Introduction: The Myth of Human Language

1.2 Core Ideas

So far in this section I have been largely concerned with comparing and contrasting my theory of the dynamic lexicon with certain standard works in the philosophy of language. It is perhaps time to now clearly lay out the central doctrines.

Meaning underdetermination.
In the introduction I gave the example of someone asking me how many books I had written, and me responding by asking what they meant by ‘book’. I had to ask the question because the meaning of ‘book’ is underdetermined. Nothing in our broadly shared understanding of the meaning of ‘book’ was sufficient for me to answer the question. Thus we had to modulate the meaning to the point where I could answer the question. As I also noted in the introduction, I think this is a pervasive fact about the words we use: They are all underdetermined to some extent or another.
It is important to see that this isn’t the same thing as vagueness. As I’ll argue in section 1.3 I think that vagueness is just a special case of meaning underdetermination where the underdetermination is along a scalar dimension. As we will also see in 1.3, meaning underdetermination is distinct from meaning underspecification and meaning indeterminacy.
Representational Austerity and Representational Neutrality
We need to be careful to distinguish between word meanings and the way that those word meanings are represented. It is entirely possible (probable, I should think) that the meaning of a word is not exhausted by the way in which we represent it. How much is represented? By some accounts, not very much. A good case can be made that the part of the meaning explicitly represented in lexical items consists of just hints and clues (like one finds in dictionary entries) that may help us to deploy cognitive resources (typically analogical reasoning) to flesh out the word meanings, and the way we flesh them out will vary according to contexts and social settings. A similar point has be made by Rayo (forthcoming), who talks about a “grab bag” theory in which we rely on a variety of resources to work out word meanings. As Putnam has stressed, elements of meaning could depend upon other members of our linguistic community and likewise facts about the world. Let’s call the view that representations of meaning underspecify meanings Representational Austerity.
As intuitive as Representational Austerity may seem, I am not climbing on board with it here because I am suspicious of there being a clear idea of what representations look like and how austere or robust they are. One idea would be that a representation is a data structure – a semi-stable, localizable structure in the computational state realized by an agent (person or machine, for example). But as I argued in Ludlow (2011; Chapter 2) I seriously doubt that there is a brute fact about systems individuated narrowly that determines what computational states they realize. Furthermore, whatever states they realize and however those states are individuated, the connection between those states and what they represent is largely mysterious. There is no reason to think that the data structure has to be isomorphic to what is represented in order for us to represent something (a meaning, for example). So what then is the connection? There is hopefully an answer to the question, but the doctrine of Representational Neutrality says that the theory of the Dynamic Lexicon is neutral on the matter.
Accordingly, in what follows, I’m not going to have much to say about how meanings are represented. Instead, I will talk directly about meanings, whether they can be broadened and narrowed, whether they are fully determinate, whether we can control them, and ways in which we make them explicit without talking about the role of representation in this. This is not to say that meanings aren’t represented or that the representation of meaning is not part of a full story of meaning; it is just to say that the theory I am proposing here is neutral on all of these matters.
This claim of neutrality may seem anomalous when we begin talking about, for example, the thematic structure of words (e.g. meaning facts involving roles like agent, patient, theme), but the anomaly is only apparent. One can perfectly well subscribe to these thematic relations without taking a stand on how these roles are to be represented (or even on whether they are represented at all).
Meaning Egalitarianism.
Sometimes people argue that there is a primary or privileged meaning for a term. It is difficult to see what this might be in the case of ‘book’, but in other cases the intent is clear enough. The privileged/primary meaning of ‘flat’ would be a meaning like absolutely flat. The privileged/primary meaning of ‘know’ would be something like knowing without any possibility of error.
