How Language Began

by Daniel L. Everett

  Every neuroscientist worries about whether their own brain is up to the task of understanding the human brain more generally. Some, like Ramón y Cajal in the quote at the beginning of this chapter, think not. Others believe that such pessimism is unwarranted and that one must press on, making gradual progress, as with any difficult object of study. Optimists believe that full understanding of the brain can be achieved homeopathically – a little at a time, learning a tiny bit more about the brain experiment by experiment, as everything in science is learned. The banal fact that the brain is necessary for language doesn’t shed much light on how the brain functions in language.

  For some of these reasons, interest in the brain has grown tremendously over the past fifty years. In 1970 the Society for Neuroscience was formed, with 500 founding members. As of this writing, the society has more than 35,000 members worldwide and holds conferences with an average audience of 14,000 for presentations and a total of 30,000 attendees. The philosophy of neuroscience is another recently popular field that, according to most of its practitioners, originates in the 1986 book Neurophilosophy by Patricia Churchland. There is tremendous variation in beliefs, theories and research interests in both of these fields of study, but one of the emerging points of consensus among many (though by no means all) is that brains are simply one organ of the entire individual’s physiology and that cognition, activity, ascriptions of abilities and so on are properties of the entire individual, as well as the culture in which that individual is found.

  Part of the reasons for neuroscepticism, the belief that humanity will never understand the human brain, is expressed in this passage:

  Gram for gram, the brain is far and away the most complex object we know of in the universe, and we simply haven’t figured out its basic plan yet – despite its supreme importance and a great deal of effort … No Mendeleyev, Einstein, or Darwin has succeeded in grasping and articulating the general principles of its architecture, nor has anyone presented a coherent theory of its functional organisation … there is not even a list of basic parts [of the brain] that neuroscientists agree on.11

  What evolved in the skull from the earliest primates to Homo sapiens is an organ of the body. The brain is not, nor does it contain, an ethereal entity such as a mind or a soul. The brain is nothing less nor more than the main organ of the nervous system, as the heart is the main organ of the circulatory system, or the lungs are the main organ of oxygenation, or the nose is the main organ of smell, or the eyes the main organs of vision. The brain cannot live or develop on its own. It is, like any corporeal organ, shaped and constrained by all the other physiological systems of the body, in cooperation with cultural experiences, individual apperceptions, the food we eat, the exercise we take and, generally, just the way we have lived. The brain is found in the head, which itself has evolved to fit and protect it, changing as the brain changes and essential to its proper functioning.12

  Much of what is known about the human brain has been learned from experimentation on the brains of live animals. For human brains, less-cruel methods are used, such as functional neuroimaging of different types and electroencephalogram (EEG) recordings. These more humane methods of brain study have provided a multitude of insights into how the brain supports human language.

  The brain’s three pounds consist of neurons, glial cells and blood vessels. Each plays a vital role in proper brain functioning, intelligence and other cognitive abilities of the species. There are some 100 billion neurons in the average brain. There are also non-neuronal cells of a roughly equal amount. Nearly 20 per cent of all neurons in the brain reside in the cerebral cortex, including white matter found beneath the cortex or ‘subcortical white matter’.

  The biggest part of the brain is the cerebrum (from the Latin word for ‘brain’), which sits underneath to the rear of the cortex (literally, the ‘bark’ of the brain). The cerebrum is broken into two hemispheres. When someone talks about being ‘left-brained’ or ‘right-brained’, they are referring to the two hemispheres of the cerebrum.

  As we can see from the ventral (underside) view of the brain shown in Figure 18, what in turn sits under the cerebrum is the brain stem. Behind the brain stem is the cerebellum. Nearly 69 billion of our brain neurons, 80 per cent of all our neurons, are located in the cerebellum or ‘little brain’, which sits just below the cortex.

  The cerebral cortex is convoluted (lots of ridges and valleys), a common feature of larger brains, regardless of species. The brain is soft and, were it not for the skull housing it, would be very easy to damage. Many components of the human brain are found in all vertebrate brains. These include the medulla oblongata, pons, optic tectum, thalamus, hypothalamus, basal ganglia and olfactory bulb.

  The major human brain divisions of forebrain, midbrain and hindbrain are likewise common throughout the animal kingdom. Mammalian brains are more advanced than the average vertebrate brain, however. All mammals have a six-layered cerebral cortex. As primates, humans have larger cortices than the brains of non-primates and, like other primates, their brain shape has been mildly affected by the fact that they hold their heads upright.

  But humans are not merely vertebrates, mammals and primates – they are hominins, possessors of the largest brain relative to their body size in perhaps the entire animal kingdom. But in particular they possess the densest packing of neurons of any creature. Human brains are formed through gene-environment interactions. There are barriers (some of blood and others of cerebrospinal fluid) around the brain and specialised cells (glial cells and mast cells) that offer an independent neuroimmune system, distinct from the immune system for the rest of the body. Humans share much of our brain structure with other hominins, such as Australopithecus and Homo erectus, though sapiens brains are larger, more specialised and more complex than those of other hominins.

