On Naming

On Naming

According to the classic description theory, names are synonymous with definite descriptions; they are said to abbreviate such descriptions, to be “short for” them. The two are therefore intersubstitutable. The description is said to “analyze” the name—spell out its meaning. Thus, there can’t be a language with names but no descriptions, and names work in the same way as descriptions. In principle, names are eliminable in favor of descriptions. But this synonymy claim raises an obvious question: if the two are synonymous, then aren’t descriptions synonymous with names—don’t they mean what names mean? Whenever the description occurs, we should be able to substitute the name and preserve meaning. In fact, couldn’t we construct a language in which all descriptions have associated names synonymous with them? Then we should be able to replace descriptions with names in all occurrences. The trouble is that synonymy is symmetrical, but the description theory is intended to provide an asymmetrical analysis in which descriptions are taken as primary and basic. We don’t want to end up saying that descriptions are analyzable by means of names—indeed, that they arenames. But what is to stop us saying that descriptions are just “long for” names? What if people spoke a language in which from childhood everything referred to is named but never definitely described (no expressions of the form “the F”), and then descriptions are added later? Wouldn’t these speakers naturally view the later descriptions as analyzable by means of names? For this is what goes through their minds when definite descriptions are used.

What is the right thing to say here? The first thing to say is that names and descriptions have a different function and play different linguistic roles. The name labels, the description characterizes—so the two are not semantically equivalent (strictly synonymous). The dictionary (OED) is helpful here: a name is defined as “a word or set of words by which someone or something is known, addressed, or referred to”; a description is defined as “a spoken or written account of a person, object, or event”. These are quite different concepts: descriptions are not words by which a thing is known or addressed, and names don’t provide accounts of things. The two types of expression do different jobs, answer to different needs. In fact, it is a misnomer to call definite descriptions “descriptions”—they don’t provide “accounts” of people and things. It would be better to call them singular predicative phrases; then the “description theory” would be claiming, implausibly, that words by which people and things are known or addressed are semantically indistinguishable from singular predicative phrases. A person known as “Socrates” would be defined as someone an account of whom would include the fact that he taught Plato. That sounds funny at best. The truth is that the naming relation and the describing relation are different relations; so, the corresponding phrases belong to different linguistic categories. One cannot be reduced to the other. They are not synonyms. Maybe they share sense and reference, but that doesn’t make them instances of the same linguistic type.

That objection may seem pedantic, philosophically, though it is suggestive. What about the issue of dependency—do names depend on descriptions (so-called) but not vice versa? Names certainly don’t depend on non-indexical descriptions; they can be introduced by means of suitable demonstratives. But more to the point descriptions typically work by employing names, either proper names or names of properties. We say “the capital of France” and we name the properties that things have (“red”, “capital”, “electricity”). So, descriptions depend on names not vice versa. We can’t find an asymmetry that way. What about the idea that descriptions are psychologically more basic, simpler, clearer? What if an avid Russellian went around speaking Russell-ese—using Russell’s analysis of descriptions everywhere other people use names and unanalyzed descriptions? Someone unfamiliar with this apparatus might not know what the hell he is talking about; a friend might interject saying, “He means Charles is a bit of a twerp”. Then why didn’t he just say that instead of all the rigmarole about “There is an x such that blah blah blah”? The description theory is not exactly good social psychology. Then what is it exactly? Why all the stuff about analysis, abbreviation, basicness? There is clearly a semantic relationship between names and descriptions, but that is a far cry from the rhetoric deployed in formulating the so-called description theory of names. Why not speak of the name theory of descriptions based on the same data? Indeed, descriptions are more name-like and name-involving than names are description-like and description-involving. Names don’t look and function like singular predicative phrases; if anything, these phrases are syntactically name-like (Frege’s actual theory). Among family members names are routinely employed in ignorance of the descriptions known by the general public; it would be very strange for Einstein’s relatives to think of little Alfred as “the inventor of relativity theory”. For them the name is far more salient than the description. In precisely what sense are names “less basic” than descriptions? The fundamental problem is that the description theory simply helps itself to an asymmetry claim while starting from a claim of symmetrical synonymy. That claim is dubious to begin with, but the further claim looks unwarranted. The two categories of expression are interrelated and share some semantic features, but the similarity doesn’t go much deeper than that. The idea that names are nothing but descriptions looks like an exaggeration, equally matched by the idea that descriptions are nothing but names. It is a case of over-assimilation in both directions.[1]

[1] The abbreviation claim is both implausible and necessary to the description theory. A genuine abbreviation literally shortens a word or phrase, as with “Tom” for “Thomas” and “Sue” for “Susan”, but names don’t abbreviate descriptions in that way, or in any way; they are quite different words. But the claim is necessary because without it the objection would be that names are nothing like the descriptions supposed to define them. An abbreviation of “the capital of France” would be something like “the cap of Fra”, but “Paris” is nothing like that. The description theory is really a highly revisionary analysis of names, implausible on its face. We have all the counterexamples that have been brought against it, but it is also methodologically flawed. You can’t infer semantic identity from semantic similarity. I suspect Russell, in particular, was moved to this inference by his anti-substance metaphysics and love of sense-data.