Abstracts

NB: provisional abstract

Epistemic possibilities are relative to bodies of information or, as I will call them, perspectives. To claim that something is epistemically possible is typically to claim that it is epistemically possible relative one’s own perspective (the body of information currently available to one). 

To assert epistemic possibilities we commonly use bare, unqualified epistemic possibility (EP) sentences, sentences that do not indicate the perspective from which the possibility is being considered. I mean sentences like “Maybe Floyd Landis used a banned drug” and “Perhaps your keys are still in the front door”, as opposed to ones like “As far as I know, maybe Floyd Landis used a banned drug” and “As far as you know, perhaps your keys are still in the front door”. 

In most conversational situations we don’t need to make the perspective explicit because it is evident what the relevant perspective is. That’s why we can effectively use bare EP sentences. Even so, such sentences fall short of fully expressing propositions, since they fail to mention (refer to, describe, or quantify over) perspectives. That’s because, as I said, epistemic possibilities are relative to perspectives.

Proponents of currently popular views on epistemic modals take for granted that bare EP sentences do express propositions. They implicitly take the fact that we have strong intuitions about the truth or falsity of utterances of such sentences, given particular scenarios, as strong evidence that the sentences themselves are capable of being true or false. 

Their focus is on how truth values can shift with a change in a person’s perspective or with a shift from one person’s perspective to another’s. According to Contextualism, epistemic modals are like indexicals, so that bare EP sentences are context-sensitive. That is, the specific proposition such a sentence expresses depends on its context of use, which somehow determines the perspective included in the proposition. In contrast, Relativism does not treat epistemic modals as anything like indexicals. 

It maintains that a bare EP sentence expresses a fixed proposition (assuming the sentence is not context-sensitive in some other respect, by virtue of containing any genuine indexicals). But this is a proposition whose truth or falsity depends on the perspective with respect to which it is considered.

Radical Invariantism, as I call my view, denies that the contents of bare EP sentences vary with the context. However, it also denies that their contents add up to propositions. 

These sentences express only propositional radicals, not full-fledged propositions, hence are not capable of being true or false. Nevertheless, they are perfectly capable of being used to assert propositions and of being taken as so used, because utterances of them implicitly mention the perspective with respect to which the relevant possibilities are to be considered.

To demonstrate the advantages of Radical Invariantism over Contextualism and Relativism, I will look at a variety of examples and take up various natural objections. These examples will also illustrate what we do when we use or encounter bare EP sentences. 

Typically we use them when seeking or considering answers to questions and when advising others about questions they’re looking into. We assert or at least raise certain possibilities and reject or just dismiss others. We also use these sentences to challenge knowledge claims by raising counter-possibilities, potential defeaters. These sentences also occur in embedded clauses to attribute attitudes toward epistemic possibilities and to explain people’s actions in light of them. 

In all these cases, we don’t need to make explicit the relevant perspective as long as it is evident which perspective that is. This can be so even when more than one perspective is involved, but in most cases the relevant perspective is the speaker’s, if only by default.

In order to make my case for Radical Invariantism, I will invoke several distinctions. 

The distinction between a sentence expressing a proposition and being capable of being used to convey a proposition will be used to show that bare EP sentences, like sentences of many other sorts, can be semantically incomplete and still be pragmatically effective. 

The distinction between epistemic possibilities and propositions involving them will be used to explain conflicting intuitions about the various examples that have fuelled recent debates about epistemic modals. 

And the distinction between representing such a proposition and the structure of the proposition itself will help explain how it is that one can consider an epistemic possibility proposition from a perspective without having to represent that perspective Ð one can simply be in that perspective. 

Together, these three distinctions will be used to explain, or in some cases explain away, the intuitions that have motivated Contextualism and Relativism.

The problem of epistemic modals is not as hard as contextualists and relativists have made it out to be. The trouble with their approaches is that they inaccurately describe the data and, as a result, misconstrue the problem. 

The solution doesn’t require semantic bells and whistles. The strategy behind the solution I’m proposing is to recognize the semantic slack left by bare EP sentences and to focus on how speakers who use and encounter them pick up the slack.

Speakers sometimes ask one question and thereby indirectly ask another. Suppose that Frances is working on a crossword puzzle while Greta is reading a newspaper.

