LANGUAGE ORIGINS: WHAT KANZI CAN DO AND HUMANS DO DO.

by

Iain Davidson, University of New England. (April 1990)

 

Kanzi, a bonobo or pygmy chimpanzee, can comprehend spoken English.  He may, also, be able to utter some sounds that should be recognised as words.  This startling result of research by Sue Savage-Rumbaugh and her team at the Language Research Centre, Georgia convinced some hardened sceptics at a recent conference on "Tools, Language and Intelligence: evolutionary implications".  Each of the participants at the conference will draw different conclusions from this, but there was a consensus that the new data on chimpanzees, both from Savage-Rumbaugh, and from field studies by Christophe Boesch and synthesis by Bill McGrew, go even further than previous ape studies in making us question the nature of human uniqueness.  In this summary of the events of the conference I want to record my personal impressions, and some of the first conclusions I draw.

The conference was organised in Cascais, Portugal, for the Wenner-Gren Foundation for Anthropological Research by Tim Ingold and Kathleen Gibson, and took place between 16 and 24 March, 1990.  The participants were from various disciplines: archaeology (Davidson, Graves, Hewes, Toth, Wynn); primatology (Boesch, McGrew); social anthropology (Ingold, Silverman); developmental psychology (Goldin-Meadow, Greenfield, Kempler; Langer, Lock); biological anthropology (Gibson, Parker); neurology (Calvin) and palaeoneurology (Falk); cognitive science (Lieberman); experimental psychology (Kendon, Visalberghi); and zoology (Snowdon).  There were experts on gesture (Goldin-Meadow, Kendon) and on speech production (Lieberman); on stone tools (Toth, Wynn) and on Alzheimer's disease (Kempler).  But despite the diversity, in the characteristic mode of Wenner-Gren conferences, the dialogue across disciplinary boundaries was outstanding.  We may not have founded a new discipline, but we all know much more about the relevant parts of others.  As Andy Lock put it: "there is something different about getting it from the mouth of a horse, and not the other end of a cow".

THE DEVELOPMENTAL PSYCHOLOGY OF LANGUAGE

It is clear that language is acquired and not innate: it gets there not in the genes but by guided reinvention, as Lock calls it.  Pat Greenfield and Jonas Langer both presented data and theory on the earliest stages of cognitive development in human infants.  Greenfield demonstrated that the earliest stages of the manipulation of objects and of vocal utterances should be considered homologous, not merely analogous.  The issue, then, is how and why they diverge.  For Langer, the essential disjunction, at this early stage of the ontogeny of modern humans, is that the "production" of cognition precedes its comprehension, while in language development, comprehension precedes production.  Indeed, it can be argued that the production of language is what makes the comprehension of cognition possible.  In simple language, and stripped of the assumptions involved in referring to perceptions of the world as cognitive, I take this to imply that young children explore their world through sounds and objects in similar ways.  But at some stage these explorations take different paths.  No sentient organism could survive without "production" of cognition in this sense, so it should not surprise us that it appears early in human development.

Two major points converge on these observations.  First that it may be that the combination of cognitive production and linguistic comprehension leads to cognitive comprehension.  Simplistically, we know what we know because we know what we are talking about.  Secondly, the comprehension that vocal utterances can communicate necessarily involves a relationship between persons, while object manipulating only involve the person (or infant) and the object.  As Sue Savage-Rumbaugh indicated, manipulation of objects may only involve the individual and the tools.  But in manipulation of the object the individual will invent gestures about the use of tools.  Viewing videotapes presented by Christophe Boesch of chimpanzees in the Tai forest cracking nuts, and by Susan Goldin-Meadow of deaf children of hearing parents using invented gestures to communicate, we saw "language invented in front of our eyes" in Savage-Rumbaugh's terms.  In each case, the gestures were directly within the contingency of object manipulation, real or imaginary.

In archaeology there are two approaches to evolutionary issues that derive from developmental studies.  The first takes a sequence of ontogenetic development and matches it to some sequence of changes described by archaeologists.  Lock pointed out the pitfalls of any argument that is so crudely recapitulationist, and I pointed out that many of the tales told by archaeologists have not taken into account that language is an issue, and therefore have assumed greater language abilities in hominids that we can reasonably infer.  Sequence matching therefore may be unrelated to reasonable interpretation of the archaeological record.

