Russell on Descriptions

Russell on Descriptions Bertrand Russell s analysis of descriptions is certainly one of the most famous (perhaps the most famous) theories in philosophy not just philosophy of language over the last century. For obvious reasons, our primary purpose will be to examine the theory as a work in philosophy of language, but it is worth noting that Russell s motivations were not merely linguistic: he was motivated by both metaphysical and epistemological considerations as well. The following sentence was problematic for our naive theory of language: (1) The present king of France is bald. The problem was that there is no present king of France. Therefore, we cannot provide the present king of France with a referent and we must find some other way to account for its meaning. Alexius Meinong, a contemporary of Russell s, recognized this problem and (in)famously decided to adjust his metaphysics to solve it. Meinong accepted such entities as the present king of France, the round square, and whatever else you may form a corresponding description of. Russell finds this metaphysical picture unacceptable; he is unwilling to recognize the existence of such things and he was in search of a theory which had the metaphysical benefits of avoiding such commitments. This was one of his primary motivations for the theory of descriptions. As he puts it: Logic, I should maintain, must no more admit a unicorn than zoology can (324). (Russell also had epistemological motivations for the theory, but they are more complicated and would take us too far afield. Hylton(2004) gives a nice overview of Russell s motivations, if you are really interested.) It is good to keep Russell s motivations in mind in order to remember that what we do in philosophy of language isn t an isolated enterprise. That said, from now on we ll largely set implications aside and focus on Russell s theory as a piece of linguistic theory. Even so, focusing so much energy on descriptions may seem a little strange. To a certain extent you ll have to trust me on this one, but here are two quick remarks. First, as will become clear, studying descriptions will bring to light an extraordinary number of foundational issues in philosophy of language. Second, the immense historical importance of Russell s theory, as well as the fertile discussion of it by the best philosophers of the century, makes it essential to any study of philosophy of language. Types of Descriptions Russell distinguishes two types of descriptions: definite and indefinite. Definite descriptions begin with the definite article the and indefinite descriptions, or at least the ones Russell focuses on, begin with the indefinite article a. (2)- (4) are examples of definite descriptions and (5)-(7) are examples of indefinite descriptions. (2) The table 1

(3) The best table I have over owned (4) The best table ever owned by Ghostface Killah in the days before the Wu-Tang clan hit the big-time (5) A car (6) A blue car parked in my spot (7) A super-expensive car bought by Ghostface Killah We ll work with relatively simple descriptions since they will suffice to generate the relevant issues. However, it is good to keep in mind the fact that natural language descriptions can get quite complicated, as shown by (4) and (7). Any finished theory of descriptions will have to be shown to be capable of handling the relevant type of complexities found in natural language. 1 Plausibly, there are a number of natural language descriptions that aren t covered in Russell s taxonomy. Possessives like (8) and (9) are intuitively classified as descriptions. Again, any finished theory of descriptions will have to deal with these. We ll set them aside, though. (8) Ghostface s car (9) My father s father s mother Important note: Indefiniteness vs. Ambiguity Russell repeatedly calls indefinite descriptions ambiguous. At least when it comes to our contemporary usage of ambiguous, this is a mistake. (It may be that there was another sense of ambiguous in Russell s day, I m not sure.) The word bank is ambiguous. In (10) it denotes a type of financial institution and in (11) it denotes a river bank. (10) I withdrew $100 from the bank. (11) I followed my friends to the bank but I didn t jump in the river. For a term to be ambiguous, there must be multiple distinct meanings associated with that term. This is exactly what is going on with bank but it is not what is going on with indefinite descriptions like a man. A man has a single meaning, it is indefinite in the sense that it doesn t pick out a single man, but it is not ambiguous. What is Reference? Russell s main claim will be that neither indefinite nor definite descriptions refer. He will give an alternative account of them as what he calls denoting phrases. 1 Neale (1990) contains what is, as far as I know, the most thorough attempt to generalize Russell s theory. 2