Frances: Do you know what the 13th element in the periodic table is?
Greta: Yes.

Frances asks whether Greta knows what the 13th element in the periodic table is. Greta answers that question by uttering “Yes”. But Frances clearly wants Greta to answer another question, namely the question of what the 13th element in the periodic table is. Frances asks that latter question indirectly, by directly asking a question about Greta’s knowledge. But does Frances conversationally implicate a question by asking a question?

This example seems rather similar to cases in which a speaker implicates one proposition by asserting another, yet Grice’s theory of implicature seems ill-equipped to explain it. In this paper, I explore how Grice’s theory of conversational implicature might be modified so as to accommodate acts of implicating questions with questions. Along the way, I discuss a number of fundamental issues in the theory of speech acts and implicatures.

The truth of claims such as Vulcan causes perturbations in Mercury’s orbit, if it exists, or Vulcan is identical to itself appears to require a free logic, for none of them entails Vulcan exists. 

Free logics are nowadays classified as negative (all atomic sentences are false), positive (some, a = a in particular, are true) and neutral (all are neither true nor false). We should reject negative free logic, if we reject bivalence for claims involving vacuous singular terms. As Lehmann notes in a recent survey of research into free logics, Burge’s justification for the stipulation presupposes it, and the same applies to Sainsbury’s recent discussion.

Here I would like to support a positive free logic with a supervaluationist semantics. Lehmann criticizes a philosophical justification by Bencivenga based on a “counterfactual theory of truth”: “Why should truth, which is ordinarily regarded as correspondence to fact, be reckoned in terms of what is contrary to fact? Why should we reckon that ‘Pegasus is Pegasus’ is true because it would be true if, contrary to fact, ‘Pegasus’ did refer?” (op. cit., 233), concluding, “If supervaluations make sense in free logic, I believe we do not yet know why” (ibid).

I would like to explore a way of making sense of them, by appealing to distinct notions of what is said.

Four decades ago, Grice (1967) provided the impetus for modern pragmatic theory by distinguishing what is said from what is implicated, and by further bifurcating the latter category into the conventionally and non-conventionally (in particular, conversationally) implicated. Since then, these Gricean moves and their implementation have been robustly challenged from a number of directions, and it is two of these challenges that I will focus on.

In addition to introducing the very successful and user-friendly line of conversational implicatures, Grice sponsored a somewhat inchoate class of what he termed conventional implicatures for which the reception in the scholarly market has been mixed at best. A conventional implicature associated with an expression E is a non-cancellable aspect of the meaning of E that does not affect E’s truth conditions.

While this construct has evoked much recent scepticism - Bach (1999) assigns it a chimerical status, while Potts (2005) undertakes a pyrrhic rehabilitation - Grice’s proposal for dealing with components of speaker meaning that are irrelevant to the truth conditions of the primary asserted proposition has a rich lineage. Frege’s (1892, 1918) notion of Andeutung, invoked for aspects of conventional meaning that do not “affect the thought” or “touch what is true or false”, is a direct precursor of Grice’s conventional implicature. 

I will argue that for a range of constructions including the familiar vs. formal second person singular pronouns of many modern European languages as well as particles like but and even, epithets, discourse particles, and active/passive alternations and word order variation, the approach favoured by Frege and Grice remains eminently plausible. Drawing on the differential roles of entailment and assertion, I will propose a new taxonomy of relations to deal with the diverse range of both primarily asserted, secondarily asserted, and non-asserted aspects of communicated meaning.

Within the realm of conversational implicature, the Gricean apparatus of nine implicature-generating maxims falling under four macro-categories is a plausible candidate for winnowing, as Grice himself recognised. Can the maxims be reduced to two or three general pragmatic principles as in neo-Gricean work (Horn, Levinson) or even to one overarching principle (as in Relevance theory)?

In particular, are there sufficient reasons for maintaining a fundamental distinction between upper-bounding (eg scalar) Q-based implicature and strengthening R-based implicature (à la Horn 1984), or do both simply instantiate strengthening to a more specific meaning, as Carston (2002, 2005) has argued? In support of the Manichaean model, I will present evidence from a variety of constructions indicating this is indeed a distinction with a difference.