As I reported last week, the latest issue of Human Biology is devoted to language origins, with a particular focus on "genetic and cultural" approaches. In an afterword to the issue Tecumseh Fitch addresses a theme running through the whole issue, coevolution. Fitsch sees coevolution as the way to get past relying on an overly sharp distinction between biological and cultural explanations for language origins:
All this is undoubtedly true (although just the other week, a commenter objected to the idea that human children "effortlessly master language"). And yet, if we want to understand the coevolution of genome, culture, and language, we need some techniques for thinking clearly about the differences. In fact, a central drama in the story of language origins is the arrival of this new way of coevolving.
Coevolution is not a new idea. Famously, bees and flowers coevolved. Bees adapted to flowers, becoming more efficient pollen collectors. Flowers adapted to bees, becoming more efficient pollen donors. But both sides of that process were genetic. What is new in the human story is that the cultural and verbal parts of the evolving does not include genes. Call them what you want—memes, symbols, artifacts, words, or ideas—this is a new kind of coevolution, one that is only metaphorically like the relationship between bees and flowers.
What is culture? The simplest definition is that it is any kind of non-genetic behavior that is passed along from generations without, but that kind of behavior can be classified as social learning. Culture is more complex.
A critical difference between cultural evolution and ordinary social learning is that social learning directly supports genetic appetites. I'm hungry (a biological condition); I use a stone to smash open a nut (use social learning to satisfy a biological need).
Contrast that with going to a bookstore to get a book on how to use some new piece of software I've received. Eventually this chain of dependences may lead us to some biological yen but much has come between the genes supporting that biological need and the "need" for a technical manual. Cultural and linguistic coevolution works on levels that do not satisfy an immediate biological need.
Eventually there can be so many in-between levels that people contradict biology entirely—e.g., embracing death in martyrdom, fasting to the point of near starvation, and choosing to live a life without sex.
The oldest archaeological evidence of cultural levels slipping between biology and behavior comes from about 2.8 million years ago. Oldowan technology used a hammerstone to make a cutting flake. The flake was then used to slice food off an animal. From a cultural perspective, the key invention is not the flake, but the hammerstone. The flake satisfies a biological need. The hammerstone satisfied a need for flakes. It is a small distinction, but evolutionary beginnings always reflect small steps.
Does making a hammerstone require language? Probably not. The prerequisites would seem to be (a) the use of stones to satisfy a biological need and (b) an intelligence strong enough to imagine the value of a hammerstone. Chimpanzees today appear to meet requirement (a) but not (b); however, the first Homo of 2.8 m.y.a. had slightly bigger brains than either their Australopithecus forebears or other apes. On the other hand, the maintenance of a toolmaking tradition is the sort of thing we use language for. Was the first word hammerstone?
Language's prerequisites include something analagous to Oldowan prerequisite (a): the need for some existing system for grabbing attention that can grow into verbal control over a listener's attention. Gesture, and possibly vocalization as well were already available to the Homo of 2.8 m.y.a.. As for Oldowan prerequisite (b), it probably takes no more intelligence to think of a new word than it does a new tool. But the original language users needed a third thing (c), a willingness to direct attention to a topic distinct from both the speaker or the listener. Without that you cannot have the speech triangle, which on this blog is the foundation stone of true language.
The cultural evolution of artifacts proceeded slowly (by our contemporary standards). Oldowan tools reigned for almost a million years before being replaced by Acheulean tools that were worked on both faces. These too ruled for nearly a million years before advanced Acheulean "axes" appeared, at which point the pace picked up. The transition to Acheulean axes did require the insertion of a further level into tool making. The artisan makes an edge on one side, as in Oldowan technology, and then on the other side.
During that same period of time the primary change in the Homo body was a remarkable increase in brain size. Some experimental work reported previously on this blog found that even today in modern people, making Oldowan tools uses the oldest part of the brain available to Homo habilis. (See: Toolmaking and Speech) This detail suggests what common sense also assumes, that as the brain got bigger it could perform more sophisticated tasks. The question is how precisely the changes in culture and brain are related. Does the brain develop an ability and then culture comes along to use that ability, or does culture evolve a task and then the brain tags along to provide the neural support?
It used to be taken for granted that the brain had to move first, but then came evidence that tonal languages can be learned and spoken perfectly well by people who don't have the genes that seem to support tonal languages. Dan Dediu, one of the authors of that original study, has a paper in the Human Biology special issue addressing the question "Are Languages Really Independent from Genes?" In his answer he insists "on the concept of genetically biased language change" and argues that genetic biases "are essential for a proper understanding of language evolution, change and diversity." 
This argument is a loosening of coevolution, going from genetic determinism in the individual to bias in the population. In classical coevolution (as described by Deacon), cultural change C1 leads to Brain change B1 which leads to C2 and B2 until you have a string C1--> B1 -->C2 -->B2 -->C3… It is a somewhat slow process as the gene for B1 has to be fixed in the population before we can move on to C2. If the cultural change happens before the gene has spread through the whole population we will start seeing speciation, one variety can use C2 while the other variety cannot.
But if the absence of a gene in an individual can be made up by learning and intelligence, as happens in the case of using tonal languages, then whether or not a particular individual has the gene does not matter so much. The genes of the population bias the population's culture, but do not determine an individual's cultural behavior.
This appears to me as coevolution with selection going on at the group level. For several years now this blog has sided with E.O. Wilson, David Sloan Wilson and others in favoring multi-level selection over gene-only selection. But I've never wondered about how multi-level selection might work on in coevolution.
At the individual cultural level, each of us must learn a language and skills that will enable us to get along and survive. How would group-level coevolution work?
In multi-level biological evolution, the fittest group survives. Thus there tends to be evolution of pro-group emotions (e.g., shame and guilt), pro-group behavior (e.g., loyalty and steadfastness), and shared aids to cooperation (e.g., signals). In multi-level cultural evolution, the fittest culture will survive. Thus, cultures will tend to evolve education (for the teaching of traditions), technical vocabulary (for the mutual consideration of cultural details), and respect (for the bearer of traditions),
One thing that leaps out from these lists is that the coevolution to support these things is not at all like the traditional lists of what we needed to evolve in order to speak languages, make artifacts, preach morality, etc. Instead this process suggests that coevolving genes need to lend support to teaching, considering details, and respecting the wiser. Meanwhile, cultural processes needed to lend support to group emotions, group virtues, and cooperation. We don't normally think of those things as part of an evolutionary story, but they do seem necessary for a species that lives by learning what others do.