‘There is no magic’, part 1: conceptualist semantics

I often hear the complaint that involving God in an explanation for anything that happens in the world is an appeal to ‘magic’ and therefore Not Allowed. In fact, this charge is levelled against any explanation which does not assume metaphysical naturalism. Ray Jackendoff makes this exact move in his 2002 book Foundations of Language: Meaning, Brain, Grammar, Evolution (Oxford: OUP), in the course of giving an outline of what he thinks a semantics of natural language should be like:

I propose to begin from the following surely uncontroversial postulate:

People find sentences (and other entities) meaningful because of something going on in their brains.

That is, we are ultimately interested not in the question: What is meaning? but rather: what makes things meaningful to people? This anchors the enterprise both in the theory of psychology and in ordinary human experience.
A second postulate is:

There is no magic.

That is, we seek a thoroughly naturalistic explanation that ultimately can be embedded in our understanding of the physical world.
Such an explanation comes at a heavy price. (p268)

Now, I could say something about Jackendoff’s first postulate, but for present purposes I want simply to take it as an introduction to the second. Of course, Jackendoff isn’t talking about God – the ‘magic’ charge is one he will come to level at rival theorists in semantics. I want to take a brief look at whether, by his own lights, Jackendoff’s own theory avoids being magical. I hope this will cast some light on the question of whether or not ‘a thoroughly naturalistic explanation that ultimately can be embedded in our understanding of the physical world’ is what we really should be after, or indeed what it would look like. This will take two or three posts; I haven’t decided yet.

Jackendoff thinks that realist views of language end up relying on ‘magic’:

Frege and much of the tradition following him take language to be independent of its human users: it relates directly to the world […] language is indeed “out in the world” and it refers to “objects in the world”; but people use language by virtue of their grasp of it, where “grasp” is a transparent metaphor for the “the mind holding/understanding/making contact with” something in the world. […]

One might interpret Katz’s program this way. he is personally interested only in the part of language that is an abstract object “in the world” […] But an abstract object by definition has no physical manifestations that can impinge on the nervous system. So how does the nervous system “grasp” them? Without a careful exegesis of the term – which no-one provides – we are ineluctably led toward a quasi-mystical interpretation of “grasping,” a scientific dead end. (pp296-9)

So much for those thinkers. Anyway, this is Jackendoff’s own view of word meaning:

Linguistic semantics per se is the study of the interface between conceptualization and linguistic form (phonology and syntax) […] In particular, lexical semantics studies the organizations of conceptualization that can be bundled up in a single word (or to be clearer, in an interface rule whose other end is a morpheme). (p293)

I’ll unpack that a bit. The idea is that a word has a particular meaning – for an individual, remember – by virtue of the fact that the word groups together particular concepts in the individual’s mind. So, what is a concept? This is a fraught question, but I’ll try to give as untendentious a characterisation as possible. Very roughly, a concept is a way of dividing up experience. That is to say, it’s because I have the concept BOOK¹ that I’m able to categorise items around me into books and non-books, and because I have the concept READ that I’m able to categorise actions into those of reading and those of not-reading (or, more accurately, that I’m able to categorise anything into an action of reading or not-an-action-of-reading), and so on. Which is not to say that the concepts themselves just are abilities. Some people say this, but I think that Jackendoff has in mind, rather, that concepts are parts of thoughts. This should become clearer as we go on.

OK, but what is it that gives concepts meaning, then? How can a concept be about books, or reading, or Ray Jackendoff, or natural language semantics? Concepts are in the mind, and these things are out in the world. What is the nature of the connection between them? Jackendoff doesn’t think much of Jerry Fodor’s suggestions for dealing with the problem:

For [Fodor], language is a mental faculty that accesses concepts […] In turn, concepts have a semantics; they are connected to the world by virtue of being “intentional.” The trouble […] is that one cannot make naturalistic sense of intentionality. If suffers from precisely the same difficulty as “grasping” language […]: there is no physically reliable causal connection between concepts and objects. (p300)²

So much the worse for naturalism, you might say. But Jackendoff doesn’t. Instead, he proceeds by ‘pushing “the world” down into the mind of the language-user too, right along with language’ (p303). His solution is that, actually, not only concepts, but also what concepts are about, are in the mind. In fact, concepts just are what they are about:

We must explicitly deny that conceptual structures are symbols or representations of anything in the world, that they mean anything. Rather, we want to say that they are meaning […] Language is meaningful, then, because it connects to conceptual structures. (p306)

Such an explanation comes at a heavy price, all right. What Jackendoff is saying is that the concept BOOK can be about books because books are in the mind – because the book I see ‘in front of me’ is, in fact, a set of sense impressions that are entirely mental. But without some additional qualification this looks like idealism, which I hardly think comports with ‘our understanding of the physical world’, which I’m quite sure Jackendoff takes to be, well, physicalist. Idealism, by his lights, would be ‘magic’.

Jackendoff doesn’t quite go down this route, although at points he seems to be toying with it³. Instead, the trick is to be found in the construction of concepts. That’s what I’ll focus on next time.

Notes
1 It's conventional to give the names of concepts in capital letters. I'm going to follow this convention.
2 The disagreements between Jackendoff and Fodor as to the nature of concepts, and their relation to word meaning, go much deeper than I have time, space or energy for here, although I may return to this issue in the next post. See for example Fodor’s 1998 book Concepts: Where Cognitive Science Went Wrong (Oxford: OUP), pp49-56.
3 See the remarks on Berkeley on p305, and on reality ‘independent of human cognition’ on p309.