Sometimes people argue that in expressions like ‘six foot six’ or ‘3:00 o’clock’ have privileged meanings that are their “on the nose” meanings. When we say ‘we will be there at 3:00’, there is a privileged/primary meaning of that expression which means precisely 3:00 o’clock down to the nanosecond. We may be entitled to diverge from that meaning, but the core meaning is the absolute or on the nose meaning, and any divergence from that core meaning is just that – divergent.
Meaning egalitarianism is the doctrine that there are no privileged/primary meanings – no “absolute” senses like absolutely flat or “on the nose” senses like precisely six foot six. The absolute and on the nose senses are simply two modulations among many others that are, at the outset, equals. The absolute and on-the-nose senses do not come first, and they need not be the goal of a proper lexical modulation. This does not mean that all modulations are equally good. As we sill see, there are norms governing how we modulate word meanings, and the best modulations will turn on our interests and the important properties that anchor our interests. They will not turn on the approximation of some absolute or on-the-nose meaning. In fact, such meanings may not even exist, which leads to my next doctrine.
Meaning Imperfection.
Some people believe that the core meaning of a term like ‘flat’ is a sense like absolutely flat and that the core meaning of an expression like ‘six foot six’ or ‘3:00 on the nose’ are precise spatial lengths and temporal locations. As I said, meaning egalitarianism rejects the idea that these are the core meanings, but the doctrine of meaning imperfection suggests that such meanings may not exist at all. Is there a coherent notion of absolute flatness or exactly six foot six in mathematics or physics? There is certainly reason to doubt it.
Just to illustrate, consider a predicate like ‘flat’. Obviously there are no surfaces that are absolutely flat in the physical world (presumably everything is bumpy at the micro level), but it isn’t clear that there is a stable notion of flatness in the mathematical realm. We might think, for example, that a two dimensional plane is absolutely flat, but what then happens when the plane is in a non-Euclidian geometry. In some such geometries a plane can be quite “bumpy.” Of course we could stipulate that a plane in a particular axiomatic system like Euclid’s is flat, but this misses the point. The question is whether there is a notion of absolute flatness (or absolute anything) that is stable across contexts. As we will see, this is doubtful.
Meaning Control.
The doctrine of meaning control says that we (and our conversational partners) in principle have control over what our words mean. The meanings are not fixed by convention, nor by our conversational setting alone. If our conversational partners are willing to go with us, we can modulate word meanings as we see fit. This isn’t the Humpty Dumpty theory of word meaning because Humpty needs Alice to play along. If she doesn’t play along, the word meanings are not modulated as Humpty wishes. It does not follow, however, that there isn’t a right way and a wrong way to modulate word meanings. There are still norms that govern this process of modulation. We will take up these norms in Chapter 4.
Meaning Control does not exclude the possibility of externalism about content – either environmental externalism or social externalism as in Burge (1979) – nor does it preclude the possibility of a division of linguistic labor as in Putnam (1975). The idea is that it is within our control to defer to others on elements of the meaning of our words (for example a doctor on the meaning of ‘arthritis’ and a botanist on the referents of ‘beech’ and ‘elm’) and it is also within our control to be receptive to discoveries about the underlying physical structure of the things we refer to (for example the discovery that ‘water’ refers to H20 and not XYZ). The theory of the dynamic lexicon is largely neutral on theories of content; my point here is simply that the dynamic lexicon is compatible with existing theories that employ some version of externalism about content. And conveniently so, since I happen to be an externalist about content.
Concepts as Linguistic Meanings
Clark (1992) talks about the result of lexical entrainment as being a “conceptual pact.” I rather like this idea. The basic thrust of it is that concepts are not things floating in Plato’s heaven or etched in some data structure of our mind/brains; rather they are simply word meanings. That by itself may sound like a standard thing to say (one could reverse the equivalence and argue that word meanings are objects in Plato’s heaven or are things etched in data structures in the mind/brain after all), but when it is linked with the idea that we collaborate to modulate word meanings it suggests that concepts are the sorts of things that we build together on the fly. I won’t be defending this thesis in the book, but it is a doctrine worth some reflection along the way: Concepts are underdetermined, modulable, and often the product of collaborative effort.