Find Enlightenment

Wittgenstein's Investigations §79
by Bryan Register

Date: 1 Aug 98
Forum: University of Texas at Austin
Copyright: Bryan Register

Note: The author may or may not still agree with the views expressed in this paper.

Wittgenstein sketches a theory of proper names in section 79 of the Philosophical Investigations. The purpose of this paper is to explicate that theory, attending to any philosophical difficulties which may arise on the way.

Wittgenstein asks us to consider the meaning of the example sentence "Moses did not exist". Immediately a problem arises.

On Searle's account of assertive propositions, such sentences can be divided into two parts, the subject and the predicate. The subject selects some object(s) in the world, while the predicate specifies that the selected object(s) possesses some property or properties. (I take this element of Searle's theory to be commonsensical and not in doubt; merely a way of saying the obvious.) If the sentence fails to have an object selected for predication, then it is not properly a sentence. If I were to say "Is red", I will have failed to say anything. If I say "Glurg is red", I have again failed to say anything, because 'glurg' is no better than a blank space at being a meaningful word.

This seems to be the situation of the sentence "Moses did not exist". If the sentence were true, then there is no Moses for the word "Moses" to select and thus the sentence fails to say anything. But of course the sentence does say something. Otherwise, my clause "If the sentence were true" would make little sense, because a failed sentence doesn't get to be true or false but is meaningless. However, we have no difficulty in understanding my clause. Moreover, Wittgenstein proposes a number of possible meanings for the sentence: "The Israelites did not have a single leader when they withdrew from Egypt", and so forth. Any of these are things someone might mean to say by uttering the sentence "Moses did not exist".

We know that the sentence does mean something. However, its meaning cannot be simply its reference (the object in the world it is the name of) or else we enter the problem mentioned above. If the meaning of a proper name is not its reference, then perhaps it is its sense. Thus Wittgenstein looks immediately to Russell's theory of descriptions: "We may say, following Russell: the name 'Moses' can be defined by means of various descriptions"(emphasis added).

Russell had said the following: "And so, when we ask whether Homer existed, we are using the word 'Homer' as an abbreviated description: we may replace it by (say) 'the author of the Iliad and the Odyssey .' The same considerations apply to almost all uses of what look like proper names." (Introduction to Mathematical Philosophy, p. 179) On a Russellian account of names as abbreviated descriptions, we can treat "Moses" more like an ordinary kind term and less like something unusual, and give it a definition such as "the man who led the Israelites through the wilderness". Thus, through substitution, we arrive at "The man who led the Israelites through the wilderness did not exist", which can in turn be rephrased "No one man led the Israelites through the wilderness".

In retrospect we can see that section 79 has precious little to do with the problem of negative existentials (the problem of sentences which, through denying the existence of the referent of the subject, make themselves fail as sentences). Rather, Wittgenstein could have just as easily asked us to consider the example sentence "Moses existed". He chooses the negative utterance, perhaps, because it seems more problematic and because there are obviously many different things which it could mean (see his list of examples). We are not tempted to think that "Moses existed" has any more than the one surface meaning. However, the positive sentence does in fact bring on the problem of how proper names mean. Any sentence which includes a name brings on this question. Section 79 is a partial critique of Russell's theory of the meaning of proper names, even as it develops from that theory. Wittgenstein thinks that Russell is wrong to say that a proper name can be precisely defined. It is worth noting that Wittgenstein has also argued that ordinary kind terms cannot be precisely defined, either. There is a similarity here which I will exploit below.

Wittgenstein's critique of the notion that proper names can be precisely defined is as follows. There is no one description of Moses which we would be willing to substitute for "Moses". Nor is there some class of descriptions all of which must be true, because we would probably accept someone as Moses who had done only a few of the things the historical Moses is said to have done. Some of these descriptions would turn out to be irrelevant in comparison with others. But we cannot say which are and which are not relevant, at least not precisely.

Let me express this in a different way. Consider a set of possible worlds. These worlds are identical, but for the difference between each world's Moses. (There will also be other, trivial, differences. For instance, there will be the fact about me in this world that I am in such-and-such a Moses world, and not another one. But since this difference is a rider on the main difference, is causally irrelevant, and is a consequence primarily of our way of talking about facts and about possible worlds, I will ignore it.) In one world, Moses led the Israelites, was rescued from the Nile, and received the Ten Commandments from God (if God is possible). In another world, God doesn't exist and Moses wrote the Ten Commandments. But the possible worlds differ only in facts about Moses (and irrelevant riders). All possible worlds with a Moses include one common fact: they include only one person who qualifies as Moses.

