Terminating Ambiguity: The Perplexing Case of “The”
Maybe they should never have called the first movie The Terminator. After all, there’s more than one Terminator. That may seem like a picky point, but, believe it or not, philosophers have long been obsessed with trying to determine the meaning of the word “the.” Indeed, much controversy swirls around this seemingly innocuous definite article. Specifically, the controversy focuses on whether or not definite descriptions are ambiguous.
A definite description is a phrase that begins with the word “the” like “the Terminator,” “the leader of the resistance in 2029,” and “the mother of John Connor,” just to name a few examples. These kinds of phrases are called “definite” descriptions since they pick out one unique thing; the thing that fits the description. [i] A word, or phrase, is ambiguous when it has multiple meanings. There are at least two kinds of ambiguity. The first kind is syntactic ambiguity, as in the sentence, “Visiting Terminators can be dangerous.” This sentence has two meanings depending on how we understand it. It could mean that having a Terminator as a visiting houseguest can be dangerous or it could mean that going to visit a Terminator could be dangerous (both are likely true!). But notice that no word or phrase in this sentence has multiple meanings; it is the sentence as a whole that’s ambiguous. Contrast the following sentence: “The Terminator went to the bank.” The word “bank” has at least two meanings and so the sentence could mean either that the Terminator went to the bank of some river, or that it went to some financial institution. The question, then, is: is the word “the” semantically ambiguous like “bank,” thus admitting multiple meanings?
T1: Russell versus Strawson
“Listen, and understand. That Terminator is out there. It can't be bargained with. It can't be reasoned with. It doesn't feel pity, or remorse, or fear. And it absolutely will not stop, ever, until you are dead.” –Kyle Reese
Consider this sentence: “The real-life Terminator can’t be bargained with.” Is it true, false, or meaningless? The sentence certainly seems false as it stands, since there is no real-life Terminator. But if the sentence is false then its opposite, “The real-life Terminator can be bargained with,” should be true. This sentence, however, seems just as false as the first one. We typically think that for any pair of sentences, one of which affirms (“The real-life Terminator can’t be bargained with”) and one of which denies (“The real-life Terminator can be bargained with”), one of them must be true and the other must be false. So we’re faced with a bit of a puzzle.
Bertrand Russell (1872-1970), in his famous 1905 paper “On Denoting,” proposed an answer to this puzzle.[ii] According to Russell the grammatical structure of the sentence doesn’t really clue us into its logical structure. Logically speaking, the sentence really says something like this: “There is a unique object which is the real-life Terminator and which can’t be bargained with.” This phrasing, although clunkier, is helpful because it shows us that there are really two ways to make the sentence false. One way is the way that we previously considered, namely, to say that the Terminator can be bargained with. But we can also consider the opposite, or “negate” the first part (“There is a unique object which is a real-life Terminator”) to get “There is no unique object which is a real-life Terminator.” With this change made, our original sentence is false because there is no real-life Terminator. And its negation is true (“There is no real-life Terminator…”) so the puzzle we started with is solved. This is Russell’s famous theory of descriptions, which enjoys wide support among philosophers who are “in the know.”
Russell’s theory has an interesting implication in that phrases with the word “the” in them are not referring expressions. They do not refer to any particular individual but rather just describe the world as being some way or other. Another eminent philosopher, P.F. Strawson (1919-2006) vigorously attacked Russell’s theory in his paper “On Referring.”[iii] Suppose that the Terminator is at the door and you are about to open it when Kyle Reese springs in, shouting “Don’t open the door! The Terminator will kill you!” Kyle seems to be referring to the particular Terminator at the door, and not merely saying that there is some object (“the Terminator”) and that this object will kill you. Strawson thinks that Russell is wrong, that definite descriptions are referring expressions, like names (“Kyle,” “Sarah,” “John”) and demonstratives (words like “this” and “that”). Russell failed to notice the difference between an expression and the use of that expression.
