A Defense of Canonical Contextualism About Epistemic Modals[1]
J.L. Dowell
Draft at 2/24/09
On the canonical account, modal expressions (like “might” and “must”) are quantifiers over possibilities. Modal expressions themselves are neutral; they make a single contribution to determining the propositions expressed across a wide range of sentence-uses. What modulates the modality of the proposition expressed—as bouleic, epistemic, deontic, etc.—is context.
This ain’t the canon for nothing. Its power lies in its ability to figure in a simple and highly unified explanation of a fairly wide range language use. Recently, the canon’s neat story has come under attack. Some philosophers and linguists have defended revisions to the canon by arguing that the original account is unable to provide an adequate explanation of cases in which the modal is epistemic and there is an at least prima facie puzzle about whose knowledge could be relevant for determining the possibilities that are the modal’s domain.[2]
I’ll argue that once we have the right canonical contextualist account on the table, we can see that the puzzle really is only prima facie. The account I’ll defend takes as its inspiration Kaplan’s sketched account of denotation determination for demonstratives.[3] The resulting contextualist account will then fit with a plausible, Kaplanian account of domain-determination for quantifiers over individuals. I’ll then show how that account handles the allegedly problematic cases at least as well as rival relativist[4] and revisionary contextualist[5] accounts. In addition, unlike its main, canonical contextualist rivals,[6] the position I’ll defend does without reliance on the difficult-to-spell out notion of “epistemic reach”. Widely accepted methodological considerations favor a canonical, contextualist account over its rivals. So, providing such an account that handles the problematic cases at least as well is sufficient reason to prefer the canonical view.
Once on the table, the availability of the present position seems almost obvious. After all, it’s the result of assembling familiar and widely accepted parts of theories of other context sensitive expressions and filtering them through a general Gricean approach to semantics and pragmatics. So why has it seemed to so many philosophers and linguists that the canon is in need of revision?
The answer, I think, lies in the aims and limitations of Kratzer’s original proposal. On some ways of dividing up philosophy of language’s tasks, how context determines the extension of a context-sensitive expression is part of metasemantics, not semantics. This is a common view in the literature on quantifier domain restriction.[7] Being a semantic theory, it is no surprise that Kratzer’s original proposal is silent on this question. In her examples, it’s always clear which group is relevant for restricting the domain over which a modal gets evaluated, but how context serves to determine the relevant group is not addressed.[8] This means that when we consider new cases, it can be unclear how to identify the relevant group. The present proposal isn’t a revision of the canon, but it is an extension of it. It accepts the basic features of Kratzer’s descriptive semantics and adds to it an account of how context determines the range of possibilities modals quantify over by appeal to familiar, Gricean mechanisms.
The plan of action is as follows: Before outlining the kind of view I favor, I’ll offer a sketch of Kratzer’s canonical account and a description of the case that I think provides it’s greatest challenge. I’ll then offer a diagnosis of the sources of the apparent difficulty for the canonical account and defend a general approach to questions about the semantics of modals, one that everyone should favor regardless of whether they find the particular position I’ll defend attractive. This will involve a brief overview of some of the current thinking about two other kinds of context-sensitive expressions, quantifiers over individuals and demonstratives.
I’ll then outline the proposal I favor and show how it can accommodate the flexibility and objectivity of epistemic modals. Next I’ll turn again to the core problem case and show how the account I favor accommodates the data at least as well as its revisionary rivals. Then I’ll discuss two other kinds of case that some see as motivating departures from the canon, showing how the present account is able to handle them as well. Finally, I’ll consider how the present proposal handles two problematic cases for relativists that also, I’ll argue, can’t be explained with revisionary contextualism’s current resources. Taken together, these considerations undermine the motivation for departures from the canon.
1. The Canon, the Puzzle and its Revisionary Solutions
1.1 The Canon
On the canonical view, modal expressions like “might” and “must” are quantifiers over sets of possibilities. Their basic form is
MODAL(B)( f)
where B determines the restriction on the domain of quantification, f, the prejacent, is a sentence that gets evaluated at the possibilities in B, and MODAL determines (roughly) which or how many of the possibilities in B are possibilities in which f must come out true for MODAL(B)(f) to come out true. My focus here is on “might” and “must” which require, respectively, that some or all of the possibilities in B are possibilities in which f comes out true.[9]
In some cases, the use of an explicit, restrictor phrase determines B. Here are some examples:
In view of the local traffic laws, visitors must stop at stop signs.
In view of Sally’s preferences, she might take the turnpike to get to Boston.
In view of what Holmes knows, the gardener must be the culprit.
In other cases, though, the restriction is supplied by the context. Following von Fintel and Gillies, I’ll call such statements ‘bare epistemic modals’ or ‘BEMs’.[10] On the canonical view, in both sorts of case, the modals contained in such statements are neutral—‘must’ and ‘might’, for example, have a single meaning across all contexts in which they appear.[11] This is one of the theory’s singular advantages; it provides a simple, unified explanation of a great variety of linguistic behavior, including a speaker’s ability to generate and understand new modal statements.
The locus of the dispute are BEMs. All parties to the dispute of concern here accept that the modals in such sentences quantify over a restricted range of possibilities, where the restriction is made to those compatible with some state of information or knowledge. The disputes arise over which state of knowledge is the relevant one and how that state is determined.
1.2 The Puzzle
The most compelling challenge to canonical contextualism rests on cases that aim to show that within a single context, there is no single resolution of the contextual parameter that fits with all of the data. The problematic data comes in the form of intuitive judgments about the truth-values of what’s expressed by the use of BEMs and their linguistic appropriateness. In von Fintel’s and Gillies’ example, Alex is helping her roommate Billy search for her keys.[12] Alex asserts
(C) “You might have left them in the car.”
Billy has two available responses:
(Y) “You’re right. Let me check.”
And
(N) “No; I still had them when we came into the house”.[13]
The difficulty, according to von Fintel and Gillies, is that there doesn’t seem to be a single interpretation of the contextual parameter that preserves our intuitions both about the truth-values of (C), (Y), and (N) and about a speaker’s warrant in asserting them. A group reading, where the relevant group contains just Alex and Billy, fits with our judgments regarding the truth-value of (C) and the appropriateness of (Y) or (N). To see this, suppose first that among the possibilities compatible with what Alex and Billy together know is at least one possibility in which the keys are in the car. The group reading predicts that in this case, (C) is true—and our intuition is that it is true in that case.[14] Moreover, we have the further intuition that Billy’s response in (Y) is appropriate, as the group reading predicts. Suppose, though, that Billy knows that the keys are not in the car. The group reading then predicts that (C) is false and (N) appropriate. This fits with our intuitions that in that case (N) is appropriate and (C) false.[15]
So, it looks like there is a canonical, contextualist treatment of the case, the group reading, that fits with our intuitions regarding the truth-value of (C) in various circumstances and the appropriateness of Billy’s available responses. The difficulty for the canon, according to its foes, is that it is hard to see how Alex could be warranted in asserting (C) on the group reading. As von Fintel and Gillies claim,
She does not seem to be within her linguistic rights to be claiming that the group’s information cannot rule out the prejacent. After all, Alex does not know whether Billy has private information about the whereabouts of the keys.[16]
According to the objection, that difficulty goes away under the solipsistic reading. Alex is fully warranted in asserting (C) so long as the keys’ being in the car is compatible with what she knows. But that reading no longer preserves our sense that in each of the above scenarios, Billy’s available responses are appropriate. (Just take the affirmative response. There’s nothing in the scenario that guarantees that Billy is in a position to take a stand on what’s compatible with what Alex knows. And given that she is looking for her keys, it’s not plausible that she is affirming the prejacent.) What’s a semanticist to do?[17]
1.3 Revisionary Solutions
I’ll call this basic case “KEYS” to have a handy way of referring back to it in what follows. The interest of KEYS lies in the use and assessment of BEMs that appear unstable with respect to solipsistic and group readings. Revisionary contextualists and relativists accept the line of reasoning that suggests that BEMs in these cases are unstable and conclude that the canon needs revision. Indeed, the acceptance of this line of reasoning provides one of the central motivations for both views.[18] Below I’ll argue that if we look more carefully at the details of the case, we’ll see that there is a single, canonical contextualist interpretation that captures both intuitions about appropriateness and about truth-values in each of a series of versions of the KEYS case. But first, a quick spin through the opposition.
Relativist conclusions from KEYS
Revisionary contextualist conclusions from KEYS
2. A Diagnosis and an Approach
2.1 Diagnosis
Below I hope to show that the problem KEYS poses for the canon is only apparent. Before outlining the view I favor, it may be useful to consider two sources of this appearance. A hint of one source emerges from an interesting remark of Egan, Hawthorne, and Weatherson in their paper defending a version of relativism.[24] They consider several different ways one might flesh out the basic contextualist view. To cover the range of cases that any contextualist account needs to cover, it looks like the contextualist will need to require that different contexts pick out different kinds of groups as relevant. In some cases, it looks like the context will need to select the knowledge of some group that includes the speaker to restrict the modal’s base; in others, it will need to require the speaker’s exclusion. They rightly complain that without a principled account of why certain contexts require a speaker’s inclusion and others don’t, the resulting view will be ad hoc. This suggests one source of the problem is the absence precisely of such a principled account. This is just what a metasemantic account of domain determination for modals would provide. Being a semantic theory, Kratzer’s original proposal does not include such an account. So, it is no surprise that it is unclear what principle determines how context conspires with use to generate the needed restrictions. (It should be noted here that neither revisionary contextualism nor relativism provides the needed metasemantic account either. Those views are in the same boat as Kratzer’s original proposal in this respect. Later we’ll see that this neglect saddles each of those views with puzzles of their own.)
