Language and Species

Derek Bickerton

• Publisher: University of Chicago Press
• Year: 2018
• ISBN:
  9780226220949
• Topic: Languages & Linguistics
• Subtopic: Linguistics
• Index: Languages & Linguistics

Language and Species presents the most detailed and well-documented scenario to date of the origins of language. Drawing on "living linguistic fossils" such as "ape talk, " the "two-word" stage of small children, and pidgin languages, and on recent discoveries in paleoanthropology, Bickerton shows how a primitive "protolanguage" could have offered Homo erectus a novel ecological niche. He goes on to demonstrate how this protolanguage could have developed into the languages we speak today."You are drawn into [Bickerton's] appreciation of the dominant role language plays not only in what we say, but in what we think and, therefore, what we are."—Robert Wright, New York Times Book Review "The evolution of language is a fascinating topic, and Bickerton's Language and Species is the best introduction we have."—John C. Marshall, Nature

1

The Continuity Paradox

Anyone who sets out to describe the role played by language in the development of our species is at once confronted by an apparent paradox, the Paradox of Continuity. If such a person accepts the theory of evolution, that person must accept also that language is no more than an evolutionary adaptation—one of an unusual kind, perhaps, yet formed by the same processes that have formed countless other adaptations. If that is the case, then language cannot be as novel as it seems, for evolutionary adaptations do not emerge out of the blue.

There are two ways in which evolution can produce novel elements: by the recombination of existing genes in the course of normal breeding, or by mutations that affect genes directly. Even in the second case, absolute novelties are impossible. What happens in mutation is that the instructions for producing part of a particular type of creature are altered. Instructions for producing a new part cannot simply be added to the old recipe. There must already exist specific instructions that are capable of being altered, to a greater or lesser extent. What this means is that language cannot be wholly without antecedents of some kind.

But what kind of antecedents could language have? Since language is so widely regarded as a means of communication, the answer seems obvious: earlier systems of animal communication. It has long been known that many species communicate with one another. Some, like fireflies, have blinking lights, others, like crickets, rub legs or wingcases together, while many exude chemical signals known as pheromones. Of course such means are limited in their range of potential meaning and may signal nothing more complex than the presence of a potential mate. But the more sophisticated the creature, the more sophisticated the means—from the dances of honeybees, through the posturing of sea gulls, to the sonar of dolphins—hence, the more complex the information that can be conveyed. Could not human language be just a supersophisticated variant of these?

The trouble is that the differences between language and the most sophisticated systems of animal communication that we are so far aware of are qualitative rather than quantitative. All such systems have a fixed and finite number of topics on which information can be exchanged, whereas in language the list is open-ended, indeed infinite. All such systems have a finite and indeed strictly limited number of ways in which message components can be combined, if they can be combined at all. In language the possibilities of combination, while governed by strict principles, are (potentially at least) infinite, limited for practical purposes only by the finiteness of the immediate memory store. You do not get from a finite number to infinity merely by adding numbers. And there are subtler but equally far-reaching differences between language and animal communication that make it impossible to regard the one as antecedent to the other.

But the net result of all this is the Paradox of Continuity: language must have evolved out of some prior system, and yet there does not seem to be any such system out of which it could have evolved. Until now, arguments about the nature, origin, and function of language have remained inextricably mired in this paradox. Let us see if there is any way in which they can be released from it.

A WORD ABOUT FORMALISM

We can at least clean a little of the mud from our wheels if we begin by tackling what might seem at first an unpromising and unrelated issue: the role that formal structure plays in language. Some linguists will tell you that the formal structure of language is very important. Others will tell you that it is relatively unimportant. Who is right?

There are two very odd imbalances between the formalist and antiformalist groups. The first imbalance is in what they believe. No formalist believes that a purely formal approach is the only way to study language. Any formalist would agree that there are many aspects of language—meaning, use, interaction with other social and psychological domains—that are all worthy of study. If you ask formalists why they insist on studying formal structure in isolation from all these other factors, they will probably tell you that significant advances in knowledge have always involved focusing on particular aspects of things and abstracting away from other aspects. They can see no reason for the study of our own species to reverse this sensible procedure.

But if you ask antiformalists why they ignore the formal structure of language, you will sometimes hear a much less tolerant story. They may tell you that it is quite senseless to study the formal aspects of language in isolation from its mode of functioning in society. Quite possibly they will go on to say that since those aspects are merely uninteresting mechanisms, or superficial trimmings, or even artefacts of the method of inquiry, they can be relegated to an inferior position, if not dismissed altogether.

The second imbalance between formalists and antiformalists is that since formalists have ignored all issues involving the evolution of language, that field has been yielded without a blow to the enemies of formalism. Subsequently there has been no significant interchange between the two sides, indeed they are barely on speaking terms. This has left the antiformalists alone to grapple with the Continuity Paradox.

