When considering phrases like 'Frege is the author of the Begriffschrift', we can distinguish among the following:
- The mechanism by which the word 'Frege' might attach to a particular object in the world.
- The meaning of the term - - either construed as the mechanism by which reference is established or perhaps as the reference of the term.
- The relation between the mechanism by which reference is established, or perhaps reference, and the truth-conditions of assertive sentences containing the term.
Descriptivist theories of proper names claim a proper name like 'Frege' has an associated description - - which can vary from speaker to speaker and over time - - by which reference is accomplished. The mechanism on the basis of which reference is fixed forms, on the descriptivist theory, the meaning of the term or expression at hand.
Descriptivist views developed in opposition to Millian theories of naming, whereby the name was regarded to have no semantic content. Frege famously argued that 'The Morning Star is the Evening Star' gives informative content. Accordingly, material identity statements seem to cry out for an analysis in terms of descriptive content. 'Morning Star' and 'Evening Star' seem to mean different things, though they have a common reference. If the meaning of a term were simply its reference, then an informative identity statement is seemingly impossible. But it seems that informative identity statements are possible, therefore, by modus tollens, the meaning of a term cannot be its reference.
Or take 'Fred Flintstone does not exist'. If the meaning of 'Fred Flintstone' is its referent, than how can 'Fred Flintstone does not exist' have meaning, for the necessary condition for meaningfulness clearly does not obtain?
Or take 'Janet believes that Melanchthon, but not the author of the 1521 Loci Communes, wrote on rhetoric'. If 'Melanchthon' and 'the author of the 1521 Loci Communes' have the same referent, how can Janet, a supposed rational agent, hold the statement to be true?
On a descriptivist view, these problems seem to dissolve. For instance, it is different semantic content that makes 'The Morning Star is the Evening Star' informative. In 'Fred Flintstone does not exist', 'Fred Flintstone functions as a description that is not satisfied. In 'Janet believes that Melanchthon wrote on rhetoric, but not the author of the 1521 Loci Communes', it is because Janet associates different semantic content to each of the terms that the statement is true.
But Ruth Barcan Marcus, and later Saul Kripke, argue subsequently (and persuasively) that names do not have semantic content after all, and thus are not semantically equivalent to any description. For both thinkers, proper names refer directly without mediation of an associated description. For Kripke, not only proper names, but also definite descriptions and natural kind terms (e.g., sheep) rigidly designate their bearers. Kripke points to three problems with descriptivist views: the epistemic problem, the modal problem, and the semantic problem.
The epistemic problem is the problem of unwanted necessity. Suppose that Bob knows that the Morning Star is the Evening Star because he knows they refer to the same thing. Then to Bob, adjusting his associated descriptions, it would seem that 'Morning Star is Evening Star' is necessary. But clearly this is not necessary, thus a descriptivist account is false.
The modal problem arises when the application of descriptive content to names produces absurdities in counterfactual situations. Imagine that to the word 'Fred Flintstone' we associate the greatest cartoon figure of the 1960s. Then were Fred Flintstone not to have been invented, 'Popeye the sailor lived in Bedrock' would be true - - if the description 'greatest cartoon figure of the 1960s were now satisfied by Popeye the sailor.
The semantic problem emerges when a description is falsely associated with a name. Take, for instance, the description 'the author of the Speculative Grammar'. Tradition has it that Duns Scotus was the author. But it may have been Thomas Bradwardine. Thus, if one were to say 'Duns Scotus died in 1308' and have as the associated description 'the author of the Speculative grammar', then the statement would be false because Thomas Bradwardine died in 1349.
The fact that names have no semantic content is consistent with a causal theory of reference, a theory that allows for a neo-Millian approach to proper names. According to the theory, in a reference-fixing event there is a dubbing of a name to a bearer, a relationship that is then causally transmitted linguistically. 'Aristotle' thus names the individual writing the Metaphysics in this world, while it names the same person in other possible worlds who did not write the Metaphysics. Instead of demanding a criterion of transworld identification that would pick out the same individual in different possible worlds, the proper name rigidly designates the same individual in all possible worlds - - not by virtue of a description, but through causal reference.
How does the descriptivist account fare with respect to the term 'God'? Plausibly, we might claim an associated description of 'God' as the sum total of all positive properties to the infinite degree. Accordingly, 'God' would pick out in each and every world that being that has the available properties and degrees instantiatable in that world. Obviously, a world in which there is neither goodness, love, nor even thought would have a very different being satisfying the description 'God'. To say, 'God could have created a world without goodness' demands that this God is instantiated in possible worlds without goodness, not simply that some being or other fulfills the suitable description in other possible worlds - - some without either matter or thought.
Of course, the idea that 'God' names the same being in all possible worlds in which He exists suggests that there is some ontological contour to God, a contour projectible across possible worlds. While this individual divine essence cannot be wholly specified, the fact that one names this God rather than that one suggests an ontological contour to the divine. More to the point, to say that 'God is good' is to say that the individual referred to by 'God' in all possible worlds in which that individual exists is identical to the instantiation of goodness - - whatever "the instantiation of goodness" quite means. All properties of God are accordingly essential to God, and they apply to God necessarily.
It is critically important to disambiguate logical, conceptual and metaphysical necessity here. While the freedom of God implies that God could have done other than what God did in fact do (and thus that 'God is good' is neither a logical nor conceptual truth), the way that God is constitutes God's essence in this world and in every world where this God is. Given the choice God has made for human beings, God's contour is metaphysically necessary for God to be God. From the standpoint of the reference of 'God', God could still have been other than good, for it is logically possible for God to have done other that what God did. While human beings generally know God as 'aliquid quo nihil maius cogitare possit', in salvation history the reference of 'God' is accordingly fixed such that it is now metaphysically necessary for God to be good. This necessity, however, is intraworldly, it applies to those world's in which the conditional 'God established covenant x' is true.
Another way of saying this is that God is identical to love and goodness in all metaphysically possible worlds, but not so in all logically or conceptually possible worlds. The set of metaphysically necessary worlds is a subset of the set of logically necessary worlds. There are worlds in which God could do x that are not metaphysically accessible by the God who is who He is, and will be who He will be. The question concerns now the "bare particularity" of the divine, the One who in being other than who He is, could have still be somehow still Himself.