A second source of the puzzle is the eliding of an important distinction between kinds of norms of assertion in its original statement.[25] In some cases, an assertion is bad because irrational —we assert what we do not have good grounds to believe. In other cases, asserting is done badly because it is a manifestation of our semantic incompetence—in asserting, we manifest our misunderstanding of the meanings of some of the terms we use. Which form of goodness must an assertion exhibit to provide datum for a descriptive semantic theory? Requiring an assertion to meet the epistemic standard here seems too strong. The aim of a descriptive theory is merely to assign contents to sentences in a context, not to identify the conditions under which beliefs with those contents would be rational. Here the relevant standard in asserting is weaker than rationality: A speaker need merely manifest her semantic competence with the terms she uses in order for her assertions to provide evidence for a theory about the semantic contents of those terms. To require something stronger is to confuse the aims of a descriptive semantic theory.
2.2 An Approach and its Inspiration
2.2.1 An Approach
It’s generally agreed that simpler, more unified, semantic theories are better than more complex, less unified ones. So, we should try to give a single account of bare modal expressions, since this would give the most simple and unified account of language-use. The only well-worked out account on offer is Kratzer’s canonical contextualist one, so this is the one to beat. The first part of the approach, then, dictates that, before revising the canon, we either exhaust our canonical, contextualist options for BEM expressions or provide a rival account of bare modals as such.
Second, all parties to the dispute I consider accept that modals are quantifiers over possibilities. Moreover, all accept that in the case of epistemic modals, the domain of quantification is in some way restricted. A semantic theory that accepts that much and treats modal expressions in a way analogous to its treatment of other quantifier expressions will be a more simple, unified theory than one that doesn’t. There is disagreement over the exact mechanisms of domain restriction in the case of quantification over individuals. However, there is this much agreement: however domains get restricted, they get restricted contextually.
Putting these together suggests that before accepting revisions to the canon, we should first see whether there is a version of the canonical account that avoids the allegedly counter-intuitive consequences by taking as our inspiration contextualist treatments of quantifiers over individuals. Here I’ll argue that, if we think about one plausible story about how quantification over individuals goes, we’ll find a template for a canonical story about bare modal expressions that avoids the allegedly problematic consequences for epistemic modals in particular.
2.2.2 Its Inspiration: Quantifiers and Demonstratives
On one plausible view, quantifiers are context-sensitive not in the way that pure indexicals are, but rather in the way Kaplan holds that demonstratives are.[26] We might say that the use of a demonstrative in an appropriate context is by itself insufficient to secure a denotation. In addition, the speaker needs to manifest an appropriate referential or directing intention. This feature of demonstrative-use is constant across contexts. But which feature of the context a speaker intends her addressee to rely upon in order for the latter to recognize the referent is something that varies with context. The speaker’s intentions with regard to what it takes to be the referent also may vary with context.
This Kaplanian story about the context-sensitivity of demonstratives can be applied to other context-sensitive expressions. Here’s an illustration of how that story might work applied to quantifier domain restriction. Letisha is a philosophy professor who is in her colleague Shevon’s office, talking about the division of her household’s goods in advance of her separation from her husband, Juan, a lawyer. She says to Shevon
(A) “All of the philosophy books are mine.”
It’s clear from conversational context that she intends to restrict the domain of “all” to the philosophy books in her house and to exclude the perceptually more salient philosophy books in Shevon’s office. Contrast this use with another. Imagine that Letisha is now in her home library and Juan walks in. Out of the blue she says, “All of the philosophy books are mine”. Here it’s clear that she intends perceptual salience to settle the domain restriction and for the domain of “all” to be restricted to the philosophy books in their library.
What is constant over these two contexts is the role played by the speaker’s referential intention. What varies in each case is the content of her intention and way she intends salient features of the context to manifest her intention to her addressee. In other words, not only may the domain for a quantified sentence shift from context to context, as the speaker’s referential intentions change, what makes for salience may shift with context, as the speaker’s referential intentions change. In one context, a speaker may rely on perceptual salience to make her intentions manifest, in another, it may be conversational salience. What must be constant is that the context must work to resolve which form of salience is the one the speaker is best understood as relying upon.
On this story, quantifiers exhibit a kind of flexibility in extension-determination that makes them context-sensitive in a way that makes them more like demonstratives than true indexicals.[27] In the case of a true indexical, an expression conspires with context (when the context is non-defective) to fully resolve extension. In the case of demonstratives, in contrast, there is something more for the speaker to do: the speaker must in some way make manifest how she intends the denotation to be determined by context.
I propose to let this story about the context-sensitivity of demonstratives and quantifiers provide a model for how a canonical contextualist might account for modals in general and the epistemic ones in particular. On the resulting view, modal quantifiers get their domains determined by requiring speakers to in some way resolve the kind of salience that is relevant for domain determination in a context. One proposal relies on speakers’ referential intentions to explain domain determination; no doubt there are at least as many rival proposals as are there are rival accounts of the context-sensitivity of demonstratives and of quantifiers.
The important point here is that any canonical contextualist account of modal expressions must be allowed to exhibit the kind of flexibility theorists find perfectly acceptable in the case of other context-sensitive expressions, such as demonstratives and quantifiers. If you don’t find the above treatment of quantifiers or demonstratives attractive, you should nonetheless accept the above recipe: take your favored, contextualist account of those expressions and construct a parallel treatment of modals. Then see how the resulting view does in the challenge cases compared to the revisionist views.
3. A Canonical Contextualist Account of Modal Expressions and its
Advantages
3.1 The Account
In order to test the approach I advocate against rivals views, we need a concrete proposal on the table. Here’s the one that seems attractive to me, in light of its fits with well-motivated views of quantifiers and other context-sensitive expressions, as well as its fit with the canon. If you don’t find this view attractive, remember that there are other ways of implementing the general approach that would need to be explored before giving up on the canon.
Here’s the account: Modal expressions are quantifiers over possibilities. When a bare modal expression is used, its ‘flavor’, as bouleic, epistemic, or deontic, etc. is determined by a speaker’s referential intentions in a context of use. A referential intention is a speaker’s intention for the addressee to recognize some feature of the context as settling extension in that context. In most uses, the domain of the modal quantifier is restricted and when it is, it is the speaker’s referential intentions that do the restricting. Applied to bare, epistemic modal sentences (BEMs) in particular, we get:
H1: The proposition expressed by the use of a BEM is determined by what’s known by a contextually determined group, where the relevant ‘group’ may contain only a single member and may or may not contain the speaker.
H2: Group knowledge is distributed knowledge; it’s the set of possibilities you get by intersecting the sets of possibilities compatible with what’s known by each member of contextually determined group G.[28]
H3: Which group is contextually relevant is determined by the referential intentions of the speaker regarding what determines group membership. (More precisely, her intention is for her addressee to recognize some feature of the context as helping to manifest what she takes to determine group-membership in that context.[29]) A speaker may intend her addressee to recognize that she is the sole member of the relevant group—or she may intend her addressee to recognize her intention to include, for example, whoever is also currently engaged in the sort of inquiry she is engaged in.[30]
H4: In the case of belief-attributions (where the attribution takes widest scope), which group is contextually relevant is typically determined at least in part by the attributee’s group membership.
H5: In the case of factives like “S realizes that might-P”, the complement of the attitude report typically gets its content determined by a contextually determined group that includes the attributee and needn’t include the attributor. In addition, factives like “realizes” require that the attributor regard the complement as true. (Notice that this is compatible with the attributor’s regarding the prejacent (the sentence that goes in for f in MODALf) of the modal complement as false.)
H6: For a sincere speaker S’s assertion of a BEM to be linguistically appropriate, S must believe the proposition expressed by her use of her BEM. (This is a norm of semantic competence.)
H7: For a sincere speaker S’s assertion of a BEM to be epistemically warranted, S must be justified in believing the proposition expressed by a BEM on the H1-H3 account of which proposition its use expresses.[31]
My arguments below depend upon the claim that the distinction marked in H6 and H7 is important. So, one quick note about those hypotheses: What’s important here is the distinction between two different kinds of badness in asserting that they are to mark, not the terminology used to mark it. Here’s that distinction in illustrations, banning the expression ‘norm of assertion’.
Compare two cases:
Case 1: In various different contexts, a woman asserts “Barack Obama is a Muslim”. We quiz her about her knowledge of Islam. We bring in the world’s foremost Imam to check her answers. All of them display a keen familiarity with Islam’s central tenets. To ensure that she’s talking about Barack Obama with her uses of “Barack Obama”, we show her clear photographs of the US president and ask her to identify the man she’s referring to. Indeed she is referring to Barack Obama. Then we ask about her beliefs about Obama’s religious commitments. She asserts “Barack Obama is a Muslim”. As evidence, she cites her ‘gut feeling’ and claims that one can tell he is a Muslim ‘just from looking at him’.[32]
Compare this with
Case 2: We are eating together. As I heartily consume large quantities of meaty sauce, I loudly declare my fondness for such sauces. I then say “I don’t like gravy”, offering as a justification that diseases that are caused by vitamin deficiencies are detrimental to one’s long-term health.[33]
Both Case 1 and Case 2 are examples of bad assertions, but their badness is of a different kind. In the first case, the speaker’s assertion exhibits a failure of rationality. In the second case, I manifest a lack of mastery of all of the terms I’ve used to express my beliefs. My arguments below depend only on the claim that these are two different forms of badness in asserting and not on how we mark it.
3.2 Some Advantages: Accommodating Flexibility and Objectivity
3.2.1 Flexibility
Some of the advantages of this account have already been noted. It gives a unified account of bare modal expressions and it fits with independently well-motivated views about quantifiers. Another advantage is that it accounts for two features of epistemic modal expressions widely accepted in the literature. Epistemic modals seem to exhibit a kind of flexibility and, at least sometimes, a degree of objectivity. By ‘flexibility’ I mean that which kind of group determines a modal’s domain varies with contexts. For example, in some cases, it looks like the knowledge of a single individual that is relevant for a BEM’s assessment, in others, that of some larger group. Here’s an example Gilles and von Fintel cite from Kratzer of a solipsistic case:
MAN:
“Suppose a man is approaching both of us. You are standing over there. I am further away. I can only see the bare outlines of the man. In view of my evidence, the person approaching may be Fred. In view of your evidence, it cannot possibly be Fred, it must be Martin. If this is so, my utterance of [M] and your utterance of [U] are both true.
