Perhaps the best known fact about human evolution is that recently our brain has gotten much bigger. Starting about 2.5 million years ago, and following a reasonably straight slope, it continued growing until about 200,000 years ago. Australopithecus brains were about the size of modern chimpanzee brains. With the appearance of the Homo lineage the brain started growing and kept it up until the arrival of ours truly, Homo sapiens. Surely this development must have included something about language, but what?
The June 1 issue of The Neuroscientist gives much food for thought in a paper by Elliot D. Ross titled, “Cerebral Localization of Functions and the Neurology of Language” (abstract here). The paper reviews how the brain works and examines the production of language in detail as an extended example. Most importantly, it shows how fully integrated language production is in the brain. It makes the idea that language might be some sort of big-bang hopeful monster as almost childishly naïve, and it also supports the argument that this level of integration puts an older date to the origins of speech.
Many animals, including insects and fish, have well-established systems of vocal-acoustic communication, but only humans are endowed with … language … If one takes an evolutionary approach to human language, its functional-anatomic basis should be viewed as an adaptation resulting from the encephalization … of communication systems that in lower animals are heavily represented in non-neocortical areas (p. 228).
I have many doubts about the assumptions that underlie this paragraph. I don’t think human language is functionally equivalent to the vocal-acoustic communication of other animals, and furthermore we have the argument in favor of gestural origins that, like laughter, ape vocalizations are not voluntary and turning them into voluntary speech required more than encephalization. At the same time we have to face Dr Ross’s fact: at its most fundamental level, speech rests on subcortical auditory-vocal controls, not gestural-visual ones.
A whole course of further study suggests itself before the matter can be resolved, but part of this blog’s spirit is to take what we know now and see what we can make of it, knowing that later on we will surely make something else of it. The suggestion I take from this is that if language rests on some earlier vocal-acoustic system, it must have been a system unknown to the apes. It is not a very radical idea because we already see that in modern humans. Infants, even deaf and autistic one, go through a babbling period of making emotional sounds that encourage bonding but have no semantic function.
Taken as a whole, “human language is a distributed process that actively engages both hemispheres” of the brain . Taken in a more piecemeal form, the brain appears to divide speech into four broad functions—articulation, comprehension, semantics, and prosody—each of which is managed by a different very large portion of the brain. Ross calls them “emergent properties,” which I take to mean something that occurs as a result of some activity rather than something that is present at all times.
- Articulation: “an emergent property of a convergent-parallel (large-scale) network with Broca’s area serving as its critical nodal point” 
- Comprehension: “an emergent property of a convergent-parallel (large-scale) network with Wernicke’s area serving as its critical nodal point” 
- Semantics: articulated and comprehended meaning emerges from network activity in the brain’s left hemisphere.
- Prosody: articulated and comprehended emotions emerges from network activity in the brain’s right hemisphere.
It is important to understand that these functions are not part of a modular system in which different functions are handled by separate areas. There is extensive overlap and when speech breaks down in some pathological way the “majority of clinical syndromes are best understood as disconnection syndromes rather than primary processing deficits” . Articulation and comprehension, which in artificial intelligence studies, are typically called production and parsing and kept quite separate, overlap in many brain areas, according to Dr. Ross
The implication here is very important. We did not evolve a series of modules for language (or, really, for any major cognitive activity). Instead, we evolved more ways to form associations within a specific sense and a complex series of connections between the senses (cross-modal associations):
…humans have the capacity for processing and stories cross-modal associations between various sensory attribute areas because of the dramatic expansion of the parieto-temperol-occipital heteromodal neocortex. This expansion allows for greater cognitive flexibility compared to other primates by permitting more abstract types of representations to occur. 
Taking Ross’s paper along with Terrence Deacon’s planned presentation at Utrecht (see Grand CRU d’Utrecht) and Isler and van Schank’s work on brain growth (see The Riddle, the Mystery, and the Enigma) this work suggests to me a story about the evolution of the human brain and language:
- Before the brain began to enlarge, a system of babbling-type vocalizations emerged and was supported by subcortical regions supporting vocalization and hearing (discussed earlier in this post).
- The brain was only able to grow larger when the Homo line became a cooperative species where the default level of selection was the group rather than the individual (see post on Isler and van Schank’s work).
- The physical reason the brain increased in size was probably a duplication of some portion of the genome responsible for creating cortex that allowed neural growth and new links between regions (see post on Deacon’s work).
- The result of the growing brain was an increase in cortex for handling sensory modalities at more abstract levels and connections between modalities (discussed earlier in this post).
Still missing from the story is the way attention works (but see Voluntary Redirection of Attention) and how the brain combines separate foci of attention into a single sentence. But progress is in the air … even if the story outlined here proves to be hokum. Disproving it would be invaluable, just as disproving the localized module story has marked a great stride toward an answer.