Language and Mind
Political language—and with variations this is true of all political parties, from Conservatives to Anarchists—is designed to make lies sound truthful and murder respectable, and to give an appearance of solidity to pure wind.
—George Orwell, Politics and the English Language (1946)
In his novel Nineteen Eighty-four, Orwell painted a grim picture of a future in which the ultimate technology for thought control was the language Newspeak, which could render impossible all modes of thought other than those required by Ingsoc (English Socialism). We have struggled past 1984, but political life, at least, is still replete with euphemisms designed to make us think differently. Thus collateral damage is a way of referring to the killing of innocent people during war, underprivileged means poor, special means handicapped, liquidate means murder. An extreme movement known as General Semantics was established in 1933 by Count Alfred Kozybski, an engineer, and popularized in bestsellers such as Stuart Chase’s Tyranny of Words, and Samuel Ichiye Hayakawa’s Language in Thought and Action. Hayakawa was later president of San Francisco State College, and gained notoriety for stamping out student protest. According to General Semantics, human folly is created by semantic damage brought about by the structure of language.
The relation between language and thought is one of the most contentious issues in the history of philosophy. As we saw in chapter 2, Chomsky’s concept of I-language—the common language underlying E-languages—is essentially the language of thought. This is encapsulated also in the so-called language of thought hypothesis proposed by the philosopher Jerry Fodor, who argued that virtually all of the concepts underlying words are innate.1 Steven Pinker refers to this as the theory that “we are born with some 50,000 concepts,” based on the number of words in the typical English speaker’s vocabulary.2 Of course the actual words we use will depend on the linguistic environment a person is exposed to, but it is as though we have been already supplied with all the meanings we shall ever want, and all we need do is discover the verbal labels. It is of course difficult to believe that a person alive during the Renaissance could have been supplied with the meaning of the word helicopter, although one might perhaps make an exception of Leonardo da Vinci, who indeed developed the idea of such a contraption.
The idea of a strong connection between language and thought implies that nonhuman animals are incapable of thinking as we humans do, an idea defended by the psychologist Clive Wynn in his 2004 book Do Animals Think? If nothing else, the idea of dumb animals—dumb in both senses of the word—is a source of comfort, because it helps justify our appalling treatment of our fellow creatures, as I noted in the preface to this book. Not only do we treat them badly, but we also use them for insults, as when we characterize people as brutish, beastly, swinish, mousey, mulish, catty, hawkish, foxy, bullish, and lazy cows.
Most such expressions are unfair to the animal in question. Consider, for example, the phrase pissed as a newt—as far as I know, newts are not known for inebriation.3 There are of course counterinfluences, such as the animal rights movement, which in its extreme form seeks to reverse the exploitation, threatening to harm or even kill people who exploit animals, with the aim, I suppose, of redressing the balance. Less extreme is animal welfare legislation that seeks to minimize harm to animals. Ironically, perhaps, it was in Nazi Germany that hunting foxes with hounds was first banned, on the orders of Hermann Göring in 1934. In Britain, the Hunting Act of 2004 states that “a person commits an offence if he hunts a wild mammal with a dog, unless his hunting is exempt.”4
A modicum of thought, though, should convince us that other animals have at least a modicum of thought. The companionship of cats and dogs depends on behaviors that are not merely reflex. The behaviorists even supposed that the laws governing human action could be understood through experiments on rats or pigeons. Of course the behaviorists preferred not to speak of consciousness or mental events, so that even human activity was described in terms of behaviors, not thoughts, but the idea of a continuity between animals and humans was paramount. As John B. Watson, the founder of behaviorism, put it, “Behaviorism … recognizes no dividing line between men and brutes.”5 In chapter 3 I described how starlings were able to solve problems in parsing sequences that their experimenters believed to require recursive processing. By adopting a simpler but still clever strategy, the starlings may be said to have out-thought their human keepers. In chapter 3, too, I noted Wolfgang Köhler’s classic experiments on insight in chimpanzees.