[M] The person approaching might be Fred.
[U] The person approaching cannot be Fred.”[34]
As von Fintel and Gillies say in discussing this case,
“Kratzer claims that the BEMs in this example are claims about the speaker’s evidence. When we consider that the first speaker may well be aware that the second speaker has a better vantage point and may thus have a better idea about who is approaching, it makes sense to assume that the first speaker is only making a claim about her own somewhat limited information state.” [35]
In contrast, some uses seem to require non-solipsistic readings. Here’s an example from DeRose, modified by von Fintel and Gillies, that clearly requires the relevance of some larger group’s knowledge.
TEST:
“John has had a screening test that can rule out cancer but will not determine that he has it if he does. After the test has been run and the doctors have the results, Jane can say things like:
[K] I don’t know whether John might have cancer; only the doctors know. I’ll find that out tomorrow when the results of the tests are revealed.”[36]
[K] here is clearly at least appropriate, some might say also clearly true. If so, then it can’t just be the speaker’s knowledge that’s relevant for its assessment. On a solipsistic reading, as von Fintel and Gillies note, it won’t be appropriate. It also won’t be true. (If it were, you should be able to insert “in view of what I (Jane) know” in the complement of [K] to get something appropriate[37] and true.)
Together, MAN and TEST highlight the apparent the flexibility of modals. Sometimes they seem best interpreted solipsistically, sometimes not. Some revisionists have suggested that this flexibility poses a problem for the canonical contextualist account.[38] It seems to suggest that there is no unique, contextual parameter that can fit with our intuitions about the full range of cases. Pairing these cases can make a contextualist treatment of epistemic modals seem unstable or ad hoc. Sometimes it is the knowledge of a single individual that is relevant, sometimes that of a group. Conclusion: we should reject canonical contextualism.
With H3, no such problem arises for the present account. On that account, domain restriction requires a context, an expression (or its use), and a speaker’s intentions. Since speakers’ referential intentions are flexible, so is what determines domain restriction. In MAN, the first speaker (Kratzer) intends and is best interpreted as intending to make a claim about what is possible in view of what she knows. She is relying here on salient features of the context (features that she reasonably assumes are salient to her addressee) to make plain that this is her intention (including that it is plain to everyone that the addressee is in a perceptually better vantage point.)
In TEST, the conversational salience makes plain Jane’s intention for “…John might have cancer” to get evaluated in a way that includes the knowledge of John’s doctors. So, Jane is best understood as intending a group reading that includes them.[39] Given this, it is entirely appropriate for Jane to utter [K]; Jane doesn’t know whether John’s having cancer is compatible with what his doctors know, so she doesn’t know whether it’s compatible with what the group knows. Moreover, [K] comes out true on this reading.
Notice that it’s hard to see how a relativist can make sense of this case. It looks like Jane intends “…John might have cancer” to get evaluated in a way that somehow includes his doctors’ knowledge of the test results. But, as we’ll see, according to the relativist, that should get evaluated against what Jane knows in her context of assessment. As MacFarlane notes, for the speaker, the context of assessment is always her context, which at the time of the utterance just is the context of use. In such cases, there is no difference between the predictions of solipsistic contextualism and relativism.[40] TEST is such a case. Here the relativist will predict that [K] is inappropriate and false. But it’s not. So we have at least one case for which the present proposal provides a better explanation than actual relativists.
One option for the relativist here is to give up on the solipsistic relativism that generates the problematic prediction in favor of group relativism or flexible relativism. If he does, he’ll owe us an account of how a (metasemantic) account of how context determines the scope of the relevant group.
In both MAN and TEST, the speaker either clearly is or may be included in the contextually determined group. But not all cases are like this. Some BEMs require a reading that excludes the speaker from the relevant group. Here is such an illustrative case from Egan, Hawthorne, and Weatherson:
BUS
Ann is planning a surprise party for Bill. Unfortunately, Chris has discovered the surprise and told Bill about it. Now Bill and Chris are having fun watching Ann try to set up the party without being discovered. Currently Ann is walking past Chris’s apartment carrying a large supply of party hats. She sees a bus on which Bill frequently rides home, so she jumps into some nearby bushes to avoid being spotted. Bill, watching from Chris’s window, is quite amused, but Chris is puzzled and asks Bill why Ann is hiding in the bushes. Bill says
[(B)] “I might be on that bus”.
As Egan et. al. rightly note, supposing that Bill is in the contextually determined group won’t make sense of the appropriateness of (B) in this context. Bill, after all, knows that he’s not on the bus, so it can’t be compatible with what any group that includes him knows that he is. However, given that Bill is offering an explanation of Ann’s behavior, it’s clear from conversational salience that he intends Ann alone to be in the relevant group. Ann is jumping in the bushes because it is compatible with what she knows that Bill is on the bus. This explanation of the case is fully in keeping with the present account.
3.2.2 Objectivity
Other cases highlight the objectivity of BEMs by seeming to suggest that they are at least sometimes false, even when compatible with what’s known by what seems to be the contextually relevant group. What should the present proposal say about these cases? Here are two examples, the first from Hacking, the second from von Fintel and Gillies.
SHIP: A salvage crew is searching for a long-sunk ship. The mate relies on the ship’s old log to determine where the team should dive. On the basis of his calculations, the mate says
(H) “The hulk might be in these waters”.
Neither the mate nor any of his addressees knows anything that rules out the ship’s presence in the indicated waters, but a more careful examination of the log would show that the ship is 30 miles south.[41] Hacking reports having the intuition that in this case (H) is false.
Here is the von Fintel and Gillies case:
SCHMOLMES: Schmolmes is a detective who, unlike his more famous cousin, sometimes makes mistaken deductions. On the basis of such erroneous deductions, he declares
(G) “Ah, the gardener might be the culprit.”
Unfortunately for Schmolmes, his own interview notes conclusively rule out the gardener. Von Fintel and Gillies report that their intuition in this case is that Schmolmes has said something false.[42]
Hacking and DeRose offer canonical contextualist accounts of such cases by proposing that BEMs get evaluated against not what some contextually determined group knows, but what is within their epistemic reach.[43] Since Schmolmes’ notes are within his epistemic reach, as the logbook is within the mate’s, (G) and (H) each come out false. As others rightly note, though, the notion of ‘epistemic reach’ is difficult to fill out in any determinate and plausible way.[44]
Von Fintel and Gillies’ elegant solution is to treat stores of information as agents eligible for inclusion in a contextually determined group.[45] With this idea in hand, we get an explanation for why (G) and (H) each can seem false. In both cases, it is clear from context that the mate and Schmolmes are each basing their assertions at least in part on what the relevant information source contains. Schmolmes, we may assume, means to be drawing a conclusion from his interview notes. Likewise, we may assume that the mate means to base his assertion in part on what’s contained in the log. Given this, each is best interpreted then as intending to include, respectively, the knowledge of the note-writer (Schmolmes’ past self) and of that part of the log author’s knowledge that is expressed in the log. Once the relevant groups are extended to include these, each of (H) and (G) come out false. All of this is compatible with the present proposal, without appeal to the notion of epistemic reach.
Intuitions about these cases aren’t uniform, though. To the extent that you find nothing amiss with (H) or (G), that can be accommodated by the present account by noting that both Schmolmes and the mate may each be to some extent warranted in asserting what they do and their assertions are each linguistically appropriate, even if what they have said comes out false.
4. KEYS Revised and Puzzle Solved
So far so good. But we haven’t yet seen how the present account can handle the case that poses the greatest challenge to the canon. With KEYS, von Fintel and Gillies are trying to find a bit of dialogue that reflects a realistic usage of a BEM. In actual cases, speakers generally have a lot more information than the information we’re given in the original KEYS scenario. In fact, KEYS is described in such skeletal fashion that it can be unclear whether all intuitive reactions are responses to a single case or to different ones. In order to assess how the proposals do against our intuitions about cases, then, those cases need to be described with all the relevant details filled in so that we are sure that our intuitions are being tested against the same case.
To pose a problem for the canon, the basic KEYS case needs to be filled out in a way that generates the problematic intuitions. In some ways of filling out KEYS, though, the intuition that Alex’s assertion is warranted goes away. Here’s one such case:
KEYS 1: Let’s assume that both Alex and Billy are sincere, reasonably intelligent, and reflective. Alex is helping Billy search for her keys. It is compatible with what Alex knows that the keys are in the car. But Alex also knows that Billy is a very careful searcher and only asks for help after she has checked all of the obvious places to look for the location of the missing item. Billy admits that she is searching for her keys just after Alex sees her emerge from the garage where Alex knows the car to be located. Let’s additionally assume that Alex isn’t too tired or otherwise cognitively impaired to put these bits of information together. It isn’t lost on her that the best assumption given the evidence is that Billy has already thoroughly searched the car—and is still looking for her keys.
Suppose now Alex asserts (C) “You might have left them in the car”. Does that seem fine to you? Her assertion doesn’t seem appropriate, does it? Here is a clear case in which we don’t get the intuition that drives the moves to revisionary contextualism and relativism.
The above, canonical contextualist hypotheses have no trouble with this case. H3, together with the usual Gricean assumptions, will get us that, in a context in which Alex is helping Billy search for her keys, she is most plausibly understood as intending her modal to be evaluated against what’s known by a group that includes both herself and Billy. H7 predicts that her assertion is epistemically unwarranted—after all, Alex has every reason to believe that Billy has ruled out that the keys are in the car. H6, assuming that Alex is sincere, predicts that what Alex has said is linguistically inappropriate. Alex is bright and reflective—so how could she be understood as believing that the keys’ being in the car hasn’t been ruled out by what she and Billy together know? All of this precisely fits with what we intuitively want to say about this case—including our sense that the most sensible reply on Billy’s part would be “How can you really believe that?!” or “why did you say that!? You saw me just come out of the garage—clearly I just checked there!”
