entailed claim that he is searching for something is not taken as evidence against the verb ‘search’ introducing an intensional context.21 Similarly, when John is building a house, he is building something. So the entailed claim that he is building something should not be taken as evidence against ‘is building’ (a verb of creation with progressive aspect) introducing an intensional context.

What Szabó does claim is that although Ponce de León was searching for something he was not searching for some thing. The latter purportedly does not follow from ‘Ponce de León was searching for the fountain of youth’. On the contrary, according to Szabó, not only is John building something, he is also building some thing, and this latter entailment is evidence in favor of the predicate modifier approach.

My concern here is that Szabó's evidence does not hold up. If Ponce de León was searching for the fountain of youth, then he was searching for something. Whether we stress the entire word ‘something’, or we stress ‘some’ or ‘thing’ individually, I do not get a change in reading. He was searching for something/some-thing/some-thing. I am not sure our intuitions capture a difference in this case. Similarly for the sentence ‘John is building a house’. John is building something/some-thing/some-thing. The intuitive evidence appears to be the same in both cases, but Szabó clearly needs an evidential difference here to justify the existential entailment carried by the predicate modifier approach.

Szabó appeals to two other kinds of cases to support this entailment. First, there seems to be a distinct contrast in the intelligibility of follow-up questions about location between progressive sentences with verbs of creation, on the one hand, and sentences with typical intensional transitive verbs such as ‘seek a unicorn’, on the other.22 For example, consider the following pairs:

  • 8.
    1. —I am seeking a house.
    2. ? —Oh yeah? Where is it?
  • 9.
    1. —I am building a house.
    2. —Oh yeah? Where is it?

Assuming progressive sentences with verbs of creation block entailments to the existence of some object, then we should expect indefinite expressions to behave similarly in them, as is seen with indefinite expressions in typical intensional transitive verbs.23 However, whereas the follow-up question seems decidedly odd in (8), it is not at all odd in (9).24 The problem seems to arise from the misuse of ‘it’ in a follow-up question to a sentence with an intensional transitive verb. Since one can be seeking a house but no particular one, it does not make sense to use ‘it’ to refer to the house one is seeking. The lack of this problem in (9) suggests that ‘I am building a house’ provides an object that ‘it’ can refer to in follow-up questions.

Second, there seems to be a distinct contrast in the availability of demonstrative reference between progressive sentences with verbs ofcreation and sentences with typical intensional transitive verbs.25 As Szabó claims, if I successfully complete a general search for a house I began last October, I cannot point at it and say (10), but if I successfully complete the building of a house I began last October, I can point at it and say (11):

  • 10.
    ? This is what I was seeking since last October.
  • 11.
    This is what I was building since last October.

Assuming progressive sentences with verbs of creation block existential entailments, then we should expect indefinite noun phrases in those linguistic contexts to behave similar to indefinite expressions in typical intensional transitive verbs. However, whereas the use of a demonstrative seems decidedly odd in (10), it is not at all odd in (11). The problem seems to arise from the fact that a person can be seeking a house without any particular house being sought. In these cases, there is no house to play the referent of ‘this’ that is also the thing being sought for. The lack of oddity in (11) suggests that there is a thing to play the referent of ‘this’ that is the thing being built since last October.

I am still concerned that Szabó's evidence does not hold up. With regard to follow-up questions about location using ‘it’, the lesson does not apply across all contexts of use. Consider the following case. ‘Homestar’ is the name of a company that builds and sells prefab housing. Since they are a large company, they farm out their labor across the states: the walls, the appliances, the foundation, and the other parts are constructed and readied for shipment from different locations. John puts in an order with Homestar. He wants to live in one of their new neighborhoods that they are constructing. Since he is one of the first buyers and there are no other homes in the neighborhood, Homestar gives him the opportunity to decide where the house will go once all the pieces have arrived and are ready for assembling.

Now imagine someone talking with John a little after he has put in his order with Homestar and the process of construction is underway, but before he has decided where his home will eventually be located. The conversation goes as follows:

  • 12.
    1. —Homestar is building my house.
    2. ? —Oh yeah? Where is it?

In this situation, the follow-up question in (12) seems odd. The problem here seems to be that there is no clear way to answer the question since the major pieces of the house are in different locations. In typical cases, the building of a house happens on a particular site and the ‘it’ can be taken to refer to that specific location. But building a house on site is not essential to building a house. Once that is removed, follow-up questions such as Where is it? no longer seem acceptable. I take this to show that Szabó's examples in (8) and (9) do not provide adequate evidence for the existential entailment that the predicate modifier approach requires.26

With regard to Szabó's ‘this’ examples, I am worried about how much we can infer from this one case. Consider another example that pairs a demonstrative expression with the intensional transitive verb ‘look’. This verb displays at least one feature of intensional transitive verbs insofar as it gives rise to notional readings: I can be looking for a member of the philosophy department but no particular one. Imagine that John is engaged in a general pumpkin search for his Halloween pumpkin carving party. He has been looking for several days to no avail. He finally goes to his local pumpkin patch to look for one, without any particular one in mind. Once he picks one out, he remarks to a friend: this is what I have been looking for. In this context of use, I do not hear any oddity in John's response despite the use of a demonstrative combined with an intensional transitive verb as in (10). Szabó's evidence, therefore, does not decisively show that the acceptable follow up in (11) demands an existential entailment from progressive sentences with verbs of creation.