Now, each world has Moses, but each world's Moses is different. What accounts for each of the Moseses being Moses, in spite of being different from one another? There is not a single property, nor a well-defined and stable set of properties that account for someone's being Moses. But there is a family of properties, somewhat related, at least some of which must be possessed for someone to be Moses. If someone in a possible world possesses some of that family of properties, then he gets to be Moses. Otherwise, that world is without a Moses. Thus, were we to make the claim "Moses did not exist", we are saying that we live in a world in which too few of that family of properties are instantiated in the same person for anyone to count as Moses.

We can consider a space of possible worlds, arranged by considering the relative similarities and differences between the possible worlds and putting the most similar ones, represented by points, nearest each other. Different kinds of difference would be treated as closeness or distance along different dimensions of the space; since there are many different kinds of difference, the space would have many (very many) dimensions. The possible worlds within a certain segment of the space each include someone who qualifies as Moses. There are worlds who have an unquestionable Moses, who instantiates all of the central features we would regard as crucial for Mosesness. There are worlds which have a Moses-like figure who we might or might not regard as Moses. And there are worlds with less and less Moses-like figures who we are less and less likely to regard as Moses. (For instance, say that as the Israelites wandered through the Sinai, they kept coming upon some large stone which worked for them as a kind of direction marker - thus it leads them through the desert. Moreover, one time when they come upon it, the wind and sand have etched a set of lines on it which happen to look just like moral rules. Is the rock Moses? Certainly not.)

The edges of the set of Moses-worlds, however, are fuzzy. Consider §88, where Wittgenstein says "If I tell someone 'Stand roughly here'-may not this explanation work perfectly?" even though it is inexact? We could precisely define a space to stand in, but what would the point be? Likewise for Moses. We allow a space of possible worlds, whose residents resemble one another in various ways. "Moses" selects a person within each of the worlds within a certain vaguely defined region of this space.

When we wonder whether Moses existed, we are wondering which possible world we live in. Are we living in one of the Moses-worlds? And if we are in a non-Moses world, exactly which crucial property(ies) are not instantiated in some one person in our world? Wittgenstein's different explanations for what someone might be saying by denying Moses's existence evinces different answers to this last question.

I wish to drop at this point the 'possible worlds' way of talking about Moses simply because it is hard to speak in this way. I will hereafter speak of 'possible people', only one of whom actually existed, but all of whom are Moses. Rather than a space of possible worlds, I will talk of a space of possible people and a segment of the space whose members would qualify as Moses. Each of these Moseses corresponds to a possible world which is a Moses-world. This segment of space is, as with the group of possible worlds which are Moses-worlds, fuzzy and open-ended.

Wittgenstein's account of proper names builds from Russell's even as it critiques it. Wittgenstein agrees with Russell that proper names have a sense; that is, they are not simply attached to their referents. Proper names refer to their referents by means of their senses or definitions. Wittgenstein wishes, however, to make the 'definition' in question open, fuzzy, and imprecise. He seems to reject the notion that we would be "always ready to substitute some one ... description..." but he does not say that we should never be ready to substitute any description, so long as it were sufficiently rich and we were willing to allow for incompleteness, open-endedness, and correction in light of newly discovered facts.

He says, "And this can be expressed like this: I use the name 'N' without a fixed meaning." Wittgenstein does not say no meaning. And since in this context, we can take it that 'meaning' refers to sense or definition, rather than referent, it is an open possibility that Wittgenstein may think that a proper name's meaning is its sense. And Wittgenstein's explanation of proper names does seem to indicate just that.

