If this sentence represents a theory of language evolution in the human species, then the full stop at the end represents the fossil records that we have on which the theory rests.
It is no surprise that the Societe de Linguistique de Paris in 1866 placed a ban on the discussion of the origins of language within the journals of the day given the speculative nature of the task. The last couple of decades however has seen an explosion of ideas from linguists, anthropologists, biologists, cognitive scientists, palaeontologists and the like to try and account for the origins of this enigmatic form of communication. Language evolution is once more a respectable field of study.
Within the origins of language debate, there are several dimensions to be considered:
- The representational /communicational dimension: did language evolve primarily as a representational system to facilitate perception and behaviour or did it evolve to promote communication between mankind.
- The continuous /discontinuous dimension: does language differ from other animal communication systems in degree or kind.
- The early /late dimension: did language evolve early in the history of man (100,000 to 250,000 BP) or did it evolve late (40,000 BP). (footnote 1)
- The gradual /catastrophic dimension: did language evolve gradually with many intermediate stages or did it emerge catastrophically, thus rendering discussion of intermediate stages superfluous.
The multiple-volume tome required to comprehensively cover such a wide range of issues precludes discussion of them as a whole here. Instead, I will concentrate on item (4) – the gradual /catastrophic dimension – and, in particular, the emergence of syntax and whether this was gradual or catastrophic. In doing this, I hope I will be extracting for linguists certain notions that will prove useful for their profession and at the same time avoiding many of the speculative and unfalsifiable claims that are characteristic of the other arguments.
I will argue that there is a position which allows for the catastrophic emergence of syntax alongside the more widely-accepted gradualist account, but that this position must look toward the ‘genes of mother nature’ (a term which will become clear in due course), rather than the human genome, for its home. Current limits to knowledge, however, forestall a definite endorsement of one position over the other. Despite this inability to resolve the issue, I will show that the outcome is much the same for linguists. That is, the origins of language debate does not impinge as much on their domain as might first be thought.
I will begin by giving a brief sketch of the notion of protolanguage before moving on to consider the case of the gradualists (Pinker & Bloom, 1990; Newmeyer, 1991) and subsequently the catastrophists (Bickerton, 1995; Gould & Lewontin, 1979). By then I shall be in a position to expound upon the term ‘the genes of mother nature’ and the implications this brings with it.
A Basic Sketch of Protolanguage
Humans have language: animals do not. Short of paying homage to creationism, we are compelled to infer that language emerged sometime after the split of the hominid family from our panid cousins (chimpanzees). Further, the supposition that full and complete language did not emerged from languageless hominids in one single generation is uncontroversial. It is therefore necessary to postulate the existence of a ‘protolanguage’; a kind of reduced form of language that was the precursor to modern day language. The shape and form of this protolanguage are where most of the real questions lie for the linguist: What was the size of its lexicon? Did it have a syntactic structure or not and, if so, how complex was this?
In soliciting answers to these questions, one problem arises in that no one is certain if we have modern day examples of protolanguages which we can use for reference. Some scholars (e.g. Bickerton, 1995) have pointed to pidgins, child language and ape (sign) language as examples of modern day protolanguages, but it is difficult to say whether the conditions that pertain today and which shape these protolanguages prevailed when man first began to speak. Did, for example, ancient man possess a brain comparable in size and cognitive capacity to modern day man with which to process such a protolanguage, or was it significantly different? Moreover, even if we do take pidgins, child language etc. as modern day equivalents of protolanguages, there is still a great deal of controversy over the structure of them. Are they, for example, syntactically structured? Bickerton (1995:51) claims they are not and challenges anyone in his book to write a simple grammar for them. Others claim they are (e.g. Pinker & Bloom, 1990:723).
In trying to reconstruct the protolanguage of ancient man, it is possible to start with a basic lexicon (footnote 2) and add complexity or to start with full language and remove complexity. But in either case, the same question emerges: Is it possible to have a continuum of intermediate protolanguages between full language and no language (or simply a lexicon) with each protolanguage varying syntactically in efficiency and expressive power from the previous by one degree of complexity? Furthermore, would each protolanguage be stable such that it would remain within the species for any appreciable amount of time before yielding to the next stage?