One response is that ‘this’ in my example is referring to a type of pumpkin and not to the pumpkin itself.27 If John is actually engaged in a general pumpkin search—i.e. a search for no pumpkin in particular—then at most he can say he had a kind of pumpkin in mind, the kind that he used ‘this’ to refer to, but not a particular instance of that kind. If we interpret ‘this’ to refer to a particular instance, then its use is inappropriate.

We can use ‘this’ (and ‘that’) to refer to kinds and properties. If I am baking a cake by following a picture and recipe of a cake I have hanging on my refrigerator, I can say to a friend, ‘I am baking that’. Surely, I am not baking the picture itself or baking something that will eventually become the picture itself. I am baking what the picture represents. So ‘that’ in this context of use cannot refer to an object. It either refers to a kind of object or a property. Although it is tempting to say that ‘that’ refers to an object indirectly by way of the picture of the cake, this will not do. Imagine the picture on my wall is a drawing of a cake that does not exist. I can still be baking that, even though the picture is of nothing real at all.

Another issue that arises is that sometimes the instance of a property is demonstratively pointed at, but the property itself is what is intended as the referent of ‘this’. Assume that you want to paint your room brown. Saying this to me while I am distracted I ask you to repeat yourself. You see a brown piece of cloth and point at it and say, ‘This is how I want to paint my room’. You clearly intend ‘this’ to refer to the color brown (a property) and not the instance of brown you are pointing at.28

The problem this raises for Szabó's second example is that without further testing, it is no longer clear that ‘this’ in ‘This is what I was building since last October’ refers to an object. It could refer to either a property or a kind of object. But Szabó clearly needs ‘this’ to refer to an object, a particular concrete object, so that he can provide evidence for the claim that ‘Mary was building a house’ entails the existence of some thing. Without further examples to help his case, appealing to ‘this’ does not help.

These arguments alone are not decisive. However, there is another concern regarding metaphysical excess that strengthens my case against the predicate modifier approach. First, it is important to note that progressive sentences that are about some process can be true despite that process still being in progress and even if the process remains incomplete. For instance, suppose John starts construction on his new house. He lays the foundation, gets the four main walls up and secured, and begins work on the roofing. But in the middle of working on the roof, he is struck down by lightning. In this situation, John was building a house even though he will not finish it.

The same holds true for situations in which more than one house is at issue. Suppose John plans on building five houses. He has completed two of them, is working on a third, and has purchased the materials for the fourth and fifth. The materials for the fourth and fifth are not sorted out yet. He has not cut the drywall into separate units or cut the wood into the appropriate lengths, but he has enough drywall, wood, and other materials for building two more houses. In this situation, it is true that John is building five houses (even if he does not finish).29

As another example, consider making cookies. Josh prepares the batter for roughly ten to fifteen cookies. The oven is on, the cookie sheet is out and prepped, and all that is left is dividing out the batter into individual cookies. Consider two possible ways of filling in these details. First, suppose Josh is planning on making ten cookies. In this situation it would be correct to say that Josh is making ten cookies. Second, suppose that Josh is planning on making fifteen cookies. In this situation, it would be correct to say that Josh is making fifteen cookies.

In all of these cases, the predicate modifier approach is committed to there being a particular number of objects (or objects-in-progress, but objects nonetheless). This is metaphysically questionable. When John is building five houses but only has two done, is in progress on the third, and has not divided the materials for the fourth and fifth, it is unintuitive to say that the undivided material counts as two objects. The predicate modifier approach is committed to this entailment nonetheless. Similar concerns apply to the cookie examples.

Given how we talk about creation in cases where more then one thing is being created, it is questionable to say, in general, that there are distinct numbers of objects-in-progress corresponding to the objects being built. There might just be an undifferentiated pile of stuff out of which the objects will be created. These metaphysical commitments would not look so bad assuming the evidence for the approach were compelling. But it is not. I say it is better to find a view that is evidentially supported and whose metaphysical commitments are acceptable.

4. In Defense of the Modal Account of the Progressive

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. A Field of Views
  5. 3. Against the Predicate Modifier Approach
  6. 4. In Defense of the Modal Account of the Progressive
  7. 5. Concluding Remarks
  8. References

In the following sections I will present what I take to be the strongest arguments against the modal account, such as purported counterexamples to the progressive's modal entailment as well as concerns about combinations with unless clauses, and show that they fail. I will then argue for the modal account by showing how it handles the problematic data for the predicate modifier approach and how it is preferable to the adverbial approach on theoretical grounds.

4.1 Against Criticisms of the Modal Account of the Progressive

4.1.1 Debunking Criticisms of Failed Modal Entailments

The most prominent line of criticism against modal accounts of the progressive targets the possibility entailment it licenses. According to the modal account, ‘John is building a house’ entails that it is possible John will build a house. More explicitly, it entails that in all the best worlds compatible with the facts relevant to the truth of ‘John is building a house’, John will eventually build a house. In any of these worlds, it will eventually be true that John builds a house. Therefore, it is possible for John to build a house. This entailment, licensed by the modal account, is mistaken according to some anti-modalists.