Kripke has developed the major countertheory to the Wittgensteinian or 'cluster' view. On Kripke's view, a name means its referent directly with no sense intervening. Descriptions do not provide the definition or sense of a proper name, but serve rather to fix the reference of the name. This happens, I take it, as follows. I mention Moses in a string of discourse. You ask me "Who is Moses?" I respond with some adequate description. You attend to my description adequately. Your use of the name "Moses" has now been fixed to a certain referent, the referent of my use of the name "Moses". As Kripke puts it "An initial 'baptism' takes place. Here the object may be named by ostension, or the reference of the name may be fixed by a description. When the name is 'passed from link to link,' the receiver of the name must, I think, intend when he learns to use it with the same reference as the man from whom he heard it." (Naming and Necessity, p. 96)

Kripke does not, however, disagree merely about whether proper names have sense. Since he denies the cluster view, he adopts the view that names are rigid designators, or words which "in every possible world... designate... the same object". (Naming and Necessity, p. 48) But the object is not given by description. Since names are not descriptions but attach directly to their referents, the objects need not actually share anything in common to count as the same object other than that they are the same as one another.

This theory does not seem adequate. However, the reasons for its inadequacy are not especially Wittgensteinian but consist of puzzles of the same kind Kripke provides for the cluster view. I will raise two objections which are not intended to be conclusive, but should cast sufficient doubt on the theory to make it worthwhile to continue to consider the cluster view.

Let us say that you ask me who Moses is, and I (as a practical joke) say something wrong, like "Moses was the great Buddhist leader who led his people to freedom." According to Kripke, I have fixed the reference of the word. But, though I use the word correctly, I have fixed your reference wrongly. When you use the name "Moses", you will use the name to refer to someone who was Buddhist, not Jewish. The conventional usage of the name "Moses" will not change your idiosyncratic usage; you will not have meant the Jewish leader.

In the movie Casablanca, we are treated to the amusing spectacle of two Germans who are trying to escape to America. They have learned English and plan to use it exclusively, in preparation for their eventual arrival. One of them asks the other, "What watch?" and the other, after looking at her watch, responds "Seven watch." This pair is using the word "watch" wrongly in two different ways, quite aside from what they want to do. In English, they are pronouncing nonsense sentences. Likewise, if someone's reference is fixed wrongly, they will be using language wrongly. Kripke seems to think that a word refers to its object no matter what they user knows or believes. But this is manifestly wrong; words do not have an intrinsic link to their referents beyond any link provided by human description. If that link is broken, the word is used to refer wrongly. But if we have a criterion for wrong reference, then that criterion is the sense of the word - the very thing Kripke sets out to deny.

My second objection is a puzzle about personal identity. Let me take myself as an example. In fact, I was raised in a particular household and with a particular frame of mind. But it is surely possible that my parents could have died while driving me home from the hospital, that I could have been subsequently adopted, and that I could have then been raised in a radically different environment which would have produced a person utterly unlike myself.

I am tempted to say that I would not exist had this course of events taken place. Rather, there would have been an identical twin of mine with the same name. Consider the following hospital mix-up: I was one of a pair of identical twins (in the same world), but my twin was given to the wrong family and no one in my family knew about the twin. Let us say that the family to which my twin was given named him "Bryan Register". We are not tempted to say that his accomplishments reflect on me; we are different people. Likewise, my twin in the (possible) world in which I was adopted is not me. His name and mine are not the same name any more than two John Smiths have the same name - our names are homonyms. But for Kripke, the name "Bryan Register" would pick out my twin in the possible world, even though he is manifestly not me.

Let us apply this to Moses. Let us say that someone was born, christened "Moses", and lived an empty life devoid of accomplishment and was then totally forgotten after his death. But he was a Jew who lived at the appropriate time. In this scenario, there was no great leader of the Israelites who led them from Egypt; Moses is a myth. The pointless Moses of this world could have grown up to lead his people, but he didn't. In fact, he shares nothing with the historical Moses but a name and a genetic code.

It is not right to say that, when I use the word "Moses" in this possible world that I am referring to this person. I am trying to refer to a person who led the Israelites to freedom. This person is not Moses just like my twin is not me. But, since names are rigid designators for Kripke, and my twin and the pointless Moses are the same object as myself and Moses in another possible world, then for Kripke we use these names, against our will , to refer to these people. But surely we can't do anything with language which we do not wish to do (other than convey confusion, as with the watches above).