The second approach is to recognise from the guidedness of the reinvention of language that the context of learning is so fundamental that only when such a context had emerged could language be invented.  Tim Ingold stressed how important it is to distinguish different senses of the concept "social", as this context of learning is generally described by the word (e.g. White 1985).  Social relations of the type commonly studied by social anthropologist, like Ingold, are defined by the recognition in language of kinship relationship beyond those of immediate parenthood.  Clearly these could not have emerged before language itself.  But social interactions between individuals, particularly between mother and infant might easily have emerged as continuing relationships in any altricial species.  The context for learning, as seen in the chimpanzees of the Tai forest, may occur among wild-living primates.  The question becomes how such a context came to include language comprehension and production.  Adam Kendon stressed links between vocal and gestural pantomime as a means for the emergence of "displaced reference", a key notion given rather little prominence at the conference, despite its formulation as the issue by Davidson and Noble (1989; Noble & Davidson 1990).

 

THE RELEVANCE OF STUDIES OF NON-HUMAN PRIMATES

The extraordinary discovery of Kanzi's language comprehension, contrasted with the more general inability of chimpanzees to comprehend, must be understood in this context.  Susan Goldin-Meadow commented on the contrast between the learning environments of the deaf children she studied and of the chimpanzees in Sue Savage-Rumbaugh's project.  By not encouraging the deaf children to learn sign language, the parents were providing a less complete opportunity for development than a hearing child would get from hearing parents.  Kanzi, by contrast, has had an enriched environment, for language learning, though as Christophe Boesch pointed out, not for all aspects of learning as a chimpanzee.  We might, therefore, attribute Kanzi's learning to the enrichment of the developmental process. 

Chimpanzees in the wild, in Christophe Boesch's study, also learn from their caregivers, particularly in the processing of nuts of Panda and Coula trees.  The caregivers achieved this by leaving hammers (often stone hammers) on an anvil, leaving nuts on an anvil, or placing both nut and hammer on an anvil.  Such behaviours would be called "scaffolding" by developmental psychologists such as Greenfield.  On two occasions, chimpanzees demonstrated to their infants how to solve problems they could not overcome by themselves.  Such abilities are far in excess of those of even clever monkeys, such as South American Cebus, as shown by Elisabetta Visalberghi in her presentation and another video.  In these capuchin monkeys some problem solving was possible, but not consistently across the studied individuals.  There seemed to be no good argument that Cebus can learn by imitation.

The lesson of this is that there are differences between species, particularly in their abilities to learn.  Kathleen Gibson, one of the organisers, repeatedly stated that "every time I learnt something unique about humans, it wasn't unique".  But to emphasise this at the expense of searching for differences is to run the risk of thinking that chimpanzees are the same as humans.  I do not think any of us really believed that.  Bill McGrew came closest to expressing this view, in an attempt to encourage further thought and research.  He pointed out that none of the field studies of chimpanzees of the last 20 years has concentrated on vocalisations or other possible means of communication, despite anecdotic evidence that chimpanzee behaviour sometimes seems to require it.  I am sure that this conference will stimulate such work on vocalistations and gestures.  Chuck Snowdon reviewed the literature on language-like abilities among other species, arguing that we can sometimes learn as much from convergence as from diverging evolution.  He concluded that, despite the claim for generative grammar in the chickadee, none of the claims for symbolic communication are convincing.

Chimpanzees are clever, but they are not as clever as humans.  The central problem of all chimpanzee studies is that their human-like abilities in the wild seem to be less thatn their abilities in the laboratory.  In a developmental framework we may attribute this to the stimulus provided by human caregivers.  The issue for archaeologists seems to be that which Noble and I have identified: how could the greater abilities of humans have emerged in human evolution without the prior existence of such tutors?