Before we can understand Russell s theory, we ll have to say a little bit more about the nature of reference. Reference is a relation between words and the world. As we ve already mentioned, it is standardly assumed that proper names refer to individuals. Thus, the proper name David Liebesman refers to me. Another way of saying this is that the reference relation obtains between me and David Liebesman. Perhaps a more intuitive gloss is given by stands for. The name David Liebesman stands for me. All of this is pretty skeletal. On of the best ways to understand what philosophers have in mind is through examples. Pronouns like he and demonstratives (words that usually require an accompanying demonstration) like that are also commonly thought the be referential. He in (12) and that in (13) are referring expressions. The former stands for Bertrand, and the latter stands for a table. (12) Bertrand is thinking about philosophy; he never stops! (13) That is Ghostface s old table. Compare terms like David, he, and that with non-referring expressions that don t stand for particular individuals. Most men in (14) and Nobody in (15) don t refer. (14) Most men are pigs. (15) Nobody is listening to me. Why doesn t most men in (14) refer? Well, most men doesn t stand for anything. To see this, begin by imagining that it does. The following question then arises: which men does it stand for? There is no single group of men that counts as most men, there are lots of equally plausible groups, and there is no reason to pick one over another. This argument is quick (maybe too quick!) but we ll get back to extremely similar reasoning in a minute. Why doesn t nobody in (15) refer? Well, there is nobody that is nobody! Everybody is somebody. There seems to be nobody for nobody refer to. The same goes for phrases like no table and no dog. Recall that Meinong would disagree here. He would claim that nobody refers to a very odd type of thing. Let s stick with Russell and assume that this is a non-starter. For now that s probably the best we can do by way of distinguishing between referring and non-referring expressions. Notice that we don t have an analysis of the distinction. An analysis is what we would have at the end of the day if we did our job well; it is not what we start with. Rather, we start with an intuitive distinction and try to analyze it. Do Descriptions Refer? Russell s answer is no. To see where he is coming from, let s see how tempting it is to claim that definite descriptions do refer. Compare (16) with (17): (16) Barack Obama is in the middle east today. 3

(17) The President of the US is in the middle east today. (16) and (17) look extremely similar. They both appear to be subject/predicate sentences such that the subject picks out an individual Barack Obama and the predicate characterizes that individual. The similarity of the sentences is brought out by fact that we can often swap one for the other without batting an eyelash. Imagine a newscaster choosing between them, it seems that his only relevant considerations would be stylistic. They seem to convey the exact same information and, furthermore, they seem to have the same superficial structure. Russell s contention now seems a bit exciting. He thinks that the surface similarities between (16) and (17) are misleading. Deep down, according to Russell, the sentences have extremely different meanings. This is surprising. The difference, Russell contends, is that (16) contains a referential term in subject position, while (17) does not. (17) according to Russell contains The president of the US as a subject term and this description does not refer to (stand for/pick out) an individual. So why does Russell think this? To ease into Russell s theory begin by considering indefinite descriptions, e.g. a man. Russell gives three arguments that indefinite descriptions aren t referring expressions. Argument 1: The argument from contradiction It is clear what what I assert is not I met Jones. I may say I met a man, but it was not Jones ; in that case, though I lie, I do not contradict myself, as I should do if when I say I met a man I really mean that I met Jones. (323) The idea is that a man cannot refer to Jones because then to assert I met a man would be equivalent to asserting I met Jones. However, this equivalence does not hold because I met a man but not Jones is not an outright contradiction even if it is false while I met Jones but not Jones is. Argument 2: The argument from falsity. But we may go further: not only Jones, but no actual man, enters into my statement. This becomes obvious when the statement is false, since then there is no more reason why Jones should be supposed to enter into the proposition than why anyone else should. (323-324) Russell s thought here is that a man can t refer to any particular man because when I met a man is false, there is no reasonable candidate for a man to refer to. Let s unpack this. Assume that I met a man is false. In that case I met Jones is also false. After all, Jones is a man. Can we then assume that a man refers to Jones. We can t, and here s why: when I met a man is false there is no man that I met. I met Jones on the other hand, is compatible with my meeting lots of other men. Therefore, a man in I met a man cannot refer to Jones. This reasoning, of course, has nothing to do with Jones in 4