My general response to this criticism is that we must pay attention to the details of the case before we query intuitions about example sentences. The modal account that I presented above has two contextually sensitive parameters: the modal base and the ordering source. The modal base determines a baseline set of worlds against which the progressive is interpreted. This baseline is determined, in part, by taking the facts that are relevant to the truth of the progressive sentence in question. This set of facts will vary according to context and will not always be clear when first evaluating a particular progressive sentence. To the extent that the facts in the modal base are unclear, the truth of the progressive sentence in question will also be unclear. Similar considerations apply to the ordering source.

An example of this is the following. Consider the sentence:

  • 13.
    Sara is swimming across the Atlantic.

Asking whether (13) is true or false without giving details about Sara and her abilities, the state of the ocean, and how far into the swim she may be is completely mistaken. For instance, suppose the facts are that Sara is a superb swimmer, she has swum across each of the Great Lakes twice, she has trained years for her cross-Atlantic swim, and she is already two-thirds into her swim. In this situation, I judge (13) to be true. If the situation is such that Sara is just a regular swimmer, has not specially trained for this swim, and just started out somewhere on the coast of New Jersey, then I judge (13) to be false. Lastly, in this last situation, if Sara ends up completing her swim across the Atlantic, then it must have been true that she was swimming across the Atlantic all along. In this situation, the correct thing to say seems to be that we were wrong about the facts in the modal base. She must have had super-human capabilities all along!30 But judging (13) to be true or false without supplying sufficient details is just mistaken.31

With this said, consider the following sentences, which are presented in Wulf, 2009 as counterexamples to the progressive's modal entailment:

  • 14.
    Shannon was making a pumpkin pie, but someone had already used the last can of pumpkin.
  • 15.
    John was drawing a circle, but the pen immediately ran out of ink.32

Wulf claims that (14) and (15) can be asserted felicitously and that this poses a problem for the modal account of the progressive. Looking at (14), the first half of the sentence entails that it is possible for Shannon to make a pumpkin pie, whereas the second half seems to contradict that entailment by claiming that a key ingredient in the process (the pumpkin) is not available to Shannon. So the modal account should predict some infelicity in asserting (14), whereas there does not seem to be any according to Wulf. Similar reasoning applies to (15).

The problem with Wulf's argument is that he does not provide enough details against which to interpret (14) and (15). Once details are provided, there is no problem for the modal account. For example, the situation described by (14) alone is one in which Shannon was making a pumpkin pie and the last can of pumpkin has already been used. This leaves unspecified many details. Are there other cans of pumpkin nearby that Shannon could easily acquire? Or does she live in such a remote place (like a cabin deep in the Alaskan wilderness) where acquiring another can of pumpkin is not possible? Or is this is a dramatic situation in which the last can of pumpkin in the world is being used so it is really impossible to get any more pumpkin?

Consider a typical way of filling out the details. Shannon preps the materials for pumpkin pie, but unbeknownst to her the last can of pumpkin in her house was used the week before. However, her neighbor has plenty of pumpkin in her cupboard. In this context of use, I hear (14) as felicitous just as Wulf does. But this not a problem for the modal account since it does not predict that (14) is infelicitous. It is clearly possible for Shannon to complete making the pumpkin pie since she can easily walk over to her neighbor and acquire another can. Wulf clearly needs the modal account to predict infelicity for his argument to go through, which it does not in this case.

Now consider a fanciful way of filling out the details. Shannon is living in a post-apocalyptic world. As a way to remember the good old days of the twenty-first century, she sets out to make pumpkin pie by getting the pie crust materials and other baking tools she needs ready. But unbeknownst to her someone used the last can of pumpkin in the rationing warehouse. This was the last can of pumpkin in the entire world. In this context of use, I no longer hear (14) as felicitous. It now strikes me as incongruous that she was making pumpkin pie when the last can of pumpkin was used. But this is clearly in line with the modal account of the progressive. In this context of use, it is not possible for Shannon to complete making the pumpkin pie so it is not true that in all the best worlds Shannon makes pumpkin pie.

So whereas Wulf's argument requires that (14) is felicitous in the face of a contrary prediction by the modal account, it is clear that once the details are filled in the felicity of (14) varies accordingly with the predictions of the modal account. Wulf's argument fails because he does not pay close enough attention to the details. Simply presenting a sentence and saying it is felicitous or infelicitous is bad methodology.33

One other case is worth mentioning since it comes up repeatedly in both Szabó, 2004 and Szabó, 2008. This case concerns the sentences:

  • 16.
    Antoni is building the cathedral.
  • 17.
    Antoni can build the cathedral.

The setup given is that Antoni knows he will not be able to finish the cathedral but that others will take it up after his lifetime. In this context of use, according to Szabó, (16) is true while (17) is false, thereby providing a counterexample to the progressive's modal entailment.

First, there is the issue of details and whether (16) is true. Szabó does not provide enough contextually relevant details to determine the truth of this sentence. If Antoni is the only one working on the cathedral, then (16) no longer seems true. Perhaps it is true that Antoni intends to build a cathedral since many people intend to do things they cannot, but that he is building a cathedral is much different and not the case if he is working alone.