To see why this difference matters, consider the sentence, “The Terminator cannot be bargained with.” Kyle says this to Sarah in the The Terminator and, let’s suppose, Sarah says it to John in T2. Both Kyle and Sarah use the very same sentence, but they make very different uses of this sentence. In the first movie Kyle uses it to refer to the T-101 whereas in the second movie Sarah would be referring to the T-1000 (let’s say). It is true on both these occasions, since it is said in the movie (if I said it, in real life, the sentence would be false, of course). Strawson concludes that the sentence type (the words that Sarah, Kyle, and I all use) is neither true nor false. It only becomes true or false when someone uses it in a specific situation to refer to something. As Strawson says, “…referring is not something an expression does; it is something that someone can use an expression to do.”[iv] What this really means is that, for Strawson, the meaning of any expression is the set of general directions for using that expression. In the case of “the,” the directions command us to use this word to refer to a familiar object. So, according to Strawson, we cannot decide the truth or falsity of the sentence type at all. All we have are instructions for the use of the expression. Nor can we say who the sentence type is about. It can be used to refer to many different people on different occasions and so can be about many people.
Besides this problem, Strawson points out another flaw in Russell’s approach. The phrase “the Terminator,” according to Russell, tells us that there is a unique object that fits the description, but this just isn’t the case. The T-101 that Arnold plays is just one of many, many T-101s. So when Kyle says, “The Terminator can’t be bargained with,” he is literally saying something false because there are many T-101’s. But this seems like a counter-intuitive result because Kyle seems to be saying something that is straightforwardly true.
T2: The Ambiguity of “The”?
“I can hear it now. He's going to be called the goddamned phonebook killer.”
Philosopher Keith Donnellan (1931- ) entered into this discussion about the role of “the” by publishing a paper called “Reference and Definite Descriptions” in which he pointed out that, in a sense, both Strawson and Russell were right.[v] He argued that any given definite description can be used to refer, but it can also be used in what he called an “attributive sense.”
So consider Lieutenant Traxler’s statement above. He makes the statement when he finds out that someone is going around and killing all of the Sarah Connors in the phonebook, but he has no idea who this person is. Suppose that Lieutenant Traxler says, “The phonebook killer has no pity.” This is an attributive use of the description. It is true of whoever is correctly described as “the phonebook killer.” But now suppose that this description caught on in the press (it doesn’t) and that Sarah, after she learns about the Terminator, says the same thing as Traxler: now she is arguably referring to the Terminator. In the one kind of use, we’re merely trying to tag a property (lack of pity) to some object or other. In the other kind of use, we’re trying to refer to some object or person that we have in mind. The difference between these two kinds of uses turns on what makes them true. In Traxler’s attributive case the truth depends only on whether the description fits some individual, whereas in Sarah’s referential case the truth of the sentence depends on the person being referred to, whether or not the description is true of them.
So, suppose that Lt. Traxler (wrongly) thinks that Kyle Reese is the phonebook killer. Then if he were to say, “The phonebook killer has no pity,” intending to refer to Kyle, the truth of what he says depends on whether or not Kyle Reese has pity, regardless of the fact that Kyle is not the phonebook killer. It would then be false, since Kyle does have pity. On the other hand, if he were to use it in the attributive sense, not speaking about Kyle Reese but rather talking about anyone who would kill in the manner that the phonebook killer does, then the sentence’s truth will depend on whether or not the T-101 has any pity. It would then be true, since the Terminator has no pity.
Given these two different uses of definite descriptions, we have to decide between two competing notions about their nature. On the one hand we might say that definite descriptions are ambiguous: they genuinely have two separate kinds of meanings, a referential and an attributive one. Here, we’d be treating descriptions like “the phonebook killer” the way that we treat words like “bank.” “Bank” has at least two different meanings. So, in this way of dealing with definite descriptions, the sentence, “The phonebook killer has no pity,” will have two distinct meanings. In one of its meanings it will mean something like “the phonebook killer, whoever that is, has no pity” and in the other, it’ll mean something like “the phonebook killer, by which I mean that particular guy, has no pity.”