particular. There is no man that a man can plausibly be said to refer to, for precisely analogous reasons. Argument 3: The argument from reference failure. Indeed, the statement would remain significant, though it could not possibly be true, even if there were no man at all. (324) This should seem a little familiar: it is similar to the worries concerning on referring terms that we ve repeatedly discussed. The idea here is that a man cannot be a referring term because it remains meaningful even in the absence of any possible referent. Argument 4: The too many men argument. Socrates is a man, Plato is a man, Aristotle is a man, but we cannot infer that a man means the same as Socrates means and also the same as Plato means and also the same as Aristotle means, since these three names have different meanings. Nevertheless, when we have enumerated all the men in the world, there is nothing left of which we can say, This is a man, and not only so, but it is the a man, the quintessential entity that is just an indefinite man without being anybody in particular. It is of course quite clear that whatever there is in the world is definite: if it is a man it is one definite. (327) Assume that a man refers. Then it seems as if it must refer to Plato because it is true that Plato is a man. By the same reasoning, it seems as if it must refer to Socrates, and mutatis mutandis for every man! However, since all of these men are distinct, it cannot refer to them all. If, then, it refers, it must refer to some indefinite man that is distinct from all of the definite men. This, however, offends Russell s metaphysical sensibilities. Those are the arguments that indefinite descriptions don t refer. Russell gives an independent argument that definite descriptions don t refer. Argument 5: Substitution Failure A proposition containing a description is not identical with what that proposition becomes when a name is substituted, even if the name names the same object as the description describes. Scott is the author of Waverly is obviously a different proposition from Scott is Scott : the first is a fact in literary history, the second a trivial truism. Assume that Scott and the author of Waverly mean the same thing. It then follows that Scott is the author of Waverly and Scott is Scott will be identical in meaning. However, the latter is trivial while the former is not. 5

Therefore, they do not mean the same thing. Therefore, it follows, that they do not co-refer. Since the author of Waverly refers to Scott if it refers at all, it follows that the description does not refer. In Descriptions this is the only argument that Russell gives against the claim that definite descriptions refer. However, when he goes on to construct his theory, he elicits a number of virtues. Insofar as these virtues of his theory are not shared by a referential theory, these also speak in favor of Russell s analysis. We ll get back to the virtues. Russell s Analysis For the most part, I am going to follow Lycan s presentation, rather than Russell s own because Lycan s is much easier to follow. Start, again, with indefinite descriptions. (18) is an English sentence containing an indefinite description, (19) is an English gloss of Russell s analysis, and (20) is the analysis of (18) in Russell s favored notation. (Now may be a good time for a logic refresher!) (18) A dog is in the room. (19) There exists a dog that is in the room. (20) x(dog(x) & in the room(x)) As you can, see on Russell s analysis, a dog is not a referential expression, rather is it a quantificational expression. The analysis of definite descriptions is a little more complicated. (21) The author of Waverly was Scotch (22) At least one person authored Waverly and at most one person authored Waverly and whoever authored Waverly was Scotch. (23) x(authored Waverly(x) & Scotch(x) & y(authored Waverly(y) y=x)) From now on we ll focus exclusively on Russell s analysis of definite descriptions, as it is the more important, influential, and controversial analysis. Virtues of the Analysis Apparent Reference to Nonexistents The problem with (1) was that there was nothing for the present king of France to refer to. Therefore, it was mysterious how the sentence could be meaningful and, in fact, false. (1) The present king of France is bald. 6

