A neuroscientist and a linguist show how evolution could have given rise to structured language.
A machine for language? Certainly, say the neurophysiologists, busy studying the language specializations of the human brain and trying to identify their evolutionary antecedents. Linguists such as Noam Chomsky talk about machinelike "modules" in the brain for syntax, arguing that language is more an instinct (a complex behavior triggered by simple environmental stimuli) than an acquired skill like riding a bicycle.
But structured language presents the same evolutionary problems as feathered forelimbs for flight: you need a lot of specializations to fly even a little bit. How do you get them, if evolution has no foresight and the intermediate stages do not have intermediate payoffs? Some say that the Darwinian scheme for gradual species self-improvement cannot explain our most valued human capability, the one that sets us so far above the apes, language itself.
William Calvin and Derek Bickerton suggest that other evolutionary developments, not directly related to language, allowed language to evolve in a way that eventually promoted a Chomskian syntax. They compare these intermediate behaviors to the curb-cuts originally intended for wheelchair users. Their usefulness was soon discovered by users of strollers, shopping carts, rollerblades, and so on. The authors argue that reciprocal altruism and ballistic movement planning were "curb-cuts" that indirectly promoted the formation of structured language. Written in the form of a dialogue set in Bellagio, Italy, Lingua ex Machina presents an engaging challenge to those who view the human capacity for language as a winner-take-all war between Chomsky and Darwin.
About the Author
William H. Calvin is Affiliate Professor of Psychiatry and Behavioral Sciences at the University of Washington, Seattle. His books include The Cerebral Code (MIT Press, 1996).
Read an Excerpt
The Villa Serbelloni
People at dinner last night kept asking me what Chomsky's innate grammar is all about. Where is this language macromutation in the brain, and all that?
Wrong question, of course, but it's a sure sign they've gotten used to the amazing view of Lake Como from the terrace where we eat at the Villa Serbelloni, on a long table with several dozen interesting people. You'll see when you arrive. If there's a clear evening before I get back from Milan, remember to watch for the last of the sunset over the Dolomites.
Provided, of course, the other "residents" give you a chance. Several confessed to reading up on our subject, in anticipation of our arrival for a month of writing about the brain and language. It forcefully reminded me that Chomsky's innateness has been the intellectual spectator sport of the last four decades. I tried to explain to them that some gene-specified aspect was unsurprising to a biologist that you and I hoped to flesh it out with appropriate anthropology and neuroscience in a way that Chomsky wasn't particularly interested in doing, and to provide some evolutionary proposals that wouldn't rely on macromutations and the like.
I also tried to explain your notion of protolanguage put forth in Language and Species, with a good supply of words but with sentence length limited to only a few words by the lack of structural elements such as phrases and clauses. Protolanguage has no way of saying who did what to whom, not without anenormous effort. I emphasized that there was a large gulf between protolanguage and our full-fledged syntax without any obvious intermediate states, quite a jump from my pidgin Italian to being able to nest four verbs in saying, "I think I saw him leave to go home."
It's going to be challenging for us to try and describe how the gulf was first bridged by evolutionary processes. I hope we can avoid the deus ex machina quality of some of the previous attempts to explain the origins of language ability, the ones that finally seize upon a slender, unsupported reed as the way out of the muddied morass the equivalent of that "god machine" the ancient Greek playwrights wheeled in to solve thorny plot problems. Yet it is a language machine we're searching for, one capable of those elaborate maneuvers seen in language with syntax (you don't have to think about it; indeed, you can't turn language recognition off), but conforming to some design constraints imposed by the neurobiology (what it's possible to do with mere neural circuits) and the evolutionary history (up from apelike communication and mental powers in only five million years, each stage bootstrapping the next).
But, in a broader view, language is just our best example of the whole range of higher intellectual functions. Our lingua ex machina probably needs to be able to handle creative shaping up of quality (for instance, figuring out what to do with the leftovers in the refrigerator), long-range planning, procedural games, and even music. Solve the structural basis for one, and you might solve them all.
I think that the linguists' conceit, that syntax is what thought is all about (and that without syntax, you couldn't think with any depth or originality), reflects a useful strategy for brain researchers, simply because syntax provides a lot of useful constraints on theorizing. But other parts of higher intellectual function might be even more useful in that regard. Want to lay any bets that we would discover more about higher intellectual function via studying music in the brain? Yes, music seems likely to be a spare-time use of the neural machinery evolved for thought and language but we might be able to separate the issues of vocabulary and structuring better in music, where you have structure without predication, as the Israeli musicologist Ruth Katz reminded me at dinner! What's unmusical in any culture might tell us what the neurons can't do.