Perhaps the most eloquent advocate for the recognition of animal thought and consciousness was the late Donald R. Griffin, whose 1976 book The Question of Animal Awareness claimed that animal communication offered “a possible window on the animal mind.” Griffin continued to write on this topic, and his most recent book, Animal Minds: Beyond Cognition to Consciousness (2001) opens with the following anecdote:
A hungry chimpanzee walking through his native rain forest comes upon a large Panda oleosa nut lying on the ground under one of the widely scattered Panda trees. He knows that these nuts are much too hard to open with his hands or teeth and although he can use pieces of wood or relatively soft rocks to batter open the more abundant Coula edulis nuts, these tough Panda nuts can only be cracked by pounding them with a very hard piece of rock. Very few stones are available in the rain forest, but he walks 80 meters straight to another tree where several days ago he had cracked open a Panda nut with a large chunk of granite. He carries this rock back to the nut he has just found, places it in a crotch between two buttress roots, and cracks it open with a few well-aimed blows.6
Despite such anecdotes, it is difficult to disagree with Clive Wynn’s assertion that animals do not think as we humans do. Of course they can do lots of things that we humans cannot do; examples include echolocation in bats and global navigation in migrating birds, and our clumsy machines that fly pale beside the effortless grace of a bird on the wing. Birds that cache food perform outstanding feats of memory, remembering thousands of different locations. A recent study shows a remarkable performance of chimpanzees in remembering the locations of briefly displayed numerals; one chimp outperformed a group of university students on this task.7 So what is missing?
The evidence reviewed so far in this book indicates that nonhuman animals, even chimpanzees, are essentially incapable of theory of mind, mental time travel, or language—or are at best capable of these capacities only sporadically, and only in rudimentary fashion. I argued in chapter 7 that grammatical language depends on mental time travel, and the adaptive advantage of being able to share episodes. The final ingredient essential to the evolution of language may have been that other recursive function discussed in chapter 8—theory of mind.
Language and Theory of Mind
There is a well-known joke about a man who meets a business rival at a train station and asks where he is going. The business rival replies he is going to Minsk. The first man then says, “You’re telling me you’re going to Minsk because you want me to think you’re going to Pinsk. But I happen to know that you are going to Minsk, so why are you lying to me?”
Commenting on this conversation, Steven Pinker writes:
If a speaker and a listener were ever to work through the tacit propositions that underlie their conversation, the depth of the recursively embedded mental states would be dizzying.8
This is an exaggerated example, but in normal conversation we hardly ever spell out exactly what we mean. Instead, we depend on shared assumptions about what is going on in each other’s minds. You might meet a colleague at work one morning and say, Hey, that was a great game. You know that she watched the rugby match last night, and would have appreciated its quality. You know too that she knows that you are referring to that game, and not to some other event, and you know that she knows that you know this. If you are speaking to a less familiar colleague, though, you might offer more information: Did you watch the rugby last night? Great game, wasn’t it? Or, to a bemused visitor from the United States: Rugby, you know, is our national passion. There was a game shown on TV last night, and if you happened to watch it you would have seen rugby at its best. We therefore calibrate our conversational remarks according to theory of mind—our often implicit assumptions about what the recipient understands and knows.
Sometimes, of course, language is directed at a broader audience, and even as I write this paragraph I have in mind the minds of the audience I hope to reach. That audience, of course, includes you. I assume for a start that you understand English, and that your understanding of the world is much the same as mine, so that the various illustrations I use will make sense. I nevertheless try to spell out the arguments more than I might do if communicating with a close colleague in the same field—such colleagues won’t read the book anyway, since they know me and the stuff I write all too well. The mental work underlying our utterances, whether in casual conversation or in delivering a formal lecture, is recognized by the discipline known as cognitive linguistics. One of its foremost proponents, Gilles Fauconnier, writes that “when we engage in any language activity, we draw unconsciously on vast cognitive and cultural resources, call up models and frames, set up multiple connections, coordinate large arrays of information, and engage in creative mappings, transfers, and elaborations.”9
The essential role of theory of mind in language can be credited to the philosopher Paul Grice, who held that true language requires that the speaker has the intention of changing belief in the mind of the listener by means of the recognition of that intention. (How’s that for a recursive sentence?). Grice was much concerned with the complex reasoning that goes into decoding what a given sentence might mean. He gives an example of what might underlie an utterance, P, in relation to a certain unstated thought, Q:
He said that P; he could not have done this unless he thought that Q; he knows (and knows that I know that he knows) that I will realize that it is necessary to suppose that Q; he has done nothing to stop me thinking that Q; so he intends me to think, or is at least willing for me to think, that Q.10
Unraveling this recursive tangle, though, would seem extremely complex, and at odds with the apparent ease with which people converse. The various thought processes and intentions underlying such statements are known as implicatures, and the manner in which speakers and listeners determine implicatures is one of the goals of the branch of linguistics known as pragmatics. Dan Sperber and Deirdre Wilson have argued that the implicatures depend on a specialized theory-of-mind module, in the sense proposed by Jerry Fodor, and briefly discussed in chapter 1. Modules are assumed to be innate, and to operate automatically, so one might suppose that they carry out operations of a complexity similar to those that, say, allow us to maintain balance while walking. Even so, simply declaring a function to depend on an innate module doesn’t tell us how it actually works.
