Chapters 1 and 2 discussed the idea that meaning = reference, either in general or at least for definite descriptions. Lycan gave reasons to reject that view, but although meaning and reference can't be identified with one another, both seem to be important aspects of language: descriptions and names do refer to things in the world, and linguistics expressions in general do have meaning.
So what is meaning? What must a theory of meaning account for? The word "meaning" seems to get used for two things that may or may not go together: (1) something that determines reference; (2) something that must be in the mind of a person who understands an expression.
(See pp. 65-66.)
1. Meaningfulness. There's a difference (to be explained) between things with meaning and things without.
2. Synonymy. Some pairs of expressions are synonymous (i.e. mean the same thing).
3. Ambiguity. Some expressions mean more than one thing.
4. Containment, entailment. Some meanings seem to "contain" other meanings, e.g. the meaning of "bachelor" contains or includes the meaning of "male."
Does understanding and explaining these facts require "reifying" meanings, i.e. treating meanings as things of some sort? Not automatically. (Analogy: it's a fact that we sometimes do things for other people's sakes -- e.g. I might try to be a good parent for my children's sake. But it doesn't necessarily follow that there are things called "sakes.") On the other hand, it does seem that the simplest and perhaps most natural way to account for things like meaningfulness and synonymy is by supposing that there are things called "meanings": being meaningful is having (at least) one of them; for two expressions to be synonymous is for them to have the same one; for an expression to be ambiguous is for it to have more than one, etc.
Lycan calls this sort of view an "entity theory" of meaning: a view according to which meanings are entities of some sort. What sort? Lycan considers two possibilities: ideational theories and propositional theories. (We'll see a more sophisticated form of the propositional theory in chapter 10.)
One thing to keep in mind in Lycan's discussion of ideational theories is the distinction between tokens and types. A token is a particular, concrete object located at a particular place and time. A type is a general category that can have many tokens (e.g. the type 'circle') or none (e.g. the type 'round square'). Consider the following word on your monitor or sheet of paper: 'balloon'. How many letters does it contain? That depends on whether you are counting types or tokens: there are seven letter tokens, but only five letter types.
One thing that could make Lycan's discussion a little confusing is that he uses the term 'idea' only for idea tokens. In ordinary English, I think the word 'idea' can refer to either a token or a type. (For instance, the title of Robert Paul Wolff's book The Idea of the University presumably is meant to refer to an idea type that many people can share, not the token idea possessed by a particular person at a particular time.)
Locke: (a) meanings are ideas; (b) ideas are (or are very like) images. Problems:
1. (67) Images are too specific and detailed to be meanings for most expressions. Any image of a dog will be of a specific breed; any image of a triangle will have specific angles; etc. (Why Berkeley denied that there are abstract ideas.)
2. (67) There are no images associated with many words: 'is', 'and', 'of', 'chiliagon', 'nonentity'. (Not that you couldn't associate images with these words, but the pretty clearly wouldn't have anything much to do with the meaning of the words. As Descartes observed, you might have a sort of vague or fuzzy image of a chiliagon, but this wouldn't be any different from the image you have of a thousand-and-one-sided figure. Nabokov in his childhood apparently experienced a different color for each letter when he was reading; this would have given him vivid images for every word, but when he lost this trait later in life, words didn't lose their meanings.)
3. (68) Linguistic meanings must be public, shared (because language is); images are private, may differ from one person to the next.
4. (68) Some meanings do not correspond to any actual mental entity. (This is perhaps Lycan's most obscure objection. The idea is that there are lots of sentences that never have been or will be actually expressed or thought of. Therefore no token idea or image corresponds to them. But those sentences have meanings too. So it can't be necessary that meanings are actual (token) ideas. They could still be idea types. And that leads to the next account, the propositional theory.
According to the propositional theory, sentence meanings are propositions. These are abstract entities. (An analogy: numbers are abstract entities which are useful for measuring things. Similarly, propositions are abstract entities which are useful for "measuring" states of people's minds and contents of what they say.) They have internal structure: propositions are composed of abstract parts, concepts, which are the meanings of subsentential expressions. Problems:
1. (71) If propositions exist, they are weird entities. (They are not in space and time, they have no causal properties, etc.) Perhaps that is a reason to think that they do not exist. (But maybe they are not weirder than necessary? And of course one could raise the same objections to numbers, which seem very useful.)
2. (72) "alien to our experience." Reply 1: not at all! Whenever you understand what a declarative sentence means, you have the experience of grasping a proposition. Reply 2: even if we don't directly experience propositions, they might be explanatorily useful. Compare electrons, numbers, etc.
3. (73) Propositions do not really explain anything; they just present the data to be explained in a different terminology. (Lycan attributes this objection to Gilbert Harman.) Reply [not exactly Lycan's reply]: we should probably just admit that as so far presented, this is true; as so far presented there isn't really much of a theory here. We need to elaborate and refine the theory before it will actually explain anything. Chapters 9 and 10 essentially give more developed versions of the proposition account.
4. (73) Meanings have causal powers, but abstract entities like propositions don't. Reply [not exactly Lycan's reply]: It's not propositions themselves that have causal powers; rather, it is people believing, communicating, desiring, etc. those propositions. In this regard they're just like numbers: numbers don't cause anything, but lots of things are caused by things having a certain temperature, or a certain pressure, or a certain atomic mass, or whatever. Abstract entities (propositions or numbers) don't directly cause anything, but we use them to measure states that do cause things (like beliefs or temperatures).