Now consider Russell s analysis of (1), which requires three things in order to be true: (a) That there is a present king of france (b) that there is only one present king of France, and (c) whoever is the present king of France is bald. Since there doesn t exist a present king of france, (a) is false. Therefore, (1) is false. Since none of (a)-(c) requires the existence of a particular person in order to be meaningful, Russell has shown us how (1) can be meaningful even though the present king of France does not refer. Negative Existentials True negative existentials are really just a special case of apparent reference to non-existents. They do, however, bring the problem into sharp relief. Consider (24): (24) The present king of France doesn t exist. If descriptions are referential then it seems that we must refer to the present king of France in order to ascribe him non-existence. However, if we refer to something then it exists. Therefore, on the referential theory of descriptions, (24) must either be false or non-sensical. However, it is true. This is a problem for the referential view of non-existents. Russell s theory of descriptions nicely avoids the problem. On this theory, (24) is analyzed as (25) (25) x(present king of france (x) & exist (x) & y(present king of france(y) y=x)) Substitution Failure Recall Russell s argument that definite descriptions aren t referring expressions. It was that sentences like (26) and (27) differ insofar the former is trivial while the latter isn t. (26) Barack Obama is Barack Obama (27) Barack Obama is the president of the US. On the theory of descriptions we can explain the differences between (26) and (27). (26) is just as it appears: a trivial identity statement. (27), on the other hand, is tantamount to claiming that: (1) there is a president of the US (2) there is only one president of the US, and (3) Any president of the US is Barack Obama. The conjunction of these three claims is far from trivial. Names and Descriptions As Lycan notes, and as you probably noticed by now, many of the problems that arose for a referential theory of definite descriptions also arise for a referential theory of proper names. I ll briefly mention two. First, just as there are definite descriptions that appear not to refer, e.g. the present king of France, there are proper names that appear not to refer, Vulcan and Sherlock Holmes. If the meaning of a proper name is its referent, then it seems that these names are meaningless. However, this can t be 7

correct because they occur in meaningful sentences. Therefore, it seems that the meaning of a proper name cannot be its referent. Second, just as an identity involving a proper name and a definite description may be non trivial, e.g. (27), an identity involving two proper names may be non-trivial: (28) Dennis Coles is Ghostface Killah Again, if the meaning of a proper name is its referent then it seems as if (28) must be trivial. However, (28) may be highly informative. Therefore, the meaning of a proper name must go beyond its referent. Russell is aware of these problems and his solution is somewhat radical. He thinks that these sentences don t really contain proper names. Rather, they contain definite descriptions disguised as proper names. We will have quite a bit more to say about this when we get to Kripke, but keep it in the back of your mind for now. Potential Shortcomings of Russell s Theory Notice that when discussing definite descriptions, we have followed Russell in focusing exclusively on singular definite descriptions that closely resemble referential expressions. Even if Russell s analysis works for these, it is not obvious that his analysis will work for all definite descriptions. Consider the following: (29) The pandas have escaped. (30) The water is yellow. (31) The dog is a great pet. Russell s analysis will have to be adapted to fit these examples. After all, there certainly isn t a single unique panda when the pandas have escaped. Similarly, whenever we have some water it seems as if can hardly be unique, just divide it in half and you have some other water. Finally, the dog in (31) doesn t appear to pick out a single dog at all! After all, (31) doesn t state that a single dog, e.g. Fido, is a great pet. Rather, it states that dogs, in general, make great pets. Surface Structure and Logical Form We are going to spend quite a bit of additional time evaluating Russell s theory. However, whether or not Russell s theory is correct, it is a landmark piece of analytic philosophy. On of the main reasons for this is that Russell clearly distinguished between a sentence s surface structure and its logical form and hypothesized that the two may come apart. Surface structure (what Russell calls grammatical form ) is just what it sounds like: the way a sentence appears on the surface. Logical form, on the other hand, reveals a sentences deeper properties. It is called logical form because it is supposed to show all of the logically relevant properties, e.g. entailments. Once we distinguish between surface structure and logical form it is an 8

open possibility that logical forms may diverge greatly from surface structure. It then becomes our job to look past surface structure and grasp logical form if we really want to understand the meaning of a sentence. These notions of surface structure and logical form surely need additional explication. The basic idea though, that sentence meanings may have hidden complexity, is clear enough. This idea has had an enormous impact on philosophy of language, and it is certainly one you will notice throughout our readings. 9