On the other hand, if Antoni is working with a team of people (a large team, since we are talking about a cathedral), then we might say that (16) is true. But in this case (16) is used to talk about Antoni as part of a team, not Antoni alone. This is not unlike what we might say today when, for instance, building a house. One person says they are building a house, when they really mean they are part of a team of people building a house, a team including carpenters, plumbers, and electricians. So we have two readings of (16), one of which is false (if we are only talking about Antoni building a cathedral) and the other of which is true (if we are talking about Antoni as part of a team of builders).

For Szabó's argument against the modal entailment to work, it must be that (16) is true and (17) false on any one of these readings. But this is not the case. For (16) to be true on a reading where we are only talking about Antoni building a cathedral, Antoni would need to be an inhumanly strong and fast worker. But in this case, (17) would be true as well. Indeed, I cannot think of a case where (16) would be true and (17) false where we only talk about Antoni himself building the cathedral. Similar remarks apply for the alternative reading where we are talking about Antoni and a team of builders. On this reading, whenever (16) is true, (17) must also be true.

Szabó might respond as follows. Agreed, once we specify the details of the examples, (16) and (17) do not provide counterexamples to the modal entailment licensed by the modal account of the progressive. But this comes at the cost of undermining the original motivation for that account, which is captured by the slogan nothing is happening unless it can eventually happen.34 Once we appeal to details of the context of use and alternative ways of interpreting (16) and (17), we have moved from intuitive data to theoretical claims and, in the process, left behind the original motivation for the modal account.

This is not a problem we ought to be concerned about when engaged in the kind of empirically informed philosophy of language that motivates the kinds of analyses at issue in this article. Intuitive claims such as nothing is happening unless it can eventually happen are important when first approaching a semantic analysis of some linguistic expression. For the progressive, there are numerous cases of true sentences that provide evidence for this claim. Just provide the relevant details for making the following sentences true: ‘Mary is crossing the street’, ‘John is baking a cake’, ‘Buzz is flying to the moon’, and many more. But there will be cases that seem to falsify the intuitive claim as well. At this point, we are pitting intuitions about general claims against intuitions about specific cases. And which intuitions we accept is partly a function of which account captures the most intuitions in both an empirically and theoretically satisfactory way. This applies equally to the modal account as well as Szabó's predicate modifier approach. I am concerned with responding to purported counterexamples to the modal entailment of the progressive because we learn about the details of progressive sentences, the need for specifying contextually relevant details when evaluating the truth of any given sentence, and ways of interpreting sentences that might otherwise go unnoticed. In doing so, I take myself to be strengthening the case for the modal account, not undermining it by refining our understanding of why we ought to accept it in the first place.

So it seems that Szabó is both trading on a lack of information and on alternative readings of (16) and (17) when using these sentences as counterexamples to the modal account of the progressive. Once details are supplied and the two readings are carefully pulled apart, Szabó's argument no longer poses a serious challenge. Far from moving away from the intuitive motivation for the modal account, we arrive at a better understanding of the complexities that go into interpreting progressive sentences.

4.1.2 Debunking Recent Criticisms from Unless-clauses

In Wulf, 2009, a problem is posed for modal accounts of the progressive from unless-clauses. It is widely accepted that progressive sentences can be true despite the fact that the process being described is interrupted. For example, ‘John is building a house’ can be true even if John is struck by lightning while building and never finishes. But consider the sentence:

  • 18.
    Shannon is making a pumpkin pie, unless her neighbor pulls her away (again) to play bridge.

According to Wulf, who is following an analysis of unless given by Declerck and Reed, 2000, the core meaning of (18) is something along the lines of Shannon is making a pumpkin pie in a case other than one in which her neighbor pulls her away (again) to play bridge and Shannon is making a pumpkin pie except if her neighbor pulls her away (again) to play bridge. If it turns out that the neighbor did pull Shannon away, then it is false that Shannon is making a pumpkin pie.35 But these kinds of exceptions are exactly the situations modal accounts are designed to handle. So there is a problem lurking here for the modal account.36

Wulf's problem crucially rests on the assumption that if the neighbor comes to pull Shannon away, then the progressive Shannon is making a pumpkin pie is false because of the meaning of unless in (18). But this surely cannot be right. Unless sentences of the form P unless Q do carry the implicature in which case not P, but it is cancelable. For example: John is eating with his mother unless his sister shows up and they all eat together. Sometimes the content of P can force the implicature. For example: The full moon will be visible unless the clouds move in. Here the content of P (The full moon will be visible) makes it such that P and Q cannot be true together. So it does not immediately follow from the neighbor coming and pulling Shannon away (the truth of Q) that Shannon is not making pumpkin pie (the falsity of P).

What I suggest is that in these progressive cases, the implicature is canceled. Shannon is making a pumpkin pie unless her neighbor pulls her away. In this case, she was making a pumpkin pie, although she never finished because she went to play bridge. What Wulf needs to do is argue that the content of Shannon is making a pumpkin pie forces the implicature reading, in which case there might be a problem for the modal account of the progressive. But since progressives are acknowledged as the kinds of sentences that can be true despite the processes they describe being interrupted, it seems like he will have a hard time making his case. In light of this, I do not see a compelling argument against the modal account of the progressive from unless-clauses.