Scholars who hold that this synopsis is possible, such as Pinker & Bloom (1990) and Newmeyer (1991), argue for a gradual evolution of syntax over many millennia. Other scholars, such as Bickerton (1995), hold that it is futile to talk of a range of intermediate protolanguages. Instead, they propose one protolanguage that at some point changed catastrophically, gained a complete syntax and emerged as a full-fledged, modern day language within the space of a few generations (footnote 3). I shall consider the case for the gradual evolution of syntax first.
The Case for the Gradual Evolution of Syntax
For selectionists such as Pinker & Bloom (1990) and Newmeyer (1991), it is crucial to be able to show that syntax developed gradually over many millennia for it to become encoded in our genes (and hence lend support to the universal grammar hypothesis). According to the neo-Darwinian law of natural selection, anything trait as complex as language that is genetically determined requires several millions of years of evolution to have become encoded in our genes. Pinker & Bloom thus reject early estimates of the emergence of language such as that of Noble and Davidson (1991:223), who propose it emerged relatively late at about 32,000 BP during the middle-to-upper Paleolithic cultural transition. Instead, they look to the emergence of a protolanguage sometime after the split of hominids with panids. At a maximum, this would give them 4 million years for syntax to develop (Pinker 1994:353). At a minimum, they suggest several thousand years would be sufficient (Pinker & Bloom, 1990:726).
During this evolutionary time frame, they suggest that variation within the gene pool led to individuals having grammars that were marginally more competent than others and thus more useful. This ‘conferred a reproductive advantage on its speakers, and this advantage [was] large enough to become fixed in the ancestral population’ (Pinker & Bloom, 1990:721). Grammars of intermediate complexity (i.e. protolanguages), they suggest, could have had ‘symbols with a narrower range, rules that are less reliably applied, [and] modules with fewer rules’ (Pinker 1994:366).
It is indeed possible to imagine grammars of such intermediate complexity. We might, for example, postulate a grammar with an abridged binding theory module that applied only to anaphors rather than anaphors, pronouns and referring expressions. Or a grammar lacking a complete module altogether such as the subjacency principle. Furthermore, a language can easily be fashioned from symbols with narrower ranges. Proto-verbs, for example, may have subcategorized for fewer theta-roles than they do today. The verb struck, which currently subcategorizes for an agent and patient as in ‘Thorg struck Fred’ may have been only able to assign agent in an intermediate protolanguage as in ‘Thorg struck’, with the patient absorbed into the semantic space of the verb.
Specifying a grammar of intermediate complexity is not in itself particularly problematic. The difficulty is in constructing and rationalizing a mechanism that explains how an individual with a grammar of complexity i+1 secures a reproductive advantage over those individuals with a grammar of complexity i. We find today very articulate individuals who are able to use language very eloquently and persuasively but this does not automatically confer survival and reproductive advantages on such individuals (footnote 4).
Pinker & Bloom answer this by claiming that the way of life of hunter-gatherer societies using such a protolanguage would have put a premium on linguistically supported interaction and note how even ‘tiny selective advantages are sufficient for evolutionary change’ (Pinker & Bloom, 1990:724). In fact, ‘the pressure may be so small that it cannot be measured by human observers’ (ibid. 724). This immediately creates a problem because if the selective advantage is so small that it is unobservable, then it implies a theory which is not falsifiable.
The reluctance of Pinker & Bloom and many scholars to provide substantial elaboration of this area perhaps reflects their recognition that our present level of knowledge of man’s past is insufficient to furnish persuasive arguments either for or against the adaptive advantage of marginally more complex grammars and that we run the risk of embroiling ourselves in ‘just-so’ stories. I am in full agreement with the profession here and I don’t intend to pursue the point further.
Fortunately for Pinker & Bloom, the foundation of their theory is not built solely on the above thesis but has a much more substantial argument that we will explore next. This will necessitate the elaboration of the other side of the debate – that of the catastrophists – and the theory of punctuated equilibrium.
Punctuated Equilibrium and the Catastrophic Emergence of Syntax
As an alternative to neo-Darwinian gradualist orthodoxy, the theory of punctuated equilibrium has risen in popularity over the last two decades or so in an attempt to account for the lack of transitional forms in the fossil records that a gradualist account (of evolution) requires (Piattelli-Palmarini, 1989:8). Its main proponent, Steven Jay Gould (Eldredge & Gould, 1972, Gould & Lewontin, 1979), holds that the evolution of new species occurs in rapid bursts interspersed with long periods of relative stagnation – hence the name, ‘punctuated equilibrium’.