Since Wittgenstein preserves meaning (sense, definition) but unfixes it, he is continuing the Russellian tradition with the modification of a looser and changeable set of descriptions functioning as definition for the name, rather than letting a single permanent description function as definition. This is very similar in some ways to Wittgenstein's account of family resemblance. When he discusses games, Wittgenstein points out that when we examine all of the things which we would be tempted to call a "game" (all the referents of the word "game"), "we see a complicated network of similarities overlapping and criss-crossing: sometimes overall similarities, sometimes similarities of detail." Let us envision all the games, and everything else, as within a space which shows similarity relationships along a number of dimensions. Different individual things (either possible or actual) are represented as points within an n-dimensional space (like the space of possible people from before, but not like the space of possible-worlds) and more similar things are closer together. "Game" refers to anything within a certain fuzzily outlined region of the space. At the borders of the space, we begin to wonder whether things are or are not games and may include them or not. Farther out, we get to things which are not at all games. This is the same spatial picture which I exploited to explain Wittgenstein's treatment of proper names: there is a region of space which represents all the Moseses or all the games. At the edge, group membership becomes doubtful. The difference between a proper name and a kind term, it would seem, is that a proper name refers to the one actual individual within a fuzzily defined set of possible individuals, while the kind term refers to all of the actual and possible members of a fuzzily defined set of possible and actual objects (persons, attributes, etc.).

At §75 Wittgenstein asks "What does it mean to know what a game is?" If the accounts of proper names and kind terms are similar as I have argued, then this question must be similar to the question "What does it mean to know who Moses is?" At §78, Wittgenstein tells us to "Compare knowing and saying :

how many feet high Mont Blanc is-

how the word 'game' is used-

how a clarinet sounds."

Knowing how many feet high Mont Blanc is would be something like being able to pronounce mentally a certain number. But this is exactly the same as saying out loud how many feet high the mountain is. The only difference is whether the pronunciation is internal or external.

Knowing how a clarinet sounds is a non-linguistic affair (for non-musicians; musicians' precise vocabulary might allow one musician to convey the sound of a clarinet to another linguistically). This is a case where knowledge, even unto certainty, cannot be communicated.

We might be tempted to say that, were I to pull out a clarinet and play it, I would answer the question "What does a clarinet sound like?" But this is not right. Let's say the question were, instead, "What does the song of a robin sound like?" Let us say that, just as the questioner concludes, a robin begins to sing. We are not tempted to say that the robin has answered the question, though the questioner now knows what she wanted to know. Likewise for playing a clarinet. The relation between a question and a reality which shows the answer is different from the relation between a question and a response which says the answer. Even if the clarinet is played just to convey to the questioner a sense of what a clarinet sounds like, it does not say but rather shows that sound. I have answered the question, but in a different way - the way the robin answers your question about robins. (If God poked the robin and made it sing because He wanted you to know what a robin sounded like, we would not wish to say that God had said, but only that He had shown, the answer.)

Knowing how the word 'game' is used is somewhere in the middle. I can say, both to myself and to you, what a game is. But as I consider more things which I would call games, I recognize them as such even though they may not possess the defining characteristic(s) which I would say to myself or to you. I cannot convey exactly (even to myself) what is the rule by which I select things as games, but I know what "game" means if I select correctly (and correctness is defined in relation to my form of life, which is probably that of my society. Thus I am correct if I am doing what everyone else would do.) I thus cannot tell you what the word means but must train you in the use of the word just as I have been trained.

Likewise, it seems, for "Moses". I can tell both myself and you what Moses was like. But if I were to be informed by archaeologists and historians that my characterization had been mistaken, I would know whether the new historical personage was properly Moses or not even in lieu of a precise definition for Moses. I could then redefine "Moses" in accordance with the newfound facts about the referent just as I could add another disjunct to my characterization of "game" on learning about a new kind of game.

My goal has been to compare Wittgenstein's treatment of proper names with his treatment of kind terms. The referents of kind terms (all of the things which a native speaker of English would be willing to say are whatever is the kind term) share family resemblance. We can specify, but only vaguely and temporarily, what is the sense or definition of the term. On being faced with non-paradigm instances of the kind, we can expand our sense of the word. Likewise, though the referent of a proper name is only one person, just what are the properties of that person may be unknown. Yet we can know of any given person whether she is that person, even though we do not know in advance what are the properties which any person must have in order to be that person. We recognize a non-paradigm game without knowing just why; we recognize someone as Moses even though he didn't do many of the things we had thought Moses had done.

Find Enlightenment at