4.2 In Favor of the Modal Account of the Progressive

4.2.1 Accounting for Creation of Multiple Objects

Since it is unclear whether the predicate modifier approach can handle the objections I raised earlier, the modal account seems preferable to it when it comes to the semantics of the progressive. But it might be wondered whether the objections also arise for this account. I will show that they do not.

One problem for the predicate modifier approach is that a sentence such as ‘John is building five houses’ is committed to the existence of five houses-in-progress even in cases where five things-in-progress clearly do not exist. This results from the fact that the approach is fully extensional. There is nothing to block the existential entailments even if there is a special modifier that blocks how the things quantified over can be described. This is clearly a structural problem that cuts to the center of the whole predicate modifier approach to the progressive.

The modal account does not suffer from this problem. Imagine the case again. John has built two houses, is working on the third, and has purchased the materials for the fourth and fifth but not differentiated them yet into their respective piles. It seems that the sentence ‘John is building five houses’ is true in this case. Assume it is, for argument's sake. The modal account says that in all the best worlds (given the facts relevant to the interpretation of this sentence and in view of no interruptions occurring), John eventually builds five houses. Formally, this is accomplished by evaluating the proposition John builds five houses in all those worlds. The intensionality at work washes away all existential commitment to houses in the world of evaluation. So the modal account can handle the sentence ‘John is building five houses’ without licensing unpalatable entailments.

4.2.2 Preferring the Modal Approach on Theoretical Grounds

The modal approach and the adverbial approach provide distinct semantic pictures of the progressive. The modal approach treats the progressive as a modal operator whereas the adverbial approach treats it extensionally as a predicate of events. This semantic difference with respect to the progressive leads to an additional difference with respect to the interpretation of the underlying verb phrase. Where the modal approach relies on whatever account is available for the underlying verb phrase, the adverbial approach provides a unique intensional treatment. This unique treatment is necessary to avoid entailments to the existence of objects before they are created.

Although these two approaches provide different overall semantic pictures of the progressive and underlying verb phrase, they share a common intensional outlook. So it might be wondered: why favor the modal approach over the adverbial approach? If both invoke intensional operators, why favor placing it with the progressive as opposed to the underlying verb phrase?

The modal account of the progressive is preferable, but for reasons that are theoretical as opposed to data driven. To see this, consider the following pair of sentences:

  • 19.
    Mary was making a cake.
  • 20.
    Mary made a cake.

(19) does not entail (20). Additionally, (19) does not entail the existence of a cake, since Mary could have been making a cake but have been cut off during the early stages of the process before there is anything remotely like a cake in front of her. Both of these entailments hold in the reverse direction, however. (20) entails (19). It also entails the existence of a cake, since it is outright contradictory to suppose Mary made a cake but that there was no cake that she made. The change in aspectual profile from (19) to (20) (i.e. a change from the imperfect to the perfect) is reflected by a difference in entailments.

Since the modal approach treats the progressive as a modal operator, there is no special problem for it in accounting for the failed entailments from (19) to the existence of a house and from (19) to (20). Furthermore, since the modal approach leaves the underlying verb phrase to receive whatever interpretation it would otherwise receive and since it makes no claims on a semantics for the perfect, there is no conflict with accounting for the entailments from (20) to the existence of a house or from (20) to (19). The modal approach does not make predictions here, but it is not incumbent on a semantic theory of the progressive to also account for the perfect as well.

The adverbial approach, on the other hand, does not fair so well. Here are the logical forms that the adverbial approach assigns to (19) and (20) respectively, letting ‘R’ name the property denoted by ‘a cake’:

  • 21.
    et(making(e) ∧ Agent(e,Mary) ∧ Theme*(e,R)) ∧ Hold(e,t))
  • 22.
    et(making(e) ∧ Agent(e,Mary) ∧ Theme*(e,R)) ∧ Cul(e,t))37

The adverbial approach interprets the indefinite in the underlying verb phrase using a revised intensional thematic relation. The reason for revising the thematic relation is to avoid entailing the existence of a cake when progressive aspect is present. But at least on this version of the approach the progressive is not responsible for the intensional thematic relation since it is given an extensional interpretation. So the adverbial approach predicts the same intensional thematic relation in (22) as well as (21).38 This means that the adverbial approach will not be able to account for the entailed existence of a cake that (20) licenses.

There is a solution due to Zucchi, 1999, which he says originates with Montague, 1974, by way of meaning postulates. Let Theme** abbreviate λeλx[Theme*(e,^λX.X(x))]. To recapture the entailment to the existence of a house licensed by (20) but not by (19), the following principle is needed: ∀etxQ[[making(e) ∧ Agent(e,x) ∧ Cul(e,t)] [RIGHTWARDS ARROW] [Theme*(e,Q) [LEFT RIGHT ARROW] Qλy(Theme**(e,y))]]. Loosely put, when there is an event that culminates, then a property Q characterizes that event just in case it is instantiated. Returning to (20) and its logical representation (22), when it is true that Mary made a cake then it will follow that the property R is instantiated. This captures the entailment to the existence of a cake that posed a threat to the adverbial approach. Furthermore, since there are numerous verb phrases that will pose a similar threat to the adverbial approach, these meaning postulates will need to be revised and added as necessary.