  If one were to flatten out the folds of an entire anatomically modern human’s cortex on the floor, it would have a surface area of approximately 2.6 square feet, or 0.24m2. The folds of the human cerebral cortices have raised and lowered portions, ridges and grooves, each called a gyrus (a raised portion, from the Greek for ‘ring’) or a sulcus (an indented portion, from the Latin for ‘furrow’ or ‘wrinkle’).

  Each of the cerebral hemispheres is divided into four lobes (Greek for ‘pod’): frontal (Latin ‘front’), parietal (Latin ‘walls of house’), occipital (Latin ‘back of head’) and temporal (Latin ‘lasting for a time’). These are named in the main for the corresponding portion of the skull immediately above them. One sees references to these lobes frequently in the neuroscientific literature, but their primary purpose is to refer to a general region. They are too poorly understood and demarcated to attribute to them any significant function in the overall behaviour of the brain. In fact, different brain functions are commonly distributed across the supposed borders of the different lobes. An added complication is that every human brain is unique – no two brains have exactly the same pattern of gyri and sulci. Yet human brains all function similarly, in spite of gyri and sulci patterns. This indicates either that the functions of brain folds are not fully understood, or that the brain folds make little difference to brain functioning, or that there are brain characteristics that distinguish individuals in ways not yet identified. All of these options are in all probability true in some sense.

  Figure 17: Midsagittal view of the brain

  Figure 18: Ventral view of the brain

  Figure 19: Dorsal view of the brain

  However primitive our current understanding of it, it is clear that the brain’s anatomical architecture is very important to language and other cognitive functions of our species. But there is another kind of architecture beyond the gross anatomy which is equally important, if not more so, to brain functions: its cytoarchitecture.

  In the midsagittal view of the brain (Figure 17), the brain stem, the cerebellum and the cerebrum are all visible. What is worth noting in particular in this diagram, however, is the ‘lie of the land’ and the
fact that specific regions of the brain are where we most commonly find the control centres of certain physical abilities. The fact that these abilities are found in most subjects does not necessarily mean that they cannot be repurposed for other uses. The cognitive scientist Elizabeth Bates and her colleagues in fact developed a model of brain specialisation that relied on the order and manner in which things were first learned and the physical nature of the relevant cytoarchitecture, avoiding in many cases the need to call upon unidentified genes. This is not to say that genes do not play a role in the architecture of the brain. Of course they do. But they should not be invoked without understanding their role. Two other views of the brain that illustrate its overall complexity are the ventral (lower) and the dorsal (upper) views (Figures 18 and 19).

  More detailed representation of the cognitive functions of brain areas come from ‘cytoarchitecture’. This is the cover term for the differences in the construction of individual cells found in particular brain regions. The division of brain regions based on this cytoarchitecture is given in the Brodmann classification of areas of the cortex distinguished by cell forms (Figure 20). This organisation is manifested in various distinct ways – connections between cells, shapes of cells or cell parts, thickness of the cortex in a particular cellular region, according to which region of the cortex we are viewing and what function that region has. It is named after Korbinian Brodmann, the first to propose this type of division of the brain. Most of the divisions he proposed are linked to parts of the body or higher cognitive functions.

  Figure 20: Cytoarchitecture/Brodmann areas

  Based on the cytoarchitectural division of the brain into Brodmann areas, recent proposals searched for specific linguistic regions of these areas that might explain language’s uniqueness to humans. Some of these proposals assume a view of the neurological basis for language of sapiens that is somewhat problematic. With all proposals on the anatomy or evolution of the brain, one should be suspicious of attempts to interpret neurology as being either predicted by or supportive of highly specific theories of language. In particular, Brodmann’s areas offer no direct support to the ‘X-Man’ view which takes language to be a special, saltational change resulting from a mutation.13 Considering this proposal and rejecting it paves the way for a discussion of more general properties of the brain necessary for language.

  The proposal is partially built off the fact that there have been multiple evolutionary innovations in the human brain. Thus some researchers have claimed that there are differences in the sapiens brain resulting from a mutation for grammar. The mutation proponents also claim that no other species past or present, such as Homo erectus, could have had language. This is because such researchers claim that there is no evidence for symbolic representations among Homo erectus or, probably, even Homo neanderthalensis. Therefore, only sapiens could have had language. This is a doubly erroneous idea. The two problems with it are clear enough. First, erectus did have symbolic representations – tools, status symbols and etchings. Second, the method is wrong. It entails a claim based on the absence of evidence, namely that symbolic art is not found in non-sapiens Homo species. But absence of evidence is a weak argument. And in this case there is evidence. Possibly third, this idea that erectus lacked language ignores their accomplishments. Remember that erectus sailed. This activity alone demonstrates that erectus was able to think ahead, to imagine and to communicate. Even if erectus’s language lacked hierarchical syntax, this would differ only in degree and not in kind from modern languages. In other words, the language of erectus, what might be called a ‘G1’ (early grammar) language, could evolve seamlessly into a contemporary ‘G3’ (later grammar) language – one with recursion and hierarchy. And after the G1 language was invented by erectus, and quite possibly adopted subsequently by neanderthalensis and sapiens, G3 arrived after little more than a few years of tinkering (could be thousands of years or dozens of years or only a few years). Some researchers have correctly demonstrated that the evidence for language goes back at least to Homo neanderthalensis, some 500,000 years ago. But there is no reason that language could not reach back even further. Like sapiens, neanderthalensis were beaten to the invention of language by their ancestors, Homo erectus.

