Language, Representation, and Action

This was supposed to be a post where I made an argument for the Philosophical Investigations’ approach to linguistic meaning. The elements I intended to include in that argument are mostly here.

I suppose the target might be the question whether it is coherent or required to hold that language either (1) always really contains or (2) is always best represented as containing a quasi-autonomous ‘propositional’ or ‘descriptive’ or ‘factive’ element alongside all the others.

But why would anyone believe this? Don’t nursery rhymes refute the view straight away? Or is that not language?

Here’s an imaginary argument:

A: In a large number of cases, the meaning of a word is its use in the language – the way the word is woven into our actions and forms of life.

B: Well, I’m not sure what you mean by ‘woven’ here, or what a ‘form of life’ is, but I get that you think we don’t fully understand language unless we understand commands and self-expression, for example, as well as descriptions, direct reference, and so on. And I entirely agree! But it’s clear that commands and expressive uses of language are in some sense parasitic on descriptive ones. You can’t for example tell someone to do something without in some sense having represented (described) the state of affairs that obtains before and after he does it, and you can’t express your own feelings without an antecedent grasp of how the words you use for your expression describe various states of mind – otherwise how would you know whether they were an appropriate expression of what you were feeling? So this is why it’s quite correct to treat assertion, description, propositional discourse – say, ‘representational’ uses of language – as a fundamental and autonomous subsystem within language, and then, if we want to talk about expression, command, ethics, and so on, to introduce various operators to mark the various modalities employed in these different uses.

Does A have a good response to B? B’s position does not seem totally insane to me, although right off the cuff I have an objection to the treatment of expressive uses of language suggested in B’s reply. (David Finkelstein has published some good thoughts on these issues.)

We might defend B by saying: isn’t what we want a representation of different kinds of use of language? And to represent a use of language don’t we have to tie it into the world – understanding that we ourselves are part of the world – in the right sort of way? And isn’t the use of language most fundamentally connected to the world something like assertion or description?

“When grown-ups named some object and at the same time turned towards it, I perceived this, and I grasped that the thing was signified by the sound they uttered, since they meant to point it out…” In Augustine’s picture we start out as little scientists, forming hypotheses on the basis of our (antecedently possessed) abilities to individuate things and recognize human feelings and desires to figure out what things the words mean, and then once we know that we employ the words to get our own needs met.

No doubt there are many things we do, and many things our brains do, that fit Augustine’s picture tolerably well. 

I remember a period in my early childhood when I had no sense of humor. I enjoyed being tickled and hugged and figuring things out, but jokes made no sense to me. I studied what made other people laugh and eventually got fairly good at provoking laughter. Then, at some point, a switch flipped inside me, I ‘went native’, and now I laugh at things myself. (I never realized how thankful I was for this until typing those words just now.) But anyway it was something in my life that followed Augustine’s pattern – almost.

The thing that doesn’t fit the picture is acquiring a sense of humor myself – learning to participate in the ‘form of life’ of joke-telling. Some sociopaths can tell jokes but don’t find them funny themselves – though they can give the outward signs of finding things funny when they need to. They pass the Turing test, as it were.

My example here is clumsy. What it’s supposed to be an example of is a case where the very structure of my thoughts and feelings was transformed by being introduced into a language-game, that of joke-telling, rather than a case where I more or less already knew what I felt and thought and I only needed to learn the right words and/or social rituals to communicate those things to others.

Augustine’s famous quotation seems to assimilate all language-learning to the second kind of case. There are cases like that. You’re working as, say, a mechanical engineer, and you need a circuit that will control the motion of some moving part in your machine. You know what you need it to do and how much force it needs to exert, so you go talk to your electrical engineer colleague. You might need to learn a couple of basic terms of electrical engineering so you can clearly indicate what you want to her, and vice versa, but here things really do work more or less the way Augustine says.