Notice that H1-H3 also fit with our sense of when (C) is true. If Billy has already ruled out their being in the car, the group-reading predicts that (C) is false—just as we intuitively think. But if, contra to what Alex has reason to believe, Billy hasn’t ruled that out, then (C) seems true, again, just as H1-H3 predict.
Even if, contrary to the best way of making sense of her assertion, Alex intended a solipsistic reading, we still have a reading of the case that explains our intuition that there’s something wrong with her assertion. If Alex has just seen Billy come out of the garage, knows that Billy is a careful searcher and that she is still looking for her keys, then the their being in the car does not seem compatible with what Alex knows.
Contrast KEYS 1 with another way of filling out KEYS:
KEYS 2: As before, but Alex knows that Billy is quick to enlist aides when she has lost something. Alex has no reason to think that Billy has already checked the car—indeed, she has some reason to believe that Billy hasn’t. Suppose, though, that Billy has in fact ruled out that her keys are in the car. And suppose now Alex asserts
(C) “You might have left them in the car”.
Here, we think that Alex’s assertion is warranted. But, we still don’t have a KEYS case that poses a problem for the canon. The present hypotheses provide an explanation for our intuitions in this case as well. Here, too, given that Alex is helping Billy look for her keys, she is best understood as intending the group-reading. The reason we have the intuition that Alex’s assertion is warranted is because of Alex’s epistemic position at the time of her utterance with respect to what the group knows. Alex is warranted because she has reason to think that the keys’ being in the car is compatible with what she + Billy know. Moreover, her assertion is linguistically appropriate because it expresses her belief in this compatibility. All of this is compatible with Alex’s claim being nonetheless false.
What about a harder case for canonical contextualism? Here’s one:
KEYS3: As before, but Alex has no reason to think that Billy either has or has not already ruled out that the keys are in the car.[46] Suppose that Alex now asserts (C). Here’s the question: In light of the contrast with KEYS2, do we really have the intuition that Alex is warranted in asserting (C)? Remember, Alex’s job is help Billy locate her keys and we’re assuming that Alex is sincere, reasonably intelligent, and reflective. Not knowing where Billy has already checked or how thoroughly, wouldn’t it be better for Alex to simply ask,
(Q) “Could they be in the car?”
My intuition is yes—assuming that Alex is intelligent, sincere, and reflective, it is somewhat inappropriate for her to assert (C), when she could have asked (Q).[47] If that’s your intuition too, then we already have a fit with H1-H3. The reason it is odd for Alex to assert (C) in this case is that she wouldn’t be warranted in believing that the keys’ being in the car is compatible with what the group knows, since she has no reason to believe that Billy hasn’t already ruled that out. And if Alex is intelligent and reflective, then she doesn’t believe what she doesn’t have reason to believe. And if she’s sincere and doesn’t believe it, it is linguistically inappropriate for her to assert a BEM that expresses such a belief. Moreover, if you have that intuition, then you have an intuition that does not fit with either revisionary contextualism or relativism, since on those views, (C) is warranted in KEYS3.
But maybe you don’t share my intuition; maybe you think (C) is fine here. If so, make sure that you aren’t confusing KEYS3 with
KEYS4: As in KEYS3, but Alex sincere but not reflective before she asserts (C).
Before deciding what your intuitions are in KEYS4 or whether you are distinguishing it from KEYS3, don’t forget the distinction between linguistic and epistemic norms of assertion, here captured at least in part by H6 and H7. Now that you’ve got that distinction in mind, ask yourself, do you find Alex’s assertion in KEYS4 both epistemically warranted and linguistically appropriate? Remember, Alex is supposed to be helping Billy find her keys.
It seems to me that in KEYS4, what Alex says is linguistically appropriate, though epistemically unwarranted. Unfortunately, this sort of thing is not uncommon. Ordinary speakers often find themselves expressing beliefs they don’t have good evidence for. To the extent that you find what Alex says here fine, that’s explained by Alex’s being in just such a situation. She believes that the keys’ being in the car is compatible with what the group knows, though she is not epistemically warranted in having such a belief. Her having that belief is explained by her failure to reflect on what she knows. In asserting (C), she is expressing that belief, so her assertion meets one norm for linguistic appropriateness. In general, linguistic appropriateness does not require epistemic warrant. If we accepted that it does, then we would also have to accept that much of ordinary, non-philosophical conversation is linguistically inappropriate. That is just too much error to attribute to folks who are clearly competent speakers.
Now back to KEYS3. Once its been distinguished from a case in which Alex unreflectively believes that the groups’ knowledge is compatible with the keys’ being in the car, do you still find (C) appropriate in KEYS3? Think of it this way. Imagine someone prompting Alex as follows:
You: “Ok, Alex. I’m going to ask you a question and your answer is to help Billy locate her keys. But first, let’s review what you do and don’t know. You know that Billy doesn’t know where her keys are. You know where you have looked, in particular, that you haven’t thoroughly checked the car. But you don’t know where or how thoroughly Billy has looked. For all you know, Billy has already thoroughly checked the car. Do you believe the keys might be in the car? Before you answer, take several minutes to think about it. And when you answer, be sure to direct your answer to Billy.”
According to the relativist and the revisionary contextualist, it would be fine for Alex to reply
(C’) “Yes. They might be in the car.”
But, if Alex’s aim is to help Billy find her keys, this would be a pretty odd thing to say, especially in light of the fact that she’s taken the time to think about it and she could instead say
(S) “I’m not sure.”
The canonical contextualist account proposed here gets this case right. This is exactly what we should expect if (C’) is given the Alex + Billy reading. And the Alex + Billy reading is just what we should predict in a case in which the speaker is engaged in the kind of joint project Alex is engaged in and H1-H3 are true.
I’ve been discussing the KEYS cases as if the present proposal as committed to a group reading in each of them. But it should be remembered that the proposal has a way of accommodating the flexibility of BEMs that it can apply to these cases to generate alternative, sopilsistic readings. In any of these cases, it may be that Alex thinks of her answer as merely speaking to the question of what is compatible with what she knows. If she does this, she is neglecting the task of helping Billy, which is a feature of the context that suggests a group reading, but sometimes practical tasks get neglected when we reflect narrowly on questions. If that were so in a case like KEYS4, H1-H3 would assign (C’) a solipsistic reading. Then Alex is warranted in asserting (C’) and her assertion is linguistically appropriate, though her assertion fails the Gricean maxim of quantity.
To be fair to my revisionary rivals, there’s another way of filling out of the KEYS case that needs to be considered, the case of retractions.
KEYS5: Here we first imagine Billy giving a more forceful negative response to Alex than (N). Let’s suppose that in reply to (C), Billy says
(N’) “No, they can’t be in the car. I’ve already carefully checked it.”
To which we imagine Alex responding,
(W) “Oh, I guess I was wrong, then.”
According to relativists, retraction cases are difficult for canonical contextualists to explain. The difficulty comes in explaining how a speaker, such as Alex, could be warranted in her original assertion, such as in (C), while as the same time being warranted in her retraction, such as (W). Relativism explains both.[48]
Von Fintel and
Gillies contest the retraction data.
They argue that speakers do not always retract in such cases, nor do
speakers uniformly have the intuition that retraction is appropriate.[49] I agree with von Fintel and Gillies here, but
think more can be said in defense of contextualism about these cases since
sometimes it does seem that retraction is appropriate. Moreover, there do seem
to be cases about which the relativist is right to claim that both a speaker’s
original assertion and her retraction are appropriate. So, we still need a contextualist-friendly
account of these cases.
If we imagine KEYS5 as a continuation of KEYS2, we’ll have just such a case. But this poses no difficulty for the present proposal. Given that Alex asserts (C) as a part of a joint project of locating Billy’s keys, H1-H3 predicts that she is best understood as intending the Alex+Billy reading of her assertion. Here Alex’s original assertion (C) is warranted because she has every reason to think that Billy hasn’t ruled out that the keys are in the car. Since Alex knows that she hasn’t ruled that out either, she has reason to believe that the keys’ being in the car is compatible with what they both know. Since she’s reflective and reasonably intelligent, she does believe what she has reason to believe. So, by H6 and H7, her assertion is both warranted and appropriate. In KEYS2, though, it turns out that Billy has in fact already ruled out that the keys are in the car. So, though Alex was correct to assert that the keys might be in the car, what she said was nonetheless false. Billy’s response allows her to see this and she rightly retracts her earlier claim.
What about the other KEYS cases? I’ve argued that in KEYS1 and KEYS3, Alex’s asserting (C) would be odd. If that’s right, then these cases cannot be filled out in a relativist-friendly, apparently contextualist-unfriendly way. If we fill out KEYS4 as a retraction case, then, on the present account, we’ll get a case in which a speaker’s original assertion is linguistically appropriate, though unwarranted and her retraction, appropriate and warranted. Moreover, the proposition her retraction expresses will come out true. In other words, retraction phenomena don’t provide any clear cases that can’t be explained by H1-H7.
Finally, what about a case in which, instead of retracting, the speaker sticks to her guns? Von Fintel and Gillies are right to note that there are these cases as well. Here’s their example:
KEYS6:
Alex: “The keys might be in the car.”
Billy: “They’re not. I still had them when we came into the house. Why did you say that?”
Alex: “Look, I didn’t say they were in the car. I said they might be there—and they might have been. Sheesh.”[50]
They rightly note that Alex’s final assertion seems entirely appropriate. In the above cases, I said that Alex is best understood as intending the group reading and that means that H1-H3 would predict that (C) gets the group reading. But if Alex’s first assertion in KEYS6 gets a group-reading, then her second assertion is inappropriate. So, KEYS6 is a problem for the present proposal, if Alex’s first assertion requires a group-reading here.