Now to tackle a paradox, or indeed any research issue, from a one-sided position is not the best recipe for success. In large part, failure to resolve the Continuity Paradox has resulted precisely from what one might call the ‘naive continuism’ of the antiformalists, who have tried in a variety of ways to establish a direct line of development from animal communication to human language. Although all their efforts have signally failed to produce a convincing ‘origins’ story, their rejection of more formal approaches has left them without any viable alternative.

Accordingly the present work tackles the Paradox from a rather different viewpoint. This viewpoint takes as basic the assumption that formal properties of language do exist and do matter, and that without the very specific types of formal structure that language exhibits, it could not perform the social and communicative functions that it does perform, and could not convey the wealth of peculiarly human meaning that it does convey.

Those functions and that meaning should not—and, indeed, in a work of this nature literally can not—be ignored or even minimized. However, it seems reasonable to stand the antiformalist position on its head and say that it is quite senseless to study the origins and functions of language without at the same time studying the formal structures that underlie those functions. For these formal structures, abstract though they may appear, are exactly what enable language to communicate so efficiently. Nothing else that we know of (or can imagine) could have given language the unprecedented power that it proved to have: power that gave to a single primate line the mastery of the physical world and the first, and perhaps only, entry into the world of consciousness.

THE GULF BETWEEN LANGUAGE AND ANIMAL COMMUNICATION

Having established this perspective, we can now look a little more closely at the ways in which animal communication differs from language. Perhaps the most obvious is that of productivity. The calls or signs of other creatures usually occur in isolation from one another. There are as yet few, if any, clear cases where they can be combined to form longer utterances whose meaning differs from the sum of their meanings in isolation, in the way in which look out!, for instance, differs from the sum of the meanings of look and out.

It is not impossible that future research will uncover such cases. But then, if we were to parallel language, we would have to look for cases where the same calls in a different order can mean different things, like Dog bites man versus Man bites dog. Even this far from exhausts the possibilities of human syntax, which can also place similar words in different orders to mean the same thing (John gave Mary the book, Mary was given the book by John) or the same words in almost the same order to mean quite different things (The woman that saw the man kicked the dog, The woman saw the man that kicked the dog).

Note however that to achieve such effects we have to use elements like -en, by, that. Later on we shall look at such elements in more detail. For the moment it is sufficient to note that they differ from elements like John or woman in that the latter refer (if only indirectly) to some entity or class of entities in the real world, whereas the former do not really refer at all, but rather serve to express structural relations between items that do refer. The first class of elements can be described as grammatical items and the second, the class that refers, as lexical items. To which class of items do animal calls and signs belong?

Certainly there seems to be nothing in any animal communication system that corresponds even vaguely to grammatical items. But it is also questionable, in at least a large majority of cases, whether there is any true correspondence with lexical items either. We may find, for example, a particular facial expression, accompanied perhaps by a bristling of hair, that we might want to translate as I am very angry with you, or a peculiar cry that perhaps we would translate as, Look out, folks, something dangerous is coming! In other words, most elements in animal communication systems might seem to correspond, in a very rough and ready sense, with complete human utterances, rather than with single words per se. But note that the true correspondence is with utterance rather than sentence, because oftentimes a single-word utterance like Help! or Danger! would serve as well. The category complete utterance, however, is not a structural category in language, precisely because it can cover anything from a complex sentence (or even a series of such sentences) to a one-word exclamation.

It follows that, for the most part, the units in animal communication systems do not correspond with any of the units that compose human language. There is a good reason why this is so. Animal communication is holistic, that is to say it is concerned with communicating whole situations. Language, on the other hand, talks mainly about entities (whether other creatures, objects, or ideas) and things predicated of entities (whether actions, events, states, or processes).

The units of animal communication convey whole chunks of information (rough equivalents of I am angry, You may mate with me, A predator just appeared). Language breaks up those chunks in a way that, to the best of our knowledge, no animal communication system has ever done. In order to convey our anger, we must, as an absolute minimum, specify ourselves by a particular sign and the state in which we find ourselves by another sign (in English and numerous other languages we have, in addition, to use an almost meaningless verb in order to link ourselves with our current state, while in another set of languages, we would have to add a particle to indicate that our state was indeed current, not a past or future one).

If we think about it, this way of doing things may seem somewhat less natural than the animal way. Suppose that the situation we want to convey is one in which we have just seen a predator approaching. From a functional point of view, it might seem a lot quicker to let out a single call with that meaning, rather than Look out! A lion’s coming! But the oddity is not just functional. In the real situation, it is simply not the case that we would see two things: an entity (the lion) and something predicated of that entity (’coming’). If we actually were in that situation, what we would perceive would be the frontal presentation of a lion getting rapidly larger. That is, we would experience a single intact cluster of ongoing perceptions. So the animal’s representation of this would seem to be not merely more expeditious, but more in accord with reality than ours.

