To read theories of language development is to do a particular kind of anthropology. Imagine a village in some remote part of Borneo, and a total eclipse of the sun on a bright spring morning.
‘The sun is dead! The sun is dead!’ will shout the children and the more excitable women, whilst running about in panic.
‘The sun has been extinguished,’ the more earnest young men will report to the leaders, as calmly as possible, because at times you have to report to someone, even when they have just seen it for themselves.
‘It’ll be back. It always does,’ will opine someone’s grandmother with equanimity and the air of infinite sagacity that comes with knowing you’ll be dead before you can be proved wrong.
‘The Gods are angry, you must sacrifice and obey us,’ will say the priests, with slightly overtheatrical authority.
‘Listen to the priests, they know what to do,’ will proclaim the King, in the tone of one who will know what to do with the priests if anything goes wrong.
‘Gather round, children, and I’ll tell you where the sun has really gone,’ will say the only old man the young listen to, tapping the side of his nose, and mentally organizing stories about dragons and celestial fire.
And none of them in fact has the faintest idea of what is really happening.* The available information about the subject of language evolution is so limited, even by the standards of palaeoanthropology, that anything you say is as likely to be right as it is wrong, and is untestable by any normal scientific means.
Just to give a taste of how tricky this is, there are no clues in modern language as to how it developed. All known languages are equally complex, capable of expressing anything that anyone might want to say in them. There is no such thing as a primitive language, nothing that might permit us to study older, developmental stages.
Cognition doesn’t fossilize. We can measure the size of the brain by taking endocasts of skulls, but they don’t tell us much about its structure, and given that there are many animals with much larger brains than ours, size may not be as important as we tend to assume, even in the brains of our close relatives. The Neanderthals had slightly larger brains than we do, but it is not clear that they had greater cognitive ability, and the current assumption is that they didn’t. The detailed structure of the brain that allowed for levels of cognition capable of producing language might have evolved very recently.
The larynx and the hyoid can fossilize, but to say that a given hominid has a lowered larynx similar to ours, allowing greater voicing and control of sound produced, or that it has the bone that we have at the root of the tongue doesn’t mean that Homo neanderthalensis could do what we can do with it.
And the theories are mostly wild speculation, shots in the dark. A darkness the size of Wales, illuminated by half a dozen Zippos. At least we’ve got beyond the Heave Ho and the Bow Wow theories, which for some time were taken rather more seriously than their originators intended, partly, perhaps, because there wasn’t much else. Here and here are a couple of papers which summarize the state of play.
Nowadays you will hear of the attention triangle, universal grammar, the search for recursion encoded in the genome, and similar ideas. Observation of how children learn suggests the stimuli they are exposed to may be insufficient for them to identify how language works, and so it must be at least partly genetic. Others disagree, saying that the amount of information they are exposed to is far greater than is supposed, and is sufficient for the purpose.
Is language an entirely learned behaviour, a cultural artefact like art, humour or music, an accidental by-product of our high level of cognition, a spandrel? Would two babies brought up by a deaf-mute on a desert island develop a language and communicate through it? There is anecdotal evidence of children brought up in a group without any stimulus who did indeed develop communication systems. And similar anecdotal evidence suggests, on the other hand, that such a child on its own will not learn to speak, which further suggests that the mind wants to communicate with others, not with itself. But all of this is little more than speculation, and the experiments, though perfectly practicable, might meet some social resistance.
There is nothing in nature remotely comparable to human language. No other animal communicates with syntax. Chimpanzees have been taught to express their immediate desires by pointing or making signs, but they have a very limited range, no spontaneity and no ability to combine ideas or represent the abstract, nor any interest in doing so. Animal communication bears no discernible relation to ours, beyond the trivial fact that they are aware of each other’s existence. It tells us nothing about the nature or origin of speech.
Noam Chomsky’s answer to this problem was a Deus ex machine. (He does the same with politics, stating that the solution to the world’s problems is x, and ignoring reality altogether.) He decreed that there must exist a Language Acquisition Device in our brains**. However I process this idea, it refuses to become anything but a name for the problem it is supposed to solve.
So we know almost nothing about how language evolved, but it exists, and it is extraordinary, and we would very much like to know how it works and where we came from. It might take a while.
*It’s like listening to journalists talking, or reading the comment columns in newspapers. Even now, with interesting things happening in Egypt, we get people writing about it from thousands of miles away who are simply inventing stories based on their own and their paper’s prejudices and what they can pick up from Al Jazeera and Twitter. Well, we can all do that, we don’t need them. There are one or two who claim to be in Cairo itself, but they’re talking to the protestors, and hearing only what each individual protestor thinks it’s about, which is not lacking in interest but it doesn’t tell us anything about the political and social situation and gives no clue as to where it all might lead.
**The WP article isn't very gfood but I can't link to the original paper or find a better descripotion of the concept.