Within this theory, the notion of ‘exaptation’ takes on a central role. Exaptation is the adaptation of an existing structure for a new purpose, and Gould and Lewontin (1979) give the spandrels of the San Marco cathedral in Venice (see Appendix 1) as the classic example of an existing structure (the spandrel created by mounting a dome on rounded arches) being used for some other function (the display of artistic paintings). The important point is that it is not just the fact that the space (spandrel) exists as a by-product of the design of the cathedral but that it contains a complex design (painting).
Derek Bickerton (1991, 1995) has endorsed the theory of punctuated equilibrium and applied it to the evolution of language, suggesting the emergence of syntax was ‘a catastrophic event occurring within the first few generations of the [Homo sapiens sapiens] species’ (Bickerton, 1995:69), thus claiming that the change from protolanguage to language occurred in one foul swoop. The emergence of syntax was a natural consequence of an expanded brain that Homo sapiens sapiens had been bestowed with.
He suggests pidgins, child languages and ape (sign) languages are modern day versions of protolanguages with their collective feature being the absence of any syntactic structure. He thus dismisses Pinker & Bloom’s claim that we can observe today a range of protolanguages each with its own degree of syntactic complexity. A language, according to him, either has a (complete) syntax or it doesn’t; there is no in-between position. By extension, this situation must have held for ancient man (footnote 5).
His basic synopsis then is that protolanguage developed sometime during the evolution of the hominid genus and was relatively stable over many millennia. Full language with syntax then burst onto the scene within the first few generations of anatomically modern humans as a by-product of an expanded brain:
The linguistic history of the hominid line appears, accordingly, as a two-stage process: first a stage in which there was a lexicon without a syntax, then a stage in which infinitely productive mechanisms emerged to create syntax as we know it. (Bickerton, 1995:51)
The dilemma in accepting this theory, according to critics like Pinker & Bloom (1990:714), a dilemma which is at the core of the debate, is that it is vanishingly unlikely that anything as complex as language (with a syntax) would emerge as a by-product (i.e. a spandrel with a design) of an evolving organ (the brain). Their rationale can be illustrated thus: If, for example, we were to observe an object being used for a mundane job such as a paperweight, we would be hard pressed to guess what that object was since virtually anything could be adapted to do the job. On the other hand, if we saw an object being used to receive television broadcasts then we could pretty well guess that it had been designed to do the job since the chances of anything so complex just popping into place are vanishingly small.
This reasoning is not so watertight, however, because we know that complex systems sometimes spawn complex by-products. The Internet is such a case of a high-tech system that was designed to carry particular signals (hypertext documents and e-mails) but is also capable of carrying television pictures. So complex systems can yield high-tech spin-offs providing the system is high-tech in the first place. If the human brain evolved in cognitive complexity as we know it has done (why it did so is another story) then syntactically structured language could be a plausible by-product of this. Unfortunately, I don’t think the spandrels of San Marco were a suitable choice by Gould and Lewontin to illustrate this point since they (the spandrels) are passive areas without complexity onto which complexity has been affixed. This is not the case with the human brain.
Of course, Bickerton’s catastrophic theory runs into difficulty when he tries to claim that full language emerged catastrophically and was encoded in our genes as a biological trait that we can transmit to our offspring. Now that really would be a miracle for it requires a mutation to occur on a grand scale and for the resulting mutant to be able survive and reproduce. But we know that virtually all such catastrophic mutations yield new species that do not survive. For example, a mutation that created an animal with two hearts may at first sight appear to have an adaptive advantage which would be selected for. But the animal’s chances of survival would be very slim indeed since it would also require a set of correlated changes in the circulatory system of veins and arteries to accommodate the extra blood flow (together with appropriate neural mechanisms that regulate blood flow). The chances of independent genetic mutations synchronizing like this just at the right time and in the right way are very small indeed (Lieberman, 1984:7). Few evolutionists support the idea of new species being created in such a way through massive mutations .
Yet Bickerton is the main protagonist of the ‘language bioprogram hypothesis’ (Bickerton, 1984) that assumes we have a default grammar encoded in our genes. Children make use of this bioprogram when they learn their first language in impoverished conditions (e.g. children who are born into pidgin speaking communities) and emerge with a full language, namely a creole (footnote 6). The remarkable similarities of creoles across a wide range of grammatical markers (e.g. tense, mood and aspect) form the backbone of this bioprogram hypothesis (Aitchison, 1996:147). This innate bioprogram is the genetic hard-wiring that was supposedly planted in our genes in the first few generations of our species.