  Some problems attendant to the proposal that the brain has specific encapsulated modules dedicated to language can be seen by briefly reviewing research reported by neuroscientist Angela Friederici of the Max Planck Institute for Human Cognitive and Brain Sciences in Leipzig. Friederici has written extensively on her research on hypothesised brain mechanisms underlying language. She adopts assumptions on the nature of language as a specific type of complex grammar. The claim of hers worth scrutinising here is the proposal that Brodmann Area 44 (BA 44) is a functionally specific (that is dedicated to a specific task) range of tissues (BA 44 is visible in Figure 20). Friederici claims that this area is connected to the temporal cortex and that this is the neural difference that triggers or enables the capacity of Homo sapiens for syntax. Before examining this bold claim, a simple question is in order. What did the researcher believe about their subject before beginning their research? This question can be helpful for detecting the early signs of ‘confirmation bias’ (obtaining experimental results that confirm one’s pre-existing beliefs).

  This particular research begins by assuming the narrow view that language is a grammar with recursion. No communication system lacking this property can be a language, according to this perspective. This assumption is made in many experiments on the cerebral support for language. In many cases the researchers assume a grammatical rule called ‘Merge’ as the basis for all human language. Merge is simply an operation that combines two objects to form a larger object. Assume that one wishes to utter or interpret the phrase, ‘the big boy’. This phrase is not formed by simply placing the words ‘the’, ‘big’ and ‘boy’ together. In the Merge theory of grammar one takes ‘big’ then combines that with the word ‘boy’. Then ‘big boy’ is combined with ‘the’. As it is designed, the rule of Merge cannot work with more than two words or phrases at a time. This means that it, and therefore all of language is, a ‘binary’ procedure. This is a conception of language as structure, rather than as meaning and interaction.

  In making the assumption that human language = Merge, researchers arguably overestimate the importance of syntax, which plays only a minor role in human language as a means to organise information flow. As such, it acts as a filter, to help guide the hearer to the intended interpretations of an utterance.† Grammar is symbols used together. Syntax is the arrangement of those symbols as they are used together. Compared to the invention of symbols and the cultural basis of their meaning, however, along with the knowledge of how to use symbols appropriately in telling stories, conversing and using language in its many forms, syntax is a helpful tool, but arguably non-essential. The problem is that the operation Merge is not found in all human languages. In several modern languages it has emerged that there is no evidence supporting a binary syntactic operation.‡ Merge is therefore not a prerequisite for human language. Moreover, even if it were, Merge is not specific to language. It is an example of the well-known process of associative learning, of the type that made Pavlov’s dog famous. Pavlov’s pooch learned to associate a ringing bell with food. The bell was rung just as the food was served. Eventually the dog ‘merged’ these two concepts, bell and food, and would salivate at the sound of a bell.

  Yet if Merge is simply a form of associative learning, then it cannot be linguistically unique. Associative learning is commonly found both outside of syntax and language and across most species. Arguably, all animals possess a capacity for associative learning. So this is the first problem with Friederici’s claims. She has the wrong concept of syntax.

  But the conceptual neurolinguistic problems don’t stop there. It isn’t just the erroneous concept of syntax adopted that damages attempts to show language specificity in the brain. The main problem is not merely the idea that there is innate l
anguage that is a simple syntactic operation, but the idea that one could locate an innately hardwired place for language in the brain. That idea is like asking the brain, ‘Where’s the syntax?’ Well, of course, syntax is in some place or places in the brain, yet just because syntax is found in the brain does not mean that it is genetically prespecified to be there. Moreover, it is possible that some areas of the cytoarchitecture of the brain are indeed propitious for storing or operating syntax. But that is not the same as saying that BA 44 and its temporal lobe connection are the evolutionary developments that pulled language out of a primate brain. BA 44, like any other region of the brain, has a broad range of functions and is not exclusively associated with syntax. BA 44 has at least six separate functions, including sound or phonological processing, syntactic processing, understanding meaning or semantic processing, and music perception. It is also employed in responding to ‘go’ vs ‘no go’ decision-making. BA 44 is also used in hand movements. So BA 44 is not ‘for syntax’. It might be necessary for syntax, but that’s like saying that the hand is necessary for pencils. The different roles of BA 44 illustrate that we need a more general, higher-level understanding of what it does. We cannot call it a language organ.

  Work like Friederici’s errs not only because it assumes that syntax is more important for language than it in fact is but also because the relevant portions of the brain are not as specialised as this work assumes. Researchers of this persuasion also seem to suggest that language is simpler and younger than the evidence we have reviewed indicates. Not only is language much more complex than research like Friederici’s assumes but, if I am correct, it is also significantly older. Nearly 2 million years older. It predates perhaps even the evolution of relevant portions of the brain that she claims to be crucial for language.