One of the issues that arose out of the Barcelona workshop on relativism was that there is a lot of terminological confusion around the place, and it would be good to sort this all out. The point of this post is to try and set out in a systematic way what the logically possible options are, with the hope that it leads to some clarity down the line. We’re going to start off very abstract, but hopefully things will get more concrete shortly.
The general case is that we have a class of tokens that fall under several different types. Each of the tokens has (perhaps relative to something) a value. We’ll distinguish three broad classes of views you might have about the distributions of those values.
* *Invariantist* – All tokens of the same type have the same value in all contexts
* *Contextualist* – Any token has the same value in all contexts, but some tokens of the same type take different values
* *Relativist* – A token can have different values in different contexts
The salient kinds of tokens (for the debates that are currently raging) are utterances and propositions. The salient kinds of values are truth values and contents. Since I can’t make much sense of what it is for a proposition to have a variable content, that leaves three applications of this schema.
# Tokens = Utterances, Values = Contents
# Tokens = Utterances, Values = Truth Values
# Tokens = Propositions, Values = Truth Values
Call the first of these the UC-issue, the second the UT-issue and the third the PT-issue. One can be an invariantist, contextualist or relativist about each of the first two questions, and either an invariantist or a relativist about the third. (I don’t really believe in a type-token distinction for abstract entities like propositions, so a view like contextualism that relies on a type-token distinction, doesn’t make sense. Maybe others will want to reinstate this.)
For example, orthodox (Cohen-DeRose-Lewis) contextualism is contextualist about the first, contextualist about the second, and invariantist about the third. Call this a CCI view, following the initials of the three answers.
Standard temporalists about statements like “It is raining in Barcelona” are invariantists about the first, contextualist about the second, and relativist about the third. Similarly, call this an ICR view. Several people at Barcelona (John MacFarlane, Stefano Predelli, Isidora Stojanovic) were arguing that ICR views in general can do a lot of the work that some people (including your narrator) used UT-relativist tools to do. (John calls these views nonindexical contextualist views, a name I don’t like because I use a similar term to describe UC-and-UT-contextualists who don’t think there is a relatively straightforward connection between compositional meaning, context and utterance content.)
The view about epistemic modals in Egan, Hawthorne and Weatherson is (I think) contextualist about the first, relativist about the second and relativist about the third, i.e. a CRR view. The view is contextualist because whether S’s knowledge is in the relevant knowledge base depends, inter alia, on whether S is the speaker. If you think the relevant knowledge base is entirely determined by assessment context, you’d support an IRR view.
Someone who thinks there is no contextual variant of any sort in a class holds an III view.
Filling in other kinds is difficult, but possible. Consider someone who scrawls the following bit of theist graffiti on a wall “It’s not too late to repent.” One view about the kind of speech act they have performed is that relative to the context of each context of assessment, i.e. each person who walks by and reads it, it says that it is not too late *now* to repent. If we mix this view with eternalism about propositions, we get an RRI view. I’ve sometimes called this content relativism. If we mix it with temporalism, we’re back to an IRR view. (I assume here that for the sake of linguistic theory we should assume that sometimes it is too late to repent and sometimes it isn’t, however theologically plausible that may be.)
Given the formal structure I set up, there are eighteen possible classes of views. (Three answers to each of the first two questions, two to the third, and 3 x 3 x 2 = 18.) But some of these are not live options. If one is a UC-invariantist (about a given class) and a PT-invariantist, then one must be a UT-invariantist, for there is no other way for utterance truth value to vary. So ICI and IRI are off the table. Conversely, if one is a UC-invariantist, but a PT-relativist, then one cannot be a UT-invariantist, since each utterance type expresses a constant proposition, and those propositions vary in truth value, one cannot hold a UT-invariantism view. These kind of considerations seem to rule out the following types. (In some cases, such as the inference that CII is not possible, I assume the classes of utterances we have in mind are large enough that any variation in content leads to a variation in truth value somewhere. CII is obviously a possible view about a small, uninteresting class of sentences.)
IIR
ICI
IRI
CII
CRI
RII
RCI
That leaves 11 options. Of those, 3 rely on the kinds of relativism ‘cancelling out’. These are
CIR
RIR
RCR
What I mean by ‘cancelling out’ is that as the truth value of the proposition varies, the content of the utterance varies equally so the truth value of the utterance (in a context or invariantly) stays fixed. These views seem possible (temporalists about past-tense statements have a CIR (or RIR) view for example) but it is hard to think of cases where they are interesting about a wide variety of cases. That leaves us with the following eight cases, of which the starred ones have been mentioned above.
III*
ICR*
IRR*
CCI*
CCR
CRR*
RRI*
RRR
Although RRR seems possible, it is hard to imagine a case in which it isn’t overkill. So I won’t mention it much more. What does seem a little worth exploring is CCR. As I mentioned above, there is a lot of interest these days in ICR views, but I suspect in a few cases we’ll need to expand those to CCR views. (Examples are left as an exercise for the very interested reader.)
After all of this laying out of the land, I want to end on a polemical note. Although I’ve used the terms ‘invariantist’, ‘contextualist’ and ‘relativist’ as constituents of terms I’ve used, I’ve never used them on their own. I’d like to make a linguistic proposal. (Actually it was Crispin Wright’s proposal in Barcelona, and I’d like to endorse it.) That when those three terms are used on their own, we use them for what I call here UT-invariantism, UT-contextualism and UT-relativism. That is, that we take utterance truth to be the salient issue. There are three reasons for this. First, it matches most closely current usage, as close as I think a systematic definition can do. Second, it seems these capture important schools of thought, in a way that the UC-classifications and PT-classifications do not. Third, it is sociologically interesting, because it puts all the new 21st century relativists together, as relativists, and away from any views that had mainstream coverage (at least in philosophy of language) more than a few years ago.