Department of Linguistics, Macquarie University
cmatthie@ling.mq.edu.au

While speculations about the origin and evolution of language were considered outside the realm of scientific enquiry in linguistics for many years, there have in recent years been a growing number of valuable contributions to this topic from a range of scholars, including now also linguists (e.g. Beaken, 1996; Deacon, 1997; W. Foley, 1996: Ch. 2; Hawkins & Gell-Mann, 1992; Hurford, Studdert-Kennedy & Knight, 1998; Jablonski & Aiello, 1998; Noble & Davidson, 1996; Ruhlen, 1994).

There is still a certain tendency to see the beginning of the evolution of language as dependent on conditions created by the evolution of the brain (to cope with general intelligence) and the vocal organs (resulting from bipedalism, a change in foraging behaviour, and so on), thus suggesting that language is a secondary phenomenon – perhaps even a property unique to a human "language faculty" that arose because of some mutation, as suggested by e.g. Bickerton (1990), thus creating a sharp divide between humans and other primates or indeed other animals in general. Such a line of interpretation seems to be made more plausible when there is a lack of a developmental model that could be used to explain an evolutionary sequence leading up to language: language comes to be seen only by reference to the full complexity of modern human languages (though Bickerton himself has a notion of proto-language).

Against this kind of view, a number of scholars have argued that language was a much more central factor in the evolution of humans (and as part of that, in the evolution of the human brain): see e.g. Deacon (1992, 1997), Edelman (1992) and Bickerton (1995: Ch. 2). According to this line of interpretation, language and the human brain co-evolved. This is certainly the view that makes sense from the point of view of a systemic functional understanding of language and the evolution of language: see Halliday (1995). However, the accounts offered in the literature can be further supported by a general theory of language that will allow us to interpret the co-evolution of language with the brain and other aspects of modern humans as a gradual but accelerating increase in complexity. I believe Halliday’s (1975) account of the ontogenesis of language from protolanguage via a transitional phrase to (adult) language provides us with a very plausible model for how language might have evolved. Significantly, it does not force us to assume a sharp divide between (modern) humans and other primates – or indeed other mammals: Halliday’s account of ontogenesis provides a very detailed model of the ontogenetic phases; it combines both continuity and transformation and it interprets language development as children actively learning how to mean – developing a meaning potential together with the people they interact with in a growing range of critical social contexts (cf. Trevarthen’s, 1987, discussion of intersubjectivity in ontogenesis) – rather than as children "acquiring" a ready-made product.

In this paper, I will adopt a "cosmogenetic" perspective (cf. Layzer, 1990; Delsemme, 1998; Smith & Szathmáry, 1999) on the evolution of language. Language will be assumed to have evolved "emergently" as a higher-order semiotic system within an ordered hierarchy of systems of increasing complexity – physical, biological, social and semiotic systems (see Halliday & Matthiessen, 1999: Ch. 13). The evolution of language will be assumed to have been a gradual but accelerating process following general principles of increasing complexity in complex adaptive systems in general (cf. Steels, 1998) and building on earlier capacities shared with other primates (following the general evolutionary principle of new functions out of old structures); and the evolution of languages is further assumed to have been a process characterized by co-evolution with both biological systems (bipedalism, vocal apparatus, brain etc.; see Deacon, 1992, 1997; Edelman, 1992; Halliday, 1995) and social systems (division of labour, social hierarchy etc.). Through this co-evolution, there has been a gradual increase in complexity in all the systems involved as greater complexity in one creates the conditions for greater complexity in another. (I thus do not accept the arguments put forward against "gradualism" by e.g. Noble & Davidson, 1996.) In other words, the evolution of language has been an integral part of the evolution of "humanity". Importantly, the evolution of modern language has provided a new order of organization that has made human evolution possible without further brain expansion after the appearance of modern Homo sapiens around 100 K years ago.

In reviewing the evidence cited in the evolution of the human line – from the splitting off of Australopithecus, via Homo habilis, Homo erectus, archaic Homo sapiens to modern Homo sapiens, I will sort it according to the different orders of system – physical (changes in climate; changes in habitat etc.), biological (bipedalism; brain evolution; evolution of the vocal organs; state of newborn and prolonged infant dependency; etc.), social (sociality; prolonged infant dependency and social learning; recognition of individuals and group evolution; division of labour; geographical spread; settlement etc.) and semiotic (the appearance of art; evidence of burial; body adornment; but also implications from "deep" language history [cf. Nichols, 1992; Ruhlen, 1994] etc.) – while noting that the only direct evidence that remains exists at the order of physical systems (excluding backward projections based on DNA, language history etc.).

The notion of a gradual but eventually accelerating evolution of complexity fits well with Mithen’s (1996) account of the evolution of the mind, with major transitions such as the evolution of a "general intelligence" in addition to previously separate specialized intelligences (cf. the discussion below of the transformation of linguistic macro-functions into simultaneous modes of meaning). However, the account that I will explore here is language-based, with the evolution of the meaning potential as the central motif (cf. Halliday & Matthiessen, 1999). The model for the phased evolution of complexity in the meaning potential is based on Halliday’s (e.g. 1975) investigation of ontogenesis, expanded by subsequent research by Painter (1984), Torr (1997) and others. According to this model, it is possible to identify three non-discrete phases of development after a period when infant engage with semiotic instances and precursors (such as exchanges of attention – what Catherine Bateson has called proto-conversation) that have not yet crystallized into a semiotic system –

This account embodies both continuity (the functional strands) and transformation (the metafunctional re-interpretation; the stratal fission of content with the emergence of lexicogrammar) and lexicogrammar is seen as arising as an emergent formation of a new stratum (cf. Luc, 1998: 399-402, on "level formation") rather than as the result of genetic preprogramming. This puts into perspective Bickerton’s (1995: 68-69) view that the emergence of "syntax" as "a catastrophic event", interpretable in terms of Eldredge & Gould’s theory of punctuated equilibrium. (It is important to note that in ontogenesis lexis and grammar develop together. We have to revise common view is that lexis develops first, followed by grammar: the early signs of protolanguage are neither words nor lexical items – they are protolinguistic signs. It is also important to note that at the level of the phonological system, articulation begins as a kind of prosody, developing out of protolinguistic postural vocalizations.)