From the point of view of capturing the failed entailments from (19) to (20), the modal and the adverbial approach are on equal footing. But the adverbial approach makes inaccurate predications about (20) and related examples of sentences with perfect aspect that require a host of meaning postulates to correct. Once corrected, the modal and adverbial accounts are evidentially on a par once again. However, acceptance of a semantic theory is not just a matter of capturing the data. There are broader theoretical concerns to consider such as fruitfulness, integration into previous theory, elegance, and so forth. When these concerns are taken into account, it is clear that the modal approach should be accepted. The modal approach integrates well into existing semantic work on natural language modals. It provides a fruitful way of extending that framework, while remaining elegant in its formulation and economical in its additions of new semantic primitives. The adverbial approach, as is now clear, is not elegant or economical. The sheer abundance of meaning postulates necessary to capture the unacceptable entailments it predicts evidences this. In light of this excess, it is reasonable to suggest that this is not a fruitful approach to extending the extensionalist account of the progressive. What the adverbial approach seems to provide is a disjointed and clunky theory to account for data that is either easily integrated into the modal approach or need not be accounted for by a theory of progressive aspect.39

5. Concluding Remarks

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. A Field of Views
  5. 3. Against the Predicate Modifier Approach
  6. 4. In Defense of the Modal Account of the Progressive
  7. 5. Concluding Remarks
  8. References

Recent arguments against the modal account of the progressive should be rejected. Progressive sentences are contextually sensitive, but some of the arguments against the modal account fail to take this into consideration when testing intuitions about key examples. For instance, Wulf's purported counterexamples to the progressive's modal entailment are none other than sentences tested without sufficient details provided. Once details are supplied, intuitions fall in line with the modal account of the progressive. Turning to Wulf's argument from unless clauses, I argue that it is best explained as a case of canceled implicature. So it seems the modal account of the progressive is not undermined by these recent counterarguments. Given the evidential and metaphysical concerns regarding the predicate modifier approach, on the one hand, and the theoretical concerns regarding the adverbial approach, on the other, I argue that the modal account ought to be accepted.

This is significant for at least three reasons. First, it shows that the progressive is another, perhaps unexpected, way of expressing modal information in natural language. Given philosophical interest in modality, language, and meaning, the progressive and its semantic and logical properties provide yet another area to investigate these important philosophical notions.

Second, the success of the modal account of the progressive provides yet another successful application of Kratzer's semantic framework for modeling natural language modal expressions. Adding the progressive as a means for expressing modal information might seem like a cost since it is not typically considered alongside the usual modal suspects such as ‘possibly’ and ‘necessarily’. Using Kratzer's framework to account for the meaning of the progressive provides evidence that there is a deep semantic connection between this incongruous set of expressions.

Lastly, the modal account of the progressive is favorable for theoretical reasons. Although it is clear that modality will play a role in the account of the progressive and its complement verb phrase, whether we treat the progressive or its underlying verb phrase as the source is decided on theoretical rather than evidential grounds. This is significant for how we understand the meaning of natural language expressions. What would seem to be a matter for empirical study (i.e. what is the meaning of some natural language expression?) sometimes turns on theoretical virtues, which have a rich history in the philosophy of science. So although the kind of philosophy of language project engaged in this article may seem like a matter for empirical linguistics, it is clear that philosophical considerations about theories and explanatory virtues play a significant role.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. A Field of Views
  5. 3. Against the Predicate Modifier Approach
  6. 4. In Defense of the Modal Account of the Progressive
  7. 5. Concluding Remarks
  8. References
  • Asher, N. 1992: A default, truth conditional semantics for the progressive. Linguistics and Philosophy, 15(5), 463508.
  • Bonomi, A. 1997: The progressive and the structure of events. Journal of Semantics, 14, 173205.
  • Davidson, D. 2001a: Essays on Actions and Events, 2nd edn. Oxford: Oxford University Press.
  • Davidson, D. 2001b: The logical form of action sentences. In D. Davidson, Essays on Actions and Events, 2nd edn. Oxford: Oxford University Press.
  • Declerck, R. and Reed, S. 2000: The semantics and pragmatics of unless. English Language and Linguistics, 4(2), 20541.
  • Dowty, D. 1979: Word Meaning and Montague Grammar. Dordrecht: D. Reidel.
  • Dowty, D. 1991: Thematic proto-roles and argument selection. Language, 67, 547619.
  • Forbes, G. 2006: Attitude Problems: An Essay on Linguistic Intensionality. Oxford: Oxford University Press.
  • Kamp, H. and Reyle, U. 1993: From Discourse to Logic: Introduction to Modeltheoretic Semantics of Natural Language, Formal Logic, and Discourse Representation Theory. Dordrecht: Kluwer Academic Publishers.
  • Kratzer, A. 1977: What ‘must’ and ‘can’ must and can mean. Linguistics and Philosophy, 1, 33755.
  • Kratzer, A. 1981: The notional category of modality. In H. J. Eikmeyer and H. Reiser (eds), Words, Worlds, and Contexts: New Approaches in Word Semantics. Berlin: Walter de Gruyter.
  • Kratzer, A. 1991: Modality. In A. von Stechow and D. Wunderlich (eds), Semantics: An International Handbook of Contemporary Research. Berlin: Walter de Gruyter.
  • Landman, F. 1992: The progressive. Natural Language Semantics, 1, 132.
  • Montague, R. 1974: Formal Philosophy: Selected Papers of Richard Montague, R. Thomason ed. New Haven CT: Yale University Press.
  • Parsons, T. 1990: Events in the Semantics of English: A Study in Subatomic Semantics. Cambridge MA: MIT Press.
  • Portner, P. 1998: The progressive in modal semantics. Language, 74(4), 76087.
  • Szabó, Z. G. 2004: On the progressive and the perfective. Noûs, 38, 2959.
  • Szabó, Z. G. 2008: Things in progress. Philosophical Perspectives, 22, 499525.
  • Wulf, D. 2009: Two new challenges for the modal account of the progressive. Natural Language Semantics, 17 (3), 20518.
  • Zucchi, S. 1999: Incomplete events, intensionality, and imperfective aspect. Natural Language Semantics, 7, 179215.
  1. 1