On the other hand we can say that what’s going on here makes sense in terms of what’s called the semantic vs. pragmatic distinction. Semantics deals with the meaning of expressions or terms, whereas pragmatics deals with the way that people use the expression in communication. So, to take a simple example, suppose that I said “This movie is so interesting” in a sarcastic way, as to make it clear that what I really meant was that this movie (hopefully not Terminator: Salvation!) was anything but interesting. The sentence that I speak in this case would have a different meaning from how you meant to use it.
Yet Paul Grice (1913-1988) argued that there’s no need to worry about this second meaning for the sentence I said, one on which it really means the opposite of what it usually means. This is an example of Grice’s “modified Ockham’s razor.” Ockham’s razor helps us choose between two competing theories: all other things being equal, the simplest one is the best. Grice’s version tells us that we shouldn’t “multiply linguistic entities beyond necessity.” If we can find a way to explain what is going on in the above examples without having to come up with multiple meanings for the sentence, then we should do so.
Grice thinks we can distinguish between the meaning of the sentence itself (as said spoken) and the speaker’s meaning in saying the sentence. The sentence’s standard meaning is that the movie holds my attention; but the speaker means to be convincing us of the opposite. If Grice is right, then we can explain what’s going on with definite descriptions in this way: we can say that definite descriptions have their attributive meanings like Russell thought, and yet people sometimes use these in referential ways like in the examples above. But this isn’t a meaning of the sentence that I say. This is the thought that I am trying to express in saying what I did in just the same way as when I say that the movie is so interesting. What I say is not what I mean. So which is right? Do they have two meanings or just one meaning and different uses?
T3: Kripke and Devitt
Saul Kripke (1940- ) argues against treating descriptions as ambiguous in his paper “Speaker’s Reference and Semantic Reference” where he calls positing an ambiguity “the lazy man’s approach.”[vi] If referential uses of descriptions occurred in a language that was stipulated to be as Russell says then it cannot be an argument against Russell that it occurs in English. So let us imagine a fictional language, call it “Russell English,” in which we just stipulate that definite descriptions work in the way that Russell says that they do. In Russell English, “The Terminator has no pity” has just its attributive meaning, which is that there is some object or other that is the one and only Terminator and that object has no pity. In such a language, Kripke argues, people could still use “The Terminator has no pity” to refer to the Terminator, and so the existence of referential uses in actual English cannot be an argument against Russell. In this fictional language we are imagining there is no referential meaning for definite descriptions. This is true simply because we have stipulated it to be so. So there is no question about whether the descriptions in Russell English are ambiguous or not; they are not. But even so the speakers of Russell English could still use those descriptions to refer to people and objects in spite of the lack of referential meaning. If this is the case, and it seems as though it is, it can’t be a problem for Russell’s theory that people use descriptions to refer. How could it? We could, for all we know, be speaking Russell English, and if that were so we would still be able to use descriptions referentially. Kripke suggests that Grice’s way of handling these kinds of cases is all that we need. Why multiply linguistic entities that do no work for us?[vii]
Kripke also argues against treating descriptions as ambiguous by drawing our attention to anaphor. When we use anaphor, we employ pronouns that refer to some object that we previously used a name for. So, in the sentences, “The Terminator has no pity. It cannot be bargained with,” the word “it” is anaphoric. Kripke thinks this gives us good reasons not to treat definite descriptions as ambiguous. Suppose that we are watching John having lunch with someone (who is not the Terminator) and who is acting sympathetically towards John. Kripke asks us to “consider the two following dialogues,” modified for our context:
Albert: The Terminator is kind to him.
Barbara: No he isn’t. The man you’re referring to isn’t the Terminator.
Albert: The Terminator is kind to him.
Barbara: He is kind to him, but he isn’t the Terminator.
In the first dialogue, Barbara uses the word “he” to refer to the Terminator (and not to refer to the person having lunch with John) while in the second dialogue, Barbara uses “he” to refer to the person having lunch with John (and not the Terminator). However, Albert is using “The Terminator is kind to him” in the referential sense and so is referring to the person that John is having lunch with. Whether or not that person is the Terminator, the sentence depends for its truth on that person and only on that person.