Sperber and Wilson suggest that there are submodules that help narrow down the alternative meanings, and so reduce the computational demand. For instance, we have a built-in sensitivity to where others are looking, and this can establish a common focus of attention. A statement such as That’s really weird can then be quickly understood to refer to any object at that focus. More generally, Sperber and Wilson suggest that we continually maximize the relevance of available inputs, whether from the outside world or from memory, which can include knowledge of the memories and cultural habits of the person we are conversing with. This immediately narrows the possible interpretations of utterances, and may allow a conversation to proceed with minimal unraveling of the possible implicatures.
Sperber and Wilson provide an illustration of how this might work:
Suppose that Peter and Mary are walking in the park. They are engaged in conversation; there are trees, flowers, birds, and people all around them. Still, when Peter sees their acquaintance John in a group of people coming towards them, he correctly predicts that Mary will notice John, remember that he moved to Australia three months earlier, infer that there must be some reason he is back in London, and conclude that it would be appropriate to ask him about this. Peter predicts Mary’s train of thought so easily, and in such a familiar way, that it is not always appreciated how remarkable this is from a cognitive point of view.11
So-called relevance theory, as developed more fully elsewhere by Sperber and Wilson,12 suggests that our minds are focused, moment to moment, in such a way as to tune our thought processes to what is most relevant, and so reduce the linguistic demand. In conversation, we may be impervious to all but the topic under discussion, and oblivious to other events in the environment. This immediately disambiguates what might otherwise be multiply ambiguous statements. Any unexpected happening would likewise shift the shared mental processing, so participants in the conversation could immediately reach a mutual understanding of what to talk about. Language, then, is a meeting of minds, and conversation often does little more than float on the surface of shared streams of thought
At least in conversation, then, language requires what Dennett called the intentional stance, in which we treat people as mental rather than physical beings. We make implicit assumptions about the extent to which a listener is tuned into the same stream of thought, and adjust our language accordingly. Sometimes, of course, we err, as in lectures when students manifestly have no idea what I’m talking about. One might swear at a dog, or a car that won’t start, and even tell stories to them. A more extreme case is illustrated by the following lyric from the 1951 Lerner and Loewe musical Paint Your Wagon:
I talk to the trees
But they won’t listen to me
I talk to the stars
But they never hear me
The breeze hasn’t time
I talk to them all in vain.
In varying degrees, natural language is an exercise in minimalism,13 providing just enough information to direct mutual trains of thought. This is reflected in two of Grice’s famous Maxims:14
• Make your contribution as informative as is required for the current purposes of the exchange.
• Do not make your contribution more informative than is required.
Minimalism is exploited in the plays of the Irish author Samuel Beckett, and conveyed in his poem What is the word? which ends thus:
glimpse -
seem to glimpse -
need to seem to glimpse -
afaint afar away over there what -
folly for to need to seem to glimpse afaint afar away over there what -
what -
what is the word -
what is the word
We saw in the previous chapter that people with autism suffer deficits in theory of mind. If theory of mind is critical to normal language, we should expect them also to have difficulties with language. And they do.15 Even Temple Grandin, the high-functioning autistic woman we encountered in the previous chapter, struggles with language, despite the fact that she has written books and teaches in a university. She didn’t begin to speak until the age of three and a half, and then used words primarily to refer to things rather than people. In the words of W. D. Hamilton, she is a prototypical things person rather than a person person. She was teased at school for the mechanical way she talked, and was incapable of gossip or chitchat. Perhaps it’s not unfair to say she learned language in much the same way as an animal might learn to perform tricks, and not as a way to share information.
As philosophers claim that no true philosophy is possible without doubt, by the same token, one may claim that no authentic human life is possible without irony.