Intelligence (in our sense of versatility in dealing with novel situations) is a particularly intriguing part of the puzzle of higher intellectual functions. But as Ernst Mayr once said, most species are not intelligent, which suggests "that high intelligence is not at all favored by natural selection" or that it's very hard to achieve. So our look at bootstrapping syntax also needs to keep in mind this more general problem of finding indirect ways of achieving intelligence. What gives rise to syntax might also give intelligence a big boost.
Evolution, after all, is full of sidesteps, such as those conversions of function that Darwin identified. Wheelchair considerations may be what "paid for" all of those curb cuts on every corner, but most of their subsequent use involves wheeled suitcases, baby carriages, grocery carts, skateboards, bicycles, and other uses that would never have paid for it. Some of the underpinnings of language may be secondary uses as well, so we need to watch for free "curb cuts" affecting syntax.
See you soon.
Well, when I got greeted with a facetious "Calvin tells us that the two of you are going to out-Chomsky Chomsky," I started to wonder what you'd been telling them. Then I remembered that whatever you tell people about Chomsky, they seem to get hold of the wrong end of the stick. Some people get no respect, others get no comprehension. If what Chomsky said about innate capacities had been said about any species but ours, everyone would have accepted it years ago. The evidence that language is a biologically determined, species-specific, genetically transmitted capacity is simply overwhelming, no matter how many people try to chip away at isolated bits of it. But somehow humans are supposed to be special. The same rules don't apply. The idea that our prize possession, language, is just some mechanical thing seems very threatening to some people.
Unfortunately, Chomsky is unwilling to look either at the neurological infrastructure of language or at the ways it might have evolved. Why he's unwilling is neither here nor there. That's his business. No one has to do everything. But obviously, once it's established that language is biologically determined, the next step is for someone to try and find out exactly how it evolved. And once it's established that language is rooted in the structure of the brain, the next step is to go looking for it there. These three things language, evolution, and the brain it seems to me, are interlocking. You can't really look at any one of them without looking at the others. If you want to know how language evolved or how it operates via brain mechanisms, you've got to know exactly what it's like how it differs from bee dances and chimp calls. But you really can't be sure what it's like until you've seen how it evolved or how it works in the brain. All three areas of knowledge should be feeding one another, but they're not. And that's the king-size hole in our understanding of ourselves that I'm hoping, between us, we might be able to plug up a bit in the next month.
However, once we start looking at how language could have evolved and how the brain does the job, we become aware that some of the ways linguists have looked at language are more than a little awkward. Over the last decade or so, an enormous amount has been written about the evolution of language all the more enormous when you consider how little we know about it. Some of what has been written is sensible; much, alas, is not. To go through this literature, to evaluate it, to show how our ideas differ from it, would be an enormous task, one that would inevitably get in the way of the concrete proposals we have to make. So we're not going to go through it and we're not going to criticize anyone else's approach. In the endnotes we'll show where alternative answers can be found.
Well, Chomsky's term "language organ" might have been unfortunate, as was some of the early supercharger-type imagery used to describe how language might have been tacked onto an ape brain, as were the cardboard notions of how evolution works (those deus ex machina macromutations). But I have no quarrel with what I take to be the heart of Chomsky's argument, that human brains are predisposed to use certain types of syntax and not other possible schemes and that it wasn't, obvious how to do this from textbook versions of Darwinism. Today, we'd probably emphasize a baby's predisposition to discover patterns in the language (or invent, in the case of creoles) and thereby softwire a language machine in one of the neurologically possible self-organizing schemes, rather than speaking of something being innate from conception onward. But that's just the current state of the ever present nature-nurture debate.
There are lots of little brain areas, the size of a small coin, one of whose functions is particularly specialized say, naming inanimate objects. I'll give some examples when I eventually discuss where concepts are located in the temporal lobe. We still tend, following Gall's phrenology, to give functional names as if an area were exclusively concerned with the named function. But most areas are multifunctional; we merely discover one function that compels our attention and name the area after it! And so onwards to the reification fallacy (it has a name, therefore it must be a thing).
But certainly the language specializations of the brain are not exclusive; the same areas of brain have a lot to do with inventing oral-facial and hand-arm movement sequences, and with judging sound sequences and these functions probably all evolved together. Their brain real estate may well constitute a common facility, one used not just for language tasks but for any involved sequence, whether sensation or movement or thought just as curb cuts are now multiple-use though paid for by a single use.
