Discovery Guides Areas


Language Origins:
Did Language Evolve Like the Vertebrate Eye, or Was It More Like Bird Feathers?

(Released December 2003)

  by Christopher Croom  


Key Citations

Web Sites



Review Article

Linguistics and evolutionary theory share an extremely tenuous historical relationship, as linguistics was more concerned with philology, rather than scientific observation, when Darwin first published the Origin of Species.1 The most important argument within contemporary linguistics and evolutionary theory was sparked by Pinker and Bloom's (1990) seminal analysis outlining comments made by Noam Chomsky and Stephen Jay Gould that contradicted the basis of modern evolutionary theory; this article led to an enduring debate that has persisted over the last decade.2 Since Chomsky and Gould have made a number of assertions that language (the communication system unique to human beings), could not have evolved through natural selection, and natural selection has long been the prevailing theory in evolutionary biology, the challenge presented by Pinker and Bloom was to develop a theory of language origin that was compatible with the mainstream theory of evolution, the theory of natural selection.3 Since then, however, research has provided evidence that some aspects of language may have been naturally selected for, in line with Pinker and Bloom's arguments, while other aspects of language did not result because of natural selection, thus also supporting Chomsky and Gould. The following explores and integrates the history, evidence, and theories surrounding both selectionist and nonselectionist explanations of the origin of language.

A Brief History of Evolutionary Theory

Evolutionary theory has been wrought with controversy since its initial appearance in public debate in the late 19th century, at which time numerous biologists and theologians generated a wide variety of theories of the origins of humankind. Only one of those has withstood repeated tests of scientific merit and has been observed numerous times in ecological systems; this has come to be known as the theory of natural selection, first posited in Charles Darwin's Origin of Species.4 This theory shows how certain physical traits among members of a given population of organisms will be favored by any number of unpredictable environmental conditions, to the extent that these individuals experience a reproductive advantage and are thereby able to concentrate the expression of those traits in future generations. Of course, the environmental conditions that nurture the proliferation of a trait in one generation may be absent in the following generation, thus conferring no reproductive advantage to the offspring that receive the trait; this may lead to the eventual disappearance of the trait. Darwin discovered all of this before the existence of genes and DNA was confirmed, so he did not actually understand how physical traits were transferred from one generation to the next, but he knew that advantageous traits must be transmitted. Darwin took inspiration for his theory from his observations of Galapagos finches with beaks that seemed to be in exact proportions to the size of the seeds that were most frequent on their particular island, yet otherwise were anatomically identical.5 Darwinian natural selection is now used to explain a large proportion of evolutionary changes in species, ranging from the development of the vertebrate eye, to the origin of grammar, and even all of language.6

Natural Selection and Other Evolutionary Forces

When DNA was finally recognized by James Watson and Francis Crick to be the molecule responsible for transmitting genetic information in 1953, evolutionary biology had finally found the substance that made natural selection possible. This buttressed Darwin's theory of evolution and lead to the discovery of other mechanisms that are compatible with, but lie outside the scope of, Darwin's original theory. The discovery of these processes expanded the theory of evolution and the ability of biologists to understand evolutionary change. The other processes that drive the differentiation of populations, and ultimately, speciation (in addition to natural selection) are mutation, genetic drift, gene flow, the process of co-optation, and something that Darwin called "preadaptation."7

Mutation involves random changes in the DNA of individual organisms, often caused by mutagens in the environment such as ultraviolet radiation, chemicals, gamma rays, microwaves, or radioactive elements. Since mutations that profoundly change the structure or appearance of an organism can be deleterious to the organism and may prevent it from reaching reproductive age, massive changes to the phenotype of an individual can be evolutionarily difficult to pass on to future generations. Genetic drift includes the stochastic, unequal transmission of genes to subsequent generations and the effects this process has over time. It is also used to explain the effects of random events removing certain genes from the environment and preventing their expression. Gene flow involves the removal of a certain element of the gene pool, usually due to migration, and the way the source and contact populations are evolutionarily affected by the removal and addition of different genes. The process of co-optation is a nonselectionist evolutionary force that accounts for the existence of nondeleterious, yet nonadaptive (neutral) traits, explaining such phenomena as the brooding chamber in snails.8 "Preadaptation," although first identified by Darwin, was poorly explained in the Origin of Species, and is now called "exaptation."9 Consequently, exaptation is not considered part of natural selection and is therefore non-Darwinian, although it is an integral part of modern evolutionary theory.