But this presents a problem. The second dialogue is easy to explain: Barbara’s use of “he” is anaphoric and depends on Albert’s referential use of “the Terminator.” It refers to the person that John is having lunch with. But this isn’t the case in the first dialogue. If “he” is anaphoric in this dialogue, depending on “the Terminator” then it has to have the same reference. If it’s true that Albert’s line has a referential meaning then Barbara’s use of “he” must refer to the person that John is having lunch with. But this is clearly not what’s going on. Barbara is obviously using “he” to refer to the Terminator and not to the person John is having lunch with. Since it’s clear that Barbara means to use “he” to refer to the actual Terminator and not to the person who is having lunch with John, we can explain this only by either denying that “he” is anaphoric or by giving up the idea that “the terminator is kind to him” has a referential meaning.
Michael Devitt (1938- ) takes the first option, suggesting that “he” in the first dialogue is in fact not anaphoric but rather a pronoun of laziness. Let’s look at the way these things normally work. If I say, “John Connor has his paycheck directly deposited but Myles Dyson has to take it to the bank,” the occurrence of “it” here can’t be anaphoric. That is, it can’t be taken as referring to John’s check but must be taken as referring to Myles’s check. Maybe I should have said, “…Myles Dyson has to take his paycheck to the bank,” repeating “his paycheck” but, being lazy, I use “it” instead. If “he” was Devitt’s pronoun of laziness in the first dialogue, then the dialogue should be read in the following way:
Albert: The Terminator (in-the-referential-sense-referring-to-that-guy-over-there) is kind to him.
Barbara: No the Terminator (in-the-referential-sense-referring-to-the-actual- Terminator) isn’t. The man you’re referring to isn’t the Terminator.
Originally, Barbara simply used “he” as a replacement for the description “the Terminator” but we take it to be the referential meaning that refers to the actual Terminator.
Kripke objects to this because “[Barbara] may well be in no position to use [“the Terminator”] referentially. She may have merely heard that [the Terminator has no pity].” If this were the case then Barbara could not be using “he” as a pronoun of laziness and it would have to be taken as anaphoric on Albert’s use of “the Terminator.” Devitt responds that it “might be a pronoun of laziness for “[the Terminator] taken attributively, even though Albert’s use of “[the Terminator]” is referential.”[viii] So for Devitt, we should understand Barbara as saying, “No, the Terminator (in-the-attributive-sense-whoever-that-is) isn’t. The man you’re referring to isn’t the Terminator.” This would be possible if Barbara knew the person who John was having lunch with and knew that he wasn’t the Terminator.
At first glance, it does seem that we might be able to switch between the meanings of ambiguous words when using these kinds of pronouns. For instance consider the recent DVD release of a movie called Paycheck.[ix] Now suppose that I say “I put my paycheck in the bank, John put it in the DVD player.” Can “my paycheck” be taken as referring to my paycheck from work while the “it” is used as a pronoun of laziness that refers to his rented copy of the movie Paycheck? I could do this if I wanted to make a joke through a play on words, but as Devitt says, this depends on “how much laziness is acceptable,” and there are limits!
So far, all of these arguments have been inconclusive, which is always a possibility in philosophy. Kripke, however, makes another suggestion which seems to me to be decisive, if we modify it for our Terminator context:
There is no reason to suppose that in making an indirect discourse report on what someone else has said I myself must have similar intentions, or be engaged in the same kind of speech act; in fact it is clear that I am not. If I say “[John Connor says all of the police are here]” [John] may have meant it as a warning but I need not say it as a warning. If the referential-attributive distinction is neither syntactic nor semantic, there is no reason, without further argument, to suppose that my usage, in indirect discourse, should match the man on whom I report, as referential or attributive. The case is quite different for a genuine semantic ambiguity. If Jones says, “I have never been to a bank,” and I report this, saying, “Jones denied that he was ever at a bank,” the sense I give to “bank” must match Jones’ if my report is to be accurate.[x]
If I say, “The Terminator has no pity,” and I am using the description referentially, and then you report to someone else that, “Richard said that the Terminator has no pity,” you don’t have to be using the description referentially. In fact you may not even be able to use it referentially, as you may not know who “the Terminator” is. If the referential use were semantic, this would pose a problem, because we can only resolve genuine semantic ambiguities if we use words with the same sense (“paycheck” has two senses above). Since descriptions are more pragmatic than semantic, there is no ambiguity.
