Several new articles explore the question of where words come from. The basic association between sound and sense is the most obvious one and it explains why many mammals can learn to recognize their name and respond appropriately to certain ones. “Com’on boy,” you say to your dog, and you dog comes to you. We take the feat for granted, but it is no small thing to control an animal by means of speech and gesture alone. And when behaviorism ruled the psychology departments, association was considered the only way words could be learned. Even theorists now grant that the idea was as naïve as a middle-schooler in Kokomo, but what else in going on? How about some mind-reading?
The July/August issue of Child Development has a couple of articles that explore the rise of words in young children. One of them compares normally developing children with autistic ones, finding that autistic children can learn some words, although they are no good at it when mind reading is required. (See: “Children with Autism Illuminate the Role of Social Intention in Word Learning,” by Julia Parish Morris, Elizabeth A. Hennon, Kathy Hirsh-Pasek, Roberta Michnick Golinkoff, and Helen Tager-Flusberg, abstract here.)
The article does not discuss simple association, but that is no surprise even though it may well lie behind the very first words children speak . They associate the sounds they make with the people around them, but associations are so much older than speech that as far as language is concerned they seem to fall into the spandrel category of things that just happen to come along with the rest of a pile of evolutionary baggage. The ability to associate sounds with objects is necessary for speech, but is nowhere close to being sufficient.
A step further up the line is learning the names of things by paying attention. The need for attention is probably required even on the animal learning level. I would not be surprised to learn that gerbils are smart enough to catch their name, but they pay so little attention to the sounds that their supposed owners make that they never associate sound and self Dogs are much more attentive to the human voice, so not surprisingly they associate things with what we say.
The authors quote Michael Tomasello:
Sounds become language for young children when and only when they understand that the adult is making that sound with the intention that they attend to something. [p. 1266]
There you have it. Mind reading. Language seems convenient because it allows us to know what is in another’s mind without having to resort to ESP, but it seems you cannot get to language without being able to guess their intentions.
Parish-Morris et al. report a series of experiments to explore this idea. This first one was a test to determine whether young children can have their attention directed by others even without the use of words. When experimenters pointed to things and handled them, children will pay attention to the indicated objects. Even autistic children will pay attention, although not for as long a period as typically-developing children. It would be interesting to see this experiment performed with chimpanzees.
Apes will redirect their attention in response to others, but only a few things can hold their attention. Language enables speakers to select topics at random and declare them interesting, but listeners have to be willing to direct their attention away from food or predators to something neutral in the environment. This first experiment seems trivial in its way—pointing at an object in a group of 4 and watching to see if a child will pay attention to the object—but it is so contrary to the obvious pressures for survival that a willingness to explore novelty is quite unusual. Yet even very damaged, autistic, children can do it.
A second experiment presented children with two objects and tested the children for which one seemed more interesting. Sometimes the experimenter directed a child’s attention to the object it chose as more interesting, sometimes toward the other. In each case, the experimenter gave the child a name for the indicated object. Would it be any easier to learn the name of an “interesting” object than a “boring” one?
The results determined that typically developing children can learn the name for either an interesting or boring object, but autistic children did much better at learning the name of the interesting object. The normal children were about two years old, the autistic ones several years older. Typically, year-old children cannot learn the names of things that do not find intrinsically interesting either. So something happens during a normal child's second year to socialize their attention. This path suggests a series of steps in ability to learn words:
- Associate sounds the speaker makes with important people (mama)
- Learn names of interesting things (juice)
- Learn names of things not immediately interesting (string)
We cannot assume that speech’s evolutionary history followed the same history as today’s development of normal children, but we can at least see that somewhere in the story a willingness to attend to things that are not naturally interesting must have appeared.
Even more challenging than learning things according to another person’s attention is learning things by mind reading, that is by grasping the purpose of another person’s actions. Is mind reading even possible? The experimenters performed a little test. They tried to do something, such as pound pegs with a mallet, and failed, then asked for help. Could a child do what the experimenter tried to do but couldn't The trick here is for a child to see the failed action and to understand what the purpose was. For example, a person pounds a mallet near a peg but without hitting the peg. Can the child understand that the point was to hit the peg? The quick answer is that even toddlers can do that when the action is as routine and familiar as hitting toy pegs with a mallet. Autistic children are not as skilled as the normally developing ones, but they can do it too.
The more interesting responses came when the task was contrary to expectations. In these cases an adult would try to do something odd like sprinkle water from a watering can on a truck or stack a block on top of a tree. Again the adult failed and asked for help. Could a toddler realize the adult was trying to do something unfamiliar and contrary to expectations? Yes, they were not as good as they were at helping in familiar tasks, but they could do it. Autistic children, by contrast, were very poor at this task. Only 15% of them managed the task. (I suspect, however, that this score is better than one that chimpanzees or bonobos would compile. Experiments to confirm or disprove that notion would be welcome.)
The ability of normally developing human children, at a young age, to put themselves in another’s shoes and understand what others are trying to do, is—upon reflection—a remarkable skill that we should not take for granted. Could this ability to read minds lead to learning? To test the ability of children to learn words by mind reading the experimenters had an adult hunt through a purse for a parlu. What’s a parlu? The kids didn’t know. In the sack were a couple of familiar items like a ball and a hammer and four unfamiliar items. The adult would pull out, say, a ball and report, that’s not the parlu, and put the object back in the purse. In this manner both familiar objects and three unfamiliar objects were taken out of the purse, shown to the child, and declared not to be the parlu. Could the child hunt through the purse and find the parlu?
Even normally developing children could only do this one about half the time and autistic children had a success rate no better than chance would predict. But the fact that normal toddlers are already well on the way to learning words by seeing beyond the facts of the moment and reading intentions—oh, this strange object must be the thing the grown-up couldn’t find—is an important observation.
One of the most famous arguments for linguistic innateness cites “the poverty of the stimulus.” Children seem to know every so much more than the world shows them, and at the level of stimulus-response psychology, the argument is convincing. But these new experiments show how slight a “stimulus” needs to be. Empathy and attention to others lets us understand so much that is literally absent from the stimulus and yet intuitively present. The innate quality seems not linguistic but a sensitivity to others. That commonality speeds up word learning dramatically. Speech evolution was never going to get far without it.
(There are several other new articles about word learning and I will discuss them next week.)