But the present proposal doesn’t require that. The role that speaker’s intentions play in that proposal makes a solipsistic reading available in cases such as KEYS6. It is true that, as in the previous KEYS cases, Billy is reasonable to take Alex to intend the group reading, since it is reasonable for her to assume that Alex is being cooperative and aiming to say the most informative thing she can. In her reply in KEYS6, Alex is presenting herself as having intended a solipsistic reading in her original assertion. There are a variety of reasons why she might do this. She might do this because, although it would have been more practically useful to intend the group reading were she in a position to assert it, she wasn’t in such a position and so retreated to the weaker claim. A second possibility is that Alex did intend a group-reading in her original assertion, but retreated to a solipsistic reading in defending herself against Billy’s reply. In so doing, her reply to Billy is either insincere or self-deceived, but still ‘appropriate’ in the sense of not displaying linguistic incompetence. Instead, Alex displays her ability to exploit the deference commonly accorded to speakers on the question of their linguistic intentions.
Here is a good place to remember an additional advantage of the present proposal. If we think of modal expressions, as canonical contextualists, revisionary contextualists, and relativists all do, as quantifiers over possibilities, then something like the present proposal is exactly what we should expect to get if modal expressions were to function like quantifiers over individuals. Each of these proposals see the use of bare modal expressions as involving quantifier domain restriction—none of them regard the use of BEMs as quantifying over all possibilities. So, all three proposals face the question of what it is that determines how that restriction is determined. H3 of the present proposal answers this question. A nice feature of that answer is its fit with the parallel question about what determines restrictions on quantification over individuals. To illustrate:
MEETING:
Sally: “Every student was at the meeting.”
George: “What, even those that are on leave in Nicaragua?”
Sally: “No, what I meant was every student in residence.”[51]
Here, Sally’s response makes clear what implicit restriction on “every student” she intended in her original utterance. Usually, context can work to make it clear which restriction a speaker intends. (Indeed, George seems particularly thick in his response—or engaged in the kind of wordplay that passes for humor among philosophers.) That speaker’s intentions play this role is a highly plausible assumption about quantification over individuals. (Indeed, it’s hard to see how to make sense of Sally’s ‘…what I meant was…’ in this perfectly natural bit of dialogue if they didn’t.) I’m proposing we accept this plausible assumption for quantification over possibilities as well.
Notice here that assuming that Sally intended the restriction she makes explicit in her reply in making her original assertion does not require that we assume she consciously entertained that restriction prior to making that original assertion. We need only assume that in making her reply she recognizes that restriction as the one she intended all along. I propose we make a similar assumption about speaker’s intentions in the case of implicit restrictions on the set of possibilities BEMs quantify over. Showing that speakers do not consciously entertain explicit restrictions prior to asserting is not enough to show that speakers don’t have intentions sufficient to determine unique propositions expressed.[52] Typically, context works to make a speaker’s intentions clear. But if it doesn’t, speakers are often on hand to aid in clarification. That’s what we find in both KEYS6 and MEETING.
5. Other Challenge Cases: Disputes and Eavesdroppers
5.1 Disputes
Here’s a case from MacFarlane that he takes to pose a problem for nonsolipsistic contextualists. (These are contextualists who take the knowledge of some group at the context of use to fix the domain of quantification.)
Suppose two research groups are investigating whether a certain species of snail can be found in Hawaii. Neither group knows of the other’s existence. One day they end up at the same bar. The first group overhears members of the second group arguing about whether it is “possible” that the snails exist on the big island, and they join the discussion. Although the two groups have different bodies of evidence, it does not intuitively seem that they are talking past each other when they argue. Nor does it seem that the topic changes when the first group joins the discussion (from what is ruled out by the second group’s evidence to what is ruled out by both groups’ evidence). To accommodate these intuitions, the Nonsolipsistic [i.e. group] Contextualist will have to take all the possibility claims made by both groups to concern what is ruled out by the collective evidence of everyone who is investigating the question (known or unknown)—for any of these investigators could show up at the bar, in principle.[53]
Since MacFarlane’s discussion ends there, presumably the final sentence is meant as a kind of reductio of group contextualism. But it isn’t hard to make sense of how there could be case of the kind he describes on the present account. Suppose the case is like this:
SNAIL: The members of Research Team I overhear Investigator 1 from Research Team II assert
(I) “It is possible that the snails are on the big island.”
Suppose, moreover, that when Investigator 1 speaks, she intends to include in the relevant group anyone currently engaged in the kind of inquiry she and her mates are currently engaged in. She has a mistaken assumption about who is in this group; she thinks that everyone in this group is a member of Research Team II. In light of this (and assuming that the snails being on the big island is compatible with what Research Team II knows) what she says is appropriate and warranted. Nonetheless, it will be false if the snails’ being on the big island is incompatible with what Research Team I and II together know. Moreover, given her intentions and corresponding intentions on the part of the members of Research Team I when they reject her assertion, it is straightforward to see how it could be that the members of the two groups are engaged in a dispute.
5.2 EAVESDROPPERS [third-party assessments of modals]
Another sort of case that is thought to help motivate relativism involves eavesdroppers. According to relativists, eavesdroppers are able to make appropriate, true, and apparently contrary third-party assessments of BEMs asserted in conversations to which they are not a party. Here is an example from Egan [2007]:
James Bond has just returned to London after a long day of infiltrating SPECTRE’s secret base in the Swiss Alps, planting a bug in the main conference room and slipping out by night after leaving persuasive but misleading evidence of his presence in Zurich. …while monitoring the newly place bug, Bond and his CIA colleague Felix Leiter overhear a conversation between Blofeld and his second in command, Number 2.[54]
After Number 2 has discovered the misleading evidence, Bond and Leiter overhear him say to Blofeld:
(ZURICH) “Bond might be in Zurich.”
Upon hearing this, Blofeld replies:
(O) “That’s true.”
Upon hearing (ZURICH), Leiter turns to Bond and says:
(L) “That’s false.”
According to the relativists, we should have the intuitions that in making their respective assertions, each of the speakers has said something appropriate. Moreover, they argue, it looks like Number 2 and Leiter are disagreeing about a common content. They then argue that no contextualist proposal is able to accommodate all of these intuitions. The only way to accommodate all these intuitions is to allow that a common proposition can be true as assessed from one context and false as assessed from another.[55]
Elsewhere,[56] von Fintel and Gillies make a persuasive case against this putative data. Since those arguments are neutral between canonical and revisionary contextualism, I’ll focus only on the one to which MacFarlane provides a relativist reply. Von Fintel and Gillies point out that when a speaker says “that’s false” in response to a BEM containing ‘might’ or ‘must’, it’s available to the contextualist to interpret the speaker as rejecting the prejacent rather than the modalized claim itself.[57] MacFarlane [forthcoming] accepts the ambiguity of “that’s false”, but then offers a second test to distinguish between the two readings. His test involves treating ourselves as eavesdroppers or third-party evaluators of the speakers in the dialogues that are generating the data.[58] In our assessment of ZURICH, we must be careful to distinguish our rejection of the prejacent (‘Bond is in Zurich’) from the rejection of ZURICH itself. Given the story, we know that Bond isn’t in Zurich. So, we will reject the prejacent. Is that what we, and Leiter, are doing in rejecting ZURICH? If so, then we have undermined the reading of this case meant to motivate relativism.
So, instead of asking us to assess assertions like ZURICH simply by registering our inclination to use the ambiguous “that’s false”, MacFarlane asks us instead to register our inclination to say that Number 2 spoke falsely. To say that Number 2 spoke falsely is to reject his entire claim, not merely the prejacent. So, if we are inclined to say that Number 2 spoke falsely, then we have forced a relativist-friendly reading of the case.[59]
I myself am not so inclined. But maybe I’m a victim of my own theoretical tutoring. The real issue here in any case is not whether philosophers or linguists steeped in the debates find themselves so inclined, but whether ordinary speakers do. It would be nice if I had some solid empirical research addressing that issue to point to here. In absence of that, I ask you to think of the ordinary conversations you’ve had and overheard. Here’s my question: When is the last time you heard someone say of someone else “s/he spoke falsely” in an ordinary conversation? Did you get the answer “never” too? The reason for this, I’ll hazard, is that MacFarlane’s test relies on speakers marking a distinction in ordinary language that they don’t in fact mark. Speakers do say “that’s false” in ordinary conversations. But, as we’ve seen, “that’s false” as a third-party assessment of a modal claim is ambiguous. This means that, theoretically speaking, this eavesdropper data (involving the use of ‘might’) is up for grabs.[60]
A second kind of eavesdropper case cannot be explained in this contextualist-friendly way, however. BEMs that involve comparative modals don’t have prejacents. So it’s not available to the contextualist to interpret the eavesdropper’s “that’s false” as a denial of the prejacent. To illustrate: Suppose that after finding the misleading evidence, Number 2 says to Blofeld:
(ZP) “Bond’s more likely in Zurich than in Paris.”
Overhearing him, Leiter says to Miss Moneypenny:
(PZ) “That’s false. He’s more likely in Paris.”
Here we can imagine a filling out of the context in which both (ZP) and (PZ) are both fine. How might a contextualist explain this, given that Leiter can’t be understood as denying a prejacent with ‘that’s false’?[61]
Here the contextualist’s answer should depend on how the context is filled out. There seem to me to be several ways the context might be filled out such that (ZP) and (PZ) are both fine, but that none of those ways results in a case the present account can’t explain. Comparative modals require ordering sources, in the Kratzer framework. The ordering source for epistemic modals is stereotypical; the worlds in the base are ranked with respect to their ‘normality’, their alikeness to the ‘normal course of events’.[62] So, roughly speaking, (ZP) is true or false depending upon whether more of the most normal worlds in the modal base are worlds in which Bond is in Zurich or in Paris. The question here is: what could the contextualist say is in their modal bases such that (ZP) and (PZ) are both fine?
