Wolfgang Schwarz

Hi Wo,

I heartily approve of appealing to aboutness in this context (and I too would like a workable theory of that notion to be around). For what it is worth, the way I want to play a similar thought is to use the invariance of "aboutness" facts to explain away intuitions against inscrutability of reference (in the semantic-value sense).

I guess what your line turns on is what we say about our use of (seemingly) semantic vocabularly, e.g. "reference". You shouldn't insist that the semantic value of "reference" is a relation that pairs "WWII" to WWII, otherwise you'd be pulling determinacy of subsentential *semantic values* out of the air. Rather, your claim has to be that the reference of "reference" is a relation that pairs "WWII" to WWII.

Here's one question for you if what I've just said is right: what progress does the appeal to aboutness make? Putnam, Davidson and others noted that by reinterpreting "reference" (and "aboutness" and "causality" etc) in ways appropriate to a permutated interpretation of a language, we can make all the familiar disquotational principles come out true. ("reference" in this sense is penumbrally connected, in Fine's sense, to all the singular terms in the language)? Why do we need to say more than this? What exactly is the further thing as-yet-undone that requires we start talking about aboutness?

There's a really nice paper by Hartry Field, from back in the day (I think it's "Conventionalism and instrumentalism in semantics" Nous 1975). I'd be interested to know whether you think that the idea in that paper are strategically similar to the ones you're suggesting here.

There's an entirely different way of taking what you say (which I think that Brian Weatherson has suggested in the past)---and that's to say that aboutness-facts as constitutive of semantic facts in the same way as conventions of truthfulness for Lewis: that is, no semantic theory counts as successful unless it respects both facts about conventions of truthfulness and "conventions of aboutness". I took it that you weren't taking this direction since you were conceding that "semantic value" was radically inscrutable, even though "reference" turned out not to be.

Anyway, be interested in hearing whether'll you'd want to associate yourself with either of these approaches.

Hi Robbie,

yes, using the invariance of aboutness to explain away intuitions against inscrutability is exactly what I'm trying to do in the main part of the post, before the final brackets. Have you worked that out somewhere?

Re your first point: I'd rather not talk about the semantic value, nor about the reference of "reference" (because paradox is lurking here), and just say that there are certain quasi-analytic constraints on the use of "refer" that rule out that "WWII" refers to a cherry.

I don't know much about what Putnam, Davidson and Field have said about all this, but I used to believe (half-heartedly) something that sounds similar: that our concept of reference is mostly exhausted by disquotation principles and hence that what we should demand of a semantics in this respect is only that it be compatible with those principles.

But I think I'd really prefer to have more than this, though it's a little hard to explain what. In a sense, I want to respect the intuition that "Berlin is broke" is about Berlin, and not just the intuition that "'Berlin is broke' is about Berlin" is true. And I'd like to have determinate intra-linguistic aboutness: "Millionen Russen starben im zweiten Weltkrieg" should be determinately about the Second World War, too. Maybe the crucial point is that Putnam, Davidson and Field would all accept that *in some sense*, it is indeterminate whether "Berlin is broke" (in English) is about Berlin. Whatever exactly that claim amounts to, I'd rather deny it.

I definitely wouldn't want to impose a further general constraint on semantics besides conventions of use. I suspect it is just an accident that there are workable semantics for all our languages that make use of reference (rightly-so-called). So I suspect the further constraint would render some possible languages un-interpretable (or un-interpretable by any reasonable semantics). I also believe it's important to keep different semantic roles separated: it's one thing to figure in the best systematic description of our linguistic practice, it's another thing to satisfy aboutness conditions. If we can find semantic values that do both, great. But that doesn't mean these very different roles always belonged together or constitute some especially important and unified semantic notion.

That's really interesting. When you say that it's an accident that workable semantics make use of reference, is that because there might be languages with some totally "other" semantics (e.g. the Quine algebraic stuff). Or that there're languages where there's not sufficient complexity in use to lead us to detect singular term reference (Evans suggested something like that in "Identity and predication"). Or that there are languages that e.g. use quantifiers rather than singular terms, so aren't interpretable via singular-term reference? Or something else? I guess one response would be to maintain that the constraints are conditional in nature: if "t" is a singular term in sentences that are about O, then CP semantics should have assign O to "t". If there are no singular terms, then the constraint just lapses. But it does seem kinda ugly to me, especially in the first two cases.