    This is a rough and ready version of David Dowty's account of the progressive (Dowty, 1979). It is not meant to capture the details of Dowty's account, nor is it meant to be an acceptable semantics for the progressive. For the latter, see the next section in which I detail a contemporary view of the modal account of the progressive due to Paul Portner (Portner, 1998).

  2. 2

    It is easy to find high-quality surveys of the history and development of approaches to the progressive. For instance, Wulf, 2009 covers the development of modal approaches up until Fred Landman's account, while Szabó, 2004 covers the problems that prompt semantic theorizing about the progressive as well as the fixes one can make in order to account for them. See also Asher, 1992 and Bonomi, 1997. In this article, I will avoid repetition and go straight into the contemporary accounts. The outcome of this, as I will show, is that the modal account of the progressive, understood in its contemporary form, ought to be accepted over its rivals.

  3. 3

    It should be added that a number of important modal accounts of the progressive provide an interpretation where the progressive is a verb phrase operator as opposed to a sentential operator. For instance, whereas Dowty, 1979 treats the progressive as a sentential modal operator, Landman, 1992 treats the progressive as an intensional relation between events and properties. I will focus on sentential modal accounts in this article given their prominence in the literature on the progressive. This will not affect the arguments I present here.

  4. 4

    The modal account I present here is in all essential details that of Portner (1998). The aim of this presentation is not to revise his account but to show that his account provides the tools necessary to handle the objections that are currently in fashion against the modal account.

  5. 5

    As Portner (1998) suggests, either <L,w can be defined in terms of L and w (following Kratzer's suggestion) or it might be taken as semantically primitive. For the purposes of this article, I will not take a stand on this. See Portner, 1998; Kratzer, 1977, 1981, 1991, for the relevant discussion. Also, this presentation of Kratzer's semantics for modals presupposes the limit assumption, but making this assumption is not necessary and does not affect the arguments that follow.

  6. 6

    It is important to add ‘best’ since it will not be that in all the worlds in which she steals Bill's car she goes to jail. In some of those worlds she does not go to jail because she escapes the police.

  7. 7

    From here forward, the ‘in view of’ prefix to a given modal sentence will be left off. Since the modal base and ordering source are treated as contextual parameters, it is possible to account for the truth of sentences such as ‘Mary must go to jail’ even if there is no explicit ‘in view of’ phrase.

  8. 8

    How contextual parameters in general are provided by the context of use is an important question in the philosophy of language that requires a treatment that goes beyond the scope of this article. For present purposes, I follow Portner (1998, p. 771) when he says, ‘The identities of the modal base and ordering source are the central theoretical primitives of this theory. As semanticists, we do not try to give an account of how they arise … We must take these parameters as simply being provided by the extra-linguistic context.’ This is a common sentiment among philosophers and linguists working in empirically informed philosophy of language.

  9. 9

    This is not meant to be exhaustive, but to give an intuitive gloss on what seems relevant for the interpretation of ‘Mary is building a house’.

  10. 10

    This is the primary and strongest motivation for the modal account of the progressive because the modal account provides the best explanation for this intuitive conditional. I would like to specially thank an anonymous referee for pointing out that this intuitive conditional on a Kratzer-style semantics for conditionals becomes a restricted necessity modal, which is the kind of modal used in the account of the progressive being defended here. This provides more evidence for the case I make at the end of this article that the modal account of the progressive ought to be preferred on theoretical grounds and not just evidentiary grounds. Regarding the intuitive conditional and its relation to progressive sentences, see Kamp and Reyle, 1993, p. 576.

  11. 11

    To maintain continuity with Kratzer's approach, M and O take worlds as arguments, but this will need further amendment. It is clear that what counts as relevant to the successful completion of an event and what counts as interruptions to it will in part depend on the event itself. This could be made explicit by letting M and O take events as arguments instead of worlds as Portner, 1998 does. For present purposes, I will leave this refinement aside since it does not affect my argument in any way.

  12. 12

    The classic reference for this is Davidson, 2001b. See also related essays in Davidson, 2001a.

  13. 13

    The time in question will be the result of the time of utterance and the tense profile of the sentence in question. I am suppressing these details for ease of presentation.

  14. 14

    These example sentences are adapted from Portner, 1998. They are inadequate as they are formulated. For example, they do not adequately translate and capture the internal structure of definite noun phrases and the tense profile is not accounted for. However, their inadequacies will not affect the discussion of this section.