There are two contextualist-friendly interpretations of (ZP) here. On the present account, which interpretation is correct will depend upon the speaker’s intentions. First, in asserting (ZP) Number 2 may intend for the modal base to be determined by what’s compatible with what he (or he together with Blofeld) know. A second possibility is that in asserting (ZP) Number 2 intends for the modal base to be determined by what’s compatible with what everyone currently engaged in his inquiry is engaged in, where being a part of that inquiry involves addressing the question “where is Bond more likely to be, Zurich or Paris?”
There are also at two ways that Leiter can be understood to be assessing each of the resulting two possible propositions expressed by (ZP) with (PZ), depending upon whether Leiter correctly appreciates which proposition Number 2 has expressed with (ZP). Suppose that Number 2 in fact only intends to speak to the question of which is more likely, given what he and Blofeld together know. First, Leiter could correctly appreciate that this is the proposition Number 2 expresses with (ZP). Leiter would then be asserting that that proposition is false; given what Blofeld and Number 2 know, it is not more likely that Bond is in Zurich. In a case like this, the contextualist will say that both Leiter and Number 2 may have said something linguistically appropriate, but only one of them can have said something warranted or true.
A second possibility is that Number 2 only intends to be speaking to the question of which is more likely, given what he and Blofeld together know, but that Leiter mistakenly understands Number 2 to be speaking to the question of which is more likely, given what is known by all currently engaged in his inquiry. In that case, Number 2’s assertion will be linguistically appropriate and may be warranted and true, depending upon what is more likely, given what he and Blofeld know. Assuming that Leiter’s interpretation of Number 2’s use of (ZP) is reasonable, (PZ) is linguistically appropriate and (so long as Leiter is right that, given what he knows together with what Number 2 and Blofeld know, Bond is not more likely in Zurich than in Paris) the asserting itself is in some sense warranted. It is reasonable for Leiter to believe the proposition he thinks he has asserted and reasonable for him to believe that he has asserted that very proposition, though in fact, he has not. Leiter is wrong about which proposition’s truth he is rejecting when he says “that’s false”. So, if what Number 2 has said is true, what Leiter has said with “that’s false” is false.
Those are two ways of filling out both (ZP) and (PZ) such that the use of each is fine, at least in the way that matters most to semantics, namely, linguistic appropriateness. There are two more ways both can be filled out, on the supposition that Number 2 intends to be speaking to the question of which is more likely, given what’s known together by all that are currently engaged in his inquiry. Number 2 believes that there are only two such individuals, himself and Blofeld. But he is wrong about this; Leiter and Miss Moneypenny are eavesdroppers also engaged in the same inquiry. Here Number 2 has said something linguistically appropriate and also warranted, assuming that he is warranted in supposing that he and Blofeld are the only ones currently engaged in his inquiry. But, if adding what Leiter and Miss Moneypenny know makes it more likely that Bond is in Paris than Zurich, what Number 2 has said is false.
We now get two distinct ways of interpreting (PZ), depending upon whether Leiter correctly understands which proposition Number 2 has asserted with (ZP). Suppose that Leiter is correct about which proposition Number 2 has expressed. Then what Leiter has said with (PZ) is appropriate, and, if what Number 2 has said is false, what Leiter has said is true and may be warranted. Or suppose instead that Leiter is mistaken about what Number 2 said; he thinks Number 2 is merely speaking to the question of what Number 2 and Blofeld together know. Here Leiter may have said something appropriate and his asserting may be in some sense warranted, if he is warranted in supposing what he does about which proposition Number 2 expressed and about what’s likely given what Number 2 and Blofeld together know.
6. Challenges for Revisionaries: Attitude-attributions,
Factives, and Assessor Knows More than Speaker
6.1.1 Factives and Attitude-attributions
What about cases that seem to pose challenges for relativism? One such case involves presupposition-accommodation involving factives. Here is an example from von Fintel and Gillies.
HENCHMEN: Neither Number 2 nor Blofeld has yet found the misleading evidence that Bond has planted. Listening into their conversation, it is perfectly appropriate for Leiter to say to Bond:
(E) “If Blofeld realizes you might be in Zurich, you can breathe easy—he’ll send his henchman to Zurich to find you.”
And it is perfectly appropriate for Bond to reply:
(T) “That’s true.”
The standard view is that presuppositions triggered in conditional antecedents carry over to their consequents. And factives, like ‘realize’, presuppose the truth of their complements. So, (E) presupposes that Bond might be in Zurich. But Leiter knows that Bond isn’t in Zurich. So, by the relativist’s lights, he shouldn’t be able to say something appropriate with (E); instead, we should have a case of presupposition failure. Likewise, since Bond knows that he is not in Zurich, his reply (T) shouldn’t be appropriate. But it is.[63]
Though von Fintel and Gillies offer this case to illustrate a problem for relativists, its not clear how their own proposal[64] can handle it. The difficulty arises because of their failure to explain what makes a reading of a BEM ‘available’. Given this, it’s not clear what readings they take to be the ‘available’ ones for “you might be in Zurich” in the antecedent in (E). If the only available readings are Bond-, Leiter-, and Bond and Leiter-, then their proposal also predicts a case of presupposition failure, saddling them with the same problem they saddled relativists with. If Blofeld-, Number 2-, and Blofeld and Number 2- readings are also available, then Leiter is warranted in asserting (E), but, unless every reading that includes either Leiter or Bond is excluded, Bond won’t be warranted in accepting what Leiter says or, if he is, it won’t be by satisfying the only sufficient condition on being warranted that von Fintel and Gillies identify. (That’s because the ‘strongest available reading’ that Bond ‘can reasonably have an opinion about’ will be the Leiter-, Bond-, Blofeld-, and Number 2-reading, on which the complement in (E)’s antecedent comes out false.) So, von Fintel and Gillies either need a way of excluding that reading (to avoid treating the case as one of presupposition-accommodation failure) or they need a different norm of confirmation/denial to handle cases like this one.[65]
The problems don’t arise, of course, if only the Blofeld-, Number 2- and Blofeld and Number 2-readings are the only ones ‘available’ in the context. If this is their view, it would be nice to have a non-ad hoc explanation of why, since all of their examples involve restrictions of available readings to conversational participants and neither Blofeld nor Number 2 are parties to the conversation in which (E) is asserted, while both Leiter and Bond are. Here the difficulty’s source is precisely parallel to a source of the difficulty for the canonical view in the original KEYS case, namely, the lack of a metasemantics. Here, though, a metasemantic account is needed to supply an answer not to the question ‘how do contexts restrict modal bases?’ but to the question ‘how do contexts restrict the range of available readings?’
The present proposal takes this case in two steps. Consider first a case of attitude-attribution that doesn’t involve a factive. Imagine instead of saying “that’s false” to Bond as in Egan’s original eavesdropper case, Leiter instead said:
(Z) “Oh, good. Number 2 believes that you might be in Zurich.”
To which Bond can reply:
(T’) “That’s true.”
Here both the speaker and addressee regard the prejacent of the might-claim as false. But that’s OK. In attributing a belief to someone, we are trying to characterize the attributee’s state of mind. So we should expect that when the attribution takes widest scope, which proposition the complement mightf has as its content is often determined at least in part by the attributee’s group membership.[66] Leiter is then saying (and Bond is affirming) something like:
(Z’) “Oh, good. Number 2 believes that it is compatible with what he/his group[67] knows that you are in Zurich.”
Notice that the complement of this belief-attribution is one that both Leiter and Bond can regard as true. This is what we get in the case of a factive such as ‘realizes’. Here the content of complement must be such that both attributee and attributor regard it as true. Imagine instead that Leiter says to Bond:
(R) “Oh, good. Number 2 realizes you might be in Zurich.” [68]
To which Bond can reply:
(T’’) “That’s true.”
Here we can understand Leiter as asserting something like (Z’) with the presupposition that the complement clause is true. And given that he has just overheard the Blofeld-Number 2 discussion of the misleading evidence, he is fully within his linguistic—and epistemic—rights to presuppose its truth. And so is Bond.
So what about (E)? Here Bond and
Leiter know that Blofeld and Number 2 haven’t yet found the misleading
evidence. Leiter is then best understood
as making a claim about what will be the case if they do. If they do, then they will come to believe
that Bond’s being in Zurich is compatible with what they know. And if they do, Leiter is happy to presuppose
the truth of the complement of that belief attribution. Given this, it is appropriate for Leiter to
presuppose the truth of the complement in (E)’s antecedent. And it is appropriate for Bond to accommodate
that presupposition in his reply.
6.1.2 Attitude-reports: A Complication
H4 has the advantage of fitting with data that suggest that when BEMs serve as complements in attitude-attributions, the attributee is at least almost always contained in the group relevant for their evaluation.[69] However, the need for this feature in a contextualist account is a bit uncomfortable. After all, if context serves to restrict the domain of quantification for BEMs, shouldn’t there be some contexts in which attributees aren’t included in the relevant group?[70]
To see how the present account can explain this limitation on the role of context, consider first a case of an attribution with an EM complement that contains an explicit restrictor phrase. Suppose that in conversation, Leiter calmly informs Miss Moneypenny that Blofeld and Number 2 are after Bond. Alarmed, she cries
“How can you remain so calm?! Blofeld and Number 2 are dangerous men!”
To which Leiter replies
“Don’t worry. For all they know, Bond might be in Zurich.”
Later, Miss Moneypenny reports this to M, Bond’s boss, explaining that there is no need to worry about Bond’s safety since, as she says
(P) “Leiter believes that, for all they know, Bond might be in Zurich.”
Here (P) seems fine. So, it isn’t hard to get a belief report with a modal complement where that complement gets evaluated against a group’s knowledge that does not include the attributee, at least when a explicit restrictor phrase is supplied. The difficulty only arises when that restrictor phrase gets left out. Compare (P) to
(NP) “Leiter believes that Bond might be in Zurich”.