Besides natural selection, exaptation is the most important evolutionary process discussed in this analysis. One kind of exaptation happens when a population of a given organism uses a bodily structure or other adaptation for something other than its original evolutionary purpose. A classic example of exaptation from evolutionary biology, as supported by the fossil record, is the theory that birds first evolved feathers for the purpose of insulation, and that this function of the feathers was later adapted for gliding in order to catch insects more efficiently.10 Talking parrots, such as African Greys (Psittacus erithacus) also demonstrate how exaptation works when they use neural oromotor structures that were selected for within-species communication in order to produce words from human languages, even though this behavior would provide them absolutely no added benefits in their natural environment. 11

Exaptation also has another way of working, which has been coined "spandrels" by Gould and is a word taken from the architectural term for the tapering, triangular shape that occurs as a result of structural necessity when two arches are joined at a right angle in order to build something on top of them (such as a dome).12 In the context of modern evolutionary theory, a spandrel is any structure or phenomenon that exists due to a physiological by-product of something else. In the animal kingdom, it can be difficult to determine if something is a spandrel or occurs because of regular exaptation, but an example used by Pinker and Bloom is found in the case of an aquatic bird that uses its wings to shade the glare off the surface of the water in order to better see its prey.13 In this situation, the bird's wings are like spandrels because they manage to block light by sheer virtue of being solid structures that are also good for flying. Similarly, in the case of a human being using his or her mouth as a hand, the mouth becomes spandrel-like, because in order to work well for chewing and eating it must also be able to grasp bite-sized objects.14

Recent research in the origins of language suggests that certain neurological processes underlying aspects of language may be derived from cerebral structures that are not only present in hominids, such as Homo sapiens, but are also present in non-human primates whose ancestors have been on Earth for a much longer time.15 What follows from this is that parts of language were exapted from cognitive structures that our pre-human ancestors used for food gathering, rule learning, tool making, and hunting, among a vast number of other purposes.16 As a consequence of these recent discoveries, selectionist and exaptationist views represent a polarity among evolutionary biologists, between those who view language as purely the result of natural selection, like the vertebrate eye, and those who view language as the result of cognitive capacities that were originally used for something else, as bird feathers may have evolved.17

Problems with Studying Evolution

Evolution is extremely difficult to study for a number of reasons. Darwin theorized that human beings would probably never actually be able to witness evolution in action, because the changes would be so gradual that they would be imperceptible even to the trained observer. Fortunately, since Darwin's death, we have been able to observe evolution happen, from pepper moths in England, to antibiotic-resistant bacteria, to the African cichlids of the Rift Valley lakes. Human evolution is even more arduous to study due to our longer generation times which make major evolutionary changes difficult to observe; by the time our biologists would be ready to witness the changes, if any, in the next generation of offspring they would also be ready to retire, leaving less experienced observers to continue the project. We can, however, study isolated populations such as the Amish, and evaluate and compare their genetic changes (genetic drift) with larger, less insular groups of people.

The other major problem with studying human evolution is that the fossil record of our species is incomplete, often at the precise points in time that we most need data about. It is highly likely that none of our fossil collections of hominid ancestors include a statistically representative sample of their respective populations, as none of our fossils have been randomly collected and we do not yet have reasonable estimates of the total populations of these species. This means that the structural differences we have observed between various hominid ancestors could be caused by chance rather than speciation. Moreover, when it comes to the study of hominid ancestry, the only data we have as a basis for comparison are the structural differences observed from the skeletal remains of the fossil record, as organs and other "soft" tissues are extremely poorly preserved. Ideally, if we were trying to study the neurological changes that took place in human pre-history and later gave rise to language, we would want a large number of petrified whole brains representing the numerous different species of human ancestors from a wide range of time periods, and such fossils do not exist. Numerous inferences have been made in the study of human evolution that as cranial capacity, or brain size, increased in the hominid line, so did our cognitive abilities, but if brain size were always directly correlated with cognitive ability, we would have to acknowledge that cetaceans (e.g., whales and dolphins), have the greatest cognitive capacities of all animals since they have brains that are proportionately far larger than those of human beings.

