Yesterday’s post listed three ideas developed on this blog that seem open to investigation. They were:
- Attention triplet
- Cognitive advances yield syntactic advances
Yesterday’s post looked for evidence that neoteny (delayed maturation) lies behind the evolution of the human capacity for speech. Today’s post looks at the proposed basis for that capacity, the attention triplet, and asks what challenges might be raised against it.
Linguistically, the attention triplet consists of a speaker, a listener, and a topic. The triplet can be generalized for non-linguistic situations: two individuals aware of each other and a third thing. Social predators sometimes show this joint attention, as when a mama lion holds a small gazelle while a cub administers the lethal bite. It also seems to play a role when lions stealthily surround a target. There seems to be more going on here than just a group of hunters paying attention to the same thing. Their tactics seem to be taking each other into account at the same time.
Primates engage in bits of joint attention as well, and probably more often, working together for various shared tasks. But the attention triplet among non-humans is limited to routine tasks. Lions are not going to join attention for anything beyond predation. Primates too stick to their routine. They are not curious about each other’s thoughts.
In the famous effort to teach sign language to a young primate, wonderfully named Nim Chimpsky, the results found that Nim’s “utterances” (hand signs) were usually made in response to a trainer and that Nim was a very poor “listener,” interrupting his trainer much more frequently than human children do. (The abstract for that famous study is here.) In short, the Nim Chimpsky found that Nim was both a poor speaker, rarely initiating a bit of joint attention, and a poor listener, rarely joining in on the attention initiated by another. Yet most explanations for what is special about language have looked to something other than the attention triplet, notably names, symbols, syntax, and mother-infant interactions.
B.F. Skinner maintained that naming was the essence of “verbal behavior” and many other thinkers have thought that naming was language’s decisive feature. The chief argument against this idea has been that speech in infants begins holophrastically. Instead of saying names, infants speak single words that express a complete idea. When a thirteen month old smiles and says, “Daddy,” he is not naming something but expressing the complete idea, “There is Daddy.” It seems impossible to test this idea.
Name recognition does not appear to be sufficient to support language. On a trivial level, pet dogs learn their names and can learn a variety of command words, yet they make no linguistic progress. More seriously, apes have learned to use over 100 separate words appropriately and yet they do not progress to language either.
A classic study by Michael Tomasello and M. J. Farrar (abstract here) showed that joint attention precedes learning names. When an infant’s attention was directed toward an object and then given the object’s name, the infant learned the word much more readily than when the name was given before the directing the attention to the object. “See the ball,” and pointing is not as effective a teaching device as pointing and, once the infant is focused, saying, “Yes, it’s a ball.”
Symbols are names and then some. Most theorists who focus on language’s symbolic nature would probably agree that the first symbols are names of things, but then symbols move on to represent concepts, emotions, and things even more remote yet.
Many people have argued that it is impossible to reason through a problem without language. This is debatable on its own grounds because many people seem to think visually. Einstein denied that language played any part in his creative thought. In the past quarter century experimental work has also gained ample evidence that chimpanzees can do many of the things previously believed to be the fruit of linguistic thinking. They can solve simple arithmetic problems and play video games requiring an appreciation of numerical order. My favorite bit of information in the press release about this research was the news that monkeys “clearly enjoy” playing video games. (See here.)
Many philosophers distinguish between names and symbols. The name is a “token,” a physical formation, of a symbolic concept. Thus the English word “name,” the French “nom,” Latin “nomen,” and Swahili “jina” are all tokens for the same symbolic concept. This sort of subtle distinction is very appealing and seems in keeping with the subtle difficulty of language’s nature. The simplicity of the attention triplet is probably, in itself, a source of suspicion and doubt.
Many of the papers at the recent Stellenbosch conference assumed that symbolism was the tell-tale sign of language’s presence and thus argued for a very late appearance of language in human history because archaeology finds no archaic symbolism. Stone tools are over two million years old, but they did not carry markings, not even of the “William made this” or “Jack’s hand ax” variety.
Artificial-intelligence work assumes that symbolism is the necessary and sufficient foundation of language. Mechanical intelligence consists of using algorithms to manipulate symbols.
The attention triplet challenges all this work, not by dismissing it, but by saying it is insufficient. The work of investigators like David Premack has established quite clearly that primates can manipulate symbols, yet they do not speak. Plainly, the classic philosophical position that language depends upon the ability to name and use symbols is insufficient. You cannot have language without that ability, but you need more.
This blog holds that the attention triplet is that more, but tomorrow we will look at a few other alternatives that people have proposed.