Even with the context supplied, the use of (NP) does not seem to say what the use of (P) says. This needn’t, however, put pressure on the present account. On the present account, a speaker’s referential intentions must be such that she relies on some feature of the context it is reasonable for her to regard as salient to her addressee in order to manifest what she takes to determine the domain restriction. In this case, without the aid of the restrictor phrase, the use of “Leiter” makes Leiter’s mental states conversationally salient. This conversational salience makes it hard for an addressee to think of him as excluded from the relevant group, even if others have been mentioned earlier in the conversation. (This is analogous to cases involving quantifiers over individuals. Suppose I’m in your office and I say “all the philosophy books are mine.” Given the perceptual salience of your books, it’s pretty hard to hear my assertion as making anything other than the claim that all of the philosophy books in your office are mine, even if we add that I in fact intend to refer to the books in my home office.)
Other cases are able to override this difficulty. BEMs that figure in behavioral explanations or predictions of individuals or groups can serve to make the knowledge of that group conversationally salient. Typically, only an actor’s own mental states figure in the explanation of her action; this makes an actor’s own mental states more than merely salient. So, sentences expressing attitudes towards propositions about action-explanations for actors other than the attributee can provide examples of the generally problematic kind. To illustrate: Suppose I have devised a treasure hunt for a group of children I know well. I have told the children that the treasure is hidden somewhere on the house’s grounds. In fact, I have hidden it in the attic. You know that I’ve hidden it in the attic, but, like me, don’t know where the children have already looked, only that they haven’t found it yet. Not knowing the children as I do, you wonder whether they will think to check the garden. I reply:
(G) “If the treasure might be in the garden, they’ll check there.”
(G)’s antecedent can’t have the same content as “for all JD knows, the treasure is in the garden” or “for all JD and the children together know, the treasure is in the garden”. Unless I’m guiding them, it’s not my mental states, but only those of the children that are relevant for predicting the conditions under which they can be expected to check the garden. (The conditions this action-explanation implicitly appeals to are that the children are such that if the treasure’s being in the garden is compatible with what they know, they will notice this and their noticing this will (ceteris paribus) be enough to get them to check. The conditions, in this context, cannot be: the children are such that if the treasure’s being in the garden is compatible with what I or I together with the children know, they will notice this and their noticing this (ceteris paribus) be enough to get them to check.) Given this, (G) is plausibly read as having the same content as
(G’) “If, for all they know, the treasure is in the garden, the children will check there.”
So far so good, but we still need a case of an attitude-attribution. Here’s one: Suppose as I wonder off, a neighbor, Sally, joins you in watching the hunt. Sally also doesn’t know the children well and wonders whether they will think to look in the garden. You reply,
(BG) “Well, JD knows the children well and she thinks that if the treasure might be in the garden, they’ll check there.”
(BG) attributes to me the belief I expressed with (G). Since, plausibly, (G) and (G’) express the same proposition, the proposition attributed to me in (BG) is the same expressed by (G’). If so, then we have a case of a belief-attribution where the complement contains a modal clause whose content is not determined relative to the knowledge of a group that contains the attributee.
6.2 Assessor Knows More than Speaker
A second nice example from von Fintel and Gillies involves a case in which an assessor of a modal claim knows more than the speaker. In the example, Mordecai and Pascal are playing Mastermind. Mordecai has started to give Pascal hints. Pascal says:
(P) “There might be two reds.”
to which Mordecai replies
(D) “That’s right. There might be.”[71]
This example poses an interesting challenge for the relativist. There are just two cases here. Since Mordecai knows the number of reds he’s hidden, he either knows that there are two reds or he knows that there aren’t. If he knows that there aren’t, then (D) is false at his point of assessment. If he knows that there are, then (D) violates Grice’s maxim of quantity. Either way, the relativist predicts that there should be something wrong with (D), but it’s just fine.
Here too, it’s hard to see how the von Fintel and Gillies view handles this case. It doesn’t predict that either (P) or (D) would be out of order. But (D)’s being fine can’t be explained by its satisfaction of their sufficient condition on an assertion’s being warranted. On their view, an addressee’s confirmation/denial is appropriate if it is on the ‘strongest available reading he is in a position to be opinionated about’. For Mordecai, that should be the Mordecai and Pascal-reading. But Mordecai shouldn’t say (D) on that reading. After all, he knows what information Pascal has acquired as a result of his guesses so far. And he knows how many reds he’s hidden. If there aren’t two reds, Mordecai knows this and then (D) comes out false on the group reading. Morever, since Mordecai knows this, he shouldn’t assert (D) in that case. If there are two reds, Mordecai’s said something true, but violated the Gricean maxim of quantity,[72] since he’s in a position to assert “there are two reds”, but doesn’t. But, as they note, Mordecai’s asserting (D) seems intuitively fine.
Since von Fintel and Gillies confirm/denial norm is a sufficient one, it’s open for them to appeal to a different norm to govern (D) in this case.[73] It would be good to know what that norm is and why we should think it a genuine norm for an addressee’s response in cases such at this.
The present proposal, in contrast, has a straightforward way of handling this case. It should be clear to Pascal in the context that Mordecai in (D) is taking up Pascal’s epistemic perspective. The point of giving hints in a game isn’t to tell your opponent the answer, but to help him figure it out for himself. Pascal knows that Mordecai is in a position to tell him exactly how many reds there are. If Mordecai were intending a solipsistic reading of (D), that would be tantamount to telling him that there are two reds (if there are) or saying something thoroughly misleading (if there aren’t). Either way would violate the spirit of the game and the purpose of hint-giving. The proposition that Mordecai is best understood as expressing with (D), then, is that there being two reds is compatible with what Pascal knows. If he’s sincere, reflective, and reasonably intelligent, that is something Mordecai is fully warranted in asserting. And even if he’s not reflective, it’s certainly linguistically appropriate to express his belief that it is compatible.[74]
Conclusion:
The present, canonical contextualist proposal handles the challenge cases to both relativism and contextualism at least as well as and in some cases better than, its revisionary rivals. Moreover, it accommodates the flexibility and objectivity of modal expressions and does so in the context of a simple, unified theory of the semantics and pragmatics of those expressions. It does this by supplementing Kratzer’s original account with a Gricean, metasemantic account of how context determines a modal’s domain restriction. In some respects, that supplementation is hardly original. It takes a familiar, widely accepted proposal for how the denotations of demonstratives are determined and applies it to the case of domain restriction for modal expressions. In doing so, it guarantees its fit with a highly plausible account of how context serve to restrict domains in the case of other quantifier expressions. Taken together, these constitute fairly compelling reasons to prefer the present canonical contextualist account to relativism and revisionary contextualism.
Reference
Bianchi, Claudia. 2006. “’Nobody Loves Me’: Quantification and
Context.” Philosophical Studies 130: 377–397.
DeRose, Keith. 1991. “Epistemic Possibilities.” Philosophical Review
100: 581–
605.
Egan, Andy. 2007. “Epistemic Modals, Relativism, and Assertion.” Philosophical
Studies 133: 1–22.
Egan, Andy, John Hawthorne, and Brian Weatherson. 2005. “Epistemic
Modals in Context.” In Contextualism in Philosophy, ed. G.
Preyer and
G. Peter, 131–69. New York: Oxford University Press.
von Fintel, Kai & Anthony S.
Gillies. 2007. “An opinionated guide to
epistemic modality.” In Oxford Studies in Epistemology: Volume 2, ed. T. Szabo Gendler and J.
Hawthorne, 32-62. Oxford: Oxford University Press.
von Fintel, Kai & Anthony S.
Gillies. 2008. “CIA leaks.” Philosophical
Review
117: 77-98.
von Fintel,
Kai & Anthony S. Gillies.
Forthcoming. “Might Made Right”
(Ms. April 2008, URL http://mit.edu/fintel/fintel-gillies-2008-mmr.pdf
). In Epistemic Modality, ed. B. Weatherson and A. Egan. Oxford. Oxford University Press.
Grice,
Paul. 1991. “Logic and Conversation.” In
Studies in the Ways of Words, Paul
Grice, 22-40. Cambridge: Harvard
University Press.
Hacking, Ian. 1967. “Possibility.” Philosophical Review 76: 143-168.
Kaplan, David. 1989a. “Demonstratives. An Essay on the Semantics, Logic,
Metaphysics, and Epistemology of Demonstratives and Other
Indexicals.”
In Themes from Kaplan, ed. J.
Almog, J. Perry, and H. Wettstein, 481–563.
Oxford: Oxford University Press.
Kaplan, David. 1989b. “Afterthoughts.” In Themes from Kaplan, ed. J. Almog,
J. Perry, and H. Wettstein, 565–614. Oxford: Oxford University Press.
Kratzer, Angelika. 1977. “What Must and
Can must and can mean.” Linguistics and
Philosophy 1: 337-355.
Kratzer, Angelika. 1981. “The notional
category of modality.” In Words, Worlds, and Contexts: New Approaches
in Word Semantics, ed. H. J. Eikmeyer and H. Rieser, 38-74. Berlin: de Gruyter.
Kratzer, Angelika. 1986. “Conditionals.” Chicago Linguistics Society 22:
1-15.
Kratzer, Angelika. 1991. “Modality.” In
Semantics: An International Handbook of
Contemporary Research, ed. A. von Stechow and D. Wunderlich, 639-650. Berlin: de Gruyter.
Kratzer, Angelika. Forthcoming. Modals and Conditionals again. (Ms. January 2008, URL: ) Oxford: Oxford
University Press.
MacFarlane,
John. 2005. “The Assessment Sensitivity
of Knowledge Attributions.” In Oxford Studies in
Epistemology 1,
ed. T. Szabó Gendler and J. Hawthorne, 197–233. Oxford: Oxford University
Press.
MacFarlane, John. Forthcoming.