But I don’t believe they work like that all the way down. Although there may be – perhaps must be – exceptions – there are also cases where we only learn what to look for, what to feel, what to want, from the language-games and forms of life we’re trained in.

Where what we think and say emerges out of what we have learned to do. And being trained to speak and think are also often matters of being told what to do – learning calculus gives you a new way of thinking about certain things in algebra, geometry, and mechanics, but part of what learning calculus amounts to is being trained in techniques – finding limits and derivatives, integrating, etc. Learning to do those things and doing them is bedrock for seeing the world or thinking about particular problems in terms of calculus.

With these remarks as a prologue, let’s take up the question of what sort of response A could make to B.

“B,” A might say, “you everywhere interpret our language and thought as a picturing of facts. On your view we are always basically and fundamentally involved in fact-picturing (referring to objects, describing objects, proposing states of affairs). But this is at best a very peripheral use of language. The more ordinary uses of language are tied primarily into social activity and communication.

“Let’s get really concrete about this for a second. Imagine a construction site inhabited by a primitive tribe. They use only a few words – ‘slab,’ ‘block,’ ‘pillar,’ ‘beam.’ These words only have one use: the foreman shouts them out to a tribesman – say, ‘slab’ – and then the tribesman shouted to picks up the nearest slab and brings it to the foreman, who arranges it as he sees fit.

“The linguistic item ‘slab’ has a clear meaning here, and the meaning is something you do – locate the nearest object of such-and-such a type and bring it to me. The sign is correlated with a form of activity. Young tribesmen are cuffed in the head when they don’t pick up a slab, or pick up something which is not a slab, and given stew when they bring a slab to the foreman – in this way they learn what the word means. Just as in calculus class we give good grades to the students who learn to perform integration properly and bad grades to those who don’t. ‘Find the integral of…,” like “slab,” is an instruction.”

Now B finds this response sort of interesting, but he’s not really persuaded by it. “Well,” B might say, “I see your point. But look – isn’t it clear that the tribesmen can’t so much as find a slab to pick up unless they can individuate slabs and distinguish them from pillars and so on? And so don’t they have to be able to represent their environment and the things around them in certain ways even to use this very simple form of language? Not to mention the general idea of objects having spatial location and being movable. So I don’t see how you get out of the need for pictures of the facts here.”

What is it to represent an environment? Check this out:

http://www.wired.com/wiredscience/2013/12/amplituhedron-jewel-quantum-physics/

We can entertain the idea that the reality underlying a lot of the physical happenings around us is best represented by the volume of some higher-dimensional geometrical structure. Why not? Or maybe that’s completely wrong, too. The point is this – if we want to say that in order to respond to ‘slab’ we have to represent our environment, there’s a sense in which I don’t disagree. But what is this representation? Is Slab Man in any sense calculating the amplituhedron of the various bits of slab in order to bring it to the foreman? I mean, if that’s what’s really going on here, that’s what he’s really doing, right? If his thought is a picture of the facts, mustn’t it be a picture of the real facts, all the way down, and not just a picture of some arbitrary cross-section of those facts? Otherwise, in what sense is he representing the real slab he is lifting and moving around?

Hold up there a moment. Slab Man isn’t calculating amplituhedrons. But it’s not an arbitrary cross-section of the facts that his representation picks out, either. It’s the cross-section of the facts he needs to make the foreman happy in participating in the constructive process which defines the life of his tribe.

(The physical theory is a piece of language and a representation within a system too, of course. I guess the example is intended as a kind of reductio-like stimulus to thought.)

So – we do things. We encounter the world through the things we do and fail to do. (In the wrong quarry “slab” could pick out a stone no-one could lift.) And the organization of our activity, our form of life, is the main (but not the only) source both of our individuating objects and of the meaning of our words.

But it’s not the case that the activity or the form of life exists first and then the language is built up around it, either. Teaching people what to do and teaching people what to say is part of training them in an activity. This is why I keep using the ‘woven into’ phrase. Language is an aspect of our social organization.