There is, however, one way out for Bickerton that would allow him to maintain his claim that syntax emerged abruptly and catastrophically. That is for him to withdraw from the innatists’ camp and drop his claim that language is an innate trait encoded in our genes. Bates has noted how sometimes certain features of an organism’s behaviour can seem to be remarkably complex and regular but can be the logical ‘fall-out’ solution to particular problems to such an extent that ‘genetic underwriting is unnecessary’ (Bates, 1984:188). She gives the example of the perfect hexagonal structure of bee honeycombs. At first, one is tempted to assumes the bees must be innately endowed with the ability to construct such shapes. But we now know that the structure is the inevitable outcome of a mathematical packing principle that occurs when bees with hemispherical heads pack together at random pressures from all angles. All that needs to happen is that the organism is ‘placed in the right ballpark for the problem to be encountered and solved’ (Bates, 1984:189).
Thus ancient man may have been placed in the right ‘ballpark’ by several separate genetic changes (say lowering of the larynx and the appearance of Broca’s area (footnote 7)) and grammar (=language) is what fell out on purely formal grounds. This kind of ‘catastrophic’ change is more true to the original concept of punctuated equilibrium and doesn’t assume that every complex phenomena that we observe in the universe has to be genetically underwritten. In other words, the universality of a phenomenon per se cannot be taken as prima facie evidence for its innateness.
Kirby (1998), in more recent work, has validated this hypothesis through computer simulations which show compositional syntax emerging from randomness in populations over time. This system (communication) is passed on ‘culturally through observational learning by new individuals, and there is nothing else inherited by later generations from earlier ones.’ (Kirby, 1998:18). Furthermore, the work has shown how populations can rapidly innovate a syntactic structure from a pidgin-like input thus providing a non-innatest account of the intra-generational creolization of pidgins by children (Senghas, 1995).
The two poles of the gradual /catastrophic dimension then can be summarized thus:
- Syntax evolved gradually by being naturally selected for according to the neo-Darwinian mechanism. This is the only mechanism that can account for its complexity if we discount massive mutations. Syntax is thus encoded in the genes of modern man.
- Syntax evolved catastrophically when ancient man was pushed into the right ‘ballpark’ by a number of independent changes in his make-up which may or may not have been genetically determined. The structure of syntax is the ‘fall-out’ (logical) solution of the problem-space; the ‘genes of mother nature’ if you like. Syntax is thus not encoded in the genes of modern man.
Of course, there can be other scenarios that embody various facets of these two extremes. Syntax may have emerged via mechanism (2) but over many millennia. Or it may have evolved through process (1) but in punctuated steps. Or (as is often the case in these situations) we may find it is a combination of the two, with the core grammar being generated by process (1) and the peripheral grammar arising from process (2). However, what is for certain is that our current level of knowledge of evolutionary mechanisms, man’s past and language universals is insufficient to conclusively declare which process led to modern day language.
Functionalism, Formalism: Where’s the debate?
Of course, the two poles of the debate outlined above supposedly exemplify the formalist and functionalist’s perspective. But it is interesting to note that Pinker & Bloom’s solution seem to be more functionally motivated than Bates’ solution which she herself admits arises ‘on purely formal grounds’ (Bates, 1984:189). In fact, Bates and MacWhinney (1990:727) in their peer commentary title, ‘Welcome to functionalism’, applaud Pinker and Bloom for their apparent shift toward the functionalist’s camp.
So is there really a need for a formalist /functionalist divide in linguistics as we are led to believe. Whether we look toward man’s own genome to explain the recurring structure of syntax in language or the ‘genes of mother nature’, it seems that the task on both sides of the debate is the same: to discover the structure of syntax and capture its shape in as logically precise and complete way as possible. It is difficult to conceive of a task more central to linguistics than this, yet the very same statement seems to divide the field so sharply and leads to pointed battles amongst its leading scholars.
For Chomsky and the generative grammarians, this appears to be the impetus behind their frantic effort to work on two levels simultaneously: a highly technical level to develop a principles and parameters framework to account for the syntactic structure of language and a highly theoretical (and at times speculative) level to postulate a genetic underpinning for their framework. Hence the work of Pinker & Bloom (1990) and others. But the very fact that grammar exists as a complex structure (which no one doubts) is sufficient grounds for its study as an autonomous system. It doesn’t require a pact with the biologists and the human genome to be made for its legitimization.