Structure is one way to look for the physical basis of real language, but you can also ask how each individual develops the functionality during early life. Part of the language instinct could turn out to be something very simple say, a real fascination on the part of the young human with discovering any hidden patterns in the sensory environment, such as the repeated strings of speech sounds that we call words. We may happen upon crystal-like self-organizing tendencies in the neural circuitry that preserve them tendencies that aren't likely to come from experience. That way, after discovering words in the auditory barrage, you can go on to discover the pattern of words that we call a "question." There could be one stage after another of searching for higher- and higher-order patterns, each making use of the same automata propensities in the neural circuitry.
So language acquisition might consist of the discovery of patterns in the environment, some of which can be remembered by patterns in the brain. Just as some types of crystal are more common than others, so syntax might settle into certain patterns more than others. Those patterns are what, I take it, "Universal Grammar" refers to. Rather than a gene for a language machine, you might have an epigenetic tendency to seek hidden patterns in your sensory environment which, together with the brain's potential for creating varied "crystals" shaped up by previous evolution, would give you the syntax that makes us so different from the apes.
Now, Derek, let me summarize what we said about book organization at breakfast and afterwards, when we walked up to the castle on the cleaver. I need a little aide memoir for my fallible brain, the sort of thing that politicians write in their diaries to save for the day when they write their memoirs.
We're not trying to write the book about language origins, one that covers the landscape of interesting ideas that float around at conferences on language origins. We'll be happy just to show several powerful ways of getting from ape behaviors to syntax without relying on the usefulness of communication per se. We're after invention, not improvement.
Our imagined audience is not unlike the other residents here at the Villa Serbelloni: the typical serious reader, but not necessarily in the sciences (the artists and poets hereabouts ought to find it of interest and be able to follow our explanations). As for content, well, as Ernst Mayr likes to say, the big scientific questions tend to resolve around what, how, and why. And they're all interlocking; one's incomplete without the others (though we often pretend otherwise, as when we focus on one aspect at a time as "the answer"). So we might want to structure our Bellagio book around the relevant what-how-why questions.
What's a word, anyway? What's a simple utterance of a few words? And, since a longer sentence is not just a heap of words any more than a house is merely a heap of construction materials, what's all this argument structure and phrase structure that constitutes syntax or used to, back before minimalism struck? And what about all those little closed-class words of grammar, such as the articles and prepositions? What are the stages of a child's development of language?
How does the brain represent a word? How does it link words up? How does it store new memories, retrieve them? How does the brain invent a novel utterance, without it being complete nonsense most of the time? How does language deteriorate in strokes?
But all the linguistic what's and neurophysiological how's are incomplete without the evolutionary why's, those step-by-step explanations about how things came to be the way they work now, explanations involving Darwin's bootstrap. Why is it unlikely that words evolved out of primate cries and calls? Why did our particular kind of syntax evolve? Does it have anything to do with the brain expanding fourfold during the ice ages?
What's our best scenario for a step from primate cries to protolanguage? From lots of vocabulary up to the use of syntax for facilitating long sentences like this one? We'll need to talk about the relationship between evolving language and all the other changes evolution made in the typical ape (I know you want to discuss the extensive advance in sharing food and doing favors for friends). Then, with some examples in hand, we ought to discuss what would constitute a really satisfying explanation, covering the whole spectrum of such questions about language and the rest of the higher intellectual functions that separate us from the smartest apes. The "unfinished agenda," as it were.
Though often considered as a gradual series of improvements in efficiency, evolution is also characterized by a string of Good Tricks conserved by evolution and reused in different contexts. Many biological structures are multipurpose, and the most obvious "function" of a structure may change over time. Darwin's example was the swim bladder of the fish, most obviously useful for adjusting buoyancy via filling up a balloon with blood gases, but also useful as an interface for exchanging blood gases with the atmosphere, a simple lung that might allow the fish to crawl ashore. Darwin may not have known about curb cuts, but he spoke of a conversion of function, and warned that selection favoring one function might well benefit another function. We'd probably say that selection for language abilities benefitted musical abilities, because it's so hard to figure out what evolutionary circumstances would have rewarded four-part harmony. There may be no free lunches in some ultimate sense, but there sure is a lot of bundling of products; pay for one, get something else "free." And the minor product may turn into the major one in the long run, enormously aided by the initial natural selection that paid in a different coin for the other.