From A’s point of view B seems to think we need something we don’t actually need. That something is an explicitly picturing-descriptive thought or element of meaning which somehow yokes our actual forms of speaking and acting to reality. From A’s point of view B is a sort of Semantic Representational Internalist, for whom in order for words to have meaning in they must be connected to some sort of representation which in turn connects those words to the world in the right sort of way.

But what do you need that for? How a speaker/thinker/doer represents the world and is connected to it shows itself in the kinds of things that speaker/thinker/doer says, thinks, and does.

“Well, if a language-user doesn’t have such an internal representation, how do you know that his or her words have any connection to reality at all?” Picking up and moving that big slab of rock wasn’t enough?

One persistent theme in the Investigations is that what isn’t actually in mind when a person does something – isn’t actually in mind. Wittgenstein doesn’t deny, for instance, that you might imagine that a chasm opened up in front of your door every day before you left the house, or that one could worry about such a thing, or even that it could happen. But such doubts are not relevant to going out the door in the normal case – and there’s no sense in which a person who does not have them is overlooking something.

Likewise with the remarks on reading: we can slow down our thought and notice some of the mechanisms of subvocalization and mental association that occur during that process. And maybe processes structurally similar to these even happen in the brain, ‘subconsciously,’ when we do not notice them! But in normal reading such mental processes are not relevant to what you are doing. (Even if they make it physically possible for you to do it.) Having received the proper training and practice – you read.

There are cases where doubt arises and cases where it doesn’t; cases where the fine texture of experience is deeply relevant to what we are doing and cases where it is not. Call ‘everyday’ cases, ‘workaday’ cases, the ones where doubt and fine texture alike are off to one side. Utterances are every bit as linguistically meaningful in the everyday cases as elsewhere, if not in many ways more so. You know what to do with them; they have a use in the language; and so on.

So the objection is not to the idea that representing is going on. It’s to the picture where the representation comes first and the action second. There’s a temptation to want to reply in kind (Im Anfang war die Tat). But mightn’t it be better to say that what we call representing and what we call doing are of a piece, that representing is itself a kind of doing and that even independent practices of representing tie into what we do in various ways, and likewise that what carves out pieces of this immensely complex reality for us is largely what we’ve learned to do with them?

So yeah, sure, Slab Man has a representation of reality. You can see that by the way he picks out slabs as opposed to pillars when he hears “slab.” And that representation even occurs ‘within a certain overall theory of reality’ – you know, slabs as opposed to blocks or pillars. Was there a problem here?

There is a standard analysis/grammar of action, often referred to as ‘folk psychology’, where beliefs (about the environment, and about which actions undertaken within that environment will satisfy one’s desires) and desires combine with an agent’s environment to yield actions. If this analysis/grammar is accepted/in place, it is a necessary truth that acting involves representing one’s environment. But our primary criteria for attributing beliefs and desires to people are the things they do and say.

So it doesn’t seem unreasonable to look at meaning primarily from the point of view of organizing action – the things we do with words – and to read off representation and value in the first place from that.

To think that there is something special or separate or initial or independent about the representational/belief component of this is to ignore that the necessary connections go both ways. (And that they are necessary connections, not psychological or physical ones.)

A man runs into a burning building and saves a young child. “I just did what anyone would do” – that is, it didn’t occur to him to do otherwise. He heard the yelling, ran into the building, picked up the child and fled. The child could have died! He didn’t stop to consider the danger to him of the flames and creaking floorboards, or whether he ought to wait for a professional.

Was his representation of the scenario inadequate? In different contexts I would give different answers. But it is not inadequate simply because he does not fully model the situation to himself before going in – even though in some cases doing that might be called for, there are others where it is clearly not, and where even hesitating to represent the situation might be counted against him. It is clear to me, though, that there are cases in which his representation of the situation was perfectly adequate – even if he didn’t notice that the building was on fire at all.

What else does B want?