LTM-CNRS
ENS, 1 rue Maurice Arnoux, F-92120 Montrouge, France
Bernard.Victorri@ens.fr

What is at stake in this paper is the emergence of the human species-specific language from an evolutionary point of view. Following Bickerton (1990), I shall assume that pre-human hominids were endowed with a protolanguage without syntax and some other important properties of human language but which was fitted for simple acts of communication about 'here-and-now' perceptible reality, in which pragmatic considerations can compensate for the lack of these properties. I will leave aside the problem of the emergence of protolanguage as such: several authors (Dunbar 1996, Dessalles 1998, among others) have shown that this problem (bringing up the 'Machiavellian Intelligence' paradox - see Byrne & Whiten 1988) can be dealt with from a Darwinian perspective, even if there is no easy solution readily available. The point I want to make in the article deals with the transition from protolanguage to language. I will argue in favour of the emergence of a novel function of communication which could have started the process that led to the acquisition of the very specific properties the human language is endowed with. In this respect, what I suggest belongs to the category of the 'function-centred' scenarios, in which approaches such as Donald (1991) and Knight (1998) can be placed, contrasting with the 'mechanism-centred' scenarios which can be found in Pinker (1994) or in Bickerton (1998).

Human language differs from other communication systems by many specific syntactic and semantic features. Many authors focus exclusively on syntax, but semantics deserves as much consideration.

To give only one example, all human languages possess complex systems of markers for expressing aspect (durative, perfective, etc.). The category of aspect is very different from that of tense: aspect refers, not to the time relation between a situation and the moment of its being mentioned in speech, but rather, to how the situation itself is being viewed with respect to its own internal makeup (cf. Comrie 1976). The importance of aspect in human languages (Chinese, for instance, has no grammatical markers for tense, though it does have aspect morphemes) points to the central place of the narrative function of language.

Actually, aspect is not really needed when the goal of the communication is reduced simply to transmission of factual information. It is also worth noticing that there is no trace of aspectual notions in logical languages, even in those dealing with temporal logic. But aspect is absolutely necessary for the purposes of narration: it allows the narrator to present an event from various points of view, depending on the relation between the event and the characters on which the narrator focuses.

As a matter of fact, semantics of tense itself is best explained in terms of narrative mechanisms. As pointed out by Turner (1996:149-154), tense expresses the relation between a temporal focus and a temporal viewpoint, and the temporal viewpoint depends on the choice of the narrator: nothing in language forces the narrator to choose the moment of speech as the temporal viewpoint. This is why the present tense can so easily be used to evoke past or forthcoming events.

Observations of the same kind can be made about other semantics domains, including expression of spatial relations, modalities, animacy and agency, etc. Semantic systems may seem imperfect to a logician or an information theorist, but they are perfect tools for a narrator who wants to present in his own way past or imaginary scenes and events, with no direct link to the 'here-and-now' situation.

At last, syntax as such can be explained in the same framework. Turner (1996:143-148) shows that grammatical constructions, with thematic roles, can be viewed as projection of "basic abstract stories". Embedding and recursion can also be viewed as projections of story nesting, i.e. combining one basic abstract story inside another.

On the basis of such considerations, which fit into the general framework of cognitive linguistics (Lakoff 1987, Langacker 1986, Talmy 1988, Fauconnier 1997, Goldberg 1995, Sweetser 1990, Fauconnier & Sweetser 1996), Turner (1996:160-168) proposes a novel scenario of the emergence of language. He assumes that narrative imagining first arose as an individual cognitive ability, resulting from the increase of hominid general-purpose intelligence. Then, the narrative structure, already present in individual minds, has been projected into the communication system, transforming it into human language.

Turner's scenario is appealing because it can be put in relation with the last steps of hominid evolution. One of the most puzzling facts about hominisation is the extinction of almost all the archaic Homo Sapiens, such as Neanderthals. According to contemporary theories supported by both paleoanthropology and population genetics (see for instance Lewin 1989, Cavalli-Sforza 1996), all the descendants of Homo Erectus, who dispersed all over the Ancient World one million years ago, and who evolved locally into different groups of archaic Homo Sapiens, disappeared subsequently (some thirty thousand years ago for Neanderthals), except, of course, our species, coming from a small group living in East Africa one hundred thousand years ago. It is hard to understand why such intelligent beings (the brain size of Neanderthals was even slightly larger than ours), well adapted to various environmental conditions, were erased from the earth without any descent. External reasons, such as climatic changes, epidemics, or competition with our own species, are not really convincing: in particular, they cannot explain why this extinction took place all over the Ancient World.

So we must look at endogenous reasons, such as social disturbances (Victorri 1997). At one stage, increase in individual intelligence could have become contradictory with social organisation. It is well known that intelligence is not at all a defence against antisocial behaviour. In animals behaviours that are dangerous for the survival of the species, such as killing siblings or eating offspring, are prevented by instinctive reactions. But one consequence of the great development of the neocortex that characterises hominisation is precisely the control of instinctive behaviour: even if we feel that an action is repulsive, we can carry it out anyway. Killing his brother or his father to replace him as the chief of the tribe, killing and eating children during a lengthy period of food shortage, are examples of intelligent behaviour that ensures a short-term individual well-being, which prove fatal to the group prosperity in the long range. Of course it does not go without conflicts, internal conflict for the one who is ready to act, and social conflict within the group as each member senses and reacts to the forthcoming drama. So such acts could provoke real crises, triggering off uncontrollable violence with disastrous consequences to the life of the group.

Then, the hypothesis that can be put forward is that language emerged in such situations of crisis, allowing to avoid them. We can assume that when a crisis was about to break out, most members of the group remembered the preceding crises, adding the recollection of past disasters to the instinctive repulsion. If an individual was able to evoke what happened in the past by his voice and his gestures, he had a chance of winning the support of the whole group and stopping the imminent catastrophe. Expressing what happened in the past was also expressing what could happen again and what should never happen any more. This behaviour would bring a new cohesion to the group, constructing a new collective conscience able to offset individual desires. It would open the path to a new social order, with laws imposed from above by the awareness of belonging to a same group endowed with a collective history. All the myths and religions of humanity rest the fundamental prohibitions upon narrative accounts of origins that put precisely on stage the prohibited behaviours.

Thus, the first use of the narrative function could have consisted in expressing specific social events, creating a totally novel social organisation in the animal kingdom, which permitted our species to control the social disturbances that could explain the extinction of the other archaic Homo Sapiens. Human language, with all its syntactic and semantic properties, would then have stemmed from the needs of narration, leading to what Donald (1991) calls 'the mythic culture'.

We can even assume that language proper was first limited to this particular function, and that during a long period its development was independent (to a certain extent) from the utilitarian protolanguage that our species must have possessed, like the other archaic Homo Sapiens. Afterwards, the extension of the use of language to all aspects of the life could have triggered the "symbolic revolution" that took place some forty thousand years ago (cf. Tattersall 1998).

The hypothesis of the emergence of the narrative function as the starting point of the emergence of language takes three main points into account:

- first, it can explain the unique syntactic and semantic properties of language, specially those semantic properties that make human language different not only from animal communication systems but also from logical and other formal languages.

- second, it is compatible with our current knowledge of the last steps of hominid evolution, including the puzzling extinction of almost all the archaic Homo Sapiens, which contrasts with the great evolutionary success of our species.

- third, it permits to understand the development of a novel level of organisation, specific to our species, in which socio-cultural laws replace, to a large extent, the socio-biological constraints governing all the rest of the animal kingdom.

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