Devitt has tentatively responded[xi] that whether or not you know who the Terminator is, this would not prevent you from using “the Terminator” referentially. This is due to Devitt’s notion of reference borrowing. This is easiest to illustrate in the case of names. Suppose that you have never seen any of the Terminator movies (for shame!) and I tell you that the movie is about a guy named John Connor who will eventually lead the human resistance against an army of cyborgs created by Skynet. You then acquire the ability to refer to John Connor from me. Devitt thinks that our ability to reference borrow doesn’t take much. All you have to do is to hear the name from someone who is in a position to refer to the person in question and you then acquire the ability to refer to that person as well, whether you know anything about them or not.
But even if you were able to use “the Terminator” referentially, Kripke’s point is that you don’t have to. You could be using the description in an attributive sense when you report what I said. Nothing forces you to use the description in the same sense that I did in order to successfully report what I said. But this is very different from the case of actual semantic ambiguities. In the case of an actual semantic ambiguity, if you do not use the word with the same sense that I used it, then you are not accurately reporting what I said. So, if I say, “I like dogs,” meaning hot dogs and you report, “Richard said that he likes dogs,” meaning the animal canis familiaris, you haven’t accurately reported what I said.
T4: Ambiguity Salvation
So what then is the score? From what we’ve seen the balance seems to be tilted slightly in favor of Grice. Kripke’s argument from indirect quotation doesn’t have an answer and so I think we can safely say: “Ambiguity, you’ve been terminated.” But what does this tell us about the question we started with? Should the first movie have been called The Terminator? If what we have said is right then the title literally means what Russell said that it did: there is one and only one object which is the Terminator. Given this the title is literally false; there are many Terminators. Nonetheless James Cameron most likely meant to be taken as referring to the T-101. This is the way that everyone in the films uses the phrase as well. This is a perfectly legitimate use of the phrase and so the title is apt despite its literal falsity. Sheesh! Can you believe people actually get paid to think about this stuff? Well, if it’s any consolation, philosophers don’t get paid much![xii]
[i] They are contrasted with indefinite descriptions like “a Terminator” or “a leader of the resistance in 2029.” I will not have anything to say about these kinds of phrases though there is a fair amount of controversy that swirls around them as well.
[ii] Russell, “On Denoting,” in The Philosophy of Language, ed. A. P. Martinich (Oxford University Press, 1985).
[iii] P.F. Strawson, “On Referring,” in The Philosophy of Language.
[iv] Strawson, “On Referring,” p. 249
[v] Keith Donnellan, “Reference and Definite Descriptions,” in The Philosophy of Language.
[vi] Saul Kripke, “Speaker's Reference and Semantic Reference,” in Pragmatics, ed. S. Davies, (Oxford: Oxford University Press, 1991).
[vii] Michael Devitt’s response to this is that in Russell English there would not be a convention for expressing singular thoughts the way there is in English. There would have to be careful stage setting in order for someone to make a referential use. For a careful defense of the ambiguity of definite descriptions see Devitt, “The Case for Referential Descriptions,” in Descriptions: Semantic and Pragmatic Perspectives, eds. Marga Reimer and Anne Bezuidenhout, (Oxford: Oxford University Press, 2004), pp. 280-305
[viii] Michael Devitt, “Donnellan's Distinction,” in Foundations of Analytic Philosophy: Midwest Studies in Philosophy, eds. French, Uehling and Wettstein (Minneapolis: University of Minnesota Press, 1981), p. 522.
[ix] Based on a story by Philip K. Dick.
[x] Kripke, “Speaker’s Reference and Semantic Reference,” p. 83.
[xi] In a personal communication to the author.
[xii] I would like to thank Frank Pupa, Kevin S. Decker and Bill Irwin for valuable help in revising earlier drafts.