—Søren Kierkegaard, The Concept of Irony (1841)
Irony provides an excellent example of the role of theory of mind.16 It refers to those occasions on which we say the precise opposite of what we mean, with the understanding that the listener will understand what we are really getting at. Imagine you get stuck in a traffic jam on the way home. “Terrific,” you say to your companion, “that means we won’t have to sit through the game on TV tonight.” If you are a sports fan, this is irony. It is used when there is a discrepancy between what we want or expect, and reality, and occurs widely in such everyday expressions as clear as mud, or Oh great! when you learn that they’ll need your car for yet another day before it’s fixed. Election time in my country seems to bring out an epidemic of the expression Yeah, right.17
Irony is also a well-known literary device. At the beginning of Pride and Prejudice Jane Austen was being ironic when she wrote “It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife.” She really means the opposite—it’s women or their mothers who desperately seek a rich single man for marriage. Jonathan Swift was being ironic when he put forward A Modest Proposal to solve the problems of starvation and overpopulation in Ireland by eating babies. In The Ransom of Red Chief, the short-story writer O. Henry refers to a city: “As flat as a flannel-cake, and called Summit, of course.” A crude form of irony is sarcasm, signaled by a sneering tone of voice. Dostoyevsky called it “the last refuge of modest and chaste-souled people when the privacy of their soul is coarsely and intrusively invaded.”18
Irony depends on theory of mind, the secure knowledge that the listener understands one’s true intent. It is perhaps most commonly used among friends, who share common attitudes and threads of thought; indeed it has been estimated that irony is used in some 8 percent of conversational exchanges between friends.19 Irony can be dangerous if one moves outside one’s circle of friends and acquaintances—as a New Zealander I occasionally find myself misunderstood, roughly as a function of cultural distance. I have no doubt that people in Italy or Japan have their own acute sense of irony, but I have the sense that I may have left in those countries a trail of perceived outrageousness.
Irony seems to pose special difficulties for people with autism. Szilvia Papp tells of an otherwise bright 16-year-old boy diagnosed with autistic spectrum disorder who undertook five GCSE exams in Britain, and became extremely upset when his father jokingly said, “If you don’t pass you’ll have to do them again.”20 Such difficulties extend to other kinds of nonliteral statements, such as the use of metaphor. Papp observes that if told it is raining cats and dogs, the autistic boy will look to the sky to see where these unlikely animals are coming from. Francesca Happé notes that a discussion with a bright autistic child can reveal just how much we use metaphor in everyday speech.21 For example, asking the child to “give you a hand” meets with the serious reply that she needs both hands and cannot cut one off. Asking her to “stick her coat down over there” is met with a request for glue. Or telling an autistic boy that his sister is “crying her eyes out” leads to anxious peering at the floor to find out where her eyes have gone. Because of the ubiquity of metaphor in everyday language, autistic individuals are unable to follow soap operas, and prefer learning lists of train times to reading fiction.
Theory of mind also allows normal individuals to use language in a loose way that tends not to be understood by those with autism. Most of us, if asked the question “Would you mind telling me the time?” would probably answer with the time, but an autistic individual would be more inclined to give a literal answer, which might be something like “No, I don’t mind.” Or if you ask someone whether she can reach a certain book, you might expect her to reach for the book and hand it to you, but an autistic person might simply respond yes or no. This reminds me that I once made the mistake of asking a philosopher, “Is it raining or snowing outside?”—wanting to know whether I should grab an umbrella or a warm coat. He said, “Yes.” Theory of mind allows us to use language flexibly and loosely precisely because we share unspoken thoughts, which serve to clarify or amplify the actual spoken message.22
The language deficits in autism apply primarily to pragmatics—the adaptation of language to social or real-world contexts. In other respects, autistic language may be relatively normal, especially in individuals with the high-functioning form of autism known as Asperger’s syndrome. The books that Temple Grandin has written are grammatically correct and reveal a vocabulary that is if anything superior to normal. Her deficits in the use of language are primarily social, although she has learned to compensate by means of dogged attention to how people behave. There may even be respects in which language in people with Asperger’s syndrome is superior to normal—in the use of technical language, perhaps. One study shows high-functioning autistic boys to be superior to normal in the naming of pictures of objects.23 But it is the social function of language, its role in storytelling, gossip, and group bonding that was probably critical to the evolution of language in the first place. Individuals with autism appear to use language primarily to acquire information rather than to share it.
The understanding that true language involves theory of mind, and the sharing of information, now allows us to take a closer look at gestural communication in chimpanzees, and see what extra is needed to turn this into language.
Gestural Origins
I have argued in chapter 4 that the origins of language lie in manual gestures, and the most language-like behavior in nonhuman species is gestural. But it is still not language. Most authors suggest that the deficit is specific to language, a lack of recursive grammar or of what has been rather clumsily called the faculty of language in the narrow sense (FLN).24 In chapter 7 I suggested that this might in turn derive from the inability of nonhuman species, including apes, to travel mentally in time, and generate sequences of past or imagined future events. Here I consider the further possibility that apes may also be incapable of the recursive theory of mind that underlies human conversational language. To examine this, we need to look more closely at how apes communicate using gestures.
