In Part One, I described reference as “rationalism’s reality problem.” How does a belief connect with the thing it is about?
Rationalism proposes a correspondence relation, mediated by propositions. Analysis showed that these are necessarily non-physical. Correspondence and propositions are metaphysical entities without causal power; they might as well be a system of magic mediated by obliging fairies. I promised to provide an alternative, naturalistic understanding of reference.
In this chapter, we’ll perform an ethnomethodological flip. We’ll replace rationalism’s metaphysical questions about how reference works in theory with concrete observations of the everyday hassle of achieving references in practice.1
Part Three will rely on this chapter’s explanation to understand how we use mere reasonableness to connect systematic rationality with reality. This chapter also sets up for a discussion of reasonable instruction use, which Part Three relies on to understand how we perform rational procedures.
Stupid referring tricks
In a naturalistic account, physical things refer. Principally, physical people refer. We have to do the job that the correspondence theory attributed to metaphysical fairies. Reference is a task that we achieve with concrete work over time in a specific situation for a specific purpose.
According to tradition, high correspondence magic is extraordinarily elegant, and the fairies are astonishingly beautiful; but they are only visible to highly-trained philosophers.2 Nice for them.
The rest of us have to get by with stupid referring tricks.3 We don’t know any general method for referring, so not only do we have to do the fairies’ work for them, we have to figure out how each time. This involves actual work, actually doing things, not “mental computation.” We resort to ungainly, amateurish kludges. Sometimes they’re one-offs that work just well enough to get a job done before falling apart. We’ve seen one example already: referring to a restaurant patron as “the straw hat.”
The rest of this chapter is a collection of stupid referring tricks.4 Referring is accomplished by whatever means is available, and improvised methods are unenumerable, so there can’t be any systematic theory or rational taxonomy of reference, only an unsystematic catalog of special cases.
We’ll mostly look at referring accomplished by spoken language, because that’s observable. The theory will be that written language, and also unobservable referrings, such as by beliefs, work pretty much the same way.
Good enough referring
“The Spanish Inquisition” explained how the open-ended richness of reality, with its unenumerable potentially relevant details, defeats rationalism.
In “The National Omelet Registry,” we saw that propositions, since they live in a context-free Neverland outside of space and time, have difficulty relating to beliefs like “that dog is digging up the garden.” Which dog?
Referring is possible only because the specificity of our purposes limit relevance. We don’t go around referring to arbitrary sets of atoms or random regions of spacetime. In everyday activity, we only want to refer to something because it’s relevant to our purposes. The way in which it is relevant usually provides adequate resources for referring.
Reasonable purposes are never absolute. Good enough is good enough. Sort-of truth is generally all we need. That’s good because it’s generally all we can get. An omelet doesn’t need to come out perfectly (if that were even meaningful); it only needs to look and taste good enough.
Spoken language referring is adequately accomplished if the hearer can adequately determine the referent when the time comes. What counts as “adequately” depend on the purpose of the activity.
The forklift operator approaches the building foreman with a load of drywall. “Yeah, put it over there,” says the foreman, nodding in the general direction of an empty bit of the construction site. Where exactly is “there”? What are its boundaries? How does the forklift driver know whether she’s dumped it in the right place? It doesn’t matter. The foreman probably has only a nebulous “there” in mind, anyway. All that matters is to put it somewhere out of the way but easily available when it will be needed tomorrow.
If someone exclaims “the President of Burkina Fasso has been shot!” you may or may not need to know who specifically that is for reference to have been accomplished. It depends on how and whether and why you care.
If you are making breakfast on your first day in an AirBnB kitchen and think “I’ll use the eggbeater,” you haven’t accomplished reference until you’ve rooted through several drawers full of clutter and found it. If you can’t, “the eggbeater” was a failed attempt at referring.
A referring expression only has to be specific enough to get the job done, here, this time. How, as the speaker, do you know how specific that is? Referring may be analyzed as giving instructions for how to access the referent. Referring is more a matter of knowing-how than knowing-that. You’ve generally got a pretty good sense of the hearer’s know-how, because it’s pretty much the same as yours. But you can’t always get this right, and the hearer won’t always get it right. If it doesn’t work, you repair by trying again: “No, the other one.”
Referring relative to the context
Referring is always context-dependent.5 This is a brute fact of physics. There is no absolute coordinate system, no (0, 0, 0) point to start from, and no Cosmic Eggbeater Registry assigning unique identification numbers. You could try to refer to this eggbeater by a description you hope is unique in the entire universe, but that would require an unwieldy amount of detail. Exactly which features distinguish this eggbeater from every other eggbeater? And how can you be sure? You could check every eggbeater on the planet, but can you be absolutely certain there is no similar one somewhere else in the universe? Unenumerable galaxies potentially containing eggbeaters are potentially relevant.
In this kitchen, here, where you and I are, there is probably only one eggbeater. If I say “pass the eggbeater,” you know which one I mean. This is highly efficient; I don’t have to find a unique description for it, and neither of us has to know its serial number. Plus, the single simple expression “the eggbeater” automatically gets the job done in pretty much any kitchen. I don’t need to figure out a different way of referring to each different eggbeater each time.
Ultimately, all referring has to be relative to me. I’m here, now, and that defines the context. I’m doing this, and that defines the purpose. You are you, not just Julius Quimby Alexis Featherstonehaugh IX, because I’m talking to you now. The eggbeater you hand me is the eggbeater because it’s the one I’m going to use.