“Epistemic modals are assessment-sensitive” (Ms of May 9, 2008, URL http://johnmacfarlane.net/epistmod.pdf.). In Epistemic Modality, ed. B. Weatherson and A. Egan.
Oxford. Oxford University Press.
Stanley, Jason and Szabo, Zoltan. 2000. “On Quantifier Domain
Restriction.”
Mind
and Language 15(2), 219–261.
Stephenson, Tamina. 2007. “Judge
dependence, epistemic modals, and
predicates of personal taste.” Linguistics and Philosophy 30(4),
487-525.
Swanson, Eric 2006. Interactions with Context. Ph.D dissertation, MIT.
[1] Thanks to Derek Ball, John Gibbons, Errol Lord, Jennifer McKitrick, and Sarah Moss for comments on an earlier draft of this paper and to audiences at the Arche Contextualism and Relativism Workshop II, the University of Edinburgh, and the University of Stirling. Thanks also to Torfinn Huvenes, Dilip Ninan, Francois Recanati, and David Sobel for discussion and to Michael Dowell and Marguerite Walter for sharing their intuitions.
[2] Swanson 2006, MacFarlane forthcoming, and von Fintel and Gillies forthcoming.
[3] Kaplan 1991b.
[4] Egan 2007; MacFarlane forthcoming.
[5] von Fintel and Gillies forthcoming.
[6] DeRose 1991; Hacking 1967.
[7]Stanley and Szabo 2000; Bianchi 2006.
[8] See, for example, Kratzer 1986; Kratzer 1981; Kratzer forthcoming.
[9] This focus on ‘might’ and ‘must’ allows me at least initially to ignore the complexities that arise from the introduction of ordering sources, which rank worlds. I consider complications that require their consideration below.
[10] This summary of the canonical view owes much to the clear and concise presentation in von Fintel and Gillies forthcoming.
[11] See Kratzer [1991] and Kratzer [forthcoming].
[12] The original example of this kind is from Swanson 2006.
[13] The example is from von Fintel and Gillies forthcoming, pp.9-12. Of course, Billy needn’t cite these particular grounds for rejection. But, as von Fintel and Gillies note, she does need to cite some.
[14] Maybe you don’t have the intuition that (C) is true in these circumstances. If so, the puzzle can be stated solely in terms of the joint appropriateness of Alex’s asserting (C) and Billy’s asserting (Y).
[15] Maybe you don’t have the intuition that (C) is false in these circumstances. If so, then, the puzzle can be stated solely in terms of the joint appropriateness of Alex’s asserting (C) and Billy’s asserting (N).
[16] Ibid p. 11. See also Swanson 2006.
[17] MacFarlane considers a set of cases meant to bring out precisely the same tension. He writes:
“The problem is that we have two kinds of data, and they seem to point in different directions. If we attend to facts about when speakers take themselves to be warranted in asserting that something is ‘possible’, Solipsistic Contextualism looks like the right view. Unfortunately, it cannot account for the data about speakers’ assessments of epistemic modal claims…We can account for these data by making our Contextualism less solipsistic, but then we can no longer account for the data that originally motivated Solipsistic Contextualism.” (forthcoming, p.14.)
The advantage of the von Fintel and Gillies forthcoming and Swanson 2006 examples are that they each involve a single context, so the contextualist is blocked from claiming that context shift is involved to account for both the solipsistic-friendly and group-friendly intuitions.
[18] See both von Fintel and Gillies forthcoming, pp.9-12 and MacFarlane forthcoming, pp. 5-7 and 14.
[19] MacFarlane forthcoming.
[20] It’s not clear from von Fintel and Gillies [forthcoming] what, on their view, makes a reading ‘available’, but its clear from their discussion that in the KEYS case, these are the three readings they regard as the available ones. In order to generate predictions, the view will need to address this issue. That’s what the present account does in H3 below.
[21] Ibid. p.16.
[22] Ibid. p.17.
[23] von Fintel and Gillies forthcoming.
[24] Egan, Hawthorne, and Weatherson 2005.
[25] I’ll consider an objection to this terminological choice below. What’s important here is the distinction marked, not how we mark it.
[26] See Kaplan 1991a and 1991b.
[27] For a different Kaplanian account of quantifier restriction, see Bianchi 2006.
[28] Formally, [[B]]c,i= Ç fx(i), where [[B]]c,i is the extension of the modal at c, i, a context-
xÎGc
index pair, and fx(i) the function that takes group member x and world i into
xÎGc
what’s known by x in i. (See von Fintel and Gillies [forthcoming].) Here I plunk for this account of group-knowledge partly because it seems to get the cases right and partly just to have a concrete proposal on the table so that the view can be tested. There may be other accounts of group knowledge that do just as well and perhaps even better. Nothing here hangs on the present choice.
[29] On H3, referential intentions must satisfy a publicity constraint—they must be such that it is reasonable for a speaker to expect that the context can make her intention manifest to a reasonable addressee. For the most part, nothing hangs on this way of thinking of referential intentions.
[30] The present view may be counted as a kind of “flexible contextualism” in the Egan/Hawthorne/Weatherson 2005 sense. However, it avoids their objection to that view by including a principled answer to the question of how the relevant group is contextually determined.
[31] Any adequate account of BEMs must also account for their evidential features. These, I believe, can be accounted for within the basic features of the present proposal with (something like)
H8: In asserting a BEM, a speaker typically conversationally implicates that her evidence is indirect.
But I won’t defend that claim here.
[32] This is a fictionally filled out version of an actual interview with a voter from York, Pennsylvania. [NPR, date of broadcast, 9/11/08]
[33] This is a variant of a case from Macfarlane 2005.
[34] Kratzer 1986.
[35] Von Fintel and Gillies forthcoming, p.5.
[36] Ibid.
[37] Ibid.
[38] See von Fintel and Gillies 2008, pp.95-96; Egan, Hawthorne, and Weatherson 2005; and MacFarlane forthcoming.
[39] Its interesting that von Fintel and Gillies see a role for speaker’s intentions to play in this case in justifying the claim that [T] gets a group reading, but don’t seem to apply this reasoning across the board. (They write “…it is plausible that Jane fully intends her might to be, in part, about the information John’s doctors have.” (forthcoming, p.5.)
[40] MacFarlane forthcoming.
[41] Hacking 1967.
[42] von Fintel and Gillies forthcoming, pp.5-6.
[43] Hacking 1967; DeRose 1991. “Epistemic reach” is Egan’s nice phrasing. (Egan 2007.)
[44] MacFarlane forthcoming; von Fintel and Gillies forthcoming.
[45] See their forthcoming, p.6 footnote 8.
[46] This seems to be closer than KEYS1 or KEYS2 to the case that von Fintel and Gillies forthcoming intend and the spirit of Swanson 2006.
[47] This isn’t to say that we generally require that reflective speakers say what is maximally appropriate. Here, though, it’s reasonable to expect that Alex, since she’s reflective, will attend to a feature of her general knowledge that the conversation makes quite salient, namely, that she has no information about where Billy has already searched.
[48] See MacFarlane forthcoming.
[49] von Fintel and Gillies 2008.
[50] von Fintel and Gillies forthcoming, p.20.
[51] von Fintel and Gillies forthcoming pp.20-21. There they argue that their account of modal expressions fits with a plausible account of contextual restriction on quantifiers by suggesting that ‘the precise delimitation of the contextual domain of quantification for [“every student was at the meeting”] can often be indeterminate in a realistic context.” In MEETING, however, the intended restriction seems quite clear.
[52] The propositions determined here are determinate ‘up to vagueness’. Von Fintel and Gillies’ proposal that there is semantic underdetermination involved should not be confused with the common view that most of language, including modal language, contains some vagueness.
[53] MacFarlane forthcoming, p.10.
[54] Egan 2007, p.2.
[55] See Egan 2007; MacFarlane forthcoming.
[56] von Fintel and Gillies 2008, pp.81-83.
[57] von Fintel and Gillies 2008, p.83.
[58] MacFarlane forthcoming, pp.3-5.
[59] Ibid. p. 5.
[60] Other relativists deploy the eavesdropper case in a slightly different way. In their 2005, Egan, Hawthorne, and Weatherson claim that if contextualism were true, it should be appropriate for someone in Leiter’s position to say “that’s right” in response to Number 2’s assertion of ZURRICH. But, they claim, it’s not appropriate, so contextualism is false. But, first, intuitions may vary as to the appropriateness of such a response and, second, insofar as it may sound inappropriate, that can be explained by it’s sounding like a reusing of the same sentence in a different context in which it would express a falsehood.
[61] I owe both this objection and the example to Sarah Moss (pc).
[62] Kratzer 1991 p.644.
[63] von Fintel and Gillies 2008.
[64] von Fintel and Gillies forthcoming.
[65] Thanks to Derek Ball for discussion here.
[66] The assumption here is that it is the default case that one’s own modal beliefs are beliefs about what’s compatible with what one or one’s group knows.
[67] Whether ‘you might be in Zurich’ gets its content determined by what Number 2 knows or by what’s known by some group that includes him is determined by Leiter’s intentions in asserting (B’).
[68] Similar considerations allow the present view to avoid the Egan, Hawthorne, Weatherson 2005 objection to contextualism with a Speaker-Inclusion Constraint (SIC-Contextualism).
[69] Tamina Stephenson 2007, p.498.
[70] Thanks to Francois Recanati for discussion here.
[71] von Fintel and Gillies 2008, p.90.
[72] Grice 1991.
[73] Thanks to Derek Ball for discussion here.
[74] An example from Egan, Hawthorne, and Weatherson [2005] is in some respects similar to this one. In their example, Sally and Tom are lost in a maze. Sally knows the way out, but Tom does not. Tom asks her whether the exit is to the left. In reply, Sally says “it might be; it might not be”. Here we may also think of Sally as taking up Tom’s epistemic position and refusing to rule out any possibility that is compatible with it.