According to Michael Tomasello, there are two kinds of ape gesture.25 One is designed to get another individual to do something. These gestures are small rituals, such as lightly hitting another animal to initiate play, or touching another beneath the mouth to request food, or touching the back of another to initiate piggyback riding. The other kind of gesture is designed to attract attention. A chimp may point to an item of food that’s just out of reach. By attracting another individual’s attention to it, the chimp may hope that the individual will pick up the food and hand it across. Other attention-getting actions include slapping the ground, throwing things, or poking another chimp.
These behaviors suggest that chimps are at least somewhat sensitive to what’s going on in the minds of others, or to what others may do. A chimp may point to an object out of reach to get a person to fetch it for her, but will only do this if she can see that the human is paying attention, which does suggest some awareness of the human’s attentional state. Tomasello asserts that chimpanzees will only point for humans, and the failure to observe pointing among chimps in the wild has led to the belief that chimpanzees don’t point at all. He suggests, though, that they don’t see the point of pointing to each other since they know it won’t work. They have learned that humans are cooperative, at least in research settings, and that pointing brings its rewards.26 Other kinds of gestures between chimps may work, especially if both have something to gain, as in mutual play. In making gestures, then, chimpanzees do seem to be aware of the attentional states of others, whether human or ape, and also aware of the intentions of others. And their gestures are used flexibly and intentionally, rather than as responses induced by events. These qualities are prerequisites for language. Tomasello even refers to them as “the original font from which the richness and complexities of human communication and language have flowed.”27
But they are not sufficient. Michael Tomasello suggests that the missing ingredient is that of sharing. Chimpanzee gestures are essentially imperative, designed to bring reward or advantage to the gesturer. That is, the chimp is requesting something rather than making a statement. Studies of the use of signs by chimpanzees28 and bonobos29 in their interactions with humans have shown that 96–98 percent of their signs are imperative, with the remaining 2–4 percent serving no apparent function—except perhaps one of greeting, or scratching an itch. In marked contrast, human language includes declarative statements as well as imperative ones. We talk in order to share information, rather than merely request something for ourselves.30
The declarative function may be evident even in one-year-old human infants, who sometimes point to objects that an adult is already looking at, indicating the understanding that attention to the object is shared. Tomasello gives a number of other examples where the intention is to share rather than to receive gratification. A 13-month-old child watches as her father arranges the Christmas tree. Her grandfather comes into the room, and the child points to the tree for him, as if to say, “Look at the tree, isn’t it great?”31 At 13.5 months, while her mother is looking for a missing refrigerator magnet, a child points to a basket of fruit, under which the magnet is hidden. Such gestures form the basis of language in that they are designed to share information. They also demonstrate that, in development as in evolution, language originates in manual gestures.
This morning I am sitting on a bus. A woman in the seat in front of me is holding a wriggling 18-month-old boy. She tells me that this is his first bus ride, and he is excited. He catches my eye and begins to point to things, including cars and houses, the bus driver, the cord to signal that people want to get off at the next stop. As he points he looks at me. He does not want these things; he wants to share his delight in them with me. I don’t think chimpanzees point in this way. I think that children only point like this when there’s someone to share the information with.32
The toddler has not yet developed full theory of mind, which develops progressively until the age of about four. Pointing to share information nevertheless seems to be an early stage in the emergence of both language and theory of mind.33 It is remarkable that human children already differ from chimpanzees in that this kind of communication seems to be present in one-year-old human infants, but there is no evidence for it in great apes at all. This work provides further evidence that language is built on mental capacity, rather than mental capacity being dependent on language. The mental prerequisites for language, moreover, begin to emerge well before language itself develops. But the emerging theory of mind may be the spur that allows language to unfold.
Figure 11. Frequencies of gestural and vocal communication in infants between the ages of 9 and 24 months, based on the work of Volterra et al. (2005). In ontogeny, as I have suggested in phylogeny, intentional gesturers appear well before intentional vocalizations. (Reprinted with permission of Taylor & Francis Group. Figure kindly supplied by Virginia Volterra.)
Chimpanzees, along with bonobos, are our closest living nonhuman relatives, and provide the best estimate of what communication was like prior to the emergence of true language. On the basis of current evidence, then, it seems likely that the additional step that allowed humans to share their thoughts arose in the course of hominin evolution itself, after the split from the apes. Again, the critical ingredient may be recursion, the glue that seems to unite theory of mind, mental time travel, and language itself.
In the next chapter, I place recursion in the context of the debate about whether there is a profound discontinuity between humans and other animals, or whether the difference is one of degree rather than kind.