Also, because structures are so easily duplicated, once you have the genes for one, it is possible to specialize in several directions at once. Our chromosomes are filled with nonfunctional near-duplicates of the functional genes; that's exactly how any computer programmer would operate, doing experiments on copies of the functioning program, eventually using the new one more and more of the time as the bugs are eliminated.
Simple rules generate complex patterns. (The big lesson of fractals and chaos!) Some variants on the existing rules are stable (most are nonsense, others quickly undo themselves), and so one sees self-organizing systems bootstrapping themselves in what Jacob Bronowski calls stratified stability. The stabilities are, of course, somewhat confining just as the steep fjordlike walls of the Como valley made it easier for the ancient glacier to go certain directions rather than in others.
That's the sort of thing we ought to see with language evolution: experimental advances above a plateau of stable function (like your protolanguage), sometimes discovering a new stable level (like structured utterances), but with confines developing as you go.
Levels of organization are, fortunately for us, a commonplace in technology. As an example of four levels, fleece is organized into yarn, which is woven into cloth, which can be arranged into clothing. Each of these levels of organization is transiently stable, with ratchetlike mechanisms that prevent backsliding: fabrics are woven to prevent their disorganization into so much yarn; yarn is spun to keep it from backsliding into fleece.
A proper level is also characterized by "causal decoupling" from adjacent levels. For example, you can weave without understanding how to spin yarn (or make clothing). Many of the branches of science are founded around a single level of organization. Mendeleyev figured out the table of elements and predicted the weight and binding properties of undiscovered elements, long before anyone knew about atomic spectra or biochemistry. As a chemist, it helps to know the electron orbits that underlie chemical bonds, and it may help to understand an overlying level such as stereochemistry, but most of chemistry is a set of relationships within a level just like weaving, a subject in its own right.
Within the brain sciences, we have to cope with close to a dozen levels of organization (and so we frequently argue about whether learning is a matter of gene expression, ion channel, synaptic, neuron, or circuit-level alterations). We can even invent new levels on the fly, such as analogies, though most of them don't last for very long.
But some do. Among the major tasks of early childhood are the discovery of four levels of organization in the apparent chaos of the surrounding environment. Infants discover phonemes and create standard categories for them. With a set of basic speech sounds, babies start discovering patterns amid strings of phonemes, averaging nine new words every day. Between 18 and 36 months of age, they start to discover patterns of words called phrases and clauses, adding -s for plural, adding -ed for past tense. After syntax, they then go on to discover Aristotle's rule about narratives having a beginning, middle, and end. Thus in four years, children "pyramid" four levels of organization, each with its own rules that are causally decoupled from the underlying level's rules. I'd caution that levels don't mean orderly hierarchies: you might have several different levels taking off from an earlier one, more like a tree or a web than a ladder.
It is tempting to treat consciousness as the highest level of organization that you've currently got cooking. When you first contemplate the toothpaste in the morning, the level of consciousness might not be very high, operating merely at the level of objects or simple actions. Handling relations (such as speaking in sentences) may become possible only after your morning coffee. The relations between relationships level (analogies) may require a double espresso. Poets, of course, have to compare metaphors, which requires a series of stage-setting preliminaries. Writers attempt to dramatically shape their materials to result in, as Sven Birkerts said in The Gutenberg Elegies, "a kindled-up sort of high."
Understanding such staging might allow us to spend more time at more abstract levels or even invent a new level in this house of cards, if the prior levels can be sufficiently shored up. I can almost imagine a meta-poet taking a long walk here at the Villa Serbelloni, trying to stage manage yet another level atop the earlier shaky edifice, inventing meta-metaphors.
So, Derek, I wonder if your protolanguage isn't just going to be a level of relationships mostly associations between a few objects and a verb atop which syntax can operate as a new, more structured level. And that some sort of meta-syntax could operate atop it, in turn.
What you want the neuro to provide, as I understand it, is a nice clear step up from protolanguage to syntax, the brain finally getting its act together because of one important improvement that, together with what's already in place, provides an emergent property, syntax. The committee can finally do something that all the separate parts couldn't. It might be like adding a capstone to an arch, which permits the other stones to support themselves without scaffolding as a committee, they can defy gravity. Our task as scientists is, in part, to imagine the scaffolding that could have put such a stable structure in place initially.