Such a paucity of data can lead one to argue that any theories about language evolution are highly conjectural; indeed, a number of inappropriate uses of adaptive theorizing have appeared throughout the literature. Pinker and Bloom warn against these kinds of arguments and characterize them as "ad hoc" and "arbitrary." The example they provide is an anecdote about Voltaire ridiculing a colleague who is wondering why spandrels exist at the San Marco Basilica in Venice; Voltaire declares that the spandrels are there because they have to be, just as noses were made for our glasses to perch on and legs were designed for pants.18 This clearly illustrates the dangers of positing ad hoc or arbitrary evolutionary hypotheses, as cause and effect can become completely obscured.

Ad hoc and arbitrary evolutionary theories can also become self-fulfilling in the sense that if a researcher is actively looking for adaptations that perform some evolutionarily advantageous function, he or she can be lead by their biases into almost always finding examples. This is because practically every structure that is not vestigial or blatantly deleterious can ultimately be argued to perform some kind of function that is perceived to be advantageous to the organism that possesses it.19 By this kind of reasoning (also known as naive adaptationism), the extremely slow reproduction rate of the Giant Panda could be argued to be an advantageous adaptation because it might allow the mother Panda to extend her reproductive lifespan by conserving more energy, even though their slow reproductive rate (coupled with habitat destruction) is now contributing to the decline of the species. But ultimately in science, hypotheses are just hypotheses until definitive data have been observed, published, and debated in peer-reviewed journals. In short, the lack of data available to evolutionary biologists has certainly been the largest blight to research on the origins of language.20

Language and Human Pre-History

The study of the evolution of language, however, is not relegated completely to making hypothetical statements that may ultimately be impossible to validate. We have found a large number of fossils of human ancestors and pre-historic Homo sapiens, and many of the artifacts found with these fossils have given us volumes of information about the subsistence patterns and lifestyles of human ancestors and early humans. One prevalent theme in the hominid archaeological record is that large amounts of tools first started appearing in human history around 2.3 million years ago in the context of the first member of the genus Homo, who was actually named after these tool associations. Homo habilis literally means 'habile man', or handyman, as tools have never been found in such high concentrations with the earlier members of the hominid line, the Australopithecines.21

The evidence that Homo habilis was probably the first of the hominids to make sturdy, lithic tools strongly suggests that Homo habilis may have had greater cognitive capabilities than his precursors. Because of this, researchers have looked for any signs that the structure of the brain of Homo habilis is different, or more complex than the brain structure of Australopithecus. ince the soft organs have long since biodegraded away, the best current method to evaluate structural neural changes in hominid fossils is by taking cranial endocasts, or plaster casts of the inside of the brain case of fossilized hominid skulls. Botha and Givón both discuss research that suggests that certain cerebral structures started appearing for the first time in Homo habilis that are now known to be necessary for producing language.22 Apparently, according to the cranial endocasts, the parietal, occipital, and temporal lobes of the brain merge for the first time in Homo habilis, creating an area of the brain called the POT, or Wernicke's Area.23 Broca's Area, one of the other highly important areas involved in language production and processing, can supposedly also be observed for the first time in the hominid line from endocasts of Homo habilis skulls. The argument that follows from this evidence is that the Broca's and Wernicke's areas were selected for in Homo habilis in order to make tools (and probably also for gathering food and hunting), but were subsequently exapted by later humans for the purposes of language production and processing.24

Unfortunately, cranial endocasts cannot usually provide any conclusive evidence about structural evolutionary changes in the human brain due to the lack of representative populations of fossils (mentioned earlier), and because cranial endocasts can only be made well from craniums that can be pieced back together or are more or less complete, which are not as frequent in the archaeological record as disarticulated skull fragments. Moreover, cranial endocasts only show the structure of the surface of the brain, and only insofar as it is reflected by the contours of the inside of the skull. They tell us nothing of structural changes that may or may not have taken place deep inside the cerebrum itself, and they cannot show whether the individual possessing the skull had any lesions, necrosis, or other areas of damage caused by something that would not be shown on the skull (i.e., a stroke, as opposed to blunt head trauma).