 

LANGUAGE AND THE BRAIN

This begins to sound like an evolutionary question.  The difficulty that many people, some anthropologists included, seem to have with evolutionary arguments is that they cannot divest themselves of notions of progress.  They seem to believe that there are positive pressures creating variability to permit new functions.  Thus the social position would be caricatured as suggesting that once the relationship of caregiving of infants by adults came into being, language would somehow develop in the interaction.  Clearly it cannot be as simple as that.

Similar arguments, with perhaps more reason to be criticised by evolutionary theorists, abound in popular perceptions of the brain and its expansion during the course of hominid evolution.  One of the most robust data of the fossil record is that hominid brain size has tended to increase through time.  The rate of increase for Homo species seems to be greater than that of Australopithecus species, and greater than any documented increase for any other genus.  Falk additionally reported that statistical analysis of brain size estimates, where this is possible, suggests no saltations in brain size in the genus Homo.

It is very tempting to suggest that the brain expanded to allow language or because of language.  But the mechanism can never have been so simple, or so mystical.  Dean Falk summarised the data on brain expansion, and drew attention to gender differences in the dominance of hemispheric lateralisation.  Gender was not an issue pursued at any length in the conference.  Bill Calvin proposed his mechanism for brain expansion: selection for redundancy in the circuitry of the brain through increased advantage from accurate and powerful throwing.  This seems to Noble and myself (Davidson & Noble 1989; Noble & Davidson 1989) the currently most plausible argument, though I was more sceptical of Calvin's adoption of the "frisbee" theory of the use of the handaxe, until we threw the specimen which Nick Toth made for the conference delegates.

One of the most interesting suggestions at the conference was by Kathleen Gibson, who suggested that absolute brain size was really the critical variable.  Other mammals with large brain size are claimed to have complex behaviours associated with brain function: particularly elephants and dolphins.  Rather as Savage-Rumbaugh postulated that syntax is an inevitable result of the organisation of mor complex communicative utterances, so Gibson suggested that complexity of the organisation of the brain was a result of the increase of information processing made possible and necessary by the expanding brain.

Brains are rather difficult to understand for those of us who are not specialists.  As people, we tend to act as if it makes sense to talk of the brain "doing something", as if there were a little person, a homunculus, in there pulling strings, and turning handles.  The reality is much more complex.  But in our efforts to understand persons we tend to seek explanation either by moving outside the person into the social environment in which persons operate, or retreating into the body, and especially the brain as if it operates persons (Noble & Davidson n.d.).  In reality the person is we are interested in taking into account its interactions both with society and the brain, but the complexities of each context are so difficult that it is easy to concentrate on one or the other.  Maybe after Social Archaeology and The Archaeology of Mind it would be appropriate to have an "Archaeology of Persons".

We managed to avoid homunculi, though Broca's Area became so important in some discussions that it was tempting to suggest that that was where the homunculus lives.  Falk has demonstrated that an area that corresponds to Broca's Area can be identified on endocasts of Homo habilis, the question becomes what its significance is.  The plasticity of the brain, to recover from injury, as discussed in Dan Kempler's paper, together with the difficult of identifying precisely where key regions such as Broca's Area are, suggests that the model of the brain as independant organism cannot work.

The brain seems to be an organ that receives stimuli and responds to them.  The particular responses of an adult organism may be a limited set of all of the possible responses that that organism is capable of, and it is the developmental process that establishes those limitations.  There is essentially a difference between what Kanzi can do and what other apes do do and these differences result from the different developmental environments they have experienced.  The brain has unrevealed capacity, and that contributes to the variability in the behaviour of organisms.  But we should not take that to licence interpretations of the archaeological record which assume that capacities unrevealed in the surviving record would be identifiable in the record that has not survived.  In investigating the evolutionary emergence of novel behaviours, the behaviours, whatever they are, that distinguish humans from other species, we cannot rely on arguments that the Olduvai hominids painted "Sunflowers" or typed "The Tempest, but the evidence has just not survived.  That sort of story telling sounds altogether too post-processual, and I assume we are more serious about our science than that.  The real problem, as Bill McGrew identified, is to develop methodology for the interpretation of the archaeological record which will allow us to identify the features of the emergence of such evanescent yet fundamental behaviours as language.