So “the” is a referring trick, the simplest one. We are not always so lucky that there is an obvious, unique thing that we might be referring to. What then?
Referring is feasible only because the difficulties posed by unenumerable details of a situation are offset by the richness of the unenumerable tools available in that same situation. We can exploit those for stupid tricks.
Here’s one: referring to an object relative to another one that is easier to refer to.
“That’s a Hudsonian godwit!” says the more experienced birder, gesturing toward a mixed-species group of sandpipers wading in the shallows. “Which?” asks the less experienced one. “Just to the left of that piece of driftwood? … I mean, the log sticking out of the sandbar into the bay?” “Oh! Wow! Yeah, I see the godwit now!”
This took three tries. The less experienced birder couldn’t immediately pick out the exciting individual from the flock of similar waders. But it’s just left of that piece of driftwood. Which piece? The one sticking out of the sandbar.
Let’s go through a bunch of stupid referring tricks quickly. Our job here is just to get a sense of their diversity, not to do proper ethnomethodology, much less science.
Pointing is a common way of referring. I’ve given two examples above already: the foreman nodding “over there” and the birder gesturing at the flock.
Looking is a common way of finding a referent. You can locate and grasp the mouse on your desk without even taking your eyes away from the screen. (Note how you shifted your visual attention to it, though!) If someone tells you there’s water in the refrigerator, you poke around inside to find it.
“Yikes!” I yell. You are playing a video game you are new to, and I’m an expert watching over your shoulder. “Yikes!” refers to the thing that just happened: it was significantly bad, in a way I suspect you don’t recognize. Describing what it was might be complicated, and the time taken would blur the instant. A sharp “Yikes!” refers to right then. It’s like pointing, but in time rather than space.
Names are a common way of referring. Names don’t have to be absolutely unique. You may be the only Julius Quimby Alexis Featherstonehaugh IX, on this planet at least, but there’s plenty of John Smiths, and that only rarely gives them trouble. There are many dogs named “Samantha,” but probably only one in your neighborhood, and that’s usually good enough.
I refer to my car just as “my car”: relative to me, but via possession rather than location. I know which one it is. It’s not likely I’d get it confused with someone else’s, especially because it’s old enough to have some distinctive dents on it.
Still, it has a unique Vehicle Identification Number, mandated by the National Highway Traffic Safety Administration. Labeling is a way of modifying objects to make them easier to refer to. The VIN is stamped on the car in several places. I don’t remember what it is, but if it came to an ownership dispute, I could look it up.
Conversely, it’s wise to ensure that indistinguishable (or nearly indistinguishable) objects are causally equivalent (or sufficiently so). If every machine screw was unique, it would cause a lot of trouble. Ones with different thread pitches had better have pitches that are different enough that it’s easy to tell them apart.
And you’d do well to keep screws of similar sizes in separate containers. More generally, keep everything in its proper place, so you can refer to “the serving dish on the top shelf.”
You can delay locating a referent. If you are told “turn right at the ornate Vietnamese restaurant and then the theater will be a couple blocks down on the left,” you won’t recognize “the restaurant” or “the theater” until you get to them.
You can delegate the ability to locate a referent to someone else. If you remember reading that Ouagadougou is the capital of Burkina Fasso, you know that fact, even though you might not have a clue where you were if you were parachuted into it. “Um, excuse me, where am I?” “Ouagadougou, mais ici on parle français!”
You can delegate referring itself, including to inanimate objects acting as your proxy. Writing instructions is an example.
As with all things reasonable, referring can’t be guaranteed to work. If it fails, as in other routine activity, one can usually repair it. “Pick up the amulet,” I advise, watching over your shoulder as you play the sword-and-sorcery game. And then, as you head in the wrong direction, “No, the other one!”6 I wrongly assumed you knew that one was cursed; I meant the one on the left, not the one on the right.
Cognitive science assumes that beliefs are mental representations. There’s problems with that idea, but let’s go with it for a moment. We don’t know how it would work in any detail, but it seems that, inasmuch as beliefs refer, it must also be by enabling you to locate the referent well enough for the belief to hold. Your belief “Samantha is white” depends on the ability to check, potentially at least. Maybe you’ve actually seen the dog, knew it was Samantha, and saw her color. Or maybe someone has only told you she was white, and you wouldn’t recognize her. But then, as with Ouagadougou, you are relying on someone else’s ability to say which dog it was. And that is fine.
- 1. For further discussion, see Chapters 11 and 12 in Philip E. Agre’s Computation and Human Experience (1997).
- 2. I am referring here to model theory and its attempted application to natural language semantics. Model theory is, indeed, extraordinarily elegant and beautiful; I recommend learning and appreciating it if you have the opportunity. Formal semantics as a subfield of linguistics… not.
- 3. This is a reference to “stupid human tricks,” a segment of comedian David Letterman’s television show, in which amateurs would demonstrate remarkable but dubious skills such as playing tunes on disposable razors or sticking grapes up their nose with their tongue.
- 4. In ethnomethodology, the polite term for stupid human tricks is “ethnomethods”: methods used by a people (ethnos in Greek). Those are the actual topic of ethnomethodology. You might think “ethnomethodology” would be a theory of what technical methods should be used by ethnographers, but it’s not. Ethnomethodology itself has no technical methods; it finds out what ethnomethods people use by any means available.
- 5. This is a major theme in linguistic pragmatics and in ethnomethodology. In both fields it is termed indexicality.
- 6. This was a central example in my PhD thesis book: Vision, Instruction, and Action, 1991.