The Brains of Other Animals: Exaptation and Language

There are better ways to get a sense of the kinds of cognitive abilities that early humans might have had that are more observable in both the fossil record and in modern, living animals. Hauser et al. have shown that one of our more genetically distant primate relatives, the cotton-top tamarin (Saguius oedipus), can learn to generalize algebraic rules, which is also a skill that human infants use in acquiring language.25 If this skill is important for cotton-top tamarins (which have a brain capacity about the size of a squirrel or a rat) for some aspect of living in their environment, it can be posited that some of the cognitive processes underlying language probably existed in the most distant human/primate ancestors. Since the common ancestor that humans share with tamarins has been extinct for millions of years, subjecting it to a battery of cognitive tests is impossible. Also, since this ancestor occurred so early in the history of primates, one of two outcomes becomes possible: either the human/tamarin coancestor needed the ability to generalize rules, or humans and tamarins (and all the other primates, for that matter) developed this adaptation independently because it has been favored in all the different environments in which primates live.26

Since the latter scenario is statistically far less likely and genetic testing has confirmed taxonomists' suspicions in pointing out a common ancestor of humans and tamarins, it is highly possible that the primate lineage had rule learning sorted out tens of millions of years ago. Most importantly, as tamarins and humans appeal to similar cognitive structures for this rule learning ability, and humans have language while tamarins do not, language definitely relies on cognitive processes that are useful for other things. This increases the likelihood that language, or at least the rule learning parts of it, were exapted the way bird feathers probably were.27

Another similar argument for the exaptation of certain aspects of language has been discussed only recently and comes from the optic tectum, a structure located in the midbrain of many reptiles and birds, such as barn owls (Tyto alba).28 The barn owl needs to be able to create a mental map localizing the source of prey sounds from purely auditory information, in order to effectively hunt in almost total darkness. An auditory-visual modality crossover similar to the barn owl's optic tectum is found in humans and not only enables us to visualize where a sound might be located in a space, but is also necessary for the modality shift required to code our lexicon.29 While bird and human brains are organized very differently, we have evolved similar cognitive processes for situating and relating sounds in space, which further suggests that the neural structures or pathways that make our language possible would surely be useful for other purposes.30

In fact, numerous cognitive processes that humans use for language are also found in non-human primates.31 Studies of evoked responses in human and macaque brains have shown that we use two basic neural pathways for language and cognition, and both pathways are generated from a point very close to the same basic region in the occipital lobe, known as Wernicke's Area in humans and the F5 region in other primates. The ventral pathway (or ventral stream) shared by humans and primates emanates from the occipital and proceeds through the temporal lobe, terminating in a region of the brain that includes the paleo-cortex and the amygdala. Both humans and primates use this pathway for object recognition and actions, although humans use it expressly for semantic memory, verbs, and the lexicon. The other pathway, which projects from the occipital region and progresses to the posterior part of the parietal lobe is called the dorsal stream, and it terminates in the subcortical limbic region of the hippocampus. This neural pathway is used by both humans and primates for processing motion and spatial relationships, but humans also use it for processing episodic memory and prepositions.32

If the neural hardware for so many aspects of language exists in non-human primates, then one should ask why humans are the only animals that have language. A casual observer might note that since we are more than 98% genetically identical to chimpanzees, then chimpanzees should be able to develop language also. Pinker points out that the hyrax (Procavia capensis), a very small rodent-like mammal that is 98% genetically identical to the African elephant, not only lacks a trunk, but scarcely resembles an elephant at all.33 Chimpanzees do have a rich system of communication, do use tools, and do have something that resembles human culture in the sense that different communities of chimpanzees teach their offspring how to subsist in different ways, depending on how their parents were taught, but they do not have language, a system that enables an infinite possibility of utterances.

