There is an attractive and commonsensical idea of what meaning is -- so attractive that most of us think of it by the time we are ten or eleven years old.  The idea is that linguistic expressions have the meanings they do because they stand for things; what they mean is what they stand for.  On this view, words are like labels; they are symbols that represent, designate, name, denote or refer to items in the world: the name “Adolf Hitler” denotes (the person) Hitler; the noun “dog” refers to dogs, as do the French “chien” and the German “Hund.”  The sentence “The cat sat on the mat” represents some cat’s sitting on some mat, presumably in virtue of “The cat”‘s designating that cat, “the mat”‘s designating the mat in question, and “sat on” denoting (if you like) the relation of sitting-on.  Sentences thus mirror the states of affairs they describe, and that is how they get to mean those things.  For the most part, of course, words are arbitrarily associated with the things they refer to; someone simply decided that Hitler was to be called “Adolf,” and the inscription or sound “dog” could have been used to mean anything.
     This Referential theory of linguistic meaning would explain the significance of all expressions in terms of their having been conventionally associated with things or states of affairs in the world, and it would explain a human being’s understanding a sentence in terms of that person’s knowing what the sentence’s component words refer to.  It is a natural and appealing view.  Indeed it may seem obviously correct, at least so far as it goes.  And one would have a hard time denying that reference or naming is our cleanest-cut and most familiar relation between a word and the world.  Yet when examined, the Referential theory very soon runs into serious objections.
 
     Objection 1.  Not every word does name or denote any actual object.
     First, there are the “names” of nonexistent items like Pegasus or the Easter Bunny.  “Pegasus” does not denote anything, because there is in reality no winged horse for it to denote.  Or consider pronouns of quantification, as in
 
                     (5) I saw nobody.
 
It would be a tired joke to take “nobody” as a name and respond, “You must have very good eyesight, then.”  (Lewis Carroll: “Who did you pass on the road?”.... “Nobody[.]”...  “...So of course nobody walks slower than you.”<1>   And e.e. cummings’ poem, “Anyone lived in a pretty how town,”<2>  makes little sense to the reader until s/he figures out that cummings is perversely using expressions like “anyone” and “no one” as names of individual persons.)
     Second, consider a simple subject-predicate sentence:
 
                     (6) Ralph is fat.
 
Though “Ralph” may name a person, what does “fat” name or denote?  Not an individual.  Certainly it does not name Ralph, but describes or characterizes him (fairly or no).
     We might suggest that “fat” denotes something abstract; for example, it and other adjectives might be said to refer to qualities (or “properties,” “attributes,” “features,” “characteristics,” etc.) of things.  “Fat” might be said to name fatness in the abstract, or as Plato would have called it, The Fat Itself.  Perhaps what (6) says is that Ralph has or exemplifies or is an instance of the quality fatness.  But that suggestion leaves the copula “is” untreated.  If we try to think of subject-predicate meaning as a matter of concatenating the name of a property with the name of an individual, we would need a second abstract entity for the “is” to stand for, say the relation of “having,” as in the individual’s having the property.  But then we would need a third abstract entity to relate that relation to the original individual and property, and so on -- and on, and on, forever and ever.  (The infinite regress here was pointed out by F.H. Bradley 1930:17-18.)
     Third, there are words that grammatically are nouns but do not, intuitively, name either individual things or kinds of things -- not even nonexistent “things” or abstract items such as qualities.  Quine (1960) gives the examples of “sake,” “behalf,” and “dint.”  One sometimes does something for someone else’s sake or on that person’s behalf, but not as if a sake or a behalf were a kind of object the beneficiary carried around on a leash.  Or one achieves something by dint of hard work; but a dint is not a thing or kind of thing.  Despite being nouns, words like these surely do not have their meanings by referring to particular kinds of objects.  They seem to have meaning only by dint of occurring in longer constructions.  By themselves they barely can be said to mean anything at all, though they are words and meaningful words at that.
     Fourth, many parts of speech other than nouns do not even seem to refer to things of any sort or in any way at all: “very,” “of,” “and,” “the,” “a,” “yes,” and for that matter “hey” and “alas.”  Yet of course such words are meaningful and occur in sentences that any competent speaker of English understands.
     (Not everyone is convinced that the Referential theory is so decisively refuted, even in regard to that last group of the most clearly nonreferential words there are.  In fact, Richard Montague (1960) set out to construct a very sophisticated, highly technical theory in which even words like those are assigned referents of a highly abstract sort, and do mean, at least in part, by referring to what they supposedly refer to.)
 
     Objection 2:  According to the Referential theory, a sentence is a list of names.  But a mere list of names does not say anything.
 
                     (7) Fred Martha Irving Phyllis
 
cannot be used to assert anything, even if Martha or Irving is an abstract entity rather than a physical object.  One might suppose that if the name of an individual is concatenated with the name of a quality, as in
 
                     (8) Ralph fatness,
 
the resulting string would have normal subject-predicate meaning, say that Ralph is fat.  But in fact, (8) is ungrammatical.  For it to take on normal subject-predicate meaning, a verb would have to be inserted:
 
                     (9) Ralph {has/exemplifies} fatness,
 
which would launch Bradley’s regress again.
 
     Objection 3:  There are specific linguistic phenomena that seem to show that there is more to meaning than reference.  In particular, coreferring terms are often not synonymous; that is, two terms can share their referent but differ in meaning -- “John Paul” and “the Pope,” for example.
 
     It looks as though we should conclude that there must be at least one way of being a meaningful expression other than by naming something, possibly even for some expressions that do name things.  There are a number of theories of meaning that surpass the Referential theory, even though each theory faces difficulties of its own.
 
 

1  Alice’s Adventures in Wonderland and Through The Looking Glass (London: Methuen, 1978), p.180.

2  Complete Poems, 1913-1962 (New York: Harcourt, Brace, Jovanovich, 1972).