But there are, even in this limited example, compensating features. A generalized predator warning call, or even a specific lion warning call, could not be modified so as to become A lion was coming (as in the context of a story), A lion may come (to propose caution in advance), No lions are coming (to convey reassurance), Many lions are coming! (to prompt still more vigorous evasive measures), and so on. To achieve this kind of flexibility, any utterance has to be composed of a number of different units each of which may be modified or replaced so as to transmit a wide range of different messages. And after all, if we want a rapid response, the possession of language in no way inhibits use of the human call system. In a tight corner, we can still just yell.

Still, you might argue, language had to begin somewhere, and where is it most likely to have begun than in some particular call whose meaning was progressively narrowed until it now covers about the same semantic range as does some noun in a language? Once the species had acquired a short list of entities—lions, snakes, or whatever—it needed only to attribute states or actions to those entities and it would then already have the essential subject-predicate core of language, to which all other properties could subsequently have been added.

You might then point to creatures such as the vervet monkey which have highly developed alarm calls. The vervet, a species that lives in East Africa, has at least three distinct alarm calls that might seem to refer to three species that are likely to prey upon vervets: pythons, martial eagles, and leopards. That it is the calls themselves that have this reference, and not any other behavioral or environmental feature, has been experimentally established by playing recordings of the calls to troops of vervets in the absence of any of the predators concerned. On hearing these recordings, most vervets within earshot respond just as they would to a natural, predator-stimulated call. They look at the ground around them on hearing the snake warning, run up trees on hearing the leopard warning, and descend from trees to hide among bushes on hearing the eagle warning.

We might therefore think that these calls were, in embryo at least, the vervet ‘words’ for the species concerned. But in fact, a warning call about pythons differs from a word like python in a variety of ways. Even though python is only a single word, it can be modified, just as we saw the sentence A lion’s coming! could be modified. It can, for instance, be given at least four different intonations, each of which has a distinct meaning. With a rising intonation it can mean ‘Is that a python there?’ or ‘Did you just say python?’. With a neutral intonation, it merely names a particular variety of snake, as in a list of snake species, for example. With a sustained high-pitch intonation it can mean that there’s a python right there, right now. With an intonation that starts high and ends low, especially if delivered in a sneering, sarcastic tone, it can mean ‘How ridiculous to suppose that there’s a python there!’

Assuming that all these are used without intent to deceive, only in the third case is there a python there for sure. But with the vervet call, there is always a python there. At least, with one rare exception, the vervet involved genuinely believes there is a python there. (Just like human children, young vervets have to learn the semantic range of their calls, and again like children they tend to overgeneralize and sometimes give calls in inappropriate circumstances).

In order to understand further differences between humans and vervets, certain aspects of meaning must first be clarified. We might suppose that any relation between events in the world and meaningful utterances could be characterized as a mapping relation, that is to say, an operation that matches features of the environment with features of a (more or less arbitrary) representational system. We might begin by saying that a python in the real world is matched with a particular call in the vervet system and a particular noun in a given human language. This would be not very far from Bertrand Russell’s theory of meaning and reference, for Russell believed, and got into terrible difficulties through believing, that nouns referred directly to entities in the real world.

Linguists, at least since Russell’s contemporary de Saussure, have known that this is not so for human language. As noted above, grammatical items do not refer at all, and lexical items refer to real-world entities only indirectly. This is because not one, but at least two mapping operations lie between the real world and language. First our sense perceptions of the world are mapped onto a conceptual representation, and then this conceptual representation is mapped onto a linguistic representation.

Indeed, even in the animal case there cannot be a direct relationship between external object and call. Every now and then, even adult vervets will use, say, an eagle call for something that is not an eagle. It is no help to say that the vervet merely made a mistake. Why did it make that mistake? Because it thought that what it saw really was an eagle. In other words, if the vervet is wrong, it is wrong because it is responding to its own act of identification, rather than to the object itself. But are we then to say that the vervet responds to its own identification when it happens to be wrong and to the real object when it happens to be right? Obviously not. Vervets respond to their own identifications under all circumstances. But in that case there cannot be a direct link between call and object. The call labels an act of identification: the placing of some phenomenon in a particular category. In some sense, vervets too must have concepts.

That the things words refer to are not external entities is even clearer in our own case. One piece of evidence is the very existence of expressions like a unicorn or the golden mountain that gave Russell so much trouble. Since such expressions cannot refer to real-world entities, they must refer to a level of r...

Table of contents

• Cover
• Copyright
• Title Page
• Dedication
• Epigraph
• Contents
• Preface
• Introduction
• 1. The Continuity Paradox
• 2. Language as Representation: The Atlas
• 3. Language as Representation: The Itineraries
• 4. The Origins of Representational Systems
• 5. The Fossils of Language
• 6. The World of the Protolanguage
• 7. From Protolanguage to Language
• 8. Mind, Consciousness, and Knowledge
• 9. The Nature of the Species
• Epilogue
• Notes
• References
• Index