Each ontogenetic phase has critical implications for both social and biological developments in the growing child. For example, while protolanguage goes with crawling and primary consciousness, language goes with walking and higher-order consciousness (Halliday, 1998). In addition, the account also has clear socio-cultural implications: each phase represents a distinct relationship between language and context, from the constrained phase of protolanguage where the microfunctions are directly tied to particular contexts of use in a one-to-one mapping to the very complex and open-ended relationship that characterizes phase III. This will make it possible at least to explore the question of what our ancestors would have been able to achieve with the linguistic potential characteristic of each particular phase.

The ontogenetic model can, I believe, serve as an explicit and detailed outline for exploring phylogenesis, with one central difference. (This is of course an analytical strategy; as far as language development is concerned, it is the other way around: just as in the biological development of an individual, ontogenetic strategies recapitulate phylogenetic ones – up to a point.) Children learn how to mean in interaction with their immediate caregivers who have a full-fledged linguistic meaning potential and possibly with older siblings. Thus when they have learned the basic principles through their own protolanguages and begin the transition into the mother tongue spoken around them, they have an existing model to draw on in building up their own meaning potentials. This was of course not the case in human evolution: there were no models; and all members of a given social group were at the same stage of language evolution. As a result, I believe the transitional phase must have extended over a very long period of time in phylogenesis even though it is fairly brief in ontogenesis. But this seems to be what the general picture of human evolution suggests – a long period of relative phylogenetic stasis for the duration of Homo erectus from around 1.8 m years ago until the appearance of archaic Homo sapiens some 400-200 K years ago. There would thus have been a gradual build-up in semiotic complexity during the transition, followed by a more rapid evolution once the break-through to language had taken place.

It seems plausible that some form of phase I – protolanguage – is very old indeed – predating the evolution of the hominid line. It would have been multifunctional from the beginning, evolving in regulatory, interactional, instrumental and personal contexts of use rather than being restricted to something like social grooming or bonding or linked to something like tool production and use (which can be achieved through non-linguistic apprenticeship; cf. Savage-Rumbaugh & Lewin, 1994: 246). It is very likely to be a semiotic potential shared with apes in the wild (as I would interpret accounts of the natural semiotic systems of modern chimpanzees [cf. Goodall, 1986], bonobos [Savage-Rumbaugh & Lewin, 1994: e.g. 106-107; 112-113, 119] and orang-utans [cf. Kaplan & Rogers, 1999]) – perhaps even one that goes back to the evolution of primates some 60 million years ago, correlating with R. Foley’s (1997: 173-174) observation that "primates are the social order par excellence" and that "sociality is really part of the primate core adaptation". In human infants, protolanguages seem to develop to the size of 60-70 signs; but it seems quite plausible that such inventories would have been further extended during phylogenesis until the point where the protolinguistic system had to evolve a more complex form of organization to cope with the meaning-making pressures it had helped to create. Protolanguages of this kind would probably have been vocal-gestural combinations in expression (including facial expression of the kind we find with modern apes), just as with human infants, and would thus not have been subject to any of the requirements on vocal production characteristic of modern languages (as emphasized by Philip Lieberman); these requirements would only have emerged much later in the evolution of language. Such protolanguages would have enabled members of groups to share attention and togetherness, to exercise simple control, to issue warnings – that is, they would have provided these members with microfunctional modes of meaning situated within very specific types of context; but they would not have enabled them to engage in dialogue (as opposed to simple turn taking), to construe their experience (e.g. by naming plants and animals independently of the microfunctional act), to create and exchange information; and it would not have enabled them to mean more than one thing at the same time. These semiotic capacities only begin to become possible with the transitional phase.

In the paper, I will explore the question of when the transition might have started – relating it to observations about both biological and social changes (against the background of material changes); and I will explore the question of when the transitional phase may have been evolving into the modern linguistic phase. While it seems plausible that the transitional phase would have begun already with Homo habilis (2 to 2.5 million years ago) – and certainly with Homo erectus (around 1.8 million years ago), the particular challenge is to identify the possible period during which the transitional phase evolved into the modern linguistic phase: did this happen when archaic Homo sapiens appears (200-400 K years ago) or when anatomically modern Homo sapiens appears (very roughly somewhere around 100 K years ago); had it happened by the time people migrated to Australia (perhaps even 60 K years ago); what linguistic changes correlate with the "Upper Paleolithic revolution" when there is an apparent explosion of cultural artifacts (perhaps around 40-30 K years ago)? Based on evidence from existing languages and assumptions about branching and stock density, Nichols (1998: 138-139) estimates "the linguistic age of the world" to be around 132,000 years: this is the time it would take "to populate the traditionally inhabited world with language families at the density that can be reached when circumstances are favourable – that of New Guinea". This is one possible source of evidence, pushing the date back before the first traces of anatomically modern humans; but it does not extend further back into the long transitional phase of language evolution.

 Conference site: http://www.infres.enst.fr/confs/evolang/