What I was thinking of for Putnam and Davidson is really just the "just more theory" moves. One way of thinking of that argument is as making the point that you don't need determinate semantic value, to have sensible reference-talk (and, one might add, to fulfill at least some of the explanatory roles associated with reference). That seems to me the interesting core of the "just more theory" considerations, anyway.

I think that reading of the "just more theory" move is found in the Field article. And Field (at the time) was still wanting a reductive theory of truth-conditions in terms of a causal theory of reference (he wasn't yet into his deflationist phase). He wanted to (a) resist the idea that the "just more theory" considerations undermined what he was doing in "Tarski's theory of truth" (b) concede that there's no "deep" reason why we should talk about the reference-relation (i.e. the Fieldian causal relation) rather than some permuted variant, for the purpose of semantics: there's no explanatory job that could be done with the first that couldn't be done with the latter. It's a pretty delicate line to tread: did seem to me similiar in spirit to what you were proposing.

I see the point about the inter-linguistic usage of "reference". That does seem to be something gained over straight appeal to the disquotational notion.

p.s.
I haven't worked out this stuff anywhere: there's about a paragraph about Aboutness in this context in my thesis, and there's a paper I'm working on where I repeat that stuff. (It's page 6 of http://www.personal.leeds.ac.uk/~phljrgw/wip/thepriceofinscrutability.pdf)
Really, it's just an idea that intrigued me and that I want to think more about in the future. In particular, I wanted to make sure that I understood "aboutness": I think I picked up the characterization I use from Lewis's "Events". I wonder whether something more sophisticated is going on in "Statements..." and "Relevant Implication".

I was thinking of appealing to facts on a sentence-by-sentence basis: e.g. appealing to the fact that whatever goes on sub-sententially, "Socrates is wise" is about Socrates. One problem with that is that e.g. "water is H20" and "1+1=2" aren't distinctively about water or the number 2, on the formulation I had in mind (necessarily true statements turn out to be about everything). And I was also concerned with statements that contain names referring to people that don't exist in this possible world... So it all seemed a bit messy around the edges. One nice thing about your way of doing things (construed as explaining-away intuitions, rather than anything stronger) is that by dropping the sentence-by-sentence appeal, at least some of these problems go away: it's because of the fact that "water is wet" is about Water, that we relate "water is H20" to Water.

What I had in mind as a semantics that makes no use of reference were 1) semantics for a language about things with very unclear temporal and modal boundaries, like feelings and weather conditions: think of a language that contains only sentences translating "it's raining", "it's snowing heavily", "yesterday, it rained all day", etc. It isn't obvious that a workable semantics here will inevitably involve referents; 2) semantics for a language whose sentences directly express possible distributions of fundamental properties in a world, without naming any hooks for these properties to attach to; 3) something in the direction of a semantics that interprets all singular terms as generalized quantifiers or as functions from various indices (world, time, place) to properties or the like.

I have sympathies with what you describe as Field's view (I still haven't looked up the article), but it's not what I'd like to promise now. Which is: First, even if semantic values of subsentential expressions are radically indeterminate, what the sentences are about is not; second, one should only speak of "reference" (as opposed to "semantic value") if, in normal cases, the terms in a sentence refer to objects the sentence is about. ("In normal cases" excludes, for instance, necessary truths.) It follows that while semantic values are underdetermined, reference is not.

I'm not sure "water is H20" and "1+1=2" are distinctively about water or the number 2. I think there's something right about the logical empiricists' claim that necessary/a priori truths aren't distinctively about anything. Nevertheless, if in virtue of its occurrence in other sentences, "water" *refers* to water, one could say that "water is H2O" is about water in another sense of aboutness, by containing a name that refers to water. One could also define aboutness in terms of A-intensions rather than C-intensions. Neither move will help for numbers, though.

Weird coincidence. I just (about a 2 months ago) came across Goodman's paper "About" (Mind, Vol. 70) and Ryle's with the same title but placed in single quotes. Now that I do a search, I notice that there are probably tons of papers on the subject (e.g. Putnam & Ullian's, Thalheimer's, Rescher's, etc.). It just struck me as strange that there is an entire enterprise built around giving an analysis of 'about'.

Continue on!