Difficulties with Indeterminacy
What is the doctrine of the indeterminacy of meaning? It isn’t easy to get a firm handle on it: are we limited to saying, uninformatively, that it is the doctrine that meaning is indeterminate? But what kind of indeterminacy is at issue? Three analogies may be cited: the quantum world, fictional entities, and vagueness. In the quantum case (itself obscure) we are told that particles don’t have definite properties such as position and momentum—there is no fact of the matter about where they are and how fast they are moving. Everything is blurry, multiple, unfixed, merely potential. In the case of fictional entities we must accept that there is no answer to every question we can ask about what their properties are: does Hamlet have a mole on his back? Nothing in the text settles the matter, and Hamlet has no being independently of the text. Fictional characters are gappy, sketchy, only partially formed. In vagueness we have the phenomenon of borderline cases in which no definite classification can be made: no fact of the matter about whether a certain color is red or not, say. These all warrant the adjective “indeterminate”, but it is unclear that they fit the intended import of the indeterminacy of meaning thesis. It isn’t a matter of one determination of meaning rendering another determination impossible, or of ontological incompleteness in meanings, or of the borders between meanings being blurred. The idea seems rather to be that meanings have no fixed identity: they are neither one thing nor the other—they are inherently mushy, a kind of fluid nothingness. They can’t make up their mind about what they are, like ghostly shape shifters, a kind of undifferentiated mass. That, at least, seems to the general idea. The underlying thought is that meanings are inherently formless—considered in themselves they have no determinate being.
But that is not the only way to understand the doctrine, or even the standard way: instead it can be understood as concerning the relation between words and meanings. The meanings in themselves are perfectly determinate, no less so than other things, but what is indeterminate is which meaning a given word expresses. Does “gavagai” express the meaning rabbit or does it express the meaning undetached rabbit part? It is assumed that these two meanings are discrete entities, that each is a well-defined and determinate semantic unit, but what is not determinate is the expression relation. This type of position fits the indeterminacy of reference thesis quite naturally: references, construed as objects, are not inherently indeterminate—they are just ordinary objects—but it is indeterminate what objects words refer to. It is the reference relation that is said to be indeterminate not references themselves. Ditto for meanings, under this interpretation of the doctrine: the realm of meanings (propositions, concepts) is determinate, but it is indeterminate which of these determinate entities a given word expresses (connotes, signifies). Without discussing individual authors, it may be remarked that these two doctrines are not always clearly distinguished. One feels that there is sometimes an unannounced transition from the latter doctrine to the former. In any case, we should clearly distinguish them, and give them different names. We could call the first doctrine “strong indeterminacy” and the second “weak indeterminacy”, but that understates the logical difference between the two doctrines—they are not just on a scale of strength. It would be better to call the first “meaning indeterminacy” and the second “expression indeterminacy”, though these labels are not very descriptive either. The important point is that neither entails the other: expression indeterminacy certainly doesn’t entail meaning indeterminacy, but neither is it true that meaning indeterminacy entails expression indeterminacy (words might determinately express indeterminate meanings—as is the case for vague predicates like “bald”).
The main point I want to make is that expression indeterminacy actually entails meaning determinacy, as standard presentations of it demonstrate. For we need to have clearly defined discrete meanings in order to set up a case in which it is indeterminate which meaning is expressed. For example, we need to assume that we have two meanings in play in order to ask whether it is determinate whether “gavagai” expresses one or the other—as it might be, the meanings associated with “rabbit” and “undetached rabbit part”. If we use italics to indicate reference to meanings, the question at issue is whether “gavagai” expresses the meaning rabbit or the meaning undetached rabbit part. This only makes sense if we have two well-defined determinate meanings to work with; if we don’t, we can’t even raise the question. So the question of the indeterminacy of word meaning can only be discussed by presupposing that meanings themselves have a determinate nature.  The same goes for indeterminacy of reference: in order to ask whether words determinately refer we have to assume a totality of references with well-defined identity conditions—the usual objects that we deal with. If not, the question becomes moot. Thus no drastic ontological thesis about the nature of meanings (or references) can follow from considerations about the semantic relations linking words and things. In other words, meanings are determinate if expressing them is indeterminate. The actual argument for indeterminacy assumes that meanings are determinate entities—which one would never guess from standard expositions. In order to argue for the indeterminacy of meanings as such one would need a quite different line of attack, whose nature I cannot even conjecture. Meanings certainly don’t resemble any of the three areas I mentioned earlier: no Uncertainty Principle, as in physics; no gappiness, as in fictional characters; no blurry borderline cases, as in vagueness. Meaning seems about on a par with the rest of reality as far as intrinsic determinacy is concerned. The meaning of “rabbit”—the thing itself, whatever it may be–is no more indeterminate than rabbits, which is perhaps not surprising. One might even venture to say that meanings are paradigms of clearly defined entities. This is why they are so useful in acts of communication. 
 To put it another way, synonymy relations can’t be indeterminate or else there would be no fact of the matter about whether “rabbit” means the same as “undetached rabbit part”. And they can’t be synonymous as used in the argument for the indeterminacy of “gavagai”: the meanings here have to be distinct in order to constitute alternative assignments of meaning.
 Wouldn’t this be Frege’s position? The realm of sense is a well-defined totality of discrete entities, just like the realm of reference. Thoughts can therefore be passed on without fear of indeterminacy in what is passed on. Human knowledge depends on such clear and distinct thoughts. Thinking is not wallowing in a formless mass of slippery semantic goo. Thought is determinate if anything is. (Of course, behaviorism might not be able to respect that fact, but so much the worse for behaviorism.)