 

ARCHAEOLOGY AND LANGUAGE

There were four papers by archaeologists at the conference:  Gordon Hewes, whose gestural origin theory was one of the early contributions to recent work on language origins; Nick Toth, who spoke from the point of view of his work on replication of stone artefacts; Thomas Wynn, pioneer of application of Piagettian analysis to the individual artefacts of the Lower Palaeolithic; and myself.  There was general agreement that stone tools do not yield obvious information about language competence or performance.  Toth and Schick, however, made the observation that as hominid populations moved out of Africa for the first time they seemed to lose the technology of the Acheulean.  They suggested that this was a result of difficulty in locating suitable raw material supplies, together with a short term mode of communicating the necessary techniques.  In other words, when the first hominids got to a new environment, the difficulty of locating suitable raw material was enough to make them forget how to make handaxes.  And theymay not have reinvented them.  Rather, the Acheulean only spread later when new populations leaving Africa would have had the advantage of interaction with existing hominids to learn suitable stone supplies.  I have advanced a similar argument about the first colonisation of Australia, rather later in date, by fully modern humans.  Stone raw material supply is a major variable in understanding the nature of stone industries.

Wynn was concerned, in his early work, with the operational skills necessary to manufacture artefacts.  By seeking in stone artefacts indications of the necessary spatial competence to manufacture them, Wynn succeeded in identifying some aspects of the developmental stage that the manufacturers must have reached.  This operationalising of the concepts defined by Piaget seems to me far more satisfactory than the published attempts (e.g. by Parker 1985) to match the sequence of stages defined by Piaget with some sequence of stages of human evolution taken from the standard archaeological account.  Nevertheless, it leaves hominids at stages of childhood which seems to describe them as incomplete humans, rather than complete hominids.

A serious problem with Wynn's approach, however, is the degree of necessity in the operations he describes.  Certainly rotation of the core is necessary to manufacture a biface in one sequence (or chain) of action, but was there such a sequence?  My paper with Noble for the conference addressed this sort of issue directly, by pointing to what we call the "finished artefact fallacy".  This is the, often subconscious, attitude which sees every product of prehistoric people as intentional.  The typologies by which we describe stone industries are absolutely wedded to this position.  Elaboration of tool types, defined as flakes with modified or retouched edges, is a conceit of the classifiers rather than necessarily of the manufacturers.  And in our formulation of this as a problem the question goes straight to the issue of language and its recognition in the archaeological record.  The question is about intention, time-depth of intention and hence displacement.

Noble and I have been tempted in our published and unpublished papers to deny any intention at all to all non-human animals.  I don't think we can sustain that.  You see the dog pounce on the ball, drop it at your feet and stand looking at you wagging its tail.  Even I have some difficulty denying that there is some intention there, although I may wish to be slightly more circumspect than some dog-lovers about the precision with which the dog is expressing that intent.  But get out a piece of meat and the dog forgets the ball immediately.  It is a trivial example, but it is informative about the understanding of language.  The dog clearly has intentions and can, in a sense, judge between them - or can it?

It is often suggested that there can be "proto-language" or that there can be more or less complex languages and that there is a continuum between the vocalisations of primates and language.  It is vain to deny that , in an evolutionary process, there are breaks so sharp that a continuum cannot be seen.  nevertheless, we contend that there is a step in the evolution of communication systems that is so sharp that it could be seen as fundamental.  This is the notion of displaced reference, referred to at the conference by Kendon.

Language, by any definition, is a system of symbolic communication capable of referring to things in the absence of those referents.  There is no necessary relationship between the symbol (word) and the referent - this is the notion of arbitrariness.  But there is a convention about the relationship between sign and signified, so that a community of language users all understand the same thing (or nearly the same thing) by the sign.  This arbitrariness defined by convention allows the possibility, and is allowed by the possibility of meaning.  I suggest that it is in the process of infant development that children acquire this sense of meaning, and it is the growth of meaning that joins the comprehension of language to the production of language, and which makes the comprehension of cognition possible.  In the distinction made by Ingold, after Ryle, "knowing how" is transformed into "knowing that".  This is such a fundamental difference between human communication and the communication systems of all other animals that I think it is useful to consider it as a sharp break along the continuum - a high step, such that once it was surmounted people could see further in the real world, and create imaginary worlds and share them with others.  They could engage in what Parker, at the conference, called declarative planning.