I can imagine some Good Tricks that might provide that big step, allowing for the recursive nature of embedded phrases and a considerable improvement in speed of operation. A big step doesn't necessarily mean that performance suddenly flowers. Graded improvement of function can still occur via the amount of time that you utilize the Good Trick, or the number of situations to which you apply it, or the intensification of culture occasioned by widespread use (more vocabulary invented, etc.). But I think that I can give you something without intermediate syntax levels, something that will deteriorate back to protolanguage in a fairly obvious manner without intermediate stops (I've never heard of an aphasic patient able to embed two deep, but not three). And that recursive phrases and clauses will emerge in our lingua ex machina as neatly as they do in the child's third year.
* * *
DB: I know that you like to make fun (as do I) of attempts to leap from the sub-basement of quantum mechanics to the penthouse of consciousness, but doesn't biological psychiatry attempt to leap from a gene to a psychosis?
WHC: Ah, the "gene" for schizophrenia. But such things merely show that a high level is dependent on the whole edifice. A crack can indeed propagate upwards through a dozen levels. Just as sparkplug failure causes an occasional traffic jam, so a misread gene can occasionally set up a psychosis. But if you want to understand the typical traffic jam in the middle of nowhere, you have to understand how the packing density of vehicles moving at somewhat different speeds can, in combination with a hill to climb, create a traffic jam even where there is no traffic entering or leaving the freeway. So, too, understanding delusions and hallucinations means knowing how thought builds on the immediately-underlying layers of the edifice. We'll need to "stand under" thought and appreciate how network dynamics are structured by syntaxlike processes, so we can nest sentences. That a bad gene can disrupt it simply doesn't explain very much. Useful explanations require relevant foundations, not just another "everything is connected to everything" demonstration.
Table of Contents
|1.||The Villa Serbelloni||1|
|2.||What Are Words?||13|
|3.||Why Putting Words Together Isn't Easy||27|
|4.||Bigger than a Word, Smaller than a Sentence||41|
|5.||Language in the Brain||55|
|6.||How Are Memories Stored?||67|
|7.||Hexagonal Mosaics and Darwin Machines||75|
|8.||A Common Code: The Brain"s "Esperanto" Problem||93|
|10.||Reciprocal Altruism as the Predecessor of Argument Structure||123|
|11.||Role Links for Words||135|
|12.||The Word Tree as a Secondary Use of Throwing's Segmented Movement Planner||151|
|13.||Corticocortical Coherence Promotes a Many-Voiced Symphonic Sentence||169|
|14.||The Pump and the Slingshot||183|
|15.||Darwin and Chomsky Together at Last||195|
|About the Authors||281|
What People are Saying About This
In a lively back-and-forth across disciplines, Calvin and Bickerton wrestle language to the mat in a way no single author could. Why did language evolve, and kind of existing blueprints in the brain did it enlist? Rather than depicting language as a stand-alone instinct, the authors see it as the culmination of many different developments, some of which started long before our species appeared on the planet, such as tit-for-tat cooperation. The result is a delightful introduction to the language origins debate Chomsky's agnosticism on the matter not withstandingby two vetern debaters.
"This book is witty, opinionated and deeply clever, a wonderfulintroduction to one of the most controversial issues in the study ofmind." The New York Times BookReview (April 16, 2000)"Lingua ex Machina is the result of a fascinating and unlikely collaborationbetween two highly original thinkers—a linguist and a theoreticalneurophysiologist—who have spent their careers considering the evolution ofthe human mind from these very different perspectives. The result issomething halfway between a synthesis and a dialogue, that leads the readeron a challenging ride through some of the most interesting and controversialtopics in the science of mind. Terrence W. Deacon, BostonUniversity
" Lingua ex Machina is the result of a fascinating and unlikely collaboration between two highly original thinkers—a linguist and a theoretical neurophysiologist—who have spent their careers considering the evolution of the human mind from these very different perspectives. The result issomething halfway between a synthesis and a dialogue, that leads the readeron a challenging ride through some of the most interesting and controversialtopics in the science of mind. Terrence W. Deacon , BostonUniversity
In Lingua ex Machina, Calvin and Bickerton, in a refreshing spirit of courage and wonder, advance directly and insightfully into the amazing details of the hardest question of language.
Lingua ex Machina is the result of a fascinating and unlikely collaboration between two highly original thinkersa linguist and a theoretical neurophysiologistwho have spent their careers considering the evolution of the human mind from these very different perspectives. The result is something halfway between a synthesis and a dialogue, that leads the reader on a challenging ride through some of the most interesting and controversial topics in the science of mind.