Natural Selection and Language

Although Pinker asserts that the process of natural selection is more than sufficient to explain the evolution of the entirety of language, the recent research from Givón mentioned above shows that a large portion of our language abilities are derived from neural structures that an ancestor of ours probably already had, meaning they were probably exapted.34 However, exaptation cannot account for all of language, and the best evolutionary theory to explain the evolution of the facet of language known as grammar is regular old Darwinian natural selection.35

Studies of Pidgin languages and pre-grammatical child speech have shown us that vocabulary is always acquired before grammar;36 moreover, in cases of agrammatic aphasia (also known as Broca's aphasia, because the Broca's area is affected), the lexicon is preserved, allowing the patient to comprehend and actually produce connected discourse without grammar, a capability unique to the agrammatic form of aphasia. This may help explain our abilities to successfully communicate more ideas to chimpanzees and gorillas with a limited sign vocabulary than we can communicate to dogs (for instance). This mutual primate ability stems from our shared pathways for object recognition and episodic memory; but since natural selection has never made the ability to acquire grammar a favorable adaptation in chimpanzees, they will never be able to generate grammatically correct sentences.37 Accordingly, evoked responses are not activated in the primate brain in a similar way to the human brain when it processes grammatical information.

Of course, there are a number of arbitrary interpretations one can conjure up in order to explain how grammar was selected for, and many of these arguments take the form of explaining why humans have language and animals do not. Although many of these observations are purely speculative at this stage in our evaluations of language origins, hypotheses need to be forwarded so that they can ultimately be tested and challenged, and there are some very clear selective advantages that language provides. While it is clear that language enables its users to communicate salient events, it is not clear that communicating narrative events is the most important function of language, nor that the evolution of language made civilization itself possible, as an agrammatic complex society could theoretically exist.38 However, language has a vast number of attributes that seem to confirm its obvious adaptive benefits. Having language streamlines the teaching process and makes it possible without props, which is what chimpanzees have to use to teach their offspring.39 Language does not always need to rely on facial expressions to transmit meaning, so verbal language can take place easily in the dark (this also explains why verbal languages may have appeared chronologically later and lasted longer than gesture-based language systems).40 Language can be used to send or leave information without requiring the physical presence of a group member, especially if it is written down. Because of its ability to generate an infinite number of utterances, language can potentially describe or explain anything that happens.

Most importantly, human grammar greatly increased linguistic complexity, making it infinitely descriptive, converting our communicative system based on static labels and verbs into language. Among the enormous number of conceivable advantages that language brought our species undoubtedly lie some reasons that are not ad hoc and arbitrary and that would help explain why natural selection favored the probable evolution of grammar in language sometime around the appearance of the first archaic Homo sapiens, ca. 400,000 years ago or less.41

The most current evidence illustrates that language has characteristics that appear to have been exapted from primate neural pathways, and other characteristics that appear to have been naturally selected for by the prehistoric human environment.42 Our abilities for object recognition and semantic memory, in addition to spatial relations and episodic memory evolved much the same way as bird feathers did, meaning we first used these cognitive skills for finding food and mates and escaping predators, the central concerns of most vertebrates. Human grammar, which makes human language possible, conversely appears to have originated because it provided those who possessed it with a decisive reproductive advantage over those who did not, just as the vertebrate eye bestowed a definite advantage over the first vertebrates who could see.43 The broad thrust of theorists responding to Pinker and Bloom, then, casts language as a remarkable product of exaptation, leading to the development of a complex lexicon, and natural selection, leading to the development of grammar. If future empirical evidence can provide concrete validation for these theories, not only will evolutionary biology be provided with a marvelous example of how varying evolutionary processes can interact in order to produce complex traits, but we will gain valuable insight into exactly where and how human cognitive capacities have diverged from those of our closest primate relatives.

© Copyright 2003, All Rights Reserved, CSA