I hope, then, that the fresh rise in popularity of the origins of language debate does not serve to deepen the chasm between linguists in their efforts to develop a theory of language. For it seems that the academic and, at times, conjectural debate doesn’t appear to offer much more for the linguist that it did in 1866. Perhaps another imposition of a ban within our field is on the horizon?
1. Some scholars, such as Pinker (1994:353), put this upper limit much higher at 4 million BP when Australopithecus afarensis (‘Lucy’) appeared.
2. Note, the question of how a lexicon emerged is also a question for those studying protolanguages. We will not look at that issue here but the reader is referred to Wray (1998) for an up-to-date discussion on this topic.
3. A few others, such as Aitchison (1996) take an in-between position where protolanguage develops along an S-curve, slowly at first, then through a rapid phase where 90% of the development takes place and finally through another slow phase where the final language emerges.
4. The natural inference to make would be that linguistically eloquent speakers tend to have fewer offsprings than the average, although this is extrapolating from a modern society perspective. As Fouts (1991:42) says of Newmeyer, ‘It seems crucial to me that he [Newmeyer] demonstrate how a human male or female who uses Chomskian perfect grammar has a better chance of breeding than one who failed English 101 and is noted for ungrammatical monosyllabic utterances yet has bedroom eyes and drives a BMW.’
5. As further evidence for the abrupt divide between protolanguages and languages, Bickerton (1995: 29-32) notes a number of features of languages that protolanguages don’t have: languages have longer utterances, are fluent, are always interpretable and contain units that make reference to the structure of the utterance (e.g. subordinating conjunctions).
AITCHISON, J., (1996), The Seeds of Speech Language Origin and Evolution, Cambridge University Press, Cambridge.
BATES, E., (1984), ‘Bioprograms and the innateness hypothesis.’, Behavioral and Brain Sciences, 7:2, 188-190.
BATES, E., & MaCWHINNEY, B., (1990), ‘Welcome to functionalism.’, Behavioural and Brain Sciences, 13, 727-728.
BICKERTON, D., (1984), ‘The language bioprogram hypothesis.’, Behavioral and Brain Sciences, 7:2, 173-221.
BICKERTON, D., (1995), Language and Human Behaviour, UCL Press.
ELDREDGE, N., & GOULD, S.J., (1972), ‘Punctuated equilibrium: an alternative to phyletic gradualism’, in Schopf, T.J.M., (ed), Models in Palaeobiology. 82-115, Freeman & Cooper.
FOUTS, R.S., (1991), ‘Dirty bathwater, innateness neonates and the dating game.’, Language and Communication, 11:1-2, 41-43.
KIRBY, S., (1998 in preparation), ‘Language evolution without natural selection: From vocabulary to syntax in a population of learners.’, Edinburgh Occasional Papers in Linguistics. (Download from http://www.ling.ed.ac.uk/~simon/ )
NEWMEYER, F.J., (1991), ‘Functional Explanation in Linguistics and Their Origins of Language’, Language and Communication, 11:1-2, 3-28.
GOULD, S.J., & LEWONTIN, R., (1979), ‘The spandrels of San Marco and the Panglossian paradigm. A critique of the adaptionist programme.’, Proceedings of the Royal Society of London, 205, 281-8.
LIEBERMAN, P., (1984), The Biology and Evolution of Language, Harvard University Press, Cambridge, Massachusetts.
NOBLE, W., & DAVIDSON, I., (1991), ‘The evolutionary emergence of modern human behaviour: language and its archaeology.’, Man, 26, 223-253.
PIATTELLI-PALMARINI, M., (1989), ‘Evolution, selection, and cognition: From “learning” to parameter setting in biology and the study of language.’, Cognition, 31, 1-44.
PINKER, S., (1994), The Language Instinct, Penguin Books.
PINKER & BLOOM, P., (1990), ‘Natural language and natural selection.’, Behavioural and Brain Sciences, 13, 707-784.
SENGHAS, A., (1995), Children’s contribution to the birth of Nicaraguan sign language, PhD Thesis, Dept. of Brain and Cognitive Sciences, M.I.T.
WRAY, A., (1998), ‘Protolanguage as a holistic system for social interaction.’, Language and Communication, 18, 47-67.
A spandrel of the San Marco cathedral. (Source: Aitchison, 1996:75)