There is general agreement that this transformation had occured by the time of the Upper Palaeolithic of Europe.  Noble and I (Noble & Davidson 1989) have suggested that the imagination required to build a boat to get to the unknown shores of Australia before 40 thousand years ago required such knowledge.  The question is how much earlier it can be seen.  And this is the interest of handaxes. 

Handaxes are the most obviously designed product of the lower Palaeolithic.  Did that their production necessarily involve design, knowing the form of the end product?  John Gowlett has argued that it did, I would like to suggest otherwise.  Consider, for a moment, the proposition.  A handaxe takes about 40 minutes to make, depending on the prettiness desired, the skill of the knappper and the nature of the raw material.  The proposition is that hominids invested this sort of time in making a tool for future use.  If you could demonstrate that sort of delay (or displacement in time) between conception of need and the possibility of satisfying it, then hominids would be far more competent that I take them to be.  To adjust an Australian expression, that is far more outstanding than dogs' balls.

So how do I account for handaxes?  Was there an intention to produce the stereotyped shape so familiar to us all?  Well, the first question I have is how stereotyped are they?  Or rather, in a complete Acheulean assemblage (and how do you measure completeness?), how much can we say that the only intended product was a handaxe?  Or do archaeologists selectively define handaxes by shape or technology of production such that the stereotypy derives from the classification process?  Most fundamentally, were the flakes used?  There are some indications from Potts' recent excavations at Olorgesaile that they were.  And if flakes were used, can the production of bifaces be simplified into decisions made in the context of producing flakes, which reduces the time depth of the intentionality?  I find that these questions are rarely being asked, but they hold the key to understanding hominid evolution.  If hominids had 40 minute depth to their intentionality, I cannot imagine that they did not have language.  But if they had language, I cannot imagine that they did not create more than they did.  It seems strange that there should be no products other than tools that could be considered to have symbolic content, defined by arbitrariness and convention, earlier than the Upper Palaeolithic and the colonisation of Australia.  There is the paradox.

 

CONCLUSIONS

I began by discussing Kanzi and infant development and have ended with handaxes.  How are these related, except by being topics at the conference?

There is general agreement that by the time of the colonisation of Australia there was displaced reference, hence language.  Yet there seems to be a problem that the earliest appearance of anatomically modern Homo sapiens is accompanied by Mousterian industries, as are the latest archaic Homo sapiens - and in these industries too we might explain the variability by imitative learning, restricted patterns of motor action, and modification through or for use.  All these features we might associate with hominids lacking the creativity of language-using humans.  In the back of the mind (another of those danger words) of many people is an anxiety because their expectation is that once the skeletal morphology looks like us then they should be behaving like us.  It is a hangover, I think, from the erroneous subliminal belief that natural selection creates variability.  Certainly Lieberman seemed worried by the expectation.  I believe that this has always been unlikely - what we have to account for is why one end of the range of variation of a polymorphic species should have become more successful than another.  And here, I think, Lieberman and Kanzi hold the key. 

Lieberman long ago demonstrated that there were differences in the ability of different morphologies to vocalise, whether or not you accept all of his claims about this.  Kanzi's language comprehension and novel vocalisations demonstrate that performances that are not part of the normal repertoire of a species may be stimulated by a favourable developmental environment.  What has evolved, in fact, is development.  Once displaced reference and symbols in communication were invented, and could become part of the developmental environment of hominids, syntax and semantics could expand. 

The evidence from gestures of deaf children of hearing parents, from the gesticulations that accompany speech in normal humans, and from the invention of gestures by chimpanzees in tool use, suggests that Gordon Hewes was close to the answer - more complex communication arose through gesture.  As Sue Savage-Rumbaugh suggested, such gestures "coalesce on tools" and thus may "carry things through time".  But this will not achieve displacement.  That requires that the gestures somehow become dissociated from the context and contingency of use - and that is the scenario that Noble and I produced in our Current Anthropology paper.