  15. 15

    Thematic roles are not a necessary component of the event semantic framework. But given recent syntactic work on verbs and their arguments that investigate intuitive relations among sentences such as ‘Brutus stabbed Caesar with the knife’ and ‘Caesar was stabbed by Brutus with the knife’, thematic roles seem explanatorily desirable. For a brief introduction to the way thematic roles function in event semantics, see Parsons, 1990. For a more thorough discussion of thematic roles, see Dowty, 1991.

  16. 16

    Since the arguments I will present only depend on the use of properties by the property view, I am suppressing many logical considerations. But it should be noted that according to the property view the expression ‘a house’ denotes a second order property named by the expression ‘P’. The thematic relation ‘Theme*’ is a relation between an entity and a second order property that holds when the second order property is the thematic object of the event. For more on the logical details of the property view see Forbes, 2006 and Zucchi, 1999.

  17. 17

    Szabó, 2008 is responsible for calling the view outlined here the adverbial account.

  18. 18

    The main proponents of the adverbial approach are Forbes, 2006 and Zucchi, 1999. Forbes (2006) provides a further analysis of Theme* (which he introduces as Char) in terms of modality and other semantic primitives. For the purposes of my arguments, these additional details are not necessary.

  19. 19

    More formally put: ∃x(IP(house)(x) ∧ blue(x)) entails ∃x(IP(house)(x)) but ∃x(IP(house)(x) ∧ blue(x)) does not entail ∃x(house(x)).

  20. 20

    This is slightly different than Szabó's presentation. He does not represent tense and he leaves the indefinite expression in situ. Given the typical existential commitment of indefinites, my introduction of an existential quantifier does not significantly alter Szabó's presentation of his proposal.

  21. 21

    See Szabó, 2008, p. 500 for his account of this entailment with the verb ‘search’ and how he uses it to motivate the predicate modifier approach.

  22. 22

    These examples are from Szabó, 2008, p. 508.

  23. 23

    By ‘similar behavior’ I mean similar semantic behavior of the indefinite expression itself as well as similar semantic effects on surrounding expressions. I am assuming this is what Szabó has in mind when he presents his argument.

  24. 24

    It is important for this example as well as similar examples using intensional transitive verbs to focus on the non-specific, no particular one reading. On this reading, the follow-up question in (8) is decidedly odd.

  25. 25

    The presentation here follows Szabó, 2008, p. 508.

  26. 26

    Rich Thomason first gave me the idea of an example such as the one I just presented.

  27. 27

    Szabó suggested this response to me (p.c.).

  28. 28

    I came across an example like this one while reading an article on related issues. But I cannot remember who provided the example. So I note here that I owe someone credit without being able to supply their name.

  29. 29

    As I am imagining the situation, John has a single plan to build five houses. Intuitions about the truth or falsity of ‘John is building five houses’ become much more complicated if we assume John has five individual plans for building one house. Assuming it is possible to have a single plan to build five houses, there is no need to consider the more complicated case. Similar remarks apply to the cookie-example that follows. (Thank you to an anonymous reviewer for pointing this out.)

  30. 30

    My response is inspired by Portner, 1998, p. 775 where he gives a similar response to a related example.

  31. 31

    This is almost what Szabó, 2008, p. 502 does. Although he does not completely endorse that (13) is false, he does suggest that any view on which the progressive carries the modal entailment under discussion can explain the falsity of (13) by the falsity of ‘Sara can swim across the Atlantic’.

  32. 32

    For earlier attempts to discredit the modal account of the progressive in ways similar to Wulf, 2009, see Szabó, 2004 and 2008.

  33. 33

    Similar remarks apply to (15) and whether it poses any problems for the modal account.

  34. 34

    See Szabó, 2008, pp. 502–3 for this formulation of the motivating slogan for the modal account of the progressive and a version of this response.

  35. 35

    Wulf, 2009, p. 214 states this as follows: ‘Likewise for [(18)] … Shannon is making a pumpkin pie is true in a case other than if she is pulled away (though other obstacles could arise) … Shannon is making a pumpkin pie is true except if she is pulled away (though perhaps not except iff she is pulled away). Again, it crucially follows that if [(18)] is true and it is also true that Shannon is pulled away to play bridge, then Shannon is making a pumpkin pie is false.’

  36. 36

    Indeed, there is a purported problem here for all current accounts of the progressive, but I will leave this detail aside and only focus on the modal account since I do not think this is a serious problem to begin with.

  37. 37

    I am omitting the representation of tense in these logical forms. This omission does not affect the argumentative point being made here.

  38. 38

    This is especially the case for progressive sentences with verbs of creation (such as ‘make a cake’ and ‘build a house’), where the presence of Theme* is attributed to the verb of creation instead of the progressive. See Zucchi, 1999, p. 188 for details.

  39. 39

    This applies even if the adverbial approach is revised and ties the introduction of Theme* to the progressive. On this refinement, the intensionality would be due to the progressive but not by assigning it a modal semantics. Although this refined theory would be much closer to the modal approach, it would still be theoretically undesirable. This refined approach would be inelegant (why not just make the progressive intensional?) and out of step with previous work on modality and intensionality in natural language. In the absence of additional evidence that clearly shows the progressive not to be intensional, the modal approach is preferable to this refined version of the adverbial account.